Casein Kinase II Regulation of the Hot1 Transcription Factor Promotes Stochastic Gene Expression*

Background: Dynamic signaling events are required for cell-to-cell gene expression differences during responses to environmental stress. Results: During hyperosmotic stress, casein kinase II (CK2) interacts with and phosphorylates the Hot1 transcription factor. Conclusion: CK2 negatively regulates transcriptional activation to promote cell-to-cell variability in gene expression. Significance: Multiple kinase inputs contribute to stochastic gene activity in response to environmental stress. In Saccharomyces cerevisiae, Hog1 MAPK is activated and induces a transcriptional program in response to hyperosmotic stress. Several Hog1-responsive genes exhibit stochastic transcription, resulting in cell-to-cell variability in mRNA and protein levels. However, the mechanisms governing stochastic gene activity are not fully defined. Here we uncover a novel role for casein kinase II (CK2) in the cellular response to hyperosmotic stress. CK2 interacts with and phosphorylates the Hot1 transcription factor; however, Hot1 phosphorylation is not sufficient for controlling the stochastic response. The CK2 protein itself is required to negatively regulate mRNA expression of Hot1-responsive genes and Hot1 enrichment at target promoters. Single-cell gene expression analysis reveals altered activation of Hot1-targeted STL1 in ck2 mutants, resulting in a bimodal to unimodal shift in expression. Together, this work reveals a novel CK2 function during the hyperosmotic stress response that promotes cell-to-cell variability in gene expression.

In Saccharomyces cerevisiae, Hog1 MAPK is activated and induces a transcriptional program in response to hyperosmotic stress. Several Hog1-responsive genes exhibit stochastic transcription, resulting in cell-to-cell variability in mRNA and protein levels. However, the mechanisms governing stochastic gene activity are not fully defined. Here we uncover a novel role for casein kinase II (CK2) in the cellular response to hyperosmotic stress. CK2 interacts with and phosphorylates the Hot1 transcription factor; however, Hot1 phosphorylation is not sufficient for controlling the stochastic response. The CK2 protein itself is required to negatively regulate mRNA expression of Hot1-responsive genes and Hot1 enrichment at target promoters. Single-cell gene expression analysis reveals altered activation of Hot1-targeted STL1 in ck2 mutants, resulting in a bimodal to unimodal shift in expression. Together, this work reveals a novel CK2 function during the hyperosmotic stress response that promotes cell-to-cell variability in gene expression.
Cells respond to fluctuating environmental conditions through global alterations in gene expression to facilitate adaptation. In Saccharomyces cerevisiae, the activation of the Hog1 MAPK signaling cascade coordinates the transcriptional response to several types of osmotic stress, such as increasing extracellular concentrations of NaCl (herein referred to as salt stress) (1,2). Under conditions of salt stress, rapid signaling events activate the Hog1 MAPK, which then enters the nucleus and directs a combination of transcriptional activators to initiate the transcriptional response (3,4). Hundreds of gene targets exhibit altered levels of transcription within minutes after exposure to moderate salt stress (0.4 M NaCl). Numerous efforts have extensively characterized the Hog1 MAPK signaling response, with most recent reports uncovering significant cell-to-cell variability in salt-induced gene expression among populations of identical cells (5,6). This cell variability in expression is due to the stochastic activation of transcription of several salt-induced gene targets. Factors known to contribute to stochastic gene activity include the duration of Hog1 nuclear activity, the intracellular concentration of the Hot1 transcription factor, and the chromatin remodeling and modifying events required to transition the promoter to an active state (5,6). The current models for stochastic gene activation in response to salt stress account for a singular kinase input from Hog1 MAPK (5,6).
Sko1 and Hot1 target a subset of overlapping genes, which are among the most highly induced (1,16). Binding of Sko1 and Hot1 to cis-promoter elements leads to the recruitment of Hog1 and the reallocation of RNA polymerase II from housekeeping genes to salt-responsive genes (16). The molecular details orchestrating these events have not been fully defined. In the case of Sko1, Hog1 phosphorylation switches the transcription factor from a repressor to an activator (7). However, for Hot1, the effects of Hog1 phosphorylation remain unclear. The Hot1 target, STL1, requires a Hog1-Hot1 physical interaction for salt-induced expression; however, a Hog1-phosphodead version of Hot1 shows no defects in STL1 expression or recruitment of RNA polymerase II (17,18). Unlike Msn2/Msn4 and Sko1, there is no evidence for PKA regulation of Hot1 or changes in Hot1 nuclear-cytoplasmic localization upon salt stress. Thus, we set out to investigate additional Hog1-inde-pendent kinase inputs regulating Hot1 function in the transcriptional response to salt stress.
Here we identify a novel role for casein kinase II (CK2) 2 in the mechanism for stochastic gene expression during salt stress in S. cerevisiae. In a wild type population of cells exposed to moderate salt stress, stochastic gene activity results in bimodal expression of the Hot1 target STL1. However, in ck2 mutant cells, this bimodality is lost, and STL1 is expressed in all cells. Taken together, CK2 plays a previously undefined role in saltinduced gene expression by negatively regulating the Hot1 transcription factor.

EXPERIMENTAL PROCEDURES
Yeast Strains, Growth Conditions, and Plasmids-Yeast strains listed in Table 1 are of the BY4743 designer deletion S288C background. Yeast mating, sporulation, dissection, and transformations were conducted according to standard procedures (19). Endogenous point mutations were introduced with the delitto perfetto method (20). All S. cerevisiae strains were grown in YPD (2% peptone, 2% glucose, 1% yeast extract) or SC dropout medium at 30°C. To induce salt stress, a stock solution of YPD ϩ 4 M NaCl was added to the final concentrations indi-cated. Plasmids used in this study are listed in Table 2. Molecular cloning was performed as described previously (21).
Immunoprecipitations and Immunoblotting-Cultures were grown to an A 600 of 0.5 and treated with and without 0.4 M NaCl for 10 min. Cells were immediately washed in ice-cold doubledistilled H 2 O, and cell pellets were snap-frozen in liquid nitrogen. Cells were lysed by bead beating in lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA 0.1% Triton X-100, 10% glycerol, 1ϫ protease inhibitors (Roche Applied Science), 0.1 mM PMSF, 1 mM Na 3 NO 4 , 50 mM NaF). Lysates were clarified at 13,000 rpm for 6 min, and supernatant was incubated with camelid GFP-binding protein-conjugated Sepharose beads at 4°C for 1 h. The GFP-binding protein-conjugated beads were washed three times in wash buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 0.1% Triton X-100) and then boiled in 2ϫ SDS buffer. Further analysis with SDS-PAGE and immunoblotting was performed with mouse anti-rabbit (TAP) and rabbit anti-GFP (GFP) antibodies. Blots were imaged using a LI-COR Odyssey system.
Microscopy-Cultures for imaging were diluted from saturated overnight starter cultures to an A 600 of 0.05 and then grown at 30°C for 5 h to an A 600 of 0.4 -0.6. Quantification of nuclear Hog1-GFP was performed in untreated cells and at 5-min intervals up to 60 min after shifting to 0.4 M NaCl. Images were acquired with a standard microscope (BX50, Olympus) This study equipped with a motorized stage (model 999000, Ludl), UPlanF1 ϫ100, numerical aperture 1.30 oil immersion objective, and digital charge-coupled device camera (Orca-R2, Hamamatsu). Quantification of nuclear Hog1-GFP was performed in ImageJ (National Institutes of Health).
Isolation of CK2-Cka1-TAP-expressing S. cerevisiae were grown in 2 liters of YPD at 30°C to an A 600 of 2-3. Cells were harvested, washed in 50 ml of cold PBS, and then further washed with resuspension buffer (20 mM HEPES, pH 7.4, 1.2% polyvinylpyrrolidone, 1 mM dithiothreitol (DTT), 1 mM phenylmethylsulfonyl fluoride (PMSF), 4 g/ml pepstatin A) before freezing in liquid nitrogen. Frozen cells were lysed in the solid phase by milling, using a planetary ball mill (23). Frozen lysate was thawed, resuspended in 50 ml of buffer D (40 mM Hepes, pH 8, 150 mM NaCl, 1 mM PMSF, 1 mM DTT, 20% glycerol), and clarified by centrifugation at 20,000 rpm for 20 min at 4°C. Supernatant was supplemented with Triton X-100 (final concentration 0.1%) and incubated with 200 l of IgG-Sepharose for 2 h at 4°C under rotation. The IgG-Sepharose beads were washed twice with IPP150 buffer (10 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.1% Triton X-100) and twice with TEV cleavage buffer (10 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.5 mM EDTA, 1 mM DTT). TEV protease cleavage was carried out in 200 l of TEV cleavage buffer with 100 units of purified TEV protease for 2 h at room temperature and then placed on ice. Bound ϪTEV, bound ϩTEV, and cleaved fractions were analyzed by SDS-PAGE and Coomassie staining.
In Vitro Kinase Assay-Plasmids used to express maltosebinding protein (MBP) fusions (MBP-Hot1-FL, MBP-Hot1-Nterm, MBP-Hot1-C-term, MBP-hot1-C-term-3A, and MBP) were transformed into Rosetta cells and induced with 400 M isopropyl 1-thio-␤-D-galactopyranoside at 16°C overnight. The cells were lysed in binding buffer (20 mM HEPES, pH 7.5, 145 mM NaCl, 5 mM KCl, 10 mM EGTA, 10 mM EDTA, 1ϫ protease inhibitors (Roche Applied Science), and 0.1 mM PMSF) with sonication. Lysates were centrifuged for 15 min at 15,000 rpm, and the soluble fraction was incubated with amylose resin (New England Biolabs) for 2 h at 4°C. The column was washed with 200 ml of binding buffer and eluted with binding buffer ϩ 10 mM maltose. Purified proteins were dialyzed into kinase buffer (20 mM Tris-HCl, 50 mM KCl, 10 mM MgCl 2 , pH 7.5). In vitro kinase reactions were performed in kinase buffer supplemented with 20 M cold ATP, 5 Ci of [␥-32 P]ATP, and 100 ng of Cka1-TAP (purified yeast CK2) in a 30-l total reaction volume for 1 h at 30°C. Reactions were terminated by boiling in 2ϫ SDS buffer. Samples were analyzed by SDS-PAGE, Coomassie staining, and autoradiography. Mass Spectrometry-Glutathione S-transferase (GST)-Hot1 samples, with or without the addition of CK2, were digested with 1 g of Trypsin Gold (Promega, Madison, WI) overnight at 37°C. The peptides were loaded onto a reverse-phase capillary trap column (360-m outer diameter ϫ 150-m inner diameter) packed with 4 cm of C18 reverse phase material (Jupiter, 5-m beads, 300 Å; Phenomenex) in line with a capillary analytical column (360-m outer diameter ϫ 100-m inner diameter) packed with 20 cm of C18 material (Jupiter C18, 3-m beads or Aqua C18, 3-m beads, Phenomenex). Peptides were gradient-eluted at a flow rate of 500 nl/min using an Eksigent NanoLC Ultra HPLC. Mobile phase solvents consisted of 0.1% formic acid, 99.9% water (solvent A) and 0.1% formic acid, 99.9% acetonitrile (solvent B). A 90-min LC-coupled tandem mass spectrometry (LC-MS/MS) analysis was performed, and the LC gradient consisted of the following: 0 -12 min, 2% B; 12-50 min, 2-40% B; 50 -60 min, 40 -90% B; 60 -65 min, 90% B, 65-70 min, 90 -2% B. This was followed by column equilibration at 2% B from 70 to 90 min. Tandem mass spectra were collected in a data-dependent manner with dynamic exclusion using an LTQ Orbitrap Velos mass spectrometer (Thermo Scientific), equipped with a nanoelectrospray ionization source. For peptide identification, tandem mass spectra were searched against an S. cerevisiae subset of the UniprotKB protein database, appended with the GST-tagged Hot1 protein sequence. Database searches were performed using Sequest (24) (Thermo Scientific) on the Vanderbilt ACCRE Linux cluster, and results were assembled in Scaffold version 3.6.4 (Proteome Software). All searches were configured to include variable modifications of oxidation on methionine, carbamidomethylation on cysteine, and phosphorylation on serine and threonine. The results were searched with 95% filtering criteria, and spectra corresponding to potential sites of phosphorylation were confirmed by manual evaluation. For selected LC-MS/MS analyses, the LTQ Orbitrap Velos was operated using a method consisting of targeted scan events, for which specific m/z values corresponding to Hot1 phosphorylated peptides were provided in the data acquisition method to facilitate collection of targeted MS/MS spectra despite the low intensity of peptide precursors.
Flow Cytometry-Yeast cultures grown to late log phase were diluted and grown for at least 15 h to an A 600 of 0.5. These cultures were treated for various times at the indicated NaCl and sorbitol concentrations. Samples were harvested by diluting into 10 mM Tris, 1 mM EDTA pH 8.0 ϩ 1 g/ml cyclohexamide and immediately measured for GFP fluorescence using a Guava easyCyte flow cytometer. 20,000 cells were collected within gated SSC and FSC populations that excluded doublets and small debris. Data were graphed using FlowJo software.

Hot1 in Vivo Phosphorylation and Activity Is Regulated by
CK2-Previous work has identified CK2 as a kinase required for cellular responses to salt stress (25). Deletion of the CK2 ␤ regulatory subunits results in growth sensitivities to Na ϩ and Liϩ (26); however, functions for CK2 in the response to salt stress remain largely unknown. CK2 targets many cellular proteins, including signaling proteins, transcription factors, and a number of DNA/RNA regulatory factors, poising CK2 to impact multiple stages of the gene expression pathways (27). It was known that Hog1 is not the only kinase regulating Hot1 function because residual phosphorylation is observed in a hog1⌬ and Hog1-phosphodead version of Hot1 (18). Thus, we set out to investigate the requirements for both Hog1 and CK2 in the in vivo phosphorylation of Hot1.
To determine whether CK2 impacts Hot1 phosphorylation, wild type, hog1⌬, cka2⌬, and hog1⌬ cka2⌬ mutant strains expressing Hot1-GFP were grown to early logarithmic phase and shifted to conditions of moderate salt stress (0.4 M NaCl). Whole cell lysates were prepared every 15 min for up to 1 h. Samples were resolved by SDS gel electrophoresis, and immunoblotting was conducted with anti-GFP antibodies. In wild type cells, Hot1-GFP electrophoretic mobility changed with the presence of slower migrating form(s), peaking between 30 and 45 min (Fig. 1A, top, lanes 3 and 4), and then was reduced after 60 min of 0.4 M NaCl stress (lane 5). To test whether the slower migrating forms were due to phosphorylation modifications, the samples from cells shifted to salt stress for 60 min were treated with -phosphatase (PP). Only a single slower migrating band was observed (Fig. 1A, top, lane 6), similar to that before the shift (lane 1), which confirms that this band shift represents Hot1-GFP phosphorylation in response to salt stress. In cka2⌬ cells, a phosphorylated form was detected at 15 min and persisted throughout the time course (Fig. 1A). In contrast, although a slower migrating form was detected in hog1⌬ cells, the shift was not as prominent, and it was only apparent later in the time course at 45-60 min (Fig. 1A). Interestingly, Hot1-mediated transcription is inactivated by 30 min after 0.4 M NaCl stress induction (see below; Fig. 1C). We speculated that the slower migrating forms in the hog1⌬ cells were due to CK2 phosphorylation events. Hot1-GFP phosphorylation was quantified in wild type, hog1⌬, cka2⌬, and the hog1⌬ cka2⌬ double mutant samples by measuring the amount of shifted Hot1-GFP in a set gel area (the same for each lane) above that for the respective position of the prominent Hot1-GFP band in the untreated lysate (lane 1). Phosphorylation was detected in wild type, hog1⌬, and cka2⌬ lysates at all time points (lanes 2-5); however, Hot1-GFP phosphorylation was reduced to levels observed in untreated cells (lane 1) and -phosphatasetreated (lane 6) lysates (Fig. 1A). Taken together, both Hog1 and CK2 inputs regulate the levels of Hot1 phosphorylation in response to 0.4 M NaCl.
Given that CK2 influenced Hot1 phosphorylation at time points when Hot1 transcription activity is inactivated (Fig. 1A), we asked whether CK2 negatively regulates Hot1 functions in transcription. Previous studies report that Hot1 interactions with chromatin occur at some promoters under normal growth conditions (Group 1: STL1 and GPD1), whereas, at other promoters, Hot1 is only significantly enriched under salt stress (Group 2: HSP12, RHR2, and RTC3) (1). To investigate whether altered phosphorylation in ck2 mutants correlated with changes in Hot1-chromatin interactions, chromatin immunoprecipitation (ChIP) experiments were performed with these validated Hot1 gene targets in wild type and cka2⌬ cells. Compared with wild type, Hot1-GFP was significantly enriched at the Group 1 promoters in untreated (Ϫ) cka2⌬ mutant cells (Fig. 1B), and this enrichment persisted after shifting to 0.4 M NaCl for 10 min (Fig. 1B, ϩ samples). Similarly, Hot1-GFP was also more enriched at Group 2 genes in untreated cka2⌬ cells and, in this context, occupied these gene targets in the absence of stress (Fig. 1B, Ϫ samples).
The Hog1 MAPK is strictly required for the activation of Hot1 target genes by directing chromatin remodeling and modifying machinery to alter the local chromatin environment and recruiting RNA polymerase II (17, 28 -30). Therefore, in unstressed cells, the absence of CK2 should not misregulate expression of Hot1 target genes. However, Hot1 enrichment at the gene promoters is a rate-limiting step during the activation of Hot1 gene targets (17). We hypothesized that Hot1 should be primed for transcriptional activation in ck2 mutants. Real-time quantitative PCR experiments were performed to compare Hot1-dependent gene induction in wild type and cka2⌬ mutant cells. As shown in Fig. 1D, the STL1 mRNA level peaked earlier and was more robust in the cka2⌬ mutant cells. RTC3 and HSP12 genes also showed a more robust activation in cka2⌬ mutant cells. The duration of RTC3 and HSP12 mRNA expression extended to later time points, consistent with the accumulation of CK2-dependent phosphorylation of Hot1 from 30 to 60 min after 0.4 M NaCl stress (Fig. 1, A and C). Surprisingly, RTC3 and HSP12 mRNAs were also elevated in untreated cka2⌬ mutant cells in the absence of Hog1 MAPK signaling. This suggested that loss of CK2 negative regulatory input is sufficient to lead to Hog1-independent activation of Hot1 targets. An alternative possibility was that ck2 mutants alter Hog1 MAPK signaling, leading to the inappropriate activation of Hot1 target genes. To distinguish between these two possibilities, Hog1 activity was quantified in wild type and cka2⌬ mutant cells by the percentage of cells with nuclear localized Hog1-GFP (Fig. 1D). Over a time course of 0.4 M NaCl stress, the dynamics of Hog1-GFP nuclear localization in cka2⌬ mutant cells was the same as that for wild type cells. Therefore, altered Hog1 activity did not account for ck2 mutant phenotypes. Overall, we concluded that CK2 inhibits Hot1 binding to chromatin and serves to negatively regulate Hot1-dependent gene expression.
Hot1 Physically Interacts with CK2-We next tested whether Hot1 physically associates with CK2 because in vivo interactions have previously provided a predictive measure for potential CK2 substrates (27). Therefore, co-isolation of epitopetagged Cka1 and Hot1 was tested by immunoprecipitation. Cells expressing both Cka1-GFP and Hot1-TAP were grown to early logarithmic phase, and total cell lysate fractions were isolated before and after shifting to growth in 0.4 M NaCl medium. As shown in Fig. 2A, isolation of Cka1-GFP with GFP-binding protein-Sepharose beads resulted in the co-isolation of Hot1-TAP under both normal and salt stress conditions (lanes 5 and 6). In the absence of Cka1-GFP, Hot1-TAP was not detected (lane 4). The reciprocal isolation was performed with Hot1-GFP and similarly resulted in the co-isolation of Cka1-TAP (Fig. 2B). As shown in Fig. 2C, Hot1-GFP immunoprecipitations also isolated Cka2-TAP, indicating that a CK2 complex composed of both Cka1 and Cka2 subunits is able to interact with Hot1-GFP. Moreover, these interactions were maintained upon Hot1 phosphorylation (gel shift in lanes 6 versus lanes 5), indicating that these interactions occur under conditions of 0.4 M NaCl stress (Fig. 2, B and C). We further tested whether the Cka1 and Hot1 interaction was altered in the absence of the Cka2 subunit. In Fig. 2, D and E, co-isolation of Cka1 and Hot1 in a cka2⌬ strain was reduced when compared with a wild type strain with Cka2 present. Taken together, these results indicated that Hot1 and CK2 interactions occur in unstressed and salt-stressed cells and require an intact Cka2 subunit.
Hot1 Is Directly Phosphorylated by CK2-The previous results demonstrate that CK2 regulates Hot1 activity in vivo. To determine whether Hot1 is a direct substrate for CK2, in vitro phosphorylation assays were performed. The N-terminal region of Hot1 has a predicted coiled-coil structure and contains several known Hog1 phosphorylation sites, whereas the C-terminal region contains a DNA-binding domain (residues 615-695) with homology to Msn1 and Gcr1 transcription factors (Fig. 3A). Bacterially expressed recombinant MBP-tagged Hot1, MBP-Hot1-C-term, MBP-Hot1-N-term, and MBP alone were purified and incubated with [␥-32 P]ATP, and CK2 was isolated through tandem affinity purification from yeast (CK2). Reactions were separated by SDS-gel electrophoresis and analyzed by Coomassie staining and autoradiography (Fig. 3B). Only MBP-Hot1-FL and MBP-Hot1-C-term were phosphorylated by CK2 (lanes 3 and 5).
To determine the sites of CK2 phosphorylation, in vitro phosphorylated GST-tagged Hot1 as well as GST-Hot1 alone were proteolytically digested and analyzed by LC-MS/MS. Following database searching, the approach yielded 79 and 73% sequence coverage of GST-Hot1 in the ϪCK2 and ϩCK2 samples, respectively. In parallel, an in silico approach (31) was also used to scan Hot1 primary sequence for CK2 consensus sites. Analysis of CK2 phosphorylation sites indicates that it is a dualspecificity kinase that targets serine and threonine amino acids upstream of acidic amino acids ((pS/T)XX(E/D)) or previously phosphorylated residues ((pS/T)XX(pS/T)) (27). The CK2 minimum consensus is SXX(E/D) and often includes an enrichment of aspartic acid and glutamic acid residues spanning amino acids n ϩ 4 through n Ϫ 7 (27). Three putative highly acidic CK2 consensus sites (Ser-532, Ser-560, and Thr-561) were identified in the C-terminal Hot1 region just upstream of the annotated DNA-binding domain (residues 615-695) (Fig. 3A). Strikingly, the Hot1 region spanning amino acids 555-585 was readily observed in the ϪCK2 sample, with 45 assigned spectra being identified as peptides in this region, yet this region was completely absent in the Hot1 sequence coverage map for the ϪCK1 sample. These data suggest that phosphorylation of one or multiple CK2 consensus sites precludes the detection of this region with data-dependent acquisition LC-MS/MS. Subsequently, in a targeted LC-MS/MS approach, where the phosphorylated Hot1 peptides were targeted and isolated for tandem mass spectrometric analysis, both singly and doubly phosphorylated Ser(P)-560/Thr(P)-561 Hot1 peptides were identified following database searching.
To test whether these sites are phosphorylated by CK2, sitedirected mutagenesis was used to generate a hot1-3A gene, wherein alanine substitutions would result at each CK2 consensus site (S532A/S560A/T561A). MBP-hot1-C-term-3A protein was purified and tested in an in vitro kinase assay with CK2. Normalizing for the same relative protein levels, the MBP-hot1-C-term-3A protein had a lower level of 32 P incorporation than wild type MBP-Hot1-C-term (Fig. 3, C and D). Thus, one or more of these three sites was modified by CK2.
CK2 Impacts Stochastic Expression of STL1-To test whether CK2 regulation of Hot1 plays a role in the stochastic transcription of STL1 in a population of salt-stressed cells, single cell analysis of STL1 expression was conducted using flow cytometry to detect the production of Stl1-GFP protein from the endogenous STL1 locus. Under 0.4 M NaCl stress, a clear bimodal expression pattern was observed in a wild type cell population with a fraction of the cells lacking detectible levels of Stl1-GFP (Fig. 4A). As expected, in hot1⌬ cells and Hog1 analog-sensitive (hog1-as) cells treated with a 5 M concentration of the selective inhibitor 1NM-PP1, no production of Stl1-GFP was detected (Fig. 4A). Interestingly, in the absence of either CK2 catalytic subunits (Cka1 or Cka2), the cell populations shifted to a greater percentage of Stl1-GFP-expressing cells. A dose response for Stl1-GFP expression was also performed to compare wild type and cka2⌬ cells. At all concentrations of NaCl tested, cka2⌬ cells exhibited a unimodal response (Fig.  4B). Similarly, in a time course experiment comparing wild type and cka2⌬ cells, the cka2⌬ population was unimodal within the time intervals of 15, 30, 45, and 60 min after 0.4 M NaCl stress (Fig. 4C). To assess whether the CK2 input was specific to salt stress or also occurred under other hyperosmotic stresses, Stl1-GFP expression was analyzed after sorbitol treatment (Fig. 4D). In a wild type population of cells, after 60 min of 1 M sorbitol stress, a bimodal distribution was observed under 1 M sorbitol stress after 60 min. However, in the sorbitol, the cka2⌬ population was more responsive and unimodal in distribution. Thus, both salt and sorbitol stress responses for Stl1-GFP expression were altered in cka2⌬ cells. Overall, CK2 acted as a negative  regulatory input under hyperosmotic stress that impacts the stochastic STL1 expression.
To determine whether direct CK2 phosphorylation of Hot1 influences bimodal expression, hot1-3A mutant cells were analyzed (Fig. 4, C and D). As in the cka2⌬ cell population, the hot1-3A cells did not show a strong peak of non-expressing cells and were shifted to a unimodal pattern of Stl1-GFP expression. However, after 60 min of 0.4 M NaCl stress, the percentage of non-responsive cells in wild type and the hot1-3A mutant pop-ulations was 31.7 and 27.5, respectively, as compared with 7.4% in the cka2⌬ mutant (Fig. 4E). Similar results were obtained in 1 M sorbitol (Fig. 4E). Consistent with Stl1-GFP expression in the hot1-3A mutant, the association of hot1-3A-GFP with STL1   CK2-dependent rate-limiting event is required for the proper stochastic activation of the Hog1-responsive, Hot1-dependent genes. CK2 Influences Hot1-dependent Transcription Upstream of Chromatin Alterations at STL1-Hog1-dependent events that transition chromatin at the gene promoters to an "active" state require the activities of the SAGA complex, the Rpd3 histone deacetylase complex, and RSC chromatin-remodeling complex (28 -30). As described previously, deletion of GCN5, which encodes the histone acetyltransferase of SAGA, results in reduced expression of STL1 and a shift to a more bimodal responding population (5). To further investigate the role for CK2 and the epistasis of the distinct CK2-and Hog1-regulated events in Hot1-dependent gene expression, single cell analysis of Stl1-GFP expression was compared for single cka2⌬ and gcn5⌬ mutants and a cka2⌬ gcn5⌬ double mutant. As previously reported (5,6), increased levels for the non-responsive cell population were observed in the gcn5⌬ cultures in both the time course (Fig. 5A) and dose-response experiments (Fig. 5B). Furthermore, the gcn5⌬ mutant and the cka2⌬ gcn5⌬ mutant exhibited similar STL1 expression patterns over the time course of treatment with 0.4 M NaCl. This suggested that CK2dependent inhibition of Hot1 occurs upstream to the Hog1-de-pendent transition of chromatin at the promoters to a transcriptionally active state.

DISCUSSION
This work reveals compelling evidence for a novel regulatory input from CK2 in the Hog1-MAPK transcriptional response. We show that CK2 interacts with and directly phosphorylates the Hot1 transcription factor. We conclude that CK2 is required to negatively regulate Hot1 activity and targeting to chromatin, thereby enhancing modulation of expression for the Hot1-dependent gene, STL1. Thus, CK2-dependent negative regulation of the Hot1 transcription factor influences cell-tocell variations in gene expression.
We identified CK2-mediated phosphorylation of Hot1 within a highly acidic region containing three CK2 consensus sites (Ser-532, Ser-561, and Thr-562). Prior studies have identified two small amino acid elements KR4 (KRRRR, 381-385) and ED5 (EDDDDD, 541-546) that are required for Hot1 activation of transcription (32). The KR4 element mediates the Hog1-Hot1 interaction necessary for targeting Hog1 to saltresponsive promoters. The precise function of the ED5 element remains unclear. CK2 recognition of substrates occurs through a docking interaction at amino acid residues neighboring CK2   phosphorylation sites (27). We speculate that the ED5 element is a docking site for CK2, similar to the KR4 docking site for Hog1. Although the CK2 phosphorylation within the Hot1 acidic region occurs in close proximity to the DNA binding domain (residues 615-695), ChIP experiments with the hot1-3A-GFP protein indicate that phosphorylation of these sites does not alter Hot1 targeting to promoters. However, Hot1-GFP is enriched to promoters in both salt-stressed and unstressed cka2⌬ cells. Considering these results, we conclude that the presence of the Cka2 protein itself is required to negatively regulate Hot1 activity and enrichment to target promoters. Our results provide evidence for an in vivo interaction between Hot1 and CK2. We predict that the formation of a Hot1-CK2 complex leads to alterations in Hot1 chromatin associations and activity in a mechanism that is independent from the phosphorylation of amino acids Ser-532, Ser-560, and Thr-561. In a similar manner, the association of Hot1 with gene promoters occurs independently of its phosphorylation by Hog1 (17,18). Formation of a Hot1-Hog1 complex is predicted to allow Hog1 phosphorylation of other key factors involved in transitioning Hot1 gene targets to an active state (16,28). We speculate that the Hot1-CK2 complex positions CK2 to target additional factors that promote an inactive state for Hot1 gene targets. Future studies are needed to address this intriguing possibility. Hog1 signaling occurs in response to a variety of cellular stresses that activate distinct subsets of genes. Nucleocytoplasmic shuttling of Sko1 provides an additional layer of regulation to ensure the appropriate context of transcriptional activity (15). Unlike Sko1, Hot1 remains localized to the nucleus in the presence and absence of salt stress. Therefore, we propose that additional input from the nuclear localized CK2 provides the appropriate context for Hot1 transcriptional activation.
Others have reported an undefined role for CK2 during salt stress (25,26,33). We have shed light on this CK2 function; however, the mechanisms that relieve CK2 inhibition of Hot1 remain unknown. It is possible that a phosphatase acts to reverse CK2 phosphorylation of Hot1 upon salt stress, resulting in the inducible Hot1 activity. Alternatively, the constitutive activity of CK2 might be modulated under salt stress. A global phosphoproteomicapproachtoidentifysalt-inducedphosphorylation events revealed an enrichment of Hog1 and PKA consensus motifs, confirming previous observations for the saltinduced activity of these respective kinases (34). However, CK2 consensus motifs were not enriched (34), indicating that CK2 activity does not increase during early time points of salt stress. Alternatively, it is possible that salt stress results in a temporary attenuation of CK2 activity. In this scenario, CK2-targeted substrates receive negative regulatory input under normal physiological conditions that are relieved under conditions of salt stress. Mechanisms for regulation of the Ycf1 yeast vacuolar ABC transporter suggest that this may be the case, whereby salt-dependent attenuation of CK2 phosphorylation results in a temporary relief of inhibition Ycf1 (35). If a salt-dependent inhibition of CK2 occurs, then reactivating CK2 would provide negative feedback and a mechanism to return to normal signaling and gene expression. Our results reveal a peak in CK2 phosphorylation of Hot1 at 30 -45 min after exposure to moderate salt stress and support salt-dependent modulation of CK2 activity.
Multiple mechanisms contribute to stochastic gene expression (reviewed in Refs. 36 and 37). Taking this study into account, we propose a model wherein two independent kinase inputs contribute to the stochastic activation of Hot1-dependent genes. A major input from the Hog1 MAPK is required for activation of Hot1-dependent transcription. Elegant studies show that the Hog1 MAPK transitions chromatin structure at salt-responsive promoters and that these events contribute to the stochastic transition of genes from a repressed to activated state (5,6). Here, the CK2-dependent input involves negative regulation of Hot1 activity and localization to target promoters. Potentially, this CK2 input dampens the transcriptional response when conditions are less favorable or provides a mechanism to turn off the Hog1 MAPK transcriptional response once cells have adapted to the change in extracellular salinity. Through epistasis experiments, we find that cells lacking the negative regulatory input from CK2 still show requirements for the histone acetyltransferase activity of SAGA. Overall, in this model, two kinase inputs feed into the regulation of Hot1 transcriptional activity, leading to cell-to-cell variability in salt-induced gene expression.
In summary, we have identified a previously undescribed mechanism of Hot1 inhibition during salt stress. It is intriguing to consider the potential advantages conferred to cells by CK2 regulation of Hot1 activity. This might serve to dampen the Hog1-induced transcriptional response, which is toxic to cells when constitutively activated (38). By promoting variations in cell-to-cell gene expression, CK2 might also contribute to allowing a range of phenotypes within a population for a given environmental insult. This positions CK2 to enhance the fitness of a subset of cells and play a role in ensuring survival in a changing environment. In humans, CK2 targets numerous transcription factors, including known oncogenes and tumor suppressors as well as several that function in the immune system (27). Altered CK2 activity is strongly associated with human cancers (39 -42), and increased consumption of salt induces a population of T cells that contribute to autoimmune disease in mice (43,44). It will be important to understand the conserved functions for CK2 in regulating stochastic transcription in multicellular organisms and whether these functions influence subsets of cells during tumorigenesis and autoimmune responses.