PerspectiveCell Signaling

Protein Kinases Seek Close Encounters with Active Genes

See allHide authors and affiliations

Science  28 Jul 2006:
Vol. 313, Issue 5786, pp. 449-451
DOI: 10.1126/science.1131158

Upon exposure to changes in the environment or to developmental cues during differentiation, a cell reprograms transcription in its nucleus through a circuitry of signals that ultimately alters gene expression. Many of the steps of such signal-transducing cascades are executed by kinases, enzymes that transfer phosphate molecules onto target substrates. Often, kinases at the end of such cascades (terminal kinases) trigger the necessary response by directly phosphorylating transcription factors, coregulatory proteins, or the proteins that, with DNA, make up chromatin. Until recently, the prevailing view has been that terminal kinases operate enzymatically, without stable association with the chromatin that harbors target genes of a signaling pathway. But an alternative model whereby such kinases also play a structural role by binding to factors within transcription complexes at target genes has been slowly gathering support (1). On page 533 of this issue, Pokholok et al. (2) report a global analysis in yeast of the association of kinases with genes that they regulate, further supporting this model. Their findings suggest that such interactions can be observed not only with sequence-specific transcription factors positioned at regulatory (promoter) regions lying upstream of target genes, but also with the coding region of genes in some cases.

The yeast HOG mitogen-activated protein kinase (MAPK) pathway responds to changes in external osmolarity by activating the Hog1p MAPK, which then regulates expression of osmoresponsive genes (3, 4). The necessity of its transcription factor substrate to retain Hog1p in the nucleus after cellular exposure to osmotic stress suggested that Hog1p might form stable interactions with its substrates, and experiments that identified potential binding partners for Hog1p indicated the same (5, 6). A breakthrough came when chromatin immunoprecipitation (ChIP) experiments showed that in response to osmotic stress, Hog1p is recruited to particular target genes by transcription factors (78). Further work showed that Hog1p not only functions as a kinase at such genes, but also forms an integral component of transcription complexes involved in the recruitment of transcription factors, components of the general transcription machinery, RNA polymerase II (Pol II), and chromatin remodeling/modifying activities (710). This opened up the possibility that terminal kinases might have dual functions: a structural role, by mediating crucial protein-protein interactions within various transcription complexes, and an enzymatic role, by phosphorylating target proteins in such complexes to turn them on or off (1). Indeed, the finding that p38 MAPK—the mammalian homolog of Hog1p—associates with RNA Pol II (9) and also with the enhancer region of muscle-specific genes during myogenic differentiation (11) supports this model. Furthermore, MSK1/2, the kinase that p38 MAPK phosphorylates and activates in mammals, is a nuclear kinase that phosphorylates proteins associated with chromatin, including histone H3 and CREB (3′,5′-cyclic adenosine monophosphate response element-binding protein) (1213). The MSK1/2-related kinase in Drosophila melanogaster, Jil-1, is reported to be chromatin associated (14). Thus, the physical and functional association of Hog1p/p38 MAPK with chromatin is quite well established. What about other gene-regulatory kinases?

Pokholok et al. extend this concept to other such kinases and a greater multitude of genes by combining the ChIP assay with DNA microarrays—so called “ChIP-on-chip” technology. The authors expand the subset of genes known to bind Hog1p in response to osmotic stress from 7 to 39, and they use a mutant yeast strain devoid of Hog1p to show that normal expression of most of these genes requires Hog1p. Binding is highest at the promoter region of these genes but is also detectable to a lesser extent at coding regions. Curiously, only 39 genes were found in this study (an array spanning 85% of the yeast genome), even though there are ∼600 Hog1p-controlled osmoresponsive genes (1517). Thus, perhaps only a subset of Hog1p-regulated genes requires Hog1p to stably bind to chromatin.

Pokholok et al. also show that Fus3p and Kss1p, kinases of the mating pheromone signaling pathway, physically associate with the coding regions of eight pheromone-responsive genes. Strikingly, the scaffold protein Ste5p, which interacts with Fus3p at the cell membrane, occupies the same gene coding regions, which suggests that adaptor proteins might be involved at specific genes in the indirect recruitment of additional factors by kinases. Finally, the authors show that the different catalytic subunits of protein kinase A (Tpk1p and Tpk2p) associate with particular genes. Tpk1p associates with the coding regions of most actively transcribed genes of yeast under normal conditions. Furthermore, the amount of Tpk1p binding to chromatin positively correlates with the transcription rate of the target genes. Loss of Tpk1p binding was observed when particular genes were repressed (increased Tpk1p binding was observed when these genes were activated). Tpk2p was observed largely at the promoter region of genes encoding ribosomal proteins, and this enrichment did not correlate with gene activity.

Kinase recruitment

(First panel) In response to cellular stimuli, some kinases are recruited to target genes. (Second panel) Hog1p is recruited by a transcription factor (Hot1p) to the promoter region of the STL1 osmoresponsive target gene. Hog1p then recruits RNA Pol II and a histone deacetylase complex (Rpd3-Sin3) to control gene expression. (Third and fourth panels) The Tpk2p catalytic subunit of protein kinase A (PKA) is recruited to the promoter region of target genes, whereas the Tpk1p PKA catalytic subunit, Fus3p, and Kss1p are recruited to the coding regions. Although the mechanism and purpose of recruitment of such kinases are not known, they may involve factors that share similar intragenic locations. CMC, chromatin modifying complex; GTM, general transcription machinery; TF, transcription factor; EF, elongation factor.


This study raises several interesting issues. One quantitative aspect that deserves comment is the difference in the relative enrichment of chromatin-associated factors as determined through ChIP-based analysis. The enrichment varies from about 40× for the transcription factor Gcn4p to about 10× or less for the Hog1p and Tpk1p kinases (2). If all other experimental variables during ChIP experiments [such as antibody recovery differences (18)] are accounted for, this variation may indicate that the residence times of these proteins at these locations differ. For example, a stable interaction between a transcription factor and its target DNA is expected to give a higher recovery in ChIP-based analysis of the promoter region of a gene than the transient interaction of RNA Pol II at the coding region of the gene would recover coding sequences. Interpretation of quantitative differences in recovery by ChIP assays is fraught with complications but is unavoidable if we are to extract the full value of these data (18).

Differences in the types of genes and regions of genes with which these different kinases bind may reflect the mechanisms by which they are recruited and/or the functions that they carry out. For example, Hog1p localizes mostly to the promoter region of genes, where we would expect to find specific transcription factors, transcription initiation factors, and promoter-associated coregulatory proteins. This provides an obvious mechanism of protein-protein interaction for the specific recruitment of kinases. Previous findings have shown Hog1p to be recruited by promoter-bound transcription factors and that it functions in the recruitment of RNA Pol II (79). Similarly, Pokholok et al. show good correlation between the genic locations of Tpk2p, the Rap1p transcription factor, and the Esa1p subunit of the NuA4 chromatin-modifying complex (2). Thus, one could speculate that Rap1p recruits Tpk2p and/or Tpk2p aids in the recruitment of the NuA4 complex.

Less obvious with respect to mechanism is the finding of a correlation between the genic distribution of Tpk1p with RNA Pol II and specific histone H3 posttranslational modifications at the coding regions of some genes (2). There is no clear evidence that Tpk1p binds directly to posttranslationally modified histone tails at active genes. One speculation is that RNA Pol II and transcription are involved in the recruitment of Tpk1p to specific genes. This idea is supported by the positive correlation between transcription rate and Tpk1p gene association; if true, it raises the question of how Tpk1p is recruited specifically to particular genes and not to others that are being simultaneously transcribed by RNA Pol II. The presence of Hog1p in the coding regions of specific genes is easier to explain as Hog1p is also recruited to the promoters of these genes, and perhaps enters the coding regions by “piggybacking” with RNA Pol II. Nonetheless, in this important study, Pokholok et al. widen the circumstances in which kinases may be found as a relatively stable constituent of chromatin at both promoter and coding regions of active genes. This may be a more widespread and general phenomenon than is currently appreciated.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
View Abstract

Stay Connected to Science

Navigate This Article