Rat inositol 1,4,5-trisphosphate 3-kinase C is enzymatically specialized for basal cellular inositol trisphosphate phosphorylation and shuttles actively between nucleus and cytoplasm.

The calcium-liberating second messenger inositol 1,4,5-trisphosphate (Ins(1,4,5)P3) is converted to inositol 1,3,4,5-tetrakisphosphate (Ins(1,3,4,5)P4) by Ins(1,4,5)P3 3-kinases (IP3Ks) that add a fourth phosphate group to the 3-position of the inositol ring. Two isoforms of IP3Ks (named A and B) from different vertebrate species have been well studied. Recently the cloning and examination of a human full-length cDNA encoding a novel isoform, termed human IP3K-C (HsIP3K-C), has been reported. In the present study we report the cloning of a full-length cDNA encoding a rat homologue of HsIP3K-C with a unique mRNA expression pattern, which differs remarkably from the tissue distribution of HsIP3K-C. Of the rat tissues examined, rat IP3K-C (RnIP3K-C) is mainly present in heart, brain, and testis and shows the strongest expression in an epidermal tissue, namely tongue epithelium. RnIP3K-C has a calculated molecular mass of approximately 74.5 kDa and shows an overall identity of approximately 75% with HsIP3K-C. A bacterially expressed, enzymatically active and Ca2+-calmodulin-regulated fragment of this isoform displays remarkable enzymatic properties like a very low Km for Ins(1,4,5)P3 ( approximately 0.2 microm), substrate inhibition by high concentrations of Ins(1,4,5)P3, allosteric product activation by Ins(1,3,4,5)P4 in absence of Ca2+-calmodulin (Ka(app) 0.52 microm), and the ability to efficiently phosphorylate a second InsP3 substrate, inositol 2,4,5-trisphosphate, to inositol 2,4,5,6-tetrakisphosphate in the presence of Ins(1,3,4,5)P4. Furthermore, the RnIP3K-C fused with a fluorescent protein tag is actively transported into and out of the nucleus when transiently expressed in mammalian cells. A leucine-rich nuclear export signal and an uncharacterized nuclear import activity are localized in the N-terminal domain of the protein and determine its nucleocytoplasmic shuttling. These findings point to a particular role of RnIP3K-C in nuclear inositol trisphosphate phosphorylation and cellular growth.

For the conversion of the calcium-liberating second messenger inositol 1,4,5-trisphosphate (Ins(1,4,5)P 3 ) 1 (1), all metazoan cells examined so far seem to possess two types of enzyme activities: first, the 5-phosphatases (2), which degrade it to inositol 1,4-bisphosphate and thereby initiate the recycling of the inositol moiety into membrane phospholipids; and second, the 3-kinases (IP3Ks), 2 which use ATP to add a fourth phosphate group to the 3-position of the inositol ring to generate the signaling molecule inositol 1,3,4,5-tetrakisphosphate (Ins (1,3,4,5)P 4 ). This InsP 4 isomer has been implicated in signaling functions concerning the regulation of a Ras GTPase-activating protein (3), isosteric inhibition of Ins(1,4,5)P 3 5-phosphatases (30) and thus prolongation of Ins(1,4,5)P 3 signals, as well as calcium entry through the plasma membrane (4), and other potential functions (5,6). Moreover, Ins(1,3,4,5)P 4 is the first product derived from Ins(1,4,5)P 3 that can be converted to a plethora of inositol phosphate isomers found in metazoan cells (7), ranging from different InsP 4 isomers through InsP 5 isomers and InsP 6 to the pyrophosphate-containing InsP 7 and InsP 8 isomers. For IP3K in invertebrates three splice variants of one single IP3K-gene have been characterized in Caenorhabditis elegans (8). In vertebrates on the cDNA level, two isoforms from different genes have so far been characterized from different species (see Ref. 9 and citations within). Recently the cDNA of a third human isoform, termed HsIP3K-C, with a calculated molecular mass of 75.2 kDa and a broad range of tissues expressing this isoform has been published (10). The stimulation of IP3K activity by Ca 2ϩ -CaM might be an important factor in the control of calcium oscillations in mammalian cells. However, the degree of stimulation by Ca 2ϩ -CaM found for HsIP3K-C was much lower than that observed for the A and B isoforms (see Ref. 10 and citations therein). Although the A isoform is stimulated by Ca 2ϩ -CaM by a factor of ϳ2-3, in particular the B isoform is strongly activated by Ca 2ϩ -CaM by a factor of ϳ10. For the brain-and testis-specific isoform A, an F-actin targeting function of its N-terminal domain (24), a domain obviously not necessary for the enzymatic function (9), has been reported. Dewaste et al. (10) reported that the HsIP3K-C is a predominantly cytosolic enzyme when transiently expressed in COS-7 cells, but did not demonstrate a specific targeting function of its N-terminal domain. Thus, the genuine function of IP3K-C and the regulatory or targeting function of its N-terminal domain remained unclear up to now.
In this study we highlighted possible special roles of IP3K-C in the cellular and in particular in nuclear inositol phosphate metabolism. A rat homologue of HsIP3K-C, termed rat Ins(1,4,5)P 3 3-kinase C (RnIP3K-C) was identified. Its mRNA expression pattern in different tissues, the enzymatic properties and the allosteric regulation of bacterially expressed enzyme, and the nucleocytoplasmic shuttling of EGFP fusion proteins derived from RnIP3K-C expressed in mammalian cells were examined. The most interesting findings of our study are that the RnIP3K-C enzyme is strongly Ca 2ϩ -CaM-activated in a substrate concentration-dependent manner, is enzymatically optimized for an efficient conversion of basal cellular concentrations of Ins(1,4,5)P 3 in presence of Ins(1,3,4,5)P 4 , and can convert Ins(2,4,5)P 3 to Ins(2,4,5,6)P 4 under these conditions. Furthermore, RnIP3K-C undergoes active nucleocytoplasmic shuttling as a result of a hitherto unidentified nuclear localization activity and an identified nuclear export signal (NES) both residing in the N-terminal domain of the enzyme. Active transport of proteins between the nucleus and cytoplasm is mediated mainly by the intensively studied canonical nuclear localization signals (NLS) and by the recently characterized nuclear export signal (both reviewed in Refs. 11 and 12). Leptomycin B (LMB), an antifungal antibiotic and an inhibitor of the cell cycle in mammalian and fission yeast cells (13), binds to CRM1 (for chromosome region maintenance), an NES receptor in the nuclear pore complex (14), and thereby inhibits NES-dependent nuclear export (15). By employing LMB inhibition and deleting the canonical NES in the N-terminal domain of RnIP3K-C, we could prove the functionality of this site and the nuclear shuttling activities of the enzyme when expressed in cells.

EXPERIMENTAL PROCEDURES
Cloning of Rat Inositol 1,4,5-Trisphosphate 3-Kinase Isoform C (RnIP3K-C) cDNA-Degenerate primers (forward, 5Ј-CAA(A/G)CC-(A/G/T/C)CG(A/G/T/C)TA(C/T)ATGCA(A/G)TGG-3Ј; reverse, 5Ј-AGCC-(G/A)TCCTC(C/A/G)CG(G/A)TT(G/C)CCCTC-3Ј) were designed according to the completely conserved amino acid sequences KPRYMQW and EGNREDG, respectively, in the catalytic domain of rat and human IP3K-A and IP3K-B, and IP3K from chicken (see Ref. 9 and citations within). A PCR using 3 l of a ZAP Express cDNA library prepared from rat circumvallate papilla (16) as template in a total volume of 50 l was carried out as follows: 94°C for 3 min, 40 cycles of 94°C for 1 min, 60°C for 1 min, 72°C for 1 min, and finally 72°C for 10 min. PCR products isolated from the single band of ϳ400 bp were subcloned into pUC18 using the SureClone ligation kit (Amersham Biosciences). The nucleotide sequences were determined by the dideoxy termination method with an ABI Prism 373A sequence analyzer (Applied Biosystems, Foster City, CA). 22 phage pools of the circumvallate cDNA library, each consisting of ϳ5 ϫ 10 4 independent plaque-forming units were screened using the PCR with the primer pair and under conditions as described above. 1 ϫ 10 5 plaque-forming units of each of the three pools that were positive in the PCR pre-selection were plated onto Escherichia coli XL-1 MRFЈ cells; the DNA was transferred to nitrocellulose filters and hybridized conventionally with the 32 P-labeled subcloned PCR fragment. The hybridizing 11 plaques were purified by three rounds of plating and hybridization and finally converted into recombinant pBK-CMV plasmids by in vivo excision according to the protocol from the supplier (Stratagene, La Jolla, CA). After diagnostic restriction fragment analysis, the clone exhibiting the longest insert of ϳ3.3 kb was sequenced on both strands using a set of gene-specific primers.
5Ј-RACE-mRNA, purified by oligo(dT) columns (Amersham Biosciences) from total RNA prepared from rat brain (strain Wistar) as described (17), was used for 5Ј-RACE employing the Marathon TM kit (Clontech) according to the instructions from the manufacturer. For reverse transcription, carried out for 1 h at 42°C, a gene-specific primer (5Ј-GGGCCTCGGCTGGCAATG-3Ј) was used in a reaction containing 3.5 g of mRNA. With the generated double-stranded cDNA, a first round of PCR was performed using a gene-specific primer (5Ј-CGGCA-CGGGCAGCGCCTCATAC-3Ј) and primer AP1 (Marathon TM kit) with an annealing temperature of 56°C. On the diluted PCR product, nested PCR was performed using a gene-specific primer (5Ј-CTTCCCCTTTA-AATCCTACAAG-3Ј) and primer AP2 (Marathon TM kit) and an annealing temperature of 56°C. A second round of RACE PCR was done on the original cDNA template with a gene-specific primer (5Ј-GCTCCA-GGCCGCTCGGATATGTAG-3Ј) and primer AP1 using an annealing temperature of 56°C. From this reaction a nested PCR was performed with a gene-specific primer (5Ј-TATGTAGGGGTTAACCGATACTC-3Ј) and primer AP2 at 59°C annealing temperature.
Northern Blot Analysis-A Northern blot containing ϳ2 g of poly(A) ϩ RNA/lane from eight different rat tissues was purchased and hybridized according to the instructions from the supplier (Clontech). The 3.3-kb DNA fragment representing the full-length RnIP3K-C cDNA labeled with [ 32 P]dCTP by random priming was used for hybridization in ExpressHyb solution at 68°C for 90 min, followed by washing two times in 0.1ϫ SSC, 0.1% SDS for 40 min at 50°C.
To compare specifically RnIP3K-C mRNA in tongue and taste epithelium with other tissues, 15 g of glyoxylated total RNA/lane was separated by agarose gel electrophoresis, transferred onto a Hybond N membrane (Amersham Biosciences) and hybridized with the radioactive RnIP3K-C probe described above. Hybridization in ULTRAHyb solution at 42°C overnight and washing in 0.2ϫ SSC, 0.1% SDS at 45°C were carried out according to the instructions from the manufacturer (Ambion, Austin, TX). As control for RNA loading and integrity, the blot was stripped and hybridized with a 32 P-labeled actin probe under the same conditions. Blots were exposed to a phosphorimaging screen for up to 24 h and analyzed in a Fujix Bio-Imaging Analyzer (BAS 2000).
Reverse Transcription-PCR-1 g of total RNA from various rat tissues including enzymatically prepared tongue and taste epithelium was reverse transcribed using Superscript II reverse transcriptase (Invitrogen) and random hexamer primers (Amersham Biosciences). A pair of RnIP3K-C-specific primers (forward, 5Ј-TTCACAGACCTGACC-TCC-3Ј; reverse, 5Ј-GAAGCCATCGGTGTCATGTG-3Ј) was designed to amplify a 271-bp fragment from 1/20 of the reverse transcribed total RNA using the Taq PCR Master Mix (Qiagen). PCR was performed in 25 cycles (94°C, 1 min for the initial denaturation, 94°C, 30 s; 55°C, 30 s; 72°C, 30 s for each cycle and 72°C, 10 min for the final elongation). Following the same protocol, a hypoxanthine-guanine-phosphoribosyltransferase fragment of 526 bp was amplified as control. PCR fragments were visualized with ethidium bromide after separation of equal volumes of each reaction on a 1.8% agarose gel.
Bacterial Expression of a RnIP3K-C Fragment Comprising the CaM Binding and the Catalytic Domain (CBD-RnIP3K-C) and a Fragment Comprising Only the Catalytic Domain (⌬CBD-RnIP3K-C)-A fragment of the protein coding region of RnIP3K-C (termed CBD-RnIP3K-C) comprising CaM binding and catalytic domain (amino acids 366 -678) was amplified with the following primer pair (5Ј-GAGGAT-GACCATATGGCTGGGGGCGGAGGTACCAG-3Ј; 5Ј-TCCTCACTAGT-GAGCCTAGCAGTGGCAGCTG-3Ј) in a PCR (25 cycles) using 2.5 units of Pfu polymerase (Promega) in a 100-l standard reaction mixture according to the manufacturer (annealing temperature of 65°C and 4-min elongation at 72°C/cycle). The PCR product was first cloned into the pGEM T-Easy vector (Promega). The cloned PCR product was cleaved out with NdeI and SpeI and re-ligated into pET17b vector (Novagen) previously cut with the same restriction enzymes. The recombinant fragment of RnIP3K-C was expressed in E. coli BL21(DE3)pRIL cells (Novagen) transformed with the resulting expression vector and purified by phosphocellulose and CaM affinity chromatography essentially as described previously for chicken IP3K (9).
A fragment of RnIP3K-C (termed ⌬CBD-RnIP3K-C) comprising only the catalytic domain (amino acids 410 -678) was amplified with the following primer pair (5Ј-AACAGCTATGACCATGATTACGCC-3Ј; 5Ј-A-CGGCGCCTCCGGACATGCTGGGAACTTCC-3Ј) in a PCR as described above. The PCR product cloned into the pGEM T-Easy vector was cleaved out with EheI and NsiI and re-ligated into pET17b vector (Novagen). The fragment of RnIP3K-C was expressed in E. coli BL21(DE3)pLysS,pREP cells (Novagen) transformed with the resulting expression vector and purified by phosphocellulose chromatography essentially as described previously for chicken IP3K (9). No addition of lysozyme to the cells was necessary, because of the lysozyme expression by the bacteria. The destruction of DNA by DNase treatment was replaced by a sonification step.
Enzymatic Analysis of CBD-RnIP3K-C and ⌬CBD-RnIP3K-C-Enzymatic activities were measured under various substrate, product, activator, and inhibitor concentrations using a coupled enzymatic optical assay essentially as described previously (9). A Lambda 20 UVvisible spectrometer (PerkinElmer Life Sciences) equipped with thermostatted cuvette holder and numerical data storage and derivation device (program UV Winlab, PerkinElmer Life Sciences) was employed. The standard reaction conditions used in the assays were: 10 mM triethanolamine, pH 7.5, 5 mM MgCl 2 , 30 mM KCl, 1 mM dithiothreitol, 0.2 mM NADH, 1 mM phospho(enol) pyruvate, 0.5 mM ATP, 5 units/ml lactate dehydrogenase, 2.5 units/ml PK in the presence of 6 mM ammonium sulfate. All assays were performed at 30°C after a 15-min preincubation of the reaction mixture. After following background NADH consumption in presence of PK, lactate dehydrogenase, and IP3K for 5-10 min, the IP3K assay was started with inositol phosphate substrate. All assays were performed in absence and presence of 0.1 M Ca 2ϩ -CaM. In the latter case, 20 M Ca 2ϩ was present. The dependence of enzymatic activity on the concentration of Ins(1,4,5)P 3 (bought from Alexis, Woburn, MA) was assayed by single transients. They were started at differing initial InsP 3 concentrations (varying between 1 and 35 M,) and followed in the optical assay until InsP 3 was completely consumed. The K m for ATP was assayed by varying the initial ATP concentration at a fixed initial Ins(1,4,5)P 3 concentration of 5 M and measuring initial activities. Activation by Ca 2ϩ -CaM was also assayed by measuring initial activities at 5 M initial Ins(1,4,5)P 3 ; the concentration of Ca 2ϩ was 20 M, and that of CaM was varied. Activation or inhibition by Ins(1,3,4,5)P 4 was assayed by determining the initial enzyme activity at 1 M Ins(1,4,5)P 3 and increasing concentrations of Ins(1,3,4,5)P 4 present before starting the optical assays. In case of the enzyme form ⌬CBD-RnIP3K-C, the true K m for Ins(1,4,5)P 3 of this enzyme form was determined from the K m(app) values of a series of single transients started at different initial Ins(1,3,4,5)P 4 concentrations by plotting these parameters against Ins(1,3,4,5)P 4 concentration. The y intercept of the resulting linear regression curve corresponds to K m , the x intercept to ϪK i for Ins(1,3,4,5)P 4 . The reaction products generated by conversion of Ins(2,4,5)P 3 (pure synthetic Ins(2,4,5)P 3 was a gift from Barry V. L. Potter) and from assays of the activation of IP3K-C by Ins(1,3,4,5)P 4 were analyzed by metal dye detection (MDD)-HPLC (18 -20). For that, aliquots from the optical assay mixture were trichloroacetic acid-precipitated and charcoal-treated to remove ATP and NAD(H) as described (18 -20).

Construction of Fusion Genes and Fusion
Gene Derivates-The EGFP fusion genes N-tag full and C-tag full were created by PCR techniques. The open reading frame of RnIP3K-C was amplified using primer pairs (N-tag full : 5Ј-GTCGACATGAGGCGCTGCCCGTGCCG-3Ј and 5Ј-GGA-TCCTAGCTCTGGGCCAGGCCCTGAA-3Ј; C-tag full : 5Ј-AGCGCTATG-AGGCGCTGCCCGTGCCG-3Ј and 5Ј-CTCGAGGCTCTGGGCCAGGC-CCTGAA-3Ј). The PCR products were initially cloned into the pGEM T-Easy vector (Promega, Mannheim, Germany). The open reading frame was completely sequenced, and then the fragment was subcloned by the introduced restriction sites (SalI/BamHI for N-tag full , Eco47III/XhoI for C-tag full ) into the expression vectors pEGFP-C1 and pEGFP-N1 (Clontech), respectively.
Cell Culture, Transient Gene Expression, and Fluorescence Microscopy-The cell lines were purchased from Cell Lines Service (Heidelberg, Germany). NRK 52E cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal calf serum (Invitrogen). PC12 cells were grown in RPMI medium with Glutamax I supplemented with 10% horse serum and 5% fetal calf serum (Invitrogen). Both cell lines were cultured at 37°C in a humidified atmosphere in the presence of 5% CO 2 .
Cell transfections were performed using the Metafectene method according to the instructions from the manufacturer (Biontex Laboratories, Munich, Germany). Fixation, staining with 4Ј,6-diamidino-2phenylindole dihydrochloride, and examination of the intracellular localization by fluorescence microscopy were carried out as described (22). A minimum of 100 undamaged cells/experiment were examined. LMB was purchased from Sigma and added to the complete medium to a final concentration of 11 ng/ml (20 nM). Statistics and Enzyme Kinetic Data Evaluation-Unpaired t test was performed using GraphPad InStat version 3.05 (GraphPad Software, San Diego, CA). A value of p Ͻ 0.05 was considered statistically significant.
Evaluation of enzyme kinetic parameters was performed from primary UV Winlab data sets as follows. In each assay data set (NADH consumption versus time), a correction for the initially determined background ATPase activity (coming from the indicator enzymes employed), a smoothing of corrected NADH versus time data, and a final derivation of specific activity versus substrate concentration data (v versus S data) was performed by exporting the data sets into Excel and employing special macro routines. For all enzymatic parameter and goodness of fit estimates performed with the final v versus S data, non-linear least squares techniques were employed. For that, the standard non-linear fitting and initial parameter definition techniques implied in the enzyme kinetics program GraphPad PRISM version 2.1 were used. Special kinetic models employed were formulated as described by Segel (47), and corresponding novel functions were defined in the non-linear function fitting module.

RESULTS
Identification and Cloning of RnIP3K-C-To detect novel isoforms of Ins(1,4,5)P 3 3-kinases (IP3Ks) from rat in addition to the known A and B forms, degenerate oligonucleotide primers were derived from highly conserved protein regions of the catalytic domains of all IP3K isoforms found so far in vertebrates (see Ref. 9 and citations within) and used in a polymerase chain reaction. The phage cDNA library used as template in the PCR was prepared from rat tongue epithelium containing predominantly keratinocytes and taste bud cells from the circumvallate papilla. After subcloning of the PCR products in the 400-bp band, restriction enzyme mapping indicated the presence of two types of clones (data not shown). Although the restriction pattern of one type was identical with the pattern expected for the rat isoform B, the other could not be assigned to a known isoform. Sequence analysis of several of those clones showed 62 and 65% identity on the nucleotide level to isoform A and B, respectively (data not shown). When the novel sequence was used to screen a circumvallate papilla cDNA library under stringent conditions, a 3309-bp cDNA clone containing a single long open reading frame and a poly(A) tail of 18 bp was isolated. The translated protein consists of 678 amino acids with a predicted molecular mass of 74,463 Da. Because the first methionine residue was not preceded by an in-frame translational stop codon, 5Ј-RACE experiments were performed to extend the sequence further to the 5Ј end of the mRNA. In the 246 bp of sequence added by the analysis of the 5Ј-RACE products, two in-frame stop codons and no further in-frame initiation codon were identified. The presence of the 5Ј-sequence was confirmed independently by analysis of RT-PCR fragments that span the junction between the RACE products and the original cDNA clone (data not shown). Therefore, the novel sequence has 353 bp of 5Ј-untranslated region and the 5Ј-most ATG found in the cDNA clone most probably represents the initiation codon, although the neighboring residues do not agree well with the Kozak consensus sequence (from Ϫ9 to ϩ4: [GCC]GCC(A/G)CCATGG) (23). The 3Ј-untranslated region of 1146 bp exhibits the polyadenylation signal AATAAA 18 nucleotides upstream of the poly(A) tail. The deduced protein sequence of RnIP3K-C shows significant identity with other IP3Ks and essentially the same domain structure (Fig. 1). The highest overall identity of ϳ75% is found with the HsIP3K-C (10). Although the amino acid identities in the more conserved catalytic, calmodulin binding, and PEST do-mains are ϳ90% and higher, the less conserved N-terminal domain shows only ϳ50% identity.
Tissue Distribution of RnIP3K-C-Northern blot analysis showed that RnIP3K-C is expressed in a tissue-specific fashion at relatively low abundance indicated by the fact that only poly(A) ϩ RNA blots produced reasonable strong signals after overnight exposure. A single transcript of 3.4 kb was present predominantly in heart, brain, and testis; at lower levels in lung, liver, and kidney; and almost undetectably in spleen and skeletal muscle ( Fig. 2A). The size of the transcript is consistent with the length of the cloned cDNA. The expression pattern of RnIP3K-C mRNA in rat tissues differs remarkably from that found in human organs (10). In humans, skeletal muscle showed highest expression, whereas brain and kidney contained almost no specific mRNA. Specific rat tissues such as tongue and taste epithelium exhibited amounts of RnIP3K-C mRNA that were considerably higher than those found in, e.g., rat brain (Fig. 2B). In these epithelial tissues, the 3.4-kb transcript could be detected on Northern blots even when using total RNA. The relatively high abundance of RnIP3K-C mRNA in taste and tongue epithelium compared with other tissues was confirmed by semiquantitative RT-PCR with a set of cDNAs prepared separately (Fig. 2C). The RT-PCR data correlate exactly with the results of both Northern blot experiments.
Enzymatic Activities and Calmodulin Regulation of Bacterially Expressed RnIP3K-C Fragments-Measurements of the enzymatic activity of a recombinant RnIP3K-C fragment comprising the catalytic and calmodulin binding domains (CBD-RnIP3K-C) revealed an unexpected behavior of this enzyme regarding its dependence on the substrate concentration. The enzymatic activity of the fragment is already very high at low Ins(1,4,5)P 3 concentrations (Ͻ100 nM), displays a maximum at ϳ0.3 M, and decreases with higher substrate concentrations. This substrate inhibition effect is particularly evident in the absence of the activator Ca 2ϩ -CaM, whereas its presence leads to a higher relative activity at high substrate concentrations, i.e. a lower relative substrate inhibition. This has the consequence that the degree of Ca 2ϩ -CaM activation increases from ϳ4-fold at 0.3 M Ins(1,4,5)P 3 to ϳ8-fold at 5 M Ins(1,4,5)P 3 (Fig. 3, A and B). These data sets can be fitted almost equally well with a simplified model for general non-competitive substrate inhibition or a model for an ordered Bi-Bi reaction (for parameters and explanations, see subscripts of Fig. 3 (A and B), and Ref. 47). Only in the case without Ca 2ϩ -CaM does the ordered Bi-Bi reaction model seem to fit slightly better to the data points (Fig. 3A). As compared with the known K m values of IP3Ks ranging from 0.  Fig. 3(A and B, respectively). These maximum values are lower than the "true" V max values derived for the reaction models (see parameters to Fig. 3 (A and B)) as a result of the substrate inhibition phenomenon. A K D value for the binding of and activation by Ca 2ϩ -CaM of 9.4 nM was determined at 5 M Ins(1,4,5)P 3 and 0.5 mM ATP by the Ca 2ϩ -CaM-induced enzyme activation data (Fig. 3C). The K m value for ATP, determined at 5 M Ins(1,4,5)P 3 , was 33 M in the absence and 52 M in the presence of 0.1 M Ca 2ϩ -CaM (data not shown).
Unusual is the effect that the product Ins(1,3,4,5)P 4 exerted on the enzyme in absence of Ca 2ϩ -CaM. Although both IP3K-A and IP3K-B show a marked competitive inhibition by Ins(1,3,4,5)P 4 with respect of the substrate Ins(1,4,5)P 3 both in absence and in presence of Ca 2ϩ -CaM (our own data for IP3K-B (not shown); Ref. 29 for IP3K-A), the rat C-isoform displays a distinct activation by Ins(1,3,4,5)P 4 only in the absence of Ca 2ϩ -CaM. The degree of activation by Ins(1,3,4,5)P 4 determined at 1 M Ins(1,4,5)P 3 is more than 2-fold, and an activation plateau is reached at ϳ3 M Ins(1,3,4,5)P 4 . An apparent K D for the binding of and activation by Ins(1,3,4,5)P 4 of 0.52 M was derived from the data by assuming simple hyperbolic binding kinetics (Fig. 4A). MDD-HPLC analysis showed that no InsP 5 product was formed in presence of Ins(1,3,4,5)P 4 (data not shown); therefore, the product activation observed is a true allosteric activation of the conversion of Ins(1,4,5)P 3 . This allosteric effect is apparently abolished by Ca 2ϩ -CaM binding (Fig. 4B), which per se activates the enzyme.
To test the hypothesis that the product activation that is not observed in presence of Ca 2ϩ -CaM is really a regulatory property residing in the CaM binding domain of IP3K-C and is caused by binding of Ins(1,3,4,5)P 4 to this domain in the absence of Ca 2ϩ -CaM, we constructed a bacterial expression vector termed ⌬CBD-RnIP3K-C, lacking the cDNA sequence coding for the CaM binding domain. This enzyme showed a more than 10-fold increased maximal specific activity in absence of  (Fig. 4B). Apparently the presence of the CaM binding domain is not only directly responsible for the observed product activation but also for the strong substrate inhibition of the Ca 2ϩ -CaM free enzyme and an increased substrate affinity (see above). The obvious isosteric interaction between substrate and product at the substrate/product binding site of the catalytic domain devoid of the CaM binding domain is characterized by markedly lower apparent binding affinities for InsP 4  Unexpectedly, this isoform is also able to phosphorylate a second biological InsP 3 isomer, namely Ins(2,4,5)P 3 , with relatively high efficiency. In assays where we used Ins(2,4,5)P 3 instead of Ins(1,4,5)P 3 as a substrate, we were able to detect a low but consistent enzyme activity of ϳ50 milliunits/mg (Fig.  5A). The maximal activity was twice as high in the presence of Ca 2ϩ -CaM as in the absence, but also the K m value increased from 1.5 to ϳ5 M under these conditions (Fig. 5B). Although these maximal activities are ϳ50 and 100-fold lower, respectively, than the corresponding V max(app) values for Ins(1,4,5)P 3 (see above), they are in the range of activities reported for the conversion of Ins(1,3,4)P 3 to Ins(1,3,4,6)P 4 and Ins(1,3,4,5)P 4 by purified Ins(1,3,4)P 3 5/6-kinase (46). To identify the product of this reaction, MDD-HPLC analysis of the reaction mixture after 0, 1, and 2 h of incubation with the enzyme was performed. The comparison of the inositol phosphates present in these reaction mixtures against a mixture of standard inositol phosphate isomers revealed a predominating reaction product co-eluting with Ins(2,4,5,6)P 4 (Fig. 6). Because both the substrate and the product appear to be Ͼ95% pure, this clearly indicates the phosphorylation of Ins(2,4,5)P 3 exclusively to Ins(2,4,5,6)P 4 . This Ins(2,4,5)P 3 6-kinase thus is an authentic side activity of RnIP3K-C. Small peaks of Ins(1,3,4,5)P 4 and Ins(3,4,5,6)P 4 also detected by MDD-HPLC analysis after incubation of Ins(2,4,5)P 3 with RnIP3K-C can be explained (i) by a small contamination of Ins(2,4,5)P 3 with Ins(1,4,5)P 3 (being converted to Ins(1,3,4,5)P 4 ), as well as (ii) by a further contamination of Ins(2,4,5)P 3 with a small amount of Ins(3,4,5)P 3 , the latter one apparently also phosphorylated, like Ins(2,4,5)P 3 , at the 6-hydroxy group to Ins(3,4,5,6)P 4 .
To find out whether the ability of IP3K-C to convert Ins(2,4,5)P 3 to Ins(2,4,5,6)P 4 is an exclusive property of this isoform of IP3K, we expressed catalytic forms of rat IP3K-B and avian IP3K-A, both also containing the catalytic and the adjacent CaM binding domain but no N-terminal domain (data not shown). In both cases we found the same type of specific phosphorylation of this substrate at the hydroxyl group 6. In IP3K-A, which is only expressed in neurons, testes, and avian red blood cells (9,27), a high V max of 382 milliunits/mg and a  (lower panel of B). C, the relative strong expression of RnIP3K-C in tongue and taste epithelium compared with other tissues as suggested by the Northern experiment in B was confirmed by semiquantitative RT-PCR. In accordance with both Northern blots, the signals for brain and kidney are weaker whereas again skeletal muscle contains the lowest amount of RnIP3K-C mRNA. Hypoxanthine-guanine-phosphoribosyltransferase amplification served as control for RT-PCR and gel analysis (lower panel of C). Details of the experimental procedures are described under "Experimental Procedures." uitously expressed IP3K isoforms, IP3K-B and IP3K-C, the latter exhibits ϳ4.5-fold higher specific activity toward Ins(2,4,5)P 3 in absence and in presence of Ca 2ϩ -CaM, whereas the K m values for Ins(2,4,5)P 3 are not significantly different between these two isoforms. We also tested whether the initial presence of activator Ins(1,3,4,5)P 4 together with Ins(2,4,5)P 3 or its addition during the reaction could further increase the specific activity of IP3K-C for this alternative substrate but no significant effect, either activation or inhibition, was observed (data not shown). However, in the other two isoforms, A and B, Ins(1,3,4,5)P 4 exhibited a strong competitive inhibition of conversion of Ins(2,4,5)P 3 . IP3K-C thus is the isoform better suited to perform this reaction in tissues and cell types expressing both isoform B and C because of its 5-fold higher V max against this substrate and the absence of competitive inhibition by Ins(1,3,4,5)P 4 . We determined in different tissues and cell types by direct mass analysis of inositol phosphates using MDD-HPLC whether Ins(2,4,5,6)P 4 is resent and is increased after phospholipase C stimulation and could confirm both phenomena (data not shown).
We performed several optical assays to confirm the reversibility and the degree of Ca 2ϩ -CaM activation of the enzyme. One of these experiments is shown in Fig. 7. The reactions were started at high Ins(1,4,5)P 3 in the presence of 20 M free Ca 2ϩ , but in absence of CaM. After 1-2 min, when the enzyme exhibited basal substrate inhibited activity (see above), 0.1 M CaM was added and the reaction followed. Thus, both the degree of CaM activation (7-fold in the example shown in Fig. 7) and the product activation and decrease of substrate inhibition could be directly demonstrated (see the activity derivations plotted together with the optically assayed decrease of NADH, which was converted into an equimolar decrease of Ins(1,4,5)P 3 in Fig. 7). The addition of 125 M EGTA apparently led to the dissociation of CaM and the observed re-decrease of activity. Finally, by addition of ϳ100 M Ca 2ϩ a second (incomplete) CaM activation was induced and activity was followed until the substrate was completely consumed. After the second Ca 2ϩ addition, the degree of Ca 2ϩ -CaM activation was incomplete as a result of only partial replacement of the complexed Ca 2ϩ .
RnIP3K-C Tagged with EGFP Shuttles Actively between the Nucleus and Cytoplasm-The intracellular localization of IP3K-C has been investigated by Dewaste et al. (10). In transfected COS-7 cells the enzyme seems to be localized predominantly in the cytoplasm, but no cellular targeting mechanisms are known so far. To further investigate the intracellular localization of RnIP3K-C, the coding region of its full-length cDNA was fused with an N-terminal EGFP tag (N-tag full ) and a Cterminal EGFP tag (C-tag full ), respectively. These fusion genes were transiently expressed in NRK 52E cells, and the localization of the fusion proteins was determined by inspection using fluorescence microscopy. (Typical images of the different types of localization are shown in Fig. 8A.) Both arrangements of protein sequences in the fusion protein were used to examine steric effects of the fused fluorescent protein influencing the localization of the fusion protein. In most of the transfected cells, the fusion proteins were detected evenly distributed between nucleus and cytoplasm or exclusively in the cytoplasm. In a small minority of cells, they were localized predominantly in the nucleus (Fig. 8B; N-tag full , C-tag full in Table I). This localization pattern was not an artifact of EGFP tagging, because the arrangement of protein sequences had no obvious impact on the localization pattern (N-tag full versus C-tag full in Table I). In control experiments where we expressed EGFP alone, all transfected NRK 52E cells showed the same localization, namely an even distribution between nucleus and cytoplasm (data not shown), ruling out a targeting activity of EGFP itself. Small proteins can enter the nucleus of a cell by passive diffusion, but proteins larger than ϳ40 kDa require an NLS for active translocation through the nuclear pore complex (11). The molecular weight of the fusion protein (101.4 kDa) clearly exceeds this limit. This obviously active translocation of IP3K-C into the nucleus seems to be in contradiction to the exclusively cytoplasmic localization described by Dewaste et al. (10). Therefore, we examined the possibility of an additional nuclear export activity by incubation of NRK 52E cells expressing the fusion protein N-tag full with the export inhibitor LMB. After 6 h of LMB treatment, no cells with an exclusively cytoplasmic localization of the fusion protein were observed ( Fig. 8B; Ntag full /no LMB versus N-tag full /ϩLMB in Table I). The increase of cells showing an even distribution and the decrease of cells showing an exclusively cytoplasmic localization of the fusion protein are extremely significant (p Ͻ 0.001). Furthermore, the proportion of cells with a predominantly nuclear localization seems to be increased (0.05Ͻ p Ͻ 0.1). Therefore, the fusion protein is obviously transported out of the nucleus by an active, LMB-sensitive mechanism. Similar results were also obtained by additional transfection experiments using PC12 cells (data not shown). In summing up, the RnIP3K-C seems to be a nucleocytoplasmic shuttling protein with both nuclear import and nuclear export activity.
The N-terminal Domain Determines the Nucleocytoplasmic Shuttling of RnIP3K-C-Our results (see above) indicate that RnIP3K-C is a nucleocytoplasmic shuttling protein, but the positions of potential sites possessing nuclear import and export activity, respectively, are unknown. The N-terminal domains of the three IP3K isoforms are highly diverse and mainly uncharacterized in their function(s), whereas the other parts of the enzymes are conserved and well studied (Fig. 1). Therefore, these domains are preferred candidates to determine the intracellular targeting of the different isoforms. Indeed, the IP3K isoform A is localized to F-actin and dendritic spines by its N terminus (24). To narrow down sites exhibiting nuclear import or export activity, we fused the N-terminal domain of RnIP3K-C (aa 1-379) with an N-terminal EGFP tag (Ntag Nterm ) and a C-terminal EGFP tag (C-tag Nterm ), respectively. The localization pattern of both EGFP-tagged fragments is comparable with that of the full-length fusion proteins (Ntag Nterm versus N-tag full ; C-tag Nterm versus C-tag full in Table I), as determined by fluorescence microscopy of transiently transfected NRK 52E cells. An effect of EGFP tagging can be ruled out, because the pattern is obviously not influenced by the arrangement of protein sequences in the fusion protein (Ntag Nterm versus C-tag Nterm in Table I). The nuclear entry of the EGFP-tagged fragments still requires an active import, be- cause their molecular size (67.2 kDa) rules out a passive diffusion into the nucleus. Thus, the N-terminal domain of RnIP3K-C tagged with EGFP seems to possess a nuclear import activity comparable with that of the full-length fusion protein, although no classical NLS (25) was revealed by consensus sequence search (data not shown). To further examine potential targeting mechanisms, NRK 52E cells expressing an EGFP-tagged fragment (N-tag Nterm ) were incubated with LMB.
After an LMB incubation of 6 h, no cells with an exclusively cytoplasmic localization were observed (N-tag Nterm /no LMB versus N-tag Nterm /ϩLMB in Table I). Both the decrease of the cell number with an exclusively cytoplasmic localization and the increase of cells with an even distribution of the fragment are extremely significant (p Ͻ 0.001), whereas the proportion of cells showing a predominantly nuclear localization seems to be increased (0.05 Ͻ p Ͻ 0.10). Therefore, the EGFP-tagged N- Recombinant enzyme (CBD-RnIP3K-C) was incubated with Ins(2,4,5)P 3 under the standard optical assay conditions, and products were analyzed after 0, 1, and 2 h of incubation by MDD-HPLC as described under "Experimental Procedures." The conversion of Ins(2,4,5)P 3 was nearly complete after 2 h of incubation. A standard mixture of inositol phosphates generated by limited acid hydrolysis of InsP 6 was separated with the same gradient (upper chromatogram) and used as an isomeric standard mixture. The same type of analysis was also performed with the reaction product(s) of Ins(1,4,5)P 3 , and the only isomer formed was Ins(1,3,4,5)P 4 (chromatograms not shown).
terminal domain of RnIP3K-C demonstrates an LMB-sensitive nuclear export activity comparable with that of the full length fusion protein. To further narrow down potential sites possessing nuclear export activity, we analyzed the sequence of the N-terminal domain of RnIP3K-C. One candidate sequence (aa 318 -326) precisely fits the NES consensus ( Fig. 9) (11,12,26) and is completely conserved between human and rat protein (data not shown). This putative NES was deleted in an EGFPtagged full-length fusion gene to further investigate its role in the intracellular targeting of IP3K-C. The NES deletion mutant was transiently expressed in NRK 52E cells, and its localization was examined by fluorescence microscopy. Now almost all cells showed an even distribution of the fusion protein between nucleus and cytoplasm comparable with LMB-treated cells expressing EGFP-tagged full-length proteins (Ctag full,⌬NES in Fig. 8B and Table I). The deletion of an internal sequence is a drastic operation, which can lead to misfolding and thus changes in the three-dimensional structure of a pro- The almost instantaneous increase of activity was followed by a slow further increase of activity. After another 7.5 min (second arrow), EGTA was added to a final concentration of 125 M, leading to a rapid decrease of activity back to somewhat more of the initial activity. After another 3 min (third arrow), CaCl 2 was added to increase the total Ca 2ϩ to 120 M, and now the reaction was followed until InsP 3 was completely converted, i.e. no more ATP and thus NADH consumption occurred. An apparently incomplete Ca 2ϩ -CaM activation was induced because of [Ca 2ϩ ] Ͻ [EGTA], but in this phase the activity also slowly increased before substrate consumption led to a rapid fall in activity.
FIG. 8. Transiently expressed EGFP/ RnIP3K-C fusion protein shuttles between nucleus and cytoplasm of NRK 52E cells. NRK 52E cells were transiently transfected, fixed, 4Ј,6-diamidino-2-phenylindole dihydrochloride-stained, and examined by fluorescence microscopy as described under "Experimental Procedures." A, typical images of the different types of nuclear and cytoplasmic localization of EGFP-tagged RnIP3K-C are shown (N, nuclear; NC, nuclear/cytoplasmic; C, cytoplasmic). B, LMB treatment (6 h) induces the nuclear accumulation of transiently expressed EGFP-tagged RnIP3K-C (N-tag full ). A comparable effect is also caused by deletion of a potential NES motif in the N-terminal domain (Ctag full,⌬NES ). Graphs show the results (mean Ϯ S.D.) of three independent experiments in which the proportion of cells showing nuclear (N), nuclear/cytoplasmic (N/C), or cytoplasmic (C) localization of the fusion protein was scored by inspection. More than 100 transfected cells were examined per experiment.
tein. These changes can mask an NES without its direct deletion. A less drastic strategy to eliminate a putative NES is the mutation of conserved hydrophobic amino acids (Fig. 9) to alanine residues. Almost all cells expressing such an NES point mutant (L 318 CPV mutated to ACPA) showed an even distribution of the fusion protein between nucleus and cytoplasm comparable with the deletion mutant (C-tag full,⌬NES versus C-tagfull,mNES in Fig. 8B and Table I). Thus, the identified motif is necessary for the observed active exclusion of the EGFP-tagged RnIP3K-C from the nucleus. Additional experiments with PC12 cells showed comparable results (data not shown). In summary, both an NES and an uncharacterized nuclear import activity are localized in the N-terminal domain of RnIP3K-C and seem to determine the nucleocytoplasmic shuttling of the enzyme. (1,4,5)P 3 Phosphorylation in Unstimulated Cells-Comparison of the RnIP3K-C amino acid sequence to the other known vertebrate IP3K sequences yields a higher degree of identity to its human orthologue than to the other two known rat isoforms (our results). A result that is also true if A and B isoform sequences are compared within and between species (data not shown). Therefore, it seems plausible that the three isoforms diverged early in vertebrate evolution to fulfill different tasks within the inositol phosphate metabolism. Since then, considerable selective pressure to conserve these specific functions of each isoform must have kept the sequence of the individual isoforms relatively unchanged during successive speciation events. One possibility for the formation of isoforms is their adaptation to cell-or tissue-specific expression. This type of specialization has obviously been adopted by the A isoform, which shows a unique expression in neurons and in testis (27). On the other hand, the available evidence for the B (28) and, with some restrictions, C isoform (10) instead suggests a relatively broad distribution of both isoforms across tissues and cell types, which directs toward other kinds of functional specialization of these two isoforms. Still, the hitherto known tissue distribution of the C isoform indicates variation between species. Our results obtained with rat tissues show expression of the C isoform in heart, brain, lung, liver, kidney, and testis and most abundantly in an epidermal tissue, namely tongue and taste bud epithelium. Almost no RnIP3K-C mRNA was found in skeletal muscle. In contrast, Dewaste et al. (10) detected the human C isoform mRNA in skeletal muscle but not in brain and kidney. Assuming that A, B, and C isoforms have at least overlapping expression patterns, a specialization of these isoforms could pertain either to their intracellular localization or to their catalytic properties. In these respects our study has highlighted considerable differences between A and B isoforms on the one hand and the C isoform on the other hand. Although A and B isoforms are markedly inhibited by the product of the IP3K reaction Ins(1,3,4,5)P 4 (our own results for IP3K-B (data not shown); see Ref. 29 for IP3K-A), this enzyme product is in fact an allosteric activator of the C isoform in the absence of Ca 2ϩ -CaM (our results). Because the whole N-terminal domain upstream of the CaM binding domain was missing in the enzyme form employed for the kinetic analyses, the allosteric site where Ins(1,3,4,5)P 4 binds can only reside in the catalytic domain or in the CaM binding domain. The strong degree of sequence identity between all three isoforms in their catalytic domain does not make it likely that such an allosteric site is residing in a segment of the catalytic domain. Rather, the fact that this product activation by Ins(1,3,4,5)P 4 is completely abolished by addition of Ca 2ϩ -CaM (our results) or by deletion of the CaM binding domain (our results) directs toward a localization of this allosteric site in this domain. An inspection of the CaM binding domains of all three isoforms reveals that the C isoform contains an additional basic residue at position Lys 391 , which might be responsible for a specific interaction of the uncomplexed CaM binding domain with both the active site and the two inositol phosphates Ins(1,4,5)P 3 and Ins(1,3,4,5)P 4 in a way inhibiting or activating substrate conversion, respectively. Because the allosteric activation by Ins(1,3,4,5)P 4 is half-maximal at a concentration that this product easily reaches intracellularly (0.2-20 M depending on cell type and stimulation of corresponding cells; our own data not shown), this activation is likely to represent a physiological "feed-back" activation mechanism being active only under resting Ca 2ϩ concentrations, i.e. in cells exhibiting a slow release of InsP 3 insufficient to release Ca 2ϩ . The low K m /high affinity for Ins(1,4,5)P 3 would further enhance this basal Ins(1,4,5)P 3 conversion. On the other hand, the unusual phenomenon of substrate inhibition that is seen with the Ca 2ϩ -CaM free enzyme at higher Ins(1,4,5)P 3 concentrations in vitro and is also likely to be mediated by the CaM binding domain (see "Results") does not imply that the enzyme is switched off physiologically at a higher InsP 3 level. Namely, this normally induces a cellular  to Christina Deschermeier for help with cloning of expression vectors and to Casimir Bamberger for discussion and help with the Northern blot.