Overexpression of c-Myc in β-Cells of Transgenic Mice Causes Proliferation and Apoptosis, Downregulation of Insulin Gene Expression, and Diabetes

A 2.7-kb XbaI-XhoI fragment of cDNA containing exons 2 and 3 of the coding sequence of the mouse c-myc gene was cloned between 0.6 kb of the rat insulin II promoter (20,21) and a 0.3-kb fragment containing a 3′ untranslated region of the human growth hormone (hGH) gene encoding the polyadenylation signal (Fig. 1A). The 3.6-kb HindIII fragment of the RIP-II/c-myc construct was microinjected into fertilized eggs of FVB mice, and viable embryos were implanted into the oviducts of pseudopregnant Swiss Webster mice in the Joslin Diabetes Center Transgenic Core facility to generate founder transgenic mice. At 4 weeks of age, the offspring were tested for the presence of the transgene using Southern blot and PCR analysis on DNA extracted from tail snips. Copy number of integrated transgene was estimated from Southern blot signal intensity compared to indicator bands of a plasmid-transgene construct. Transgenic F₀ founder mice expressing the RIP-II/c-myc chimeric gene were backcrossed with FVB and C57BL6 mice (Taconic Farms, Germantown, NY). All animal procedures were approved by the Harvard Microbiological Safety Committee and the Joslin Diabetes Center Animal Care Committee.

Southern blots were performed with 5 μg genomic DNA digested with SspI. A [α³²P]-labeled probe in exon 3 of the c-myc gene was hybridized to a 1.4-kb fragment specific for the RIP-II/c-myc transgene (Fig. 1A). Genotyping by PCR was performed with 250 ng DNA using oligonucleotide primers designed to hybridize to the RIP-II promoter and exon 2 of the c-myc gene (Fig. 1B) or, alternatively, to exon 3 of the c-myc gene and the hGH 3′ untranslated sequence. The transgene-specific products were amplified with 40 cycles of PCR and visualized using ethidium bromide in 1.5% agarose gels (Fig. 1B).

Pancreata from embryonic (embryonic day 17–18 [E17–18]) and neonate (day of birth and 1 and 2 days after birth) F₁ progeny of RIP-II/c-myc transgenic mice were dissected and fixed by immersion in 4% buffered formaldehyde. After embedding in paraffin, 5- to 7-μm sections were used for histology and immunochemistry. Hematoxylin-stained sections were used for direct microscopic examination. Islet relative volume was measured by point-counting morphometry. All sections were read by one observer (S.B.-W.). Apoptotic and mitotic cells were quantified in hematoxylin-stained pancreatic sections as mitotic figures or characteristic condensed or fragmented nuclei of apoptotic cells. Proliferation was assessed with K_i-67 antibody (1:200; PharMingen) counterstained with hematoxylin. Terminal deoxynucleotidyl transferase–mediated dUTP nick-end labeling (TUNEL) staining was performed according to the manufacturer’s instructions (in situ cell death detection POD kit; Roche) on sections of blocks used for K_i-67 staining and apoptosis counting by morphologic criteria. For immunohistochemistry, primary antibodies were guinea-pig anti-human insulin (1:200; Linco Research, St. Charles, MO); a cocktail of rabbit anti-bovine glucagon (1:2,000, gift of M. Appel, Worcester, MA), rabbit anti-bovine pancreatic polypeptide (1:3,000, gift of R.E. Chance, Eli Lilly, Indianapolis, IN), and rabbit antisynthetic somatostatin (1:300, made in our laboratory) for identifying non–β-cell hormones; rabbit anti-mouse GLUT2 (gift of B. Thorens, Lausanne, Switzerland); anti–PDX-1 (1:7,500, gift of J. Habener, Boston, MA); and anti-Nkx6.1 (1:2,000, gift of P. Serup, Hagedorn, Gentofte, Denmark). The secondary antibodies used for immunofluorescence were, for insulin, Texas red–conjugated Affinipure donkey anti–guinea pig IgG (1:100); for the non–β-cell hormones, fluorescein isothiocyanate (FITC)-conjugated donkey anti-rabbit IgG and streptavidin-conjugated FITC (1:100); for GLUT2, FITC-conjugated donkey anti-rabbit IgG (1:400); for PDX-1, donkey biotinylated anti-rabbit IgG (1:400) followed by streptavidin-conjugated Texas red (1:400); and for Nkx6.1, donkey biotinylated anti-rabbit IgG (1:400) followed by streptavidin-conjugated FITC (1:400) (all from Jackson ImmunoResearch). Immunofluorescent images were taken on a Zeiss 410 microscope in nonconfocal mode. Fresh frozen blocks of pancreata in tissue freezing medium were prepared in molds immersed in chilled isopentane. Frozen sections were stained with an antibody for c-Myc (C-33; Santa Cruz). Pancreatic tissue was processed for electron microscopy by fixing in 2.5% glutaraldehyde in 0.1 mol/l phosphate buffer and embedding in plastic resin (Araldite; E.F. Fullam, Lanthan, NY). Micrographs were taken on a Phillips 301 electron microscope.

Pancreata were sonicated in 2 ml acid ethanol and stored overnight at 4° C. The next day, the homogenate was centrifuged (2,500 rpm for 10 min), and the supernatant was stored at −20°C until insulin concentration was measured by radioimmunoassay using commercially available kits (Linco Research). Total protein content of the extract was measured by the bicinchoninic acid protein assay kit (Pierce, Rockford, IL) using BSA as standard.

Total RNA was extracted from pancreata of F₁ progeny using Ultraspec RNA isolation reagent following the manufacturer-suggested protocols (Biotecx Laboratories, Houston, TX). RNA (500 ng) was then reverse-transcribed to cDNA using Supercript II RNase H⁻ reverse transcriptase (Life Technologies), as previously described (6). Aliquots of diluted cDNA (1.5–3 μl, equivalent to 10–20 ng RNA) were amplified in 25-μl reactions containing 1× GeneAmp PCR buffer, 1.5 mmol/l MgCl₂ (Perkin-Elmer), 80–200 μmol/l dNTPs (Life Technologies), 10 pmol of appropriate oligonucleotide primers (Genosys, Sigma), 2.5 U AmpliTaq Gold DNA polymerase (Perkin-Elmer), and 0.125 μCi [α-³²P]dCTP (3,000 Ci/mmol; New England Nuclear). Oligonucleotide primers were designed with Eugene software (Version 2.2; Daniben Systems, Cincinnati, OH). PCR amplification was performed in a Perkin-Elmer 9700 Thermocycler in which samples underwent a 3-min initial denaturing step, followed by cycling of 1 min at 94°C, 1 min at the annealing temperature indicated in Table 1, and 1 min at 72°C. The final extension step was 10 min at 72°C. Amplimers were resolved by 6% PAGE in 1× Tris borate EDTA buffer. The amount of [α-³²P]dCTP incorporated into amplimers was measured with a PhosphoImager and quantified with Imagequant software (Molecular Dynamics, Sunnyvale, CA). Preliminary reactions were performed to optimize PCR conditions such that the product of each set of primers was linearly amplified.

Blood samples were collected from the F₁ progeny before death for determination of blood glucose and plasma insulin concentrations. Blood glucose was measured with a Medisense Precision QID glucometer (Abbott Laboratories, Bedford, MA) and plasma insulin by radioimmunoassay (Linco Research).

The RIP-II/c-myc transgene resulted in neonatal lethality (see results), so to study long-term effects of c-Myc overexpression in β-cells, we transplanted freshly excised pancreata from 1-day-old transgenic and wild-type mice under the kidney capsules of 6-week-old male immune-deficient Swiss nude mice and followed graft development. After implantation, blood glucose levels and body weights of recipient mice were measured weekly. At 6 weeks after implantation, grafts were excised under anesthesia, trimmed of extraneous tissue, and either fixed in 4% buffered formaldehyde for histology or lysed in RNA isolation solution for analysis of gene expression by RT-PCR as described above.

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

Of 20 mice born from embryos microinjected with the RIP-II/c-myc chimeric gene, 4 male mice were identified as carriers of the transgene by Southern blot and PCR analysis (Fig. 1). The copy numbers of the founders, as determined by Southern blotting, were ∼2, 7, 15, and 40. The studies presented here are from two of these founders (copy numbers 7 and 15) whose transgene was transmitted to F₁ progeny. (The founder with two copies was infertile, and the founder with 40 copies did not transmit the transgene to offspring.) The studies in each line were performed in parallel and showed a similar phenotype. Presumably, the founders were mosaic and thus had some normal β-cells.

F₁ generation mice were studied at four stages of life: at an embryonic age of 17–18 days, on the day of birth (within 15 h of birth), 1 day after birth, and 2 days after birth. Litters were killed at each stage for blood sampling and pancreas collection. The blood glucose levels at each of these stages are shown in Fig. 2A. Embryonic glycemia was variable among litters, and although marked hyperglycemia was absent at this stage of development, the RIP-II/c-myc transgenic embryos had blood glucose levels that were statistically elevated. When evaluated within each litter, blood glucose levels were modestly (36%) higher in transgenic compared with wild-type embryos (P < 0.001). Initially after birth, blood glucose levels of all mice decreased to <20 mg/dl. Within a few hours, the mice had fed normally, as indicated by milk in their stomachs. Within 12 h, milk was seen in the duodenum and ileum, and the blood glucose of transgenic mice had increased to levels greater than in wild-type mice. By the first day after birth, transgenic mice had developed severe diabetes. Marked hyperglycemia was present at days 1 and 2 after birth; blood glucose levels were fivefold higher than those of wild-type littermates (Fig. 2A). Furthermore, plasma of transgenic neonates was milky and the liver pale, indicative of elevated circulating triglycerides and liver steatosis. Transgenic mice not studied at the above time points died on day 3 after birth. The diabetic phenotype developed when the founders were bred with either FVB or C57BL6 female mice. Transgenic pups were statistically lighter than their wild-type littermates (Fig. 2B). When body weights after birth were normalized between litters, the transgenic mice were 12% lighter than wild-type littermates (P < 0.01). The difference in weight was exacerbated after the onset of hyperglycemia; transgenic mice gained less weight by day 1 after birth and had no significant body weight change from day 1 to day 2 (Fig. 2B). Transgenic neonates had plasma insulin concentrations that were on average 65% lower than those of wild-type neonates (P < 0.001) but still within the sensitivity range of the assay (Table 2). Plasma insulin concentrations shown in Table 2 represent pooled data from day of birth and from 1- and 2-day-old mice. At each stage, plasma insulin levels were similarly reduced in transgenic mice.

Pancreatic insulin concentration of transgenic mice was only 0.1% of the wild-type level (Table 2), whereas plasma insulin concentrations were only moderately reduced (65%) (Table 2). Normally, insulin stores in the pancreas are in vast excess of the circulating insulin content. Based on the measurements of pancreas and plasma insulin, we estimate that insulin content in the pancreata of transgenic mice was >50-fold the insulin content in the circulation. Hence, the insulin produced in the pancreata of transgenic mice can account for the presence of insulin in plasma, but clearly at a concentration insufficient to achieve euglycemia. To clarify the mechanism responsible for the development of diabetes and to determine the β-cell phenotype of RIP-II/c-myc transgenic mice, the following histological and gene expression analyses were performed.

Transgenic mice had strikingly altered pancreatic morphology. Endocrine cells were viewed as an amorphous, compact mass around the ducts (Figs. 3B and D). The transgenic phenotype was in stark contrast to that of wild-type mice, in which endocrine cells were less compact and formed as distinct ovoid bodies (Figs. 3A and C). The proportion of islet cells (relative volume) in the pancreas was greatly increased. Morphometric analysis of hematoxylin-stained sections was performed to quantify the increase in islet relative volume (Table 3). Transgenic neonates had an endocrine-cell area threefold larger than that of wildtype neonates: endocrine cells comprised 7.6 ± 1.2% of the pancreatic area in transgenic neonates compared with 2.5 ± 0.5% in wild-type neonates (P < 0.05).

Consistent with the profound reduction of insulin content, very few cells stained for insulin, and these were scattered within the “islets” (Fig. 3D). The expanded mass of cells in transgenic mice stained for GLUT2 (Fig. 4B), a protein commonly used as a β-cell marker. The intensity of GLUT2 staining in pancreatic sections from transgenic mice was reduced and showed a more diffuse pattern, mainly cytoplasmic (Fig. 4B) instead of the usual plasma membrane location (Fig. 4A). Other β -cell markers were expressed in cells from transgenic mice. Staining for the homeodomain protein transcription factor PDX-1 (Fig. 4F) was unremarkable, with typical nuclear staining. Nkx6.1 expression is normally restricted to β-cells in the mature islet and is downstream of PDX-1 in the hierarchy of transcription factors that regulate β-cell differentiation (22,23,24). In some cells of transgenic mice, Nkx6.1 staining was nuclear, but in most cells, staining was cytoplasmic (Fig. 4D) rather than the expected nuclear seen in wild type (Fig. 4C). The observation was made in separate stainings of pancreata from three different litters from the same founder. The pattern of expression of non–β-cell endocrine hormones was assessed by immunostaining with a mixture of antibodies for glucagon, somatostatin, and pancreatic polypeptide (Figs. 3C and D; Figs. 5B and C). Whereas the wild-type mice showed a characteristic pattern of well-defined islets with a β-cell core and the non–β-cell hormones distributed around the mantle (Fig. 3C), the islet architecture in transgenic mice was perturbed, with the non–β-cell hormones forming only clumps at the periphery (Fig. 3D; Figs. 5B and C). Furthermore, most of the insulin-positive cells in transgenic mice also stained for the non–β-cell hormones (Fig. 5C). This phenomenon was apparent at both fetal and neonatal stages, precluding hyperglycemia as a contributing factor. From frozen sections of neonatal pancreata, an increased number of c-Myc–staining β-cells was observed in transgenic mice (data not shown).

Ultrastructural analysis of pancreata from transgenic mice revealed the presence of large euchromatic nuclei, nucleolar hypertrophy, and the almost complete absence of β granules (Fig. 6). Many insulin granules were observed in the cytoplasm of wild-type neonatal β-cells, whereas only a few granules per cell were observed in transgenic mice. Other organelles, such as mitochondria, rough endoplasmic reticulum, and Golgi, did not appear to be altered. The ultrastructure of these cells indicated that they were β-cells and not duct cells. The increased euchromatin and nucleolar hypertrophy were also revealed in hematoxylin-stained sections (Figs. 7B and C).

K_i-67 is a marker of cell proliferation that stains in early to mid-G₁ phase of the cell cycle through mitosis. Interestingly, there was a massive increase in the frequency of K_i-67–stained β -cells of transgenic mice within 3 h of birth (Fig. 8B) that was not seen in 2-day-old transgenic mice (Fig. 8C). An increased incidence of apoptotic nuclei was observed in hematoxylin-stained sections from transgenic mice (Figs. 7B and C), and they were confirmed as being in β-cells in sections double-stained with propidium iodide and GLUT2 (data not shown). Apoptotic nuclei were fourfold more abundant in transgenic compared with wild-type neonates (Table 3). As confirmation, an increase in TUNEL-positive staining was also observed in transgenic mice (Fig. 9). The TUNEL-positive cells were, as expected, more numerous than apoptotic bodies because of the indefinite but longer time of DNA fragmentation. Apoptotic nuclei and bodies are thought to be short-lived in vivo. There was a tendency for an increased frequency of mitotic nuclei in transgenic compared with wild-type neonates (Table 3). Mitotic figures are evident for only about 30 min, so any increase in frequency is indicative of a greatly increased replication. The frequency of mitotic nuclei and apoptotic figures was already increased in transgenic embryos; mitotic nuclei were increased 2-fold (wild type, 0.67 ± 0.17%; transgenic, 1.24 ± 0.27%), and apoptotic figures were 13-fold more abundant (wild type, 0.17 ± 0.17%; transgenic, 2.25 ± 0.45%) in transgenic compared with wild-type E17–18 embryos. The embryonic data reveal the effects of c-Myc overexpression without any confounding influence of hyperglycemia. The data suggest an increased turnover of cells in the islets of transgenic mice. This raises the possibility that the β-cells may have been replicating continuously at a very fast rate, leading to impaired differentiation.

To further assess the extent of β-cell differentiation, mRNA levels were compared in transgenic and wild-type pancreata by semiquantitative RT-PCR. Representative gels for each gene are shown in Fig. 10. After normalization of the specific gene to an invariant control gene (cyclophilin, TBP, or 18S rRNA), mRNA levels are expressed as a percent of wild type (Fig. 10). The transgenic mice showed c-Myc mRNA levels that were increased three- to fourfold. Because these measurements represent c-Myc expression in all cells of the pancreas and the transgene was under control of the insulin promoter and should be expressed only in β-cells, the actual increase in c-Myc mRNA level in β-cells should be even greater. Further analysis revealed a dramatic decrease in insulin mRNA in the pancreata of both fetal (Fig. 10A) and neonatal (Fig. 10B) transgenic mice, findings consistent with the scarcity of insulin-positive cells observed by immunostaining. In contrast, islet amyloid polypeptide (IAPP) mRNA levels were equivalent in transgenic embryos and tended to be elevated, although not significantly, in transgenic neonates, suggesting a selective downregulation of insulin gene expression in β-cells. GLUT2 mRNA levels tended to be lower in the pancreata of transgenic embryos (Fig. 10A) and neonates (Fig. 10B), consistent with the diminished intensity of GLUT2 immunostaining. The increased mRNA levels for the transcription factors BETA2 and Nkx6.1 in the transgenic mice (Figs. 10A and B) probably reflect the expansion in β-cell volume because, when normalized for the expanded mass, these mRNA levels are near normal. The level of PDX-1 mRNA was similar in transgenic and wild-type mice (Figs. 10A and B), perhaps because of its expression in pancreatic duct cells. Alternatively, because the β-cell volume was increased, this lack of change may indicate a decrease in PDX-1 mRNA per β-cell. With respect to LDH-A mRNA levels (Figs. 10A and B), the relatively high expression in neonatal ducts and acinar tissue (seen by immunostaining; data not shown) would be expected to obscure any effect of the transgene in β-cell expression. The level of glucagon mRNA was similar in transgenic and wild-type embryos (Fig. 10A). In neonates, the glucagon mRNA level was increased approximately twofold (Fig. 10B). This increase could have come from either α-cells or immature β-cells that coexpress glucagon.

Because c-Myc overexpression caused neonatal lethality, the question remained whether the β-cells would continue to proliferate with time. We transplanted pancreata from 1-day-old transgenic and wild-type mice under the kidney capsules of normal nude mice to assess the effects of long-term c-Myc overexpression in β-cells. Pancreata from transgenic and wild-type mice were of comparable sizes when transplanted. Over the 6-week posttransplantation period, the presence of the grafted pancreas (transgenic or wild type) had no effect on blood glucose levels or weight gain of recipient mice (data not shown). At 6 weeks after transplantation, the grafts were well-vascularized and had increased in size, but the transgenic grafts were approximately half the size of the wild-type grafts. Histological examination revealed well-defined islets in the wild-type grafts, but only ductal profiles and no islets in transgenic grafts (images not shown). Indirectly, this suggests that with sustained c-Myc activation, apoptosis eventually predominates over proliferation, causing β -cell loss. Further evidence for this was obtained with RT-PCR analysis of mRNAs extracted from the pancreatic grafts (Fig. 10C). Not surprisingly, insulin mRNA levels in grafts from transgenic mice remained markedly suppressed. However, IAPP, PDX-1, BETA2, and Nkx6.1 mRNA levels were also markedly reduced in the 6-week-old grafts from transgenic mice. In particular, IAPP and Nkx6.1 mRNA were present with barely detectable levels, whereas BETA2 and PDX-1 mRNA levels were reduced to 20% of the wild-type levels. This residual expression may be due to BETA2 expressed in other endocrine cells and PDX-1 in ductal tissue. GLUT2 mRNA in the graft most likely represents expression derived from contaminating kidney tissue. Grafts retrieved from the kidney capsule after transplantation can be contaminated with the kidney cortical tissue, which also expresses GLUT2 (25). We tested this by checking for a kidney-specific marker, NKT (26); it was expressed in the grafts but not in neonatal pancreata (data not shown). Glucagon mRNA was maintained in the transgenic grafts, indicating that the loss of mass was specific to the β-cells. The maintenance of β-cell gene expression in the grafts from wild-type mice precludes immune rejection as a mechanism for the loss of β-cells in grafts from transgenic mice.

These studies show that overexpression of c-Myc in β-cells of transgenic mice markedly perturbs islet development, with alteration in β-cell proliferation and turnover, disruption of islet formation, and downregulation of insulin gene expression that result in early-onset diabetes and neonatal lethality. The enhanced β-cell turnover in transgenic mice resulted, in the short term, in β-cell hyperplasia. However, the expanded β-cell mass was not organized into discrete islets, and the pancreas had an altered gene expression pattern, with markedly suppressed insulin levels, less severe reduction of GLUT2 expression, and abnormal cytoplasmic expression of GLUT2 and Nkx6.1. The stimulation of apoptosis by overexpression of c-Myc was striking. There was a complete loss of β-cells in transplanted pancreases of transgenic mice, indirectly suggesting that apoptosis eventually predominated over proliferation.

In normal islets, c-Myc expression is low, but it is induced in diabetic rats, rats made hyperglycemic by glucose clamp, and isolated islets cultured under high-glucose conditions (6,19). The impetus for making these transgenic mice was to see if the β-cell changes found after partial pancreatectomy would be recapitulated by transgenic overexpression of c-Myc. We postulated that the increased c-Myc found in β-cells after Px contributed to the observed changes in differentiation and β-cell hypertrophy (6). Interestingly, at 4 weeks after Px, no increase in replication or apoptosis could be measured (27). The failure of c-Myc to cause such a drastic change after Px is not understood, but perhaps less c-Myc was produced in the Px model. It is also possible that the persistent hyperglycemia after Px led to a gradual change in β-cell phenotype, which included genes that inhibit entry into the cell cycle and others that protect against apoptosis. Consistent with this suggestion, we have found an increase in the expression of a variety of stress genes, some of which might be protective, in islets after Px (28). Nonetheless, in both situations, c-Myc was associated with activation of β-cells to enter the cell cycle. In the transgenic mice, rapid cell proliferation occurred, whereas hypertrophy was seen in the Px rats, which may have been due to simultaneous expression of cell-cycle inhibitors that could trap β-cells in G₁. Indeed, we have found that p21^cip1 is induced after Px (G. Xu and S.B.-W, unpublished data). The characteristics of the transgenic mice closely resembled those recently reported in adult mice with conditional expression of c-Myc in β-cells. Activation of c-Myc by tamoxifen led to a rapid increase in β-cell proliferation that was soon outstripped by apoptosis, resulting in diabetes in only a few days (15).

We hypothesized that an increase in β-cell mass in the prediabetic stage of type 2 diabetes is an important compensatory response to insulin resistance (1,2). In support of this hypothesis, expansion in the number (hyperplasia) and size (hypertrophy) of β-cells with increased insulin demand has been observed in Px rats (6,27), female Zucker diabetic fatty (ZDF) rats with impaired glucose tolerance (4), and rats after 96-h glucose infusion (29); in pregnancy (30); and with the insulin resistance in mice induced by ablation of one allele of the insulin receptor and insulin receptor substrate-1 (3). Once the capacity for compensatory β-cell expansion is overwhelmed, diabetes develops (1,2,4, 6).

Our findings are consistent with a role for c-Myc in the compensatory growth of β-cells. Elevated c-Myc expression is associated with glucose stimulation of DNA synthesis in islet cell lines (31) and in insulinomas in combination with another oncogene such as ras (18,32). In combination with ras, c-Myc has been shown to stimulate β-cell replication in vitro (33), and as mentioned above, conditional expression of c-Myc in mice leads to increased β-cell proliferation (15). The finding of increased euchromatin and prominent nuclei and nucleoli in the transgenic β-cells is consistent with the effects of c-Myc in the liver (34) and other cells (35) to induce ribosomal biogenesis and increase cell mass.

Results obtained in these transgenic mice indicate that c-Myc can promote apoptosis in β-cells. Importantly, this contrasts with the effects of c-Myc in different cell types (15). In epidermis and lymph tissue, overexpression of c-Myc leads to proliferation with little apoptosis (15,36,37). When c-Myc overexpression is targeted to acinar cells of the pancreas, mixed acinar/ductal adenocarcinomas develop (38). In the late fetal period of the RIP-II/c-myc transgenic mice, there was an apparent link between replication and apoptosis that was quickly lost in the postnatal period, as evidenced by the marked decrease in K_i-67–positive cells in 2-day-old mice (Fig. 8); c-Myc stimulation of apoptosis appeared to be maintained. This later dominance of apoptosis is consistent with the loss of β-cells noted in pancreatic tissue transplanted under the kidney capsules of athymic mice. The susceptibility of β-cells to apoptosis may be related to low levels of expression of the protective antioxidant genes in islets (39). To transform β-cells, it is presumably necessary to inhibit the apoptotic process. The development of insulinomas with activated c-Myc may occur only with concomitant mutation in genes implicated in c-Myc–induced pro-apoptotic pathways (p53, MEN1, or p19^ARF) or upregulation of anti-apoptotic genes (Bcl family) (15,32,40).

The β-cells of the transgenic mice exhibited abnormal differentiation, which is not surprising considering their rapid rate of proliferation, a condition well known to lead to dedifferentiation. This has been well shown in insulin-producing cell lines produced by overexpression of the simian virus 40 (SV40) tumor antigen gene under control of the rat insulin promoter. An inverse relationship between growth rate and insulin gene expression has been demonstrated (41–43). However, these cell lines can maintain some features of differentiated β -cells after many passages. It is impossible to say how much of the change in β-cell differentiation in the transgenic mice is directly due to c-Myc and how much is due to some general process that accompanies proliferation. Likewise, in the Px model, it is unclear how much of the change of gene expression is directly due to c-Myc and how much is caused by other hyperglycemia-induced mechanisms.

It is of interest that most of the few insulin-positive cells in transgenic mice also stained for the non–β-cell hormones. Coexpression of insulin and glucagon has been detected in the earliest stages of endocrine cell development (44–46), whereas a recent study showed that β-cell progenitor cells transcribe the pancreatic polypeptide gene and never glucagon (47). These dual-staining cells could be regarded as transient/immature cells that can also be called protodifferentiated. Normally, these cells are thought to produce non– β-cell hormones only briefly before full β-cell differentiation takes over. In these transgenic mice, the protodifferentiated cells may represent very young cells that maintain insulin expression only briefly before it is shut off by c-Myc. Other aspects of β-cell differentiation are at least partially maintained. The RIP-II/c-myc transgenic mice had typical nuclear staining of PDX-1. The equivalent levels of PDX-1 mRNA in the pancreas of transgenic mice may indicate an actual reduction per β-cell (see results), but PDX-1 expression in the ducts may obscure any change (48). Important elements of the transcriptional machinery that determine islet development and activate β -cell transcription (BETA2, Nkx6.1, PDX-1) are present in spite of c-Myc overexpression. However, finding staining of Nkx6.1 in the cytoplasm rather than in the nucleus raises questions about its function (23,24).

We doubt that the marked reduction of insulin gene expression can be explained by competition between the endogenous insulin promoter and the transgene insulin promoter for transcription factors, because this RIP-II sequence has been used in the past without evident interference (20,21,43, 49–51).

A potential mechanism is suggested by studies showing that c-Myc can suppress insulin gene transcription by inhibiting BETA2-mediated transcriptional activation (52). If c-Myc does exert a potent inhibitory influence on insulin gene expression, why are a few β-cells positively stained for insulin? We suspect that the rare insulin-positive β-cells in RIP-II/c-myc transgenic mice represent newly formed β-cells still expressing insulin protein after transcription has been suppressed, this being facilitated by the long half-life of insulin mRNA (∼ 24 h). It remains to be determined whether c-Myc downregulation of insulin promoter activity involves a direct or indirect mechanism. c-Myc could act by binding directly to the insulin promoter, by physically interacting with an insulin gene-activating complex, or by inducing the expression of other proteins that repress insulin gene transcription. The ability of c-Myc to bind E-boxes suggests a possible physical interaction of c-Myc with this important element that binds heterodimers belonging to the bHLH family of transcription factors. The complex consists of BETA2, the tissue-restricted member of the family, and its heterodimerizing, ubiquitously expressed bHLH family members E2A, E47/ E12, or HEB (53). The binding of the Myc-Max complex to the insulin E-box may compete for binding by these bHLH proteins that might, in turn, inhibit insulin gene transcription.

In conclusion, our results suggest that the transcription factor c-Myc has major influence on the regulation of β-cell mass, islet formation, gene expression, and glucose homeostasis. The studies from RIP-II/c-myc transgenic mice demonstrate the effects of c-Myc overexpression on 1) initial proliferation and expansion of β-cells; 2) marked suppression of insulin gene expression, reduced GLUT2 expression, and the development of diabetes; and 3) ultimate depletion of β-cell mass through increased rates of apoptosis. We suggest that c-Myc is a potentially important regulator of β-cell growth, although its precise contributions to normal β-cell development, adaptation to insulin resistance, and the development of diabetes remain to be determined.

The Diabetes Endocrinology Research Center Core facilities were supported by Grant DK-36836 from the National Institutes of Health. This work was supported by National Institutes of Health Grant DK-35449 and the Diabetes Research and Wellness Foundation. D.R.L. is a recipient of a postdoctoral fellowship from the Juvenile Diabetes Foundation. A.S. is a recipient of a career development award from the American Diabetes Association.

We thank Adam Groff for expert technical help; Nitin Trivedi and Jennifer Hollister-Lock for performing pancreas transplantation; Sanjoy Dutta for experimental advice; Chris Cahill of the Joslin Advanced Microscopy Core; the Joslin RIA Core for insulin assay; the Joslin Transgenic Core; and the Animal Care Core.