Endocrinology and Metabolism

A general and islet cell-enriched overexpression of IGF-I results in normal islet cell growth, hypoglycemia, and significant resistance to experimental diabetes

Katie Robertson, Yarong Lu, Kristine De Jesus, Bing Li, Qing Su, P. Kay Lund, Jun-Li Liu


Insulin-like growth factor I (IGF-I) is normally produced from hepatocytes and various other cells and tissues, including the pancreas, and is known to stimulate islet cell replication in vitro, prevent Fas-mediated β-cell destruction and delay the onset of diabetes in nonobese diabetic mice. Recently, however, the notion that IGF-I stimulates islet cell growth has been challenged by the results of IGF-I and receptor gene targeting. To test the effects of a general, more profound increase in circulating IGF-I on islet cell growth and glucose homeostasis, we have characterized MT-IGF mice, which overexpress the IGF-I gene under the metallothionein I promoter. In early reports, a 1.5-fold-elevated serum IGF-I level caused accelerated somatic growth and pancreatic enlargement. We demonstrated that the transgene expression, although widespread, was highly concentrated in the β-cells of the pancreatic islets. Yet, islet cell percent and pancreatic morphology were unaffected. IGF-I overexpression resulted in significant hypoglycemia, hypoinsulinemia, and improved glucose tolerance but normal insulin secretion and sensitivity. Pyruvate tolerance test indicated significantly suppressed hepatic gluconeogenesis, which might explain the severe hypoglycemia after fasting. Finally, due to a partial prevention of β-cell death against onset of diabetes and/or the insulin-like effects of IGF-I overexpression, MT-IGF mice (which overexpress the IGF-I gene under the metallothionein I promoter) were significantly resistant to streptozotocin-induced diabetes, with diminished hyperglycemia and prevention of weight loss and death. Although IGF-I might not promote islet cell growth, its overexpression is clearly antidiabetic by improving islet cell survival and/or providing insulin-like effects.

  • pancreatic islets
  • streptozotocin
  • metallothionein promoter
  • gluconeogenesis
  • insulin-like growth factor I

insulin-like growth factor I (IGF-I) is normally produced from hepatocytes and various other cells and tissues, including the pancreas. Acting through its receptor, IGF-IR, IGF-I promotes embryonic development, postnatal growth, and maturation of major organ systems. For two decades, IGF-I has been known to stimulate islet cell replication in vitro, prevent Fas-mediated autoimmune β-cell destruction and delay the onset of diabetes in nonobese diabetic (NOD) mice (6, 23, 47). Although pancreatic islet-specific IGF-I overexpression in rat insulin promoter I (RIP)-IGF mice did not cause an islet hyperplasia, it promoted islet cell regeneration, thus a faster recovery from diabetes, and prevented cytokine-induced lymphocytic infiltration and insulitis (10, 17). Moreover, in MODY3 diabetes, which displays reduced β-cell mass, diminished IGF-I expression seems to play a key role (59). Together these results indicate that IGF-I is a growth factor for pancreatic islet cells. However, our laboratory recently discovered that liver- and pancreatic-specific IGF-I gene deficiency (in liver-specific IGF-I gene-deficient and pancreatic-specific IGF-I gene-deficient PID mice, respectively) caused increased islet β-cell mass, suggesting that IGF-I exerts an inhibitory effect on islet cell growth, albeit indirectly, involving growth hormone and Reg family proteins (31, 32, 58). Furthermore, mice with β-cell-specific IGF-IR gene deficiency exhibited normal islet cell mass, indicating that IGFs are not required for normal islet cell growth (26, 57). Although β-cell-specific dual deficiency of insulin receptor and IGF-IR genes resulted in diminished islet cell growth and early diabetes, insulin signaling seems to play more of a major role than IGF-IR (53). These discrepancies, derived from mouse genetic manipulations, with the classic view of IGF-I actions encourage for further evaluations of its role on islet cell growth in more effective systems. Previously in pancreatic islet-specific RIP-IGF mice, IGF-I expression was limited to the islet cells and/or only at a modest scale (17, 19). To test whether a general and more profound increase in local production and in circulating IGF-I might increase islet cell growth, we have characterized MT-IGF mice, which overexpress the IGF-I gene under the metallothionein I promoter (5, 34, 39, 42, 60). Normally, metallothioneins are synthesized primarily in the liver and kidney. However, in an early characterization, another independent transgenic line of MT-IGF mice displayed IGF-I overexpression not only in the liver and kidney but also in the pancreas (34). In fact, the resulting level of IGF-I mRNA was 31-fold higher in the pancreas than in the liver; IGF-I content was increased 5,200-fold in the pancreas, with the absolute content 344-fold higher than in the liver (34). Because of this robust overexpression, serum IGF-I level was increased 1.5-fold, with a corresponding reduction in the level of growth hormone. Moreover, overexpressed IGF-I resulted in selective organomegaly in the spleen and pancreas (2.0- and 1.8-fold weight increases, respectively) and a 1.4-fold increase in the total body mass (9, 34, 39).

Furthermore, IGF-I has established insulin-like effects in adipose tissues and skeletal muscles resulting in increased glucose transport, lipogenesis, and glycogenesis and decreased lipolysis. In the past decade, this has led to the explorations of IGF-I as a diabetic intervention for improving insulin responsiveness and activating insulin receptor substrates directly (1, 13, 37, 43, 44). IGF-I and insulin-like growth factor-binding protein-3 complex reduced basal glucose production and peripheral glucose uptake during a hyperglycemic clamp in subjects of type 1 diabetes (T1DM) (44). However, the effect of a long-term, systemic elevation of circulating IGF-I level on glucose homeostasis has not been characterized in transgenic mice. Unexpectedly in this study, we demonstrated a general, robust yet islet β-cell-enriched IGF-I overexpression in MT-IGF mice, which had no effect on islet cell growth but caused severe hypoglycemia due to suppressed gluconeogenesis and enabled mice to be resistant to streptozotocin-induced β-cell damage and diabetes.


The MT-IGF mice.

Mice with germline integration of a human IGF-I cDNA driven by the mouse metallothionein 1 promoter (MT-IGF) were studied along with their wild-type littermates, on a mixed C57BL/6 background (39, 60). The animals were maintained in 12:12-h dark-light cycles at room temperature with free access to food and water. As previously reported, zinc supplement in food was not required for the transgene induction (34). To determine genotype, genomic DNA was isolated from tail clips with standard methods. Primers MT-1 (5′-GCA TGT CAC TCT TCA CTC CTC AGG) and MT-2 (5′-TCT GCA TCG TCC TGG CTT TG) were used in PCR reactions, which yield a 0.5-kb band for the transgenic allele. At various ages (2.5–6 mo), the mice were anesthetized with a cocktail of ketamine-xylazine-acepromazine and killed by cervical dislocation. The serum was collected, and the pancreas was removed to perform pancreatic RNA analysis and/or histology. Glycogen content in liver and muscles under fasted conditions was quantified as previously reported (3). The McGill University Animal Care Committee approved all animal-handling procedures.

RNA isolation, Northern blots, and real-time PCR.

Total RNA was isolated by acid guanidinium isothiocyanate-phenol-chloroform extraction. To minimize RNA degradation, the mice were anesthetized; the pancreas was dissected and homogenized immediately before the animals were killed. Northern blot analysis was performed as reported using rat IGF-I, mouse insulin, and β-actin probes (19, 31, 32). For real-time PCR, 1 μg of liver RNA was transcribed into first-strand cDNA using M-MLV reverse transcriptase and oligo(dT)12–18 primers (Invitrogen). The reaction was performed using ABI 7500 System (Applied Biosystems) and QuantiTect SYBRgreen RT-PCR kit (Qiagen) and Qiagen customized primers phosphenolpyruvate carboxykinase (PEPCK) and glucose-6-phosphatase (G6Pase), following the manufacturer's instructions. Five separate samples were amplified from each group (wild-type and MT-IGF mice, 5 mo old, 20-h fasted) in duplicates. The results of each sample were normalized using β-actin mRNA.

Blood chemistry and in vivo procedures.

Serum or plasma concentrations of insulin and glucagon were determined using RIA kits (Linco Research, St. Charles, MO). Blood glucose levels were measured using the OneTouch blood glucose meter (LifeScan Canada, Burnaby, BC, Canada). For insulin tolerance testing, random-fed animals were injected with recombinant human insulin (0.75 IU/kg ip; Roche); for glucose tolerance testing, mice were fasted 24 h and injected with glucose (1 and 3 g/kg ip); for pyruvate and glutamine tolerance testing, mice were fasted for 24 h and injected with pyruvate (2 g/kg ip) or l-glutamine (1.5 g/kg ip), respectively. For glucose stimulated insulin secretion, mice were fasted for 19 h. Blood was collected from tail clips at 0 min, mice were injected with glucose (3 g/kg) through proximal tail veins, and blood was further collected at 2 and 5 min at the tip of the tails.

Immunohistochemistry and islet cell mass measurement.

Pancreatic sections were stained with insulin, glucagon (Monosan, Uden, The Netherlands) or IGF-I antibodies (clone Sm1.2, Upstate USA, Charlottesville, VA) using diaminobenzidine substrate. Microscopic images were captured using a Retiga 1300 digital camera (Q Imaging, Burnaby, BC) and Northern Eclipse software, version 6.0 (Empix Imaging, Mississauga, ON, Canada). The islet density (numbers per tissue surface are), average size of individual islet cells, and β-cell percent (per tissue surface area) were determined as previously reported (31). Specifically, we have measured 30–40 islets per pancreas from 5 MT-IGF mice and 5 wild-type littermates. The β-cell percent was measured by tracing the insulin stained area and using the threshold option to measure the surface area of insulin. The total pancreatic tissue area was measured as well using the threshold option to measure only the stained tissue area. The β-cell percent was measured by dividing the total insulin positive area by the total tissue area for each sample. To establish the source of IGF-I production in islet cells, a double-labeled immunofluorescence against IGF-I and glucagon was performed as previously reported (3032).

Streptozotocin-induced islet cell damage and diabetes.

MT-IGF and wild-type littermates, 3-mo-old males and females, were injected daily for 5 days with streptozotocin (Sigma; 80 mg/kg for females and 75 mg/kg for males; ip) prepared fresh in 0.1 M sodium citrate, pH 4.5 (31). Blood glucose levels from tail vein and body weight were measured every 3 days after the initial injection. As mice became diabetic by 22 days, they were killed to determine serum insulin level and perform pancreatic immunohistochemistry. To detect islet cell apoptosis before the onset of hyperglycemia, a separate set of mice were injected with streptozotocin twice at 0 and 24 h and killed at 48 h; dewaxed paraffin sections of the pancreas were labeled with an in situ cell death detection kit (Roche) and insulin antibody by immunofluorescence, as previously reported (31).

Statistical analysis.

Data are expressed as mean ± SE. The graphs were prepared using SigmaPlot software, version 10 (Systat, San Jose, CA). The Student's t-test (unpaired and paired) and one-way ANOVA followed by Dunnett posttest comparisons were performed using InStat software version 3 (GraphPad Software, San Diego, CA).


A general, robust yet islet β-cell-enriched IGF-I overexpression in MT-IGF mice.

Consistent with previous reports, including the one using the same transgenic line as in this study (39), serum concentration of IGF-I increased 1.3- to 1.5-fold in adult MT-IGF mice vs. wild-type littermates (Table 1). By Northern blot analysis, the level of IGF-I mRNA exhibited a 422-fold increase in the pancreas, because the basal level was virtually undetectable, confirming a pancreatic-enriched expression (Fig. 1, A and C). To reveal what types of cells in the pancreas express the transgene, immunohistochemistry confirmed IGF-I expression in the pancreas and surprisingly, highly concentrated in the endocrine islets, resembling that of insulin (data not shown), while in wild-type pancreas the staining was essentially undetectable in the pancreas or islets, with some nonspecific signals in blood cells. To further establish IGF-I production in either α- or β-cells of the islets, a double-labeled immunofluorescence was performed. In Fig. 2, the pancreatic islets were recognized by glucagon antibody (red Cy-3; left panels), in α-cells distributed at the peripheral of the islets of either wild-type or MT-IGF mice. In wild-type islets, the IGF-I staining (green fluorescein; middle panels) was extremely rare and scattered. On the contrary, IGF-I was detected in all other islet areas except α-cells of the MT-IGF mice. No IGF-I was detectable in α-cells, judging from the lack of signal overlap (would be yellow; right panels). Although the metallothionein promoter is active in islet cells, such a high-level and β-cell-specific activity was not expected. Finally, consistent with previous reports, the general and robust IGF-I overexpression accelerated somatic growth in MT-IGF mice (34, 42). Adult body weights were significantly increased, although the increase in length did not reach statistical significance in males (Table 1). Our tests not only confirmed IGF-I overexpression but also for the first time revealed a very strong β-cell-specific effect, which made the following evaluation of islet cell growth more interesting.

Fig. 1.

Increased Insulin-like growth factor I (IGF-I) but decreased insulin mRNA levels in the pancreas of MT-IGF mice(which overexpress the IGF-I gene under the metallothionein I promoter). A: a representative Northern blot was depicted from 6 mice/samples from each group. The level of β-actin mRNA was used as a loading control. B and C: the relative mRNA abundances in wild-type control and MT-IGF mice, after β-actin-corrected densitometry. n = 6. ***P < 0.001 vs. wild-type littermates.

Fig. 2.

Islet β-cell-specific IGF-I expression revealed by double-labeled immunofluorescence. Pancreatic sections prepared from 3-mo-old mice of wild-type and MT-IGF genotypes were labeled against glucagon (red Cy-3; left panels) and IGF-I (green fluorescein; middle panels) consequently using specific primary and secondary antibodies. Right panels represent merged images to reveal potential overlap in signals. Images were representatives of at least 10 mature islets from each group and have been captured at ×400. Scale bar = 50 μm.

View this table:
Table 1.

Changes in the growth rate, serum chemistry, glycogen content, liver mRNA levels and islet histology in MT-IGF mice

Normal pancreatic islets, hypoinsulinemia, and hypoglycemia.

As a prominent growth factor, IGF-I overexpression by the islet β-cells was expected to accelerate islet cell growth and boost insulin production. However, serum insulin level was markedly decreased by 49% (Table 1) and pancreatic insulin mRNA by 44% (Fig. 1, A and B) in MT-IGF mice vs. wild-type littermates. Evaluation of hematoxylin-eosin-stained pancreatic sections and immunohistochemistry for insulin revealed no obvious abnormality in islet morphology. There was no change in the percent distribution of small, medium, or large islets (Fig. 3, A and B), in the average islet cell size (surface area), or the density of the islets per tissue area (Table 1). As an indication of β-cell mass, we measured β-cell percentage and found no change in MT-IGF mice (Fig. 3C). To explore possible effect of IGF-I on the rate of cell replication, we performed immunohistochemistry against Ki67 (as reported) (30), a known marker for β-cell replication. At the age of 2 wk, islet cells from MT-IGF mice exhibited comparable rate of Ki67 staining to that of wild-type littermates (data not shown). In adult islets (from mice of 2–4 mo old), Ki67 was hardly detectable, and no meaningful change in MT-IGF mice was observed (49). Thus MT-IGF mice did not display any increase in islet cell growth but exhibited decreased (rather than increased) insulin gene expression and plasma insulin level. The absolute β-cell mass was not used as it would be differentially affected by the increases in pancreatic and total body masses in MT-IGF mice. To evaluate β-cell function, insulin release in response to an intravenous injection of glucose showed no significant decrease in MT-IGF-I mice (Fig. 3D) (62).

Fig. 3.

MT-IGF mice displayed normal pancreatic islets. Mice of 3-mo-old, mixed sex, were used for histology. A: immunohistochemistry using insulin antibody. Images were recorded in ×100 magnification and analyzed using Northern Eclipse software. Representative small, medium, and large islets were illustrated from wild-type and MT-IGF littermates. Scale bar =100 μm. B: summary of islet size distribution. In both wild-type and MT-IGF mice, islets were divided into 3 groups according to their surface area (in 1,000 μm2). The y-axis represents percentage of total islets distributed in each of 3 size groups. (n ≥ 35) C: summary of β-cell percentage, as means ± SE. n = 5. D: glucose-stimulated insulin release in MT-IGF vs. wild-type littermates. Male mice 6 mo of age were fasted for 19 h and then stimulated with glucose (3 g/kg iv). Serum insulin concentrations were measured at 0, 2, and 5 min. *P < 0.05 vs. 0 min in wild-type mice. n = 5–8.

In association with increased IGF-I and its insulin-like effects, adult MT-IGF mice exhibited lower blood glucose level vs. wild-type littermates at random-fed status (Table 1; by 14% and significant in male mice). After being fasted for 24 h, the difference was greatly exaggerated, such that MT-IGF mice had on average a 47% decrease in blood glucose level, which even induced hypoglycemic coma in some animals (Table 1). This effect has not been reported in MT-IGF mice and occurred despite decreased serum insulin level and its normal response to glucose stimulation, suggesting a direct consequence of IGF-I production and its insulin-like effects.

Normal insulin sensitivity but significantly improved glucose tolerance.

A possible cause of the hypoglycemia is elevated insulin sensitivity, as our laboratory has reported in mice deficient in growth hormone receptor gene (30). However, MT-IGF mice showed no change in insulin tolerance (Fig. 4A). To further explore possible changes in glucose homeostasis, a glucose tolerance test was performed. As shown in Fig. 4B, following glucose injection wild-type mice exhibited a sharp increase in blood glucose level, peaking between 15 and 30 min, which was not normalized before 120 min. In contrast, MT-IGF mice started with a significantly lower glucose level, displayed a much reduced glucose escalation after the injection, and have a faster decline to baseline by 60 min. This significant contrast indicates the presence of a much more efficient mechanism of glucose disposal in MT-IGF mice, which is not associated with increased insulin secretion or improved insulin sensitivity.

Fig. 4.

Normal insulin sensitivity, increased glucose clearance, and decreased gluconeogenesis in MT-IGF mice: results of insulin, glucose, pyruvate-, and glutamine tolerance tests. Mice aged 2.5 mo old, mixed male and female, were used. A: insulin tolerance test. Wild-type and MT-IGF mice in random-fed state were injected with insulin (0.75 IU/kg ip). Blood glucose levels were measured at 0, 20, 40, and 60 min. Percent values relative to time 0 were expressed as means ± SE. Result of 1-way ANOVA in both wild-type and MT-IGF-I mice, calculated separately: P < 0.0001; posttest result: #P < 0.05, ##P < 0.01 vs. time 0, respectively (n = 4). B: glucose tolerance tests. Mice were fasted for 24 h and injected with glucose (1.0 g/kg ip). Blood glucose levels were measured at 0, 15, 30, 60, and 120 min after injection and expressed as means ± SE. Result of 1-way ANOVA: wild-type mice, P < 0.0001; MT-IGF mice, P < 0.0001. Similar result was obtained using 3 g/kg glucose (n = 4–6). C: pyruvate tolerance tests. Mice were fasted for 24 h and injected with pyruvate (2.0 g/kg ip). Result of 1-way ANOVA: wild-type mice, P < 0.0001; MT-IGF mice, P < 0.01. (n = 4–5). D: glutamine tolerance tests. Mice were fasted for 24 h and injected with l-glutamine (1.5 g/kg ip). Result of 1-way ANOVA: wild-type mice, P < 0.0001; MT-IGF mice, P < 0.01 (n = 7–12). In addition, the results of all unpaired t-tests were indicated as *P < 0.05, **P < 0.01, ***P < 0.001 vs. wild-type littermates.

Suppressed hepatic gluconeogenesis by pyruvate tolerance test.

The increased glucose disposal as a consequence of “insulin-mimicking” IGF-I stimulation would explain the slightly decreased glucose level under random fed status. The more severe hypoglycemia after 24 h fasting in MT-IGF mice suggests additional consequences such as decreased gluconeogenesis and/or glycogenolysis. The conversions of pyruvate into glucose in the liver and glutamine in the kidney and the small intestines are key steps in gluconeogenesis and can be measured by tolerance tests (41). As shown in Fig. 4C, after pyruvate injection, wild-type mice showed a rapid increase in blood glucose level, peaking at 30 min, which remained elevated beyond 120 min. On the other hand, MT-IGF mice started with a significantly lower glucose level, and they displayed a much diminished elevation following pyruvate injection, which rapidly returned to baseline by 120 min. In contrast to the wild-type mice, except the peak change at 30 min after pyruvate, the elevations in the blood glucose level at all other time points were not significant in MT-IGF mice analyzed using one-way ANOVA. Considering the different basal glucose levels, MT-IGF mice displayed a significantly smaller peak increase of 2.2- vs. 2.7-fold in wild-type littermates (P < 0.01 by t-test). The result of the glutamine tolerance test is illustrated in Fig. 4D. In wild-type mice, blood glucose levels were elevated stepwise after glutamine injection, reached a peak level of 200 mg/dl, and did not recede significantly within 2 h. In comparison, MT-IGF mice displayed a smaller and gradual increase, peaked only at 100 mg/dl, which was considered less significant using one-way ANOVA (Fig. 4D). Nevertheless, when corrected by their respective basal glucose levels, the relative fold increases were actually very close, i.e., 2.8-fold in MT-IGF vs. 3.1-fold in wild-type mice. The obvious difference in glucose production induced by pyruvate suggests a suppressed hepatic gluconeogenesis, although an increased glucose disposal (as reflected in Fig. 4B) may also contribute to the hypoglycemia, to a smaller extent because the animals were fasted. Because insulin level was decreased and insulin sensitivity remained normal, this effect is likely a direct consequence of IGF-I overexpression. We have measured possible change in the mRNA levels of two key enzymes, PEPCK and G6Pase, in the liver of those mice using real-time PCR. Under IGF-I overexpression in MT-IGF mice, PEPCK and G6Pase mRNA levels were not decreased to explain the reduced gluconeogenesis activity (Table 1). On the other hand, the rate of glycogenolysis seemed to be normal because no change was detected in the glycogen contents in either the liver or skeletal muscles in MT-IGF mice after 24 h fasting (Table 1).

Resistance to streptozotocin-induced β-cell death and diabetes.

Because IGF-I is known to inhibit cell apoptosis, the robust β-cell-specific overexpression in MT-IGF mice should provide a significant protection against islet damage and prevent the onset of diabetes; additionally, the enhanced “insulin-like actions” as a result of IGF-I overexpression should relieve the symptoms once the diabetes has been induced. To test these possibilities, we have challenged MT-IGF mice with a multiple-low-dose injection of streptozotocin. In wild-type mice, both male and female, streptozotocin induced a rapid onset of hyperglycemia from day 3, with continued increases up to day 18 or 21 (Fig. 5, A and B). The mean peak glucose levels reached 550–600 mg/dl. In three wk after streptozotocin, the wild-type mice lost ∼20% of body weight, 30–38% of them actually died toward the latter half of the study (Fig. 5, C, D, E, and F). In contrast, MT-IGF mice exhibited significantly smaller increases in the glucose level at most time points (using t-test), from day 3 until the end of the 21 days, peaked at ∼450 mg/dl, indicating a delayed onset of diabetes and/or improved symptoms of the disease (Fig. 5, A and B). The weight loss was only ∼8% in both male and female MT-IGF mice, which did not reach significance using one-way ANOVA; and there was no death caused by streptozotocin administration (Fig. 5, C, D, E, and F), further indicating a significant protection and/or relief of the diabetic symptoms. To confirm islet damage at the end of 21 days, pancreatic immunohistochemistry was performed using insulin antibody (Fig. 5G). Without streptozotocin treatment, there was no significant difference in islet histology and insulin staining in MT-IGF and wild-type littermates. Streptozotocin caused a drastic decrease in islet size and in the level of insulin staining in wild-type mice; however, the islets in MT-IGF mice appeared to be better preserved, e.g., larger in size (β-cell percents per total tissue area: wild-type 0.11 ± 0.03% vs. MT-IGF mice 0.18 ± 0.05%, n = 5), although with similarly reduced level of insulin staining.

Fig. 5.

MT-IGF mice were resistant to streptozotocin-induced diabetes. MT-IGF mice and wild-type littermates, 3 mo old, were injected with streptozotocin (female 80 mg/kg qd, ip; male 75 mg/kg) for 5 days. Blood glucose levels from tail vein were measured at 0, 3, 6, 9, 12, 15, 18, and 21 days after the initial injection. A and B: changes in blood glucose levels, illustrated in means ± SE, in female and male mice, respectively. In most time points, MT-IGF mice displayed continuous and significantly lower glucose levels than wild-type littermates. Results of 1-way ANOVA: A: wild-type mice, P < 0.0001; at days 3 and 6, the P values were >0.05; at all other time points, P < 0.001 vs. day 0, respectively. MT-IGF mice, P < 0.0001; at days 3 and 6, the P values were >0.05; at days 9–18, P < 0.001; at day 21, P < 0.01 vs. day 0. (B) Wild-type mice, P < 0.0001; at all other time points, P < 0.001 vs. day 0 respectively. MT-IGF mice, P < 0.0001; except P > 0.05 at day 3 and P < 0.01 at day 6, at all other time points, P < 0.001 vs. day 0. C and D: changes in body weight in female and male mice respectively, expressed as percent of initial weights. MT-IGF mice exhibited more stabilized body weight than wild-type littermates after streptozotocin. Results of one-way ANOVA: C: wild-type mice, P < 0.0001; at day 3, 6, 9, 12, 15, 18, and 21, P values were >0.05, <0.01, <0.001, <0.001, <0.001, <0.01 and <0.001 vs. day 0. Wild-type mice, P = 0.60, the slight decrease in body weight was considered insignificant. D: wild-type mice, P < 0.0001; except P > 0.05 at day 3 and P < 0.01 at day 6, at all other time points, P < 0.001 vs. day 0. Wild-type mice, P = 0.75, the slight decrease in body weight was considered insignificant. In addition, the results of all unpaired t-tests were indicated as *P < 0.05 vs. wild-type littermates. E & F, Differences in accumulated survival rates in female and male mice, respectively. Compared with >30% death rates in wild-type mice after streptozotocin, MT-IGF mice had all survived the experiment. Data in panels A, C, and E are from females, n = 7; data in panels B, D, and F are from males, n = 7–9. G: pancreatic islet damage in MT-IGF and wild-type mice. Twenty-two days after streptozotocin administration, the mice were killed, and their pancreatic sections were studied by immunohistochemistry using insulin antibody. Representative islets from each group (bottom panels) and untreated control mice (upper panels) were recorded as ×100 images. On average, MT-IGF mice exhibited better preserved islets (larger) and the same ratio of insulin staining vs. their wild-type littermates after streptozotocin. Images were representatives of at least 20 islets from 3 mice of each group. Similar results were obtained from both male and female animals. Scale bar =100 μm.

Cellular apoptosis as a result of DNA damage is a key aspect of streptozotocin-induced islet β-cell death (31, 35, 36, 38). To demonstrate that the delayed onset of diabetes was due to IGF-I-induced islet cell survival, we studied islet cell apoptosis at an early time point after streptozotocin, before the onset of hyperglycemia. Pancreatic sections were prepared from wild-type and MT-IGF mice, 48 h after streptozotocin administration (or no administration), and double-stained for immunofluorescence against insulin (red Cy-3) and 2-deoxyuridine 5-triphosphate nick end labeling (TUNEL; green fluorescein). As shown in Fig. 6, no apoptotic nucleus can be detected in untreated wild-type mice (top middle panel). After streptozotocin, many islet cells underwent apoptosis in wild-type mice, at 2.0 ± 0.3 cells per 1,000 μm2 islet surface area (n = 22). In contrast, fewer cells were apoptotic in MT-IGF mice, with the ratio decreased to 1.2 ± 0.2 (n = 21; P < 0.01). This significant protection of islet cells against apoptosis by IGF-I overexpression seems to explain why MT-IGF mice had a delayed onset of hyperglycemia after streptozotocin. Thus IGF-I overexpression in MT-IGF mice delayed the onset of diabetes and/or improved the diabetic symptoms, at least partly due to improved survival and function of the islet cells.

Fig. 6.

IGF-I overexpression protected the β-cells from streptozotocin (STZ)-induced cell death. Pancreatic sections were prepared from 3-mo-old wild-type (WT) or MT-IGF (MT) mice 48 h after STZ administration or without treatment (Ctrl). Double-labled immunofluorescence against insulin (red Cy-3; left panels) and 2-deoxyridine 5-triphosphate nick end labeling (TUNEL; green fluorescein; middle panels) was performed. Representative islets from each group were illustrated as × 400 images, with the bar indicating 50 μm. The white arrow in the middle panel indicates TUNEL signal. Compared with untreated wild-type (WT-Ctrl, top row) islets, streptozotocin treatment (WT-STZ, middle row) caused decrease insulin staining and appearance of apoptotic islet cells; in MT-IGF mice (MT-STZ, bottom row), the insulin level was preserved and apoptotic rate diminished significantly. Images were representative of at least 20 islets taken from 3 mice from each group.


Although the role of IGF-I in insulin actions and diabetes has been addressed extensively, it is still controversial on whether IGF-I is a physiological islet growth factor (29). On one hand, T1DM is associated with reduced serum IGF-I level (8); MODY3 diabetes, caused by a mutation of hepatocyte nuclear factor-1α gene, exhibits reduced IGF-I expression and β-cell growth (59); IGF-I treatment is effective in both T1DM and type 2 diabetes (T2DM) (37, 44). Decreased islet β-cell mass is a key element in the development of autoimmunity-induced T1DM and in compensating insulin resistance in T2DM. It has long been known that IGF-I stimulates islet cell growth and promotes the survival of transplanted islet cells in rodents (2, 11, 17, 18, 23, 47). Recently, however, this notion was challenged by several reports of tissue-specific gene targeting, including those from our laboratory (25, 26, 31, 57). They indicate that IGF-I is not involved in normal islet cell growth and undermine its potential applications in islet cell expansion, protection, and transplantation. Current studies using a robust and islet β-cell-enriched IGF-I overexpression further support the new consensus that IGF-I does not stimulate islet cell growth in vivo. Because of the high level expression, the MT-IGF mice provide a unique opportunity to reevaluate the in vivo insulin-like effects, the effect of IGF-I on insulin secretion, and islet cell survival against damage.

A seeming difference of the recent in vivo reports from early in vitro studies is the presence of high level glucose in the latter (23, 47). IGF-I promotes β-cell replication and survival via receptor activation and recruitment and phosphorylation of insulin receptor substrate (IRS)-2 (55); IRS-2 plays a pivotal role in the maintenance of a normal β-cell population (15, 20, 54, 56). In recent reports, IRS-2 mRNA and protein are relatively unstable in islet cells with half-lives of 1.5–2 h. Short-term (<24 h) and physiological levels of glucose (5–15 mM) increases IRS-2 mRNA and protein levels four- to sixfold (4, 28). It is thus conceivable that high level glucose is required for IGF-I effect by maintaining an adequate level of IRS-2. It has also been reported that glucose controls β-cell mass and function by stimulating an autocrine insulin release and resulting phosphorylation and nuclear exclusion of transcription factor Foxo1 (33, 40). Furthermore, unlike in cultured islet cells, the effect of increased free IGF-I in transgenic mice would be largely neutralized by corresponding increases in binding proteins (7).

Several reports indicate that IGF-I inhibits insulin secretion, involving activations of phosphodiesterase 3B and protein kinase B (22, 27, 62). Consistently, MT-IGF mice exhibited a significantly lower level of insulin and a reduction in insulin gene expression. Moreover, severe hypoglycemia and/or the insulin receptor upregulation in the liver and muscles (K. Robertson and J. L. Liu, unpublished observation) in MT-IGF mice might cause decreased insulin secretion and/or gene expression indirectly. In addition, the unexpected high level of IGF-I production in the islet β-cells of MT-IGF mice may create a nonspecific competition for the same transcription/translation/secretion machinery against insulin secretion. Although metallothionein may be involved in pancreatic hormone synthesis and secretion, in addition to zinc homeostasis and detoxification (52), it has never been reported that the promoter in a transgenic system would have such a high-level expression in islet β-cells (Fig. 2). The abundant zinc ions in β-cells might have boosted the promoter activities. Despite all those possibilities, we have not detected a significant reduction in glucose-stimulated insulin release in vivo (Fig. 3E).

Although excessive IGF-II such as in some endocrine tumors is known to cause hypoglycemia (46), those caused by a chronic IGF-I excess, either in rodents or human are very rare (12). Previously, 6- to 14-day IGF-I treatment caused no change in glycemia or only a transient hypoglycemia in normal, hypophysectomized or diabetic rats (45, 50, 51, 61). The severe hypoglycemia in this study is likely the result of IGF-I cross-activating insulin receptors, as well as its cognate effect mediated by IGF-IR itself, such as in the liver (K. Robertson and J. L. Liu, unpublished observation). In this study, the total serum IGF-I level in wild-type mice, in molar concentrations, was 475-fold of that of insulin (Table 1); considering only ∼4% of total IGF-I being free, the level of free hormone would only be 19-fold higher than insulin and insufficient to significantly activate insulin receptor (Kd for IGF-I 10–100 nM vs. for insulin 0.1–0.2 nM). In MT-IGF mice, with concurrent increase in IGF-I and decrease in insulin levels, the free IGF-I would become 55-fold higher than the insulin levels, reaching a threshold required for activating the insulin receptor with a lower affinity. The excess in free IGF-I is likely <55-fold because of the reported 2.1- to 2.9-fold increases in serum IGFBP-3 and IGFBP-2 levels in these mice (7). Acting through the insulin receptors on skeletal muscles, hepatocytes and other insulin targets, overexpressed IGF-I would stimulate glucose uptake and inhibit glucose production.

As counterregulatory hormones, growth hormone and glucagon stimulate gluconeogenesis. Although there was no change in serum glucagon level, the decrease in growth hormone level that occurs in MT-IGF mice might play some role (34). In this study, decreased glucose production in MT-IGF mice seems to be caused by suppressed hepatic gluconeogenesis rather than a change in glycogenolysis. Additionally, part of the reduction in gluconeogenesis could also be caused by decreased proteolysis due to IGF-I overexpression, thus reduced supply of amino acids (16). Fasting hyperglycemia as a result of enhanced overnight gluconeogenesis is a problem in T2DM that cannot be well managed even by bedtime insulin. Our results suggest a therapeutic potential of IGF-I. Because IGF-IR is capable of mediating insulin-like effects, including direct inhibition on gluconeogenesis (14, 41, 44, 48), it would be worthwhile to determine to what extent the hypoglycemia in this model was caused by IGF-IR-mediated events.

In this report, MT-IGF mice were significantly resistant to streptozotocin-induced diabetes. It is expected that once the diabetes has been induced, persistent elevated IGF-I in the circulation would manifest an “insulin-like treatment” and thus improve the symptoms, including hyperglycemia, weight loss, or animal death. In a longer term, overexpressed IGF-I promoted a faster recovery from diabetes through accelerated islet cell regeneration (17). Our present study supports another possibility, that IGF-I may help to prevent β-cell destruction through an apparent antiapoptotic effect. This was indicated by the findings that in the first 6 (female) or 3 days (male) after streptozotocin treatment, MT-IGF mice displayed a significantly delayed onset of hyperglycemia (Fig. 5, A and B). More directly, streptozotocin-induced islet cell apoptosis, i.e., DNA fragmentation, was significantly diminished in IGF-I overexpression (Fig. 6). Streptozotocin is known to cause β-cell apoptosis in vivo and in vitro (35, 36, 38). It has been known that IGF-I prevents islet cell apoptosis, prevents autoimmune destruction of islets, and delays onset of diabetes in NOD mice (6, 10, 21, 24). Thus, although IGF-I may not be effective in promoting islet cell growth in vivo, it can still be potentially useful in combating T1DM. Either at diabetic prevention, treatment, or islet recovery, IGF-I-based therapy should be effective. As for the concerns of side effects and tumor incidence, we should work on strategies of targeting specific cells and/or intracellular substrates.

In summary, although MT-IGF mice have been created almost two decades ago, this study revealed a novel, islet β-cell-enriched IGF-I overexpression. The resulting lack of islet hyperplasia supports the recent findings from conditional targeting of the IGF-I and receptor genes that argued against the notion that IGF-I is an in vivo islet growth factor. We further discovered significant alterations to glucose homeostasis, including hypoglycemia, hypoinsulinemia and improved glucose tolerance. Improved pyruvate tolerance indicated that IGF-I suppressed hepatic gluconeogenesis. Chronic IGF-I overexpression thus confirmed strong, persistent insulin-like effects on glucose disposal and especially on glucose production. Finally, either due to prevention of β-cell damage or the insulin-like treatment, MT-IGF mice were significantly resistant to streptozotocin-induced diabetes, exemplified by the prevention of weight loss and death. Our results are consistent with the new consensus that IGF-I does not promote islet cell growth normally but inhibits β-cell apoptosis, insulin secretion and hepatic glucose production.


This work was supported by funding from McGill University Health Center (MUHC) Research Institute and Department of Medicine; the J. R. and C. M. Fraser Memorial Fund; Canadian Institutes of Health Research Grants NMD-83124, MOP-84389, and CCI-85675; Natural Sciences and Engineering Research Council of Canada (Grant RGPIN 341205-07; and Canadian Diabetes Association Grant IG-1-07-2305-JL) (to J.-L. Liu). K. Robertson and Y. Lu received McGill Graduate fellowship and studentship award from MUHC Research Institute.


Dr. J. D′Ercole of University North Carolina provided the MT-IGF mice. Dr. A. F. Parlow of the National Hormone and Peptide Program, Harbor-UCLA Medical Center performed the IGF-I assay.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract