Focal agonist stimulation results in spatially restricted Ca²⁺ release and capacitative Ca²⁺ entry in bovine vascular endothelial cells

Abstract

1. Intracellular Ca²⁺ ([Ca²⁺]_i) signals were studied with spatial resolution in bovine vascular endothelial cells using the fluorescent Ca²⁺ indicator fluo-3 and confocal laser scanning microscopy. Single cells were stimulated with the purinergic receptor agonist ATP resulting in an increase of [Ca²⁺]_i due to intracellular Ca²⁺ release from inositol 1,4,5-trisphosphate (IP₃)-sensitive stores. ATP-induced Ca²⁺ release was quantal, i.e. submaximal concentrations mobilized only a fraction of the intracellularly stored Ca²⁺.

2. Focal receptor stimulation in Ca²⁺-free solution by pressure application of agonist-containing solution through a fine glass micropipette resulted in a spatially restricted increase in [Ca²⁺]_i. Ca²⁺ release was initiated at the site of stimulation and frequently propagated some tens of micrometres into non-stimulated regions.

3. Local Ca²⁺ release caused activation of capacitative Ca²⁺ entry (CCE). CCE was initially colocalized with Ca²⁺ release. Following repetitive focal stimulation, however, CCE became detectable at remote sites where no Ca²⁺ release had been observed. In addition, the rate of Ca²⁺ store depletion with repetitive local activation of release in Ca²⁺-free solution was markedly slower than that elicited by ATP stimulation of the entire cell.

4. From these experiments it is concluded that both intracellular IP₃-dependent Ca²⁺ release and activation of CCE are controlled locally at the subcellular level. Moreover, redistribution of intracellular Ca²⁺ stored within the endoplasmic reticulum efficiently counteracts local store depletion and accounts for the spatial spread of CCE activation.

Stimulation of vascular endothelial cells with the vasoactive agonist adenosine triphosphate (ATP) evokes a rise in intracellular free Ca²⁺ concentration ([Ca²⁺]_i). ATP acts through purinergic receptors to stimulate the formation of inositol 1,4,5-trisphosphate (IP₃) and subsequent Ca²⁺ release from intracellular storage sites of the endoplasmic reticulum (ER) (Chen et al. 1996; Madge et al. 1997). Depletion of the IP₃-sensitive Ca²⁺ store results in Ca²⁺ entry from the extracellular environment, which plays a key role in maintaining the ER Ca²⁺ load (Schilling et al. 1992; Madge et al. 1997). The activation mechanism of this ‘capacitative Ca²⁺ entry’ (CCE) pathway residing in the plasma membrane and its molecular identity in endothelial cells remains elusive. Models explaining the signalling mechanism which functionally links the ER Ca²⁺ store to stimulation of Ca²⁺ entry across the plasma membrane postulate either the involvement of a diffusible messenger or a direct physical link between an ER Ca²⁺-sensing protein and the plasma membrane CCE pathway (for a review see Berridge, 1995; see also Parekh & Penner, 1997; Putney, 1997; Holda et al. 1998). Recent reports of CCE in Xenopus laevis oocytes (Petersen & Berridge, 1996; Jaconi et al. 1997) have revealed a colocalization of Ca²⁺ entry with the sites of IP₃-induced Ca²⁺ release. A similar membrane-restricted regulation of the Ca²⁺ entry pathway has been reported for polarized airway epithelial cells (Paradiso et al. 1995). Moreover, CCE but not the agonist-induced Ca²⁺ release was suppressed in endothelial cells treated with cytochalasin D to disrupt the actin microfilaments (Holda & Blatter, 1997). Taken together, these findings suggest that CCE may be controlled in a subcellularly restricted fashion. Elements of the cytoskeleton might participate either directly or by providing a structural scaffold that ensures the physical proximity of the two membranes involved in the signalling mechanism.

In the present study, we have used confocal Ca²⁺ imaging and a focal agonist application system to investigate the spatial interrelation between agonist-induced Ca²⁺ release from the ER and CCE across the plasma membrane in cultured vascular endothelial cells. The results suggest that Ca²⁺ release and CCE are controlled through a subcellularly restricted mechanism which does not involve ‘long-range’ intracellular messengers. Moreover, intra-ER redistribution of stored Ca²⁺ from non-stimulated regions of the cell to the site of Ca²⁺ release prevents local store depletion and is responsible for activation of CCE at sites distant from Ca²⁺ release.

METHODS

Cell culture and solutions

Experiments were performed on single cultured CPAE vascular endothelial cells in non-confluent cultures. The CPAE cell line (American Type Culture Collection, Rockville, MD, USA) was originally derived from calf pulmonary artery endothelium. Details of the culture conditions have been published previously (Holda & Blatter, 1997; Hüser & Blatter, 1997). Cells from passages 3–6 were used for experimentation. The cells were superfused continuously with standard physiological salt solution, containing (mM): 140 NaCl, 5 KCl, 1 MgCl₂, 2 CaCl₂, 10 glucose and 10 Hepes, adjusted to pH 7.3 with NaOH. In the Ca²⁺-free solution CaCl₂ was omitted. All experiments were carried out at room temperature (20–22°C).

Ca²⁺ measurements

Endothelial cells were loaded with the fluorescent Ca²⁺ indicator fluo-3 by exposure to 5 μm of the membrane permeant acetoxymethyl ester of the dye (fluo-3 AM; Molecular Probes, Eugene, OR, USA) for 20 min at 37°C. Following transfer of the coverslip onto the stage of an inverted microscope (Axiovert 100, Carl Zeiss, Germany) equipped with a × 40 objective lens (Plan-Neofluar, oil immersion, 1.3 NA, Carl Zeiss), the cells were washed for 10 min with standard physiological salt solution to allow sufficient time for de-esterification. The microscope was attached to a confocal laser scanning unit (LSM 410, Carl Zeiss). The fluo-3 fluorescence was excited using the 488 nm line of an argon-ion laser while emitted fluorescence was recorded simultaneously at wavelengths ≥ 515 nm. In the recordings shown, changes in Ca²⁺-sensitive fluorescence were averaged over small subcellular regions of interest (ROIs), about 200–400 μm² in size. Relative changes in [Ca²⁺]_i were expressed as changes of fluo-3 fluorescence normalized to resting fluorescence (F/F₀).

For staining the ER, cells were incubated in 500 nM DiOC₆ (Molecular Probes) for 5 min and subsequently washed in standard physiological salt solution for 10 min. Fluorescence was excited at 488 nm and simultaneously recorded at wavelengths ≥ 515 nm.

Focal application system

During experimentation the cells were placed in the laminar flow from a fast solenoid-operated perfusion system allowing rapid exchange of the extracellular solution (half-time, t_0.5≡ 100 ms). For focal application of agonist (ATP)-containing solution a glass micropipette (tip diameter, 1–3 μm) was positioned close to a cell process, which was located downstream of the cell body with regard to the direction of bulk flow. The ATP solution was ejected by pressure application. Control experiments using fluorescent solutions revealed that focal stimulation at any pressure setting was restricted to cellular regions located downstream from the pipette opening. Furthermore, pressure application of agonist-free solution did not induce Ca²⁺ release or Ca²⁺ entry (not shown).

RESULTS

Ca²⁺ release and CCE induced by whole-cell stimulation

Figure 1 summarizes the characteristics of intracellular Ca²⁺ release and CCE upon stimulation of the whole cell with ATP. Prolonged stimulation of the cell with a high agonist concentration (Fig. 1A) resulted in a rapid rise of [Ca²⁺]_i due to Ca²⁺ release from IP₃-sensitive stores (Schilling et al. 1992). [Ca²⁺]_i subsequently declined to an elevated plateau level, which was determined by the rate of CCE and the counteracting cellular Ca²⁺ extrusion mechanisms (Schilling et al. 1992; Madge et al. 1997; Klishin et al. 1998). Consequently, removal of extracellular Ca²⁺ resulted in a decrease in [Ca²⁺]_i to resting levels. Subsequent readdition of Ca²⁺ to the extracellular solution caused a CCE signal with a rapid rise and a marked overshoot followed by a decrease in [Ca²⁺]_i to the previously described plateau level. Klishin and co-workers (Klishin et al. 1998) have attributed this overshoot of CCE to the delayed Ca²⁺-calmodulin-dependent activation of the plasma membrane Ca²⁺ pump. The overshoot, however, was not seen in all recordings, most probably because CCE was not always maximally activated under our experimental conditions. Figure 1B shows a recording from a cell repetitively stimulated with short (10 s) pulses of ATP (1 μm) in Ca²⁺-free solution. Because of the net cellular Ca²⁺ loss associated with each stimulation, the amplitude of the resulting [Ca²⁺]_i transients decreased progressively. Following store depletion by this protocol readdition of extracellular Ca²⁺ elicited a large CCE transient.

Figure 1. Ca²⁺ release and capacitative Ca²⁺ entry in response to global agonist stimulation.

A, superfusion with ATP (1 μm) of the whole cell in the presence of extracellular Ca²⁺ evoked a biphasic [Ca²⁺]_i transient. The initial phase was caused by Ca²⁺ release from intracellular stores, while the maintained plateau phase was primarily carried by CCE. Consequently, removal of extracellular Ca²⁺ resulted in a decrease of [Ca²⁺]_i to resting levels. Readdition of Ca²⁺ typically elicited a biphasic rise in [Ca²⁺]_i carried exclusively by CCE. B, repetitive agonist stimulation (1 μm ATP, 10 s pulses) in Ca²⁺-free solution elicited [Ca²⁺]_i transients with successively decreasing amplitudes caused by progressive store depletion. The marked rise in [Ca²⁺]_i upon readdition of extracellular Ca²⁺ indicated the activation of CCE.

The biphasic time course of the [Ca²⁺]_i transient evoked by 1 μm ATP with its pronounced plateau phase in the presence of extracellular Ca²⁺ (Fig. 1A) was the first indication that this agonist concentration maximally stimulated Ca²⁺ release. To further test the concentration dependence of intracellular Ca²⁺ mobilization by the purinergic agonist we superfused endothelial cells with increasing ATP concentrations in Ca²⁺-free solution. The result is shown in Fig. 2. The ensemble trace (Fig. 2, top trace) which represents the mean fluorescence changes of 20 cells subjected to the same experimental protocol revealed that submaximal ATP concentrations did not mobilize all the releasable Ca²⁺. This phenomenon which is commonly referred to as quantal Ca²⁺ release (Muallem et al. 1989) has been observed in many electrically non-excitable cells (for a review see Bootman et al. 1994; Parys et al. 1996). The concentration dependence for ATP was steep, with a threshold concentration for Ca²⁺ release of about 150 nM (not shown in this set of recordings). In most cells (90 %) superfusion with 1 μm ATP resulted in the complete discharging of the agonist-sensitive intracellular Ca²⁺ pool. Further increase of the agonist concentration was without any effect on [Ca²⁺]_i.

Figure 2. Concentration dependence of ATP-induced Ca²⁺ release.

Endothelial cells were superfused with increasing concentrations of ATP in Ca²⁺-free solution. The top trace shows the mean fluorescence ([Ca²⁺]_i) changes obtained from 20 cells subjected to the same experimental protocol. The 3 lower traces were derived from individual cells. Very low ATP concentrations caused single [Ca²⁺]_i spikes or locally restricted release events (marked by asterisks). Submaximal ATP concentrations (0.2–0.4 μm) released only a fraction of the total intracellularly stored Ca²⁺. Typically, Ca²⁺ release evoked by intermediate agonist concentrations was oscillatory. A further step increase in [ATP] resulted in an additional [Ca²⁺]_i transient or enhanced oscillations. Maximally stimulatory agonist concentrations (≥1μM) evoked only a single slowly declining [Ca²⁺]_i transient.

Focal agonist stimulation triggers subcellularly restricted Ca²⁺ release

In the experiment depicted in Fig. 3 agonist stimulation was confined to one end of a narrow cell process (Fig. 3A, position 1). We used 1 μm ATP for local stimulation rather than a supramaximal agonist concentration to ensure that Ca²⁺ release remained confined to the stimulated region. Furthermore, 1 μm ATP should mobilize most or all intracellular Ca²⁺ at the site of stimulation (see Fig. 2) resulting in a marked and rapid depletion of intracellular Ca²⁺ stores. Purinergic stimulation rapidly initiated Ca²⁺ release at the site of stimulation (Fig. 3B, trace 1). Ca²⁺ release subsequently extended some 10 μm (Fig. 3B, trace 2) into non-stimulated regions but failed to propagate further than ∼60 μm (traces 3 and 4). It should be noted that the spatio-temporal pattern of Ca²⁺ release elicited by focal agonist stimulation critically depended on the geometry of the cell and the subcellular region subjected to stimulation. Spatially restricted Ca²⁺ release could be induced most reliably when stimulating at narrow cell processes. In most cells release failed to propagate beyond the point where the cell process fused into the larger cell body. In contrast to global agonist stimulation in Ca²⁺-free solution (Fig. 1B), repetitive Ca²⁺ release transients could be evoked by focal ATP stimulation without pronounced signs of store depletion. In the example shown in Fig. 3C (trace 1), the amplitude of the 15th ATP-induced [Ca²⁺]_i transient was still about 50 % that of the 1st transient. Thus, a mechanism must exist which attenuates store depletion when Ca²⁺ release is confined to a small subcellular region. A probable mechanism accounting for the continuous replenishment of Ca²⁺ stores is the redistribution of intracellularly stored Ca²⁺ ions within the ER from non-excited regions to the site where Ca²⁺ release and Ca²⁺ extrusion across the plasma membrane had occurred. Because in these particular experiments the cellular volume affected by focal agonist-induced Ca²⁺ release was small compared with the total cell volume, repetitive Ca²⁺ release resulted only in moderate signs of local store depletion. Finally, readdition of extracellular Ca²⁺ following repetitive focal ATP stimulation (Fig. 3C) resulted in a clearly detectable CCE signal at the site of stimulation (trace 1) and at a distant site where no Ca²⁺ release was observed (trace 4). The differences in the rate of rise of the CCE signal were most probably due to differences in the regional surface-to-volume (s/v) ratio, i.e. the rate of rise of the Ca²⁺ entry signal is higher in regions with a larger s/v ratio (Hüser & Blatter, 1997; Holda et al. 1998). These differences were most pronounced when comparing Ca²⁺ entry signals from regions located in the cell periphery or a fine cell process with those recorded in the cell body or in the perinuclear space.

Figure 3. Focal agonist stimulation results in spatially restricted Ca²⁺ release.

A, contour image of the cell. The shaded area represents the nuclear and perinuclear space. Focal ATP stimulation was confined to the end of a narrow cell process (position 1). Positions 2–4 are successively greater distances from the focal stimulation point into the main body of the cell. B, focal stimulation triggered Ca²⁺ release starting at the site of stimulation (trace 1). The region of elevated [Ca²⁺]_i subsequently expanded in a wave-like fashion towards more central regions of the cell (trace 2). No release of Ca²⁺ could be detected at distances ≥ 75 μm (traces 3 and 4) from the site of stimulation. C, series of [Ca²⁺]_i transients evoked by repetitive focal stimulation in Ca²⁺-free solution (trace 1). Notably, locally restricted Ca²⁺ release was less affected by store depletion when compared with global agonist stimulation (Fig. 1B). Note no detectable Ca²⁺ release at position 4.

The data presented in Fig. 3 demonstrate that CCE can be activated in subcellular regions that do not show intracellular Ca²⁺ release. During the period of repetitive local agonist stimulation (17 min), however, the whole cell was superfused with Ca²⁺-free solution. It was likely that during this period stores in non-stimulated regions lost some Ca²⁺ due to agonist-independent leakage of Ca²⁺ from the intracellular store. To test whether such passive Ca²⁺ leakage from intracellular Ca²⁺ stores can account for the time-dependent activation of CCE at remote sites, we measured the amount of Ca²⁺ releasable by ATP (10 μm) as a function of time of preincubation in Ca²⁺-free solution (Fig. 4). The results revealed that in endothelial cells the passive discharge of the agonist-sensitive intracellular Ca²⁺ store was very slow. Furthermore, CCE was not detectable as a change of [Ca²⁺]_i in cells incubated for less than 90 min in Ca²⁺-free solution (not shown here). It should be noted here that the sensitivity of [Ca²⁺]_i measurements to CCE is limited. The rate of CCE has to be sufficiently high to overcome cellular Ca²⁺ buffering and ER reuptake to be detected by the fluorescence indicator. Therefore failure to record CCE signals at times ≤ 90 min does not necessarily indicate the absence of Ca²⁺ entry. The experiment demonstrates, however, that passive Ca²⁺ leakage from the ER alone was not sufficient to account for a significant activation of CCE in cells superfused with Ca²⁺-free solution for periods of less than 20 min.

Figure 4. Passive store depletion by Ca²⁺ leakage.

Endothelial cells were superfused with Ca²⁺-free solution and stimulated with 10 μm ATP to release Ca²⁺ from the entire agonist-sensitive Ca²⁺ pool. The plot of the mean amplitude (each point represents the mean ±s.d. of 4–8 cells) of the resulting [Ca²⁺]_i transient vs. the time of preincubation in Ca²⁺-free solution revealed only a slow loss of releasable Ca²⁺ from intracellular Ca²⁺ stores in the absence of agonist. [Ca²⁺]_i has been calculated according to the formula: [Ca²⁺]_i=K_d (F/F₀)/(K_d/[Ca²⁺]_rest - (F/F₀) + 1), where K_d= 1 μm and [Ca²⁺]_rest= 50 nM.

Colocalization of CCE with Ca²⁺ release

From the experiments presented it became clear that CCE could be activated in subcellular regions where no previous Ca²⁺ release had occurred. Moreover, the distance between sites of intracellular Ca²⁺ release (and concomitant Ca²⁺ loss in Ca²⁺-free solution) and sites of capacitative Ca²⁺ entry could be large suggesting that locally restricted Ca²⁺ depletion could lead to global activation of CCE across the entire plasma membrane. In the experiment shown in Fig. 5 conditions were chosen to further resolve the spatial and temporal dependence of CCE on agonist-induced intracellular Ca²⁺ release. As in previous experiments, focal ATP stimulation was restricted to a cell process (Fig. 5, inset, position 1). The resulting Ca²⁺ release propagated to the point where the cell process fused into the larger cell body. No Ca²⁺ release was observed at a site ∼100 μm away (position 2) from the region showing Ca²⁺ release. Readdition of extracellular Ca²⁺ following the first short ATP pulse revealed a clear CCE transient in the cell process (Fig. 5, trace 1). No appreciable CCE signal, however, could be recorded at the distant unstimulated site (Fig. 5, trace 2). Importantly, repeating this protocol caused CCE to become visible and progressively larger at the remote site where at no point in time had any Ca²⁺ release occurred. Thus, while CCE was initially colocalized with Ca²⁺ release in the cell process, repetitive local Ca²⁺ release caused activation of CCE at sites not participating in IP₃-induced Ca²⁺ release. Initial colocalization of CCE with Ca²⁺ release was observed in all five cells subjected to comparable stimulation protocols. The lack of changes in [Ca²⁺]_i at the remote site during agonist stimulation suggested that a possible mechanism for the slow spatial spread of CCE from the site of Ca²⁺ release/loss to non-excited regions might be provided by the redistribution of Ca²⁺ within the ER Ca²⁺ store.

Figure 5. Colocalization of CCE with Ca²⁺ release.

Comparison of the time course of [Ca²⁺]_i at the cell process (inset, position 1) and a remote site (inset, position 2) not participating in intracellular Ca²⁺ release. Following the first ATP pulse (1 μm for 10 s) in Ca²⁺-free solution, readdition of extracellular Ca²⁺ revealed activation of CCE at the site of Ca²⁺ release (trace 1). At this time, no Ca²⁺ entry could be detected at the remote site. With repetitive focal stimulation of Ca²⁺ release, however, CCE became progressively visible at the distant non-stimulated site (trace 2).

Staining of endothelial cells with the carbocyanine dye DiOC₆, which has been used to stain intracellular endo- and sarcoplasmic Ca²⁺ stores in various cell types (e.g. Terasaki et al. 1985; Blatter, 1995; Golovina & Blaustein, 1997), revealed heteromorphic intracellular structures (Fig. 6). Image resolution was best in the cell periphery, where the cell thickness is likely to be less than one confocal section. Besides the round or ellipsoid structures (Fig. 6B), DiOC₆ labelled a structure of narrow interconnected tubules. Because of their shape and higher fluorescence intensity the non-reticular structures were most probably mitochondria. The reticular structure closely resembled the ER morphology reported for example in kidney cells (Terasaki et al. 1985; Lee et al. 1989). The interconnection of the ER tubules supports the notion of a continuous intracellular Ca²⁺ storage organelle and provides the structural basis for a Ca²⁺ redistribution process within the ER lumen.

Figure 6. Visualization of the endoplasmic reticulum with DiOC₆.

The carbocyanine dye DiOC₆ stained heteromorphic structures in vascular endothelial cells. The images show detailed subcellular sections from the perinuclear region (left) and the cell periphery (right) at high electronic magnification (average of 8 consecutively recorded frames). The vesicular structures in the cell periphery most probably represent mitochondria, since their fluorescence intensity markedly exceeds the staining of the reticular structure. In addition, similar structures were labelled when tetramethylrhodamine ethyl ester (TMRE), which selectively stains mitochondria, was applied (not shown). The ER was visible as a network of narrow interconnected tubules. Scale bars represent 5 μm.

DISCUSSION

The experiments presented in this paper used spatially resolved confocal [Ca²⁺]_i measurements together with focal agonist stimulation to study the subcellular regulation of intracellular Ca²⁺ release and capacitative Ca²⁺ entry in vascular endothelial cells. The main results of this study were threefold. First, focal agonist stimulation at a small cell process caused spatially restricted release of Ca²⁺ from intracellular stores. Release frequently propagated some tens of micrometres into non-stimulated regions. Second, Ca²⁺ store depletion by repetitive focal stimulation of a small subcellular region had a slower time course than store depletion caused by stimulation of the entire cell. Third, CCE was initially restricted to those subcellular regions where Ca²⁺ release had occurred previously. Following repetitive local release and concomitant Ca²⁺ extrusion at the site of release, however, capacitative Ca²⁺ entry became detectable at remote sites, which did not participate in Ca²⁺ release.

The purinergic agonist ATP binds to surface membrane receptors to stimulate the membrane bound intracellular formation of IP₃ by phospholipase C (Madge et al. 1997). IP₃ is a diffusible messenger that binds to the IP₃ receptor in the ER membrane to initiate Ca²⁺ release into the cytosol (Berridge, 1997). Confining agonist stimulation to a small subcellular region by focal agonist application is expected to result in marked subcellular [IP₃] gradients. IP₃ diffusion away from its site of synthesis most probably accounted for the spread of Ca²⁺ release into regions excluded from extracellular stimulation in our experiments. The range of action of IP₃, which is determined by its diffusion coefficient and its rate of degradation, has been estimated to be about 17 μm (Kasai & Petersen, 1994). This value, however, should vary significantly with cell geometry. In particular, when stimulating a narrow cell process with a high s/v ratio, the resulting rise in [IP₃] will be faster and reach higher concentrations than in the cell body (compare Schaff et al. 1997). As a result the diffusional spread of IP₃ is faster and is expected to extend over larger distances in the narrow processes than in the cell body. It is therefore conceivable that the diffusion of IP₃ from its site of formation underlies the observed propagation of Ca²⁺ release of about 40–80 μm into non-stimulated regions. [Ca²⁺]_i waves most frequently failed to propagate beyond the point where the cell process fused into the larger cell body. During whole-cell agonist stimulation [Ca²⁺]_i waves also started in the cell processes before propagating towards the centre of the cell. Under these conditions, the wave propagation velocity abruptly slowed down at the points where the narrow process merged with the large volume cell body (Hüser & Blatter, 1997).

Agonist-induced Ca²⁺ release in Ca²⁺-free solution was associated with a pronounced depletion of intracellular Ca²⁺ stores. The degree of depletion depended on the agonist concentration. Submaximal concentrations released only a fraction of the intracellularly stored Ca²⁺. Thus agonist- induced Ca²⁺ release in vascular endothelial cells displayed quantal behaviour typical for many non-excitable cells (Parys et al. 1996). The ATP concentration used for local and global stimulation (1 μm) in our experiments was high enough to ensure local release of the major fraction of stored Ca²⁺ and subsequent Ca²⁺ extrusion. Depletion of the Ca²⁺ store, in turn, resulted in the activation of CCE (Fig. 1B; see also Schilling et al. 1992). The rate of Ca²⁺ loss from the ATP-sensitive intracellular store during superfusion with Ca²⁺-free solution in the absence of agonist stimulation was very slow (Fig. 4). In fact, after 90 min preincubation in Ca²⁺-free medium the ATP-induced [Ca²⁺]_i transient was reduced by only 50 %. At this point Ca²⁺ entry was not detectable as a change of [Ca²⁺]_i upon readdition of extracellular Ca²⁺. Ca²⁺ extrusion and the resulting store depletion were markedly facilitated during agonist stimulation. The rate of store depletion with repetitive agonist stimulation, however, was considerably less when stimulation was restricted to a small cell process. Therefore, a mechanism must exist which compensates for the local loss of stored Ca²⁺. An attractive candidate for such a mechanism is the intraluminal redistribution of stored Ca²⁺ within the ER. The visualization of the ER as narrow interconnected tubules (Fig. 6; compare Lee et al. 1989) supports the notion that the ER forms a continuous intracellular membrane network. Moreover, the ER continuity has been assessed functionally using fluorescence recovery after photobleaching (FRAP) in basophilic leukaemia cells expressing a luminal 60 kDa protein conjugated with the green fluorescent protein (Subramanian & Meyer, 1997). The FRAP experiments revealed a high mobility of the protein within the ER lumen with an apparent diffusion coefficient of ∼0.5 μm² s⁻¹. Thus local Ca²⁺ depletion can be effectively counteracted by redistribution of stored Ca²⁺ within the ER. Thus, locally restricted Ca²⁺ release and depletion would eventually result in global store depletion (Fig. 3). Again, the slow rate of passive store depletion ruled out the possibility that incubation of the cell in Ca²⁺-free solution alone accounts for the time-dependent activation of CCE at non-stimulated sites.

Mogami and co-workers (Mogami et al. 1997) have presented evidence that Ca²⁺ entry and reuptake into the Ca²⁺ store restricted to the basolateral membrane of pancreatic acinar cells resulted in refilling of Ca²⁺ stores at the apical pole of the cell. Because the refilling of apical stores was not accompanied by any detectable increase in cytosolic Ca²⁺ concentration, these authors concluded that Ca²⁺ ions were transported from the site of uptake into basolateral stores via ‘tunnels’ to release sites at the opposite pole of the cell. Similarly, the redistribution of stored Ca²⁺ in our experiments was not accompanied by any detectable change in [Ca²⁺]_i. Such a mechanism necessitates a continuous ER-Ca²⁺ store allowing effective intraluminal diffusion of Ca²⁺. Generally, the ER-Ca²⁺ store appears to be physically continuous. Intraluminal Ca²⁺ diffusion tends to equilibrate concentration gradients within the store. The rate of redistribution, however, is slow when compared with the rate of Ca²⁺ release and/or uptake. Therefore, segregation of release channel proteins (IP₃ or ryanodine receptors) and Ca²⁺ pumps into discrete subcellular ER regions may account for the functional compartmentalization of cytosolic and intraluminal [Ca²⁺] directly visualized in imaging experiments (e.g. Button & Eidsath, 1996; Golovina & Blaustein, 1997; Berridge, 1997; Hüser & Blatter, 1997).

As pointed out above, the intraluminal redistribution of stored Ca²⁺ provides a means for the spatial spread of store depletion. This mechanism would account for the activation of CCE at a site where no Ca²⁺ release preceded Ca²⁺ entry (Figs 3 and 5). Following readdition of extracellular Ca²⁺ to probe for activation of CCE after a single focal agonist stimulation (Fig. 5) the CCE-mediated rise in [Ca²⁺]_i strictly colocalized with the site of Ca²⁺ release. A similar colocalization of CCE with Ca²⁺ release has been reported for Xenopus laevis oocytes (Petersen & Berridge, 1996; Jaconi et al. 1997). Although these experiments do not exclude the participation of a diffusible intracellular messenger in the activation of CCE, they suggest that the effective range of such a factor does not exceed that of IP₃. Both Ca²⁺ release induced by IP₃ and activation of CCE are controlled locally at the subcellular level. In this model, local store depletion would lead to activation of CCE only in plasma membrane areas in the proximity of the store. Furthermore, disruption of the structural arrangement of the participating signalling components by degradation of cytoskeletal elements has been shown to selectively suppress CCE but not agonist-induced Ca²⁺ release (Holda & Blatter, 1997). Thus, the spatial range of the CCE-activating signal is clearly more restricted than the receptor-induced intracellular IP₃ signal. In addition, a possible direct involvement of the cytoskeleton in the CCE signalling mechanism presents an attractive hypothesis which, however, needs further experimental support. It should be noted that the requirement of an intact cytoskeleton for CCE activation may be cell type-specific, since in NIH 3T3 fibroblasts disruption of actin microfilaments exerted opposite effects (Ribeiro et al. 1998). According to these authors, CCE in cytochalasin D-treated cells remained functional but IP₃-dependent Ca²⁺ release was abolished.

In summary, subcellularly localized activation of the inositolphosphate metabolism by focal application of the purinergic receptor agonist ATP resulted in spatially restricted Ca²⁺ release in vascular endothelial cells. Due to the diffusional spread of IP₃, Ca²⁺ release expanded some tens of micrometres into non-stimulated regions. Local store depletion caused activation of CCE that was initially restricted to the site of Ca²⁺ release. Furthermore, subsequent redistribution of stored Ca²⁺ within the ER-Ca²⁺ store from non-stimulated regions to sites of local depletion served to replenish the Ca²⁺ store in the stimulated region and to spread the signal for CCE activation throughout the cell.