myo-Inositol 1,4,5-Trisphosphate A SECOND MESSENGER FOR THE HORMONAL MOBILIZATION OF INTRACELLULAR Ca2+ IN LIVER*

The stimulation of hepatocytes by al-adrenergic ag- onists and vasoactive peptides results in a mobilization of intracelluiar Ca2+ which is accompanied by break- down of phosphatidylinositol 4,5-bisphosphate to release myo-inositol 1,4,5-trisphosphate (Zns(1,4,5)P3). The possible involvement of Ins(1,4,5)Ps in intracel- lular Ca2+ mobilization was tested using a preparation of saponin-permeabilized hepatocytes. Added Ca2+ was sequestered by intracellular organelles in the presence of ATP until the medium free Ca2+ concentration was lowered to a new steady state level. The subsequent addition of Ins(1,4,5)P3 caused a rapid Ca2+ release, which was complete within 5 s. Half-maximal and maximal Ca2+ release were obtained at concentrations of Ins(1,4,5)P3 of 0.1 and 0.5 PM, respectively. The maximal amount of Ca2+ mobilized was 450 pmol/mg of cell dry weight. Using experimental conditions de- signed to permit selective Ca2+ accumulation into mitochondrial or non-mitochondrial stores, it was deter- mined that all of the Ca" released by Ins(1,4,5)P3 originated from non-mitochondrial, essentially as described by Tsien et al. (16). Unlabeled Ins(1,4,5)P3 was prepared by hydrolysis of the naturally occurring PtdIns(4,5)Pz in erythrocyte ghosts or ox brain extracts. The endogenous inositol lipid phosphodiesterase (phospholipase C) activity of these tissues was stimulated by addition of 2 mM CaC12 and the related Ins(1,4,5)P3 was separated by anion exchange chro-matography. The details of the preparative procedure and authenti- cation of Ins(1,4,5)P3 were essentially the same as those given else-where (17, 18). The concentration of Ins(1,4,5)P3 was determined by measurement of organic phosphorus (19). 32P-labeled Ins(1,4,5)P3 was prepared in a similar manner from erythrocyte ghosts after prelabel- ing the erythrocytes with "Pi for 12 h at 30 "C. Under these conditions, a constant specific activity of PtdIns(4,5)Pz was obtained which presumably reflects an equal distribution of 3'P between the 4 and 5 positions on the inositol ring. Most other chemicals and biochemicals were obtained from the sources given previously (4,9).

in phosphorylase a activity can be observed in the complete absence of extracellular Caz+, indicating that the initial increase in cytosolic free Ca2+ is due to the mobilization of Ca2+ from intracellular compartments (3,4). Although the identity of these compartments remains controversial, evidence has been presented to suggest that both mitochondrial and nonmitochondrial compartments may be involved (4)(5)(6). It is still not known how a hormone, which binds at the cell surface, can release Ca2+ from compartments located within the cell interior.
An increased hydrolysis of inositol phospholipids is associated with the action of calcium-mobilizing stimuli in a number of different cells (for reviews see Refs. 7 and 8). In the liver, such stimuli lead to the breakdown of PtdIns(4,5)Pz' by a phosphodiesterase with the release of Ins(1,4,5)P3 and diacylglycerol (9)(10)(11): This breakdown of PtdIns(4,5)P2 is not secondary to the elevation of cytosolic Caz* and occurs sufficiently rapidly to play a role in generating a messenger for mobilizing intracellular Ca2+ (9, lo).' Since pathways exist for the rapid production and degradation of Ins(1,4,5)P3 in liver and this compound is water-soluble, it appears to be the most promising intermediate of hormone-stimulated inositol lipid metabolism potentially capable of acting as a messenger for the mobilization of calcium from intracellular compartments. We have tested this hypothesis and data are presented in this communication to show that low concentrations of Ins(l,4,5)P3 can cause a rapid release of Ca2+ from a nonmitochondrial store in saponin-permeabilized rat hepatocytes. Our data confirm and extend similar observations that have been made recently on the action of lns (1,4,5)P3 in a preparation of "Ieaky" pancreatic acinar cells (12) and in saponinpermeabilized guinea pig hepatocyte^.^

MATERIALS AND METHODS
Isolated hepatocytes were prepared as described previously (4) (13). Measurements were made with a remote calomel reference electrode (Radiometer K4040). The Ca2+ electrode was calibrated using the procedure described by Bers (14). Calcium movements, measured with the fluorescent Ca2+ indicator Quin 2 (15), were carried out with cells at 5 mg dry weight/ml in the medium described above supplemented with 75 p~ Quin 2 (free acid form; Calbiochem) plus oligomycin (3 pg/ml) and rotenone (0.5 pg/ ml). Substrates such as succinate (5 mM) and MgATP plus a creatine phosphate ATP regenerating system were added as required after saponin permeabilization was complete. Quin 2-Ca fluorescence was measured using a fluorimeter (Johnson Research Foundation, Electronics Shop) with excitation at 335 nm and emission at 490-570 nm selected by specific filters.' The cell suspension was continuously stirred in the cuvette chamber maintained at 37 "C. The required free Ca'+ in the medium was obtained by adding known amounts of Ca2+ to the Quin 2 buffer. Calibration of the Quin 2-Ca fluorescence signal was made by adding 2.5 mM EGTA (plus Tris base to give pH 8.3) followed by 5 mM CaC12, essentially as described by Tsien et al. (16).
Unlabeled Ins(1,4,5)P3 was prepared by hydrolysis of the naturally occurring PtdIns(4,5)Pz in erythrocyte ghosts or ox brain extracts. The endogenous inositol lipid phosphodiesterase (phospholipase C) activity of these tissues was stimulated by addition of 2 mM CaC12 and the related Ins(1,4,5)P3 was separated by anion exchange chromatography. The details of the preparative procedure and authentication of Ins(1,4,5)P3 were essentially the same as those given elsewhere (17, 18). The concentration of Ins(1,4,5)P3 was determined by measurement of organic phosphorus (19). 32P-labeled Ins(1,4,5)P3 was prepared in a similar manner from erythrocyte ghosts after prelabeling the erythrocytes with "Pi for 12 h at 30 "C. Under these conditions, a constant specific activity of PtdIns(4,5)Pz was obtained which presumably reflects an equal distribution of 3' P between the 4 and 5 positions on the inositol ring. Most other chemicals and biochemicals were obtained from the sources given previously (4,9).

RESULTS AND DISCUSSION
ins(1,4,5)P3 Specific Ca2+ Release from Permeabilized Liver Cells-In order to examine possible effects of Ins( 1,4,5)P3 on the release of Ca2+ from intracellular pools, a preparation of hepatocytes was used in which the plasma membrane was permeabilized by saponin treatment. For measurement of Ca2+ movements using a Ca2+-sensitive electrode, isolated liver cells were incubated in a medium designed to mimic the ionic composition of cytosol containing high K+, low Na+, and no added Ca" (giving a free Ca2+ of 1-2 p~) .
When ATP was present in the medium prior to disruption of the plasma membrane by saponin, the Ca2+ concentration of the medium was reduced by sequestration into the newly exposed intracellular Ca2+ pools. After approximately 5 min, a steady state free Ca2+ of 0.45 p~ (pCa2+ = 6.35) was established as measured with the Caz+ electrode. Addition of Ins(1,4,5)Pa under these conditions resulted in a rapid release of Ca2+ which was then slowly reaccumulated until the original steady state free Ca2+ was restored (Fig. la). No Ca2+ release was observed with Ins(4,5)P2, L-myoinositol 2-phosphate, or fructose 2,6bisphosphate. No significant response of the electrode was observed when Ins( 1,4,5)Pa was added to the incubation buffer in the absence of hepatocytes (Fig. l b ) .
Sequential additions of Ins(1,4,5)P3 produced cycles of Ca2+ release and reuptake in which the amount of Ca2+ released was dependent on the concentration of added Ins(1,4,5)P~. The large difference in electrode response upon Ca2+ additions in the presence and absence of permeabilized hepatocytes indicates considerable buffering by cellular Ca2+ binding sites  ( Fig. 1). Consequently, the amount of Ca2+ released by Ins(1,4,5)P3 was quantitated by making appropriate additions of known amounts of Ca2+ into the electrode chamber at the end of each trace.
The relationship between the concentration of Ins(1,4,5)P3 added and the amount of Ca2+ released is shown in Fig. 2. A maximum value of about 450 pmol of calcium/mg of cell dry weight was released by Ins(1,4,5)P3. This value is within the range of values reported for the amounts of Ca2+ released from intracellular stores in the perfused liver and isolated hepatocytes by concentrations of vasoactive peptides and cy1adrenergic agonists giving a maximum glycogenolytic response (3,4,20). Half-maximal and maximal amounts of Ca2+ release were obtained at Ins(1,4,5)P3 Concentrations of 0.1 and 0.5 p~, respectively. These concentrations compare very well with our recent measurements of Ins(1,4,5)P3 in intact hepatocytes after vasopressin treatment? In that study, the increase of Ins(1,4,5)P3 which coincided with peak cytosolic free Ca2+ elevation was calculated to be 0.6 p~. Thus, both the amount of Ca2+ released by Ins(1,4,5)P3 and the concentration dependence of this release observed in the permeabilized hepatocyte are compatible with a role for Ins (1,4, (l,4,5)P3-The nature of the Ca2+ pool from which Ins(1,4,5)P3 releases Ca2+ in the saponin-permeabilized hepatocytes has been further investigated by using the fluorescent Ca'+ indicator Quin 2 as both a Ca2+ buffer and a Ca2+ indicator. This approach has some advantages over the Ca2+ electrode in that it allows the medium Ca'+ to be buffered at a fixed, predetermined concentration. In addition, the fluorimeter used for these studies has a time resolution of about 0.2 s (based on Ca2+ pulse additions) which allows resolution of the very rapid kinetics of Ins(l,4,5)P3-induced Ca2+ release (see below). For experiments using Quin 2, hepatocytes were permeabilized with saponin in the K+ based buffer described above but with the addition of 75 PM Quin 2, which reduced the concentration of free Ca2+ in the medium to about 20 nM. Under these conditions, in the absence of added substrates and in the presence of rotenone and oligomycin to prevent endogenous substrate formation and utilization, there was no uptake of medium Ca2+ during saponin treatment, and the endogenous Ca2+ content of intracellular organelles was minimal. When cell permeabilization was complete, Ca'+ could be added to give the required starting free Ca'+ concentration.
As shown in Fig. 3, when succinate was added to energize the mitochondria of saponin-permeabilized hepatocytes, there was no significant uptake of Ca2+ if the medium free Ca2+ was in the range of 150-200 nM, which is believed to be the basal free Ca2+ in the cytosol of intact hepatocytes (21,22).' Addition of MgATP, however, caused a substantial sequestration of Ca2+ which amounted to about 2 nmol/mg of cell dry weight. Dissipation of the mitochondrial electrochemical proton gradient with the uncoupler, 1799 (23), caused only a small amount of Ca2+ release after ATP-dependent Ca2+ uptake, while essentially all of the sequestered Ca'+ could be released by the subsequent addition of the Ca'+ ionophore ionomycin (Fig. 3B). In the absence of ATP, the ionomycin-sensitive Ca2+ pool was negligible (Fig. 3C). These results are similar to those obtained in other permeabilized cell preparations (24,25) where it has been shown that at Ca2+ concentrations believed to occur in the cytosol of unstimulated cells, the major intracellular site of Ca2+ sequestration is a non-mitochondrial, ATP-dependent calcium pool. The finding that this pool of Ca'+ can be released by ionomycin indicates that it is a vesicular pool, presumably associated with the reticular system of the cells.
In agreement with the results obtained using the Ca'+ electrode (Fig. l), addition of Ins(1,4,5)P3 after ATP-dependent Ca'+ uptake into the saponized hepatocytes resulted in a rapid Ca2+ release followed by a slower reaccumulation of this Ca2+ (Fig. 3A). Ins(1,4,5)P3 did not release Ca2+ after addition of ionomycin (Fig. 3B) or in the absence of ATP (Fig. 3C) and there was no alteration in the distribution of Ca2+ between the mitochondrial and non-mitochondrial pools after Ins(1,4,5)P3 treatment (Fig. 3, A uersus B ) . These data indicate that Ins(1,4,5)P3 releases ca'+ from a vesicular ATPdependent, non-mitochondrial pool while reuptake of Ca2+ occurs into the same or a similar pool. Our finding that multiple cycles of Caz+ release and reuptake can be obtained even after addition of a saturating concentration of Ins(1,4,5)P3 (Fig. 1) also suggests that the Ins(l,4,5)P3-sensitive Ca2+ pool refills during the reuptake phase.
Using both cell fractionation and nondisruptive techniques, it has previously been found that mitochondria contain substantial amounts of Ca2+ in the intact hepatocyte and that, in addition to the non-mitochondrial calcium pool that can be of Ca2+ between mitochondrial and non-mitochondrial vesicular Ca2+ pools was determined by sequential additions of 5 p~ uncoupler 1799 followed by 5 rg/ml of ionomycin. In truce A , Ins(1,4,5)P3 was added after substrate-induced Ca2+ sequestration, in trace B, vesicular Ca2+ pools were released by ionophores before Ins(1,4,5)P3 addition, and in trace C, ATP was omitted. mobilized by hormones, Ca'+ can also be released from the mitochondrial pool after hormone treatment (4,5 ) . Since the mitochondria of the saponin-permeabilized hepatocytes do not take up Ca2+ under the conditions used in Fig. 3, it was not possible to determine whether Ins(1,4,5)P3 can also release Ca'+ from the mitochondrial Ca2+ pool. Under these conditions, the discrepancy in mitochondrial calcium content between the saponin-permeabilized hepatocyte and our previous measurements with intact hepatocytes could result from the dilution of some cytosolic effector of mitochondrial Ca2+ transport. This and other possibilities are currently under investigation.
As shown in Figs. l a and 3A, in saponin-treated cells which had taken up Ca'+ to a steady state in the presence of ATP, subsequent addition of Ca2+ did not result in further Ca'+ uptake. This indicates that the ATP-dependent Ca2+ pool was saturated. However, after further additions of Ca2+ had elevated the free Ca" to greater than 0.7 p~, a second lower affinity Ca'+ pool was revealed and the added Ca2+ was again sequestered (Fig. la). The ability of this second pool to completely remove sequential pulses of Ca2+ and maintain a constant steady state free Ca2+ concentration in the medium

Second Messenger
Role for Inositol Trisphosphate resembles the "set point" behavior of isolated mitochondria (26). Inhibition of Ca2+ uptake into this pool by ruthenium red, uncouplers, and respiratory chain inhibitors, suggests that the lower affinity Ca2+ pool represents Ca2+ uptake by mitochondria. In order to assess possible effects of Ins(1,4,5)P3 on mitochondrial Ca2+ release, the action of this compound was tested under conditions where the free Ca2+ of the medium was elevated to a level where appreciable amounts of Ca2+ were sequestered within the mitochondria.
In the experiment shown in Fig. 4A, after the completion of ATP-dependent Ca2+ uptake, further Ca2+ was added and the mitochondria were allowed to accumulate 2-3 nmol of Ca2+/mg of cell dry weight (equivalent to about 10 nmol of Ca2+/mg of mitochondrial protein). The addition of a small Ca2+ pulse clearly shows that the mitochondria in these cells maintained the medium free Ca2+ at approximately 0.7 pM. Under these conditions, Ins(1,4,5)P3 was able to release a similar amount of Ca2+ to that observed at the lower medium free Ca2+ used in Figs. 1-3. However, all of this Ca2+ was apparently still derived from the non-mitochondrial pool because no Ins(l,4,5)P3-induced Ca2+ release was observed in the absence of ATP (Fig. 4B). Furthermore, the Ca2+ release observed when mitochondrial Ca2+ uptake was blocked by the addition of ruthenium red and omission of succinate (Fig. 4C) was identical to that when the mitochondria did contain Ca2+ (Fig. 44). Thus, the data presented above clearly show that Ins( 1,4,5)P3 does not release Ca2+ from the mitochondrial pool in permeabilized hepatocytes and that the observed Ca2+ Ps on mitochondria1 Caz+ content. Ca2+ uptake and release in saponin-permeabilized hepatocytes was measured using Quin 2 exactly as described in the legend to Fig.  3 except that the medium free ea2+ concentration was elevated to a point where mitochondrial Ca" uptake could occur. In the presence of succinate, mitochondria were allowed to accumulate 2-3 nmol of Ca2'/mg of cell dry weight. A calibration addition of 3 nmol/ml of CaCI2 (Ca"') was then added, and after a steady state free Ca2+ had been re-established, 0.5 FM Ins(1,4,5)P3 (IF',) was added. The presence of Ca2+ within the mitochondrial matrix was verified by addition of the uncoupler 1799. In trace A , succinate and MgATP were present to allow Ca2+ uptake into both mitochondrial and non-mitochondrial Ca2+ pools. In trace B , MgATP was omitted, and in trace C, succinate was omitted and 1 p~ ruthenium red was added. 5. Degradation of 32P-Ins(1,4,5)P3 in saponin-permeabilized hepatocytes. The non-mitochondrial Ca2+ pool of saponinpermeabilized hepatocytes was loaded with Ca2+ in the presence of MgATP exactly as described for Fig. 3, trace A. At steady state, 1 p~ "P-labeled Ins(1,4,5)P3 (IF',) was added and the resulting CaZ+ release and subsequent reuptake was measured by using Quin 2-Ca fluorescence. The kinetics of Ca2+ release were resolved in this experiment by using a fast chart speed during the first 10 s after Ins(1,4,5)P3 addition. Ins(1,4,5)P3 breakdown was measured in a parallel incubation where samples (250 J) were removed at the times indicated and quenched in 250 pl of 20% trichloroacetic acid. For the zero time point, 32P-Ins(1,4,5)P3 was added into a trichloroacetic acid extract. The deprotonized samples were neutralized and diluted to 2 ml before separation of inositol phosphates on Dowex 1-X8 anion exchange columns as described by Berridge et al. (18). Each point is the mean of values from two separate incubations. release originates entirely from a non-mitochondrial, vesicular compartment.

200-
Significance of the Ca2+ Reuptake Response-In all cases where Ca2+ release after Ins(1,4,5)P3 treatment has been observed, at least part of the released Ca2+ was always reaccumulated within the following 2-5 min. The transient nature of the Ins(l,4,5)P3-induced Ca2+ release could result from some intrinsic autoregulation of the Ca2+ release mechanism. However, Fig. l a shows that multiple cycles of Ca" release and reuptake can be obtained with sequential Ins(1,4,5)P3 additions. Taken together with the observation that the time taken for Ca2+ reuptake increases with increasing Ins( 1,4,5)P3 concentrations beyond that required for maximal Caz+ release (Fig. la), these results suggested that the reuptake phase might be the result of degradation of Ins(1,4,5)P3. In order to investigate this possibility, 32P-labeled Ins( 1,4,5)P3 was used to measure its rate of disappearance during the Ca2+ releaseuptake cycle. Fig. 5 shows the Ca2+ released induced by 32Plabeled Ins(1,4,5)P3, as measured using Quin 2, under conditions similar to those used in Fig. 3A. The time scale has been expanded to show the rapidity of the Ca2+ release, which is complete within 5 s and occurs at an initial rate of about 250 pmol/mg of cell dry weight.s-l. The amount of 32P-Ins(1,4,5)P3 remaining at various times is shown superimposed on the Quin 2-Ca trace. It is clear from Fig. 5 that Ins(1,4,5)P3 degradation coincides with the reuptake of Ca2+. Although degradation of the messenger partially explains the transient nature of Ins(l,4,5)P3-induced CaZ+ release, other factors are probably also involved since the ability of increasing Ins(1,4,5)P3 concentrations to delay Ca2+ reuptake is rather limited. It was found that the half-time for reuptake of released Ca2+ did not increase further when concentrations of Ins( 1,4,5)P3 in excess of 6 PM were used (data not shown).
Ins(1,4,5)P3 did not release Ca" from purified microsomal vesicles, derived from a high speed spin (80,000 x g) of a postmitochondrial supernatant (27), and also did not release Ca2+ from purified plasma membrane vesicles4 or mitochondria

Second Messenger
Role for Inositol Trisphosphate 3081 (data not shown). However, if Ins(1,4,5)P3 is added to various subcellular fractions of rat liver, Ca2+ release can be obtained from a crude mitochondrial fraction that is heavily contaminated by endoplasmic reticulum and which is obtained at much lower centrifugal forces than used to prepare microsomal vesicles. The exact nature of this fraction and the mechanisms of Ins(l,4,5)P3-induced Ca2+ release is currently under investigation.