Formation of Inositol 1,4,5-Trisphosphate and Inositol 1,3,4-Trisphosphate from Inositol 1,3,4,5-Tetrakisphosphate and Their Pathways of Degradation in RBL-2H3 Cells*

Previous studies with antigen-stimulated rat baso-philic leukemia (RBL-2H3) cells indicated the formation of multiple isomers of each of the various cate-gories of inositol phosphates. The identities of the different isomers have been elucidated by selective labeling of [SH]inositol 1,3,4,5-tetrakisphosphate with [SaP]phosphate in the 3’- or 4‘,5’-positions and by fol-lowing the metabolism of different radiolabeled inositol phosphates in extracts of RBL-2H3 cells. We report here that inositol 1,3,4,5-tetrakisphosphate, when incubated with the membrane fraction of extracts of RBL-2H3 cells, was converted to inositol 1,4,5-tris-phosphate and inositol 1,3,4-trisphosphate. Further dephosphorylation of the inositol polyphosphates proceeded rapidly in whole extracts of cells, although the process was significantly retarded when ATP (2 mM) levels were maintained by an ATP-regenerating system. The degradation of inositol 1,4,5-trisphosphate proceeded with the sequential formation

with the authentic labeled compounds on HPLC' columns. Inositol 1,4,-bisphosphate was identified in a similar manner; the other two bisphosphate isomers were not identified although they did not co-elute with the two authentic inositol bisphosphate standards (2,4-and 4,5-bisphosphate) that were available to us at that time.
The multiple isomers of inositol bisphosphate and inositol monophosphate in intact cells could indicate degradation of the more highly phosphorylated inositol metabolites through more than one pathway. In attempting to elucidate these pathways we found unexpectedly that [3H]inositol 1,3,4,5-tetrakisphosphate, when incubated with the membrane fraction of extracts of RBL-2H3 cells in the presence of ATP, was converted to both [3H]inositol 1,4,5-trisphosphate and [3H]inositol 1,3,4-trisphosphate. We describe here the identification of these two isomers, as well as that of the previously unidentified isomers of [3H]inositol bisphosphate, by selectively labeling the 3-position of [3H]in~~itol 1,3,4,5-tetrakisphosphate with [32P]phosphate for some experiments and the 4-and 5-positions for other experiments.

MATERIALS AND METHODS
General Procedures-The RBL-2H3 cells were grown in suspension culture or as monolayers in 35-mm diameter Petri dishes as described in previous publications (2,8,9). When cells were permeabilized, cultures in Petri dishes (1 X lo6 cells/ml of rnedium/dish) were washed twice with a phosphate-free buffered salt solution (8) before permeabilizing the cells with streptolysin 0 (10) as described elsewhere? The permeabilized cells (which remained attached to the dish) were washed once more, and 1 ml of fresh buffered salt solution was added to the cultures.
Studies of the Metabolism of fH]Inositol-labeled Inositol Phosphates in Cell Extracts-Cells were harvested from suspension culture by centrifugation. They were washed twice and then disrupted by sonication to yield an extract of IO' cells/ml (whole cell extract) (2).
The medium (pH 7.2) used for these experiments contained (in The abbreviations used are: HPLC, high performance liquid chromatography; EGTA, [ethylenebis(oxyethylenenitrilo)tetraacetic acid; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid. In the figures and table the designations of the inositol phosphates are indicated by the number (subscript) of phosphate (P) groups and their position on the inositol (I) ring, e.g. inositol 1,4,5-trisphosphate is 1 (1,4,5)PZ. The asterisks indicate the position of radiolabel. in RBL-2H3 Cells millimolar): NaCl,30;potassium glutamate,70;KCl,30;MgS04,4; Hepes, 10; EGTA, 1; ATP, 2, in deionized water (2). The calculated concentration of free Ca2+ was < 5 nM. This medium was nominally referred to as "0 pM CaZ+." Alternatively, sufficient CaC12 was added to give a calculated concentration of free CaZ+ of 2 p~ where indicated. In some experiments, 3 mM phosphopyruvate (Sigma) and 2 units/ ml pyruvate kinase (Sigma, type 111 from rabbit muscle) were included as an ATP-regenerating system. The extract was centrifuged (Beckman Airfuge, 10' X g for 60 min) to obtain the supernatant (cytosol) and particulate (membrane) fractions. Aliquots (100 pl) of these extracts were incubated with the various labeled inositol phosphate (20 or 50 nCi) as indicated. The reaction was terminated at the stated times by the addition of 150 pl of ice-cold trichloroacetic acid (15% w/v) which contained a mixture of unlabeled nucleotides-adenosine, AMP, ADP, and ATP (50 p M each)-as internal standards (2). The nucleotides were omitted as necessary for the assay of residual ATP and ADP in the cell extracts. Permeabilized cells (in Petri dishes) were treated the same way with appropriate adjustments in the volumes of reagents. Insoluble material was removed by centrifugation and the supernatant fraction was washed with l ml of diethyl ether five times. The ether was removed by evaporation under a stream of nitrogen, after which 200 pl of the extract was applied to the HPLC columns.
Separation of 3H-Labeled Water-soluble Metabolites by HPLC Techniques-The procedures (2,11) utilized Whatman Partisil 10 SAX anion exchange columns (25 cm X 0.46 cm) and Beckman Model llOA two-pump system. The appearance of unlabeled nucleotides (see above) in the eluted fractions was monitored by a Spectroflow 773 (Kratos Analytical Instrument) flow detector. The absorbance readings (254 nM) for each peak were integrated (Hewlett Packard 3390A integrator) to calculate the recovery of the individual nucleotides.
The solvents for elution consisted of 0.01 M, and 1.00 M ammonium phosphate (pH 3.8). The protocol employed for the elution of the labeled metabolites from the column was a modification of one that was described previously (11). Typically, the gradients of ammonium phosphate used were (flow rate, 1 ml/min); initially 0.01-0.05 M over 30 min (to separate 3H-labeled inositol and the isomers of inositol monophosphate); then 0.22-0.25 M over 30 min (to separate the isomers of inositol bisphosphate); 0.54-0.57 M over 30 min (to separate the isomers of inositol trisphosphate); and finally an isocratic solution of 1.0 M ammonium phosphate (to elute inositol tetrakisphosphate). The stepwise increments in concentration were accomplished in 1 min. The protocol was modified slightly as required for changes in performance of the columns (11). The eluates were collected in 0.5-ml fractions and assayed for radioactivity. The efficiency of counting was determined by addition of labeled standards to fractions with low radioactivity. The ratio of 3H to 32P in each peak was determined by counting the disintegrations/minute in the three or four fractions with the highest activity for 100 min, or less if the level of activity were sufficient to achieve a P value of >0.95.
Preparation of D-my0-[2-~H]Znositol 1,3,4-Trisphsphate-This compound was prepared from D-my0-[2-~H]inositol 1,4,5-trisphosphate by phosphorylation of the 3'-position followed by dephosphorylation at the 5-position by use of the homogenate of rat brain (see above) which served as a source of both the kinase and phosphatase activities (3). Five pCi of D-my0-[2-~H]inositol 1,4,5-trisphosphate (1 Ci/mmol, Amersham Corp.) was evaporated to dryness and then dissolved in the solution of Tricine, 100 mM, and MgCl, 10 mM (800 pl). To this was added 100 pl of a solution of ATP, 100 mM, and 150 pl of the rat brain homogenate. The mixture was incubated for 45 min (37 "C) before terminating the reaction with 500 pl of trichloroacetic acid (30%). An extract of the supernatant fraction was prepared exactly as described above.
Purification of Radiolabeled Compounds-The extracts described above were applied individually to a Whatman Partisil 10 SAX column (25 X 0.46 cm) to which had been attached a "precolumn" of Whatman pellicular anion exchange resin. The radiolabeled inositol phosphates were eluted with ammonium phosphate (pH 3.8) at a flow rate of 1 ml/min (11). The radiolabeled inositol 1,3,4-trisphosphate was separated by the protocol described by Dean and Moyer (11). The radiolabeled inositol 1,3,4,5-tetrakisphosphate was separated on a gradient of 0.01-0.6 M of the ammonium phosphate buffer over 15 min to elute [32P]phosphate. The concentration of the buffer was then held at 0.6 M for 30 min to elute inositol 1,4,5-trisphosphate, inositol 1,3,4-trisphosphate, and ATP. The concentration of the buffer was then increased to 1.0 M and the elution continued for 30 min to elute the inositol tetrakisphosphate. The fractions that contained only the desired radiolabeled compound were pooled. The pooled fractions were diluted five times with distilled water before application to a column of 5 ml of AG 1-X8 (formate form, 100-200 mesh) anion exchange resin (Bio-Rad). Inorganic phosphate was eluted with 30 ml of a solution of 0.2 M ammonium formate and 0.1 M formic acid. The inositol 1,3,4-trisphosphate was eluted with 20 ml of a solution of 1.0 M ammonium formate and 0.1 M formic acid. The inositol tetrakisphosphate was eluted with 30 ml of a solution of 1.4 M ammonium formate and 0.1 M formic acid. The fractions that contained the labeled compounds were passed over 25 ml of acidwashed Dowex HCR-W2 (hydrogen form) cation exchange resin (Sigma). The eluate was then frozen and dried by lyophilization. The various purified radiolabeled inositol compounds were dissolved in distilled water and stored at -20 "C.
The two minor peaks of the [2-3H]inositol bisphosphates that were generated in antigen-stimulated intact-cells (see Ref. 2) had retention times identical to those of the inositol 1,3-bisphosphate and inositol 3,4-bisphosphate described above. The same two peaks were obtained when permeabilized cells were incubated with [2-3H]inositol 1,3,4,5-tetrakisphosphate but not with [2-3H]inositol 1,4,5-trisphosphate (data not shown). As the peaks had similar retention times in all cases, we presumed that the inositol bisphosphates that are generated from either endogenous or exogenous substrates are the same.
Kinetics and Pathways of Metabolism of the Inositol Polyphosphates in RBL-2H3 Cell Extracts-The relative proportions of the 1,3,4-trisphosphate and the 1,4,5-trisphosphate isomers that were generated from the inositol tetrakisphosphate varied (e.g. Fig. 1B uersus IC). The studies, described below, indicated that such variation was attributable to kinetic considerations, and that factors such as the presence or absence of ATP or Ca2+ influenced the time course and pathways of metabolism of the inositol polyphosphates.
As was observed in our preliminary experiments, incubation Calculated from the ratio [3'P]/[3H] (Table I) [4,5-32P]inositol 1,4,5-trisphosphate (from which the radiolabeled inositol 1,3,4,5-tetrakisphosphate was made) has been reported by others (7,20). (12) and the concentration of ammonium phosphate used to elute them are shown in A . For B and C the double-labeled inositol tetrakisphosphates were incubated with the membrane fraction of extract of cells (10' cells/ml) and ATP (2 mM) for 5 min at 37 "C. The concentration of the tetrakisphosphate used was 0.1 and 0.2 PM for B and C , respectively. The identification of the different inositol phosphate is discussed in the text. The data from this and an additional experiment are summarized in Table I.
In contrast to the above, the amounts of inositol 1,4,5trisphosphate generated in the cytosolic fraction of RBL-2H3 cells (e.g. Fig. 2) accounted for, at most, 7% (range 4-14%) of the total 3H label in the three experiments. We did not determine whether these small amounts were generated by a soluble 3-phosphatase activity or by contaminating membrane enzyme activity.
The appearance of the inositol 1,4,5-trisphosphate occurred before the levels of ATP had declined substantially, although in the membrane fractions the levels of ATP declined by 50% over 5 min (upper panels, Fig. 2). Other experiments with whole cell extracts showed, however, that the rates of dephosphorylation of either the inositol tetrakisphosphate or its metabolites were reduced in the presence of ATP especially when ATP levels were maintained by an ATP-regenerating system. In the absence of ATP, the predominant pathway of dephosphorylation of inositol 1,3,4,5-tetrakisphosphate was through inositol 1,3,4-trisphosphate, inositol 1,3-bisphosphate, inositol 3,4-bisphosphate, inositol l-monophosphate,4 and inositol. Small amounts of inositol 1,4-bisphosphate and inositol 1-monophosphate were also generated. In the presence of ATP, the predominant pathway was inositol 1,4,5trisphosphate, inositol 1,4-bisphosphate, inositol 4-monophosphate, and inositol. Studies with the 3H-labeled inositol trisphosphates showed, in addition, that in the absence of ATP the formation of inositol 1,3-bisphosphate from inositol 1,3,4-trisphosphate was a relatively minor pathway and that inositol 1,4-bisphosphate was largely, but not exclusively, degraded to inositol 4-monophosphate.

DISCUSSION
In this study we report the first example of the dephosphorylation of inositol 1,3,4,5-tetrakisphosphate via the removal of the 3"phosphate to yield inositol 1,4,5-trisphosphate and of the 5"phosphate to yield inositol 1,3,4-trisphosphate. The specific loss of the 32P label in the 3'-position in the inositol 1,4,5-trisphosphate peak excludes the misidentification of the product as either inositol 3,4,5-trisphosphate or inositol 1,3,5trisphosphate. The 3'-phosphomonoesterase activity appears to be largely, possibly exclusively, a membrane enzyme, whereas the 5'-phosphomonoesterase activity is present in both the membrane and soluble fractions of the cell (Fig. 2). The absence of a 3"kinase activity in the membrane fraction ' Either D-or L-myoinositol 1-monophosphate.
of RBL-2H3 cells (2) suggests also that the formation of inositol 1,4,5-trisphosphate is not attributable to a reversal of the 3'-kinase reaction due to a low ratio of ATP to ADP. The presence of a 3'-phosphomonoesterase was examined subsequently in another cell line (GH3), but the amounts of inositol 1,4,5-trisphosphate formed were small (-2% of the inositol trisphosphate fraction) in whole cell extracts (12), as indeed was the case for whole extracts of RBL-2H3 cells (3-6% of total 3H).
The studies illustrate the difficulties in extrapolating findings with cell extracts, or subfractions thereof, to those with intact cells. The studies with cell extracts indicate, at best, the potential pathways for the metabolism of the various inositol phosphates. The formation of inositol 1,4,5-trisphosphate from inositol 1,3,4,5-tetrakisphosphate might be dismissed as trivial in experiments with whole cell extracts, whereas the experiments with cell membranes could suggest that the 3'-phosphomonoesterase activity helps maintain, in the vicinity of the membrane, an equilibrium between inositol 1,3,4,5-tetrakisphosphate and inositol 1,4,5-trisphosphate, both of which, as will be discussed later, might serve as comessengers for influx of ca'+ across the plasma membrane. However, the promotion of additional phosphorylation of the inositol phosphates and the suppression of the various dephosphorylation reactions by ATP (Ref. 14 and this paper) makes it difficult to extrapolate what the situation might be in the intact cell. The levels of ATP in RBL-2H3 cells (9) are much higher (-10 mM) than those used here (2 mM), although the ratio of ATP to ADP (-4:l)' in antigen-stimulated intact cells is similar to that observed in extracts with an ATPregenerating system (5:l).
While this work was in progress, other groups identified in GH, cells and tissue preparations the same novel products of inositol 1,3,4-trisphosphate, namely inositol 1,3-bisphosphate and inositol 3,4-bisphosphate (14-17) that we had observed previously (2) and subsequently identified (this paper) in RBL-2H3 cells. The identity of the two metabolites in RBL-2H3 cells was established deductively (e.g. retention of the 3'-[32P]phosphate label; Fig. 2B), and they are, in all probability, identical to those described by Irvine and colleagues (15). The phosphorylation of inositol 1,3,4-trisphosphate in cultured adrenal glomerulosa cells (18) and rat liver (14) to form inositol 1,3,4,6-tetrakisphosphate has been reported recently. Inositol 1,3,4-trisphosphate is transformed to a more polar metabolite in the presence of ATP in RBL-2H3 cells but this metabolite does not account for a third inositol tetrakisphosphate isomer that appears in small amounts in stimulated RBL-2H3 cells5 and which could be a candidate for the L-myoinositol 1,4,5,6-tetrakisphosphate (19).
Although the pathways for the metabolism of the inositol phosphates in RBL-2H3 cells are, in general, similar to those previously identified in other types of cells (e.g. 3-7, 15, 17, 20) there may be differences. Others have shown that, at least in rat parotid gland (21) and hepatocytes (22,23), the hydrolysis of inositol 1,3,4,5-tetrakisphosphate occurs exclusively through removal of the 5"phosphate. Also in contrast to extracts of calf brain in which inositol, 1,4-bisphosphate is converted exclusively to inositol 4-monophosphate (20), we observed some conversion to inositol 1-monophosphate. The overwhelming predominance of inositol 1-monophosphate in antigen-stimulated RBL-2H3 cells (2), however, suggests that this monophosphate is derived directly from phosphatidylinositol in concurrence with the conclusion of Majerus and colleagues (20). It should be noted that stimulation of RBL-2H3 cells requires the aggregation of plasma membrane receptors for immunoglobulin E (1) to induce hydrolysis of membrane inositol phospholipids, an increase in concentration of cytosol Ca", and the secretion of histamine (2,8,24). These reactions are dependent on the number of receptors clustered (8) and on the type of receptor cross-linking agent used (2). The mechanism of activation of RBL-2H3 cells (and mast cells) may, therefore, differ in some respects from that in which cell activation requires the binding of a single hormone to a receptor. A unique property of the RBL-2H3 cell line is that, in the absence of external Ca'+, the stimulatory signals are weak or undetectable (2) and external Ca2+ is required for amplification of the stimulatory responses and for secretion to occur (2).
The question has been raised as to whether the formation of inositol 1,3,4,5-tetrakisphosphate by a Ca2+-modulated 3'kinase (5), which appears to be operative in RBL-2H3 cells,' allows inactivation of inositol 1,4,5-trisphosphate or provides a second species of messenger. Recent studies by microinjection into sea urchin eggs suggests that inositol 1,3,4,5tetrakisphosphate in coordination with the action of inositol 1,4,5-trisphosphate promotes influx of Ca2+ ions across the plasma membrane (25). If the same situation exists in RBL-2H3 cells, especially because of the substantial dependence of RBL-2H3 cells on external Ca2+, the 3"kinase (3,5) and the 3'-phosphomonoesterase (this paper) enzyme activities in RBL-2H3 cells could conceivably regulate the balance between both messengers. As both substances are substrates for the inositol 5'-phosphomonoesterase, whose activity is modulated by protein kinase C activity (26), one enzyme may inactivate simultaneously the two messengers. An interesting point emerging from the present work is this: if the presence of a 3'-phosphomonoesterase in membranes of RBL-2H3 cells is related to their reliance on extracellular Ca'+, is the enzyme less prominent in cell lines that can generate a Ca'+ signal by mobilization of intracellular Ca"? If so, information as to whether or not the enzyme is restricted to the plasma membrane would be instructive.
Since submission of this manuscript, Doughney and coworkers (27) have reported that human erythrocyte membranes metabolize inositol 1,3,4,5-tetrakisphosphate to inositol 1,4,5-trisphosphate, as well as inositol 1,3,4-trisphosphate, in the absence of M8+. These authors suggest, as we have done, that such a reaction in hormonally sensitive cells may provide a mechanism for the maintenance of constant concentrations of inositol 1,4,5-trisphosphate and inositol 1,3,4,5tetrakisphosphate and may thus be important for stimulation of Ca2+ entry after Ca" mobilization.