A critical perspective of the diverse roles of O-GlcNAc transferase in chromatin

Abstract

O-linked β-N-Acetylglucosamine (O-GlcNAc) is a posttranslational modification that is catalyzed by O-GlcNAc transferase (Ogt) and found on a plethora of nuclear and cytosolic proteins in animals and plants. Studies in different model organisms revealed that while O-GlcNAc is required for selected processes in Caenorhabditis elegans and Drosophila, it has evolved to become required for cell viability in mice, and this has challenged investigations to identify cellular functions that critically require this modification in mammals. Nevertheless, a principal cellular process that engages O-GlcNAcylation in all of these species is the regulation of gene transcription. Here, we revisit several of the primary experimental observations that led to current models of how O-GlcNAcylation affects gene expression. In particular, we discuss the role of the stable association of Ogt with the transcription factors Hcf1 and Tet, the two main Ogt-interacting proteins in nuclei of mammalian cells. We also critically evaluate the evidence that specific residues on core histones, including serine 112 of histone 2B (H2B-S112), are O-GlcNAcylated in vivo and discuss possible physiological effects of these modifications. Finally, we review our understanding of the role of O-GlcNAcylation in Drosophila, where recent studies suggest that the developmental defects in Ogt mutants are all caused by lack of O-GlcNAcylation of a single transcriptional regulator, the Polycomb repressor protein Polyhomeotic (Ph). Collectively, this reexamination of the experimental evidence suggests that a number of recently propagated models about the role of O-GlcNAcylation in transcriptional control should be treated cautiously.

Fig. 1

Reported O-GlcNAcylated residues in histone proteins mapped onto the nucleosome structure. a Top view of the human nucleosome crystal structure (PDB 3AFA, Tachiwana et al. 2010). The nucleosome is the basic unit of chromatin and is composed of a tetramer of histones H3 and H4 and two H2A-H2B dimers around which 147 bp of DNA is wrapped. N- and C-terminal unstructured extensions are schematized and labeled. Serines or threonines of canonical histones that have been proposed to be O-GlcNAc modified in vivo are highlighted in red (summarized in Table 3). b Lateral view of the nucleosome. Zoomed-in views of candidate O-GlcNAcylated histone residues predicted either to be inaccessible to Ogt in the context of the assembled nucleosome (top row) or accessible because exposed at the nucleosome surface (bottom row)

Fig. 2

Published ChIP-seq profiles of H2B-S112GlcNAc in human cells. Published ChIP-seq profiles (Fujiki et al. 2011) at six genes for which the authors reported significant H2B-S112GlcNAc enrichment near their TSSs in HeLa cells. The profile at GSK3B was reported in Fig. 4e of Fujiki et al (2011); the five other genes were randomly chosen among the first 20 genes at the top of the list showing H2B-S112GlcNAc enrichment in Table S3A (Fujiki et al 2011). The ChIP-seq profiles are centered on each TSS (highlighted in pink) and extend 25 kb upstream and downstream (50 kb windows are shown in total), RefSeq genes are indicated with exons (boxes) and introns (thin lines), and genome coordinates are indicated above (version hg19). For all profiles, we used the same scale on the y-axis. Note that unlike the enrichment of H2B O-GlcNAcylation ChIP signal at GSK3B, the signals at the other positively scored genes are approaching background ChIP signals

Fig. 3

Model of O-GlcNAcylation function in Polycomb repression in Drosophila. a Schematic representation of the fly Ph protein that is O-GlcNAcylated on an S/T-rich region, located close to the C-terminal SAM domain. b Model illustrating how O-GlcNAcylation of Ph allows the formation of ordered SAM-SAM assemblies that are needed to silence Polycomb target genes (top). Ph is bound at PREs as part of the PRC1 complex (other PRC1 subunits are not shown) in both wild-type and Ogt mutant animals. In the absence of O-GlcNAcylation of the S/T stretch, Ph molecules aggregate through their SAM domains (Gambetta and Müller 2014) (bottom). The exact molecular mechanism through which O-GlcNAcylation of the S/T-rich stretch prevents Ph molecules from engaging in non-productive contacts with other SAM domains is not known, but it might involve intramolecular contacts (illustrated here as small loops) between the S/T stretch and the SAM domain that alter SAM conformation in a way that favors aggregation with other SAM domains in a similar conformation (Gambetta and Müller 2014)