Inositol 1,3,4-Trisphosphate Acts in Vivo as a Specific Regulator of Cellular Signaling by Inositol 3,4,5,6-Tetrakisphosphate*

Ca2+-activated Cl− channels are inhibited by inositol 3,4,5,6-tetrakisphosphate (Ins(3,4,5,6)P4) (Xie, W., Kaetzel, M. A., Bruzik, K. S., Dedman, J. R., Shears, S. B., and Nelson, D. J. (1996) J. Biol. Chem. 271, 14092–14097), a novel second messenger that is formed after stimulus-dependent activation of phospholipase C (PLC). In this study, we show that inositol 1,3,4-trisphosphate (Ins(1,3,4)P3) is the specific signal that ties increased cellular levels of Ins(3,4,5,6)P4 to changes in PLC activity. We first demonstrated that Ins(1,3,4)P3 inhibited Ins(3,4,5,6)P4 1-kinase activity that was either (i) in lysates of AR4–2J pancreatoma cells or (ii) purified 22,500-fold (yield = 13%) from bovine aorta. Next, we incubated [3H]inositol-labeled AR4–2J cells with cell permeant and non-radiolabeled 2,5,6-tri-O-butyryl-myo-inositol 1,3,4-trisphosphate-hexakis(acetoxymethyl) ester. This treatment increased cellular levels of Ins(1,3,4)P3 2.7-fold, while [3H]Ins(3,4,5,6)P4 levels increased 2-fold; there were no changes to levels of other 3H-labeled inositol phosphates. This experiment provides the first direct evidence that levels of Ins(3,4,5,6)P4 are regulated by Ins(1,3,4)P3 in vivo, independently of Ins(1,3,4)P3 being metabolized to Ins(3,4,5,6)P4. In addition, we found that the Ins(1,3,4)P3 metabolites, namely Ins(1,3)P2 and Ins(3,4)P2, were >100-fold weaker inhibitors of the 1-kinase compared with Ins(1,3,4)P3 itself (IC50 = 0.17 μm). This result shows that dephosphorylation of Ins(1,3,4)P3 in vivo is an efficient mechanism to “switch-off” the cellular regulation of Ins(3,4,5,6)P4 levels that comes from Ins(1,3,4)P3-mediated inhibition of the 1-kinase. We also found that Ins(1,3,6)P3 and Ins(1,4,6)P3 were poor inhibitors of the 1-kinase (IC50 = 17 and >30 μm, respectively). The non-physiological trisphosphates,d/l-Ins(1,2,4)P3, inhibited 1-kinase relatively potently (IC50 = 0.7 μm), thereby suggesting a new strategy for the rational design of therapeutically useful kinase inhibitors. Overall, our data provide new information to support the idea that Ins(1,3,4)P3 acts in an important signaling cascade.

and Ins(1,3,4,5)P 4 ( Fig. 1) act in a co-ordinated manner as mediators of stimulus-dependent Ca 2ϩ mobilization (1,2). This has naturally led us to consider that the 5-phosphatases that degrade Ins(1,4,5)P 3 and Ins(1,3,4,5)P 4 (3) are signaling "offswitches." This in turn has created the impression that the pathway by which these two inositol phosphates are dephosphorylated serves only as a conduit that replenishes the free inositol pool. In contrast, we have recently suggested that one of these downstream products, namely Ins(1,3,4)P 3 , should be viewed in an important cell-signaling context (4). This new hypothesis comes from the observation that a rat hepatic Ins(3,4,5,6)P 4 1-kinase was inhibited in vitro by Ins(1,3,4)P 3 (4,5). The reason that this effect of Ins(1,3,4)P 3 upon Ins(3,4,5,6)P 4 metabolism is of such interest is that Ins(3,4,5,6)P 4 is an inhibitor of the conductance of the calciumactivated Cl Ϫ channels in the plasma membrane (6 -9). These ion channels make important contributions to salt and fluid secretion, and in addition they may participate in osmoregulation, pH balance, and smooth muscle excitability (10 -13).
The cellular accumulation of Ins(3,4,5,6)P 4 is known to correlate well with receptor-dependent changes in PLC activity, but the molecular mechanisms that link these two events have not been fully elucidated (14). Our current hypothesis (15,16) is that cellular levels of Ins(3,4,5,6)P 4 depend upon a dynamic balance between two competing enzyme activities acting in a closed substrate cycle: Ins(1,3,4,5,6)P 5 1-phosphatase and Ins(3,4,5,6)P 4 1-kinase (Fig. 1). The poise of this cycle is proposed to be regulated in such a manner that it can shift in favor of Ins(3,4,5,6)P 4 accumulation whenever PLC is activated, perhaps through inhibition of the Ins(3,4,5,6)P 4 1-kinase by Ins(1,3,4)P 3 (Fig. 1). However, to date such inhibition has only been observed in studies with the purified rat hepatic kinase (4,5). No direct evidence has previously been published that indicates Ins(1,3,4)P 3 can regulate Ins(3,4,5,6)P 4 1-kinase activity in intact cells; it was a goal of the current study to explore this issue.
Another feature of an effective signal transduction process relates to its specificity. If the biological effects of a signaling compound cannot be imitated by its products and precursors, this provides sensitivity in the signaling "on" and "off" switches. In the case of signaling by Ins(1,3,4)P 3 , the "onswitch" is dephosphorylation of Ins(1,3,4,5)P 4 (3). This process is particularly sensitive, as Ins(1,3,4,5)P 4 is a 290-fold weaker inhibitor of the 1-kinase than is Ins(1,3,4)P 3 (4). We have now turned our attention to considering how effective is the dephosphorylation of Ins(1,3,4)P 3 as a signaling off-switch. In vivo, both 4-and 1-phosphatases actively degrade Ins(1,3,4)P 3 to Ins(1,3)P 2 and Ins(3,4)P 2 , respectively (24 -26). We have therefore determined the potency with which these bisphosphate degradation products inhibit the 1-kinase.
Bombesin, bovine serum albumin, phosphocreatine, phosphocreatine kinase, heparin agarose resin (type II and IIIs), and protease inhibitors were purchased from Sigma. The calmodulin-dependent protein kinase (CaM KII) was obtained from New England Biolabs. Protein kinases A and C, and their assay kits (SpinZyme Format), were the products of Pierce. The UNO Q12 anion exchange column was acquired from Bio-Rad Laboratories. Polyethylene glycol 4000 was purchased from Fluka. Frozen bovine aorta were purchased from Pel-Freez Biological.
Assay of Ins (3,4,5,6)P 4 1-Kinase-The 1-kinase activity was assayed as described before (4). Briefly, 10 -20 l of enzyme was incubated at 37°C in a final volume of 100 l containing about 4000 dpm M ammonium formate, and 0.1 M formic acid. The quenched reactions were diluted to 10 ml with deionized water, and chromatographed on Bio-Rad gravity-fed columns using AG 1-X8 ion exchange resin.
For some assays, the 1-kinase was preincubated at 30°C for 10 min with (a) 125 units of the catalytic subunit of protein kinase A, (b) 0.2 unit of protein kinase C, (c) 600 units of calmodulin, or (d) 500 units of CaM KII, preactivated with calmodulin/Ca 2ϩ (New England Biolabs). The protein kinases used in these experiments were all shown to be active in control experiments (assay kits for protein kinases A and C were supplied by Pierce; the CaM KII was checked using a kit purchased from Upstate Biochemicals).
The 1-kinase was also used as a diagnostic tool to verify the nature of HPLC-purified [ 3 H]Ins(3,4,5,6)P 4 . In these incubations, 45 l of purified 1-kinase was added to 225 l of medium containing 67 mM HEPES (pH 8.0 with KOH), 0.7 mM EDTA, 8.7 mM MgSO 4 , 6.7 mM ATP, 13.3 mM phosphocreatine, 1.33 M InsP 6 , and 6 Sigma units of phosphocreatine kinase. Then, 30 l of the appropriate HPLC fraction was added (which brought the final pH to approximately 6.5). Reactions (at 37°C) were allowed to proceed to completion (over a 3-h period), and then the amount of [ 3 H]InsP 5 formed was determined using gravity-fed ion-exchange columns, as described above.
Purification of Ins (3,4,5,6)P 4 1-Kinase-Frozen bovine aortas were thawed on ice, the attached fat was removed, and then the aorta were pulverized in a meat grinder. In a typical preparation, 300 -350 g of ground aortas were homogenized in two volumes of 50 mM bis-Tris (pH 7.0), 1 mM EGTA, 1 g/ml leupeptin, 1 g/ml pepstatin A, and 1 mM phenylmethylsulfonyl fluoride in a tissue blender.
The homogenate was filtered through four layers of cheesecloth, and a 10 -30% (w/v) polyethylene glycol 4000 precipitate was prepared. The resultant pellet was resuspended in 100 ml of Buffer A containing 50 mM bis-Tris (pH 7.0), 1 mM EGTA, 1 g/ml leupeptin, and 1 g/ml pepstatin A. The suspension was filtered and loaded at a flow rate of 1 ml/min onto a heparin-agarose type II column (3.2 ϫ 24 cm). After washing with 300 ml of Buffer A at flow rate of 1.5 ml/min, the bound protein was eluted with a linear gradient of 0 -30 mM of sodium pyrophosphate in Buffer A.
The peak fractions of enzyme activity eluted from the heparin column were pooled, then frozen and stored at Ϫ70°C. Either two or three preparations were subsequently thawed and combined, dialyzed against 2 liters of 25 mM bis-Tris (pH 7.0) at 4°C for 3 h, and loaded onto a UNO Q12 anion exchange column (1.5 ϫ 6.8 cm), which was pre-equilibrated with 100 ml of Buffer A. A constant flow rate of 0.5 ml/min was maintained throughout the chromatography. After washing with 60 ml of Buffer A, the bound protein was eluted with a linear gradient of Buffer A plus 0 -300 mM NaCl, followed by 60 ml of Buffer A plus1 M NaCl.
Peak fractions of enzyme activity eluted from the UNO Q12 column were pooled, dialyzed against 2 liters of 25 mM bis-Tris (pH 7.0) at 4°C for 3 h, and loaded on to heparin-agarose type IIIs (1.1 ϫ 13.5 cm), which was pre-equilibrated with 50 ml of Buffer A. A constant flow rate of 0.5 ml/min was maintained throughout. After washing with 60 ml of Buffer A, the bound protein was eluted with a linear gradient of 0 -300 mM NaCl in Buffer A, followed by 60 ml of 1 M NaCl in Buffer A.
The protein concentration of the 1-kinase preparation was determined using Bio-Rad Protein Assay Dye Reagent with bovine serum albumin as standard. Final enzyme preparations were stored in 10% glycerol plus 1 mg/ml bovine serum albumin at Ϫ70°C.
Gel Filtration-A 1-ml aliquot of a resuspension of a 10 -30% PEG precipitation was loaded at a flow rate of 0.25 ml/min to Sephacryl S100 column (2.0 ϫ 86 cm), which was pre-equilibrated with 600 ml of bis-Tris buffer containing 50 mM bis-Tris (pH 7.0), 1 mM EGTA, 1 g/ml leupeptin, 1 g/ml pepstatin A, and 100 mM NaCl. The protein was then chromatographed using the same buffer at a constant 0.25 ml/min flow rate. Fractions (5ml) were collected and assayed for enzyme activity. The column was calibrated under the exactly same conditions using bovine serum albumin, chicken ovalbumin, equine myoglobin, and vitamin B-12.
Culturing and Incubation of AR4 -2J Cells-The AR4 -2J pancreatoma cells were cultured in Dulbecco's modified Eagle's medium containing 4.5 g/liter glucose, 10% fetal bovine serum, 2 mM glutamine, 500 units/ml penicillin, and 500 g/ml streptomycin, with 10% conditioned medium, and harvested by brief trypsinization. Either 2.0 ϫ 10 5 or 1.2 ϫ 10 6 cells were seeded in 24-well or 6-well tissue culture plates, respectively. Cells were labeled with 75-150 Ci/ml [ 3 H]myo-inositol for 4 days (medium was replaced on the 3rd day) in 700 l (for 24-well plates) or 3 ml (for 6-well plates) of the above culture medium. After completion of the labeling protocol, the culture medium was aspirated and the cells were washed twice with Krebs/Ringer/HEPES solution (15). Cells were then incubated in 300 l (for 24-well) or 1 ml (for 6-well) of Krebs/Ringer/HEPES solution for 2 h. Then 20 mM LiCl was added, and 20 min later cells were treated for the indicated time with (i) a cell-permeant inositol phosphate, (ii) vehicle, or (iii) bombesin.
Cells were quenched and neutralized, and the inositol phosphates were separated by HPLC as described elsewhere (36). Radioactivity was either counted on-line, using a Radiomatic Flo-1, or recovered in 1-ml FIG. 2. Purification of Ins(3,4,5,6)P 4 1-kinase. The 1-kinase was purified as described under "Experimental Procedures" by subjecting a 10 -30% PEG precipitate to Heparin II-agarose affinity chromatography (panel A), followed by UNO Q12 anion exchange chromatography (panel B) and Heparin IIIS affinity chromatography (panel C). Data are representative of three experiments.

RESULTS
Purification and Properties of the Ins(3,4,5,6)P 4 1-Kinase from Bovine Aorta-In our earlier study with the Ins(3,4,5,6)P 4 1-kinase in rat liver, the enzyme was found to be extremely labile, and we were only able to elicit a 1600-fold purification with a 1% yield (4). No other laboratory has published a purification protocol for this enzyme. We developed a new strategy for the current study, the most notable aspect of which was the efficiency of an affinity purification step using heparin IIIS ( Fig. 2; Table I). Thus, using homogenates of bovine aortas as starting material, we purified the 1-kinase 22,500-fold with a 13% yield (Fig. 2; Table I). Our preparations of 1-kinase have an affinity for Ins(3,4,5,6)P 4 (0.1-0.2 M, data not shown) that is very similar to the substrate affinity of the rat hepatic enzyme (4). Gel filtration indicated the size of the enzyme to be 46 kDa (Fig. 3), which is also similar to that of the rat liver enzyme (4). The 1-kinase was strongly inhibited by Ins(1,3,4)P 3 (IC 50 ϭ 0.17 M, Fig. 4A). The enantiomer of Ins(1,3,4)P 3 , namely Ins(1,3,6)P 3 , was a 100-fold weaker inhibitor of the Ins(3,4,5,6)P 4 1-kinase (Table II). The activity of the purified Ins(3,4,5,6)P 4 1-kinase was unaffected when 20 mM KCl in the incubation buffer was substituted with 20 mM LiCl (data not shown).
The purified 1-kinase was reconstituted with either protein kinase A, protein kinase C, Ca 2ϩ /calmodulin, or CaM KII. In no case was there any modification to 1-kinase activity, nor was there any effect upon the potency of inhibition by Ins(1,3,4)P 3 (data not shown). Positive controls for each of these protein kinases were obtained by verifying their activities using appropriate assay kits (see "Experimental Procedures").
A second aspect of our experimental protocol that is worth emphasizing is that we preincubated cells with [ 3 H]inositol for 4 days. At this point the cellular pool of Ins(3,4,5,6)P 4 was radiolabeled to steady state (16,39). Thus, any increases in [ 3 H]Ins(3,4,5,6)P 4 that we observed truly reflect elevated mass levels of this polyphosphate. Third, we were concerned that cell-permeant analogues of inositol phosphates are typically de-esterified relatively slowly (17), such that a rapid rate of Ins(1,3,4)P 3 dephosphorylation would act to prevent the accumulation of this compound inside cells. Our cell incubation medium was therefore supplemented with lithium, so as to inhibit the Ins(1,3,4)P 3 1-phosphatase (40). However, it should be noted that this is only a partial solution of this particular problem, since lithium does not inhibit the less active, alternative pathway of Ins(1,3,4)P 3 dephosphorylation by a 4-phosphatase (41). Control experiments indicated that this lithium treatment did not affect levels of [ 3 H]Ins(3,4,5,6)P 4 (data not shown). Extracts of [ 3 H]inositol-labeled control cells were resolved by HPLC, and the various 3 H-labeled inositol phosphates were assayed using an on-line scintillation counter (Fig. 5, upper  panel). We also analyzed extracts from cells treated with 200 M D/L-Bt 3 Ins(1,3,4)P 3 /AM (Fig. 5, lower panel). In these experiments, any metabolic conversion of Ins(1,3,4)P 3 to Ins(3,4,5,6)P 4 would, by a pulse-chase effect, tend to decrease the amount of [ 3 H]label in the Ins(3,4,5,6)P 4 pool (15). In fact, the opposite result was obtained; the size of the [ 3 H] Ins(3,4,5,6)P 4 peak increased about 2-fold (Fig. 5, Table III We were only able to synthesize limited amounts of Bt 3 Ins(1,3,4)P 3 /AM, and so we did not have sufficient material to perform detailed dose-response curves or time courses. However, we did observe that the treatment of cells with 400 M D/L-Bt 3 Ins(1,3,4)P 3 /AM approximately doubled the elevation in  (3,4,5,6)P 4 levels (data not shown). One factor that must be taken into account when using these types of cell-permeant analogues is the relatively slow rate of their activation by intracellular esterases (17).
The Ins(3,4,5,6)P 4 1-kinase was used as a diagnostic tool so as to confirm the identity of the [ 3 H]Ins(3,4,5,6)P 4 peak that eluted from the HPLC. For these experiments, we did not assay [ 3 H]inositol phosphates by on-line scintillation counting. Instead, individual HPLC fractions were collected. The amount of material loaded onto the HPLC column was increased by culturing greater numbers of cells in larger wells (see "Experimental Procedures"). Thus, the quantity of [ 3 H]Ins(3,4,5,6)P 4 was larger than in the experiments described by Fig. 5. Aliquots of the putative [ 3 H]Ins(3,4,5,6)P 4 peak were incubated with the purified 1-kinase (see "Experimental Procedures"). The [ 3 H]InsP 5 formed, after the assays had been allowed to proceed to completion, was used to identify the amounts of   The potencies with which various inositol phosphates inhibited the  Ins(3,4 We next investigated if the treatment of AR4 -2J cells with Bt 3 Ins(1,3,4)P 3 /AM elicited increases in levels of Ins(1,3,4)P 3 and Ins(3,4,5,6)P 4 that were physiologically relevant. For these experiments, we compared the effects of the cell-permeant analogue with those of receptor-dependent activation of PLC, using bombesin as the agonist. [ 3 H]Inositol-labeled AR4 -2J cells were treated for 20 min with 200 nM bombesin, or for 60 min with either 100 M D-Bt 3 Ins(1,3,4)P 3 /AM, or vehicle. The cell-permeant derivative used in these experiments was from a batch that was different from that used in the experiments described above. This particular batch of the analogue was also enantiomerically pure, and therefore it was used at half the concentration of the D/L-Bt 3 Ins(1,3,4)P 3 /AM used in the experiments described above. In three experiments, 60-min treatment of AR4 -2J cells with 100 M D-Bt 3 Ins(1,3,4)P 3 /AM elevated [ 3 H]Ins(3,4,5,6)P 4 levels 1.9 Ϯ 0.3-fold (Fig. 6), which is not significantly different from the results obtained with 200 M D/L-Bt 3 Ins(1,3,4)P 3 /AM (Table III). The fact that these changes in Ins(3,4,5,6)P 4 levels were within a physiologically relevant range was confirmed by comparison with the effects of bombesin, which led to a nearly 5-fold increase in [ 3 H]Ins(3,4,5,6)P 4 levels (Fig. 6).
The product of 5-phosphatase attack upon Ins(1,4,5)P 3 , namely Ins(1,4)P 2 , was found to be a poor inhibitor of the 1-kinase (Fig. 7A). This is also an important observation that demonstrates that it specifically requires Ins(1,4,5)P 3 metabolism through the 3-kinase pathway to yield an inhibitor of the Ins(3,4,5,6)P 4 1-kinase. In addition, this result shows that all three phosphates of Ins(1,3,4)P 3 contribute substantially to the specificity of inhibition of the 1-kinase.
The Contribution of the 2-Phosphate Group to Inhibition of the 1-Kinase-We found that, in an appropriate context, the addition of a 2-phosphate to the inositol ring could make a positive contribution to inhibitory potency; this is illustrated by the observation that Ins(1,2,3)P 3 was a more potent inhibitor of the 1-kinase (IC 50 ϭ 4.2 M, Table II) than was Ins(1,3)P 2 (IC 50 Ͼ 30 M, Fig. 7A). D/L-Ins(1,2,3,4)P 4 was also a relatively potent 1-kinase inhibitor (IC 50 ϭ 1.6 M, Table II). D-and/or L-Ins(1,2,3,4)P 4 and Ins(1,2,3)P 3 are found in mammalian cells, the latter at concentrations of up to 10 M (15,42,43). These polyphosphates may be formed by dephosphorylation of InsP 6 (15,42,43). Among the inositol phosphates that we tested that contain a 2-phosphate (Table II), D/L-Ins(1,2,4)P 3 was the most potent inhibitor of the 1-kinase (IC 50 ϭ 0.7 M, Fig. 7B). Neither D-nor L-Ins(1,2,4)P 3 have been detected in mammalian cells, even under circumstances where, had they been present, they should have revealed themselves to some detailed structural analyses (43). As the inhibitory action of Ins(1,3,4)P 3 upon the 1-kinase was so effectively imitated by D/L-Ins(1,2,4)P 3 (Table II), the latter could be a useful new starting point for developing drugs that might intervene in the 1-kinase/1-phosphatase cycle with therapeutic benefit.
Ins(1,3,4)P 3 would seem to be particularly well suited to its task as an intracellular mediator that links Ins(3,4,5,6)P 4 levels to changes in PLC activity, since cellular levels of Ins(1,3,4)P 3 quite closely follow both the extent and the duration of PLC activation (45,46). In addition, we have shown the relative ineffectiveness with which the 1-kinase is inhibited by both of the InsP 2 products of Ins(1,3,4)P 3 metabolism, namely Ins(1,3)P 2 and Ins(3,4)P 2 (Fig. 7). Thus, the dephosphorylation of Ins(1,3,4)P 3 comprises an efficient signaling off-switch. In this context, it now seems more significant that both the 1-and 4-phosphatases that attack Ins(1,3,4)P 3 may be regulated. This is indicated first by the amino acid sequence of the 4-phosphatase containing consensus motifs for phosphorylation by protein kinases (47). Second, and more directly, Ca 2ϩ activates the 1-phosphatase (40).
As a result of this study, we are wiser as to the structural determinants of Ins(1,3,4)P 3 -mediated inhibition of the 1-kinase (Table II). We also made the intriguing observation that a 2-phosphate could increase potency of 1-kinase inhibition, depending upon which other phosphate groups were also present. A practical outcome of this increased information concerning specificity is that the non-physiological material, D/L-Ins(1,2,4)P 3 , proved to be a particularly potent 1-kinase inhibitor; this could be a productive starting point for the rational design of therapeutically useful drugs that might inhibit the Ins(3,4,5,6)P 4 1-kinase in vivo. This provides an alternative to the approach of designing drugs that act at the site of action of Ins(3,4,5,6)P 4 (27).
The very existence of the Ins(1,4,5)P 3 3-kinase, but more so its complex regulation through cross-talk from other signaling pathways, are observations that have been used to bolster the teleological argument that Ins(1,3,4,5)P 4 must be functionally significant (2). Indeed, there is a large body of evidence that Ins(1,3,4,5)P 4 does indeed perform a valuable role inside cells (2). The 3-kinase also has the role of inactivating Ca 2ϩ signaling by Ins(1,4,5)P 3 . Our new data assign additional significance to this metabolic pathway: control over the production of Ins(1,3,4)P 3 , which in turn regulates cellular levels of Ins(3,4,5,6)P 4 (an inhibitor of Ca 2ϩ -activated Cl Ϫ channels) (8). The acknowledgment that the Ins(1,4,5)P 3 3-kinase has several important roles provides us with a better appreciation of why so many cellular control processes converge on the regulation of this enzyme's activity (48).