Compartmentalization of Ca2+ Signaling and ... - Semantic Scholar

4 downloads 0 Views 3MB Size Report
Streptolysin 0-permeabilized pancreatic acini were used to study compartmentalization of Ca2+ signaling and Ca2+ pools. In these cells, the inositol 1,4 ...
Vol. 269, No. 47, Issue of November 25, pp. 29621-29628, 1994 Printed in U.S.A.

THEJOURNAL OF BIOLOGICAL CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biolom, Inc

Compartmentalization of Ca2+ Signaling and Ca2+ Pools in Pancreatic Acini IMPLICATIONS FOR THE QUANTAL BEHAVIOR OF Ca2+RELEASE* (Received forpublication, March 17, 1994, and in revised form, August 22, 1994)

Giovanna TortoriciSO, Bin-Xian Zhangl, Xin Xu, and Shmuel Muallemn From the Department of Physiology, The University of Texas Southwestern Medical Center, Dallas, Texas 75235-9040

Signal transduction from the plasma membrane to the cell Streptolysin 0-permeabilized pancreatic acini were used to study compartmentalization of Ca2+signaling interior occurs through various G protein-coupled signaling and Ca2+ pools. In these cells, the inositol 1,4,5-trisphos- pathways. A signaling complex in the plasma membrane inphate (IPS)-dependentCa2+channels could be activated cludes a receptor, a heterotrimeric G protein, and effector enby a number of agonists (carbachol, cholecystokinin, or zymes such asadenylyl cyclases (1-4) or phospholipases (4,5). bombesin) or by activation of the entire cellular phos- Many cell types expressa combination of these signaling pathpholipase C pool with GTPyS. Surprisingly, each of the ways. Furthermore, it is common that cells express several antagonists interacting with acinar cells inactivated the different receptors that activate the same signaling pathway. channels after stimulation with GTP#. In addition, For example, pancreatic acinarcells respond to cholecystokinin when permeabilized cells were stimulated with more (CCK),l acetylcholine, bombesin, and epidermal growthfactor than one agonist, any antagonist to the specific agonists (6) by activation of phospholipase C (PLC) to generate IP, and employed inactivated the channels. Theaberrant behavior of the antagonists in permeable cells was notrelated mobilize Ca2+from internal stores (IS). The response to CCK to a loss of specificity since (a)when added before involves binding of the hormone to two states of the receptor, GTPyS, the antagonists had no effect on Ca2+ release and which trigger state-distinctive Ca2+signals (7). How the different signaling complexes are organized and ( b )when cells were stimulated with a single agonist,the antagonists prevented only the effect of their specific their activity coordinated is not known. I t is likely that signalt o achieve specificity and agonist. The differential behavior of the antagonists in ing pathways are compartmentalized intact and permeable cells suggestsa compartmentaliza- control of activation. Compartmentalization can be inferred tion of Ca2+signalinginto separate, agonist-specificunits from the tightcoupling of receptor-G protein-effector complexes that is modified by cell permeabilization. Further evi- (8) and from the stimulated generation of high localized condence for compartmentalization of signaling was obcentration of the second messengers (9). However, there is no tained by showing that the partial agonist (the CCK oc- direct evidence that the signalingcomplexes function as sepatapeptide analogue JMV-180)can access and release only rate and compartmentalized units. 50% of the cholecystokinin- or IPS-mobilizable Ca2+ pool Another form of compartmentalization resides in the action in intactand permeable cells. Kineticmeasurements re- of the second messenger. In thecase of IP,, there isevidence in vealed a multiphasic time course of agonist-evoked Ca2+favor of focal or global action within the same single cells. release in permeable cells. At high agonist concentra- Localized regions of high [Ca2+lidue to Ca2+ release from IS tions, all phases were fast and merged into an apparent have been reported in several cells (5-16). Measurement of single event of Ca2+ release. The phases were separated mitochondrial [Ca2+lsuggests theexistence of microdomains of by three independent protocols: reduction in agonist high Ca2+close to the IPS-sensitive channels (17). These localconcentrations, addition of heparin, or addition of ized effects have been attributed to a heterogeneous distribuguanosine-5'-0-(thio)diphosphate.Since all protocols that caused phase separation reduce IPS-mediated Ca2+tion of the Ca2+release channels in subcellular compartments release, these findings demonstrate heterogeneity in the (15, 16) that may have different affinities for IP, (16). Such affinity for IPSof channels present in compartmentalized organization of pools and channels is reflected in a quantal Ca2+ pools of the same cells. Compartmentalizationof sig- behavior of Ca2+ release(18).These findings contrast with the naling and the heterogeneity in the affinity for IPSre- observations that incells expressing receptors for several Ca2+ mobilizing agonists, any one agonist can mobilize the entire sulted in a quantal agonist-evoked Ca2+ release. The overall findings are discussed in thecontext of an integrated IP,-sensitive Ca2+pool (5, 6, 19). This would suggest that IP, model of compartmentalization of signaling complexes, production and action maynot be compartmentalizedand, Ca2+ pools, and IPS-activatedCa2+ channels. more importantly, that all agonists may access the Ca2+ pool in the same manner.Evidently, in this case,selectivity and spec* This work was supported by National Institutes of Health Grants ificity of agonist-evoked signaling is lost. To reconcile the varDK38938 and DK46591.The costs of publication of this article were iousobservations and obtain evidence of agonist-dependent defrayed in part by the payment of page charges. This article must specificity in signal generation and transduction, the organizatherefore be hereby marked ''advertisement" in accordance with 18 tion and action of signaling complexes need to be determined. U.S.C. Section 1734 solely to indicate this fact. Recently, we have shown that pancreatic acini permeabilized $ Present address: Instituto Venezolano de Investigaciones Cientificas, Centro de Biofisica y Bioquimica, Laboratoriode Fisiologia Gastrointestinal, Caracas, Venezuela. ' The abbreviations used are: CCK, cholecystokinin;'CCK8, cholecys8 Contributed equally to these studies. tokinin octapeptide; IP,, inositol 1,4,5-trisphosphate; IS, intracellular 1 To whom correspondence should be addressed: Physiology Dept., stores; SLO, streptolysin 0; PLC, phospholipase C; CCKJ, the CCK The University of Texas, SouthwesternMedical Center at Dallas, 5323 octapeptide analogue JMV-180; L18, CCK receptor antagonist Harry Hines Blvd., Dallas, TX 75235-9040. Tel.: 214-648-2593; Fax: L-364,718 (Merck, Sharp, and Dohme); GTPyS, guanosine 5'-3-0214-648-8685. (thi0)triphosphate;GDPBS, guanosine-5'-O-(thio)diphosphate.

2962 1

29622

Compartmentalization of Ca2+Signaling and

with the toxin streptolysin 0 (SLO) retain intact signaling complexes that can be activated by agonists and inactivated by antagonists (20). In thepresent studies, we used this system t o examine the organization of Ca2+signaling and Ca2+pools. By comparing selective andtotal activation of theG proteincoupled phospholipaseC pools and antagonist-induced channel inactivation, we were able to show an organization of signaling systems into units that include the generation of IP, and its action t o release Ca2+.Measurement of the kinetics of Ca2+ release evoked by CCK or its partial agonist, CCKJ, and their sensitivity to heparin and GDPpS revealed a heterogeneity in localization and sensitivity to IP, of the Ca2+pools. Thus, both the signaling systems and the Ca2+pools are compartmentalized. Such compartmentalization probably accounts for the quantal behavior of Ca2+release. EXPERIMENTALPROCEDURES Preparation of Pancreatic Acini-Rat pancreatic acini were prepared by limited collagenase digestion as previously described (21, 22). In brief, the pancreas of a 100-150-g rat was removed, injected with solution A containing (in mM) 140 NaC1, 5 KCl, 1MgCl,, 1CaCl,, 10 Hepes (pH 7.4, with NaOH), 10glucose, 0.1% bovine serum albumin, and supplemented with 10 m~ pyruvate and 0.02%soybean trypsin inhibitor. The tissue was minced and digested for 5-6 min with collagenase P (Boehringer Mannheim). The acini were washed twice in solution A, filtered through a nylon mesh, and suspended in about 8 ml of solution A. Measurement of Ca2+ Uptake and Release in Permeable CellsPermeabilization of acini with SLO was performed as previously described (20).About 100 mg (wet weight) of acini were washed twice with a solution containing 145 mM KC1 and 10 mM Hepes (pH 7.4, with NaOH) and once with the same solution that was treated with Chelax 100. The acini were added to a chelax-treated solution containing (final concentrations)0.02%soybean trypsin inhibitor, 3 m~ ATP, 5 mM MgCl,, 10 mM creatin phosphate, 5 unitdm1 creatin phosphokinase, 10 p~ antimycin A,10 p~ oligomycin, 1 p~ Fluo 3, and 0.4 unitdml SLO (permeabilization medium). Recordingof Fluo 3 fluorescence was initiated on addition of cells to a warm incubation medium in afluorometer cuvette and was at anexcitation wavelength of 488 nm and an emission wavelength of 530 nm. The cells were allowed to incorporate Ca2' and reduced the [Ca"] in theincubation medium to 50-100 nM before stimulation. The Fluo 3 fluorescence signals were calibrated by addition of 2 mM CaCl, to the incubation medium (FmJ,which was followed by the addition of 10 mM EGTA and 20 mM NaOH to obtain Fmin. [Ca2'l, was calculated as described before(23) using a Kd of 370 nM at 37 "C. Measurement of[Ca2+1,-[Ca2+l, was measured in cell suspension as described before(9) so that measurements in intact and permeable cells can be comparedand to initiate rapid cell activation or inhibition. Acini were loadedwith Fura 2 by incubation with 5 p~ Fura 2/AM for 30min at 37 "C. The cells were washed once, resuspended in solution A, and kept at room temperature until use. For measurements of [Ca2+l,,a sample of cells was transferred to a warm solution A in a fluorometer cuvette, and fluorescencewas measured at excitation and emission wavelengths of 340 and 500 nm, respectively. Formeasurements in the absence of external Ca2+,the cells were washed twice with a solution containing 140 mM NaC1, 5 mM KC1, and 10 mM Hepes (solution B) supplemented with 1 m~ MgCl, and 10 mM glucose and once with solution B, which was treated with Chelax 100 before the addition of MgCl, and glucose. The cells were added to a similar warm Chelaxtreated solution before stimulation with agonists. The fluorescence signals were calibrated to obtain [Ca2+l,as described before(9). Mass Measurementof IP,--For measurements in intact cells, acini in solution A were stimulated with carbachol and then inhibited with atropine. At the indicated times, the reactions were stopped by transferring 100 p1 of cells to a cold solution containing 15%perchloric acid, 0.1 mM ATP, and 2 mM Pi. These samples were processed to determine IP, as detailed below for measurements with permeable cells. To measure IP, in permeable cells, the acini were incubated for 2 min at 37 "C in thepermeabilization medium as described above formeasurement of Ca2+release. The permeabilized acini were stimulated with GTPyS or treated with GTPyS for 30 s and then inhibited with atropine. At the indicated intervals, 100-pl samples were transferred to 100 of a cold solution containing 15%perchloric acid, vigorously mixed, and kept on ice. At the end of the experimental incubations, the tubes were centrifuged for 2 min at 10,000 x g, and the supernatants were transferred to

I

\

Pools

\

\

1 min

\

u CCK8

lo-* M

FIG.1. Effect of agonist stimulationand antagonist inhibition on IPS-activatedCa" release. Acini were incubated in permeabilization medium. Whereindicated, the permeabilized cells werestimulated with lo-@M CCK8 (a-c) and then exposed to50 p~ of the cholecystokinin antagonist L18 ( b ) or 200 pg/ml heparin. clean tubes. Standards of IP, were prepared in solution A or permeabilization medium and processed in asimilar manner. Perchloric acidwas removed, and IP,in thesamples was extracted by the addition of 0.2 ml of tri-n-octylamine and 0.2 ml ofFreon. After mixingand centrifugation, between 15 and 25 pl of the upper layer was used for measurement of IP, content with a radioligand binding assay as described before (20). The concentration of IP, in each sample was read from the appropriate calibration curve constructed from the standard samples of IP,. RESULTS

Compartmentalization of Ca2' Signaling-To define the signaling unit, we determined the effect of agonists and antagonists on Ca2+release from IS. Fig. 1 shows the basic protocol using the SLO-permeabilized pancreatic acini. Stimulation of these acini with 10 n~ CCK8 increased medium [Ca2+]from about 55 to 425 n~ (Fig. la). The CCK8-evoked Ca2+release was inhibited by addition of heparin before (see below) or after cell stimulation (Fig. IC),indicating the mobilization of Ca2+ from the IPS-sensitivepool. Comparable rates of Ca2+uptake before and after cell stimulation indicate that 200 pg/ml heparin completely blocked Ca2+release (Fig. IC).Termination of cell stimulation with the specific CCK8antagonist L18 induced rapid Ca2+uptake into IS (Fig. lb). Inhibition of Ca2+release by L18 was due to inactivation of the IP,-dependent Ca2+channel, since L18inactivated Ca2+release in the presence of saturating IP, or inositol 2,4,5-trisphosphate concentrations (not shown, but see Ref. 20). Similar observations were made when SLO-permeabilized acini were stimulated with bombesin and inhibited with a bombesinreceptor-specific antagonist. Accordingly, the IPSmediated Ca2+release channels can be activated by any Ca2+ mobilizing agonist acting on pancreatic acinar cells and inactivated by the subsequent addition of the specific antagonist. Ca2+release can also be evoked by global stimulation of the G protein-coupled PLC pool with GTPyS (Fig. 2). 40 p~ GTPyS increased medium [Ca2+]from 60 to 345 n~ (Fig. 2a), and this effect was inhibited by heparin added before (Fig. 2 b ) or after (Fig. 2 c ) stimulation. The ability to stimulate IP, production and the IP,-activated Ca2+channels by different specific agonists or by stimulation of the entire cellular PLC pool with GTPyS allowed US to study the relationship between signaling units. Fig. 3 shows that after activation of Ca2+release by GTPyS, the release channel was inactivated by any of the antagonists that bind t o specific receptors on acinar cells. Thus, the CCK receptor antagonist L18 (Fig. 3b) and the muscarinic receptor antagonist atropine (Fig. 3c) inactivated the IP,-dependent Ca2+channel to induce rapid Ca2+uptake into IS similar to that observed after inhibi-

Compartmentalization of Ca2+Signaling and Pools

29623

345 -

Fc .-

+-

(u

a

0

60 G T P 6 40 pM

Heparin 200 pg/rnl

Heparin 200 pg/rni

~

GTP* 40 pM

FIG.2. Activation of IP,-mediated Ca2+ release by GTPyS. Acini in permeabilization medium werestimulated with 40 p~ GTPyS ( a ) .As indicated, 200 pg/ml heparin were added before ( b )or after (c) stimulation with GTPyS.

O'

lb

io

io

310

io

$0

:o

Time (sec)

"1

T

I30

!30

'

0 Control 0 Carbachol A

Carbachol +Atropin

- - -. [Ca2+Ii 60

GTP@ 40 pM

2 Time (sec)

A Control C.

'

so

- x

%E

0 G T p F 40 pM

0 GTW+Atmpine 2 mM F=O) (T40sec)

Atropine

Time (rnin)

FIG.4. Antagonist-mediatedinactivation of Cas+ releasein intact cells. Fura 2-loaded cells (panel A were stimulated with 0.2 mM carbachol (a+) for 10 ( b )or 60 s (c) before inhibition with 20 atropine.In panel B , acini insolution A were stimulated with 0.2 mM carbachol (0,A), and after 10 or 60 s (A), acini were inhibited with20 p~ atropine.At the indicated times, samples were removed todetermine the levels of IP, as described under "Experimental Procedures." The dashed lines show the time courses of the reduction in [Ca"], due to atropine, which were taken from traces b and c in the upper panel.

FIG.3. Antagonists induce inactivation of Caa+release in acini stimulated withGTP#. In experimentsa+, the effect of stimulation with 40 p~ GTPyS and subsequent exposure to 50 PM L18 ( b )or 2 m~ atropine (c) on Ca2+release was tested. In experiment d, the effects of plateaued after 40 s of stimulation at about 6-fold above basal. GTPyS andatropine on IP, levels were measured. Acini were incubatedAddition of atropine increased the rateof reduction in IP, levin the presence (circles) or absence (triangles)of 40 p~ GTPyS. After30 els. However, the reduction in IP, levels was not completed s of stimulation with GTPyS,some of the acini were treated with 2 mM within the time of the onset of[Ca"], reduction and lagged atropine (open circles).At the indicated times, samples were removed cells, the antagoto determine the levels of IP, as detailed under "Experimental behind [CaZ+li reduction.Thus, also in intact nist probably inactivates the IP,-dependent Ca'' channels inProcedures." dependent of IP, levels. tion of the channel with heparin (see Fig. 2c). This was comInactivation of Ca2+ release by the antagonists in GTPySpletely unexpected since activation of G proteins by GTPyS is stimulated cells (Fig. 3) raised the question as to whether the irreversible and should not be affected by any one receptor antagonists retained theirspecificity in the permeabilized cells antagonist (1).Indeed, atropine had no effect on the rate or and whether channel activation was requiredfor inactivation extent of IP, production stimulated by GTPyS (Fig. 3d). This by the antagonists. This was tested in experiments similarto experiment indicates that the antagonists inactivated the IP,those shown in Figs. 5 and 6. Fig. 5b shows that addition of dependent Ca" channel despite the stable activation of G pro- atropine inhibited the abilityof carbachol to activate Ca'' reteins and the cellularPLC pool by GTPyS and the high levels lease.When added before cell stimulation, atropine had no of IP,. effect on the abilityof CCK8 (not shown)or GTPyS (Fig.5 b ) to The concept of antagonist-initiated channel inactivation in- increase [Ca"]. This was thecase also when carbachol was not dependent of IP, concentration (Fig. 3 and Ref. 20) was ex- added to the incubation medium prior t o stimulationwith tended to intactcells by measuring theeffect of the antagonist CCK8 or GTPyS (not shown). The presence of atropine did not on [Ca2+liand [IP,] in stimulated intact cells. Fig. 4A shows that change theeffectiveness of L18 to inactivate channels activated addition of atropine tocells stimulated withcarbachol for 10 or by GTPyS (Fig. 5b). The converse experiments showed that, 60 s triggered a rapid reduction in [Ca"],.We have shown when addedbefore stimulation, L18 blocked the effect of CCK8 to reuptake of [Ca"], into the IS (24). without affecting channel activation by carbachol (not shown) previously that this is due The onset of [Ca"], reduction requires inactivation of the IP,- or GTP@ (Fig. 5c) or the ability of atropine to inactivate the dependent Ca2+ release channels. This occurred within the time channel (Fig. 5c). resolution of our measurements (about 2 s). Reduction of [Ca"], Further demonstrationof the needfor activation of a signalto near basal level required a 7-10-s incubation period with ing unitt o observe efficient channel inactivationby an antagoatropine. Thebehavior of cellular IP, levels under similar con- nist interaction with this unit is demonstratedFig. in 6. Fig. 6a ditions is illustratedin Fig. 4B. Carbachol stimulation caused shows that atropine had little effect on the Ca2+release chana transient increase in IP, levels with a peak at 10 s, which nels after maximal activation with CCK8. Carbachol had no

Compartmentalization of Ca2+Signaling Pools and

29624

GTP@ - 40 pM 1 min H

;3 0 m m[

I.

75

I

Atropine 2mM

Carbachol 2mM

FIG.5. Specificity of theantagonistsinSLO-permeabilized cells. In allexperiments, the cells were stimulated with 40 p~ GTPyS. Dace ais the control. In trace b, the acini wereexposed t o 2 mM atropine and 2 mM carbachol before GTPyS and then inhibited with50 PM L18. In truce c , the acini were exposed to 50 p~ L18 and 0.1p~ CCKS before stimulation with GTPyS and inhibition with 2 mM atropine. Carbachol

t 4

t

I

I

'r

4

0.6

t

0.4

LI

t

0.2

t

0.1

CCKJ Concentration in WM

FIG.7. Effect of differentconcentrations of CCKJ on Ca2* release. The permeabilized aciniwere allowed to reduce medium[Ca''] to the same level before stimulation with the indicated concentrationsof CCKJ. Note that the rateof the rapid phase of Ca2+release is similar between 0.2 and 4 p~ CCKJ.

centration did not sufficiently resolve the different phases to determine theirexact dependenceon [CCKJ]. However, the fact that reducing the intensity of the stimulus resulted in phase 1 min separation indicates that thetwo phases have different sensitivities t o CCKJ, where the second phase maximizes at higher N agonist concentration. m Y To determine the reasonsfor the multiple kinetic phases of Ca2+ release,it was necessary to identify the pool from which CCKJ mobilizes Ca2+. Severalstudies havereported that CCKJ b. 40 stimulates IP, production poorly and may release Ca2+from an IP,-insensitive pool (27-30). A possible explanation for the mulccK8 10-7M FIG.6. Effect of sequential addition of agonists and antago- tiple phases is thatcell permeabilization might have changed nists on Caa+ release in permeabilized cells. In experiment a , the the ability of CCKJ to activate PLC. Fig. 8 shows that this is cells were stimulated with 0.1 PM CCK8 and then treated with 2 mM not thecase. As in intactcells (see Refs. 27 and 28 and data not atropine. In experiment 6 , the acini were stimulated with 0.1p~ CCKS shown), CCK8 was much more effective than CCKJ in stimuand then with 2 mM carbachol. Note that carbachol had no effect on [Ca"] of cells maximally stimulated with CCK8. Where indicated, the lating the production of IP, in SLO-permeabilized cells. The concentrations of CCKJ causing maximal Ca2+ release (4-10 acini were inhibited with 2 mM atropine. PM) increased cellular IP, levels 20-fold less than that caused effect on Ca2+release in acini stimulated with 0.1 p~ CCK8 by 0.1 VM CCK8. Previous studies concluded that IP, and CCKJ mobilize Ca2+ (Fig. 6b). However, addition of atropine now completely inactivated Ca2+ release. Direct mass measurements showed that from different pools in SLO-permeabilized cells (29). In view of under theconditions of Fig. 6, a and b, atropine hadno effect on recent reports showing that injection of heparin into acinar IP, levels (not shown), indicating thecontinuous stimulation of cells inhibited the CCKJ-mediated activation of Ca2+-dependent K+channels (31)and CCKJ-dependent Ca2+release (32), we PLC by CCK8. The behavior of the antagonists in permeable cells demon- reexamined the relationship between the pools mobilized by strated inFigs. 5 and 6 is different from the behavior in intact IP,, CCKJ, carbachol, and CCK8 in SLO-permeabilized cells. cells in which the antagonists inactivate only the signaling Fig. 9 (upperpanel) shows that discharge of the IP,-mobilizable system coupled to theirspecific receptors (24,251. Thisis prob- pool by 2 1.1~IP, prevented the effect of CCKJ (Fig. gal, carbaably due to the loss of signaling compartmentalization,which chol, and GTP-yS or CCK8 (not shown) on [Ca2+l.Furthermore, stimulation of the cells with CCKJ proportionally reduced the exists in intactcells. Compartmentalization of ea2+ Pools-Another level of com- ability of IP, (Fig. 9b) and maximal agonist stimulation with partmentalization studied is that of the Ca2+pool. For this, we carbachol and GTP-yS (Fig. 9c) to release Ca2+from IS. Finally, took advantage of the uniqueproperties of Ca2+release evoked stimulation of the cells with 0.1 PM CCK (Fig. 9d) or carbachol by the CCK analogue, CCKJ. This analogue was shown to bind and GTP-yS (not shown) increased medium [Ca2+lto the same to one state of the CCK receptor (7, 26) and to exclusively level as a maximal concentration of IP, (Fig. 9a) and prevented induce [Ca2+],oscillations in intact cells (27). As a first step in any further Ca2+release by CCKJ. Evidently, CCKJ mobilized characterizing the effects of CCKJ, we measured the concen- part of the Ca2+stored in the IP, pool, and Ca2+mobilization tration dependence of its effect. This is illustrated in Fig. 7. from an IP,-insensitive pool cannot account for the multiphasic Even at very high CCKJ concentrations and therelatively poor kinetic of Ca" release. Another point illustrated inFig. 9 is that maximal or supertime resolution possible with these measurements, the kinetics of Ca2+release were multiphasic (Fig. 7). This is reflected as maximal concentrations of CCKJ released only about one-half to medium tCa2+l by about 205 abrupt changes in the rate of Ca2+release. A better resolution of of the Ca2+present in IS increase the timecourse was obtained by reducing CCKJ concentration. -c 13 nM (n = 27). In comparison, maximal [CCK8] increased At 0.4 V M CCKJ, a fast phase of Ca2+release was completed medium [Ca2+]by 473 +. 15 rm (n = 48). Hence, in permeabilized within 10 s of stimulation. A partially overlapping second event cells, CCKJ can access about 50% of the IP,-mobilizable pool. started sometime after and was maximal about 45 s after cell Fig. 9, e andf, show similar behavior in intactcells. Stimulation stimulation. It is clear from Fig. 7 that reducing agonist con- of cells maintained in low Ca2+ medium with supermaximal

"3

29625

Compartmentalization of Ca2+Signaling and Pools A=lO pM CCKJ

B

1 rnin

H

T

f

.-C

a

m

e Control

N

70

B=0.2 pM

CCKJ

Heparin in pg/ml

A 0.1 pM CCK8

m

o 10pM CCKJ

0

20-

Time (min)

Agonist concentration p[M]

FIG.8. Time and concentration dependence of the effects of CCK8 and CCKJ on IP,. In panel A, the acini were incubated in permeabilizationmedium for 2 min at 37 "C (closed circles) before stimulation with 0.1 PM CCK8 (open triangles) or 10 1.1~CCKJ (open circles). At the indicated times, samples were removed to estimate the levels of IP,. In panel B , acini incubated for 2 min at 37 "C in permeabilization medium were stimulated for 30 s with the indicatedconcentrations of CCK8 (open triangZes) or 1 min with CCKJ(open circles).

release. Inpanel FIG.10.Effect of heparin on CCKJ-evoked CaZ+ A (experiments a-c), the acini werestimulatedwith 10 PM CCKJ, whereas in panelB (experiments d-g), the acini were stimulated with 0.2 p~ CCKJ. Where indicated, the acini were treated with different concentrations of heparin before stimulation with CCKJ. CCKJ was added after the first ( a and d ) or second (all other times) breaks in recording.

evoked by carbachol in untreated cells (Fig. 9e). A better resolution of the Ca'' release phases a t low agonist concentration suggests thata possible explanation for the multiphasic timecourse is thatCCKJ releases Ca'' from multiple, compartmentalized Ca'' pools with differentsensitivities to 1 min 2mM Carbachol H +GTP(*) 40pM IP,. One approach t o test this possibility is to determine the I effect of the IP, antagonist, heparin, on the CCKJ-evoked Ca'' 520 release. The resultsof such experimentsat two concentrations 430 of CCKJ are shown in Fig. 10. Comparing Fig. 10a with 10b I reveals that 5 pg/ml heparin caused phase separation inacini E 245 ._ stimulated with 10 p CCKJ. Panel B of Fig. 10 shows that further phase separation canbe obtained at lower CCKJ cono_ centration. At 0.2 p CCKJ, the usual two phases of Ca'' release were observed (Fig. 10d). Heparin at 5 and 10 pg/ml 65 reduced the extent of Ca'' release of all phases, including the initial phase. However, underthese conditions, thetime CCKB 0.1pM courses of Ca'' release had three phases(Fig. 10, e and f). The lP3 2$M first phase ended 18.7 2 0.6 s after stimulation, and the cells partially reduced medium[Ca"] before the second event of Ca2+ release reached maximum. The second phase ended after about 56 * 1.7 s of cell stimulation, and the third phase reached maximum after 103 2 7.6 s ( n = 3) of stimulation. Thus,Fig. 10 shows that partial inhibition of IP,-mediated Ca2+ release at constant stimulation intensity was as effective as reducing agonist concentration in resolving the phases of Ca" release. CCK8 increases IP, levels 20-fold (Fig. 7) and mobilizes f f about twice more Ca2+than CCKJ (Fig. 1)t o deplete the IP,Carbachol CCKJ Carbachol 0.2mM 10 MM 0.2 mM mobilizable pool (Fig. 9). Thus, the question arises of whether FIG.9. Overlap between the Ca2*mobilized byCCKJ,CCKS, CCK8 also mobilizes Ca'' from multiple IP,-sensitive pools. carbachol, and IPS. Duces a d are from permeable cells, whereas This question was addressedby testing theeffect of heparin on truces e and f are from experiments with intact cells. In trace a , the acini CCK8-mediated Ca'' release. Although relatively highconcenwere exposed to saturating concentrationof IP, (2 p ~ before ) stimulac, trations of heparin were needed to block Ca'' release by maxition with 4 PM CCKJ. Dace b shows the reverse experiment. In trace the acini were stimulated with 4 p~ CCKJ. When[Ca"] stabilized, PLC mal CCK8, heparin uncovered multiple phases of Ca'+ release was maximally activated by a mixture of 40 PM GTPyS and 2 mM from cells stimulated with different concentrations of CCK8 carbachol. For comparison, trace d shows the effect of 0.1 1.1~CCK8 on Ca'' release. In traces e and f, Fura 2-loaded cells were washed and (not shown). Another protocol used to demonstrate the multiphasic naresuspended in a chelax-treated Ca2+-free medium. Where indicated, they were stimulated with0.2 mM carbachol or 10 p~ CCKJ. ture of Ca'' release and theheterogeneity of the agonist-mobilizable Ca'' pool was to attenuate PLC activation with GDPPS. concentration of CCKJ increased [Ca"], to about 600 IIM. Sta- Fig. 11 shows that both CCKJ- and CCK8-evoked Ca" release bilization of[Ca"], above resting levels in CCKJ-stimulated can be inhibited by GDPpS, although a much higher concencells reflects the continuous oscillations of [Ca2+Iiunder these tration of GDPPS was required to inhibit the effect of 10 n~ conditions (33). Stimulation of these cells with carbachol re- CCK8 than that of 10 p~ CCKJ. At 10 p GDPpS, the CCKJsulted ina second increase in [Ca"],, which was similar to that evoked Ca" release was clearly separated into two phases. evoked by CCKJ (Fig. Sf) but about 53 4% ( n = 3) of that Increasing GDPpS to 25 had little further effect on the first

-

N

+

29626

Compartmentalization of ea2+ Signaling and Pools A=l OpM CCKJ

step downstream of IP, production to inactivate thechannels. Hence, the action of antagonists on stimulated cells should not be viewed merely as the end of cell stimulation but rather as the initiation of cellular events leading to the unstimulated state. Another implication of the experiments with GTPyS and those in Figs. 5and 6 is that the inhibitory signal generatedby any one antagonist can access channels activated by multiple agonists. Evidently, this is an aberrant behavior observed in SLO-permeabilized cells, since in intact cells the antagonists reverse only the action of their own agonists. The importance of these findings is not the abnormalbehavior of permeable cells but rather what they imply for the organization of Ca2+ signaling complexes in intactcells. One way to reconcile the behavior of the antagonists in intact and permeable cells is to suggest that Ca2+signaling is organized into autonomous units, and spreading of the antagonist-dependent inactivation signal is restricted in intact cells (See the model in Fig. 12). FIG.11. Effect of GDPSS on CaZ+release evoked by CCKJ or The present studies also provide evidence for compartmenCCKB. The aciniin panel A were stimulated with 10 p~ CCKJ, and the acini in panel B were stimulated with 10 nM CCK8. The indicated talization and heterogeneity of the intracellular Ca" pool by concentrations of GDPPS were added to the permeabilization medium studying theeffect of full (CCK8) and partial (CCKJ) agonists before addition of the acini. of CCK on Ca2+ release.For this it was necessary to define the pool from which CCKJ mobilizes Ca", sinceprevious work phase while sufficiently inhibiting the second phase t o allow (26-31) was inconclusive. Our results show that CCKJ mobipartial reuptake of Ca2+at the end of the first phase and to lized Ca2+exclusively from the IP, pool. The evidence includes GDPPS observe a third phaseof Ca2+ release.In thecase of stimulation the following: (a)inhibition of G protein activation with with 10 nM CCK8, 50-200 VM GDPPS separated Ca2+release inhibited Ca2+ release by CCKJ; ( b ) the effect of CCKJ was into at least two distinctive phases. At 500 PM GDPPS, only the blocked by heparin; ( e ) depletion of the IP, pool with a saturatfirst fast phase of Ca2+release is observed (Fig. 11,panel I?). ing concentration of IP, prevented the effect of all agonists, When the cells were stimulated with 0.1 p~ CCK8, between including CCKJ, on Ca2+release (Fig. 9a); ( d )stimulation with 200-500 VM GDPPS were required to separate the phases, and CCKJ reduced the effect of IP, (Fig. 9b) or maximal concentra2 mM GDPPS were required to maximally inhibit Ca2+release tions of carbachol and GTPyS (Fig. 9c) on Ca2+release. The combined results clearly show that CCKJ mobilizes Ca2+from (not shown). the IP,-sensitive pool. DISCUSSION Examining the kinetics and extent of CCK8- and CCKJThe organization of cellular Ca2+signaling and the agonist- induced Ca2+release reveals that the Ca2+mobilized by IP, is mobilizable Ca2+pool is only partially defined. An emerging stored in at least three general compartments.Ca2+stored in picture is thatof compartmentalization of Ca2+signaling in all the first compartment is rapidly mobilized by high and low levels (19). This isbased on findings of localized hot spots from [CCKJ]. The second compartment is rapidly emptied by high, which a Ca2+ spike emanates t o spread into the entire cell but slowly by low, CCKJ concentrations (Fig. 7).Ca2+stores in the third compartment cannotbe or are only slowly mobilized (10-16) and the quantal behavior of Ca2+release (18, 19, 3438). The possible compartmentalization of signaling and Ca2+ by CCKJ but can be rapidly mobilized by high concentrations of pools raises thequestion as to how they are related. Pancreatic CCK8, carbachol, or IP, (Fig. 9). The Caz+pool is more extenacinar cells appear tobe a suitable systemt o study suchques- sively compartmentalized since frequently three or more overtions because they respond to a battery of Ca2+mobilizing ago- lapping components of Ca2+ release at low CCKJ can be renists (61, which act on the same Ca2+ pool (6, 19, 39). The solved by heparin or partial inhibition of PLC with GDPPS. The resolution of the techniques agonists mobilize Ca2+in a quantal manner(18),although with temporal and, in particular, spatial intense stimulation each agonist can mobilize the entire IPS- available limit further separation between the phases and thus sensitive Ca2+pool (24,391. Furthermore, ina recent study, we the ability to directly estimate the minimal size of a Ca2+ overall number of these showed that terminationof cell stimulation with an antagoniststorage/mobilizable unitandthe caused rapid inactivation of the IP,-dependent Ca2+channel compartments. independent of IP, metabolism (20). Here, we extend these An important question is whether the multiple phases of to spatial localization of the pools findings to show that itoccurs in intactcells and with any pair Ca2+release are due different IP,. The evidence of agonist and antagonist so that the relationships between or due t o heterogeneity in their sensitivity to favors the latter explanation. Every procedure used to reduce signaling units canbe explored. the effectiveness of IP,-mediated Ca2+release during CCKJ The most striking finding of the present studies was the antagonist-induced inactivation of Ca2+ release stimulated by stimulation inhibited the second more than the first phase. GTPyS. GTPyS activates G proteinsin a stable andirreversible This indicates that the Ca2+pool released during the second manner (1).Indeed, GTPyS increased IP, levels about %fold phase has significantly lower affinity to IP, than the pool recells also contain a Ca2+ more than any agonist(Ref. 20, Figs. 3 and 81, and addition of leased during the first fast phase. The atropine toGTPyS-stimulated cells had no effect on the rateor pool that cannot be mobilized by CCKJ. Intense stimulation extent of IP, production (Fig. 3) while causing thehydrolysis of with CCK8, carbachol, or GTPyS, all of which increase IP, to IP, in carbachol-stimulated cells (20). This implies that the levels much higher than CCKJ, was required to mobilize this lowest afflnity to IP,. If antagonist-mediated inactivation of the IP,-dependent Ca2+ pool. Therefore, this pool must have the channel may notproceed through a G-protein-coupled pathway. differential localization was responsible for the multiphasic In addition, the antagonist-evoked inhibitory signal acts on a rates of Ca2+release, then heparin should have inhibited the

Compartmentalization of Ca2' Signaling and

Pools

29627

+

FIG.12. A model for the arrangement of Ca2+signaling complexes, Ca2+pools, and distribution of IPS-activated Ca" channels. The model assumes autonomous signaling units, each of which includes receptor, coupling, and effector proteins (RI-R4) and a group of

IPS-activatedCa" channels with different thresholds for activation by IP3. Thechannels are distributed among different noncommunicating compartmentsof the CaY+ pool. Subcompartments of different signaling units with similar thresholds for activation by IP, communicates across the boundaries of the signaling units. Positive and negative regulatory signals emanate from a signaling complex can access only

the Ca" pools and the IPS-activated Ca" channels within the signaling units.

a

first phase more than the second phase, and reduction of ago- lus releases part of the Ca2+from all stores. The reduction in nist concentration and addition of GDPpS were expected to Ca2+ content of the storesreduces the affinity of the channels to inhibit all phases to the same extent without causing phase IP,, which prevents further Ca2+ release (6, 9). It has been separation. difficult to obtain unequivocal evidence in supportof any of the The question that arises is whether thepools with differen- models. The results provided here aredirect evidence for comtial sensitivity to IP, reside in the same or different acinar cell. partmentalization of the pool, a heterogeneity in theaffinity to The fact that quantal behavior of Ca2+ release is observed in IP,, and, as a consequence, quantal Ca2+ release. The impact of single intact cells (30, 33)suggests that thepools reside in the functionally independent storeswith variable sensitivityto IP, same cells. Further, CCKJ and CCK8, which mobilize Ca2+ on quantal Ca2+release has been mathematically analyzed pools with different affinities to IP, (Figs. 7-11), act on the (35). same single cells (30, 32, 33). It isdifficult to believe that after The present results and our previous studies (9, 18, 20, 33, permeabilization with SLO, some of the cells responded exclu- 42) examined several aspects of the organization and regulasively to CCKJ and some to CCK8. Finally, if the pools reside in tion of Ca2+signaling. The model in Fig. 12 is an attempt to different cells, then to obtain phase separationby modification integrate thefollowing findings. (a)Agonist-stimulated IP, proof three distinctive steps in the signaling pathways at several duction is highly localized (9); ( b )antagonists acting on intact (up to 9) heparin and GDPpS concentrations with 4 different or permeable stimulated cells inactivate the IP,-dependent Ca2+ channels independent of their effect on IP, levels (Ref. 20 agonists that were used at several (5-7) concentrations, requires several thousand (3000-5000) populations of acinar cells and Figs. 3 and 4); ( c ) in permeabilized cells, each antagonist with different affinitiesto IP,. Not only is thisvery unlikely, but inactivates all the IP,-dependent Ca2+channels even when the in this case we willnot be able to separate the phases since Ca2+ entire pool of PLC is activated (Figs.3 and 5);( d )in intactcells, release from one of these populations will be too small to be each antagonist inactivatesonly the Ca2' channels activated by detected. Hence, pools with different affinitiesto IP, must exist its own agonist; ( e ) each agonist can mobilize the entire IP,in the same cells. sensitive pool in permeabilized (Figs. 6, 7, and 9) and intact An important aspect of the heterogeneous sensitivity to IP, cells (6,7, 18, 19); ( f , the cellular Ca2+pool is heterogeneous among compartmentalizedCa2+pools of the samecells is that itwith respect to activation by IP, (Figs. 7-11); (g)Ca2+release provides a straightforward explanation for the phenomena of evoked by agonist or IP, is a quantal process (Ref. 18 and Figs. quantal behavior of Ca2+ release (18, 19, 34-38). Two major 7-11); and ( h )CCKJ and CCK8 acting on different states of the mechanisms have been proposed to explain the quantal nature CCK receptor communicate with a different and separate porof Ca2+release. The first is that submaximal stimulation with tion of the Ca2+pool (Figs. 7-11). The model assumes compartagonist or IP, releases all the Ca2+from a fraction of the IS, mentalization and heterogeneity of both Ca2+signaling and the while none of the Ca2+ isreleased from the remaining stores Ca2+pool. Each signaling complex is assigned a group of IP,(18, 37). This requires compartmentalization of the pool into activated Ca2+ channels and a subpool to form a signaling unit. units expressing IP,-activated Ca2+channels with different af- The channels of each group have a different threshold for acfinities or thresholdsfor activation by IP, (18).The alternative tivation by IP,. Positive and negative regulatory signals initimodel is based on a dynamiccontrol of the affinity for IP, by the ated by each signaling complex can access only the channels in Ca2+content of the stores(40,411. The stores are homogeneous the signaling units. In SLO-permeabilized cells, this barrier is with respect to their interaction with IP,. Submaximal stimu- breached to allow negative signals generated by antagonists to

29628

Compartmentalization ofSignaling Ca2+ and

Pools

12. Inagaki, N., Fukui, H., Ito, S., Yamatodani,A,, and Wade, H.(1991)Proc. Natl. Acad. Sci. U. S. A. 88, 4215-4219 13. Brundage, R. A., Fogerty, K. E., Tuft, R. A,, and Fay, E S. (1991)Science 254, 703-706 14. Llinas, R., Sugimori, M., and Silver, R. B. (1992)Science 256,677479 15. Kasai, H., Li, Y. X., and Miyashita, Y. (1993)Cell 74, 669-677 16. Thorn, P., Lawrie, A. M., Smith, P.M., Gallacher, D. V., and Petersen, 0.H. (1993)Cell 74,661-668 17. Rizzoto, R., Brini, M., Murgia, M., and Pozzan, T. (1993)Science 262, 744-746 18. Muallem, S., Pandol, S. J., and Beeker, T. G. (1989)J. Biol.Chem. 264, 205-212 19. Muallem, S. (1992)in Advances in Second Messenger and Phosphoprotein Research: Inositol Phosphates and Calcium Signaling (Putney, J. W., Jr., ed) pp. 351-368,Raven Press, Ltd., New York 20. Zhang, B.-X., Tortorici, G., and Muallem, S. (1994)J. Biol. Chen. 269,1713217135 21. Xu, X., Star, R. A., Tortorici, G., and Muallem, S. (1994)J. Biol. Chem. 269, 12645-12653 22. Stauffer, P.L. A.,Zhao,H., Luby-Phelps, K . , Moss, R. L., Star, R.A,, and Muallem, S. (1993)J . Biol. Chem. 268,19769-19775 23. Minta, A., Kao, J. P. Y., and Tsien, R. Y.(1989)J. Biol. Chem. 264,81714178 24. Muallem, S., Pandol, S., and Beeker, T.G. (1988)Biochem. J. 256, 301-307 25. Zhang, B.-X., Zhao,H., Loessberg, P. A., and Muallem, S. (1992)J. Biol. Chem. 267,15419-15425 26. Matozaki, T.,Martinez, J., and Williams, J. A. (1989)Am. J. Physiol. 257, G594-G600 27. Matozaki, T., Goke, B.,Tsunoda, Y., Rodriquez, M., Martinez, J.,and Williams, J. A. (1990)J. Biol. Chem. 265,62474254 28. Saluja, A. K., Powers, R. E., and Steer, M. L. (1989)Biochem. Biophys. Res. Commun. 164,8-13 Acknowledgments-We thank Drs. Qler Miller and Don Hilgemann 29. Saluja,A. K., Dawra, R. K., Lerch, M. M., and Steer, M. L. (1992)J . Biol. Chem. for constructive comments and Mary Vaughn for expert administrative 267,11202-11207 assistance. 30. Yule, D. I., and Williams, J. A. (1992)J . Biol. Chen. 267, 13830-13835 31. Thorn, P., and Petersen, 0.H. (1993)J . Biol. Chem. 268,23219-23221 REFERENCES 32. Gaisano, H. Y., Wong, D., Sheu, L., and Foskett, K. (1994)Am.J. Physiol. 267, C220-C228 1. Gilman, A. G. (1987)Annu. Rev. Biochem. 56,615-649 33. Zhang, B.-X., and Muallem, S. (1992)J. Biol. Chem. 267,24387-24393 2. Tang, W.J., and Gilman, A. G. (1992)Cell 70,869-872 34. Parker, I., and Ivorra, I. (1990)Science 250,977-979 3. Stryer, L., and Bourne, H. R. (1986)Annu.Reu. Cell Biol. 2, 391-419 4. Simon, M. I., Strathman, M. P., and Gautam, N. (1991)Science 252,802308 35. Kindman, L. A,, and Meyer, T. (1993)Biochemistry 32, 1270-1277 36. Meyer, T.,and Stryer, L. (1990)Proc. Nutl. Acud. Sci. U.S. A. 87,384-3845 5. Berridge, M. J. (1993)Nature 361,315-325 37. Oldershaw, K. A,, Nunn, D. L., and Taylor, C. W. (1991)Biochem. J. 278, 6. Williams, J. A,, and Yule, D. I. (1993)in The Exocrine Pancreas (Go, V. L. W., 705-708 ed) pp. 167-189,Raven Press, Ltd.,New York 38. Bootman, M. K., Berridge, M. J., and Taylor, C. W.(1992)J. Physiol. (Lond.) 7. Williams, J. A,, and Blevins, G. T.,Jr. (1993)Physiol. Reu. 73,701-723 450,163-178 8. Harden, K. T. (1992)in Advances in Second Messenger and Phosphoprotein 39. Pandol, S. J., SchoeEeld, M. S., Sachs, G., and Muallem, S. (1985)J . Biol. Research: Inositol Phosphates and CalciumSignuling (Putney, J.W., Jr., ed) Chem. 260,10081-10086 pp. 11-34,Raven F'ress, Ltd., New York 40. Imine, R. F. (1990)FEBS Lett. 263,5-9 9. Zhao, H.,Khademazad, M., and Muallem, S. (1990)J. Biol. Chen. 265,1482241. Missiaen, L., DeSmedt, H., Droogmans, G., and Casteels, R. (1992)Nature 14827 357,59%602 10. Miyazaki, S. (1988)Deu. Growth & Diffec 30,603410 11. Rooney, T.A,, Sass, E., and Thomas, A.P. (1990)J . Biol. Chem. 265, 10792- 42. Zhang, B.-X., Zhao, H., and Muallem, S. (1993)J . Biol. Chem. 268, 1099711001 10796

spread between signaling units. Ca2+pools with a similar threshold of activation by IP, communicate with each other across the boundaries of the signaling units. This communication is controlled by agonist stimulation. At low agonist concentration or during stimulation with partial agonist such as CCKJ, IP, is produced in a restricted area and reaches low concentrations sufficient to mobilize only the pools with high affinity to IP,. Additional Ca2+pools with high affinity for IP, and coupled to other signaling complexes can be released by increasing the communication between these pools without activating their channels. As the intensity of the stimulus is increased, an increasing fraction of the pool is mobilized due to increased steady state levels of IP,, which can access pools with low sensitivity to IP,. The model also allows for quantal Ca" release, which is due to the combined effect of the restricted spreading of the regulatory signals, the compartmentalization of the Ca2+pool, and theheterogeneity in theresponsiveness t o IP,. This arrangement predicts a complex form of quantal response, the type which is observed at intermediate concentrations of CCKJ (0.41p d . The SLO system appears to be suitable for further testing of several aspects of this model.