A Novel Receptor-mediated Regulation Mechanism of Type I Inositol Polyphosphate 5-Phosphatase by Calcium/Calmodulin-dependent Protein Kinase II Phosphorylation*

d-myo-inositol 1,4,5-trisphosphate (Ins(1,4,5)P3) andd-myo-inositol 1,3,4,5-tetrakisphosphate (Ins(1,3,4,5)P4) are both substrates of the 43-kDa type I inositol polyphosphate 5-phosphatase. Transient and okadaic acid-sensitive inhibition by 70–85% of Ins(1,4,5)P3 and Ins(1,3,4,5)P4 5-phosphatase activities was observed in homogenates from rat cortical astrocytes, human astrocytoma 1321N1 cells, and rat basophilic leukemia RBL-2H3 cells after incubation with carbachol. The effect was reproduced in response to UTP in rat astrocytic cells and Chinese hamster ovary cells overexpressing human type I 5-phosphatase. Immunodetection as well as mass spectrometric peptide mass fingerprinting and post-source decay (PSD) sequence data analysis after immunoprecipitation permitted unambiguous identification of the major native 5-phosphatase isoform hydrolyzing Ins(1,4,5)P3 and Ins(1,3,4,5)P4 as type I inositol polyphosphate 5-phosphatase. Inortho-32P-preincubated cells, the phosphorylated 43 kDa-enzyme could be identified after receptor activation by immunoprecipitation followed by electrophoretic separation. Phosphorylation of type I 5-phosphatase was blocked after cell preincubation in the presence of Ca2+/calmodulin kinase II inhibitors (i.e. KN-93 and KN-62). In vitro phosphorylation of recombinant type I enzyme by Ca2+/calmodulin kinase II resulted in an inhibition (i.e. 60–80%) of 5-phosphatase activity. In this study, we demonstrated for the first time a novel regulation mechanism of type I 5-phosphatase by phosphorylation in intact cells.

The critical role of type I inositol 5-phosphatase in controlling the Ca 2ϩ response and the Ca 2ϩ oscillations in intact cells has been supported by the construction of mutants and their transfection in CHO-K1 cells stimulated by ATP. The pattern of Ca 2ϩ oscillations in cells transfected with a phosphatase mutant that could not be prenylated remains unchanged in contrast to cells transfected with the wild type enzyme where the Ca 2ϩ oscillations were totally abolished (16). Underexpression of the 43-kDa 5-phosphatase in rat kidney cells provoked an increase in Ins(1,4,5)P 3 and Ins(1,3,4,5)P 4 levels as well as in basal intracellular calcium concentration (17). Cells underexpressing the 43-kDa 5-phosphatase demonstrated a transformed phenotype, and antisense-transfected cells formed tumors in nude mice (17).
Despite the presence of several potential phosphorylation sites based on consensus phosphorylation site sequences for protein kinase C (PKC) as well as for Ca 2ϩ /calmodulin (Ca 2ϩ / CaM)-dependent protein kinase II (Ca 2ϩ /CaM kinase II) and cAMP-dependent protein kinase (PKA) (5,6,18), there is no evidence for PKC-dependent phosphorylation of recombinant human enzyme produced in bacteria (19). Nevertheless, two groups reported that the enzyme can be activated by interaction with PKC-phosphorylated pleckstrin or with another platelet protein, i.e. 14-3-3-, in vitro (20,21). Phosphorylation of the enzyme by Ca 2ϩ /CaM kinase II has not been investigated yet.
It has been previously reported that Ins(1,4,5)P 3 3-kinase activities of the isoenzymes A and B are sensitive to the Ca 2ϩ / CaM complex (22)(23)(24)(25). In intact cells, the two isoforms are phosphorylated and activated by Ca 2ϩ /CaM kinase II (for isoform A, see Ref. 26) or by PKC and Ca 2ϩ /CaM kinase II (for isoform B, see Ref. 27). In this paper, we report for the first time that the 43-kDa 5-phosphatase is the target of a regulatory mechanism involving Ca 2ϩ /CaM kinase II-mediated phosphorylation both in vitro and in intact cells in response to FIG. 1. Immunodetection and immunoprecipitation of type I 5-phosphatase from astrocytic and basophilic mast cells by antihuman 43 kDa 5-phosphatase antibodies. A, a 5-l aliquot of crude cell lysate (ϳ25 g of proteins) was separated by SDS/PAGE, and the 5-phosphatase (indicated by an arrow) was immunodetected in the presence of anti-43-kDa 5-phosphatase antibodies in rat brain cortical astrocytes (lane 1), 1321N1 cells (lane 2), and RBL-2H3 cells (lane 3). B, immunoprecipitated 5-phosphatase from 1321N1 cells (100 l of crude cell lysate, i.e. ϳ700 g of proteins, with 2 l of immune serum) was separated by SDS/PAGE, and the protein was stained by colloidal Coomassie Blue (lanes 1 and 2) and then excised (circle, lane 1) for mass spectrometry analysis. The standard molecular masses (kDa) are indicated in the margin.  19.02 R -R a Tryptic peptides are single-charged, and therefore, masses include one proton. Experimental and theoretical masses are monoisotopic masses, respectively. Mass accuracy is typically 0.1% or better. Theoretical masses were calculated considering normal cysteine and methionine and tryptic cleavage after lysine and arginine but not before proline. These seven peptides cover 88 out of the 412 amino acids (i.e. 21%) from the human primary structure. The peptide that has been analyzed by PSD is indicated by an asterisk.
b The peptide with an experimental monoisotopic mass of 1605.777 Da has been selected as a parent ion for fragmentation by PSD generating the typical C-terminal 18 O-labeled y-ion series. In the PSD spectrum, the parent ion presented a mass of 1605.22 Da. Theoretical monoisotopic masses were calculated considering normal cysteine and methionine. The dashes mean not detected. Analysis of the raw PSD data by using the Mascot algorithm gave the type I 5-phosphatase primary structure as the top-scoring protein.
Ca 2ϩ -raising agents. This effect appears quite general; indeed, treatment of rat brain cortical astrocytes, human astrocytoma 1321N1 cell line, and rat basophilic leukemia RBL-2H3 cell line with carbachol provoked a transient increase in the phosphorylation of the 43-kDa isoform and a corresponding decrease in enzymic activity.

EXPERIMENTAL PROCEDURES
Generation of Antibodies Dressed against the N-terminal Part of Human 43-kDa Inositol Polyphosphate 5-Phosphatase-An immune serum recognizing a 16-amino acid peptide (i.e. MALHCQEFGGKN-YEAS) corresponding to amino acids 49 -64 of human 43-kDa 5-phosphatase (28) coupled to hemocyanine has been generated in rabbits. It was verified that the antibodies (dilution 1:1000) immunodetected and immunoprecipitated recombinant type I 5-phosphatase. An aliquot of 2 l of crude immune serum was used for immunoprecipitation experiments. For competition studies, 5 g of the corresponding peptide were added for 1 h to the diluted serum before immunodetection.
Preparation of Rat Brain Cortical Astrocytes-Primary cultures of rat cerebral cortex astrocytes were established using dissociated rat cerebral tissue at 2 or 3 days after birth, according to methods previously described (29). Briefly, dissected cortical tissue was washed and then dissociated by gentle repeated pipetting in modified Eagle's medium containing 1 mM sodium pyruvate, 10% fetal calf serum, 2% penicillin/streptomycin, and 1% fungizone. Cells were decanted by gravity for 5 min, and the supernatant was saved. Cells were diluted in complemented cell medium and plated in 10-cm-diameter or 22 ϫ 22-cm-square dishes at 37°C with 5% CO 2 . When astrocytes were adherent, dishes were agitated overnight, and the medium was changed  (1,4,5)P 3 5-phosphatase activity in rat brain cortical astrocytes and RBL-2H3 cells Enzyme activity was assayed at 10 M Ins(1,4,5)P 3 . 100% enzyme activity corresponded to 0.15 mol/min ⅐ mg and 0.22 mol/min ⅐ mg in rat cortical astrocytes and RBL-2H3 cells, respectively. O.A., okadaic acid. Each value is the mean of duplicates Ϯ S.D. The results are from one representative experiment out of four.

,5)P 4 5-phosphatase activity in rat brain cortical astrocytes and RBL-2H3 cells
Enzyme activity was assayed at 5 M Ins(1,3,4,5)P 4 . 100% enzyme activity corresponded to 20 nmol/min ⅐ mg and 28 nmol/min ⅐ mg in rat cortical astrocytes and RBL-2H3 cells, respectively. O.A., okadaic acid. Each value is the mean of duplicates Ϯ S.D. The results are from one representative experiment out of three.  after two washes. Astrocytes reached confluence after 7 days in culture and could be trypsinized 3 to 4 times (27). Human astrocytoma 1321N1 and rat basophilic leukemia RBL-2H3 cell lines were grown in the same complemented modified Eagle's medium. CHO-K1 cells were grown as reported before (28). Cell culture medium, dishes, and antibiotics were from Life Technologies, Inc.
Cell Labeling and Enzyme Immunoprecipitation-When cells were ϳ80% confluent in 6-cm-diameter culture dishes, they were washed two  (1,4,5)P 3 5-phosphatase activity in rat brain cortical astrocytes and RBL-2H3 cells Enzyme activity was assayed at 10 M Ins(1,4,5)P 3 . 100% enzyme activity corresponded to 0.15 mol/min ⅐ mg and 0.22 mol/min ⅐ mg in rat cortical astrocytes and RBL-2H3 cells, respectively. Each value is the mean of duplicates Ϯ S.D. The results are from one representative experiment out of three.  The cells were subsequently washed in prewarmed KRH medium, and an aliquot of 2 ml of this medium containing the agent(s) was pipetted onto each culture dish for an incubation with agonist. Crude cell extracts were prepared as described above. Type I 5-phosphatase was immunoprecipitated using protein A-Sepharose (Amersham Pharmacia Biotech) coupled to anti-rabbit IgG (Sigma) and the rabbit polyclonal anti-human brain type I 5-phosphatase antibodies. An aliquot of 90 l of cell extract (ϳ600 g of protein) was immunoprecipitated in the presence of 25 l of pretreated protein A-Sepharose and 2 l of immune serum. Immune complexes were separated by SDS/PAGE and detected by autoradiography using a Hyperfilm-MP (Amersham Pharmacia Biotech) exposed for 24 -32 h. In Vitro Enzyme Phosphorylation by Ca 2ϩ /CaM Kinase II, PKC, and PKA-Recombinant human brain type I 5-phosphatase was affinitypurified by metal chelation as described previously (30). Phosphorylation of purified recombinant human brain type I 5-phosphatase by Ca 2ϩ /CaM kinase II was performed at 37°C for 5 min in Hepes/NaOH, itated, and washed in the presence of 75 mM phosphoric acid before counting radioactivity (31).

Peptide Mapping and Sequence Analysis by MALDI-Reflection Time of Flight Mass Spectrometry-
The excised Coomassie Blue-stained gel band was treated as previously described (32,33), except that the protein was denatured in 8 M urea/Tris 0.1 M, pH 8.5, and diluted to 2 M urea before tryptic digestion in 50% H 2 18 0 to label the C-terminal carboxyl moiety of the generated peptides. Tryptic digestion was performed overnight in the presence of 500 ng of modified trypsin (Promega). Adsorption onto Poros R2 beads (PerSeptive Biosystems) as well as mass spectrometry experiments for peptide mapping and sequence analysis by PSD on a Bruker Reflex III Instrument (Bruker-Franzen Analytik GmbH) were performed as previously reported (33). Briefly, the dried peptide-bound beads were mixed with 0.7 l of 4 mg of ␣-cyano-4-hydroxycinnamic acid, 0.8 mg of 3,5-dihydroxybenzoic acid in 50% acetonitrile/water with 0.1% trifluoroacetic acid. After air-drying onto the MALDI target, MALDI mass spectra were recorded after ionization, achieved using a 337-nm nitrogen laser (attenuation between 15 and 25). In the reflectron mode, positive (monocharged) ions were accelerated to a total voltage of 26.3 kV. After selection of the interesting peptide, PSD spectra were acquired in segments, with a 15% decrease in reflectron voltage for each. PeptideSearch algorithm was used to search the non-redundant protein data base for direct peptide monoisotopic mass fingerprinting and PSD-derived sequence tag search. The Mascot algorithm (available at www.matrixscience.com) was also used for protein identification using peptide mass fingerprints and peptide fragmentation spectra. Type I Ins(1,4,5)P 3 /Ins(1,3,4,5)P 4

5-Phosphatase by Immunodetection, Immunoprecipitation, and Mass
Spectrometry-A 43-kDa band was observed after immunodetection with anti-type I 5-phosphatase antibodies in human astrocytic 1321N1 cells, rat brain cortical astrocytes, and RBL-2H3 cells (Fig. 1A). The presence of the initial immunogenic peptide with the same immune serum during the immunodetection abolished the signal at 43 kDa (data not shown). After immunoprecipitation from a crude homogenate of 1321N1 cells and colloidal Coomassie Blue staining (Fig. 1B), the protein was excised from the gel and digested in the presence of trypsin, and the generated peptides were analyzed by mass spectrometry. Direct mass fingerprinting and sequence data obtained by PSD showed that the excised 43-kDa protein corresponded to the type I 5-phosphatase (Table I). In the supernatant obtained after immunoprecipitation from both cell types, Ins(1,4,5)P 3 and Ins(1,3,4,5)P 4 5-phosphatase activities were decreased by at least 90% as compared with unprecipitated enzyme (data not shown).
Phosphorylation by Ca 2ϩ /CaM Kinase II Was Necessary for Maximal Inhibition of 43-kDa 5-Phosphatase-Rat cortical astrocytes and RBL-2H3 cells were prelabeled with ortho-32 P and incubated with carbachol to inhibit the enzymic activity. Type I 5-phosphatase was immunoprecipitated and analyzed by SDS/PAGE. Enzyme inhibition coincided with phosphate incorporation into the 43-kDa protein band (Fig. 2). Mass spectrometric analysis after excision and tryptic digestion of the radioactive 43-kDa band showed that it corresponded to type I 5-phosphatase (data not shown). 32  and without any kinase (lane 6). As a control, 0.75 g of purified Ins(1,4,5)P 3 3-kinase A (26) was phosphorylated under the same conditions in the presence of Ca 2ϩ /CaM kinase II and its cofactors (lane 1). After SDS/PAGE, the gel was stained with colloidal Coomassie Blue and then autoradiographed for 12 h. The standard molecular masses (kDa) are indicated. B, Ca 2ϩ /CaM kinase II-provoked 32 P incorporation into human type I 5-phosphatase was measured by phosphorylating enzyme (5 g) with Ca 2ϩ /CaM kinase II (50 ng) in the presence (E) or absence (ⅷ) of 10 M free Ca 2ϩ , 2 M CaM, and 100 M [␥-32 P]ATP (final activity ϳ250 Ci/ml) for various times (0 -10 min). It was also performed in the presence of PKA (f) and PKC (Ⅺ) and its cofactors. After each incubation time, the enzyme was precipitated onto P81 phosphocellulose with 75 mM phosphoric acid before counting radioactivity. Results are the means of triplicates Ϯ S.D. 30 s. Preincubation with okadaic acid before receptor activation potentiated the phosphate incorporation into the 43-kDa enzyme in both cell systems, which was more sustained between 15 s and 1 min (Fig. 2). The same results were obtained after stimulation of human astrocytic 1321N1 cells by carbachol (data not shown). Moreover, preincubation with KN-93 or KN-62 before receptor activation prevented 32 P incorporation into the enzyme in a dose-dependent manner, with a maximal effect at 2 M. The two inactive analogs KN-92 and KN-04 did not prevent the incorporation of 32 P (Fig. 3A). TPA, calphostin C, and forskolin had no effect on enzyme phosphorylation (data not shown).

In Vitro Phosphorylation and Inhibition of Recombinant 43-kDa 5-Phosphatase by Ca 2ϩ /CaM Kinase II-Since in vivo
phosphorylation of type I 5-phosphatase was prevented by potent Ca 2ϩ /CaM kinase II inhibitors, we investigated in vitro phosphorylation by Ca 2ϩ /CaM kinase II of recombinant human type I 5-phosphatase produced in bacteria. Ca 2ϩ /CaM kinase II-catalyzed phosphorylation resulted in a decrease in Ins(1,4,5)P 3 (Fig. 4A) and Ins(1,3,4,5)P 4 (Fig. 4B) 5-phosphatase activities in the presence of 10 M free Ca 2ϩ and 2 M CaM, i.e. 50 -70% as compared with basal activity measured after preincubation in the absence of Ca 2ϩ /CaM kinase II, Ca 2ϩ , and CaM. In vitro phosphorylation of type I 5-phosphatase by Ca 2ϩ /CaM kinase II did not provoke any change in sensitivity toward magnesium ions (data not shown). In agreement with previously reported data (19), PKA-and PKC-dependent phosphorylation did not provoke any modulation in Ins(1,4,5)P 3 and Ins(1,3,4,5)P 4 5-phosphatase activities in vitro (Fig. 4, A and B). Inhibition of purified type I 5-phosphatase by Ca 2ϩ /CaM kinase II in vitro was correlated with an incorporation of 32 P into the 43-kDa enzyme (Fig. 5A). This was not observed after incubation in the presence of PKA or PKC (Fig.  5A). Additionally, stoichiometric measurements indicated that the enzyme was phosphorylated by Ca 2ϩ /CaM kinase II at two major residues and reached a plateau in a time course study after 4 min at 37°C (Fig. 5B). DISCUSSION The apparent requirement of Ins(1,3,4,5)P 4 in order for Ins(1,4,5)P 3 to stimulate Ca 2ϩ entry and mobilization has been reported in mouse lachrymal cells (34). Application of Ins(1,3,4,5)P 4 in an extracellular manner in normal hippocampal slices mimicked the deterioration of ischemic neurons, so it was suggested that the formation of Ins(1,3,4,5)P 4 plays a critical role in neuronal death and that Ins(1,3,4,5)P 4 acts as a signal for inducing Ca 2ϩ entry (35). In rat hippocampus, Ins(1,3,4,5)P 4 enhances long term potentiation by regulating Ca 2ϩ entry through up-regulation of voltage-gated calcium channels. So it was proposed that a possible physiological function of Ins(1,3,4,5)P 4 could be to activate postsynaptic Ins(1,4,5)P 3 -dependent Ca 2ϩ release and, therefore, to participate in the induction of the long term potentiation that is normally dependent on the Ca 2ϩ -induced Ca 2ϩ release (36). In another recent study, it has been shown that the activation of the store-operated Ca 2ϩ current I CRAC in the rat mast cell line RBL-2H3 was facilitated by Ins(1,3,4,5)P 4 -mediated inhibition of Ins(1,4,5)P 3 metabolism through type I Ins(1,4,5)P 3 5-phosphatase (37). Depending on its cellular concentrations, Ins(1,3,4,5)P 4 could act as a bi-modal regulator of calcium signaling, inhibiting Ins(1,4,5)P 3 receptors at high concentrations and inhibiting Ins(1,4,5)P 3 5-phosphatase at low concentrations by acting as a co-substrate. Such data support a second messenger function for Ins(1,3,4,5)P 4 (discussed in Ref. 38).
Evidence has been provided that enzymes that phosphorylate Ins(1,4,5)P 3 to Ins(1,3,4,5)P 4 , i.e. Ins(1,4,5)P 3 3-kinase isoforms, are regulated by phosphorylation in intact cells. Ins(1,4,5)P 3 3-kinase A, which is largely expressed in the dendrites and spines of hippocampal CA1 cells (39 -41), is activated upon phosphorylation by Ca 2ϩ /CaM kinase II (26). Ca 2ϩ / CaM kinase II is also particularly abundant in the hippocampus, and it has been proposed that Ins(1,4,5)P 3 3-kinase A may function as a co-incidence detector as it may be an effector of the "frequency reading" Ca 2ϩ /CaM kinase II (42)(43)(44)(45). Such a regulation mechanism has also been proposed for the isoform B in primary cultures of rat cortical astrocytes and in the human 1321N1 astrocytoma cell line, where this isoenzyme could be activated through a Ca 2ϩ /CaM kinase II and protein kinase C phosphorylation mechanism (27). The data presented here support the idea that both the Ins(1,4,5)P 3 kinase and the major Ins(1,4,5)P 3 /Ins(1,3,4,5)P 4 phosphatase, i.e. type I 5-phosphatase, are coordinated to generate high levels of Ins(1,3,4,5)P 4 . Indeed, as it has been shown for the Ins(1,4,5)P 3 3-kinases A and B, we show now for the first time that type I 5-phosphatase is a target of a regulation mechanism through Ca 2ϩ /CaM kinase II in intact cells and in vitro. This was shown in rat cerebral cortex astrocytes, 1321N1 and RBL-2H3 cell lines stimulated by carbachol (as well as UTP, in the case of rat astrocytes), and in type I 5-phosphatase-transfected CHO cells stimulated by UTP. Ca 2ϩ mobilization resulting from an increased production of Ins(1,4,5)P 3 had been shown in CHO cells in response to purinoreceptor P2Y2 activation (46). Activation of P2Y2 receptor with UTP also provoked a rapid decrease in Ins (1,4,5)  2H3 cell line has been shown in this study by immunoprecipitation and mass spectrometry analysis. As for astrocytes (47), the production of inositol phosphates, especially Ins(1,3,4,5)P 4 , has been established in the rat basophilic leukemia RBL-2H3 cell line (48,49). In this cell line, the two candidate Ins(1,3,4,5)P 4 receptors GAP1(IP4BP) and GAP1(m), which belong to the GAP1 family of Ras GTPase-activating proteins, have been shown to be primarily localized to the plasma membrane and to the cytoplasm, respectively (50). When cells were preincubated with 32 P, evidence was provided that Ca 2ϩ /CaM kinase II inhibitors (i.e. KN-93 and KN-62) prevented phosphate incorporation into the 43-kDa protein after immunoprecipitation. Phosphorylation in both cell systems was protected in the presence of okadaic acid, suggesting that the phosphorylated enzyme could be a substrate for protein phosphatase 1 or 2A (51). Non-effective analogs of Ca 2ϩ / CaM kinase II inhibitors, i.e. KN-92 and KN-04, did not prevent phosphorylation of the enzyme in intact cells, supporting the supposition that the effects of KN-93 and KN-62 were specific toward inhibition of Ca 2ϩ /CaM kinase II. TPA, calphostin C, and forskolin did not induce any change in Ins(1,4,5)P 3 / Ins(1,3,4,5)P 4 5-phosphatase activity nor 32 P incorporation in any tested cell types. More directly, purified type I 5-phosphatase was phosphorylated in vitro by Ca 2ϩ /CaM kinase II. Direct in vitro 32 P incorporation showed that at least two residues are involved. Four consensus sites (Arg/Lys)-X-X-(Ser/Thr) for protein phosphorylation by Ca 2ϩ /CaM kinase II (52) are found in the primary structure of human type I Ins(1,4,5)P 3 5-phosphatase, i.e. serines 239, 310, 366, and threonine 372, and are conserved in the orthologuous dog primary structure (5). These four residues are located in the catalytic domain of the enzyme, which has been so defined on the basis of the conservation of this domain sequence between the different members of the 5-phosphatase family and on the identification of critical residues involved in substrate binding and/or catalysis (30,(53)(54)(55). Phosphorylation of type I 5-phosphatase decreased enzyme activity but without affecting the sensitivity to magnesium ions (in intact cells nor in vitro).
CaM kinase II thus appears a central regulatory enzyme in phosphorylating both enzymes controlling Ins(1,4,5)P 3 and Ins(1,3,4,5)P 4 levels. It has also been reported previously that Ins(1,4,5)P 3 receptor could be a target of phosphorylation by Ca 2ϩ /CaM kinase II (56). Ca 2ϩ /CaM kinase II, a multifunctional enzyme that catalyzes phosphorylation of many proteins, has a wide tissue distribution but is particularly abundant in brain (57). Regulation of several CaM-dependent proteins by Ca 2ϩ /CaM kinase II has been reported. For example, phosphorylation of calcineurin decreased phosphatase activity by decreasing the V max or by increasing the K m , depending on the substrate utilized (58). Our data suggest that in different cell types (e.g. astrocytic cells), Ins(1,3,4,5)P 4 levels could be controlled by Ca 2ϩ /CaM kinase II-dependent phosphorylation of Ins(1,4,5)P 3 3-kinase and Ins(1,4,5)P 3 /Ins(1,3,4,5)P 4 5-phosphatase. Generation of oscillations in Ins(1,4,5)P 3 is thought to require Ca 2ϩ -dependent activation of PLC (59 -61). We previously reported the importance of the kinetic parameters of type I 5-phosphatase in the control of Ca 2ϩ oscillations in response to any extracellular signal (16,62). Ca 2ϩ -dependent phosphorylation of the two Ins(1,4,5)P 3 -and Ins(1,3,4,5)P 4 -metabolizing enzymes with opposite change in activity could provide an alternative important mechanism in the generation or control of Ca 2ϩ oscillations (Table VI).
Recent data obtained in primary hippocampal cultures indicated that Ins(1,4,5)P 3 3-kinase A associates with F-actin and is also colocalized postsynaptically with Ca 2ϩ /CaM kinase II (63). The data therefore suggest a direct link between the production of Ins(1,3,4,5)P 4 and calcium signaling via Ca 2ϩ / CaM kinase II. The latter enzyme has been shown to be an integrator of pulsatile Ca 2ϩ signals (44). The data presented here provide an additional mechanism of control of Ins(1,3,4,5)P 4 levels by decreasing type I 5-phosphatase activity in stimulated cells. The mechanism also involves Ca 2ϩ /CaM kinase II phosphorylation. Since type I 5-phosphatase is concentrated in neurons (6), it is tempting to speculate that both Ins(1,4,5)P 3 3-kinase and 5-phosphatase could localize in some cells and at some time. Whether type I 5-phosphatase would be associated with F-actin or be present in dendritic spines is not known. However, the enzyme can be phosphorylated (this study) and is particularly abundant in Purkinje cells, and calcium released selectively from the spine apparatus of Purkinje cells is crucial for the establishment of long term depression (64).
The coordination of the two enzymes responsible of the Ins(1,3,4,5)P 4 metabolism could be generalized to other second messengers. For example, a subset of olfactory neurons selectively express cGMP-stimulated phosphodiesterase and guanyl cyclase D and define a unique olfactory signal transduction pathway (65). Targeting of protein kinases and phosphatases to the cytoskeleton enhances the regulation of signal transduction events. Targeting of the PKA to the cytoskeleton is achieved through interaction with protein kinase A-anchoring proteins, AKAPs, which maintain multivalent signaling complexes by binding additional enzymes, including kinases and phosphatases (66). The SH2 domain-containing 5-phosphatase SHIP1, when phosphorylated on tyrosine, has been shown to recruit the regulatory subunit of the phosphatidylinositol 3,4,5trisphosphate synthetic enzyme p85 (67), arguing in favor of a coordinated action of phosphatidylinositol 3,4,5-trisphosphate synthesis and degradation.