The pathway for the production of inositol hexakisphosphate in human cells.

The yeast and Drosophila pathways leading to the production of inositol hexakisphosphate (InsP(6)) have been elucidated recently. The in vivo pathway in humans has been assumed to be similar. Here we show that overexpression of Ins(1,3,4)P(3) 5/6-kinase in human cell lines results in an increase of inositol tetrakisphosphate (InsP(4)) isomers, inositol pentakisphosphate (InsP(5)) and InsP(6), whereas its depletion by RNA interference decreases the amounts of these inositol phosphates. Expression of Ins(1,3,4,6)P(4) 5-kinase does not increase the amount of InsP(5) and InsP(6), although its depletion does block InsP(5) and InsP(6) production, showing that it is necessary for production of InsP(5) and InsP(6). Expression of Ins(1,3,4,5,6)P(5) 2-kinase increases the amount of InsP(6) by depleting the InsP(5) in the cell, and depletion of 2-kinase decreases the amount of InsP(6) and causes an increase in InsP(5). These results are consistent with a pathway that produces InsP(6) through the sequential action of Ins(1,3,4)P(3) 5/6-kinase, Ins(1,3,4,6)P(4) 5-kinase, and Ins(1,3,4,5,6)P5 2-kinase to convert Ins(1,3,4)P(3) to InsP(6). Furthermore, the evidence implicates 5/6-kinase as the rate-limiting enzyme in this pathway.

Ins(1,2,3,4,5,6)P 6 (InsP 6 ) 1 has been implicated in many cellular processes. It is required for mRNA export from the nucleus in yeast (1) and human cells (2). InsP 6 binds to the clathrin assembly proteins AP2 and AP180 (3,4) and inhibits clathrin cage assembly in vitro (5,6). InsP 6 inhibits serine and threonine protein phosphatases, which are thought to regulate L type Ca 2ϩ channels in pancreatic islet cells (7). Nonhomologous DNA end joining of double strand breaks is stimulated by InsP 6 through its binding to the Ku70/80 subunits of DNA-PK (8,9). Most recently, InsP 6 has been suggested to stimulate endocytosis, possibly by the activation of protein kinase C and inhibition of synaptojanin (10). The many roles for InsP 6 necessitates an understanding of the pathway leading to its production.
An alternate pathway to the one discussed above was proposed when the yeast pathway was discovered through genetic screens (Fig. 1B). In yeast, Ins(1,4,5)P 3 is converted directly to InsP 6 by the sequential action of two proteins: Ipk2, which produces InsP 5 in a two-step phosphorylation of Ins(1,4,5)P 3 , first to Ins(1,4,5,6)P 4 and then to InsP 5 ; and Ipk1, which produces InsP 6 (1,15). This pathway differs from that proposed by Menniti et al. (12) in that no isomerization of Ins(1,4,5)P 3 to Ins(1,3,4)P 3 is required and in that the intermediate InsP 4 isomer is not Ins(1,3,4,6)P 4 but Ins(1,4,5,6)P 4 (15). Deletion of IPK2 causes an increase in Ins(1,4,5)P 3 , whereas loss of IPK1 causes a loss of InsP 6 and an accumulation of InsP 5 , proving that there is no other pathway to go from InsP 3 to InsP 6 in yeast. In addition, yeast cells do not possess a 5/6-kinase, nor has an Ins(1,3,4)P 3 isomer been seen in metabolically labeled yeast extracts. The Drosophila and Arabidopsis homologs of IPK2 have been cloned and can complement the yeast Ipk2 deletion mutant (16,17). The Drosophila and Arabidopsis homologs first produce Ins(1,4,5,6)P 4 from Ins(1,4,5)P 3 , and then InsP 5 from this InsP 4 isomer, as does yeast. Interestingly, when the authors were determining the activities of the Drosophila and Arabidopsis homologs of Ipk2, they found a significant 5-kinase activity on Ins(1,3,4,6)P 4 . This activity is necessary for the pathway proposed in human cells. Because the yeast pathway works through Ins(1,4,5,6)P 4 , which is already phosphorylated at the D-5 position, they suggest that this activity is not relevant for the production of InsP 6 in these organisms.

MATERIALS AND METHODS
All chemicals were reagent grade or better. Restriction endonucleases, DNA-modifying enzymes, and general reagents were from Amersham Biochemicals, Roche Applied Science, Fisher, Invitrogen, New England Biolabs, Promega Corp., Sigma, and Stratagene unless stated otherwise. PCR was performed using the Pfu DNA polymerase per the protocol from Stratagene. Oligonucleotide synthesis and DNA sequencing were performed by the Protein and Nucleic Acid Chemistry Laboratory, Washington University, St. Louis, MO. Acrylamide solution, Bio-Safe Coomassie Blue stain, and the Bradford protein assay kit used for protein work were purchased from Bio-Rad. A SuperSignal West Pico kit used for detection of Western transfer blots was from Pierce. Radiolabeled inositol phosphates H]InsP 6 were purchased from PerkinElmer Life Sciences and Amersham Biosciences.
Strains, Plasmids, and Growth Conditions-Methods for Escherichia coli growth and selection were described previously (22,23). E. coli strain XL-1Blue (Stratagene) was used as the bacterial host for all plasmids unless stated otherwise. Bacterial strains were cultured in LB (1% tryptone, 0.5% yeast extract, 1% NaCl, pH 7.4) medium supple- mented with 100 g/ml ampicillin where appropriate and were transformed by standard methods (22,23). All bacterial strains were grown at 37°C.
Western Blot Analysis-Specified tissue culture cells were treated with trypsin, washed with phosphate-buffered saline, and resuspended in 25 mM HEPES, pH 7.5, 3 mM MgCl 2 , 150 mM NaCl, 1 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, 1 mM ATP, and proteinase inhibitors (Complete Mini EDTA-free, Roche Applied Science). Cells were lysed by two freeze-thaw cycles in an ethanol-dry ice bath, and particulate debris was removed by centrifugation at 10,000 ϫ g in an Eppendorf centrifuge at 4°C. The protein concentration of the clarified lysate was determined using the Bradford assay (Bio-Rad) as per the manufacturer's protocol. Samples (5 g) were loaded onto a 10% gel for SDS-PAGE and subsequently electroblotted onto polyvinylidene difluoride membranes (Immobilon-P, Millipore). For detection of 5-kinase overexpression, anti-Myc antibody (Cell Signaling) was used, and for a loading control, anti-AP1 antibody was used. The appropriate horseradish peroxidase-conjugated secondary antibody and the SuperSignal West Pico Chemiluminescent Substrate (Pierce) were used to visualize the appropriate band.
5-Kinase Enzyme Activity Assay-For each sample, the clarified cellular lysate was prepared as described above. Enzymatic activity was determined as described previously (21) and summarized below. Enzyme was added to 50 mM HEPES, pH 7.2, 100 mM KCl, 100 g/ml bovine serum albumin, 8 mM MgCl 2 , 5 mM ATP, and 1,000 -2,000 cpm of [ 3 H]Ins(1,3,4,6) to a total reaction volume of 50 l at 37°C for the desired times. The reaction was stopped by the addition of 1 ml of H 2 O, and the sample was loaded onto a 500-l Dowex column (AG 1-X8 formate, mesh 200 -400, Bio-Rad) equilibrated in water. The column was washed eight times with 1 ml of 1 M ammonium formate and 0.1 M formic acid to elute the substrate. The product was eluted with 2 ml of 2 M ammonium formate and 0.1 M formic acid and counted in a liquid scintillation counter.
Construction of Stably Transfected Cell Lines Overexpressing Inositol Phosphate Kinases--HEK-293 cells expressing 5/6-kinase were constructed as reported previously (28). The 2-kinase-expressing cells and 5-kinase-expressing cells were constructed using the tetracycline-inducible vector pcDNA4/TO (Invitrogen). The 2-kinase gene used previously for expression in Sf9 cells (29) was subcloned into the pcDNA4/TO vector using the BamHI and NotI restriction sites; the resulting construct has 2-kinase preceded by a FLAG tag. The human 5-kinase cDNA was subcloned from the plasmid host pTrcHisA (Invitrogen) (21) by PCR with 5Ј-CATGCCATGGCAACAGAGCCACCATCCCCCCTC-3Ј and 5Ј-CGGGGTACCAATTGTCTAAAATACTTCGAAGTAC-3Ј primers. The resulting PCR product has 5-prime NcoI and 3-prime KpnI restriction sites and is inserted into the NcoI/KpnI sites of the plasmid pTrcHis2B (Invitrogen) in-frame with a 3-prime Myc tag. A second PCR with 5Ј-CGGGGTACCTCCGTTATGGCAACAGAGCCACCATCCCC-C-3Ј and 5Ј-CGCGGATCCTCAATTCAGATCCTCTTCTGAGATGAG-3Ј primers using the pTrcHis2B construct of the human 5-kinase generated a product with 5-prime KpnI and 3-prime BamHI restriction sites flanking the human 5-kinase cDNA in-frame with a 3-prime Myc tag; this construct was ligated into the mammalian expression plasmid pcDNA4/TO. All PCR products were verified by DNA sequencing.
Stable cell lines were constructed by transfecting these constructs into HEK-293 TRex cells (Invitrogen) using Lipofectamine 2000 (Invitrogen), plating serial dilutions of the transfected cells, and selecting them in Dulbecco's modified Eagle's medium with 10% fetal bovine serum (Tet system approved, Clontech), 2 mM glutamine, 100 units/ml penicillin G, 10 g/ml streptomycin, 0.25 g/ml amphotericin B, 5 g/ml blasticidin (Invitrogen), and 0.4 mg/ml Zeocin (Invitrogen) until single colonies were obtained. Stable cell lines were maintained in medium containing Dulbecco's modified Eagle's medium with 10% fetal bovine serum (Tet system approved, Clontech), 2 mM glutamine, 5 g/ml blas-ticidin, and 0.3 mg/ml Zeocin. These gene products were induced with 0.2 g/ml tetracycline unless otherwise noted. Expression of the human 5-kinase was verified by Western blot analysis using monoclonal antibodies recognizing the Myc epitope (Cell Signaling) and by enzyme activity assays, whereas expression of 2-kinase was verified by Western blot analysis using monoclonal antibodies recognizing the FLAG epitope (Sigma) of 2-kinase.
Gene Silencing of the Inositol Phosphate Kinases-The preannealed siRNA oligonucleotides used for gene silencing of the inositol phosphate kinases were synthesized by Dharmacon RNA Technologies. The oligonucleotides used for the human 5-kinase were sense 5Ј-GGAUGGAG-UCUCCAGAUUAdTdT-3Ј and antisense 5Ј-AAAUCUGGAGACUCCA-UCCdTdT-3Ј, for the luciferase control were sense 5Ј-CUUACGCUGA-GUACUUCGAdTdT-3Ј and antisense 5Ј-UCGAAGUACUCAGCGUAA-GdTdT-3Ј, and for 2-kinase were sense 5Ј-GAAGACCUCGGAAGAGA-UUdTdT-3Ј and antisense 5Ј-UAUCUCUUCCGAGGUCUUCdTdT-3Ј. Tissue culture cells were grown to 85-95% confluence in 12-well plates with or without tetracycline in standard media and transfected with 1.6 g of siRNA oligonucleotide using Lipofectamine 2000 per the manufacturer's protocol. After 6 h, the cells were treated with trypsin, transferred to 6-well plates, and then grown overnight in 5% CO 2 at 37°C. A second transfection with 4 g of siRNA oligonucleotide was performed using Lipofectamine 2000, and after a 6-h incubation, the cells were transferred to 10-cm plates containing media with or without 10 Ci/ml [ 3 H]inositol and with or without 0.2 g/ml tetracycline. After 3 days, cells were harvested for Western blot analysis and HPLC of soluble inositol phosphates. Densitometry of the bands from the Western blot analysis was performed using the Kodak one-dimensional 3.5 software with the Kodak Image Station 440CF. Construction of the 5/6-kinase RNAi stable cell line was described previously (30).
Analysis of Soluble Inositol Phosphates-Cells were grown in complete media containing 10 Ci/ml [ 3 H]inositol for at least 72 h with or without tetracycline induction. Cells were lysed in methanol and 0.5 N HCl (2:1) and extracted with chloroform. The aqueous phase was separated, dried, and resuspended in water. 32 P-Labeled standards were added to each sample. The samples were applied to a 4.6 ϫ 250-mm Adsorbosphere SAX HPLC column and eluted with a linear gradient of 0 -1.0 M ammonium phosphate, pH 3.5, over 120 min at a flow rate of 1 ml/min, using a modification of the method of Hughes et al. (31). For 2-kinase overexpression or RNAi lines a Whatman Partisphere SAX strong anion exchange column (4.6 ϫ 125 mm) was used running a 30-min gradient from 0 to 1.7 M ammonium phosphate, pH 3.5, followed by a 30-min step of 1.7 M ammonium phosphate, pH 3.5. Radioactivity was measured using the inline detector ␤-RAM (IN/US System Inc.), and the identity of the individual inositol phosphates was assigned on the basis of coelution with known standards.

Overexpression of the Inositol 5/6-Kinase Results in Elevated
Levels of InsP 4 , InsP 5 , and InsP 6 -Stable HEK-293 cells expressing a tetracycline-inducible 5/6-kinase were labeled with [ 3 H]inositol for 3-4 days, their soluble inositol phosphates extracted, and equal counts of soluble inositol phosphates were separated on an Adsorbosphere SAX HPLC column. Two sets of labeling experiments were performed, one of log phase growing cells (Fig. 2, A and B) and another of confluent cells (Fig. 2, C  and D). 5/6-Kinase-expressing lines were either uninduced (Fig. 2, A and C) or induced with 0.2 g/ml tetracycline (Fig. 2,  B and D). Induced cells had elevated levels of Ins(1,3,4,6)P 4 (7-fold and 2-fold increase in log phase and confluently growing cells, respectively), and they also showed elevated levels of InsP 5 (3-fold and 5-fold, respectively) and InsP 6 (1.7-fold and 2-fold, respectively). This suggests that the product of the 5/6kinase reaction, Ins(1,3,4,6)P 4 , is phosphorylated to InsP 5 , which is then phosphorylated to InsP 6 . A large amount of Ins(3,4,5,6)P 4 is seen in labeling of confluently growing 5/6kinase-overexpressing cell lines (Fig. 2, C and D), which is not seen consistently in labeling of log phase growing cells (Fig. 2,  A and B). Ins(3,4,5,6)P 4 is thought to arise from the action of a 1-phosphatase on InsP 5 (32). It therefore mirrors the rise in InsP 5 .
Cells Expressing the 5-Kinase Show No Increase in InsP 5 or InsP 6 Levels-HEK-293 cells were stably transfected with the cDNA encoding 5-kinase under the regulation of a tetracyclineinducible system (T-REx, Invitrogen) as described under "Materials and Methods." In the presence of tetracycline, a 10-fold increase in the 5-kinase enzymatic activity was observed, whereas without tetracycline, the activity is similar to that of the control HEK-293 cells stably transfected with vector DNA with or without tetracycline (Fig. 4A). Induction of 5-kinase was confirmed by the presence of Myc-tagged protein only in the presence of tetracycline in the stable cell line, whereas no recombinant 5-kinase protein was observed in vector cells (Fig.  4B) or in the uninduced stable cell line.
To determine the effect of overexpression of 5-kinase on soluble inositol phosphates in vivo, 5-kinase stable HEK-293 cells were labeled with 10 Ci/ml [ 3 H]inositol for 3 days in the presence or absence of tetracycline. Soluble inositol phosphates were extracted and resolved by Adsorbosphere SAX HPLC. Interestingly, unlike the 5/6-kinase results, overexpression of 5-kinase does not increase the level of InsP 5 , the product of the 5-kinase enzyme reaction, or InsP 6 (Fig. 4D) compared with uninduced cells (Fig. 4C). The soluble inositol phosphate pro- files are identical to those of the control vector cells with or without tetracycline (data not shown), and the lipid inositol phosphate profiles are not altered by the overexpression of 5-kinase (data not shown). These experiments were performed multiple times with the same results. Hence, increased 5-kinase activity does not alter the level of InsP 5 and InsP 6 in HEK-293 cells, suggesting that production of InsP 5 is limited by the availability of the substrate of 5-kinase, Ins(1,3,4,6)P 4 .
Gene Silencing of the 5-Kinase Results in Decreased InsP 5 and InsP 6 Levels-Because the overexpression experiments did not produce a change in higher inositol phosphates, we performed gene silencing experiments using synthetic siRNA oligonucleotides to confirm that the 5-kinase protein was necessary for synthesis of InsP 5 and InsP 6 in vivo. HEK-293 cells were transfected with no siRNA oligonucleotide, siRNA oligo-nucleotide directed against the luciferase gene, or siRNA oligonucleotide directed against the 5-kinase gene. To determine the magnitude of gene silencing, we used the overexpressing 5-kinase HEK-293 stable cell line because we do not have antibodies that can consistently detect endogenous levels of 5-kinase. When 5-kinase stable cells induced with tetracycline were transfected with a siRNA oligonucleotide directed against the 5-kinase gene, the overexpressed 5-kinase protein is decreased significantly on Western blot analysis compared with either no siRNA oligonucleotide or siRNA oligonucleotide directed against the luciferase gene (Fig. 5A). To determine the magnitude of gene silencing, we compared the relative optical density of the 5-kinase band on the Western blot with that of the loading control AP1; we observed a ϳ90% decrease in the 5-kinase protein (Fig. 5B). Naive HEK-293 cells were used for labeling experiments so that we could determine the effect of silencing the endogenous 5-kinase. These cells were transfected with siRNA oligonucleotides, labeled with 10 Ci/ml [ 3 H]inositol, the soluble inositol phosphates were extracted, and equal counts were resolved on an Adsorbosphere SAX HPLC column. In cells transfected with 5-kinase siRNA oligonucleotide, the level of InsP 5 was decreased to 29.4% and 33.1% compared with the level observed in cells mock transfected or FIG. 5. Gene silencing of the human 5-kinase. A, Western blot analysis of 5 g of lysate from induced 5-kinase cells transfected with no siRNA oligonucleotide, the luciferase siRNA oligonucleotide, or 5-kinase siRNA oligonucleotide as described under "Materials and Methods." The 5-kinase was visualized with anti-Myc antibody, and the loading control was determined by anti-AP1 antibody. B, relative densitometry of bands in A determined by Kodak Image Station 440CF. The optical density ratio of 5-kinase to AP1 was plotted on the y axis. C-E, HPLC profiles of naive HEK-293 cells transfected with no siRNA oligonucleotide (C), the luciferase siRNA oligonucleotide (D), or 5-kinase siRNA oligonucleotide (E), and labeled with [ 3 H]inositol as described under "Materials and Methods." Soluble inositol phosphates were extracted and separated on an Adsorbosphere SAX HPLC column, and the HPLC chromatograph is shown. The reference locations of isomers of InsP 4 , InsP 5 , and InsP 6 were determined by the addition of known 32 P-labeled standards. The radioactivity (cpm) of each sample was normalized to protein.
transfected with the luciferase siRNA oligonucleotide, respectively, and the level of InsP 6 was decreased to 26.8% and 28.4%, respectively (Fig. 5, C-E). Similar results were observed in 5-kinase stable cells with or without overexpression of the 5-kinase protein and in naive HeLa cells (data not shown). These findings are consistent with 5-kinase functioning in the pathway for the synthesis of InsP 5 . Interestingly, HEK-293 cells transfected with the 5-kinase siRNA oligonucleotide did not accumulate Ins(1,3,4,6)P 4 , the preferred substrate (21); however, in HeLa cells transfected with the 5-kinase siRNA oligonucleotide, the Ins(1,3,4,6)P 4 level increased by 70% (data not shown). Additionally, the peak representing the isomer Ins(3,4,5,6)P 4 decreased in both HEK-293 and HeLa cells transfected with 5-kinase siRNA oligonucleotide, which is expected because this isomer is derived from InsP 5 .
Cells Overexpressing the 2-Kinase Produce InsP 6 by Depleting the Available InsP 5 -To assess further the in vivo pathways to InsP 6 , stable HEK-293 cell lines were constructed with a tetracycline-inducible 2-kinase gene and labeled with [ 3 H]inositol for 4 days in the presence of 0.1 g/ml tetracycline. The soluble inositol phosphates were extracted, and equal counts were resolved on a Partisphere HPLC column as described under "Materials and Methods." Compared with the cells stably transfected with an empty vector (Fig. 6A), the soluble inositol phosphate profile of the cells expressing 2-kinase showed a decrease of InsP 5 (to 20% of vector lines) and concomitant increase of InsP 6 (1.5-fold) (Fig. 6B). This result was consistent from run to run and from multiple clones. In all cases, expression of 2-kinase decreased the amount of InsP 5 . Interestingly, InsP 5 does not seem to be replenished. The results from the 5-kinase-expressing lines would suggest that this is because of a lack of substrate for 5-kinase. These results confirm that 2-kinase is sufficient for production of InsP 6 in vivo.
Cells Transfected with a siRNA to the 2-Kinase Block Production of InsP 6 and Accumulate InsP 5 -2-Kinase is necessary for InsP 6 production, as shown by silencing 2-kinase in naive HEK-293 cells using siRNA oligonucleotides (Fig. 7B). As a control we used oligonucleotides to the luciferase gene (Fig.  7A). When these cell lines were labeled with [ 3 H]inositol and their soluble inositol phosphates purified and separated by HPLC, there was a decrease in InsP 6 relative to the controls by about 50%, confirming that 2-kinase is necessary for production of InsP 6 in vivo. In addition, these lines had a 4-fold increase of InsP 5 . Stable cell lines expressing 2-kinase RNAi and labeled with [ 3 H]inositol showed a decrease of InsP 6 to 30% of controls (data not shown). The increase in InsP 5 and relatively smaller decrease of InsP 6 suggest that cells conserve InsP 6 . During our attempts to produce stable knock-outs of 2-kinase using a RNAi expression construct, we were only able to produce a few stable cell lines, none of which had a complete depletion of InsP 6 (data not shown); we had a similar experience trying to produce stable RNAi cell lines of 5/6-kinase. Feng et al. (2) showed that depleting the higher inositol phosphates by expressing the Salmonella protein, SopB, a phosphatase that breaks down inositol phosphates, caused the cells to ball up and stop dividing. Thus it may be difficult to get a more complete silencing of 2-kinase because the effects are toxic to cells.

DISCUSSION
Here we provide in vivo evidence that the human pathway to the higher inositol phosphates proceeds through the Ins(1,3,4)P 3 isomer via the action of 5/6-kinase to produce Ins(1,3,4,6)P 4 , 5-kinase to produce InsP 5 , and 2-kinase to produce InsP 6  concomitant loss of InsP 5 , whereas depleting 2-kinase results in a decrease in InsP 6 and an accumulation of InsP 5 . These results confirm that the pathway proposed for the production of InsP 6 in rat cells by Menniti et al. (12) is the pathway used for the production of InsP 6 in human cells.
Because overexpression of 5/6-kinase results in an increase in Ins(1,3,4,6)P 4 , InsP 5 , and InsP 6 , whereas overexpression of 5-kinase alone does not result in an increase of InsP 5 , the production of InsP 5 must be limited by the availability of substrate, Ins(1,3,4,6)P 4 . This is supported by expression of 2-kinase, during which InsP 6 is produced at the expense of all of the available InsP 5 , which is not replenished. Therefore, our data suggest that the rate-limiting step in this pathway is the production of Ins(1,3,4,6)P 4 by 5/6-kinase. Overexpression of Ins(1,3,4)P 3 5/6-kinase showed that this protein is sufficient to cause increases in both InsP 5 and InsP 6 . Silencing this gene resulted in an increase in its substrate, Ins(1,3,4)P 3 , and a decrease in its product, Ins(1,3,4,6)P 3 , as expected, and also a decrease in InsP 5 and InsP 6 , showing that 5/6-kinase is also necessary for their production in vivo. The results from the RNAi experiments confirm that the activities described for 5-kinase and 2-kinase in vitro are likewise necessary for production of InsP 6 in vivo.
In yeast, Arabidopsis, and Drosophila, the pathway can operate directly through Ins(1,4,5)P 3 without isomerization to Ins(1,3,4)P 3 , but the evidence argues against such a pathway in human cells. First, Seeds et al. (16) showed that expression of the Drosophila Ipk2 gene from the actin promoter in Drosophila resulted in a 5-fold increase in InsP 5 and InsP 6 , but the expression of 5-kinase, the human homolog of Ipk2, does not alter the levels of InsP 5 and InsP6 in human cells. Thus, in Drosophila Ipk2 is sufficient for production of InsP 5 from Ins(1,4,5)P 3 , whereas in human cell lines 5-kinase is not. Also, silencing of 5/6-kinase should not affect the higher inositol phosphates if its activity were uninvolved in the pathway, but we show that silencing 5/6-kinase blocks higher inositol phosphate production. Furthermore, the expression of Drosophila Ipk2 in an ipk2-null yeast strain restores InsP 5 production, whereas expression of the human 5-kinase in yeast does not (21). Therefore, the human 5-kinase cannot convert Ins(1,4,5)P 3 to InsP 5 . These data show that these organisms use different pathways.
The formation of Ins(3,4,5,6)P 4 seen in these experiments is also relevant to the control of this pathway. Ins(3,4,5,6)P 4 is a potent inhibitor of 5/6-kinase (31). When the supply of InsP 5 is sufficient for InsP 6 production and the metabolic needs of the cell, Ins(3,4,5,6)P 4 builds up and inhibits 5/6-kinase. This would presumably shut down the synthesis of InsP 5 and, therefore, InsP 6 . Another observation seen in these experiments is the ability of the cell to preserve InsP 6 at the expense of InsP 5 . This is seen in the RNAi experiments of 5/6-kinase, and it was seen during numerous 5-kinase siRNA transfections, during which InsP 5 levels are less than InsP 6 . Given the number of functions attributed to InsP 6 , this is not a surprising result. When we attempted to silence 2-kinase and 5/6-kinase using stable transfections of the RNAi constructs, we had difficulty generating 2-kinase and 5/6-kinase RNAi lines, whereas the control RNAi lines were abundant. None of the lines we did generate had complete depletion, suggesting that the loss of InsP 6 is toxic to cells.
We have determined the in vivo pathway for the production of InsP 6 in human cells. Although the pathway we have determined differs from the yeast, Drosophila, and Arabidopsis pathways, it does follow the pathway proposed for rat cells by Menniti et al. (12). Humans may have evolved a pathway that requires 5/6-kinase to produce InsP 6 , possibly for the greater control of the pathway; in support of this, we believe that 5/6-kinase is the rate-limiting enzyme in the human pathway. It is interesting that the activity of 5-kinase on Ins(1,3,4,6)P 4 , which is necessary in the human pathway, has been described for both the Drosophila and Arabidopsis Ipk2 proteins and thus seems to be conserved. What has changed is the predominance of 5/6-kinase in the human pathway. Some questions do remain. Is there a 5/6-kinase in Drosophila? Why does Arabidopsis have three copies of 5/6-kinase if they are unnecessary for production of InsP 6 ?