A Role for Nuclear Inositol 1,4,5-Trisphosphate Kinase in Transcriptional Control

See allHide authors and affiliations

Science  17 Mar 2000:
Vol. 287, Issue 5460, pp. 2026-2029
DOI: 10.1126/science.287.5460.2026


Phospholipase C and two inositol polyphosphate (IP) kinases constitute a signaling pathway that regulates nuclear messenger RNA export through production of inositol hexakisphosphate (IP6). The inositol 1,4,5-trisphosphate kinase of this pathway in Saccharomyces cerevisiae, designated Ipk2, was found to be identical to Arg82, a regulator of the transcriptional complex ArgR-Mcm1. Synthesis of inositol 1,4,5,6-tetrakisphosphate, but not IP6, was required for gene regulation through ArgR-Mcm1. Thus, the phospholipase C pathway produces multiple IP messengers that modulate distinct nuclear processes. The results reveal a direct mechanism by which activation of IP signaling may control gene expression.

How diverse extracellular stimuli elicit selective cellular responses through the activation of IP signaling pathways remains an important question in eukaryotic biology. Central to this process is phosphatidylinositol-specific phospholipase C, activation of which produces second messenger molecules such as inositol 1,4,5-triphosphate (IP3), a regulator of calcium release (1, 2). Modification of IP3 to form higher phosphorylated inositols such as IP4, IP5, and IP6, the functions of which are less well understood, indicates that the regulatory effects of IP messengers may not be limited to calcium signaling (2, 3). Further diversity may be achieved through the spatial localization of IP pathways to cellular organelles (4). For example, a phospholipase C–dependent IP kinase signaling pathway contributes to IP6-mediated mRNA export from the nucleus. In this pathway, phosphatidylinositol 4,5-bisphosphate is hydrolyzed to IP3, which is then sequentially phosphorylated to IP6 by two IP kinases: a putative kinase with dual specificity for IP3 and IP4, whose locus has not been mapped, and an IP5 2-kinase (Ipk1) that is localized to the nuclear envelope–pore complex (5).

Many agonists that activate IP signaling also initiate specific changes in gene expression. Tight control of transcription is facilitated through the assembly of site-specific protein complexes on DNA promoter elements that regulate the activity of the general transcriptional machinery. One such multiprotein assembly is the arginine-responsive ArgR-Mcm1 complex, which comprises four proteins—Arg80, Arg81, Arg82, and Mcm1—each of which is required for proper transcriptional control (6, 7). Arg80 and Arg81 function as arginine-specific transcription factors, whereas Arg82 and Mcm1 are pleiotropic regulators. Mutants ofarg82 have defects in responses to nutrients, sporulation, mating, and stress (8). Mcm1 is a versatile transcription factor required for regulation of diverse gene sets including those involved in cell-cycle control, cell-type specificity, and pheromone and nutrient responses (6, 9). As a canonical member of the MADS (MCM1, agamous, deficiens, serum response factor) box transcription factor family, Mcm1 typically acts in concert with factors that have more limited cellular roles to achieve specificity in transcriptional control (10). Thus, combinatorial interplay between the dedicated arginine-specific transcription factors Arg80 and Arg81 and the more general factors Mcm1 and Arg82 allows interpretation of changes in extracellular arginine levels into transcriptional control of genes that participate in biosynthesis and catabolism of arginine. We find that Arg82 is a dual-specificity IP3-IP4 kinase and thus establish a potential link between phospholipase C–activated IP signaling and transcriptional control.

The phospholipase C–dependent IP signaling pathway apparently uses a single gene product to convert IP3 to IP5. IP3 and IP4 kinase activities copurify from both budding and fission yeast extracts through multiple chromatography steps (11). Also, a mutation of a single locus in Saccharomyces cerevisiae, GSL3, results in the loss of IP4 and IP5 production in vivo (5). To identify the IP3-IP4 kinase, we speculated that an evolutionary relation might exist among IP3 kinases. In metazoans, a family of IP33-kinases has been identified (12). Multisequence alignment of this family was used to identify several conserved elements. These elements were then used to search the yeast database and revealed Arg82 as a candidate IP kinase with 30% identity and 55% similarity throughout two 30–amino acid conserved elements (Fig. 1). Arg82 has a predicted molecular mass similar to that of the purified yeast IP3-IP4kinase activity (11).

Figure 1

A yeast gene product, Arg82, with similarity to conserved elements in an inositol phosphate kinase gene family. Amino acid numbering is provided to the left of the sequences. Identical residues, black; residues in which the nature of the side chain is preserved, gray. The sequences are as follows: ce3K,Caenorhabditis elegans, accession number (acc) AAC38962; h3Ka, Homo sapiens, acc Swissprot P23677; h3Kb, H. sapiens, acc Swissprot P27987; h3Kc, H. sapiens, acc Swissprot D38169; Arg82, S. cerevisiae, acc Swissprot P07250.

To test directly whether ARG82 encodes an intrinsic IP3 kinase, we analyzed purified, recombinant Arg82. Incubation of recombinant protein with [3H]inositol 1,4,5-trisphosphate and adenosine triphosphate (ATP) resulted in the formation of inositol 1,3,4,5,6-pentakisphosphate, indicating that Arg82 performs two phosphorylation steps (Fig. 2A). Recombinant Arg82 in the presence of pure inositol 1,4,5-trisphosphate and trace mass amounts of [γ-32P]ATP generated a single 32P-labeled IP peak, which coeluted with inositol 1,4,5,6-tetrakisphosphate and was resistant to hydrolysis by type I inositol polyphosphate 5-phosphatase (Fig. 2B). With pure d-inositol 1,4,5,6-tetrakisphosphate, a single radiolabeled product peak of inositol 1,3,4,5,6-pentakisphosphate was observed (Fig. 2B). Seven other pure IP isomers were also tested as possible substrates, and under these conditions, kinase activity was not detected (13). Furthermore, the entire pathway from IP3to IP6 could be recapitulated in vitro by incubation of IP3, ATP, and recombinant Ipk2 and Ipk1. Thus, we conclude that Arg82 is a dual-specificity IP3 6-kinase or IP4 3-kinase, which we designate Ipk2 (for inositol polyphosphate kinase).

Figure 2

Arg82, renamed Ipk2, an intrinsic IP3-IP4 kinase. The coding sequence ofIPK2 was fused in frame to glutathioneS-transferase (GST), expressed in bacteria and purified. (A) GST-Ipk2 (10 ng) was incubated in 50 mM Hepes (pH 7.3), 50 mM NaCl, and 10 mM MgCl2 with 10 μM [3H]I(1,4,5)P3 (DuPont Biotechnology Systems, Boston, Massachusetts) and 2 mM adenosine triphosphate (ATP) at 37°C for 30 min, and reaction products were resolved by HPLC. Conversion was time- and dose-dependent with approximate activity of 1 μmol/min per milligram. (B) GST-Ipk2 (10 ng) was incubated as above except with trace amounts of [γ-32P]ATP and either 10 μM pure I(1,4,5)P3 (top) or I(1,4,5,6)P4 (bottom) (Matreya, Pleasant Gap, Pennsylvania). Incubation of GST (10 μg) alone with any IP isomer yielded no detectable radiolabeled IP product. All reaction products were analyzed by Partisphere (Whatman) strong anion exchange (SAX) (4.6 mm by 125 mm) HPLC using a linear gradient from 10 mM to 1.7 M ammonium phosphate (pH 3.5) over 25 min, followed by further elution with 1.7 M ammonium phosphate for 25 min. Individual IP isomers were assigned on the basis of coelution with known IP standards and sensitivity or resistance to various inositol phosphate phosphatase enzymes.

The functional role of Ipk2 in vivo was determined by analysis of ipk2 mutant strains. One copy ofIPK2 (ARG82) was disrupted in diploid yeast with the HIS3 gene by homologous recombination. Haploidipk2 null (Δ) mutants were recovered by sporulation and dissection of the heterozygous IPK2/ipk2Δ strain. All four progeny were viable when grown at 30°C; however, ipk2Δ mutants grew slowly at 30°C and were not viable at 37°C. Soluble inositol phosphates were measured in mutant and wild-type cells by steady-state [3H]inositol labeling and high-performance liquid chromatography (HPLC). The predominant IP isomer in wild-type yeast under these conditions is inositol hexakisphosphate (IP6), the end product of a Plc1–dependent inositol kinase pathway (5). Cells lacking Ipk2 showed a loss of IP4, IP5, and IP6, and had increased amounts of IP3 and an unidentified isomer of IP2 (IP2x) (Fig. 3). This is consistent with a blockade of the major IP3 kinase activity in ipk2Δ cells. Production of IP6 was restored to a state similar to that in wild-type cells by reintroduction of IPK2 into mutant cells (Fig. 3).

Figure 3

Ipk2, the primary I(1,4,5)P3 kinase in yeast. HPLC analysis of soluble extracts prepared after steady-state3H-labeling of ipk2Δ (top) oripk2Δ + pIpK2 (bottom) in which an ipk2Δ strain expressed a fluorescent GFP-Ipk2 fusion protein (26). Cells were incubated in complete minimal medium containing [3H]inositol (20 μCi/ml) and harvested during logarithmic growth as described (27). Soluble fractions from approximately 1 × 107 cells were resolved by HPLC as described above.

To further examine this IP pathway, we analyzed the relation between Ipk2 and Plc1, the sole phosphatidylinositol-specific phospholipase C from yeast (14). In wild-type cells, overproduction of Plc1 resulted in large increases in the levels of IP3, IP4, IP5, and IP6(5). Overexpression of Plc1 in ipk2Δ cells resulted in 20-fold increases of IP3 and IP2x above those observed in ipk2Δ strains; however, IP4, IP5, and IP6 remained essentially undetectable (15). To test whether IP2x and IP3 arise solely from Plc1 activation, we analyzed IP levels in an ipk2Δplc1Δ double mutant. As was the case in plc1Δ cells, IPs were not detected in the double mutant, indicating that Plc1 activity produces the relevant IP3 substrate for Ipk2 in vivo.

Given the kinase selectivity of Ipk2, and the similar IP profiles of the ipk2Δ strain and previously described gsl3mutants, we speculated that the gsl3 mutants had mutations in IPK2. A plasmid harboring IPK2 was found to complement the gsl3-3 gle1-2 synthetic lethal phenotype and the gsl3-3 temperature-sensitive phenotype and resulted in a strain that produced IP4, IP5, and IP6 (16). When the gsl3-3 and the ipk2Δ strains were mated, the resulting diploid was temperature-sensitive, and the alleles did not cosegregate in tetrad dissection analysis (16). These results confirm thatgsl3-3 is an allele of IPK2.

Previous studies of Arg82 (Ipk2) indicate that it regulates cellular processes at the level of gene transcription (8). An in vitro assay for ArgR-Mcm1–mediated control is the ability of cellular extracts to form multiprotein assemblies on site-specific DNA promoter elements (17). Inability to form such complexes has been correlated with a loss of transcriptional control (18). For this reason, we characterized the role of Ipk2-mediated IP production in formation of ArgR-Mcm1 protein–DNA complexes. Extracts prepared from wild-type cells were incubated with a32P-radiolabeled fragment of the ARG5,6 promoter element that can be bound by the ArgR-Mcm1 complex. The binding mixture was resolved on nondenaturing polyacrylamide gels, such that a band of large apparent molecular size was observed in wild-type extracts that corresponded to the multiprotein complex bound to DNA (Fig. 4A). When we used extracts of mutant cells in which IPK2 was deleted, no such complex was detected (17). Reintroduction ofIPK2 in an ipk2Δ strain restored complex formation. Because mutations in IPK2 impair production of IP6, we examined whether failure to form the ArgR-Mcm1 transcriptional complex reflected a defect in IP6-regulated mRNA export. ArgR-Mcm1 complexes did form in extracts from theipk1Δ strain, indicating that defects in cellular IP6 production and mRNA export do not hinder ArgR-Mcm1 complex formation.

Figure 4

Requirement of inositol polyphosphate production for function, but not formation, of the ArgR-Mcm1 transcriptional complex. (A) ArgR-Mcm1 complex formation using DNA mobility shift assay. Cells (5 × 108) of (1) wild-type, (2) ipk2Δ, (3) ipk2Δ + pIPK2, (4) ipk1Δ, (5) plc1Δ, (6) ipk2-3, and (7) ipk2Δ + pIpk2kin strains were grown to early logarithmic growth phase in rich medium or complete minimal medium lacking tryptophan (pIPK2 and pIpk2kinstrains). Crude extracts were obtained by mechanical lysis of yeast cells with glass beads (28). The relevant promoter region of the ARG5,6 gene was amplified by the polymerase chain reaction, gel-purified, and end-labeled with [γ-32P]ATP. Soluble protein (50 μg) from each strain was incubated with 32P-labeled probe; complexes were resolved by electrophoresis and visualized by autoradiography (28). (B) Phenotypic assay of ArgR-Mcm1 transcriptional activity. Freshly grown single colonies of appropriate strains (as labeled) were streaked onto rich medium (left) versus plates containing ornithine as the sole nitrogen source supplemented with limiting amounts (20 μg/ml) of nitrogen compounds for which our strain is auxotrophic (histidine, adenine, tryptophan, uracil, and leucine) (7). Plates were incubated at 30°C for 3 days. (C) Ipk2-kinase activity is required for ArgR-Mcm1 function. Freshly grown single colonies of ipk2Δ + pIPK2 oripk2Δ + pIpk2kin were streaked onto either complete minimal (left) or ornithine medium (both of which lacked tryptophan) and incubated as above.

Phosphatidylinositol-specific phospholipase C (Plc1) activity produces the IP3 substrate for Ipk2, andplc1Δ cells do not produce detectable IPs (5). Therefore, if any products of the Ipk2 reactions were required for complex formation, binding should be abolished in theplc1Δ strain as well. In extracts from aplc1Δ strain, complexes were formed (Fig. 4A). Additionally, in extracts prepared from ipk2Δ strains expressing a kinase-inactive Ipk2 mutant (ipk2kin ), complexes were also able to form (19) (Fig. 4A). Thus, cellular production of IPs through phospholipase C and Ipk2 kinase activity are not necessary to maintain components of the ArgR-Mcm1 transcriptional complex, whereas the cellular presence of the Ipk2 protein is required.

Because the DNA mobility shift experiments do not distinguish between active and inactive transcriptional complexes, we also used a phenotypic assay for transcriptional control. In wild-type yeast strains provided with arginine (or its biosynthetic precursor, ornithine) as the sole nitrogen source, activation of the ArgR-Mcm1 complex results in induction of two arginine catabolic genes and repression of six arginine anabolic genes (7, 20). Therefore, strains that do not form or activate the ArgR-Mcm1 transcriptional complex are inviable on medium with arginine or ornithine as the sole nitrogen source. Wild-type,ipk2Δ, ipk2Δ + pIPK2,ipk1Δ, and plc1Δ strains were plated at 30°C on rich medium or medium containing ornithine as the sole nitrogen source (21) (Fig. 4B). Wild-type, ipk1Δ, andipk2Δ + pIPK2 strains grew normally, butipk2Δ and plc1Δ strains failed to grow. We also examined the ipk2Δ + pIpk2kinstrain and found that, despite their ability to form ArgR-Mcm1 complexes, these cells were unable to grow on ornithine medium (Fig. 4C). We conclude that production of inositol polyphosphates may be dispensable for ArgR-Mcm1 complex assembly, but that phospholipase C-induced, Ipk2-mediated production of IP4-IP5, but not IP6, is required in vivo for the function of the ArgR-Mcm1 transcriptional complex.

Because Plc1 and Ipk2 but not Ipk1 appear to be required for transcriptional activity, the biologically relevant IP molecule may be IP4 or IP5. To further define the isomer or isomers required for regulation of the ArgR-Mcm1 complex, we examined an allele of IPK2, ipk2-3, which produces IP4 but displays a significant reduction in the levels of IP5 and IP6 in cells grown at 30°C. In extracts from ipk2-3 mutants, complex formation with theARG5,6 promoter was similar to that in wild-type (Fig. 4A), and the cells grew normally on ornithine (Fig. 4B). Thus, it appears that compromising the production of IP5 and IP6does not significantly influence ArgR-Mcm1 function.

Several enzymes in IP signaling pathways are found in the nucleus, including Ipk1 and phospholipase C (4, 5, 22). Moreover, an Ipk2-lacZ fusion protein fractionates with the nucleus (23). To visualize the subcellular localization of Ipk2, a green fluorescent protein (GFP)-tagged Ipk2 was expressed in ipk2Δ cells and analyzed by direct fluorescence microscopy. Living cells expressing GFP-Ipk2 exhibited concentrated fluorescence in the nucleus and weak cytoplasmic staining (24). Thus, localized production of IPs in the nucleus may influence transcriptional control.

Our results emphasize that Ipk2 influences transcriptional responses by two mechanisms. First, Ipk2 protein but not IP synthesis is needed to enable formation of ArgR-Mcm1 complexes on DNA promoter elements. Second, production of I(1,4,5,6)P4 and possibly I(1,3,4,5,6)P5 through both phospholipase C and Ipk2 kinase activity is required to properly execute transcriptional control as measured by growth on ornithine as the sole nitrogen source. Together, this pathway produces distinct messengers involved in regulating nuclear processes [see Web fig. 1 (25)]. The coordination of gene expression through Ipk2 and the regulation of mRNA export through Ipk1 may contribute to the mechanisms by which diverse cellular agonists induce selective responses.

  • * To whom correspondence should be addressed. E-mail: yorkj{at}


View Abstract

Navigate This Article