Spacial compartmentalization of Ca2+ signaling complexes in pancreatic acini.

Imaging [Ca2+]i at high temporal resolution and measuring the properties of Ca2+ signaling in streptolysin O (SLO)-permeabilized cells were used to study the spacial organization of signaling complexes. Sequential stimulation of single cells within pancreatic acini with several Ca2+-mobilizing agonists revealed an agonist-specific pattern and propagation rate of Ca2+ waves in the same cells, with CCK8 stimulating the fastest and bombesin the slowest waves. More importantly, each agonist initiated the wave in a different region of the same cell. On the other hand, repetitive stimulation with the same agonist induced Ca2+ waves of the same pattern that were initiated from the same region of the cell. The agonist-specific Ca2+ signaling does not appear to be the result of coupling to different G proteins as infusion of an anti-Gαq antibody into the cells through a patch pipette equally inhibited Ca2+ signaling by all agonists. Further evidence for compartmentalization of signaling complexes was developed in permeabilized cells. The time-dependent loss of Ca2+ signaling due to SLO permeabilization occurred in an agonist-specific manner in the sequence cabachol > bombesin > cholecystokinin. Signaling by all agonists could be completely restored with as low as 2 μM guanosine 5′-3-O-(thio)triphosphate (GTPγS). At this low concentration GTPγS recoupled inositol 1,4,5-trisphosphate production and Ca2+ release, rather than enhancing phospholipase C activity. Priming of Ca2+ signaling by GTPγS was agonist-specific. Guanosine 5′-O-(thio)diphosphate (GDPβS) uncoupled the ability of signaling complexes to release Ca2+ much better than stimulating inositol 1,4,5-trisphosphate production. The uncoupling of Ca2+ signaling by GDPβS was also agonist-specific. The combined findings of agonist-specific initiation sites of the Ca2+ wave and differential access of guanine nucleotides to signaling complexes suggest spacial compartmentalization of Ca2+ signaling complexes. Each complex must include a receptor, G protein, and phospholipase C that are coupled to a specific portion of the Ca2+ pool.

Transduction of the Ca 2ϩ signal can occur through G proteincoupled or G protein-independent receptor complexes (1,2). The G protein-independent complexes are proteins containing SH2 and SH3 domains which couple plasma membrane-local-ized receptors to the ␥ isoform of phospholipase C (PLC). 1 The G protein-dependent receptor complexes use heterotrimeric G proteins from the Gq/11 family that couples plasma membrane receptors to the ␤ isoform of PLC (1)(2)(3)(4). Like other heterotrimeric G proteins, Gq/11 is composed of G␣ and G ␤␥ subunits (2,4). Both subunits can activate various PLC␤ isoforms (4 -10) to potentially transduce different types or patterns of Ca 2ϩ signaling. Additional diversity of the system is provided by the ability of a single receptor to couple to several combinations of G␣ and G␤␥ (11)(12)(13)(14) to transduce different information.
A central question in recent years is how the diversity in signal transduction is manifested while maintaining specificity in cellular signaling. Many cells express a combination of signaling pathways and several receptors that activate the same signaling pathway (4). In the case of the exocrine pancreas, acinar cells respond to cholecystokinin (CCK), acetylcholine, and bombesin (BS) (15) by activation of PLC to generate IP 3 and mobilize Ca 2ϩ from internal stores. How the different Ca 2ϩ signaling complexes are organized and their activity coordinated is not known. Compartmentalization of signaling complexes and their communication with specific portions of the Ca 2ϩ pool can be used to achieve signal specificity and control of activation. Indeed, recently we inferred such compartmentalization from the aberrant behavior of antagonists acting on streptolysin O-permeabilized pancreatic acini stimulated with GTP␥S (16). To obtain further evidence for compartmentalization of signaling complexes, we measured the patterns of the [Ca 2ϩ ] i changes evoked by several Ca 2ϩ -mobilizing agonists and their sensitivity to G␣q/11 antibodies in the same intact single cells. We also used SLO-permeable cells to determine the sensitivity of the signaling complexes to GTP␥S and GDP␤S. The clear differences observed with the various agonists show that agonist-specific [Ca 2ϩ ] i signaling can be attributed to spacial compartmentalization of signaling complexes.

MATERIALS AND METHODS
Preparation of Pancreatic Acini and Single Acinar Cells-Acini were prepared from the pancreas of 100 -150-g rats by limited collagenase digestion as described previously (17). The acini were resuspended in a standard solution A containing (in mM) 140 NaCl, 5 KCl, 1 MgCl 2 , 1 CaCl 2 , 10 Hepes (pH 7.4 with NaOH), 10 glucose, and 0.1% bovine serum albumin, and kept on ice until use for [Ca 2ϩ ] i measurement in intact cells or Ca 2ϩ release in permeable cells. Single acinar cells were obtained by incubation of a minced pancreas in an 0.025% trypsin, 0.02% EDTA solution for 5 min at 37°C. After washing the tissue with solution A supplemented with 0.02% soybean trypsin inhibitor, single cells were liberated by a 7-min incubation at 37°C in the same solution that also contained 160 units/ml pure collagenase. The cells were washed with solution A and kept on ice until use. kept on ice until plating on glass coverslips that formed the bottom of a perfusion chamber. The plated cells were incubated for 2 min at room temperature to allow attachment to coverslips and then perfused at a rate of 10 -12 ml/min for at least 10 min at 37°C prior to the first stimulation. Acinar cell clusters consisting of three to five cells and forming a single layer in the bottom of the coverslip were selected for experimentation. Fura 2 fluorescence was measured by a PTI Int. image acquisition and analysis system as described elsewhere (18). To maximize the temporal resolution, fluorescence was measured at a single excitation wavelength of 380 nm and without image averaging. The light was directed by a dichroic mirror to an image intensifier and a video camera. The images were digitized and stored in CASH memory for further analysis. Under these conditions using a frame size of 256 ϫ 240 pixels allowed recording the fluorescence of the entire cluster at a resolution of 2.0 Hz. A resolution of 4.7 Hz was obtained by recording the fluorescence from one cell within the cluster which could be fitted in a frame size of 128 ϫ 120 pixels. During the perfusion with the control solution and just before the first stimulation, the image of the resting cells was acquired by averaging eight consecutive images. This was taken as the fluorescence at time 0 (F 0 ). All subsequent images were divided by this image, and the traces in Figs. 1-3 are the calculated Measurement of Ca 2ϩ Release in Permeable Cells-Agonist-mediated Ca 2ϩ release from SLO-permeabilized pancreatic acini was measured exactly as described previously (16,19). The SLO permeabilization medium was composed of Chelax-treated 145 mM KCl and 10 mM Hepes solution and supplemented with 0.02% soybean trypsin inhibitor, 3 mM ATP, 5 mM MgCl 2 , 10 mM creatine phosphate, 5 units/ml creatine phosphokinase, 10 M antimycin A, 10 M oligomycin, 1 M Fluo 3, and 3 mg/ml SLO (Difco). Fluo 3 fluorescence was recorded and calibrated as detailed previously (19).
Measurement of IP 3 -IP 3 levels were measured by a radioligand assay as described elsewhere (19). Acini washed with the Chelaxtreated solution and incubated in the SLO permeabilization medium with or without GDP␤S, were stimulated with the indicated agonist for 20 s at 37°C. The reactions were stopped by the transfer of 100-l samples to 100 l of an ice-cold 15% perchloric acid, vigorous mixing, and incubation on ice for at least 10 min to allow precipitation of proteins. The supernatants were then collected by 2-min centrifugation at 10,000 ϫ g and transferred to clean tubes. Standards of IP 3 were prepared in permeabilization medium and processed in the same manner. The perchloric acid was removed and IP 3 extracted by the addition of 0.2 ml of Freon and 0.2 ml of tri-n-octylamine. After mixing and phase separation by centrifugation, between 15 and 25 l of the upper layer were used to measure IP 3 content with microsomes prepared from bovine brain cerebella.
Whole-cell Current Recording-The tight-seal whole-cell current recording of the patch clamp technique was used (20) for measurement of Ca 2ϩ -activated Cl Ϫ current, which directly correlates with changes in [Ca 2ϩ ] i (21). The experiments were performed with single acinar cells perfused with solution A. The standard pipette solution contained (in mM): KCl 140, MgCl 2 1, EGTA 0.2, Na 2 ATP 5, and Hepes 10 (pH 7.3 with KOH) as described in previous studies (21). In some experiments this solution also contained either 80 g/ml preimmune serum or IgG anti-G␣q/11 antibodies, both of which were a generous gift from Dr. Paul Sternweis, University of Texas Southwestern Medical Center, Dallas (22). Seals of 6 -10 gigohms were produced on the cell membrane, and the whole cell configuration was obtained by gentle suction or voltage pulses of 0.5 V for 0.3-1 ms. The patch clamp output (Axopatch-1B, Axon Instruments) was filtered at 20 Hz. Recording was performed with patch clamp 6 and a Digi-Data 1200 interface (Axon Instruments). The cells were voltage clamped at Ϫ30 mV, and depolarizing voltage jumps of 100-ms duration at a frequency of 1 Hz to a membrane potential of ϩ10 mV were applied throughout the experiments to measure the leak current. In all experiments the Cl Ϫ and cation equilibrium potentials were about 0 mV. All the traces shown were at a holding potential of Ϫ30 mV and were corrected for leak currents.

RESULTS
Agonist Specific Initiation Sites of Ca 2ϩ Release-Pancreatic acini express several distinct receptors for Ca 2ϩ mobilizing agonists in the basolateral membrane, which stimulate the exocytotic secretion of digestive enzymes through the apical membrane (15). The most potent Ca 2ϩ -mobilizing agonists are acetylcholine, bombesin, and CCK, all of which cause a large increase in IP 3 to mobilize all the agonist-sensitive intracellu-lar Ca 2ϩ pool (15,(23)(24)(25)(26). Mobilization of Ca 2ϩ from internal stores by one of these agonists, acetylcholine, was shown to be initiated in the apical and spread to the basolateral pole of the cell (27)(28)(29). To determine whether all agonists acting on acinar cells initiate the Ca 2ϩ signal at the same site, we compared the effects of the three agonists on [Ca 2ϩ ] i of the same cells. All the experiments were performed on clusters of three to five cells to ensure the maintenance of intact tight junctions and cellular polarity, and then [Ca 2ϩ ] i was analyzed in single cells within the cluster. Fig. 1 shows the experimental protocol used. The cells were stimulated with supermaximal concentration of agonists to mobilize all the Ca 2ϩ in the intracellular stores. Shortly after stimulation, the cells were inhibited with the relevant antagonist to force Ca 2ϩ back into the intracellular stores and cause rapid and maximal reloading (25,30,31). In the first set of experiments images were collected from the entire acinus at a resolution of 2.0 Hz, which was sufficient to determine the rate of the Ca 2ϩ wave. The upper panel of Fig To further analyze the differences between the agonists, the same protocol was used but now [Ca 2ϩ ] i was imaged in one cell and only during the rising phase at a resolution of 4.7 Hz (maximal possible with our recording system). corded at a magnification of 1000 ϫ and at an excitation wavelength of 380 nm. The top left image was recorded under resting conditions. The second row of images shows the first detected [Ca 2ϩ ] i increase when the same cell was stimulated with carbachol (top), bombesin (middle), or CCK8 (bottom). The next row shows images for each agonist when the [Ca 2ϩ ] i spread through about one-half of the cell. The row on the right shows the first images when maximal Ca 2ϩ increase was achieved for each agonist, which demonstrates that each agonist was able to access the entire cellular Ca 2ϩ pool. It is clear that in the experiment shown each agonist initiated the [Ca 2ϩ ] i increase in different areas of the cell. In five of seven similar experiments, all three, and in two of seven experiments, two of the three, agonists initiated the [Ca 2ϩ ] i increase in different areas of the same cell. Since the spread of the Ca 2ϩ wave (see above) is significantly slower than the recording resolution, it is unlikely that the agonist-specific initiation sites were due to inadequate spacial resolution. Agonist-specific initiation sites were frequently observed when the recording resolution was under the condition of Fig. 1. Under this condition the fluorescence of the entire cell cluster could be recorded and again in 34 cells from nine clusters at least two of the three agonists initiated the [Ca 2ϩ ] i increase in different areas of the same cell.
Another difference between the agonists is the pattern of the Ca 2ϩ wave as illustrated in the traces in the bottom panels of Fig. 2. The traces show the pattern of [Ca 2ϩ ] i changes during the first 1-2 s of cell stimulation for each agonist. The carbachol-induced [Ca 2ϩ ] i wave propagated from areas 1 to 2, 3, 4, and 5. The bombesin-induced wave appears to progress simultaneously in two directions increasing first at area 2 and then at areas 1 and 3 at about the same time and then at areas 4 and 5. The CCK8-induced wave was almost in the opposite direction from that of carbachol, starting at area 4 and rapidly progressing through areas 3, 2, 1, and then 5. Similar agonist-specific patterns were recorded in the experiments mentioned above that were designed to identify the initial site of [Ca 2ϩ ] i increase for each agonist. As reported before for acetylcholine (27)(28)(29), carbachol and bombesin always initiated the [Ca 2ϩ ] i increase in the apical region (marked with A in bright-field images) and the wave tended to spread through the cell periphery to the basal region (marked with B). In the case of CCK8 the [Ca 2ϩ ] i increase was initiated either in the apical (Fig. 1) or basolateral (Fig. 2) regions, although in most experiments (ϳ76%) it started in the apical region of the cell.
An important control experiment in assigning a specific portion of the pool to a Ca 2ϩ -mobilizing agonist is to demonstrate that repetitive stimulation with the same agonist initiates Ca 2ϩ waves of the same pattern and from the same region of the cytoplasm. To demonstrate this behavior, we selected carbachol, since stimulation by this agonist can be readily reversed with a very low concentration of atropine, and atropine can then be easily washed to allow repetitive stimulations. 3 shows the results of such an experiment. The acinus (top image) was stimulated once with maximal and twice with supermaximal concentrations of carbachol. At the [Ca 2ϩ ] i peak, stimulation was terminated with 2 nM atropine, and then atropine was washed for at least 5 min before a second stimulation was attempted. Recording and analysis of [Ca 2ϩ ] i changes in different regions of one of the cells (bottom image) clearly showed that, on repetitive stimulation, carbachol always increased [Ca 2ϩ ] i first at region 1 of the apical pole and then the wave propagated in the direction 1, 2, 3, 4, and 5. In addition, the rate of wave propagation was very similar at each stimulation. Identical results in terms of the initiation site and wave pattern were obtained in four additional experiments. At submaximal agonist concentrations that still accessed the entire Ca 2ϩ pool, the rate of wave propagation was slower than at maximal agonist concentration, but the initiation site and wave pattern were identical at all agonist concentrations (not shown, n ϭ 4). Hence, in all experiments performed with no exception, repetitive stimulation with the same agonist resulted in wave initiation from the same site.
Coupling to G␣q-The agonist-specific initiation, pattern, and speed of the [Ca 2ϩ ] i wave can be due to spacial compartmentalization of signaling complexes and/or coupling to different ␣ subunits of the heterotrimeric G proteins. We examined to some extent the role of G␣ subunits in conferring specificity by determining the sensitivity of Ca 2ϩ signaling by the various agonists to inhibition of the action of G␣q/11. Fig. 4 shows the results of such experiments. An IgG anti-G␣q/11 was infused into the cell through a patch pipette, and [Ca 2ϩ ] i was followed by measuring the activity of the Ca 2ϩ -activated Cl Ϫ channel. This technique was extensively used to follow [Ca 2ϩ ] i changes next to the plasma membrane of various secretory cells (21). Preliminary experiments showed that when the pipette solution contained between 80 -240 g/ml IgG anti-G␣q/11 (the highest concentration tested) maximal inhibition of Ca 2ϩ sig-naling could be achieved. Concentrations of the antibody between 80 and 240 g/ml affected only the time of dialysis required to achieve maximal inhibition. In Fig. 3, 80 g/ml IgG was used. Trace b shows an experiment in which less than maximal inhibition was achieved. The trace is shown to demonstrate that the G␣q/11 antibody inhibited the signal induced by the three agonists to a similar extent. The effect of carbachol was inhibited by 82%, that of bombesin by 86%, and that of CCK8 by 87%. In trace c the anti-G␣q/11 antibody inhibited the effect of all the agonists by more than 95%. The antibody inhibited Ca 2ϩ signaling by all agonists and to a similar extent in all 17 experiments tested. Thus, it appears that all three receptors are coupled to G␣q.
Compartmentalized Signaling Complexes in Permeable Cells-Spacial compartmentalization of signaling complexes was supported by testing the time-dependent effect of cell permeabilization with SLO toxin on Ca 2ϩ signaling by the three agonists. As we reported before (16,19), after permeabilization with SLO, pancreatic acinar cells retain intact signaling systems, including the stimulated generation of IP 3 and Ca 2ϩ release from intracellular stores. However, continuous exposure to SLO resulted in a time-dependent loss of signaling, which was agonist-specific. Fig. 5 shows individual experiments with carbachol and CCK8, whereas Fig. 6 (closed symbols) summarizes the results of three to six experiments with all agonists. The ability of carbachol to signal Ca 2ϩ release was lost within 10 min of exposure to SLO. The lose of signaling by bombesin was slower with 60% inhibition after 10-min incubation with SLO, while CCK8 signaling was quite resistant to permeabilization by SLO with only about a 20% loss of signaling after 10 min of permeabilization.
The loss of Ca 2ϩ signaling could be completely restored by the addition of very low concentration of GTP␥S. Fig. 7 shows individual experiments in which the cells were stimulated with carbachol or CCK8 and 2 M GTP␥S after 2 or 10 min of permeabilization with SLO. Fig. 6 (open symbols) summarizes the results obtained with all agonists after 10 min permeabilization. It is clear that the low concentration of GTP␥S was able to restore maximal signaling by all agonists. G proteins can affect Ca 2ϩ signaling either by regulating PLC and IP 3 production (4), or by directly regulating the IP 3 -activated Ca 2ϩ chan- nel (32). To distinguish between the two possibilities we measured the effect of SLO permeabilization on IP 3 production. Table I shows that after 2-min permeabilization with SLO all agonists increase IP 3 levels to the same extent and 2 M GTP␥S had no effect on IP 3 production. A more significant finding is that after 10-min permeabilization IP 3 production by all agonist was equally reduced by about 25-30% and GTP␥S did not restore IP 3 production to control levels. Thus, the loss or uncoupling of a GTP-dependent mechanism downstream from PLC stimulation was specific for the signaling complex of each agonist.
The effects of GTP␥S described above prompted us to test the sensitivity of signaling by the three agonists to GDP␤S. Individual experiments with carbachol and bombesin are shown in Fig. 8, while Fig. 9A summarizes the results of four experiments. In all experiments the sensitivity to GDP␤S was tested after 2 min of permeabilization, and GDP␤S was included in the permeabilization medium to ensure uniformity of conditions. The differences between the agonists are quite clear. GDP␤S inhibited carbachol-induced Ca 2ϩ release with an IC 50 of about 0.042 mM, bombesin-induced Ca 2ϩ release with an IC 50 of about 0.53 mM, and CCK8-induced Ca 2ϩ with an IC 50 of about 1.48 mM. Hence, GDP␤S has different access to the Ca 2ϩ signaling complexes coupled to the carbachol, bombesin, and CCK receptors.
To determine if the inhibition of Ca 2ϩ signaling by GDP␤S was due to inhibition of PLC we measured the effect of GDP␤S on IP 3 production. Fig. 9B shows that inhibition by GDP␤S was agonist-specific. However, much higher GDP␤S concentrations were needed to inhibit IP 3 production (Fig. 9B) than Ca 2ϩ release (Fig. 9A). Even at 2.5 mM, GDP␤S had no effect on IP 3 production stimulated by CCK8. GDP␤S only partially inhibited the effect of bombesin with about 32% inhibition at 2.5 mM. In the case of carbachol, GDP␤S inhibited IP 3 production by approximately 65% with an IC 50 of 0.53 mM, which is about 13-fold higher than that required for 100% inhibition of Ca 2ϩ release (Fig. 9A). DISCUSSION Multiple receptor types that use the same transduction system and second messengers are present in virtually every cell type (1,2,4). Cross-talk between the same and different signaling pathways has been known for many years (1,4). Most recently such cross-talk was demonstrated between heterotrimeric G proteins and tyrosine kinase-dependent pathways (33,34). The extensive and coupled network of signaling pathways requires mechanisms to achieve specificity and precision in signal transduction. Compartmentalization of signaling by localization of several (or all) components of signaling complexes is emerging as a control mechanism in achieving signaling specificity. For example, many components of the mitogenactivated protein kinase and other signaling pathways are clustered in caveolae (35,36). In the case of Ca 2ϩ signaling, differential localization of protein kinase C isoforms is used to regulate different cellular activities (37). Clustering of specific Ca 2ϩ channels by binding to components of the exocytotic path-  (c and d). The changes in medium [Ca 2ϩ ] during the 10-min incubation at 37°C is shown only for trace b. All the experiments performed are summarized in Figure 6.

TABLE I Effect of incubation time in SLO on agonist-stimulated IP 3 production
Acini incubated in the SLO-permeabilization medium for either 2 or 10 min at 37°C were stimulated with the indicated agonist and with or without 2 M GTP␥S for 20 s.

IP 3 in pmol/mg protein
Values are the mean Ϯ standard error of the mean of triplicate determinations. Similar results were obtained in three additional experiments. way is believed to control the site of neurotransmitter release (38,39).
The problem of Ca 2ϩ signaling specificity is somewhat more acute in polarized secretory cells such as pancreatic and other acinar cells. For the most part receptors in the basolateral membrane transduce their signal to the apical membrane, where exocytosis of digestive enzymes and fluid secretion occurs (15). A partial solution to signaling specificity appears to be compartmentalization of the Ca 2ϩ pool which results in quantal behavior of Ca 2ϩ release (40) and localized changes in [Ca 2ϩ ] i next to the apical membrane (27,28). In the present studies we extended the previous findings (27)(28)(29) to show that all major Ca 2ϩ -mobilizing agonists release Ca 2ϩ first in the apical region which then propagates to the basolateral region. A difference between our studies and those reported before (27,28) is that we used clusters of three to five cells to ensure intact tight junctions and the maintenance of cell polarity, whereas previous studies (27,28) used single isolated cells. The use of cell clusters showed that the initial increase in Ca 2ϩ tended to occur in the boundary between the basolateral and the apical membrane. In addition we find that, at least in the case of CCK receptors, the [Ca 2ϩ ] i wave can start in several regions of the cell, such as the apical, apical-lateral or lateral-basal regions (Fig. 2). These features suggest that the receptors for all agonists can be clustered in regions of the basal or lateral membranes from which they initiate localized Ca 2ϩ signal close to the apical membrane. Thus, the Ca 2ϩ liberating messenger IP 3 does not need to diffuse across the cell to initiate the Ca 2ϩ signal in the apical region. Rather, it can act at the site of its production by the clustered receptors. Clustering of the receptors can be viewed as the first order of signal compartmentalization.
Another aspect of signal specificity was revealed when the initiation, pattern and propagation of the Ca 2ϩ signal evoked by each agonist was studied in the same cell. Previous studies have already shown that carbachol and CCK can initiate different patterns of Ca 2ϩ spiking in acinar cells (41). At low agonist concentration this Ca 2ϩ spiking can be confined to one pole of the cells (27,28). Here we used supermaximal concentrations of agonists to access all the Ca 2ϩ pool by each agonist to determine whether a) agonist specificity is maintained when IP 3 production is not limited and b) to be able to follow the initiation and propagation of the wave in the entire cell. The protocol of Fig. 1 revealed that each agonist initiates the Ca 2ϩ signal in different region of the cell, even when high agonist concentrations are used. In addition, the pattern and propagation rate for each Ca 2ϩ wave differs among the agonists used. The simplest interpretation of these findings is compartmentalization of the Ca 2ϩ pool and coupling of individual compartments to specific signaling complexes. This coupling must be very tight as repetitive stimulation of the same cell with maximal or different concentrations of the same agonist always initiated the [Ca 2ϩ ] i wave at the same cellular site. In previous studies we provided evidence for such an arrangement in SLOpermeabilized cells (16,19). Compartmentalized pools with variable affinities for IP 3 could be accessed by all antagonists in permeabilized but not intact cells (16). Furthermore, in salivary gland ducts isoprenaline, carbachol and ATP access non overlapping and limited portions of the internal Ca 2ϩ pool, whereas isoproterenol and forskolin access the same overlapping portion of the pool (42). Compartmentalization of the pool and its coupling to specific signaling complexes can be viewed as the second order of signal compartmentalization.
A more direct evidence for spacial compartmentalization of signaling was obtained by studying the properties of Ca 2ϩ signaling in permeable cells. Cell permeabilization with SLO caused time-dependent lose of Ca 2ϩ signaling with the sequence (first to last) carbachol, bombesin, CCK. Signaling by all agonists could be completely restored at any time after permeabilization with low concentration of GTP␥S. GTP was much less efficient than GTP␥S in restoring signaling (not shown) despite the presence of an ATP/GTP regeneration system in the reaction medium. A tentative interpretation of these experiments is that the signaling complexes lose GTP needed for activation of G proteins at different rates. This is, however, not likely since ATP and IP 3 have complete access to the Ca 2ϩ pump and the Ca 2ϩ channel, respectively, a short time after cell permeabilization (16,19). Heparin with a molecular mass of 6 kDa completely inhibits Ca 2ϩ release induced by IP 3 and all agonists (ref 16 for carbachol and CCK8 and not shown for bombesin). The 5-kDa ␤ thymosins and a 15-kDa S1 fragment of gelsolin can access the actin in the basal and luminal membranes (43). In addition, it is difficult to comprehend how GDP␤S can access the signaling complexes (albeit at high concentrations) while GTP, which is likely to be in the M range, will be retained by the CCK signaling complex. Indeed, the loss of Ca 2ϩ signaling by carbachol and bombesin was not accompanied by a loss in the ability to activate a G␣q-coupled PLC and generate IP 3 , nor the ability of GTP␥S to restore signaling was reflected in an increased agonist stimulated IP 3 production (Table I).
A more likely explanation for the loss of signaling and its restoration by GTP␥S may lie in our recent findings, showing a dual role for G proteins in Ca 2ϩ signaling; activation of PLC and regulation of the coupling between IP 3 production and Ca 2ϩ release (32). Thus, it is possible that incubation in SLO leads to a time-dependent uncoupling between agonist-stimulated IP 3 production and Ca 2ϩ release, which can be restored by low [GTP␥S] and inhibited by GDP␤S. The rate of uncoupling in SLO is different for the various agonists. The agonistspecific uncoupling points to strict spacial compartmentalization of signaling complexes which must include, for each agonist, a receptor, a G protein, an effector (PLC), IP 3 , and its access to specific portion of the internal pool.
In summary, we used the measurement of [Ca 2ϩ ] i at relatively high time resolution in the same single cells stimulated with several Ca 2ϩ mobilizing agonists and characterized several properties of Ca 2ϩ signaling by the various agonists in SLO-permeabilized cells to conclude the spacial compartmentalization of Ca 2ϩ signaling complexes. Such compartmentalization is likely to be important in determining signaling specificity. Thus, the site of action of second messengers can be determined by compartmentalization so that the same second messenger can modulate different cellular activities or the same activities at different sites. We also provide evidence for multiple roles for G proteins in Ca 2ϩ signaling, which regulate IP 3 production and the ability of IP 3 to release Ca 2ϩ from internal stores.