O-GlcNAcylation and 5-Methylcytosine Oxidation: An Unexpected Association between OGT and TETs

Abstract

Three recent studies, including one in this issue of Molecular Cell, document unexpected physical and functional interactions between two unrelated enzymes: OGT, which transfers O-GlcNAc to serine/threonine residues of numerous cellular proteins, and TET-family dioxygenases, which successively oxidize 5-methylcytosine in DNA.

“Epigenetic” modifications of DNA and histones modulate gene expression. DNA methyltransferases generate 5mC, which is successively oxidized by TET proteins to yield 5hmC, 5fC, and 5caC (Tahiliani et al., 2009; Ito et al., 2011; He et al., 2011) (Figure 1A). OGT (O-linked N-acetylglucosamine [O-GlcNAc] transferase) is a recent addition to this growing list. OGT adds a single GlcNAc moiety to serine/threonine residues in numerous cellular proteins including histones (Hart et al., 2011) (Figure 1B). OGT and OGA (O-GlcNAcase), the enzyme that opposes OGT activity by removing O-GlcNAc moieties, are both encoded by single copy genes and are essential for embryonic development in mammals (Hart et al., 2011). Three groups have now demonstrated physical and functional interactions between OGT and TET proteins (Chen et al., 2013; Deplus et al., 2013; Vella et al., 2013), uncovering a novel connection between these unrelated enzymes.

Figure 1. TET-OGT Complexes Modulate Chromatin Conformation and Gene Expression at CpG-Rich TSSs in Mouse ESCs.

(A) TET proteins catalyze sequential oxidation of 5mC to 5hmC, 5fC, and 5caC, potentially facilitating DNA demethylation. Ineffective recognition of the oxidized forms of 5mC by the maintenance DNA methyltransferase Dnmt1 could facilitate passive DNA demethylation, whereas TDG-dependent excision of 5fC and 5caC could mediate active DNA demethylation.

(B) OGT and OGA act as terminal effectors of the hexosamine biosynthesis pathway by modulating O-GlcNAcylation of serine/threonine residues on proteins.

(C) Left panel: In mESCs, Tet1-Ogt complexes are predominantly recruited to CpG-rich TSSs that correspond to transcriptionally active genes. Tet1 and Tet2 are depicted as tightly and loosely bound to chromatin, respectively; both interact with Ogt. Hcfc1 binds Ogt directly, but whether it is part of the OGT-TET complex is not clear. Middle panel: Approximately 70% of the sites that corecruit Tet1 and Ogt also recruit mSin3a, which cooperates with Tet1 to maintain transcriptional repression. Right panel: TET1 also binds many sites that are not co-occupied by OGT, many of them potentially distal cis-regulatory regions (e.g., enhancers). The TET1 partners at these sites are unknown but could include transcription factors and other transcriptional regulators (Stadler et al., 2011).

The first link between O-GlcNAcylation and chromatin regulation emerged from studies on Drosophila Supersexcombs (Sxc). Sxc was originally annotated as a Polycomb group (PcG) protein, but genetic mapping of sxc mutations identified Sxc as the Drosophila homolog of OGT (Gambetta et al., 2009). The authors showed that O-GlcNAcylated proteins were localized at polycomb response elements and that Sxc O-GlcNAcylated the PcG protein Polyhomeotic. The link between OGT and Polycomb proteins was confirmed in mouse embryonic stem cells (mESCs) (Myers et al., 2011): lack of Eed or Suz12 (components of polycomb repressor complex 2) was associated with substantially decreased levels of Ogt and O-GlcNAc-modified proteins in the nucleus. This study also showed that Tet1 was O-GlcNAcylated in mESCs.

Tet1 and Tet2 have now been identified as two of the most abundant proteins present after affinity purification of biotin-tagged Ogt from mESCs (Vella et al., 2013). They form independent high-molecular-weight nuclear complexes with Ogt, and both are O-GlcNAcylated in mESCs (Figure 1C). Reciprocally, OGT was identified as an abundant protein in affinity purifications of TET2 and TET3 ectopically overexpressed in HEK293T cells (Chen et al., 2013; Deplus et al., 2013).

Remarkably, Ogt is recruited to chromatin in mESCs almost exclusively through its interaction with Tet1 (Vella et al., 2013). In ChIP-seq experiments, effectively all Ogt binding sites in mESCs were co-occupied by Tet1. Tet1 depletion resulted in a striking reduction of Ogt recruitment; conversely, acute depletion of Ogt decreased Tet1 occupancy at the co-occupied sites (Vella et al., 2013).

The three studies differ in a major point: the relative roles of Tet1 and Tet2 in modulating Ogt. Vella et al. clearly document Tet1-Ogt interactions, but neither Chen et al. nor Deplus et al. found a strong interaction between TET1 and OGT. The reason for this discrepancy is unclear, but could reflect the experimental conditions used: Vella et al. probed Tet-Ogt interactions in mESCs that expressed low levels of tagged Ogt, whereas Chen et al. and Deplus et al. overexpressed TET proteins at relatively high levels in cell lines. Chen et al. report that Tet2 depletion in mESCs impaired OGT chromatin association and O-GlcNAcylation of core histones, but Vella et al. find no such effect: instead they show that Tet2 is localized in the nucleus, but is loosely chromatin associated compared to Tet1.

If the identity of the TET protein is ignored, the three studies come to relatively similar conclusions. TET, OGT, and the H2B Ser112 O-GlcNAc mark are colocalized in the genome, largely at transcription start sites (TSS) that contain CpGislands and correspond to promoters of transcriptionally active genes. TET-OGT co-occupied promoters show high H3K4me3, and the associated genes are expressed at increased levels compared to all genes; conversely, depletion of TET2, TET3, or OGT decreased gene expression (Chen et al., 2013; Deplus et al., 2013). Together these findings suggest that TET-OGT interaction is related to positive regulation of gene expression.

In summary, by documenting TET-OGT interactions, the three studies have established a direct link between two previously unrelated enzymatic activities in transcriptional regulation. Several important questions remain to be addressed. Does O-GlcNAcylation of TET proteins influence their enzymatic activity? Are TET proteins regulated differently at genomic sites co-occupied or not by OGT? Does the presence of TET-OGT complexes at genomic regions result in a consistent and meaningful change in chromatin structure and/or transcriptional activity? Potentially, other proteins in Tet-Ogt complexes (such as Hcfc1 and Sin3a; Vella et al., 2013) might alter Tet-Ogt activity in a context-dependent manner (Figure 1C). Finally, what is the function of histone O-GlcNAcylation? All four core histones can be O-GlcNAcylated, but O-GlcNAcylation of H2B at Ser112 is the only O-GlcNAc mark that has been studied to some extent: its presence is thought to facilitate the ubiquitination of H2B at Lys120 (Fujiki et al., 2011). Akin to histone lysine methylation that correlates with transcriptional activation or gene silencing depending on the lysine in question, O-GlcNAcylation of distinct histone residues might have distinct consequences for transcriptional activity. The coming years should provide some major insights.