The Synthesis of Inositol Hexakisphosphate

The enzyme(s) responsible for the production of inositol hexakisphosphate (InsP6) in vertebrate cells are unknown. In fungal cells, a 2-kinase designated Ipk1 is responsible for synthesis of InsP6 by phosphorylation of inositol 1,3,4,5,6-pentakisphosphate (InsP5). Based on limited conserved sequence motifs among five Ipk1 proteins from different fungal species, we have identified a human genomic DNA sequence on chromosome 9 that encodes human inositol 1,3,4,5,6-pentakisphosphate 2-kinase (InsP5 2-kinase). Recombinant human enzyme was produced in Sf21 cells, purified, and shown to catalyze the synthesis of InsP6 or phytic acid in vitro. The recombinant protein converted 31 nmol of InsP5 to InsP6/min/mg of protein (V max). The Michaelis-Menten constant for InsP5 was 0.4 μm and for ATP was 21 μm. Saccharomyces cerevisiae lackingIPK1 do not produce InsP6 and show lethality in combination with a gle1 mutant allele. Here we show that expression of the human InsP5 2-kinase in a yeastipk1 null strain restored the synthesis of InsP6 and rescued the gle1–2 ipk1–4 lethal phenotype. Northern analysis on human tissues showed expression of the human InsP5 2-kinase mRNA predominantly in brain, heart, placenta, and testis. The isolation of the gene responsible for InsP6 synthesis in mammalian cells will allow for further studies of the InsP6 signaling functions.

Cells amplify and regulate signals through the generation of a variety of second messengers. The inositol polyphosphate family of second messengers has grown in complexity with the discovery of new functions for the soluble, more highly phosphorylated inositols. The common precursor of all soluble inositol phosphates in mammalian cells is Ins(1,4,5)P 3 , which is produced when phospholipase C cleaves phosphatidylinositol 4,5-bisphosphate yielding InsP 3 and diacylglycerol. InsP 3 is then metabolized to a number of more highly phosphorylated inositol species through the actions of several phosphatases and kinases; the cellular functions of these inositol polyphosphates are beginning to be elucidated (1)(2)(3)(4). An understanding of the enzymes responsible for the production of soluble inositol polyphosphates will be critical to establishing their roles in cellular physiology.
The major inositol pentakisphosphate isomer in eukaryotic cells, inositol 1,3,4,5,6-pentakisphosphate (InsP 5 ), 1 is converted to inositol hexakisphosphate (InsP 6 ) by phosphorylation at the D2 position of the inositol ring. A role for the product of the 2-kinase, InsP 6 or phytic acid, has been implicated in many cellular processes. InsP 6 has been shown to bind the clathrin assembly proteins AP2 and AP3 (5,6) and to inhibit clathrin cage assembly in vitro (7,8). InsP 6 inhibits serine and threonine protein phosphatases, which are thought to regulate Ltype Ca 2ϩ channels in pancreatic islet cells (9). Nonhomologous DNA end joining of double strand breaks is stimulated by InsP 6 (10) through its binding to the Ku70/80 subunits of DNA-PK (11,12). Most recently, InsP 6 has been suggested to stimulate endocytosis, possibly by the activation of protein kinase C and inhibition of synaptojanin (13).
The first role for InsP 6 in vivo was revealed by studies in the budding yeast Saccharomyces cerevisiae (14), in which the production of InsP 6 was shown to be required for efficient messenger RNA (mRNA) export. This is based on the results of a genetic screen for mutations that were lethal in combination with a temperature-sensitive gle1 mutant defective for the essential mRNA export factor, Gle1 (14). The synthetic lethal screen specifically identified the three gene products that together are responsible for converting phosphatidylinositol 4,5bisphosphate to InsP 6 (14,15). This included the previously characterized Plc1 (16) and two inositol polyphosphate kinases, Ipk1 and Ipk2 (14,15). Besides the genetic linkage between mutants defective in InsP 6 production and the gle1 mRNA export mutant, strains lacking the IPK1 gene alone show a marked accumulation of mRNA in their nuclei (14). This directly implicates the enzyme that produces InsP 6 in mRNA export.
Our ongoing studies have focused on testing whether the Ipk1 protein and InsP 6 function are conserved across species. Recent studies have identified IPK1 genes from two other fungi, Schizosaccharomyces pombe and Candida albicans (17). Although functionally conserved, the sequence identity is limited to a few small regions with high homology. However, there is, overall, less than 24% identity in all pairwise combinations across the fungal InsP 5 2-kinase domains. This lack of signifi-cant homology initially impeded the discovery of a nonfungal InsP 5 2-kinase. In particular, the enzymes for de novo synthesis of InsP 6 in mammalian cells are unknown. In this report, we have used a strategy of searching for proteins in data bases that share small, nearly identical amino acid sequences found in the fungal kinase domains. We have identified and characterized a protein from human cells that represents the first nonfungal InsP 5 2-kinase.

MATERIALS AND METHODS
Strains and Media-Escherichia coli strain DH5␣ was used as the bacterial host for all plasmids. Bacterial strains were cultured in LB medium and transformed by standard methods. 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 (18). 5-Fluoroorotic acid (5-FOA) was obtained from United States Biologicals and used at a concentration of 0.5 mg/ml. The S. cerevisiae strains used in this study include: SWY2105 (MATa ade2 ade3 ura3 his3 leu2 trp1 can1 ipk1::KAN r ) (kindly provided by S. Johnson) and SWY1793 (MATa ade2 ade3 ura3 his3 leu2 trp1 can1 gle1-2 ipk1-4 pSW611 (GLE1/URA3/ADE3)) (14).
Cloning of Human InsP 5 2-Kinase-The gene encoding human InsP 5 2-kinase was isolated by nested PCR amplification using a Marathon spleen cDNA library (CLONTECH) as template. First-round PCR primers were chosen 90 bp upstream of the initiator methionine and 25 bp downstream from the stop codon (upstream primer: 5Ј-AGCTCCGTC-CCCGAGTCCTAGC-3Ј; downstream primer 5Ј-AAAGACACTGCAGG-GAAAGAGTTAGACC-3Ј). This product was then used as a template for second-round PCR using a sense primer encoding a BamHI site followed by the sequence starting from amino acid number 2 (sense, 5Ј-CGCGGATCCGAAGAGGGGAAGTTGGACGAGAATGAATGG-3Ј; antisense, 5Ј-AAGCTTGGGGACCTTGTGGAGAACTAATGTGCAATC-TTCGC-3Ј). PCR was performed with Taq polymerase (Fisher Scientific) using standard protocols. The PCR product was inserted into a TOPO-TA cloning vector (Invitrogen) using the manufacturer's instructions.
Analysis of Human InsP 5

2-Kinase Expression in Yeast Mutant
Strains-The sequence encoding the human InsP 5 2-kinase was inserted into a yeast expression vector by replacing the NcoI/SnaBI fragment of the plasmid pSW747, which contains most of the GLE1 gene (17) with a cDNA fragment amplified from the TOPO-TA cDNA InsP 5 2-kinase clone as template (sense primer, 5Ј-CATGCCATGGGG-AAGAGGGGAAGTTGGACGAG-3Ј; antisense primer, 5Ј-CCGGAATT-CGGGAAAGAGTTAGACCTTGTGGAG-3Ј). The resulting construct places the human InsP 5 2-kinase gene behind the GLE1 promoter and the resulting protein is a fusion to the first eight amino acids of Gle1. Yeast strains were transformed using standard protocols (18) and grown on synthetic minimal media lacking leucine (19). The ipk1 null strains (SWY2105) containing either an empty LEU2/CEN plasmid (pRS315) (20) or the LEU2/CEN plasmid containing human InsP 5 2-kinase were labeled with [ 3 H]inositol (30 Ci/ml for 36 h). Soluble inositol phosphates were isolated from the yeast cells as described (17). Equal amounts of radioactivity from the two strains were loaded onto a Partisphere SAX (4.5 ϫ 126 mm) strong anion exchange HPLC column along with a [ 32 P]InsP 6 standard (see below) and separated at 1 ml/min with a 0 -1.7 M gradient of ammonium phosphate (pH 3.5) over a period of 20 min followed by 30 min at 1.7 M ammonium phosphate. To test complementation of the gle1-2 ipk1-4 synthetic lethality, the yeast strain SWY1793 was transformed with the appropriate LEU2 plasmids. The resulting strains were streaked onto 5-FOA plates and grown for 4 days at 23°C.
Purification of Human InsP 5 2-Kinase from Sf21 Cells-Full-length human InsP 5 2-kinase was subcloned into the pBacPAK9 vector (CLONTECH) using the restriction sites EcoRI and NotI. The resulting FIG. 1. Alignment of the putative InsP 5 2-kinases. Yeast InsP 5 2-kinases (denoted by their yeast gene name, IPK1) and the human InsP 5 2-kinases were assembled using the Clustal program. The human, S. cerevisiae, and C. albicans sequences are full-length, whereas the S. pombe Ipk1 sequence is restricted to the C-terminal kinase region. The N-terminal first 374 amino acid residues of S. pombe Ipk1 comprise a domain that is unrelated to the other Ipk1 proteins and show significant amount of homology to an unrelated protein (see Ives et al. (17)). Black boxes represent identity between at least two amino acids. Boxed residues represent similarity between amino acids. Bars designate Box A and Box B as discussed by Ives et al. (17) and the newly defined Box C and Box D (see "Results"). The asterisk denotes the conserved cysteine residue required for activity of the S. cerevisiae Ipk1.
Kinetic Analysis of InsP 5 2-Kinase Activity-The optimal conditions for assay of FLAG purified human InsP 5 2-kinase were determined to be 50 mM HEPES (pH 7.0), 100 mM KCl, 1 mM DTT, 10 mM MgCl 2 , and 0.1-0.5 g of human InsP 5 2-kinase in 100 l at 37°C for 30 -60 min. Assays were performed using [ 3 H]InsP 5 mixed with unlabeled InsP 5 at a constant specific activity for each assay. The K m for InsP 5 was determined at 0.5 mM ATP while varying the InsP 5 concentration from 0.25 to 4 M; the K m for ATP was determined at 10 M InsP 5 while varying the ATP from 5 to 200 M. Reactions were performed using 0.18 g of enzyme for 1 h and were stopped with an equal volume of 60 mM ammonium phosphate (pH 3.5) and 2 mM InsP 6 to aid in recovery from a Whatman Partisphere SAX strong anion exchange column (4.6 ϫ 125 mm). InsP 5 was separated from InsP 6 at 1 ml/min with a 10-min gradient from 0 to 1.7 M ammonium phosphate (pH 3.5) followed by 30 min at 1.7 M. Kinetic data are the result of three independent experiments. Because the assay regularly converted more than 20% of InsP 5 to InsP 6 its concentration was calculated by taking the log average of the starting and ending InsP 5 concentrations.
Northern Analysis-The full-length cDNA encoding human InsP 5 2-kinase was gel-purified using Qiaex II gel extraction kit and labeled with [␣-32 P]dCTP using a random hexamer labeling kit (Rediprime II, Amersham Biosciences). The probe was hybridized to a multiple tissue Northern blot following the manufacturer's instructions (CLONTECH).

Identification and Isolation of a Gene Encoding the Putative
Human InsP 5 2-Kinase-Based on the Ipk1 protein sequence from S. cerevisiae, genes encoding InsP 5 2-kinases in S. pombe and C. albicans were identified previously (17). Further use of these sequences and data base searching algorithms allowed us to identify partial clones that likely encode two additional fungal InsP 5 2-kinases, one each from Kluyveromyces lactis and Saccharomyces servazzi. The alignment of all the fungal sequences allowed the identification of a number of regions of short though significant homology throughout the sequence (designated Boxes A-D, Figs. 1 and 6) even though the overall level of conservation is limited. Using the BLAST program for short, nearly exact matches (NCBI), we found that amino acids D(L/V)DLK(P/S)X(E/M) of Box D from the fungal Ipk1s matched a predicted gene and a number of human ESTs that all mapped to chromosome nine. A consensus from an alignment of all the ESTs was compared with the fungal Ipk1 sequences using the Clustal method (25). The resulting align-ment is shown in Fig. 1. Although the overall identity was quite low (less than 20%), the human sequence did share significant identity with all of the fungal Ipk1 cDNA sequences. In particular, it was similar to all three of the four boxes of homology defined by the fungal Ipk1 sequences. The human sequence also included the conserved cysteine in Box C, the point mutation of which renders the enzyme inactive in yeast (17). We therefore postulated that this gene might be a candidate human orthologue of the fungal Ipk1. The fulllength cDNA, isolated as described under "Materials and Methods," was used for further studies confirming identification of a human InsP 5 2-kinase.
The Expression of Human InsP 5 2-Kinase Complements Yeast Mutants Deficient in Endogenous Ipk1-To test for whether the gene encoding the potential human InsP 5 2-kinase is functional, the protein was expressed in S. cerevisiae cells. A LEU2/CEN yeast expression plasmid was constructed with the full-length human cDNA under the control of the S. cerevisiae GLE1 promoter. The plasmid expressing the human InsP 5 2-kinase was then transformed into two mutant S. cerevisiae yeast strains: SWY2105, a null of the yeast IPK1 gene (ipk1::KAN r ), and SWY1789, the gle1-2 ipk1-4 double mutant synthetic lethal strain. As controls, strains were independently transformed with an empty LEU2/CEN plasmid. To analyze the production of inositol polyphosphates, the ipk1 null strains were grown in the presence of [ 3 H]inositol, the inositol phosphates were isolated from total cell lysates, and samples were separated by HPLC. As previously reported, the ipk1 null strain produced no InsP 6 and accumulated InsP 5 and an undefined PP-InsP 4 metabolite ((14); Fig. 2A). In contrast, samples from the cells with the putative human InsP 5 2-kinase plasmid resulted in an elution profile with reduced InsP 5 and PP-InsP 4 peaks, as well as a new peak that ran immediately after the PP-InsP 4 peak and co-eluted with the 32 P InsP 6 standard (Fig.  2B). Thus, the potential human InsP 5 2-kinase can produce InsP 6 in a S. cerevisiae strain lacking the yeast Ipk1.
To test directly whether the human InsP 5 2-kinase functionally complements the yeast IPK1 gene, we tested for whether the plasmid expressing the human InsP 5 2-kinase could rescue the growth defect of a gle1-2 ipk1-4 double mutant. The gle1-2 ipk1-4 double mutant is lethal unless a plasmid harboring either wild type GLE1 or IPK1 is present (14). Using the double mutant strains maintained by a GLE1/URA3/CEN plasmid (SWY1793) and transformed with either the human InsP 5 2-kinase, LEU2/CEN, or empty LEU2/CEN plasmid, growth was tested on media containing 5-FOA. 5-FOA is metabolized into a toxic product by the Ura3 protein, and only strains that lack or can lose a plasmid containing the URA3 gene will grow on 5-FOA (26). Thus, growth will happen only if the human InsP 5 2-kinase can rescue the lethal phenotype and allow cells to grow without the GLE1/URA3/CEN plasmid. The strain containing the empty LEU2/CEN plasmid did not form colonies, reflective of the gle1-2 ipk1-4 lethal phenotype (Fig. 3). However, the strain expressing the potential human InsP 5 2-kinase gene did grow on 5-FOA, suggesting that the human protein is functionally conserved. Based on its ability to both produce InsP 6 in yeast and rescue the gle1-2 ipk1-4 synthetic lethality, we conclude that we have identified the human InsP 5 2-kinase.
Expression of Human InsP 5 2-Kinase in Sf21 Cells and Kinetic Studies-Repeated attempts to express the human InsP 5 2-kinase in bacterial expression systems yielded mostly insoluble protein; therefore, a baculovirus expression system was employed. The gene encoding the human InsP 5 2-kinase was inserted into a baculovirus expression vector containing Nterminal polyhistidine and FLAG peptide epitope tags. The expression plasmid was used to transfect Sf21 cells. Three days post-infection, the cells were harvested and lysed, and protein was purified by affinity absorption with M2 anti-FLAG-agarose beads and elution with FLAG peptide. The Michaelis-Menten kinetic parameters were determined by following the conversion of [ 3 H]InsP 5 to InsP 6 at multiple concentrations of InsP 5 and ATP (Fig. 4). These assays directly confirmed the in vitro activity of the protein as an InsP 5 2-kinase. The apparent K m measurements for InsP 5 and ATP were determined to be 0.43 and 21 M, respectively, by double reciprocal plots of 1/V versus 1/S (Fig. 4, insets). The apparent V max was confirmed by both sets of kinetic data to be ϳ31 nmol of InsP 6 formed/min/mg of protein.
Northern Analysis of InsP 5 2-Kinase-The full-length InsP 5 2-kinase gene was labeled and used to probe a human, multitissue Northern blot, showing the ubiquitous expression across all tissues with high levels of transcript in heart, brain, testis, and placenta (Fig. 5).
Alignments of the Sequences of Putative InsP 5 2-Kinases-Using the sequence of human InsP 5 2-kinase, we searched both genomic and EST data bases to compile alignments of our cloned and the putative InsP 5 2-kinases. Our searches have yielded a number of putative InsP 5 2-kinases across many nonfungal species (Fig. 6). All of the putative InsP 5 2-kinases contained most of the boxes of homology defined by the yeast alignments (Fig. 1). Box A is almost perfectly conserved among all InsP 5 2-kinases with the consensus sequence EXKPK. Box C was initially defined by the sequence CRXC found in all the fungal and human sequences. The addition of the nonfungal InsP 5 2-kinases expanded the region of homology of Box C, exposing the sequence (F/Y)CPLDL, which was found almost perfectly represented in all sequences except for the fungi C. albicans, K. lactis, and S. servazzi. Of the other two boxes of homology, Box B seems less well conserved; when the nonfungal homologues are included in the alignment, the residues of the human InsP 5 2-kinase that line up with this box shift (compare Fig. 1 with Fig. 6). Box D, which was used to identify the human InsP 5 2-kinase, is well conserved between human, fungal sequences (except S. pombe), corn, and Anopheles; because the amino acid sequences from C. elegans and Drosophila are both predicted from genomic sequences, it is possible that they are lacking the homology found in Box D because of errors in gene prediction. DISCUSSION Although a role for InsP 6 in mRNA export has been clearly evinced through studies in budding yeast (14), the other proposed cellular functions of InsP 6 require the study of higher eukaryotes. For instance, studies of the role of InsP 6 on L-type Ca 2ϩ channels were performed on insulin-secreting B cells (9). In addition, although the mammalian Ku proteins were shown to be the binding partners for InsP 6 in the DNA-PK complex, the yeast Ku proteins do not bind InsP 6 under the same conditions (11). This suggests that the role of InsP 6 in nonhomologous end joining DNA repair may be specific to mammalian cells. Whereas previous studies of the role of InsP 6 in higher eukaryotes have been conducted by in vitro experiments or by correlating a specific function with an inositol phosphate profile, studies directed at the enzymatic source of InsP 6 have been limited by the lack of a InsP 5 2-kinase gene or protein. Here we have identified the first nonfungal InsP 5 2-kinase.
One strategy that we have used previously to analyze the roles of inositol polyphosphates in mammalian cells has been to deplete cells of inositol metabolites by heterologous expression of an inositol phosphate phosphatase from Salmonella dublin SopB (27). When SopB is overexpressed in a tetracycline-dependent system in human cells, there are rapid perturbations in cellular levels of multiple soluble inositol phosphates. In particular, total cellular InsP 5 and InsP 6 levels are depleted. Coincidentally, polyadenylated RNA accumulates in the nucleus, and protein synthesis is markedly inhibited, thus suggesting a conserved role for the InsP 5 2-kinase of yeast and mammalian cells in mRNA export. However, this study could not distinguish the relative importance of different inositol phosphates in the mRNA export mechanism, because several inositol polyphosphates are perturbed. Isolation of the human InsP 5 2-kinase will now allow direct gene deletion studies in which only InsP 6 production is depleted. This is a goal of our future studies.
In this report, we have confirmed the functional conservation between InsP 5 2-kinase of yeast and mammalian cells. We show that human InsP 5 2-kinase can complement the deficiencies of InsP 6 production in a S. cerevisiae ipk1 null mutant strain and the synthetic lethality of the gle1-2 ipk1-4 mutant strain. Such broad cross-species complementation is unusual, especially when the overall protein sequence homology is so low. This suggests that enzymatic activity and InsP 6 production alone is sufficient for the role of Ipk1 in mRNA export. It remains possible that protein-protein interaction motifs and targeting sequences could also be conserved and remains to be discovered.
It is notable that the apparent V max of human InsP 5 2-kinase is quite low at 30 nmol of InsP 6 formed/min/mg of protein.
However the V max values for the fungal proteins are also quite low (17) as is the value for the mammalian Ins(1,3,4)P 3 5/6kinase (60 nmol of InsP 4 formed/min/mg of protein) (22). The latter is the first enzyme in the pathway of synthesis of InsP 6 in mammalian cells. Whether these low levels of activity result from some as yet unidentified cofactor or post-translational modification that is missing from the in vitro assays remains to be determined. FIG. 6. The alignment of the putative and the cloned InsP 6 2-kinases. Sequences were isolated from BLAST searches of EST and genomic data bases and aligned with the fungal and mammalian sequences using the Clustal method. The regions listed correspond to the boxes of homology as discussed in the text and in the legend for Fig. 1. The sequences are as follows: sp, S. pombe, sc, S. cerevisiae; ca, C. albicans; kl, K. lactis; ss, S. servassi; hs, Homo sapiens; dm, Drosophila melanogaster; ce, Caenorhabditis elegans; ag, Anopheles gambiae; zm, Zea mays. The D. melanogaster and C. elegans sequences are predicted from genomic sequences, and therefore the boxes near the C terminus could not be predicted with confidence. Those residues that are almost perfectly conserved are shaded.
The human InsP 5 2-kinase mRNA is found in most tissues, although to varying degrees. The abundance in the brain is consistent with its proposed function in endocytosis and its possible effect on the synaptic protein synaptojanin (13). The mRNA level matches that of the InsP 3 5/6 kinase (22), both showing robust expression in heart and brain. This is not surprising because they both act in the pathway that produces InsP 6 . The apparent paucity of the InsP 5 2-kinase in some tissues is surprising, considering the range of functions proposed for InsP 6 . Gene deletion experiments will allow us to assess the role of InsP6 in vivo.
The discovery of the putative, nonfungal InsP 5 2-kinases will also aid in our understanding of the function of InsP 6 . The alignment of the fungal, human, and the putative InsP 5 2-kinases from corn, Drosophila, Anopheles, and C. elegans highlights the most conserved residues among the InsP 5 2-kinases; these will provide good targets for mutagenesis to discover which residues are necessary for enzyme activity, its regulation, or possibly its interaction with other proteins. Interestingly, the cysteines in Box C are perfectly conserved among all of the sequences except for corn. ESTs of other plants (potato, rose, wheat) also lack the conserved cysteines, even though Box C itself is maintained. It will be interesting to inquire whether there is a connection between this residue substitution and the peculiarities of the plant enzymes. The discovery of a putative plant InsP 5 2-kinase is also interesting because of the desire for a low phytate seed to mitigate the problems associated with high InsP 6 levels in seeds, i.e. pollution and malnutrition (28). The identification of the human Ins 5 P 2-kinase sets the stage for breakthroughs in analyzing the role of InsP 6 production in animal development and disease.