Effects of mutations in the insulin-like growth factor signaling system on embryonic pancreas development and beta-cell compensation to insulin resistance.

Insulin and insulin-like growth factors (IGF) play overlapping and complementary roles in pancreatic beta-cell function and peripheral metabolism. In this study, we have analyzed mice bearing loss-of-function mutations of the insulin/IGF signaling systems. Combined inactivation of insulin receptor (Insr) and Igf1 receptor (Igf1r), but not of either receptor alone, resulted in a 90% decrease in the size of the exocrine pancreas, because of decreased cellular proliferation. In contrast to the findings in the exocrine compartment, endocrine alpha- and beta-cell development was unperturbed. Combined ablation of Igf1 and Igf2, the ligands for these two receptors, resulted in an identical phenotype. We also examined the effect of heterozygous null Igf1r mutations on glucose homeostasis in adult mice. Igf1r haploinsufficiency did not affect insulin action and compensatory beta-cell growth in insulin-resistant mice with combined Insr and Igf1r heterozygous null mutations, resulting in a considerably milder phenotype than combined haploinsufficiency for Insr and its main signaling substrates, Irs1 and Irs2. We conclude that Igf1r and Insr are required for embryonic development of the exocrine but not of the endocrine pancreas and that defects of Igf1r do not alter glucose homeostasis as long as the insulin receptor system remains intact.


Insulin and insulin-like growth factors (IGF) play overlapping and complementary roles in pancreatic
␤-cell function and peripheral metabolism. In this study, we have analyzed mice bearing loss-of-function mutations of the insulin/IGF signaling systems. Combined inactivation of insulin receptor (Insr) and Igf1 receptor (Igf1r), but not of either receptor alone, resulted in a 90% decrease in the size of the exocrine pancreas, because of decreased cellular proliferation. In contrast to the findings in the exocrine compartment, endocrine ␣and ␤-cell development was unperturbed. Combined ablation of Igf1 and Igf2, the ligands for these two receptors, resulted in an identical phenotype. We also examined the effect of heterozygous null Igf1r mutations on glucose homeostasis in adult mice. Igf1r haploinsufficiency did not affect insulin action and compensatory ␤-cell growth in insulin-resistant mice with combined Insr and Igf1r heterozygous null mutations, resulting in a considerably milder phenotype than combined haploinsufficiency for Insr and its main signaling substrates, Irs1 and Irs2. We conclude that Igf1r and Insr are required for embryonic development of the exocrine but not of the endocrine pancreas and that defects of Igf1r do not alter glucose homeostasis as long as the insulin receptor system remains intact.
Type 2 diabetes is a complex metabolic disease characterized by impaired insulin action in peripheral tissues and perturbed pancreatic ␤-cell function (1,2). Although these two abnormalities are interdependent, and lead in concert to the onset of hyperglycemia, it is unclear whether they share a common pathogenesis.
Increasing evidence indicates that insulin and IGF 1 signalings play overlapping roles in metabolism and pancreatic ␤-cell function. For example, ablation of Insr function in skeletal muscle, a key target tissue in glucose disposal, does not cause significant metabolic abnormalities (3,4). In contrast, combined inhibition of Insr and Igf1r function by a dominant negative Igf1r transgene results in insulin-resistant diabetes (5). Moreover, selective ablation of Insr (6) or Igf1r (8) 2 in ␤-cells results in a similar phenotype of abnormal insulin secretion, as does lack of the INSR substrate insulin receptor substrate (Irs) 1 (9). On the other hand, ablation of Irs2, a closely related INSR substrate, results in impaired ␤-cell proliferation and diabetes (10). At present, however, the details of the involvement of INSR and IGF1R in ␤-cell proliferation and insulin secretion and also in compensatory functions in response to insulin resistance remain unclear (11).
The phenotypes of mice carrying targeted null mutations of Insr and Igf1r indicate that the two genes have fundamentally different physiologic roles (12)(13)(14). Nevertheless, some data suggest that there is functional overlap between the growthpromoting functions of the two receptors (15). As we have previously shown, IGF2 stimulates embryonic growth in mice not only through its cognate receptor (IGF1R) but also in part through INSR (16). Moreover, mice lacking both IGF receptors, IGF1R and IGF2R, are rescued from perinatal lethality by IGF2 acting through INSR to promote growth (17). In contrast, the ability of IGF1R to compensate for the metabolic actions of INSR is unclear (18 -24). In addition to its putative role in ␤-cells (25), 2 IGF1R has a potential metabolic role in skeletal muscle, where IGF1 stimulates glucose uptake at concentrations that do not activate insulin receptors (3, 4, 26 -30).
To investigate further the functional overlap between INSR and IGF1R signaling in development, especially as it pertains to the pancreas, and also in metabolic control, we examined different mouse models. First, we analyzed pancreatic development in embryos homozygous for null mutations of the genes encoding Igf1r and Insr, alone or in combination. We also studied double mutant embryos lacking both IGF ligands, Igf1 and Igf2. Perinatal lethality precludes a metabolic analysis of nullizygotes for Insr or Igf1r mutations in adulthood. Thus, using Insr ϩ/Ϫ Igf1r ϩ/Ϫ double heterozygotes, we asked whether the insulin resistance caused by the Insr mutation (31,32) is exacerbated by Igf1r haploinsufficiency. Because we have previously shown that Insr haploinsufficiency has a strong diabetogenic effect in appropriate predisposing backgrounds (Irs1 ϩ/Ϫ or Irs2 ϩ/Ϫ ), it was important to examine whether a similar predisposing effect of Igf1r haploinsufficiency would be detected by this experimental approach.

EXPERIMENTAL PROCEDURES
Animal Husbandry-Generation and genotyping of mice with null alleles of Igf1, Igf2, Insr, and Igf1r have been reported in previous publications (12,14,33). Double heterozygous Insr ϩ/Ϫ Igf1r ϩ/Ϫ mice were obtained by crossing the two strains as described previously (16). Timed pregnancies were set up to retrieve Insr Ϫ/Ϫ and Igf1r Ϫ/Ϫ embryos at gestational stages E12.5 and E18.5. Mice were housed in a barrier facility in ventilated cages with filtered air in a 12-h light-dark cycle.
Phenotypic Analysis-Mice were fed regular chow ad libitum or fasted for 16 h. Blood was drawn from the retroorbital sinus of anesthetized animals between 0900 and 1100 h. For glucose tolerance test, mice were fasted starting at 1800 for 16 h. Thereafter, mice were injected intraperitoneally with glucose (2 g/kg body weight). Blood samples were obtained immediately before and at the indicated time points after injection. Glucose levels were measured using a glucometer (Accucheck; Roche Diagnostics). Insulin was measured in plasma samples by radioimmunoassay using a rat insulin standard (Linco, St. Charles, MO).
Immunoprecipitation, Immunoblotting, and PI 3-Kinase Assays-Experiments were carried out in overnight fasted, 4 -6-month-old male mice. Animals were anesthetized by the intraperitoneal administration of sodium pentobarbital (65 mg/kg), and human insulin (5 units) was injected through the inferior vena cava. Extracts of liver and hindlimb muscles (gluteus and soleus) were prepared as described (32). Immunoprecipitation, immunoblotting, and measurements of PI 3-kinase activity in skeletal muscle were performed as described previously (32).
Pancreatic Immunohistochemistry and Morphometry: Adult Mice-Pancreata were removed from adult mice, cleared of fat and spleen, weighed, and fixed overnight in Bouin's solution. Tissues were embedded in paraffin, and consecutive 5-m sections were mounted on slides (32). Sections at 200-m intervals were immunostained with various antibodies used at the dilutions indicated above. The cross-sectional area of the pancreas was covered systematically by accumulating images from a series of non-overlapping fields at a ϫ40 magnification. Images were recorded on a digital camera Nikon Coolpix 950 (Nikon) and analyzed using the "image analysis" macro of the NIH Image software, version 1.60. Areas corresponding to the whole pancreas, ␣and ␤-cells were measured by manual tracing of the acquired images. ␣and ␤-cell areas were calculated by dividing the area of glucagon (␣cells) and insulin-positive (␤-cells) immunoreactivity by whole pancreas area (32).
Embryonic Studies-To measure pancreatic size, E18.5 embryos were sacrificed; the abdominal cavity was cut open to facilitate tissue fixation, and serial transverse abdominal sections covering the entire pancreatic length were obtained. Total pancreatic size was estimated from the number of pancreas-containing sections, because pancreatic weights could not be reliably ascertained due to the small size of some mutant embryos. However, total embryo weight could be used to normalize the various comparisons, as we have shown that in normal embryos the total number of pancreatic sections obtained correlated with whole embryo weight (r 2 ϭ 0.8, p Ͻ 0.001). Three sections, spaced at least 50 m apart, were chosen for morphometric measurements. To closely match the position of these sections among different genotypes, we used the duodenum, liver, and spine as reference points. Three non-overlapping microscopic fields at low magnification (ϫ10) were scored on each section, and four to six embryos were analyzed per genotype. Islet area was defined as the sum of ␣and ␤-cells areas after staining with insulin and glucagon antisera, respectively, because the number of somatostatin-and pancreatic polypeptide-positive cells was too small to be quantitated reliably.
BrdUrd Staining-Pregnant females were injected intraperitoneally with BrdUrd (100 mg/kg body weight) and 5-fluoro-2Ј-deoxyuridine (6.7 mg/kg body weight) at E18.5 and sacrificed 1 h later. Embryos were dissected, fixed, and analyzed by immunohistochemistry using anti-BrdUrd antibody as described (34). The number of BrdUrd-positive and total nuclei within a microscopic field was determined using the "cell counting" macro of the NIH Image 1.60 software. Ten sections from three different WT animals and eight sections from three different Igf1 Ϫ/Ϫ Igf2 ϩ/pϪ mice were analyzed.
Statistical Analysis-Descriptive statistics and analysis of variance (ANOVA) were calculated using the Statsview software (Abacus Concepts, Inc., Berkeley, CA).

Effect of Null Insr and Igf1r Mutations on Fetal Pancreatic
Development-To determine the consequences of ablation of INSR or IGF1R signaling on pancreatic development, we analyzed the morphology, size, and composition of pancreata in embryos lacking INSR, IGF1R, or both. Embryos were examined at embryonic day E18.5, corresponding to the developmental stage in which aggregation of endocrine cells into distinct islet structures was first recognizable (35). A representative immunohistochemical analysis for markers of endocrine (insulin) and exocrine (amylase) differentiation is shown in Fig. 1, a and b. As we have shown in the past, mutations of the IGF signaling system result in embryonic growth retardation. Thus, the approximate weights of mutant embryos relative to normal were Insr Ϫ/Ϫ 90% (16), Igf1r Ϫ/Ϫ 45% (14), and Insr Ϫ/Ϫ -Igf1r Ϫ/Ϫ 30% (16). The weight values were used to normalize pancreatic sizes, determined as described under "Experimental Procedures" and expressed in arbitrary units (Fig. 1c). This morphometric analysis indicated that the normalized total pancreatic size in Insr Ϫ/Ϫ embryos was similar to WT controls, whereas it was increased by ϳ2-fold in Igf1r Ϫ/Ϫ and was decreased by ϳ50% in Insr Ϫ/Ϫ Igf1r Ϫ/Ϫ embryos (Fig. 1c). Immunostaining with anti-amylase antiserum indicated that the decrease in pancreatic size observed in the double nullizygous embryos could be accounted for by a decrease of exocrine pancreatic size (Fig. 1b) (p Ͻ 0.05 by ANOVA). In contrast, ␣and ␤-cell areas, as assessed by immunohistochemistry with antiinsulin and anti-glucagon antisera, were not significantly different from WT in Insr Ϫ/Ϫ and Igf1r Ϫ/Ϫ embryos (Fig. 1d). In double mutant embryos, however, relative ␣and ␤-cell areas were increased ϳ3-fold, as a result of the decreased area of exocrine pancreas (Fig. 1d). The ␣/␤-cell area ratio within islets was unchanged in all embryos except Igf1r Ϫ/Ϫ , in which a 60% increase in the relative area of ␣-cells was detected (p Ͻ 0.05 by ANOVA). Notably, islet architecture appeared to be disrupted in double Insr Ϫ/Ϫ Igf1r Ϫ/Ϫ mutants, apparently as result of the overall impairment of exocrine pancreas development.
These analyses indicated that, in Insr Ϫ/Ϫ embryos, pancreatic acinar size is reduced proportionately to body weight (isometric change). However, in Igf1r Ϫ/Ϫ and double nullizygous embryos the relationship of pancreatic size to body size was altered in opposite directions (i.e. increased and decreased, respectively; allometric changes). To obtain an absolute measure of exocrine pancreatic size (i.e. independently of body size), we measured islet and acinar diameters in pancreata of each genotype. Mean acinar diameter was reduced by ϳ40% in Insr Ϫ/Ϫ Igf1r Ϫ/Ϫ embryos, as compared with wild-type mice, with no changes in the two types of single mutants (Fig. 1e). In contrast, islet diameter was similar in all four genotypes (Fig.  1f). These data indicate that exocrine pancreas growth is selectively blunted in Insr Ϫ/Ϫ Igf1r Ϫ/Ϫ embryos.
Mice Lacking Both Igf1 and Igf2 Exhibit the Same Phenotype of Reduced Exocrine Pancreatic Growth-Because of the known interaction of IGF2 with both IGF1R and INSR, the decrease of pancreatic exocrine tissue in double mutants Insr Ϫ/Ϫ Igf1r Ϫ/Ϫ , but not in single Insr Ϫ/Ϫ or Igf1r Ϫ/Ϫ mutant embryos, is consistent with the hypothesis that acinar tissue growth is mediated by the IGF2 ligand (16). Because the mutational consequences of the lack of IGF2 begin manifesting at E12.5 and precede those due to lack of IGF1 by 1 embryonic day (33), we examined pancreatic development in E12.5 embryos. Interestingly, the change in pancreatic size was already apparent at this early stage, consistent with an IGF2dependent effect (Fig. 2).
For further verification, we examined embryos lacking both IGF1 and IGF2 ligands, and we showed that they are phenotypically indistinguishable in their histopathology from Insr Ϫ/Ϫ Igf1r Ϫ/Ϫ mice. Thus, they display normal endocrine pancreas architecture but a decrease of exocrine pancreatic size (Fig. 3a).
Because loss of IGF function results in reduced cell proliferation (36), we examined the proliferation potential of Igf1 Ϫ/Ϫ -Igf2 ϩ/pϪ mice using BrdUrd labeling. The ratio of BrdUrdpositive cells to total number of cells scored in Igf1 Ϫ/Ϫ Igf2 ϩ/pϪ FIG. 1. Pancreatic morphology and morphometry in E18.5 embryos with Insr and Igf1r mutations. Pancreatic sections were processed as indicated under "Experimental Procedures." Immunohistochemistry was performed either with anti-insulin (a) or anti-amylase antibodies (b), and a representative section is shown. c, evaluation of total pancreatic size was performed by measuring pancreatic area in transverse abdominal sections and dividing the result by embryo weight. An asterisk denotes a p Ͻ 0.05 for the following comparisons: Igf1r Ϫ/Ϫ versus WT, Insr Ϫ/Ϫ and Insr Ϫ/Ϫ Igf1r Ϫ/Ϫ ; Insr Ϫ/Ϫ Igf1r Ϫ/Ϫ versus WT, Insr Ϫ/Ϫ and Igf1r Ϫ/Ϫ . d, islet morphometry. ␤-Cell and ␣-cell areas are expressed as percentages of total pancreatic area surveyed. The ratio of ␣/␤-cells was calculated by counting nuclei of insulin-and glucagon-positive cells within each islet. An asterisk denotes a statistically significant difference (p Ͻ 0.05) in ␤-cell or ␣-cell area between Insr Ϫ/Ϫ Igf1r Ϫ/Ϫ and WT, Insr Ϫ/Ϫ and Igf1r Ϫ/Ϫ mice, and in ␣/␤-cell ratio between Igf1r Ϫ/Ϫ and WT, Insr Ϫ/Ϫ and Insr Ϫ/Ϫ Igf1r Ϫ/Ϫ mice. Values represent means Ϯ S.E. derived from three sections for each embryo and four to six animals for each genotype. ␣/␤-Cell ratio was determined by counting nuclei in ϳ100 islets/genotype. e, hematoxylin and eosin-stained pancreatic sections from mice with the genotypes indicated were analyzed using the NIH Image 1.60 software to determine the diameters of exocrine acini (e) and endocrine islets (f). The maximal apparent diameter was manually traced and measured using the image analysis macro. The number of acini and islets analyzed for each genotype is indicated. For each genotype, three animals were analyzed in three sections 50 m apart. For each section, three microscopic fields were scored. *, p Ͻ 0.05 by ANOVA. mice was reduced by 35% (p Ͻ 0.04), consistent with previous observations (36) (Fig. 3b).
Effects of IGF1R Haploinsufficiency on Mouse Growth-We next examined adult mice with heterozygous Igf1r mutations, either alone or in combination with Insr heterozygous mutations. Determination of body weights at different postnatal ages indicated that there was no statistically significant difference between WT and Insr ϩ/Ϫ mutant mice or between Igf1r ϩ/Ϫ and double heterozygous Insr ϩ/Ϫ Igf1r ϩ/Ϫ mutant mice. In agreement with the results of a detailed analysis of post-natal growth of Igf1r ϩ/Ϫ mice, 3 we observed that the relative weights of mice carrying the heterozygous Igf1r mutation, alone or in combination with Insr heterozygosity, was ϳ85% of WT (Fig. 4a). Accordingly, we conclude that there is growth haploinsufficiency of IGF1R function, which is not aggravated by concomitant heterozygosity for Insr. Haploinsufficiency results from a gene dosage effect, as evidenced by the decrease in the expression of IGF1R, at least as assayed in muscle extracts (Fig. 4b).
Effect of a Null Igf1r Mutation on Metabolic Control in Insulin-resistant Insr ϩ/Ϫ Mice-To determine the metabolic ef-fects of the combined Insr/Igf1r mutations, various parameters were analyzed in 6-month-old mice. Insulin levels were measured in the fasted and fed states (Fig. 5, a and b). Igf1r ϩ/Ϫ mice had normal insulin levels. In contrast, insulin levels in Insr ϩ/Ϫ mice were 1.8-and 2.9-fold higher than in fasted and fed WT mice, respectively. Combined Insr ϩ/Ϫ Igf1r ϩ/Ϫ mutations caused a further increase to 2.2-and 4.5-fold higher than WT (p Ͻ 0.05 by ANOVA versus WT and Ig1r ϩ/Ϫ ).
To examine the effect of these mutations on pancreatic islets, we performed immunohistochemistry of pancreatic sections with anti-insulin and anti-glucagon antisera to analyze islet morphology and morphometry. Islets were moderately enlarged in Insr ϩ/Ϫ and Insr ϩ/Ϫ Igf1r ϩ/Ϫ mice compared with 3 E. Chiao and A. Efstratiadis, unpublished results. To determine the time course of the growth defect in mice, embryos were collected at E12.5 and analyzed by staining with anti-amylase antiserum. A representative E12.5 embryo indicates that the growth defect is already evident at this early stage of pancreatic morphogenesis.
FIG. 3. Decreased exocrine pancreatic size in mutant mouse embryos lacking IGF1 and IGF2 is due to decreased proliferation. a, double mutant embryos Igf1 Ϫ/Ϫ Igf2 ϩ/pϪ were collected at E18.5, and pancreatic sections were processed as indicated above. A representative section stained with anti-insulin (top) and anti-amylase anti-serum (bottom) is shown. b, BrdUrd labeling of proliferating cells was performed on E18.5 embryos as indicated (34). Thereafter, the percentage of BrdUrd-positive nuclei was scored by selecting representative microscopic fields and counting the number of labeled and unlabeled nuclei using the cell counting macro of the NIH Image 1.60 software. Three animals per genotype were analyzed. Eight sections were analyzed for WT mice and 10 for Igf1 Ϫ/Ϫ Igf2 ϩ/pϪ mice. *, p Ͻ 0.04 by ANOVA.
Fasting glucose levels were maintained within a normal range in all mice (Fig. 5e). Measurements of glucose levels in the fed state indicated that ϳ10% of Insr ϩ/Ϫ Igf1r ϩ/Ϫ mice had developed diabetes (defined as an elevation of glucose values more than the mean ϩ 2 S.D. on at least two separate measurements), as had 1 of 10 Insr ϩ/Ϫ mice analyzed in this cross (Fig. 5f). These data are consistent with larger data sets reported in previous analyses (3,4,31,32,37). An intraperitoneal glucose tolerance test showed that Insr ϩ/Ϫ and Igf1r ϩ/Ϫ mice had normal glucose tolerance. In contrast, double heterozygous Insr ϩ/Ϫ Igf1r ϩ/Ϫ mice had abnormal glucose tolerance, with a 2-h glucose value of 250 mg/ml (p Ͻ 0.05 by ANOVA versus WT and Igf1r ϩ/Ϫ mice) (Fig. 5). However, intraperitoneal insulin tolerance tests failed to demonstrate a difference in insulin sensitivity among the four genotypes studied (data not shown), consistent with previous observations in Insr ϩ/Ϫ mice (3,4,32,37). These results suggest that Igf1r haploinsufficiency does not cause a major deterioration of metabolic homeostasis in Insr ϩ/Ϫ mice. These data stand in sharp contrast to those we reported previously (32) in mice with combined haploinsufficiency of Insr/Irs1 or Insr/Irs2. We next examined the effect of the combined Insr ϩ/Ϫ -Igf1r ϩ/Ϫ mutations on insulin signaling in muscle, because muscle is the main site of insulin-dependent glucose disposal, and combined ablation of INSR and IGF1R by a dominant negative Igf1r transgene has been shown to cause diabetes (5). To this end, we analyzed patterns of Igf1r expression and phosphorylation, as well as insulin-induced activation of PI 3-kinase in skeletal muscle extracts. Western blotting analysis revealed that Igf1r was expressed in skeletal muscle and that expression was reduced in Igf1r ϩ/Ϫ mice compared with WT, as expected based on the presence of a null allele (Fig. 5b). Expression of Igf1r was not detected in liver extracts (not shown). In skeletal muscle extracts, both insulin and IGF1 induced tyrosine phosphorylation of Igf1r (data not shown). Insulinstimulated PI 3-kinase activity was assessed in vivo in skeletal muscle from mice with various combinations of mutations. Igf1r ϩ/Ϫ mice showed normal insulin-stimulated PI 3-kinase activity. In contrast, Insr ϩ/Ϫ mice showed a ϳ30% decrease and Insr ϩ/Ϫ Igf1r ϩ/Ϫ mice a 37% decrease (Fig. 6). However, the difference between Insr ϩ/Ϫ mice and Insr ϩ/Ϫ Igf1r ϩ/Ϫ mice was not statistically significant.

DISCUSSION
The IGF System in Endocrine Pancreatic Development-The current consensus regarding development of endocrine islet cells is that they derive from Ngn3-positive precursors (38), because genetic ablation of Ngn3 results in the absence of all four endocrine cell types (39). Specification of ␣-, ␤-, ␦-, and pancreatic polypeptide-cell fate requires expression of additional transcription factors, including the homeobox genes Pax-4 for ␤and ␦-cells (40) and Pax-6 for ␣-cells (41). ␤-Cell development is also arrested in mice lacking the insulin gene transcription factors NeuroD/␤2 (42) and Nkx2.2 (43). The growth factors required for islet cell proliferation are unknown, although there is evidence for a role of fibroblast growth factors (44), IGFs (45)(46)(47), hepatic growth factor (48), and placental lactogen (49). Because islet cells are still present in embryos lacking IGF signaling, our data indicate that INSR and IGF1R are not required for embryonic islet development and proliferation. It remains to be seen whether they participate in postnatal ␤-cell growth (50).
Exocrine pancreatic development requires the activity of PTF1 (51), a heterotrimeric protein that directs exocrine pancreas-specific gene expression. Mice lacking the p48 subunit of PTF-1 die shortly after birth due to the complete absence of exocrine pancreatic tissue and display abnormal localization of endocrine cells to the spleen (52). Pancreatic agenesis also results from null mutations of the homeodomain transcription factor PDX1, both in humans (53) and in mice (54 -56). PDX1 plays a critical role in the process of pancreatic differentiation and in the maintenance of a ␤-cell-specific phenotype (57).
In contrast to the wealth of information available on the transcription factors required to initiate pancreatic differentiation, little is known about growth factors that mediate exocrine pancreatic proliferation (35). Several studies have shown that exocrine pancreatic development is crucially dependent upon mesenchymal/epithelial interactions, suggesting that locally produced growth factors play an essential role in this process (55, 58 -60). However, whereas most growth factors can promote growth of pancreatic explant cultures in vitro (61), the factors acting physiologically in vivo are unknown. Recent studies have implicated fibroblast growth factor receptors in exocrine pancreatic growth (62,63), whereas transforming growth factor-␤ (64) and epidermal growth factor (58,61) have been shown to affect the rate of exocrine and endocrine tissue formation. In addition, follistatin has also been reported to promote exocrine pancreas growth (65,66). 4. a, body weights of control and mutant mice. Mice were weighed at 1, 3, and 6 months (n Ն 10 for each genotype). The differences between WT or Insr ϩ/Ϫ versus Igf1r ϩ/Ϫ or Insr ϩ/Ϫ /Igf1r ϩ/Ϫ were statistically significant (p Ͻ 0.05 by ANOVA) at all time points. b, detection of IGF1R in skeletal muscle extracts. Detergent extracts were prepared from mice of the various genotypes and subjected to immunoprecipitation with anti-IGF1R antiserum (C-20) prior to immunoblotting with the same antiserum (upper panel). On a separate filter (lower panel), total extracts were blotted with anti-albumin antiserum to normalize protein levels Our developmental analysis indicates that exocrine pancreas growth is blunted in embryos with combined Insr Ϫ/Ϫ Igf1r Ϫ/Ϫ mutations. Because this phenotype is not observed in single Insr Ϫ/Ϫ or Igf1r Ϫ/Ϫ mutant embryos, it is unlikely to result from impaired insulin or IGF1 signaling, which activates only the cognate receptor. Moreover, the phenotype is already apparent in E12.5 embryos, a stage at which absence of IGF2, but not of IGF1 or insulin, has manifestations (15). We conclude that this growth effect is mediated by IGF2, which binds to both receptors with high affinity (67). These findings are consistent with the observation that IGF signaling promotes growth of cultured acinar cells and amylase gene expression (68 -70). Further studies will be required to examine mecha-Means Ϯ S.E. of ␤-cell area in pancreata from 4-week-old mice. ␤-Cell area is expressed as percentage of the total pancreatic area surveyed. Values are derived from three separate sections for each animal and four animals for each genotype. Sections were spaced Ͼ200 m apart to avoid counting the same islets multiple times. d, classification of islets according to size. Following morphometric evaluation, islets were arbitrarily subdivided into small (Ͻ1,000 arbitrary area units), intermediate (1,000 -3,000 arbitrary area units), or large (Ͼ3,000 arbitrary area units). Values represent the % of small, intermediate, or large islets in each genotype. A total of 200 islets was scored for each genotype. *, p Ͻ 0.05 Insr ϩ/Ϫ /Igf1r ϩ/Ϫ versus WT and Igf1r ϩ/Ϫ mice. e, fasting and fed (f) blood glucose levels were measured in 6-month-old animals of the indicated genotype. Each bar represents means Ϯ S.E. of at least 10 WT, Insr ϩ/Ϫ , Igf1r ϩ/Ϫ , and Insr ϩ/Ϫ Igf1r ϩ/Ϫ mice. g, intraperitoneal glucose tolerance tests were performed in fasted 6-month-old WT (open circles) (n ϭ 6), Insr ϩ/Ϫ (closed squares) (n ϭ 9), Igf1r ϩ/Ϫ (open squares) (n ϭ 3), and Insr ϩ/Ϫ /Igf1r ϩ/Ϫ mice (closed triangles) (n ϭ 12). *, p Ͻ 0.05 Insr ϩ/Ϫ Igf1r ϩ/Ϫ versus WT and Igf1r ϩ/Ϫ mice at 120 min by ANOVA. h, morphologic analysis of pancreatic islets in Insr ϩ/Ϫ Igf1r ϩ/Ϫ mice. Immunohistochemical analysis of islets was performed by staining fixed pancreatic sections with anti-insulin antiserum (green pseudocolor) as described under "Experimental Procedures." FIG. 5. Effect of combined heterozygous Insr and Igf1r mutations on metabolic control. Fasting (a) and fed (b) plasma insulin levels were measured in 6-month-old animals of the indicated genotype. Each bar represents means Ϯ S.E. of at least 10 WT, Insr ϩ/Ϫ , Igf1r ϩ/Ϫ , and Insr ϩ/Ϫ Igf1r ϩ/Ϫ mice. c, morphometric analysis of ␤-cell area.
FIG. 6. Insulin-stimulated PI 3-kinase activity in muscle. Insulin (5 units/kg) was injected intravenously via the inferior vena cava in anesthetized 8 -12-week-old mice. Thereafter, muscle extracts were prepared for immunoprecipitation with anti-phosphotyrosine antibody and subjected to PI 3-kinase assay as described (32). The results are expressed as % of PI 3-kinase activity in insulin-treated WT mice. The data represent means Ϯ S.E. from four independent experiments and four animals for each genotype. A representative autoradiogram is shown on top of the bar graph. nistic aspects of the role of IGF2 in exocrine pancreatic growth and to determine whether this factor is the long-sought and elusive promoter of acinar development (71).
Role of IGF1R in Metabolic Control-The generation of mice with combined mutations of Insr and Igf1r has enabled us to begin to address the role of IGF1R in metabolic control. Although conclusive evidence indicates that INSR plays a role in promoting embryonic growth (16,17), the metabolic role of IGF1R remains elusive. We have shown previously (18) that IGF1 administration to Insr Ϫ/Ϫ mice decreases glucose levels, without affecting overall survival. We have suggested that this results from the ability of IGF1R to compensate for INSR in skeletal muscle but not in liver (19 -21). This interpretation is supported by a recent study (5) in which a dominant negative igf1r transgene, overexpressed in skeletal muscle, caused diabetes as a result of combined inhibition of INSR and IGF1R function, probably mediated by heterotetrameric receptors composed of an INSR monomer and an IGF1R monomer (72).
From our characterization, there does not appear to be further deterioration of insulin sensitivity in Insr ϩ/Ϫ Igf1r ϩ/Ϫ mice compared with Insr ϩ/Ϫ mice. The lack of a marked synergism between the Insr and Igf1r mutations stands in sharp contrast to similar experiments in which Insr ϩ/Ϫ mice were crossed with mice bearing Irs1 and Irs2 mutations (32,37). These data suggest that the role of Igf1r in metabolism is accessory to Insr and not essential for metabolic regulation. This conclusion is entirely consistent with the findings of Fernandez et al. (5), who showed that simultaneous disruption of INSR and IGF1R in muscle is required to cause insulin-resistant diabetes. We suggest that the two receptors have the ability to compensate for each other, but neither is necessary for the metabolic response in muscle. These data should not be interpreted to suggest that muscle metabolism is unimportant for overall fuel homeostasis, because the phenotype of mice with a conditional Glut4 knockout in muscle shows that blocking glucose uptake causes severe insulin resistance (73). Rather, it suggests that there are converging pathways leading to the metabolic response in muscle, which can be activated either by INSR/IGF1R or by AMP kinase-dependent pathways (7).
In summary, we present an extensive analysis of the interactions between INSR and IGF1R in mice. The data reveal a novel role for IGF signaling in exocrine pancreatic development and indicate that IGF1R signaling plays an ancillary role to insulin signaling in adult mice.