Cellular Responses to Stimulation of the M5 Muscarinic Acetylcholine Receptor as Seen in Murine L Cells*

The membrane signaling properties of the neuronal type-5 muscarinic acetylcholine receptor (M5 AChR) as expressed in murine L cells were studied. Recipient Ltk-cells responded to ATP acting through a P2- purinergic receptor by increasing phosphoinositide hy- drolysis a-fold but were unresponsive to 17 receptor agonists that are stimulatory in other cells. L cells expressing the M5 AChR responded to carbachol (CCh) with an -20-fold increase in phospholipase C activity, mobilization of Ca2+ from endogenous stores, causing a transient peak increase in the intracellular concen- tration of Ca2+ ([Ca2+li), influx of extracellular Ca2+, causing a sustained increase in [Ca2+]i dependent on extracellular Ca’+, and release of [3H]arachidonic acid from prelabeled cells, without altering resting or pros- taglandin El-elevated intracellular CAMP levels. None of the effects of the M5 AChR were inhibited by pertussis

in pertussis toxin-treated cells. This is consistent with existence of a kinetic control of the sustained [Ca'+]i change by a receptor-G protein-dependent mechanism independent of the IP3 effector site(s) (e.g. pulsatile activation of phospholipase C and/or pulsatile activation of a receptor/G protein-operated plasma membrane Ca2+ channel). Thus, the non-excitable L cell may be a good model for studying [Ca2+]i regulations, as may occur in other nonexcitable cells of which established cell lines do not exist, and for studying of receptors that as yet cannot be studied in their natural environment.
The neurotransmitter acetylcholine acting through musca-*Supported in part by National

Institute of Health Research
Grants HL-34077, DK-19318, and HD-09581 (to L. B.), DK-41244 (to M. B.) and HL- rinic receptors has a wide variety of effects which are conditioned by the cell type and the molecular identity of both the receptor with which acetylcholine interacts and of the coupling proteins with which the receptors interact. Thus, acetylcholine may cause at the biochemical level the stimulation of phosphoinositide breakdown (2), which in turn leads through formation of the second messenger inositol trisphosphate to release Ca*+ from intracellular Ca2+ stores (3) or the inhibition of adenylyl cyclase (4) and, if present, concomitant stimulation of K' channels (5,6). Other cellular responses to muscarinic receptor stimulation are inhibition of the M-type K+ current in neuronal cells (7), inhibition of Ca2+ channels (8) and stimulation of Ca*+ influx (9), and may include increases in CAMP levels (10) and release of arachidonic acid (11). All these effects of acetylcholine are mediated by a set of muscarinic receptors which were subdivided on the basis of their pharmacological behavior with respect to blockers such as pirenzepine and AF-DX 116 and the primary tissues in which they are expressed into Ml or neuronal, M2 or cardiac, and M3 or glandular (12)(13)(14). More recently, purification and molecular cloning revealed the existence of a family of five distinct muscarinic receptors, termed Ml through M5 (reviewed in Ref. 15). Based on their structural similarity and biochemical effects (or lack thereof) on inhibition of adenylyl cyclase and on phosphoinositide turnover these receptors have been grouped as M2-like, which include the M2 and M4 receptors and cause inhibition of adenylyl cyclase in isolated membranes, and Ml-like, which comprise the Ml, M3 and M5 receptors and cause stimulation of phospholipase C. Ml, M2, and M3 receptors are found rather widely distributed in both neuronal and non-neuronal tissues (16,17). The M4 and M5 receptors have thus far been found only in neuronal tissues, the expression being most restricted for the M5 receptors (16-19). This makes the M5 receptor essentially inaccessible to biochemical analysis of its mode of action. In a previous publication we showed that the M5 receptor' when expressed in murine L cells stimulates phospholipase C activity as seen in intact cell studies but does not affect adenylyl cyclase as measured in isolated membranes (15). The present work was undertaken to explore other potential signaling pathways used by the M5 receptor as seen in the murine L cell, such as causing the release of arachidonic acid, as shown for Ml and M3 receptors (ll), and changing intracellular Ca2+ concentrations.
We also tested whether the M5 receptor would alter CAMP levels in the L intact cell, for it was reported have this effect in Chinese hamster ovary cells (20). '  Earlier investigations studied 45Ca2+ fluxes (e.g. 21,22) or "Rb+ efflux, which reflected the activity of Ca'+-dependent K' channels (23). With the advent of fluorescent Ca*+ indicators such as quin-2 (24) and fura-(25), [Ca*+]i transients in response to agonists have been assessed more directly and shown to follow a common biphasic pattern formed of a large and transient peak response and a lasting or sustained response. The peak response is independent of extracellular Ca2+ and is accepted as being the result of inositol trisphosphate-induced release from an internal Ca2+ store. In contrast to the peak response, the sustained response is dependent on extracellular Ca2+ and due to influx of Ca2+ into the cell by mechanisms that are not well understood (reviewed in Ref. 26). The decreases in [Ca*+]i after the rise to the initial peak level is thought to be due to active extrusion of Ca2+ from the cells as deduced from 45Ca2+ efflux studies (22,27). Quantitatively the two phases of the response observed in the various cells vary remarkably in their details, the sustained response arising in some cells faster than the peak response (28) and lagging behind the peak response in other cells (29). Further, the sustained response may maintain [Ca"'], as high as 70% of that of the peak response and decay so slowly as to appear constant (29), or it may be but 40% of the peak response and decay to close to control values within a few minutes (uide infru).
The mechanism by which Ca*+ enters cells during the sustained phase of the response varies as well with cell type. It may involve voltage-gated Ca*+ channels, as seen through their blockade with Ca2+ channel blockers (e.g. Ref. 30) or their oscillations concordant with changes in membrane potential (31), the activation of a sodium influx pathway in tandem with a compensatory Na'/ Ca2+ exchange mechanism (32), and it may enter through unknown pathways not involving electrogenically active ionic channels (e.g. Ref. 33). In view of the ability of the M5 receptor to stimulate phosphoinositide hydrolysis, we characterized its effects on intracellular Ca*+ transients as seen in murine L cells.
Murine Ltk-cells were developed in 1963 by Kit et al. (34) while studying the functional importance of thymidine kinase and are easily transfected with genomic or cloned DNA (35). They have therefore been used rather widely as expression systems either to clone new genes (e.g. Refs. 36,37)  3 The non-excitable, thymidine kinase-negative L cells (38) developed by Kit et al. (34) and used in our studies, differ from the electrically active L cells (74), also referred to as A9 and A9 L cells, developed by Littlefield (75) and used by Axelrod, and collaborators to characterize electrical (e.g. 76, 77) and phospholipase responses (e.g. 11, 52) to transfected muscarinic receptors.
intrinsic activities such as binding and/or ionic channel activity of transfected molecules but also in studying the signaling pathways that such molecules may affect. With regards to G protein-coupled receptors, L cells contain both cholera and pertussis toxin substrates, a prostaglandin (PG) stimulatable adenylyl cyclase, and a purinergic (ATP) stimulatable phospholipase C system which we showed in our previous publication to respond also to the transfected M5 acetylcholine receptor (15). We show below that L cells also contain the biochemical machinery that allows it to respond to M5  with NaOH to 7.4 at room temperature) at a density of 2 x lo6 cells/ml.
Two ml of the cell suspension were saved as nonloaded control.
The remainded of the cells were pelleted and resuspended in 2.5 ml of ECS buffer prewarmed to 37 "C and all subsequent manipulations were carried out keeping the cells in the dark to prevent bleaching of the fluorescent dve. To load cells with fura-2. the suspension (20)(21)(22)(23)(24)(25) x lo6 cells/mi) was incubated for 30 min at 37 "C with 20 pM of the acetoxymethyl ester of fura-(fura-P/AM, added in 25 ~1 of dimethyl sulfoxide), diluted with ECS buffer to a density of 2 x lo6 cells/ml, and incubated for another 30 min at 37 "C. The cells were then pelleted, resuspended at a density of 2 X 106/ml in ECS at room temperature, distributed in 2-ml fractions into 15-ml polyethylene centrifuge tubes, and kept at room temperature until used. Prior to use the cells were washed twice with 2 ml of ECS buffer (100 X g, 4 min), resuspended in the same volume, and added to a quartz cuvette, Fluorescence  cells were grown to confluence in 6-well plates in the presence of [3H]inositol, washed, and exposed to 10 mM LiCl and 100 pM CCh as outlined on the figure. After the times indicated the reactions were stopped, deproteinized, and the [3H]inositol phosphates separated by Dowex 1 chromatography as described (15). Main graph, sum of the [3H] inositol phosphates accumulated. Inset, distribution of radioactivity in fractions eluting from the Dowex 1 columns with standard inositol phosphate (ZP), inositol bisphosphate (ZPp), and inositol trisphosphate (ZP3). For further details see Ref. 15. Results such as these were obtained in two additional experiments.
saturating carbachol than in its absence. Since accumulation of inositol phosphates under these conditions is a reflection of the activity of the receptor-sensitive phospholipase C, this indicates that the M5 receptor occupancy is able to stimulate this enzyme 20-fold over basal.
We tested whether in addition to the transfected receptor L cells express receptors for other agonists acting similarly to stimulate formation of inositol phosphates. These agonists were the purinergic receptor ligands ATP and AMP-P(NH)P, a variety of peptides, thrombin, prostaglandin Fza, and biogenie amines, serotonin, and epinephrine, which are all known in other systems to either stimulate phosphoinositide hydrolysis or to promote increases in [Ca*+]i. Significant effects were obtained only with ATP (Table I) and was mimicked by AMP-P(NH)P (not shown). Inositol phosphates accumulated in response to the purinergic ligands 2-2.3-fold with respect to control in the parent Ltk-cell line and a similar 2.3-2.8-fold in the transfected LM5.36 cell line (not shown). Since phosphoinositide hydrolysis can be increased in these cells by as much as 20-fold, this indicates that the resident purinergic receptor(s) affect but a fraction of the intrinsic cellular potential for hydrolyzing phosphoinositides. The potential of M5 and purinergic receptors to affect arachidonic acid release from cells prelabeled with ['%]arachidonic acid was assessed for both LM5 and Ltk-cells. As illustrated in Fig. 2, both types of receptors caused release of free arachidonic acid, and like phosphoinositide hydrolysis, M5 receptors had a much higher efficacy eliciting this response than the purinergic receptors. Incubation of cells for 15 min (Fig. 2) or 30 min (not shown) with protein kinase stimulator PMA did not mimic the effects of receptor stimulation on arachidonic acid release by LM5 (Fig. 2) or by Ltlzcells (not shown). This indicated that stimulation of this enzyme alone cannot be solely responsible for the receptorstimulated arachidonic acid release. In our previous report we reported that the M5 receptor did where PGEl increased intracellular CAMP levels -lo-fold, and tested for an effect of CCh either increasing or decreasing basal or PGE1-stimulated CAMP levels. We found that CCh, and by inference the M5 receptor, was without effect. Stimulation of the purinergic receptor was also without effect (not shown).
Because both M5 and purinergic receptors promote IPS formation, and in other cell systems this leads to release of Ca2+ from endogenous stores (3), we loaded Ltk-and LM5 cells with the fluorescent Ca2+ indicator dye fura-and tested the effect of ATP and CCh. As shown in panels A-D of Fig.  3, we indeed found that intracellular Ca*+ levels were transiently elevated in response stimulation of these receptors, following a time course that is typical for these type of responses (e.g. 27,28,33). As expected from the phosphoinositide hydrolysis and arachidonic acid release studies, only the LM5 cells responded to CCh while ATP triggered a response in both the Ltk-and the LM5 cells. In agreement with the known pharmacological properties of these type of receptors, the effect of CCh but not that of ATP was blocked by the muscarinic blocker atropine (Fig. 6, B and E). However, while the ratio of responsiveness CCh/ATP was about the same for phosphoinositide hydrolysis and arachidonic acid release (CCh stimulating about &fold more than ATP), the peak concentrations of intracellular Ca*+ ion ([Ca"']J in the LM5.36 cell in response to saturating CCh were about 3-fold higher than those obtained with ATP. Fig. 4 presents the results from four experiments in which the accumulation of inositol phosphates, arachidonic acid release, and the maximal peak increase in [Ca'+]i were studied as a function of CCh concentration.
Half-maximal effects for the last two of these effects were obtained at 0.4-0.5 pM CCh. In contrast between 1.5 and 2 pM CCh were required for half-maximal effects on accumulation of inositol phosphates.
The above experiments are consistent with the interpretation that both the L cell purinergic receptor and the neuronal M5 receptors trigger cellular responses by stimulating phospholipase C with consequential formation of IP, and that the IPs thus formed causes release of Ca*+ from intracellular stores, and that in addition these receptors stimulate the liberation of free arachidonic acid. The nature of the lipid from which arachidonic acid is hydrolyzed has not been investigated, but similar studies from other laboratories suggest that it is a phospholipid and the release is due to stimulation of phospholipase A2 activity (47). The relative responses to purinergic and M5 receptors suggest that if the increase in [Ca*+]i is required for triggering the arachidonic acid release response, it is not the sole intracellular second mediator of this effect. Our data are consistent with the proposal of Burch et al. (44) that the arachidonic acid release responses triggered may be mediated through a signaling pathway parallel to that responsible for inositol trisphosphate production.
Complexity of the Ca*+ Response Triggered by the M5 and Purinergic Receptors in L Cells The [Ca"], transients caused by CCh and ATP in Fig. 3 agree with those described for other phospholipase C-stimulating agonists in other cell systems in that they are formed of a transient peak and a sustained phase. The peak response has a rapid rise and a relatively fast decay (tsh -30 s) that is slower than that of the ATP-triggered response (see Fig. 3, A uersus D and E). The sustained response appears to rise more slowly with CCh than ATP (see third and fourth panels from the top in  Several lines of evidence were obtained that indicate that the fast decay in the initial [Ca2+]i is associated with an exhaustion of the pool of Ca2+ from which the initial peak had been derived. The first was that postaddition of saturating (100 pM) ATP after a saturating concentration (100 pM) of CCh (Figs. 3A and 5) had no effect, the second was that stimulation of the less effective purinergic receptor with 100 pM ATP decreased but did not abolish the response to subsequent stimulation of the more effective M5 receptor (Fig.  3C), and the third and perhaps most conclusive was that if after 100 pM CCh excess atropine was added, so as to block further action of CCh through the M5 receptor, postaddition of saturating ATP 4 to 5 min later had the same effect as in na'ive cells (cf. Figs. 3 and 6).
The left panels of Fig. 3 show the effect of varying the concentrations of CCh on the first and second phase responses as well as on the availability of Ca*+ for a subsequent response to saturating ATP. In these particular experiments, carried out 4 months after the dose-response curves shown in Fig. 4 using cells grown to confluence in 150-mm Petri dishes instead of 6-well plates, the concentrations of CCh giving halfmaximal effects were &M, 1, and 0.6 pM for eliciting, respectively 1) the initial peak increase of [Ca*+]i, 2) the peak increase of [Ca2+]i of the second phase response (as seen 1.2 min after CCh with 0.3 and 1.0 PM cCCh), and 3) the "inhibition" of the ATP response. While in this and two repeat experiments we always found a slight (2-2.5-fold) left shift in the dose-response curve for inhibition of a second first-peak response in [Ca2+li, as compared with that with which the first peak [Ca2+]; response was obtained, the dose-response curves for the first and second peak responses to the initial addition of CCh did not differ in a statistically significant manner.
We obtained an idea as to the time required for full reappearance of a first phase response (as seen with ATP) after causing its partial disappearance with an initial dose of CCh when we blocked the sustained response to CCh with excess atropine, and then added saturating ATP at varying times thereafter. Fig. 6 shows the results of one of three such experiments indicating that the depletable pool of intracellu-lar Ca2+ is refilled in 3-4 min under the standard incubation conditions used here. The data are consistent with the first peak response being due to IPB-mediated release of a Ca2+ from a limiting intracellular pool. Panels G and H are from another experiment and demonstrate that the replenishment of this limiting pool does not occur for as long as agonist is present, consistent with continued production and action of IP3.
The experiments of Fig. 6 (see also Fig. 7C) indicate further that the [Ca2+]i response is continuously dependent on agonist occupancy of the receptor. As has been seen in other systems, the peak height of the first peak response is essentially unaltered or diminished only slightly by removal of Ca2+ from the medium as obtained when, for example, 4 mM EGTA is added together with the agonist (not shown). In contrast, the sustained response is abolished. Continued incubation in the presence of high levels of EGTA led to a reduction in [Ca'+]i to levels below control. The dependence of the sustained response on extracellular Ca2+ was established further in the experiments shown in Fig. 7, in which fura-2-loaded cells were resuspended in low, nominally Ca2+-free buffer containing 0.3 mM EGTA. Under these conditions, basal levels of [Ca"]i, while lower than in Ca2+ containing buffers, were nevertheless stable for up to 12-15 min. It can be seen that under these conditions the sustained response is absent, both at high (Fig.  7B) and low (Fig. 7C) agonist, but can be readily restored by addition of extracellular Ca2+. The experiment in Fig. 7C confirms the conclusion drawn on the basis of results shown in Fig. 6 that the sustained phase Ca2+ entry is dependent on continued receptor activation by agonist.

Factors That Do and Do Not Alter the Agonist-induced [Ca"+]i Transients in Murine L Cells
Nature of the Agonist-The right panels of Fig. 5 show the results of experiments akin to that shown in the right panels but in which cells were stimulated with varying concentrations of the P2 purinergic agonist instead of CCh. It is clear that regardless of the concentration of ATP used patterns of sustained [Ca2+li increases such as seen with CCh are not obtained with ATP. These results indicate that the temporal pattern of [Ca2+11 changes may have receptor specific components.
Effect of Cholera Toxin (CTX) Treatment-CTX (3 pg/ml for 24 h) reduced both the peak and the sustained responses to subsaturating as well as saturating concentrations of ATP or CCh. This is shown for the response to 1 pM CCh in Fig.  8B. Although reduced in magnitude, the Ca2+ entry was still agonist-dependent.
The effect of CTX was mimicked by forskolin treatment (Fig. 8C), indicating that it is most likely mediated by CAMP rather than being due to an adenylyl cyclase-independent effect of G.. In contrast, determination of the total number of N-methyl-scopolamine-specific binding sites were either unchanged (n = 1) or increased by 10 f 4% (n = 1) upon treatment with CTX and were unchanged upon treatment with forskolin (n = 2) (not shown). When tested for effects on phosphoinositide hydrolysis, CTX had a minor (23%) but significant stimulatory effect on the basal rate (agonist absent) and did not affect the agonist-stimulated rates of phosphoinositide hydrolysis (not shown), indicating that the CTX-and forskolin-induced reduction in the responses of [Ca2+li to agonists is unrelated to IP3 formation. to -0.05% of control (Fig. 9). Phosphoinositide hydrolysis in such PTX-treated cells was unaltered in the absence of agonist and slightly (26%) enhanced in the presence of CCh (not shown) and specific N-methyl-scopolamine binding was reduced by 12 f 3% (n = 2) (not shown). Under these conditions, we found that while the peak [Ca"']i responses to CCh or ATP were essentially unchanged, the sustained responses to the two agonists were clearly enhanced (Fig. 10). This last effect is best seen in Fig. lOB On the Mechankm of the Sustained Agonist-stimulated Ca*' Entry into L Cell.+-Agonist-stimulated Ca*' entry into L cells was blocked by La3', suggesting that it enters through a Ca'+specific pathway but was unaffected by nitrendipine, a dihydropyridine with Ca*+ channel blocking activity (not shown). This indicated that Ca2+ entry during the sustained response was not secondary to voltage-dependent Ca*+ channel activation and thus different from the mechanism by which Traces obtained with two batches of furaloaded LM5.36 cells are shown. Note that the time for recovery is about the same whether the sustained response is interrupted early @an& A-F) or late (panels G and H). The results are representative of a total of four experiments of this type.
sustained Ca*+ entry is promoted in pituitary GH& cells (30, 31). The finding that Ca2+ entry was unaffected by the dihydropyridine blocker is consistent with our independent observation that L cells do not exhibit voltage-gated Ca*+ currents ary to Na' influx as may be the case in smooth muscle cells (32).
Other-Attempts to dissociate arachidonic acid release from phosphoinositide hydrolysis have failed. These included addition of 1 mM neomycin sulfate (n = 3), shown to inhibit inositol phosphate production in Fisher rat thyroid cell line cells in response to al-adrenergic receptor stimulation (44) and in Madin-Darby canine kidney cells in response to bradykinin (51) and PTX treatment of L cells, which inhibited the agonist-stimulated arachidonic acid release in Fisher rat thyroid cell line cells (44) and macrophages (43). Neither had an effect on the L cells studied in the present report. i.e. neomycin failed to affect CCh-stimulated phosphoinositide hydrolysis and PTX failed to affect basal or CCh-stimulated arachidonic acid release (not shown).
Addition of 20-40 F~M arachidonic acid 2-4 min prior to agonist was without effect on the [Ca"+], transients (not shown). Addition of PMA (1 pM) added 1 or 10 min prior to 1 pM CCh had no effect on basal [Ca'+], or on the CChinduced [Ca'+], transients (not shown).

DISCUSSION
On the Signaling Properties of the M5 Receptor and the Use of L Cells to Study Cellular Signaling-The experiments presented here were carried out to characterize whether in addition to stimulating phosphoinositide turnover (15, 50), the M5 receptor is able to trigger other cellular responses. Of these, some, like IPa-mediated release of Ca*+ from intracellular stores seems to be an obligatory consequence if the cellular biochemistry supports it, while others such as the arachidonic acid release, or elevation of CAMP levels are often but not always associated with receptors capable of stimulating phospholipase C. For a receptor with a cellular distribution as limited as that of the M5 receptor, this may be the only way to learn about its properties. We found the receptor to trigger a complex Ca*+ mobilization response and arachidonic acid release, but it caused no changes in CAMP levels. Unfortunately, we have been unable to discern whether any of the effects other than inositol phosphate accumulation are due to an effect of the M5 receptor independent of phospholipase C stimulation. Thus, while in some studies it has been possible by the use of neomycin to dissociate the arachidonic acid release, i.e. phospholipase A2 stimulation, from phospholipase C stimulation (44, 51), our studies were not informative in this regard. It may be that in L cells the release of arachidonic acid is a consequence of increased Ca*+ levels plus a protein kinase C-mediated phosphorylation event, as opposed to a true receptor G-protein-mediated stimulation of phospholipase A2 activity. Consistent with this interpretation is the finding that arachidonic acid release was stimulated by 60-100% with CCh and not more than 15-18% with ATP which is in close proportion to their relative effect on phosphoinositide hydrolysis. Likewise, PTX, which abolishes the arachidonic acid response in FRTL cells without affecting phospholipase C stimulation by al-adrenergic receptors (44), was without effect on arachidonic acid release in our studies.
We failed to observe effects of CCh on intracellular CAMP levels in the LM5 cells. This indicated that it is unlikely that either the transfected M5 or the resident purinergic receptors are direct regulators of adenylyl cyclase and suggests that the CAMP-elevating role of the M5 receptor in other cells (20) is likely to be indirect and conditioned by the particular biochemistry of that cell. Indeed while this manuscript was under consideration for publication, Felder et al. (52) reported that accumulation of CAMP in A9 cells in response to Ml receptor stimulation is indirect, depending on phosphatidylinositol hydrolysis and occurring possibly via an increase in cytosolic Ca'+ as a result of IP, formation, followed by an action of Ca'+-calmodulin on calmodulin-dependent adenylyl cyclase.
Although we have no evidence that M5 receptors may alter CAMP levels or have an effect on arachidonic acid formation in L cells, this is not to say that in its natural neuronal environment it may not have these effects. Both effects are dependent not only on the nature of the receptor but also on the biochemistry of the effector cell. Thus, our descriptions of M5 receptor properties is limited by the nature of the cell in which we have chosen to express it. In view of the paucity of L cell receptors of the type that would be expected to promote phospholipase C activation and Ca*+ mobilization (Table I) and the robust response in these two parameters obtained with the transfected M5 receptor, we believe that the non-excitable murine L cell may be a good model for the characterization of receptors that may have this type of effects but which cannot as yet be studied in their natural environment. Overall the M5 receptor resembles in its cellular signaling properties the Ml and M3 receptors, which also trigger intracellular Ca*+ transients (53) and promote arachidonic acid release (52). Our studies indicate further, the L cell may also be good to study [Ca'+]i regulations, such as may occur in other non-excitable cells, none of which exists as an established cell line.
On the Interpretation of the Changes in [Ca"], Observed upon Stimulation of the L Cell by Agonists-The pattern of change in [Ca"'], observed in L cells upon addition of agonists could be described as being the result of two distinct effects of receptors, one to release via IP3 formation Ca*+ from a depletable intracellular compartment and the second to cause entry of Ca2+ from the extracellular space possibly through activation of receptor (G-protein?) or IPs/IP, operated Ca*+ channels located on the plasma membrane of the cell. However, as summarized in Fig. 11, an analysis of the literature suggests that the two phases of the Ca*+ mobilization response may have a single underlying mechanism consisting at all times of an IP,-mediated release of Ca*+ from an internal store. This is based on both the oscillatory responses to agonists as seen in single cells and the biochemical resolution of distinct types of Ca*+ accumulating and releasing vesicles.
It has been shown that application of low concentrations of agonists to cells results in sustained trains of periodic increases of [Ca'+]i that are independent of changes in voltage-gated ion channels (54-57). These periodic increases, referred to as oscillations (58) or spikes (54), are of relatively constant amplitudes and may vary in frequency between 0.251 min to 4/min, depending on agonist concentration (54). At high agonist concentrations these spikes become too frequent and coalesce into what probably corresponds to the transient or peak response observed in our studies. As predicted from studies of cell populations such as reported here, the [Ca'+]i oscillations become dependent on extracellular Ca*+ in cells from which the intracellular IPs-releasable pool has been depleted (26,55). Thus a sustained response at the cell population level, having as its basis in nonsynchronized cells discrete periodic [Ca2+li oscillations, could simply be the result of a pulsatile activation of an entry mechanism through the plasma membrane from the extracellular space into the cytoplasm (e.g. via receptor-operated (59) or IPJIP.,-operated (60, 61) Ca2+ channels). Arguments have been presented that the intrinsic oscillator lies in the microsomal site of action of IP, (56,(60)(61)(62)(63)(64)(65).
Refilling of the IPB-sensitive Ca2+ store has been proposed to occur by a path that bypasses the cytosolic [Ca*+]; com-partment (26), triggered by the lowering of its Ca2+ content. Biochemically, the refilling process that bypasses the cytosolic Ca*+ compartment appears to involve two non-mitochondrial Ca*+ compartments of which one is IP3 sensitive and the other is 67). These have been proposed by  to constitute a vesicular Ca*+ transport system, akin to the vesicular translocation system operating between Golgi stacks (68), in which the IP3-sensitive compartment receives Ca*+ both from the plasma membrane and from the cytosol and transfers its contents into the IPBsensitive compartment (67). These two subcompartments have recently been localized to separable membrane vesicles (67).
Based on the findings with single cells, it is thus possible to propose that the sustained response, dependent on extracellular Ca*+ for its continuity, is due to accumulation of threshold levels of IPB (56, 63, 64), release of Ca*+ from the IPB-sensitive store (3), Ca*+-mediated inhibition of continued action of IPS (65), and refilling of the Ca*+ store triggered by lowering of its Ca2+ content (69), lowering of cytosolic Ca2+ by ATP-dependent extrusion from the cell (21,22), ATPdependent re-uptake into both the transfer pool and the IPBreleasable pool (66,67,70), and reinitiation of the cycle (Fig.  11). The lowering of [Ca*+], through ATP-dependent extrusion, could be responsible for a time-and agonist-dependent shift in the source of Ca2+ for re-refilling of the IPs-sensitive pool from cytosolic to extracellular and for a gradual increase in the role for the transfer pool, thought to be submembranous (66,67,70), in maintaining of continuously oscillating response.
Although many details of this cycle are missing and subject of intense work in several laboratories, many of our results are in good agreement with this scheme. It predicts the existence of a delimited and exhaustible pool of IPB-releasable Ca*+ (Fig. 3, 5, and 6) and requires a continued role for IPS throughout the agonist response such as seen here (Fig. 6). Our finding that increasing CAMP levels in cells, a condition that affects the affinity of IPS for its receptor and increases its EC& for Ca*+ release from microsomal stores (48), results in a decrease in the agonist response for both the peak and the sustained phases of the Ca*+ mobilization response (Fig.  8) supports a role for IP3 throughout the sustained phase of the agonist response.
However, some of the patterns of the sustained responses obtained in the present study are difficult to interpret in terms of amounts of IP3 formed acting on a single pool of stored [Ca*+]i and lead us to the argument that oscillations seen at very low agonist concentrations may in fact be due to a combination of the above mechanisms plus a pulsatile entry of Ca2+ through the plasma membrane into the cytosol by a mechanism that does not involve IP,.
First, we observed that the pattern of the Ca** mobilization response varied with the agonist (compare the shape of the [Ca*+], responses to CCh to those obtained with ATP in Fig.  5). This variation could not be compensated for by varying the concentration of the agonist and or be correlated with the degree of depletion of the IPB-sensitive pool of Ca*+. A similar result was reported by Cobbold and collaborators (71), who found with aqueorin-loaded hepatocytes that the individual Ca*+ spikes had shapes (e.g. narrow and smooth versus wider and jagged, varying rise and relaxation times) that differed depending on whether they were elicited by vasopressin, phenylephrine, or angiotensin II. Our results therefore indicate that the Ca*+ response depends not only on the absolute level of cytosolic IP, but also the rate and/or cellular site of formation of the second messenger. An explanation for this The pathway by which the M5 and the P2 receptors are thought to activate phospholipase C with resultant formation of DAG and IP3 are denoted along the upper surface of the cell as is the metabolic conversion of IP3 to IPI. Three forms of Ca*+ entry through non-voltage-dependent Ca" channels, activated either directly by the receptor or indirectly through a G-protein or products of phospholipase C activation (60,61), which would all behave according to the macroscopic properties of the sustained response and fall under the classification of receptor-operated Ca*+ channels or ROCCs (59), are depicted along the left side of the cell. Two possible modes of Ca2+ entry, stimulation of Na+ entry coupled to Na+/Ca'+ antiport activity, and direct or indirect stimulation of voltage-gated Ca2+ channels or VOCCs (58), which were ruled by the experiments reported here, are depicted along the right side of the cell. The IP3-sensitive intracellular Ca2+ pool is represented in the center of the cell delimited by a membrane containing the IP, receptor/ Ca*+ release channel, a Ca2+ pump responsible at least partially for the refilling of the pool, and a GTP-dependent transfer mechanism (63, 70) also responsible for the refilling of the IPa-sensitive Ca2+ pool by transfer from a second vesicular, IPg-insensitive pool of Ca'+ derived which can derive its Ca*+ both from extracellular Ca*' space (lower edge of the ceH) and from the cytosolic space (Ca'+ pump). The figure also depicts locations of Ca'+ pumps responsible for the lowering of [Ca'+], by extrusion from the cell and uptake into both the IPa-insensitive transfer and the IP+ensitive storage pools but does not detail mechanisms by which the muscarinic and purinergic receptors cause arachidonic acid release. Abbreuiutions: AC%, acetylcholine; M5 AChR, muscarinic acetylcholine receptor subtype M5; P,R, purinergic receptor subtype Pz; Gp, G-protein responsible for mediating the activation of phospholipase C by the muscarinic and purinergic receptors; (Y*, activated (Y subunit of G, protein; PIP,, phosphatidyl inositol 4,5-bisphosphate; PhL C, phospholipase C; DAG, diacylglycerol; Ca2+e, Ca2+a, Ca2+S, extracellular, intracellular, and stored Ca'+; Nafe, Na',, extracellular, and intracellular Na+; DHP, dihydropyridine.
For further details see text.
may be that different receptors use a different complement of G-proteins to stimulate IP, formation. Precedents for this were provided by Ewald et al. (72), who have shown in sensory rat dorsal root ganglion cells that bradykinin and type-B yaminobutyric acid receptors differ in their ability to use exogenously added Gi and G, to cause inhibition of Ca*+ currents, a common effector system not unlike IPs-mediated Ca2+ release. Likewise, Ashkenazi et al. (49) showed in Chinese hamster ovary cells that to stimulate phosphoinositide hydrolysis different receptors use a different complement of G-proteins (subclassified in terms of pertussis toxin sensitivity).
Second, we found that the shape of the Ca" response, the underlying nature of which we are assuming to be a sum of individual Ca*+ spikes, is altered upon treatment of cells with PTX (Fig. 10). This is also in agreement with the possibility that different G-proteins may be involved in the activation of phospholipase C by one agonist as compared with another. ADP-ribosylation of all PTX-sensitive G-protein molecules is likely to alter the activation/deactivation kinetics of other G-proteins that share the same pool of /3r dimers (for review see Ref. 73), even if they are themselves insensitive to PTX. For example we have noted that PTX treatment of cells tends to potentiate stimulatory responses of adenylyl cyclase. Alternatively, rather than being a reflection of temporal or kinetic aspects of IPS formation, the agonist-specific aspects of the Ca2+ response could also be due to existence of a Ca*+ entry path that is regulated by one or more G-proteins or by a site of IPB and/or IP, action distinct from that causing the release of Ca*+ from the internal store.
Third, we noticed that readmission of Ca*' to cells in which we had depleted the IPB-sensitive Ca*+ compartment in the absence of extracellular Ca2+ (Fig. 7) results within seconds in restoration of the average intracellular Ca2+ levels, interpreted as resumption of periodic oscillations of [Ca2+11, equivalent to what would have been seen if the agonist response would have been allowed to run its course in the presence of extracellular Ca*+. This would indicate that the refilling process reaches steady state very rapidly and that the turnover of agonist-sensitive intracellular Ca2+ is very fast. In contrast, in Fig. 6, in which the action of CCh is interrupted shortly after the peak response, and ATP is added at varying times thereafter to measure the refilling of depletable IPa-sensitive pool of Ca'+, we found that this pool is refilled only gradually. This would indicate that the turnover of agonist-sensitive intracellular Ca*+ is very slow. Delaying the interruption of the sustained response did not alter this result (Fig. 6, A -F versus G and H). This raises the question as to whether indeed there is only one Ca2+ store responsible for both the sustained elevation of [Ca2+]i, which presumably represents the sum of non-synchronous low frequency [Ca2+]i oscillations, and the peak increases in [Ca2+];, which presumably represents increases in [Ca2+]i under conditions where individual [Ca'+]i oscillations have coalesced, are one and the same.
Taken together our results indicate that the Ca" response of L cells is both under continuous dependence of IP3 and under a continuous control either of the temporal pattern of IPS formation or of how Ca2+ entry is regulated. Although the overall sources of Ca*+ (intraversus extracellular) are known, the immediate origin of the free intracellular Ca2+ ([Ca'Q is not clear and deserving of further investigation.