Evidence for Bile Acid-evoked Oscillations of Ca2+-dependent K+ Permeability Unrelated to a D-myo-Inositol 1,4,5-Triiphosphate Effect in Isolated Guinea Pig Liver Cells*

In single liver cells, the D-myo-inositol 1,4,5-tri-phosphate (InsP3)-dependent agonists such as nor- adrenaline and angiotensin I1 evoke oscillations in intracellular calcium [Ca2+Ii resulting mostly from the periodic release and reuptake of calcium from intra- cellular stores. In the present work, we have reexam-ined the effects of these agonists and investigated whether the natural bile acid taurolithocholic acid 3-sulfate (TLC-S), which permeabilizes the endoplasmic reticulum, could initiate oscillations of [Ca2+Ii. Oscillations of [Ca2+Ii were monitored with the Ca2+-de- pendent K+ permeability in whole-cell voltage-clamped guinea pig liver cells. Our results confirm the presence of two types of oscillations induced by hormones. They could be distinguished by their frequency periods. The fast (type I) had periods ranging from 5 to 12 s and the slow (type 11) from 60 to 240 s. They have been re-spectively attributed to second messenger- and recep- tor-controlled oscillations, respectively. Our results also show that TLC-S, as noradrenaline and angioten- sin 11, induced the activation of this Ca2+-dependent K+ current and was able to reproduce

In single liver cells, the D-myo-inositol 1,4,5-triphosphate (InsP3)-dependent agonists such as noradrenaline and angiotensin I1 evoke oscillations in intracellular calcium [Ca2+Ii resulting mostly from the periodic release and reuptake of calcium from intracellular stores. In the present work, we have reexamined the effects of these agonists and investigated whether the natural bile acid taurolithocholic acid 3sulfate (TLC-S), which permeabilizes the endoplasmic reticulum, could initiate oscillations of [Ca2+Ii. Oscillations of [Ca2+Ii were monitored with the Ca2+-dependent K+ permeability in whole-cell voltage-clamped guinea pig liver cells. Our results confirm the presence of two types of oscillations induced by hormones. They could be distinguished by their frequency periods. The fast (type I) had periods ranging from 5 to 12 s and the slow (type 11) from 60 to 240 s. They have been respectively attributed to second messenger-and receptor-controlled oscillations, respectively. Our results also show that TLC-S, as noradrenaline and angiotensin 11, induced the activation of this Ca2+-dependent K+ current and was able to reproduce both types of oscillations. The bile acid effect was not blocked by intracellular perfusion of heparin known to inhibit both InsP3 binding and InsP3-evoked Ca2+ release in several tissues. In these conditions, TLC-S only evoked type I oscillations, suggesting that these fluctuations could originate from a mechanism that is independent of InsP, and is an intrinsic property of internal Ca2+ stores.
In many nonexcitable tissues elevation of cytosolic free calcium ([Ca"],) is a central event in the mechanisms linking the receptor binding of transmitter substances to consequent changes in cell function. A biphasic elevation of [Ca2+]; has been shown in a wide range of cell types and involves a primary release of Ca2+ from intracellular stores by InsP31 and a sustained plateau phase that has been ascribed to an influx of extracellular Ca2+ (Williamson and Monck, 1989). When measured in a single cells loaded with Ca2+ indicators or monitored from indirect measurements of Ca2+-dependent K+ or C1-currents, the hormone-induced rise in [Ca2+]; may display repetitive oscillations. This has been observed originally (Green et al., 1972) and carefully analyzed (Woods et al., 1986(Woods et al., , 1987 in isolated hepatocytes and then found in a wide variety of other nonexcitable cells (for review, see . The observation of periodic releases of Ca2+ from hormone-stimulated perfused liver (Graf et al., 1987), the fact that the oscillation frequency may be controlled by the concentration of the agonist (Woods et al., 1986(Woods et al., , 1987Rooney et al., 1989), and the prevalence of oscillations in a wide variety of cells have led to speculations Jacob, 1990) that repetitive [Ca2+Ii signals may be a general mode of cell frequency coding of information in hormonal responses. These oscillations can be divided in two types (Berridge and Galione, 1988). Type I consists of oscillations of the Caz+ signal around a mean value during the plateau phase of the response with a rather high frequency (5-12 min"). Type I1 oscillations are made up of responses of maximal amplitude which occur in cycles of low frequency (0.25-1 min") separated by periods of no activity.
In the present work, we investigated the possibility that other natural molecules able to release Ca2+ from the endoplasmic reticulum (ER) in suspensions of liver cells could generate oscillations of [Ca2+], in single hepatocytes. We used the sulfated derivative of the monohydroxylated bile acid taurolithocholate (TLC-S), which is known for its remarkable ability in mobilizing Caz+ from the ER (Combettes et al., 1988(Combettes et al., , 1989. [Ca2+Ii was monitored with the Ca2+-dependent K+ membrane conductance, a method that provides a good measure of the agonist-mediated rises in [Ca2+Ii (Marty et al., 1984;Capiod et al., 1987;Gray, 1989;Ogden, 1989a, 1989b). In guinea pig hepatocytes, the apamin-sensitive K+ channels are a fast and sensitive tool to detect rapid changes in [Ca2+]; near the plasma membrane (Ogden et al., 1990). Our results show that TLC-S is able to promote fast (type I) or slow (type 11) oscillations of [Ca2+]; similar to those seen with the InsP3-dependent hormones noradrenaline and angiotensin 11. This indicates that these oscillations originate from a mechanism that is an intrinsic property of internal Ca2+ pools and which can be triggered in the absence of InsP, action.

EXPERIMENTAL PROCEDURES
Guinea pig (males of the Hartley strain) liver cells were isolated by collagenase digestion and mechanical dispersion (Field and Jenkinson, 1987). Thirty minutes after preparation, hepatocytes were plated in 35-mm-diameter Falcon plastic Petri dishes at a density of approximately 400,000 cells in 2 ml of Earle's minimum essential medium (GIBCO) supplemented with newborn calf serum (lo%), penicillin (200,000 units/ml), and streptomycin (100 mg/ml). The dishes were incubated at 37 "C in an atmosphere of COS in O2 (19/1) for at least 1 h and were then used within the next 8 h.
Standard tight-seal, whole-cell recording techniques were used Bile Acid-evoked [Ca"Ii Oscillations 269 (Hamill et al., 1981). Patch pipette resistances were typically 5 megohms. Whole-cell currents and potentials were measured with an RK300 patch-clamp amplifier (Biologic). Signals from whole-cell recordings were digitized by a CED 1401 interface (CED Limited). Rise times were calculated by computer using the VCAN package supplied by J. Dempster (University of Strathclyde, United Kingdom).
Movements of Caz+ in Saponin-treated Guinea Pig Hepatocytes-Hepatocytes were permeabilized by saponin as reported by Champeil et al. (1989). Shortly, a volume of 30-50 ml of cells was centrifuged and washed twice (50 X g for 1 min) with the above mentioned Earle's medium lacking Ca2+ and then resuspended at 37 "C for 5 min in a cytosolic-like medium containing (in mM): KC1,lOO; NaC1,20; MgClZ, 5; NaH2P04, 0.96; Hepes, 25 (pH 7.25 at 20 "C); Na2ATP, 1.5; creatine phosphate, 5; creatine kinase, 5 gg/ml; and 20 g~ fluorescent Ca2+ buffer quin2. Saponin was added at a final concentration of 50 gg/ ml; this allowed permeabilization of the plasma membrane but not of internal Ca2+ stores, which thus retained their ability for active accumulation of Ca2+. After this 5-min period, the permeabilized cells were transferred on ice and used within a few h. More than 97% of the cells were freely permeable to solutes as estimated by exclusion of trypan blue. Before experiments, samples of cell suspension (2 ml) were transferred to the spectrofluorometer cuvette and rewarmed to 37 "C for 5 min (which allowed them to accumulate Ca2+). Calibration of quin2 was performed as in Binet et al. (1985).
Chemicals-Collagenase was obtained from Worthington or Boehringer Mannheim; heparin, from Prolabo; and all other reagents from Sigma or Boehringer Mannheim.

Properties of the Ca2+-dependent K+ Current in Guinea Pig
Liver Cells-It has been shown previously that macroscopic K+ current (whole-cell) and the underlying unitary channels (isolated membrane patch) are steeply dependent on cytosolic Ca2+ in the range of 0.3-1.2 p~ in guinea pig liver cells. The channels require the binding of two to three ea2+ ions for activation. They do not show desensitization in the continued presence of e a 2 + or regenerative behavior at the level of the activation by ea2+. These K+ channels also appear to be insensitive to membrane potential (Capiod and Ogden, 1989a, 198913). Photolytic release of ea2+ from the calcium-Nitr5 complex (~l ms) has shown that the response t o a step of Ca2+ has a half-time of 250 ms at 0.5 p~ final free [Ca2+Ii (Ogden et al., 1990). These properties make these K+ channels a very valuable indicator of the variations of the [Ca2+Ii occurring during a hormonal stimulation. In the present experiments, the K+ current was optimized by holding the membrane potential at 0 mV and by eliminating the C1conductance (using gluconate ions in both external and internal solutions). Under these conditions, the current was recorded under a large driving force (EM -EK = -85 mV) and was essentially carried by K+ ions Ogden, 1989a, 1989b).

Patterns of Responses Elicited by
Agonists-The cell responses t o noradrenaline or TLC-S displayed patterns that differed markedly from cell to cell (Fig. 1). This is consistent with results reported previously in hepatocytes using fluorescent Ca2+ indicators (Monck et al., 1988;Rooney et al., 1989;Kawanishi et al., 1989) or in other cells, in which they have been referred to as "fingerprints" (Prentki et al., 1988;Berridge and Galione, 1988;Jacob, 1990). These different patterns were not dependent on the agonist concentrations used in these experiments, and the mode of dose-frequency dependence of agonist response initially described by Woods et al. (1986Woods et al. ( , 1987 in aequorin-loaded hepatocytes was not examined in this study. Therefore, maximal concentrations of noradrenaline (1-10 PM) and TLC-S (200-300 p M ) were used throughout. A majority of guinea pig hepatocytes (80% of 70 cells tested) displayed partial (type I, Fig. 1, A and B ) or sustained oscillations (type 11, Fig. 1, C and D ) in response to either TLC-S or noradrenaline. 90% of the oscillating cells (72% of the total) were of type I. Two striking effects of the bile acid-evoked response must be raised. First, an external application of TLC-S can apparently reproduce any effect on the ea2+-dependent K' permeability activated by noradrenaline; and second, when applied sequentially with noradrenaline (before or after) on the same cell, no evident difference was observed between the two traces. The amplitude of the responses to bile acids or Ca2+-mobilizing hormones was maximal, indicating that [Ca2+Ii was rising over 1 pM with all the effectors. Furthermore, the rate of rise of the Ca2+-dependent K' permeability, which represents the rate at which [Ca2+Ii was rising near the K+ channels in the plasma membrane (Ogden et al., 1990), was the same with TLC-S and noradrenaline (Table I). These results support the view that a common pathway was used by both effectors in order to obtain comparable responses in liver cells. The fast oscillations were sometimes recorded for periods longer than 6 min with a fast recovery when the effector was withdrawn (Fig. 2 A ) . The type I1 oscillations were also observed for long periods either with the bile acid of the Ca2+-mobilizing hormones (Fig. 2   trations of TLC-S on the same cell. The cell appeared to be nonresponsive to 50 pM TLC-S whereas 100 and 200 pM elicited full responses. A delay preceding the onset of the Ca2+-dependent K+ current was present. For the same concentration of TLC-S (200 pM), it ranged from 10 to 60 s, revealing the marked asynchrony of hepatocyte responses due to cell-to-cell variations as observed in hepatocytes treated with InsP3-dependent hormones (Field and Jenkinson, 1987;Woods et al., 1987;Monck et al., 1988). However, in the case of bile acids, the variable delay of single cells should reflect the period of time required to accumulate the threshold concentration of TLC-S able to trigger the process of Ca2+ release from the ER (Combettes et al., 1988). This is consistent with the fact that the delay was dependent on the concentration of TLC-S used (for example, Fig. 3). It also must be noted that in our experimental conditions, this delay was augmented further by the temperature used (28-32 "C). Another striking characteristic of the TLC-S-evoked response was that the current declined quickly back to the resting level after the bile acid was washed off the external medium. This suggests that the permeabilization of an internal store to Ca2+ by the bile acid was not permanent and was totally reversible.
TLC-S-evoked Oscillations and Cell Membrane Potential-It has been reported recently in single rat hepatocytes loaded with the fluorescent indicator fura2 that depolarization of the cell plasma membrane by isotonic substitution of external Na+ by K' slows the frequency of oscillations induced by the InsP3-dependent agonist phenylephrine with little effect on their amplitude (Kawanishi et al., 1989). The putative role of potential on oscillations induced by the bile acid was examined by determining the Ca2+-dependent K' current at different membrane potentials. Fig. 4 shows that the TLC-S-evoked oscillations of the plateau phase of the response were not affected over a range of potential from -60 to +60 mV. These results seem to indicate that both the uptake of TLC-S and its resulting permeabilizing action of the ER as well as the mechanism that regulates the oscillatory behavior of the response are not controlled by the membrane polarization of the guinea pig hepatocyte. It must be noted also that this range of potential includes the resting membrane potential (from -25 to -50 mV) which has been generally reported in guinea pig hepatocytes (Haylett and Jenkinson, 1972;Egashira, 1980;Field and Jenkinson, 1987).

Effects of TLC-S on I m p 3 Fraction in Guinea Pig Liver
Cells-In saponin-treated rat liver cells, it has been shown that the monohydroxylated bile acid, taurolithocholic acid releases Ca2+ from an internal store even in the presence of the InsPa chelator neomycin or of the receptor binding competitor heparin, which abolish InsP3-mediated Ca2+ release (Combettes et al., 1988(Combettes et al., , 1989. However, in our experiments on guinea pig hepatocytes, the similarities between the responses to TLC-S and noradrenaline could suggest a common pathway, possibly involving InsP3. To test that hypothesis, we determined the effects of the sulfated derivative of this bile acid on the [3H]InsP3 fraction (isomers and InsP4) production in guinea pig liver cells. In contrast to rat liver cells, which display taurolithocholic acid-mediated rises in [Ca2+], independent of InsP3, we were surprised to observe that maximal concentrations of TLC-S (200-300 pM) elicited a significant increase in the cell [3H]InsP3 fraction. The in-

271
crease was not significantly different from those measured with 10 p~ noradrenaline and 100 nM angiotensin 11 (Fig. 5 ) .
The fact that the calcium ionophore ionomycin (10 pM) could reproduce these effects led us to propose that the action of TLC-S on InsP3 metabolism was probably mediated by an indirect activation of phospholipase C by Ca2+. Although phospholipase C of isolated plasma membrane or purified from rat liver displays a certain Ca2+ sensitivity (Nakanishi et al., 1985;Taylor and Exton, 1987), in situ phospholipase C is poorly activated by the appropriate range of cytosolic Ca2+ in that species (Thomas et al., 1984;Renard et al., 1987). This may explain why the calcium ionophore (1-20 p M ) as well as taurolithocholic acid (100 p~) do not increase InsPs metabolism in rat liver cells (Creba et al., 1983;Combettes et al., 1988). Therefore, our results indicate that phospholipase C of intact guinea pig liver cells possesses the Ca2+-stimulated-InsP3 synthesis loop that is apparently lacking in rat hepatocytes.

Dissociation of InSP3 and TLC-S Actions by Intracellular Perfused
Heparin-In order to demonstrate that the InsP3 concentration increase is a consequence and not a cause of the bile acid-evoked mobilization of Ca", we used intracellularly applied heparin to confirm the noninvolvement of InsPs in the TLC-S-evoked activation of the Ca2+-dependent K+ permeability. In preliminary experiments, guinea pig hepatocytes were permeabilized by saponin in an internal medium containing contaminant Ca2+ (2-3 p~) , ATP and the regenerative ATP system, and 5 p~ carbonyl cyanide p-chlorophenylhydrazone to eliminate the contribution of mitochondria to Ca2+ uptake. As illustrated in Fig. 6, after the cells were allowed to accumulate Ca2+, the addition of heparin (200 pg/ml) did not block the TLC-S-induced release of Ca2+ from internal stores although the InsP3 effect was dramatically abolished. Therefore, in the next experiment, heparin was applied into the cell via the patch pipette. However, the slow rate of access of active agents by diffusion from the patch pipette (Pusch and Neher, 1988) constrained us to use concentrations higher than those that inhibit InsP3 binding (Worley et al., 1987) and Ca2+ release (Cullen et al., 1988;Combettes et al., 1989) in permeabilized liver cells. The results are shown in Fig. 7. The time indicated above each trace refers to the whole-cell formation when heparin started to diffuse into the cell. This cell was sequentially superfused with maximal concentrations of angiotensin 11 (100 nM) or  TLC-S (300 pM). A first addition of angiotensin I1 4 min after the beginning of the intracellular perfusion elicited the usual Ca2+-dependent K+ current (Fig. 7, far left). After 10 min (Fig.  7, right center), the response to a new application of this agonist was blocked, indicating that heparin had reached the concentration able to inhibit InsP3 binding to its receptors efficiently. As it may be expected from the experiments reported in Fig. 6, the response to TLC-S remained unaffected by heparin. Despite the loss of hormonal effect, the cell was effectively responding to TLC-S 7 min as well as 13 min after perfusion of the antagonist (Fig. 7, left center and far right). This effect could not be ascribed to a greater InsP3 production by TLC-S than by angiotensin 11, which could have overcome the capacity of heparin to block InsP3 receptors (Fig. 5 ) .
Similar results were obtained for cells perfused intracellularly with higher concentrations of heparin (2 mg/ml) in order to reduce the time required to block the effect of InsP3-dependent agonists. In another experiment, noradrenaline (10 PM) was also tested to show that the lack of response to a second application of angiotensin I1 was not due to a desensitization of angiotensin 11 receptors or to an indirect effect of heparin

Bile Acid-evoked [Ca"Ii Oscillations
on the coupling of these receptors to the InsP3 metabolism.
As for angiotensin 11, the response to noradrenaline was totally blocked by heparin (not shown). It must be noted that in this cell, TLC-S was not only able to activate the Ca2+dependent K+ permeability but also was able to elicit an oscillatory plateau (Fig. 8) similar to those observed in the absence of intracellular heparin (Fig. 2). This showed that type I oscillations of [Ca2+Ii can be obtained in the absence of InsP3 binding to its receptor although InsP, probably interferes in the frequency of these oscillations in the absence of heparin. Unfortunately, none of the six heparin-loaded cells displayed the type I1 oscillations in response to TLC-S. We therefore cannot make conclusions on the role of InsP3 in this type of oscillation. Lack of Noradrenaline-induced Ca2+ Influx in Heparintreated Cells-It has been shown already in rat hepatocytes (Berthon et al., 1984) that the rise in [Ca2+Ii in response to InsP3-dependent agonists such as noradrenaline and angiotensin I1 is biphasic. This was also observed with our results on the Ca2+-dependent K+ permeability in guinea pig liver cells. The first phase of rapid elevation has been attributed to a release of Ca2+ from the InsP3-sensitive intracellular store (Burgess et al., 1984;Joseph et al., 1984) and the second phase, which immediately follows, to an influx of calcium from the external medium (Mauger et al., 1985). The calcium influx has been ascribed to three different putative pathways linked to InsP3-activated calcium channels in the plasma membrane (Kuno and Gardner, 1987;Penner et al., 1989;, InsP, (Irvine and Moor, 1986), or the capacitive model that might involve both of them (Putney, 1986). Surprisingly, in presence of heparin, we observed that both phases of the response to the Ca2+-mobilizing hormones were totally inhibited (Fig. 7, right center). This showed that both the release of Ca2+ from the internal store and the influx of Ca2+ from the external medium are sensitive to an intracellular perfusion of heparin. It has been reported that heparin inhibits the InsPa kinase at a lower dose than that used to inhibit the InsP, binding to its receptors (Guillemette et al., 1989). Although there is no evidence of an effect of InsP, alone or together with InsPa in these cells (Ogden et al., 1990), these results could suggest that inositol polyphosphates might be involved in the sustained phase of the response to noradrenaline and angiotensin 11. On the other hand, it has been proposed that the Ca2+ influx is a consequence of the discharge of Ca2+ from intracellular stores (Putney, 1986). In this case, intracellular heparin would block the release of Ca2+ from these stores and consequently Ca2+ influx. Our results do not allow us to discriminate between these two hypotheses be-

FIG. 8. Recording of type I oscillations induced by 200 pM
TLC-S in the presence of intracellular heparin. The pipette concentration of heparin was 2 mg/ml. The response was recorded 8 min after the membrane rupture. 10 nM angiotensin 11 and 5 p M noradrenaline tested on the same cell did not elicit any activation of the Ca2+-dependent K' current. Holding potential, 0 mV. The trace is representative of four out of six heparin-loaded cells tried. cause of the recurrent production of Imp3 (and probably InsP,) due to the Ca2+ activation of phospholipase C in guinea pig liver cells.

DISCUSSION
Our experiments show that there are at least three different patterns of activation of Ca2+-dependent K+ channels to either bile acids or Ca2+-mobilizing hormones in single guinea pig hepatocytes. We have observed that most of the cells (80%) displayed an oscillatory behavior when challenged with one of these effectors. The fast oscillations (type I, frequency of about 8 min") represent the major type of responses observed in guinea pig hepatocytes for the maximal concentrations of effectors used. But, although the form and frequency of oscillations vary from cell to cell, they are remarkably constant for individual cells and, therefore, have been referred as fingerprints (Prentki et al., 1988). This characteristic probably reflects the intrinsic mechanism of regulation of cellular [Ca2+Ii in a single cell. It is still unclear whether these three individual patterns of responses represent three different mechanisms or whether they just depend on intracellular variations of the cell content in receptors, enzymes, and the size of the Ca2+ stores (Rooney et al., 1989). However, the results reported in this study are the first demonstration that [Ca2+Ii oscillations usually observed in single cells treated with InsP3-dependent hormones can be generated in the absence of InsP3 action in nonexcitable cells.
The mechanism by which TLC-S can elicit the same responses as Ca2+-mobilizing hormones is not clear. The effects of monohydroxylated bile acids on liver cells have already been studied extensively. External application of taurolithocholic acid induces a large rise in [Ca2+Ii. In saponin-treated cells, its effects on Ca2+ release are not abolished by the messenger chelator, neomycin, and the receptor competitor, heparin (Combettes et al., 1988(Combettes et al., , 1989. However, the mechanism of action of TLC-S on the permeabilization of an internal Ca2+ store has not yet been elucidated. But it is known that the bile acid does not affect the basal permeability to Ca2+ of natural membranes other than that of ER, i.e. those of mitochondria and plasma membrane, and does not alter that of the artificial membrane of liposomes (Combettes et al., 1988(Combettes et al., ,1989. The bile acid-induced oscillations of the Ca2+dependent K+ permeability were also observed in the absence of external Ca2+ (data not shown) confirming that the rise in [CaZ+li results from a primary mobilization of internal Ca". An important observation in these experiments is that after an external application of TLC-S, the rate of rise of the Ca2+dependent K+ permeability, which is a better indicator of the rate of Ca2+ release from the ER than the amplitude of the response (Ogden et al., 1990), is as fast as those measured with InsP3-dependent hormones. This strongly suggests that although InsPs is not involved in the bile acid-induced response, TLC-S and noradrenaline are acting through a common mechanism, which accounts for the rapid early phase of the response. But, as the kinetics of InsP3-induced Ca2+ release from the ER (Champeil et al., 1989;Meyer et al., 1990) and activation of the Ca2+-dependent K+ channels by photolytic release of large concentrations of InsP3 (Ogden et al., 1990) are much faster (90-200 ms) than that measured with noradrenalineor angiotensin 11-evoked activation of the K+ permeability in liver cells (see Table I), we cannot conclude that TLC-S acts through the same Ca2+ channels that are opened by InsP,. However, we can propose several mechanisms of action of TLC-S on the activation of the Ca2+dependent K+ channels. TLC-S, as InsP,, could provide a constant release of Ca2+ which can act as a primer to drive a Bile Acid-evoked [Ca"]i Oscillations 273 process of Ca2+-induced Ca2+ release (see Rooney et al., 1989;Marty and Tan, 1989;Berridge, 1990). TLC-S would operate on a Ca2+ store, which also can be InsPs sensitive, to promote the same recurrent mechanism. Alternatively, TLC-S could release Ca2+ from the same store as InsP3 through the same channels, and oscillations would then be explained by a Ca2+ regulation of these Ca2+ channels located in the ER membrane (Supattapone et al., 1988;Joseph et al., 1989). Anyway, this indicates that this permeabilization process to Ca2+ tightly controls the resulting oscillatory response. Our results do not intend to explain from where and how the calcium signal is starting and then spreading throughout the cell generating oscillating activation of the Ca2+-dependent K+ channels in the plasma membrane. However, our study shows that this signal can originate and occur in an InsP3-independent manner. It has been already shown that type I oscillations could take place with a constant concentration of InsPs within the cell (Capiod et al., 1987;Wakui et al., 1989). We now suggest that the whole mechanism of these oscillations can be independent of the presence of InsP3 binding to its receptor. Consistent with this view, InsP3-independent fluctuations of [Ca2+Ii have been reported in pancreatic acinar cells (Osipchuk et al., 1990) and chromaffin cells (Malgoroli et al., 1990). The type I1 oscillations occur with slower frequency (about 1 min-') and have also been attributed to cyclical production of the second messenger (Woods et al., 1987;Meyer and Stryer, 1988). During our study on the effect of heparin, none of the cells that we tried presented slow oscillations, and we cannot therefore conclude that slow oscillations are independent of InsP3. In fact, TLC-S may function as a trigger to set up the system, the main part of the transient spike being due to the recurrent InsP3 production. However, the initial release determined in permeabilized cells is large, and oscillations described here are not inhibited by heparin, suggesting that InsP3 production is not responsible for the action of TLC-S. This suggests that the mechanisms that control the [Ca2+Ii in the liver cells are probably very complex and may involve spatial gradients of Ca2+ concentration between the plasma membrane and the InsP3-sensitive store.