Permanent neonatal diabetes in INS(C94Y) transgenic pigs.

Mutations in the insulin (INS) gene may cause permanent neonatal diabetes mellitus (PNDM). Ins2 mutant mouse models provided important insights into the disease mechanisms of PNDM but have limitations for translational research. To establish a large animal model of PNDM, we generated INS(C94Y) transgenic pigs. A line expressing high levels of INS(C94Y) mRNA (70-86% of wild-type INS transcripts) exhibited elevated blood glucose soon after birth but unaltered β-cell mass at the age of 8 days. At 4.5 months, INS(C94Y) transgenic pigs exhibited 41% reduced body weight, 72% decreased β-cell mass (-53% relative to body weight), and 60% lower fasting insulin levels compared with littermate controls. β-cells of INS(C94Y) transgenic pigs showed a marked reduction of insulin secretory granules and severe dilation of the endoplasmic reticulum. Cataract development was already visible in 8-day-old INS(C94Y) transgenic pigs and became more severe with increasing age. Diabetes-associated pathological alterations of kidney and nervous tissue were not detected during the observation period of 1 year. The stable diabetic phenotype and its rescue by insulin treatment make the INS(C94Y) transgenic pig an attractive model for insulin supplementation and islet transplantation trials, and for studying developmental consequences of maternal diabetes mellitus.

Heterozygous mutations of the insulin (INS) gene have been identified as a cause of permanent neonatal diabetes mellitus (PNDM) in humans (1). Ins2C96Y (Akita) and Munich Ins2C95S mutant mice have development of hyperglycemia early in life, reduced insulin secretion after glucose challenge, ultrastructural β-cell alterations, and progressive β-cell loss (2,3). These models contributed substantially to the clarification of disease mechanisms induced by misfolded (pre)proinsulin, including impaired trafficking of normal proinsulin by formation of high-molecular-weight complexes with mutant (pro)insulin (4), accumulation of misfolded insulin in the endoplasmic reticulum (ER), and ER stress (5). ER stress triggers β-cell apoptosis and diabetes mellitus when it exceeds a certain level (6). Although there is increased degradation of misfolded insulin molecules (7), β-cell failure finally occurs.

Whereas Ins2 mutant mouse models provided important mechanistic insights, they have limitations for translational studies. To establish a large animal model of PNDM, we generated transgenic pigs expressing a mutant porcine INS gene (INSC94Y) that resembles the human INSC96Y mutation and the mutation of the Akita mouse.

RESEARCH DESIGN AND METHODS

Generation of INSC94Y transgenic pigs.

All animal experiments were performed according to the German Animal Welfare Act with permission of the responsible animal welfare authority. The INSC94Y expression vector including the porcine INS gene with the G→A transition leading to a Cys→Tyr exchange at amino acid position 94 as well as its essential regulatory elements and a neomycin selection cassette (neo) (Fig. 1) were nucleofected into male primary porcine fetal fibroblasts, and pools of stable nucleofected cell clones were used for nuclear transfer to generate transgenic founder pigs (8). Genotyping was performed by PCR using the transgene-specific primers 5′-CAG CTG TGC TCG ACG TTG TC-3′ and 5′-GAG TCA ACT AGT CCT CAG AAG AAA G-3′, and Southern blot analysis using a 32P-CTP–labeled probe specific for the neo cassette (Fig. 1). For transgene expression studies, total RNA was extracted from pancreatic tissue (RNeasy Kit; Qiagen), DNaseI-treated (Roche), and reverse-transcribed with SuperScriptII (Life Technologies) using random hexamer primers. INSC94Y and wild-type INS cDNAs were amplified by PCR using insulin-specific primers 5′-TTT TTA TCC GCA TTC TTA CAC GG-3′ and 5′-ATC TTC CTC AGC TCC TTC CAG G-3′, and their ratio was determined by next-generation sequencing of RT-PCR amplicons (Genome Analyzer IIx; Illumina; >10,000 reads per sample).

FIG. 1.

Generation of diabetic INSC94Y transgenic (tg) pigs (expression construct, Southern blot, glucose levels, transgene expression). A: Expression cassette including the porcine INS gene with the Cys→Tyr exchange at amino acid position 94, essential regulatory elements, and a neomycin selection cassette (neo). Restriction site of BamHI and binding site of the probe (bar) used for Southern blot analysis are indicated. B: Southern blot analysis of BamHI digested genomic DNA from INSC94Y tg pigs and littermate control animals (wt) showing different integration sites in founders 9725, 9726, 9727, 9728, 9745, 9746, and 9747. Tg offspring (1334, 1336, 1340, 1341) of founder 9747 show the same integration pattern as founder 9747, demonstrating a single integration site. C: Fasting blood glucose levels of INSC94Y tg founder boars showing hyperglycemia in founder 9747 progressively deteriorating over time, whereas blood glucose levels of the six other founders remain within the reference range (70–115 mg/dL, depending on the laboratory); arrow indicates start of insulin therapy. D: Quantification of INSC94Y and wt INS transcripts in pancreatic tissue of INSC94Y tg pigs by next- generation sequencing of RT-PCR amplicons. Founder 9747 shows at least five-fold higher expression of the mutant INSC94Y compared with the other six founders (F0) and similar expression compared with its F1 offspring (F1; n = 3). Inset: RT-PCR products of INSC94Y/INS transcripts in pancreatic tissue of all INSC94Y tg founders (left) and founder 9747, as well as its offspring (right). M, pUC Mix Marker; gDNA, genomic DNA; H2O, Aqua bidest.

Animal husbandry, in vivo parameters, and insulin treatment.

Animals were housed under controlled conditions, had a once-daily feeding regimen, and had free access to water. Body weight was recorded regularly. Random and fasting blood glucose levels were determined regularly using a Precision Xceed glucometer and Precision XtraPlus test stripes (Abbott). Intravenous glucose tolerance, homeostasis model assessment of β-cell function, and homeostasis model assessment of insulin resistance were determined as described previously (9,10). Continuous glucose monitoring was performed using the Guardian REAL-Time System (Medtronic) whereby a glucose sensor was inserted subcutaneously behind the ear and data were transferred via a connected transmitter to an external computer. Plasma insulin levels were determined by radioimmunoassay (Millipore). Insulin-treated pigs received a combination of long-acting insulin (Lantus; Sanofi) and short-acting insulin (NovoRapid; NovoNordisk). Blood glucose levels were determined once or twice daily using a glucometer to control treatment.

Necropsy, pancreas sampling, quantitative stereological analyses, and electron microscopy.

The 8-day-old and 4.5-month-old pigs were killed and subjected to routine necropsy. Selected organs were weighed. Pancreas samples for quantitative stereological analyses were chosen by systematic random sampling and were routinely processed for paraffin histology (9). β-cells and non-β-cells were visualized as described (9) using a polyclonal guinea pig antiporcine insulin antibody or polyclonal rabbit anti-human glucagon, somatostatin, and pancreatic polypeptide antibodies, respectively. The volume densities and the total volumes of β-cells and non-β-cells were determined as described previously (9). For transmission electron microscopy of β-cells, pancreas samples were fixed in 6.25% glutaraldehyde in Sorensen phosphate buffer (pH 7.4) for 24 h and processed as described (3). The left kidney was fixed by orthograde vascular perfusion with 4% formaldehyde. Six specimens of the renal cortex obtained by systematic random sampling (11) were processed for plastic embedded light microscopy (11) and transmission electron microscopy as described previously (11). The mean glomerular volume was determined from the mean glomerular profile area, as described (11). The true harmonic mean thickness of the glomerular basement membrane (GBM) was determined by the orthogonal intercept method (11). Tibial nerve fascicles were immersed in 2.5% glutaraldehyde in Soerensen phosphate buffer (pH 7.4) for 1 h and postfixed in 2% OsO4 for 2 h. After a graded alcohol series, they were embedded in epoxy resin. Semithin sections (0.5 μm) were stained with azur II-methylene blue-safranin. Some nerve fascicles were prepared for fiber teasing as reported previously (12).

Statistics.

All data are presented as means ± SEM unless indicated otherwise. Longitudinal data of blood glucose levels and body weight gain were statistically evaluated by ANOVA (PROC MIXED; SAS 8.2), taking the fixed effects of group (transgenic versus control) and age as well as the random effect of animal into account (9). For other parameters, the two-tailed unpaired Mann-Whitney U test in combination with an exact test procedure (SPSS 19.0) or two-tailed unpaired Student t test was used unless indicated otherwise. P < 0.05 was considered significant.

RESULTS

A total of 562 embryos reconstructed from INSC94Y nucleofected cell clones were transferred to five recipient gilts, resulting in three pregnancies, two of which went to term. Southern blot analysis revealed that all seven offspring were INSC94Y transgenic, with different transgene integration patterns (Fig. 1). Founder boar 9747 exhibited elevated fasting blood glucose levels at the age of 85 days, further increasing with age. The boar showed distinct growth restriction, which could be substantially improved by insulin treatment. Transgenic offspring were delivered after mating 9747 to wild-type sows and showed the same transgene integration pattern as the sire, arguing for a single integration site in the genome. The other six INSC94Y transgenic founder boars had normal development and exhibited unaltered glucose levels (Fig. 1) and intravenous glucose tolerance (data not shown) within the observation period of 213 days. The INSC94Y-to-INS transcript ratios in pancreas samples of 9747 and his diabetic offspring (n = 3) were 0.75 and 0.78 ± 0.05. In the other founder boars, this ratio was at least five-fold lower (Fig. 1), indicating that the diabetogenic effect of INSC94Y is dose dependent.

Glucose control and insulin secretion.

To evaluate the effect of INSC94Y expression on glucose homeostasis, blood glucose levels were measured regularly beginning at birth before the first food intake. Transgenic offspring exhibited significantly (P < 0.05) elevated random blood glucose levels within 24 h after birth compared with their nontransgenic littermates (Fig. 2). However, divergence in basal plasma insulin levels of 8-day-old INSC94Y transgenic pigs compared with controls was less pronounced (Fig. 2). In contrast, fasting insulin levels of 4.5-month-old INSC94Y transgenic pigs were significantly (P < 0.01) reduced compared with nontransgenic littermates (Fig. 2). Homeostasis model of assessment of β-cells index was significantly reduced in 4.5-month-old INSC94Y transgenic pigs (1.8 ± 0.4 vs. 47.2 ± 10.5; P < 0.01, one-tailed), indicating reduced steady-state β-cell function. Homeostasis model of assessment of insulin resistance index was significantly elevated in 4.5-month-old INSC94Y transgenic pigs (2.4 ± 0.48 vs. 1.2 ± 0.24; P < 0.05, one-tailed), suggesting insulin resistance. For long-term survival, INSC94Y transgenic pigs were treated with insulin starting at the age of 4.5 months. Glucose control under insulin treatment was evaluated over a period of 42 h, and the average blood glucose concentration in the INSC94Y transgenic pigs was 123.2 ± 41.9 (SD) vs. 92 ± 11.9 (SD) mg/dL in the control pigs (Fig. 2). Variations in blood glucose levels were generally greater in the INSC94Y transgenic (40–198 mg/dL) than in the control pigs (62–114 mg/dL).

FIG. 2.

Glucose control and insulin secretion in INSC94Y transgenic (tg) pigs. A: Random blood glucose levels (left) of INSC94Y tg pigs and littermate controls (wt) up to age 8 days. Glucose levels of tg pigs are significantly elevated 24 h after birth. Random and fasting blood glucose levels (right) of tg and wt pigs show deterioration of glucose control in tg pigs with increasing age. B: Fasting insulin levels of 8-day-old and 4.5-month-old INSC94Y tg pigs and littermate controls reveal hypoinsulinemia in 4.5-month-old INSC94Y tg pigs vs. littermate controls. C: The 42-h continuous glucose monitoring of an insulin-treated INSC94Y tg pig and a nontransgenic control (wt) pig. Insulin dose per day: 0.5 to 1 units per kilogram of body weight. IGC, interstitial glucose concentration. Data are means ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001 vs. control.

Loss of β-cell mass and ultrastructural alterations in β-cells.

Pancreas weight did not differ between INSC94Y transgenic pigs and controls at 8 days of age (4.4 ± 0.6 vs. 4.9 ± 0.4 g; P = 0.5) but was significantly reduced in 4.5-month-old INSC94Y transgenic pigs compared with nontransgenic littermates (79 ± 6 vs. 134 ± 5 g; P = 0.001), proportionate with body weight reduction. Qualitative histological evaluation of the pancreas revealed single β-cells and small islet clusters in 8-day-old INSC94Y transgenic pigs to the same extent as in controls (Fig. 3). Compared with nontransgenic littermates, pancreas sections of 4.5-month-old INSC94Y transgenic pigs revealed a visibly reduced β-cell mass and insulin immunostaining intensity (Fig. 3). The volume density and total volume of β-cells in the pancreas of 8-day-old INSC94Y transgenic pigs were unaltered compared with controls. However, in 4.5-month-old INSC94Y transgenic pigs, the volume density of β-cells in the pancreas, total volume, and the total volume of β-cells related to body weight were diminished by 54, 72, and 53%, respectively (Fig. 3). Electron microscopy revealed unaltered secretory granules but dilation of the ER in 8-day-old transgenic piglets and a low number of insulin secretory granules and severe dilation of the ER in β-cells of 4.5-month-old transgenic pigs (Fig. 3). The volume density of non-β-cells in the pancreas was not altered in 4.5-month-old INSC94Y transgenic pigs. The total non-β-cell volume of INSC94Y transgenic pigs was reduced by 38% (290 ± 39 vs. 472 ± 59 mm3; P < 0.05); however, in relation to body weight, their total non-β-cell volume was not different from that of controls (6.5 ± 0.8 vs. 6.0 ± 0.5; P = 0.9).

FIG. 3.

Reduced β-cell mass and ultrastructural changes in INSC94Y transgenic (tg) pigs. A: Immunohistochemical staining for insulin in 8-day-old pigs and double immunohistochemical staining for insulin (pink) and glucagon, somatostatin, and pancreatic polypeptide (brown) in 4.5-month-old pigs. Representative histological sections of pancreatic tissue from a control (wt) and an INSC94Y tg pig. Scale bar = 50 µm. B: Quantitative stereological analyses of the pancreas. Unaltered total β-cell volume (V(β-cell, Pan)) in 8-day-old INSC94Y tg pigs (n = 4 per group) and significant reduction of V(β-cell, Pan) in 4.5-month-old INSC94Y tg pigs (n = 7 per group) compared with littermate controls. Data are means ± SEM. ***P < 0.001 vs. control. C: Transmission electron microscopy of pancreatic tissue from a representative 8-day-old and 4.5-month-old INSC94Y tg and a littermate control (wt) pig. In 4.5-month-old INSC94Y tg pigs, only a few insulin granules (arrows) and severe dilation of endoplasmic reticulum (*) are visible, whereas in 8-day-old tg pigs insulin granules are present and dilation of the endoplasmic reticulum is less severe.

Secondary alterations.

INSC94Y transgenic pigs grew normally up to an age of 2 months; thereafter, their growth rate decreased, resulting in a 41% lower body weight at the age of 4.5 months (Fig. 4). The weights of most organs, including the pancreas, were proportionately reduced (data not shown). A notable exception was kidney weight, which was reduced by only 15%. Thus, relative kidney weight was significantly increased (Fig. 4). Further, the ratio of glomerular volume to body weight was significantly increased in 4.5-month-old INSC94Y transgenic pigs (Fig. 4). However, there were no histological signs of diabetic kidney disease in 4.5-month-old diabetic pigs (Fig. 4) or in 1-year-old cloned diabetic INSC94Y transgenic pigs (n = 5; kindly provided by Minitube of America). Also, there was no histological evidence for diabetic neuropathy in 4.5-month-old INSC94Y transgenic pigs (Fig. 4). In contrast, cataract was observed at the edges of the eye lenses in 8-day-old INSC94Y transgenic pigs, whereas at 4.5 months of age the entire lens was affected (Fig. 4).

FIG. 4.

Diabetes-related alterations of INSC94Y transgenic (tg) pigs. A: Body weight gain in INSC94Y tg pigs (red circles; n = 9) and controls (blue circles; n = 7) up to age 4.5 months. B: Increased relative (rel.) kidney weight (percent body weight) and ratio of mean glomerular volume to body weight (BW) (v[glom] / BW [µm3/kg] × 103) in 4.5-month-old INSC94Y tg pigs vs. littermate controls (wt; n = 7 per group). C: Unaltered true harmonic mean thickness of the GBM in 4.5-month-old INSC94Y tg pigs compared with littermate controls (n = 3 per group) and representative transmission electron microscopic sections. Scale bar, 500 nm. D: Representative histological sections (left) of a tibial nerve fascicle of a 4.5-month-old INSC94Y tg pig and a littermate control demonstrating unchanged fiber density without structural changes in axons and myelin. Azur II-methylene blue-safranin staining. Scale bar, 100 µm. Single fiber teasings (right) showing unchanged fiber myelinisation and morphology of nodes of Ranvier. Scale bar, 100 µm. E: Eye lenses of 8-day-old and 4.5-month-old INSC94Y tg pigs and littermate controls (wt) show a progressive cataract with increasing opacity in tg pigs, whereas lenses of wt pigs remain clear. Data are means ± SEM. *P < 0.05 vs. control; ***P < 0.001 vs. control. US, urinary space; CL, glomerular capillary lumen; p, podocyte foot process.

DISCUSSION

In this study, we established the first transgenic pig model of PNDM triggered by an INS mutation. Induction of insulin-deficient diabetes by other methods has substantial drawbacks. Chemical diabetes induction, e.g., by streptozotocin treatment, is hampered by variable outcomes, including age, sex, or individual differences in susceptibility or spontaneous reversal of diabetes (13). Pancreatectomy is highly invasive, time consuming, and expensive, and it requires supplementation of the pancreatic enzymes (14). In contrast, INSC94Y transgenic pigs have development of stable diabetes because of a specific clinically relevant impairment of β-cells. Unaltered blood glucose levels and body weight at birth suggest postnatal disturbance of β-cells, although impairment of (pro)insulin action to some extent during fetal development cannot be excluded. Like the Ins2 mutant mouse models, the INSC94Y transgenic pig model does not reflect all disease mechanisms of human type 1 diabetes because of the lack of the autoimmune component. Importantly, INSC94Y transgenic pigs can be propagated by conventional breeding when treated with insulin. However, the costs for maintenance of pigs ($3.30 per animal and per day in our facility) and, in particular, of diabetic pigs ($12 per day) are markedly higher than for rodent models.

Expression of INS transcripts at a level of 75% of the endogenous INS mRNA was sufficient to cause hyperglycemia already in newborn transgenic pigs, whereas β-cell mass was initially unaltered. Nevertheless, structural alterations of β-cells suggested ER stress already at this early stage. At the age of 4.5 months, ratio of β-cell mass to body weight of transgenic pigs was significantly reduced and was associated with marked ultrastructural evidence for loss of insulin granules and ER stress. Because insulin staining was used to identify β-cells, we may have overestimated the reduction of β-cell mass attributable to loss of insulin granules. Nevertheless, our findings indicate that the INSC94Y transgenic pig model reflects the disease mechanisms demonstrated in Ins2 mutant mice and imposed in human PNDM. Binding affinity of INSC94Y might differ from endogenous INS, which possibly could have led to accompanying insulin resistance in cases of higher binding affinity of INSC94Y (15). However, in vitro studies using des mutants of human insulin showed that each of the three disulfide bonds is essential for receptor binding affinity, and that lack of the A7B7 interchain disulfide bond has the greatest impact on binding affinity (16). In addition, it remains to be clarified if INSC94Y is at all secreted from β-cells because in vitro studies suggest failure of INSC96Y to undergo exocytosis in response to glucose (17–19) and that mutant misfolded proinsulin also affects bystander proinsulin in the ER (4).

The growth rate of untreated INSC94Y transgenic pigs was markedly reduced as compared with controls. This is in line with stunted growth of diabetic children with insufficient insulin therapy (20) and underlines the essential role of insulin as growth factor and anabolic hormone. Whereas the weights of most organs of INSC94Y transgenic pigs revealed proportional reduction in body weight, the relative kidney weight was increased by 29%. Renal and glomerular hypertrophy is a common early-stage finding in diabetic patients (21); however, histopathological examination did not reveal alterations diagnostic for diabetic nephropathy in INSC94Y transgenic pigs. Whereas the thickness of the GBM is an early and sensitive indicator of diabetic kidney disease in mice (11), it takes years until thickening of the GBM can be detected in human diabetic patients, and even longer until clinical diabetic nephropathy is present (22). Therefore, it is not surprising that GBM thickness is still unaltered in 4.5-month-old INSC94Y transgenic pigs. Future studies need to address if time or complicating factors such as high-protein diet or elevated blood pressure (23) will provoke kidney lesions in diabetic INSC94Y transgenic pigs.

The lack of kidney pathology in young INSC94Y transgenic pigs is in contrast to findings in a diabetic transgenic pig model expressing mutant hepatocyte nuclear factor-1α. Immature renal development and liver alterations were detected in two hepatocyte nuclear factor-1α transgenic-cloned piglets, whereas two others revealed glomerular hypertrophy and sclerosis. Whether these lesions were attributable to the dominant-negative effects of mutant hepatocyte nuclear factor-1α, secondary changes attributable to chronic hyperglycemia, or anomalies associated with somatic cell nuclear transfer remains unclear (24). Structural changes in peripheral nerves possibly will develop with increasing age in INSC94Y transgenic pigs because chronic sensorimotor diabetic polyneuropathy is highly dependent on diabetes duration in humans (25). Similar to humans, INSC94Y transgenic pigs have development of remarkable cataracts, which never has been reported in Ins2 mutant mice. This might be related to a lower aldose reductase activity in lenses of mice compared with humans and rats (26). Therefore, pigs seem to be a favorable model for diabetic cataracts, although the underlying molecular mechanism has to be clarified.

A number of interesting questions can be addressed with this novel pig model for PNDM. A genetically diabetic pig model facilitates testing the functionality of transplanted pancreatic islets, e.g., after microencapsulation or in specific macrodevices (27), in an organism of human size. Also, INSC94Y transgenic pigs seem to be a valuable model for preclinical testing of novel treatment strategies to improve glucose control. The reproducible loss of β-cells makes the INSC94Y transgenic pig an attractive model for the development of in vivo imaging techniques to monitor β-cell mass. Developmental consequences of maternal diabetes on oocytes and embryos and their interaction with the maternal environment are other interesting research topics. Because of the similarities in oocyte maturation and early development of human and pig embryos (28), the INSC94Y transgenic pig is an attractive model to address these questions.