Delayed "All-or-None" Activation of Inositol 1,4,5-Trisphosphate-dependent Calcium Signaling in Single Rat Hepatocytes*

When single rat hepatocytes were stimulated with the phospholipase C-activating hormone, vasopressin (from 300 PM to 1 p ~ ) , the [Ca2+], signals were always "all-or- none" responses. At low concentrations of vasopressin, Ca2+ release was maximal because liberation of additional inositol 1,4,5-trisphosphate (IP,) by photolysis of its caged precursor at the top of the [Ca2+li spike failed to increase [Ca2+], further. However, if IP, was generated by photolysis of caged IP, in previously unstimulated cells, [Ca2+], increased immediately, and the magnitude of the response was a graded function of the quantity of IP, released. We also analyzed the kinetics of activation of intracellular IP, receptor/Ca2+ channels by monitoring the quench of sequestered dye by the entry of cytoplas- mic Mn2+ into fura-2-loaded intracellular IPS-sensitive organelles. This Mn2+-induced quench was precipitous and always preceded by a delay inversely related to the vasopressin concentration. In hepatocytes stimulated with 10 n~ vasopressin, IP, increased slowly, and the half-time of the IP, rise was comparable with the latency for the release of intracellular calcium. The slow rise in IP, would be predicted to produce accelerating Ca2' re- lease. This is consistent with the results of the Mn2+ quench

In hepatocytes, submaximal concentrations of phospholipase C-linked hormones induce regular and repetitive spikes, or base-line oscillations of [Ca2+],,l first described by Woods et al. (1). Increasing the agonist concentration does not modify the amplitude of the spikes, but it enhances their frequency. Similarly, the latency before the first [Ca2+], spike is inversely related to agonist concentration (2). This constancy of amplitude with varying latency is indicative of "all-or-none" behavior such as that seen in the regenerative process underlying action po-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

EXPERIMENTAL PROCEDURES
Preparation of Hepatocytes, Fura-2IAM Loading and Fluorescence Measurements-Procedures for preparing hepatocytes and measurement of [Ca' ' ], in single fura-2/AM loaded cells were similar to those described by Glennon et al. (5). Hepatocytes were prepared by collagenase perfusion of livers from male Sprague-Dawley rats (200-300 9). Cells were maintained in ice-cold Dulbecco's modified Eagle's medium (pH 7.4 with 2% bovine serum albumin) equilibrated with 5% CO, and 95% 0,. Hepatocytes were plated on polylysine-coated glass coverslips (500 pg/ml) and incubated in the presence of 2 p~ fura-2/AM at 37 "C under 5% CO, for 15-30 min for measurement of [Ca"],. These conditions resulted in substantial dye present in the cytoplasm, whereas longer incubations (45-60 min) resulted in more compartmentalized fura-2 (for example, Fig. 1) and were thus better suited for Mn2+ quench experiments (described below). After the incubation period, the coverslip with attached fura-2-loaded cells was mounted in a Teflon microscope chamber (Bionique Laboratories, Lake Saranac, N Y ) and washed three times with 1 ml of a Hepes-buffered Krebs-Ringer media (KRH) containing, in mM: NaCl (120), KC1 (5.41, MgSO, (0.8), CaCI, (1.8), glucose (111, Hepes (201, and 0.2% bovine serum albumin at pH 7.4. All experiments in this study were then carried out in medium lacking added Ca2+. Single cell fluorescence measurements were obtained by using a Nikon Diaphot inverted microscope equipped with a 40x Neofluor objective and a Delta-Scan 1 dual excitation system (Photon Technologies International, Princeton, NJ). The excitation wavelengths were 340 and 380 nm, and fluorescence emission was measured at 530 nm. All experiments were performed at room temperature, and experimental agents were applied by exchanging the solutions of the microscope chamber.
Because fura-2 is partly compartmentalized in hepatocytes after loading with the acetoxymethylester form of the dye, we did not calculate apparent values for [Ca''], from the ratios. The data are presented in one of two ways. In experiments examining the latency to the rise in [Ca2+li, ratios of fluorescence with excitation at 340 and 380 nm (F34d which increase as [Ca2+], increases, are shown. In experiments determining the latency to the activation of intracellular IP, receptors in the absence of Ca2+ mobilization, intracellular Ca2+ pools were discharged by thapsigargin, and Mn' ' was allowed to enter the cytoplasm until all cytoplasmic fura-2 was quenched and free Mn' ' was present in excess. After a wash to remove all extracellular Mn", activation by vasopressin of intracellular IP, receptors resulted in the penetration of cytoplasmic Mn2+ into the IP,-sensitive, Ca2+ storage organelle (presum- A single hepatocyte loaded with fura-WAM for 45 min as described under "Experimental Procedures" was treated with 2 m M MnZ' in the absence of external Ca". This caused a quench of fura-2, which was measured by monitoring F,,, (see "Experimental Procedures"). The initial drop in Fim was due to Mn2+ quench of cytoplasmic dye because permeabilization of the plasma membrane with 50 pg/ml saponin produces no further quench. Addition of 10 p~ of the divalent cationophore, ionomycin, resulted in quench of the remaining indicator, indicating its presence in intracellular organelles.
ably through the IP,-regulated channel), and quench of compartmentalized fura-2 occurred. Rather than monitor fluorescence directly with excitation at an isosbestic wavelength, an algebraic summation of the fluorescence values with excitation at 340 nm and 380 nm (Fim = x.F340 + F380, where x = 3-4, determined empirically for each preparation) was used to obtain a measure of Ca2+-independent fluorescence.
Microinjection and Photolysis of Caged-IP, and Caged-GPIP, in Hepatocytes-Cells were microinjected as described previously (6). . The excitation wavelength used was 495 nm, the emission was collected at 530 nm. Since calcium green is not a ratioing indicator, tCa2+li values were not calculated. During microinjection, Ca2+ was omitted from the KRH. After microinjection, cells were allowed to recover for at least 20 min in KRH containing Ca2+ (1.8 mM) to insure that intracellular Ca2+ stores were filled. Then, Ca2+ was removed again at least 10 min before the beginning of the experiment.
Photolysis of Caged-IP, and Caged-GPIP,-Photolysis of caged compounds was achieved by using techniques described previously (7). PHIIP, Accumulation-For measurement of inositol phosphate formation, freshly isolated hepatocytes (at least 4 ml of packed cells) were incubated 90 min with agitation at 37 "C with 5 mCi [3HIinositol (American Radiolabeled Chemicals, St Louis, MO) and 1% bovine serum albumin in Dulbecco's modified Eagle's medium as described previously (8). The final volume of incubation was 10 ml. Cells were then centrifuged at 600 rpm for 5 min, and resuspended in KRH (with 1.8 m M Ca2+ and 0.2% bovine serum albumin). Hepatocytes were washed two more times, and were resuspended at 25 "C in KRH without Ca". After 20 min, suspensions were treated either with the vehicle (dimethyl sulfoxide) or thapsigargin (2 PM). In some experiments Mn2+ (2 mM) was also added with thapsigargin. After 10 min, vasopressin (10 nM) was added. The reaction was stopped by pipetteting 500 pl of cells at the appropriate time into an ice-cold mixture of 12% perchloric acid, phytate (1 mg/ml), and EDTA (4 mM). Samples were subsequently processed and analyzed by HPLC for [,H]IP, as described previously (8).
Material~-[~H]Inositol was purchased from American Radiolabeled Chemicals; fura-P/AM and calcium green were from Molecular Probes. Caged-IP, and caged-GPIP, were from Calbiochem, and collagenase A was from Boeringer Mannheim. Thapsigargin was from LC Services, and vasopressin and all other chemicals were from Sigma.

RESULTS
When fura-2IAM loaded-hepatocytes were stimulated with different concentrations of vasopressin, an initial spike of calcium (F34,jF380 increase) was observed after a delay that was inversely related to the hormone concentration (Fig. 2, top; Fig.  3, closed circles). Note that all of the experiments shown in this study were carried out with freshly isolated hepatocytes and in nominally calcium-free medium; in our hands under these conditions we seldom saw continuous base-line oscillations or even a second spike. Statistical analysis confirmed that the maximal rate of rise of the [Ca''], spike was not significantly affected by increasing the concentration of vasopressin from 300 PM (TH spike = 1.77 * 0.25 s; mean 2 S.E., n = 11) to 1 PM (Tv, = 1.31 2 0.13 s; mean * S.E., n = 5). Furthermore, as reported previously (91, the release of intracellular Ca2+ was all-or-none in that, at the lowest concentrations tested, it appeared that only a percentage of cells showed a [CaZ+l, response. As shown in Fig. 4, in marked contrast to the situation with vasopressin (all-or-none [CaZ+l, spike), graded photolysis of the caged precursor produced "quantal" [Ca2+l, signals that occurred with no measurable delay. Similar results were obtained The cell, in a medium lacking external Ca2+, was exposed to increasing flash intervals when indicated, and then a near threshold concentration of vasopressin was added. On top of the delayed Ca2+ peak, a subsequent ments). Saponin (50 pg/ml) and Ca2+ (10 mM) were added at the end of supramaximal photolysis had no effect ( n = 31 independent experithe experiment to insure that the dye was not saturated. with a caged derivative of a poorly metabolizable analog of IP,, GPIP, (7) (data not shown). However, a low concentration of vasopressin (300 PM) activated a maximal [Ca2+], spike with a considerable delay, and when IP, was released photolytically at the top of the spike, no further Ca2+ release was observed (Fig. 4).
To attempt to understand this apparent paradox (graded versus all-or-none behavior), we next exploited the ability of fura-2 to be compartmentalized in IP,-sensitive organelles ( 5 , 10, 11). In compartmentalized cells, the kinetics and extent of activation of the IP, receptor-channel can be examined under conditions where Ca2+ fluxes are eliminated by pretreatment with thapsigargin. We used Mn2+ as a surrogate for Ca2+, not in the conventional manner to indicate divalent cation movement across the plasma membrane but rather to report divalent cation permeability of the intracellular Ca2+ channel controlled by IP, ( 5 , 9,12,13). A representative experiment performed with an hepatocyte loaded with 2 VM fura-a/AM for 45 min is shown in Fig. 5 . Fluorescence data for excitation at both 340 and 380 nm are shown on the top panel, and both F34dF380 and F,,, are displayed in the bottom panel. As expected, thapsigargin increased F34dF380 ( i e . [Ca"],) but had no effect on F,,,. Upon the addition of 2 mM Mn2+, F34,jF380 increased and Fi,, decreased, indicating that the cytoplasmic portion of the dye was quenched. The ratio increased because the [Ca2+] within most organelles is higher than [Ca"],. Once the cytosolic dye was completely quenched, the bathing medium was changed t o one lacking Mn2+, and then the addition of 1 nM vasopressin induced a further precipitous quench of the fura-2 fluorescence after a delay of about 1 min. This latency, although somewhat variable, was observed in all such experiments. Since no extracellular Mn2+ was present when vasopressin was added, the quench indicates movement of cytoplasmic Mn2+ into the IP,sensitive pool, consistent with previously published observations (5, 9,12,13). The lag in the onset of the quench thus represents a delay in the activation of the IP,-regulated channels, allowing Mn2+ entry, rather than a delay in the quenching of the dye by Mn2+ because (i) the latency depends on vasopressin concentration (see below) and (ii) Mn" has a very high affinity for fura-2 (Kd = 3 nM, Ref. 14), and thus it quenches the dye almost instantaneously and with high binding affinity when it enters the pool. Furthermore, with this experimental protocol, Ca", which might compete with Mn2+ for binding to fura-2, is depleted from the intracellular IPS-sensitive store by prior treatment with thapsigargin.
As previously observed for the experiments measuring the delay to the rise in [Ca2+], (Fig. 2, top; Fig. 3, closed circles), and in confirmation of the findings of Hajn6czky et al. (131, the delay preceding the quench of sequestered dye by Mn2+ was also inversely related to the concentration of vasopressin (Fig. 2  bottom; Fig. 3, open circles). As for the [Ca"], spiking experiments, the Mn2+ quench was all-or-none. As shown in Fig. 6, even low concentrations of vasopressin caused a complete quench of the fura-2 trapped in the IP,-sensitive pool. Note that for the cell in Fig. 6 A , a very low concentration of vasopressin was capable of activating quench, and when this was complete, 1 PM vasopressin produced little or no additional quench. The cell in Fig. 6B was apparently less sensitive to vasopressin; the near threshold concentrations of 300 PM and 1 nM did not cause significant quench, but 10 nM vasopressin induced a precipitous quench. Again, a supramaximal concentration of vasopressin caused no additional quench. However, unlike the case for the [Ca"], spikes, the maximal rate of Mn" quench increased as a function of vasopressin (TY> at 300 PM = 46.5 f 8.30 s, mean * S.E., n = 5 , and T s a t 1 VM = 19.36 2 2.74 s, mean S.E., n = 5). The ability of Mn2+ to quench the entire agonist-sensitive calcium pool, at a rate that depends on the agonist concentration, has been reported previously (13). Finally, the data summarized in Fig. 3 show that the average latency for full activa- were added when indicated. This experiment was carried out three times with similar results. B , another hepatocyte, with a different sensitivity to vasopressin was monitored in the same manner. In this case 300 PM, 1 m, 10 m, and 1 p~ vasopressin were added. Note that for both cells, 1 p~ vasopressin does not produce a further quench, but some dye is still quenched by ionomycin addition. tion of the IP, receptor (see legend to Fig. 3 for method of measurement of latency) in Ca2+-depleted cells was more than twice as long as the latency to the [Ca2+], spike in cells with intracellular calcium stores intact. This is especially clear in experiments such as the one summarized in Fig. 7, in which we were able to demonstrate both Ca2+ mobilization and, subsequently, Mn2+ quench of compartmentalized fura-2 in a single hepatocyte. Note that we did not attempt to measure the latency of response to photolyzed IP, with the Mn2+ quench protocol because photolysis experiments require indicators with excitation wavelengths in the visible region and thus cannot be carried out in cells with compartmentalized fura-2.
We reasoned that the slowly developing Mn2+ quench, reflecting an accelerating increase in divalent cation permeability, might result from a slow accumulation of IP,. Thus, we measured the time course of IP, production due to 10 nM vasopressin in cells with intracellular calcium stores either intact or in cells previously depleted by thapsigargin. Fig. 8 shows that in both cases, IP, increased rather slowly, requiring about 30 s to reach a maximal level. Interestingly, in cells with intact intracellular stores, and thus capable of producing an initial calcium spike, IP, increased more rapidly and was at least transiently higher than in the cells previously depleted of their stores by thapsigargin. However, it appeared that in the steady state, the levels of IP, were not significantly different. This result suggests that the accelerating rate of Mn2+ quench could result from a slow accumulation of IP,. Also, differences in the kinetics of IP, accumulation may explain the different latencies shown in Figs. 2 and 3.

DISCUSSION
The purpose of this work was t o gain insight into the mechanisms underlying the delayed, all-or-none intracellular Ca2+ mobilization observed when hepatocytes are stimulated with the phospholipase C-activating hormone, vasopressin. In confirmation of earlier studies (2), we found that the initial [Ca"], spike due to low concentrations of vasopressin occurred after a delay that was inversely related to the concentration of vaso-  mobilization (F34,jF380). The cell was washed, and allowed to recover for 30 min. It was then challenged with thapsigargin and Mn2+, and washed as in Fig. 5, and then a second stimulation with vasopressin (1 m) induced a further quench (Fiao). Both responses are aligned such that the additions of vasopressin are at time zero (dotted line). This experiment was carried out three times with similar results. [3Hlinositol-labeled cells were treated with 10 m vasopressin (at t = 0) in calcium-free medium and were then analyzed by HPLC for content of C3H1IP3 at the times indicated. In the experiments indicated by open circles, the cells were pretreated with 2 p~ thapsigargin as in the Mn" quench experiments (n = 4). In two experiments, 2 m~ Mn2+ was added along with thapsigargin (filled triangles), with no significant effect. Control cells, indicated by filled circles, were treated with a similar quantity of vehicle (dimethyl sulfoxide) ( n = 6). The data are expressed as cpm [3HlIP&pm [3H]inositol (~1,000) (means f S.E.). pressin applied. However, the maximal rate of rise of [Ca2+], was independent of the vasopressin concentration, and the extent of release appeared to be maximal; i.e. in responsive cells, all IP3-releasable calcium was mobilized, even by very low concentrations of vasopressin. In contrast, when IP, was increased either by photolysis of caged-IP, or caged-GPIP,, the calcium signal always occurred without delay, and the extent of release was graded with the quantity of inositol trisphosphate agonist released.
To determine the reasons for this apparent paradox, we examined the kinetics of the IP3-induced increase in permeability of intracellular calcium stores by using the intracellular Mn2+ quench method first described by Glennon et al. (5) as an assay. This technique relies on the fact that in some cells types, such as hepatocytes, loading with the acetoxymethylester derivative of fura-2 leads to dye accumulation in intracellular organelles, including the agonist and IPS-sensitive stores. When Mn2+ is present in the cytoplasm, activation of the intracellular IP, receptorkhannel allows retrograde flow of Mn2+ into the store, resulting in quench of the sequestered dye. By using this approach, we found that, following agonist stimulation, the permeability of the intracellular stores does not immediately increase, as would be predicted by the photolysis experiments, but increases gradually and then precipitously after a considerable delay that again depends on the concentration of vasopressin.
The simplest explanation for this behavior is that the permeability of the intracellular stores begins to significantly increase only when the level of IP, reaches some critical level. If IP, rises rapidly, and almost instantaneously as in photolysis of caged precursors, the permeability will rapidly increase and remain constant for a period, and CaZ+ will exit at a constant rate. However, if IP, levels increase more slowly, as happens when low concentrations of vasopressin are used, then the permeability of the stores will also increase with a similar long time course, causing the apparent accelerating rate of Mn2+ quench. In addition, the binding of IP, to its receptor is known to be a cooperative process ( E ) , and this could also contribute to the rapid acceleration of activation.
In the Ca2+ replete condition, it would also be expected that the rate of Ca2+ release would gradually increase. This acceleration could be important in the eventual initiation of a calcium-induced calcium release-triggered (16), all-or-none [Ca"], spike. Indeed, when we measured IP, accumulation in hepatocytes, we found that with 10 I " vasopressin, about 30 s were required for IP, levels to reach their peak. Since the cellular half-time for IP, is much shorter (for example, see Ref. 171, this most likely reflects the time course of activation of the phospholipase C-signaling system by vasopressin. In support of the interpretation that this slow rate of IP, accumulation determines the latency to the [CaZ+], spike, prior depletion of intracellular stores with thapsigargin increased the delay to full activation (measured as intracellular Mn2+ quench) and also prolonged the rise in IP, to about the same extent (Fig. 8). Note that this may indicate a role for [Ca2+], in the initial kinetics of activation of phospholipase C by vasopressin. A role of [Ca2+li in activating phospholipase C is a component of the model proposed by Meyer and Stryer (18) for calcium spiking. However, our data indicate that only the rate of increase in IP, is facilitated by intracellular [Ca2+l, release; the steady-state level of IP, was apparently unaffected. Additional experimental work will be required to ascertain whether or not [Ca2+li activation of phospholipase C and slowly oscillating levels of IP, play a role in the maintenance of sustained [Ca"], spiking in hepatocytes.
Prior reports have pointed out the necessity of calcium priming mechanisms to achieve a delayed, all-or-none calcium response (19). As a result of the current study, we suggest that an important determinant of this calcium signaling pattern is the kinetics of IP, accumulation. In fact, previous experimental results suggesting a role of calcium in priming the spiking behavior may reflect calcium effects on the rate of rise of IP,. Since the kinetics of IP, accumulation will differ for different agonists and receptor types, the likelihood of obtaining all-ornone [Ca2+], spikes and continuous base-line spike oscillations may also similarly vary. An interesting case is the activation of the muscarinic cholinergic receptor, which often does not give delayed all-or-none responses; rather, as was the case in this study with caged compounds, muscarinic stimuli induce immediate, graded rises in [Ca2+], (although sometimes accompanied by rapid sinusoidal oscillations) (4, 20, 21). It has been considered paradoxical that even in the same cell type, the pancreatic acinar cell, a peptide agonist, cholecystokinin, causes delayed all-or-none base-line spikes, while activation of the muscarinic pathway does not (21). We note that the agonists used to activate muscarinic receptors typically have receptor affinities orders of magnitude less than peptide hormones for their receptors. It is expected, therefore, that the rate of equilibration of muscarinic agonists with their binding sites would be much faster than for peptide agonists. Thus, the reason that musca-rinic agonists do not generally give rise t o delayed, base-line spikes could result from the rapid accumulation of IP, to steady-state levels; i e . the IP, kinetics may be more similar to those artificially generated in the flash photolysis experiments.
What might be the physiological consequences of these kinetic distinctions? It has been argued that base-line spiking, generated by very low but prolonged stimulation by peptide hormones, allows for detectable signal-to-noise and thus may be important for long term responses such as those involved in cell growth and differentiation (22). With the neurotransmitter acetylcholine, perhaps it is more important to provide rapidly increasing, graded, and rapidly reversible changes in [Ca2+l, solely to regulate acute responses, such as secretion or smooth muscle contraction.
In single corticotropin pituitary cells (231, it has been shown that corticotropin releasing factor, a CAMP-stimulating agent, stimulates corticotropin secretion in a concentration-dependent manner. With higher concentrations of corticotropin releasing factor, each single cell releases more corticotropin (i.e. a graded response). In contrast, with the IP, generating hormone vasopressin, the extent of corticotropin secretion (measured in single cells with the reverse hemolytic plaque assay method) is not concentration dependent (all-or-none response). Rather, as is the case in hepatocytes, the percentage of responding cells was increased with higher concentrations.
In summary, our data provide evidence that the kinetics of IP, accumulation is an important determinant of delayed allor-none calcium spiking behavior. However, there are likely other significant factors that regulate the kinetics of [Ca2+l, spiking in oscillating cells. The phenomenon of calcium-induced calcium release by the IP, receptor (24, 25) is probably responsible for the all-or-none nature of the spike. Consistent with this suggestion, we observed that the rate of rise of [Ca2+1, spikes was independent of agonist concentration, but the rate of Mn2+ quench, when no Ca2+ release occurred, was dependent of the agonist concentration. Based on our data, we suggest that the delay preceding the first spike occurs primarily because initially the IP, level is low and the rate of Ca2+ release is slow due to the various brakes and negative feedbacks on cytoplasmic calcium (for example, intracellular calcium buffers, pumps, and the inhibitory effects of decreasing luminal calcium concentration). However, as the level of IP, increases, the rate of Ca2+ release will at some point exceed these negative influences, [Ca2+], will begin t o rise, and the process of calcium induced calcium release will trigger an all-or-none spike.