Chronic Hyperglycemia, Independent of Plasma Lipid Levels, Is Sufficient for the Loss of β-Cell Differentiation and Secretory Function in the db/db Mouse Model of Diabetes

C57BL/KsJ db/db mice and their age-matched lean db/+ littermates (control) were taken from the Garvan Institute breeding colony, which was originally obtained from The Jackson Laboratories (Bar Harbor, ME). Animals were kept under conventional conditions with free access to water and standard pelleted food. All procedures were approved by the Garvan Institute/St. Vincent’s Hospital Animal Experimentation Ethics Committee, following guidelines issued by the National Health and Medical Research Council of Australia. To assess the effects of the duration of diabetes exposure on β-cell differentiation, islets from db/+ and db/db mice were studied at 10 and 16 weeks of age. Blood was collected from the tail tip of fed mice to determine blood glucose and plasma nonesterified fatty acid (NEFA), triglyceride, and insulin concentrations. Mice were anesthetized, and their islets were isolated with liberase RI (Roche) digestion of the pancreas. Islets were further separated with a Ficoll-Paque PLUS gradient (Amersham Biosciences, Uppsala, Sweden) and handpicked under a stereomicroscope. Immediately after collection, islets were used for extraction of RNA.

Total RNA was extracted from islets using RNeasy mini kit (Qiagen) according to manufacturer-suggested protocols. After quantification by spectrophotometry, 200 ng RNA was reverse-transcribed into cDNA in a 20-μl solution containing 10× reverse-transcribed buffer (200 mmol/l Tris-HCl, pH 8.4, and 500 mmol/l KCl), 5 mmol/l MgCl₂, 10 mmol/l dithiothreitol, 0.5 mmol/l dNTP, 0.5 μg of Oligo(dT)_12–18, 40 units of RNaseOUT recombinant ribonuclease inhibitor, and 50 units of SuperScript II RT (Invitrogen Life Technologies). Reverse-transcribed reactions were incubated for 50 min at 42°C and 15 min at 70°C and then chilled on ice. Residual RNA was digested with 2 units RNase H incubated for 20 min at 37°C. cDNA products were stored at −20°C.

Real-time PCR was carried out in a volume of 10 μl consisting of 1 μl cDNA, 1× LightCycler enzyme and reaction mix (SYBR Green I dye, TaqDNA polymerase, and dNTP; Roche), 1.5 mmol/l MgCl₂, and 600 nmol oligonucleotide primers (Proligo). Oligonucleotide primer sequences (Table 1) were designed with MacVector software (Oxford Molecular). All reactions were performed in a LightCycler (Roche) in which samples underwent 40 cycles of PCR with an annealing temperature of 55°C. Standards for each transcript were prepared in a conventional PCR and purified using High Pure PCR Product Purification kit (Roche). The value obtained for each specific product was expressed relative to the control gene (cyclophilin) for each sample (ratio of specific product to cyclophilin). These ratios were then expressed as a percentage of the ratio in db/+ islet extracts. PCR products were confirmed as the gene of interest on 1% agarose gels.

Starting at 10 weeks of age, db/db mice were either untreated or treated with phlorizin or bezafibrate for 2 weeks. Phlorizin (Sigma, St. Louis, MO) was dissolved in 1,2 propanediol and injected intraperitoneally twice a day at a dose of 0.4 g/kg. Bezafibrate (Sigma) was added to pulverized diet and thoroughly mixed to a concentration of 200 mg/100 g food (0.2%). After the 2-week treatment period, islets were isolated, and RNA was extracted for gene expression analysis by real-time RT-PCR (LightCycler; Roche) as described above.

Whole-blood glucose levels were measured with a portable glucometer (ACCU-CHEK II; Roche Diagnostics). Plasma insulin concentrations were measured using an ultrasensitive enzyme-linked immunosorbent assay kit (Mercodia AB). Plasma lipids were measured from samples collected in EDTA. Plasma NEFA were measured by a acyl-CoA oxidase-based colorimetric method (Wako Pure Chemical Industries). Plasma triglyceride were measured with a triglyceride assay kit (GPO Trinder; Sigma) using glycerol as standard.

Islets isolated from db/+ and untreated, phlorizin-treated, and bezafibrate-treated db/db mice were washed in Krebs-Ringer HEPES buffer (KRHB; containing 5 mmol/l NaHCO₃, 1 mmol/l CaCl₂, 2.8 mmol/l glucose, 10 mmol/l HEPES, and 10% FCS). Groups of five islets were preincubated for 30 min at 37°C in KRHB and then incubated for 1 h in KRHB containing 2.8 or 25 mmol/l glucose. Insulin was measured in an aliquot of the buffer by radioimmunoassay (Linco Research, St. Louis, MO) with rat insulin standard.

All results are presented as means ± SE. Statistical analyses were performed using unpaired Student’s t test or one-way ANOVA.

Untreated db/db mice displayed higher blood glucose, plasma NEFA, and plasma triglyceride levels at both 10 and 16 weeks of age compared with age-matched db/+ (control) mice (Fig. 1); and blood glucose levels tended to be higher, although not significantly (P = 0.087), at 16 weeks compared with 10 weeks of age. Plasma NEFA and triglycerides were elevated to similar levels in db/db mice at 10 and 16 weeks of age. Having established these metabolic indexes, we next sought to measure the expression of islet-associated transcription factors, specialized β-cell metabolism enzymes, ion channels/pumps, and other normally suppressed metabolism enzymes and stress response genes in islets from 10- and 16-week-old obese diabetic db/db and lean nondiabetic db/+ mice.

After normalization of the gene of interest to a control gene (cyclophilin), mRNA levels in db/db islets were quantitated as a percentage of db/+ (Table 2). Note, similar results were obtained comparing expression levels with another housekeeping gene, α-tubulin (not shown). The expression levels of several transcription factors important for pancreas and islet development and the maintenance of β-cell differentiation were significantly altered. mRNA levels of Nkx6.1, β-cell E-box trans-activator 2 (Β2/NeuroD), pancreatic duodenal homeobox-1 (PDX-1), and Pax6 were significantly reduced in islets from db/db mice. Interestingly, Nkx6.1mRNA levels were more severely affected than the other transcription factors tested, whereas PDX-1 and Pax6 showed only modest, and in the case of Pax6 delayed, reduction (Table 2). Β2/NeuroD mRNA levels were significantly reduced at 10 weeks and were further reduced at 16 weeks of age, showing a significant time-dependent effect (P < 0.05). Reduced Nkx6.1 and PDX-1 in db/db islets showed a similar, although nonsignificant, tendency for a time-dependent deterioration in mRNA levels. Not all transcription factors considered important for β-cell function were altered in db/db islets. Hepatocyte nuclear factor 1α (HNF1α) mRNA levels were unchanged in db/db mice at both ages. We also tested inhibitor of differentiation/DNA binding 1 (ID-1), a negative regulator of basic helix-loop-helix transcription factors normally expressed at low levels in islets (11) and known to be associated with dedifferentiation in other cell types (12). Through interactions with other non–basic helix-loop-helix proteins, ID-1 can also influence cell proliferation and apoptosis. ID-1 mRNA levels were significantly increased in db/db islets at 10 weeks and were increased further at 16 weeks (P < 0.01, 10- vs. 16-week-old db/db mice). Thus, the upregulation of ID-1 in islets of db/db mice was associated with the duration of diabetes. The role of peroxisome proliferator–activated receptor γ (PPARγ) in β-cells, if any, is poorly understood (13), but it is an important regulator of lipid and glucose homeostasis via well-characterized effects in peripheral tissues (14). The amount of mRNA for PPARγ was significantly increased in diabetic db/db mice at 10 weeks of age. Surprisingly, however, at 16 weeks of age, PPARγ mRNA was normal, which raises questions about its role in the β-cell dysfunction associated with diabetes.

Several genes involved in β-cell glucose metabolism were evaluated in db/db islets. The glucose transporter, GLUT2; the anaplerotic enzyme, pyruvate carboxylase; the rate-limiting enzyme of the glycerol-phosphate shuttle, mitochondrial glycerol phosphate dehydrogenase (mGPDH); and the enzyme responsible for the majority of β-cell glucose phosphorylation, glucokinase, were downregulated in islets of db/db mice. Downregulation of GLUT2 expression has been previously identified in islets of db/db mice (15) and in other models of diabetes (4,16), and here, reduced mRNA levels showed a significant time-dependent deterioration (Table 2). GLUT2, pyruvate carboxylase, and mGPDH were significantly reduced in db/db islets at 10 weeks and were reduced further at 16 weeks. In contrast, glucokinase mRNA levels were only affected with the more prolonged diabetes exposure at 16 weeks of age. Conversely, hexokinase I, an enzyme normally suppressed in β-cells, was increased in db/db mouse islets. Also increased were the mitochondrial lipid transport enzyme, carnitine palmitoyl transferase 1 (CPT-1), as well as the mitochondrial inner membrane proton carrier, uncoupling protein 2 (UCP-2), which has been shown to be capable of negatively regulating ATP production and GIIS (17,18).

The expression of several ion channels important for the stimulation of insulin secretion was also evaluated (Table 2). The mRNA levels of Kir6.2, the pore-forming subunit of the ATP-sensitive K⁺ channel and SERCA2b, a sarco/endoplasmic reticulum Ca²⁺-ATPase pump, were significantly reduced in db/db mice at 10 weeks and were reduced further at 16 weeks. mRNA levels of a related isoform, SERCA3, were unaltered at 10 weeks but were reduced in db/db islets at 16 weeks of age. This is consistent with findings of reduced SERCA protein and activity in islets of db/db mice (19). In contrast, the mRNA levels of VDCCα_1D, the neuro-endocrine isoform of the α-subunit of voltage-dependent Ca²⁺ channels, were unchanged in db/db mice at 10 and 16 weeks.

At the time points studied, the proportion of β-cells to non–β-cells was reported as equal or even increased in db/db compared with normal islets due to β-cell hyperplasia (20). Thus, the observed changes in gene expression are indicative of dedifferentiation of the β-cells. However, despite the downregulation of several important islet-associated transcription factors, insulin mRNA levels were preserved in diabetic db/db mice at 10 and 16 weeks of age (Table 2), consistent with results in 12-week-old db/db mice (21). mRNA levels for islet amyloid polypeptide (IAPP) and the central processing enzyme prohormone convertase 2 were significantly increased in 10- and 16-week-old db/db mice. Also increased in db/db islets were mRNA levels for genes implicated in the control of distal secretory processes that are commonly used by all secretogogues: Calcyclin, a Ca²⁺-binding protein, has been implicated in Ca²⁺-induced release of insulin (22); and synaptosomal-associated protein of 25 kDa (SNAP25), a t-SNARE protein, is thought to be an essential component of the exocytotic machinery (23), and its altered expression will likely impact insulin granule movement (24). Thus, in contrast to the downregulation of genes that confer the “glucose-sensing” ability of the β-cell, transcript levels of β-cell hormones and genes implicated in distal secretory mechanisms are better maintained or even increased.

The role of cellular stress and stress response mediators in the failure of β-cells in diabetes has been the subject of much recent attention (25–27). Activating transcription factor 3 (ATF-3) is a stress-inducible gene normally expressed at low levels in β-cells (28). mRNA levels of ATF-3 were markedly induced in islets of db/db mice. Similarly, mRNA levels of the antioxidant enzymes heme oxygenase-1 and glutathione peroxidase were increased in islets of db/db mice. Also increased was the cell-surface protein, Fas, which in β-cells can promote both apoptosis and proliferation (29). These results suggest that induction of stress gene expression is part of the adaptive response of β-cells exposed to hyperglycemia.

We next tested whether the changes in mRNA levels in db/db mice were more related to chronic hyperglycemia or hyperlipidemia. Hyperglycemia was selectively normalized in db/db mice by using phlorizin, an inhibitor of glucose reabsorption in the kidney. Phlorizin blocks the Na⁺/glucose cotransporter in the proximal tubules of the kidney causing glucosuria. Circulating hyperlipidemia was selectively normalized in db/db mice by using bezafibrate. The actions of bezafibrate to lower plasma lipids involve activation of PPARα and enhanced expression of genes involved in fatty acid uptake and oxidation in the liver (30).

As shown in Fig. 2, hyperglycemia was normalized in phlorizin-treated db/db mice (blood glucose levels were not different from db/+ mice); whereas, plasma lipid levels remained elevated (compared with db/+ mice) and were not different from untreated db/db mice. In bezafibrate-treated db/db mice, hyperlipidemia was reversed (plasma NEFA and triglyceride levels were not different from db/+ mice), whereas hyperglycemia was unaffected (blood glucose levels in bezafibrate-treated db/db mice were not significantly different compared with untreated db/db mice; Fig. 2). db/db mice displayed marked hyperinsulinemia (Fig. 2) as previously described (7). Interestingly, plasma insulin concentrations did not decrease in phlorizin-treated db/db mice despite the normalization of blood glucose (Fig. 2). Similarly, plasma insulin concentrations were not affected by lowering lipids (bezafibrate treatment).

At the end of the 2-week treatment period, mRNA levels were analyzed in islets isolated from db/+, untreated db/db, phlorizin-treated db/db, and bezafibrate-treated db/db mice. Strikingly, the normalization of blood glucose in phlorizin-treated db/db mice restored the abundance of the islet-associated transcription factors to the levels apparent in islets from control db/+ mice (Fig. 3); islet mRNA levels for Nkx6.1 and Β2/NeuroD were reduced in untreated db/db mice and were completely restored after phlorizin treatment. In contrast, lowering hyperlipidemia without affecting hyperglycemia (bezafibrate treatment) had no effect on islet transcription factor mRNA levels; mRNA levels for Nkx6.1 and Β2/NeuroD in bezafibrate-treated db/db mice were reduced compared with db/+ mice and not significantly different from untreated db/db mice (Fig. 3). ID-1 upregulation in db/db islets also correlated with the presence of hyperglycemia and was not related to lipidemia; ID-1 mRNA levels were normalized in phlorizin-treated db/db mice to control db/+ values, whereas levels in bezafibrate-treated db/db mice were increased to values similar to untreated db/db mice (Fig. 3). In contrast, the upregulation of PPARγ in db/db islets did not correlate with glycemia or lipidmia; mRNA levels were increased in the untreated, phlorizin-treated, and bezafibrate-treated groups of db/db mice (Fig. 3).

As shown in Fig. 4, the normalization of blood glucose in phlorizin-treated db/db mice reversed the changes in expression of glucose metabolism enzymes, whereas the normalization of plasma lipids was without affect; reduced mRNA levels for GLUT2, pyruvate carboxylase, and mGPDH in untreated db/db mice were completely normalized by lowering glucose (phlorizin) but were unaffected by lowering lipids (bezafibrate). A similar association was observed between blood glucose and the mRNA levels of the K⁺ channel subunit Kir6.2 and the Ca²⁺-ATPase pump SERCA2b; both were completely normalized by lowering glucose (phlorizin) but were unaffected by lowering lipids (bezafibrate). Increased mRNA levels of the Ca²⁺-binding protein calcyclin were also normalized by lowering glucose but were unaffected by lowering lipids. Of all the genes tested, only CPT-1, the rate-limiting mitochondrial lipid transport enzyme, was regulated by both glycemia and lipidemia in islets of db/db mice; the higher levels of islet CPT-1 mRNA in untreated db/db mice were significantly reduced toward normal in both phlorizin- and bezafibrate-treated db/db mice. In contrast, the higher levels of UCP-2 mRNA in db/db mice were not affected by the lowering of either glucose or lipid (Fig. 4). Thus, our studies show that changes in UCP-2 are disassociated with changes in insulin secretion (see insulin secretory assay data in Fig. 6), in conflict with evidence supporting a role of UCP-2 as a negative regulator of insulin secretion (17,18), but in agreement with a recent study (31), suggesting that moderate increase in UCP-2 does not impair GIIS.

As shown in Fig. 5, the increases in stress gene mRNA levels were reversed by 2-week treatment of db/db mice with phlorizin, strongly suggesting the specificity of hyperglycemia in eliciting the β-cell stress response in db/db mice. The normalization of blood glucose levels led to a reversal toward normal of the changes in ATF-3, heme oxygenase-1, glutathione peroxidase, and Fas expression. In contrast, normalization of plasma lipid levels was without affect on stress gene expression (Fig. 5).

In islets isolated from lean nondiabetic db/+ mice, insulin secretion was significantly increased from 2.8 to 25 mmol/l glucose (Fig. 6), indicative of robust GIIS. Insulin release at basal glucose (2.8 mmol/l) was higher in islets isolated from db/db mice compared with db/+ mice. However, GIIS was almost completely abolished in db/db islets (Fig. 6). Thus, the secretory dysfunction in islets of db/db mice was similar to that found in humans with type 2 diabetes. Strikingly, insulin secretion at 25 mmol/l glucose was restored in islets isolated from phlorizin-treated db/db mice but remained severely blunted in bezafibrate-treated db/db mice (Fig. 6). These data imply a causal relationship between chronic hyperglycemia and the loss of GIIS in db/db mice.

Here, we show using the db/db mouse model of type 2 diabetes that β-cell dysfunction is associated with a widespread disruption of gene expression with the induction of several normally suppressed genes and decreased expression of a panel of genes that optimize GIIS. This disruption in the pattern of gene expression further deteriorates over time, demonstrating the importance of diabetes duration in the loss of β-cell phenotype. The change in β-cell phenotype is reversed by phlorizin treatment, which normalized blood glucose but not hyperlipidemia, suggesting that hyperglycemia, independent of plasma lipid levels, is sufficient for the loss of β-cell differentiation and secretory function in db/db mice.

The findings of this study and others (4,32,33) greatly strengthen the hypothesis that a gradual loss of differentiation underlies the dysfunction of β-cells in diabetes. mRNA levels of several transcription factors important for the maintenance of β-cell differentiation (PDX-1, Β2/NeuroD, Nkx6.1, and Pax6) were reduced with time in db/db islets. Interestingly, expression of Nkx6.1, the most severely affected, lies downstream in the hierachy of transcription factor control of β-cell differentiation (34). Normal expression is restricted to β-cells of the adult pancreas, and its absence leads to profound inhibition of β-cell formation (34). The downregulation of Nkx6.1 and other potentially important islet transcription factors could contribute to the altered expression of genes essential for GIIS (35). The expression of multiple genes that confer the glucose-sensing ability of the β-cell, namely the glucose metabolism genes (GLUT2, pyruvate carboxylase, mGPDH, and glucokinase), and ion channels/pumps (Kir6.2, SERCA2b, and SERCA3) were downregulated. Furthermore, parallel increases were observed for a gene capable of inhibiting differentiation, ID-1, as well as other normally suppressed genes. In normal islets, ID-1 expression is low, but it is induced in isolated islets and β-cell lines cultured under high glucose conditions, with little effect in other cell types (11), and its forced expression can inhibit insulin promoter activity (36,37). These findings raise the possibility that increased ID-1 plays a role linking hyperglycemia to β-cell dedifferentiation in db/db mice.

Insulin mRNA levels were preserved in db/db islets, indicating that adaptation of β-cells is accompanied by abnormalities in the glucose-sensing machinery, whereas transcript levels of the β-cell hormone are better maintained. A similar phenomenon was observed in a rat partial pancreatectomy model with mild β-cell deficiency in which several glucose-sensing genes and islet transcription factors were significantly reduced, whereas insulin levels were only affected with more severe β-cell depletion (32). In the db/db mouse, hyperglycemia and hyperlipidemia appear as a result of the failure of their β-cells to meet the secretory demand generated by increasing obesity and insulin resistance. In the partial pancreatectomy rat, increased secretory demand is imposed by β-cell deficiency in the absence of obesity or hyperlipidemia. In both models, the persistent hyperglycemia leads to a gradual change in β-cell phenotype, with secretory defects and changes in islet gene expression that are remarkably similar with a few notable exceptions. Thus, IAPP was increased in islets of db/db mice, whereas in partial pancreatectomy rats, it was increased in animals with low-level hyperglycemia and reduced in animals with more severe diabetes. LDH-A (lactate dehydrogenase A subunit) was not significantly altered in db/db islets, unlike in partial pancreatectomy islets, which showed a severalfold induction (32). Species differences in islet susceptibility to stress will likely influence the severity of β-cell dedifferentiation under conditions of increased secretory demand (38), with constitutive differences in defense mechanisms likely to play an influential role (39,40). It is also important to recognize that leptin signaling, which may impair insulin secretion (41) and induce β-cell apoptosis (42), is defective in db/db mice. The progression to diabetes in ZDF rats is associated with a similar global alteration in β-cell gene expression (4), even though the ZDF rat carries an inherited defect in insulin gene transcription (43), which is presumably not carried in the db/db mice. Taken together, these studies in animal models highlight the importance of the severity and duration of hyperglycemia as critical factors in the deterioration of the β-cell phenotype. Importantly, similar gene expression alterations have been recently identified in islets from human type 2 diabetic patients (44).

The loss of β-cell function in humans has been linked to increases in fasting plasma glucose levels, even in the nondiabetic range (45). Because of its similarities with human type 2 diabetes, the C57BL/KsJ db/db mouse model seems particularly suitable to address the question of the relative influence of hyperglycemia versus hyperlipidemia in the progressive loss of β-cell function. Using independent variation of circulating glucose on the one hand and fatty acids and triglyceride on the other, we show here that hyperglycemia independent of hyperlipidemia plays a critical role in the loss of β-cell differentiation in db/db mice. This is in accordance with results from a study in ZDF rats, in which high glucose levels, and not high plasma triglycerides, appear to be associated with the decreased insulin gene expression and increased islet triglyceride content in this model (46). However, other studies in ZDF rats have proposed a causative role of high fatty acids in the development of β-cell dysfunction (47), and others have shown that fatty acids can directly lead to alterations in β-cell gene expression broadly consistent with the dedifferentiation observed here (48–50,51). Given the likelihood that intracellular lipid pathways play an important role in the β-cell dysfunction of diabetes (52), it is possible that hyperglycemia in vivo leads to alterations in β-cell lipid metabolism that recapitulate the effects on gene expression of fatty acid addition in vitro. Additionally, circulating fatty acids might exert synergistic toxic effects in concert with hyperglycemia (53,54).

In conclusion, the finding in C57BL/KsJ db/db mice of time-dependent alterations in the pattern of islet gene expression is consistent with the hypothesis that a gradual loss of differentiation contributes to the dysfunction of β-cells in diabetes. The validity of this relationship is further strengthened by studies in phlorizin-treated db/db mice in which the restoration of GIIS occurs in parallel with the normalization of the majority of the changes in islet gene expression. Phlorizin treatment normalized blood glucose without affecting plasma NEFA or triglyceride levels, strongly suggesting that hyperglycemia, independent of hyperlipidemia, plays a causal role in the progressive loss of β-cell differentiation and secretion in db/db mice. Taken with other studies (4,32,55,56), it is apparent that there are similarities in the abnormal patterns of gene expression that accompany β-cell dysfunction in several rodent models of diabetes, suggestive of common mechanisms. Similar mechanisms may be responsible for β-cell dysfunction in human type 2 diabetes (44).

This work was supported by a grant from Diabetes Australia Research Trust. T.J.B. is the recipient of a grant from the National Health and Medical Research Council (NHMRC) of Australia. D.R.L. is a recipient of a Career Development Award from the NHMRC and a grant from the NHMRC of Australia.

We thank Ebru Gurisik, Zehra Elgundi, and Eric Schmied for islet isolation and Greg Cooney and Ted Kraegen for advice and support.