Store-operated influx of Ca²⁺ in pancreatic β-cells exhibits graded dependence on the filling of the endoplasmic reticulum

A rise of the cytoplasmic Ca²⁺ concentration ([Ca²⁺]_i) is the key trigger of insulin secretion from pancreatic β-cells exposed to glucose and other nutrient stimuli (Wollheim and Sharp, 1981; Hellman and Gylfe, 1986). The signal transduction involves metabolism of the stimulus causing a rise of the ATP/ADP ratio and subsequent closure of the ATP-sensitive K⁺ (K_ATP) channels, resulting in depolarization and influx of Ca²⁺ through voltage-dependent channels (Ashcroft and Rorsman, 1989). However, there is increasing evidence that intracellular sequestration and release of Ca²⁺ are also important in the regulation of insulin secretion. Glucose is consequently a potent stimulus for Ca²⁺ accumulation in the endoplasmic reticulum (ER) of pancreatic (Hellman et al., 1986; Gylfe, 1991; Tengholm et al., 1999) as well as clonal β-cells (Gylfe and Hellman, 1986; Maechler et al., 1999). Moreover, depolarization during glucose stimulation triggers formation of inositol 1,4,5 trisphosphate (Ins(1,4,5)P₃) (Roe et al., 1993; Liu et al., 1996), which seems to be the major messenger for Ca²⁺ mobilization from the β-cell ER (Tengholm et al., 1998). Apart from affecting [Ca²⁺]_i directly, the intracellular release has indirect actions resulting from changes in membrane potential. The Ins(1,4,5)P₃-induced rise of [Ca²⁺]_i can activate a hyperpolarizing K⁺ current shutting off the voltage-dependent entry of the cation (Ämmälä et al., 1991; Liu et al., 1998; Dryselius et al., 1999). The associated emptying of the ER has the opposite effect, activating a depolarizing store-operated current carried by Ca²⁺ or Na⁺ (Worley et al., 1994; Bertram et al., 1995; Liu and Gylfe, 1997; Gilon et al., 1999).

The role of the store-operated current in the physiology of the glucose-stimulated β-cell ultimately depends on how it is regulated by store filling. In some types of cells Ca²⁺ influx is activated in an all-or-none fashion after almost complete emptying of the intracellular Ca²⁺ stores (Fierro and Parekh, 2000; Fierro et al., 2000), whereas in others there is gradual activation with increasing depletion of the stores (Hofer et al., 1998; Sedova et al., 2000). The present study provides the first evidence that the store-operated entry of Ca²⁺ into the β-cell exhibits a graded dependence on Ca²⁺ filling of the ER. Small variations in the ER Ca²⁺ concentration may consequently contribute to the regulation of the membrane potential and [Ca²⁺]_i determining insulin release.

Reagents of analytical grade and deionized water were used. Fura-2 and its acetoxymethyl ester were from Molecular Probes Inc. (Eugene, OR). EGTA and carbachol were provided by Sigma (St Louis, MO), 2-aminoethoxydiphenyl borate (2-APB) by Aldrich (Gillingham, UK), cyclopiazonic acid (CPA) by Calbiochem (La Jolla, CA). Collagenase and HEPES were bought from Boehringer Mannheim GmbH (Mannheim, Germany). Schering (Kenilworth, NJ) and Knoll AG (Ludwigshafen, Germany) kindly donated diazoxide and methoxyverapamil, respectively. Staphylococcus aureus α-toxin was a product of BioSys Inova (Stockholm, Sweden).

Islets of Langerhans were isolated by collagenase digestion from the pancreas of adult ob/ob mice from a local colony (Hellman, 1965). These islets consist of more than 90% β-cells, which respond normally to glucose and other regulators of insulin release (Hahn et al., 1974). Free cells were prepared by shaking the islets in a Ca²⁺-deficient medium (Lernmark, 1974). The cells were suspended in RPMI 1640 medium containing 11 mM glucose and supplemented with 10% fetal calf serum, 100 IU/ml penicillin, 100 μg/ml streptomycin and 30 μg/ml gentamicin and allowed to attach to circular 25 mm coverslips during culture for 4-7 days in an atmosphere of 5% CO₂ in humidified air. Further experimental handling of cells was performed with a medium containing 25 mM HEPES (pH 7.40), 1 mg/ml bovine serum albumin, 137 mM Na⁺, 5.9 mM K⁺, 1.2 mM Mg²⁺, and <1 nM, 1.28 or 10 mM Ca²⁺ with Cl⁻ as the sole anion. The lowest Ca²⁺ concentration was obtained by including 2 mM EGTA in a Ca²⁺-deficient medium. When testing the effects of Gd³⁺, bovine serum albumin and EGTA were omitted.

In most experiments, loading of the cells with the fluorescent indicator fura-2 was performed in the presence of 1.28 mM Ca²⁺ during a 40 minute incubation at 37°C in a medium supplemented with 1 μM fura-2 acetoxymethyl ester, 400 μM diazoxide and 20 mM glucose. However, when testing the effect of K⁺ depolarization, fura-2 loading was made in medium lacking diazoxide and containing 3 mM glucose. With these procedures 90±0.5% (n=4) of the fura-2 is cytoplasmic as judged from the release of indicator in response to plasma membrane permeabilization using a previously described technique (Tengholm et al., 2000) with 1250 hemolytic units/ml α-toxin. Calculations of [Ca²⁺]_i and [Mn²⁺]_i (see below) were compensated for this compartmentalization of fura-2. The coverslips with attached cells were used as exchangeable bottoms of an open chamber containing 50 μl medium. The chamber was placed on the stage of an inverted microscope (Nikon Diaphot) within a climate box maintained at 37°C by an air stream incubator, and the cells were superfused at a rate of 0.3 ml/minute with similar indicator-free medium. When studying store-operated Ca²⁺ influx this medium was supplemented with 50 μM methoxyverapamil.

The microscope was equipped with an epifluorescence illuminator and a 100× UV fluorite objective. A filter changer of a time-sharing multichannel spectrophotofluorometer (Chance et al., 1975) provided excitation light flashes of 1 millisecond duration every 10 milliseconds at 340 and 380 nm, and the emission was measured at 510 nm with a photomultiplier. A computer recorded the electronically separated fluorescence signals at the two wavelengths.

[Ca²⁺]_i values were obtained according to a previously described method (Grynkiewicz et al., 1985) using Equation 1:

K_DCa²⁺ is 224 nM. F₀ and R_min are the fura-2 fluorescence at 380 nm and the 340/380 nm fluorescence excitation ratio, respectively, in an ‘intracellular’ K⁺-rich calibration solution lacking Ca²⁺. F_S and R_max are the corresponding data obtained with saturating concentrations of Ca²⁺.

Variations in the influx through the store-operated pathway were estimated more directly by a Mn²⁺ quench approach. However, instead of measuring only the reduction in fura-2 fluorescence in β-cells exposed to this cation (Liu and Gylfe, 1997) we introduced a novel approach linearizing the data by calculating the cytoplasmic Mn²⁺ concentration ([Mn²⁺]_i). Because Mn²⁺ quenches the fluorescence of fura-2, irrespective of excitation wavelength, a single wavelength technique is used. To make such measurements independent of changes in [Ca²⁺]_i the isosbestic wavelength of fura-2 is utilized. However, instead of measuring the fluorescence excited at the isosbestic wavelength, a Ca²⁺ insensitive ‘isosbestic’ fluorescence signal was calculated as F_i=F₃₄₀+α·F₃₈₀. In this equation, α is the isocoefficient that scales the negative F₃₈₀ response to compensate exactly for the positive F₃₄₀ response (fluorescence excited at 340 and 380 nm, respectively) when [Ca²⁺]_i is increased (Zhou and Neher, 1993). The effectiveness of this procedure is illustrated in Fig. 1, in which panel C shows lack of effect of carbachol on the calculated ‘isosbestic’ fluorescence despite a pronounced carbachol-induced [Ca²⁺]_i response (panel D). Owing to photobleaching and loss of indicator from the cells there is a slow gradual decrease of the Ca²⁺-independent fluorescence even in the absence of Mn²⁺ (Fig. 1C, broken line 0). After compensating for this decrease (Fig. 1B), [Mn²⁺]_i can be calculated in analogy to the method previously described (Grynkiewicz et al., 1985) using Equation 2:

K_DMn²⁺ is 2.8 nM (Kwan and Putney, 1990). F_max is the unquenched (Fig. 1B, line 0) and F_min the maximally quenched fura-2 fluorescence in the presence of Mn²⁺, which was set to 1% of F_max (Kwan and Putney, 1990). Fig. 1A illustrates the slow rise of [Mn²⁺]_i upon introduction of the ion (broken line 1) and acceleration of this effect after stimulation with carbachol (broken line 2). Although the apparent K_DMn²⁺ may be expected to change slightly with the Ca²⁺ concentration, we found no evidence for such interference because [Ca²⁺]_i peaks occurred without fluctuations in the Mn²⁺ signal.

Results are presented as means±s.e.m. Differences were statistically evaluated by the two-tailed Student’s t test. The dose-response data (Fig. 2B; Fig. 4B) were fitted to a sigmoidal equation (logistic function) using the Marquart-Levenberg algorithm (SigmaPlot, SPSS Inc. Chicago, IL). The linear curve fits (Fig. 1; Fig. 3; Fig. 5) and all illustrations were made with the Igor Pro software (Wavemetrics Inc., Lake Oswego, OR).

Omission of extracellular Ca²⁺ (reduction from 1.28 to <1 nM) resulted in a modest slow lowering of [Ca²⁺]_i in the hyperpolarized β-cells exposed to 20 mM glucose (Fig. 2A). Subsequent introduction of 10 mM Ca²⁺ caused a rapid but modest increase of [Ca²⁺]_i above the baseline. This effect was considered to be due to leakage of Ca²⁺ through pathways other than the store-operated and voltage-dependent Ca²⁺ channels, which are inhibited by glucose exposure (Liu and Gylfe, 1997) and hyperpolarization, respectively. After return to 1.28 mM Ca²⁺ for a few minutes, Ca²⁺ was again omitted and carbachol added to mobilize Ca²⁺ from the ER. When Ca²⁺ was subsequently increased to 10 mM in the continued presence of carbachol, there was a marked increase in [Ca²⁺]_i owing to contribution of the store-operated Ca²⁺ channels. Carbachol was then omitted and Ca²⁺ lowered to 1.28 mM. Similar cycles were then repeated with increasing concentrations of carbachol (Fig. 2A). When 0.3-1 μM carbachol was introduced in the absence of Ca²⁺ there were no detectable changes in [Ca²⁺]_i and sequentially adding higher concentrations of the drug resulted in small temporary elevations. The effect of 10 mM Ca²⁺ depended on the prevailing carbachol concentration in a graded fashion, the maximal increase in [Ca²⁺]_i being quite pronounced. Fig. 2B shows the dose-response relationship for carbachol-induced elevation of [Ca²⁺]_i in the presence of 10 mM Ca²⁺. Half-maximal and maximal effects were reached at 2.48±0.31 and 30 μM carbachol, respectively, whereas 100 μM gave a slightly smaller response.

After exposure to Mn²⁺ in the absence of agonist there was a linear rise in [Mn²⁺]_i in the hyperpolarized and glucose-exposed β-cells. This rise can be expected to represent entry of the ion through pathways other than the voltage-dependent and store-operated Ca²⁺ channels (Fig. 3A). Subsequent addition of carbachol dose-dependently accelerated the rate of [Mn²⁺]_i increase, owing to activation of the store-operated pathway. In a series of three experiments, the acceleration obtained with 3.6 μM carbachol was 29.9±9.5% of the maximal activation obtained with 30 μM of the drug.

During maximal SERCA inhibition the ER is rapidly depleted owing to leakage of Ca²⁺ (Liu and Gylfe, 1997; Tengholm et al., 1999). We then used increasing concentrations of the SERCA inhibitor CPA to gradually deplete the ER in individual β-cells. The protocol shown in Fig. 4A is similar to that used for carbachol in Fig. 2A except that extracellular Ca²⁺ was varied only between <1 nM and 1.28 mM. Like carbachol, CPA caused some increase of [Ca²⁺]_i in Ca²⁺-deficient medium and a dose-dependent, more pronounced rise in the presence of extracellular cation. The latter effect was half-maximal and maximal at 1.94±0.23 and 10 μM CPA, respectively, whereas 30 μM gave a slightly smaller response (Fig. 3B). Using influx of Mn²⁺ as measure of the store-operated pathway, 2 μM CPA accelerated the influx by 23.6±2.7% (n=5) of the maximal activation obtained with 20 μM of the drug (Fig. 5A).

Gd³⁺ at a concentration of 1 μM has been found to inhibit the store-operated Ca²⁺ entry in a smooth muscle cell line without affecting vasopressin-stimulated influx of the ion (Broad et al., 1999). We now find that 1 μM Gd³⁺ does not interfere with mobilization of ER Ca²⁺ in response to 100 μM carbachol (Fig. 6A) or 50 μM CPA (Fig. 6B) in hyperpolarized β-cells exposed to Ca²⁺-deficient medium. In accordance with an inhibitory effect on store-operated Ca²⁺ influx, subsequent restoration of a physiological Ca²⁺ concentration (1.28 mM) in the continued presence of carbachol or CPA resulted in elevation of [Ca²⁺]_i only when Gd³⁺ was absent. When the store-operated influx in response to carbachol was monitored with Mn²⁺, it was completely abolished by Gd³⁺, which even reduced the basal Mn²⁺ influx (Fig. 7). In addition, Gd³⁺ was an effective blocker of the voltage-dependent rise of [Ca²⁺]_i in response to K⁺ depolarization (Fig. 6C). Other experiments indicated that the effect of Gd³⁺ is not reversible and that 0.1-5 μM of this ion fails to discriminate between the store-operated and voltage-dependent entry of Ca²⁺ (data not shown).

Studies with the cell-permeable Ins(1,4,5)P₃ receptor inhibitor 2-APB have indicated that Ins(1,4,5)P₃ receptors are important for activation and maintenance of store-operated Ca²⁺ entry by a mechanism other than merely emptying the ER (Maruyama et al., 1997; Ma et al., 2000; van Rossum et al., 2000). Using an experimental approach similar to that in Fig. 6A,B, we found that 2-APB diminishes not only intracellular Ca²⁺ mobilization and store-operated influx of the ion in response to the Ins(1,4,5)P₃-elevating agonist carbachol (Fig. 8A), but also the store-operated influx induced by SERCA inhibition with CPA (Fig. 8B). The small increase in basal [Ca²⁺]_i upon introduction of 2-APB (Fig. 8A,B,D) might be caused by the slight SERCA inhibition (Maruyama et al., 1997). Fig. 9 indicates that 2-APB blocks the store-operated influx of Mn²⁺ in response to carbachol as well as the basal influx. Because [Mn²⁺]_i even seems to decrease in the presence of 2-APB, it is not excluded that extrusion of Mn²⁺ dominates under these conditions. Although the presently used concentration of 2-APB has been reported to block store-operated Ca²⁺ influx without effect on L-type Ca²⁺ channels (Maruyama et al., 1997), we find a marked inhibitory action on the voltage-dependent rise of [Ca²⁺]_i obtained with K⁺ depolarization (Fig. 8C,D). This effect is evidently not readily reversible, as there was only a modest temporary rise in [Ca²⁺]_i when omitting 2-APB during K⁺ depolarization (Fig. 8C), and the initial response to K⁺ depolarization was not fully restored even 8 minutes after 2-APB omission (Fig. 8D).

The presence of a store-operated or capacitative pathway was first suggested by Putney (Putney, 1990; Putney, 1986) and has since proved to be the most important mechanism for Ca²⁺ entry into non-excitable cells. In the excitable pancreatic β-cell, this mechanism seems to have only modest direct effects on [Ca²⁺]_i (Liu and Gylfe, 1997) but, by modulating the membrane potential, store-operated fluxes of Ca²⁺ and Na⁺ may be significant for the more pronounced Ca²⁺ influx through the voltage-dependent channels (Worley et al., 1994; Bertram et al., 1995; Liu and Gylfe, 1997; Gilon et al., 1999). The molecular events coupling Ca²⁺ emptying of the ER to activation of Ca²⁺ influx have not yet been unequivocally identified (Putney, 1999). More than one mechanism may be involved, explaining why the Ca²⁺ influx is activated in an all-or-none fashion after almost complete emptying of the intracellular Ca²⁺ stores in some types of cells (Fierro and Parekh, 2000; Fierro et al., 2000), whereas there is gradual activation with increasing depletion of the stores in others (Hofer et al., 1998; Sedova et al., 2000). To date, it is not known how the store-operated Ca²⁺ entry depends on store filling in the pancreatic β-cell, although such knowledge is a prerequisite for current models attributing important functions to the store-operated pathway in the regulation of insulin release (Worley et al., 1994; Bertram et al., 1995; Liu and Gylfe, 1997; Gilon et al., 1999).

To selectively study the store-operated pathway in individual β-cells without interference from voltage-dependent Ca²⁺ entry we employed a previously developed technique (Gylfe, 1991; Liu and Gylfe, 1997). In this approach the β-cells are hyperpolarized with diazoxide, which activates the K_ATP channels (Trube et al., 1986). As an extra precaution, the medium was supplemented with methoxyverapamil, a voltage-dependent Ca²⁺ channel blocker lacking effects on the store-operated entry (Gylfe, 1991; Liu and Gylfe, 1997). Maximal filling of the Ins(1,4,5)P₃-sensitive store of ER Ca²⁺ was ascertained by pre-exposure to 20 mM glucose (Gylfe, 1988; Gylfe, 1991; Tengholm et al., 1999), which was present throughout the experiments. In every cell we found that increasing concentrations of the Ca²⁺-mobilizing carbachol cause gradual elevation of [Ca²⁺]_i depending on store-operated influx. In most experiments, one observation point was on the steepest part of the dose-response curve, contrary to what can be expected if the entry is regulated in an all-or-none fashion. Similar results were obtained with increasing concentrations of the SERCA inhibitor CPA, which empties the ER via a leakage pathway after inhibition of Ca²⁺ uptake. Unlike a previous study (Liu and Gylfe, 1997), we did not attempt to correlate the effects of carbachol and CPA on mobilization of intracellular Ca²⁺ with the magnitude of the store-operated influx. Such an approach requires separate experiments at each concentration to ascertain that the ER is completely filled when introducing the test substance.

Mn²⁺ quenching of the fura-2 fluorescence is a potent technique for more direct studies of fluxes through the voltage-dependent (Dryselius et al., 1999) and store-operated (Liu and Gylfe, 1997) pathways in the β-cell. Because quenching exhibits a non-linear dependence on Mn²⁺ concentration, we introduced a novel approach calculating actual [Mn²⁺]_i levels from the quenching curve. In all situations studied, [Mn²⁺]_i increased linearly throughout the observation periods, although the rate varied depending on stimulation. The rate of increase can therefore be taken as a measure of influx with little interference from outward transport. Using this approach we found that 3.6 μM carbachol and 2 μM CPA, concentrations slightly higher than those giving half-maximal elevation of [Ca²⁺]_i, induced only 30 and 24% activation of the store-operated influx, respectively. Consequently, there is no linear relationship between influx rate and elevation of [Ca²⁺]_i. An explanation may be that, at high agonist concentrations, fura-2 in the submembrane space becomes saturated with Ca²⁺ resulting in underestimation of [Ca²⁺]_i and a left shift of the dose-response relationships.

Individual pancreatic β-cells respond to glucose with slow [Ca²⁺]_i oscillations with a frequency of 0.2-0.5/minute (Grapengiesser et al., 1988). Similar oscillations are observed in pancreatic islets but, within the islets, the β-cell response is dominated by about tenfold faster oscillations (Valdeolmillos et al., 1989; Bergsten et al., 1994; Gilon et al., 1994). It was previously shown that the fast oscillations depend on cAMP and that they can be transformed into slow oscillations by SERCA inhibition (Liu et al., 1998). Modeling the generation of the fast oscillatory pattern it has been suggested that release of Ca²⁺ from the ER causes a hyperpolarizing current, which shuts off the voltage-dependent entry of Ca²⁺ (Ämmälä et al., 1991; Liu et al., 1998; Dryselius et al., 1999). However, the associated emptying of the ER has been suggested to generate the fast oscillations by activating a depolarizing store-operated current (Worley et al., 1994; Bertram et al., 1995; Gilon et al., 1999). To discriminate between these seemingly inconsistent alternatives it would be valuable to have an inhibitor of the store-operated pathway, which does not affect mobilization of ER Ca²⁺ or voltage-dependent entry of the ion. Testing suggested inhibitors we found that Gd³⁺ and 2-APB lack the required Ca²⁺ channel specificity. The usefulness of 2-APB is limited because this Ins(1,4,5)P₃ receptor inhibitor will interfere with Ca²⁺ mobilization from the ER.

Taken together, this study provides the first evidence that the store-operated entry of Ca²⁺ into the β-cell exhibits a graded dependence on Ca²⁺ filling of the ER. Small variations in the ER Ca²⁺ concentration may consequently contribute to the regulation of the membrane potential and [Ca²⁺]_i determining insulin release.

This work was supported by grants from the Swedish Medical Research Council (12X-6240), the Swedish Foundation for Strategic Research, the Swedish Foundation for International Cooperation in Research and Higher Education, the Wenner-Gren Center Foundation, the Swedish Diabetes Association, Novo-Nordisk Foundation, Family Ernfors foundation, Åke Wiberg’s Foundation and the Swedish Society for Medical Research.