Transcriptional Control of Gene Expression By O-Glcnacylation during Erythropoiesis

Understanding the pathophysiological defects during erythropoiesis resulting in sickle cell disease, thalassemias, and other hemoglobinopathies is critical to the discovery of targeted therapies or cures. An understudied aspect of these disorders is how post-translational modifications (PTMs) affect gene transcription during erythroid development. We established that O-GlcNAcylation is essential for erythropoiesis. O-GlcNAcylation regulates numerous cellular functions including stress response, transcription, and cell cycle progression. O-GlcNAc is a single O-linked b-N-acetyl-D-glucosamine moiety added to serine/threonine amino acids of nuclear, cytoplasmic, and mitochondrial proteins. O-GlcNAc transferase (OGT), which adds the modification, and O-GlcNAcase (OGA), which removes the modification, are responsible for the dynamic processing of the PTM.

Our published data showed that GATA-1 interacts with OGT and OGA and delivers them to GATA-1-regulated genes. Perturbation of this process resulted in erythroid defects, most strikingly, a shift in commitment towards other hematopoietic lineages. We hypothesize that at the onset of erythroid lineage commitment, GATA-1 functions as an adaptor protein to deliver these enzymes to erythroid-specific cis-regulatory DNA elements (CREs), where the O-GlcNAc status of bound protein complexes is modified to direct transcriptional networks necessary for normal erythroid development. More recently, we developed erythroid-specific OGT conditional knockout mice that show OGT is essential for fetal definitive erythropoiesis, but not for primitive erythropoiesis, although primitive erythrocytes exhibit hallmarks of ineffective erythropoiesis.

Our previous work identified a GATA-1-FOG-1-NuRD repressor that silences γ-globin by binding GATA sites located at -566 or -567 relative to the ^Aγ-globin or ^Gγ-globin transcription start sites, respectively. O-GlcNAc plays a role in formation of this repressor complex. OGT and OGA interact with this repressor complex. O-GlcNAcylation of the NuRD protein CHD4 modulates the assembly of the NuRD complex. Removal of this PTM by OGA activates g-globin gene expression. We hypothesize that CHD4 O-GlcNAcylation promotes CHD4 interaction with a variety of proteins to form functionally different multi-protein complexes. To identify novel O-GlcNAcylated CHD4 interacting proteins, we treated a gene-edited erythroleukemia cell line (UAP1:F383G K562) with a sugar analogue containing a photo-crosslinkable diazirine group (GlcNDAz). OGT adds the sugar analogue to proteins. After crosslinking, we used CHD4 immunoprecipitation (IP) coupled with mass spectrometry to identify CHD4-interacting proteins. O-GlcNDAzylated CHD4 was associated with proteins involved in gene regulation. Following induction of terminal differentiation, we identified >150 novel interacting proteins. GlcNDAz CHD4 interacted preferentially with proteins in the mini-chromosome complex (MCM), mitotic spindle, and with specific NuRD complex members. Using cell cycle-synchronized HeLa cells, we validated an S phase interaction of CHD4 with MCM7 and a M phase interaction with the mitotic spindle. These data reveal novel functions for CHD4 and suggest these they are preferentially generated by O-GlcNAcylation of CHD4.

Manipulation of O-GlcNAcylation at single unique CREs has been technologically challenging. We solved this problem by developing novel CRISPR-based gene targeting tools to probe the function of O-GlcNAc at CREs. A catalytically dead Cas9 endonuclease (dCas9) was fused to OGT or OGA to ascertain their function at GATA-1 regulated genes in K562 cells. Negative controls were generated by mutating a catalytic residue of OGT (H558F) or OGA (D174A). The CBP/p300 acetyltransferase core was fused to dCas9 as a positive control. We targeted the γ-globin promoters using single guide RNAs (sgRNAs) that do not disrupt CREs in K562 cells. When OGA-dCas9 was targeted to the -1 to -223 bp region relative to the mRNA start site of the γ-globin promoters, we observed a significant increase in γ-globin expression. OGT-dCas9 had a significant repressive effect when targeted to the same regions. We validated these results by performing chromatin immunoprecipitation (ChIP) and Hit-and-Run ChIP to verify occupancy of the targeted regions by our constructs and the presence or absence of O-GlcNAc.