The endoplasmic reticulum is a glucose-modulated high-affinity sink for Ca²⁺ in mouse pancreatic β-cells

Abstract

1. The regulation of organelle free Ca²⁺ was analysed in individual mouse pancreatic β-cells loaded with the fluorescent low-affinity indicator furaptra.

2. Removal of the cytoplasmic indicator by controlled digitonin permeabilization of the plasma membrane resulted in a sudden increase of the 340 nm/380 nm fluorescence excitation ratio followed by a gradual decay, reflecting the emptying of Ca²⁺ from organelle pools.

3. Subsequent introduction of 3 mM ATP caused rapid refilling of a Ca²⁺ pool, which represented the endoplasmic reticulum (ER) in being mobilized with inositol 1,4,5-trisphosphate (IP₃) and the sarco(endo)plasmic reticulum Ca²⁺-ATPase inhibitor thapsigargin.

4. The concentration of Ca²⁺ in the ER observed immediately after permeabilization depended on the glucose concentration in a hyperbolic fashion with half-maximal filling at about 6 mM of the sugar.

5. Glucose promotion of Ca²⁺ sequestration in the ER involved a high-affinity mechanism not requiring but accelerated by a rise of the cytoplasmic Ca²⁺ concentration.

6. Glucose also exerted a long-term action on the ER storage of Ca²⁺, maintaining the set-point for its maximal concentration and preserving the response to IP₃.

7. The results indicate that the ER has an important role in the glucose-stimulated β-cell by serving as a high-affinity sink for Ca²⁺, irrespective of the prevailing concentration of cytoplasmic Ca²⁺.

Glucose is the major natural stimulator of insulin release from the pancreatic β-cell. Metabolism of the sugar induces closure of ATP-regulated K⁺ channels in the plasma membrane, resulting in depolarization with elevation of the cytoplasmic Ca²⁺ concentration ([Ca²⁺]_i) and activation of exocytosis (Wollheim & Sharp, 1981; Hellman & Gylfe, 1986a; Ashcroft & Rorsman, 1989; Hellman et al. 1992). Although these events at the plasma membrane are the most important determinants for insulin secretion, there is evidence that intracellular sequestration and release of Ca²⁺ can also modulate β-cell function (Worley et al. 1994; Bertram et al. 1995; Liu et al. 1998; Gilon et al. 1999). Early studies of ⁴⁵Ca fluxes indicated that glucose, in addition to promoting voltage-dependent Ca²⁺ entry, stimulates the sequestration of the ion in inositol 1,4,5trisphosphate (IP₃)-sensitive stores (Hellman et al. 1986). The store filling enables the β-cells to respond to muscarinic (Hellman & Gylfe, 1986b) and purinergic (Gylfe & Hellman, 1987) stimuli with mobilization of intracellular Ca²⁺.

By measuring organelle free Ca²⁺ concentration with the fluorescent low-affinity indicator furaptra, we recently demonstrated that glucose stimulates the uptake of Ca²⁺ in the endoplasmic reticulum (ER) of the pancreatic β-cell (Tengholm et al. 1999). The experiments were performed in the presence of the hyperpolarizing sulphonamide diazoxide, indicating that elevation of [Ca²⁺]_i is not required for the action of the sugar. In contrast to this conclusion, studies of clonal insulin-releasing INS-1 cells indicated that an increase of [Ca²⁺]_i is the major determinant and ATP a permissive factor for glucose-stimulated Ca²⁺ sequestration in the ER (Maechler et al. 1999). The proposed function of the ER as a passive sink for Ca²⁺ became the basis for a model explaining the generation of the electrophysiological burst pattern in glucose-stimulated β-cells (Gilon et al. 1999). In the present study, we have extended the direct measurement of ER free Ca²⁺ concentration in individual pancreatic β-cells to clarify the role of [Ca²⁺]_i in the effect of glucose. We show that the glucose-stimulated uptake of Ca²⁺ in the ER is a high-affinity process, not requiring but accelerated by an elevation of [Ca²⁺]_i. Moreover, we provide evidence that glucose exerts a long-term action on the ER storage of Ca²⁺, maintaining the set-point for its maximal concentration and preserving the mobilization in response to IP₃.

METHODS

Materials

Reagents of analytical grade and deionized water were used. The acetoxymethyl ester form of the Ca²⁺ indicator furaptra, thapsigargin and IP₃ were purchased from Molecular Probes (Eugene, OR, USA). Collagenase, Hepes and ATP were from Boehringer Mannheim (Mannheim, Germany) and digitonin was from Calbiochem (San Diego, CA, USA). The Ca²⁺ chelator EGTA was obtained from Sigma Chemical Co. Diazoxide and tolbutamide were kind gifts from Schering (Kenilworth, NJ, USA) and Hoechst Marion Roussel (Frankfurt/Main, Germany), respectively. Unless otherwise stated, intact cells were exposed to a medium containing (mm): NaCl 125, KCl 5.9, MgCl₂ 1.2, CaCl₂ 1.3 and Hepes 25 with pH adjusted to 7.40 with NaOH. Permeabilized cells were superfused with an ‘intracellular’ medium containing (mm): KCl 140, Na₂ATP 0 or 3 and Hepes 10 with pH adjusted to 7.00 with KOH. Free Mg²⁺ was maintained at 0.1 mm by adding appropriate amounts of MgCl₂ depending on the ATP concentration and free Ca²⁺ was buffered to 50 nm or 1 μm with 2 mm EGTA. The ion concentrations were calculated using the Maxchelator program (Bers et al. 1994).

Preparation of pancreatic β-cells

Islets of Langerhans were isolated from the pancreas of adult ob/ob mice taken from a non-inbred colony (Hellman, 1965). The experimental procedures were approved by the Uppsala Animal Ethics Committee. The animals were placed in a sealed container into which a stream of CO₂ was delivered. When the animals became unconscious they were killed by decapitation. The peritoneal cavity was opened and the pancreas was excised and cut into small pieces, which were digested with collagenase to obtain free islets of Langerhans. Single cells were then prepared by shaking the islets in a Ca²⁺-deficient medium (Lernmark, 1974). After suspension in RPMI 1640 medium containing 11 mm glucose, 10 % fetal calf serum, 100 i.u. ml⁻¹ penicillin, 100 μg ml⁻¹ streptomycin and 30 μg ml⁻¹ gentamicin, the cells were allowed to attach to circular coverslips during culture for 1-6 days at 37 °C in an atmosphere of 5 % CO₂ in humidified air. The ob/ob mouse islets contain more than 90 %β-cells (Hellman, 1965), known to respond normally to glucose and other stimulators of insulin release (Hahn et al. 1974). Non-β-cells were avoided by selecting cells of a large size and low nuclear/cytoplasmic ratio for analyses (Berts et al. 1995).

Loading with Ca²⁺ indicator, preincubation and permeabilization

The experimental protocols used for indicator loading and preincubation are shown in Table 1. After isolation and initial culture in RPMI 1640 medium for 0-4 days, the cells were exposed to different glucose concentrations or other substances. For studies of the long-term regulation of intracellular Ca²⁺ sequestration, the cells were cultured for 2 h to 6 days in the presence of different test agents. They were then loaded with 4 μm furaptra acetoxymethyl ester in a medium containing 3 mm glucose during 60 min incubation at 37 °C. When evaluating more acute effects, the cells were exposed for 5-60 min to the test agents after loading with furaptra.

Table 1.

Experimental protocols used for loading β-cells with Ca²⁺ indicator and exposure to test substances

Protocol	Initial culture →	Furaptra loading →	Exposure to test substances →	Furaptra loading
I	1–4 days	yes	5–60 min	no
II	1–4 days	no	2–18 h	yes
III	0–1 days	no	3–6 days	yes

After isolation and an initial culture period in RPMI 1640 medium containing 11 mm glucose the cells were exposed to different glucose concentrations or other substances. For the shortest incubation times (protocol I), the test substances were added to the extracellular medium after 60 min loading with the Ca²⁺ indicator furaptra in the presence of 3 mm glucose. For more prolonged exposure times (protocols II and III), glucose and other test substances were present in RPMI 1640 medium before furaptra loading in the presence of 3 mm glucose.

The coverslips with the attached cells were used as exchangeable bottoms of an open superfusion chamber thermostatically regulated at 37 °C. The plasma membrane was permeabilized in an ATP-free intracellular medium supplemented with 4 μm digitonin during simultaneous measurements of the fluorescence obtained by excitation at 340 and 380 nm (Tengholm et al. 1998). The detergent was immediately removed after the sudden drop in fluorescence, caused by the loss of cytoplasmic furaptra. In most experiments the Ca²⁺ concentration of the permeabilization medium was 50 nm. When testing the effect of depolarization on Ca²⁺ filling of organelles, the permeabilization medium contained 1 μm Ca²⁺ to counteract the loss of Ca²⁺. In the latter case, the Ca²⁺ concentration was lowered to 50 nm immediately after permeabilization (Tengholm et al. 1999). Being below the detection limit for the indicator, the Ca²⁺ concentration of the media did not affect the furaptra fluorescence during permeabilization.

Measurements of Ca²⁺ in intracellular stores

Organelle free Ca²⁺ concentration was measured with a dual wavelength microfluorometric system (Deltascan, Photon Technology International Inc., Princeton, NJ, USA). The excitation light was alternately directed to two monochromators by a chopper mirror spinning at 50 Hz. The monochromator outputs were connected via a bifurcated optical fibre to the epifluorescence attachment of an inverted microscope (Nikon Diaphot) equipped with a x100 objective (NA 1.3). Fluorescence was recorded at 535 nm with a photomultiplier using a 25 nm half-bandwidth interference filter. The background-subtracted signals, obtained by excitation at 340 and 380 nm (1 nm half-bandwidth), were recorded at 2 Hz using FeliX software (Photon Technology International Inc.). We have previously demonstrated that the 340 nm/380 nm fluorescence excitation ratio of furaptra provides information about the organelle Ca²⁺ concentration in permeabilized β-cells without interference from changes in Mg²⁺ concentration (Tengholm et al. 1998). In the present study, free Ca²⁺ concentration was calculated from the 340 nm/380 nm fluorescence excitation ratio according to Grynkiewicz et al. (1985) using the equation:

F₀ and R_min are the furaptra fluorescence excited at 380 nm and the 340 nm/380 nm fluorescence excitation ratio, respectively, in a Ca²⁺-deficient intracellular-like medium. F_S and R_max are the corresponding values obtained in the presence of a saturating concentration of Ca²⁺. K_D, the dissociation constant for furaptra binding of Ca²⁺, was assumed to be 53 μm (Raju et al. 1989). Linearization of the data by conversion from the 340 nm/380 nm fluorescence excitation ratio to the calibrated Ca²⁺ concentration had only marginal effects on the results. Accordingly, the dose-response relationship shown in Fig. 1C was almost identical irrespective of whether it was based on fluorescence ratios or calibrated Ca²⁺ concentration values. With the presently used calibration approach the range of maximal Ca²⁺ concentrations was 100-500 μm, comparing favourably with the 200-500 μm estimated by in situ titration (Tengholm et al. 1998, Tengholm 1999). The latter studies also demonstrated that furaptra is not saturated when the organelles are maximally filled with Ca²⁺.

Figure 1. Effect of glucose on organelle Ca²⁺ concentration in individual pancreatic β-cells.

After 60 min loading with furaptra in the presence of 3 mm glucose, the cells were incubated for 30 min in an indicator-free medium containing 3 (A) or 11 mm(B) glucose (Table 1, protocol I). After rinsing in similar medium containing only 50 nm Ca²⁺, the cells were exposed to 4 μm digitonin in an ATP-free intracellular medium with maintenance of the Ca²⁺ and glucose concentrations. Digitonin was immediately withdrawn upon permeabilization (arrows). ATP (3 mm), IP₃ (10 μm) and thapsigargin (Th; 200 nm) were subsequently introduced as indicated. The filling of the organelle pool immediately after permeabilization (ini), expressed as a percentage of the filling reached after the addition of ATP (max), is plotted as a function of the glucose concentration in C. Each point is the mean of 4-5 experiments ±s.e.m. and the continuous line shows a fit of the 23 individual data points to a hyperbolic function plus a constant representing glucose-independent uptake (r = 0.93; P < 0.001).

Presentation of data

The relative Ca²⁺ filling of the organelle pool in intact β-cells was estimated by comparing the Ca²⁺ concentration recorded immediately upon permeabilization in an ATP-free medium with that obtained after subsequent addition of ATP. Since glucose was found to have a long-term effect on the maximal filling of the ER, the figures presented may not be entirely representative. However, it is likely that this is a minor problem, since the effect of glucose on the maximal filling of the ER did not reach significance within 30 min incubation. Results are expressed as mean values ±s.e.m. Differences were statistically evaluated by Student's two-tailed t test. All illustrations were made with Igor Pro software (Wavemetrics Inc., Lake Oswego, OR, USA).

RESULTS

When furaptra-loaded cells were incubated for 30 min in an indicator-free medium containing 0-20 mm glucose, subsequent permeabilization in an ATP-free medium containing 50 nm Ca²⁺ resulted in a sudden increase of the 340 nm/380 nm fluorescence excitation ratio followed by a gradual decay reflecting emptying of Ca²⁺ from organelle stores (Fig. 1). Both the magnitude of the initial elevation and the decay were considerably smaller when the cells had been incubated in 3 (Fig. 1A) compared to 11 mm (Fig. 1B) glucose. The emptied organelles were rapidly refilled after addition of a saturating concentration of ATP (3 mm). Most of the accumulated Ca²⁺ could be mobilized by the sarco(endo)plasmic reticulum Ca²⁺-ATPase (SERCA) inhibitor thapsigargin, and 67 ± 3 %(n = 23) of this Ca²⁺ was sensitive to IP₃. Pre-exposure to thapsigargin prevented both the initial rise upon permeabilization and the effect of ATP (data not shown). Figure 1C illustrates how pre-exposure to glucose affects the initial organelle filling, expressed as a percentage of that reached after subsequent addition of ATP. After subtraction of the glucose-independent filling, the data fitted a hyperbolic function with a half-maximal effect at 6.3 mm of the sugar (Fig. 1C). This value is similar to the previously reported 5.5 mm obtained using the hyperpolarizing sulphonamide diazoxide to prevent the glucose-induced elevation of [Ca²⁺]_i (Tengholm et al. 1999).

The glucose stimulation of organelle Ca²⁺ uptake was apparently due to a high-affinity mechanism, since there was also an efficient accumulation during 30 min pre-exposure to 20 mm glucose in a medium deprived of Ca²⁺ and supplemented with both 2 mm EGTA and 400 μm diazoxide (Fig. 2A). Under such conditions, the glucose-induced Ca²⁺ filling corresponded to 79 ± 10 %(n = 7) of that reached with 3 mm ATP. However, elevation of [Ca²⁺]_i also stimulated the organelle sequestration of Ca²⁺. The effect of 3 mm glucose alone (Fig. 2A Fig. 1A) was markedly amplified after elevation of [Ca²⁺]_i by depolarization with 30.9 mm K⁺ plus 400 μm diazoxide (data not shown) or with 100 μm tolbutamide (Fig.2B). The organelle filling observed under these conditions was 64 ± 11 %(n = 6) after K⁺ and 75 ± 5 %(n = 8) after tolbutamide depolarization. Despite this stimulatory effect of depolarization, the dose-response relationships for the glucose-induced Ca²⁺ sequestration in the ER were almost identical when [Ca²⁺]_i was allowed to increase in response to glucose (Fig. 1C) or was clamped at a resting level (Tengholm et al. 1999). We therefore studied whether [Ca²⁺]_i affects the rate of organelle Ca²⁺ filling in glucose-stimulated β-cells. This was accomplished by comparing the effect of 20 mm glucose on organelle Ca²⁺ filling during 5-60 min incubation in the absence and presence of hyperpolarizing diazoxide. Although similar steady-state levels were reached after 30 min, the sequestration was markedly faster when [Ca²⁺]_i was allowed to increase (Fig. 3). At 15 min the filling, expressed as a percentage of that obtained after addition of ATP, corresponded to 91 ± 5 %(n = 7) in the absence and 52 ± 7 %(n = 13;P < 0.001) in the presence of diazoxide.

Figure 2. Influence of cytoplasmic Ca²⁺ on the effect of glucose on organelle Ca²⁺ concentration in individual pancreatic β-cells.

After 60 min loading with furaptra in the presence of 3 mm glucose, the cells were incubated for 30 min in an indicator-free medium deficient in Ca²⁺ and containing 20 mm glucose, 2 mm EGTA and 400 μm diazoxide (A; Table 1, protocol I) or in a medium containing 3 mm glucose, 1.3 mm Ca²⁺ and 100 μm tolbutamide (B). After rinsing in similar medium containing 50 nm(A) or 1 μm(B) Ca²⁺, the cells were exposed to 4 μm digitonin in an ATP-free intracellular medium with maintenance of the concentrations of Ca²⁺, glucose and tolbutamide. Immediately upon permeabilization the medium was changed to one lacking digitonin and containing 50 nm Ca²⁺ (arrows). ATP (3 mm), IP₃ (10 μm) and thapsigargin (Th; 200 nm) were subsequently introduced as indicated. Representative of 7 (A) and 6 (B) experiments.

Figure 3. Time dependence of the effect of glucose on organelle Ca²⁺ concentration in individual pancreatic β-cells.

After 60 min loading with furaptra the cells were incubated in the absence or presence of 400 μm diazoxide for 0-60 min in an indicator-free medium containing 20 mm glucose (Table 1, protocol I). After rinsing in similar medium containing 50 nm Ca²⁺, the cells were exposed to 4 μm digitonin in an ATP-free intracellular medium with maintenance of the Ca²⁺ and glucose concentrations. Digitonin was immediately withdrawn upon permeabilization (arrows) and ATP (3 mm) was subsequently introduced as indicated. The upper panels show experiments with cells incubated for 15 min in the absence (A) and presence (B) of diazoxide. C shows the filling of the organelle Ca²⁺ pool immediately upon permeabilization in cells incubated without (○) or with (□) diazoxide. The data are expressed as a percentage of the filling reached after subsequent addition of 3 mm ATP, and plotted as a function of the incubation time. Data are mean values ±s.e.m. for 5-13 experiments.

Pre-exposure to glucose also affected the maximal filling of the organelle Ca²⁺ stores reached in response to subsequent addition of 3 mm ATP (Fig. 4A and B). A higher organelle Ca²⁺ concentration was obtained in β-cells pre-exposed to 20 as compared to 3 mm glucose. This difference was detectable after 1 h and became more pronounced with prolongation of the exposure time (Fig. 4C and D). It is not entirely clear whether the increasing difference in organelle Ca²⁺ concentration was due to a reduction of the maximal ER Ca²⁺ concentration in the cells exposed to 3 mm glucose or an increase in those exposed to 20 mm of the sugar, since different experimental protocols were used for the different exposure times and variations over time were observed at both glucose concentrations. Although a tolbutamide-induced rise of [Ca²⁺]_i stimulated the organelle filling (see above), 100 μm of this sulphonylurea did not mimic the action of a high glucose concentration in maintaining the set-point for the maximal organelle Ca²⁺ concentration in 1-2 h experiments (data not shown).

Figure 4. Glucose priming of ATP-dependent Ca²⁺ uptake in individual pancreatic β-cells.

Short-term effects of glucose were evaluated after 15, 30 or 60 min incubation in indicator-free medium containing 3 or 20 mm glucose following loading with furaptra in the presence of 3 mm glucose (Table 1, protocol I). For longer exposure, the sugar was added to the culture medium prior to loading with indicator in the presence of 3 mm glucose (Table 1, protocols II and III). The cells were then exposed to 4 μm digitonin in an ATP-free intracellular medium containing 50 nm Ca²⁺. Digitonin was immediately withdrawn upon permeabilization (arrows) and ATP (3 mm) was subsequently introduced as indicated. A and B show experiments with cells incubated for 2 h with 3 or 20 mm glucose, respectively. C shows the steady-state concentrations of organelle Ca²⁺ reached after exposure to ATP in cells incubated for different periods of time in medium containing 3 (□) or 20 mm (○) glucose. D shows the difference in steady-state concentration of organelle Ca²⁺ between cells exposed to 3 and 20 mm glucose (d in A and B) at different incubation times. Data are mean values ±s.e.m. for comparisons based on a separate series of 6-13 experiments at each time point. Asterisks denote statistical significance for the effect of glucose: * P < 0.05; ** P < 0.01; *** P < 0.001.

The IP₃ sensitivity of the Ca²⁺ stores was investigated after 3-6 days of culture in different concentrations of glucose. Figure 5A shows the design of the experiments with exposure to 0.5-10 μm IP₃ before a complete emptying of the Ca²⁺ stores with thapsigargin. The dose-response relationship for the IP₃-induced reduction of organelle Ca²⁺ concentration is illustrated in Fig. 5B. Irrespective of the culture conditions there were no differences in the IP₃ sensitivity, which was half-maximal and maximal at about 1 and 10 μm of the messenger, respectively. However, the magnitude of the IP₃ response was much lower after culture in 3 than in 11 (Fig. 5B) or 20 mm (not shown) glucose, whereas there was no difference between cells cultured in 11 and 20 mm of the sugar (not shown). Also, when expressed relative to the release obtained with thapsigargin, 10 μm IP₃ mobilized less Ca²⁺ in cells cultured in 3 (58 ± 4 %;P < 0.001; n = 11) than in 11 (78 ± 1 %; n = 8) or 20 mm (79 ± 2 %; n = 9) glucose.

Figure 5. Effects of IP₃ on organelle Ca²⁺ concentration in individual pancreatic β-cells.

The cells were cultured for 3-6 days in RPMI 1640 medium containing 3 or 11 mm glucose prior to loading with furaptra in the presence of 3 mm glucose (Table 1, protocol III). The cells were then exposed to 4 μm digitonin in an ATP-free intracellular medium. Digitonin was withdrawn and ATP (3 mm) added immediately upon permeabilization. A shows the design of an experiment with additions of IP₃ (0.5-10 μm) and thapsigargin (Th; 200 nm) to a cell cultured in 11 mm glucose. B shows the dose-response relationship for the IP₃-induced changes of organelle Ca²⁺ concentration in cells cultured with 3 (□) or 11 mm (○) glucose. The trace in A is representative of 4 experiments and the data points in B are mean values ±s.e.m. for 4-11 experiments.

DISCUSSION

The organelle handling of Ca²⁺ can be studied in intact cells by targeted expression of Ca²⁺-sensitive luminescent (Brini et al. 1999) or fluorescent proteins (Miyawaki et al. 1997). This technique is essentially limited to clonal cell lines, which often lack the specific features of differentiated cells. Organelle Ca²⁺ concentration can be measured in normal cells loaded with low-affinity non-protein Ca²⁺ indicators, like furaptra, provided that cytoplasmic indicator is removed by permeabilization of the plasma membrane (Hofer & Machen, 1993; Hajnóczky & Thomas, 1997; Mogami et al. 1998). In pancreatic β-cells, most Ca²⁺ monitored by furaptra is sensitive to IP₃ and almost the entire pool is released by thapsigargin, indicating that the signal is representative for the ER (Tengholm et al. 1998, Tengholm 1999, Tengholm 2000). The Ca²⁺ pool measured with furaptra will therefore subsequently be referred to as the ER.

The ER has a central role in intracellular Ca²⁺ homeostasis. It was recently reported that glucose stimulates Ca²⁺ uptake into the ER of insulin-secreting INS-1 cells expressing ER-targeted aequorin, and that the effect of the sugar is mediated by the elevation of cytoplasmic Ca²⁺ concentration rather than by ATP (Maechler et al. 1999). By studying the Ca²⁺ concentration detected with furaptra at the time of permeabilization, elevation of [Ca²⁺]_i has also been found to promote ER sequestration of the ion in normal β-cells (Tengholm et al. 1999). However, ATP appeared to be more important as a regulatory factor, since glucose stimulation resulted in maximal Ca²⁺ filling in the ER even when [Ca²⁺]_i was kept at resting levels by hyperpolarization with diazoxide. We now provide evidence that the glucose action is mediated by a high-affinity mechanism, the stimulation being maintained after omission of extracellular Ca²⁺. Another important observation is that the dose-response relationship for the glucose-induced accumulation of Ca²⁺ in the ER is almost identical when [Ca²⁺]_i is clamped at a resting concentration with diazoxide (Tengholm et al. 1999) or allowed to increase in response to glucose (Fig. 1C). Although these data indicate that the ER fills to the same extent irrespective of glucose-induced elevation of [Ca²⁺]_i, such elevation was found to considerably accelerate the filling. It has been reported that excessive diazoxide suppresses mitochondrial metabolism in the β-cell (Hellman et al. 1974; Grimmsmann & Rustenbeck, 1998). It is therefore important to note that the concentration used in the present study does not affect glucose oxidation in the mouse islets (Bergsten & Hellman, 1987). The 15-30 min period required for glucose-stimulated Ca²⁺ filling of the ER is in good agreement with observations on clonal insulin-releasing RINm5F cells (Gylfe & Hellman, 1986).

The Ca²⁺ filling of the ER is a determinant for store-operated cation currents in the plasma membrane proposed to be involved in the regulation of the membrane potential of β-cells (Worley et al. 1994; Bertram et al. 1995). Glucose-stimulated β-cells within pancreatic islets exhibit an oscillatory electrical activity with action potentials grouped in bursts separated by electrically silent and repolarized intervals. In a discussion of the mechanisms for this electrophysiological burst pattern, it was recently proposed that the ER acts as a passive buffer for Ca²⁺ (Gilon et al. 1999). When [Ca²⁺]_i is high during depolarization, the ER becomes filled, shutting off the store-operated current that contributes to the depolarization. The resulting hyperpolarization is associated with a lowering of [Ca²⁺]_i and a gradual leak of Ca²⁺ from the ER, until the store-operated current activates to depolarize the cell and start a new cycle. In demonstrating a high-affinity accumulation of Ca²⁺ in the ER the present data question this model, since it is unlikely that Ca²⁺ is passively lost from the ER during glucose stimulation. However, our results do not contradict a role for intracellular Ca²⁺ stores in regulating the burst pattern when Ca²⁺ is mobilized by active mechanisms. Besides activating a store-operated current (Worley et al. 1994; Bertram et al. 1995), the [Ca²⁺]_i transients, resulting from IP₃-induced mobilization of ER Ca²⁺, have been proposed to hyperpolarize the β-cell during bursting (ÄmmÄLÄ C, Larssonet al. 1991; Liu et al. 1998; Dryselius et al. 1999).

Apart from increasing the Ca²⁺ uptake into the ER, glucose acted to maintain the set-point for the maximal filling obtained when permeabilized β-cells were exposed to 3 mm ATP. Since it was not mimicked by sulphonylurea-induced elevation of [Ca²⁺]_i, it is likely that the effect of glucose on the maximal Ca²⁺ filling of the ER is determined by factors other than the cytoplasmic concentration of the ion. In various types of cell glucose starvation, Ca²⁺ depletion and other types of ER stress can induce the expression of ER chaperones (Chapman et al. 1998; Pahl, 1999). Some of these chaperones are identical to the low-affinity, high-capacity Ca²⁺ buffers in the ER lumen (Llewellyn et al. 1996; Lièevremontet al. 1997; Waser et al. 1997; Lucero et al. 1998). Glucose-dependent alterations in the expression of these proteins may therefore affect the steady-state concentration of Ca²⁺ in the ER. Whether such a mechanism underlies the effect of glucose on the maximal filling of the ER remains to be established.

The pancreatic β-cell expresses at least three isoforms of the IP₃ receptor (De Smedt H, Missiaenet al. 1994; Lee et al. 1999) with different functional properties (Patel et al. 1999). Long-term exposure to 20 mm glucose has been found to alter the expression of the IP₃ receptors in insulin-releasing cells and pancreatic islets (Lee et al. 1998, Lee 1999). The differences observed here in the effect of IP₃ on ER Ca²⁺ concentration after culture in 3 compared to 11 or 20 mm glucose may well involve changes in IP₃ receptor expression.

We now provide evidence that glucose promotes high-affinity Ca²⁺ sequestration in the ER of β-cells. The accumulation is hyperbolically related to the glucose concentration with half-maximal and maximal effects at about 6 and > 20 mm, respectively. The dose-response relationship resembles that for the initial lowering of [Ca²⁺]_i below the resting level seen when raising the glucose concentration (Gylfe, 1988), a phenomenon requiring functional SERCA pumps (Chow et al. 1995). Also, the ATP/ADP ratio, which may determine the activity of the SERCA pumps (see Corkey et al. 1988), exhibits a similar dependence on the glucose concentration (Detimary et al. 1995). Taken together, the data indicate that the ER has an important role in the glucose-stimulated β-cell by serving as a high-affinity sink for Ca²⁺ irrespective of the prevailing [Ca²⁺]_i.