Influence of Membrane Potential Changes on Cytoplasmic Ca2 Concentration in an Electrically Excitable Cell, the Insulin-secreting Pancreatic B-cell*

Transcription

1 THE JOURNAL OF BIOLOGICAL CHEMISTRY by The American Society for Biochemistry and Molecular Biology, Inc. VOl. 267, No 29, Issue of October 15. PP ,1992 Printed in U. S. A. Influence of Membrane Potential Changes on Cytoplasmic Ca2 Concentration in an Electrically Excitable Cell, the Insulin-secreting Pancreatic B-cell* (Received for publication, February 14, 1992) Patrick Gilontj and Jean-Claude Henquintll 11 From the $ Unit4 de Diabetologie et Nutrition, University of Louuain Faculty of Medicine, UCL 54.74, avenue Hippocrate 54, B-1200 Brussels, Belgium and the ltphysio1ogische.s Znstitut, University of Saarland, HomburglSaar, Germany Glucose stimulation of insulin release involves me- voltage-dependent Ca2+ channels by repetitive depolarizations tabolism of the sugar and elevation of cytoplasmic of the membrane. However, this distinction is not always very calcium (Ca2+i) in pancreatic B-cells. We compared the clear cut (3, 4). dynamic changes of metabolism (fluorescence of endog- Pancreatic B-cells are electrically excitable (5). On stimuenous reduced pyridine nucleotides, NAD(P)H), mem- lation by glucose, the physiologically most important regulator brane potential (intracellular microelectrodes), and of insulin release, B-cells depolarize and exhibit Ca2+-depend- Ca2+i (fura-2 technique), in intact mouse islets. Glucose ent electrical activity (6). A similar response can also be (16 mm) sequentially triggered an increase triggered by a number of other physiological or pharmacolog- NAD(P)H fluorescence, a depolarization with electri- ical agents. An important characteristic of this electrical cal activity, and a rise in Ca2+i. The change NAD(P)H in activity is its rhythmicity. In the presence of mm was monophasic and regular, whereas the changes in glucose, it consists of regular slow waves of the membrane membrane potential and Ca2+i were multiphasic, with potential (2-2.5/min) with bursts of Ca2+ spikes superimposed steady-state regular oscillations of similar average freon the plateau potential (6, 7). Pancreatic B-cells are thus a quencies (about 2.2/min). Digital image analysis revealed that Ca2+i oscillations were synchronous in all good model to test whether Ca2+i changes are essentially regions of the islets. Omission of extracellular Ca2+ driven by membrane potential changes in excitable cells. abolished the rise in Caz+i but not the increase in Recent studies using microspectrofluorimetric techniques NAD(P)H. Both electrical and Ca2+i oscillations disap- have indeed identified oscillations of Ca2+, in single rat or peared in low external Ca2+ (1 mm), and became larger mouse B-cells. However, these oscillations had a considerably but slower in high Ca2+ (10 mm). Sustained depolari- lower frequency ( /min) than the slow waves of memzation (by tolbutamide, arginine, or high K+) and hy- brane potential (8-10). It was also noted that the Ca2+L reperpolarization (by diazoxide) of B-cells caused sus- sponse to glucose was highly variable from B-cell to B-cell tained increases and decreases of Ca2+i, respectively. (11). On the other hand, a good correlation was found between In conclusion, the changes in membrane potential in- the oscillations of Ca, and B-cell membrane potential measduced by various secretagogues trigger synchronous ured simultaneously in intact mouse islets perifused with a changes in Ca2+i in all B-cells of the islets. The oscil- medium containing glucose alone or with tetraethylammo- latory pattern of the electrical and Ca2+i responses induced by glucose is not accompanied by and thus probably not due to similar oscillations of metabolism. Calcium plays an essential role in stimulus-secretion coupling (1). Stimulation of electrically nonexcitable cells often induces oscillations of free cytoplasmic Ca2+ concentration (Ca2+;)l by modulating release and uptake of the cation in intracellular stores (2-4). In electrically excitable cells, Ca2+< oscillations could rather be due to periodic activation of * This work was supported by Grant from the Fonds de la Recherche Scientifique Midicale, Brussels, by Grant F from the Fonds National de la Recherche Scientifique, Brussels, by Grant SPPS-AC 89/135 from the Ministry of Scientific Policy, Brussels, and by the Deutsche Forschungsgemeinschaft, Bonn-Bad Godesberg. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Chargi de Recherches of the University of Louvain. 11 Directeur de Recherches of the Fonds National de la Recherche Scientifique, Brussels. To whom correspondence should be addressed Unit6 de Diabitologie et Nutrition, UCL 54.74, ave. Hippocrate 54, B-1200 Brussels, Belgium. The abbreviation used is: Ca2+<, free cytoplasmic calcium nium ions (12). In another recent study using intact islets, rat only slow oscillations of Ca2+; (0.2/min) were observed during glucose stimulation (13). It remains thus unclear whether changes in membrane potential are the major drive of the changes in Ca2+i in B-cells. This possibility has even been questioned recently (14). In the present study, intact mouse islets loaded with the Ca2+, indicator fura-2 were used to characterize Ca*+, changes in B-cells. These changes were compared to those of the membrane potential separately recorded with intracellular microelectrodes. The islets were submitted to various types of stimulations to ascertain that possible correlations were not fortuitous. Since the effects of glucose on the membrane potential are mediated by changes in B-cell metabolism, we also continuously monitored the fluorescence of reduced pyridine nucleotides (NAD(P)H) to determine whether it is the pattern of the metabolic changes that eventually determines the pattern of the electrical and Ca2+, changes. Finally, the use of digital image analysis made it possible to assess the synchronization of Ca2+i changes in different regions of the islets. EXPERIMENTAL PROCEDURES Solutions The medium used was a bicarbonate-buffered solution which contained 120 mm NaC1, 4.8 mm KC], 2.5 mm CaC12, 1.2 mm MgCl,, and

2 20714 Cytosolic Ca2+i and Membrane Potential 24 mm NaHC03, and which was gassed with 02/COz (94:6) to maintain ph 7.4. When the concentration of KC1 was increased to 30 mm, that of NaCl was decreased accordingly. In certain experiments, the concentration of CaClZ was either decreased to 1 mm or increased to 10 mm without compensation for the change in osmolarity because control experiments had shown that this did not influence the results. The solutions used for membrane potential measurements did not contain albumin. Those used for fluorescence measurements were supplemented with bovine serum albumin (1 mg/ml) except in experiments where tolbutamide or diazoxide were tested, to avoid changes in free drug concentration. Animak All experiments were performed with pancreatic islets of fed female NMRI mice (25-30 g) killed by decapitation. changes in NAD(P)H fluorescence were expressed as a percentage of control values by dividing the integrated gray levels at a given time by those obtained during the last min preceding stimulation. Presentation of Results All measurements of membrane potential, Ca2+;, and NAD(P)H fluorescence are illustrated by recordings which are representative of results obtained with the indicated number of islets. For membrane potential measurements, this number of islets corresponds to different mice. For Ca2+i and NAD(P)H measurements, several islets from the same culture were tested with the same protocol, but each protocol was repeated with islets from at least three different cultures. Recordings of B-cell Membrane Potential RESULTS A piece of pancreas was taken immediately after the death of the animal and fixed in a perifusion chamber. A few islets were then Of fluorescence and were partially microdissected by hand. The membrane potential of a single made Under similar conditions (except for the absence offuracell within the islet was continuously measured with a high resistance 2 loading in the former case), whereas membrane potential microelectrode (15). B-cells were identified by the typical electrical recordings were obtained under different conditions (nonculactivity that they display in the Presence of 15 mm glucose. The tured islets, other perifusion system). The changes in the time temperature was 37 "c and the flow rate about 7 ml/min. The dead course of the first two events Can thus be directly compared, space of the system corresponds to approximately 10 s and has been corrected in the figures. but the comparison cannot be so strict with the membrane potential. This reservation should be born in mind through- Measurements of Ca" and Reduced Pyridine Nucleotides out. Preparation-Islets were isolated from the pancreas after collagen- Control Experiments-We first determined to which extent ase digestion of the tissue. They were then cultured for h at the autofluorescence of the islets contaminates the fura-2 37 "C in RPMI 1640 medium (Flow, Asse-Relegem, Belgium) con- signal. To this end, one islet that was not loaded with fura-2 taining 10% heat-inactivated fetal calf serum, 100 IU/ml penicillin, was studied simultaneously with one loaded islet, under optiand 100 pg/ml streptomycin. The concentration of glucose was 10 mal conditions for recording the fura-2 signal (excitation at mm. CQ"; Measurements-Cultured islets were loaded with fura-2 dur- 340 and 380 nm and recording at 510 nm). As shown in Fig. ing 35 min of incubation at 37 "C in 2 ml of control medium containing 1, increasing the concentration of glucose from 3 to 15 mm 10 mm glucose and 1 pm fura-2 acetoxymethylester (Molecular produced antiparallel changes in the fluorescence emitted by Probes, Eugene, OR, added from a 1 mm stock solution in MezSO). the islet loaded with fura-2 (upper traces) and parallel changes Loaded islets were then transferred into a temperature-controlled in the fluorescence emitted by the control nonloaded islet perifusion chamber (Applied Imaging, Sunderland, United Kingdom (lower truces). The intensity of the signals was also much (U. K.)) with a bottom made of a glass coverslip. They were held in place by gentle suction with a glass micropipette and perifused at a smaller in the control than in the loaded islet. In eight similar flow rate of 1.3 ml/min. The dead space of the system corresponds to 2 min and has been corrected in the figures. Perifusion solutions were Glucose 15rnM kept at 37 "C in a water bath, and the temperature controller ensured a temperature of 37.2 "C (+ 0.3 "C) close to the islet as monitored by a thermistor placed near the tissue. The perifusion chamber was mounted on the stage of an inverted microscope (Nikon Diaphot) used in the epifluorescence mode with a X20 objective. Fura-2 was successively excited at 340 and 380 nm by means of two narrow band-pass filters mounted on a computercontrolled motorized filter wheel placed in front of a 75-watt xenon lamp. A dichroic mirror centered at 430 nm reflected the UV light to the perifusion chamber and transmitted the emitted fluorescence which was further filtered through a 510-nm filter. Fluorescent images were obtained with a CCD video camera (Photonic Science Ltd., Tunbridge Wells, U. K.) at a resolution of 256 X 256 pixels. They were then digitized into 256 gray levels and analyzed with the system Magical of Applied Imaging. To improve the signal-to-noise ratio, eight consecutive 40-ms frames were averaged at each wavelength before ratioing. The time interval between successive series of images was 3.2 s. During these intervals, the excitation light was stopped to avoid photobleaching of fura-2. The concentration of Ca2+i was calculated by comparing the ratio of the fluorescence at each pixel to a calibration curve based on the equation of Grynkiewicz et al. (16). The KD for Ca2+-fura-2-complex employed was 224 nm. The mean Ca2+ concentration in the islet was then calculated by averaging the Caz+ concentration at all pixels of the islet. Measurements of Reduced Pyridine Nucleotide Fluorescence-Cultured islets were first preincubated for 35 min at 37 "C in 2 ml of control medium containing 10 mm glucose. They were then transferred to the same experimental set-up as for fura-2 fluorescence. The reduced forms ofnad and NADP, referred to as NAD(P)H, were excited at 360 nm, and the fluorescence emitted through a dichroic mirror centered at 400 nm was filtered at 470 nm. Eight consecutive 40-ms frames were acquired and averaged every 3.2 s. The resolution was thus the same as for fura-2 experiments. The 1 rnin FIG. 1. Effects of an increase of the glucose (G) concentration from 3 to 15 mm on the fluorescence emitted at 510 nm by two islets studied simultaneously and excited at 340 and 380 nm. The upper two traces were recorded from an islet that was loaded with fura-2 whereas the lower two traces were recorded from a control, nonloaded islet. To avoid overlapping, the 340-nm trace from the islet loaded with fura-2 was shifted upward by 10 gray levels. These traces are representative of results obtained in eight similar experiments.

3 experiments, the autofluorescence signal averaged 14.5 f 1 and 18.2 f 1.2% of the signal in the islet loaded with fura-2 at excitation wavelengths of 340 and 380 nm, respectively, when the concentration of glucose was 3 mm. Under steadystate stimulation with 15 mm glucose, the contamination by the autofluorescence signal averaged 19.9 f 1.1% at 340 nm and 29.5 & 1.8% at 380 nm. As the contamination is relatively larger at 380 than at 340 nm, the Ca2+i calculated from the ratio is marginally underestimated in the basal state and slightly more during glucose stimulation. It is important, therefore, to regard the presented Ca2+, values as indicative. In a second series of control experiments, we assessed that the autofluorescence signal recorded at 470 nm from nonloaded islets excited at 360 nm can be used to monitor cell metabolism. As shown in Fig. 2, raising the concentration of glucose from 3 to 15 mm caused a marked increase in autofluorescence. Subsequent addition of azide, which interferes with site 3 of the mitochondrial respiratory chain and thus prevents oxidation of pyridine nucleotides, caused a rapid and reversible increase in the fluorescence. On the other hand, mannoheptulose, which prevents glucose metabolism by inhibiting the phosphorylation by glucokinase, caused a marked decrease in the fluorescence signal (Fig. 2). The islet autofluorescence recorded under these conditions thus reflects NAD(P)H levels and, as in previous studies (11, 17), can be used to monitor the rate of metabolism. Effects ofglucose-fig. 3 illustrates typical changes in reduced pyridine nucleotides, membrane potential, and Caz+i produced in single islets by a change in glucose from a nonstimulatory (3 mm) to a stimulatory (15 mm) concentration in a control medium containing 2.5 mm Ca2+. The NAD(P)H fluorescence significantly increased after 46 f 3 s. This increase was smooth and reached 155 f 1% of the basal value after 8 min (n = 20). In low glucose, the resting potential of B-cells was between -60 and -70 mv. Approximately 1 min after the change in glucose concentration, the membrane slowly depolarized to a threshold potential. When this threshold was reached, the membrane rapidly depolarized to a plateau potential onto which spikes (action potentials) appeared. NADIPIH 1% I I loo} - U Glucose 15mM Azide + Glucose 15mM + Cytosolic Ca2+i and Membrane Potential MH * I U 2 min FIG. 2. Effects of azide (2 mm) and mannoheptulose (MH, 20 mm) on NAD(P)H fluorescence in mouse islets. The concentration of glucose (G) was increased from 3 to 15 mm, and the test substances were added athe times indicated by the arrows. NAD(P)H fluorescence is expressed as a percentage of the signal measured during the last minute preceding the change in glucose concentration. The traces correspond to the mean responses (+ S.E.) in six islets. NADIPIH 100. G 3mM, Glucose 15mM. Ca 2SmM 50 l" / - 1 rnin FIG. 3. Correlations between the changes in reduced pyridine nucleotides (NAD(P)H), B-cell membrane potential (ME'), and Ca2+in mouse islets stimulated by an increase of glucose (C) concentration from 3 to 15 mm. NAD(P)H fluorescence is expressed as a percentage of the signal measured during the last minute preceding the change in glucose concentration. These recordings are representative of results obtained in 20 islets (NAD(P,JH), more than 30 islets (MP), and 30 islets (Ca",) for experiments made in the presence of 2.5 mm extracellular Caz+, and in 15 islets (NADfP)H) and nine islets (Ca2+,) for experiments made in the absence of extracellular Caz+. After some time, the spike activity stopped, the membrane partially repolarized, and the membrane potential started to oscillate in regular slow waves with spikes superimposed on the plateau. In the presence of 3 mm glucose, the average Ca2+i concentration in islet cells was 75 f 3 nm (n = 42). Upon stimulation with 15 mm glucose, Ca2+i changed in three phases. An initial, small but consistent decrease averaging 6.5 f 0.4 nm (n = 30, p < by paired t test) occurred after 69 f 3 s. It is not an artifact caused by the change in autofluorescence, as shown by the antiparallel changes in the 510 nm signal, which increases during excitation at 380 nm and decreases during excitation at 340 nm (Fig. 1, upper truces). The initial decrease in Ca", was followed by a marked increase (Fig. 3) that started after 96 f 3 s and reached a maximum of 213 f 6 nm after 142 f 3 s. This first increase usually lasted about 1 min, sometimes more (Fig. 3,lower punel), and was followed by oscillations of Ca2+*. Similar experiments were performed in the absence of extracellular Ca2+i except for membrane potential recordings which we are not able to perform under these conditions. In the Ca2+-free medium, basal Ca2+i concentration in 3 mm glucose averaged 37 f 2 nm, a value lower than in the presence of extracellular Ca2+ (n = 18, p < by unpaired t test). Increasing the concentration of glucose caused a small lowering of Ca2+i (4.0 f 0.3 nm, n = 9, p C by paired t test). The increase in NAD(P)H levels brought about by 15 mm glucose persisted in the absence of Ca2+ (Fig. 3) and, relative to basal values in 3 mm glucose, appeared similar to that occurring in the presence of Ca2+. These experiments, there-

4 20716 Potential Cytosolic Membrane and fore, establish that Ca2+ is not necessary for glucose to increase NAD(P)H levels, but do not exclude the possibility that Ca2+ influences the magnitude of this increase because basal fluorescence may be affected by Ca2+ omission. Fig. 4A shows that raising the concentration of glucose from 3 to 15 mm in a Ca2'-free medium increased NAD(P)H levels in islet cells (from 100 to 156 f 2%, n = 6) and that the subsequent readmission of Ca2+ was followed by a further increase to 184 k 2%. To permit a comparison of the effects of Ca2+ in low and high glucose, islets were perifused with a medium containing 3 or 15 mm glucose and no Ca2+ for 18 min. As shown in Fig. 4, B and C, Ca2+ readmission then caused an increase in NAD(P)H levels by 4.0 f 0.7% (n = 12) in 3 mm glucose and 14.9 f 1.2% (n = 9) in 15 mm glucose. Effects of Tolbutamide and Other Agents-Addition of 100 pm tolbutamide to a medium containing 3 mm glucose was followed by a delayed and slight increase in NAD(P)H fluorescence (Fig. 5). This increase was significant after 124 f 11 s and reached 108 f 1% (n = 9) of basal value after 7-8 min. The depolarization brought about by tolbutamide was rapid and sustained, leading to continuous spike activity. The attending increase in CaZfi was also rapid (14 f 2 s, n = 9), reached a maximum of213 f 11 nm after s, and remained sustained (204 f 11 nm after 7-8 min). In a Ca2+free medium, tolbutamide increased neither Ca2+L nor NAD(P)H levels (Fig. 5). During steady-state stimulation with 15 mm glucose, B- cells exhibited a rhythmic electrical activity consisting of regular slow waves of the membrane potential with superimposed spikes (Fig. 6). Under these conditions, Ca2+i was also found to oscillate in a regular manner. The rhythmic electrical activity was transformed into a sustained depolarization at the plateau potential, with continuous spike activity by addition of tolbutamide or arginine to the medium. These addi r l2or L 3mln Ca 2.5mM CaO Ca 25mM - FIG. 4. Influence of CaZ+on NAD(P)H fluorescence in mouse islets. Panel A, the concentration of glucose (G) was increased from 3 to 15 mm in a Caz+-free medium, and 2.5 mm Ca2+ was subsequently introduced into the medium. Panels B and C, Ca2+ (2.5 mm) was added to a medium containing 3 or 15 mm glucose after 18 min of perifusion in the absence of Ca2+. NAD(P)H fluorescence is expressed as a percentage of the signal recorded during the last minute preceding the change in glucose concentration (panel A) or the addition of Ca2+ (panels B and C). The traces correspond to the mean responses (+ S.E.) in six (panel A), 12 (panel B), and nine (panel C) islets G 3mM Tolbutamide 100pM. NADIPIH. Ca 25mM l%l 4. "\, MP (mvi,-. ".. "_ (a? 2o011Ca 25mM (nm I,_ I - 1 mln " FIG. 5. Correlations between the changes in reduced pyridine nucleotides (NAD(P)H), B-cell membrane potential (MP), and Ca2+; in mouse islets stimulated by 100 p~ tolbutamide in the presence of 3 mm glucose. NAD(P)H fluorescence is expressed as a percentage of the signal measured during the last minute preceding the change in glucose concentration. These recordings are representative of results obtained in nine islets (NAD(P)H), more than 20 islets (MP), and nine islets (Ca",) for experiments made in the presence of 2.5 mm extracellular (!a2+, and in nine islets (NAD(P)H) and nine islets (Ca",) for experiments made in the absence of extracellular Ca". tions were also followed by sustained increases in Ca2+i; this increase was, however, more stable after tolbutamide (237 f 10 and 223 f 9 nm after 1 and 6-7 min, respectively) than after arginine (272 f 11 and 226 f 8 nm after 1 and 6-7 min, respectively) (Fig. 6). Increasing the concentration of extracellular K' to 30 mm not only produced a continuous depolarization of the B-cell membrane, but also decreased the plateau potential by about 10 mv and caused a progressive decrease in spike amplitude. The attending rise in Ca2+i was larger (348 f 16 nm at the peak), but not so sustained (269 f 12 nm after 6-7 min) as those produced by tolbutamide or arginine (Fig. 6). Diazoxide exerts effects opposite of those of tolbutamide. It inhibits insulin release by increasing K+ conductance of the B-cell membrane through opening of ATP-sensitive K+-channels (18, 19). Addition of 50 PM diazoxide to a medium containing 15 mm glucose abolished electrical activity in B- cells and repolarized the membrane to the resting potential (Fig. 7). It also suppressed the oscillations of Ca2+* and decreased the average Ca2+, to levels (70 f 4, n = 12) not different from those measured in 3 mm glucose. Subsequent addition of 25 PM tolbutamide reversed the effects of 50 PM diazoxide on both membrane potential and Ca2+i. No oscillations were seen under these conditions. However, when the concentration of diazoxide was raised to 250 PM, the membrane partially repolarized and slow waves appeared, which were shorter than those seen in 15 mm glucose alone. Ca2'i also decreased but remained higher than in low glucose and started to oscillate rapidly. We next compared the influence of the concentration of extracellular Ca2+ on the membrane potential and on Ca2+i (Fig. 8). When external Ca2' was lowered from 2.5 to 1 mm, slowwaves of the membrane potential were progressively replaced by a sustained depolarization and continuous spike activity. These spikes were larger than in the presence of 2.5

5 Cytosolic Ca2+i and Potential Membrane G lsmm Tolbutamide 100pM b G 15mM - Ca 25mM Ca 1mM -601'uuuuuuuu L_J I 150 InM) 250- caf. InMl 100- ~ G 15mM C Argmne lomm Or Ca 25mM, Ca lomm b t [a: 300! 150 InM1 - G lsmm n, - 1 mln K 30mM + b FIG. 6. Correlations between the changes in B-cell membrane potential (MP) and Caz+i in mouse islets stimulated by 100 pm tolbutamide, 10 mm arginine, or 30 mm K+ in the presence of 15 mm glucose. These recordings are representative of results obtained in five to seven islets (MP), and in islets (Ca",). Note that the scale for the changes in Caz+, in the lower panel is not the same as above. FIG. 7. Correlations between the changes in B-cell membrane potential (MP) and Ca2+i produced in mouse islets by various combinations of diazoxide and tolbutamide. The concentration of glucose (G) was 15 mm throughout. These recordings are representative of results obtained in four islets (MP) and in 12 islets (Ca":). mm Ca2+ because they started from a more negative potential; their frequency was lower than that of the spikes appearing on the plateau of control slow waves. The decrease in external Ca2+ was followed by an initial fall of Ca2+z to an average of 1 mln FIG. 8. Correlations between the changes in B-cell membrane potential (MP) and Ca2+i produced in mouse islets by a decrease to 1 mm or by an increase to 10 mm of the Ca2+ concentration in the medium. The concentration of glucose (G) was 15 mm throughout. These recordings are representative of results obtained in five to six islets (MP) and in nine and 17 islets (Ca2+;) nm (n = 9). Subsequently, Ca2+increased and stabilized at an average level of 108 f 9 nm. No clear oscillations of Ca"; were observed in the presence of 15 mm glucose and 1 mm extracellular Ca2+ (Fig. 8). Raising the concentration of external Ca2+; to 10 mm also altered glucose-induced electrical activity in B-cells (Fig. 8). The most prominent changes were a slight depolarization of the plateau of the slow waves and a marked lengthening of the intervals between the slow waves, with a decrease in slow wave frequency from 2.2 +: O.l/min to 1.1 f O.l/min; n = 7. The oscillations of Ca2+, were also profoundly changed. Their peak increased, their frequency dropped from 2.5 f 0.3/min to 0.6 f O.OS/min (n = 17), and their nadir during the intervals was much lower than under control conditions. The descending phase of the Ca2+; oscillations clearly displayed two components: an initial fast one and a second much slower one. Characteristics of the Ca", Oscillations Induced by Glucose- When islets were perifused with 15 mm glucose, the membrane potential was found to oscillate in slow waves in all tested B- cells. Unambiguous oscillations of Ca2+i were observed in 93.5% of the islets (129:138). The remaining nine islets displayed fast fluctuations of Ca2+; which could not be reliably resolved with our system (less than four ratioed images/ fluctuation). The amplitude of the unambiguous Ca2+, oscillations ranged from 20 to 130 nm. Their frequency was also variable. Fig. 9 shows examples of oscillations of membrane potential and Ca2+, with particularly low and high frequencies. It also shows distribution histograms of the frequency of these oscillations. The distribution is almost perfectly gaussian for the slow waves, but slightly asymmetric, with a higher proportion of low frequencies for Ca2+i oscillations. The mean and median values are 2.25 and 2.27/min for slow waves (n = 200) and 2.19 and 2.14/min for oscillations of Ca2+; (n = 129).

6 20718,::,"t /"v+vv 1 min O 0 S Cytosolic Ca2+i and Membrane Potential Frequency (per rninl FIG. 9. Upper left, examples of B-cells exhibiting membrane potential oscillations (slow waves) with low or high frequency in the presence of 15 mm glucose. Lower left, examples of B-cells exhibiting cytosolic Ca2+, oscillations with low or high frequency in the presence of 15 mm glucose. Right, distribution histograms of the frequency of membrane potential and Ca2+, oscillations in (n) islets stimulated by 15 mm glucose. FIG. 10. Synchronism of Ca2+i oscillations in various regions of islets stimulated by 15 mm glucose. The mean Ca2+i concentration in whole islet (I) or in regions of the islet (2-7) was calculated by averaging the Ca2+ concentration at all pixels of the whole islet or of each separate region. Panels A-C correspond to three islets displaying Ca2+, oscillations with different frequencies. Note the different time scales. The uertical calibration bars correspond to 100 nm Ca2+i. We finally investigated whether Ca2+; oscillations are similar in all regions of the islet. This could be made by integrating the signal captured by the camera over the whole islet and over subregions of the islet. This type of analysis was performed in 27 islets displaying Ca2+i oscillations of variable frequencies. As shown by three examples in Fig. 10, Ca2+i was found to oscillate in synchrony in all regions of the islets. In some islets, the amplitude of the oscillations was also fairly similar in all regions (Fig. loa), whereas differences were noted in other islets: compare traces B2 and B4, or C3 and C5 (the largest observed difference). In certain islets, the amplitude of the oscillations was higher at the periphery than in the center (Fig. 10, B6 and B7), but this was not always the case (islets A and C). DISCUSSION Although the measurements were made separately and under distinct conditions dictated by technical constraints, an excellent correlation was found between the changes in membrane potential and Ca2+in normal mouse pancreatic B-cells challenged with a number of stimulators or inhibitors. On the other hand, recordings of NAD(P)H fluorescence, which pro- vide a dynamic measure of metabolic changes in B-cells, did not reveal clear oscillations, similar to those of membrane potential and Ca2+; during stimulation with glucose. The NAD(P)H fluorescence of the islets increased after stimulation byglucose. This increase was rapid, occurring sooner than the changes in membrane potential and Ca2+,, persisted in the absence of extracellular Ca2+, was monophasic, and did not display oscillations in the steady-state. The last three characteristics have been noted before in experiments using oblob mouse islets (17), whereas the increase in NAD(P)H fluorescence was often biphasic in single B-cells (11). That the changes in NAD(P)H precede the increase in Ca2+i has also been shown at the level of individual B-cells (11). An essential finding of the present study is that no oscillations of NAD(P)H fluorescence are observed under conditions where Ca2+i oscillations are easily seen. This does not necessarily mean that oscillations of metabolism do not occur. These could be present in certain cells only, be asynchronous in different cells, be very fast and escape detection by our system, or be slow and detectable only during longer periods of observation than those used here. Oscillations of O2 consumption with a period of about 5 min have been noted in intact rat islets (13). However, our data do not support the conclusion that the biphasic and oscillatory pattern of electrical activity and Ca2+; produced by glucose is secondary to similar synchronous oscillations of glucose metabolism in all B-cells. On the other hand, oscillations of the concentration of metabolic second messengers (e.g. the ATP/ADP ratio) beneath the plasma membrane might be involved (20). In contrast to glucose, tolbutamide only caused a very modest increase in NAD(P)H fluorescence that had not been detected in previous studies (17). This increase started after that of Ca2+, and did not occur in the absence of extracellular Caz+. We suggest that the small change in NAD(P)H fluorescence triggered by tolbutamide is the consequence of Ca2+ influx. It is indeed known that tolbutamide causes a Ca2+dependent increase in O2 consumption by islets perifused in the presence of glucose (21). This interpretation is also in keeping with our observation that introduction of Ca2+ into a Ca2+-free medium increases NAD(P)H levels in islets perifused with low or high glucose. In single B-cells a decrease in NAD(P)H fluorescence (11) and in Ca2+i (8, 11, 22) sometimes precedes the increases produced by glucose. A small, but consistent decrease in Ca2+; also initially occurred in whole islets stimulated by a rise in glucose concentration from 3 to 15 mm, but we did not observe any initial decrease in NAD(P)H fluorescence. On the contrary, NAD(P)H fluorescence was already elevated when the small fall in Ca2+; occurred. This fall is currently attributed to metabolism-induced sequestration of Ca2+ in cellular organelles (23). It has been shown previously that the increase in Ca2+; produced by glucose or tolbutamide in dispersed B-cells is prevented by omission of extracellular Ca2+ (23, 24). The present study establishes that the same holds true for intact islets. Moreover, our observations that the rise in Ca2+i brought about by glucose is biphasic and dependent on external Ca2+ during both phases lends support to the following proposal. On stimulation by glucose, B-cells respond by a biphasic increase in electrical activity (25), which reflects a biphasic increase in Ca2+ influx that leads to a biphasic rise in Ca2+i with, eventually, a biphasic release of insulin (26). The fact that both phases of glucose-induced insulin release depend on Ca2+ influx (27) is also in keeping with this pro- posal. However, we do not wish to suggest that there is always a tight relationship between Ca2+i and the rate of insulin

7 Cytosolic Potential Membrane Ca2+i and release. Wollheim and Biden (28) have already pointed out numbers of experiments (n = 200 and 129), it is hard to that Ca"; is not the sole regulator of secretion, and we recently believe that this similarity is simply fortuitous. We rather presented evidence that once Ca2+i has been increased in B- suggest that it reinforces the conclusion that Ca2+, oscillations cells, glucose can modulate insulin release without further are secondary to the influx of Ca2+ occurring during the slow changing Caz+i (29). waves. There was an excellent correspondence between the In earlier experiments using single B-cells, glucose produced changes in B-cell membrane potential and the changes in a relatively sustained increase in CaZfi without regular oscil- Ca2+i under practically all tested conditions. Oscillations of lations (11). This could be due to the use of only 1 mm CaC12 Ca2+; were seen whenever the Ca2+-dependent electrical activ- in the medium. In other experiments, glucose induced large ity was rhythmic, and they disappeared when the membrane oscillations ofca", which had a very low frequency (0.05- became persistently depolarized. This steady elevation of 0.5/min) (8-10). Oscillations of Ca2+i and membrane potential Ca2+; was independent of the mechanism by which the mem- with a frequency of less than 0.5/min were exceptional in the brane was depolarized a blockade of ATP-sensitive K+ chan- present and other (33) studies with whole mouse islets. This nels by tolbutamide (18, 19), the entry of positive charges difference might be due to the fact that glucose-induced with arginine (30) or the shift of the equilibrium potential of electrical activity is altered in single B-cells (34, 35), which K+ in high K+ solutions. The fact that the increase in Ca2+* must be electrically coupled to display regular slow waves was less stable in high K+ or arginine than with tolbutamide (36). may reflect the progressive disappearance of the spikes in The use of a camera to record the fluorescence signal made high K+ and the decrease in spike frequency in the presence it possible to compare Ca2+, oscillations in different regions of arginine. Diazoxide and tolbutamide exert oppositeffects of the islets. It was found that the oscillations, were they of on the membrane potential of B-cells by opening and closing low or high frequency, were synchronous in all regions, but ATP-sensitive K+ channels, respectively (18, 19). These that their amplitude was sometimes variable. It is unlikely changes in membrane potential were faithfully paralleled by that this variability simply reflects an optical artifact linked changes in Ca2+;. The changes in insulin release also follow a to different depths of tissue because the amplitude of the similar pattern (18). oscillations was not consistently higher at the periphery than The only apparent dissociation between the changes in at the center. We are presently unable to distinguish between membrane potential and in Ca2+, was observed when the concentration of extracellular Ca2+ was lowered from 2.5 to 1 mm. The membrane rapidly remained depolarized and exhibited continuous spike activity, while Ca2+, first fell markedly. This suggests that the homeostatic mechanisms which ex- of the presence of non-b or nonresponsive B-cells. In any trude Ca" from the cytoplasm transiently overcompensated case, we note that the variability of the Ca2+; response between the decrease in Ca2+ influx resulting from the decrease in Ca2+ subregions of an islet is much smaller than the variability electrochemical gradient. It may seem surprising that steady- between individual B-cells after dispersion (11). We did not state Ca", was lower in 1 mm than in 2.5 mm Ca2+, while the observe propagation of clear Ca", waves throughout the islets. electrical activity was continuous. There are two explanations. However, the resolution of the system is not sufficient to First, the plateau, which itself corresponds to an inward Ca2+ exclude that such waves occur within small territories of current (6, 7), is more negative in 1 than in 2.5 mm Ca2+. neighboring cells (10) and that the signal starts in pacemaker Second, the frequency of the spikes occurring on the plateau cells. Moreover, our results showing that Ca", apparently is much lower than that of the spikes occurring on the plateau increases in all B-cells of an islet do not necessarily contradict of slow waves (31). the concept of B-cell functional heterogeneity (37, 38). It has The changes in Ca2+i produced by an increase in the concentration of extracellular Ca2+ from 2.5 to 10 mm corresponded well with the changes in B-cell membrane potential. They revealed two interesting features. First, the concentration of Ca2+, kept on decreasing during the interval between two oscillations while the membrane slowly depolarized during that interval. This indicates that the slow depolarization is unlikely to be due to a pacemaker inward Caz+ current. Second, there was a marked difference between the concentrations of Ca2+; at the start of an oscillation in the presence of 2.5 and 10 mm Ca2+ in the medium. This suggests that the opening of Ca2+ channels at the beginning of each slow wave is not directly controlled by Ca2+;. It has been pointed out previously that the frequency of the slow waves of membrane potential induced by 11 mm glucose is highly variable from islet to islet and displays a bimodal distribution with peaks around 2 and 4 slow waves/min (32). We also observed a substantial variability of the slow wave frequency in the presence of 15 mm glucose, but the distribution was unimodal, with a mean of 2.25 f O.O6/min. The frequency of Ca2+i oscillations also varied from islet to islet, but the distribution was also unimodal with a mean of 2.19 f O.lO/min. The frequency distributions of the oscillations of membrane potential and Ca2+1 were thus strikingly similar despite the different conditions of recording. With so large two possibilities. First, certain regions of certain islets respond with a synchronous but smaller rise in Ca"; because they possess fewer functional Ca2+ channels. Second, the amplitude of the Ca2+i oscillations is smaller in certain regions because been suggested recently that B-cells which do not secrete insulin in response to glucose have normal ion channels and a normal electrical response to the stimulus (39). In conclusion, there is an excellent correlation between the changes in membrane potential and in Ca2+; in normal mouse B-cells within intact islets challenged by different stimulators or inhibitors of insulin release. This strongly suggests that Ca2+t changes are driven by similar oscillations of the membrane potential and of voltage-dependent Ca2+ channel activ- ity. The oscillations of the Ca2+i responses induced by glucose are synchronous in all B-cells of an islet. They are unlikely to result from similar oscillations of glucose metabolism. Acknowledgments-We are grateful to M. Nicaise and W. Schmeer for skillful assistance and to M. Nenquin for editorial help. REFERENCES 1. Douglas, W. W. (1968) Br. J. Pharmacol. 34, Berridge, M. J. (1990) J. Biol. Chem. 266, Jacob, R. (1990) Biochim. Biophys. Acta 1062, Putney, J. W. (1990) Cell Calcium 11, Dean, P. M., and Matthews, E. K. (1968) Nature 219, Henquin, J. C., and Meissner, H. P. (1984) Experientia 40, Cook, D. L., Satin, L. S., and Hopkins, W. F. (1991) Trends Neurosci. 14, Grapengiesser, E., Gylfe, E., and Hellman, B. (1988) Biochem. Biophys. Res Commun. 151, Wang, J.-L., and McDaniel, M. L. (1990) Biochem. Biophys. Res. Commun. 166, Gylfe, E., Grapengiesser, E., and Hellman, B. (1991) Cell Calcium 12,

8 20720 Cytosolic Potential Membrane Ca2+i and 11. Pralong, W.-F., Bartley, C., and Wollheim, C. B. (1990) EMBO J. 9, Santos, R. M., Rosario, L. M., Nadal, A,, Garcia-Sancho, J., Soria, B., and Valdeolmillos, M. (1991) Pfliigers Arch. 418, Longo, E. A., Tornheim, K., Deeney, J. T., Varnum, B. A,, Tillotson, D., Prentki, M., and Corkey, B. E. (1991) J. Biol. Chem. 266, Ammala, C., Larsson, O., Berggren, P.-0.. Bokvist, K., Juntti-Berggren, L., Kindmark, H., and Rorsman, P. (1991) Nature 363, Meissner, H. P. (1990) Methods Enzymol. 192, Grynkiewicz, G., Poenie, M., and Tsien, R. Y. (1985) J. Biol. Chem. 260, Panten, U., Christians, J., Kriegstein, E. V., Poser, W., and Hasselblatt, A. (1973) Diabetologia 9, Henquin, J. C., and Meissner, H. P. (1982) Biochem. Phrmacol. 31, Trube, G., Rorsman, P., and Ohno-Shosaku, T. (1986) Pfliigers Arch. 407, Henquin, J. C. (1990) Diabetes 39, Panten. U.. Zunkler. B. J.. Scheit. S.. Kirchhoff, K., and Lenzen, S. (1986) Diabetologia 29, Yada, T., Kakei, M., and Tanaka, H. (1992) Cell Calcium 13, Gvlfe. Gylfe, E. (1988) J. Biol. Chem , G;apengiesser,'E., Grapengiesser, E., Gylfe, E., and Hellman, B. (1990) Mol. Phrmacol. 37, Meissner, H. P. (1976) J. Physiol. (Paris) 72, Curry, D. L., Bennett, L. L., and Grodsky, G. M. (1968) Endocrinology 83, Henquin, J. C. (1978) Endocrinology 102, Wollheim, C. B., and Biden, T. J. (1986) Ann. N. Y. Acad. Sei. 488, Gembal, M., Gilon, P., and Henquin, J. C. (1992) J. Clin. Inuest. 89, Henqum, J. C., and Meissner, H. P. (1981) Am. J. Physiol. 240, E245- E Ribalet, B., and Beigelman, P. M. (1980) Am. J. Physiol. 239, C124-C Atwater, I., Dawson, C. M., Scott. A,, Eddlestone, G., and Rojas, E. (1980) Horm. Metab. Res. (Suppl. 10) Valdeolmillos, M., Santos, R. M., Contreras, D., Soria, B., and Rosario, L. M. (1989) FEBS Lett. 259, Smith, P. A,, Ashcroft, F. M., and Rorsman, P. (1990) FEBS Lett. 261, Dunne, M. J., and Petersen, 0. H. (1991) Biochim. Biophys. Acta Sherman, A,, Rinzel, J., and Keizer, J. (1988) Biophys. J. 54, Salomon, D., and Meda, P. (1986) Ex. Cell Res. 162, Hiriart, M., andramlrez-mendeles, d C. (1991) Endocrznology 128,3193- I1 OP 39. Soria, B., Chanson, M., Giordano, E., Bosco, D., and Meda, P. (1991) Diabetes 40,