Spontaneous Diabetes in Hemizygous Human Amylin Transgenic Mice That Developed Neither Islet Amyloid nor Peripheral Insulin Resistance

Lines of human amylin transgenic mice were developed in a FVB/n genetic background through pronuclear microinjection of the following transgenic construct into fertilized oocytes.

The transgene construct consisted of a RT-PCR–generated human amylin cDNA under the regulatory control of the rat insulin-II promoter. Two lines (L13 and L9) were found to develop spontaneous diabetes. One of these, L13, was selected for intensive characterization and is that which is mainly described in this paper. Southern blotting of genomic DNA demonstrated that L13 hemizygous mice have ∼36 ± 7 copies of the transgene. Pancreas-specific expression of human amylin was confirmed by Northern blotting analysis of RNA transcripts from various organs.

[^25,28,29Pro]human amylin transgenic mice (L44) were generated in the FVB/n strain, as was L13, using a similar transgene construct with the wild-type human amylin cDNA replaced by a [^25,28,29Pro]human amylin cDNA. The modified human amylin variant has proline residues at positions 25, 28, and 29. This nonamyloidogenic human amylin transgenic line was generated to provide a control to monitor the nonspecific effects of overexpression of a peptide within islet β-cells. Pancreas-specific expression of [^25,28,29Pro]human amylin was confirmed by Northern blotting and immunohistochemical analysis (results not shown).

Hemizygous L13 and L44 male animals were backcrossed with wild-type FVB/n females. Hemizygous male offspring were selected for this study, and nontransgenic male littermates were used as controls. Animals from L13 were divided into three groups based on weekly nonfasted blood glucose reading: pre-diabetic, mid-diabetic, and end-diabetic phases. Control animals and those from L44 were analyzed at similar time points. L13 animals that were normoglycemic at 3 and 4 weeks of age and those that had been hyperglycemic for 5 consecutive weeks were categorized into the pre- and mid-diabetic groups, respectively. End-diabetic animals were at the terminal stages of diabetes and displayed loss of exploratory behavior, relative immobility, and failure to groom. All animal experiments were approved by the institution's animal ethics committee.

After an overnight fast, glucose was administered into the intraperitoneal cavity (1g/kg body wt). Blood glucose was determined using a glucose meter (Roche Diagnostics) before injection and at 5, 15, 30, 60, 90, and 120 min after injection. The procedure for the insulin tolerance test (ITT) was identical to that of the glucose tolerance test (GTT) with the exception that soluble insulin (Novo Nordisk Pharmaceuticals) was injected into the intraperitoneal cavity (1 mU/g body wt) after a 4-h fast.

Soleus muscles were first preincubated for 30 min in Krebs-Henseleit bicarbonate buffer (KHB) supplemented with 8 mmol/l glucose and 32 mmol/l d-mannitol with or without 2 mU/ml insulin. Muscles were then rinsed and incubated in KHB buffer with 2 μCi/ml [³H]2-deoxyglucose (ICN Pharmaceuticals), 0.3 μCi/ml [¹⁴C]d-mannitol (ICN Pharmaceuticals), and 2 mmol/l sodium pyruvate for 30 min. All incubations were performed under 95% O₂/5% CO₂ with gentle agitation at 30°C followed by processing. Muscles were boiled (1 mol/l NaOH, 10 min), and muscle extracts were neutralized with 5 mol/l HCl. Radioactivity of muscle extracts and incubation media was determined in duplicates for dual labels.

Total protein was extracted from frozen pancreatic tissue by acid-ethanol (1.5% HCl, 75% ethanol, and 20 μl/mg pancreatic wt). After overnight incubation (4°C on shaking), homogenates were centrifuged, supernatant was removed, and pancreatic extracts were neutralized (pH 7) with 1 mol/l Tris base. Total pancreatic protein was quantified using BCA assay (Sigma-Aldrich).

Immunoreactive insulin (Mercodia) and human amylin (Linco Research) were measured in plasma and pancreatic extracts using enzyme-linked immunosorbent assays (ELISAs). Total amylin (Peninsula Laboratories) and glucagon (Linco Research) were determined using radioimmunoassay (RIA) kits. Somatostatin was measured using an enzyme immunosorbent assay kit (Phoenix Pharmaceutical). All kits were used according to manufacturers’ instructions.

Pancreatic tissues were fixed in 4% paraformaldehyde, embedded in paraffin, and sectioned at 7 μm. Serial sections were stained with Congo red. Amyloid-positive pancreatic sections were included with every Congo red staining run as positive controls. Sections were also immunostained with anti-glucagon (ICN Pharmaceuticals), anti-insulin (Dako), anti-amylin (in-house derived), and/or anti-somatostatin (Santa Cruz Biotechnology) antibodies. Appropriate second antibodies conjugated to fluorescein isothiocyanate, Rhodamine Red, or Cy5 (Jackson ImmunoResearch Laboratories) were used for the covisualization of α-, β-, and δ-cells within the islets (TCS SP2; Leica Microsystems). Cross-sectional islet areas and total pancreatic section areas were determined in hematoxylin-eosin sections under light microscopy using image analyses software (AxioVision 2; Carl Zeiss). β-Cell mass was estimated using a previously described method (24).

For immunofluorescence staining for cleaved caspase-3, pancreatic tissues were immediately frozen and embedded in OCT. Eight-micrometer sections were cut and initially blocked with 10% (vol/vol) normal goat serum for 2 h and then incubated with rabbit anti–cleaved caspase-3 (Cell Signaling) followed by detection using goat anti-rabbit Rhodamine Red-X–conjugated secondary antibody (Jackson ImmunoResearch Laboratories). Stained sections were viewed and photographed using inverted-phase fluorescence microscopy (Nikon).

Pancreases were cut into 1-mm³ pieces and fixed in 2% glutaraldehyde and 0.1 mol/l Sorenson's phosphate buffer (pH 7.2, 4°C, overnight). Specimens were postfixed with 0.1% osmium tetroxide and 0.1 mol/l Sorenson's phosphate buffer (pH 7.2, 1 h), followed by dehydration in increasing concentrations of ethanol solutions. Specimens were then infiltrated with a 1:1 mixture of epoxy resin and 100% ethanol (1 h, room temperature) followed by 100% epoxy resin infiltration overnight. Specimens were then embedded with fresh resin and cured (48 h, 60°C). Ultrathin sections were collected onto 200-mesh copper grids and double stained with 2% (wt/vol) aqueous uranyl acetate followed by lead citrate before viewing (CM-12; Phillips).

All data are expressed as means ± SE. Independent sample t tests were performed using SPSS v11.5. Two-way ANOVA, general linear model, and post hoc Tukey's honestly significant difference analyses were performed using Statistica 6.0. P values <0.05 were accepted as statistically significant.

Hemizygous human amylin transgenic mice developed hyperglycemia from 6 weeks after birth (Fig. 1A), from which time blood glucose values were elevated compared with controls (P < 0.01). Blood glucose remained significantly elevated throughout the entire 31-week monitoring period (Fig. 1A). Ninety-six percent of the transgenic mice developed diabetes during this monitoring period, with a mean onset time of 11.2 weeks (Fig. 1B). Intraperitoneal GTTs (IPGTTs) performed in fasted animals at the mid-diabetic stage (Fig. 2A) demonstrated impaired glucose tolerance and elevated fasting blood glucose (Fig. 2A) in human amylin transgenic mice (P < 0.05). The lack of glucose-stimulated insulin secretion during GTT present in human amylin transgenic mice (t₀ 29.9 ± 15.1 cf. t₁₅ 39.1 ± 18.5 pmol/l; NS), is consistent with impaired β-cell function. In contrast, we detected a positive insulin response to glucose in the control nontransgenic mice (t₀ 26.2 ± 4.3 cf. t₁₅ 128.6 ± 55.7 pmol/l; P < 0.05). Neither fasting hyperglycemia nor impaired glucose tolerance was present in [^25,28,29Pro]human amylin transgenic mice (Fig. 2A), wherein blood glucose concentrations remained equivalent to control levels at all times during the observation period (from birth to 24 weeks of age) (Fig. 1A).

Systemic insulin administration (intraperitoneal ITT [IPITT]) at the mid-diabetic stage (Fig. 2B) showed that human amylin transgenic mice were more responsive to insulin, consistent with the absence of peripheral insulin resistance (P < 0.01). Absence of peripheral insulin resistance in human amylin transgenic mice was confirmed by ex vivo glucose uptake analysis in isolated, stripped soleus muscles (Fig. 2C), which showed similar rates of glucose uptake between transgenic and nontransgenic animals in either the basal or the insulin-stimulated states. This finding is consistent with that from the ITT analysis and confirms that peripheral insulin resistance is not implicated in the pathogenesis of diabetes in this model.

Plasma and pancreatic hormone levels were determined at the pre-diabetic, mid-diabetic, and end-diabetic phases and at equivalent times in nondiabetic mice. Elevated pancreatic human amylin (P < 0.001) and, hence, total amylin (P < 0.01) was present before diabetes developed (Table 1). The development of diabetes in these mice (at the mid- and end-diabetic stages) was associated with concomitant decrements in pancreatic β-cell hormones: insulin (P < 0.01) and total amylin (P < 0.001). By contrast, in [^25,28,29Pro]human amylin transgenic mice, which did not develop diabetes, the pancreatic total amylin and insulin levels were significantly higher than in nondiabetic animals at all three phases (Table 1). This suggests that overexpression of human amylin results in β-cell destruction and impairment of β-cell function, which is consistent with our histological evidence of β-cell loss during diabetes progression (Fig. 3A–F). The decrease in β-cell mass in human amylin transgenic mice (Fig. 3H) confirms that the loss of β-cell mass contributes to the mechanism of diabetes in these human amylin mice, at least in part. Pancreatic glucagon was also elevated at the end-diabetic stage in transgenic animals (P < 0.05), whereas no significant alteration in pancreatic somatostatin was observed (Table 1).

Plasma hormone concentrations also showed responses similar to those of their respective pancreatic contents (Table 2). As expected, human amylin was detected in plasma of transgenic animals only (at the pre-, mid-, and end-diabetic phases; P < 0.05 in each case). At the end-diabetic stage, plasma insulin was decreased in transgenic mice (P < 0.05); by contrast, increased plasma glucagon was observed in transgenic animals at this stage (P < 0.001). Plasma total amylin and somatostatin could not be determined because of insufficient sample volumes.

The specificity of the amylin assay systems used in this study enabled us to calculate estimated values for pancreatic mouse amylin by subtraction of values for pancreatic human amylin from corresponding measurements of total pancreatic amylin (Table 1). Both the total amylin RIA and the human amylin ELISA systems were validated to confirm that antibodies in each recognized the appropriate amylin species. Our validation experiments demonstrated that antibodies used in the ELISA specifically detected human amylin and did not cross-react with mouse amylin (Fig. 4A). Antibodies used in the RIA system were 100% cross-reactive with human amylin, murine amylin, and [^25,28,29Pro]human amylin, thus recognizing human and murine amylin at equimolar concentrations but detecting [^25,28,29Pro]human amylin at lower sensitivity compared with the former (Fig. 4B). Antibodies in both assay systems also required that amylin be amidated at its COOH terminus for recognition.

Elevated levels of mouse amylin were observed in transgenic animals before diabetes onset (P < 0.01). Thus, overexpression of human amylin may have contributed to the elevated production of endogenous amylin. However, decreased pancreatic mouse amylin concentrations were present at the mid- and end-diabetic stages in human amylin transgenic mice, consistent with the progressive decline in β-cell function that occurred during the course of diabetes progression in these animals.

Immuno-labelling of α-, β-, and δ-cells at the end-diabetic stage showed markedly decreased β-cell numbers in transgenic islets, consistent with decreased numbers of functional β-cells and decreased insulin production in these animals (Fig. 3F). Electron microscopic studies, in which at least 30 islets were examined from each animal, further confirmed the presence of typical β-cell secretory granules in nondiabetic islets (Fig. 5G) that, by contrast, were largely lacking in those β-cells (defined as endocrine cells present in the body of the pancreas but lacking typical α- or δ-cell granules) remaining in the body of the pancreas in human amylin transgenic mice at the end stage (Fig. 5H).

β-Cells from end-stage diabetic mice frequently showed invagination of their nuclear membranes and chromatin margination, both characteristics of cells undergoing apoptosis (Fig. 5H). In addition, apoptosis was also detected and quantified by immunofluorescence staining for cleaved caspase-3 (Fig. 5I and J). About 60 islets from each study group (20 islets per animal and 3 animals per group) were examined, and the total number of cleaved caspase-3–positive cells was determined. Numbers of apoptotic islet cells were found to be significantly increased in the transgenic mice at the mid-diabetic stage (Fig. 5I) compared with matched nondiabetic, nontransgenic mice (Fig. 5K). These findings are consistent with in vitro studies demonstrating human amylin-evoked apoptosis in cultured β-cells (15). In contrast, equivalent β-cell loss was never observed in islets of [^25,28,29Pro]human amylin transgenic mice, although alterations in the relative positioning of α- and β-cells similar to those demonstrated in human amylin transgenic mice were present (Fig. 3G). This latter observation indicates that the modifications in islet cell distribution were insufficient in themselves to cause β-cell loss or defective insulin secretion.

Examination under polarized light of Congo red-stained sections revealed the absence of green birefringence at all three stages of the diabetic process in human amylin transgenic mice (Fig. 5A–F). Moreover, Congo red–stained islet amyloid was never detected in [^25,28,29Pro]human amylin transgenic mice (data not shown). These findings indicate that Congo red–stainable islet amyloid is not necessary for and cannot have played a role in the development of diabetes in this model. We note that islet amyloid–positive sections as positive controls were included with every Congo red staining run.

Here, we report characterization of a hemizygous line of human amylin transgenic mice (L13) that synthesizes human amylin in its pancreatic islets and demonstrates increased release of human amylin into the plasma (Table 2). In the current study, 96% of L13 transgenic mice developed diabetes (sustained hyperglycemia, Fig. 1A) and ∼95% were diabetic by 22.5 weeks of age; by contrast, corresponding rates of diabetes in nontransgenic littermates were 0% (Fig. 1). Diabetes in L13 animals is associated with decreased β-cell function (Table 1). These animals have never been found to be hyperinsulinemic (Table 2) but show significant impairment of glucose tolerance (Fig. 2A) in the absence of peripheral insulin resistance (Fig. 2C) or islet amyloid formation (Fig. 5A–F).

L13 transgenic mice had unaltered levels of pancreatic and plasma insulin during the pre-diabetic stage (Tables 1 and 2). Despite the lack of significant differences in plasma insulin between transgenic and nontransgenic mice at the mid-diabetic stage (Table 2), a fourfold decrease in pancreatic insulin was present in transgenic mice during this period (Table 1). At the end-diabetic stage, however, >17-fold decreases in pancreatic and plasma insulin were present (Tables 1 and 2). The progressive development of β-cell dysfunction over the diabetic process was paralleled by a progressive loss of insulin immunoreactive cells between the mid- (Fig. 3D) and end-diabetic stages (Fig. 3F) in transgenic islets, indicating progressive falls in the number of functional β-cells in hemizygous L13 mice. The disappearance of islet β-cells with concomitant development of impaired insulin secretion is thus responsible for the observed diabetic phenotype in this model.

Lines of [^25,28,29Pro]human amylin transgenic mice were also constructed to address the possible nonspecific causation of diabetes by overexpression of protein within islet β-cells. The aggregation potential of [^25,28,29Pro]human amylin has been eliminated by substitution of three prolyl residues at positions 25, 28, and 29 within the wild-type human amylin molecule. Mice expressing the [^25,28,29Pro]human amylin transgene in their islet β-cells do not develop either islet β-cell disappearance or insulin-deficient diabetes (Fig. 3G), demonstrating that the insertion of this transgene and its expression in islet β-cells of a nonfibrillogenic amylin variant is not in itself sufficient to cause β-cell destruction. Given that [^25,28,29Pro]human amylin mice expressing a nonaggregating amylin variant do not develop diabetes (Fig. 2A) or islet β-cell degeneration (Fig. 3G), findings presented here support the idea that the development of diabetes in mice expressing wild-type human amylin (L13) is caused specifically by the expression of a fibrillogenic amylin variant in islet β-cells rather than simply a nonspecific event caused by overexpression of any protein from this transgene in this location. It thus appears that the amyloidogenic property of human amylin is necessary for its ability to cause β-cell disappearance and diabetes, just as it is for its in vitro cytotoxicity (25,26).

Human amylin transgenic mouse models have previously been investigated by other laboratories (16–19). Phenotypic observations reported in those mice have been wide ranging; these have varied from a complete lack of spontaneous diabetes (16–18) to the appearance of severe early-onset diabetes in one line (19). The data reported here are closest to those reported by Janson et al. (19) and show the presence of early-onset diabetes without the formation of detectable islet amyloid. However, our findings were demonstrated in hemizygous animals, whereas those of Janson et al. (19) were reported in homozygous mice, although it has yet to be determined whether these two lines have comparable human amylin expression levels. Differences in amylin expression levels are one possible explanation for phenotypic variation between these strains of transgenic mice.

Reported phenotypic differences between strains of human amylin transgenic mice might also be attributable, at least in part, to the genetics of the background strain in which they were generated. Such transgenic mice have mainly been developed in two genetic backgrounds, FVB/n, in the current study and that of Janson et al. (19), and C57BL/6J (16,18). Severe spontaneous diabetes has previously been reported in homozygous individuals from a line generated in the FVB/n background (19) but has reportedly been absent in C57BL/6J-based strains (16,18). The influence of genetic background on phenotypic outcomes is well recognized. For example, Haluzik et al. (2004) reported the effect of the ob/ob mutation in both FVB/n and C57BL/6J backgrounds. Mice from the FVB/n strain were severely hyperglycemic, had lower triglyceride clearance, and were more insulin resistant than their C57BL/6J counterparts (27). This implies that the FVB/n background could be more susceptible to abnormal lipid and glucose metabolism and thus might have a greater predisposition to the development of diabetes. The L13 transgenic mouse model reported here was generated in the FVB/n background and demonstrated development of spontaneous diabetes, consistent with this hypothesis.

Previous in vitro studies are also consistent with the idea that the cytotoxicity of amyloid-forming amylin peptides, such as human amylin per se, is conferred by soluble species rather than mature, aggregated fibrils (2,25,28). Activation of an apoptotic JNK1/c-Jun/caspase-8/caspase-3 pathway is a related downstream mechanism by which amylin-induced destruction of islet β-cells occurs in vitro (29–31). Here, we have shown that caspase-3 activation also occurs in pancreatic islets of L13 hemizygous mice but not in the islets of their nontransgenic littermates; both findings are consistent with β-cell destruction through such an apoptotic pathway in these transgenic mice. Although both non–β-cells and β-cells could theoretically have contributed to our analysis of islet cell apoptosis by activated caspase 3 immunofluorescence, we believe that it was mostly due to β-cells because of the known β-cell targeting properties of the human amylin construct used herein and the demonstrated loss of β-cells from the islets of diabetic transgenic animals, whereas non–β-cells (glucagon- and somatostatin-containing) were relatively increased in the islets of late-diabetic animals (Fig. 3F). In addition, transmission electron microscopy (TEM) findings reported herein also provide structural evidence for the occurrence of β-cell apoptosis in these transgenic mice (Fig. 5H), consistent with other reports (2,15,32).

Despite the overexpression of human amylin in pancreatic islet β-cells, Congo red–stainable islet amyloid did not contribute to the pathogenesis of diabetes in L13 transgenic mice (Fig. 5A–F). However, we cannot exclude the possibility of a presence of small human amylin aggregates/oligomers that is not detectable by Congo red staining. An absence of islet amyloid has previously been reported in another line of human amylin transgenic mice that develop spontaneous diabetes (19). The greater than twofold increase in pancreatic mouse amylin observed in L13 mice (Table 1) may play a role in the lack of islet amyloid formation in these mice. Westermark et al. (2000) reported that mice expressing human amylin without murine amylin (hA^+/+/mA^−/−) developed extracellular fibrillar amyloid deposits by 12 months of age, compared with transgenic human amylin mice (hA^+/+/mA^+/+), which had developed them by 16 months of age. The earlier onset of islet amyloid formation in hA^+/+/mA^−/− mice compared with hA^+/+/mA^+/+ mice was reportedly interpreted as consistent with an inhibitory effect of mouse amylin on fibril formation (33).

In summary, islet-specific overexpression of human amylin in L13 mice caused diabetes associated with degeneration and loss of islet β-cells with progressive impairment of insulin secretion. The disease mechanism in these mice is independent of both large islet amyloid deposits and peripheral insulin resistance. Overexpression of both human amylin and [^25,28,29Pro]human amylin caused altered islet-cell type distribution, consistent with prior reports (34). The pathophysiological relevance of this alteration is unclear, but it may be a nonspecific effect of overexpression of a protein in β-cells and is unlikely in itself to be causative of impaired insulin secretion. When combined, these two lines of transgenic mice form a composite model that may well have utility in the search for compounds with the pharmacological properties of ameliorating or suppressing human amylin–mediated β-cell death and type 2 diabetes evoked by human amylin–mediated mechanisms.

W.P.S.W. has received doctoral scholarships from the Foundation for Research, Science and Technology, ViaLactia Biosciences (New Zealand), and Protemix. This work has received support from the Endocore Research Trust; the Health Research Council of New Zealand; the Foundation for Research, Science, and Technology of New Zealand; Lottery Health (New Zealand); the Maurice and Phyllis Paykel Trust; and the University of Auckland.

We thank B. Davy, X.L. Li, L. Munoz, J. Ross, and V. Tintinger for outstanding technical assistance; D. Brunton for statistical advice; and M. Broadhurst and P. L'Huillier for generation of the transgenic mice.