Perturbations to S phase that lead to replication stress can induce clinically relevant changes in cell state, such as differentiation in leukemia, but the mechanism underlying these changes remains unclear. Recent work indicates that in the hematopoietic system, poised genes, which are epigenetically silenced but inhabit euchromatin regions, are especially vulnerable to derepression during replication stress. Emerging evidence also suggests the increased presence of repressive epigenetic machinery at stalled replication forks. Whether these two observations are mechanistically linked is unknown. Here, we study reactivation of fetal hemoglobin (HbF) by the replication stress inducer hydroxyurea (HU) as a model to understand how poised genes are activated by replication stress. The histone methyltransferase G9a and other components of the G9a/GLP complex are normally required for silencing of the gamma-globin genes HBG1/2, which comprise HbF. We provide evidence that G9a, which dimethylates histone H3 at lysine 9 (H3K9me2), is redistributed from HBG1/2 and other poised genes towards stalled replication forks upon replication stress. Our findings therefore provide a mechanism for HbF reactivation by hydroxyurea, the current first-line treatment for sickle cell disease (SCD).

We developed a fluorescent reporter for HBG1/2 transcription in HUDEP-2 erythroid progenitor cells. Using this system, we first demonstrate that HU induces HBG1/2 at the expense of reduced proliferation. HBG1/2 is also activated by other compounds that induce replication stress, such as aphidicolin and etoposide. Using CUT&RUN, we show that HU causes decreased binding of G9a at the HBG1/2 locus, accompanied by decreased H3K9me2. The gamma-globin genes are examples of poised loci that are actively repressed within a stretch of euchromatin (the globin locus); at many other poised genes throughout the genome, we also observe decreased chromatin association of G9a and H3K9me2 along with increased levels of activating histone marks upon HU treatment. However, using cellular fractionation followed by immunoblotting, we do not observe changes in total cellular or chromatin-bound levels of G9a or H3K9me2. Using proximity ligation assays and pulldown of replication fork-associated proteins through isolation of proteins on nascent DNA (iPOND), we find that replication forks undergoing replication stress accumulate both G9a and H3K9me2. This effect is independent of replication stress signaling through the kinase ATR and is suppressed by a small-molecule degrader of WIZ, a member of the G9a/GLP complex that contains a DNA-binding domain important for G9a recruitment to chromatin. Lastly, activation of HBG1/2 is independent of any effects on differentiation state induced by HU, which we confirm in CD34+ hematopoietic stem and progenitor cells. Because these effects occur without global changes in protein levels for G9a or H3K9me2, our data indicate that stressed replication forks recruit corepressors away from poised genes such as HBG1/2, which in turn facilitates their reactivation.

Our work provides insights into the mechanism of action of hydroxyurea, which remains poorly understood despite decades of use for the treatment of SCD. A better understanding of how replication stress alters the corepressor repertoire bound to gamma-globin may open new doors for more effective and less toxic strategies to treat SCD. More broadly, we provide a model for how replication stress can lead to epigenetic changes that facilitate lineage-specific changes in cell state.

This content is only available as a PDF.
2025
Sign in via your Institution