Biochemical and Functional Characterization of Inositol 1,3,4,5,6-Pentakisphosphate 2-Kinases*

Synthesis of inositol 1,2,3,4,5,6-hexakisphosphate (IP6), also known as phytate, is integral to cellular function in all eukaryotes. Production of IP6 predominately occurs through phosphorylation of inositol 1,3,4,5,6-pentakisphosphate (IP5) by a 2-kinase. Recent cloning of the gene encoding this kinase fromSaccharomyces cerevisiae, designated scIpk1, has identified a cellular role for IP6 production in the regulation of mRNA export from the nucleus. In this report, we characterize the biochemical and functional parameters of recombinant scIpk1. Purified recombinant scIpk1 kinase activity is highly selective for IP5 substrate and exhibits apparent K m values of 644 nm and 62.8 μm for IP5 and ATP, respectively. The observed apparent catalytic efficiency (k cat/K m ) of scIpk1 is 31,610 s− 1 m − 1. A sequence similarity search was used to identify an IP5 2-kinase from the fission yeastSchizosaccharomyces pombe. Recombinant spIpk1 has similar substrate selectivity and catalytic efficiency to its budding yeast counterpart, despite sharing only 24% sequence identity. Cells lackingsc-IPK1 are deficient in IP6 production and exhibit lethality in combination with a gle1 mutant allele. Both of these phenotypes are complemented by expression of thespIPK1 gene in the sc-ipk1 cells. Analysis of several inactive mutants and multiple sequence alignment of scIpk1, spIpk1, and a putative Candida albicans Ipk1 have identified residues involved in catalysis. This includes two conserved motifs: E(i/l/m)KPKWL(t/y) and LXMTLRDV(t/g)(l/c)(f/y)I. Our data suggest that the mechanism for IP6 production is conserved across species.

Inositol polyphosphates (IPs) 1 in eukaryotic cells are key regulatory molecules whose levels transiently fluctuate in response to diverse cellular stimuli (1,2). A major route for synthesis of IPs is through activation of phosphatidylinositolspecific phospholipase C. Phospholipase C cleaves lipids such as phosphatidylinositol 4,5-bisphosphate to generate inositol 1,4,5-trisphosphate, a regulator of calcium efflux from the endoplasmic reticulum. The release of a soluble inositol head group from its anchoring lipid also represents the first step in the pathway for generation of more highly phosphorylated inositols (3). The most abundant of these is inositol 1,2,3,4,5,6hexakisphosphate (IP 6 ), also known as phytate. IP 6 can represent up to 1% of the mass of a plant seed, where it may serve as an antioxidant and a phosphate storage source (4,5). The role of IP 6 is less clear in mammalian cells, although there is evidence suggesting that it may regulate inflammation, neurotransmission, and cell growth (reviewed in Ref. 3).
Recently, a metabolic pathway converting inositol 1,4,5-trisphosphate to IP 6 was delineated in budding yeast Saccharomyces cerevisiae cells (6)(7)(8)(9). It has been shown that IP 6 also serves as a precursor for diphosphorylated inositols, such as diphosphoryl inositol 1,3,4,5,6-pentakisphosphate (PP-IP 5 ), in both yeast and vertebrate cells (3,10). Combined in vivo and in vitro studies show that three genes (PLC1, IPK1, and IPK2/ARG82) account for the pathway converting phosphatidylinositol 4,5-bisphosphate to IP 6 . Ipk2 is a dual-specificity inositol-1,4,5-trisphosphate 6-kinase and inositol-1,4,5,6-tetrakisphosphate (IP 4 ) 3-kinase (7). Ipk1 is an IP 5 2-kinase (6). Physiological roles for soluble IPs in at least two nuclear functions have been established: mRNA export and transcription. Mutant alleles of plc1, ipk1, and ipk2 were all identified in a synthetic lethal genetic screen with a temperature-sensitive gle1-2 mutant defective for an essential mRNA export factor (6,11). Moreover, the plc1, ipk1, and ipk2 mutants each individually have mRNA export defects at a restrictive growth temperature (6). These mutants do not show defects in nuclear envelope morphology, nuclear protein import, or protein export (6). Overexpression of PLC1 in cells suppresses the temperature sensitivity of the gle1-4 mutant in a manner that depends on a functional IP 5 2-kinase (6). 2 Taken together, these results suggest a direct role for this IP pathway, specifically IP 6 production, in either stimulating or regulating mRNA export. Interestingly, Ipk2 is identical to Arg82, a regulator of the ArgR⅐Mcm1 transcription complex (7,12,13). Ipk2 may control transcriptional responses by at least two mechanisms. First, the Ipk2 protein itself, but not IP generation, is necessary for assembly of ArgR⅐Mcm1 complexes on DNA promoter elements (7). Second, the production of IP 4 /IP 5 through both phospholipase C and Ipk2 kinase activities is required for proper execution of ArgR⅐Mcm1-mediated gene expression (7). In this way, complexes may be poised on DNA awaiting activation of inositol signaling for direct regulation of gene expression. Subcellular localization of Ipk2 in the nucleus and Ipk1 in the nucleus and at the nuclear envelope further suggests that these enzymes constitute a nuclear signaling pathway (6,7).
Besides S. cerevisiae, IP 5 2-kinase activities from other organisms have been also been characterized (summarized in Refs. 3 and 14). However, Ipk1 from S. cerevisiae, designated scIpk1, represents the first to be characterized at the molecular level (6). The 281-residue polypeptide sequence of scIpk1 does not show detectable similarity to other IP kinases. To further understand the structure and function of IP 5 2-kinases, we report the biochemical and functional characterization of scIpk1 and the identification of an IP 5 2-kinase from the fission yeast Schizosaccharomyces pombe. Our studies have defined key residues important for catalysis and demonstrate that the budding and fission Ipk1 yeast proteins are functionally interchangeable despite sharing minimal sequence similarity. Overall, Ipk1 may represent a prototypic member of a new class of IP kinase.

MATERIALS AND METHODS
Strains and Media-Yeast strains were grown either in 1% yeast extract and 2% peptone or in synthetic minimal medium plus appropriate amino acids supplemented with 2% glucose. Yeast transformations were completed by the lithium acetate method (15), and general genetic manipulations of yeast cells were conducted as described (16). 5-Fluoroorotic acid (5-FOA) was obtained from United States Biologicals and used at a working concentration of 0.5 mg/ml. The S. cerevisiae strains used in this study are described in Table I. DH5␣ was used as the bacterial host for all plasmids. Bacterial strains were cultured in LB medium and transformed by standard methods (17).
Bacterial Glutathione S-Transferase (GST) Expression Plasmids-The plasmid for expression of GST-scIpk1 fusion protein in bacteria was described previously (6). For the GST-spIpk1-C plasmid, sequence encoding spIPK1 residues 357-640 was made by PCR amplification of purified S. pombe genomic DNA (the generous gift of Dr. Paul R. Russell, Scripps Research Institute, La Jolla, CA) using 4 ng/l sense and antisense primers (5Ј-atgcatgcagaattcgTTATATCGAATCCGCTT-TTTAAATCC-3Ј and 5Ј-atgcatgcagtcgaccTTACTTTATGCTGCTCCGG-CATG3-Ј, respectively; uppercase letters denote spIpk1 coding sequence) with Taq polymerase and manufacturer's buffer (Roche Molecular Biochemicals). The reactions were cycled 30 times under the following conditions: denaturing at 94°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 2 min. The resulting product was digested with EcoRI and SalI, gel-purified, and inserted into the EcoRI and SalI sites of pGEX-4T. The regions generated by PCR were sequenced to confirm the product. The sequences of the oligonucleotides for cloning spIPK1-C were based on a candidate S. pombe IP 5 2-kinase (283-residue partial-length gene product; GenBank TM /EBI Data Bank accession number D89240). The candidate was identified by sequence comparison of the primary amino acid sequence of scIpk1 with the GenBank TM /EBI non-redundant data base (November 1998 release) using the gapped BLASTP program (18).
The plasmids harboring GST fusion proteins of the five sc-ipk1 mutant alleles were generated as follows. The sc-ipk1 mutant strains were isolated from a synthetic lethal screen with a gle1-2 mutant (Table I) (6). Genomic DNA was prepared from wild-type and mutant cells for use as a PCR template using 4 ng/l sense and antisense oligonucleotides (5Ј-GCACATGTAGAATTCCATATGCAAGTCATCGGACGTGGTGGG-GCA-3Ј and 5Ј-GCACATGTACGTCGACCTGCCAGTACCAAAGGTGG-AAAG-3Ј, respectively). To generate IPK1 fragments, KlenTaq or Taq polymerase was used with the respective manufacturer's buffer (Sigma and Roche Molecular Biochemicals). The reactions were conducted using a decremental annealing temperature program for 25 total cycles with denaturing at 96°C for 1 min, annealing over a range from 59 to 49°C (decreasing by 0.4°C/cycle) for 1 min, and extension at 72°C for 2 min, followed by 10 additional cycles at 48°C annealing. The resulting product was purified and digested with EcoRI and SalI and inserted into the EcoRI and SalI sites of pGEX-4T. The integrity of this and all other constructs was verified by DNA sequencing using the ABI Prism dGTP BigDye Terminator Ready Reaction kit (PE BioSystems). The GST expression plasmids for each of the sc-ipk1 alleles were as follows: sc-ipk1-1, pSW1283; sc-ipk1-2, pSW1284; sc-ipk1-3, pSW1292; sc-ipk1-4, and pSW1285; sc-ipk1-5, pSW1287. In the course of constructing the pGST-scIpk1 plasmid, a construct with a PCR mutation was isolated and further characterized (pSW1288). The mutation results in a change of the codon AAT for Asn 10 to GAT for Asp, designated sc-ipk1-6.
Protein Purification of GST Fusion Proteins-All proteins were expressed in DH5␣. For large-scale purifications, 1-liter cultures of bacteria in LB medium were allowed to grow at 37°C to A 600 ϭ 0.6. Fusion protein expression was induced by addition of 0.3 mM isopropyl-␤-Dthiogalactopyranoside (final concentration), and growth was continued for 4 h at 37°C. Cells were pelleted and resuspended in 20 ml of lysis buffer (50 mM Hepes (pH 7.5), 50 mM KCl, 8 mM dithiothreitol, and 1 mM phenylmethylsulfonyl fluoride). After incubation on ice for 15 min, the cells were disrupted by sonication. The lysate was clarified by a 10,000 ϫ g spin for 30 min, and the resulting supernatant was incu- bated with a 1-ml bed volume of glutathione-Sepharose beads at 4°C for 1 h (with agitation). The beads were washed with 20 ml of lysis buffer and eluted in glutathione buffer (10 mM glutathione, 5 mM dithiothreitol, 50 mM Tris-HCl (pH 8.0), and 50 mM NaCl.). For smallscale purifications, 50-ml cultures of cells were grown in LB medium and induced for 3 h as described above. The cells were collected, resuspended in 0.5 ml of lysis buffer, incubated on ice for 15 min, and disrupted by sonication. Lysates were clarified by a 13,000 ϫ g spin for 5 min at 4°C. The resulting supernatant was incubated with 50 l of glutathione-Sepharose for 1 h at 4°C. The beads were washed with 3 ml of lysis buffer and eluted in glutathione buffer.
Enzymatic Analysis of IP 5 2-Kinase Activity-Enzymatic activity was measured using [ 32 P]IP 5 . For kinetic analysis, standard reactions were conducted in a total volume of 10 l. Purified GST fusion proteins were diluted in buffer containing 50 mM Hepes (pH 7.5), 2 mM MgCl 2 , 20 mM KCl, and 0.2 mg/ml bovine serum albumin and added in a 2-l volume to an 8-l reaction mixture of 100 mM Hepes (pH 7.5), 10 mM MgCl 2 , ATP, and IP 5 . Individual kinetic parameters for scIpk1 were determined by varying the concentration of ATP (0.02-5 mM) while holding IP 5 constant (20 M). Alternatively, IP 5 was varied (0.125-32 M) while holding ATP constant (5 mM). Parameters for spIpk1-C were determined using a range of IP 5 concentrations from 0.5 to 32 M and a range of ATP concentrations from 0 to 400 M. The amounts of recombinant protein added to individual assays and the time of incubation at 30°C were varied to maintain substrate conversion within a linear range (5-50% conversion). The reactions were stopped by addition of 2 l of TLC running buffer (1.08 M KH 2 PO 4 , 0.64 M K 2 HPO 4 , and 1.84 M HCl). The samples were spotted onto Baker-flex polyethyleneimine cellulose thin-layer chromatography sheets (J. T. Baker Inc.) and allowed to dry. Separation of IP 5 and IP 6 by TLC was achieved by incubation of the sheet in a tank of TLC running buffer for ϳ45 min. Sheets were air-dried, and the radioactivity was quantified using a PhosphorImager and ImageQuant software (Molecular Dynamics, Inc.).
S. cerevisiae Expression Plasmids-For the LEU2/CEN plasmid with scIPK1 under the control of the GLE1 promoter (pSW898), sequence encoding scIpk1 residues 2-281 was made by PCR amplification of a library plasmid harboring scIPK1 (pSW832) using 4 ng/l sense and antisense oligonucleotides (5Ј-AGATCCCATGGAGTCATCGGACGTG-GTGGG-3Ј and 5Ј-ACCAGTCGACGAAAGAAAAGTATACAG-3Ј, respectively) with KlenTaq polymerase and manufacturer's buffer (Sigma). The reactions were cycled 30 times under the following conditions: denaturing at 96°C for 30 s, annealing at 48°C for 2 min, and extension at 72°C for 1.5 min. The 875-base pair product was digested with NcoI and SalI, gel-purified, and inserted into the NcoI and SalI sites of pSW747 (19). This resulted in the sequence encoding full-length scIpk1 fused in frame to the sequence encoding the first 6 amino acid residues of Gle1. The spIPK1/LEU2/CEN expression construct (pSW1255) was generated by ligating the EcoRI/SalI fragment from the GST-spIpk1 construct into the NcoI and SalI sites of pSW747 with the linker oligonucleotides 5Ј-CATGGGAGGTGGAG-3Ј and 5Ј-AATTCTCCAC-CTCC-3Ј. The GLE1/LEU2/CEN expression plasmid was described previously (pSW406) (11), as was the LEU2/CEN vector (pRS315) (20).
Inositol Radiolabeling of S. cerevisiae Cells and Separation of Radiolabeled IPs-The IP profile of yeast strains was determined as described previously (6,21). Briefly, cells were grown in complete synthetic medium containing 40 Ci/ml [ 3 H]inositol to late logarithmic phase. Approximately 3 ϫ 10 7 cells were harvested and resuspended in 100 l of 0.5 N HCl. Soluble IPs were extracted by adding 372 l of chloroform/methanol (1:2, v/v) and 100 g of glass beads. The mixture was aggressively agitated for 2 min, followed by addition of 125 l of chloroform and 125 l of 2 M potassium chloride and another 2 min of agitation. The lysates were clarified by a 13,000 ϫ g spin for 5 min, and the supernatant was recovered. Samples were analyzed by HPLC, with the IPs resolved by a Partisphere strong-anion exchange column (4.6 ϫ 125 mm) and a linear gradient from 10 mM to 1.7 M ammonium phosphate (pH 3.5) over 25 min, followed by elution for 45 min with 1.7 M ammonium phosphate.

RESULTS
Biochemical Characterization of Recombinant scIpk1-To characterize the biosynthesis of IP 6 by scIpk1, kinetic experiments were performed with bacterially expressed and purified GST-scIpk1. The ability of the enzyme to synthesize IP 6 was assayed by incubation with [ 32 P]IP 5 and ATP in a buffered system. The reaction products were analyzed by polyethyleneimine cellulose thin-layer chromatography, which readily resolved IP 5 and IP 6 isomers (Fig. 1A). No IP 6 was formed at time 0 incubation, in the absence of ATP, in the absence of GST-scIpk1, or in the presence of the stopping buffer. Reactions were determined to be dose-and time-dependent ( Fig. 1A; data not shown). The optimal pH range, temperature, and MgCl 2 concentrations were determined under zero-order conditions to be pH 6.7-7.5, 30°C (89% maximal activity was observed at 37°C), and 10 mM, respectively (data not shown).
Michaelis-Menten kinetic parameters for GST-scIpk1 were determined by monitoring the conversion of IP 5 to IP 6 under a variety of conditions. Data from three experiments are shown as saturation curves with specific activity (micromoles of IP 6 formed per min/mg of protein) plotted versus the IP 5 substrate concentration (Fig. 1B). Apparent kinetic parameters were calculated from the best fit line derived from double-reciprocal plots of 1/v versus 1/S (shown in the inset).The average apparent K m and V max for IP 5 were found to be 644 nM and 0.020 mol/min/mg, respectively, using 40 ng of recombinant protein and 5 mM ATP. This corresponds to an apparent catalytic efficiency (K cat /K m ) of 31,610 s Ϫ1 M Ϫ1 . Similar experiments were performed with constant GST-scIpk1 and IP 5 levels (80 ng and 20 M, respectively) and varying concentrations of ATP (0.020 to 5 mM). Data from the experiments are shown in Fig.  1C. The apparent K m for ATP was determined to be 62.8 M.
Identification of S. pombe and C. albicans Ipk1 Homologues-To determine whether homologues of scIpk1 were present in the available sequence data bases, we conducted a series of searches with the gapped BLASTP program (18). Computer analysis identified putative proteins from S. pombe (GenBank™/EBI accession number CAB60684.1) and C. albicans (Stanford DNA Sequencing and Technology Center accession number Contig5-2849) with high probability scores (Fig.  2). The C-terminal region of the S. pombe protein (residues 375-640) has 24% identity and 53% similarity to full-length scIpk1. Based on the subsequent characterization described below, this protein has been designated spIpk1. The sequence in the C. albicans database represents a candidate IP 5 2-kinase (designated caIpk1). The 361-amino acid residue sequence from C. albicans completely overlaps with the sequences for scIpk1 and the C-terminal region of spIpk1. The three sequences share ϳ24% identity in all pairwise comparisons. The pairwise alignments (data not shown) and an overall alignment (Fig. 2B)  . Numbers indicate the amino acid positions. The C-terminal region of spIpk1 shows high similarity to full-length scIpk1 and putative caIpk1 (dark-gray boxes), including two conserved amino acid spans designated Boxes A and B (hatched boxes). The N-terminal region of spIpk1 has a region (residues 114 -253; light-gray boxes) with homology to scRvs167 (19% identity; residues 128 -272) and scRvs161 (20% identity; residues 116 -214). A span within the Rvs homology region of spIpk1 is predicted to form a coiled-coil domain (residues 159 -188; black box). B, an alignment of the related sequences from scIpk1, spIpk1-C, and putative caIpk1 was generated using multiple sequence alignment analysis (22). Black boxes indicate residues that are identical between at least two of the proteins, and gray boxes highlight similar residues. The spans of amino acid sequences for Boxes A and B are shown. The asterisks designate residues that were changed in the sc-ipk1-6 (N10D) and sc-ipk1-2 (C139Y) mutants. The carets note the points of truncation for the sc-ipk1-4 (Trp 128 to a stop codon), sc-ipk1-5 (Trp 237 to a stop codon), and sc-ipk1-1/3 (Gln 268 to a stop codon) mutants. of spIpk1, and residues 250 -261 of putative caIpk1. The consensus sequence for each motif was deduced as E(i/l/m)KPK-WL(t/y) for Box A and as LXMTLRDV(t/g)(l/c)(f/y)I for Box B.
Interestingly, the N-terminal region of spIpk1 is not present in scIpk1 or putative caIpk1. The open reading frame for spIpk1 encodes a predicted 640-amino acid protein with a molecular mass of ϳ73 kDa. This compares with 281 residues and ϳ33 kDa for scIpk1 and 361 residues and ϳ43 kDa for putative caIpk1. A search of the S. cerevisiae data base by BLASTP Version 2.0 with the spIpk1 sequence revealed that a span in the N-terminal region of spIpk1 is related to scRvs161 and the N-terminal region of scRvs167 ( Fig. 2A). The scRvs161 and scRvs167 proteins may play roles in endocytosis, actin function, and cell signaling (23)(24)(25)(26)(27). These regions of the yeast Rvs167 and Rvs161 proteins as well as the related region in the mammalian amphiphysins are predicted to form coiled-coil domains that mediate the formation of hetero-and homodimers (28 -32). Using the SMART program (33,34) to identify modular domains in spIpk1, residues 102-131 were designated as having a high probability of forming a coiled-coil structure. There were no regions in scIpk1 or putative caIpk1 that were predicted to form a coiled-coil domain.
The C-terminal Region of spIpk1 is an IP 5

2-Kinase That Functionally Complements S. cerevisiae ipk1 Mutants in
Vivo-To directly test if the C-terminal region of spIpk1 (spIpk1-C) is an intrinsic IP 5 2-kinase, recombinant spIpk1-C was purified as a GST fusion protein and analyzed for biochemical activity. Incubation of GST-spIpk1-C with pure IP 5 and [␥-32 P]ATP resulted in the specific conversion of IP 5 to IP 6 (data not shown). For the kinetic characterization, the experimental conditions were as described for GST-scIpk1, except that purified GST-spIpk1-C at 6 ng/reaction was used with substrate concentration ranges appropriate for the given K m . Varying the IP 5 concentration by 2-fold dilutions between a range of 0.5 and 32 M resulted in the determination of an average apparent K m for IP 5 of 5.89 M and V max of 0.240 mol/min/mg (n ϭ 3) (Table II). The apparent K m for ATP was 9.3 M (determined with 30 ng of protein/reaction). Under these conditions, the S. pombe enzyme had a lower affinity for IP 5 substrate, but a higher maximal velocity compared with the S. cerevisiae enzyme; however, the apparent catalytic efficiencies of the enzymes are similar (31,610 versus 42,105 s Ϫ1 M Ϫ1 ) (Table II).
To test if the IP 5 2-kinase region of spIpk1 is functional in budding yeast, a LEU2/CEN plasmid expressing the carboxylterminal region of spIPK1 under the control of the S. cerevisiae GLE1 promoter was constructed (spIPK1-C). The spIPK1-C plasmid, an scIPK1 expression plasmid constructed in an identical manner, and the empty control vector were each transformed into the S. cerevisiae ipk1 null strain. The cells were grown in the presence of [ 3 H]inositol, and whole cell lysates were extracted and analyzed by HPLC (Fig. 3). As expected, the empty vector control strain showed no IP 6 production and strong peaks of IP 5 and PP-IP 4 . The spIPK1-C strain showed a significant increase in the IP 6 level; and therefore, spIpk1-C has activity in vivo. Peaks of IP 5 and PP-IP 4 were still present in the sc-ipk1⌬spIPK1-C strain. However, it should be noted that the scIPK1 construct also did not restore full IP 6 production. This suggests that the plasmid-based expression by the GLE1 promoter was not sufficient. The subtle differences between the spIPK1-C and scIPK1 profiles in terms of the levels of IP 5 and PP-IP 4 may also reflect differences in the kinetic properties of the proteins or subcellular localization.
To further examine potential functional overlap between spIPK1-C and scIPK1, we tested if expression of spIPK1-C could complement the gle1-2 ipk1-4 synthetic lethal phenotype. The LEU2/CEN expression plasmids harboring spIPK1-C, scIPK1, or scGLE1 were transformed into the gle1-2 ipk1-4 tester strain containing plasmid-borne (URA3/CEN) wild-type GLE1. The assay for complementation of synthetic lethality was based on the toxicity of the drug 5-FOA in a URA3 background. Only strains that can lose the URA3 plasmid and survive without the expressed GLE1 gene will be viable on 5-FOA medium. As expected, the tester strain carrying either the GLE1/LEU2 or scIPK1/LEU2 plasmid was viable, and the strain with the empty LEU2 vector was dead (Fig. 4). Strikingly, the strain with the spIPK1-C plasmid was viable at 23°C. Thus, spIPK1-C rescued the synthetic lethality and fulfilled the functional role of the IP 5 2-kinase required for Gle1 and mRNA export. Overall, these results indicate that the  Characterization of sc-ipk1 Mutant Strains-Five mutant alleles of sc-ipk1 were identified in the gle1-2 synthetic lethal screen (6) and designated sc-ipk1-1 to sc-ipk1-5. To investigate potential phenotypic differences between the isolated mutant strains, cells were assayed for their ability to generate IP 6 (Fig.  5). Strains were grown in synthetic medium in the presence of [ 3 H]inositol (40 Ci/ml) until late logarithmic phase. Soluble cell lysates were prepared and separated by HPLC to detect specific IP species. Extracts from wild-type scIPK1 cells yielded a strong IP 6 peak (Fig. 5A), but showed no accumulation of the precursor IP 5 or the isomer PP-IP 4 . In all the mutant sc-ipk1 strains (Fig. 5), IP 6 production was greatly reduced, and the levels of PP-IP 4 and IP 5 were coincidentally increased. However, there were distinct differences between the mutant strains. There was no detectable IP 6 production in the sc-ipk1-2 and sc-ipk1-4 mutants (Fig. 5, C and E, respectively), and the profiles were similar to the profile of the previously reported sc-ipk1 null strain (Fig. 3A). In contrast, low levels of IP 6 were detected in the sc-ipk1-1 and sc-ipk1-3 mutant strains (Fig. 5, B  and D, respectively). The sc-ipk1-5 strain yielded the highest relative IP 6 production level among the mutants (Fig. 5F). There was also a significant accumulation of IP 5 in the sc-ipk1-5 cells, which is not present in the wild-type cells. The different levels of apparent enzymatic activity among the mutant alleles suggested that the mutations were distinct and potentially resided in distinct regions of the gene. In addition, the fact that all five are lethal in combination with a gle1 allele further indicated that the basis for the synthetic lethal phenotypes may be mechanistically distinct. The phenotypes could be due to either the lack of enzymatic activity or a perturbed spatial/temporal regulation of IP 6 production.
Cloning and Identification of sc-ipk1 Mutants-To delineate the structural basis for the mutant phenotypes, the full-length genes for the sc-ipk1 alleles were sequenced (see "Materials and Methods"). Isolates of each construct were tested to identify and verify the positions of the nucleotide changes (diagrammed in Fig. 6). Four of the mutant alleles (sc-ipk1-1, sc-ipk1-3, sc-ipk1-4, and sc-ipk1-5) contained nucleotide changes that resulted in nonsense mutations and premature stop codons. Alleles sc-ipk1-1 and sc-ipk1-3 (henceforth designated sc-ipk1-1/3) were identical according to the sequence data and coded for a protein truncated 14 amino acid residues before the carboxyl-terminal end of the wild-type protein. The fact that sc-ipk1-1 and sc-ipk1-3 are the same mutation correlates with their same relative in vivo activity (Fig. 5). Allele sc-ipk1-4 encoded a protein truncating at residue 127, resulting in the loss of over half of the wild-type sequence; and sc-ipk1-5 was a truncation of the carboxyl-terminal 45 residues. The fact that the truncation mutation in the sc-ipk1-5 allele was more extensive than that in the sc-ipk1-1/3 allele was surprising given that the sc-ipk1-5 strain appeared more enzymatically active in terms of IP 6 production levels in vivo (Fig. 5). This suggested that the 30 amino acid residues spanning the difference between the respective truncation points in the mutants may act as a negative regulatory region (amino acids 236 -268). Interestingly, allele sc-ipk1-2 contained a single point mutation that resulted in a substitution of the cysteine at residue 139 for a tyrosine (C139Y) (Fig. 6A). Combined with the in vivo data (Fig. 5), this suggested that this cysteine residue may play a key role in the catalytic mechanism. Interestingly, a cysteine residue at this position is conserved in the spIpk1 and putative caIpk1 sequences (Fig. 2B).
In Vitro Analysis of scIpk1 Mutant Proteins-To determine the effect of the mutations on the in vitro catalytic function of scIpk1, GST fusions of the mutant proteins were expressed and isolated from bacteria. An independent mutant allele was generated by PCR misincorporation and was designated sc-ipk1-6. The sc-ipk1-6 allele was the result of a change in the codon for asparagine at position 10 to that for aspartic acid (N10D). Initially, small-scale purifications were performed for each of the GST fusions (see "Materials and Methods"), and catalytic assays were conducted at set substrate concentrations (20 M IP 5 and 5 mM ATP) under the optimal conditions for the wildtype enzyme. These reactions were allowed to proceed for a total time of 2 h, and the percent total conversion was determined. Under these conditions, only GST-scIpk1-1/3 generated detectable amounts of IP 6 . After an 8-h incubation period, GST-scIpk1-6 did not yield IP 6 . Thus, the mutant proteins all appeared catalytically inactive.
To more accurately assess their activity, large-scale purifications were conducted for the GST-scIpk1-1/3, GST-scIpk1-2, and GST-scIpk1-5 proteins. Roughly equivalent amounts of mutant protein (ϳ20 ng/l) and various concentrations of wildtype protein were incubated under the above conditions for 1and 10-h time periods. The wild-type protein was catalytically active for at least 160 min under these assay conditions. In the 10-h incubation, IP 6 production was detected for the GST-scIpk1-1/3 and GST-scIpk1-2 proteins; however, the rates of production were less than ϳ4 and ϳ1% of the wild-type protein, respectively (Fig. 6B). Surprisingly, despite the apparent function of the mutant protein in vivo (Fig. 5), the GST-scIpk1-5 protein was catalytically inactive in vitro under all assay conditions tried to date. The lack of in vitro activity may reflect an instability or folding defect when GST-scIpk1-5 is expressed and purified from bacteria compared with the endogenous scIpk1-5 protein. DISCUSSION Previous molecular and biochemical analysis of inositol kinases and phosphatases has revealed high degrees of functional and structural conservation among respective enzyme family members across species. For example, similar inositol 5-phosphatases have been identified in mammals, S. cerevisiae, plants, Caenorhabditis elegans, and Drosophila (35). We recently reported the first molecular analysis of a gene encoding an IP 5 2-kinase, that from S. cerevisiae, termed scIpk1 (6). Here we establish that genes encoding putative IP 5 2-kinases with homology to scIpk1 exist in at least two other organisms, S. pombe and C. albicans. The S. pombe protein is a bona fide IP 5 2-kinase and is designated spIpk1. This conclusion is based on at least three pieces of evidence independent of the sequence homology. Biochemical characterization of purified recombinant GST-spIpk1-C showed that the protein was an in vitro catalyst for converting IP 5 and ATP to IP 6 . Expression of spIPK1-C in S. cerevisiae cells lacking scIpk1 restored the production of IP 6 in vivo. Finally, expression of spIPK1-C com- FIG. 4. Complementation of gle1-2 ipk1-4 synthetic lethality by expression of spIPK1-C. The S. cerevisiae gle1-2 ipk1-4 strain harboring GLE1 on a URA3/CEN plasmid (SWY1793) was transformed with the designated LEU2/CEN plasmids expressing GLE1, scIPK1, empty vector, or spIPK1-C (SWY2229, SWY2233, SWY2227, and SWY2228, respectively) (clockwise). Cells were streaked onto plates containing 5-FOA and incubated at 23°C for 4 days. Expression of GLE1, scIPK1, or spIPK1-C supported colony formation and growth on 5-FOA medium, whereas vector alone did not.
plemented the synthetic lethality of a gle1-2 sc-ipk1-4 double mutant. Thus, spIpk1 functioned in vivo and in vitro in a manner analogous to scIpk1. Given the phylogenetic span between S. pombe and S. cerevisiae, this conservation suggests that specific IP 5 2-kinase enzymes for generating the second messenger IP 6 may exist in all eukaryotes.
Despite sharing only 24% sequence identity, the recombinant scIpk1 and spIpk1 proteins have similar substrate selectivities and catalytic efficiencies (Table II). The apparent kinetic parameters for scIpk1 and spIpk1-C are also within the range of the reported parameters for the reported IP 5 2-kinase activity fractionated from immature soy beans: apparent K m values for IP 5 and ATP of 2.3 M and 8.4 mM, respectively, with an apparent V max of 0.243 mol/min/mg (36). Although others had documented that activities for producing IP 6 were present in cell extracts and fractions from S. pombe and C. albicans cells (14,37), the genes encoding the respective enzymes had not been identified, and the kinetic parameters for purified recombinant proteins had not been established.
The homology between the S. pombe, S. cerevisiae, and C. albicans proteins reveals at least two blocks of highly conserved amino acid sequence. We have designated these Box A (E(i/l/m)KPKWL(t/y)) and Box B (LXMTLRDV(t/g)(l/c)(f/y)I). These two motifs are separated by similar lengths of amino acid spans (93 residues for S. cerevisiae, 78 for S. pombe, and 104 for C. albicans). The Box A and B motifs in Ipk1 enzymes may form the catalytic and/or substrate-binding sites.
The location of the sc-ipk1 mutant alleles also highlights amino acid residues that are potentially important for catalysis or structure. In particular, the two point mutants that result in A, a schematic diagram of scIpk1 and the relative location of the changes for each allele is shown. The mutation in the respective codon is diagrammed, as is the corresponding predicted change in the amino acid sequence. B, the percentage of enzymatic activity for purified recombinant GST-scIpk1 mutant proteins was determined as compared with wild-type GST-scIpk1 (wt). n.d., not detected (indicates that the activity level was below the level of detection (Ͻ0.1%) for the assay conditions described under "Materials and Methods"). a complete loss of in vitro catalysis, C139Y and N10D, are conserved in spIpk1 and putative caIpk1. Three of the mutant sc-ipk1 alleles are C-terminal truncations of various lengths. The longest truncation (sc-ipk1-4) removed Box B and eliminated all detectable IP 5 2-kinase activity in vivo and in vitro. In contrast, the truncations that retained both Boxes A and B possessed some in vivo activity. This supports the role of the conserved motifs in catalysis. Interestingly, the removal of only the C-terminal 14 residues of scIpk1 (allele sc-ipk1-1/3) resulted in a significant decrease in catalytic activity. However, the sc-ipk1-5 mutant, with a further truncation and missing the C-terminal 45 residues, is actually more active in vivo in terms of IP 6 production. This suggests that the region between the sc-ipk1-1/3 and sc-ipk1-5 truncations (residues 237-268) is an inhibitory domain whose removal allows for a higher rate of activity.
The most striking difference between scIpk1 and spIpk1 is the unique N-terminal region of the S. pombe protein. We reanalyzed any potential open reading frames in the promoter region of scIPK1 and confirmed that there are no potential coding regions that would have homology to the N-terminal region of spIpk1 or that would extend the N-terminal portion of scIpk1. In addition, a chromosomally expressed, epitope-tagged version of scIpk1 was generated by homologous recombination at the C terminus. This results in expression of a fusion protein of the correct predicted molecular mass with the designated initiation methionine for scIpk1 (data not shown). Therefore, we believe that this is a true difference between these Ipk1 enzymes. However, as the S. pombe genome is not yet completely sequenced, it is possible that there are other IP 5 The N-terminal domain of spIpk1 contains a region that is predicted to form a coiled-coil structure and that is homologous to the coiled-coil domains of the mammalian amphiphysin and yeast Rvs161/167 proteins. The coiled-coil regions of the Rvs/ amphiphysin proteins have been shown to mediate dimerization (28 -32). We predict that the homology to spIpk1 reflects a conserved structural fold and that this region of spIpk1 is a protein-protein interaction domain for homo-or heterodimerization. It will be interesting to determine if this domain has any effect on the catalytic activity of spIpk1 and to identify protein interaction partners.
The amphiphysin and Rvs proteins have been implicated in a number of cellular processes, including endocytosis, actin function, and signaling (reviewed in Refs. 30, 38, and 39). There are also several intriguing possible connections between these proteins and inositol metabolism and IP signaling. Cells lacking scRvs167 or scRvs161 have altered phospholipid compositions (40). The role of amphiphysin-1 in clathrin-mediated endocytosis during synaptic vesicle reformation involves interactions with synaptojanin-1 (an inositol-polyphosphate 5-phosphatase), clathrin, and AP-2 (28,(41)(42)(43)(44)(45)(46). Interestingly, AP-2 has been characterized in vitro as an IP 6 -binding protein (47)(48)(49), and IP 6 alters the in vitro binding of AP-2 to synaptotagmin (50). Based on these reports, it was possible that scRvs167 and scRvs161 might have some function in the IP pathway. We therefore analyzed IP levels in S. cerevisiae cells lacking Rvs167 or Rvs161; however, no changes were observed (data not shown).
Our previous studies in S. cerevisiae have defined an in vivo role for IP 6 production in mediating mRNA export through the essential nuclear pore complex-associated factor Gle1 (6). In this report, we have conducted a complete in vivo and in vitro analysis of the IP 5 2-kinase activity for all the sc-ipk1 mutant alleles that were isolated in the original gle1 synthetic lethal screen. All showed marked in vivo reductions in IP 6 levels.
However, the low residual production of IP 6 by the sc-ipk1-5 mutant was surprising. This may indicate that the spatial and temporal regulation of IP 6 production is important to the mechanism of regulating mRNA export. A role for IP 6 in mRNA export in other eukaryotes has not yet been established. The existence of a human GlE1 (19) and a probable S. pombe Gle1 homologue (GenBank™/EBI accession number CAB39139.1) supports the hypothesis that the interrelationship between IP 6 production and Gle1 will be conserved.
The mechanism by which scIpk1 kinase activity and presumably IP 6 production regulate Gle1-mediated mRNA export is unknown. IP 6 levels have been shown to be relatively stable in cells, although recent work in yeast suggests that levels change in response to stress and other cellular perturbations (14). The turnover of IP 6 has not been carefully analyzed in S. cerevisiae, and is it not known whether or not Ipk1 functions as an ATP synthetase in vivo (the reverse reaction). It has also been demonstrated that IP 6 is the precursor for diphosphorylated inositols in both yeast and vertebrate cells (3,10). Others have suggested that production of PP-IP 5 or bis(diphospho)IP 4 (PP 2 -IP 4 ) may be required for mRNA export. However, the gene encoding an IP 6 kinase (KCS1) was not identified in the gle1 synthetic lethal screen (6), and mutations in KCS1 do not have defects in mRNA export (10). 3 Thus, it remains possible that IP 6 functions as a classic signaling molecule that directly binds an effector protein in the mRNA export machinery.
In summary, our characterization of scIpk1 and the new spIpk1 has defined their in vitro kinetic parameters, pinpointed residues important for catalysis, and documented cross-species functionality. The conservation of Ipk1 across species strongly indicates that the production of IP 6 is an important cellular function. Based on these results, we propose that a family of conserved IP 5 2-kinase enzymes exists in all eukaryotes. Given the low degree of sequence identity between the S. cerevisiae, S. pombe, and C. albicans enzymes, it may be difficult to identify putative mammalian IP 5 2-kinases based on BLAST data base searches. Our results showing heterologous complementation of an sc-ipk1 mutant background by expression of spIpk1 argues that genetic strategies to clone by heterologous complementation will be potentially powerful approaches for isolating other cross-species family members.