Credit: PhotoDisc/Getty Images

Developmental genes in embryonic stem (ES) cells are 'poised' to rapidly induce transcription and gene expression. This poised state includes the presence of histone modifications that are associated with transcriptional activation and transcriptional repression (for example, trimethylation of Lys27 on histone 3 (H3K27), which is propagated by the Polycomb repressive complex 2 (PRC2)). It also involves a poised form of RNA polymerase II (RNAPII) that is phosphorylated at Ser5 (RNAPII(S5P)). Tee et al. now show that ERK2 promotes this poised state in mouse ES cells by potentiating PRC2-mediated trimethylation of H3K27 and phosphorylating RNAPII at Ser5.

ERK1 and ERK2 (ERK1/2) signalling is important for the lineage commitment of stem cells, and evidence suggests that PRC2 might also have a role in this process. Thus, the authors assessed whether ERK1/2 signalling and PRC2 work together during mouse ES cell differentiation. Impairing PRC2 activity by knocking down Jarid2, which regulates the catalytic subunit of PRC2 or the PRC2 subunit Eed, reduced the phosphorylation of ERK1/2 and their upstream kinases MEK1 (also known as MAPKK1) and MEK2 (MEK1/2). In addition, use of a MEK1/2 inhibitor to prevent MEK1/2–ERK1/2 signalling reduced the level of trimethylated H3K27 at developmentally regulated genes. Hence, ERK1/2 and PRC2 have a functional relationship.

The authors next generated ERK1/2-null ES cells in which they observed a reduced accumulation of trimethylated H3K27 and JARID2 at many developmental genes, including the Hoxa gene cluster; interestingly, trimethylated H3K27 and JARID2 levels were restored by the overexpression of ERK2. Therefore, ERK1/2 activation potentiates PRC2-mediated trimethylation of H3K27 at developmental promoters, but how does it do this? Experiments using chromatin immunoprecipitation followed by sequencing showed that ERK2 binds to developmental genes, including the Hoxa and Hoxb gene clusters, at motifs that are characteristic of JARID2 binding. Furthermore, most regions that are enriched with ERK2 binding overlapped with regions of trimethylated H3K27 and JARID2 binding, and trimethylated H3K27 and JARID2 binding were deficient at ERK2-binding regions in ERK1/2-null cells. Thus, in vivo, ERK2 seems to bind to specific repressive genomic loci, in a sequence-specific manner, to promote H3K27 trimethylation.

ERK2 ... promotes PRC2-mediated trimethylation of H3K27 and converts RNAPII to RNAPII(S5P).

Interestingly, the authors observed that, although PRC2 occupancy was decreased at developmental genes in the absence of ERK1/2 signalling, many of these genes remained repressed. This was owing to reduced RNAPII(S5P) levels at the promoters of PRC2 target genes. For example, in ERK1/2-null cells or ES cells treated with a MEK1/2 inhibitor, less RNAPII(S5P) was observed at ERK2 and PRC2 target promoters. The authors also found that ERK2 phosphorylates Ser5 of RNAPII in vitro and that ERK2 interacts with RNAPII in vivo. Furthermore, the overexpression of wild-type, but not kinase-dead, ERK2 in ERK1/2-null cells increased the level of RNAPII(S5P) and restored PRC2 occupancy on developmental genes. Thus, ERK2 phosphorylates RNAPII at Ser5 to help to establish the poised chromatin state at developmental genes in ES cells. Also of note, a subunit of the transcription initiation factor IIH (TFIIH) complex, which is known to phosphorylate RNAPII at Ser5 and Ser7 and hence initiates transcription, was absent from developmental promoters but enriched on highly active genes.

Thus, ERK2 is a key regulator of the poised state of developmental genes, as it promotes PRC2-mediated trimethylation of H3K27 and converts RNAPII to RNAPII(S5P). Indeed, the authors observed that the level of ERK2 at developmental genes decreased after ES cell differentiation.