The Interconversion of Inositol 1,3,4,5,6-Pentakisphosphate and Inositol Tetrakisphosphates in AR4-2J Cells*

Data from several cell types have indicated that ac- tivation of hormone receptors promotes the metabolism of inositol 1,3,4,5,6-pentakisphosphate (IPS) to inositol 3,4,5,6-tetrakisphosphate ((3,4,5,6)IP4). However, to date, metabolism of IPS by cell-free preparations has resulted in the formation of only inositol 1,4,5,6-tet- rakisphosphate ((1,4,5,6)IP4). Thus, the metabolic relationships of IPS with various inositol tetrakisphosphate (IP,) isomers have been investigated in both intact cells and cell homogenates of the rat pancreatoma cell line, AR4-2J. The steady-state concentration of IPS was estimated to be 65 pM, while the combined concentration of (3,4,5,6)IP4 and (1,4,5,6)IP4 was ap- proximately 1.0 p ~ . AR4-2J cell homogenates converted (1,3,4,6)IP4, (3,4,5,6)IP4, and (1,4,5,6)IP4 to IPS. (1,4,5,6)IP4 previously has not been demonstrated to be a precursor of IPS. To alter steady-state levels of

catabolic removal of the (1,4,5)IP3 signal and subsequent salvage of inositol for reassimilation into inositol lipids have been investigated extensively (2). However, the recent recognition of other inositol phosphates, apparently not directly involved in regulation of intracellular free Ca2+, indicates that these inositol phosphates may be involved with other regulatory events in cell physiology. For example, IPS and IPS are the most abundant inositol phosphates found in mammalian cells and have been suggested to act as extracellular signalling molecules (3)(4)(5)(6)(7) and regulators of aldolase function (8). In spite of these apparent effects, the regulation of intracellular IPS and IPS concentrations in intact cells is poorly understood.
Previous studies have indicated that regulation of IPS and IPS metabolism may occur in part through phosphorylationdephosphorylation cycles (9,lO). In particular, the hypothesis of an agonist-sensitive IPS 1-phosphatase/(3,4,5,6)IP4 l-kinase phosphorylation-dephosphorylation cycle (9) has developed from experiments with intact AR4-2J cells which demonstrated a close metabolic relationship between (3,4,5,6)IP4 and IPS (9) and an apparent agonist-stimulated conversion of IPS to (3,4,5,6)IP4 (9,(11)(12)(13). The agonist sensitivity of this proposed cycle is intriguing and suggests a physiologically significant function. Nevertheless, these proposals would be considerably strengthened if the elements of the putative cycle could be demonstrated in cell-free systems. Although a (3,4,5,6)IP4 1-kinase has been observed in cell homogenates (14), there is no direct evidence in cell-free systems for the existence of an IPS 1-phosphatase. In fact, previous studies have reported that cell-free preparations only metabolize IPS to (1,4,5,6)IP4 (15)(16)(17). Therefore, to resolve this observed conflict between IP5 metabolism by intact cells and cell homogenates, the synthesis and catabolism of IPS have been investigated in intact AR4-2J cells and AR4-2J cell homogenates. The results indicate that, in unstimulated AR4-2J cells, IPS comprises a relatively stable, slowly turning over inositol phosphate pool. Yet under appropriate experimental conditions, interconversion of IPS with both (1,4,5,6)IP4 and (3,4,5,6)IP4 could be demonstrated in both intact and broken cell preparations. These studies reveal novel pathways of inositol polyphosphate phosphorylation and dephosphorylation and indicate that hormone-regulated interconversion of IPS and various IP4 isomers may be more complex than previously envisioned.
Cell Culture-The AR4-2J pancreatoma cell line was kindly provided by Dr. C. Logsdon, University of California, San Francisco, and was grown in DMEM containing high glucose (4.5 g/liter), 10% FBS, L-glutamine (2 mM), penicillin/streptomycin (500 units/ml and 500 gg/ml), and 10% conditioned media. Cells were incubated in 6-well tissue culture plates with [3H]myo-inositol (50 pCi/ml) in 2 ml of the above medium with the exception that the FBS was replaced with 10% dialyzed FBS. Cells were cultured with [3H]myo-inositol for 4 days unless otherwise indicated. Cell viability was estimated by trypan blue exclusion.
For structural analysis, 0.5-0.8-ml fractions eluted from the column were collected, and aliquots were taken to identify peaks of interest. Individual peaks were pooled, and IP, isomers were desalted with ammonium bicarbonate (21) while IPS was desalted using either HC1 (21) or triethylammonium bicarbonate (22). fHllnosito1 Lipid Measurements-Following removal of the acidic supernatant from the tissue culture plates, the remaining precipitate was scraped from the plates, and the [3H]inositol lipids were extracted according to the method of Schacht (23).
Metabolism of Inositol Phosphates by Intact AR4-W Cells or Hornogenates-[3H]Inositolphosphates were added to either intact AR4-25 cells, homogenates, or cell fractions (0.5-1.0 mg of cellular protein/ ml) and were incubated at 37 "C in an intracellular-like medium (final pH 7.2) with the following composition (mM): NaC1, 20.0; KC1, 100; MgSO,, 2.0; HEPES, 20.0; EGTA, 1.0; and total Ca2+ to result in 150 nM free Ca". When indicated, an ATP regenerating system with the following composition was added 5 mM MgATP, 10 mM phosphocreatine, 16 units/ml creatine phosphokinase. AR4-2J cell homogenates were prepared by Dounce homogenization. The unbroken cells and nuclei were removed from the homogenate by low speed centrifugation (200 X g for 5 min). For some experiments, soluble and particulate fractions were separated from the homogenate by high speed centrifugation (100,000 X g for 1 h). Metabolites of ['HIIP, were separated by their elution on an Adsorbospbere SAX column and on-line monitoring of the [3H]inositol phosphates (see above). Phosphorylation of different ['HIIP, isomers was monitored by chromatography on Bio-Rad AgX-1 resin as previously described (21).
IP5 was structurally characterized by two independent HPLC techniques which separate individual IPS isomers: 1) the method of Phillippy and Bland (28) using a Dionex AS7 column eluted with nitric acid, and 2) the method of Stephens et al. (25) using a Partisphere 5 p-SAX column eluted with diammonium hydrogen phosphate (pH 3.8 with phosphoric acid).
IPS did not attain steady-state incorporation of radioactivity after 4 days of incubation with [3H]myo-inositol, yet at that time there was nearly 9 times more [3H]IPs than [3H] (1,4,5)IP, resulting in a minimal concentration of 17 p M for IPS. These data indicate that in unstimulated cells IP, and IPS constitute large, slowly turning over pools of inositol phosphates which are kinetically far removed from the polyphosphoinositides. Similar conclusions have been reached previously by using different experimental protocols (9, 30).
IPS Metabolism in Intact AR4-2J Cells-Antimycin A rapidly reduces cellular ATP levels and greatly reduces the levels of phosphatidylinositol 4-phosphate and phosphatidylinositol 4,5-bisphosphate, thereby demonstrating the rapid turnover rate of the polyphosphoinositides (32, 33). With these results in mind, the effect of antimycin A treatment on IPS, IPB, and IP, isomers has been studied in intact AR4-2J cells. Since multiple IP4 isomers and IPS may be associated with IPS through phosphorylation-dephosphorylation cycles (9, lo), lowering ATP levels should perturb the phosphorylation pathways within these cycles and reveal precursors and pathways of IPS metabolism.  (1,4,5,6)IP, (panel C) were incubated with AR4-2J cell homogenates in the presence of an ATP regenerating system as described under "Experimental Procedures." Inositol phosphates were separated on Bio-Rad AgX-1 resin (21), and the amount of [3H]IP5 formed is shown as a percent of total counts added. Less than 1% IP' accumulated under these conditions.
Shown are the means of triplicate determinations from one experiment of three.
Antimycin A treatment of AR4-2J cells reduces ATP levels by >90% within 10 min (33). As shown in Fig. 3, antimycin A treatment for 1 h resulted in a 4-fold increase in a single IP4 peak, D/L- (3,4,5,6)IP4 (peak C). This increase was significant within 10 min after antimycin A treatment as compared to zero time basal levels and increased to greater than 6-fold over basal levels by 2 h (Fig. 4). We believe that this increase in D/L- (3,4,5,6)IP4 from 1.0 p M (see above) to 6 pM was due to an antimycin A-mediated decrease in the rate of flux from IP, to IPS (see "Discussion"). A corresponding decrease in IPS levels (ie. 5 p~) was observed but, due to the high level of IPS, this proportionately small decrease was not statistically significant.
ation (data not shown). Thus, by decreasing ATP levels, antimycin A induces a similar increase in the levels of both (1,4,5,6)IP4 and (3,4,5,6)IP4 and indicates that AR4-2J cells are capable of metabolizing IP6 to both (1,4,5,6)IP4 and (3,4,5,6)IP4. Structural analysis of the IP5 peak following up to 2 h of antimycin A treatment demonstrated that the primary IPS isomer present was (1,3,4,5,6)IP5 (>97%) and that there was no change in the small amount of D/L- (1,2,4,5,6)IP5 present (data not shown). The lack of change in IPS levels may indicate that futile cycles between IP5 and IPS, which have been demonstrated in Dictyostelium (lo), do not exist or proceed very slowly in mammalian cells. Therefore, the rapid depletion of the polyphosphoinositides (32, 33), compared to the minimal effect on levels of IPS and IPS following antimycin A treatment, further supports our above conclusion that IPS and IPS exist as pools of inositol phosphates which are independent and kinetically distinguishable from the inositol lipids.
Antimycin A treatment induces two additional changes in the observed basal pattern of inositol phosphates in AR4-2J cells. Firstly, in Fig. 3 there is a ,H-labeled peak present in basal AR4-2J cells that is more polar than IPS. A peak eluting after IPS also has been noted in Dictyostelium (34). While this peak is depleted by 1 h of antimycin A treatment (Fig. 3), other experiments (not shown) demonstrated that it had an approximate half-life of 5 min. Thus, this compound apparently has a rapid turnover rate. Secondly, after 1 h of antimycin A treatment, an additional IP3 peak occasionally is observed ( 4 0 % of HPLC separations) that elutes less than 1 min after (1,4,5)IPa. Our failure to reproducibly observe this IP, isomer is most likely a result of the inability of our HPLC system to consistently separate this peak from ( 1,4,5)IP3. This is supported by our previous observation of a similarly eluting IPa peak in basal AR4-2J cells that was revealed only after removal of (1,4,5)IP3 by treatment with purified 5-phosphatase (35). Structural characterization of this peak would be useful but presently is hampered by its low levels.
PHJIP, Dephosphorylation by AR4-2J Cell Homogenates-The rate of IPS hydrolysis by cell homogenates was compared to that by 100,000 x g particulate and soluble fractions. The relatively high level of IPS in AR4-2J cells could contribute significant and variable amounts of IPS to our incubations (up to 0.4 pM)'. Therefore, to minimize this potential variability in IPS amounts, nonradioactive IPS was added to the ['HIIPS to result in a final concentration of 10 PM. This maneuver should also reduce the ability of endogenous IPS (0.1 p~)~ to competitively inhibit IPS dephosphorylation (36).
Further experiments to characterize this IPS 3-phosphatase a The protein concentration used (1.0 mg/ml) results in an approximate 166-fold dilution of the cytoplasm (volume of intact AR4-2J cells is approximately 6 pl/mg of protein (29)); thus, the contributions of endogenous IP, and IPe would be approximately 0.4 p~ and 0.1 p~, respectively. activity revealed that it was not inhibited by Li+ (20 mM) or by removal of M e , and it was only activated by Ca2+ at concentrations above 500 p~ (data not shown). The dephosphorylation rate of the IPS 3-phosphatase was pH-dependent, was nearly tripled under alkaline conditions (pH 8.5), and was diminished by one-half under acidic conditions (pH 5.5) as compared to neutral conditions (pH 7.2). Even under these different pH conditions, the only IP4 detected was the (1,4,5,6)IP4 isomer (data not shown).
Effect of ATP on the IP4 Products of IPS Dephosphorylation-In these experiments, [3H]IP5 was incubated with AR4-25 cell homogenates in the presence or absence of 5 mM MgATP. Since no nonradioactive IPS was added in these incubations, the starting concentration of IPS was approximately 0.4 p~ due to the contribution of endogenous IPS from the cell homogenates.' Following various times of incubation, structural analysis was done on any D/L- (3,4,5,6)IP4 formed.
Determination of the involvement of these three IP, isomers in phosphorylation-dephosphorylation cycles in intact cells was approached by treatment with antimycin A. Lowering of cellular ATP levels to inhibit phosphorylation and thus expose dephosphorylation reactions resulted in an increase in both (3,4,5,6)IP4 and (1,4,5,6)IP4. Since no agonist activation is involved, these changes presumably reflect processes involved in maintaining equilibrium levels of IPS and D/L- (3,4,5,6)IP4 in these cells. This indicates that both the previously proposed agonist-sensitive (3,4,5,6)IP4-IPs cycle (9) and a novel (1,4,5,6)IP4-IPS cycle may exist in intact AR4-2J cells.
In addition to the phosphorylation of both (3,4,5,6)IP4 and (1,4,5,6)IP4 to IPS, it was also possible to demonstrate for the first time conversion of IPS to both of these isomers by broken cell preparations. However, (3,4,5,6)IP4 could only be produced in the presence of ATP. This requirement for ATP provides an explanation for the failure of earlier studies to detect this reaction (15,16). Mattingly et al. (17) also have studied IPS dephosphorylation in the presence of ATP, but under their assay conditions only (1,4,5,6)IP4 was detected. A potential explanation for the difference in their results and ours is that their experiments were performed with cytosolic fractions, and our experiments were performed with homogenates. This implies that the putative 1-phosphatase may be exclusively a particulate enzyme or alternatively that there may be a factor present in particulate fractions that is necessary for the formation of (3,4,5,6)IP4. Additionally, the utilization of a different cell type in our experiments may account for our different results.
The ability of different physiological changes within the cell (v-src transformation (17), chronic agonist stimulation (9,12,13), and cell differentiation (38)) to perturb either IPS concentrations or individual IP4-IP5 cycles indicates that, in addition to the more traditional agonist-sensitive inositol phosphates ((1,4,5)1PS and its metabolites), IPS and its metabolites also may be involved in regulation of physiological processes. Theoretically, flux through IP4-IP5 cycles may be differentially regulated by a variety of agonists in order to elicit individual cellular responses (38). Thus, the complex regulation of IPS metabolism may be analogous to the inositol lipid pool which can act as a substrate for phospholipase C to form (1,4,5)IP3 and also can be phosphorylated by a 3-kinase to form a less well characterized pool of inositol lipids that may be important in growth factor action or cell transformation (39). In addition to the well characterized, rapid effects of (1,4,5)IP~ on cellular Ca2+ metabolism, it seems likely that other important, albeit more subtle, cellular functions may be controlled through the regulated metabolism of IPS to specific Inositol Pentakisphosphate inositol tetrakisphosphate isomers.