Dynamic Properties of an Inositol 1,4,5-Trisphosphate– and Thapsigargin-insensitive Calcium Pool in Mammalian Cell Lines

Tissue culture medium and complements were purchased from Technogenetics (Milan, Italy); indo-1 and indo-1/AM were from Molecular Probes (Eugene, OR); S202791 was a gift of Dr. D. Pietrobon (University of Padova); and all other reagents were from Sigma Chemical Co. (St. Louis, MO).

Rat pituitary GH3 cells (from Dr. Hescheler, University of Berlin, Germany) were grown in Ham's F-10 medium supplemented with 15% horse serum, 2.5% FCS, nonessential amino acids, and penicillin/streptomycin. The clonal cell lines GC and GH3B6 (from Dr. Argenton, University of Padova), derived from GH3 cells, were maintained in the same conditions (with the addition of 10⁻⁷ M estradiol for the latter cell type). The pheochromocytoma cell line PC12 was grown in RPMI supplemented with 12.5% horse serum, 2.5% FCS, and gentamycin. The rat basophilic leukemia cell line RBL-1 (from Dr. Penner, Max-Planck-Institut, Göttingen, Germany) was cultured in DME supplemented with 10% FCS and penicillin/streptomycin. For Ca²⁺ measurements, all cells were allowed to attach to poly-l-lysine–coated glass coverslips and grow for 1 d.

Transient transfection with recombinant cytosolic or mitochondrial aequorin (cytAEQ or mtAEQ) (Rizzuto et al., 1992) was made by electroporation (Kodak, IBI, Rochester, NY). Cells were harvested and resuspended in fresh medium in 4-mm cuvettes in the presence of 10 μg of cytAEQ/VR1012 or mtAEQ/VR1012 plasmid/10⁶ cells. Cells were subjected to a single pulse characterized by an electric field of 300 V, 1,500 μF. Transfected cells were transferred to 13-mm-diam poly-l-lysine–coated glass coverslips (10⁶ cells/coverslip), and after an overnight incubation, the medium was changed and the incubation continued under the same conditions. After 2 d, the cells were used for Ca²⁺ measurements, according to the procedure described in detail elsewhere (Rizzuto et al., 1994).

Cells, grown on coverslips (24-mm-diam), were loaded with indo-1 by incubation with 5 μM indo-1/AM at 37°C for ∼30 min in Ham's F-10 medium containing 3% FCS and 0.04% pluronic acid. After washing the cells with a modified Krebs-Ringer medium (mKRB: in mM, 125 NaCl, 5 KCl, 1 MgSO₄, 1 Na₃PO₄, 1 CaCl₂, 20 Hepes, 5.5 glucose, pH 7.4), the coverslips were mounted in a chamber and placed on the stage of an inverted microscope (model Diaphot 300; Nikon Europe B.V., Badhoevedorp, The Netherlands), equipped with a 40× water immersion objective (NA=1.1; Nikon) and connected with a real-time UV confocal system (model RCM8000; Nikon). The 351-nm band of the argon ion laser was used for excitation and the emitted light, separated into its two components (405 and 485 nm) by a dichroic mirror, was collected by two photomultipliers. The ratio of the intensity of the light emitted at the two wavelengths (F405/F485), a function of [Ca²⁺]_i, was displayed as a pseudocolor scale. Unless otherwise indicated, all experiments were performed at room temperature and time series were acquired with a frame interval of 2 s, averaging 16 images for each frame. Online analysis of the ratio was obtained from the signals of individual cells. For presentation, ratios were off-line averaged and normalized to the average value obtained in Ca²⁺-free medium before ionomycin addition.

For measurements of cytosolic or mitochondrial free Ca²⁺ concentration ([Ca²⁺]_m) with recombinant aequorins, cells were transiently transfected with cytAEQ or mtAEQ and seeded onto 13-mm-diam poly-l-lysine– coated glass coverslips (10⁶ cells/coverslip) 2 d before the experiment. Aequorin was reconstituted by adding 5 μM coelenterazine to the culture medium 1–2 h before the experiment. During the experiment, cells were continuously perfused with mKRB containing different stimuli. At the end of each experiment, cells were lysed by perfusion with an ipoosmotic solution containing only digitonin (100 μM) and 10 mM CaCl₂ to expose all the cellular aequorin to a high [Ca²⁺]. Light emission was measured by a purpose-built luminometer and calibrated in terms of [Ca²⁺]_i or [Ca²⁺]_m as described (Rizzuto et al., 1994; Brini et al., 1995).

The total content of cellular Ca²⁺ was assayed by atomic absorption spectrophotometry. Cells were suspended in mKRB (10⁷ cells/ml) and challenged with 30 mM KCl (GH3 cells) or 1 μM Tg (RBL-1 cells) for 3 min before addition of 4 mM EGTA. Control experiments were carried out under the same conditions in Ca²⁺-free mKRB containing 1 mM EGTA. After centrifugation, cells were resuspended in Ca²⁺-free mKRH containing 1 mM EGTA and centrifuged for 1 min at 14,000 rpm in Eppendorf tubes containing 100 μl sucrose 12.5% and 400 μl silicon oil. The pellet was resuspended in 0.05% Triton plus 0.2 N NaOH before measurement.

GH3 cells were fixed and immunostained as described previously (Brini et al., 1995). The antibody, used at 1:200 dilution (kind gift of Dr. O. Rossetto, University of Padova), was against the type I membrane protein TGN38, a marker of the trans-Golgi network (Reaves et al., 1996). Binding of the antibody was revealed with an FITC-labeled anti–rabbit IgG antibody. Fluorescence was analyzed with a microscope (model RCM 8000; Nikon) and photographed (Technical Pan film; Kodak) after digital subtraction of the background.

The existence of different Ca²⁺ storage compartments whose content can be discharged in the cytoplasm and/or in the extracellular medium by treatment with a variety of agents appears to be a widespread characteristic of eukaryotic cells. Initially in the neuroendocrine cell line PC12 (Fasolato et al., 1991; Zacchetti et al., 1991), and later in several other cell types, we have functionally distinguished three types of intracellular Ca²⁺ pools: (a) an InsP₃-sensitive pool, largely identified with the ER and endowed with a Tg-sensitive Ca²⁺–ATPase; (b) an acidic pool, whose Ca²⁺ content could be released into the cytoplasm by Ca²⁺ ionophores (A23187 or ionomycin), but only after neutralization of the luminal acidic pH with monensin or NH₄Cl; and (c) a Ca²⁺ pool that could be discharged by treatment only with Ca²⁺ ionophores. This last Ca²⁺ store, presumably heterogeneous, is the most mysterious in terms of cytological nature, mechanism of loading, and physiological role. The experiments presented below were aimed at the characterization of this elusive Ca²⁺ compartment, using as a model system primarily the rat pituitary cell line GH3.

Fig. 1, a and b, shows the typical protocol used to ascertain the existence of the different Ca²⁺ pools in GH3 cells. Cells, loaded with the fluorescent Ca²⁺ indicator indo-1, were incubated in mKRB medium and analyzed by confocal microscopy. The cells were treated in sequence with EGTA, to chelate extracellular Ca²⁺; thyrotropin- releasing hormone (TRH), an agonist coupled to InsP₃ generation; a SERCA inhibitor, such as Tg; ionomycin, a Ca²⁺ ionophore; and monensin, a Na⁺-K⁺/H⁺ ionophore. Addition of TRH induced a rapid and transient increase in [Ca²⁺]_i, while Tg, after TRH, caused a barely detectable rise. The subsequent challenge with ionomycin resulted in a further small increase of [Ca²⁺]_i, and finally monensin discharged the acidic pool.

The kinetics of the [Ca²⁺]_i peaks reflect the balance between Ca²⁺ efflux from the stores, pumping out of the cell and reuptake into the stores themselves. However, the integral of the curve underlying the [Ca²⁺]_i rise, induced by each stimulus, mirrors, to a reasonable level of approximation, the Ca²⁺ content of the different stores as estimated by ⁴⁵Ca²⁺ measurements (Fasolato et al., 1991; Bastianutto et al., 1995; Mery et al., 1996) and can be used to calculate the size of the different pools. Accordingly (see Table I), the TRH- and Tg-releasable pools represent altogether 38%, the ionomycin-pool 17%, and the monensin-pool 45% of total mobilizable Ca²⁺. In Fig. 1 a, the trace is the mean of 30 cells, while in b the kinetic changes of [Ca²⁺]_i in four typical cells are presented. As previously reported, in Ca²⁺-containing medium, GH3 cells underwent spontaneous oscillations that ceased immediately upon addition of EGTA (Schlegel et al., 1987). Noteworthily, the size of the various Ca²⁺ pools was quite homogeneous in the cell population.

No [Ca²⁺]_i increase was detected upon application of 10 mM caffeine (not shown), although in another pituitary cell line (GH4C1), a caffeine-sensitive store has been identified (Law et al., 1990; Tanaka and Tashjian, 1993).

Fig. 1, c and d, shows that 1-min depolarization with 30 mM KCl in Ca²⁺-containing medium affected the size of the different pools. In particular, c, mean of 30 cells, shows that addition of KCl induced a sharp increase of [Ca²⁺]_i that rapidly subsided upon chelation of extracellular Ca²⁺ with EGTA (or addition of an L channel blocker such as nifedipine; not shown). The kinetics and amplitude of TRH- and Tg-induced [Ca²⁺]_i rise were hardly different from those of control cells (Fig. 1,a and Table I), while a massive increase occurred in the ionomycin-releasable pool. When the other Ca²⁺ ionophore A23187 was used instead of ionomycin, the results were indistinguishable (see below). This Ca²⁺ ionophore-sensitive pool induced by KCl stimulation will be referred to, for simplicity, as stimulus-induced-calcium (SIC) pool. Finally, the monensin pool appeared decreased by KCl-induced depolarization, as expected if it is accounted for, at least in part, by Ca²⁺ sequestered within secretory compartments. Similar results for both controls and treated cells have been obtained when the experiments were carried out at 37°C (not shown).

Fig. 1,d shows the behavior of four typical cells. While all the examined cells reacted to KCl stimulation with high and sustained [Ca²⁺]_i rises, the increase of the SIC pool was consistently observed in >50% of the cells. In particular, out of the 30 cells analyzed in this experiment, in a fraction of them (10%) the ionomycin peak and the integrated area were indistinguishable from those of controls, while in ∼30% the increase in the integrated area was two to sixfold higher than in controls, and the increase in the remaining 60% was between 6- and 15-fold. On average, in this experiment, there was a 10-fold increase in the area of the SIC pool (Fig. 1, a and c). In a series of similar experiments (n = 33), in 16 trials the averaged area of this pool was between 6–10-fold the control value, in 15 trials the area was three to fivefold, and the area hardly increased in the remaining two trials.

The possibility was considered that the increase of the SIC pool was only apparent and due to an impairment of the Ca²⁺ extrusion mechanisms. Two types of evidence argue against this possibility: (a) The peaks and kinetics of decay of the TRH-induced [Ca²⁺]_i increase were similar in cells that have been treated with KCl when compared to controls (1.90 ± 0.17 and 1.88 ± 0.04 normalized ratio U for the peak, and 14.9 ± 0.9 and 9.8 ± 0.3 s for the half time of decay in control and KCl-treated cells, respectively [mean ± SD]) and (b) the substitution of Na⁺ with N-methyl- glucamine⁺ in the medium (to inhibit any contribution of the Na⁺/Ca²⁺ exchanger) resulted in no alteration of the kinetics and amplitude of the [Ca²⁺]_i transients caused by the different stimuli (not shown). These results suggest that, overall, extrusion of Ca²⁺ into the medium or reaccumulation into the stores is not grossly affected by depolarization. In addition, these experiments indicate that, in GH3 cells, the Na⁺/Ca²⁺ exchanger has minor relevance in Ca²⁺ extrusion, at least under our experimental conditions.

Ca²⁺ indicators, such as indo-1, when loaded as acetoxymethyl-esters, are not only found in the cytosol but also trapped within intracellular organelles. In addition, given that their final cytosolic concentration is in the order of 20–100 μM, they substantially increase the cytosolic Ca²⁺ buffering capacity. To determine whether the appearance of the SIC pool was at least in part an artifact of the methodology used, the [Ca²⁺] in the cytosol was measured with recombinant aequorin. In this latter case, the increase in Ca²⁺ buffering capacity is negligible and the transfected protein is excluded from organelles (Brini et al., 1995). Fig. 2 shows that, while in control cells, A23187 added after TRH and tBHQ caused a marginal increase in [Ca²⁺]_i (Fig. 2,b), a very large rise was caused by the ionophore in cells pretreated with KCl (Fig. 2 a). In the experiments with recombinant aequorin (see also below), tBHQ and A23187 were used instead of Tg and ionomycin because the latter compounds tend to adhere strongly to the tubes of the perfusion system (Rizzuto et al., 1994).

To further confirm that the increase in the ionomycin response was indeed due to Ca²⁺ accumulation and not to impairment of the extrusion mechanisms, the total cellular Ca²⁺ content was measured by atomic absorption (see Materials and Methods). After 3 min of exposure to 30 mM KCl in mKRB medium, the total Ca²⁺ content increased by 43% compared to controls (5.65 ± 0.74 and 3.94 ± 0.4 nmol/mg protein, respectively; n = 5 for each condition, P < 0.002, Student's t test).

One of the key features of the SIC pool, besides its InsP₃ insensitivity, is its resistance to unloading by classical SERCA inhibitors. Fig. 3 a demonstrates that Tg, when added before depolarization, even at doses as high as 10 μM, while inhibiting the TRH-induced [Ca²⁺]_i changes, did not significantly decrease the amplitude and kinetics of the Ca²⁺ rise induced by ionomycin. Other SERCA inhibitors such as tBHQ (30 μM) and CA (10 μM) were similarly ineffective (not shown). Thus, not only are SERCA inhibitors unable to deplete this pool, but they are also ineffective in preventing its loading.

The SIC pool was also unaffected by caffeine (10 mM) or ryanodine (10 μM), whether added before or after KCl depolarization (not shown). Last but not least, preincubation with monensin resulted in an increase in the ionomycin-dependent [Ca²⁺]_i rise of both controls and KCltreated cells. The integrals of the curves, however, were equivalent to the sum of the individual ionomycin- and monensin-induced [Ca²⁺]_i increases. Thus, alkalinization of luminal pH does not prevent loading of the SIC pool.

The insensitivity to SERCA inhibitors and caffeine/ryanodine, as well as the ability to load efficiently only as a consequence of large [Ca²⁺]_i increases, appear compatible with the identification of the SIC pool with mitochondria (for reviews see Pietrobon et al., 1990; Gunter et al., 1994; Pozzan et al., 1994). Two lines of evidence, however, argue against this hypothesis. Fig. 3,b shows that in cells subjected to the same protocol of Fig. 1 c (i.e., KCl for 1 min followed by EGTA, TRH, and Tg), the addition of the mitochondrial uncoupler p-(trifluoro-methoxy)phenylhydrazone (FCCP, 10 μM) caused no increase of [Ca²⁺]_i per se and, more importantly, hardly affected the amplitude and kinetics of the ionomycin-induced [Ca²⁺]_i rise. The area of the SIC pool was 57.1 ± 12.3 and 60.5 ± 24.8 normalized ratio U (mean ± SD, n = 3) in the presence and absence of FCCP, respectively. At these concentrations, FCCP completely collapses the mitochondrial membrane potential and, in isolated organelles, causes rapid release of accumulated Ca²⁺. Accordingly, the inefficacy of the uncoupler in reducing the size of the SIC pool strongly argues against the possibility that mitochondria represent a significant part of this store. Similar results were obtained when the mitochondrial membrane potential was collapsed by a combination of the ATP-synthase inhibitor oligomycin (2 μM) and the mitochondrial complex–I inhibitor rotenone (4 μM) (not shown).

The role of mitochondria was directly tested in the experiment presented in Fig. 4. GH3 cells were transiently transfected with the cDNA coding for recombinant mtAEQ (Rizzuto et al., 1992). Fig. 4 shows that the mean resting [Ca²⁺]_m of GH3 cells was somewhat higher than that reported previously in other cell types, such as HeLa or endothelial cells (∼500 vs. ∼200 nM) (Rizzuto et al., 1994; Lawrie et al., 1996). This, however, was not unexpected since, as shown in Fig. 1 b, GH3 cells undergo spontaneous [Ca²⁺]_i oscillations. Addition of KCl resulted in a very sharp and large increase of [Ca²⁺]_m, with a peak around or above 10 μM. After the peak, [Ca²⁺]_m rapidly returned toward basal and addition of EGTA caused a further decrease to, and below, resting level. While perfused with Ca²⁺-free, EGTA-containing medium, addition of TRH + tBHQ resulted in an increase of [Ca²⁺]_m to ∼3 μM, and finally the Ca²⁺ ionophore A23187 caused a further slow and reproducible increase. Thus, the ionophore caused an increase in [Ca²⁺]_m (presumably by equilibrating the Ca²⁺ concentration of the cytosol with that of the mitochondrial matrix) and not a decrease, as expected if mitochondria represented a significant part of the SIC pool.

The experiments shown in Fig. 5 were undertaken to define the kinetics of loading and unloading of the SIC pool. A 10-s exposure to 30 mM KCl (Fig. 5,a, continuous trace) failed to induce a reproducible increase, compared to controls, of the ionomycin-induced rise of [Ca²⁺]_i, while a 20-s depolarization caused, in 50% of the trials (n = 13), a significant increase in the size of this store (Fig. 5,a, dotted trace). After 60 s of depolarization, the increase of [Ca²⁺]_i caused by the Ca²⁺ ionophore was similar to that observed after 5 min of KCl exposure (Fig. 5,a, long- and shortdashed traces, respectively). In addition, the SIC pool loaded faster and was larger if the depolarization was carried out in high external [Ca²⁺] (5 mM) (compare Fig. 5 b vs. a). In nondepolarizing conditions, the size of the pool was not appreciably affected by the increase in [Ca²⁺]_out (not shown).

In the experiments presented in Figs. 1–4, loading of the SIC pool was caused by a prolonged depolarization with KCl. The question then arises as to whether other stimuli causing Ca²⁺ influx can similarly induce an increase in the size of this Ca²⁺ store. The amplitude and kinetics of the ionomycin-sensitive pool were thus investigated in cells pretreated for 3 min with TRH in Ca²⁺-containing medium, a condition that discharges the InsP₃-sensitive store and activates voltage-operated Ca²⁺ channels (VOCCs) (Gollasch et al., 1991; Mantegazza et al., 1995) and, presumably, store-dependent Ca²⁺ influx. Fig. 5,c shows that TRH addition in Ca²⁺-containing medium resulted in a biphasic Ca²⁺ transient, a sharp rise followed by a sustained plateau that is much smaller, however, than that induced by KCl. Upon chelation of extracellular Ca²⁺ with EGTA, the addition of ionomycin, after Tg, resulted in a transient [Ca²⁺]_i increase, which was, however, indistinguishable from that of cells treated with TRH and Tg in Ca²⁺-free, EGTA-containing medium (Fig. 5 c, continuous and dashed traces, respectively). Thus, the influx of Ca²⁺ induced by TRH is insufficient to induce a significant increase in the size of the SIC pool.

The kinetics of the unloading of the SIC pool were next investigated. The protocol was the following: the pool was first loaded with Ca²⁺ through a 1-min exposure to KCl depolarization, followed by EGTA addition, to block Ca²⁺ influx through VOCCs. The cells were then washed and kept, for different times, in Ca²⁺-containing medium. Finally, after readdition of EGTA, the size of the pools was tested, as before, by addition of TRH + Tg followed by ionomycin. Fig. 5 d shows that complete unloading of the SIC pool required a total of 20–30 min between the end of KCl to the ionomycin treatment.

The insensitivity of the SIC pool to Tg and monensin indicates that Ca²⁺ uptake in this pool is not driven by H⁺ or Na⁺ gradients and does not use classical SERCAs. We considered the possibility that Ca²⁺ accumulation in the SIC pool occurred by a mechanism of vesicle endocytosis after KCl-induced secretion. However, the lack of a significant uptake of Lucifer yellow, a marker of fluid phase endocytosis, in KCl-treated cells (not shown) makes this possibility unlikely. If it occurred through endocytosis, the amount of Ca²⁺ increase (≅0.4 mmol/l of cell water) caused by depolarization would have, in fact, implied an accumulation of extracellular medium that was not only rapid but also exceedingly vast, i.e., corresponding to about 1/4 of the cell volume.

A Ca²⁺ pool with characteristics similar to those identified in GH3 cells was found also in other neuroendocrine cell lines, i.e., two pituitary cell lines, GC and GH3B6 (not shown), and in PC12 cells, derived from a rat pheochromocytoma. In this latter cell line, however, the SIC pool was detectable only when depolarization with KCl was carried out in the presence of an agonist of L type VOCCs, S202791 (1 μM) (Fig. 6, a and b). In the absence of this drug, in fact, KCl-induced depolarization resulted in a plateau level of [Ca²⁺]_i consistently lower than in GH3 cells. Thus, considerable Ca²⁺ accumulation in the SIC pool requires a [Ca²⁺]_i rise of large amplitude and long duration.

A low-affinity, high-capacity SIC pool was also found in nonexcitable cells, such as the rat basophilic leukemia cell line, RBL-1. In these cells, large loading of this pool could not be achieved by membrane potential depolarization, since they do not possess VOCCs, but rather through activation of another Ca²⁺-selective current, named I_CRAC, for Ca²⁺ release-activated Ca²⁺ current (Hoth and Penner, 1992; Fasolato et al., 1993), maximally activated by Tg (for review see Fasolato et al., 1994). Fig. 6 d shows that Tg induced a long-lasting elevation of [Ca²⁺]_i that rapidly returned to basal level upon EGTA addition. Under these conditions, ionomycin induced a large, transient increase in [Ca²⁺]_i, severalfold higher and more prolonged than that caused by the ionophore in cells treated with Tg in EGTA-containing medium (c). Similarly to GH3 cells, the increase in total cellular Ca²⁺ content, as estimated by atomic absorption, accounted for ∼33% (n = 3). Moreover, in these cells addition of monensin (1 μM) after ionomycin never elicited a [Ca²⁺]_i rise, either in control or stimulated cells (not shown). Finally, also in the RBL-1 cells the SIC pool was insensitive to caffeine (10 mM) and ryanodine (10 μM), added either before or after Tg stimulation, and to the mitochondrial uncoupler FCCP (10 μM) (not shown).

A final property of the SIC pool here investigated is its intracellular distribution. In both GH3 and RBL-1 cells, high-resolution (67 ms/ratio frame) analysis by confocal microscopy failed to reveal a localized origin of the ionomycin-induced response. In fact, Fig. 7 shows that, upon ionomycin addition in both GH3 (a) and RBL-1 cells (b), Ca²⁺ increased rather homogeneously in the cytosol and quickly diffused to the central nuclear region, with no evidence for a localized origin of the [Ca²⁺]_i rise. These observations would tend to exclude the Golgi complex, which has been previously suggested to accumulate Ca²⁺ (Virk et al., 1985; Chandra et al., 1991; Wahl et al., 1992) as the major source of the ionomycin-releasable Ca²⁺. As confirmed by immunofluorescence, the Golgi network is, in fact, specifically localized to a distinct perinuclear region (Fig. 8 c). It may be argued, however, that the increase in [Ca²⁺]_i caused by the ionophore is relatively slow, i.e., 8.1 ± 2.9 s to the peak (mean ± SD, in 12 similar experiments with GH3 cells), and thus the localization of the increase may be masked by diffusion of Ca²⁺ and/or of the indicator.

To obtain further insights into the possible involvement of the Golgi complex, induction of the SIC pool was monitored in the continuous presence of brefeldin A (BFA) (10 μg/ml, after a pretreatment of 15 min at 37°C). BFA is known to disassemble the Golgi network (LippincottSchwartz et al., 1989) and to cause a major intermixing of membrane and lumenal Golgi components with the ER. Thus, the prediction will be that if the SIC pool was largely identifiable with the Golgi, BFA treatment should make it largely sensitive to Tg. Fig. 8 shows that in both RBL-1 (a) and GH3 cells (b), BFA did not significantly affect the size of the SIC pool, although it caused the known fragmentation of the Golgi complex (c).

The Ca²⁺ concentration in the cytoplasm controls a variety of cellular functions and is in dynamic equilibrium with that of intracellular organelles. Within the lumen of different compartments, Ca²⁺ plays different roles, such as that of a reservoir to be mobilized upon cell activation (e.g., in the ER and its subcompartments), of a modulator of enzymatic reactions (e.g., in the mitochondria), and of a cofactor in the packaging of secretory products (e.g., in the Golgi complex and secretory vesicles). With the exception of the ER and mitochondria, however, the mechanism of uptake into and release from other organelles is still largely obscure. In the cell lines PC12, HeLa, mouse fibroblasts, and human lymphocytes, three major pools (or groups of pools) have been functionally characterized on the basis of their loading and release properties: the first released by InsP₃ and Tg, the second insensitive to the above agents and released by Ca²⁺ ionophores, and the third released by the combination of the latter compounds with a drug that dissipates H⁺ gradients, such as monensin and NH₄Cl (Fasolato et al., 1991; Clementi et al., 1994; Bastianutto et al., 1995; Mery et al., 1996).

The Ca²⁺ pool released by ionomycin in the presence of monensin was undetectable in RBL-1 cells, while it was consistently observed in the neurosecretory cell types. In GH3 cells and, as previously observed, in PC12 cells (Fasolato et al., 1991), the size of the acidic compartment is reduced after stimulation of secretion and may suggest its identification, at least in part, with secretory compartments.

The key observation of this contribution is that only a fraction of the nonacidic, InsP₃- and Tg-insensitive pool, which in unstimulated cells is rather heterogeneous (including mitochondria, Golgi, etc.), is highly dynamic and dramatically increases its Ca²⁺ content upon prolonged rises of [Ca²⁺]_i.

The two major questions generated by our data concern, on the one hand, the cytological nature of the SIC pool, and on the other hand, the physiological role of this compartment. As to the first question, our approach has been that of comparing the pharmacological sensitivity and subcellular distribution of the Ca²⁺ uptake and release into and from this pool with that of known organelles. The first compartment that we considered as a candidate was the mitochondrion, since some of the characteristics of the SIC pool, i.e., low affinity and high capacity, resemble those traditionally attributed to this organelle. In particular, several authors have suggested that mitochondria play a key role in buffering large and prolonged increases in [Ca²⁺]_i (Carafoli and Crompton, 1976; Bygrave, 1978; Brinley et al., 1979; Friel and Tsien, 1994; Werth and Thayer, 1994; Budd and Nicholls, 1996; Herrington et al., 1996;). However: (a) The Ca²⁺ accumulated in the SIC pool cannot be released into the cytosol by mitochondrial uncouplers; and (b) direct monitoring of [Ca²⁺]_m with targeted recombinant aequorin demonstrates that Ca²⁺ ionophores cause an increase and not a decrease in [Ca²⁺]_m. Thus, though mitochondria may well contribute to the ionomycin-sensitive pool of resting cells, they do not represent a major part of the pool that accumulates Ca²⁺ after prolonged [Ca²⁺]_i increases. Rather than Ca²⁺ sinks, mitochondria appear therefore to be in dynamic equilibrium with the changes in [Ca²⁺]_i, with dramatic transient increases of their matrix [Ca²⁺] during cell activation.

Similarly, identification of the SIC pool with acidic compartments of the cells, i.e., secretory vesicles, granules, lysosomes, endosomes, etc., appears unlikely since the kinetics, amplitude, and duration of the [Ca²⁺]_i increases induced by ionophores are unaffected by agents such as monensin or NH₄Cl, which collapse internal pH gradients. Moreover, Ca²⁺ ionophores, such as A23187 and even more ionomycin, are highly inefficient at releasing Ca²⁺ from acidic compartments (Liu and Hermann, 1978; Fasolato and Pozzan, 1989).

Another possible candidate is the Golgi complex, given that several lines of evidence, though mainly indirect, suggest that this compartment may contain large amounts of Ca²⁺ (Chandra et al., 1991; Cui et al., 1995). Since neither the mechanism of Ca²⁺ uptake nor that of release from the Golgi complex are known to date, we have tested this idea by taking advantage of the high spatial and temporal resolution of the confocal microscope. In fact, if the Ca²⁺ released by ionomycin was largely coming from the Golgi, the response was expected to get started in a region of high Ca²⁺ close to the nucleus, where the Golgi complex is known to be localized. On the contrary, the [Ca²⁺]_i increase caused by the ionophore appeared from the very beginning to distribute in the whole cytosol. In addition, the increase in [Ca²⁺]_i caused by ionomycin was unaffected by BFA pretreatment. This drug is known to cause dissolution of the Golgi complex and intermixing of its membrane and content with the ER (Lippincott-Schwartz et al., 1989).

Among the other known organelles, peroxisomes were considered unlikely candidates since they represent at the most 1% of the cell volume (Lazarow, 1989), while the increase in total Ca²⁺ in GH3 cells treated with KCl is about 0.4 mmol/l of cell water. Thus, if peroxisomes accounted for the majority of the SIC pool, their Ca²⁺ content should exceed 40 mM, a value in the order of that estimated for the cell organelle with the highest Ca²⁺ content reported so far: the terminal cysternae of sarcoplasmic reticulum (Volpe and Simon, 1991; Grohovaz et al., 1996).

As far as the ER is concerned, the insensitivity to SERCA inhibitors, InsP₃, caffeine, and ryanodine would tend to exclude also this compartment as the candidate organelle. SERCAs are in fact believed to be widespread in this endomembrane system. However, several lines of evidence indicate not only that the ER itself may be heterogeneous (for review see Sitia and Meldolesi, 1992), but also that Ca²⁺ pumps other than those sensitive to the classical inhibitors (Tg, CA, and tBHQ) may be present in this organelle (Bian et al., 1991, Chen et al., 1993). Ca²⁺–ATPases with molecular weight different from those of classical SERCAs have been described by Burgoyne et al. (1989) in adrenal chromaffin cells (but see also Poulsen et al., 1995) and by Rooney and Meldolesi (1996) in PC12 cells. Finally, in subcellular fractionation experiments, the existence of a microsomal fraction whose Ca²⁺ content is insensitive to InsP₃, caffeine-ryanodine, and also to SERCA inhibitors has been repetitively reported (Nori et al., 1993, 1996; Hussain et al., 1995). If indeed the SIC pool represents a subcompartment of the ER, it must be concluded, however, that this part of the organelle equilibrates very slowly, or not at all, with the rest of the membrane network. Until now, its existence might have gone unnoticed given that under resting conditions the Ca²⁺ content of this subcompartment is relatively small and cannot be released into the cytosol by known physiological stimuli. It should be mentioned that a low-affinity Ca²⁺ storage compartment, which loaded only upon large and prolonged increases in [Ca²⁺]_i caused by KCl depolarization, had been previously described in a clone of PC12 cells by Reuter's group. Unlike the experiments reported here, however, this compartment was sensitive to caffeine, while its Tg sensitivity was not tested (Reber and Reuter, 1991; Reber et al., 1993).

As to the physiological role of the SIC pool, the interest depends not only on its large capacity (it can account, on average, for >40% of total cell Ca²⁺) but also on the fact that it is inducible (Clementi et al., 1994; Hussain et al., 1995), and its expression and loading properties appear widespread, if not ubiquitous, as documented by its presence in numerous cell types of different origin (GH3, GC, GH3B6, PC12, RBL-1, and lymphocytes; Clementi et al., 1994; Pizzo, P., C. Fasolato, and T. Pozzan, unpublished).

Admittedly, the conditions needed for loading of this Ca²⁺ pool may appear extreme. However, long-lasting increases in [Ca²⁺]_i can be elicited by high agonist doses or under pathological conditions (Choi, 1989). Furthermore, the observation that loading and unloading of this Ca²⁺ pool can be relatively fast suggests the possibility that this store plays a role also under physiological conditions, but its high-capacity is unraveled by large and long-lasting [Ca²⁺]_i increases, when its size overwhelms that of the other Ca²⁺ storage compartments. In other words, it can be suggested that a Ca²⁺ pool, endowed with a Ca²⁺ accumulation mechanism distinct from that of mitochondria and acidic compartments and independent from classical SERCAs, is a normal component of the Ca²⁺ homeostatic machinery of the cell. The role of this previously unrecognized compartment could be twofold: on the one hand, by accumulating part of the Ca²⁺ coming from the extracellular medium or released from other stores, it becomes the most relevant Ca²⁺ sink under conditions of stress, a function traditionally attributed to mitochondria; on the other hand, by slowly releasing its Ca²⁺ content into the cytosol after a prolonged stimulation, it may also participate in signaling integration.

This work was supported by grants from the Italian Research Council (CNR) Special Project “Oncology,” from the Italian University Ministry, from “Telethon” (grant N.495), from the EU programs “Biomed2,” “Human Capital and Mobility,” and “Copernicus,” and from the “Human Frontier Science Program” to T. Pozzan.

BFA

brefeldin A

CA

cyclopiazonic acid

[Ca²⁺]_i

intracellular free Ca²⁺ concentration

[Ca²⁺]_m

intramitochondrial free Ca²⁺ concentration

cytAEQ

cytosolic aequorin

FCCP

p-(trifluoro-methoxy) phenylhydrazone

InsP₃

inositol 1,4,5-trisphosphate

mKRB

modified Krebs Ringer Buffer

mtAEQ

mitochondrial aequorin

SERCAs

sarco/endoplasmic reticulum Ca²⁺–ATPases

SIC

stimulus-induced-calcium

tBHQ

2,5-di(tert-butyl)-1,4-benzohydroquinone

Tg

thapsigargin

TRH

thyrotropin-releasing hormone

VOCCs

voltageoperated Ca²⁺ channels