The role of insulin receptor substrate 2 in hypothalamic and β cell function

Insulin receptor substrate 2 (Irs2) plays complex roles in energy homeostasis. We generated mice lacking Irs2 in β cells and a population of hypothalamic neurons (RIPCreIrs2KO), in all neurons (NesCreIrs2KO), and in proopiomelanocortin neurons (POMCCreIrs2KO) to determine the role of Irs2 in the CNS and β cell. RIPCreIrs2KO mice displayed impaired glucose tolerance and reduced β cell mass. Overt diabetes did not ensue, because β cells escaping Cre-mediated recombination progressively populated islets. RIPCreIrs2KO and NesCreIrs2KO mice displayed hyperphagia, obesity, and increased body length, which suggests altered melanocortin action. POMCCreIrs2KO mice did not display this phenotype. RIPCreIrs2KO and NesCreIrs2KO mice retained leptin sensitivity, which suggests that CNS Irs2 pathways are not required for leptin action. NesCreIrs2KO and POMCCreIrs2KO mice did not display reduced β cell mass, but NesCreIrs2KO mice displayed mild abnormalities of glucose homeostasis. RIPCre neurons did not express POMC or neuropeptide Y. Insulin and a melanocortin agonist depolarized RIPCre neurons, whereas leptin was ineffective. Insulin hyperpolarized and leptin depolarized POMC neurons. Our findings demonstrate a critical role for IRS2 in β cell and hypothalamic function and provide insights into the role of RIPCre neurons, a distinct hypothalamic neuronal population, in growth and energy homeostasis.

Insulin regulates peripheral energy homeostasis by acting on multiple tissues to control carbohydrate, lipid, and protein metabolism (1). Gene targeting in mice has shown that β cell deletion of the insulin receptor causes reduced first-phase insulin release, reduced β cell insulin content, and progressive deterioration in glucose tolerance (2). Early studies of the effects of insulin in the CNS demonstrated a role for intracerebroventricularly administered insulin in the control of food intake and body weight (3). Mouse brain insulin receptor deletion causes mild hyperphagia and adiposity in female mice, diet-sensitive obesity, and defects in reproductive function (4). Results from studies in which insulinomimetics and insulin receptor antisense were centrally administered also support a role for CNS insulin signaling in energy homeostasis regulation (5, 6). Insulin signaling mechanisms therefore regulate β cell and CNS function, but it is unclear which postreceptor components mediate which physiological effects and, in the case of the CNS, which neuronal populations are involved. It is also unclear how insulin signaling components interact with other molecules involved in energy homeostasis, such as leptin.

Insulin receptor substrate (Irs) proteins lie downstream of the activated insulin and type 1 insulin-like growth factor receptor (7). Gene targeting studies have revealed distinct physiological roles for the 4 major Irs proteins (7). Mice lacking Irs1 display profound growth retardation and insulin resistance but, due to β cell compensation, do not develop diabetes (8, 9). Irs3- and Irs4-null mice have minimal metabolic, endocrine, and growth phenotypes (7). In contrast, mice lacking Irs2 develop diabetes due to insulin resistance and pancreatic β cell dysfunction (10). These studies have suggested that Irs2 is the major mediator of the metabolic effects of insulin and have identified a novel role for Irs2 signaling in the maintenance of β cell mass. Irs2 signaling also plays complex roles in neuroendocrine function. Female Irs2-null mice are infertile, hyperphagic, and develop obesity (11). Irs2-null mice are reported to be leptin resistant (12), which suggests that Irs2 acts as a point of convergence for leptin and insulin signaling.

It has recently been reported that mice lacking Irs2 in both pancreatic β cells and a poorly defined hypothalamic neuronal population (generated using a rat insulin 2 promoter Cre [RIPCre] recombinase transgene) display reduced islet mass, impaired glucose tolerance, and hypothalamic dysfunction (13, 14). Unlike mice with global deletion of Irs2, these mice did not develop progressive diabetes. One study (13) reported that they displayed a hypothalamic phenotype suggesting melanocortin dysfunction, proposed that RIPCre neurons are proopiomelanocortin (POMC) neurons, and noted β cell recovery with time through an undetermined mechanism. Another study (14) demonstrated reduced islet mass, obesity, and leptin resistance but did not report other features of hypothalamic dysfunction or long-term analysis of islet function. Consequently, these studies have raised a number of critical but unanswered questions (15). In particular it is unclear (a) why there is a relative recovery of the β cell phenotype; (b) what role the CNS plays in the β cell phenotype and the impairment of glucose homeostasis; (c) what is the role of Irs2 in leptin action; and (d) what are the identity and characteristics of the hypothalamic neuronal population involved (15).

To address these questions, we generated mice lacking Irs2 in specific cell types: RIPCreIrs2KO mice, lacking Irs2 in β cells and a poorly characterized population of hypothalamic neurons; NesCreIrs2KO mice, lacking Irs2 in all neurons but not the endocrine pancreas; and POMCCreIrs2KO mice, lacking Irs2 in POMCCre-expressing cells.

Generation of mice with a floxed allele of Irs2. We generated mice with a floxed allele of Irs2 (Irs2lox mice) to permit its deletion in different cell types and tissues (Figure 1, A–E). Irs2lox mice, when bred to homozygosity, were phenotypically indistinguishable from wild-type animals and displayed normal Irs2 expression (data not shown).

Figure 1

Generation of Irs2flox mice and characteristics of RIPCre and POMCCre mice. (A) Schema of targeting construct design, simplified restriction map of the Irs2 locus, the locus after homologous recombination and the deletion of neomycin cassette (Neo), and Southern blotting and PCR genotyping strategies used to identify these events. External probe A was used to identify homologous recombination (HR), probe B to detect the selection cassette, and probe C to detect the coding region of Irs2. HSV-tk, herpes simplex virus thymidine kinase. (B) Southern blot analysis with probe A demonstrating homologous recombination after targeting. (C and D) Southern blots using probe B after Cre-mediated recombination demonstrating deletion (Del) of the neomycin cassette and using probe C to demonstrate retention of Irs2 coding region confirming type 2 recombination. (E) PCR analysis with primers P1 and P2 of HR clone that has lost the neomycin cassette but retained the loxP site downstream of the Irs2 coding region. (F–H) We examined Cre expression in RIPCre mice by analyzing GFP expression in pancreatic sections costained with insulin and hypothalamic sections from RIPCreZEG mice. (I) We examined Cre expression in POMCCre mice by analyzing hypothalamic sections from POMCCreZEG mice. (J) Combined ISH for POMC and ICC for GFP was performed in POMCCreZEG mice to confirm Cre expression in POMC neurons. (K) Immunohistochemistry for IRS2 (red) was performed in hypothalamic sections from POMCCreZEG mice. Scale bars: 50 μm (H, I, and K) and 10 μm (J). 3V, third ventricle; ME, median eminence.

Using 3 Cre recombinase transgenic lines, we generated RIPCreIrs2KO, NesCreIrs2KO, and POMCCreIrs2KO mice. We also intercrossed RIPCre and POMCCre mice with ZEG (16) and Rosa26lacZ (17) indicator mice on both wild-type and Irs2lox/lox backgrounds. This allowed identification of cell types in which recombination occurred and permitted Irs2 deletion and GFP or lacZ expression in the same cell types. RIPCreZEG mice expressed GFP in pancreatic islets, hypothalamic structures such as the arcuate, lateral, and ventromedial hypothalamus (VMH), and a discrete population of forebrain neurons (Figure 1, F–H, and data not shown). In POMCCreZEG mice, GFP expression was found in a discrete population of arcuate nucleus neurons (Figure 1I). Combined in situ hybridization (ISH) and immunocytochemistry (ICC) studies demonstrated colocalization of GFP and POMC (Figure 1J) in more than 90% of cells examined, which indicated that most POMC neurons also expressed functional Cre recombinase. No GFP expression was found in pancreatic β cells (data not shown). NesCre mice delete in neurons but not pancreatic endocrine cells (18) and have been used to manipulate insulin signaling in the CNS (4). Irs2 expression was found in pancreatic islets and in the arcuate, the VMH, and the paraventricular nucleus (Figure 1K and Figure 2B and data not shown).

Figure 2

Analysis of deletion of Irs2 in islets and hypothalami from RIPCreIrs2KO, POMCCreIrs2KO, and NesCreIrs2KO mice. (A) We performed PCR analysis to detect recombination of the Irs2 locus in DNA from islets and hypothalami (hypo) of control and knockout mice. The presence of a 1.3-kb PCR product indicates recombination and deletion of the Irs2 gene. (B and C) Western blot analysis of Irs2 in islets from control and RIPCreIrs2KO mice (B) and whole brain and hypothalamic lysates from control and NesCreIrs2KO mice (C). (D) Immunofluorescence analysis for Irs2 expression in the hypothalami of RIPCreZEG and RIPCreIrs2KOZEG mice. Colocalization of GFP (green) and Irs2 (red) is seen in RIPCreZEG mice (indicated by white arrows) and no colocalization in RIPCreIrs2KOZEG mice. RIPCreGFP, GFP expression in RIPCre cells. (E) Colocalization of POMCCre expression and Irs2 in POMCCreZEG mice. (F) Colocalization of POMCCre expression and Irs2 in POMCCrelacZ mice and no colocalization of GFP and Irs2 in POMCCreIrs2KOlacZ mice. Confocal images of representative arcuate nucleus fields are shown in D–F. Scale bars: 10 μm.

To confirm deletion of Irs2 in the RIPCreIrs2KO, NesCreIrs2KO, and POMCCreIrs2KO mice, we analyzed recombination of the Irs2flox allele by PCR in islet and hypothalamic DNA. In 4-week-old mice, the recombined allele was detected in both tissues in RIPCreIrs2KO mice (Figure 2A) but only in the hypothalami of NesCreIrs2KO and POMCCreIrs2KO mice (Figure 2A). Immunoprecipitation and Western blotting confirmed the lack of Irs2 expression in isolated islets from RIPCreIrs2KO and in either whole brain or hypothalamic lysates from NesCreIrs2KO mice (Figure 2, B and C). ICC analysis of RIPCreZEG mice demonstrated Irs2 colocalization with GFP in some but not all hypothalamic arcuate neurons (Figure 2D, left panel). In RIPCreIrs2KOZEG mice, there was no colocalization of Irs2 and GFP, which confirmed Irs2 deletion in RIPCre neurons (Figure 2D, right). Irs2 expression remained in other hypothalamic neurons. Analysis of Irs2 expression in POMCCreZEG mice demonstrated colocalization of Irs2 with GFP in some POMC neurons (Figure 2E). Irs2 deletion was confirmed in POMCCreIrs2KOlacZ mice (we were unable to breed POMCCreIrs2KOZEG mice), but Irs2 expression remained in other hypothalamic arcuate neurons (Figure 2F). Normal Irs2 expression was detected in muscle, liver, adipose, and other tissues in mice of all 3 lines (data not shown).

Glucose homeostasis in RIPCreIrs2KO, NesCreIrs2KO, and POMCCreIrs2KO mice. In 4-week-old male RIPCreIrs2KO, NesCreIrs2KO, and POMCCreIrs2KO mice, fasting blood glucose levels were indistinguishable from those of control animals (data not shown). At 12 weeks of age, when male Irs2-null mice are diabetic, RIPCreIrs2KO mice had higher fasting blood glucose levels than control mice, and glucose tolerance tests revealed defective glucose disposal (Figure 3, A and B). However, the elevation in fasting blood glucose levels was modest, and thus RIPCreIrs2KO mice did not display the overt diabetes and catabolic state that Irs2-null mice develop. Fasting glucose levels were also elevated in NesCreIrs2KO mice, and glucose tolerance was mildly impaired (Figure 3, A and C). RIPCreIrs2KO and NesCreIrs2KO mice at 6 months of age and older did not develop markedly elevated fasting blood glucose levels but remained glucose intolerant (Figure 3D and data not shown). In contrast, fasting glucose levels and glucose tolerance were equivalent to those of controls in POMCCreIrs2KO mice at 12 weeks and older (Figure 3A and data not shown).

Figure 3

Glucose homeostasis in RIPCreIrs2KO, NesCreIrs2KO, and POMCCreIrs2KO mice. (A) Fasting blood glucose levels of 12-week-old male mice of the indicated genotypes were measured after a 16-hour overnight fast. (B). Glucose tolerance tests were performed on 12-week-old male RIPCreIrs2KO (filled squares) and control mice (open squares). (C) Glucose tolerance tests were performed on 12-week-old male NesCreIrs2KO (filled squares) and control mice (open squares). (D) Fasting blood glucose levels of 6-month-old male mice of the indicated genotypes were measured after a 16-hour overnight fast. (E) Fasting blood insulin levels of 12-week-old male mice of the indicated genotypes were measured after a 16-hour overnight fast. Data in A–E represent the mean ± SEM for 8–10 animals of each genotype. (F) We calculated the percentage of the total pancreatic area occupied by β cells in 12-week-old male mice of the indicated genotypes using insulin-stained pancreatic sections. The right side shows data for 9-month-old RIPCreIrs2KO and control mice. Four pancreata were analyzed per genotype at each time point, and for each pancreas, 4 sections were analyzed. The data presented are mean ± SEM for 4 mice of each genotype. *P < 0.05, **P < 0.01, and ***P < 0.001. Cont, control.

At 12 weeks, RIPCreIrs2KO and NesCreIrs2KO mice, but not POMCCreIrs2KO mice, displayed fasting hyperinsulinemia compared with control animals, which suggests development of insulin resistance (Figure 3E). At 6 months, both RIPCreIrs2KO and NesCreIrs2KO mice remained hyperinsulinemic (data not shown). At 4 weeks, when there was a 50% reduction in β cell mass in Irs2-null mice (10), no decrease in β cell mass was seen in RIPCreIrs2KO compared with control animals (data not shown). However, by 12 weeks, RIPCreIrs2KO mice had a 40% reduction in β cell mass compared with control mice (Figure 3F, left), consistent with the deterioration in glucose homeostasis at this age. In contrast, 12-week-old NesCreIrs2KO mice exhibited an increased β cell mass, evidence of a compensatory response to insulin resistance (Figure 3F, left). POMCCreIrs2KO did not display altered β cell mass compared with age-matched controls (data not shown).

To investigate the preservation of glucose homeostasis compared with Irs2-null mice, we examined β cell mass in 9-month-old RIPCreIrs2KO mice. RIPCreIrs2KO mice had a reduced islet area compared with control animals (Figure 3F, right), but the area had increased from that seen in 12-week-old RIPCreIrs2KO mice, which suggests partial recovery in β cell mass (Figure 3F, right).

Hypothalamic function in RIPCreIrs2KO, NesCreIrs2KO, and POMCCreIrs2KO mice. RIPCreIrs2KO and NesCreIrs2KO mice displayed increased body weight compared with control mice starting around 4–5 weeks of age. By 12 weeks of age, they were 20% heavier than control animals (Figure 4A). RIPCreIrs2KO and NesCreIrs2KO mice were significantly longer than control mice (Supplemental Figure 1; supplemental material available online with this article; doi:10.1172/JCI200524445DS1). MRI scanning performed at 16 weeks showed increased fat mass that was more significant in NesCreIrs2KO mice (Figure 4B). At 12 weeks, 24-hour food consumption was greater in both RIPCreIrs2KO and NesCreIrs2KO mice than in control animals (Figure 4C). In contrast, analysis of resting metabolic rate, core temperature, and activity revealed no differences between RIPCreIrs2KO or NesCreIrs2KO mice and control animals (Supplemental Table 1). Furthermore POMCCreIrs2KO mice did not display increased body weight (Figure 4A), were of the same length as controls (Supplemental Figure 1), and did not eat more than control mice (Figure 4C).

Figure 4

Hypothalamic function in RIPCreIrs2KO, NesCreIrs2KO, and POMCCreIrs2KO mice. (A) Body weight was measured in 12-week-old male mice of the indicated genotypes. (B) Total body fat was determined by MRI in 16-week-old male mice of the indicated genotypes. (C) Cumulative 24-hour food intake was measured in 12-week-old male mice of the indicated genotypes. (D) Fasting blood leptin levels of 12-week-old male mice of the indicated genotypes were measured after a 16-hour overnight fast. (E–G) Cumulative food intake, percent reduction in food intake compared with baseline, and reduction in bodyweight were measured over a 3-day period of treatment with leptin (5 mg/kg) or vehicle in 6–8-week-old male mice of the indicated genotypes. (H) Food intake and fasting leptin levels were determined in 5-week-old male mice of the indicated genotypes. Data in A–H data represent the mean ± SEM for 8–10 animals of each genotype. *P < 0.05, **P < 0.01, and ***P < 0.001.

At 12 weeks of age when RIPCreIrs2KO and NesCreIrs2KO mice displayed marked obesity, leptin levels were significantly elevated in these mice but not in POMCCreIrs2KO mice (Figure 4D). We administered leptin (5 mg/kg) peripherally for 3 days to acclimatized 6-week-old mice. This treatment inhibited cumulative 3-day food intake in control animals by 20% compared with vehicle-treated animals (Figure 4, E and F). RIPCreIrs2KO and NesCreIrs2KO mice, despite being hyperphagic, also responded to leptin treatment with a 20% decrease in cumulative food intake (Figure 4, E and F). The reduction in food intake was accompanied in control, RIPCreIrs2KO, and NesCreIrs2KO mice by a significant reduction in body weight (Figure 4G). To further investigate the temporal development of the feeding phenotype, we measured leptin levels and food intake in 5-week-old male RIPCreIrs2KO mice. At this age, RIPCreIrs2KO mice were hyperphagic but did not display hyperleptinemia (Figure 4H). Findings for NesCreIrs2KO mice were similar (data not shown).

Characterization of β cell recovery in RIPCreIrs2KO mice. Our studies and others (13, 15, 19) raise the question of the mechanism for the relative recovery of the β cell mass in RIPCreIrs2KO mice with age. Analysis of 4-week-old RIPCreIrs2KOZEG mice demonstrated that a vast majority of insulin-positive cells coexpressed GFP, which suggests that recombination was occurring in most β cells (Figure 5, A and B). However, we detected small numbers of insulin-expressing cells that were GFP negative (Figure 5B), which indicates lack of recombination and that the RIPCre promoter was not active in all β cells. It would be predicted that Irs2 would not be deleted in these cells and therefore they are involved in the preservation or recovery of β cell mass. Indeed, analysis of 9-month-old RIPCreIrs2KOZEG mice demonstrated that in many islets, a significant proportion of the insulin-positive β cells no longer expressed GFP (Figure 5, C and D). Comparison of the numbers of GFP-positive β cells in 9-month-old mice demonstrated that GFP expression loss occurred significantly more often in RIPCreIrs2KOZEG than control mice (percentage of GFP-negative β cells: RIPCreIrs2KOZEG, 50.9% ± 1.2% vs. RIPCreZEG, 10.1% ± 1.1%; P < 0.001; 2,062 β cells sampled in 54 islets from 4 RIPCreIrs2KOZEG mice and 2,987 β cells sampled in 45 islets from 3 RIPCreZEG mice).

Figure 5

Analysis of recombination and duct-marker expression in islets from RIPCreIrs2KOZEG mice. (A–D) Immunofluorescence staining for insulin (red) in islets from 4-week-old (A and B) and 9-month-old (C and D) RIPCreIrs2KOZEG mice stained for DNA (DAPI, blue). White arrows indicate β cells that do not express GFP and have not undergone recombination. Many cells in the 9-month-old islet do not express GFP. (E–G) Labeling of ducts with DBA in islets from 9-month-old RIPCreIrs2KOZEG mice stained blue for either DNA (E and F) or insulin (G). No duct marker is expressed in GFP-negative β cells. Scale bars: 100 μm.

To further characterize the mechanism of β cell recovery, we used a duct-specific lectin (Dolichos biflorus agglutinin [DBA]) to identify duct epithelial cells (20). GFP or insulin expression was not detected in DBA-positive duct structures in RIPCreIrs2KOZEG mice at 9 months of age (Figure 5, E–G). Likewise, in 9-month-old RIPCreIrs2KOZEG mice, in 296 islets sampled from 4 mice, no DBA-positive cells were found within islets (Figure 5, E–G). These findings suggest that duct cells that have differentiated to insulin-producing cells within ducts or duct cells that migrate to islet structures before converting to β cells do not make a significant contribution to the β cell repopulation. These findings do not exclude the possibility that ductal precursor cells that express neither duct markers nor insulin are involved in this process.

Characterization of RIPCre neurons and comparison with POMC neurons. Our studies demonstrate that RIPCre neurons play a role in energy homeostasis and growth and that deletion of Irs2 in these cells generates a phenotype with similarities to that of melanocortin pathway mutants. However, our genetic evidence suggests that deletion of Irs2 in POMC neurons does not account for the hypothalamic phenotype seen in RIPCreIrs2KO or NesCreIrs2KO mice. RT-PCR analysis of expression of hypothalamic feeding peptides in RIPCreIrs2KO and NesCreIrs2KO mice demonstrated reduced POMC expression (POMC mRNA: 42% reduction in NesCreIrs2KO mice normalized to control mice, P < 0.05, n = 5; 48% reduction in RIPCreIrs2KO mice normalized to control mice, P < 0.05, n = 8) but normal neuropeptide Y (NPY) and agouti-related protein (AgRP) expression (data not shown), which suggests an effect on melanocortin circuits. No alteration in hypothalamic POMC, AgRP, and NPY expression was seen in POMCCreIrs2KO mice. Therefore, to characterize the relationship between RIPCre neurons and melanocortin pathways, we performed ISH/ICC analysis and electrophysiological studies of RIPCre and POMCCre neurons. POMC- and GFP-expressing neurons in the hypothalamus were clearly identified by ISH and ICC, respectively, in the same slices (Figure 6, A and B), and no GFP-positive neurons colocalized with POMC mRNA. POMC and RIPCre neurons, however, are closely associated with each other. Similarly, no coexpression of GFP was demonstrated in NPY neurons, but close apposition of these neuronal populations was observed (Figure 6, C and D). Combined ICC for α melanocyte-stimulating hormone (αMSH) and GFP demonstrated no coexpression, but many GFP neurons were in close proximity to αMSH-positive fibers, which suggests that they may be targets for POMC neurons (Figure 6E).

Figure 6

Characterization of RIPCre neurons by ICC, ISH, and electrophysiological analysis. (A and B) Dual ISH for POMC and ICC for GFP was performed on hypothalamic sections from RIPCreZEG mice. Representative low- (A) and high-power (B) views are shown. (C and D) Dual ISH for NPY and ICC for GFP was performed on hypothalamic sections from RIPCreZEG mice. Representative low- (C) and high-power (D) views are shown. ARC, arcuate nucleus. (E) Fluorescence ICC for POMC was performed on hypothalamic sections from RIPCreZEG mice. Neurons expressing POMC are red and those expressing GFP are green. Arrows indicate noncolocalized neurons and GFP neurons that are abutted by POMC fibers. A representative confocal image is shown. Scale bars: 50 μm (A and C) and 10 μm (B, D, and E). (F–I) Current-clamp recordings were made from RIPCreZEG neurons in the absence and presence of the mixed melanocortin 3/4 receptor agonist MTII, as indicated above the traces. (F) Representative recording demonstrating MTII depolarization, resulting in a small increase in spike frequency (G). This depolarization is more clearly represented in neurons that have sub-spike thresholds, achieved by application of a constant hyperpolarizing current (H). MTII also depolarizes neurons in the presence of TTX (I), indicative of a direct action. The action potentials in panel H have been truncated to show clearly the depolarizing effect of MTII. Note that the MTII depolarization was irreversible over the time course of the recordings in cells represented in H and I.

Whole-cell current-clamp recordings were made from GFP-positive hypothalamic arcuate neurons from RIPCreZEG mice. Under control recording conditions, RIPCre neurons spontaneously fired action potentials and had a mean resting potential of –50 ± 1 mV and input resistance of 1,862 ± 197 MΩ (n = 62). Bath application of the αMSH analog melanotan II (MTII) (a nonselective melanocortin 3/4 receptor agonist) at 100 nM caused long-lasting depolarization in the majority of cells tested (from –53 ± 3 mV to –48 ± 2 mV; P < 0.01; n = 9/13), associated with increased spike firing (Figure 6, F and G).

To demonstrate the MTII excitatory effect more clearly, we applied a small hyperpolarizing current to individual neurons, such that action potential frequency was markedly reduced. Under these conditions, MTII (100 nM) caused depolarization and increased firing frequency (Figure 6H). Furthermore, the application of MTII (100 nM) to RIPCre neurons, following incubation of the slice with tetrodotoxin (TTX; 500 nM) to block spike firing and prevent synaptic transmission, caused significant depolarization (n = 4), which indicates a direct action (Figure 6I). To further characterize RIPCre neurons and differentiate them from POMC (and NPY) neurons, we determined the actions of insulin and leptin on the electrical properties of RIPCre and POMC neurons. Leptin has been demonstrated to depolarize POMC neurons (21), and our current results support this finding. Pressure ejection of leptin (50 nM) for 60 seconds directly above the recorded neuron caused an increase in spike frequency and a small depolarization (from –47 ± 2 mV to –44 ± 3 mV; P < 0.04) of a minority population of POMC neurons (n = 5/13) (Figure 7, A–C). In contrast, leptin (50 nM) did not affect action potential frequency or resting membrane potential (–46 ± 3 mV vs. –47 ± 3 mV; n = 10; Figure 7, D and E) in RIPCre neurons. Pressure ejection of insulin (10 nM) for 60 seconds produced a significant increase in action potential frequency, concomitant with cell depolarization (from –49 ± 4 to –40 ± 4 mV; P < 0.01) in the majority of recordings (n = 6/9) from RIPCre neurons (Figure 7, F–H). In contrast, pressure injection of insulin (10 nM) onto POMC neurons resulted in hyperpolarization (from –48 ± 1 mV to –55 ± 2 mV; P < 0.001) and reduced action potential frequency in most (n = 9/11) neurons (Figure 7, I and J). The effect of insulin, MTII, and leptin on neuronal activity was extremely slow to reverse and was, in many cases, irreversible over the time course of the experiments. These experiments suggest that RIPCre neurons are a population distinct from either POMC or NPY neurons but lie within a melanocortin- and insulin-regulated neuronal mechanism.

Figure 7

Electrophysiological characteristics of RIPCre and POMC neurons in response to insulin and leptin. (A) Representative recordings from individual arcuate neurons are shown for POMCCreZEG (A–C, I, and J) and RIPCreZEG (D–H). Leptin (50 nM) and insulin (10 nM) were pressure-ejected (2–5 psi) for 60 seconds (as denoted by arrows). (A) Leptin depolarizes the POMCCreZEG neuron, and this is associated with significant attenuation of spike amplitude. The recording from the neuron shown in B was hyperpolarized to subthreshold potentials by injection of constant current (2–5 pA) through the recording electrode. Application of leptin depolarized and increased the firing rate of the neuron. (C) Diary plot of spike frequency for the neuron shown in B. In contrast, leptin had no effect on membrane potential (D) and spike frequency (E) in RIPCreZEG neurons. Insulin caused depolarization of the RIPCreZEG neuron (F) and increased spike frequency (G), with spike attenuation. As shown in H, the neuron was hyperpolarized to subthreshold potentials as above, and application of insulin clearly depolarizes this neuron with a large increase in spike frequency. In contrast, insulin caused hyperpolarization of POMCCreZEG neurons (I), which was accompanied by a reduction in firing rate (J).

Through global gene deletion, we and others have previously implicated Irs2 in the regulation of β cell function and peripheral glucose metabolism and the central regulation of energy homeostasis (10, 11, 19, 22). These studies have suggested but not proved that Irs2 mediates both β cell growth and leptin and insulin action in the CNS. To elucidate the contribution of Irs2 to this complex phenotype in specific tissues, we generated mice with a conditional allele of Irs2 to permit deletion in a number of cell types. We initially used RIPCre transgenic mice to generate RIPCreIrs2KO mice lacking Irs2 in β cells and a population of hypothalamic neurons, and indeed it has recently been reported that RIPCreIrs2KO mice display phenotypes due to this mixed pattern of deletion (13, 14). These mice developed reduced β cell mass, obesity, somatic overgrowth, hyperinsulinemia, and hyperleptinemia but did not progress to an overt diabetic phenotype because of an unexplained recovery in β cell mass. However, because of the inability to dissociate β cell and hypothalamic phenotypes, a number of questions regarding the role of Irs2 remain unanswered (15). To directly address such issues that we encountered with our own RIPCreIrs2KO mice, we also generated NesCreIrs2KO mice, lacking Irs2 in all neurons but not the endocrine pancreas, and POMCCreIrs2KO mice, lacking Irs2 in hypothalamic POMC neurons, which have been implicated in mediating leptin’s effects on energy homeostasis (23).

Consistent with recent reports (13, 14), 12-week-old RIPCreIrs2KO mice had reduced β cell mass and displayed mildly elevated fasting blood glucose levels, impaired glucose tolerance, and hyperinsulinemia. They developed a marked hypothalamic phenotype resembling that of mice with defects in melanocortin action. In contrast, NesCreIrs2KO mice, which have normal β cell Irs2 expression, had increased β cell mass in response to insulin resistance and displayed the obesity and overgrowth syndrome. These findings demonstrate that β cell mass reduction is due to Irs2 deletion solely in β cells and not in the CNS. However, NesCreIrs2KO did develop elevated fasting blood glucose levels and hyperinsulinemia, which suggests that neuronal Irs2 pathways are required for normal glucose homeostasis but that the increase in β cell mass seen in NesCreIrs2KO mice could compensate for insulin resistance.

In contrast to Lin et al. (13), we found that in 9-month-old RIPCreIrs2KO mice, there was a persistent reduction in β cell mass compared with that of control mice, although the mass was increased compared with that of 3-month-old RIPCreIrs2KO mice. At later ages, therefore, RIPCreIrs2KO mice had glucose intolerance but did not display marked fasting hyperglycemia and never developed the marked progressive diabetic phenotype of the global Irs2-null mouse. Anatomical studies of recombination using a GFP indicator mouse in either a wild-type or Irs2flox background revealed that as early as 4 weeks, small numbers of β cells did not harbor the recombination event. By 9 months of age, significant numbers of β cells in RIPCreIrs2KO mice not expressing GFP had repopulated the islets, but many GFP-positive cells undergoing recombination were still present, which explains the persistent reduction in β cell mass. Our studies demonstrate that the reduction in recombination occurs predominantly in RIPCreIrs2KOZEG as opposed to RIPCreZEG islets, which suggests that Irs2 signaling itself is required for expression of the RIPCre transgene. Indeed, insulin gene expression appears to require an intact insulin signaling pathway in the β cell (2, 24, 25). With Irs2 largely absent within β cells, under conditions of insulin resistance there may be potential selection for preexisting β cells not expressing the RIPCre transgene and therefore not undergoing deletion of Irs2 to expand and repopulate the islet. In 9-month-old mice, we did not detect GFP expression and Cre-mediated recombination in pancreatic ducts, and so potentially the GFP-negative β cells could have arisen from duct cells. However, we could not detect expression of a duct cell marker within islets of 9-month-old RIPCreIrs2KO mice nor could we find insulin expression in ducts. The absence of deletion in ducts also suggests that the β cell phenotype in RIPCreIrs2KO mice results from loss of Irs2 in mature insulin-producing cells rather than any precursor or stem cell population. Taken together, these findings provide anatomical evidence that a small number of preexisting β cells that escape Irs2 deletion at an early stage in RIPCreIrs2KO mice repopulate the islets. Our studies cannot categorically exclude the possibility that duct-derived precursors expressing neither insulin nor duct markers contribute to the repopulation process. However, recent lineage tracing studies (20) suggest that proliferation of mature β cells is the major mechanism for β cell expansion in adult mice. Our findings suggest that persistent Irs2 expression in mature β cells is required for the compensatory response to insulin resistance.

The preservation of β cell mass in NesCreIrs2KO mice dissociates hypothalamic dysfunction from reduced β cell mass. It has been suggested that nestin expression is a characteristic of pancreatic islet stem cells (26, 27), but recently lineage tracing studies, utilizing the same NesCre mouse used in our studies, demonstrated that pancreatic exocrine, ductal, and endothelial cells but not islet endocrine cells express nestin (18, 28). The finding of increased islet mass in NesCreIrs2KO mice further supports the concept that nestin-positive cells do not contribute to the endocrine pancreas, as deletion of Irs2 in these cells would have been anticipated to reduce β cell mass.

The hypothalamic phenotypes seen in RIPCreIrs2KO and NesCreIrs2KO mice suggested defects in melanocortin function, and in RIPCreIrs2KO and NesCreIrs2KO mice, we detected reduced hypothalamic POMC expression. Deletion of the leptin receptor in POMC neurons generates mice that have some phenotypic features in common with RIPCreIrs2KO mice, including obesity (23). It has been suggested that RIPCre neurons are POMC neurons and that Irs2 signaling events are required for central leptin action (14, 29). Our results, however, do not confirm these findings and provide evidence to define the role and sites of Irs2 pathways in the CNS. We demonstrate that deletion of Irs2 in POMC neurons did not recapitulate the hypothalamic phenotype and that RIPCre neurons do not express POMC or NPY. Furthermore, RIPCreIRS2KO and NesCreIRSKO mice were sensitive to peripherally administered leptin at 6 weeks of age. Five-week-old RIPCreIRS2KO and NesCreIRSKO mice are hyperphagic but not hyperleptinemic, which suggests that elevated leptin levels seen at later ages result from greater adiposity due to increased energy intake rather than primary central leptin resistance. This is consistent with the observation that central leptin administration to Irs2-null mice causes reduction in fat mass (12). Young nonobese hyperphagic melanocortin 4 receptor–null mice, which resemble RIPCreIRS2KO and NesCreIRSKO mice, are sensitive to the anorexigenic effects of leptin but develop leptin resistance as they become obese (30). In preliminary observations, mice lacking Irs2 in AgRP neurons, generated using an AgRPCre transgenic mouse (31), also do not develop a marked hypothalamic phenotype (H. Al-Qassab et al., unpublished observations). Therefore, these studies implicate Irs2 signaling in neurons distinct from POMC and AgRP/NPY populations in the regulation of energy homeostasis and growth and suggest that Irs2 is not an obligatory component of CNS leptin signaling pathways.

Electrophysiological analysis showed that RIPCre neurons depolarized in response to insulin and MTII, which suggests that they could be mediating the effects of insulin in the CNS in a melanocortin-sensitive neuronal pathway. Furthermore, we found no electrophysiological effect of leptin upon arcuate RIPCre neurons, which suggests that this is not a leptin-sensitive neuronal population and therefore may explain why RIPCreIRS2KO mice are not leptin resistant. In contrast, leptin depolarized and insulin hyperpolarized POMC neurons. These opposite effects upon neuronal activity perhaps explain the lack of phenotype in POMCCreIrs2KO mice and distinguish RIPCre neurons electrophysiologically from both POMC and NPY neurons (21, 32). It remains possible that the RIPCre neurons in other hypothalamic regions such as the VMH may be mediating the effects of leptin, and we have started to examine this possibility using additional mouse mutants. However, our observation of leptin sensitivity in the NesCreIrs2KO mice suggests that CNS Irs2 is not required for leptin action.

Our studies have demonstrated that CNS Irs2 pathways acting in a neuronal population distinct from POMC and NPY neurons regulate energy homeostasis and growth. Arcuate RIPCre neurons are targets for insulin and melanocortin but not leptin action, but it remains to be determined what neuropeptides and transmitters are expressed in this population. Irs2 in β cells is required for the maintenance of β cell mass, but β cells that escape Cre-mediated recombination are able to repopulate islets with time. Our observations clarify the role of Irs2 in β cell function and energy homeostasis and suggest that modulation of Irs2 function may be of use in the treatment of diabetes and obesity.

View Supplemental data

We thank Jonathan Godwin for blastocyst injections, Corinne Lobe for providing ZEG mice, Marika Charalambous for advice and discussion, Steven Lingard for technical assistance, and Andrew Tinker and Matthew Glyn for assistance with confocal microscopy. We thank the Biological Imaging Centre (BIC), Imaging Sciences Department, Imperial College London, for assistance with the MRI studies. The work was supported by grants from the Wellcome Trust (to A. Choudhury, C. Selman, M.A. Smith, M. Simmgen, K. Hisadome, M.L.J. Ashford, and D.J. Withers); the Biotechnology and Biological Sciences Research Council (BBSRC) (to M. Claret); the Medical Research Council (MRC; to H. Al-Qassab, I. Diakonov, J.D. Bell, R.L. Batterham, and D.J. Withers); Diabetes UK (to A. Choudhury and D.C. Bedford); the NIH (DK-48506, to G.S. Barsh); the Stanford Bio-X Interdisciplinary Initiatives Program (to G.S. Barsh); and by an AstraZeneca BBSRC CASE award (to H. Heffron). Part of this work was conducted within the BetaCellTherapy consortium that is supported as an integrated project by the sixth European Union–framework program. R.L. Batterham is an MRC Clinician Scientist. D.J. Withers is an MRC Senior Clinical Fellow.

Agharul I. Choudhury, Helen Heffron, Mark A. Smith, and Hind Al-Qassab contributed equally to this work.

Nonstandard abbreviations used: AgRP, agouti-related protein; DBA, Dolichos biflorus agglutinin; DIG, digoxigenin; ICC, immunocytochemistry; Irs, insulin receptor substrate; Irs2lox mice, mice with a floxed allele of Irs2; ISH, in situ hybridization; αMSH, α melanocyte-stimulating hormone; MTII, melanotan II; NesCreIrs2KO mice, mice lacking Irs2 in nestin Cre-expressing cells; NPY, neuropeptide Y; POMC, proopiomelanocortin; POMCCreIrs2KO mice, mice lacking Irs2 in POMC-expressing cells; RIP, rat insulin 2 promoter; RIPCre mice, mice expressing Cre under the control of the rat insulin 2 promoter; RIPCreIrs2KO mice, mice lacking Irs2 in β cells and a poorly characterized population of hypothalamic neurons; TTX, tetrodotoxin; VMH, ventromedial hypothalamus.

Conflict of interest: The authors have declared that no conflict of interest exists.

1. DeFronzo, RA. Pathogenesis of type 2 diabetes: metabolic and molecular implications for identifying diabetes genes. Diabetes Rev. 1997. 5:177-268.
2. Kulkarni, RN, et al. Tissue-specific knockout of the insulin receptor in pancreatic beta cells creates an insulin secretory defect similar to that in type 2 diabetes. Cell. 1999. 96:329-339.
   View this article via: PubMed CrossRef Google Scholar
3. Woods, SC, Lotter, EC, McKay, LD, Porte (Jr), D. Chronic intracerebroventricular infusion of insulin reduces food intake and body weight of baboons. Nature. 1979. 282:503-505.
   View this article via: PubMed CrossRef Google Scholar
4. Bruning, JC, et al. Role of brain insulin receptor in control of body weight and reproduction. Science. 2000. 289:2122-2125.
   View this article via: PubMed CrossRef Google Scholar
5. Obici, S, Feng, Z, Karkanias, G, Baskin, DG, Rossetti, L. Decreasing hypothalamic insulin receptors causes hyperphagia and insulin resistance in rats. Nat. Neurosci. 2002. 5:566-572.
   View this article via: PubMed CrossRef Google Scholar
6. Air, EL, et al. Small molecule insulin mimetics reduce food intake and body weight and prevent development of obesity. Nat. Med. 2002. 8:179-183.
   View this article via: PubMed CrossRef Google Scholar
7. Withers, DJ. Insulin receptor substrate proteins and neuroendocrine function. Biochem. Soc. Trans. 2001. 29:525-529.
   View this article via: PubMed CrossRef Google Scholar
8. Tamemoto, H, et al. Insulin resistance and growth retardation in mice lacking insulin receptor substrate-1. Nature. 1994. 372:182-186.
   View this article via: PubMed CrossRef Google Scholar
9. Araki, E, et al. Alternative pathway of insulin signalling in mice with targeted disruption of the IRS-1 gene. Nature. 1994. 372:186-190.
   View this article via: PubMed CrossRef Google Scholar
10. Withers, DJ, et al. Disruption of IRS-2 causes type 2 diabetes in mice. Nature. 1998. 391:900-904.
    View this article via: PubMed CrossRef Google Scholar
11. Burks, DJ, et al. IRS-2 pathways integrate female reproduction and energy homeostasis. Nature. 2000. 407:377-382.
    View this article via: PubMed CrossRef Google Scholar
12. Suzuki, R, et al. Both insulin signaling defects in the liver and obesity contribute to insulin resistance and cause diabetes in Irs2(–/–) mice. J. Biol. Chem. 2004. 279:25039-25049.
    View this article via: PubMed CrossRef Google Scholar
13. Lin, X, et al. Dysregulation of insulin receptor substrate 2 in β cells and brain causes obesity and diabetes. J. Clin. Invest. 2004. 114:908-916. doi:10.1172/JCI200422217..
    View this article via: JCI PubMed Google Scholar
14. Kubota, N, et al. Insulin receptor substrate 2 plays a crucial role in β cells and the hypothalamus. J. Clin. Invest. 2004. 114:917-927. doi:10.1172/JCI200421484..
    View this article via: JCI PubMed Google Scholar
15. Brady, MJ. IRS2 takes center stage in the development of type 2 diabetes. J. Clin. Invest. 2004. 114:886-888. doi:10.1172/JCI200423108..
    View this article via: JCI PubMed Google Scholar
16. Novak, A, Guo, C, Yang, W, Nagy, A, Lobe, CG. Z/EG, a double reporter mouse line that expresses enhanced green fluorescent protein upon Cre-mediated excision. Genesis. 2000. 28:147-155.
    View this article via: PubMed CrossRef Google Scholar
17. Soriano, P. Generalized lacZ expression with the ROSA26 Cre reporter strain. Nat. Genet. 1999. 21:70-71.
    View this article via: PubMed CrossRef Google Scholar
18. Delacour, A, Nepote, V, Trumpp, A, Herrera, PL. Nestin expression in pancreatic exocrine cell lineages. Mech. Dev. 2004. 121:3-14.
    View this article via: PubMed CrossRef Google Scholar
19. Kubota, N, et al. Disruption of insulin receptor substrate 2 causes type 2 diabetes because of liver insulin resistance and lack of compensatory beta-cell hyperplasia. Diabetes. 2000. 49:1880-1889.
    View this article via: PubMed CrossRef Google Scholar
20. Dor, Y, Brown, J, Martinez, OI, Melton, DA. Adult pancreatic beta-cells are formed by self-duplication rather than stem-cell differentiation. Nature. 2004. 429:41-46.
    View this article via: PubMed CrossRef Google Scholar
21. Cowley, MA, et al. Leptin activates anorexigenic POMC neurons through a neural network in the arcuate nucleus. Nature. 2001. 411:480-484.
    View this article via: PubMed CrossRef Google Scholar
22. Withers, DJ, et al. Irs-2 coordinates Igf-1 receptor-mediated beta-cell development and peripheral insulin signalling. Nat. Genet. 1999. 23:32-40.
    View this article via: PubMed Google Scholar
23. Balthasar, N, et al. Leptin receptor signaling in POMC neurons is required for normal body weight homeostasis. Neuron. 2004. 42:983-991.
    View this article via: PubMed CrossRef Google Scholar
24. Leibiger, IB, Leibiger, B, Moede, T, Berggren, PO. Exocytosis of insulin promotes insulin gene transcription via the insulin receptor/PI-3 kinase/p70 s6 kinase and CaM kinase pathways. Mol. Cell. 1998. 1:933-938.
    View this article via: PubMed CrossRef Google Scholar
25. Kulkarni, RN, et al. Altered function of insulin receptor substrate-1-deficient mouse islets and cultured beta-cell lines. J. Clin. Invest. 1999. 104:R69-R75.
    View this article via: JCI PubMed CrossRef Google Scholar
26. Zulewski, H, et al. Multipotential nestin-positive stem cells isolated from adult pancreatic islets differentiate ex vivo into pancreatic endocrine, exocrine, and hepatic phenotypes. Diabetes. 2001. 50:521-533.
    View this article via: PubMed CrossRef Google Scholar
27. Lumelsky, N, et al. Differentiation of embryonic stem cells to insulin-secreting structures similar to pancreatic islets. Science. 2001. 292:1389-1394.
    View this article via: PubMed CrossRef Google Scholar
28. Treutelaar, MK, et al. Nestin-lineage cells contribute to the microvasculature but not endocrine cells of the islet. Diabetes. 2003. 52:2503-2512.
    View this article via: PubMed CrossRef Google Scholar
29. Tobe, K, et al. Increased expression of the sterol regulatory element-binding protein-1 gene in insulin receptor substrate-2(–/–) mouse liver. J. Biol. Chem. 2001. 276:38337-38340.
    View this article via: PubMed CrossRef Google Scholar
30. Marsh, DJ, et al. Response of melanocortin-4 receptor-deficient mice to anorectic and orexigenic peptides. Nat. Genet. 1999. 21:119-122.
    View this article via: PubMed CrossRef Google Scholar
31. Kaelin, CB, Xu, AW, Lu, XY, Barsh, GS. Transcriptional regulation of agouti-related protein (Agrp) in transgenic mice. Endocrinology. 2004. 145:5798-5806.
    View this article via: PubMed CrossRef Google Scholar
32. Roseberry, AG, Liu, H, Jackson, AC, Cai, X, Friedman, JM. Neuropeptide Y-mediated inhibition of proopiomelanocortin neurons in the arcuate nucleus shows enhanced desensitization in ob/ob mice. Neuron. 2004. 41:711-722.
    View this article via: PubMed CrossRef Google Scholar
33. Postic, C, et al. Dual roles for glucokinase in glucose homeostasis as determined by liver and pancreatic beta cell-specific gene knock-outs using Cre recombinase. J. Biol. Chem. 1999. 274:305-315.
    View this article via: PubMed CrossRef Google Scholar
34. Tronche, F, et al. Disruption of the glucocorticoid receptor gene in the nervous system results in reduced anxiety. Nat. Genet. 1999. 23:99-103.
    View this article via: PubMed CrossRef Google Scholar
35. Xu, AW, et al. PI3K integrates the action of insulin and leptin on hypothalamic neurons. J. Clin. Invest. 2005. 115:951-958. doi:10.1172/JCI200524301..
    View this article via: JCI PubMed Google Scholar
36. Livak, KJ, Schmittgen, TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods. 2001. 25:402-408.
    View this article via: PubMed CrossRef Google Scholar
37. Ross, R, Leger, L, Guardo, R, De Guise, J, Pike, BG. Adipose tissue volume measured by magnetic resonance imaging and computerized tomography in rats. J. Appl. Physiol. 1991. 70:2164-2172.
    View this article via: PubMed Google Scholar
38. Tang, H, Vasselli, JR, Wu, EX, Boozer, CN, Gallagher, D. High-resolution magnetic resonance imaging tracks changes in organ and tissue mass in obese and aging rats. Am. J. Physiol. Regul. Integr. Comp. Physiol. 2002. 282:R890-R899.
    View this article via: PubMed Google Scholar