MOLECULAR DEFINITION OF A NOVEL INOSITOL POLYPHOSPHATE METABOLIC PATHWAY INITIATED BY INOSITOL 1,4,5-TRISPHOSPHATE 3-KINASE ACTIVITY IN SACCHAROMYCES CEREVISIAE

The production of inositol polyphosphate (IPs) and pyrophosphates (PP-IPs) from inositol 1,4,5-trisphosphate (I(1,4,5)P3) requires the 6-/3-/5-kinase activity of Ipk2 (also known as Arg82 and inositol polyphosphate multikinase). Here, we probed the distinct roles for I(1,4,5)P3 6- versus 3-kinase activities in IP metabolism and cellular functions reported for Ipk2. Expression of either I(1,4,5)P3 6- or 3-kinase activity rescued growth of ipk2-deficient yeast at high temperatures, whereas only 6-kinase activity enabled growth on ornithine as the sole nitrogen source. Analysis of IP metabolism revealed that the 3-kinase initiated the synthesis of novel pathway consisting of over eleven IPs and PP-IPs. This pathway was present in wild-type and ipk2 null cells, albeit at low levels as compared with inositol hexakisphosphate synthesis. The primary route of synthesis was: I(1,4,5)P3 --> I(1,3,4,5)P4 --> I(1,2,3,4,5)P5 --> PP-IP4 --> PP2-IP3 and required Kcs1 (or possibly Ipk2), Ipk1, a novel inositol pyrophosphate synthase, and then Kcs1 again, respectively. Mutation of kcs1 ablated this pathway in ipk2 null cells and overexpression of Kcs1 in ipk2 mutant cells phenocopied IP3K expression, confirming it harbors a novel 3-kinase activity. Our work provides a revised genetic map of IP metabolism in yeast and evidence for dosage compensation between IPs and PP-IPs downstream of I(1,4,5)P3 in the regulation of nucleocytoplasmic processes.

The enzymatic promiscuity Ipk2 and its action early in complex IP metabolic pathways may account for the pleiotropic biologic defects observed in ipk2 mutant yeast. To further dissect the roles of Ipk2's kinase activity we have utilized heterologous complementation analysis in ipk2 mutant yeast cells (20,21,23). Here, in order to specifically assess the role of 3-kinase activity, we studied the effects of expression of a Drosophila I(1,4,5)P 3 3-kinase beta isoform (dmIP3K), whose only reported enzymatic function was to generate inositol 1,3,4,5-tetrakisphosphate [I(1,3,4,5)P 4 ] (21,28). This work has led to an unexpected finding that expression of dmIP3K initiates a novel IP pathway, whose molecular basis we describe. Additionally, we provide evidence for dosagecompensation among IP species in the regulation of Ipk2-mediated nuclear and cytoplasmic processes.

MATERIALS AND METHODS
Strains and Media -Yeast were grown in either rich medium (yeast-peptone-dextrose), or complete minimal (CM) medium lacking the appropriate nutrients for maintenance of plasmids containing markers. Yeast strains used in this study were from previous studies or generated by mating strains from previous studies (5,8,10). Ornithine plates were made as previously described (5,29).

High Performance Liquid Chromatography (HPLC) Columns And Gradients -Two different
Partisphere SAX columns and elution gradients were used in this study. Method 1 utilized a custom-made narrow-bore column was obtained from Capital HPLC Limited (12.5 cm x 2.1 mm). IPs were eluted with a linear gradient of ammonium phosphate (pH 3.5) (AP) from 10 mM to 1.7 M over the course of 12 min. and isocratic elution at 1.7 M AP for 23 min. (flow rate of 0.4 ml/min). Method 2 achieved higher resolution IPs by using a wider-bore Whatman column (12.5 cm x 4.6 mm) and a longer elution gradient as follows: a linear gradient from 10 mM to 85 mM AP over 5 min., then 85 mM to 1.7 M AP over 65 min. and then isocratic elution at 1.7 M AP for 30 min. all at a flow rate of 1 ml/min.
Plasmid Construction -Construction of plasmids pRS314-dmIpk2 and pRS314-dmIP3Kβ (which we will refer to as dmIP3K throughout the remainder of this manuscript) and Kcs1 expression constructs (pRS-Kcs1 and pRS-kcs1kin-) was described previously (10,21). The plasmids were then transformed into different yeast strains using standard yeast transformation techniques. The pCR ® 2.1 vector (Invitrogen) containing the entire scIpk1 coding region with a PCR-generated SalI site at the 5' end was provided by Dr. Makoto Fujii (York lab, Duke University). The scIpk1 coding region was subcloned from pCR ® 2.1 using SalI and EcoRI and ligated into the pGEX-2T vector.
The vector was transformed into Escherichia coli (DH5α ) for expression of recombinant protein.
In Vivo Labeling Of Yeast Cultures -Yeast cultures were incubated at 30°C in minimal medium lacking the appropriate amino acids and 150 µM CuSO 4 to late logarithmic phase. [ 3 H]inositol (American Radiolabeled Chemicals) was added to a final concentration of 80 µCi/ml. For pulse-chase analysis: yeast strains were grown to late logarithmic phase in 50 ml of unlabeled CM medium. The cells were washed and resuspended in 500 µl of inositol-free CM medium supplemented with 1 mCi/ml [ 3 H]-inositol. After labeling for ten minutes the cells were washed, resuspended in 50 ml inositol-replete CM medium without label and incubated at 30˚C. Aliquots were taken at various time points and the cells were frozen on dry ice until they could be harvested and analyzed by Partisphere stronganion exchange HPLC.
Soluble IPs were harvested and analyzed by HPLC using a stronganion exchange column as described previously (8). Alternatively, the IPs were harvested for enzyme treatment as described below.

Enzyme Analysis Of [ 3 H]-Inositol Labeled Yeast
Extracts -Labeled yeast strains were resuspended in 50 mM Tris pH 7.5 and lysed for 30 seconds with a bead beater (Biospec Products) using glass beads (B. Braun Biotech International). The lysate was immediately boiled for 5 min. and then the entire procedure was repeated a second time. Extracts containing soluble IPs were recovered by centrifugation. Reactions were carried out by incubating the extracts for 1 hr at 37 ˚C in a buffer containing 10 mM Tris pH 7.5 and 10 mM NaCl with 500 ng of human Type I 5-phosphatase and/or human diphosphoryl inositol polyphosphate phosphohydrolase (DIPP). The reactions were stopped by addition of 200 µl of 10 mM ammonium phosphate pH 3.5 and analyzed by Partisphere strong-anion exchange HPLC.
Bacterial Expression of dmIP3 and scIpk1 -Transformed Escherichia coli (DH5α) were grown at 37 ˚C to an OD 600 of 0.6 and induced with 0.1 mM isopropyl-1-thio-β-D-galactopyranoside for 4 h at 30 ˚C. The 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.). The lysates were cleared by centrifugation at 14,000xg. The GST fusion proteins were purified over glutathione Sepharose (Amersham Biosciences) according to the manufacturer's instructions. A buffer containing 50 mM Tris-HCl pH 8.0, 50 mM NaCl, 5 mM dithiothreithol, and 20 mM glutathione was used to elute the proteins from the glutathione Sepharose. Proteins were quantified by modified Bradford method and SDS-PAGE analysis.
Kinetic Assays -The K m and V max of the enzymatic interaction between scIpk1 and I(1,3,4,5)P 4 and various substrates were determined.
The following reaction mixture was prepared: 10 mM Tris pH 7.5, 10 mM NaCl, 4 mM ATP, 20 mM MgCl 2 , 10,000 cpm/µ l [ 32 P]-I(1,3,4,5)P 4 , 500ng scIpk1 and various concentrations of unlabeled I(1,3,4,5)P 4 in a 20 µl reaction volume. The reaction was stopped by the addition of 4 µl of 1M KH 2 PO 4 . The amount of product formed was quantified by thin layer chromatography. The K m and V max values were obtained from a nonlinear curve fit to the Michaelis-Menten equation using GraphPad Prism version 4.01.
In order to further probe the kinasedependent roles for Ipk2, we compared I(1,4,5)P 3 6-versus 3-kinase activities for functional complementation. To accomplish this, we genetically added back I(1,4,5,6)P 4 or I(1,3,4,5)P 4 production by heterologously expressing Drosophila IPKs, either dmIpk2 or dmIP3K, in ipk2 deficient yeast. Both kinases have been shown to be members of a family of IP kinases bearing a signature PxxxDxKxG motif that includes: Ipk2 6-/3-/5-kinases; IP 6 kinases (IP6K) which function as inositol pyrophosphate synthases and utilize IP 5 and IP 6 substrates; and IP3Ks which were not found in budding yeast or plants, but were present in fly, mouse and human genomes (17,32). We recently reported the cloning and biochemical characterization of dmIP3K and dmIpk2 and showed they possess I(1,4,5)P 3 3-kinase and I(1,4,5)P 3 6-kinase activities, respectively (21). Functional analysis of ipk2 deficient cells that expressed either dmIpk2 or dmIP3K revealed that 6-kinase, but not 3-kinase activity could rescue growth on ornithine as a sole nitrogen source (Fig. 1A). This data, coupled with our previous analysis of the ipk2-3 mutant (5), indicated that production of I(1,4,5,6)P 4 , but not I(1,3,4,5)P 4 , was necessary for activation of the ArgR-Mcm1 transcriptional complex. When we analyzed the transgenic strains for growth at high temperatures, we found that either 6-or 3-kinase activity was able to complement the temperature sensitive phenotype of ipk2 null yeast (Fig. 1B).
In order to rule out that complementation analysis may be related to downstream metabolites, such as those required for mRNA export pathways as described for Ipk1 (8), we tested the expression in ipk2 ipk1double mutant cells. Rescue of growth on ornithine as the sole nitrogen source or by guest on March 24, 2020 http://www.jbc.org/ Downloaded from temperature sensitivity by dmIpk2 did not require the presence of Ipk1; however, the rescue of temperature sensitivity by dmIP3K did require Ipk1 ( Fig. 1A and B). Additionally, to determine if 6-or 3-kinase complementation of temperature sensitivity required pyrophosphate synthesis, we analyzed ipk2 kcs1 double mutants. We report that dmIpk2 and IP3K were able to partially (not fully) rescue under these conditions (Fig. 1C). Our data indicate that I(1,2,3,4,5)P 5 production and a downstream PP-IP were required for sustaining temperature sensitive complementation. Of interest, the ipk1 dependent temperature sensitive phenotype may be of use for forward genetic strategies aimed at identification of regulators and receptors of Ipk1 pathways. Of note, we have previously published that ipk1 mutation is not temperature sensitive by plating assays (only after 20 generations -see York et al, Science 1999). 3 3-kinase activity -We next analyzed the effect of heterologous expression of dmIpk2 and dmIP3K on IP metabolism in ipk2 null yeast. We previously reported that d m I p k 2 expression was able to fully complement Ipk2 enzyme function in cells by converting I(1,4,5)P 3 to I(1,3,4,5,6)P 5 (21). Remarkably, analysis of ipk2 deficient cells expressing dmIP3K revealed the synthesis of several new IP metabolites including novel IP 3 , IP 4 , IP 5 , and PP-IP species (Fig. 2).

Discovery of a novel IP metabolic pathway initiated by I(1,4,5)P
In order to determine the genes required for synthesis of the new species downstream of I(1,3,4,5)P 4 , we examined the role of inositol pyrophosphate synthase and 2-kinase activities. Kcs1 has been shown to synthesize PP-IP 4 α and PP-IP 5 α from I(1,3,4,5,6)P 5 and IP 6 precursors (14,33). Of note, we have implemented a symbolbased nomenclature (alpha, beta and gamma) to enable the distinction of the growing list of PP-IP isomers and species. As we describe below, the basis for assigning distinct isomers was that they had unique HPLC elution profiles and unique genetic routes of synthesis. At this time we do not have chemical structures that enable definition of the ring positions harboring pyrophosphates. Loss of Kcs1 in ipk2Δ cells expressing dmIP3K resulted in the obvious elimination of the most polar PP-IP species detected, PP 2 -IP 3 β (Fig. 2). We used the "beta" designation due to its distinct elution profile from PP 2 -IP 3 α (which is synthesized downstream of I(1,3,4,5,6)P 5  PP-IP 4 α) and fact that it is synthesized through I(1,3,4,5)P 4  I(1,2,3,4,5)P 5  PP-IP 4 β (also see below). These data indicate the existence of a second pyrophosphate synthase gene product, designated here as Ips1, required for the synthesis of PP-IP 4 β. It is unclear at this point whether or not Ips1 is similar or identical to the activity identified previously as Ids1 which is required for the synthesis of PP-IP 5 β (10). We next tested whether yeast Ipk1 was required for the pathway. When dmIP3K was expressed in ipk2Δ ipk1Δ cells, there was a loss of IP 5 and all PP-IPs, along with the accumulation of unique IP 3 and IP 4 species (Fig.  2). We therefore conclude that phosphorylation of I(1,4,5)P 3 on the D-3 position initiates a novel IP synthesis pathway that requires Ipk1, a novel inositol pyrophosphate synthase and Kcs1.
Does this novel pathway exist in wild-type or ipk2 mutant cells? Earlier work by our lab and the Shears lab indicated that IPs with similar HPLC elution profiles to those we observed in Fig  2 exist at low levels in ipk2 deficient and wildtype cells (8,18). We therefore re-examined several combinations of kinase mutants using a high-resolution and high-sensitivity radiolabeling system (Fig. 3). Using this method, examination of ipk2 null cells revealed a similar pattern of IPs as those observed in the dmIP3K expressing cells (compare Fig 3 top and second traces). Of note, this method also exposed that the number of unique IP and PP-IP species in ipk2 deficient cells expressing dmIP3K was substantially greater than we previously thought (compare Fig. 3, top trace to Fig. 2 second trace). The presence of these IPs in ipk2 deficient cells led us to speculate that another yeast gene product harbors I(1,4,5)P 3 3kinase activity. We therefore analyzed ipk2 ipk1 and ipk2 kcs1 double mutant cells (Fig. 3, bottom two traces). Similar to results shown above, loss of Ipk1 in i p k 2 null cells resulted in the disappearance of IP 5 and PP-IP species, and a corresponding accumulation of IP 3 and IP 4 . Loss of Kcs1 in ipk2 null cells ablated the synthesis of nearly all IP and PP-IPs detectable using this high sensitivity method. This indicates that Kcs1 may regulate or act as a I(1,4,5)P 3 3-kinase and that Ipk1 functions in the conversion of IP 4 to the higher IPs of this pathway. High sensitivity labeling of wild-type cells indicated similar species were present (data not shown) consistent with results of others (18).
Characterization of the IP species produced by the I(1,4,5,)P 3 3-kinase pathway -To understand the molecular structure of the species produced by this novel pathway, we used enzymatic digestion of the novel IPs derived from mutant cells. Our typical method for preparing IPs from radiolabeled extracts utilized 0.5 M HCl, which impaired enzymatic analysis. To circumvent this problem, we rapid boiled radiolabeled extracts isolated under neutral pH conditions to ensure the inactivation of endogenous enzymes. Extracts prepared from ipk2 null cells expressing dmIP3K in this manner appeared similar to those prepared by the acid-chloroform extraction method used in previous analysis (compare Fig. 2, second trace with Fig. 5, top trace). To determine if any of the species were I(1,4,5)P 3 , I(1,3,4,5)P 4 o r I(1,2,3,4,5)P 5 we treated extracts with recombinant human Type I 5-phosphatase, which dephosphorylates the D-5 these substrates. Surprisingly, none of the IPs or PP-IPs in this extract were hydrolyzed by the 5-phosphatase (Fig. 5, second trace). To confirm that the 5phosphatase was active under these conditions, we spiked these extracts with [ 3 H]-I(1,3,4,5)P 4 standard, and observed that it was completely hydrolyzed to I(1,3,4)P 3 by the 5-phosphatase (not shown). We next treated extracts prepared from ipk2 ipk1 double mutant cells expressing dmIP3K and found that while the IP 4 species was sensitive to 5-phosphatase, consistent with it being I(1,3,4,5)P 4 , the IP 3 species was not (data not shown). Thus, the IP 3 species found in either extract expressing dmIP3K, was likely I(3,4,5)P 3 based on its elution profile. Additionally, we have partially purified an IP 1-phosphatase activity from yeast, which we have designated Inp1, encoded by an unknown gene (York lab, unpublished results). Given that he IP 4 species in ipk2 null cells expressing dmIP3K was not 5Ptase sensitive, the elution position relative to other IP 4 standards, and the inability to be phosphorylated by either recombinant Ipk1 or Ipk2 (not shown) suggested that it was likely I(2,3,4,5)P 4 . This species may arise from either Inp1 cleavage of I(1,2,3,4,5)P 5 or by phosphorylation of I(3,4,5)P 3 by a 2-kinase. Of note, we were unsuccessful in our attempts to phosphorylate I(3,4,5)P 3 with by guest on March 24, 2020 http://www.jbc.org/ Downloaded from recombinant yeast Ipk1 (not shown); however, it was possible that the specific activity of the kinase towards this substrate was low or that we did not have proper conditions that mimicked those in cells.
We next examined the identities of PP-IP species in ipk2Δ cells expressing dmIP3K. DIPP specifically removes the beta-phosphate from PP-IPs and has no reported activity against the IP monophosphates (34). Recombinant human DIPP completely dephosphorylated the two major species that eluted at 20 and 30 min. with a concomitant increase in the levels of I(1,2,3,4,5)P 5 (Fig. 5, third trace). The observed increase in IP 4 levels in DIPP treated extracts confirmed that a PP-IP 3 α and PP 2 -IP 2 α were present as shown in Fig 3, top trace (note under low resolution HPLC this species co-eluted with IP 5 ). We did not observe evidence of IP 6 after treatment. Additionally, we simultaneously treated extracts with both DIPP and 5-phosphatase. We observed a disappearance of PP-IP 4 β and PP 2 -IP 3 β, the reduction of a major peak of I(1,2,3,4,5)P 5 , and the formation of I(1,3,4)P 3 and I(1,2,3,4)P 4 (Fig 5,  bottom trace). We conclude that the major IP 5 released by DIPP treatment was (1,2,3,4,5)P 5 based on: 1) this IP 5 species co-eluted exactly with authentic I(1,2,3,4,5)P 5 ; 2) it was fully susceptible to 5-phosphatase treatment generating a product that co-eluted with the I(1,2,3,4)P 4 standard and 3) that its synthesis in cells required Ipk1. Furthermore, recombinant type I 5-phosphatase was unable to hydrolyze any additional IP 5 species tested, and we are not aware of any literature demonstrating that 5-phosphatases have ringposition promiscuity (ie 4-or 3-phosphatase activities). Collectively, these data corroborated our genetic evidence that PP-IP 4 β and PP 2 -IP 3 β arose through sequential phosphorylation of I(1,2,3,4,5)P 5 , via the action of two pyrophosphate synthase steps, likely through Ips1 or Ids1 and then Kcs1. Additionally they suggested that I(1,3,4,5)P 4 was the likely precursor for PP-IP 3 α and PP 2 -IP 2 α. A second IP 5 (IP 5x ) species was identified through this experiment, based on the observation of an IP 5 species that was not susceptible to 5-phosphatase treatment (compare Fig. 5, second and bottom traces).
Pulse chase analysis of the IP3K-dependent synthesis pathway -To further examine the order of cellular IP synthesis we used pulse-chase analysis of ipk2 null cells expressing dmIP3K. Overnight cultures were grown to late logarithmic phase, pulse labeled with medium labeled with [ 3 H]-inositol for ten min., washed, and then chased with medium supplemented with excess cold inositol. Examination of IP profiles revealed that IP 3 and IP 4 pools accumulated first, and were followed by the synthesis of IP 5 , PP-IP 4 β and PP 2 -IP 3 β (Fig. 6A). These data were consistent with the kinetics we observed for scIpk1 phosphorylation of I(1,3,4,5)P 4 . The relatively high K m that scIpk1 exhibits for I(1,3,4,5)P 4 may help explain our observation that the 2-kinase's substrate (I(1,3,4,5)P 4 ) accumulated before a significant amount of I(1,2,3,4,5)P 5 was synthesized.
In ipk2Δ ipk1Δ cells expressing dmIP3K I(1,3,4,5)P 4 was the only IP 4 species detected at any of the tested time points. At later time points the IP 3 species was not susceptible to 5phosphatase hydrolysis indicating it was I(3,4,5)P 3 . Based on these findings, together with the genetic and pulse labeling data presented above, we propose the primary synthesis pathway to be I(1,4,5)P 3  I(1,3,4,5)P 4  I(1,2,3,4,5)P 5  PP-IP 4 β  PP 2 -IP 3 β. Given the complexity of additional species, it also appears that there were several branches to the pathway, which may arise from phosphatase and promiscuity of IPK kinase activities.
Our results indicate that Kcs1 is both a pyrophosphate synthase and a I(1,4,5)P 3 3-kinase. Previous experiments examining the pyrophosphate synthase activity of Kcs1 demonstrated that its activity requires a highly conserved PxxxDxKxG motif that is required for ATP binding (10,35). To determine if this site is also important for the 3-kinase activity of Kcs1, we generated a double point mutant (D786A; K788A) that lacks its diphosphoryl synthase activity. When this mutant was expressed in ipk2Δ cells, we did not observe the Kcs1dependent synthesis pathway, indicating that the 3-kinase activity uses the same catalytic domain as the pyrophosphate synthase activity (data not shown).

DISCUSSION
Our results may be summarized into four main findings: 1) we distinguish between the roles of 6-versus 3-kinase activities associated with Ipk2 function in cells; 2) we serendipitously have discovered and defined a molecular basis for a new IP/PP-IP pathway in yeast that is initiated by I(1,4,5)P 3 3-kinase activity; 3) we demonstrate novel in vivo activities for Kcs1 and Ipk1 which have ramifications for re-interpreting previously published work; and 4) we provide evidence for new phosphatase and inositol pyrophosphate synthase activities in yeast: designated Inp1 and Ips1, respectively.
Our initial hypothesis for the heterologous expression of Drosophila IPKs in ipk2 mutant yeast was that such experiments would allow us to distinguish between 6-and 3-kinase activities. These experiments also provide a means to "addback" I(1,4,5,6)P 4 and I(1,3,4,5)P 4 production in cells without complications due to protein components, as the Drosophila enzymes have less than 50 residues, out of over 350, in common with yeast Ipk2 (note: the concept of "add-back" was first proposed in the York lab by Drs. Audrey Odom and Jill Stevenson-Paulik while studying the Arabidopsis thaliana Ipk2). While dmIpk2, and inferred 6-kinase activity, was able to complement growth on ornithine as a sole nitrogen source and growth at high temperatures, dmIP3K I(1,4,5)P 3 3-kinase activity was only able to complement temperature growth. Thus, providing the first evidence that synthesis of I(1,3,4,5)P 4 was unable to restore regulation of gene expression as judged by growth on ornithine. These data further support our initial evidence that I(1,4,5,6)P 4 production plays a role in gene expression through studies of the ipk2-3 mutant which appears to be a 6-kinase selective enzyme in cells (5). This work also demonstrates that production of I(1,4,5,6)P 4 , but not I(1,3,4,5)P 4 , was able to bypass a kinaseindependent function of Ipk2 in transcriptional control as proposed by Messenguy and colleagues (30). Subsequent work of this group (31), and work of the O'Shea and Wu labs (6,7), further support our initial claim that Ipk2's kinase activity was and still is required for its role in gene expression and biological processes (for example, ArgR-Mcm1 transcription, chromatin remodeling, by guest on March 24, 2020 http://www.jbc.org/ Downloaded from growth on arginine or ornithine as a sole nitrogen source; and growth at high-temperatures).
Perhaps the most surprising aspect of this work arose from the metabolic analysis of ipk2 deficient cells expressing dmIP3K. In these cells, we expected to observe conversion of I(1,4,5)P 3 and stoichiometric accumulation of I(1,3,4,5)P 4 . However as Figs. 2 and 3 illustrate, we instead found over 11 new species of IPs and PP-IPs that were downstream metabolites of I(1,3,4,5)P 4 production. Of interest, when we expressed dmIP3K in wild-type yeast, we did not observe stimulation of this pathway and the metabolic profiles of these cells were identical to wild-type, having a signature major peak of IP 6 (not shown). These data indicate that when Ipk2 was present the 6-kinase pathway was the major route of metabolism for I(1,4,5)P 3 . Having said this, the 3kinase pathway was present in wild-type and ipk2 deficient cells, albeit at low levels as compared to IP 6 synthesis.
Our data helps interpret the observations that we and others made previously that several other low abundance IPs are present in ipk2 null and wild-type cells (5,8,18). In contrast to the report of Saiardi et al (18), we found that ipk2Δ cells do not generate I(1,3,4,5,6)P 5 , IP 6 , PP-IP 5 α and PP 2 -IP 4 α . Rather, they synthesize I(1,2,3,4,5)P 5 , PP-IP 3 α, PP-IP 4 β , PP 2 IP 2 α , and PP 2 IP 3 β (see Fig 3). Of note, when using a Partisphere SAX HPLC column, the PP 2 -IP 2 α species that is generated co-elutes with IP 6 , which may have led to its misidentification. We therefore, confirmed our results through treatment of the ipk2 null IP extracts with DIPP and found that the PP 2 -IP 2 α peak was completely hydrolyzed (not shown). Additionally, DIPP treatment of the extracts did not result in the formation of IP 6 , providing evidence that the PP-IPs were only synthesized from IP 4 and/or I(1,2,3,4,5)P 5 . We note that the data presented in this study do not reveal the chemical structures of the PP-IP species produced in the pathway, specifically which ringpositions harbor the pyrophosphate. However their unique elution profiles and their synthesis via different precursors provided evidence that they appear structurally distinct (see Fig 8).
The molecular and biochemical analysis of the 3-kinase pathway has allowed us to define most species and the gene products required for their synthesis. We now provide a revised genetic and metabolic map of IP/PP-IP metabolism in budding yeast (Fig. 8). Unlike the Ipk2/Ipk1dependent pathway that synthesized IP 6 , the IP3Kdependent pathway required Kcs1 (possibly Ipk2), Ipk1, and Ips1 (possibly Ids1). Interestingly, despite that role of these kinases in both yeast pathways, their sequential order, substrates and, in at least one case, activities were unique to each pathway.
Through our studies we have assigned a novel activities and specificities to Kcs1 and Ipk1. We assign an I(1,4,5,)P 3 3-kinase activity to Kcs1 that was functional in cells. This extends previous work of Dubois and co-workers that ascribed an undetermined I(1,4,5)P 3 kinase activity to Kcs1 (13). The metabolic phenocopy of dmIP3K and Kcs1 expression in ipk2 null cells, and the loss of I(1,3,4,5)P 4 production observed in a ipk2 kcs1 double mutant demonstrate this new activity is present in cells. It is not entirely surprising that Kcs1 can function as an I(1,4,5)P 3 kinase given that it is evolutionarily related to two I(1,4,5)P 3 kinases, Ipk2/IPMKs and IP3Ks (17,35), thus it may have retained I(1,4,5)P 3 kinase activity, in addition to acquiring pyrophosphate synthase activities. Also, we note in the context of wildtype cells, our data do not exclude a role for Ipk2 in initiating the 3-kinase pathway. We show for the first time, that Ipk1 from yeast acts as a 2kinase on I(1,3,4,5)P 4 . Analysis of the kinetic properties of this reaction indicated that this is not as efficient as those described for other Ipk1 substrates, but clearly under conditions of I(1,3,4,5)P 4 accumulations in cells, this activity is relevant. Thus, it is possible that cellular alterations in Ipk2, Ipk1 and/or Kcs1 activity or specificity would regulate flux to either the 6-or 3-kinase pathways.
The discovery of the 3-kinase pathway has also enabled the identification of new IP phosphatase and inositol pyrophosphate synthase activities. Recently, we reported the existence of Ids1, an inositol pyrophosphate synthase present in cells lacking Kcs1 and DIPP whose activity appeared to convert IP 6 to PP-IP 5 β (10). Here we provide evidence of a inositol pyrophosphate synthase, designated Ips1, capable of phosphorylating I(1,2,3,4,5)P 5 to generate PP-IP 4 β. At this point we have not determined if these are two distinct enzymes or a single gene product, nor have we determined which phosphate position serves as an acceptor for the synthesis of the pyrophosphate. Lastly, we find evidence for an IP 1-phosphatase activity, designated Inp1, which does not appear to be related to INPP1, a lithium-inhibited inositol polyphosphate 1phosphatase (36,37). Our previous studies of ipk2 mutant cells showed evidence for this activity towards I(1,4,5)P 3 substrates (5,8) and here we find evidence that such an activity also utilizes I(1,3,4,5)P 4 to generate I(3,4,5)P 3 . The molecular identity of Inp1 has yet to be established. However, we have ruled out that this activity was encoded by Inp5's and S A C 1-like inositol phosphatases (5,8,(38)(39)(40) based on genetic and biochemical analysis (Bryan Spiegelberg and John York, unpublished results).
The discovery of the 3-kinase pathway may have important ramifications for interpreting genetic evidence. The 6-kinase and IP 6 synthesis pathway has been implicated in the regulation of cellular functions including transcription, chromatin remodeling, RNA export, vacuole function, DNA metabolism and telomere maintenance (10,15,17). Clearly, the 3-kinase pathway that remains in ipk2 deficient cells does not appear to compensate for loss of 6-kinase, thereby supporting the parsimonious explanation that the 6-kinase pathway is most relevant to these functions. However, upregulation of the 3-kinase activity by expression of d m IP3K or Kcs1 indicated that some of the IP/PP-IPs generated by this pathway were able to dosage-compensate for at least some, but not all, of the functions attributed to 6-kinase metabolites. It is intriguing to speculate the alterations in Ipk2 specificity (i.e. 6-versus 3-kinase), and/or Kcs1/Ipk1/Ids1/Ips1 activity may provide the yeast cell with a complex repertoire of signaling molecules to enable adaptation to changes in cellular environment.  1 The abbreviations used are: PLC, phospholipase C; I(1,4,5)P 3 , inositol 1,4,5-trisphosphate HPLC, high pressure liquid chromatography; V max , maximal velocity attained with excess substrate; K m , substrate concentration permitting a half-maximal velocity, PP-IP 4 , diphosphoinositol tetrakisphosphate; IP, inositol polyphosphate; IP 3 , inositol trisphosphate; IP 4 , inositol tetrakisphosphate; IP 5 , inositol pentakisphosphate; IP 6 , inositol hexakisphosphate; Ipk2, inositol polyphosphate kinase 2; dmIP3K, Drosophila I(1,4,5)P 3 3-kinase beta isoform; DIPP, diphosphoryl inositol polyphosphate phosphohydrolase.     100ng recombinant dmIP3K (B), or both dmIP3K and scIpk1 (C) in a buffer containing 50 mM HEPES (pH 7.5), 50 mM NaCl, 20 mM MgCl 2 , and 4 mM ATP. A duplicate reaction to C was treated with human Type I 5-phosphatase at 37 ˚C (D). In panels A-D, reaction products were resolved by Partisphere strong anion exchange HPLC and elution profile of a I(1,3,4,5,6)P 5 standard was superimposed in panel C -gray line.
(E) The kinetic parameters of scIpk1 phosphorylation of I(1,3,4,5)P 4 were determined using the above reaction conditions while the I(1,3,4,5)P 4 concentration was varied. K m and V max values were obtained from a nonlinear curve fit to the Michaelis-Menten equation using GraphPad Prism version 4.01. The R 2 value was 0.9697. Fig. 5. Enzymatic identification of IP and PP-IP species present in the I(1,4,5)P 3 3-kinase pathway. DIPP and Type I 5-phosphatase treatment of labeled yeast extracts. Radiolabeled IP extracts were prepared from ipk2 deficient cells expressing dmIP3K and digested with control (no enz), 500 ng recombinant human type I inositol polyphosphate 5-phosphatase (5Ptase), 500 ng recombinant human diphosphoryl inositol polyphosphate phosphatase (DIPP) or both (DIPP/5Ptase). The resulting reactants were then separated by HPLC as described in figure 2. Elution positions of various IP and PP-IP species are indicated. Fig. 6. Pulse-chase analysis of IP metabolism in yeast mutants expressing dmIP3K. ipk2Δ (A) or ipk2Δ ipk1Δ (B) expressing pRS314 vector containing dmIP3Kβ were grown overnight to logarithmic phase, pulse labeled with 1 mCi/ml [ 3 H]-inositol for 10 min., washed and chased for indicated times with medium containing excess cold-inositol. Radiolabeled extracts were harvested and analyzed by Partisphere strong-anion exchange HPLC as described for Fig. 2. Elution positions of various IP and PP-IP species are indicated.   --+ -+ +, ≥50% of the total IP 3 , IP 4 , or IP 5 content is 5-phosphatase susceptible; -, ≥50% of total IP 3 , IP 4 , or IP 5 content is 5-phosphatase resistant; ND, not determined. 5phosphatase susceptibility was determined by comparing HPLC traces of extracts incubated with or without 5-phosphatase.