|Home | About | Journals | Submit | Contact Us | Français|
Phosphatidylinositol phosphates are involved in signal transduction, cytoskeletal organization, and membrane trafficking. Inositol polyphosphates, produced from phosphatidylinositol phosphates by the phospholipase C-dependent pathway, regulate chromatin remodeling. We used genome-wide expression analysis to further investigate the roles of Plc1p (phosphoinositide-specific phospholipase C in Saccharomyces cerevisiae) and inositol polyphosphates in transcriptional regulation. Plc1p contributes to the regulation of approximately 2% of yeast genes in cells grown in rich medium. Most of these genes are induced by nutrient limitation and other environmental stresses and are derepressed in plc1Δ cells. Surprisingly, genes regulated by Plc1p do not correlate with gene sets regulated by Swi/Snf or RSC chromatin remodeling complexes but show correlation with genes controlled by Msn2p. Our results suggest that the increased expression of stress-responsive genes in plc1Δ cells is mediated by decreased cyclic AMP synthesis and protein kinase A (PKA)-mediated phosphorylation of Msn2p and increased binding of Msn2p to stress-responsive promoters. Accordingly, plc1Δ cells display other phenotypes characteristic of cells with decreased PKA activity. Our results are consistent with a model in which Plc1p acts together with the membrane receptor Gpr1p and associated Gα protein Gpa2p in a pathway separate from Ras1p/Ras2p and converging on PKA.
The hydrolysis of phosphatidylinositol-4,5-bisphosphate (PIP2) by phospholipase C (PLC) yields two prominent eukaryotic second messengers: 1,2-diacylglycerol and inositol 1,4,5-trisphosphate (IP3) (16, 41, 50). In higher eukaryotes, the hydrophilic IP3 triggers the release of calcium from internal stores and thus modulates Ca2+/calmodulin-regulated pathways (5), while the hydrophobic 1,2-diacylglycerol activates the phospholipid- and Ca2+-dependent protein kinase C (PKC) (59).
In yeast cells, phospholipase C (Plc1p encoded by PLC1) and four inositol polyphosphate kinases (Ipk2p/Arg82p, Ipk1p, Kcs1p, and Vip1p) constitute a nuclear signaling pathway (see Fig. Fig.1A)1A) that affects transcriptional control (60), export of mRNA from the nucleus (78), homologous DNA recombination (49), cell death, and telomere length (65, 79). Ipk2p/Arg82p is a bifunctional inositol polyphosphate kinase that converts IP3 to IP4 and IP5 (60). Ipk2p/Arg82p is a component of the transcriptional complex ArgR-Mcm1 and regulates transcription of arginine-responsive genes (18). Inositol hexakisphosphate (IP6), produced from IP5 by Ipk1p, regulates mRNA export from the nucleus (78). Kcs1p and Vip1p produce inositol pyrophosphates that are involved in homologous DNA recombination (49), inhibition of Pho80p-Pho85p complex by Pho81p (42), and regulation of telomere length (65, 79).
Recently, inositol polyphosphates (InsPs) have been shown to regulate the activity of chromatin remodeling complexes in vivo and in vitro (68, 71). The induction of the phosphate-responsive PHO5 gene, chromatin remodeling of its promoter, as well as recruitment of Swi/Snf and Ino80 chromatin remodeling complexes are impaired in the ipk2/arg82 mutant strain (71). In vitro, nucleosome mobilization by the yeast Swi/Snf complex is stimulated by IP4 and IP5, while IP6 inhibits nucleosome mobilization by yeast Isw2 and Ino80 complexes and the Drosophila NURF complex (68). Since recombinant Nurf and Isw1 proteins can bind IP6, the possible mechanism by which InsPs affect chromatin remodeling may involve effects on protein conformation of the chromatin remodeling complexes (68). Alternatively, IP4 or IP5 might affect the interaction between chromatin remodeling complexes and chromatin, as has been shown for PIP2 and the Swi/Snf complex (81).
To identify genes transcriptionally regulated by InsPs on a genome-wide level, we performed DNA array experiments with plc1Δ, ipk2Δ, and ipk1Δ strains. In plc1Δ cells grown in rich medium, the expression of about 2% of genes is affected. Most of these genes are repressed by Plc1p. Surprisingly, genes regulated by Plc1p do not correlate with sets of genes regulated by Swi/Snf or other chromatin remodeling complexes but show correlation with genes induced by diauxic shift and controlled by Msn2p/Msn4p.
Msn2p, together with its partially redundant homologue Msn4p, is a zinc finger transcription factor that plays an important role in transcriptional response to starvation and other forms of environmental stress (19, 20, 53). Transcriptional activation of Msn2p-dependent genes is very complex. Msn2p is regulated by nuclear translocation (4, 10, 24, 25) or by increased binding of Msn2p to the STRE elements in the promoters of stress-responsive genes (33). The cyclic AMP (cAMP)-dependent protein kinase A (PKA) negatively regulates transcription of Msn2p-dependent genes by phosphorylating Msn2p and keeping it in the cytosol (24, 25, 70). A low level of PKA activity is associated with increased nuclear localization of Msn2p and transcriptional activation of Msn2p-dependent genes, while Msn2p is translocated to the cytoplasm upon PKA activation. In this context, it is important to note that Plc1p was found to mediate the interaction between Gpr1p and Gpa2p that function upstream of adenylate cyclase and are thus involved in regulation of PKA activity (2). This finding, together with the fact that plc1Δ and ras2Δ mutations genetically interact, implicated Plc1p in the regulation of the PKA pathway (2).
Subcellular localization of Msn2p is also regulated by the TOR (target of rapamycin) pathway. TOR promotes interaction of Msn2p with the 14-3-3 protein Bmh2p that acts as a cytosolic anchor of Msn2p (4). Rapamycin treatment or growth conditions that result in TOR inactivation result in translocation of Msn2p to the nucleus. In addition, inactivation of SRB10 compromises nuclear exclusion of Msn2p in unstressed cells (10). GSK-3 (glycogen synthase kinase 3) protein kinase does not affect nucleocytoplasmic distribution of Msn2p but regulates binding of Msn2p to the STRE element (33). Changes in chromatin structure also do not affect nucleocytoplasmic distribution but may facilitate recruitment of Msn2p to target promoters (46). It appears that Msn2p is also regulated by other mechanisms that are not so clearly defined (17, 21, 36, 40, 43, 54).
Here we report that Plc1p is involved in PKA signaling. In plc1Δ cells, the increased expression of stress-responsive genes is mediated by decreased PKA-mediated phosphorylation of Msn2p and increased binding of Msn2p to stress-responsive promoters. In addition, plc1Δ cells display other phenotypes characteristic of cells with decreased PKA activity. On the basis of our results as well as previously published work (2, 48, 75), we propose a model in which Plc1p acts together with the membrane receptor Gpr1p and associated Gα protein Gpa2p in a pathway separate from Ras1p/Ras2p and converging on PKA.
All S. cerevisiae strains used in this study are listed in Table Table1.1. Except for S. cerevisiae strains CEP A stre, YC17, and TS141, the strains are isogenic to W303. Standard genetic techniques were used to manipulate yeast strains and to introduce mutations from non-W303 strains into the W303 background (69). Cells were grown in rich medium (yeast extract-peptone-dextrose [YPD] [1% yeast extract, 2% Bacto peptone, 2% glucose]) or under selection in synthetic complete (SC) medium containing 2% glucose and, when appropriate, lacking specific nutrients in order to select for a plasmid or strain with a particular genotype. Meiosis was induced in diploid cells by incubation in 1% potassium acetate. In order to eliminate the possibility that the cells used for transcriptional studies (microarray analysis, S1 nuclease analysis, and real-time reverse transcription-PCR [RT-PCR] analysis) were in exponential phase when harvested, we determined the growth curves and kinetics of glucose consumption for wild-type (WT), plc1Δ, ipk2Δ, and ipk1Δ strains in batch cultures in YPD medium. Cells were harvested when the cultures reached an absorbance at 600 nm (A600) of 1.0; the diauxic shift started to occur for the WT and ipk1Δ strains at an A600 of 3.5 to 3.8, while for plc1Δ and ipk2Δ strains, it started to occur at an A600 of 3.1 to 3.6.
Oligonucleotides complementary to the genes assayed by S1 nuclease analysis are as follows: HSP12, 5′-CCAGCGACCTTGTCGGCCTTGTCAGTGATGTATTCCTTACCTTGTTCAGCGTATGACAAAAC T-3′; HSP26, 5′-GCCAGTAGAATCCTTTGCGGGTGTGTTTGCTAACTGACGTCTTGGTGCGTAGCCAGAATT-3′; HIS4, 5′-GGAGAACTGGAGAATCTCTTCATTACTCAGGCTCGAGCCATCCAAAAGTACCTGACCTT GTTC-3′; LEU2, 5′-CGGCATCAGCCTTCTTGGAGGCTTCCAGCGCCTCATCTGGAAGTGGAACACCTGTAGCTAGCTA-3′; RPL7B, 5′-CGAAGACCAACTTGTGTTGAGCTTCGACGTAGTAGGAACCAGCAGCCTT GGCATCACGCAATCGA-3′; RPS22B, 5′-CCTCATGGTCCATAATACCCGCAGAAGTAGTCAAAATAACGTAACCGAATTGTCTGGCTGGCTTA TTC-3′; RPL28, 5′-CCAGAAATGAGCTTGTTGCTTGTGGAAGTATCTCATACCAACCTTACCGAAATAACCTGGATGGATAAAT-3′; RPL3, 5′-GCCTTGTAACCCAAGAAGGAAGTTAGAGCAACTGGCTTGGATCTGT CATCCTTTGGAAAAGCCAATTGA-3′; and ACT1, 5′-GCTTCAGTCAAAAGAACAGGGTGTTCTTCTGGGGCAACTCTCAATTCGTTGTAGAA GGATACTA-3′.
Total RNA was isolated from cultures grown in YPD medium to an optical density or A600 of 1.0 by the hot phenol method as described previously (35). S1 probes were end labeled in a 25-μl reaction mixture (5 pmol oligonucleotide, 125 μCi [γ-32P]ATP [6,000 Ci/mmol, PerkinElmer], 1× T4 polynucleotide kinase buffer, and 20 units T4 polynucleotide kinase [New England Biolabs]) at 37°C for 1 h. The reaction mixture was diluted with 25 μl of water, T4 polynucleotide kinase was inactivated at 65°C for 20 min, and the labeled oligonucleotides were purified using MicroSpin G-25 columns (Amersham Biosciences). The labeled oligonucleotides (0.5 pmol) were hybridized with 20 to 40 μg of total RNA in 50-μl reaction mixtures (0.3 M NaCl, 1 mM EDTA, 40 mM HEPES [pH 7.0], and 0.1% Triton X-100) for 12 h at 55°C and treated with S1 nuclease (Life Sciences) as described previously (35). The samples were analyzed on 20% denaturing polyacrylamide gels, and quantification was performed using a phosphorimager (PerkinElmer).
Total RNA was isolated from WT (W303-1a), plc1Δ (HL1-1), ipk2Δ (YND1060), and ipk1Δ (YND1018) cultures grown in YPD medium to an optical density or A600 of ~1.0 by the hot phenol method and purified with an RNeasy mini kit (Qiagen). Each strain was grown as three independent cultures, and cRNA prepared from each culture was independently analyzed. Total RNA (5 μg) was used to prepare biotin-labeled cRNA with MessageAmp aRNA kit (Ambion). This method amplifies only mRNA. The cRNA was fragmented according to the manufacturer's instructions (1) and hybridized to the Affymetrix yeast genome S98 arrays at 45°C overnight. Hybridization, washing, and streptavidin staining were performed in the Affymetrix GeneChip fluidics station 400. Gene chips were scanned in a Hewlett-Packard G2500 gene array scanner, and expression data for mRNAs were analyzed with the Affymetrix software Microarray Suite 5.0. A comparison analysis for the mutant and WT strains was performed for each experiment. Upregulated genes with a change in P value of less than or equal to 0.001 and downregulated genes with a change in P value of greater than or equal to 0.999 were considered. A change in expression level was considered significant only when a particular gene was increased twofold or more or decreased twofold or more in all hybridizations. Results reported in the lists of upregulated and downregulated genes represent averages of the three values (see supplemental material). To determine the degree of overlap between plc1Δ, ipk2Δ, ipk1Δ, and published data sets, comparisons of twofold-increased and twofold-decreased genes were performed by using Microsoft Excel.
In vivo chromatin cross-linking and immunoprecipitation were performed essentially as described previously (23) with several minor modifications. Briefly, yeast cells were grown in 600 ml YPD medium to an A600 of 1.0, at which point they were fixed for 15 min by the addition of formaldehyde to a concentration of 1%. Subsequently, the cells were converted to spheroplasts with zymolyase. Spheroplasts were washed in 40 ml of ice-cold Tris-buffered saline (25 mM Tris-HCl [pH 7.4], 137 mM NaCl) and 1 ml of FA lysis buffer (50 mM HEPES [pH 7.5], 150 mM NaCl, 1 mM EDTA, 0.1% deoxycholate, 1 mM phenylmethylsulfonyl fluoride) containing protease inhibitors (Complete protease inhibitors; Roche) for each aliquot of original 50-ml culture. Finally, the spheroplasts were resuspended in 200 μl of FA lysis buffer, 300 μl of glass beads was added, and then the samples were vortexed 20 times in 15-second bursts at the highest setting. The samples were pooled, and the suspension was then sonicated 10 times for 10 seconds each time to fragment chromosomal DNA to an average size of ~500 bp. The suspension was centrifuged 30 min at 12,000 × g, and the supernatant was diluted with FA buffer to provide 1-ml aliquots of the resultant solubilized chromatin solution per immunoprecipitation and 100 μl for total input DNA. Each aliquot was precleared by adding 50 μl of 50% protein A/G-agarose slurry (Santa Cruz Biotechnology) and incubating 30 min at 4°C with gentle rocking. Beads were then harvested by centrifugation, and the supernatant was incubated with 100 μl of 25% protein A/G slurry, which had been previously incubated for 8 h with 6 μg of antibody (anti-myc polyclonal antibody A-14 or anti-hemagglutinin [anti-HA] monoclonal antibody F7 [both from Santa Cruz Biotechnology] or anti-RNA polymerase II monoclonal antibody 8WG16 from Covance). Beads were then harvested and washed, and the DNA was released and extracted as described previously (23). Total input DNA and coimmunoprecipitated DNA was then analyzed by real-time PCR (25-μl reaction mixture) using the iQ SYBR green supermix and the Bio-Rad MyIQ single-color real-time PCR detection system (Bio-Rad). Each PCR mixture was used to detect the presence of a protein at a particular locus. The MyIQ software generates a threshold count for each reaction mixture, which can be used to determine the enrichment of a protein at a given locus. Each immunoprecipitation was performed at least three times using different chromatin samples, and the occupancy was calculated using the POL1 coding sequence as a negative control and corrected for the efficiency of the primers. The sequences of the oligonucleotides used for real-time PCR analysis are available upon request.
Denatured proteins were separated on 12% denaturing polyacrylamide gels, and Western blotting with anti-c-myc monoclonal antibody (clone 9E10; Roche), anti-HA monoclonal antibody (clone F7; Santa Cruz Biotechnology), or anti-phospho-CREB (Ser133) polyclonal antibody (Cell Signaling) was carried out as described previously (8, 38). To confirm equivalent amounts of loaded proteins, the membrane was stripped and incubated with anti-actin monoclonal antibody (clone C4; MP Biochemicals) or with anti-green fluorescent protein (anti-GFP) antibody (B-2; Santa Cruz Biotechnology).
Total RNA was isolated from cultures grown in YPD medium to an optical density or A600 of 1.0 by the hot phenol method, treated with RNase-free DNase (Qiagen), and purified with an RNeasy mini kit (Qiagen). RT and real-time PCR amplification were performed with iScript kit (Bio-Rad) using 100 ng of RNA and the following primers: ACT1 (5′-TATGTGTAAAGCCGGTTTTGC-3′ and 5′-GACAATACCGTGTTCAATTGGG-3′), HSP26 (5′-AAGACGTCAGTTAGCAAACACACC-3′ and 5′-CATTGTCGAACCAATCATCTAAGG-3′), HSP12 (5′-AGTCATACGCTGAACAAGGTAAGG-3′ and 5′-AGTCGTGGACACCTTGGAAGAC-3′), and CTT1 (5′-TCACCCATACGCTTCTCAATACTC-3′ and 5′-TCGAACTCCAGTCTACAACCACC-3′).
Cells were grown in YPD medium to an optical density or A600 of 1.0, harvested by centrifugation, washed three times with YP medium (no glucose), resuspended in YP medium, and incubated for 2 h to induce glucose starvation. Glucose was subsequently added to a final concentration of 2%, and samples were removed at indicated time points by rapid filtration through a nitrocellulose filter. Cells were immediately transferred to trichloroacetic acid, cell debris was removed by centrifugation, and cAMP was determined in neutralized lysates by a cAMP enzyme immunoassay kit (GE Healthcare).
Cells were grown in YPD medium to an optical density or A600 of 1.0, and 30 A600 units were harvested by centrifugation and lysed with glass beads in perchloric acid. The lysate was neutralized, and IP3 was assayed with IP3 radioligand binding kit (GE Healthcare).
Glycogen was determined in cells grown in YPD or SC medium at 30°C to an A600 of 1.0. Samples were prepared, and glycogen was assayed after its quantitative conversion to glucose as described previously (61).
InsPs regulate activity of chromatin remodeling complexes in vitro, chromatin structure in vivo, and expression of PHO5 and INO1, two genes known to be regulated by the chromatin structure of their promoters (68, 71). To determine whether InsPs affect expression of all genes controlled by the Swi/Snf chromatin remodeling complex, we identified transcriptional targets of InsPs by genome-wide expression analysis. Three independent hybridizations were performed using RNA isolated from three independent cultures of WT, plc1Δ, ipk2Δ, and ipk1Δ strains. In rich medium, plc1Δ, ipk2Δ, and ipk1Δ mutations affect expression of 133, 229, and 11 genes, respectively. Within this set, expression of 111, 109, and 9 genes is increased and expression of 22, 120, and 2 genes is decreased in plc1Δ, ipk2Δ, and ipk1Δ mutants, respectively (Fig. 1B and C) (see supplemental material). Since the affected genes are scattered throughout the genome, it appears that InsPs control transcription at the level of individual genes, rather than affecting larger chromosomal domains, such as telomere-proximal genes, and are able to both activate and inhibit transcription. The results indicate that Ipk1p inactivation affects expression of only a small number of genes. Plc1p activity results in decreased transcription of the majority of the affected genes (111 genes display increased transcription and 22 genes display decreased transcription in the plc1Δ strain), while Ipk2p is involved in the activation and repression of approximately the same number of genes. Since we used strains with deletion mutations in the genome-wide analysis, the results necessarily reflect both direct and indirect effects of the loss of Plc1p, Ipk2p, and Ipk1p. For example, the reason that the number of genes with decreased expression in the ipk2Δ strain is significantly larger than in the plc1Δ strain (Fig. (Fig.1C)1C) may be due to the fact that ipk2Δ cells display somewhat decreased stability of Mcm1p (16), which is required for full expression of certain genes.
To assess the validity and accuracy of our genome-wide analysis, we examined the expression of selected genes using quantitative S1 analysis (Fig. (Fig.1D).1D). In general, the S1 assay results for the selected genes are in good agreement with the results of genome-wide expression analysis. However, we noted several relatively minor differences. For example, expression of RPL7B and RPS22B genes in the plc1Δ strain is reduced 7.7- and 4.8-fold, respectively, according to the DNA array analysis and 10- and 9-fold, respectively, according to the S1 assay. Similarly, expression of HSP12 and HSP26 genes in the plc1Δ strain is increased 8.3- and 3-fold, respectively, as measured by DNA array analysis, and 6.8- and 5-fold, respectively, as measured by the S1 assay.
When we compared sets of genes affected in plc1Δ, ipk2Δ, and ipk1Δ strains with genome-wide expression analysis of other mutants, we found significant correlation between genes upregulated in the plc1Δ mutant and genes that are activated by increased levels of Msn2p, genes that are repressed by Srb10p, and genes that are derepressed by diauxic shift (Table (Table22 and Fig. Fig.2).2). Some correlation was also found between genes upregulated in plc1Δ and ipk2Δ mutants and genes repressed by Mot1p and Bur6p (Table (Table2).2). It appears that the increased expression of stress-responsive genes in plc1Δ cells occurs irrespectively of the media, since the expression of HSP12 and HSP26 is also elevated in plc1Δ cells grown in SC medium (data not shown).
InsPs were shown to regulate chromatin remodeling and transcription (68, 71). However, comparison of sets of genes affected in plc1Δ, ipk2Δ, and ipk1Δ strains with sets of genes regulated by Swi/Snf or RSC complexes did not show significant overlap, indicating that most genes regulated by these chromatin remodeling complexes in rich medium do not require InsPs for proper expression (Table (Table3).3). Individual InsPs can either activate or inhibit different chromatin remodeling complexes (68), and in turn, chromatin remodeling complexes can either increase or decrease expression of different sets of genes (1a, 34, 51, 72). This mechanism can provide for a very complex mode of regulation of gene expression by InsPs. To consider the possibility that genes upregulated in plc1Δ, ipk2Δ, and ipk1Δ strains might correlate with genes downregulated in swi2, rsc3, and rsc30 mutants, we compared the sets of all affected genes (upregulated and downregulated) and genes that were only upregulated and genes that were only downregulated in the corresponding strains (Table (Table3).3). However, even these comparisons did not suggest strong correlation between the sets of affected genes.
The levels of individual mRNAs are affected by transcriptional initiation, elongation, and mRNA stability, among other factors. One possible explanation for the increased abundance of mRNAs encoded by stress-responsive genes in plc1Δ cells is increased transcriptional initiation due to facilitated assembly of the preinitiation complex at the corresponding promoters. To test this possibility, we performed chromatin immunoprecipitation (ChIP) experiments and determined recruitment of RNA polymerase II and the TATA binding protein (TBP) to several stress-responsive promoters in WT and plc1Δ cells (Fig. (Fig.3).3). For a control, we also determined recruitment of RNA polymerase II and TBP to several other promoters that are not regulated by stress. As shown in Fig. Fig.3,3, the TBP and RNA polymerase II occupancy of ADH1, ACT1, and PYK1 promoters in the WT and plc1Δ cells does not significantly differ. As expected from our DNA array and S1 assay data, RNA polymerase II occupancy at the RPS22B and RPL7B promoters is significantly decreased in plc1Δ cells and thus correlates with transcriptional output. Importantly, the recruitment of both TBP and RNA polymerase II to the stress-responsive promoters of HSP26, HSP12, PIR3, and TKL2 genes is significantly elevated in plc1Δ cells (Fig. (Fig.3).3). These results thus suggest that the increased expression of the stress-responsive genes in plc1Δ cells is caused by facilitated assembly of the preinitiation complex.
We found significant correlation between genes that are upregulated in the plc1Δ strain and genes that are either repressed by Srb10p or are activated by an elevated level of Msn2p (Fig. (Fig.22 and Table Table2).2). To determine the mechanism responsible for increased expression of stress-responsive genes in plc1Δ cells, we considered two mechanisms, one involving Srb10p and one involving Msn2p.
Srb10p is a cyclin-dependent kinase that regulates transcription at specific promoters by phosphorylating and affecting activities of several transcription factors, including Msn2p (10). Srb10p is a component of the Srb mediator complex that is tightly associated with RNA polymerase II to form the holoenzyme (38, 56). Srb8, -9, -10, and -11 proteins form a holoenzyme subcomplex (32, 44) that negatively affects transcription of a subset of genes (34, 44). The cellular steady-state level of Srb10p is significantly decreased during diauxic shift, and Srb10p is almost undetectable during stationary phase (34). This mechanism is likely responsible for the induction of Srb10p-repressed genes during diauxic shift (34). The level of Srb10p is also significantly decreased when cells are grown in medium with a poor nitrogen source (57). To test the possibility that Plc1p regulates the expression of Srb10p-repressed genes by affecting the steady-state level of Srb10p, we determined the level of Srb10p in WT and plc1Δ cells. As shown in Fig. Fig.4A,4A, deletion of PLC1 does not result in decreased level of Srb10p, and therefore, Plc1p regulates expression of Srb10p-repressed genes by a different mechanism.
To assess occupancy of Msn2p at Msn2p-regulated promoters, we performed ChIP analysis with cells expressing Msn2p tagged at the C terminus with three copies of the HA epitope (Msn2p-3xHA). Msn2p-3xHA was immunoprecipitated from cleared preparations of cross-linked and sheared chromatin, and the amount of target DNA loci in the immunoprecipitates was determined by real-time PCR. The amounts of Msn2p recruited to HSP26, HSP12, PIR3, TKL2, and SIP18 promoters were about two to five times higher in the plc1Δ cells than in the control cells (Fig. (Fig.4B).4B). These results suggest that the increased expression and facilitated assembly of the preinitiation complex at the stress-responsive promoters in plc1Δ cells are mediated by increased recruitment of Msn2p.
To further demonstrate that in plc1Δ cells the increased expression of the stress-responsive genes HSP26, HSP12, and CTT1 depends on Msn2p, we determined the expression of the three genes in WT, plc1Δ, msn2Δ msn4Δ, and plc1Δ msn2Δ msn4Δ cells (Fig. (Fig.4C).4C). The increased expression in plc1Δ cells is almost completely abolished by deleting the MSN2 and MSN4 genes, further supporting the model in which increased recruitment of Msn2p to stress-responsive promoters in plc1Δ cells triggers assembly of the preinitiation complex and transcriptional activation. However, unlike msn2Δ msn4Δ cells, expression of HSP26, HSP12, and CTT1 is detectable in plc1Δ msn2Δ msn4Δ cells, suggesting that Plc1p also affects expression of stress-responsive genes by an Msn2p/Msn4p-independent mechanism.
Transcriptional activation of Msn2p-dependent genes is mediated by nuclear translocation and by increased binding of Msn2p to the STRE elements in the promoters of stress-responsive genes (4, 10, 24, 25, 33). Protein kinase A (PKA) negatively regulates transcription of Msn2p-dependent genes by phosphorylating Msn2p and promoting its cytosolic localization (24). One of the amino acid residues of Msn2p that is phosphorylated by PKA is serine 620. Phosphorylation of S620 by PKA creates an epitope that is recognized by an antibody specific for the phosphorylated form of mammalian CREB (25). To determine whether PKA-mediated phosphorylation of Msn2p on S620 is affected in plc1Δ cells, WT and plc1Δ cells were transformed with a plasmid expressing an Msn2p-GFP fusion (24), and the level of Msn2p phosphorylation was assessed by Western blotting with anti-phospho-CREB antibody (25, 48, 75). A significant decrease in Msn2p phosphorylation at S620 was found in plc1Δ cells (Fig. (Fig.5A).5A). This result suggests that Plc1p promotes phosphorylation of at least one of the PKA substrates. The increased expression of Msn2p-dependent stress-responsive genes and decreased phosphorylation of Msn2p in plc1Δ cells are not caused by altered level of Msn2p, as indicated by Western blot analysis (Fig. (Fig.5B5B).
To test whether plc1Δ cells display other phenotypes characteristic of low PKA activity, we determined the level of glycogen accumulated (Fig. (Fig.5C).5C). High PKA activity inhibits glycogen accumulation (7), and low PKA activity results in an increased level of glycogen (11, 75, 82). We found that plc1Δ cells accumulate glycogen to a significantly higher level than WT cells, similar to ras2Δ cells (Fig. (Fig.5C).5C). This result is consistent with decreased level of PKA signaling in plc1Δ cells. Not surprisingly, in a genome-wide survey, the plc1Δ/plc1Δ homozygous diploid was found to accumulate glycogen to a significantly higher level than the corresponding WT diploid strain (76).
The most obvious mechanism that would explain lower PKA activity in plc1Δ cells is a defect in glucose-induced cAMP synthesis. Wild-type and plc1Δ cells were grown to exponential phase, washed, and starved for glucose. cAMP levels were determined following the readdition of glucose. Indeed, it appears that in plc1Δ cells cAMP synthesis following the addition of glucose was somewhat attenuated (Fig. (Fig.5D).5D). This result suggests that the lower level of PKA signaling in plc1Δ cells can be explained by decreased synthesis of cAMP.
To determine genetically whether Plc1p is involved in regulation of PKA, we tested genetic interactions between plc1Δ, ras2Δ, gpa2Δ, and gpr1Δ mutations. Ras1p and Ras2p activate adenylate cyclase (encoded by the CDC35 gene), resulting in an increased synthesis of cAMP and PKA activation (6). cAMP synthesis and PKA activity are also regulated by a G-protein α-subunit Gpa2p that is coupled to a cell surface receptor Gpr1p (48, 75). Synthetic lethality interaction was observed between ras2Δ and gpa2Δ mutations (39). Plc1p was previously found to physically interact with both Gpr1p and Gpa2p and to be required for the interaction between Gpr1p and Gpa2p (2). Our results indicate that while plc1Δ gpa2Δ and plc1Δ gpr1Δ double mutants grow as well as the plc1Δ single mutant, the plc1Δ ras2Δ mutant displays a strong synthetic growth defect (Fig. (Fig.5E).5E). This result is not entirely surprising since genetic interaction between plc1Δ and ras2Δ was also observed in a strain with Σ12785 genetic background (2).
To gain additional support for a role of Plc1p in the Gpr1p-Gpa2p signaling pathway, we determined IP3 levels in WT, plc1Δ, gpr1Δ, and gpa2Δ cells (Fig. (Fig.5F).5F). Since Plc1p-mediated hydrolysis of PIP2 is the only pathway for IP3 synthesis in S. cerevisiae (60, 78), the level of IP3 in cells is a good measure of in vivo Plc1p activity. Interestingly, the level of IP3 in gpr1Δ cells was reduced to about 16% of the WT level. This result suggests that Gpr1p enhances PIP2 hydrolysis by Plc1p and implicates Plc1p as a possible component of the Gpr1p-Gpa2p signaling module. The mechanism may include both direct regulation of Plc1p by Gpr1p or Gpr1p-mediated recruitment of Plc1p to the plasma membrane, where Plc1p has access to its substrate, PIP2.
The genetic interactions are thus in agreement with our findings indicating a role of Plc1p in PKA regulation and suggest that Plc1p acts together with Gpr1p and Gpa2p in a pathway separate from the Ras1p/Ras2p pathway, and the two pathways converge on PKA. If this model were correct, then activation of PKA would be predicted to suppress increased expression of stress-inducible genes and glycogen accumulation in plc1Δ cells. We constructed plc1Δ bcy1Δ and plc1Δ gpb1Δ gpb2Δ strains and determined expression of stress-inducible genes and glycogen accumulation (Fig. (Fig.6A6A and and5C).5C). Bcy1p is a regulatory subunit of PKA that dissociates from the catalytic subunits upon binding cAMP (73). PKA activity in bcy1Δ cells is thus independent of cAMP level. Gpb1p/Krh2p and Gpb2p/Krh1p are kelch repeat proteins that negatively regulate PKA activity (3, 28, 29, 30, 58, 62). Increased expression of stress-inducible genes and accumulation of glycogen in plc1Δ cells are suppressed by deletion of BCY1 or GPB1 and GPB2 (Fig. (Fig.6A),6A), thus indicating that Plc1p functions upstream of PKA. These phenotypes of plc1Δ cells were also suppressed by activated alleles of RAS2 and GPA2 that stimulate PKA activity (Fig. 6B and C). Since expression of stress-inducible genes and accumulation of glycogen in plc1Δ bcy1Δ cells is higher than in bcy1Δ cells, it appears that plc1Δ mutation affects expression of the stress-inducible genes and glycogen accumulation by cAMP-dependent and cAMP-independent mechanisms.
Our cumulative results thus support a model in which Plc1p is involved in PKA signaling by affecting cAMP synthesis, probably through the Gpr1p/Gpa2p pathway. In addition, Plc1p appears to affect transcription of stress-inducible genes and accumulation of glycogen by a cAMP-independent mechanism (Fig. (Fig.77).
While at least some of the roles of phosphatidylinositol phosphates in many aspects of cell physiology have been recognized and studied (13, 16, 52), the roles of InsPs have only begun to emerge. Interestingly, some of the processes regulated by InsPs are nuclear and associated with DNA metabolism and transcription. In mammalian cells, IP6 binds to the Ku70 and Ku80 subunits of the DNA-dependent protein kinase DNA-PK and stimulates double-strand break repair (26, 27). In yeast cells, IP6 is required for efficient export of mRNA from the nucleus (78), while IP4 and IP5 regulate chromatin remodeling and transcription (68, 71) and inositol pyrophosphates mediate homologous DNA recombination (49), cell death, and telomere length (65, 79). In addition, inositol pyrophosphates are able to phosphorylate proteins in vivo by a nonenzymatic mechanism (64).
The roles of InsPs in the regulation of both recruitment and activity of chromatin remodeling complexes are well established (68, 71). However, our DNA array results suggest that InsPs may not affect expression of all genes regulated by the Swi/Snf complex or by other chromatin remodeling complexes. We did find a significant correlation between the plc1Δ data set and the set of stress-responsive genes that are activated by increased levels of Msn2p. Since synthesis of IP6 is required for efficient export of mRNA from the nucleus (78), increased expression of stress-responsive genes in plc1Δ cells could be a consequence of cellular stress caused by a block in mRNA export. However, this does not seem to be the case, since ipk1Δ cells also suffer from lack of IP6 and accumulation of mRNA in the nucleus (78) but do not exhibit increased expression of the stress-responsive genes (Table (Table22).
The mechanism responsible for increased expression of stress-inducible genes in plc1Δ cells involves decreased cAMP synthesis and PKA activity and decreased level of PKA-mediated phosphorylation of Msn2p. This ultimately results in increased binding of Msn2p to the STRE elements in the promoters of stress-responsive genes and their transcriptional activation (4, 10, 24, 25, 33). This notion is supported by several lines of evidence. First, and perhaps most importantly, plc1Δ cells display decreased PKA-mediated phosphorylation of Msn2p at serine 620 that allows Msn2p in plc1Δ cells to translocate to the nucleus, bind to the STRE element, and activate transcription of stress-responsive genes. Second, somewhat decreased basal and glucose-induced cAMP levels were found in plc1Δ cells. Third, plc1Δ cells accumulate glycogen to a higher level than WT cells do, a phenotype characteristic of mutants with low PKA activity. The increased glycogen accumulation was previously detected for a plc1Δ homozygous diploid strain in a genome-wide screening (76). Fourth, plc1Δ demonstrates a strong synthetic growth defect with the ras2Δ mutation, but not with the gpr1Δ or gpa2Δ mutation. This result is also in agreement with previous work that demonstrated physical interaction between Plc1p, Gpr1p, and Gpa2p (2). In addition, plc1Δ cells display decreased transcription of several genes encoding ribosomal proteins (Fig. (Fig.1D)1D) (see supplemental material), another hallmark of cells with low PKA activity.
Some of these phenotypes, characteristic of cells with low PKA activity, are also elicited by depletion or inactivation of Tor kinases (9, 31, 67, 82). Under favorable nutrient conditions, Tor represses starvation-specific and stress-responsive genes by sequestering several transcriptional factors, such as Msn2p/Msn4p in the cytoplasm (4). Thus, the phenotypes of plc1Δ cells could be also explained by decreased Tor signaling, which would suggest that Plc1p functions upstream of Tor. However, this possibility does not seem likely, since inhibition of Tor by rapamycin does not result in decreased cAMP synthesis (82) and plc1Δ cells display reduced PKA-mediated phosphorylation of Msn2p on S620 (Fig. (Fig.5A5A).
On the basis of our results and previously published work (2), we propose a model in which Plc1p functions together with Gpr1p and Gpa2p in a signaling pathway that is parallel to Ras1p/Ras2p and upstream of PKA (Fig. (Fig.7).7). While Ras proteins stimulate adenylate cyclase to produce cAMP, which then activates PKA (37, 74), Gpa2p appears to activate PKA by a cAMP-dependent (11, 47) and cAMP-independent mechanism (48, 77). Gpa2p stimulates adenylate cyclase, resulting in an increased synthesis of cAMP and PKA activation (11, 47). The mechanism by which Gpa2p regulates PKA also involves kelch repeat-containing proteins Gpb1p and Gpb2p (28, 48). Gpb1p and Gpb2p bind Ira1p and Ira2p, two GTPase activating proteins of Ras. Deletion of GPB1 or GPB2 destabilizes Ira1p/Ira2p, which results in an increased Ras activation, cAMP synthesis, and PKA signaling (30). In addition, Gpb1p/Gpb2p stimulates the association between PKA regulatory and catalytic subunits, thus inhibiting PKA signaling (62).
The simplest model that explains the role of Plc1p in Gpr1p-Gpa2p-Gpb1p/Gpb2p signaling to PKA would be that Plc1p is required for transmission of signal from Gpr1p to Gpa2p and normal level of cAMP synthesis (Fig. (Fig.7).7). This seems quite possible, given the fact that Plc1p was reported to be required for physical interaction between Gpr1p and Gpa2p (2) and that full Plc1p function and synthesis of IP3 requires Gpr1p (Fig. (Fig.5F).5F). However, since ipk2Δ cells also display increased transcription of Msn2p-dependent stress-responsive genes (Table (Table2)2) (see supplemental material), it appears that Ipk2p and synthesis of InsPs are required for normal PKA signaling. It will be important to determine whether the catalytic activity of Plc1p is required for normal regulation of cAMP synthesis and PKA signaling or whether Plc1p serves only as a scaffold protein that mediates the interaction between Gpr1p and Gpa2p independently of its catalytic activity. Since the increased expression of stress-inducible genes and glycogen accumulation in plc1Δ cells are not fully suppressed by bcy1Δ or gpb1Δ gpb2Δ mutations (Fig. (Fig.5C5C and and6A),6A), it suggests that cAMP-dependent and cAMP-independent mechanisms operate in plc1Δ cells. This notion is in agreement with the observation that expression of HSP26, HSP12, and CTT1 in plc1Δ cells is partly independent of Msn2p/Msn4p (Fig. (Fig.4C).4C). A similar situation was found for the Ccr4-Not complex that regulates expression of stress-responsive genes by an Msn2p-dependent and Msn2p-independent mechanism (43).
The molecular mechanisms of Plc1p's role in the regulation of the cAMP/PKA pathway remain to be determined. Because PLC and PKA enzymes are present in all eukaryotes, it will be of interest to determine whether PLC is also involved in PKA regulation in other organisms.
This work was supported by grants from the National Institutes of Health (GM076075) and the American Cancer Society (RSG-01-145-01-CCG) to A.V. A.D. and J.C. were supported by U.S. Department of Education grant P200A010130.
We thank M. N. Hall, J. P. Hirsch, A. Kikuchi, L. C. Myers, M. Proft, H. Ronne, C. Schüller, D. J. Stillman, S. R. Wente, F. Winston, and J. D. York for strains and plasmids and I. Vancurova for helpful comments.
Published ahead of print on 28 March 2008.
†Supplemental material for this article may be found at http://ec.asm.org/.