Hyperinsulinism of Infancy: The Regulated Release of Insulin by K_ATP Channel—Independent Pathways Free

Studies were performed on three Caucasian patients with HI. N15 and N38 were siblings (family 1); N16 was unrelated (family 2). No other family members were affected. Each of the patients presented with typical symptoms of HI: severe hypoglycemia (blood glucose <2.6 mmol/l; normal range 3.5-5.5 mmol/l) in association with hyperinsulinism (insulin >12 mU/ml in association with a raised C-peptide at the time of hypoglycemia, normal insulin value <4 mU/ml) within the first few minutes/hours after birth. These and other established criteria(1) confirmed the diagnosis of HI.

Patient N15 was the first child of unrelated Caucasian parents and was born at 40 weeks' gestation after a normal pregnancy (birth weight 3.5 kg, 50th percentile). When the child was 3 weeks old, the parents noticed twitching of limbs and eyes in the morning, and at 11 weeks the child was admitted to the hospital after three generalized seizures. HI was confirmed on admission,following a blood glucose measurement of 0.5 mmol/l in association with hyperinsulinism. Patient N38 was the second child of these unrelated Caucasian parents. Delivery was induced at 37 weeks' gestation on suspicion of rhesus incompatibility (birth weight 4.51 kg, >99th percentile). The child was noted to be hypoglycemic (blood glucose 1.9 mmol/l) on the first day of life. Figure 1 details clinical data for these patients; similar findings were also observed for N16. Note that blood glucose levels could be raised only by increases in the rate of glucose administration (normal range <6-8 mg · kg^-1 ·min^-1) and that treatment with medical agents either was ineffective or provided only temporary relief from the need for high rates of glucose infusion. Because all patients therefore failed to respond adequately to medical therapy involving nifedipine, chlorothiazide, diazoxide,somatostatin (Octreotide), or glucagon combined with an elevated glucose infusion rate to maintain normoglycemia (>12 mg · kg body wt^-1 · min^-1), a near total (95%) pancreatectomy was performed to prevent persistent hypoglycemia (see arrows in Fig. 1). Patient N15 subsequently underwent a second resection of the pancreas because of continuing hyperinsulinemic hypoglycemia. Histopathological examination of each of the pancreata revealed that all three patients had diffuse abnormalities of endocrine tissue. Postoperatively, all three patients achieved normoglycemia at normal glucose administration rates. N15 developed clinical diabetes during the postoperative period and has continued to require regular insulin therapy. Genotyping studies using standard procedures(3,5,7)revealed that all three patients were heterozygous for a single base change within intron 16 of the SUR1 gene, 2154+3 A to G (GenBank accession number L78207), that appears to result in aberrant splicing. In family 1, both patients had the mutation on the paternal allele, whereas in family 2 the mutation was carried on the maternal allele. The 2154+3 A-to-G SUR1gene mutation was not identified in 50 control subjects or in any other HI patients tested (i.e., it is not a common polymorphism). No defects in the Kir6.2 gene were detected.

Preparation of tissue for in vitro studies. Intact islets of Langerhans were isolated from transplantable adult donor tissue (with permission) and from the pancreata of all three HI subjects using a controlled collagenase digestion procedure, as previously described(9). Islets were maintained under standard tissue culture conditions using RPMI-1640 medium (Sigma, Poole,U.K.) supplemented with 10% (vol/vol) fetal calf serum, 100 U/ml penicillin,and 100 μg/ml streptomycin in a humidified atmosphere of 5% CO₂in air at 37°C. For electrophysiology studies, intact islets were mechanically dispersed into individual cells and isolated cell clusters before experimentation. Insulin release and Ca²⁺ microfluorimetry experiments were carried out using intact islets.

Insulin release measurements from intact islets. Insulin release was measured under static incubation conditions using batches of five islets and triplicate determinations. The islets were preincubated at 37°C in Krebs-Ringer HEPES buffer (composition in millimoles per liter: 137 NaCl, 5.36 KCl, 0.81 MgSO₄, 0.34 Na₂HPO₄, 0.44 KH₂PO₄, 4.17 NaHCO₃, 10 HEPES, and 1.26 CaCl₂) containing 2 or 2.8 mmol/l glucose at pH 7.4 for 30 min, as described previously(22,23). Subsequently, the islets were exposed to the test substances for 30 min. At the end of the incubation period, aliquots of the samples were removed and kept at -20°C until radioimmunoassay was performed using a charcoal separation method (26). For insulin release studies in Ca²⁺-free medium, Ca²⁺ was omitted, and the Krebs-Ringer HEPES buffer was supplemented with 1 mmol/l EGTA.

Electrophysiology. All data were obtained from primary cultured human tissue, as previously described(9). β-cells were selected on the basis of their larger size and granular appearance, and recordings were made using inside-out, cell-attached, perforated whole-cell, and the standard whole-cell configurations of the patch-clamp technique(27). Recording pipettes were fabricated from boro-silicate glass and had a series resistance of 2-5 MΩ when filled with the internal solution for whole-cell experiments and 5-10 MΩ for single-channel current recordings. The zero potential was adjusted with the pipette in the bath, and no correction for liquid junction potentials was made for single-channel current recordings. Data were recorded using an EPC-7 patch-clamp amplifier (List Electronic, Darmstadt, Germany)with hardware (TL-1 A/D-D/A Interface) and software (pClamp 5.5) supplied by Axon Instruments (Foster City, CA). For the capacitance measurements, data were recorded using an AxoPatch 200B amplifier with a Digidata 1200B A/D-D/A interface (Axon Instruments). Changes in membrane capacitance were tracked using in-house software run on a Pentium computer (Dell, Round Rock, TX). Whole-cell currents were filtered at 0.5 kHz and digitized at 1.6 kHz for measurements of whole-cell voltage-activated K⁺ currents and stored on a 386 personal computer (Tandon, Moorpark, CA). Single K⁺channel and whole-cell K_ATP channel data was stored on digital tape using a DAT recorder (Biologic, Echirolles, France) for subsequent replay and analysis. All ion channel experiments were carried out at room temperature. Changes in K_ATP channel activity have been expressed as either changes in open-state probability (Po) or as a function of N.Po, where N is the number of operational channels(28).

Secretion studies from single cells. Capacitance measurements were used to monitor exocytosis in isolated cells, as previously described(29). All experiments were performed using either the perforated patch or the standard whole-cell configuration of the patch-clamp technique. With a temporal resolution of 100 ms, recordings of changes in the cell membrane capacitance reflect increases or decreases in the area of the cell membrane during exocytosis or endocytosis of secretory granules, respectively. For perforated whole-cell experiments,the patch-clamp recording pipette solution contained (in millimoles per liter)76 Cs₂SO₄, 10 KCl, 10 NaCl, 5 HEPES, 1 MgCl₂,and 0.24 mg/ml amphotericin, pH 7.35 with CsOH; for standard whole-cell experiments, the patch-clamp pipette was filled with (in mmol/l) 125 potassium glutamate, 10 KCl, 1 MgCl₂, 3 Mg-ATP, 0.1 cAMP, 5 HEPES, 10 EGTA,pH 7.15 with KOH. Where necessary, CaCl₂ was added at 5 mmol/l to achieve final concentrations of 0.17 μmol/l(30). All cells were bathed in a solution containing (in mmol/l) 118 NaCl, 20 TEACl, 5.6 KCl, 2.6 CaCl₂, 1.2 MgCl₂, 5 HEPES, and 5 glucose, pH 7.4 with NaOH. Where indicated, 2 μmol/l forskolin and 10 nmol/l phorbol myristate acetate (PMA) were added to the bathing solution. All capacitance experiments were performed at 32-34°C.

K_ATP channels. Ionic currents were recorded in the cell-attached, whole-cell, and inside-out patch configurations of the patch-clamp techniques (27). For all experiments undertaken using quasiphysiological cation gradients, the standard extracellular 140 mmol/l NaCl-rich bathing solution contained (in mmol/l) 140 NaCl, 4.7 KCl, 2.5 CaCl₂, 1.13 MgCl₂, 10 HEPES, 2.5 glucose (pH 7.4 with NaOH). For whole-cell experiments, the patch-clamp pipette contained (in mmol/l) 140 KCl, 10 EDTA, and 10 HEPES. For inside-out patch recordings, the internal face of the membrane was bathed with a KCl-rich solution containing (in mmol/l) 140 KCl, 10 NaCl, 1.13 MgCl₂, 1 EGTA, 2.5 glucose, 10 HEPES (pH 7.2 with KOH), and the patch-clamp recording pipette was filled with the 140 mmol/l NaCl-rich solution. For cell-attached patch recordings, the patch-clamp pipette contained either the 140 mmol/l NaCl-rich solution or the 140 mmol/l KCl-rich solution used for inside-out patch recordings, as indicated in the text.

Macroscopic K_ATP channel currents recorded using the whole-cell configuration were performed by holding the cells at -70 mV and then applying 20 mV depolarizing and hyperpolarizing pulses at a frequency of 0.5 Hz. This procedure reveals K_ATP channel currents (I_KATP) without the activation of voltage-dependent K⁺ channels. Because the vast majority of K_ATP channels in β-cells are tonically inhibited by endogenous ATP levels, dialysis of the cell interior with a solution containing a low concentration of ATP results in the apparent“activation” of channels and the generation of a significant“washout” current, as previously described(31). For single-channel current events, the procedures used for data collection and analysis have been described previously(9,28).

Voltage-gated K⁺ channels. Ionic currents were recorded using the standard whole-cell configuration. Patch-clamp pipettes were filled with a 140 mmol/l KCl-rich solution containing (in mmol/l) 125 KCl, 30 KOH, 10 EGTA, 1 MgCl₂, 2 CaCl₂, 5 HEPES, and 3 MgATP (pH 7.15 with KOH), and cells were bathed with the 140 mmol/l Na⁺-rich solution. Macroscopic voltage-activated whole-cell K⁺ currents were recorded under voltage-clamp conditions with the cell held at -70 mV and stimulated by 20 consecutive depolarizing pulses,which were increased stepwise by 10 mV from -60 to 130 mV for a duration of 500 ms and applied to the cell with a frequency of 0.5 Hz. This protocol activates outward K⁺ currents from both the delayed rectifier(K_V) and the Ca²⁺- and voltage-gated K⁺channel (K_Ca) observed as a macroscopic whole-cell K⁺current (I_K).

All illustrated records are displayed according to the convention with upward deflections representing outward currents.

Tolbutamide, diazoxide, forskolin, ATP, and ADP were obtained from Sigma,and PMA was obtained from CalBiochem. All agents were added at concentrations as stated in the text. Tolbutamide, diazoxide, and PMA were added as concentrated stock solutions dissolved in DMSO (final concentration<0.1%).

Ca²⁺ microfluorimetry. Changes in[Ca²⁺]_i were estimated in HI islets using microfluorimetry procedures with fura-2(9,28). Intact islets were cultured on poly-D-lysine—coated cover slips (50μg/ml) for up to 58 h in RPMI medium supplemented with 10% fetal bovine serum, 100 U/ml penicillin, and 100 μg/ml streptomycin, as above. They were loaded with 0.5 μmol/l fura-2-AM (Sigma) in 1 ml RPMI-1640 medium for 7-10 min at 37°C. The composition of the basic perifusion medium was (in mmol/l) 137 NaCl, 5.36 KCl, 0.81 MgSO₄, 0.34 Na₂HPO₄, 0.44 KH₂PO₄, 4.17 NaHCO₃, 10 HEPES, and 1.26 CaCl₂. This medium was gassed with air for 20 min; glucose was added to a final concentration of 2 mmol/l,and pH was adjusted to 7.4 with NaOH. Where the perifusion medium contained 40 mmol/l KCl, the concentration of NaCl was adjusted accordingly. One series of experiments was carried out using the HI-derived β-cell line NES2Y,maintained under the same conditions as previously described(28). For all experiments, an in vitro calibration procedure(32) was used to estimate changes in [Ca²⁺]_i. Basal values of[Ca²⁺]_i have been established in the presence of the basic perifusion medium. Changes in cytosolic Ca²⁺ are reported as the difference between basal [Ca²⁺]_i and the peak rise in [Ca²⁺]_i obtained for the agent tested.

Statistic analysis. Statistical comparisons were made using Student's t tests, with values of P < 0.05 considered to be significant.

In vivo data and genotype studies. In this study, three Caucasian patients with HI were investigated. Because two of the three patients were siblings (family 1), this represents the first report of familial HI among nonconsanguineous Caucasians identified in the U.K. All three patients were unresponsive to medical therapy with diazoxide, chlorothiazide, somatostatin,nifedipine, or glucagon (Fig. 1). Because these observations implicated defects in the K_ATP channel genes as the cause of HI, genotyping studies were used to screen for defects in SUR1 and Kir6.2. Each patient was found to carry a novel gene defect in intron 16 of SUR1 (2154+3 A to G) and was compound heterozygous for this mutation. Analysis of parental DNA documented a paternal defect at the same locus in family 1 and a maternal defect in family 2. No defects in Kir6.2 were found in any of the patients. The defective region of SUR1 is thought to encode part of the first nucleotide-binding domain of the sulfonylurea receptor(2). Electrophysiological studies of β-cells isolated from patient tissue confirmed that this mutation is associated with defects in K_ATP channels.

Loss of operational K_ATP channels in HI β-cells.Previous studies of HI β-cells have described complete loss of functional K_ATP channels(6,7,8,9,10). In insulin-secreting cells from HI-SUR1 patients, we now report that the pathophysiology is associated with a marked decrease in the number of functional channels, coupled with a loss of ADP-dependent regulation.

Figure 2 shows whole-cell current data from control and HI-SUR1 β-cells obtained under the same experimental conditions. In these experiments, β-cells were dialyzed using a low ATP—containing pipette solution and whole-cell K_ATP channel currents generated by repetitively depolarizing and hyperpolarizing the cell (see RESEARCH DESIGN AND METHODS). Because in normal cells K_ATP channels are inhibited by intracellular ATP(15), cell dialysis normally leads to a marked washout K_ATP channel current (n = 8; Fig. 2). By contrast, only a modest increase in the macroscopic K_ATP channel current was found in HI-SUR1 β-cells (n = 17; Fig. 2), suggesting that although K_ATP channels were present, the magnitude of the current was significantly reduced.

This was further investigated using single-current recordings made from inside-out patches. In control cells, K_ATP channels were found to be consistently open in the absence of intracellular ATP(15), and the average N.Po value was 24.9 ± 4.2 (n = 25). By contrast, K_ATPchannels were recorded in only 76% of HI-SUR1 β-cells (n = 29 of 38) and the average N.Po value was 1.2 ± 0.6 (n = 8)(Fig. 3A). These data are consistent with macroscopic K_ATP channel current recordings(Fig. 2) and therefore define the loss of functional K_ATP channels in these β-cells. In addition, we found that K_ATP channels in HI-SUR1 β-cells were unaffected by ADP (n = 26 of 29, 500 μmol/l) or diazoxide(n = 20 of 23, 200-500 μmol/l) in the presence of ATP (500μmol/l) (Fig. 3B-D)and that there were no effects of an “activation cocktail”consisting of potassium fluoride (10 mmol/l), diazoxide (200 μmol/l), ADP(100 μmol/l), UDP (100 μmol/l), and GDP (500 μmol/l) (n = 7)(Fig. 3D). In controlβ-cells, these procedures consistently led to a marked and sustained increase in K_ATP channel activity(28).

Selective loss of K_ATP channels in HI-SUR1 β-cells.Because different K⁺ channels contribute to the ionic control of insulin release (15), we also examined whether K_ATP channel defects altered the activity of voltage-gated K⁺ channels in HI-SUR1 β-cells. Whole-cell K⁺ channel currents were therefore recorded following a depolarization of the membrane from -60 to 130 mV in 10-mV steps. This led to the activation of both Ca²⁺- and voltage-gated K⁺channels and delayed-rectifier K⁺ channels. Figure 4 shows that in control(n = 25) and HI-SUR1 (n = 20) β-cells, the magnitudes of voltage dependent K⁺ channel currents were similar at all membrane potentials and that there were no differences in the voltage dependencies of channel activation.

K_ATP channel defects and cytosolic Ca²⁺signaling. Cell-attached patch-clamp recordings and measurements of cytosolic Ca²⁺ concentrations were used to examine the effects of K_ATP channel defects on the regulation of the cell membrane potential. Under basal conditions, control β-cells were electrically silent because of the presence of open K_ATP channels(12,28). In marked contrast, 66% of intact HI-SUR1 β-cell recordings were spontaneously generating action currents in the absence of glucose stimulation(n = 14 of 21; Fig. 5A). Ca²⁺ action currents were largely unaffected by somatostatin (n = 8 of 10, 100 nmol/l) or diazoxide(n = 17 of 20, 500 μmol/l)(Fig. 5A), and diazoxide had no effect on K_ATP channels in intact-cell recordings(n = 19 of 19; Fig. 5B).

Despite the appearance of spontaneous action potentials in HI-SUR1β-cells, we found the average basal cytosolic Ca²⁺concentration was similar to that in control cells: 80 ± 5 nmol/l(n = 68) vs. 78 ± 3 nmol/l (n = 141), respectively. However, as a result of the depolarized membrane potential, HI-SUR1 islets responded poorly to depolarization-dependent agonists. Thus, glucose (20 mmol/l, n = 16 of 16) and the antidiabetic sulfonylurea tolbutamide(100 μmol/l, n = 14 of 16) were unable to further elevate[Ca²⁺]_i, whereas KCl (40 mmol/l) caused only a modest increase in [Ca²⁺]_i in 7 out of 16 experiments(Δ[Ca²⁺]_i = 30 ± 5 nmol/l [n = 7]compared with 158 ± 18 nmol/l [n = 14 of 14] in control cells[9]; Fig. 6). By contrast, agents that do not elevate [Ca²⁺]_i through a change in the membrane potential, such as ATP (100 μmol/l), UTP (100 μmol/l), and acetylcholine (100 μmol/l), readily raised [Ca²⁺]_i in HI-SUR1 islets (Δ[Ca²⁺]_i = 93 ± 8 nmol/l, n = 37 of 38; Fig. 6).

Ca²⁺-dependent exocytosis in HI-SUR1 β-cells.Because K_ATP channel defects in HI-SUR1 β-cells were causally related to unregulated Ca²⁺ channel activity, we next examined the role of Ca²⁺ in determining the regulation of exocytosis using capacitance measurements. Figure 7 summarizes experiments whereby either control or HI-SUR1β-cells were dialyzed using the whole-cell configuration with either 170 nmol/l or no Ca²⁺ in the pipette solution. In both control and HI-SUR1 β-cells, we were unable to detect increases in membrane capacitance when [Ca²⁺]_i was absent (n = 12),whereas there was a marked increase in exocytosis when cells were dialyzed with [Ca²⁺]_i at 170 nmol/l Ca²⁺ (n =10) (Fig. 7). Overall, there were no significant differences between control and HI-SUR1 β-cells under these conditions.

Because these data indicate that exocytosis in HI-SUR1 β-cells is governed by increases in intracellular Ca²⁺ levels, we next examined the relationship between voltage-dependent Ca²⁺ influx and control of exocytosis (Fig. 8). Experiments were carried out using the perforated patch technique under voltage clamp conditions, and measurements of whole-cell Ca²⁺currents and changes in cell capacitance were recorded in real time. Because voltage-gated Ca²⁺ channels were spontaneously active in intact HI-SUR1 β-cells (Fig. 5),their activity was suppressed by voltage clamping the cells at -70 mV. A stepped depolarization was then made from -70 to 0 mV to generate voltage-gated Ca²⁺ influx [Fig. 8A(i)], which was associated with a transient increase in the rate of exocytosis (n = 6)[Fig. 8A(ii)]. Note that under these conditions, exocytosis was then followed by a decrease in cell capacitance, which was consistent with the stimulation of endocytosis(33). Ca²⁺-dependent exocytosis was also investigated after activation of protein kinase A (PKA) by 2 μmol/l forskolin and protein kinase C (PKC)using 10 nmol/l PMA. Figure 8shows that both forskolin and PMA enhanced exocytosis in a cumulative manner and that the overall rates of endocytosis under these conditions were markedly decreased (n = 6)(Fig.8). Neither forskolin nor PMA had any significant effect on the magnitude of voltage-dependent Ca²⁺ currents [Fig. 8A(i)]. Similar findings have also been reported in rodent β-cells (34).

K_ATP channel-independent pathway of regulated insulin release in HI-SUR1 β-cells. We have previously documented the activity of both the K_ATP channel—dependent and the K_ATPchannel—independent pathways of glucose-induced insulin release in normal human β-cells(22,23). Because HI-SUR1 β-cells lack K_ATP channels yet retain Ca²⁺-dependent exocytotic events, we used intact isolated islets to investigate the role of glucose in governing insulin secretion independently of K_ATP channel function.

Under basal conditions (2 mmol/l glucose) the rates of insulin secretion from HI-SUR1 islets were, on average, 10-fold lower than corresponding values from human control islets: 0.23 ± 0.06 ng/5 islets/30 min (n =5) vs. 2.3 ± 0.4 ng/5 islets/30 min (n = 12), respectively(22). Furthermore, since HI-SUR1 islets were also found to have a far lower content of insulin than control islets (1.1 ± 0.3 ng/islet, n = 5, patients N15, N16[1.07 ± 0.19 ng/islet, n = 18 from 10 patients] vs. 44± 8 ng/islet, n = 12)(22), these data suggest that there is a rapid rate of insulin turnover in HI islets, which is consistent with the clinical presentation of HI.

In HI-SUR1 islets, the depolarization-dependent agonists KCl (40 mmol/l)and tolbutamide (100 μmol/l) failed to significantly elevate[Ca²⁺]_i (Fig. 6), and as a consequence, neither agent had any significant effect on insulin release (Fig. 9A). This is in contrast to the effects of glucose, which failed to raise intracellular Ca²⁺ levels(Fig. 6), yet caused a dose-dependent release of insulin (Fig. 9A). The actions of glucose were unaffected by the presence of 150 μmol/l diazoxide (Fig. 9A) and were inhibited in the absence of extracellular Ca²⁺ (Fig. 9B). Finally, exposure of HI-SUR1 β-cells to 100μmol/l acetylcholine, ATP, and UTP caused a >2.5-fold increase in insulin secretion. These experiments illustrate the activity of the K_ATP channel—independent pathway in HI islets (i.e.,augmentation of Ca²⁺-stimulated insulin release as a result of the absence of functional K_ATP channels in the β-cells).

The next series of studies examined the ability of glucose to augment insulin release in the HI-SUR1 islets in the absence of extracellular Ca²⁺, as is the case in normal human islets(22). Augmentation under these conditions is achieved in the presence of activated PKA and PKC(19,20,21)(or with the pharmacological agent mastoparan[22]). Figure 9B (left side)shows that in the presence of extracellular Ca²⁺,glucose-stimulated insulin release was increased to fivefold that of the basal rate by the addition of forskolin and PMA. (Increased Ca²⁺-stimulated exocytosis in response to forskolin and PMA was also shown by capacitance measurement in Fig. 8.) Fig. 9B (right side)shows the results obtained in the absence of extracellular Ca²⁺. Under these conditions, 11.1 mmol/l glucose failed to stimulate insulin secretion, as did the combination of forskolin, PMA, and 2 mmol/l glucose. However, in the presence of forskolin and PMA, 11.1 mmol/l glucose caused a marked stimulation of insulin release. In normal β-cells, this response to glucose is not associated with any rise in the intracellular concentration of Ca²⁺ (21), and in this study we found similar responses using the HI-derived β-cell line NES2Y(28,35)transfected with Kir6.2CΔ26, which produces Kir6.2 channel activity independently of SUR1 function(36)(Table 1).

HI is a heterogeneous entity. In addition to clinical and genetic diversity, recent pathological findings have also revealed two histopathologically distinct forms of HI. Most cases of HI arise from the diffuse involvement of defective β-cells throughout the pancreas (Di-HI);however, up to 30% of patients are now thought to have focal adenomatous hyperplasia, or focal disease (Fo-HI)(2,37,38). Fo-HI is caused by loss of maternally expressed genes as the result of somatic events. This leads to 1) insulin hypersecretion through inheritance of paternally derived SUR1 gene defects and 2) β-cell proliferation as a result of the loss of the growth-suppressing genes p57^KIP2 and H19 and expression of the growth-stimulating gene product IGF-2(38).

In this study, all three patients were found to have Di-HI, and we have provided a detailed characterization of the manner in which unregulated insulin release is coupled to a genetic lesion in SUR1. This mutation was identified in two affected families and was present in both parental and sibling DNA. All affected offspring were profoundly hypoglycemic after birth and had typical symptoms of severe Di-HI, including impaired drug responses to somatostatin, diazoxide, and nifedipine therapy(Fig. 1). As is typical in these cases, all patients underwent a 95% pancreatectomy in the first instance, which alleviates the symptoms of hyperinsulinism but will predispose the affected individuals to long-term sequelae. These may include pancreatic exocrine insufficiency and insulin-dependent diabetes(1).

In response to glucose stimulation, the regulated release of insulin is controlled by depolarization-response coupling and glucose augmentation pathways(12,13,14,15,16,17,18,19,20,21,22,23). In the former pathway, K_ATP channels play a key role in governing changes in the cell membrane potential and the regulated entry of Ca²⁺, whereas in the latter pathways, glucose metabolism governs insulin secretion independently of K_ATP channel operation and will augment responses to the raised [Ca²⁺]_i(16,17,18,21,22,23). In normal rodent and human islets, this has been demonstrated by experimentally eliminating the operation of K_ATP channels using high concentrations of either tolbutamide (which clamp K_ATPchannels closed) or diazoxide (which hyperpolarize the cell membrane by activating K_ATP channels) and then depolarizing the cell with high external KCl to provide voltage-dependent Ca²⁺ influx(16,17,18,22,23). Under these conditions, glucose has been shown to enhance insulin secretion in a dose-dependent manner by mechanisms that are both dependent and independent of Ca²⁺ influx(21). More recently, the K⁺ channel—independent pathways of glucose augmentation have also been described in transgenic mice without K_ATP channels(39,40,41).

Because HI is a K_ATP channelopathy with loss of function mutations in SUR1 and Kir6.2(9,11),we have used β-cells isolated from patients as an experimental system to further characterize glucose augmentation pathways in human insulin-secreting cells. The operation of K_ATP channel—independent pathways in HI β-cells is also clinically important, since high rates of glucose infusion are administered in vivo to maintain normoglycemia in patients(Fig. 1)(1). Here, we have documented that loss of K_ATP channel operation in HI-SUR1 β-cells (Figs. 2 and 3) is causally related to uncontrolled Ca²⁺ channel activity (Figs. 5 and 6), which is directly coupled to Ca²⁺-dependent insulin secretion (Figs. 7,8,9). Because the resting cell membrane potential in HI-SUR1 β-cells is severely compromised (9),depolarization-dependent agonists such as glucose, high external KCl, and tolbutamide failed to cause a marked increase in [Ca²⁺]_i(Fig. 6). Loss of depolarization-dependent Ca²⁺ signaling readily explains why neither KCl nor tolbutamide promoted insulin release(Fig. 9A). In addition, it has been shown that tolbutamide fails to stimulate insulin secretion in vivo in Di-HI patients (see accompanying article by Grimberg et al. [25]). In contrast,whereas glucose similarly failed to raise cytosolic Ca²⁺ levels in HI-SUR1 islets, insulin secretion was stimulated, and this was found to be both concentration dependent and governed by Ca²⁺ influx(Fig. 9). Because K_ATP channels are nonfunctional in HI-SUR1 β-cells, these data document for the first time that glucose-induced insulin release is regulated through the K_ATP channel—independent pathway in HIβ-cells. Interestingly, in these cells, glucose also augmented PKC- and PKA-dependent insulin release in the absence of extracellular Ca²⁺(Fig. 9B). Therefore,the HI-SUR1 islets exhibit both the Ca²⁺-dependent and Ca²⁺-independent glucose augmentation pathways, as do normal humanβ-cells(22,23).

In this article, we have documented for the first time the mechanisms that govern insulin release from HI β-cells. These data show that loss of K_ATP channels in β-cells leads to uncontrolled Ca²⁺channel activity. As in normal β-cells(42), this will tend to increase in [Ca²⁺]_i within the vicinity of the plasma membrane and the microdomains of the insulin-containing granules. Measurements of capacitance have documented that exocytosis is dependent on a rise in[Ca²⁺]_i and that voltage-gated Ca²⁺ influx is associated with insulin release. These data are relevant to our understanding of the control of insulin release under normal and pathological conditions,but they are also relevant to the clinical management of HI. Current medical therapy for HI is unsatisfactory because it involves regimens with agents that fail to adequately control hyperinsulinism due to the loss of K_ATPchannels (Fig. 1). It is also evident from this study that use of high rates of glucose infusion to maintain normoglycemia might stimulate insulin secretion through the K_ATPchannel—independent pathway. Unfortunately, for many patients with HI the only effective manner in which to control hypoglycemia is a near-total resection of the pancreas. Recently, it was demonstrated that pancreatic venous sampling of insulin could be used to locate focal regions ofβ-cell mass in HI-SUR1 patients and that selective surgical resection of these areas alone results in the successful management of hyperinsulinism(38). Because as many as 30%of HI patients coming to pancreatectomy have focal disease, early diagnosis of these individuals, combined with pancreatic venous sampling, would obviate the necessity for a 95% pancreatectomy in favor of a 20-30% surgical resection(38). The major clinical benefit of this is that it would alleviate long-term complications of HI for a significant number of patients. Currently, preoperative diagnosis of Fo-HI is not possible on the basis of genotyping alone; some 60% of all patients presently have no identified gene defects in SUR1 or Kir6.2(2), and even when defects have been found, it is clear that a number of different mutations can give rise to similar clinical phenotypes(1,2). Thus, based on an advanced understanding of the molecular mechanisms of insulin release from normal β-cells and the data presented in this report, it has been suggested that responses to intravenous tolbutamide in the form of a tolbutamide-stimulation test (TST) could be used to distinguish Fo-HI from Di-HI forms (2). In support of this, Grimberg et al.(25) have shown that by measuring acute insulin responses to glucose and tolbutamide, patients with Di-HI can be distinguished from control subjects by the fact that they do not respond to TST, whereas glucose causes a blunted insulin release profile. These findings are largely explained by characterization of the K_ATP channel—independent pathway of regulated insulin release in Di-HI β-cells. Loss of K_ATP channels prevented tolbutamide-induced rises in [Ca²⁺]_i and insulin release(Figs. 6 and 9A), whereas glucose promoted insulin secretion in a Ca²⁺-dependent manner(Fig. 9). Since Fo-HI patients show positive responses to TST (S. Thornton, unpublished observations) because of the presence of K_ATP channels in β-cells from the nonfocal regions (M.J.D., J.M. Saudubray, C. Junien, unpublished data), these data provide strong support for the use of intravenous tolbutamide as a diagnostic procedure for distinguishing the different entities of HI.

R.O.B. is funded by a MRC-Collaborative Award with Novo-Nordisk (M.J.D.),K.E.C. by the Medical Research Council (M.J.D.), and R.M.S. by the British Diabetic Association. G.W.G.S. was supported by a grant from the National Institutes of Health (RO1-DK42063) and by a Fogarty Senior International Fellowship (FO6 TW02105). B.G. is supported in part by grant no. 4201 from the Israel Ministry of Health and a grant from the Israel Science Foundation funded by the Academy of Sciences and Humanities. Collaborative initiatives were supported by a European Union—funded Concerted Action Grant(BMH4-CT98-3284) from the European Network for Research into Hyperinsulinism in Infancy (ENRHI).

We thank Drs. Roger James and Sue Swift at the University of Leicester Department of Surgery and Kate Hinchliffe at Sheffield University for their help in the preparation of cells.