Synchronous oscillations of cytoplasmic Ca2+ and insulin release in glucose-stimulated pancreatic islets.

The cytoplasmic Ca2+ concentration ([Ca2+]i) was measured in single pancreatic mouse islets superfused in a system allowing concomitant recordings of insulin release. When glucose was raised from 3 to 11 mM, [Ca2+]i responded by a transient lowering followed by a rise to an average level of 192 +/- 11 nM. In 77% of the islets the rise was associated with the gradual appearance of oscillations, which were either fast (2-7/min), slow (0.3-0.9/min), or a combination of both types. The characteristics of the fast [Ca2+]i oscillations were those expected from a relationship with the electrical burst activity in islets. Accordingly, in most cases the fast oscillations were remarkably regular. The slow [Ca2+]i oscillations had characteristics similar to the large amplitude ones in individual beta-cells. Whereas glucagon and dibutyryl cAMP could transform slow islet oscillations into fast ones, the alpha 2-adrenergic agonist clonidine had the opposite effect. The rapid islet oscillations were also facilitated by elevated concentrations of extracellular Ca2+. Reinforcing the arguments for [Ca2+]i oscillations as responsible for a pulsatile insulin secretion it was possible to demonstrate that the release of the hormone from single islets is synchronized with the slow [Ca2+]i oscillations.

The cytoplasmic Ca2+ concentration ([Ca2+Ii) was measured in single pancreatic mouse islets superfused in a system allowing concomitant recordings of insulin release. When glucose was raised from 3 to 11 mM, [Ca2+li responded by a transient lowering followed by a rise to an average level of 192 t 11 nM. In 77% of the islets the rise was associated with the gradual appearance of oscillations, which were either fast (2-71min), slow (0.3-0.W min), or a combination of both types. The characteristics of the fast [Ca2+li oscillations were those expected from a relationship with the electrical burst activity in islets. Accordingly, in most cases the fast oscillations were remarkably regular. The slow [Ca2+li oscillations had characteristics similar to the large amplitude ones in individual p-cells. Whereas glucagon and dibutyryl CAMP could transform slow islet oscillations into fast ones, the tu,-adrenergic agonist clonidine had the opposite effect. The rapid islet oscillations were also facilitated by elevated concentrations of extracellular Ca2+. Reinforcing the arguments for [Ca2+Ij oscillations as responsible for a pulsatile insulin secretion it was possible to demonstrate that the release of the hormone from single islets is synchronized with the slow [Ca2+li oscillations. Stimulation of insulin release by glucose and other nutrients is mediated by a rise of the cytoplasmic Ca2+ concentration ([Ca2+Ii) following increased entry of the ion into the pancreatic p-cells (1). A characteristic feature of the [Ca2+Ii response to glucose is its oscillatory nature observed both in individual p-cells (2-6) and in intact pancreatic islets (7-10). These oscillations may have important physiological implications in being responsible for a pulsatile release of insulin. In analogy to what has been proposed for the periodic release of other biologically active peptides it can be supposed that the cyclic variations of circulating insulin prevent down-regulation of the peripheral receptors (11).
It has been reported that glucose triggers the appearance of both fast (7,8, 10) and slow (7,9) oscillations of [Ca2+lj in intact pancreatic islets. In the present study these oscillatory events h a v e been characterized in the a t t e m p t t o understand the mechanisms for their generation and significance for insulin secretion. The observations support the concept that the fast type of oscillations reflects the bursts of action potentials ob-* This study was supported by Grants 12x462 and 12x-6240 from the
Loading with Ca2+ Indicator and Superfusion of Cells-Pancreatic islets were isolated by collagenase using 10-month-old oblob mice taken from a non-inbred colony (12). These islets consist of more than 90% 6-cells, which respond normally to glucose and other regulators of insulin release (13). The isolated islets (1-2 pg, dry weight) were kept overnight at 37 "C in RPMI 1640 medium supplemented with 10% fetal calf serum, 100 IU/ml penicillin, 100 pg/ml streptomycin, and 30 pg/ml gentamicin. Further experimental handling was performed with a basal medium physiologically balanced in cations with CI-as the sole anion (14) and containing 0.5 mg/ml albumin and unless otherwise stated 2.56 mM Ca2+. After loading with fura-2 during 40 min of incubation a t 37 "C with 2 PM acetoxymethyl ester together with 0.02% (w/vj Pluronic F-127 the islets were rinsed and allowed to attach to the central part of a circular 25-mm coverglass coated with poly-L-lysine. The coverglass was used as the bottom of an open chamber containing 160 pl of medium (15). The chamber was placed on the stage of an inverted microscope (Nikon Diaphot) within a climate box maintained a t 37 "C by an air stream incubator, and the islet was superfused a t a rate of 0.3 mlimin. The microscope was equipped for epifluorescence fluorometry with a 400-nm dichroic mirror and a x 100 UV fluorite objective (Nikon).
Measurements of Cytoplasmic Ca2+-[Ca2+l, was recorded with dual wavelength fluorometry with excitation a t 340 and 380 nm and emission at 510 nm as described by Grynkiewicz et al. (16). The emitted fluorescence was measured with a photomultiplier or by image analysis adhering to previously adopted procedures ( 2 , 17). The analyses were performed in an optical plane approximately midway between the core and the lower surface of the islet. In the photomultiplier approach a time-sharing spectrophotofluorometer (18) provided light flashes of I-ms duration at 340 and 380 nm every 10 ms, and the 340/380-nm fluorescence excitation ratio as well as the 380-nm fluorescence were recorded from a circular measuring area with a diameter of 50-90 pm, Image analysis was used for comparing different parts of the islets. This alternative was based on a Magiscan system (Applied Imaging, Gateshead, U.K.) and the Tardis program, collecting the images with an intensified CCD camera (Extended ISIS-M, Photonic Science, Robertsbridge, U.K.). An excitation filter wheel provided a 340/380-nm image pair every 3.5 s with 1.1 s between images. The image analysis made it possible to measure [Ca2+], in selected areas of 2-3 cells.
The fluorescence excitation ratio remained unaffected during glucose stimulation in islets lacking the fura-2 indicator. The which was <15%. Neither was there any effect of blocking the anion channels in the islets with 10 PM sulfinpyrazone, indicating that leakage of fura-2 into the extracellular space did not interfere with the present measurements.
Measurements of Insulin Release-The perifusate was collected in 18-s fractions, which were immediately put on ice. Insulin was measured adhering to a previous protocol (19) by competitive enzyme-linked immunosorbent assay with the insulin antibody immobilized directly onto the solid phase (20). The results were expressed in terms of dry weight after freeze-drying and weighing of the islet on a quartz fiber balance.

RESULTS
A resting [Ca2+Ii of 63 f 3 nM ( n = 44) was recorded at 3 m M glucose. Increase of the glucose concentration to 11 m M resulted in a slight lowering of [Ca2+Ii followed by a rise to an average plateau level of 192 5 11 nM. Different types of responses to glucose are shown in Fig. 1. In 77% of the islets (34 out of 44) the rise of [Ca2+li was associated with a gradual appearance of oscillations from an elevated level. Three major categories of oscillatory activity were identified. In 27% of the islets the oscillations had a frequency of 2-7/min (panel B ). A slower type of oscillation with a frequency of 0.3-0.9/min was seen in 27% of the islets (panel C ) . In the remaining 23% there was a mixed pattern with fast oscillations superimposed on the slow ones (panel D). Although the majority of the fast [Ca2+li oscillations displayed a remarkable regularity, there were also irregular ones. In the latter case the durations of the cycles varied periodically with a frequency similar to that of the slow oscillations (Fig. 2).
The type of oscillatory [Ca2+li response was influenced by agents modulating CAMP (Fig. 3). In islets with slow [Ca2+li oscillations the addition of 10 nM glucagon resulted in the gradual appearance of a rapid pattern in 3 out of 8 cells (panel A). Dibutyryl CAMP had a similar action in 7 out of 11 cells (panel B ). The reverse effect was more reproducible, spontaneous rapid oscillations being transformed into slow ones in 5 out of 6 cells by 100 nM a2-adrenergic agonist clonidine (panel C ) . In the remaining cell the rapid pattern was replaced by sustained increase of [Ca2+Ii. With 10 n M clonidine the transformation of fast oscillations into slow ones was less prompt (not shown). Also the extracellular Ca2+ concentration affected the type of response. Lowering Ca2+ from 2.56 to 1.28 m M during superfusion with 11 m M glucose sometimes resulted in immediate replacement of the uniform fast [Ca2+li oscillations with a pattern of irregular small ones from the elevated level (Fig. 4,  panel A). The effect of raising glucose from 11 to 20 m M in a

DISCUSSION
In recent years dual wavelength fluorometry with indicators available as membrane-permeable esters has been widely employed for measuring [Ca2+li. Although this approach is suited for detecting variations of [Ca"];, it is a matter of discussion as to how accurate are the calculated concentrations. The methodological uncertainties can be expected to be more apparent in analyses of intact pancreatic islets than with single cells. It is unclear to what extent methodological complications can explain why the glucose-induced slow oscillations of [Ca2+Ji in islets occur from a plateau rather than from the basal level as in individual &cells or monolayer clusters (17). The geometry of the measurements may also be a factor of importance for the considerable variation of the amplitudes of the synchronized [Ca2+Ji oscillations in different islet regions. Whereas autofluorescence has been insignificant in our previous studies of [Caz+Ii in p-cells (2,6, 171, the contribution of this factor to the total fluorescence has been estimated to be 15-20% in studies of islets using indo-1 (7) or fura-2 (10) as indicator. Under the present excitation conditions autofluorescence was less significant (~1 5 % ) and did not change when altering the glucose concentration. The error introduced by not compensating for this autofluorescence was an underestimate of basal [Ca2+li by 3-5% and of elevated [Ca2+li by 6-14%. There is a slow leakage of fura-2 from the p-cell cytoplasm by extrusion via an anion transporter stimulated by glucose (21). It is therefore pertinent to consider whether Ca2+-saturated fura-2 retained within the extracellular space of the islet may contribute to the Ca2+ signal. This does not seem to be the case, since the glucose-induced increase of [Ca2+li was unaffected by blocking anion transport with sulfinpyrazone. Moreover, the present levels of [Ca2+li were not higher than those recorded in individual p-cells under conditions when extruded fura-2 is rapidly diluted and washed away by the superfusion medium (2, 6, 17).
Although reacting to 11 mM glucose with a rise of [Ca2+Ji, the islets differed considerably with regard to the response pattern. In most islets the rise of [Ca2+Ij was associated with a gradual appearance of oscillations, referred to as fast (2-7/min) or slow (0.3-0.9/min). In agreement with the observations of Valdeolmillos et al. (7) there were also islets responding with mixed oscillations. In this case we did not observe any obvious clustering of the superimposed fast oscillations to specific phases of the slow ones.
It has been convincingly demonstrated that the fast [Ca2+li oscillations in the intact islets reflect the electrical burst activity of the p-cells. This is indicated not only by the presence of such oscillations under a number of situations known to induce electrical activity (10) and their prolongation with an increase of the glucose concentration (7) but also by the observation of a perfect synchronization between the two phenomena (8). Like the electrical activity (  phenomenon with regularly varying periods at the plateau of depolarization has previously been observed in 20-50% of islets (22,23). Interestingly, the frequency of these periodic variations was similar to that of the slow oscillations of [Ca2+lj.
The dominating type of [Ca2+Ii oscillations in glucose-stimulated individual p-cells also reflects their electrical burst activity (24, 25), but the frequency is about 10-fold lower than recorded in islets and similar to the presently observed slow oscillations. The mechanism behind the faster bursts of action potentials in the islets remains t o be clarified. It has been shown that intracellular mobilization of Ca2+ by muscarinic activation can interrupt firing of action potentials by evoking transient hyperpolarization (26). In the physiological situation a feedback mechanism involving intracellular mobilization of Ca2+ by purinergic activation has been anticipated (271, since ATP is released from the secretory granules together with insulin. If a Ca2+-activated hyperpolarization mediated by activation of K+ channels is involved in the repolarization during burst activity in islets (26, 281, attention  It was evident from the present study that additions of both glucagon and dibutyryl CAMP to islets with slow [Ca2+Ij oscillations sometimes induce the rapid pattern and that spontaneous rapid oscillations are transformed into slow ones with the a2-adrenergic agonist clonidine. Indeed, epinephrine, which activates P-cell a2-adrenoceptors, has previously been found t o induce a strikingly parallel transformation of the rapid electrical burst pattern in mouse islets into a slow one (30). The CAMP-elevating effect of glucagon is well established, and concentrations of clonidine similar to the present ones cause pronounced depression of this cyclic nucleotide (31). It has also been argued that clonidine activates a low conductance G protein-dependent K+ channel in the p-cells (32). However, the relevance of this effect is questionable, since it was observed at 50-500-fold higher concentrations of the agonist. The results therefore indicate that CAMP is an important factor for the establishment of the rapid regular [Ca2+lj oscillations in the islets. It is likely that secretion of glucagon from the a2-cells is significant for the glucose induction of the fast [Ca2+lj oscillations in islets. In isolated p-cells deprived of adjacent a2-cells the dominating pattern in response to glucose is slow oscillations. Also in single p-cells glucagon induces rapid [Ca2+lj oscillations, which have a similar frequency but lack the striking regularity observed in islets (33).
We have previously proposed that the burst activity recorded in intact pancreatic islets may be an artifact from the physiological point of view due to insufficient exchange of the extracellular medium in islets lacking a capillary circulation (17). Autocrine and paracrine effects can be expected to be accentuated in this situation. It is therefore pertinent to note that comparative studies of hormone secretion from the pancreas perfused in the anterograde and retrograde directions have indicated that the p-cells are perfused before the an-and &cells (34). The importance of paracrine effects of the other islet hormones in the regulation of the p-cell activity can consequently be questioned.
In about 50% of the islets exposure to 11 m M glucose resulted in the appearance of slow [Ca2+li cycles with a frequency of 0.3-0.9/min. These cycles were synchronized in different parts of the islets but differed from the slightly slower large amplitude oscillations of individual p-cells in being superimposed upon an elevated plateau rather than occurring from the basal level (2, 6, 17). Nevertheless, it seems probable that the slow

Ca2+ Oscillations and
Insulin Release 8753 cycles in the islets represent a coordination of the large amplitude oscillations in individual B-cells, since functional coupling of the 0-cells somewhat increases the oscillatory rate (17). As in individual 0-cells (2, 6) we observed transformation of the glucose-induced slow oscillations of [Ca2+Ii into a sustained increase in islets exposed to high glucose concentrations. The question arises as to whether the oscillations of [Ca2+Ii can account for the pulsatile release of insulin. So far the arguments in favor of this concept have relied on the established role of [Ca2+li in stimulating insulin secretion and the fact that the frequencies of the slow 0-cell oscillations of [Ca2+li are similar to those reported for insulin (5). We have now obtained direct evidence for such a relationship by demonstrating that insulin release from the glucose-stimulated single islet occurs in synchrony with the slow [Ca2+li oscillations. However, such a synchrony should not be taken as indicating that only 50% of the islets, in which slow [Ca2+li oscillations were detected, have a periodic release. Actually, all islets studied hitherto in our laboratory (2100) have been found to respond to 11 m M glucose with similar insulin pulses. It may seem puzzling that insulin is released in pulses also from islets lacking slow oscillations of [Caz+li. However, there are several factors that could prevent the detection of the slow [Ca2+li oscillations. In addition to limitations in the technique for measuring [Ca2+li the studies of the intact islets are complicated by regional differences in the responses and the presence of non-j3-cells. Moreover, the [Ca2+Ii recorded represents an average for the cytoplasm rather than reflecting the concentration in a submembrane space directly involved in the regulation of the secretory activity. There are reasons for believing that glucose stimulation of insulin release is mediated not only by a rise of [Ca2+1, in the 0-cells but also by an increased sensitivity of the secretory machinery to the Ca2+ signal (1). It will be a matter for further studies to decide whether such a sensitization explains why raising the glucose concentration from 11 to 20 mM increases the amplitudes of the insulin oscillations while damping those of [Ca2+Ii.