Isolation of inositol 1,3,4-trisphosphate 5/6-kinase, cDNA cloning and expression of the recombinant enzyme.

Inositol 1,3,4-trisphosphate 5/6-kinase was purified 12,900-fold from calf brain using chromatography on heparin-agarose and affinity elution with inositol hexakisphosphate. The final preparation contained proteins of 48 and 36-38 kDa. All of these proteins had the same amino-terminal sequence and were enzymatically active. The smaller species represent proteolysis products with carboxyl-terminal truncation. The Kof the enzyme for inositol 1,3,4-trisphosphate was 80 nM with a V of 60 nmol of product/min/mg of protein. The amino acid sequence of the tryptic peptide HSKLLARPAGGLVGERTCNAXP matched the protein sequence encoded by a human expressed sequence tag clone (GB T09063) at 16 of 22 residues. The expressed sequence tag clone was used to screen a human fetal brain cDNA library to obtain a cDNA clone of 1991 base pairs (bp) that predicts a protein of 46 kDa. The clone encodes the amino-terminal amino acid sequence obtained from the purified calf brain preparation, suggesting that it represents its human homologue. The cDNA was expressed as a fusion protein in Escherichia coli and was found to have inositol 1,3,4-trisphosphate 5/6-kinase activity. Remarkably, both the purified calf brain and recombinant proteins produced both inositol 1,3,4,6-tetrakisphosphate and inositol 1,3,4,5-tetrakisphosphate as products in a ratio of 2.3-5:1. This finding proves that a single kinase phosphorylates inositol in both the D5 and D6 positions. Northern blot analysis identified a transcript of 3.6 kilobases in all tissues with the highest levels in brain. The composite cDNA isolated contains 3054 bp with a poly(A) tail, suggesting that 500-600 bp of 5′ sequence remains to be identified.

The phosphatidylinositol signaling pathway involves a complex scheme in which cells use a series of kinases and phosphatases to interconvert the six known inositol lipids and the more than 20 inositol phosphates that exist in eukaryotic cells (1). These molecules have been implicated in a number of intracellular events including calcium ion mobilization (2), nuclear DNA synthesis (3), trafficking of intracellular vesicles (4), and cell proliferation in response to cytokines and growth factors (1,2). Many of the reactions in this pathway are catalyzed by several different isoenzymes, most notably phospholipase C (5) and inositol polyphosphate 5-phosphatase isoenzymes (6,7), where 8 -10 isoforms of each have been discovered to date. We have now characterized an additional kinase of this pathway that utilizes inositol 1,3,4-trisphosphate as a substrate. Inositol 1,3,4-trisphosphate (Ins(1,3,4)P 3 ) 1 is at a branch point in inositol phosphate metabolism. It is dephosphorylated by specific phosphatases to either inositol 3,4-bisphosphate or inositol 1,3-bisphosphate. Alternatively, it is phosphorylated to inositol 1,3,4,6-tetrakisphosphate [Ins(1,3,4,6)P 4 ] or inositol 1,3,4,5-tetrakisphosphate [Ins(1,3,4,5)P 4 ] by inositol trisphosphate 5/6-kinase (Ins(1,3,4)P 3 5/6-kinase) (8,9). Ins(1,3,4,6)P 4 is the first intermediate in the pathway leading to the higher inositol phosphates including other tetraphosphates, pentaphosphates, inositol hexakisphosphate, and pyrophosphate forms of these (10), all of which are ubiquitously found in tissues. Because the Ins(1,3,4)P 3 5/6-kinase enzyme is at a branch point in metabolism leading to multiple different end products, it is likely to be regulated by its various end products. This enzyme has been partially purified from rat liver (8,9), porcine brain (11), and bovine testes (11) and in each case was reported to phosphorylate Ins(1,3,4)P 3 on either the 5 or 6 position yielding a mixture of two products. This property seems remarkable since the 5 and 6 positions of myoinositol are on opposite faces of the ring. We therefore isolated the enzyme from calf brain and used the amino acid sequence that we determined to identify a human expressed sequence tag (EST) and to clone a cDNA encoding the human Ins(1,3,4)P 3 5/6-kinase.

EXPERIMENTAL PROCEDURES
Materials-Heparin-agarose type I, bacterial ATP grade I, pepstatin A, soybean trypsin inhibitor type IIS, leupeptin, bestatin, diisopropyl fluorophosphate, benzamidine, phenylmethylsulfonyl fluoride, Coomassie Blue R250, isopropyl-1-thio-␤-D-galactopyranoside, calmodulin, calmodulin-agarose, phytic acid (inositol hexaphosphoric acid), CHAPS, dithiothreitol, and ammonium sulfate were obtained from Sigma. Lysyl endopeptidase was purchased from Wako Chemicals (Dallas, TX Inositol 1,4,5-trisphosphate was prepared as described previously (12). Immobilon-P membranes were purchased from Millipore (Bedford, MA). Human multitissue Northern blots were obtained from Clontech (Palo Alto, CA). Sequencing grade trypsin, T4 polynucleotide kinase, Quick Spin columns, calf intestinal alkaline phosphatase, and the random primed labeling kit were from Boehringer Mannheim. A Zap human fetal brain cDNA library (oligo(dT)-and random-primed), Rapid Excision Kit, and E. coli strain XL1 blue were purchased from Stratagene (La Jolla, CA . Reactions were stopped by the addition of 1 ml of water, and the samples were loaded onto a 400-l Dowex-formate column equilibrated in water. Each column was washed with 8 ml of 0.8 M ammonium formate, pH 3.5 to elute the substrate. The product was eluted with 2 ml of 1.6 M ammonium formate and was counted in a liquid scintillation counter. Activity was expressed as a first-order rate constant using the equation [S] ϭ [S] o e Ϫkt , as described previously (13).
The crude extract from 12 brains was filtered sequentially through 80-, 5-, and 1-m filters prior to being applied to a 3.5-liter heparinagarose column (45 ϫ 10 cm) equilibrated in 20 mM Bis-Tris, 1 mM ATP, 1 mM DTT, and 1 mM EGTA, pH 7.2 (heparin buffer A). After washing with two column volumes of buffer A, the column was step eluted with heparin buffer A containing 0.2 M NaCl. The fractions containing Ins(1,3,4)P 3 5/6-kinase activity were pooled and precipitated with 50% ammonium sulfate. The pellet was stored at Ϫ80°C until 36 calf brains were processed in the same manner. Between uses, the heparin-agarose column was washed with two column volumes each of 2 M NaCl and 8 M urea to remove residual protein. The ammonium sulfate pellets were combined, resuspended in homogenization buffer, and dialyzed against heparin buffer A containing 3 mM MgCl 2 (heparin buffer B). The dialyzed samples were loaded onto a 500-ml heparin-agarose column (25 ϫ 5 cm), equilibrated with heparin buffer B, and eluted with 0.3 M NaCl in heparin buffer B. The fractions containing Ins(1,3,4)P 3 5/6-kinase activity were pooled, precipitated with 50% ammonium sulfate, and dialyzed against heparin buffer B.
The sample was next applied to an 85-ml heparin-agarose column (48 ϫ 1.5 cm) equilibrated in heparin buffer B. After washing with two column volumes of buffer B, the column was eluted with a 1.6-liter linear gradient of 0 -1 mM inositol hexakisphosphate (IP 6 ) in heparin buffer B. Two pools of Ins(1,3,4)P 3 5/6-kinase activity were collected from this elution and were dialyzed against 55% ammonium sulfate in heparin buffer B. After centrifugation, the pellets were dialyzed against heparin buffer B and applied separately to a 1-ml Mono Q column. Fractions were eluted using a 20-ml linear gradient of 0 -0.3 M NaCl in heparin buffer B.
Ins(1,3,4)P 3 5/6-Kinase Gel Assays-Partially purified Ins(1,3,4)P 3 5/6-kinase was mixed with SDS gel loading buffer for a final concentration of 2% SDS, 10% glycerol, 62.5 mM Tris, pH 6.8, and 50 mM DTT (14), heated to 40°C for 5 min, and loaded onto two lanes of a 12% SDS gel. One lane containing 5/6-kinase was stained with Coomassie Blue to visualize protein, and the other lane was sliced into 41 pieces, each of which was eluted into a volume of 200 l of assay buffer (described above) containing 0.96 pmol of [ 3 H]Ins(1,3,4)P 3 . Samples were incubated at 37°C for 18 h, and formation of product was determined as described above.

Cloning and Expression of Human Ins(1,3,4)P 3 5/6-Kinase-Human
EST clone GB T09063 (phagemid clone HIBB058) was obtained from American Type Culture Collection. A 200-bp HindIII fragment from this clone was purified using a Qiaex II gel extraction kit and was labeled with [␣-32 P]CTP using a random hexamer labeling kit. Unincorporated label was removed using a Quick Spin column. This probe was used to screen 0.7 ϫ 10 6 plaques from a ZAP human fetal brain cDNA library. The primary screen yielded 21 clones. One of these (clone 15) was found to contain the entire open reading frame of Ins(1,3,4)P 3 5/6-kinase including a possible initiation codon, as determined from the amino-terminal amino acid sequence of the purified calf brain enzyme.
Clone 15 was digested with EcoRI, and the resulting fragment was gel purified and ligated into the Xpress System Protein Expression TrcHis vector as per manufacturer's instructions (Invitrogen). From a 25-ml culture having an A 600 of 0.6 at the time of induction with 1 mM isopropyl-1-thio-␤-D-galactopyranoside, 1 ml was removed at various time points, and the cells were pelleted by centrifugation. Bacterial pellets were lysed by the addition of 50 l of 20 mM MES, pH 6.1, 1 mM EDTA, 1 mM ATP, 10 mM benzamidine, 40 M leupeptin, 1 mM phenylmethylsulfonyl fluoride, 40 M iodoacetamide, 1 M pepstatin A, 10 M bestatin, 1 mM DTT, 17 g of calpain inhibitor I/ml, 7 g of calpain inhibitor II/ml, and 3 mM MgCl 2 , followed by freezing on dry ice and thawing. To each pellet suspension, 5 l of 10 mg of lysozyme/ml in the above buffer was added, and samples were kept on ice until all time points were collected. Assays for 5/6-kinase activity were done as described above.
Northern Blot Analysis-The 200-bp HindIII fragment of the EST clone GB T09063 was labeled with [␣-32 P]CTP as described above. Hybridization was carried out following the manufacturer's instructions (Clontech).
The two pools of enzyme activity were applied separately to a Mono Q column. Elution of pool 2 with a linear gradient of NaCl is shown in Fig. 1D. Ins(1,3,4)P 3 5/6-kinase activity elutes at 0.12 M NaCl with a small peak of coincident protein. The bulk of protein elutes later in fractions 13-15. The specific activity of the peak fraction was 3.7 ϫ 10 5 min Ϫ1 /mg. Application of pool 1 to the Mono Q column yielded an identical profile of protein and Ins(1,3,4)P 3 5/6-kinase activity, but with lower specific activities of the fractions (data not shown). SDS-gel analysis of the Mono Q fractions from pool 2 indicated that the peak fraction contains two discrete sets of protein bands. One protein has an apparent molecular mass of 48 kDa, whereas the other is a triplet of proteins ranging in size from 36 -38 kDa (Fig. 2).
A summary of the purification is shown in Table I. From 36 calf brains, a total of 53 g of protein was obtained from the 48,000 ϫ g supernatant (crude extract). The final material (pool 2) had a specific activity of 3.7 ϫ 10 5 min Ϫ1 /mg protein.
The overall yield for this fraction was 3%, with a 12,900-fold purification.
Gel Renaturation Assay of Partially Purified Ins(1,3,4)P 3 5/6-Kinase-To determine which proteins on the SDS gel represented the Ins(1,3,4)P 3 5/6-kinase, an SDS gel renaturation assay was done using material from an IP 6 affinity elution of a heparin-agarose column, with a specific activity of 12,000 min Ϫ1 /mg protein. The enzyme had some residual activity under these conditions if the sample was heated at no more than 40°C and if the gel slices were resuspended in a volume of assay buffer no smaller than 200 l in order to dilute the SDS. Fig. 3 shows the result of a gel renaturation assay. Fig. 3A shows the Coomassie-stained gel with molecular mass markers and the lane of the sample that was used for the enzyme assay. The slices made in the gel can be seen at the far right of the Coomassie-stained gel piece. In this assay, two peaks of activity were found, corresponding to the 48-and 36 -38-kDa proteins obtained in the final preparation. Both the 48-and the 36 -38-kDa proteins were also shown to have the same amino-terminal amino acid sequence (data not shown).
Amino-terminal amino acid sequence analysis, obtained from sequencing excised bands from polyvinylidene difluoride membranes from early preparations of lower specific activity (350 -550 min Ϫ1 /mg protein), indicated that a predominant 36-kDa band was aldolase type C (17), which was a major contaminant of the preparations (data not shown). Under nonreducing conditions on a series of gel filtration columns, aldolase elutes as a tetramer with an apparent molecular mass of 140 kDa, whereas Ins(1,3,4)P 3 5/6-kinase activity was found only in fractions of lower molecular weight (data not shown). It was recently shown by Baron et al. (18) that aldolase type C binds Ins(1,4,5)P 3 and that this binding is inhibited strongly by Ins(1,3,4)P 3 , which may explain its copurification with Ins(1,3,4)P 3 5/6-kinase.
Cloning of a cDNA Encoding Ins(1,3,4)P 3 5/6-Kinase-Amino acid sequence was obtained from amino-terminal sequencing of the intact calf brain enzyme, from several partially degraded proteins, and from tryptic and lysyl endopeptidase peptides. One of the peptides obtained from trypsin digestion had the sequence HSKLLARPAGGLVGERTCNAXP and was found to match the protein sequence predicted by the DNA sequence of a human EST clone in Genbank at 16 of 22 residues. This clone (GB T09063) was used to isolate additional clones from a human fetal brain library (see "Experimental Procedures").
A schematic diagram of the EST clone GB T09063 and the overlapping human fetal brain clone is shown in Fig. 4. The consensus clone contains 3054 bp with an open reading frame of 1242 bp and 1700 bp of 3Ј-untranslated region. A polyadenylation signal is located 23 bp upstream from a poly(A) tail. There is no in-frame stop codon in the sequence 5Ј of this, thus it is possible that initiation occurs in a 5Ј site not yet obtained. There is a weak Kozak consensus sequence (19) around bp 1 shown in Fig. 5. The open reading frame encodes a protein of 45.6 kDa with a pI of 6.1 (Fig. 5). The peptide sequences obtained from the calf brain preparation presumed to match the corresponding amino acids in the human protein are underlined. They were identical at 45 of 54 residues sequenced (83%).
The amino acid sequence of Ins(1,3,4)P 3 5/6-kinase shares small regions of similarity to the epsilon isoform of protein kinase C (PKC⑀) from both rabbit and human sources, as determined from a BLAST search (20). The three regions are spaced throughout both the Ins(1,3,4)P 3 5/6-kinase sequence and the PKC⑀ sequences (Fig. 6A). In the first conserved region, comprised of 12 amino acids, there is 50% identity and 75% similarity between the two PKC⑀ isoforms and Ins(1,3,4)P 3 5/6-kinase. The second region contains 32% identity and 56% similarity, and the third region contains 72% identity and 81% similarity. In addition, the sequence obtained for Ins(1,3,4)P 3 5/6-kinase has similarity to the predicted amino acid sequences of two other ESTs in the Genbank. Fig. 6B shows a comparison between the predicted amino acid sequence of human Ins(1,3,4)P 3 5/6-kinase and the predicted amino acid sequences available for the EST clones GB Z25963 from Arabidopsis thali-  ana and GB D46351 from rice. The predicted amino acid sequence of the Arabidopsis partial cDNA is 44% identical to and 79% similar to Ins(1,3,4)P 3 5/6-kinase over 54 amino acids and contains a methionine near the putative initation methionine. The predicted amino acid sequence of the rice partial cDNA is 32% identical and 76% similar to Ins(1,3,4)P 3 5/6-kinase over 139 amino acids.
Northern Blot Analysis of Ins(1,3,4)P 3 5/6-Kinase-Northern blot analysis of human tissues disclosed that there is a transcript of 3.6 kilobases that is most abundant in brain, followed by heart Ͼ skeletal muscle Ͼ kidney ϭ pancreas ϭ liver ϭ placenta Ͼ lung (Fig. 7A). We also probed a Northern blot containing mRNA from various regions of the human brain and found expression of Ins(1,3,4)P 3 5/6-kinase in all regions of the brain represented (Fig. 7B). In addition to the 3.6-kilobase transcript, a weakly hybridizing band of 5.3 kilobases was observed in some but not all of the regions of the brain. This may represent an alternatively spliced form or other isoform of Ins(1,3,4)P 3 5/6-kinase.
Our purification of Ins(1,3,4)P 3 5/6-kinase employed chromatography in the presence and the absence of MgCl 2 and affinity elution with IP 6 . The final preparation had a peak specific activity of 3.7 ϫ 10 5 min Ϫ1 /mg protein, with two sets of protein bands of apparent molecular masses of 48 and 36 -38 kDa. Identification of these protein bands as Ins(1,3,4)P 3 5/6-kinase was accomplished by gel renaturation assays, in which bands excised from an SDS gel were shown to phosphorylate The specific activity of the preparation reported here is greater than 200 times higher than that reported for the enzyme purified from rat liver (9). There is evidence of carboxylterminal proteolysis in the calf brain preparation, and the protein purified from rat liver has a molecular mass of 36 kDa, which is consistent with proteolysis in that preparation. A similar situation occurred in the isolation of inositol 1,4,5trisphosphate 3-kinase where early reports of purification found proteolyzed and relatively inactive enzyme compared with the full-length protein (28). A major contaminant of the calf brain Ins(1,3,4)P 3 5/6-kinase at early stages of the purification is aldolase type C. Because aldolase C is an inositol polyphosphate binding protein, it may also have contaminated the rat liver preparation.
Using protein sequence obtained from the purified calf brain protein, we have cloned and expressed the human homolog of Ins(1,3,4)P 3 5/6-kinase. The cDNA obtained encodes a 46-kDa protein with regions of similarity to human and rabbit PKC⑀ isoforms. Only one of the three conserved regions, residues 237-273 of Ins(1,3,4)P 3 5/6-kinase, lies in a defined domain of PKC⑀ (reviewed by Hug and Sarre (29)). This region is in a loop that is very sensitive to proteolysis by calpain. The predicted amino acid sequence is also similar to the predicted amino acid sequence of two plant cDNA EST clones in GenBank. Although limited sequence identities are described here, complete sequencing of these genes may yield further regions of similarity. These proteins may therefore represent the plant homologs of human Ins(1,3,4)P 3 5/6-kinase. There is no sequence similarity to other inositol polyphosphate kinases or phosphatases including the phosphatases that utilize the same substrate as Ins(1,3,4)P 3 5/6-kinase.
The cloning of Ins(1,3,4)P 3 5/6-kinase allowed definitive identification of the products of this enzyme. In enzyme preparations from tissues it was conceivable that two kinases were copurified that phosphorylated either the 5 or 6 position of Ins(1,3,4)P 3 . The demonstration of both 5-and 6-kinase activities toward Ins(1,3,4)P 3 by the recombinant kinase rules out contamination of the protein preparation. This dual product formation from a single substrate makes Ins(1,3,4)P 3 5/6-kinase unique among the inositol polyphosphate kinases and phosphatases. There is a constant ratio of the two tetrakisphosphate products formed by the purified calf brain enzyme and the recombinant human enzyme, with a preference for the 6 position, which might result from the formation and ultimate hydrolysis of a cyclic phosphate intermediate. If the two products result from hydrolysis of a cyclic intermediate, it is possible that the product ratio is different in cells. Ins(1,3,4)P 3 5/6-kinase phosphorylates one of the two possible positions (i.e., IP 5 is not a product). This finding would also be consistent with a cyclic 5/6 phosphate intermediate. Alternatively, the production of dual products may serve some as yet undiscovered cellular function. With Ins(1,3,4)P 3 5/6-kinase available as a recombinant enzyme, the mechanism of its dual phosphorylation and studies of formation of higher phosphorylated inositol polyphosphates may be carried out.