Activated Signal Transduction Kinases Frequently Occupy Target Genes

See allHide authors and affiliations

Science  28 Jul 2006:
Vol. 313, Issue 5786, pp. 533-536
DOI: 10.1126/science.1127677


Cellular signal transduction pathways modify gene expression programs in response to changes in the environment, but the mechanisms by which these pathways regulate populations of genes under their control are not entirely understood. We present evidence that most mitogen-activated protein kinases and protein kinase A subunits become physically associated with the genes that they regulate in the yeast (Saccharomyces cerevisiae) genome. The ability to detect this interaction of signaling kinases with target genes can be used to more precisely and comprehensively map the regulatory circuitry that eukaryotic cells use to respond to their environment.

Signal transduction pathways mediate the cellular response to specific environmental or developmental signals. The activation of signal transduction pathways can lead to phosphorylation of transcription factors (1, 2), histones (3), chromatin-modifying complexes, and the transcription machinery (4, 5). These modifications contribute to changes in the gene expression program. Although the traditional view has been that most phosphorylation events do not occur at the genes that are ultimately controlled by signal transduction pathways, recent reports have revealed that at least one mitogen-activated protein kinase (MAPK)—high-osmolarity glycerol 1p (Hog1p) in yeast and its homolog p38 in humans—physically occupies certain genes where it regulates gene expression [(610), reviewed in (5)]. This evidence, and the knowledge that the terminal kinases of multiple signal transduction pathways can be found in the nucleus under activating conditions (11), led us to investigate the possibility that components of activated signal transduction pathways generally become associated with chromatin at the genes that they activate.

To confirm previous reports that the MAPK Hog1p in yeast occupies genes upon exposure of cells to osmotic stress and to identify the complete set of genes that were so occupied, we performed chromatin immunoprecipitation coupled with microarrays (ChIP-Chip) experiments using a yeast strain in which endogenous Hog1p has a tandem affinity purification (TAP) tag (Fig. 1). The presence of the TAP tag was verified for this and all other yeast strains used in this study (fig. S1). In cells exposed to 0.4 M NaCl for 5 min (6), we identified 36 genes that were occupied by Hog1p at high confidence and showed increased transcription after exposure to NaCl or KCl (Fig. 1B). Among these were all of the seven genes previously found to be occupied by Hog1p (6, 7, 12) (Fig. 1B). We detected little occupancy of Hog1p at genes before osmotic stress, which is consistent with evidence that Hog1p is translocated into the nucleus during osmotic stress (11, 13, 14). Most genes that were occupied by Hog1p during osmotic stress showed an altered expression pattern in cells lacking Hog1p (Fig. 1B). Hog1p occupancy was highest at the promoters of genes but was also observed throughout the entire transcribed region of these genes (Fig. 1C and fig. S2).

Fig. 1.

Recruitment of Hog1p to promoters and transcribed regions of genes activated by osmotic stress. (A) The Hog MAPK pathway in S. cerevisiae. (B) Genes that are bound by Hog1p (table S2) and induced during osmotic stress (with NaCl or KCl). The maximum ChIP enrichment of Hog1p along each gene before and 5 min after NaCl addition are shown in blue. Previously identified target genes are indicated by asterisks. Changes in expression in response to osmotic stress (red for induction and green for repression) of Hog1p-bound genes are displayed for wild-type cells (WT) and for a strain lacking Hog1p (hog1Δ). Additional analysis of Hog1-occupied genes can be found in the supporting online materials. (C) Occupancy of the STL1 gene by Hog1p in control medium [yeast extract, peptone, and dextrose (YPD)] (blue line) and 5 min after the induction of osmotic stress with 0.4 M NaCl (red line), based on merged duplicate data from genome-wide ChIP-Chip analyses. The genomic positions of probe regions and their enrichment ratios are displayed on the x and y axes, respectively. Open reading frames (ORFs) are depicted as gray rectangles, and arrows indicate the direction of transcription.

The MAPKs Fus3p and Kss1p are activated in response to pheromone exposure and induce the expression of mating genes in yeast (15) (Fig. 2A). We used genome-wide ChIP-Chip analysis to determine whether Fus3p and Kss1p occupy a specific set of genes upon activation (Fig. 2). Nine genes were occupied by Fus3p and showed increased transcription within 5 min after exposure to mating pheromone (Fig. 2B). Essentially the same set of genes was occupied by Kss1p (Fig. 2D). These genes were previously shown to be dependent on the pheromone MAPK pathway for their expression (16). Enrichment of both kinases was observed throughout the transcribed regions of these genes (Fig. 2, C and E).

Fig. 2.

Recruitment of Fus3p, Kss1p, and Ste5p to genes expressed in response to alpha pheromone treatment. (A) The pheromone response MAPK pathway in S. cerevisiae. (B, D, and F) Genes bound by Fus3p (B), Kss1p (D), and Ste5p (F) at high confidence (tables S3 to S5) that are induced after alpha pheromone treatment. The occupancy of Fus3p, Kss1p, and Ste5p is shown as maximum ChIP enrichment along each gene in blue. Changes in the expression of these genes during pheromone treatment relative to untreated control samples are displayed in red (induction) and green (repression). (C, E, and G) Occupancy of the AGA1 gene by Fus3p (C), Kss1p (E), and Ste5p (G) in control YPD medium (blue line) and 5 min after exposure to alpha pheromone (5 μg/ml) (red line), based on merged duplicate or triplicate data from genome-wide ChIP-Chip analyses.

Ste5p, the central scaffold protein of the pheromone response pathway, interacts with Fus3p and possibly Kss1p at the plasma membrane but can also be found in the nucleus (11, 15, 17). We found that TAP-tagged Ste5p occupied essentially the same mating genes that were bound by Fus3p and Kss1p (Fig. 2F). Ste5p was observed throughout the transcribed regions of these genes (Fig. 2G). These results suggest that Ste5p may function as an adaptor for protein-protein interactions both at the plasma membrane and in the nucleus.

The cyclic adenosine monophosphate (cAMP)–activated protein kinase A (PKA) is stimulated by an increased concentration of intracellular cAMP when yeast are exposed to fermentable carbon sources such as glucose (Fig. 3A) (18). There are three PKA catalytic subunits in yeast: Tpk1p, Tpk2p, and Tpk3p. Genome-wide ChIP-Chip analyses suggested that Tpk1p occupies the entire transcribed region of most actively transcribed genes in cells grown in glucose media (Fig. 3, B and C). To further test this possibility, we determined whether Tpk1p occupancy would change at genes that were dynamically repressed or activated as yeast cells were subjected to different environmental conditions. Indeed, Tpk1p occupancy was reduced at genes whose expression was reduced when cells were transferred to a nonfermentable carbon source (glycerol) (Fig. 3D). In contrast, Tpk1p became associated with genes that were activated when cells were exposed to galactose (Fig. 3E). Occupancy at these galactose-inducible genes was dependent on gene activation because it was not detected in strains lacking the transcriptional activator Gal4p (Fig. 3E). These results confirm that Tpk1p generally becomes physically associated with actively transcribed genes and that occupancy occurs throughout the transcribed portions of these genes.

Fig. 3.

Occupancy of transcribed regions of active genes by Tpk1p and of promoters of ribosomal protein genes by Tpk2p. (A) The cAMP/PKA signaling pathway in S. cerevisiae. (B) Occupancy of Tpk1p at a portion of chromosome VII, containing the PMA1 and LEU1 genes, in the presence of glucose based on data from genome-wide ChIP-Chip analyses. The transcriptional frequency of the corresponding ORF (26) is indicated as mRNA per hour underneath each ORF. (C) Average Tpk1p enrichment for classes of different transcriptional frequencies [determined by means of metagene analysis (27)]. The genome's 5324 genes were divided into five classes according to their transcriptional rate (26). A fixed length was assigned to each ORF and intergenic region, and probes were assigned to the nearest relative position and averaged for each class. (D) Tpk1p occupancy at a portion of chromosome XII, containing the YEF3 gene whose transcription is substantially reduced during growth in medium containing glycerol (blue line) as compared to that in control medium (YPD) containing glucose (red line). (E) Tpk1p occupancy at GAL1-10 locus in glucose-containing medium (blue line) after the addition of galactose (red line with solid circles) and of galactose in the absence of Gal4p (gal4Δ) (red line with open circles). (F) Tpk2p occupancy at the promoter of the RPS11B gene during oxidative stress and in control medium (YPD) containing glucose.

We then investigated whether Tpk2p occupies specific portions of the genome. Tpk2p was found almost exclusively associated with the promoters of ribosomal protein genes (Fig. 3F, fig. S3, and table S6). Gene occupancy by Tpk2p did not correlate with transcription rates throughout the genome, and Tpk2p remained associated with its target genes when cells were exposed to oxidative stress, which leads to reduced transcription of ribosomal protein genes (Fig. 3F). We did not detect Tpk3p occupancy on chromatin under the conditions used here (rich media, oxidative stress, and pheromone exposure). Although we have not shown that occupancy of genes by Tpk1p and Tpk2p regulates gene expression, previous studies have shown that PKA phosphorylates the Srb9 subunit of the Mediator complex (19) and that PKA activity regulates ribosomal gene expression (2022). The idea that some PKA family members might operate, at least in part, through occupancy of actively transcribed genes is attractive because it might provide an efficient means for cells to respond to the nutrient environment at the level of gene expression.

Our finding that most activated MAPKs and PKAs in yeast become associated with distinct target genes changes our perception of the sites at which signaling pathways act to regulate gene expression. With the exception of Hog1p and p38, studies of the effect of signal transduction pathways on gene expression have not implied that the activities of MAPKs or PKAs involve genome occupancy. Although it is still possible that the phosphorylation of transcriptional regulators also occurs elsewhere in the cell, the detection of kinases by ChIP-Chip analyses at target genes suggests a model in which regulation by signal transduction kinases often occurs at the genes themselves. In this model, kinases become physically localized at specific sites in the genome by association with transcription factors, chromatin regulators, the transcription apparatus, nucleosomes, or nuclear pore proteins that are associated with subsets of actively transcribed genes (510, 19, 2325) (fig. S4).

The kinases studied here associate with target genes in at least three different patterns, suggesting that there are different mechanisms involved in their association with genes. Tpk2p was found only at the promoter regions of its target genes. Hog1p occupancy was greatest at the promoters but also occurred to a limited extent within the transcribed regions of genes. Fus3p, Kss1p, and Tpk1p showed the greatest occupancy over the transcribed regions of genes. ChIP-Chip experiments show that DNA binding transcription factors and promoter-associated chromatin regulators generally occupy the promoters of genes, whereas transcription elongation factors, gene-associated chromatin regulators, certain histone modifications, and nuclear pore proteins are found enriched along the transcribed regions of genes (figs. S2 and S4). Preferential binding to these factors could explain the localization of the kinases.

Many features of signal transduction pathways are highly conserved in eukaryotes, so it is reasonable to expect that MAPKs and PKAs of higher eukaryotes may also be found to occupy genes that they regulate. Indeed, a human homolog of Hog1p, p38, occupies and activates the myogenin (MYOG) and muscle-creatine kinase (CKM) promoters during human myogenesis (10). The observation that components of many signal transduction pathways physically occupy their target genes upon activation should facilitate the mapping of the regulatory circuitry that eukaryotic cells use to modify gene expression in response to a broad range of environmental cues.

Supporting Online Material

SOM Text

Figs. S1 to S5

Tables S1 to S7


References and Notes

Stay Connected to Science

Navigate This Article