497
Journal of Cell Science 110, 497-504 (1997) Printed in Great Britain © The Company of Biologists Limited 1997 JCS8089
Mechanisms for the coordination of intercellular calcium signaling in insulinsecreting cells Dongrong Cao1, George Lin1, Eileen M. Westphale2, Eric C. Beyer2 and Thomas H. Steinberg1,* 1Department of Internal Medicine, Division of Infectious Diseases, Box 8051, Washington University School of Medicine, 660 South Euclid Avenue, St Louis, MO 63110, USA 2Department of Pediatrics, Box 8116, Washington University School of Medicine, 660 South Euclid Avenue, St Louis, MO 63110, USA
*Author for correspondence (e-mail:
[email protected])
SUMMARY Insulin-mediated increases in cytosolic calcium are synchronized among the cells in a pancreatic islet, and result in pulsatile secretion of insulin. Pancreatic beta cells express the gap junction protein connexin43 and are functionally coupled, making gap junctional communication a likely mechanism for the synchronization of calcium transients among islet cells. To define the mechanism by which pancreatic islet cells coordinate calcium responses, we studied mechanically-induced intercellular calcium waves in the communication-deficient rat insulinoma cell line RINm5f, and in RINm5f cells transfected with the gap junction protein connexin43. Both RINm5f and RINm5f cells transfected with connexin43 propagated calcium waves that required release of calcium from intracellular
stores, did not involve gap junctional communication, and appeared to be mediated by autocrine activity of secreted ATP acting on P2U purinergic receptors. Connexin43 transfectants also propagated calcium waves that required gap junctional communication and influx of extracellular calcium through voltage-gated calcium channels. Gap junction-dependent intercellular calcium waves were inhibited by preventing plasma membrane depolarization. These studies demonstrate two distinct pathways by which insulin-secreting cells can coordinate cytosolic calcium rises, and show that it is by ionic traffic that gap junctions synchronize calcium-dependent events in these cells.
INTRODUCTION
during sustained stimulation of β-cell functioning in vivo (Meda et al., 1991). These data suggest that gap junctional communication among pancreatic islet cells regulates insulin secretion, but the mechanism by which this occurs is not understood. Increases in cytosolic free calcium trigger insulin secretion as well as other exocytic processes (Ashcroft et al., 1994), and coordination of calcium transients is a likely mechanism by which gap junctions may facilitate insulin secretion. In this paper we investigated the means by which insulin-secreting cells synchronize cytosolic calcium transients by studying the propagation of calcium transients from cell to cell, termed intercellular calcium waves (Sanderson et al., 1994), induced by mechanical stimulation of a single cell. For these studies we used the rat insulinoma cell line RINm5f (RIN), which unlike normal pancreatic beta cells lack connexin43, and RIN cells transfected with connexin43 to restore intercellular communication. Two mechanisms for intercellular calcium waves have been identified in other cell types. In many cells it is believed that calcium waves spread from cell to cell via gap junctions, in particular by diffusion of inositol trisphosphate (IP3) through junctional pores and subsequent release of IP3-sensitive intracellular calcium stores in the neighboring cells. Thus in airway epithelial cells, intercellular calcium waves propagate in the
In response to increased extracellular glucose concentrations, pancreatic islet cells undergo plasma membrane depolarization and increases in cytosolic calcium that lead to exocytic secretion of insulin (Ashcroft et al., 1994). Synchronous oscillations of both membrane potential (Mears et al., 1995) and cytosolic calcium (Gilon et al., 1993; Longo et al., 1991; Bergsten et al., 1994; Valdeolmillos et al., 1993) have been identified in pancreatic islets, and the latter are reflected in pulsatile insulin secretion (Longo et al., 1991; Bergsten et al., 1994). Intercellular communication among islet cells appears to be necessary for normal insulin secretion. This coordination is presumed to be mediated at least in part by gap junctional intercellular communication. Gap junction structures and the gap junction protein connexin43 have been identified in pancreatic islets (Meda et al., 1991), and gap junction channels can be demonstrated in these cells by dye transfer or electrophysiologic techniques (Perez-Armendariz et al., 1991; Meda et al., 1991). Gap junctional communication appears to be necessary for normal insulin secretion and production (Vozzi et al., 1995). Studies have demonstrated that pharmacologic blockade of gap junction channels in these cells alters insulin secretion (Meda et al., 1990). Furthermore, connexin43 gene transcripts and incidence of junctional coupling are modulated in parallel
Key words: Calcium wave, Purinergic receptor, Gap junction
498
D. Cao and others
absence of extracellular calcium; can be inhibited by depleting intracellular calcium stores, blocking IP3 receptors, or inhibiting phospholipase C activity; and can be initiated by microinjection of IP3 (Hansen et al., 1995; Sanderson et al., 1992; Boitano et al., 1992). Calcium waves among glial cells, or between astrocytes and neurons, are also dependent on release of intracellular calcium (Nedergaard, 1994; Charles et al., 1991, 1993). A second mechanism for the generation of intercellular calcium waves has been identified, one that does not require gap junctional communication but instead relies on autocrine activity of ATP. When rat basophilic leukemia cells are stimulated they secrete ATP that activates purinergic receptors on neighboring cells (Osipchuk and Cahalan, 1992). This in turn triggers inositol phospholipid turnover and release of intracellular calcium stores (Dubyak and El-Moatassim, 1993; Harden et al., 1995). Insulin-secreting cells, including the RINm5f cell line, are known to respond to extracellular ATP with cytosolic calcium transients and insulin secretion (Geschwind et al., 1989; Blachier and Malaisse, 1988; Arkhammar et al., 1990; Hellman, 1987) In the current study, we show that RIN cells express both ATP-dependent and gap junction-dependent intercellular calcium waves, and that intercellular calcium waves mediated by gap junctional communication involve activation of voltagegated calcium channels rather than release of intracellular calcium stores. MATERIALS AND METHODS Cells and chemicals The RINm5f cell line was grown in RPMI 1640 supplemented with 10% heat-inactivated bovine calf serum. A cDNA containing the complete coding sequence for rat Cx43 was inserted into the pSFFVneo vector (Carel et al., 1989; Fuhlbrigge et al., 1988) and transfected into RIN cells using the lipofectin reagent as previously described (Steinberg et al., 1994). Transfected cells were selected in 0.5 mg/ml geneticin, and single cell clones were selected by limiting dilution. Fura-2/AM, Lucifer Yellow, and hydroxycoumarin carboxylic acid were purchased from Molecular Probes (Eugene, OR). Other chemicals were from Sigma (St Louis, MO). Immunofluorescence Adherent cells were fixed in methanol/acetone (1:1) for 2 minutes at room temperature, incubated in rabbit polyclonal anti-Cx43 antiserum (Beyer et al., 1989), and fluorescein-conjugated goat anti-rabbit IgG. Epifluorescence micrographs were taken with a Nikon Optiphot-2 microscope. Dye transfer Chemical coupling via gap junctions was assessed by dye transfer among cells in a monolayer. Microinjection was performed as previously described (Steinberg et al., 1994). Cells were adhered overnight to 60 mm tissue culture dishes, and mounted on a Nikon Diaphot epifluorescence microscope. Lucifer Yellow CH (1%) was injected into single cells with a Narshige picoinjector and micromanipulator and the number of cells that received dye was quantitated after 1 minute. Calcium imaging Fluorescence ratio imaging was performed with an IM-4000 system from Georgia Instruments (Roswell, GA). Cells plated on 25 mm diameter coverslips were loaded with 5 µM fura-2/AM for 30 minutes, washed, and incubated in fresh medium for 15 minutes to allow dye hydrolysis. Cells were washed and incubated in PBS with 1 mM Mg,
1 mM Ca, 6 mM glucose, 20 mM bicarbonate, pH 7.5 and 2.5 mM probenecid to inhibit dye efflux and compartmentalization (Di Virgilio et al., 1988). Coverslips were mounted in a PDMI-2 incubation chamber (Medical Systems, Greenvale, NY) on a Zeiss Axiovert 35, and ratio imaging was performed using two 150 Xenon light sources and monochromators set at excitation wavelengths of 340 nm and 380 nm. Micropipettes were affixed to an Eppendorf micromanipulator and used to mechanically stimulate single cells. The micropipette was withdrawn and image pairs were recorded at intervals using a DAGE MTI CCD 72 camera and image intensifier. Ratio images were generated by the FL-4000 software package. Cytosolic calcium was estimated by generating a calibration curve with solutions of known calcium concentration (Molecular Probes, Eugene OR).
RESULTS Transfection of RIN cells with Cx43 For these studies we used the rat insulinoma cell line RINm5f (RIN) and RIN cells transfected with the gap junction protein connexin43 (RIN/Cx43). RIN cells are communicationdeficient compared to pancreatic beta cells (Meda et al., 1990), and our RIN cells did not express connexins 26, 32, 43, or 46 by RNA blot (not shown) and did not allow intercellular passage of microinjected Lucifer Yellow (Fig. 1) or the smaller dye hydroxycoumarin carboxylic acid (not shown). Furthermore, electrophysiologic studies have concluded that RINm5f cells are not electrically coupled by gap junctions (Banach and Weingart, 1996). We transfected RIN cells with a vector containing rat Cx43 cDNA and selected clones of stable transfectants in G418. RIN/Cx43 cells expressed Cx43 in a linear punctate pattern at the plasma membrane, consistent with the assembly of Cx43 into gap junction plaques (Fig. 1). To assess functional coupling, we microinjected the fluorescent dye Lucifer Yellow into single cells in confluent monolayers, and assessed dye transfer by fluorescence microscopy. Microinjected RIN/Cx43 cells transferred Lucifer Yellow to neighboring cells (29±21 cells/injected cell, n=36), but microinjected RIN cells (0.3±0.6 cells/injected cell, n=28), or RIN/Cx43 incubated in medium containing 3.5 mM heptanol to inhibit gap junction communication (1.7±1.7 cells/injected cell, n=19) did not (Fig. 1). Thus in RIN/Cx43 cells we have reconstituted the expression of Cx43 and dye coupling that is present in pancreatic islet beta cells. Both RIN and RIN/Cx43 propagate intercellular calcium waves To assess the synchronization of calcium signaling among these cells, we studied the propagation of mechanicallyinduced intercellular calcium waves in cell monolayers. Adherent RIN and RIN/Cx43 cells were loaded with the fluorescent calcium indicator fura-2, a single cell was mechanically stimulated by micropipette, and the intracellular calcium concentration was measured by fluorescence ratio imaging (Fig. 2). Because the parent RIN cells did not express Cx43 and were not dye coupled, we were surprised to find that these cells were able to propagate cytosolic calcium transients from cell to cell. After mechanical stimulation of a single RIN cell, calcium transients spread to most of the cells within a cell grouping, reached maximum extent in 8 seconds, and subsided within 25 seconds. In confluent monolayers of cells, almost all
Calcium waves in insulin-secreting cells
499
Fig. 1. Localization of Cx43 and dye transfer in RIN cells and RIN/Cx43 transfectants. (A,B) Immunostaining of Cx43 in RIN cells (A) and in RIN/Cx43 transfectants (B). Cells were fixed and incubated in polyclonal antiCx43 and fluorescein-conjugated anti-rabbit IgG. (C,D) Dye transfer in RIN and RIN/Cx43. Single cells in adherent monolayers were microinjected with Lucifer Yellow, and micrographs were taken after 1 minute. Arrows indicate injected cells. (C) RIN cells, (D) RIN/Cx43 cells.
the cells in the field of view, approximately 30-50 cells, were frequently involved. RIN/Cx43 cells also propagated calcium waves after mechanical stimulation. These waves also reached maximum extent in 8 seconds, and involved similar numbers of cells. The only noticeable difference between RIN calcium waves and RIN/Cx43 calcium waves was that RIN/Cx43 calcium waves persisted about twice as long, and did not subside for about 45 seconds. Intercellular calcium waves in RIN cells are not mediated by gap junctional communication and involve release of intracellular calcium stores Because RIN cells did not express connexin43 and were not dye coupled, it seemed unlikely that mechanically-induced calcium waves in these cells were propagated by a gap junction-dependent mechanism. This was further suggested by the demonstration that intercellular calcium waves in RIN cells were not blocked by the gap junction inhibitor heptanol (not shown), and that calcium waves were able to spread between clusters of cells that did not appear to be in physical contact (Fig. 2). Although these islands of cells did not appear to be connected under phase microscopy, it was possible that fine processes between the cells allowed cell-cell contact. Therefore, to confirm that cell-cell contact was not required for the propagation of calcium waves in RIN cells, we scraped RIN cell monolayers with a 25 gauge needle. This created a disruption, approximately two cells wide, in the monolayer. The monolayers were allowed to recover for 1-2 hours, the cells loaded with fura-2, and a cell near the disruption in the monolayer was mechanically stimulated (Fig. 3). The calcium wave spread rapidly to the cells in contact with the stimulated cell, and also spread, although less vigorously, to the cells on the other side of the scrape. Thus, calcium waves in RIN cells did not require direct cell-cell contact, and therefore did not require gap junctional communication. For both mechanisms of calcium wave propagation that have been previously described, calcium transients in neighboring cells involve release of calcium from intracellular stores. We therefore asked whether this was true of the calcium waves we
observed in RIN cells. We depleted intracellular calcium stores by incubating monolayers of RIN cells in medium containing 50 nM thapsigargin, an inhibitor of calcium uptake into intracellular stores. After addition of thapsigargin, the intracellular calcium rapidly rose in all cells, and gradually returned nearly to baseline within 30 minutes (not shown). Intercellular calcium waves induced by mechanical stimulation of a single RIN cell were not observed after intracellular calcium stores were depleted with thapsigargin, although a rise in calcium was detected in the stimulated cell (Fig. 4). Intercellular calcium waves in RIN cells are propagated by activation of purinergic receptors Further studies demonstrated that intercellular calcium waves in RIN cells appeared to be transmitted by autocrine activity of secreted ATP on purinergic receptors as has been reported for the propagation of intercellular calcium waves in rat basophil leukemia cells (Osipchuk and Cahalan, 1992). In this pathway, ATP secreted by stimulated cells activates purinergic receptors on neighboring cells, and leads to the generation of IP3 and release of calcium from intracellular stores (Dubyak and El-Moatassim, 1993; Harden et al., 1995). The addition of extracellular ATP or UTP (1 mM) induced cytosolic calcium transients in RIN and RIN/Cx43 cells (not shown), demonstrating that these cells express purinergic receptors, probably of the P2U (P2Y2) class. Concentrations of ATP as low as 1 µM reproducibly caused calcium transients in RIN cells. Extracellular ATP did not elicit calcium transients in RIN or RIN/Cx43 cells after intracellular calcium stores were depleted, showing that the ATP-induced calcium transients were indeed initiated by release of calcium stores. Because P2U receptors can be desensitized by exposure to ATP, we examined the effects of ATP pretreatment on RIN cell calcium waves. For these experiments we compared the ability of different concentrations of ATP to induce receptor desensitization to subsequent reapplication of ATP and to block mechanically-induced calcium waves (Table 1). Addition of 1 µM to 1 mM ATP induced calcium transients, and also desensitized ATP receptors so that further application of ATP yielded
500
D. Cao and others Table 1. ATP-induced calcium transients, desensitization of ATP receptors, and ATP-induced inhibition of mechanically induced calcium wave in RIN cells [ATP]
ATP-induced calcium transients
Calcium transient after repeat addition of ATP
Mechanicallyinduced calcium wave after ATP
0.1 µM 1 µM 1 mM
7/9 5/6 5/5
5/5 0/5 0/4
5/5 0/5 0/4
Adherent RIN cells were loaded with fura-2 and ratio imaging was performed. ATP-induced calcium transients were scored as the presence of ATP-induced rise in cytosolic calcium after initial application of ATP. Desensitization of ATP receptors was assessed as the presence of calcium transients after subsequent reapplication of the same concentration of ATP. The ability of ATP to inhibit mechanically-induced intercellular calcium waves was assessed by scoring the presence of calcium waves after application of ATP. Results are expressed as responses/total attempts.
Fig. 2. Calcium waves in RIN and RIN/Cx43 cells. Adherent RIN cells (left panels) or RIN/Cx43 cells (right panels) were loaded with the calcium indicator fura-2, and single cells were mechanically stimulated. In these cells, which had not been incubated in thapsigargin, calcium waves spread between islands of cells that did not have physical contact with each other. Asterisk indicates the stimulated cell. Time after stimulation in seconds is indicated on each panel.
no calcium response. The lack of a calcium transient after the second application of ATP was not due to depletion of calcium stores, because subsequent addition of thapsigargin led to a rapid increase in cytosolic calcium (not shown). These concentrations of ATP also inhibited mechanically-induced calcium waves. Addition of 0.1 µM ATP elicited calcium transients in
Fig. 3. Calcium waves in disrupted monolayers. Monolayers of RIN cells were disrupted with a 25 gauge needle, cells were loaded with fura-2, and single cells near the disruption were mechanically stimulated. Calcium waves propagated on both sides of the disruption, although spread to the opposite side of the scrape was less pronounced. Asterisk indicates the stimulated cell. Time after stimulation in seconds is indicated on each panel.
RIN cells, but did not desensitize receptors, because subsequent readdition of ATP caused a repeated calcium transient. This concentration of ATP also did not prevent subsequent mechanically-induced calcium waves. Application of 0.01 µM ATP had no effect on RIN cells. Thus ATP-mediated desensitization of purinergic receptors and ATP-mediated inhibition of mechanically-induced calcium waves had identical dose-response characteristics. A further indication that purinergic receptors were involved in intercellular calcium waves in RIN cells is that these waves were blocked by suramin, an inhibitor of purinergic receptors. The same concentration of suramin that inhibited mechanically-induced intercellular calcium waves (100 µM) was also the concentration of suramin required to block ATPmediated calcium responses (Table 2). This concentration of suramin did not inhibit gap junction-mediated intercellular dye transfer in RIN/Cx43 cells (not shown). The above data show that intercellular calcium waves in RIN cells do not involve gap junctional communication: rather, they involved release of intracellular calcium stores initiated by activation of purinergic receptors, which occurs via an IP3mediated pathway.
Calcium waves in insulin-secreting cells
501
Fig. 4. Intercellular calcium waves in RIN/Cx43 after depletion of calcium stores with thapsigargin. Adherent RIN (left panels) and RIN/Cx43 (right panels) cells were loaded with the calcium indicator dye fura-2 and incubated in medium containing 50 nM thapsigargin for 30 minutes. A single cell was mechanically stimulated and fluorescence ratio images were recorded at 5 seconds intervals. Time after stimulation in seconds is indicated on each panel.
Table 2. Inhibition of ATP-induced calcium transients and mechanically-induced intercellular calcium waves by suramin [Suramin] in µM 10 100
ATP-induced calcium transients after suramin
Mechanically-induced intercellular calcium waves after suramin
4/4 0/4
4/4 0/4
Results are expressed as responses/total attempts.
Intercellular calcium waves in thapsigargin-treated RIN/Cx43 cells are mediated by gap junctional communication and activation of L-type calcium channels In contrast to RIN cells, RIN/Cx43 cells were able to propagate intercellular calcium waves after intracellular calcium stores were depleted with thapsigargin (Fig. 4). This finding showed that RIN cells transfected with Cx43 now expressed a new gap junction-dependent mechanism for calcium wave propagation, and that this mechanism required influx of extracellular calcium rather than release of intracellular calcium stores or activation of purinergic receptors. These conclusions were supported by further experiments performed in RIN/Cx43 cells after the depletion of intracellular calcium stores with thapsigargin to inhibit the ATPdependent mechanism described above. First, intercellular calcium waves in thapsigargin-treated RIN/Cx43 cells were blocked by addition of 2 mM heptanol (not shown), which inhibits Cx43-mediated gap junctional communication, but which did not affect ATP-mediated calcium waves in RIN cells as mentioned above. In addition, mechanically-induced calcium waves propagated by thapsigargin-treated RIN/Cx43 cells were never seen to move between clusters of cells that
were not in physical contact, providing further evidence that calcium waves in thapsigargin-treated RIN/Cx43 cells required gap junctional communication. Second, treatments which blocked calcium waves in RIN cells above by blocking purinergic receptors had no effect on calcium waves in thapsigargin treated RIN/Cx43 cells. Thus neither 100 µM suramin nor desensitization of ATP receptors prevented calcium wave propagation in RIN/Cx43 cells (not shown). We confirmed that calcium waves in thapsigargin-treated RIN/Cx43 cells required influx of extracellular calcium by assessing calcium wave propagation in calcium-depleted medium. Under these conditions, calcium wave propagation
Fig. 5. Inhibition of calcium waves in thapsigargin-treated RIN/Cx43 cells with nifedipine and diazoxide. RIN/Cx43 cells loaded with fura-2 were incubated in 50 nM thapsigargin for 30 minutes, after which time intercellular calcium waves could be generated by mechanical stimulation (A,C). Cells were then incubated in 10 µM nifedipine (B) or 150 µM diazoxide (D) and calcium waves were monitored. Pictures reflect maximum extent of calcium waves.
502
D. Cao and others
did not occur, demonstrating the requirement for calcium influx in these waves (not shown). We next sought to identify the plasma membrane calcium channel responsible for this calcium influx. RIN cells express L-type voltage-gated calcium channels, which are thought to trigger insulin secretion in pancreatic beta cells. Because gap junctions mediate electrical coupling in cells such as cardiac myocytes, it seemed likely that Cx43 propagated intercellular calcium waves in RIN/Cx43 cells by allowing ionic traffic between neighboring cells, resulting in changes in plasma membrane electrical potential, which in turn might synchronize voltage-gated calcium channel activity. We therefore determined the effect of the L-type calcium channel antagonist nifedipine on intercellular calcium waves in RIN/Cx43. After depletion of intracellular calcium stores with thapsigargin, 10 µM nifedipine blocked further attempts at eliciting calcium waves by mechanical stimulation (Fig. 5). Nifedipine also blocked calcium waves in RIN/Cx43 cells after ATP desensitization (not shown). In contrast, nifedipine had no effect on calcium waves in RIN cells or in RIN/Cx43 cells with replete intracellular stores, demonstrating that the Ltype calcium channels were not required for the propagation of calcium waves via the ATP-mediated mechanism. These studies showed that extracellular calcium influx was required for gap junction-dependent calcium waves, and that L-type voltage gated calcium channels mediated this calcium influx. Because these calcium channels are regulated by the plasma membrane electrical potential, these results suggest that the role of gap junctional communication in these calcium waves is to propagate changes in plasma membrane potential between cells via ionic coupling. To confirm that plasma membrane depolarization was the ATP
ATP ATP IP3
ATP
Ca
Ca
Ca
A: RIN cells
Ca
ψ +++
Ca
Ca Ca
+++ +++ Cx43
B: RIN/Cx43 after thapsigargin Fig. 6. Two mechanisms of intercellular calcium waves in RIN and RIN/Cx43 cells. (A) Both RIN and RIN/Cx43 transmit calcium waves that do not involve gap junctional communication. They require release of intracellular calcium stores and appear to be mediated by autocrine activity of ATP. (B) RIN/Cx43, but not RIN, propagate calcium waves after depletion of intracellular calcium stores with thapsigargin. These calcium waves involve gap junctional communication and the activation of voltage-gated calcium channels.
signal that mediated calcium channel activation, we assessed calcium waves after inhibiting membrane depolarization with the KATP channel agonist diazoxide. Because diazoxide maintains KATP channels in the open state, diazoxide inhibits plasma membrane depolarization by maintaining high plasma membrane potassium conductance. RIN/Cx43 cells were incubated in medium containing 50 nM thapsigargin for 30 minutes and, as before, mechanical stimulation of a single cell resulted in intercellular calcium waves after thapsigargin treatment. However, after subsequent addition of 150 µM diazoxide, mechanical stimulation of a single cell no longer resulted in propagation of a calcium wave, although again a calcium rise was seen in the stimulated cell (Fig. 5). We next asked whether diazoxide affected ATP-induced calcium waves, and found that RIN/Cx43 cells not pretreated with thapsigargin (i.e. with intact intracellular calcium stores) propagated mechanically-induced calcium waves both before and after the addition of diazoxide. These results demonstrate that depolarization of the plasma membrane was required for the propagation of gap-junction mediated intercellular calcium waves. Thus, the signal transmitted via gap junctions results in plasma membrane depolarization in neighboring cells. Electrical coupling mediated by ionic traffic through gap junctions would appear to be the most likely mechanism for this plasma membrane depolarization, although we cannot rule out the possibility that gap junctions allow the intercellular passage of a molecule that results in depolarization of neighboring cells. DISCUSSION In these experiments we examined the propagation of mechanically-induced intercellular calcium waves as a model for the synchronization of intracellular calcium transients induced by physiologic stimuli. We identified two distinct mechanisms for these waves in RIN and RIN/Cx43 cells (Fig. 6). In RIN/Cx43 cells, calcium waves were propagated by gap junctional communication, plasma membrane depolarization, and activation of voltage-gated calcium channels. Pancreatic beta cells are coupled by Cx43 as are our RIN/Cx43 transfectants, and inhibition of gap junctional communication blocks insulin production and secretion (Vozzi et al., 1995; Meda et al., 1990), but the mechanism of these effects is unknown. Our data provide direct evidence that Cx43-mediated gap junctional communication allows the coordinated activation of voltage-gated calcium channels; they strongly suggest that the capacity of Cx43 to mediate electrical coupling via transjunctional ionic fluxes is responsible for the intercellular spread of calcium transients. It seems likely that gap junctional communication synchronizes insulin secretion in pancreatic islets by this mechanism. In addition, these studies suggest that in other cells expressing voltage-gated channels, gap junctional communication may coordinate events by allowing electrical coupling of cells and thereby coordinating the activity of these channels. A second possibility would be that gap junctions allows passage of a molecular species that influences plasma membrane potential by some other means. In these studies, we found no evidence that Cx43 mediated mechanically-induced intercellular calcium waves by eliciting the release of intracellular calcium stores in neighboring cells,
Calcium waves in insulin-secreting cells that is, by allowing transfer of a signaling molecule such as IP3 to move from cell to cell and release intracellular stores. In studies of tracheal epithelial cells and in glioma cells, it has been postulated that intercellular spread of IP3 was responsible for calcium wave propagation. If this mechanism were operative in RIN/Cx43 cells, then calcium waves in RIN/Cx43 cells should not be inhibited by blocking both purinergic receptors (by ATP-mediated desensitization) and L-type calcium channels. However, after addition of both ATP and nifedipine, we saw no calcium wave propagation in RIN/Cx43. It has recently been shown that caged IP3 can permeate gap junctions comprised of Cx43 and Cx37 (Carter et al., 1996). It is possible that mechanical stimulation of RIN/Cx43 cells does not generate IP3; it is also possible that although RIN/Cx43 cells are able to transfer Lucifer Yellow, they are unable to transfer sufficient quantities of IP3 to propagate intercellular calcium waves. Both RIN and RIN/Cx43 transfectants propagated intercellular calcium waves by a mechanism that did not involve gap junctional communication. In this instance the calcium waves appeared to be propagated by stimulation of P2U purinergic receptors, presumably by secreted ATP. An analogous mechanism has been identified in rat basophil leukemia cells (Osipchuk and Cahalan, 1992). ATP released from the stimulated cell activates P2U purinergic receptors, resulting in phospholipase activity, IP3 generation, and release of intracellular calcium stores. This rise in intracellular calcium is presumably instrumental in mediating release of insulin-containing granules, which also contain significant quantities of ATP (Leitner et al., 1975). Therefore insulin secretion is accompanied by ATP secretion, which would act as the regenerative signal that allows propagation of the calcium rise to more distant cells. The rate and extent of this form of communication in vivo would depend on factors such as the local volume of extracellular fluid into which ATP is released, velocity of flow, and the activity of ecto-nucleotidases. Purinergic receptors in pancreatic beta cells, and indeed the possibility of regenerative signaling via secreted ATP, have been recognized for almost a decade (Geschwind et al., 1989; Blachier and Malaisse, 1988; Arkhammar et al., 1990; Hellman, 1987). Because we initiated calcium waves by mechanically stimulating single cells, it is likely that in some instances we caused transient disruptions of the plasma membrane in the stimulated cell, which would result in release of intracellular ATP. Thus it is possible that activation of purinergic receptors in neighboring cells was caused by ATP released from cell disruption, and that no regenerative process, i.e. exocytic secretion of ATP, occurred in our experiments. We do not favor this hypothesis for two reasons. First, the extent of calcium wave propagation, and the lack of signal decay over a distance of many cells, argue for the presence of a regenerative signal. Second, when we induced calcium waves in monolayers that had been scraped with a needle (Fig. 3), the calcium wave propagated more rapidly and to a greater extent in cells on the same side of the scrape as the mechanically stimulated cell, and spread much less strongly to cells on the opposite side of the disruption in the monolayer. If all of the ATP was released from the stimulated cell, then all cells at an equal radius from that cell should respond in the same fashion, regardless of which side of the scrape they were on. The likely explanation for the dif-
503
ference in propagation on the two sides of the disruption is that as each cell is activated, it secretes ATP. Across the distance of the scrape where no cells are present there is no regenerative signal, and the ATP signal becomes attenuated rather than regenerated. Mechanically induced calcium waves are not only an attribute of insulinoma cells in culture, because we have found that these waves also occur in isolated rat pancreatic islets (unpublished data). It is not clear at present how the two mechanisms for the synchronization of calcium transients interact in intact pancreatic islets. They may be activated in vivo at different times and in response to different signals. Because one of these mechanisms involves gap junctional communication and the other involves secreted ATP, they would be regulated independently. It is also unclear why synchronization of calcium responses is important in these cells. For example, because the whole islet is exposed to the same fluctuations in extracellular glucose, each cell within the islet should be capable of generating a similar response to an increase in the serum glucose concentration. Nevertheless, gap junctions appear to be important for insulin secretion, and synchronization of calcium responses by both gap junction dependent and independent mechanisms may influence insulin secretion. This work was supported by NIH grant DK46686, Pilot and Feasibility Funds from the Diabetes Research and Training Center at Washington University School of Medicine, and a grant from the American Diabetes Association. We thank Lynn Obermoeller for technical assistance, and Drs Eric Brown, Michael Koval, Michael McDaniel, Stanley Misler and Colin Nichols for critical reading of the manuscript.
REFERENCES Arkhammar, P., Hallberg, A., Kindmark, H., Nilsson, T., Rorsman, P. and Berggren, P. O. (1990). Extracellular ATP increases cytoplasmic free Ca2+ concentration in clonal insulin-producing RINm5F cells. A mechanism involving direct interaction with both release and refilling of the inositol 1,4,5-trisphosphate-sensitive Ca2+ pool. Biochem. J. 265, 203-211. Ashcroft, F. M., Proks, P., Smith, P. A., Ammala, C., Bokvist, K. and Rorsman, P. (1994). Stimulus-secretion coupling in pancreatic β cells. J. Cell. Biochem. 55S, 54-65. Banach, K. and Weingart, R. (1996). Connexin43 gap junctions exhibit asymmetrical gating properties. Pflugers Arch. 431, 775-785. Bergsten, P., Grapengiesser, E., Gylfe, E., Tengholm, A. and Hellman, B. (1994). Synchronous oscillations of cytoplasmic Ca2+ and insulin release in glucose-stimulated pancreatic islets. J. Biol. Chem. 269, 8749-8753. Beyer, E. C., Kistler, J., Paul, D. L. and Goodenough, D. A. (1989). Antisera directed against connexin43 peptides react with a 43-kD protein localized to gap junctions in myocardium and other tissues. J. Cell Biol. 108, 595-605. Blachier, F. and Malaisse, W. J. (1988). Effect of exogenous ATP upon inositol phosphate production, cationic fluxes and insulin release in pancreatic islet cells. Biochim. Biophys. Acta 970, 222-229. Boitano, S., Dirksen, E. R. and Sanderson, M. J. (1992). Intercellular propagation of calcium waves mediated by inositol trisphosphate. Science 258, 292-295. Carel, J.-C., Frazier, B., Ley, T. J. and Holers, V. M. (1989). Analysis of epitope expression and the functional repertoire of recombinant complement receptor 2 (CR2/CD21) in mouse and human cells. J. Immunol. 143, 923930. Carter, T. D., Chen, X. Y., Carlile, G., Kalapothakis, E., Ogden, D. and Evans, W. H. (1996). Porcine aortic endothelial gap junctions: identification and permeation by caged InsP3. J. Cell Sci. 109, 1765-1773. Charles, A. C., Merrill, J. E., Dirksen, E. R. and Sanderson, M. J. (1991). Intercellular signaling in glial cells: calcium waves and oscillations in response to mechanical stimulation and glutamate. Neuron 6, 983-992. Charles, A. C., Dirksen, E. R., Merrill, J. E. and Sanderson, M. J. (1993).
504
D. Cao and others
Mechanisms of intercellular calcium signaling in glial cells studied with dantrolene and thapsigargin. GLIA 7, 134-145. Di Virgilio, F., Steinberg, T. H., Swanson, J. A. and Silverstein, S. C. (1988). Fura-2 secretion and sequestration in macrophages. A blocker of organic anion transport reveals that these processes occur via a membrane transport system for organic anions. J. Immunol. 140, 915-920. Dubyak, G. R. and El-Moatassim, C. (1993). Signal transduction via P2purinergic receptors for extracellular ATP and other nucleotides. Am. J. Physiol. 265, C577-C606. Fuhlbrigge, R. C., Fine, S. M., Unanue, E. R. and Chaplin, D. D. (1988). Expression of membrane interleukin 1 by fibroblasts transfected with murine pro-interleukin 1α cDNA. Proc. Nat. Acad. Sci. USA 85, 5649-5653. Geschwind, J. F., Hiriart, M., Glennon, M. C., Najafi, H., Corkey, B. E., Matschinsky, F. M. and Prentki, M. (1989). Selective activation of Ca2+ influx by extracellular ATP in a pancreatic beta-cell line (HIT). Biochim. Biophys. Acta 1012, 107-115. Gilon, P., Shepherd, R. M. and Henquin, J.-C. (1993). Oscillations of secretion driven by oscillations of cytoplasmic Ca2+ as evidenced in single pancreatic islets. J. Biol. Chem. 268, 22265-22268. Hansen, M., Boitano, S., Dirksen, E. R. and Sanderson, M. J. (1995). A role for phospholipase C activity but not ryanodine receptors in the initiation and propagation of intercellular calcium waves. J. Cell Sci. 108, 2583-2590. Harden, T. K., Boyer, J. L. and Nicholas, R. A. (1995). P2-purinergic receptors: subtype-associated signaling responses and structure. Annu. Rev. Pharmacol. Toxicol. 35, 541-579. Hellman, B. (1987). External ATP mimics carbachol in initiating calcium mobilization from pancreatic beta-cells conditioned by previous exposure to glucose. Br. J. Pharmacol. 92, 281-289. Leitner, J. W., Sussman, K. E., Vatter, A. E. and Schneider, F. H. (1975). Adenine nucleotides in the secretory granule fraction of rat islets. Endocrinology 96, 662-677. Longo, E. A., Tornheim, K., Deeney, J. T., Varnum, B. A., Tillotson, D., Prentki, M. and Corkey, B. E. (1991). Oscillations of cytosolic free Ca2+, oxygen consumption, and insulin secretion in glucose-stimulated rat pancreatic islets. J. Biol. Chem. 266, 9314-9319.
Mears, D., Sheppard, N. F. Jr, Atwater, I. and Rojas, E. (1995). Magnitude and modulation of pancreatic β-cell gap junction electrical conductance in situ. J. Membr. Biol. 146, 163-176. Meda, P., Bosco, D., Chanson, M., Giordano, E., Vallar, L., Wollheim, C. and Orci, L. (1990). Rapid and reversible secretion changes during uncoupling of rat insulin-producing cells. J. Clin. Invest. 86, 759-768. Meda, P., Chanson, M., Pepper, M., Giordano, E., Bosco, D., Traub, O., Willecke, K., El Aoumari, A., Gros, D., Beyer, E. C., Orci, L. and Spray, D. C. (1991). In vivo modulation of connexin 43 gene expression and junctional coupling of pancreatic β-cells. Exp. Cell Res. 192, 469-480. Nedergaard, M. (1994). Direct signaling from astrocytes to neurons in cultures of mammalian brain cells. Science 263, 1768-1771. Osipchuk, Y. and Cahalan, M. (1992). Cell-to-cell spread of calcium signals mediated by ATP receptors in mast cells. Nature 359, 241-244. Perez-Armendariz, M., Roy, C., Spray, D. C. and Bennett, M. V. (1991). Biophysical properties of gap junctions between freshly dispersed pairs of mouse pancreatic beta cells. Biophys. J. 59, 76-92. Sanderson, M. J., Charles, A. C. and Dirksen, E. R. (1992). Mechanical stimulation and intercellular communication increases intracellular Ca2+ in epithelial cells. Cell Regul. 1, 585-596. Sanderson, M. J., Charles, A. C., Boitano, S. and Dirksen, E. R. (1994). Mechanisms and function of intercellular calcium signaling. Mol. Cell. Endocrinol. 98, 173-187. Steinberg, T. H., Civitelli, R., Geist, S. T., Robertson, A. J., Veenstra, R. D., Wang, H.-Z., Warlow, P. M., Hick, E., Laing, J. G. and Beyer, E.C. (1994). Connexin43 and connexin45 form gap junctions with different molecular permeabilities in osteoblastic cells. EMBO J. 13, 744-750. Valdeolmillos, M., Nadal, A., Soria, B. and Garcia-Sancho, J. (1993). Fluorescence digital image analysis of glucose-induced [Ca2+]i oscillations in mouse pancreatic islets of Langerhans. Diabetes 42, 1210-1214. Vozzi, C., Ullrich, S., Charollais, A., Philippe, J., Orci, L. and Meda, P. (1995). Adequate connexin-mediated coupling is required for proper insulin production. J. Cell Biol. 131, 1561-1572. (Received 21 August 1996 - Accepted 4 December 1996)