The Inositol Hexakisphosphate Kinase Family

Saiardi et al. (Saiardi, A., Erdjument-Bromage, H., Snowman, A., Tempst, P., and Snyder, S. H. (1999) Curr. Biol. 9, 1323–1326) previously described the cloning of a kinase from yeast and two kinases from mammals (types 1 and 2), which phosphorylate inositol hexakisphosphate (InsP6) to diphosphoinositol pentakisphosphate, a “high energy” candidate regulator of cellular trafficking. We have now studied the significance of InsP6 kinase activity inSaccharomyces cerevisiae by disrupting the kinase gene. These ip6kΔ cells grew more slowly, their levels of diphosphoinositol polyphosphates were 60–80% lower than wild-type cells, and the cells contained abnormally small and fragmented vacuoles. Novel activities of the mammalian and yeast InsP6kinases were identified; inositol pentakisphosphate (InsP5) was phosphorylated to diphosphoinositol tetrakisphosphate (PP-InsP4), which was further metabolized to a novel compound, tentatively identified as bis-diphosphoinositol trisphosphate. The latter is a new substrate for human diphosphoinositol polyphosphate phosphohydrolase. Kinetic parameters for the mammalian type 1 kinase indicate that InsP5(K m = 1.2 μm) and InsP6(K m = 6.7 μm) compete for phosphorylation in vivo. This is the first time a PP-InsP4 synthase has been identified. The mammalian type 2 kinase and the yeast kinase are more specialized for the phosphorylation of InsP6. Synthesis of the diphosphorylated inositol phosphates is thus revealed to be more complex and interdependent than previously envisaged.

The very dynamic turnover of the "high energy" diphosphorylated inositol polyphosphates (PP-InsP 4 , PP-InsP 5 , and [PP] 2 -InsP 4 ) 1 may represent a molecular switching activity that regulates intracellular trafficking (see Ref. 1 for a review). For example, PP-InsP 5 represents the most potent known inhibitor of AP180-mediated assembly of clathrin cages, a key step in the endocytic retrieval of discharged synaptosomal vesicles (2). Other proteins that participate in intracellular trafficking can bind PP-InsP 5 very tightly, including coatomer (3,4) and AP2 (5). The high affinity with which PP-InsP 5 binds to myelin proteolipid protein may be important for the vesicular delivery of the latter to the myelin sheath (6).
There are increasing efforts to characterize the activities of the enzymes that regulate the turnover of PP-InsP 4 , PP-InsP 5 , and [PP] 2 -InsP 4 . Several phosphatases (diphosphoinositol polyphosphate phosphohydrolases) that hydrolyze these compounds have been described (11,12). Two forms of InsP 6 kinase (types 1 and 2), derived from distinct genes, have been cloned from mammals (13,14). At least in mammals, the further phosphorylation of PP-InsP 5 to [PP] 2 -InsP 4 appears to be the function of a separate enzyme that has been purified from rat brain (9) but not yet cloned. A yeast InsP 6 kinase has also been cloned that shows approximately 30% sequence similarity to the two mammalian InsP 6 kinases (13). We also recently reported that PP-InsP 5 and [PP] 2 -InsP 4 are present in Saccharomyces cerevisiae (15).
A notable omission from our understanding of the intricacies of the turnover of higher inositol phosphates is any characterization of the enzyme(s) that phosphorylate Ins(1,3,4,5,6)P 5 to PP-InsP 4 . Thus, an important goal of this study was to identify this "missing link" in inositide research: the PP-InsP 4 synthase. We now report that PP-InsP 4 in mammals is synthesized by the type 1 and type 2 forms of the InsP 6 kinase. In addition, we show that these enzymes can further phosphorylate PP-InsP 4 , thereby forming a hitherto unknown inositol polyphosphate.
We have further investigated the activity of the yeast enzyme in several ways. We have characterized its substrate specificity in vitro. In this study, we demonstrate that disrupting the InsP 6 kinase gene of S. cerevisiae (ip6k⌬) influences inositol polyphosphate levels in a unique fashion. We also show that ip6k⌬ yeast cells have fragmented vacuoles. These data support an important role for diphosphoinositol polyphosphates in regulating intracellular trafficking. * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18

EXPERIMENTAL PROCEDURES
Enzyme Assays-The various InsP 6 kinases used in this study were incubated for various times at 37°C in 25 or 50 l of buffer containing 20 mM HEPES, pH 7.0, with KOH, 12 mM MgSO 4 , 1 mM dithiothreitol, 10 mM ATP, 20 mM phosphocreatine, 1 mM EDTA, 0.02 mg/ml phosphocreatine kinase (Calbiochem 238395) and 0.5 mg/ml bovine serum albumin. The appropriate 3 H-labeled inositol phosphate was added as indicated. Assays were quenched with ice-cold perchloric acid and neutralized as described previously (16).
Where specifically indicated, reactions were quenched by incubation for 3 min at 100°C. Control experiments showed that all kinase activity was inactivated by this heat treatment, but none of the diphosphorylated inositol phosphates were degraded.
HPLC Analyses-The HPLC analysis of [ 3 H]inositol-labeled yeast cells was performed as described previously (15). Assays of the activities of recombinant enzymes employed HPLC using a Partisphere SAX column (Krackler Scientific, Durham, NC) that was eluted with a gradient generated by mixing Buffer A (1 mM Na 2 EDTA) and Buffer B (Buffer A plus 1.3 M (NH 4 ) 2 HPO 4 , pH 3.85, with H 3 PO 4 ) as follows: 0 -5 min, 0% B; 5-10 min, 0 -45% B; 10 -60 min, 45-100% B; 60 -70 min, 100% B. 1-ml fractions were collected. In some experiments, particularly when a new HPLC column was installed, the percentage of B at 10 min was increased to 50%. This change in the gradient, plus the tendency of inositol phosphates to elute earlier as the column aged, means that slightly different elution properties are seen in the different figures shown in this study.
Preparations of Recombinant Proteins-Recombinant hDIPP2␣ was prepared as described previously (11). The cDNAs for mammalian InsP 6 kinase types 1 and 2 were recovered from the pCMV-glutathione Stransferase vector used previously (13). The type 1 kinase cDNA was amplified using the following primers: 5Ј-GCACTCGAGAATGTGTGT-TTGTCAAAC-3Ј and 5Ј-GCTAAGCTTAGGGCCTACTGGTTCTC-3Ј; the polymerase chain reaction product was subcloned into the XhoI and HindIII sites of the pTrcHisB expression vector (Invitrogen). The cDNA for the type 2 kinase was amplified using the following primers: 5Ј-C-GACTCGAGGATGAGCCCAGCCTTCAG-3Ј and 5Ј-GCATTCGAAC TCACTCCCCACTCTCCTC-3; the polymerase chain reaction product was subcloned into the XhoI and BstBI restriction sites of pTrcHisB. The methods used to transform Escherichia coli (strain BL21), to induce with isopropyl-1-thio-␤-D-galactopyranoside, and to isolate the poly-(His)-tagged proteins using Talon resin (CLONTECH), were all according to the manufacturer's recommendations. The recombinant yeast InsP 6 kinase was produced as described previously (13).
Creation of the ip6k⌬ Strain of S. cerevisiae-The YDR017C open reading frame in S. cerevisiae comprises the kcs1 gene that encodes the yeast InsP 6 kinase. Using strain PJ69 -2A of S. cerevisiae, the portion of the kcs1 gene that encodes the C terminus of the kcs1 protein from amino acid 575 to the C-terminal stop codon was replaced by the dominant kan r marker gene using the KanMX4 expression construct as described previously (17). The oligonucleotides used for the gene disruption were: 5Ј-GAAGGAAAAGAAACTCTAATACGACTACAATGGGA-AACCATAATGCATAGGCCACTAGTGGATCTG-3Ј and 5Ј-TAA-GCGCAGCTAAAAGAATATTCATTAGTTCTATCCTTTCTTTTCAGCT-GAAGCTTCGTACGC-3Ј.
Northern Analysis of Wild-type and ip6k⌬ Yeast-10 g of total RNA prepared from exponentially growing yeast was fractionated on a 1% agarose/MOPS-formaldehyde gel and transferred to Hybond Nϩ membranes, according to the manufacturer's instructions (Amersham Pharmacia Biotech). The blot was hybridized with a 1.2-kilobase BamHI-NotI fragment corresponding to the 3Ј-region of the gene. This fragment was obtained by the digestion of the cDNA for the yeast InsP 6 kinase (13).

Creation and Analysis of a ip6k⌬ Strain of S. cerevisiae-We
recently showed that PP-InsP 5 and [PP] 2 -InsP 4 were present in S. cerevisiae (15). One goal of this study was to determine the importance of the yeast InsP 6 kinase to the synthesis of the diphosphorylated inositol phosphates in vivo. The yeast InsP 6 kinase protein is about three times larger in mass than either of the type 1 and type 2 forms of the mammalian InsP 6 kinases (13). The domain of the yeast InsP 6 kinase that is homologous to the smaller, mammalian enzymes resides in a region close to the C terminus. The latter portion of the yeast protein was deleted by disrupting the appropriate 3Ј-region of the kinase gene ( Fig. 1 and see "Experimental Procedures"). To confirm the correct integration of the marker gene, two diagnostic Southern blots were performed. Yeast genomic DNA was digested by SnaBI, and a band of approximately 5.5 kilobases was detected only in the gene-disrupted strain upon hybridization with the kan r gene (data not shown). Coincidentally, the kan r cassette used for the gene disruption is almost identical in size to the InsP 6 kinase gene it replaces (Fig. 1). Thus, a band of approximately 5.5 kilobases was detected only in wild-type

cerevisiae and Northern blot analysis of WT and ip6k⌬ cells.
A is a schematic (drawn to scale) that illustrates the domain within the yeast InsP 6 kinase into which the KanMX4 cassette was inserted. The light gray box represents the portion of the kinase that is homologous to the mammalian type 1 and type 2 InsP 6 kinases. The black box demarcates the area identified by sequence alignments to be a candidate for the active site (13). B is a Northern analysis of the kinase transcript from wild-type (WT) and ip6k⌬ yeast. 10 g of total RNA was loaded in each lane. The blot was hybridized with a probe corresponding to the 3Ј portion of the yeast kinase gene that was replaced by the KanMX4 cassette in the ip6k⌬ strain (see under "Experimental Procedures"). The arrows indicate the migration of the 25 S and 18 S ribosomal RNA. C shows the results of ethidium bromide staining of the gel used in B to check equivalence of loading. cells upon hybridization with a probe corresponding to the deleted region of the InsP 6 kinase gene (data not shown). The success of the gene disruption was also verified by Northern analysis of wild-type and mutant yeast (Fig. 1). The genedisrupted strain is hereafter designated ip6k⌬.
We investigated the consequences of the gene disruption upon the inositol phosphate profile of wild-type and ip6k⌬ cells. Levels of PP-InsP 5 in ip6k⌬ cells were Ͼ80% lower than those of wild-type cells (Table I). This result demonstrates that this kinase is quantitatively important for the expression of PP-InsP 5 synthase activity in vivo. Nevertheless, our results indicate that there must be an alternative, hitherto unrecognized, pathway for PP-InsP 5 synthesis in yeast. Levels of [PP] 2 -InsP 4 were 60% lower in the ip6k⌬ cells compared with the wild type (Table I). This result indicates that the InsP 6 kinase plays only a partial role in the pathway of [PP] 2 -InsP 4 synthesis. Note also that the levels of Ins(1,3,4,5,6)P 5 in the ip6k⌬ cells were 80% lower than those of wild-type cells (Table I). Thus, in yeast, there is an unexpected link between the metabolism of PP-InsP 5 and Ins(1,3,4,5,6)P 5 .
The ip6k⌬ cells grew more slowly than the wild-type cells at 23 and 30°C (Fig. 2). At 37°C, the ip6k⌬ cells did not grow at all (Fig. 2). Prior to the identification of the yeast InsP 6 kinase (13), the gene that encodes this protein was known as kcs1 (21). In that earlier study, the deletion of the kcs1/InsP 6 kinase gene from S. cerevisiae yielded no apparent growth phenotype (21), possibly because that strain of yeast had a different genetic background from the strain that we have used.
Nuclear mRNA Export Is Not Impaired in ip6k⌬ Cells-We (15) and others (22) have previously shown that drastic impairment of the pathway of InsP 6 synthesis in S. cerevisiae is accompanied by a decreased efficiency in the rate of export of mRNA from the nucleus. We (15) did point out, however, that mRNA export may be regulated by metabolites of InsP 6 (such as PP-InsP 5 and [PP] 2 -InsP 4 ) rather than by InsP 6 itself. Indeed, we previously showed that one particular gene-disrupted strain of yeast (ipmk⌬), which displayed an impaired mRNA export phenotype and decreased synthesis of InsP 6 , also had 60 -80% lower levels of PP-InsP 5 and [PP] 2 -InsP 4 (15). The ip6k⌬ cells gave us the first opportunity to study whether mRNA export was affected by quantitatively similar changes in levels of PP-InsP 5 and [PP] 2 -InsP 4 , under conditions where InsP 6 levels were not significantly affected (Table I). We measured mRNA export as described previously (15) but found no significant difference between wild-type and ip6k⌬ cells (data not shown). This negative result redirected our efforts to the hypothesis (see the Introduction) that diphosphoinositol polyphosphates regulate protein trafficking.
Altered Vacuolar Morphology in ip6k⌬ Cells-In yeast cells, several different vesicle transport pathways converge upon the vacuole (23). This organelle receives endocytic traffic from the cell surface as well as biosynthetic traffic from the Golgi apparatus (23,24). Thus, we studied the effect of the deletion of the InsP 6 kinase upon vacuole morphology. The luminal interior of yeast vacuoles was identified by incubating cells with membrane-permeable carboxy-DCFDA (25). Once the probe has entered the vacuoles, the acetate groups are hydrolyzed by nonspecific esterases, forming the less membrane permeable and fluorescent carboxy-DCF (25). Fluorescence detection of the vacuolar space in wild-type cells revealed the usual complement (26) of one large vacuole (Fig. 3). In contrast, the ip6k⌬ strain contained several smaller, fragmented vacuoles (Fig. 3). This altered vacuolar morphology may reflect defects in the fusion into vacuoles of small vesicles derived from endocytosis or from the trans-Golgi network. The intensity of staining is also decreased in the null strain (Fig. 3), possibly because of less active intraluminal esterase activity in the ip6k⌬ cells.
Substrate Specificity of the Yeast InsP 6 Kinase-Could the lower levels of [PP] 2 -InsP 4 in the ip6k⌬ yeast cells (Table I) reflect an unexpected ability of the yeast InsP 6 kinase to phosphorylate PP-InsP 5 ? To examine this question, we compared the rates of phosphorylation of both InsP 6 and PP-InsP 5 (Fig. 4C), and in any case, even this slow reaction required 5-fold higher concentrations of enzyme than were used for the assays of InsP 6 phosphorylation (Fig. 4A). Furthermore, levels of InsP 6 in S. cerevisiae are 180-fold higher than those of PP-InsP 5 (Table I). These data indicate that the yeast InsP 6 kinase does not provide an efficient route for [PP] 2 -InsP 4 synthesis in vivo. We must therefore search for a separate yeast PP-InsP 5 kinase with greater catalytic efficiency that can account for [PP] 2 -InsP 4 synthesis in vivo.
We also found that the yeast InsP 6 kinase phosphorylated Ins(1,3,4,5,6)P 5 (Fig. 4B). The V max values for InsP 5 and InsP 6 were each approximately 2 mol/mg/min. The affinity of the enzyme for InsP 6 (mean K m ϭ 3.3 M) was about 3-fold less than the affinity for InsP 5 (mean K m ϭ 1.2 M). This is the first time an enzyme with PP-InsP 4 synthase activity has been identified. In the case of S. cerevisiae, cellular levels of InsP 6 are 50-fold higher than those of InsP 5 (Table I), so we would not anticipate substantial phosphorylation of InsP 5 by this kinase in intact yeast cells. However, in mammalian cells, levels of Ins(1,3,4,5,6)P 5 and InsP 6 are very similar to each other (27). We therefore next investigated whether either of the two mam- Phosphorylation of InsP 5 by the Mammalian Type 1 InsP 6 Kinase-The type 1 InsP 6 kinase was expressed in E. coli as a His-tagged protein and purified using Talon resin (Fig. 5). The enzyme was incubated with trace amounts of [ 3 H]InsP 6 and an ATP regeneration system (see "Experimental Procedures"). The InsP 6 was found to be completely phosphorylated to PP-[ 3 H]InsP 5 ( Fig. 6A and Ref. 13). No [PP] 2 -InsP 4 was formed (data not shown and see Ref. 13). Several other inositol polyphosphates, namely inositol 1,4-bisphosphate, inositol 1,4,5-trisphosphate, and inositol 1,3,4,5-tetrakisphosphate, have also previously been found not to be significant substrates for this enzyme (13). However, Ins(1,3,4,5,6)P 5 was not previously tested as a substrate. We now studied this issue, in view of the observation that the yeast kinase phosphorylated both InsP 6 and Ins(1,3,4,5,6)P 5 (Fig. 4).
InsP 5 Is Converted to a Novel Inositol Polyphosphate by Mammalian Type 1 InsP 6 Kinase-Another unexpected result to emerge from the studies of InsP 5 phosphorylation by the type 1 mammalian kinase was the accumulation of a novel product (Fig. 6B, peak X), at a rate that was approximately 10% of the rate of accumulation of PP-InsP 4 (Fig. 6B, inset). A comparison of panels A and B in Fig. 6 also shows that the elution position of peak X was between those of InsP 6 and PP-InsP 5 .
We considered the possibility that X represents one of two different PP-InsP 4 isomers, each formed by direct phosphorylation of InsP 5 . This option seemed unlikely, because in our HPLC experiments, peak X eluted 10 min later than did PP-InsP 4 (Fig. 6B); this HPLC system can only resolve isomers of diphosphoinositol polyphosphates by 1-2 min even when a considerably more shallow elution gradient is employed (5). Furthermore, when the kinase was separately incubated with PP-[ 3 H]InsP 4 , approximately 50% of this substrate was phosphorylated to peak X (Fig. 7). In other words PP-InsP 4 , and not InsP 5 , is the immediate precursor of peak X. We next considered whether peak X might be another isomer of PP-InsP 5 . This explanation also seems an unlikely possibility, because the synthesis of PP-InsP 5 from PP-InsP 4 would not involve the formation of a diphosphate group. Nevertheless, we checked the nature of peak X by using, as a diagnostic tool, one of the several hDIPP isoforms that we have cloned (hDIPP2␣) (11). The hDIPP2␣ enzyme specifically removes ␤-phosphates from diphosphorylated inositol polyphosphates; hDIPP2␣ does not hydrolyze monoester phosphates (11). When hDIPP2␣ was incubated with a mixture of PP-InsP 4 and peak X, both were completely dephosphorylated back to InsP 5 (Fig. 8). Thus, peak X cannot be PP-InsP 5 , because the latter is dephosphorylated to InsP 6 by hDIPP2␣ (11).
We therefore propose that peak X may be a hitherto unknown, doubly diphosphorylated derivative of InsP 5 (Fig. 9).
Although PP-InsP 4 is a substrate for the type-1 kinase, this enzyme was very ineffective at further phosphorylating the PP-InsP 5 that was formed from InsP 6 (Fig. 4, A and C). The difference between PP-InsP 5 and PP-InsP 4 lies in the latter having a 2-OH instead of a 2-phosphate group, and this clearly has a substantial impact upon substrate specificity.
Comparisons between the Type 1 and Type 2 Mammalian InsP 6 Kinases-The amino acid sequences of the mammalian type 1 and type 2 InsP 6 kinases are about 60% similar (13). We therefore compared the activities of these two enzymes in more detail. We found that the purified, recombinant type 2 kinase (Fig. 5) was able to phosphorylate both InsP 5 and InsP 6 (Table

FIG. 8. Dephosphorylation of PP-[ 3 H]InsP 4 and peak X by hDIPP2␣.
A mixture of PP-[ 3 H]InsP 4 and 3 H-labeled peak X was prepared as described in the legend to Fig. 7, except that assays were quenched by exposure of the reaction tubes to 100°C for 3 min. The tubes were cooled on ice and then divided into two equal aliquots which received either vehicle (open circles) or 1.9 g of purified hDIPP2␣ (closed circles). Reaction tubes were incubated for a further 2 h at 37°C, and these assays were finally acid-quenched and analyzed by HPLC as described under "Experimental Procedures." The upper panel shows the profile of 3 H-labeled material. The lower panel shows the elution position of a [ 14 C]InsP 5 standard, which was determined in a parallel HPLC run. These data are representative of three experiments. II). In kinetic experiments, we established that V max values for both Ins(1,3,4,5,6)P 5 and InsP 6 were very similar (Table II). The affinity of the type 2 kinase for Ins(1,3,4,5,6)P 5 (K m ϭ 8.4 M, Table II) was approximately 20-fold lower than the affinity for InsP 6 (K m ϭ 0.43 M, Table II). These data therefore identify an important difference between the two kinases, namely, that Ins(1,3,4,5,6)P 5 will compete with InsP 6 much less effectively for phosphorylation by the type 2 kinase, compared with the type 1 enzyme.
We also compared the rates at which the two kinases phosphorylated PP-InsP 4 to [PP] 2 -InsP 3 . We could not prepare sufficient mass amounts of PP-InsP 4 to ascertain K m and V max values. Instead we compared rates of PP-InsP 4 phosphorylation under first order conditions. The activity of the type 2 kinase toward PP-InsP 4 was approximately 7-fold lower than the activity of the type 1 kinase (Table III). The two enzymes showed only a 2-fold difference in activity toward Ins(1,3,4,5,6)P 5 (Table III). DISCUSSION In this study we have obtained important new information concerning the physiological significance of the InsP 6 kinase family. We have discovered that Ins(1,3,4,5,6)P 5 can also be phosphorylated by these enzymes (Fig. 9), and we have obtained data indicating that the yeast InsP 6 kinase plays an important role in vacuole biogenesis (Fig. 3).
Ins(1,3,4,5,6)P 5 and PP-InsP 4 have previously been allocated to a metabolic pool that is different from the pool in which InsP 6 , PP-InsP 5 , and [PP] 2 -InsP 4 are contained (7). It would, therefore, not have been surprising if the turnover of these two metabolically distinct groups of compounds had turned out to be independently regulated and also functionally discrete. As a result of the work described in the current study, we can now appreciate for the first time just how interdependent are these metabolic cycles, in mammalian cells at least (Fig. 9). Levels of InsP 6 and Ins(1,3,4,5,6)P 5 in mammalian cells range from 15 to 50 M (27), and in some cells Ins(1,3,4,5,6)P 5 levels are 4-fold higher than those of InsP 6 (7). This information, together with the new data on the kinetic parameters for Ins(1,3,4,5,6)P 5 and InsP 6 phosphorylation by mammalian type 1 InsP 6 kinase (Table II), indicates that both of these substrates will compete for phosphorylation by this enzyme in vivo. If the size of the metabolic reservoir of either substrate is modified, as happens for example during passage through the cell cycle (30,31), the metabolism of all of the diphosphorylated compounds will be affected. From this intertwined metabolism, we can further conclude that the physiological activities of all of the diphosphorylated inositol phosphates are likely to be closely linked.
Our studies also provide new insight into the structural determinants of InsP 6 kinase specificity. With regards to the mammalian type 1 InsP 6 kinase, the V max and K m values were similar for phosphorylation of both Ins(1,3,4,5,6)P 5 and InsP 6 (Table II). Thus, for these two substrates, there is only a minor impact on the catalytic activity of this enzyme when a phosphate group substitutes for a hydroxyl group at the 2-position on the inositol ring. On the other hand, PP-InsP 4 can be further phosphorylated much more readily than can PP-InsP 5 (Figs. 6 and 7). The difference between PP-InsP 4 and PP-InsP 5 is that the latter has a 2-phosphate, which in this context imposes an important constraint upon catalytic activity. Thus, the impact of the 2-phosphate group on enzyme activity is critically dependent upon the nature of the substrate. Note also that two very weak substrates for the kinases are Ins(3,4,5,6)P 4 (Table  III) and inositol 1,3,4,5-tetrakisphosphate (13), so phosphate groups at both the 1-and 6-positions are also very important to substrate recognition.
Our identification of a novel diphosphorylated inositol phosphate ( Fig. 6, peak X) may also provide an unexpected new direction for future research into this area. This compound, tentatively identified as [PP] 2 -InsP 3 , was synthesized by both mammalian InsP 6 kinases (Table III), and by the yeast InsP 6 kinase (data not shown), suggesting that the catalytic site of these enzymes has not had some flexibility in the positions on the inositol ring at which the diphosphate groups are added. The mammalian type 1 kinase had the most active [PP] 2 -InsP 3 synthase activity (Table III). It would therefore be useful to ascertain whether [PP] 2 -InsP 3 can be detected in vivo. However, the identification of any of the diphosphorylated inositol phosphates inside intact cells is problematic, because their steady-state levels are so low. Indeed, these compounds escaped detection for several years, even after the discovery of Ins(1,3,4,5,6)P 5 and InsP 6 inside animal cells (32,33 (7), primary cultured rat hepatocytes (34), or hamster DDT 1 MF-2 smooth muscle cells (35). However, this search has so far only involved a small number of cell types. Additionally, the metabolism of the diphosphorylated inositol phosphates is carefully regulated (1), and so we may not have incubated cells under the appropriate conditions. By analogy, it is worth noting that it was some years after the identification of PtdIns(4,5)P 2 before the more minor lipids such as PtdIns(3,4,5)P 3 and PtdIns(3,5)P 2 were detected and found to be significant (36).
Our studies have also provided new insights into the activities of the yeast InsP 6 kinase in vivo. The persistence of significant levels of diphosphorylated inositol phosphates in ip6k⌬ cells (Table I) indicates that the InsP 6 kinase we have studied plays a major, but not an exclusive, role in the synthesis of this group of compounds in vivo. Thus, we need to search for an independent (albeit more minor) metabolic pathway that yeast can use to synthesize PP-InsP 5 . Moreover, this alternative pathway can sustain a certain amount of [PP] 2 -InsP 4 synthesis ( Table I). The newly discovered competition of Ins(1,3,4,5,6)P 5 and InsP 6 for phosphorylation by the InsP 6 kinase, also speaks to a previously puzzling observation that cellular levels of PP-InsP 4 increased dramatically in yeast when the Ins(1,3,4,5,6)P 5 2-kinase activity was deleted (22). Because the elimination of the Ins(1,3,4,5,6)P 5 2-kinase was accompanied by loss of cellular InsP 6 (22), we can now appreciate that this would have removed an inhibitory constraint upon the conversion of Ins(1,3,4,5,6)P 5 to PP-InsP 4 . Another interesting phenomeonon in the ip6k⌬ cells was their dramatically (approximately 80%) reduced levels of Ins(1,3,4,5,6)P 5 compared with wild-type cells (Table I). Understanding the molecular mechanisms that link Ins(1,3,4,5,6)P 5 metabolism to InsP 6 kinase activity will be an interesting topic for future studies. At the very least, we should be cognizant that the deletion of the InsP 6 kinase has unexpected repercussions that may contribute to the growth-impaired phenotype of ip6k⌬ cells.
In addition to the decrease in PP-InsP 5 levels (Table I), the ip6k⌬ yeast contain abnormally small and fragmented vesicles (Fig. 3). Yeast vacuoles are homologues of mammalian lysosomes, possessing acidic interiors where protein degradation takes place (23). The vacuoles derive from fusion of cytoplasmderived vesicles as well as clathrin-coated vesicles associated with the endocytic apparatus or the trans-Golgi network (23,24). PP-InsP 5 binds tightly to adaptor proteins that help assemble both clathrin-coated and non-clathrin-coated vesicles (2)(3)(4)(5). Thus, the altered vacuolar morphology in the ip6k⌬ yeast may reflect some abnormality in the inositol polyphosphate-dependent assembly of those vesicles that normally form vacuoles.
By determining the structural and functional relationships that exist between proteins from different organisms, we can gain insight into their evolutionary origins and physiological roles. Clearly, the catalytic activities of the yeast and mammalian InsP 6 kinases are all rather similar, suggesting that there has been evolutionary conservation of their active site domains. The ability of Ins(1,3,4,5,6)P 5 to compete with InsP 6 for phosphorylation by these kinases does differ between the type 1 and type 2 mammalian kinases. This is suggestive of some divergence in the function of these two proteins. This conclusion is pertinent to understanding the physiological consequences that might arise from tissue-dependent differences in the ex-pression of the type 1 and 2 isoforms, as has been described in the rat, for example (13). Overall, the dual specificity of the type-1 kinase toward Ins(1,3,4,5,6)P 5 and InsP 6 has uncovered an unexpectedly close metabolic and functional linkage between the turnover of all of the diphosphorylated inositol phosphates.