Alterations in calcium channel currents underlie defective insulin secretion in a transgenic mouse.

A transgenic mouse overexpressing a mutant form of calmodulin (CaM-8) that is selectively targeted to pancreatic beta-cells has an impaired ability to secrete insulin in response to elevated blood glucose. Fluorescence measurements of cytosolic Ca2+ concentration ([Ca2+]i) showed that intracellular Ca2+ rises produced by glucose were smaller than normal in beta-cells of CaM-8 mice. Glucose utilization rates were not different between the CaM-8 and control beta-cells, suggesting that glucose metabolism was unperturbed by CaM-8. Ion channel defects were implicated in the phenotype of CaM-8 beta-cells because treatment of these cells with tolbutamide, a blocker of ATP-sensitive K+ channels, produced smaller than normal amounts of insulin secretion and Ca2+ rises. Depolarization with elevated extracellular K+ also produced smaller Ca2+ rises in beta-cells from CaM-8 mice. Whole-cell patch-clamp recordings revealed that Ca2+ channel currents of beta-cells from CaM-8 mice were half as large as Ca2+ currents in control cells, while the currents carried by delayed rectifier and ATP-sensitive K+ channels were similar in magnitude in both cell types. We conclude that expression of the CaM-8 form of calmodulin causes a down-regulation of Ca2+ channel currents, which reduces Ca2+ entry and accumulation when glucose stimulates closure of the ATP-sensitive K+ channels. The reduction in intracellular Ca2+ accumulation then prevents an adequate amount of insulin from being secreted from beta-cells of CaM-8 mice.

A transgenic mouse overexpressing a mutant form of calmodulin (CaM-8) that is selectively targeted to pancreatic beta-cells has an impaired ability to secrete insulin in response to elevated blood glucose. Fluorescence measurements of cytosolic Ca 2؉ concentration ([Ca 2؉ ] i ) showed that intracellular Ca 2؉ rises produced by glucose were smaller than normal in beta-cells of CaM-8 mice. Glucose utilization rates were not different between the CaM-8 and control beta-cells, suggesting that glucose metabolism was unperturbed by CaM-8. Ion channel defects were implicated in the phenotype of CaM-8 beta-cells because treatment of these cells with tolbutamide, a blocker of ATP-sensitive K ؉ channels, produced smaller than normal amounts of insulin secretion and Ca 2؉ rises. Depolarization with elevated extracellular K ؉ also produced smaller Ca 2؉ rises in betacells from CaM-8 mice. Whole-cell patch-clamp recordings revealed that Ca 2؉ channel currents of betacells from CaM-8 mice were half as large as Ca 2؉ currents in control cells, while the currents carried by delayed rectifier and ATP-sensitive K ؉ channels were similar in magnitude in both cell types. We conclude that expression of the CaM-8 form of calmodulin causes a down-regulation of Ca 2؉ channel currents, which reduces Ca 2؉ entry and accumulation when glucose stimulates closure of the ATP-sensitive K ؉ channels. The reduction in intracellular Ca 2؉ accumulation then prevents an adequate amount of insulin from being secreted from beta-cells of CaM-8 mice.
The exocytotic secretion of insulin from pancreatic beta-cells requires the transduction of numerous internal and external signals, beginning with the sensing of elevated glucose levels (1,2). The transduction of the glucose signal through metabolic pathways results in the generation of ATP which closes ATPdependent K ϩ channels in the plasma membrane (3,4). This depolarizes the membrane and opens voltage-gated Ca 2ϩ channels, elevating the intracellular Ca 2ϩ concentration ([Ca 2ϩ ] i ) 1 and triggering exocytotic insulin secretion (5)(6)(7). The molecular basis of the action of Ca 2ϩ in insulin secretion is not clear, although it has been proposed that the intracellular Ca 2ϩ -receptor protein, calmodulin (CaM), may play a key role in this process (8 -10).
As part of an effort to elucidate the functional role of calmodulin in pancreatic beta-cells, transgenic mouse lines have been established that specifically overexpress CaM in their betacells (11). These mice have defective insulin secretion at an early age, due to an impairment in the metabolism of glucose and the subsequent generation of ATP (12,13). Another mouse line, referred to as CaM-8, expresses in its beta-cells a mutant form of CaM that lacks 8 amino acids in its central helix and does not activate calmodulin-dependent proteins in vitro even though this protein binds Ca 2ϩ with normal affinity (14). These mice also demonstrate a reduced insulin secretion in response to fuel secretagogues such as glucose (15), which is surprising because overexpression of this mutant form of CaM in cardiac atrial cells produces no phenotypic abnormalities (16).
Subtle differences in pancreatic beta-cell responses of 6 -8day-old postnatal CaM-8 and CaM mice suggest that the secretory defects may be different in the two lines of mice. For example, the onset of diabetes begins later in CaM-8 mice than in CaM mice and the kinetics of the glucose-induced secretion response is different in these two mice at this age (15). In addition, depolarization of the beta-cell membrane potential increases insulin secretion much more in CaM mice than in CaM-8 mice (reviewed in Ref. 15). Here we examine several key signaling events that underlie insulin secretion (1,2,17) in pancreatic beta-cells of CaM-8 mice. We find that the primary defect in beta-cells from CaM-8 mice is a remarkable reduction in the amount of Ca 2ϩ flowing through voltage-gated Ca 2ϩ channels, which results in abnormally small rises in [Ca 2ϩ ] i during glucose stimulation and presumably leads to or directly accounts for the diabetic phenotype of these animals. These results confirm that the nature and origin of the insulin secretory defect exhibited in CaM-8 mice is quite different from that of CaM mice (13) and provide a useful experimental system for the study of intracellular Ca 2ϩ signaling and Ca 2ϩ channel regulation.

MATERIALS AND METHODS
Chemical Reagents-CsOH was from Mallinckrodt Chemical Co., West Germany. All other chemical reagents were from Sigma, unless otherwise indicated.
Isolation of Islets and Beta-cells-6 -8-day-old postnatal mice were obtained from active breeding colonies of transgenic CaM-8 and control mice. All animals were cared for in accordance with the rules and regulations set forth by the United States National Institutes of Health. Isolated islets were prepared from normal and CaM-8 mice by collagenase digestion and cultured in fully supplemented RPMI 1640 (Life Technologies, Inc.) at 37°C in an atmosphere of 95% O 2 , 5% CO 2 as described previously (12,15). Single beta-cells used for patch-clamp analysis were prepared from acutely isolated islets using EDTA and trypsin treatment as described by Ribar et al. (13). Isolated beta-cells were then suspended in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) containing 10% fetal bovine serum, 11.1 mM glucose, 100 units/ml penicillin, and 100 g/ml streptomycin, plated into 24-well plates (Becton Dickinson, Oxnard, CA) containing 12-mm round glass coverslips (Fischer), and incubated at 37°C in a humidified atmosphere of 95% O 2 , 5% CO 2 for 1-3 days until used for experiments.
Optical Measurements of [Ca 2ϩ ] i -The membrane-permeant fluorescent dyes, fura-2-AM and indo-1-AM (18), were used to measure [Ca 2ϩ ] i in islets and individual beta-cells, respectively. Dyes were loaded into either islets or cells as described previously (13). Prior to recording, islets or beta-cells were allowed to equilibrate in Krebs-Ringer bicarbonate (133 mM NaCl, 2.5 mM CaCl 2 , 5 mM KCl, 1.2 mM MgSO 4 , 1.2 mM KH 2 PO 4 , and 5 mM NaHCO 3 ) supplemented with 0.1 mg/ml bovine serum albumin, 2.8 mM glucose, and 15 mM HEPES (pH 7.4) for 45 min at 30 -37°C that was also equilibrated with 95% O 2 , 5% CO 2 , Fura-2 (Molecular Probes, Junction City, OR) fluorescence was measured using a Zeiss Standard microscope equipped with a Leitz 50ϫ water-immersion objective (1.0 numerical aperture) and the spinning wheel photomultiplier-based system described by Neher (19). Fura-2 was excited with light alternated between 360 and 390 nm, while a photomultiplier was used to measure dye emission at wavelengths greater than 510 nm.
[Ca 2ϩ ] i was calculated from the ratio of the light emitted when the dye was excited by the two excitation wavelengths as described previously (19,20), using software written in AxoBasic by Hui Qiu (Duke University). Indo-1 (Molecular Probes) fluorescence was measured on a Nikon Optiphot microscope equipped with an Odyssey real-time laser-scanning confocal microscope (Noran, Middleton, WI). This microscope was equipped with a water-cooled argon ion laser to excite the dye at 350 -360 nm; emitted fluorescence was measured at two emission wavelengths (400 -450 nm and 450 -500 nm), and fluorescence images were processed with Image-1 software (Universal Imaging, Philadelphia, PA).
Insulin Secretion Measurements-Insulin secretion from isolated islets was measured with the procedures described by Epstein et al. (12) and Ribar et al. (15). Briefly, 100 -200 islets that were cultured short term (24 h) were placed in sealed perifusion chambers (0.5-ml volume) on 0.65 mM Milipore DVPP membranes (Milipore Corp., Bedford, MA). Krebs Ringer bicarbonate (see [Ca 2ϩ ] i measurement description) containing the appropriate secretagogue was pumped through the chamber at a rate of 300 l/min at 37°C, collected, and frozen at Ϫ20°C until analyzed. Secreted insulin was measured in the eluate fractions using a double-antibody mouse insulin radioimmunoassay kit (Linco, St. Louis, MO) which uses rat insulin as a standard.
Glucose utilization assays were done as previously detailed (13,21)  Patch-clamp Recordings-Conventional whole-cell patch-clamp methods (22) were used to measure and control the membrane potential of single beta-cells, as well as to measure transmembrane currents and control the ionic composition of the cell interior. Pipettes were pulled from filamented thin-walled borosilicate glass (World Precision Instruments, Sarasota, FL), and their tips were fire-polished with a microforge. After filling with various internal solutions (see below), pipette resistances ranged from 3 to 5 megohms. The pipette was connected to a patch-clamp amplifier (EPC-9, HEKA Elektronik, Lambrecht, West Germany) and a tight seal (Ͼ1 gegaohm) was established between the pipette tip and the cell membrane by gentle suction. The whole-cell recording configuration was then obtained by applying a brief pulse of negative pressure to break the membrane patch spanning the pipette tip. All experiments were performed at a temperature of 30 -37°C.
Statistical Analysis-All values are reported as the mean Ϯ S.E. Statistical comparisons were done by utilizing the Student's t test. A significant difference was accepted when p Ͻ 0.05.

Reduced [Ca 2ϩ ] i Signaling in Beta-cells from CaM-8 Mice-
Given the essential role of intracellular calcium in glucoseinduced insulin secretion (23,24), we examined the ability of glucose to elevate cytosolic [Ca 2ϩ ] i in beta-cells of CaM-8 and normal mice. For these experiments, isolated pancreatic islets were loaded with the membrane-permeant, fluorescent indicator dye, fura-2-AM (18), to monitor [Ca 2ϩ ] i . Fig. 1 shows examples of experiments illustrating that perifusion of control and CaM-8 islets with Krebs-Ringer bicarbonate saline (KRB) containing a low concentration of glucose (2.8 mM) produced resting [Ca 2ϩ ] i levels that were similar for both types of islets. Resting [Ca 2ϩ ] i , measured in KRB containing 2.8 mM glucose, was approximately 100 nM for both control and CaM-8 islets ( Fig. 2A) and did not differ significantly between the two groups (p Ͼ 0.60). Elevation of the extracellular glucose concentration to 16.8 mM produced robust rises in [Ca 2ϩ ] i from control cells (Fig. 1, top), as reported in earlier work (24). In contrast, glucose-induced [Ca 2ϩ ] i rises were much smaller in islets isolated from CaM-8 mice (Fig. 1, bottom). While [Ca 2ϩ ] i in normal islets increased by approximately 200 nM above the resting level upon exposure to KRB containing 16.8 mM glucose, this rise in [Ca 2ϩ ] i was only about one-third as large in islets from CaM-8 mice (Fig. 2B). The difference between the mean rises in [Ca 2ϩ ] i induced by glucose in the two types of islets was significantly different (p Ͻ 0.05). Thus, the ability of glucose to elevate [Ca 2ϩ ] i is impaired in islets from CaM-8 mice.
Because it has been suggested that dissociated beta-cells can produce glucose-induced [Ca 2ϩ ] i responses which differ from those of intact islets (25), we also measured [Ca 2ϩ ] i in dissociated beta-cells loaded with indo-1-AM, another membranepermeant indicator dye. We found that glucose-induced [Ca 2ϩ ] i transients also were smaller in beta-cells from CaM-8 mice than those in beta-cells from normal mice (data not shown). Thus, in our case the [Ca 2ϩ ] i responses of isolated beta-cells paralleled those of intact islets, with both indicating that betacells from CaM-8 mice produced smaller glucose-induced rises in [Ca 2ϩ ] i than cells from control mice.
In summary, our measurements of [Ca 2ϩ ] i reveal significant deficiencies in the glucose-induced calcium signaling pathway of CaM-8 beta-cells. Because of the essential role of this pathway for glucose-dependent insulin secretion, it is likely that these deficiencies account for the defective insulin secretion characteristic of the CaM-8 mouse. We next determined the mechanisms underlying this lesion in the calcium signaling pathway.
Glucose Metabolism Is Normal in CaM-8 Mice-The glucoseinduced [Ca 2ϩ ] i signaling pathway begins with extracellular glucose being transported into ␤-cells and converted into ATP. To determine whether these early steps in the pathway were altered in CaM-8 mice, we measured the rate of glucose utilization of CaM-8 islets. Fig. 3 shows that under basal metabolic conditions, CaM-8 islets (n ϭ 23) utilized glucose at a rate that was not significantly different (p Ͼ 0.10) from that of control islets (n ϭ 32). Elevating the glucose concentration to 16.8 mM resulted in an approximately 4-fold increase in glucose utilization by both CaM-8 and control islets (p Ͼ 0.70). These measurements show that glucose metabolism is normal in CaM-8 mice and indicate that the defect in glucose-induced Ca 2ϩ signaling must be downstream of this early step in the signaling cascade. Further, because glucose utilization can be used to predict whether sufficient ATP will be produced to close ATPsensitive K ϩ channels (21), our results suggest that these channels should experience normal ATP levels in CaM-8 cells.
Ion Channel Abnormalities in CaM-8 Beta-cells-The re-duced glucose-induced [Ca 2ϩ ] i signaling of CaM-8 beta-cells could result from a defect in any (or all) of the ion channels that regulate the membrane potential of beta-cells. We next performed experiments designed to identify the ion channel abnormalities that underlie this defect in the glucose-induced calcium signaling pathway. Glucose metabolism in beta-cells induces insulin secretion by closing ATP-sensitive K ϩ channels in the plasma membrane (4,26). As an initial test of the function of these channels and their downstream effectors, we examined the sensitivity of CaM-8 islets to tolbutamide, a sulfonylurea drug that blocks ATPsensitive K ϩ channels and induces insulin secretion by depolarizating the beta-cell membrane potential (4,27). Fig. 4 shows that, under basal conditions (2.8 mM glucose), CaM-8 islets and control islets secreted similar amounts of insulin. Addition of tolbutamide (1 mM) induced a 4-fold increase in insulin secretion in the presence of 2.8 mM glucose in control islets. However, while exposure of CaM-8 islets to this concentration of tolbutamide produced an initial burst of insulin secretion, subsequent secretion was only about twice the basal rate (p Ͻ 0.05). Furthermore, whereas addition of a high concentration (16.8 mM) of glucose to the tolbutamide-containing saline caused control islets to produce an additional 3-fold increase in insulin secretion, CaM-8 islets (Fig. 4) failed to respond further (p Ͻ 0.001). Thus, tolbutamide was less capable of evoking insulin secretion from islets from CaM-8 mice.
The differential effects of tolbutamide on insulin secretion in control and CaM-8 islets were paralleled by the actions of this drug upon [Ca 2ϩ ] i . For both control and CaM-8 islets, treatment with 1 mM tolbutamide (in the presence of 2.8 mM glucose) resulted in sustained and reversible increases in [Ca 2ϩ ] i , as measured with fura-2 (Fig. 5A). However, the mean rise in [Ca 2ϩ ] i induced by tolbutamide was approximately half as large in CaM-8 islets as in islets from normal mice (Fig. 5B); this difference was statistically significant (p Ͻ 0.05). Taken together, these results indicate that the secretory defect in CaM-8 cells could reside at the level of the ATP-sensitive K ϩ channels or at some subsequent step in the pathway that normally leads to a rise in [Ca 2ϩ ] i and resultant secretion of insulin.
We next used whole-cell patch-clamp methods (22) to examine directly the properties of ATP-sensitive K ϩ channels in CaM-8 beta-cells. Ionic currents flowing through ATP-sensitive K ϩ channels were measured using the paradigm illustrated in Fig. 6A. These currents were activated by progressively lowering the cytosolic concentration of ATP by dialyzing the cell interior with a patch pipette solution containing a low concentration (0.3 mM) of ATP (5,13). The amplitude of these currents reached a plateau and then slowly declined after about 10 min (not shown), presumably because the ATP-sensitive K ϩ channels exhibit "wash-out" during prolonged intracellular dialysis (5). These dialysis-induced currents were unequivocally identified as ATP-sensitive K ϩ currents because they were rapidly blocked by tolbutamide (0.1 mM; Fig. 6A). While the ATPsensitive K ϩ channels were opening during removal of intracellular ATP, the beta-cell membrane potential was alternated between Ϫ60, Ϫ70, and Ϫ80 mV to study the voltage-dependence of the induced currents. These currents were inward at Ϫ80 mV and outward at Ϫ60 mV; since the equilibrium potential for K ϩ is approximately Ϫ76 mV under our experimental conditions, this behavior is as expected for a current carried by K ϩ ions. The slope of the relationship between the magnitude of the ATP-sensitive K ϩ currents and the membrane potential gives the ATP-sensitive K ϩ conductance; this value was normalized by dividing by the membrane capacitance measured for each cell, to take into account variations in cell membrane area (22). When measured in this way, as shown in Fig. 6B, the average ATP-sensitive K ϩ conductance of CaM-8 beta-cells (n ϭ 19) was found to be significantly larger (p Ͻ 0.05) than that of control beta-cells (n ϭ 27). This indicates that beta-cells of CaM-8 mice have a relatively high density of ATP-sensitive K ϩ channels that are able to function normally. Thus, the defects in glucose-induced and tolbutamide-induced insulin secretion and [Ca 2ϩ ] i signaling must have a source other than loss of ATP-sensitive K ϩ channels.
The delayed rectifier K ϩ channel is another type of ion channel that regulates the Ca 2ϩ entry that triggers insulin secretion (5,17). Currents flowing through this type of K ϩ channel were measured in isolation by including the Ca 2ϩ buffer, EGTA, in the pipette solution to prevent the rise in [Ca 2ϩ ] i that would normally activate Ca 2ϩ -dependent K ϩ currents (28), as well as ATP to close the ATP-sensitive K ϩ channels. Fig. 7A illustrates the ionic currents produced under these conditions when beta-cells were held at a membrane potential of Ϫ70 mV and briefly depolarized to a variety of more positive potentials.
The resultant currents were outward in polarity, activated rapidly and did not inactivate during 200 ms long depolarizations. Further, they were eliminated by extracellular application of tetraethylammonium ions (data not shown), a blocker of delayed rectifier channels in these cells (28). These properties indicate that the currents measured were K ϩ currents flowing through delayed rectifier channels.
The time course and magnitude of delayed rectifier currents were very similar in beta-cells from control and CaM-8 mice. Membrane current density was quantified for each individual cell by dividing current amplitude by the membrane capacitance. The voltage-dependence of K ϩ current density was very similar for the two types of beta-cells (Fig. 7B). In both types of cells, delayed rectifier K ϩ currents activated at membrane potentials of Ϫ60 mV and increased in magnitude at more positive potentials. The peak delayed rectifier K ϩ conductance, measured at ϩ30 mV, was 0.92 Ϯ 0.14 nanosiemens/picofarad in CaM-8 cells (n ϭ 12) and 1.23 Ϯ 0.25 nanosiemens/picofarad The reduced glucose-induced [Ca 2ϩ ] i signaling of CaM-8 beta-cells could result from a defect in voltage-gated Ca 2ϩ channels. We initially tested this possibility by using fura-2 to measure the rises in [Ca 2ϩ ] i produced when these channels were opened by depolarizing the beta-cell membrane potential by treatment with saline containing 50 mM K ϩ . As has been reported previously (7,13), this treatment produced a prompt elevation of [Ca 2ϩ ] i in islets from normal mice (right side of Fig.  1A). These responses were substantially reduced in islets from CaM-8 mice (Fig. 1B). The peak amplitude of the [Ca 2ϩ ]change produced by depolarization of control islets was approx-imately 400 nM above resting [Ca 2ϩ ] i but was approximately half this value in CaM-8 mice (Fig. 2C). Although this difference was not statistically significant (p Ͼ 0.30), these data suggest possible changes in the Ca 2ϩ channels of CaM-8 betacells. Experiments on dissociated beta-cells loaded with indo-1 also indicated similar reductions in the K ϩ induced [Ca 2ϩ ] i rise in CaM-8 cells (data not shown).
To examine Ca 2ϩ channel properties more directly, we used patch-clamp methods to measure currents flowing through these channels. Ca 2ϩ channel currents were examined in isolation by including cesium ions in the intracellular pipette solution to block K ϩ channels (29). Families of Ca 2ϩ currents recorded from control and CaM-8 cells under such conditions are shown in Fig. 8A. In both cases, the membrane potential initially was held at Ϫ70 mV and then briefly stepped to more depolarized voltages to open voltage-gated Ca 2ϩ channels. While the kinetics of these currents were similar in the two populations of cells, Ca 2ϩ current amplitude appeared to be smaller in the CaM-8 cells. This was quantified by calculating Ca 2ϩ current density, using the procedure already described for K ϩ currents, and determining this parameter over a range of membrane potentials (Fig. 8B). There were no obvious differences in the voltage dependence of Ca 2ϩ currents in betacells from the two types of mice; in both cases, Ca 2ϩ currents were activated at Ϫ50 mV, increased in magnitude as voltage was increased up to 0 mV, and decreased at more positive potentials. For both cell types, the currents reversed polarity at ϩ50 mV, though efflux of Cs ϩ makes this value an underestimate of the true reversal potential for Ca 2ϩ currents (30). However, the peak magnitude of Ca 2ϩ currents recorded from CaM-8 cells was about 50% smaller than that of control cells at all membrane potentials (Fig. 8B). For example, at 0 mV the mean Ca 2ϩ current density was 8.9 Ϯ 0.9 pA/picofarad (n ϭ 34) in CaM-8 cells, slightly less than half that measured in control cells (18.1 Ϯ 1.3 pA/picofarad; n ϭ 23). The difference between these two means is significant (p Ͻ 0.05), indicating much smaller amounts of Ca 2ϩ current in CaM-8 cells. This reduction in Ca 2ϩ current is qualitatively sufficient to account for the defects in glucose-induced Ca 2ϩ signaling and insulin secretion that lead to the diabetic phenotype of CaM-8 mice. DISCUSSION In this study we have identified the physiological defect that leads to the nonimmune diabetic condition of the CaM-8 mouse.
We have found that the glucose-regulated Ca 2ϩ signaling pathway is deficient in pancreatic beta-cells of this animal. Specif-ically, depolarization of CaM-8 beta-cells, in response to extracellular glucose, tolbutamide, or potassium, resulted in smaller rises in [Ca 2ϩ ] i than in beta-cells from normal mice. These same treatments also consistently produced less insulin secretion from CaM-8 beta-cells than from normal beta-cells. Given that insulin secretion requires elevation of [Ca 2ϩ ] i (23,31,32), it is likely that this defect in Ca 2ϩ signaling accounts for the mild hyperglycemic phenotype demonstrated by the CaM-8 mice at the age we have examined (6 -8 days old). Prolonged elevation of blood glucose levels in these young mice acts as a catalyst to further desensitize the beta-cells to external stimuli, so that the diabetic condition will become more severe as the animals get older (33).
The underlying cause of the Ca 2ϩ signaling defect of CaM-8 beta-cells appears to be a defect in the voltage-gated Ca 2ϩ channels of their plasma membrane. Patch-clamp measurements revealed a significant decline in the peak amplitude of voltage-gated Ca 2ϩ channel currents while the voltage dependence and kinetics of these currents were similar to normal. Though multiple types of Ca 2ϩ channels have been found in beta-cells and may contribute to [Ca 2ϩ ] i elevation and insulin secretion (34,35), the Ca 2ϩ channels affected by the CaM-8 mutation are the L-type Ca 2ϩ channel currents because dihydropyridines blocked these currents. 2  The current amplitude is divided by capacitance to normalize differences in cell size (*p Ͻ 0.05). and reduced insulin secretion when CaM-8 beta-cells are depolarized. The fact that insulin secretion evoked by tolbutamide (Fig. 4) or elevated external K ϩ (15) (both treatments that trigger insulin secretion by opening voltage-gated Ca 2ϩ channels) was also reduced in CaM-8 mice provides independent evidence for this interpretation.
Currents flowing through the other types of ion channels that regulate insulin secretion were largely normal in CaM-8 cells. Delayed rectifier K ϩ channels, which aid in repolarization of the membrane potential during each action potential (5,17), were unaffected by expression of the CaM-8 protein. However, ATP-sensitive K ϩ channels, which initiate the electrical component of the secretory response by depolarizing the betacell (4,26), carried larger than normal currents in the CaM-8 cells. This increase in ATP-sensitive K ϩ current might somehow contribute to the CaM-8 phenotype by altering the betacell membrane potential. While the cause of this increase in ATP-sensitive K ϩ channel current is not known, it may represent some sort of compensatory mechanism. For example, Yan et al. (36) have demonstrated that changes in the level of expression of one ␤-cell protein can result in a compensatory change in other associated proteins. An up-regulation of the number of ATP-sensitive K ϩ channels would make the resting potential more negative and the glucose-induced depolarization larger, which might compensate for the CaM-8 phenotype by allowing glucose to open a larger fraction of available voltage-gated Ca 2ϩ channels. Currents carried by the Ca 2ϩ -activated K ϩ channels, which hyperpolarize the beta-cell membrane potential between bursts of action potentials (17), were not examined in our experiments.
The decreased magnitude of voltage-gated Ca 2ϩ channel currents in CaM-8 cells could arise from a number of causes. For example, a reduction in Ca 2ϩ channel currents could result from a decrease in the rate of Ca 2ϩ channel synthesis and/or membrane insertion, an increased rate of Ca 2ϩ channel degradation, post-translational changes in the channel, or changes in the regulation of the channel. We do not yet know which of these processes are involved in the CaM-8 mice.
The CaM-8 gene was originally designed to produce a protein that would serve as a control for overexpression of CaM (12,15). The rationale was that the CaM-8 protein should mimic the ability of CaM to bind Ca 2ϩ , yet not activate effector proteins such as protein kinases and phosphatases (14 -16). However, our data indicate that the CaM-8 protein must possess some intrinsic biological activity because its prescence in betacells results in decreased Ca 2ϩ channel currents. Overexpression of another E-F hand Ca 2ϩ binding protein, calbindin-D 28K , in GH 3 pituitary cells also results in reduced Ca 2ϩ channel currents through an unknown mechanism (37). Like CaM-8, calbindin-D 28K has not been shown to bind to other proteins, but has been suggested to regulate [Ca 2ϩ ] i of the cells (37). Thus, it is possible that the Ca 2ϩ binding properties of both calbindin-D 28K and CaM-8 are sufficient in themselves to reduce Ca 2ϩ channel currents through an unknown pathway. If so, this must not be an acute effect of Ca 2ϩ binding because dialysis of the Ca 2ϩ chelator EGTA during patch-clamp experiments does not inhibit Ca 2ϩ currents in normal cells.
It is possible that truncating the central helix of CaM may yield an unusual conformation that allows the CaM-8 protein to bind to or otherwise block the voltage-gated Ca 2ϩ channels. In Paramecium, changes in single amino acids in the C-terminal half of CaM prevents regulation of Ca 2ϩ -activated K ϩ channels (38). Additionally, changes in single amino acids in the N-terminal half of CaM lead to down-regulation of Ca 2ϩ -activated Na ϩ channels (38). Such studies suggest that CaM may modulate ion channels through a direct binding mechanism (39). If this were true for the voltage-dependent Ca 2ϩ channels of beta-cells, then CaM-8 may interfere with the binding of Ca 2ϩ /CaM to these channels. However, currently there is no evidence that either CaM-8 or CaM participate in modulating this channel through a direct binding mechanism (10,40).
Although there is some controversy (41), Ca 2ϩ /calmodulindependent kinase, type II has been reported to modulate the activity of beta-cell Ca 2ϩ channels through phosphorylation (40). While the CaM-8 protein conceivably could compete with native CaM for binding to this kinase in the CaM-8 cells, the fact that CaM-8 protein is about 100 times less potent in binding to CaM-binding proteins (15) makes this possibility unlikely. Alternatively, it is conceivable that CaM-8 could bind to other proteins involved in regulation of the voltage-gated Ca 2ϩ channels.
There are also similarities between the reduction of the Ca 2ϩ channel currents in the CaM-8 beta-cells and the reductions in Ca 2ϩ currents seen in beta-cells depleted of protein kinase C (42). In both instances Ca 2ϩ currents are substantially reduced in the beta-cells while the voltage dependence and kinetics of Ca 2ϩ channel gating do not appear to be altered (Fig. 8) (42). In normal beta-cells activation of protein kinase C results from a Ca 2ϩ dependent hydrolysis of phosphoinosotides by a phosphoinositide-specific phospholipase C (43). This hydrolysis generates inositol 1,4,5-trisphosphate, which rapidly releases internal Ca 2ϩ stores, and diacylglycerol which activates protein kinase C (43). It has been suggested that one of the consequences of activating protein kinase C in beta-cells is to shift the voltage dependence of the Ca 2ϩ channel so that it activates at more negative potentials, which could enhance/and or sustain long term glucose induced insulin secretion (42,43). It is possible that the action of CaM-8 protein is inhibiting some component of this protein kinase C-mediated pathway. Consistent with the possibility that the reduction in Ca 2ϩ currents seen in CaM-8 mice may be mediated via the protein kinase C pathway comes from the fact that stimulating islets with a muscarinic agonist (carbachol) that activates protein kinase C can restore the secretory response in islets from CaM-8 mice (15).
One of the original aims in developing mouse lines which overexpress CaM and CaM-8 was to explore the role of CaM in cell growth and differentiated cellular function (10,44,45) in the pancreatic beta-cell. Initial analysis of these transgenic mice suggested that CaM might be part of the glucose-regulated signaling pathway that triggers the exocytotic secretion of insulin (11,12,15). However, detailed analysis of the lesions produced by these genetic manipulations indicates that this approach is not as straightforward as intended. The results presented here show that the presence of CaM-8 yields a lesion in Ca 2ϩ channel currents which not only is upstream of the reactions directly responsible for insulin secretion but also is completely unanticipated based on the known properties of the CaM-8 protein and the voltage-gated Ca 2ϩ channel. Likewise, overexpression of CaM yields unpredicted upstream effects on ATP utilization in the beta-cell (13). Taken together, these results indicate that considerable caution must be exercised in interpreting the experimental consequences of long term manipulation of gene expression and cast some uncertainty on the utility of this approach for sorting out the molecular constituents of the secretory pathway (46). However, given that the nature of the signaling defect that now has been identified in the CaM-8 mouse, this mouse line could be useful for studying Ca 2ϩ signaling pathways and the long term regulation of Ca 2ϩ channels. Further, having diabetic mice with defined physiological lesions should be valuable for understanding the mechanisms involved in diabetes.