Sulfonylurea Receptor Type 1 Knock-out Mice Have Intact Feeding-stimulated Insulin Secretion despite Marked Impairment in Their Response to Glucose*

The ATP-sensitive potassium channel is a key molecular complex for glucose-stimulated insulin secretion in pancreatic β cells. In humans, mutations in either of the two subunits for this channel, the sulfonylurea type 1 receptor (Sur1) or Kir6.2, cause persistent hyperinsulinemic hypoglycemia of infancy. We have generated and characterized Sur1 null mice. Interestingly, these animals remain euglycemic for a large portion of their life despite constant depolarization of membrane, elevated cytoplasmic free Ca2+ concentrations, and intact sensitivity of the exocytotic machinery to Ca2+. A comparison of glucose- and meal-stimulated insulin secretion showed that, although Sur1 null mice do not secrete insulin in response to glucose, they secrete nearly normal amounts of insulin in response to feeding. Because Sur1 null mice lack an insulin secretory response to GLP-1, even though their islets exhibit a normal rise in cAMP by GLP-1, we tested their response to cholinergic stimulation. We found that perfused Sur1 null pancreata secreted insulin in response to the cholinergic agonist carbachol in a glucose-dependent manner. Together, these findings suggest that cholinergic stimulation is one of the mechanisms that compensate for the severely impaired response to glucose and GLP-1 brought on by the absence of Sur1, thereby allowing euglycemia to be maintained.

Glucose-stimulated insulin secretion by the pancreatic ␤ cell requires the coupling of changes in glucose metabolism to alterations in membrane potential (1)(2)(3)(4). In response to a rise in the intracellular ATP/ADP ratio that occurs with glucose metabolism the closure of ATP-sensitive potassium (K ATP ) 1 channels causes the ␤ cell membrane to depolarize. This, in turn, leads to the opening of voltage-gated L-type Ca 2ϩ channels, a rise in the cytoplasmic free Ca 2ϩ concentration ([Ca 2ϩ ] i ), and the subsequent exocytosis of insulin (5). The ␤ cell K ATP channel is an octameric complex of two proteins: an inward-rectifier K ϩ channel, Kir6.2, and the sulfonylurea receptor type 1 (Sur1), which are present in a 4:4 stoichiometry (6, 7). Kir6.2, which forms the channel pore, possesses intrinsic ATP sensitivity (8,9), whereas Sur1, a member of a superfamily of ATPbinding cassette transporter proteins, provides sites for interaction with Mg-ADP (10). Sulfonylureas, which are widely used for treatment of patients with type 2 diabetes mellitus, act by binding to K ATP channels and stimulating their closure (10).
Mutations in Sur1 are a frequent cause of persistent hyperinsulinemic hypoglycemia of infancy (PHHI), an autosomal recessive disorder characterized by excess and unregulated secretion of insulin. Because initial identification of Sur1 as a candidate gene for PHHI by Aguilar-Bryan et al. (11), more than 50 different mutations in this gene, as well as 2 mutations in Kir6.2, have been identified in PHHI patients (12). Analyses of pancreatic ␤ cells from PHHI patients, as well as functional studies of mutated K ATP channels introduced into cultured cells, suggest that impaired K ATP channel function and/or expression is a common mechanism underlying PHHI (13)(14)(15).
Both Kir6. 2 and Sur1 knock-out mice have previously been shown to maintain euglycemia, despite the absence of functional K ATP channel in their islets (16,17). However, the mechanisms that enable mice lacking K ATP channels to maintain regulated insulin secretion are not known. Because identification of these mechanisms may lead to new insights into the regulation of insulin secretion, we have also generated mice that lack Sur1. We found that Sur1 null mice secrete normal amounts of insulin upon meal ingestion, largely through intact second phase insulin secretory responses, and that an intact response to cholinergic stimulation may explain the ability of  these mice to regulate their insulin secretion in the absence of K ATP channels. These results suggest that, although K ATP channels play an important role in glucose-stimulated insulin secretion, other insulin secretagogues act to stimulate secretion of this hormone through K ATP channel-independent mechanisms.

EXPERIMENTAL PROCEDURES
Cloning and Targeted Disruption of the Sur1 Gene-Both BAC (129/ SvJ) and P1 (129/Ola) clones containing the Sur1/Kir6.2 gene locus were isolated by PCR screening (Genome Systems, Inc., St. Louis, MO). The gene targeting strategy involved the placement of tandemly oriented loxP sites around a 1.01-kbp DNA fragment that contains the proximal promoter and exon 1 of Sur1, and its subsequent removal with cre recombinase (Fig. 1B). Positive selection was achieved with a phosphoglycerol kinase-neomycin resistance gene (pgk-neo) cassette, and negative selection with a phosphoglycerol kinase-herpes simplex virus type I thymidine kinase gene (pgk-tk) cassette placed outside the 3Ј arm of the targeting vector. After electroporation of RW4 ES cells (Genome Systems, Inc., St. Louis, MO), 21 of 321 clones (6.5%) were found to be resistant to both G418 and gancyclovir and to have the desired 10.1-kb band after digestion with KpnI and hybridization with the BamHI/BglII probe (Fig. 1C). Chimeric animals were generated by microinjection of two different ES cell clones into C57BL/6 blastocysts. Germline transmission of the Sur1 loxϩneo allele was obtained with ES cell clone 2G4. The Sur1 loxϩneo allele was converted to the Sur1 neo allele by pronuclear microinjection of 1 ng/l CMV-cre expression vector (pBS185) into embryos obtained from matings of Sur1 loxϩneo/w and Sur1 w/w mice.
Husbandry and Genotyping-Mice were fed a standard rodent chow diet, maintained on 12-h light/dark cycle, and were specific pathogenfree. The Sur1 loxϩneo , Sur1 neo , and Sur1 w alleles were distinguished by PCR using two different sets of primers. The primers SUR.1 (5Ј-CAAT-TCCTCAACTGAGGCTCTTAA) and SUR.2 (5Ј-TCGCAGAGTGACCT-CACAGCCTGT) amplify a 530-bp DNA fragment from the Sur1 loxϩneo allele and a 412-bp fragment from the Sur1 w allele. The primers SUR.1 and Neo-5Ј (5Ј-AGCCTCGTTCCACATACACTTCA) generate a 414-bp fragment from the Sur1 neo allele. The predicted band of 1566 bp from the Sur1 loxϩneo allele is not detected using standard PCR conditions. Mice with a mixed genetic background (129/SvJ ϫ C57BL/6) were used in all studies except for pancreas perfusion study, with wild-type littermates serving as the controls. In the pancreas perfusion study, congenic animals that were produced by backcrossing for nine generations with C57BL/6J mice were used. In this case, the controls were either wildtype littermates or age-matched C57BL/6J animals.
Preparation of Islets and Single ␤ Cells-Islets of Langerhans were isolated from 8 -10-week-old mice by a collagenase technique (18) and dispersed into single cells in Ca 2ϩ -and Mg 2ϩ -deficient medium as previously described (19).
Electrophysiology-Whole-cell K ϩ currents were recorded using the perforated-patch configuration of the patch-clamp technique as described previously (20). The rate of exocytosis was determined as changes in cell capacitance (21). Cell capacitance was measured for Ͼ2 min with 0.2-s intervals after establishment of the standard whole-cell configuration. The pipette solution (intracellular-like solution) contained 125 mM potassium glutamate, 10 mM KCl, 10 mM NaCl, 1 mM MgCl 2 , 5 mM Hepes, 3 mM Mg-ATP, 10 mM EGTA, and 0, 2, 5, 8, 9, or 9.8 mM CaCl 2 . The free Ca 2ϩ concentrations in these solutions were estimated to 0, 0.05, 0.22, 0.9, 2.0, and 10 M, respectively, using the binding constants of Martell and Smith (22). Rates of exocytosis are presented as the increase in cell capacitance observed during the first 60 s following establishment of the whole-cell configuration, excluding any rapid changes occurring during the first 10 s required for equilibration of the pipette solution with cytosol.
Measurements of [Ca 2ϩ ] i -Cells were loaded with 2 M fura-2/AM in culture medium containing 3 mM glucose at 37°C for 30 min. Fluorometry using two excitation wavelengths at 340 and 380 nm was performed on either an Axiovert 35M microscope (Zeiss, Oberkerchen, Germany) equipped with an imaging system from Life Science Resources (Cambridge, United Kingdom), or an Axiovert 135TV microscope (Zeiss) equipped with a cooled CCD camera (Photometrics Ltd., Tucson, AZ) and an Inovision imaging system (Inovision Corp., Durham, NC).
Measurements of cAMP in Islets-Groups of 10 islets were preincubated in Krebs-Ringer bicarbonate buffer containing 115 mM NaCl, 4.7 mM KCl, 2.6 mM CaCl 2 , 1.2 mM KH 2 PO 4 , 1.2 mM MgSO 4 , 20 mM NaHCO 3 , 16 mM HEPES, 2 mg/ml bovine serum albumin with 1 mM isobutylmethylxanthine at 37°C for 1h and incubated for another 15 min in same buffer with or without 10 mM GLP-1-(7-36). The reaction was stopped by 50 mM HCl and neutralized with NaOH. Concentration of cAMP was determined by a cAMP 125 I scintillation proximity assay (Amersham Biosciences).
In Vivo Studies-Hyperglycemic clamp studies were performed using chronically cannulated, 12-20-week-old conscious mice as described previously (23). Mice were fasted for 6 h prior to experimentation. A 4-Ci bolus of tracer ([3-3 H]glucose, PerkinElmer Life Sciences) was given at Ϫ100 min, followed by a constant infusion at 0.04 Ci/min for the remainder of the 220-min study. The tracer and glucose solution were infused through a cannula implanted into the right jugular vein. The glucose turnover rate (mg⅐kg Ϫ1 ⅐min Ϫ1 ) was calculated as the rate of tracer infusion (dpm/mg) divided by the corrected plasma glucose specific activity (dpm/mg) per kilogram body weight of the mouse. The glucose clearance (ml⅐kg Ϫ1 ⅐min Ϫ1 ) was calculated as the rate of glucose turnover rate divided by the plasma glucose concentration (mg/ml). Blood samples were taken from the cannula implanted into the left carotid artery at indicated time.
Pancreas Perfusion-Mice at 8 -16 weeks of age were fasted for 16 h prior to experimentation. In situ pancreas perfusion was performed according to the method of Bonnevie-Neilsen et al. (24) with some modifications. After ligating the superior mesenteric, hepatic, splenic, and right and left renal arteries, and tying off the aorta just below the diaphragm, the celiac trunk was perfused with oxygenated Krebs-Ringer bicarbonate buffer containing 1% bovine serum albumin and 3% Dextran T70 (Amersham Biosciences) at 1 ml/min through a catheter placed in the aorta. The effluent was collected at 1-min intervals from the portal vein.
Biochemical Analyses-Blood glucose levels were determined by a glucose oxidase method using a blood glucose analyzer (Hemocue, Mission Viejo, CA). Plasma glucose levels were determined by the UV method (25). Plasma insulin concentrations were measured using either RIA kit (Linco Research, St. Louis, MO) or ELISA kit (Crystal Chem, Chicago, IL). Insulin concentrations in the pancreas perfusate were determined by RIA using anti-insulin coated tubes (ICN, Orangeburg, NY) and radiolabeled insulin (Diagnostic Products, Los Angeles, CA). Plasma GLP-1 (active form) concentrations were determined using ELISA kit (Linco Research, St. Louis, MO).
Statistical Analysis-Results are expressed as mean Ϯ S.E. Differences between groups were evaluated using Student's t test. p Ͻ 0.05 was considered to be significant.

Sur1 and Kir6.2 Gene Structure-Overlapping BAC and P1
clones containing the mouse Sur1/Kir6.2 gene locus were obtained and analyzed by DNA sequencing to determine intron/ exon structure. The mouse Sur1 gene was found to consist of 39 exons ranging in size from 33 to 243 nucleotides and to span ϳ80 kbp (see Fig. 1A). In contrast, the Kir6.2 gene is intronless, spans only 1.5 kbp, and lies ϳ5 kbp downstream of the 3Ј end of the Sur1 gene.
Sur1 Knock-out Mice-The gene targeting strategy shown in Fig. 1 (B-D) was used to generate mice containing the Sur1 loxϩneo allele. To create a null allele for Sur1 (Sur1 neo ), a ϳ1-kbp gene segment containing both promoter and exon 1 sequences was removed by cre-mediated recombination. Analysis of 199 offspring from the intercrossing of mice that were heterozygous for the Sur1 neo allele (e.g. Sur1 neo/w , where w indicates the wild type allele) showed a Mendelian distribution of genotypes (50 Sur1 w/w , 107 Sur1 neo/w , and 42 Sur1 neo/neo ). Mice lacking Sur1 (e.g. Sur1 neo/neo ) were fertile and indistinguishable from their wild-type littermates in term of growth rate (data not shown).
Elevated [Ca 2ϩ ] i in Sur1-deficient Mouse ␤ Cells-Pancreatic ␤ cells from mice lacking Sur1 exhibited a continuous train of action potentials that were blocked by an inhibitor of the L-type voltage-dependent Ca 2ϩ channel, similar to that reported by Seghers et al. (17). This is consistent with the lack of functional K ATP channels in Sur1 neo/neo ␤ cells as assessed by measurements of whole-cell K ϩ currents. By using the standard wholecell configuration of the patch-clamp technique, cells were depolarized to potentials between Ϫ60 and ϩ50 mV in 10-mV steps from a holding potential of Ϫ70 mV. This protocol was applied twice, immediately and at 3 min after establishment of configuration using a pipette solution devoid of Mg-ATP. The whole-cell K ϩ currents were significantly increased as intracellular ATP was washed out in control (Sur1 w/w ) ␤ cells, indicating the activation of ATP-dependent K ϩ currents ( Fig. 2A). However, wash out of ATP did not affect the corresponding currents in Sur1 neo/neo ␤ cells (Fig. 2B). We also observed elevated basal [Ca 2ϩ ] i in Sur1 neo/neo ␤ cells, as seen previously in Kir6.2 knock-out mouse ␤ cells (16). Sur1 w/w ␤ cells typically responded with a prompt increase in [Ca 2ϩ ] i as glucose was increased from 3 to 17 mM (n ϭ 13) (Fig. 2C) (Fig. 2, D and E), whereas other Sur1 neo/neo ␤ cells showed increased but non-oscillating levels of [Ca 2ϩ ] i (Fig. 2F). Upon stimulation with 17 mM glucose, the majority of the Sur1 neo/neo ␤ cells responded with a transient decrease in [Ca 2ϩ ] i , which later on was increased to a higher sustained level (Fig. 2, D-F).
The Machinery for Exocytosis in Sur1 neo/neo ␤ Cells Is Not Impaired-Despite of elevation of basal [Ca 2ϩ ] i in ␤ cells, we did not observe hyperinsulinemia in Sur1 neo/neo mice. Rather, the animals remained euglycemic for a large part of their life (Fig. 3A). Older animals showed mild hypoglycemia compared with the control animals; however, this did not appear to be the result of hyperinsulinemia, as seen in PHHI patients, because there was no difference in plasma insulin level between Sur1 w/w and Sur1 neo/neo mice (Fig. 3B). To test the possibility that the exocytotic machinery in Sur1 neo/neo ␤ cells is impaired by the chronically elevated [Ca 2ϩ ] i , we examined the sensitivity of the exocytotic machinery to Ca 2ϩ by measuring an increment of cell capacitance at different Ca 2ϩ concentrations. Either 0.9 or 10 M free Ca 2ϩ produced a more pronounced increase in cell capacitance in Sur1 neo/neo ␤ cells compared with Sur1 w/w ␤ cells, indicating a higher sensitivity of the exocytotic machinery to Ca 2ϩ in Sur1 neo/neo ␤ cells than Sur1 w/w ␤ cells (Fig. 4A). Approximating the Hill equation to the average rates of exocytosis in response to various concentrations of free Ca 2ϩ (Fig.  4B) yielded EC 50 values of 0.47 M for Sur1 w/w ␤ cells and 0.43 M for Sur1 neo/neo ␤ cells. The results demonstrate that sensitivity of the exocytotic machinery to Ca 2ϩ is not impaired in Sur1 neo/neo ␤ cells.
Glucose-stimulated Insulin Secretion in Sur1 neo/neo Mice Is Markedly Attenuated-We next examined glucose-stimulated insulin secretion in Sur1 neo/neo mice during a hyperglycemic clamp. The plasma glucose concentration was held at ϳ300 mg/dl in conscious animals for 2 h by infusion of glucose solution through an implanted cannula. In response to this prolonged hyperglycemic stimulus Sur1 w/w mice showed an increase in their plasma insulin concentrations of 5.8-fold (0.55 Ϯ 0.13 to 2.72 Ϯ 0.70 ng/ml, p ϭ 0.025). In contrast, the Sur1 neo/neo mice showed only a 1.5-fold increase in their plasma insulin concentrations (0.55 Ϯ 0.13 to 0.81 Ϯ 0.26 ng/ml, p ϭ 0.115) (Fig. 5). The glucose infusion rate necessary to maintain hyperglycemia, as well as the glucose turnover and glucose clearance rates in Sur1 neo/neo mice, were all lower than in Sur1 w/w mice (37, 58, and 58%, respectively; see Fig. 5 and Table I). The reduction of all of these parameters indicates FIG. 1. Sur1 gene structure and targeting strategy. A, Sur1 gene structure. B, Sur1 gene targeting strategy and structure of three different Sur1 gene alleles. The targeting vector contains two loxP sites surrounding the proximal promoter and exon 1. Sur1 W allele represents the wild type Sur1 gene. The Sur1 loxϩneo allele was created by homologous recombination in ES cells, which were then used to generate mice. The Sur1 neo allele, which is null, was created by cre-mediated recombination in single cell mouse embryos. B, BamHI; K, KpnI. C, Southern blot analysis using KpnI-digested DNA and the probe shown in B of tail-biopsy DNA from Sur1 w/w , Sur1 loxϩneo/w , and Sur1 loxϩneo/loxϩneo mice. D, PCR analysis showing conversion of the Sur1 loxϩneo allele to Sur1 neo allele by cre. In both cases these animals also contain a wild type allele. The corresponding regions of gene with PCR fragments amplified are shown in B.

FIG. 2. Sur1 neo/neo ␤ cells lack K ATP channels and exhibit elevated [Ca 2؉ ] i .
A and B, whole-cell K ϩ currents induced by depolarizing voltage pulses between Ϫ60 and 50 mV in 10-mV steps from a holding potential of Ϫ70 mV were measured twice, immediately and at 3 min after the whole-cell mode was established using a pipette containing an intracellular-like buffer without Mg-ATP. There was a significant increase in K ϩ currents in Sur1 w/w cells (n ϭ 7) after 3 min (A), whereas no effect was seen in diminished glucose tolerance in the Sur1 neo/neo mice because of impaired glucose-stimulated insulin secretion.
Secretogogue-specific Impairment of Insulin Secretion in Sur1 neo/neo Mice-The finding that mice without Sur1 are euglycemic for much of their life was in apparent conflict with their markedly impaired glucose-stimulated insulin secretion. Therefore, to test the possibility that non-glucose secretogogues were acting to stimulate insulin secretion in the absence of functional K ATP channels, we measured blood glucose and plasma insulin levels in both Sur1 w/w and Sur1 neo/neo mice during the postprandial period. After a 16-h fast, the mice were given free access to food and both blood glucose and insulin levels were determined. Both sets of mice were observed to begin refeeding within a few min and to eat frequently for at least 30 min. The blood glucose levels in both the Sur1 w/w and Sur1 neo/neo mice increased rapidly until about 60 min, then gradually decreased (Fig. 6A, upper graph). Although the difference in mean values between the two groups was not significant, 2 of 4 Sur1 neo/neo mice studied showed abnormally high levels of postprandial glucose concentration (Ͼ 250 mg/dl at 60 min), whereas none of the Sur1 w/w animals showed any values over 200 mg/dl through the experimental period (Fig. 6A, inset). Sur1 w/w mice exhibited a 9-fold increase in their plasma insulin   levels 30 min after initiation of refeeding and maintained plasma insulin concentrations that were within 80% of this peak level during rest of the experimental period (Fig. 6A, lower graph). Interestingly, although the plasma insulin level did not rise as rapidly during the first 30 min in Sur1 neo/neo mice, similar or higher levels were observed thereafter until the end of the 150-min test period. The oral intake of glucose is known to stimulate the secretion of incretin hormones such as glucagon-like peptide 1 (GLP-1), which augment the effect of glucose on the islet (26,27). Thus, to determine whether ingested glucose might have contributed to the increase of insulin secretion after refeeding of Sur1 neo/neo mice, we performed a gastric glucose tolerance test. After a 16-h overnight fast, glucose (2 g/kg body weight) was administered via gavage. This caused the blood glucose concentrations to increase rapidly in both the Sur1 w/w and Sur1 neo/neo mice during the first 20-min period (Fig. 6B, upper graph). In the Sur1 w/w mice the blood glucose concentration reached a peak of 374 Ϯ 10 mg/dl at 20 min before declining, whereas in the Sur1 neo/neo mice the blood glucose peaked at 433 Ϯ 26 mg/dl at 30 min. However, whereas the plasma insulin concentration increased rapidly in the Sur1 w/w mice, it did not change in Sur1 neo/neo mice (Fig. 6B, lower graph). The peak plasma insulin levels achieved after meal ingestion in the Sur1 w/w mice was over 3-fold of that observed in response to the glucose gavage. For the Sur1 neo/neo mice this difference was even more striking, with the peak plasma insulin level being ϳ13-fold higher for the meal compared with the glucose gavage.
The lack of any significant increase in plasma insulin con-centrations in Sur1 neo/neo mice after gastric glucose gavage led us to test whether incretins were being secreted, whether they were able to cause a normal rise in cAMP levels, and whether they were able to augment insulin secretion in Sur1 neo/neo mice. To explore these possibilities, we first measured the plasma concentration of GLP-1-(7-36) amide during refeeding after an overnight fast but found no differences between the Sur1 w/w and Sur1 neo/neo animals (6.6 Ϯ 2.2 pM versus 7.5 Ϯ 1.5 pM, respectively). The lack of any difference suggests that GLP-1 is secreted normally in the Sur1 neo/neo mice. We next examined the effect of GLP-1-(7-36) on insulin secretion in these animals. In Sur1 w/w mice a transient but drastic increase in plasma insulin level was observed 5 min after intravenous administration of GLP-1-(7-36) (10 nmol/kg body weight) with glucose (1g/kg body weight), whereas only a moderate increase was observed with glucose alone (Fig. 6C). In marked contrast, GLP-1-(7-36) failed to stimulate any rise in the plasma insulin concentration in Sur1 neo/neo mice at 5 min after administration. Only a slight increase in the insulin concentration, which resulted in a modest improvement of glucose tolerance, was observed 20 min after administration of GLP-1-(7-36) in these mice (Fig. 6C). Changes in blood glucose concentration during this test showed that glucose tolerance was significantly improved in Sur1 w/w mice but only slightly in Sur1 neo/neo mice by GLP-1-(7-36) administration (Fig. 6C). The inability of GLP-1 to stimulate insulin secretion in Sur1 neo/neo mice does not reflect a lack of cAMP production. As shown in Table II, we measured cAMP levels in isolated islets and observed Sur1 neo/neo islets produced a similar amount of cAMP to FIG. 6. Blood glucose and plasma insulin concentrations after food intake, gastric glucose loading, or GLP-1 administration in Sur1 neo/neo mice. A, blood glucose (upper) and plasma insulin (lower) levels after food ingestion. Mice were fasted for 16 h and then given free access to standard rodent chow. The data represent the mean Ϯ S.E. of 4 male mice. *, p Ͻ 0.05 compared with Sur1 w/w mice. The inset shows the individual blood glucose concentrations in mice examined at 60 min. B, blood glucose (upper) and plasma insulin (lower) levels after gastric glucose loading on 16-h fasted mice. The data represent the mean Ϯ S.E. of 6 male mice. *, p Ͻ 0.05; **, p Ͻ 0.005 compared with Sur1 w/w mice. C, blood glucose and plasma insulin levels were determined after intravenous injection of GLP-1-(7-36). Mice were fasted for 6 h, and then 10% glucose solution (1g/kg body weight) was injected with or without GLP-1-(7-36) (10 nmol/kg body weight). The data represent the mean Ϯ S.E. of 5 male mice. *, p Ͻ 0.05; **, p Ͻ 0.005; ***, p Ͻ 0.001 compared with GLP-1 (Ϫ). Plasma insulin concentrations in these studies were determined by ELISA.
Carbachol Stimulates Insulin Secretion in Perfused Sur1 neo/neo Pancreata-To understand the mechanisms by which Sur neo/neo mice are able to secrete nearly normal amount of insulin in response to feeding, we tested the effect of carbachol, a cholinergic agonist, on insulin secretion from perfused pancreas. Cholinergic stimulation is known to potentiate sustained insulin secretion from normal islets in glucose-dependent manner (28). When the glucose concentration was shifted from 3 to 11 mM, a clear biphasic insulin secretion was observed in Sur1 w/w pancreata, whereas only a very small amount of insulin was secreted in Sur1 neo/neo pancreata (Fig. 7). A 20-min administration of 50 M carbachol in combination with 3 mM glucose induced a small peak of insulin secretion during the first 3 min, similarly in both Sur1 w/w and Sur1 neo/neo pancreata. After this transient secretion, Sur1 w/w pancreata exhibited no additional secretion during the rest of period. In marked contrast, insulin secretion was sustained in the Sur1 neo/neo pancreata at ϳ50% of the initial peak (Fig. 7). In the presence of 11 mM glucose, carbachol potentiated both phases of insulin secretion in the Sur1 w/w pancreata. When compared with insulin secretion stimulated by 11 mM glucose alone, carbachol caused 3.5-fold increase in the first phase of secretion and 13.7-fold increase in the second phase. Insulin secretion was also observed in the Sur1 neo/neo pancreata; however, the total amount of insulin secreted during a 20-min period was about half that of the Sur1 w/w pancreata. Moreover, the insulin secretion profile of the Sur1 neo/neo pancreata was monophasic, compared with the biphasic response of the wild-type pancreata (Fig. 7). Insulin secretion caused by carbachol in Sur1 neo/neo pancreata was 3.4-fold higher in the presence of 11 mM glucose compared with in the presence of 3 mM glucose.

DISCUSSION
The K ATP channel has been thought to play a key role in insulin secretion by the ␤ cell. However, these studies demonstrate mice that lack Sur1, a key component of K ATP channel, secrete nearly a normal amount of insulin in response to a meal despite constant membrane depolarization in ␤ cell and mark-edly impaired glucose-stimulated insulin secretion. Analysis of the postprandial insulin secretion profile suggests that a greater than normal second phase of insulin secretion may compensate for the lack of a first phase of secretion in Sur1 neo/neo mice. Thus, although there are changes in the kinetics of insulin secretion after feeding, the ability of ␤ cells in Sur1 null mice to regulate the secretion of insulin remains remarkably intact. Our finding that carbachol was able to stimulate insulin secretion in the perfused Sur1 neo/neo pancreas points to cholinergic stimulation of the ␤ cell as a likely compensatory mechanism that may circumvent the markedly impaired response to glucose brought on by the lack of functional K ATP channels.
Our initial expectation was that Sur1 knock-out mice would provide a model for PHHI in humans. However, although Sur1 neo/neo ␤ cells mimic the electrical properties described for ␤ cells from PHHI-affected individuals (29,30), the absence of functional K ATP channels on insulin secretion and glucose homeostasis in mice has far less impact than in humans. Insulin secretion has long been known to comprise both a first and second phase. The first phase is a transient burst of insulin secretion from granules that exist in a readily releasable form, whereas the second phase is thought to reflect secretion from a pool of granules that must first be mobilized and docked to the plasma membrane before they can be secreted (31,32). A recent study using evanescent wave microscopy has provided additional evidence for this model (33). A first phase of insulin secretion was observed from a pool of previously docked granules, whereas the second phase of insulin secretion occurred largely from granules that were newly recruited to plasmalemmal docking sites. Indeed, although a single ␤ cell may contain more than 10,000 secretory granules (34), only 40 or so appear to exist in a readily releasable form (31). Thus, the process of granule mobilization and docking to the plasma membrane may play an important role in the regulation of insulin secretion.
Although human and rat ␤ cells respond to glucose with a progressively increasing second phase insulin release, mouse ␤ cells respond with a smaller and flatter second phase of insulin release (35). This difference has been suggested to be caused, at least in part, by the failure of high glucose to activate phospholipase C/protein kinase C signal pathway (36), and by lower production of cAMP by glucose stimulation in mouse ␤ cells (35). Thus, species-specific differences may explain why humans with defective K ATP channel function develop hypoglycemia from increased and unregulated insulin secretion, whereas mice lacking Sur1 appear to have compensatory mechanisms that prevent the hypersecretion of insulin.
Our data indicate that Sur1 neo/neo ␤ cells have elevated basal FIG. 7. Insulin secretion in perfused Sur1 neo/neo pancreata in response to carbachol. Pancreas was perfused with Krebs-Ringer bicarbonated buffer containing glucose and carbachol at concentrations indicated. The buffer and pancreas were maintained at 37°C through the experimental period. Data for males and females were not significantly different and were combined. The data represent the mean Ϯ S.E. of 6 Sur1 w/w and 5 Sur1 neo/neo mice. Insulin secretion in Sur1 neo/neo pancreata is significantly different from that in Sur1 w/w pancreata at   (29). Moreover, the sensitivity of the exocytotic machinery to Ca 2ϩ is not impaired in ␤ cells of the Sur1 neo/neo mice, but actually increased. This finding suggests that Sur1 null mice ␤ cells, like human ␤ cells with PHHI, probably secrete insulin from the granules regardless of glucose level once these granules have become a readily releasable form. The augmented second phase of insulin secretion during postprandial period in the Sur1 neo/neo mice would seem to indicate that ␤ cells in the Sur1 null mice have a sufficient reserve pool of insulin granules to meet normal metabolic demands, although we observed ϳ60% reduction in actual insulin content in Sur1 neo/neo pancreas compared with that in the Sur1 w/w pancreas (data not shown). Thus, we suggest that the lack of refilling of the readily releasable pool with granules from the reserve pool at the basal glucose level may help prevent the unregulated secretion of these insulin granules. Cholinergic muscarinic agonists, including the endogenous neurotransmitter acetylcholine and the synthetic non-hydrolyzable analogue carbachol, are known to enhance glucosestimulated insulin secretion (28). Acetylcholine is released by intrapancreatic vagal nerve endings and stimulates insulin secretion in ␤ cells mainly by activating phospholipase C/protein kinase C signal pathways (37). Vagal stimulus of the endocrine pancreas is thought to persist during the preabsorptive and absorptive phases of feeding, although there is no direct evidence because of quick degradation of acetylcholine (37). For this reason, we studied the effect of carbachol on the insulin secretion using perfused pancreas and found that carbachol stimulates insulin secretion in Sur1 neo/neo pancreas in a glucose-dependent manner. This finding suggests that the cholinergic stimulation may be one of the mechanisms whereby nearly normal insulin secretion in response to feeding is maintained in the Sur1 neo/neo mice.
Under a basal glucose concentration, Sur1 w/w pancreas responds to carbachol with only a small first phase of insulin secretion. In contrast, Sur1 neo/neo pancreas shows a sustained second phase of insulin secretion following the first phase secretion. This result suggests that carbachol stimulates granule mobilization from a reserved pool to releasable pool under a basal glucose concentration. Although it is possible that the sensitivity to cholinergic stimulation is increased in Sur1 neo/neo pancreas, it is more likely that carbachol promotes the granule processing similarly in both Sur1 w/w and Sur1 neo/neo pancreas but that insulin secretion dose not occur in Sur1 w/w pancreas because of opening of K ATP channels at this glucose concentration. This is supported by the fact that the insulinotropic effect of acetylcholine under a non-stimulating glucose concentration is unmasked by sulfonylurea treatment of mouse islets (37). Thus, we speculate that the sustained insulin secretion observed from the Sur1 neo/neo pancreas in the presence of carbachol and 3 mM glucose represents the same, unregulated mode of insulin secretion that is observed in humans with PHHI.
In this study, we also examined the effect of GLP-1 on insulin secretion in Sur1 neo/neo mice. Binding of GLP-1 to its G proteincoupled receptor leads to the activation of adenylate cyclase and generation of cAMP. In ␤ cells, cAMP, via activation of protein kinase A (PKA), affects ion channel activity, [Ca 2ϩ ] i handling, and the mobilization of granules to potentiate glucose-stimulated insulin secretion (38,39). Both subunits of K ATP channel are also target of PKA, and the function of this channel is thought to be modulated by phosphorylation by PKA (40). Furthermore, the recent study showed that the cAMPbinding protein cAMP-GEFII, by interacting with Rim2, a target of small GTP-binding protein Rab3, mediates cAMP-dependent PKA-independent exocytosis (41), and that this PKA-independent pathway is critical in the potentiation of insulin secretion by incretins (42). Thus, the late step of granule processing, i.e. downstream of granule mobilization, is a major site for the action of incretins. A bolus injection of GLP-1 together with glucose caused a strong potentiation of the first phase of insulin secretion in the control mice, but no change in the first phase and only a slight increase in the second phase of insulin secretion in Sur1 neo/neo mice. There was no defect in cAMP generation by GLP-1 in Sur1 neo/neo islets. These results provide additional evidence for the lack of a readily releasable pool of insulin granules in Sur1 neo/neo mice during the fasted condition.
Depolarization of the ␤ cell plasma membrane has long been thought to be a key step in the regulation of insulin secretion. Indeed, prevention of ␤ cell depolarization by directing the expression of a constitutively active form of Kir6.2 to pancreatic ␤ cells in transgenic mice markedly impairs insulin secretion, thereby causing severe diabetes and death of the animals within 5 days of birth (43). However, our studies, as well as those of others, clearly illustrate that mice lacking K ATP channels, as achieved either via Sur1 or Kir6.2 gene knock-outs, continue to regulate their secretion of insulin despite constant depolarization of the plasma membrane of ␤ cells in these mice. Thus, although membrane depolarization is clearly necessary for insulin secretion, the mechanisms for regulation of insulin secretion in the intact animal cannot be explained simply by the closure of K ATP channels. The precise nature of all the K ATP channel-independent mechanisms involved in regulation of insulin secretion remain to be determined.