O-GlcNAc and the Epigenetic Regulation of Gene Expression*

O-GlcNAcylation is an abundant nutrient-driven modification linked to cellular signaling and regulation of gene expression. Utilizing precursors derived from metabolic flux, O-GlcNAc functions as a homeostatic regulator. The enzymes of O-GlcNAc cycling, OGT and O-GlcNAcase, act in mitochondria, the cytoplasm, and the nucleus in association with epigenetic “writers” and “erasers” of the histone code. Both O-GlcNAc and O-phosphate modify repeats within the RNA polymerase II C-terminal domain (CTD). By communicating with the histone and CTD codes, O-GlcNAc cycling provides a link between cellular metabolic status and the epigenetic machinery. Thus, O-GlcNAcylation is poised to influence trans-generational epigenetic inheritance.

Intermediary metabolism, the process by which nutrients are converted into cellular biomass, is an interwoven network of biochemical reactions allowing reproduction, development, and response to the environment. The intermediary metabolic network is highly conserved and includes every cellular process ranging from DNA replication to transcription and translation to enzyme regulation. Epigenetics, the study of how genes may alter phenotypes beyond their ability to genetically encode information, is ultimately linked to intermediary metabolism (1). Many enzymes that participate in epigenetic gene regulation depend upon co-substrates produced by cellular metabolism, thus providing a potential link between metabolism and gene regulation (2). Mitochondria, key players in these metabolic inter-conversions, exhibit a pattern of cytoplasmic inheritance distinct from Mendelian inheritance of genes encoded in the nucleus (3). O-GlcNAcylation is a key integrator of cellular nutritional status and occurs in the nucleus, cytoplasm, and mitochondrion. O-GlcNAc transferase (OGT) 3 utilizes UDP-GlcNAc to catalyze the addition of O-GlcNAc to target proteins. UDP-GlcNAc is the end product of the hexosamine biosynthetic pathway (HSP), a series of enzymatic reactions requiring key metabolites, including glucose, glutamine, ATP, and acetyl-CoA (4). The nutrient-derived precursors render the synthesis of UDP-GlcNAc and subsequent O-GlcNAc addition by OGT nutrient-responsive. The O-GlcNAcase (OGA; MGEA5) removes the O-GlcNAc modification, and evidence suggests that its transcription and activity is highly regulated. Both enzymes of O-GlcNAc cycling contain domains that allow them to bind to epigenetic modifiers (5,6). As illustrated in Fig.  1, UDP-GlcNAc is central both to the formation of O-GlcNAc but also to the synthesis of membrane and secretory glycoproteins that perform essential roles in extracellular signaling. The intracellular O-GlcNAc modification plays a role in signaling, leading to growth and apoptosis (7)(8)(9), metabolism (10,11), and the cell cycle (12)(13)(14). In addition, O-GlcNAc has been implicated in translation (15), circadian rhythm (16), the establishment of molecular memory in neurons (17), and calmodulinkinase signaling (16).
The focus of this review is the role of O-GlcNAc in transcription and epigenetics (5, 13, 18 -20). As shown in Fig. 1A, O-GlcNAc modifies transcription factors (21), epigenetic modulators (5), and RNA polymerase II (18 -20). These targets suggest that O-GlcNAc may be poised to be a metabolically sensitive effector of gene expression and may function as an epigenetic modifier (Fig. 1B). Fig. 1B highlights the metabolic inter-conversions required for the synthesis of UDP-GlcNAc that involve a dynamic interplay between cytoplasm, mitochondria, and the nucleus. Second, the figure illustrates dynamic modification of proteins both in the nucleus and in mitochondria by differentially localized forms of OGT. Finally, we describe those epigenetic effectors, including transcription factors, histones, TET1-2, and RNA polymerase II, that are modified by O-GlcNAc to influence their function. These diverse functions of O-GlcNAc cycling suggest that the pathway plays a key role in human diseases of aging such as metabolic syndrome, diabetes, Alzheimer disease, heart disease, and cancer (16,22,23). Indeed, the O-GlcNAc gene is a susceptibility locus for obesity and type II diabetes (24). Growing evidence also suggests that OGT may play a key role in the intrauterine environment (5,22,25,26). Thus, the impact of O-GlcNAc cycling in trans-generational epigenetic phenomena is an emerging area of interest.

The Enzymes of O-GlcNAc Cycling
O-GlcNAc is added by the enzyme OGT to serine and threonine of target proteins in the cytoplasm, nucleus, and mitochondrion (27,28). The single mammalian OGT gene is located on the human X chromosome near the Xist locus and encodes multiple alternatively spliced isoforms of OGT targeted to these different intracellular locations (27,28). The three major isoforms are termed mOGT (mitochondrial OGT), ncOGT (nucleocytoplasmic OGT), and sOGT (short OGT) (Fig. 1B). The mOGT isoform is targeted to the mitochondrial matrix and is proapoptotic (28,29). The sOGT isoform, present in the nucleus and cytoplasm, is antiapoptotic (28,30) and may inhibit the action of the other isoforms (29,31). The longest isoform, ncOGT, is present in the nucleus and cytoplasm and contains 13 complete tetratricopeptide repeats (TPR) (32)(33)(34)(35)(36). The mOGT and sOGT variants contain 9 and 2 TPR, respectively (37). The TPR mediate interaction with a large number of effector proteins that target or regulate OGT (38,39). It is likely that the TPR also serve a scaffolding function, distinct from any role in enzymatic activity (38). All three isoforms share a common catalytic domain with a deep UDP-GlcNAc binding pocket and a groove for target peptide binding (40).
The O-GlcNAcase is encoded by a single gene on human chromosome 10 (26). Two major O-GlcNAcase isoforms are produced by alternative splicing of the O-GlcNAcase gene. One of these contains a domain with similarities to a histone acetyltransferase domain but lacking the critical residues for catalytic activity. This has been termed a pseudoHAT domain (41). The other major O-GlcNAcase isoform has a 14-amino extension that serves to target it to lipid droplets. This O-GlcNAcase isoform is involved in the remodeling of lipid droplet surface proteins by local activation of proteasomes on the lipid droplet surface (38,42). This study suggested a nexus between the regulation of lipid storage and hexosamine signaling through O-GlcNAc cycling (42).

The Synthesis of UDP-GlcNAc as a Metabolic Sensor Linked to O-GlcNAcylation
The levels of both enzyme substrate and product are critical regulators of the activities of most of the enzymes of intermediary metabolism. This is in contrast to the enzymes mediating intracellular signaling, which are not typically regulated in this way. Protein kinases, which use ATP as a substrate, recognize and phosphorylate target proteins, but with the sole exception of the AMP-regulated protein AMPK (43,44), kinases do not respond to ATP levels. Protein kinases bind to ATP with an affinity that is saturated by normal levels of intracellular ATP. This effectively isolates kinase signaling from normal intracellular metabolism. O-GlcNAcylation appears to be regulated quite differently. The synthesis of UDP-GlcNAc is a tightly regulated process requiring precursors derived from glucose, glutamine, acetyl-CoA, ATP, and uridine (4) (Fig. 1B). Precursor levels drive the flux through this pathway, but the overall output is limited by substrate inhibition of the key rate-limiting enzyme glutamine:fructose-6-phosphate amidotransferase (GFAT). UDP-GlcNAc inhibits this enzyme, creating a feedback loop that serves to limit the levels of total UDP-GlcNAc (4). However, the situation is much more complex in a cellular context. UDP-GlcNAc levels are also limited by epimerization to UDP-GalNAc for the synthesis of O-linked glycoproteins in the endomembrane system. Other inter-conversions can occur within nucleotide sugar pools themselves to form CMP-sialic acid. UDP-GlcNAc is transported into the endoplasmic reticulum and Golgi for glycosyltransferase reactions for complex oligosaccharide assembly in those organelles (45). In certain cell types, this pool of UDP-GlcNAc may be released extracellularly, where it may activate the broadly distributed P2X and P2Y purinergic receptors (46). Taking into account the known volumes occupied by intracellular organelles, enrichment of UDP-GlcNAc in the endomembrane system (ϳ30-fold) can result in cytoplasmic levels that are in the range expected from the known K m values for OGT (ϳ2-5 M) and UDP-GlcNAc transporters (ϳ7 M) (47). Finally, the mOGT isoform com- O-GlcNAc has been implicated in diverse processes as indicated by the arrows. These processes include metabolism, growth, apoptosis, the cell cycle, transcription, translation, the circadian clock, and the establishment of molecular memory in neurons. B, the role of hexosamine synthesis in mediating communication between the mitochondrion and nucleus. Metabolic precursors required for UDP-GlcNAc synthesis, glucose, acetyl-CoA, glutamine, uridine, and ATP are generated by the coordinated activities of cytoplasmic and mitochondrial enzyme complexes. The mitochondrion is the site of fatty acid oxidation to produce acetyl-CoA and the tricarboxylic acid (TCA) cycle utilizing glucose-derived precursors for ATP production. Glutamine is utilized by the action of glutaminase to generate mitochondrial glutamate for entry into the TCA cycle. These interconnected metabolites communicate to the epigenetic machinery by providing precursors for epigenetic modifications including O-GlcNAc addition and histone acetylation. O-GlcNAc addition is mediated by a mitochondrial variant (mOGT), a shorter isoform (sOGT), and the nuclear/cytoplasmic ncOGT. The sOGT variant is present in both cytoplasm and nucleus, although for illustrative purposes, it is shown only the cytoplasm. P indicates phosphorylated residues. petes with ncOGT in the nucleus for UDP-GlcNAc (28) (Fig.  1B). Although total UDP-GlcNAc levels have been estimated to be as high as 1 mM, the concentrations of the cytoplasmic and mitochondrial pool are likely to be much lower. Utilization of UDP-GlcNAc for hyaluronan synthesis has been proposed to limit cytoplasmic O-GlcNAc levels (48). Thus, the pool of UDP-GlcNAc available for O-GlcNAcylation is influenced by many factors including metabolic flux. At the level of metabolism, other reactions contribute to, and compete for, precursors to the hexosamine biosynthetic pathway. Glutamine is one of these key metabolites. Glutamine can be converted to glutamate in mitochondria for the production of ␣-ketoglutarate and entry into the tricarboxylic acid cycle (2). In addition, the regulated entry of glutamine may be influenced by the synthesis of UDP-GlcNAc (2). Another precursor of UDP-GlcNAc, acetyl-CoA, is primarily produced by the conversion of pyruvate into acetyl-CoA via the mitochondrial pyruvate dehydrogenase complex during the oxidation of glucose and the ␤-oxidation of fatty acids (49). There is growing evidence in both yeast and mammalian cells that acetyl-CoA levels can fluctuate to influence gene expression (44). These many factors influencing the levels of UDP-GlcNAc are summarized in Fig. 1B.  (Fig. 2) may fluctuate in response to nutrients or such cues as circadian period. These metabolites either serve as direct donors (S-adenosylmethionine (AdoMet), acetyl-CoA) or contribute to the synthesis of other key metabolites (ATP, NAD, ␤-hydroxybutyrate, and ␣-ketoglutarate) (49). The metabolites then act through key enzymes such as kinases, lysine methyltransferases, histone acetyltransferases, or O-GlcNAc transferase to modify chromatin-associated factors. The DNA methyltransferases (DNMTs) are dependent upon the methyl donor AdoMet to modify cytosine residues of DNA. The Krebs cycle intermediate ␣-ketoglutarate modulates the activity of the TET (Ten-eleven translocation) proteins (see below). These key enzymes of addition are often referred to as "writers" of an epigenetic code (49). In turn, the modifications are removed by phosphatases, lysine demethylases, histone deacetylases, and O-GlcNAcase. DNA demethylation is thought to occur both passively and through the action of the TET enzymes. Thus, the enzymes catalyzing removal are frequently termed "erasers" (49). The writers and erasers act on effector molecules that serve to maintain transcriptional homeostasis: DNA, histones, transcription factors, and RNA polymerase II. We note that O-GlcNAc cycling plays a key role in most of these phenomena that maintain homeostasis. In the following sections, we will discuss the potential role of O-GlcNAc on these effectors including the "histone code," DNA methylation, DNA demethylation, and the RNA polymerase II "CTD code."

O-GlcNAc and Epigenetic Regulation of Transcription by the Histone Code
As illustrated in Figs. 1 and 2, the enzymes of O-GlcNAc cycling have also been shown to physically and genetically interact with a number of transcriptional regulators with established roles in epigenetics (5,23,26). OGT modifies many kinases and is thought to interact with numerous kinase signaling cascades both at the level of kinase regulation and through competition for acceptor serine and threonine residues (51,52). O-GlcNAc has been shown to directly modify histones including the mitosis-specific H3 Ser-10 modification, which is conditionally phosphorylated (6,12,14,53). Other sites of O-GlcNAcylation occur on histones H2A and H2B that may be linked to histone exchange (45,53). It is noteworthy that the known sites of O-GlcNAcylation of histones are sites distinct from H3 and H4 tail modifications more directly linked to the regulation of gene expression. Perhaps the best established role for OGT in transcriptional regulation is involvement in polycomb repression (26,54,55). In Drosophila, OGT is allelic with Sxc, a gene initially characterized as a homeotic mutant (56). OGT is essential for polycomb repression in Drosophila (55) and may play a similar role in mammals (5,26,57). Polycomb repression is linked to the trimethylation of histone H3 at Lys-27 (58). Polycomb is best known for its role in regulating the Hox (homeotic) genes in Drosophila. The Hox gene clusters regulate the development of the anterior/posterior axis in Drosophila and carry out conserved functions in other metazoans from Caenorhabditis elegans to humans. In mammals, polycomb repression is involved in maintenance of pluripotency, differentiation, and imprinting (26,54,55). There is also evidence that OGT may interact with members of the trithorax group of epigenetic regulators (reviewed in Refs. 5 and 26). The trithorax group is associated with "activating histone modifications" such as histone H3 Lys-4 and Lys-36 methylation (59,60) and may involve O-GlcNAcylation as part of the regulatory network (26). In general terms, the trithorax group acts as an antirepressor that counters the action of the polycomb repressive complex. In addition to its role in polycomb repression, OGT has been shown to interact with mSin3A, a transcriptional repression complex associated with histone deacetylases (HDAC1, HDAC2) (61). The HDACs remove the acetyl residue laid down by the numerous histone acetyltransferases, which use acetyl-CoA as a precursor (1,5). The OGT-mSin3A complex is also found in association with HCF-1, a protein that is cleaved by OGT in a process linked to mitotic cell cycle progression (45,(62)(63)(64). OGT is also a central hub in the Oct4 network and associates with other factors maintaining stem cell pluripotency (26). O-GlcNAc modifies many, if not most, RNA polymerase II transcription factors (21). In addition to these numerous interactions, OGT binds to the TET class of DNA cytosine demethylases that may serve as 5-methylcytosine erasers (65)(66)(67). TET proteins were first discovered through their involvement in myeloid leukemia where the TET1 gene, located on chromosome 10, translocated with the H3K4 histone methyltransferase MLL gene on chromosome 11 (68). The TET proteins are a member of a family of dioxygenases that are dependent upon both ␣-ketoglutarate and iron(II) for their activities (38,67,69,70). Nutrient-dependent O-GlcNAcylation does not seem to play a direct role in regulating TET demethylation activity but may influence its nuclear residency (71).
Taken together, the interactions of these effector molecules (Fig. 2) with the enzymes of O-GlcNAc cycling and other mediators of epigenetic programming suggest that O-GlcNAcylation integrates metabolic information. Small molecule metabolites are central players in the regulation of these enzymes. These metabolites show complex interactions and inter-conversions providing links between diverse metabolic pathways. The changes in these metabolites from environmental influences such as circadian rhythms, nutritional flux, infection, and stress may contribute to the maintenance of homeostasis depicted in Fig. 2. From this figure, it is clear that O-GlcNAc is a central player in these processes. Although all of the effectors depicted in Fig. 2 are potentially important, we have chosen to focus this review on the impact of RNA pol II O-GlcNAcylation. pol II is downstream of all of the complex epigenetic regulatory paradigms described above in which O-GlcNAc has been implicated.

RNA Polymerase II O-GlcNAcylation and Nutrient Sensing: The Differing Forms of RNA Polymerase II
In humans, the pol II CTD consists of 52 heptad repeats of predominantly YSPTSPS. This "consensus repeat" is an accurate description of the yeast CTD and the first 26 repeats of the human CTD. However, this sequence diverges in human heptads 27-52. Notably, there are eight lysine residues (replacing serine 7), one arginine residue, and conservative switches between serine and threonine residues and asparagine and glutamic acids; all are at position 7 of the CTD repeat (72)(73)(74).
RNA polymerase II exists in many different forms dictated by the post-translational modifications on its C-terminal domain (CTD). These modifications are summarized in Fig. 3. The literature refers predominantly to two different species, pol IIA and pol IIO. pol IIA is thought to be the species required for promoter binding and initiation (75,76). It is unclear whether this species is completely unmodified or whether it is more accurately considered hypo-phosphorylated. In contrast, pol IIO is thought to be heavily phosphorylated. A third distinct species of pol II is defined by the O-GlcNAcylation of the CTD (pol II␥) (19,20). pol II␥ was found within the pol IIA population, and in fact, because of the mixing of pol IIA and pol II␥, it has never been clear whether pol IIA or pol II␥ is the initiationspecific species of pol II. Regardless, upon initiation, pol II is phosphorylated, first at serines 5 and 7, and subsequently at serine 2 (72)(73)(74).

CTD Modifications and Enzymes
The enzymes responsible for RNA pol II phosphorylation are quite numerous and have been discussed in detail elsewhere, especially in comparing yeast and human transcription systems (72,77). CDK7 of the TFIIH complex phosphorylates serine 5 and 7 residues. The phosphorylation of serine 2 is much less clear; CDK8, -9, -12, and -13 and Brd4 have all been suggested as serine 2 kinases (72,(77)(78)(79). It is clear that in vitro, these kinases can modify the CTD. However, in vivo, only CDK9, -12, and -13 have definitively been shown to be CTD kinases. Additionally, both threonine 4 and tyrosine 1 can be phosphorylated (80 -82). Two kinases are capable of modifying Thr-4, whereas c-Abl is thought to be the Tyr-1 kinase (81, 83-86). There are also several CTD phosphatases: FCP1 (serine 2), SCP (serine 5), FIGURE 3. Schematic representation of some CTD post-translational modifications. A, the YSPTSPS CTD consensus repeat is depicted along with the positions of O-GlcNAcylated (G) and phosphorylated residues (P). In addition, some of the proteins that interact with various combinations of the phosphorylated residues are shown. This is not an all-inclusive depiction; other associated factors including capping enzyme and splicing and polyadenylation factors. The reader is referred to several excellent reviews for a thorough discussion of the CTD modifications (72)(73)(74). B, the methylation of arginine in CTD repeat 31 (by CARM1 (89)). C, one of eight lysine residues that are acetylated. and RPAP2 (serine 5) (72). Pin1 (Ess1 in yeast) is unique among the CTD modification enzymes as it is a prolyl isomerase, converting proline from a predominantly trans-conformation to a cis-conformation that alters the substrate specificity of the CTD (for example, FCP1 requires Pin1 activity for CTD substrate binding) (87). Acetylation by p300 occurs on the lysine residues in the C-terminal half of the CTD (88). Lastly, the one arginine residue in the human CTD is a substrate for CARM1 arginine methyltransferase (89).
CTD O-GlcNAcylation was originally found on calf thymus RNA pol II and more recently on human pol II (19,20). Edman degradation of calf thymus pol II indicated that both Thr-4 and Ser-5 of the CTD are O-GlcNAcylated (19), whereas serine to alanine substitutions (20) and mass spectroscopy indicated that Ser-5 was O-GlcNAcylated. Additionally, O-GlcNAcylation of Ser-5 can block subsequent phosphorylation, suggesting that these are mutually exclusive events (18,20); these data argued that pol II␥ exists either before pol IIO or afterward (i.e. it is a pre-initiation species or a post-elongation species). Focus centered on pol II␥ acting before initiation after finding that OGT and OGA catalytic activity are necessary for transcription in vitro and that inhibition of these enzymes blocks transcription during pre-initiation complex assembly (20). OGT can be detected at promoters and immunoprecipitated with pol II (20). In vivo, shRNA inhibition of OGT decreased pol II occupancy at several B-cell-specific promoters (20), and both OGT and O-GlcNAc localize almost exclusively at transcription start sites (65). Similar results were obtained in C. elegans ChIP-chip analysis where OGT mutants are viable, thus providing robust controls in the ChIP-chip and ChIP-Seq experiments (13).

CTD Code and Function
The concept of a code on the CTD comprising the various PTMs was first proposed by Buratowski (90). In general, there are three ways that a code might manifest itself. The first is simply a binary code, consisting of either a modified or unmodified residue. The second is a combinatorial code, where two different residues are modified, either with the same or with different PTMs. The third is a sequential code, where one PTM establishes a second modification. There is evidence of all three (phospho-Ser-5; phospho-Ser-2; phospho-Ser-5/ phospho-Ser-7; phospho-Thr-4; arginine-methyl; phospho-Tyr-1; lysine-acetyl) (72)(73)(74)88). It is clear that specific combinations of phosphoserines in the CTD are binding sites for factors, offering definitive evidence of some form of combinatorial code (91,92).
In contrast, there is little evidence that proteins have an "O-GlcNAc binding domain" mediating an interaction with O-GlcNAcylated proteins. In the absence of such evidence, it is necessary to consider other functions of an O-GlcNAcylated pol II CTD. O-GlcNAcylation of the CTD is unlikely to be essential because C. elegans mutants lacking OGT are viable and fertile (13). Thus, CTD O-GlcNAcylation may serve a more regulatory function to sterically prevent or promote factors binding to the CTD. Secondly, pol II␥ may prevent premature, aberrant CTD phosphorylation, for which there is some evidence (18,20). Alternatively, the hydrolysis of UDP-GlcNAc, which is a high energy donor, may make free energy contributions (similar to ATP hydrolysis (96)) to promote kinetic steps during preinitiation complex formation. Lastly, because both O-GlcNAcylation and phosphorylation of CTD residue serine 5 occur in a mutually exclusive manner, and phospho-Ser-5 is a necessary prerequisite for establishing a paused pol II, we also suggest that regulation of O-GlcNAc removal might regulate the formation of paused pol II downstream of the transcriptional start site by regulating the level of phospho-Ser-5 creation. Future work is required to unravel the complex interactions of the CTD with its numerous PTM contributions. The sheer complexity of such an analysis is staggering. If the 52 repeats have 4 modification sites per repeat, there are ϳ208 sites. With the potential for O-GlcNAc, O-phosphate, or unmodified states at each of these, the total number of molecular species possible is 3 208 ϭ ϳ2 ϫ 10 99 !

The CTD as a Paradigm for Nutrient Sensing via O-GlcNAc
O-GlcNAc levels on proteins have been suggested to reflect the nutrient state of the cell (6,23,50). It is also critically important to consider the proteins that are O-GlcNAcylated and not just UDP-GlcNAc levels. For an O-GlcNAcylated protein to be considered a nutrient sensor, it should satisfy several criteria. The first is that the protein-O-GlcNAc levels must reflect changing nutrient conditions. For example, nuclear pore proteins are not likely to be nutrient sensors as their O-GlcNAc levels are very stable (49). Secondly, the O-GlcNAc levels of a protein must be dynamically regulated by either OGT and/or OGA. Finally, these changing levels of O-GlcNAc should alter the functional output of the protein in question. Examples of likely nutrient sensor proteins (NSPs) include CRTC2, RelA, and c-Rel NF-B family members, and FoxoO1 (21,97,98).

CTD as a Scaffold for Nutrient-dependent Histone Modification Enzymes
The CTD is modified by O-GlcNAcylation, phosphorylation, and acetylation that are all dependent on intracellular nutrientsensitive pools. However, we would like to suggest a second function of the CTD; it serves as a nexus for the nutrient-dependent histone methylation enzymes. These enzymes, such as MLL1/2, SET2, and Dot1, bind different phospho-CTD marks and deposit methyl groups onto Lys-4, Lys-36, and Lys-79 residues of the H3 histone tail, respectively (Fig. 3) (95). These enzymes are dependent on the AdoMet methyl donor (this nutrient dependence of the histone methyltransferases and demethylases has been documented by others (49,99)). O-GlcNAcylation is part of the information content of the CTD. The interplay between the CTD and histone codes is therefore likely to be coupled; both respond to changing nutrient states, and it is their combined state that has functional consequences. This model is summarized in Fig. 4.

Summary
The O-GlcNAc modification is unique in that it acts at every step in the cascade of effectors implicated in epigenetic regulation (68). It is a critical regulator of cell signaling, mitochondrial function, growth regulation, and apoptosis (5,24,26,100). OGT binds to the TET proteins involved in recognizing methylated CpG islands in DNA important for genomic imprinting (65)(66)(67)69). O-GlcNAc modifies histones (12,53) and participates with the polycomb and trithorax complexes responsible for histone methylation (14,26,55,101). Most RNA pol II-dependent transcription factors are O-GlcNAcylated (21). Finally, the CTD of pol II is O-GlcNAcylated, exhibiting a complex interaction with other CTD post-translational modifications (19,20). We have explored the RNA pol II CTD modification by the enzymes of O-GlcNAc cycling as a paradigm to illustrate how a nutrient-driven post-translational modification may influence a central player in transcriptional regulation. We have examined how this complex network of PTMs may be regulated and shown that the HSP is central to these metabolic interconversions. We have also shown that the enzymes of O-GlcNAc cycling can partner with other epigenetic regulators such as mSin3A, HCF-1, and Oct4. As the tools to examine O-GlcNAcylation become more widely available to the epigenetics community, this expanding field is certain to gain wider acceptance and recognition. The implications of these recent findings are profound and far-reaching. The O-GlcNAc modification provides a tangible molecular mechanism by which nutritional information can be conveyed to influence the readers, writers, and erasers of the genetic and epigenetic codes. Human diseases of aging including cancer, neurodegeneration, and metabolic disease are associated with dysregulation of O-GlcNAcylation. Advancing knowledge in this important arena is likely to fuel interest in targeted therapies and diagnostics. FIGURE 4. Integration of the CTD, histone tails, and the hexosamine biosynthetic pathway. The figure illustrates the numerous connections between nutrient inputs and the CTD, histones, and HSP. Firstly, levels of ACoA, glucose, glutamine, and uridine, as metabolic precursors, may impact synthesis of UDP-GlcNAc by the HBP, which in turn is manifested on the levels of O-GlcNAcylation of the CTD. Additionally, ACoA and AdoMet (SAM) levels likely affect the levels of methylation and acetylation of both the CTD and the histone tails, whereas UDP-GlcNAc levels may be reflected in histone H4 O-GlcNAcylation (53). Of course, any perturbation of glucose and glycogen levels possibly impinges on the extent of CTD phosphorylation (P). Finally, the differing CTD modification states then likely affect, for example, the extent of H3K4 methylation via recruitment of histone methyltransferases (HMTs) to phosphorylated CTD (PCTD) residues. These points illustrate the interconnectivity between RNA pol II, histone tails and their epigenetics, and the HBP and other nutrient sources. pY1, phospho-Tyr-1; pS1, phospho-Ser-2; pT4, phospho-Thr-4; pS5, phospho-Ser-5; pS7, phospho-Ser-7.