Main

We created mice with a β-cell specific knockout of Igf1r by breeding animals carrying Igf1r in which exon 3 was flanked with loxP sites (Igf1rtm1.1Mhz , hereafter referred to as Igf1rlox/lox )13 with mice expressing cre driven by the rat insulin promoter (TgN(ins2-cre)25Mgn, hereafter referred to as cre+/− ; Fig. 1a)14. Control (Igf1rlox/lox and cre+/− ) and knockout mice were born normally and survived equally into adulthood. Homozygous knockout of Igf1r in β-cells but not other islet cells was confirmed by RT–PCR using RNA from βIgf1r−/− islet cells separated by flow cytometry14 (Fig. 1b), and by immunostaining of pancreas sections for Igf1 receptors (Fig. 1c). Pancreas sections from six-month-old control and βIgf1r−/− mice showed no significant differences in β-cell mass (βIgf1r−/− , 1.1 ± 0.3 mg per pancreas versus control, 1.4 ± 0.2 mg per pancreas, n=4, P=NS) or non–β-cell mass (βIgf1r−/− , 0.22 ± 0.08 mg per pancreas versus control, 0.28 ± 0.06 mg per pancreas, n=4, P=NS; Fig. 2a). Similarly, insulin and glucagon content, as determined by immunostaining (Fig. 2b) or islet hormone gene expression assessed by semi-quantitative RT–PCR (Fig. 4), and acid ethanol extraction of pancreas for insulin (Fig. 2c), were not significantly different between the two groups. The lack of an alteration in islet size in the βIgf1r−/− mice, as compared with the decrease in islet mass in Igf1r−/− mice11, indicates that the islet growth defect in the Igf1r−/− mice probably results from loss of Igf1 action in cells other than the β-cell itself.

Figure 1: Conditional Igf1r targeting allows β-cell–specific Igf1r knockout.
figure 1

a , The upper panel shows a partial genomic map of wildtype Igf1r. Individual letters represent restriction sites. The lower panel shows the targeted locus before and after excision of the neomycin selection cassette (Neo) that is flanked by one loxP site upstream and two loxP sites downstream of exon (E) 3. b , RT–PCR analysis of RNA obtained from islet cells sorted by flow cytometry into β- and non–β-cells. Amplification from the knockout transcript (518 bp) is evident in the lane corresponding to the βIgf1r−/− β-cells, indicating absence of exon 3. All other lanes show the intact lox/lox allele (574 bp). c , Representative islets from pancreas sections of a Igf1rlox/lox control and a βIgf1r−/− mouse stained for Igf1 receptors. Scale bar, 25 μm.

Full size image
Figure 2: βIgf1r−/− islets show normal morphology and insulin content but have blunted glucose-stimulated insulin secretion and loss of Igf1-mediated suppression of insulin secretion.
figure 2

a , Hematoxylin and eosin staining of representative sections of pancreas isolated from control and βIgf1r−/− mice. Scale bar, 25 μm. b , Representative pancreas sections visualized by immunofluorescence after staining for insulin (red) using guinea pig anti-human insulin antibody and glucagon (green) using rabbit anti-glucagon antibody. Scale bar, 25 μm. c , Insulin content in islets extracted with acid ethanol. d , After overnight culture, islets isolated from the two groups were incubated in fresh medium for 30 min with increasing concentrations of glucose. The medium was then removed and assayed for insulin using an ELISA kit. The asterisk indicates P<0.01 versus basal (0 mM), and the cross indicates P<0.05, βIgf1r−/− versus controls (Igf1rlox/lox and cre+/− ); n=5. e , Cultured islets were placed in fresh medium containing Igf1 and different concentrations of glucose for 30 min. The medium was then removed and assayed for insulin. The asterisk indicates P<0.01 versus basal (0 mM), and the cross indicates P<0.01, βIgf1r−/− versus controls; n=5.

Full size image
Figure 3: βIgf1r−/− mice lack glucose-stimulated, acute-phase insulin secretion and have glucose intolerance.
figure 3

a , b , Acute (first) phase and second-phase insulin secretion after an i.p. injection of glucose (a) or L-arginine (b), in male βIgf1r−/− and control mice. The asterisk indicates P<0.01, βIgf1r−/− mice versus controls; n=8–12. c , Blood glucose levels following an i.p. injection of glucose. The asterisk indicates P<0.001, βIgf1r−/− mice versus controls; n=8–10. d , Insulin tolerance tests were carried out on random-fed βIgf1r−/− and control male mice at 10 wk. Blood glucose levels before and after an i.p. injection of human regular insulin are shown; n=8–12.

Full size image
Figure 4: Reduced steady-state mRNA levels of islet-enriched transcription factors and glucose-sensing proteins in βIgf1r−/− islets.
figure 4

RT–PCR analysis of islet-enriched transcription factors, glucose-sensing proteins, islet hormones and insulin receptors in islets isolated from βIgf1r−/− and control male mice. Samples were normalized using Gapd and the mitochondrial marker hypoxanthine guanosine phosphoribosyl transferase (Hprt). Quantitative measurements were obtained by densitometry and expressed as a ratio of the mean values of βIgf1r−/− mice ( −/− ) to controls (con). The single asterisk indicates P<0.05, and the double asterisk indicates P<0.001, βIgf1r−/− versus controls; n=4–5. Igf1r, Igf1 receptor; Tcf2, transcription factor 1; Foxa2, forkhead Box A2; Tcf4, transcription factor 4; Tcf1, transcription factor 1; Slc2a2, solute carrier family 2 member 2; Gck, glucokinase; Ins1/2, insulin I/II; Insr, insulin receptor.

Full size image

In addition to its potential role in islet growth, Igf1 has been shown to suppress insulin release15. To determine the presence of potential secretory defects, islets isolated from βIgf1r−/− and control mice were cultured overnight and then stimulated with glucose. Control islets showed a concentration-dependent stimulation in insulin secretion with an increase of roughly threefold at 11.1 mM glucose (Fig. 2d). In the βIgf1r−/− islets, we detected no stimulation at 5.5 mM glucose and only small increases at higher glucose concentrations (Fig. 2d). To determine the effect of Igf1 on insulin secretion in vitro, we carried out similar experiments on islets cultured in the presence of 100 nM Igf1. In the basal state and in the absence of glucose, βIgf1r−/− islets secreted higher amounts of insulin than controls, suggesting the normal suppressive effect of Igf1 is reduced in βIgf1r−/− islets (Fig. 2e). Despite the higher basal secretion, upon glucose challenge, the βIgf1r−/− islets had significantly lower insulin secretion than the controls (Fig. 2e), indicating a glucose-stimulatory defect similar to that observed in vivo.

We detected a mild increase in in vivo fasting and fed circulating insulin levels in βIgf1r−/− mice, whereas we did not find significant differences in circulating glucagon and glucose levels between the groups (Table 1). These data suggest that in vivo, similar to the results obtained in vitro, inhibition of insulin secretion by Igf1 is lost in the βIgf1r−/− mice, leading to an increase in circulating insulin. We also found significantly higher C-peptide levels, a surrogate marker for insulin secretion, in the βIgf1r−/− mice in the fasting and fed states (Table 1). The normal sensitivity of these knockout mice to injected insulin rules out the possibility of insulin resistance (Fig. 3d). In addition, to evaluate whether peripheral insulin sensitivity is altered in βIgf1r−/− mice secondary to changes in hormones and metabolites, we measured Igf1, leptin, corticosterone, free fatty acids, triglycerides and cholesterol. No significant differences were detected between knockout mice and controls at approximately two months of age (Table 1). Taken together, these results indicate that the mild elevation in circulating insulin in vivo may compensate to maintain blood glucose in the normal range in βIgf1r−/− mice.

Table 1 Metabolic parameters in control and βIgf1r−/− mice
Full size table

To assess the impact of localized Igf1 resistance on regulated insulin secretion in vivo, we examined insulin secretion in response to intraperitoneal injection of glucose or L-arginine. In agreement with the findings in vitro, we observed an absent acute (first phase) insulin response after glucose stimulation and a blunted second-phase insulin secretion pattern in the βIgf1r−/− mice, compared with robust responses in both the Igf1rlox/lox and cre+/− controls (Fig. 3a). Consequently, during the glucose tolerance test, glucose levels remained high in the βIgf1r−/− mice for up to two hours, indicating severe glucose intolerance (Fig. 3c). By contrast, we observed similar insulin secretory responses between the two groups after L-arginine injection, confirming a glucose-specific secretory defect in the knockouts (Fig. 3b).

To identify the mechanism underlying the loss of glucose-stimulated insulin secretion, we evaluated steady-state mRNA levels of glucose-sensing proteins and the hepatocyte nuclear factor (Hnf) family of transcription factors16 in islet RNA from the two groups. mRNA from βIgf1r−/− islets showed a nearly complete loss of Igf1r expression, whereas levels of insulin receptor mRNA were similar between the groups (Fig. 4). Notably, we detected decreases of 35–50% in Slc2a2 and Gck expression levels in the βIgf1r−/− mice, indicating that loss of the β-cell Igf1r negatively influences the expression of the glucose-sensing proteins (Fig. 4). Moreover, we observed significantly reduced Foxa2 (also known as Hnf3b) and Tcf2 (also known as HNF-1β) mRNA levels in the βIgf1r−/− islets, suggesting a role for Igf1r in modulating the expression of early transcription factors.

These results indicate that Igf1r in β-cells is not crucial for islet growth, but is important in differentiated function of the β-cell. These results seem to conflict with studies showing that Igf1r−/− or Irs2−/− mice, or mutants with a combination of the two genotypes, have decreased β-cell mass11. In these earlier models, however, it is not clear whether the alterations are due to a defect in Igf1 signaling in the β-cells themselves or due to defects in Igf1 signaling at all. Thus, in global knockouts (Igf1r−/− ), Igf1 might affect either primordial stem cells or ductal cells that may give rise to β-cells, or endothelial cells or other cells that support islet growth12, or might provide important paracrine or endocrine factors for islet growth and differentiation. Similarly, the defects in islet development in Irs2−/− mice might reflect defects in Igf1-independent signaling pathways, as Irs2 acts in growth hormone, prolactin and cytokine signaling17,18, all of which may influence islet growth, survival and function. In addition, genetic background may contribute to the differences in phenotypes observed between these mouse models19.

In the βIgf1r−/− model, the results are more clear-cut, as the defect is in a single signaling pathway in a single cell type. However, because the βIgf1r−/− mice were created using a cre transgene controlled by the rat insulin promoter, we cannot rule out the possibility that Igf1r is important before insulin expression at day 9 of embryogenesis20, similar to the role of Fgf-7 (refs 10,21). Another caution is that we have examined the role of Igf1 receptors in islet growth only under normal circumstances. Further study will be required to determine whether Igf1 receptors are important in the islet compensatory response observed in many diabetic states22, or in response to injury23.

Finally, it is worth comparing the effects of loss of Igf1r with loss of the insulin receptor gene (Insr) in β-cells. Although neither receptor is essential for prenatal islet growth, the role of Insr may be more important than Igf1r in postnatal growth. Mice lacking Insr in β-cells (βInsr−/− mice) show a decrease in the age-dependent growth of islets14, whereas postnatal growth is normal in βIgf1r−/− mice. By contrast, both βInsr−/− and βIgf1r−/− mice have glucose-sensing defects. It has been shown that insulin stimulation can increase intracellular calcium concentrations in β-cells24 and increase the levels of glucokinase mRNA (refs 25,26), and that insulin signaling is required for normal glucose sensing by pancreatic β-cells14,25,26. These effects depend on signaling through the Irs proteins24,25,26,27,28, involving both the PI 3-kinase and p70 S6 kinase pathways24,25,26,28. The similarity of the defects in the βIgf1r−/− and βInsr−/− mice might reflect the use of insulin/Igf1 hybrid receptors for these overlapping signaling pathways3,4. Preliminary findings in mice with β-cell–specific knockout of Insr and Igf1r indicate that the defects due to functional loss of both receptors are additive and lead to early-onset severe diabetes (R.N.K. and C.R.K., unpublished data). Thus, defects in insulin and Igf1 signaling in the β-cell indicate an important link between Igf1 and insulin action, Igf1 and insulin resistance and alterations in the function of the pancreatic β-cell.

Methods

Animals.

Igf1r floxed mice were created as described previously13 and maintained on a mixed (C57BL/6 × 129/Sv) genetic background. Mice expressing cre recombinase under a rat insulin promoter14 were maintained on a mixed (C57BL/6 × 129Sv × DBA/2) genetic background and bred with the floxed mice in the Joslin Animal Facility. All mice were housed in pathogen-free facilities on a 12-h light/dark cycle and had free access to water and food. We used male mice at 2–6 mo for all experiments. All protocols for animal use and euthanasia were reviewed and approved by the Animal Care Committee of the Joslin Diabetes Center and were in accordance with the guidelines of the National Institutes of Health. We carried out genotyping by PCR using genomic DNA isolated from the tail tip.

Islet morphology and immunohistochemistry.

Tissues were fixed in Bouin solution and 10% buffered formalin and embedded in paraffin. Sections of pancreas were stained for non–β-cell hormones using a mixture of antibodies to glucagon, somatostatin and pancreatic polypeptide14. We evaluated β-cell mass by point-counting morphometry on immunoperoxidase-stained sections of pancreas14,27. We carried out immunofluorescent staining for insulin and glucagon using guinea pig anti-human insulin antibody (Linco) and rabbit anti-glucagon antibody (Zymed). Secondary antibodies included Texas Red dye–conjugated donkey anti–guinea pig IgG (Jackson Immunoresearch) for insulin and FITC-conjugated goat anti-rabbit IgG (Zymed) for glucagon. We carried out immunofluorescent staining for Igf1r using a rabbit polyclonal antibody against the Igf1r α-subunit (Santa Cruz Biotechnology) and detected with a TRITC-conjugated goat anti-rabbit IgG (Zymed). Images were captured using a confocal microscope (Zeiss).

Analysis of metabolic parameters.

We determined blood glucose values from whole venous blood using an automated glucose monitor (Glucometer Elite, Bayer). We measured serum insulin and leptin levels by ELISA, using mouse insulin and leptin standards (Crystal Chem.). For glucagon, blood samples were collected in tubes containing aprotinin (1 μg ml−1, Bayer), and the serum was stored at −80 °C before radio-immunoassay (Linco). We measured serum levels of Igf1 using an RIA kit (ALPCO) and determined serum levels of corticosterone by RIA (ICN Biomedicals). We measured triglyceride levels in serum from fasted animals by a colorimetric enzyme method using the GPO-Trinder Assay (Sigma). Free fatty acid levels were analyzed in serum from fasted animals using the NEFA-C-Kit (WAKO). We analyzed total cholesterol levels colorimetrically, using a kit (Sigma).

Glucose and insulin tolerance tests and in vivo and in vitro insulin secretion.

We carried out glucose tolerance tests and acute insulin secretion tests on animals that had been fasted overnight for 16 h (ref. 14). Animals were injected i.p. with 2 g kg−1 body weight of glucose (glucose tolerance tests) or 3 g kg−1 body weight of glucose (acute insulin secretion) or L-arginine (10 mM in 0.2 M PBS, acute insulin secretion). For glucose tolerance tests, glucose levels were measured from blood collected from the tail immediately before, and 15, 30, 60 and 120 min after, the injection. For acute insulin secretion, blood samples were collected before and 2, 5, 15 and 30 min after injection of the stimulus, and serum insulin was measured as described above. We carried out insulin tolerance tests on animals in the random-fed state. Animals were injected i.p. with 0.75 U kg−1 body weight of human regular insulin (Eli Lilly). We collected blood samples before and 15, 30 and 60 min after injection. Islets were isolated using the intra-ductal collagenase method27. We measured in vitro insulin release from isolated islets by static incubation of islets cultured overnight. We measured insulin content in whole pancreas and islets in acid-ethanol extracts using an ELISA kit (Crystal Chem.).

Flow cytometry and RT–PCR.

Isolated islets were dispersed into β- and non–β-cell fractions by flow cytometry, and RT–PCR analysis was carried out as previously described14. We prepared RNA from the cell fractions using an RNeasy Mini Kit (Qiagen). We used three oligonucleotides to detect Igf1r mRNA from exon 3– and exon 3+ alleles with comparable efficiency in single triplex RT–PCR (Pharmacia Biotech). To measure steady-state mRNA levels, we carried out semi-quantitative RT–PCR with islet RNA (Trizol, Gibco BRL), using a previously described method29. Contaminating DNA was removed using 1 μL of RNase-free DNase-I (Boehringer) per 5 μg of RNA. cDNAs were synthesized and used as templates for PCRs using specific primers at annealing temperatures of 60–65 °C in the presence of dNTPs and Taq polymerase. We generally used 20–25 cycles for amplification in the linear range. We quantified bands using densitometry (Biorad) and calculated ratios of expression levels of knockout and control mice. Primer sequences are available on request.