A molecular basis for inositol polyphosphate synthesis in Drosophila melanogaster.

Metabolism of inositol 1,4,5-trisphosphate (I(1,4,5)P3) results in the production of diverse arrays of inositol polyphosphates (IPs), such as IP4, IP5, IP6) and PP-IP5. Insights into their synthesis in metazoans are reported here through molecular studies in the fruit fly, Drosophila melanogaster. Two I(1,4,5)P3 kinase gene products are implicated in initiating catabolism of these important IP regulators. We find dmIpk2 is a nucleocytoplasmic 6-/3-kinase that converts I(1,4,5)P3 to I(1,3,4,5,6)P5, and harbors 5-kinase activity toward I(1,3,4,6)P4, and dmIP3K is a 3-kinase that converts I(1,4,5)P3 to I(1,3,4,5)P4. To assess their relative roles in the cellular production of IPs we utilized complementation analysis, RNA interference, and overexpression studies. Heterologous expression of dmIpk2, but not dmIP3K, in ipk2 mutant yeast recapitulates phospholipase C-dependent cellular synthesis of IP6. Knockdown of dmIpk2 in Drosophila S2 cells and transgenic flies results in a significant reduction of IP6 levels; whereas depletion of dmIP3K, either alpha or beta isoforms or both, does not decrease IP6 synthesis but instead increases its production, possibly by expanding I(1,4,5)P3 pools. Similarly, knockdown of an I(1,4,5)P3 5-phosphatase results in significant increase in dmIpk2/dmIpk1-dependent IP6 synthesis. IP6 production depends on the I(1,3,4,5,6)P5 2-kinase activity of dmIpk1 and is increased in transgenic flies overexpressing dmIpk2. Our studies reveal that phosphatase and kinase regulation of I(1,4,5)P3 metabolic pools directly impinge on higher IP synthesis, and that the major route of IP6 synthesis depends on the activities of dmIpk2 and dmIpk1, but not dmIP3K, thereby challenging the role of IP3K in the genesis of higher IP messengers.

In metazoans, biochemical studies have suggested that IP 6 is synthesized from I(1,4,5)P 3 via a five-step route of several kinases and a phosphatase as shown in Fig. 1B (2,5,21). Two key differences between the proposed metazoan and defined yeast pathways are: 1) the initiation step in the synthesis of IP 6 occurs through an I(1,4,5)P 3 3-kinase (IP3K) and not Ipk2, and 2) the requirement of an I(1,3,4)P 3 5/6-kinase, orthologs of which have not yet been found in the budding yeast genome. The cloning of plant and metazoan Ipk2 (also referred to as inositol phosphate "multi-kinase"; IPMK) and Ipk1, both of which were found to possess similar activities as their yeast counterparts, has raised the possibility that synthesis of higher IP messengers in metazoans may occur similarly to that in yeast (18,20,(22)(23)(24)(25)(26).
To further probe the synthesis of higher IPs in metazoans at the molecular level we initiated studies in the fruit fly, Drosophila melanogaster. Using a bioinformatics approach, we identified orthologous kinase and phosphatase genes in the Drosophila that may contribute to higher IP synthesis in metazoans. Detailed biochemical analysis was used to characterize the I(1,4,5)P 3 kinases proposed to initiate the first phosphorylation step of higher IP production, and subsequent use of RNA interference in culture cells and transgenic flies provided gene-specific knockdowns of each activity to dissect the IP synthesis pathway. We find that the synthesis of IP 5 and IP 6 in the fruit fly is dependent on Ipk2 and Ipk1, but not IP3K, 5-phosphatase or I(1,3,4)P 3 5/6-kinase activities. Our data provide a novel molecular basis for higher IP production in a metazoan and support a role for Ipk2 as a key gatekeeper for activating this important signaling pathway.

MATERIALS AND METHODS
Plasmid Construction-Drosophila EST SD14726, SD19941 were obtained from Research Genetics and used as templates for amplification of dmIpk2 and dmIP3K␤, respectively. PCR reactions were carried out using the Expand High Fidelity PCR System (Roche Applied Science). The following primers were used: dmIP3K␤: 5Ј-ATA TCC CCG GGA ATG CCG CGG GAC TAT GGC TAC-3Ј(forward) and 5Ј-GCT CTG TTA ACG TCT AGA TTA GGG TTT GCT CTC TTC-3Ј(reverse), dmIpk2: 5Ј-ACG TCT GGA ATT CAG ATG GCC AAG AGT GAT CAG GAG-3Ј (forward) and 5Ј-TGA GCT CGA GTC GAC TCA TCG GTG GAG TAT TGA TTG-3Ј (reverse). PCR products were then cut with the following enzymes and ligated in-frame with the pUNI10 Lox recombination site or into the pGEX-KG GST expression vector. dmIP3K␤: SmaI and XbaI (for pUNI10 used SmaI and HpaI), dmIpk2: EcoRI and SalI. pGEX-KG plasmids were transformed into DH5␣-competent cells. pUNI10 constructs were recombined into a pRS314 host vector containing Lox site, MYC3, TRP1, CEN, and a copper inducible promoter (CUP1). The recombined plasmid was transformed into ipk2⌬ and ipk2⌬ ipk1⌬ yeast strains using standard yeast transformation techniques.
Bacterial Expression of dmIpk2 and dmIP3K␤-For recombinant protein expression, transformed Escherichia coli (DH5␣) were grown at 37°C to an OD 600 of 0.6 and induced with 0.1 mM isopropyl-1-thio-␤-Dgalactopyranoside for 4 h at 30°C. Cells were recovered by centrifugation at 4°C, resuspended in ice-cold 50 mM Tris-HCl pH 7.5, 50 mM KCl, 5 mM dithiothreitol, Complete Mini protease inhibitor mixture (Roche Applied Science) and lysed with four passes through a cell cracker (a high shear fluid processing system for cell rupture, Microfluidics Corp.). Lysates were then cleared by centrifugation at 14,000 ϫ g. The GST fusion proteins were purified over glutathione Sepharose (Amersham Biosciences) according to the manufacturer's instructions. Proteins were eluted from the glutathione Sepharose using 50 mM Tris-HCl pH 8.0, 50 mM NaCl, 5 mM dithiothreitol, and 20 mM glutathione. For antibody production GST-dmIpk2 was cleaved with thrombin protease according to the manufacturer's instructions (Amersham Biosciences). Purified protein was injected into rabbits and antiserum was isolated (BIOSOURCE). dmIpk2 protein was detected using standard Western blotting techniques.
Kinetic Assays-The K m and V max of the enzymatic interaction between dmIpk2 and dmIP3K␤, and various substrates were determined. The following reaction mixture was prepared: 250 mM HEPES (pH 7.5), 250 mM NaCl, 10% glycerol, 0.1% BSA, 2 mM ATP, 100 mM MgCl 2 , 80 cpm/l radiolabeled substrate, 10 ng dmIpk2 or dmIP3K␤, and various concentrations of unlabeled substrate in a 20-l reaction volume. The reaction was stopped by the addition of HPLC buffer or 1.0 M HCl. 10 mM NH 4 H 2 PO 4 (pH 3.5) was then added to the reaction and analyzed by Partisphere strong anion exchange HPLC. The kinase kinetics was calculated from measured changes in the substrate area under the curve (AUC) peak.
S2 cells were propagated in Schneider's Drosophila media (Invitrogen) containing 10% fetal calf serum dialyzed and heat inactivated (Hyclone) and 10 g/ml penicillin and streptomycin. The cells were pelleted and resuspended in Drosophila-SFM (Invitrogen) to a density of 1 ϫ 10 6 cells/ml and plated into 6-well plates (1ϫ 10 6 cells/well). 15 g of dsRNA was added to each well and incubated at room temperature for 1 h. The cells were then supplemented with 2 ml of Schneider's Drosophila media/fetal bovine serum/Penn/Strep and incubated for 4 -6 days at 25°C. In the case that RNAi knockdowns were extended over a period of 15 days, dsRNA-treated cells were counted after 8 days, split, and again treated with dsRNA as described above.
Northern Analysis of S2 Cells-Probes for Northern analysis were made using DNA templates for dsRNA as described above. DNA templates were then incubated with 32 P-labeled dCTP and the randomprimed DNA labeling kit (Amersham Biosciences) according to the manufacturer's instructions. S2 cells were treated with dsRNA for 6 days as described above. 3 ml of cells were then recovered by centrifugation at 4°C. Total RNA was then isolated using the Qiagen RNA purification system according to the manufacturer's instructions. RNA was then separated on a 2% agarose gel and transferred to hybond nitrocellulose paper (Amersham Biosciences). The DNA probe was then hybridized in ExpressHyb hybridization solution according to the manufacturer's instructions (Clontech).
In Vivo Labeling of S2 Cells and Saccharomyces cerevisiae-S2 cells were treated with dsRNA as described above or no dsRNA as a control. On day four or thirteen, [ 3 H]inositol (American Radiolabeled Chemicals) was added to the media to a final concentration of 80 Ci/ml. Cells were then incubated for 2 days at 25°C. On day 6 or 15, the cells were recovered by centrifugation and washed once in Dulbecco's phosphatebuffered saline (Invitrogen Life Technologies, Inc.). Yeast cultures were incubated in minimal media lacking tryptophan and 150 M CuSO 4 for 2 days at 30°C for 2 days. [ 3 H]Inositol was added to a final concentration of 80 Ci/ml. Soluble inositol polyphosphates were harvested and analyzed by HPLC using a Partisphere SAX strong-anion exchange column as described previously (7).
Construction and Analysis of Fly Lines-Df(2L)BCS16 was obtained from the Bloomington stock center. Nle ⌬8 was a kind gift from S. Cohen (27). For dmIpk2 RNAi, we obtained the pWIZ vector from Richard Carthew to construct a transgene containing inverted repeats of a 200-bp region of dmIpk2 separated by a 74-bp intron spacer (28). The following primers were used to generate the dmIpk2 region containing XbaI restriction sites on both ends: 5Ј-CTAGT CTAGA ACGAC TATTC AAGAC TGGCT G-3Ј (forward) and 5Ј-CTACT CTAGA TCATC ATCGG TGGAG TATTG-3Ј (reverse). The PCR product was then subcloned into the pWIZ vector using AvrII and NheI sites and checked for proper orientation. For expression of FLAG-dmIpk2 we cloned the dmIpk2 coding region into pBSIISK with sequence for an N-terminal FLAG epitope inserted (a kind gift from Rick Fehon). The following primers were used to PCR dmIpk2 from the EST containing SmaI and HindIII restriction sites: 5Ј-TCTGG ACCCG GGATG GCCAA GAGTG ATCAGG AG-3Ј (forward) and 5Ј-TGAGC T CGA A AGCTT TCATC GGTGG AGTAT TGATT G-3Ј (reverse). The FLAG epitope and dmIpk2 were then subcloned into pP [UAST]. pWIZ-Ipk2 and pP[UAST]-FLAG-dmIpk2 constructs and a P-element transposase plasmid were injected in w1118 embryos and germ line transformants were selected with the wϩ marker (29). Homozygous wϩ lines were generated with standard balancer chromosomes. For fly labeling, 3-6-day-old adult males were fed a 5% sucrose solution containing 500 Ci/ml [ 3 H]inositol for 4 days. 2 Soluble IPs were extracted as described above.
Salivary Gland Stains-FLAG-dmIpk2 expression was induced by crossing p[FLAG-dmIpk2]/cyo to Gal4 expressing flies under the actin promoter (P[FLAG-dmIpk2]/Actin). Salivary glands were dissected from 3rd instar larvae and fixed in PBS ϩ 0.1% Triton X-100 (PBS-T) and 3.7% paraformaldehyde for fifteen minutes. Glands were then washed in PBS-T and blocked overnight in PBS-T ϩ 5% goat serum. Anti-FLAG M2 monoclonal antibody (Sigma) was added to the blocking buffer and incubated for one hour with agitation. Salivary glands were washed three times with PBS-T. An anti-mouse cy3 secondary antibody (Jackson ImmunoResearch) was incubated with the salivary glands for 1 h. The glands were washed, stained with DAPI (Sigma), and mounted with ProLong Antifade (Molecular Probes). Glands were visualized with a Zeiss Axioscope equipped with Metamorph software.

Identification of Drosophila Genes Involved in Inositol
Polyphosphate Synthesis-In this study, we utilize the tractable fly model system to study mechanisms of higher IP synthesis in metazoans. Initially, our analysis focused on two I(1,4,5)P 3 kinase-dependent IP synthesis models, one derived from genetic yeast studies and one proposed from biochemical characterization of pathways metazoans (Fig. 1A). The initiation steps for each model are distinct and require I(1,4,5)P 3 kinase activities encoded by separate gene products. Ipk2 functions in yeast as a dual specificity kinase that converts I(1,4,5)P 3 to I(1,3,4,5,6)P 5 and in metazoans IP3K phosphorylates I(1,4,5)P 3 to I(1,3,4,5)P 4 . Of interest, the two I(1,4,5)P 3 kinases are members of a small family of IP kinases that share an evolutionary ancestor and common amino acid motifs, but possess divergent substrate specificities (6,8). The second and third steps of the pathway proposed in metazoans re-2 B. Elliot and V. Bankaitis, personal communication.
In order to identify various kinase and phosphatase members of these pathways, we performed in silico searches (www. ncbi.nlm.nih.gov/BLAST/and www.flybase.net/blast/) of the Drosophila genome using hallmark motifs for each gene product. Four predicted gene products were found that harbor the IP kinase motif PXXXDXKXG: CG13688 (PcvmDvKmG), CG4026 (PcvmDiKmG), CG1630 (PcvmDcKvG), and CG10082 (PcilDlKmG). Sequence analysis of CG13688 (designated here as dmIpk2) revealed that it is most similar to Ipk2 homologs from other species (Fig. 1B). dmIpk2 is a 310 amino acid polypeptide encoded by a single exon with 37% similarity and 24% identity to the human Ipk2. CG10082 (designated dmIP6K) shared the highest degree of homology with mammalian IP 6 kinases (also referred to as diphosphoryl inositol synthetases). IP6Ks use IP 5 and IP 6 as substrates to generate inositol pyrophosphate species. The putative dmIP6K open reading frame codes for an 893 amino acid protein and contains 36% similarity to human IP6K2 (GenBank TM accession number AF177145). Further studies will be required to elucidate whether this putative gene translates into a functional IP 6 kinase as it will not be discussed here. The other two PXXX-DXKXG-containing proteins (CG4026 and CG1630) share 59% similarity and are highly homologous to mammalian IP3K (Fig.  1B). Three different isoforms of IP3K (A-C) are described in higher eukaryotes that differ in sequence, tissue and subcellular localization. CG4026 (dmIP3K␣, originally annotated as dmIP3K1) was previously identified as an IP3K that confers resistance to oxidative stress when overexpressed in adult flies (31). CG1630 (dmIP3K␤, originally annotated as dmIP3K2) is most similar to the human IP3K B (45% similar and 37% identical). The putative IP3Ks do not contain calmodulin-binding sites at their N terminus as are found in mammalian I(1,4,5)P 3 3-kinases.
Type I enzymes in the 5-phosphatase family have high catalytic efficiency as D5 specific phosphatases against I(1,4,5)P 3 and I(1,3,4,5)P 4 , and contain a highly conserved motif (r/n)- (30). This pattern motif was used to search Drosophila genome and identified a predicted gene product, CG31107 (designated dm5PtaseI), of 400 amino acids having 58% similarity and 43% identity to the human type I 5-phosphatase identified. Although no other putative type I 5-phosphatases were identified in the annotated genome we found that Drosophila contain open reading frames that encode two type II 5-phosphatase isozymes (CG6805 and CG9784), a synaptojanin ortholog (CG6562), and a putative Type IV family member (CG10426). Based on our analysis, we hypothesized that Drosophila is capable of generating I(1,3,4)P 3 through the activities of IP3K and 5-phosphatase. However, it does not appear that the annotated Drosophila genome possess gene products with significant sequence similarity to mammalian I(1,3,4)P 3 5-/6-kinases, suggesting that the fly may not synthesize higher IPs through the proposed metazoan route. If this is the case then IP3K-dependent higher IP synthesis in Drosophila may employ a different set of enzymes than previously described.
To identify a Drosophila Ipk1, we searched the genome for putative 2-kinases containing conserved amino acid motifs including EXKPK (7,19,20). Using the human IP 5 2-kinase (hsIpk1) amino acid sequence, we identified a single ortholog, CG30295 (designated dmIpk1) that translated into a polypeptide sharing 34% similarity and 25% identity to hsIpk1. In summary, our genome analysis indicate that Drosophila have many of the members of both yeast and metazoan pathways required for IP 6 synthesis, with the notable exception of a I(1,3,4)P 3 5/6-kinase. We therefore postulate that flies synthesize IP 5 and IP 6 through a pathway that requires the activities of dmIpk2 and dmIpk1.
Recent work with plant and human Ipk2 homologs demonstrates that the Ipk2 possesses 5-kinase activity toward I(1,3,4,6)P 4 (18,22). Generation of IP 5 through this Ipk2-dependent reaction is hypothesized as a critical step for metazoan higher IP synthesis. We found that dmIpk2 also phosphorylates I(1,3,4,6)P 4 to generate I(1,3,4,5,6)P 5 (Fig. 3, G and H). The catalytic efficiency of dmIpk2 toward I(1,4,5,6)P 4 and I(1,3,4,6)P 4 were similar (Table I). This is in agreement with the reported properties of the human homolog of Ipk2 for its substrates (22); however, in Drosophila the significance of this reaction is unclear given the molecular route by which the fly cell produces I(1,3,4,6)P 4 does not appear to be conserved.
dmIpk2 Rescues IP 6 Production in Yeast Ipk2-deleted Cells-Given the similar substrate specificity of dmIpk2 and yeast Ipk2, we performed complementation analysis of ipk2 mutant yeast. Yeast cells lacking Ipk2 fail to generate IP 6 and accu-mulate I(1,4,5)P 3 (Fig. 5A, top trace). We observed complete restoration of IP 6 synthesis when we heterologously expressed dmIpk2 in yeast ipk2⌬ (Fig. 5A, bottom trace). Deletion of Ipk1 in yeast results in the accumulation of its substrate, I(1,3,4,5,6)P 5 , and a diphosphoryl species PP-IP 4 that is generated from IP 5 . Expression of dmIpk2 in yeast ipk1⌬ ipk2⌬ recapitulated the appearance of an ipk1⌬ mutant as IP 5 and PP-IP 4 accumulated in these cells (Fig. 5B, bottom). These data provide evidence that dmIpk2 acts as a dual specificity IP 4 /IP 5 kinase in vivo and can employ an Ipk1-dependent mechanism to generate IP 6 . Conversely, IP 6 synthesis was not restored when dmIP3K␤ was heterologously expressed in ipk2-null cells. This result is not surprising because yeast lack several enzymes that are needed to generate IP 6 from I(1,3,4,5)P 4 , including I(1,3,4,5)P 4 5-phosphatase and I(1,3,4)P 3 5-/6-kinases. Taken together, our data suggest that dmIpk2, but not dmIP3K, can cooperate with an IP 5 2-kinase to generate IP 6 in dividing cells.
RNAi in Drosophila S2 Cells Effectively Reduces IP Kinase and Phosphatase Expression-We next examined higher IP synthesis pathways in the cultured fly cell line, Schneider S2. Extracts prepared from cultured S2 cells were found to have both phosphatase and kinase activities toward [ 3 H]I(1,4,5)P 3 in the presence of ATP. Specifically, we observed the formation of IP 2 , IP 4 , IP 5 , and a small amount of IP 6 when incubated for long periods of time (data not shown). We hypothesized that endogenous dmIP3K and/or dmIpk2 contribute to the I(1,4,5)P 3 phosphorylation in the cell extracts. We further postulated that dm5PtaseI may contribute to the I(1,4,5)P 3 phosphatase activity. Therefore, we treated the S2 cells with dsRNA to generate specific knockdowns via RNA interference (RNAi) (32). The success of the RNAi on the targeted genes was confirmed by measuring the level of mRNA degradation between control and target cells by Northern analysis (Fig. 6A). dmIP3K␤, dmIpk2, and dm5PtaseI mRNA levels were reduced by at least 80%, 61%, and 82% respectively. Of note, the 61% dmIpk2 mRNA reduction may be an underestimate as degraded RNA interfered with our measurement. To this end, a Western blot analysis with a dmIpk2-specific polyclonal antibody revealed a 90% reduction in protein levels in cells treated with dmIpk2 dsRNA compared with untreated cells (Fig. 6B).
To quantify the relative contributions of dmIP3K␤ and dmIpk2 toward the phosphorylation I(1,4,5)P 3 , we carried out kinase reactions from dsRNA-treated S2 extracts. Competing kinase and phosphatase reactions in the assays made individual specific activity measurements difficult to determine using conventional time course assays. Therefore, we analyzed reaction products at early time points (and less than 5% substrate conversion) by HPLC under conditions that separated IP molecules ranging from IP 1 to IP 5 (for example Fig. 6C). In vitro phosphorylation of I(1,4,5)P 3 to IP 4 and IP 5 was significantly reduced in extracts made from either dmIP3K␤or dmIpk2depleted cells indicating 1) that each kinase was functional in extracts and 2) that the RNAi worked for both (data not shown). When dsRNA to both dmIP3K␤ and dmIpk2 were simultaneously added to cells, extracts had little observable I(1,4,5)P 3 phosphorylation as compared with an equal amount of control treated extract (Fig. 6C), suggesting that dmIpk2 and dmIP3K␤ comprise nearly all the I(1,4,5)P 3 kinase activities in S2 cells. Kinase activity derived from dmIP3K␣ may not be detectable in appreciable amounts in S2 cells. We further quantified the effect of dmIpk2 knockdown on I(1,4,5,6)P 4 kinase activity ( Fig 7A) and found that RNAi treatment reduced activity over 95%.
Additionally, we tested the effects of RNAi depletion of IP 5 2-kinase and I(1,4,5)P 3 5-phosphatase activities. Extracts pre- pared from cells treated with dsRNA toward dmIpk1 yielded a 90% reduction in activity (Fig. 7B). The specific activity in crude extracts from control-treated and dmIpk1-depleted cells was 8.6 pmol/min/mg and less than 0.7 pmol/min/mg, respectively. Moreover, when extracts from control-treated or dm5PtaseI-depleted cells were incubated with I(1,4,5)P 3 , we observed a 10 -15% decrease in I(1,4,5)P 3 dephosphorylation into IP 2 and IP (Fig. 7C). The incomplete reduction in phosphatase activity is likely due to several additional 5-phosphatase gene products present in the extracts.
Molecular Analysis of Higher IP Production in Drosophila Schneider (S2) Cells-To determine the roles of the kinases and phosphatase in the cellular production of higher IPs, we steady state-labeled various RNAi-depleted S2 cells with [ 3 H]inositol. HPLC analysis of the extracted radiolabeled IPs from controltreated S2 cells revealed two major peaks corresponding to IP 5 and IP 6 , whereas I(1,4,5)P 3 and IP 4 peaks are relatively minor (Fig. 8A). The IP 5 isomer detected co-elutes with both I(2,3,4,5,6)P 5 and I(1,2,4,5,6)P 5 standards, which under these conditions are not separable. We report that relative distribution of radioactivity into the Ins, IP 6 , and IP 5 peaks is ϳ92, 3.0, and 0.8% of the total tritiated sample recovered from the soluble extract.
We tested whether S2 cellular IP 5 and IP 6 synthesis requires the activities of dmIP3K␤ or dmIpk2 by labeling cells that were RNAi-depleted for either gene. No significant changes in IP 5 or IP 6 synthesis were found in dmIP3K␤-depleted cells after six or fifteen days of knockdown (Fig. 8, B and F). Similarly, cells treated with dsRNA for dmIP3K␣ or dmIP3K␣ and dmIP3K␤ together did not affect higher IP synthesis (data not shown). In contrast, IP 5 and IP 6 levels were significantly decreased in dmIpk2-depleted cells (Fig. 8, C and F). We found in repeated experiments that IP 5 levels were often reduced below the limit of detection on our HPLC apparatus, whereas IP 6 levels where decreased by an average of 75%. These data demonstrate that dmIpk2 is required for IP 5 and IP 6 synthesis in Drosophila S2 cells.
Our data indicate that dmIpk2 contributes to IP 6 synthesis in S2 cells by converting I(1,4,5)P 3 to I(1,3,4,5,6)P 5 , which may then be phosphorylated to IP 6 by the 2-kinase dmIpk1. Analysis of steady-state radiolabeled dmIpk1-depleted S2 cells show that IP 6 levels were reduced, as measured from four independent experiments, by an average of 44% (Fig. 8, D and F). RNAi of dmIpk1 also caused levels of I(1,2,4,5,6)P 5 or I(2,3,4,5,6)P 5 to decrease by up to 54% suggesting that the isomer is a product of the kinase via 2-phosphorylation of I(1,4,5,6)P 4 or I(3,4,5,6)P 4 . It is also possible that the D2-phosphorylated IP 5 peak arises from dephosphorylation of IP 6 at the 3-or 1-position, in which case depletion of IP 6 may also reduce this IP 5 peak by mass action. Additionally, in the dmIpk1-depleted labeled cells, we observed the accumulation of a new IP 5 isomer that co-elutes with an I(1,3,4,5,6)P 5 standard. A similar buildup of I(1,3,4,5,6)P 5 occurs in yeast ipk1⌬, supporting the hypothesis that dmIpk1 uses I(1,3,4,5,6)P 5 as a substrate in vivo (7). These data indicate that dmIpk1 contributes to IP 6 synthesis in Drosophila S2 cells by phosphorylating I(1,3,4,5,6)P 5 . To further determine if IP 6 synthesis occurs through a single pathway of phosphorylation by dmIpk2 and dmIpk1, we simultaneously treated cells with dsRNA for both kinases. IP levels from the double-depleted cells were indistinguishable from those seen when dmIpk2 was depleted alone (Fig. 8F). This suggests that dmIpk2 and dmIpk1 are epistatic in S2 cellular production of IP 5 and IP 6 .
We also tested whether the Drosophila IP 6 synthesis pathway requires a type I 5-phosphatase as predicted for step three in the metazoan model (Fig. 1A). This would be surprising in light of our results indicating that RNAi depletion of dmIP3K␣ and dmIP3K␤ (step 1 of that model) did not disrupt IP 6 synthesis. Nonetheless, we treated cells with dsRNA targeting dm5PtaseI, radiolabeled with inositol and analyzed extracts by HPLC. Interestingly, rather than a depletion, we noticed a significant increase in IP 5 and IP 6 levels as compared with control cells (Fig. 8, E and F). Likewise, co-depletion of dm5PtaseI and dmIpk2 led to an elevation of IP 5 and IP 6 as compared with dmIpk2-depleted cells alone (Fig. 8, F and G). One possible explanation for this is that depletion of dm5PtaseI leads to an increase in an I(1,4,5)P 3 pool that is shared with the dmIpk2-dependent pathway. Similar increases in higher IPs were measured when dmIP3K␤ was depleted together with dmIpk2 or dmIpk1 suggesting that the kinase may also draw from the same I(1,4,5)P 3 pool (Fig. 8, F and G).
Ipk2-mediated Higher IP Production in Adult Flies-We next tested whether higher IP synthesis in the adult fly depends on FIG. 6. Depletion of dmIP3K␤, dmIpk2, and dm5PtaseI in S2 cells by RNA interference. Drosophila S2 cells were treated with dsRNA to dmIP3K␤, dmIpk2, or dm5PtaseI for 6 days. A, Northern blot analysis of dsRNA-treated cells. RNA was isolated from dsRNA-treated cells or control cells, separated on an agarose gel and transferred to nitrocellulose paper. 32 P-labeled DNA probes specific for dmIP3K␤, dmIpk2, and dm5PtaseI were hybridized to the membrane, visualized, and quantified on a phosphorimager. B, Western blot analysis of dmIpk2 dsRNA-treated cells. Protein extracts were prepared from dsRNA-treated or untreated cells, separated on a polyacrylamide gel and transferred to nitrocellulose paper. A rabbit anti-dmIpk2 polyclonal antibody was used to detect the dmIpk2-specific band. C, I(1,4,5)P 3 -kinase activity in control and dmIpk2/dmIP3K␤ dsRNAtreated cells. Protein extracts were prepared from control or dsRNAtreated cells and incubated with 3 H-labeled I(1,4,5)P 3 for 1 h. Reactions were stopped by addition of HPLC buffer and analyzed by Partisphere strong anion exchange HPLC.
dmIpk2 through loss and gain of function analysis in transgenic insects. Wild-type (w1118) male flies were in vivo labeled by feeding them a 5% sucrose solution containing [ 3 H]inositol for 4 days. Distribution of radioactive metabolites observed by HPLC analysis of the whole fly extracts revealed an IP profile similar to S2 cells (compare Fig. 8A to Fig. 9A, top panel). To test whether the IP synthesis depends on dmIpk2, we examined IP 6 levels in fly lines that were deficient in expression of the kinase. Df(2L)BCS16 is a deficiency line that fails to complement Ush and Gsc that reside on either side of dmIpk2 at 21E2 (Kevin Cook-Bloomington stock center). Another deficiency line was obtained that is derived from a P-element excision from Nle, a neighboring gene to dmIpk2 (Nle ⌬8 ) (27).
We speculated that the break points of the Nle ⌬8 deficiency include the dmIpk2 open reading frame. Additionally, we generated a transgenic line that expresses an inverted repeat of dmIpk2-specific sequence under the Gal4 promoter for RNAi (p[WIZ-Ipk2]) (28). Western blot analysis of flies from each line revealed that dmIpk2 levels were reduced in all three deficient lines as compared with wild type (Fig. 9B). Antigen levels of dmIpk2 in both Df(2L)BCS16 and Nle ⌬8 were reduced by 50%, consistent with them being heterozygous for dmIpk2 (Fig. 9B). Ubiquitous expression of the dmIpk2 inverted repeat using an Actin-Gal4 promoter caused dmIpk2 levels to decrease 81% (Fig. 9B).
We A-E, S2 cells were treated with dsRNA for 4 days, labeled for 2 days, and harvested on day 6. A, shows the inositol polyphosphate profile of untreated S2 cells, cells treated with dmIP3K␤ dsRNA (B), dmIpk2 dsRNA (C), dmIpk1 dsRNA (D), dm5PtaseI dsRNA (E). The IP 5 isomer (IP 5x ) co-elutes on a Partisphere sax with either I(2,3,4,5,6)P 5 or I(1,2,4,5,6)P 5 standards, which are indistinguishable using this method. Changes in IP 6 (F) and IP 5 (G) levels were quantified after treating S2 cells with dsRNA for fifteen days. The percentage of each IP was calculated by dividing the area under the curve (AUC) of IP 5 or IP 6 peaks by total inositol AUC in the trace. The experiments for each bar were repeated at least 3 times and carried out in triplicate. Error bars represent S.E. The horizontal dashed line represents the limit of sensitivity of our HPLC apparatus. RNAi-treated cells were propagated for 8 days, split, and again treated with dsRNA, labeled on day 13 and harvested on day 15. from the dmIpk2-deficient lines along with wild-type flies (w1118). After labeling the flies for 4 days, their soluble IPs were extracted and analyzed by HPLC. In all three deficient flies, the relatively weak IP 5 peak observed in w1118 flies (Fig.  9A, top trace) decreased below the limit of detection suggesting a loss of synthesis (data not shown). Df(2L)BCS16 and Nle ⌬8 flies had IP 6 levels that were reduced 20 -30% as compared with w1118 flies whereas levels in the p[WIZ-Ipk2]/Actin-Gal4 flies were reduced by an average of 60% (Fig. 9C). The dosedependent reduction in IP 6 levels observed in the fly mutants, suggests that dmIpk2 activity is an important rate-limiting step in the synthesis pathway.
Transgenic flies overexpressing epitope-tagged dmIpk2 under control of the actin promoter had a 5-fold increase in higher IP production as compared with wild-type flies (Fig. 9A, bottom  trace). Interestingly, the levels of several different IP species increased in these flies, including IP 2 and IP 3 uncovering a potential catabolic component of the pathway. Collectively, our results demonstrate that regulation of higher IP production in adult flies requires a dmIpk2-dependent synthesis pathway.
The Cellular Localization of dmIpk2-Given the role and localization of yeast Ipk2 in the nucleus, we analyzed the subcellular localization of dmIpk2 in cultured cells and whole fly tissue. We observed that FLAG-tagged dmIpk2 localizes to the nucleus when expressed in Drosophila S2 cells, suggesting that Ipk2-dependent IP synthesis in the fly is nuclear (data not shown). In yeast, Ipk2 initiates a nuclear IP synthesis pathway whose products regulate events such as transcription and chromatin remodeling, raising the possibility that Ipk2 associates with active zones of chromatin DNA (7,8,11,12). The Drosophila polytene chromosomes present in giant cells of the salivary glands derived from 3rd instar larvae represent a unique system to observe nuclear proteins. FLAG-tagged dmIpk2 was expressed in larvae under control of the actin promoter and immunofluorescence studies demonstrate that the majority of protein appears localized within the nucleus (Fig. 10, top panel). Closer examination of the nuclear localization and concomitant staining of the chromosomal DNA with DAPI revealed that the bulk of dmIpk2 does not appear to be bound to chromatin DNA itself, rather it appears enriched in the non-DAPI nucleosol (Fig. 10, bottom panel). The same distribution was observed for the endogenous protein using an anti-dmIpk2 antibody, indicating the FLAG-dmIpk2 localization was not a consequence of overexpression (data not shown). These data provide new insights into a possible nuclear role for Ipk2 and its products in D. melanogaster. DISCUSSION A major finding of this work is the identification of gene products whose function is required for the genesis of higher IP synthesis in metazoans. It has been known for decades that eukaryotic cells possess an ensemble of higher phosphorylated IP messengers. Remarkably, a molecular basis for their production has only been recently determined in budding yeast but has not been clarified in higher eukaryotes. Thus, for the first FIG. 9.Inositol polyphosphate synthesis in adult flies is dmIpk2dependent. A, 3-6-day old flies (w1118, top trace) were fed 5% sucrose with 500 Ci/ml [ 3 H]inositol for 4 days. Flies were harvested by chloroform extraction, and the aqueous portion was analyzed by Partisphere strong anion exchange HPLC. Bottom trace, labeling obtained from transgenic flies that express epitope-tagged dmIpk2 under control of the actin promoter. B, protein extracts from adult male flies were prepared, separated by SDS-PAGE, and transferred for blotting. A rabbit dmIpk2 polyclonal antibody was used: Df(2L)BCS16 and Nle ⌬8 are lines that lack genomic sequence including the dmIpk2 coding region. p[WIZ-Ipk2]/cyo is a transgenic line that contains coding sequence for inverted repeat RNA specific for the Ipk2 transcript. Expression was induced by crossing p[WIZ-Ipk2]/cyo to Gal4 expressing flies under the actin promoter (WIZ2-dmIpk2/actin). w1118 or p[WIZ2-Ipk2]/cyo flies were used as controls. C, Df(2L)BCS16, Nle ⌬8 , and WIZ-Ipk2/Acin flies were labeled, harvested, and analyzed by HPLC. The levels of IP 6 in deficient flies are expressed as a percentage relative to normal (either w1118 or p[WIZ-Ipk2]/cyo). time we define the major molecular pathways used to synthesize IP 5 and IP 6 in the fruit fly D. melanogaster (summarized in Fig. 11). We find that dmIpk2 and dmIpk1 are the main contributors to higher IP synthesis. Our data prove an important role for dmIpk2 using both loss and gain of function experiments in cultured cells and the whole organism. In contrast, dmIP3K and dm5PtaseI appear to negatively regulate higher IP synthesis. This is highly significant as it dispels a widely proclaimed view in the literature that IP3K isoforms are required to initiate the synthesis of higher IPs from I(1,4,5,)P 3 .
Our data also support the hypothesis that an I(1,4,5)P 3 pool serves as a substrate source for three distinct metabolic routes through dmIP3K, dm5PtaseI, and dmIpk2. We find that regulation of any of these enzymes can impinge on Drosophila higher IP synthesis. One attractive explanation for why dmIP3K␤ or dm5PtaseI depletion causes an increase in IP levels is through expansion of an I(1,4,5)P 3 pool. Both of these enzymes are known to modulate I(1,4,5)P 3 -stimulated calcium signaling through the phosphorylation or dephosphorylation of this second messenger (4). Depletion of either of the two enzymes could lead to increases in the I(1,4,5)P 3 pool that is available for feeding into the Ipk2/Ipk1 pathway. Alternatively, separable IP synthesis pathways may exist in Drosophila that can impinge on each other when they are disrupted or downregulated. Speed et al. (33) reported similar increases in inositol polyphosphate production in mammalian cells by depleting a human Type I 5-phosphatase in NIH3T3 cells using antisense RNA (33). They showed that 5-phosphatase depletion causes I(1,4,5)P 3 , I(1,3,4,5)P 4 , IP 5 , and IP 6 levels to increase. Studies of IP3K-deficient mice indicate a phenotypic role for I(1,3,4,5)P 4 production, in neuronal and T lymphocyte function (34 -36). Consist with our data in flies, loss of IP3K B in mice did not decrease IP 5 levels (IP 6 was not shown) in labeled cells (35), although one cannot exclude the possible compensation by other IP3K isoforms.
Is the Ipk2/Ipk1 pathway a common route of IP 5 and IP 6 synthesis in all eukaryotes? Based on our data and the evolutionary conservation of Ipk2/Ipk1 across species, the parsimonious answer is "yes". However, some studies in the literature challenge the exclusivity of this pathway. Substrate specificity studies of human Ipk2 indicate that a catalytic preference for I(1,3,4,6)P 4 5-kinase versus I(1,4,5)P 3 6-/3-kinase activities (22). Accordingly, human Ipk2 would act downstream of the I(1,3,4)P 3 5-/6-kinase. Other studies of human Ipk2 show in vitro that human Ipk2 is a dual-specificity kinase that converts I(1,4,5)P 3 to IP 5 (23). Heterologous expression of human Ipk2 does not appear to restore IP 5 synthesis in ipk2 deficient yeast (22). Thus it may be that humans, but not other metazoans, have evolved a distinct more complex pathway reminiscent of the metazoan model shown in Fig. 1A. It is also possible that different cell-types or tissues utilize more than one pathway to generate these important messengers.
It is also interesting that while Ipk1 and Ipk2 are conserved among all eukaryotes, the I(1,3,4)P 3 5-/6-kinase is not found in yeast or flies. In mammalian cells, I(1,3,4)P 3 5-/6-kinase alteration through overexpression or RNAi effects tumor necrosis factor induced apoptosis possibly through alteration of higher IPs (37). In plants, Shi et al. (38) provide evidence for a role for higher IP synthesis by disruption of the maize 5-/6-kinase open reading frame, which results in a 30% reduction in IP 6 levels in seeds (38). Plant Ipk2 may therefore contribute to this pathway through phosphorylation of the D-5 position of I(1,3,4,6)P 4 to generate I(1,3,4,5,6)P 5 . Since dmIpk2 also has 5-kinase activity toward I(1,3,4,6)P 4 with a relative affinity that is comparable to the human and plant Ipk2 it is possible it serves a similar role in flies. Although, our sequence analysis of the Drosophila genome did not reveal a I(1,3,4)P 3 5-/6-kinase gene product argues against this hypothesis.
Our data is consistent with a phospholipase C-dependent sequential phosphorylation of I(1,4,5)P 3 to IP 6 . Drosophila possesses three different phospholipase C isoforms including NorpA, Plc21C, and Small wing (Sl) (39). Studies that examine these homologs may elucidate whether IP 5 and IP 6 are synthesized from I(1,4,5)P 3 and provide an understanding of how their synthesis is regulated. Phenotypic analysis of our mutant flies and cell lines are ongoing and may expose functional roles for higher IP synthesis in eukaryotes. It is also important to note that there are phospholipase C-independent routes for IP 6 synthesis reported in the literature. In the slime mold Dictyostelium, deletion of the sole phospholipase C gene does not significantly change higher IP synthesis (40). Biochemical evidence for phospholipase C-independent IP synthesis has been reported from plants and slime mold through the sequential phosphorylation of I(3)P or inositol (41)(42)(43).
It is tempting to speculate that Drosophila Ipk2 and Ipk1 comprise a nuclear inositol signaling pathway that regulates nuclear processes similar to those in budding yeast. The evolutionarily conservation of the nuclear localization of Ipk2 from a variety of species is consistent with a conserved functional role. IP 4 and IP 5 in yeast are implicated as messengers that mediate transcriptional events (8), possibly through the regulation of chromatin remodeling (11,12). IP 6 production is implicated in regulation at the nuclear pore and is required for the efficient export of mRNA from the nucleus (7,44). The localization of dmIpk2 to the nucleosol and not bulk chromatin provides the first evidence indicating that the majority of Ipk2 is not physically associated with chromatin. It is possible that activation of signaling promotes the movement of a minor pool of Ipk2 to chromatin, or that local production of IP 4 or IP 5 in the FIG. 11. Genetic roadmap of Drosophila IP synthesis. The proposed model of IP metabolism in D. melanogaster. Our data indicate that I(1,4,5)P 3 has several different metabolic fates. The primary route of higher IP synthesis requires the activities of Ipk2 and Ipk1 whereas 5PtaseI and IP3K participate in the metabolism of I(1,4,5)P 3 in other pathways. Black font highlights portions of the model that are directly supported by data presented in this study. Gray font indicates portions that are presumed based on data presented here or from previous studies. nucleosol activates transcriptional programs or chromatin remodeling. The generation of GFP-dmIpk2 transgenic flies will facilitate future studies aimed at examining potential dynamic changes in Ipk2 localization on polytene chromosomes during transcription.
If Ipk2 has an important role in regulating nuclear, or cytoplasmic function in Drosophila, we may observe phenotypic changes in transgenic flies. Currently, we have not observed a gross effect on development in either RNAi depleted (60% knockdown of cellular IP production) or overexpressing flies (5-fold increase in IP production). This may not be such a surprise since: 1) Ipk2 is conditionally essential in yeast; 2) several mutant Ipk2 lines with impaired but not complete loss of function in many cases appear normal; and 3) overexpression of Ipk2 in yeast has no observable phenotype. These data indicate that only a small amount of Ipk2 is required for regulation of nuclear function and that overproduction is well tolerated. Thus, our future phenotypic analysis will focus on specific tissues and/or the complete loss of function of Ipk2 through genetic mutation or deletion.