O-Glycosylation of Nuclear and Cytosolic Proteins

Despite the long held view that protein glycosylation occurs exclusively on extracellular or lumenal polypeptides (1), it is clear that many nuclear and cytoplasmic proteins are multiply O-glycosylated at specific serine or threonine hydroxyl groups by single b-N-acetylglucosamine moieties (O-GlcNAc) (2–4). O-GlcNAc modification is common to nearly all eukaryotes, including filamentous fungi, plants, animals, and animal parasites, as well as viruses that infect eukaryotes. Mounting evidence suggests a direct role for O-GlcNAc in cellular regulation. For example, the a-toxin of the gangrene causing bacteria Clostridium novyi is an O-GlcNAc transferase that exerts its toxic effects by the addition of O-GlcNAc to proteins in the Rho subfamily (5). Thus, the disruption of normal O-GlcNAc-regulated pathways may be responsible for the pathology of some bacteria. Moreover, disruption of the gene for OGlcNAc transferase demonstrates that O-GlcNAc modification is essential for life, even at the single cell level (6). O-GlcNAc appears to be both as abundant and as dynamic as protein phosphorylation. In several documented instances, phosphorylation and O-GlcNAc modification are reciprocal, occurring at the same or adjacent hydroxyl moieties (7, 8). Furthermore, all O-GlcNAc-modified proteins identified to date also occur as phosphorylated proteins. Nevertheless, the interrelationship between Ser/Thr O-GlcNAc modification and O-phosphorylation appears to be complex. Although there are examples of mutually exclusive O-GlcNAc modification and O-phosphorylation, it is likely that all possible combinations are represented in the complex environment of a eukaryotic cell (Fig. 1). The specific addition and removal of these two differentially regulated post-translational modifications might allow for nearly infinite modulation of protein function. The immense task of coordinating cellular activities and responding to extracellular cues with both temporal and spatial accuracy is likely to require the concerted action of both of these regulatory modifications. b-O-GlcNAc Is a Ubiquitous and Dynamic Modification Reports that alkali-induced b-elimination of adenovirus fiber proteins releases GlcNAcitol hinted at the existence of O-linked GlcNAc (9). Subsequent analysis confirmed the presence of b-OGlcNAc and suggested that the modification may be involved in adenovirus fiber assembly or stabilization (10). Although another study suggested the existence of O-glycosidically linked GlcNAc on extracellular proteins (11), later structural analyses suggested that these workers were likely observing a-linked GlcNAc, a mucin-like modification common in primitive eukaryotes (12). Another early study showed that the GlcNAc-binding lectin wheat germ agglutinin blocked ATP-dependent RNA nuclear transport (13). The characterization of b-O-GlcNAc in 1984 explained some of these preliminary observations and established O-GlcNAc as a major form of intracellular protein glycosylation (14). The nuclear pore proteins were among the first structurally characterized O-GlcNAc proteins (15, 16). Since then, several laboratories have shown that hundreds, if not thousands, of proteins in the nucleus and cytoplasm are modified with O-GlcNAc (3, 4). Given the broad spectrum of proteins that contain this modification, there are likely to be many different functions for O-GlcNAc. Studies with the transcription factor Sp1 suggest that O-GlcNAc protects the protein from proteasome degradation (17). Recent reports have shown that recognition of O-GlcNAc on peptides constitutes an important feature of major histocompatibility complex Class I antigen presentation (18). Pulse-chase analyses have shown that the O-GlcNAc modification of some proteins is highly transient, with turnover rates similar to phosphorylation (19, 20). Another study found that dynamic changes of O-GlcNAc-modified proteins are associated with lymphocyte activation (21). Several recent reports with phosphatase and kinase inhibitors have provided direct support for a relationship between O-phosphorylation and O-glycosylation of serine or threonine residues of some proteins (4, 22–24). Consistent with the hypothesis that O-GlcNAc has a regulatory role, disruptions of the O-GlcNAc transferase homolog, SPY, in Arabidopsis results in impaired gibberellin signal transduction (25). It is clear that OGlcNAc is involved in very diverse aspects of cellular physiology (Fig. 2). The challenge for the coming years is to determine the precise contribution of O-GlcNAc in the regulation of these systems.

processing or alternative RNA transcript processing. The O-Glc-NAc transferase itself is modified with O-GlcNAc as well as tyrosine phosphate (28), suggesting possible regulation through posttranslational modifications. The activity of OGT is potently inhibited by UDP, the by-product of the transfer reaction. The enzyme is sensitive to UDP-GlcNAc/UDP ratios over the entire physiologic range of sugar nucleotide concentration (micromolar to millimolar). In many cells the concentrations of UDP-GlcNAc ap-proach that of ATP with as much as 2-5% of total glucose utilization channeled to making this sugar nucleotide (32). Thus, local or global perturbations in the UDP-GlcNAc or UDP levels can modulate the activity of OGT. These features of the O-GlcNAc transferase and its environment suggest that it may be subject to complex regulatory mechanisms.
A cytosolic and nuclear ␤-N-acetylglucosaminidase (O-GlcNAcase) with a neutral pH optimum and selectivity toward O-linked GlcNAc has also been identified and purified (33). Several useful inhibitors, such as, O-(2-acetamido-2-deoxy-O-glucopyranosylidene)-amino-N-phenylcarbamate (PUGNAC) (34), have been identified. Treatment of several different cell types with this inhibitor results in an overall increase in O-GlcNAc levels on numerous proteins (35). Current data suggest that in many systems the regulation of O-GlcNAc cycling will have important consequences for the regulation of protein function. For example, a recent study showed that, when expressed in the cytoplasm and nucleus, a Golgi enzyme that caps O-GlcNAc and prevents its normal cycling is lethal to cells (36). The cloning and expression of the O-GlcNAcase will help determine the interplay between the regulated addition and removal of O-GlcNAc from proteins.

Role of ␤-O-GlcNAc in Nuclear Trafficking, Transcription, and Translation
Nucleoporins, which mediate the active transport of macromolecules into and out of the nucleus, are extensively modified with O-GlcNAc on their exposed surfaces. Several studies have shown that monoclonal antibodies or lectins that bind O-GlcNAc block nuclear transport of macromolecules at an energy-dependent step (37). Some investigators have suggested that O-GlcNAc is an alternative nuclear transport signal on some proteins (38). It is also possible that O-GlcNAc is directly involved in control of the translocation machinery. Nevertheless, the roles of O-GlcNAc in nuclear transport are presently unclear (39).
RNA polymerase II and many of its transcription factors are extensively modified with O-GlcNAc (40,41). The O-GlcNAc modification and O-phosphorylation of the C-terminal repeat domain (CTD) appear to be reciprocal (7). The glycosylation of the CTD sequence induces a turnlike structure, which could dramatically alter the conformation of the domain (42). Such conformational changes in the CTD could affect the specificity of interactions with other components of the transcription machinery. Phosphorylation of the CTD is associated with transcript elongation and RNA processing (43,44). Therefore, it is plausible that O-GlcNAc on the CTD may be involved in events prior to the elongation phase of the transcription cycle. There is evidence that O-GlcNAc plays a role in controlling both the turnover and transactivation activities of key The two modifications may be reciprocal, either at the same site (path 4) or at adjacent sites (path 5). In this case, specific cellular cues may signal interconversion between the two modification states. Alternatively, the modifications may both change independently in response to cellular signals. In this scenario, the modifications could occupy the same site (paths 1 and 2) or different sites (paths 2 and 3) on distinct subsets that represent independent functional states of the protein. Finally, the two modifications may influence each other in an ordered manner, with modulation of protein function involving sequential glycosylation and phosphorylation (path 1 to 6 or path 2 to 7). Note that the model depicts full site occupancy and thus does not take into account the subtle modulation possible through alterations of substoichiometric site occupancy. In addition, it should be noted that most of the O-GlcNAc-modified proteins identified to date have multiple sites of O-GlcNAc and phosphate. The addition of each new modification site exponentially increases the potential complexity of the interplay between O-GlcNAc and phosphate, thus dramatically increasing the possibilities for exerting multiple levels of control. transcription factors, such as SP1 and estrogen receptors (41). Some researchers have proposed that O-GlcNAc modulates or mediates the assembly of transcriptional complexes, leading to the activation of specific genes (7,22). Although it is clear that O-GlcNAc is enriched on proteins involved in eukaryotic gene transcription, further work is necessary to elucidate its specific roles in this complex process.
Gupta, Datta, and colleagues (45) have shown that O-GlcNAc plays a direct role in regulating protein translation. The eukaryotic initiation factor 2 (eIF-2) is a translation initiation factor that is inactivated when phosphorylated by any of several eIF-2 kinases. The O-GlcNAc-modified form of the eIF-2-associated protein, p67, binds to eIF-2 and prevents the action of the inhibitory eIF-2 kinases. Under conditions of serum starvation or heme depletion, a normally latent deglycosylase removes the O-GlcNAc from p67, which causes p67 to dissociate from eIF-2 and also accelerates the proteolytic degradation of p67. As a result of p67 dissociation, eIF-2 becomes subject to the action of eIF-2 kinases, hence shutting down protein translation. It will require the work of many laboratories to sort out the functions of O-GlcNAc in the regulation of protein synthesis, but it is clear that it plays a critical role in this process.

Dynamic ␤-O-GlcNAc Modification of
Cytoskeletal Proteins Many actin and tubulin regulatory proteins as well as intermediate filament proteins are O-GlcNAc-modified (22,46). Erythrocyte Band 4.1 associated with the plasma membrane appears to contain more O-GlcNAc than the total population. Many proteins that serve to bridge the cytoskeleton to cellular membranes are O-GlcNAc-modified, such as talin, vinculin, and the synapsins (3,47). O-GlcNAc site-mapping studies on neurofilaments, together with limited mutagenesis studies, suggest that the saccharide might play a role in intermediate filament fibrillogenesis. Interestingly, O-phosphorylation and O-GlcNAc modification appear to occur on distinct subsets of cytokeratins (48), with both changing during the cell cycle or after treatment with phosphatase inhibitors. Thus, on some proteins the O-GlcNAc and phosphate may not merely be reciprocal modifications but rather may represent functionally distinct isoforms of these proteins (Fig. 1).

Potential Implications of ␤-O-GlcNAc
for Human Disease O-GlcNAc and Neurodegenerative Diseases-O-GlcNAc is abundant in the brain, particularly on cytoskeletal proteins such as neurofilaments, microtubule-associated proteins, clathrin assembly protein, and the ␤-amyloid precursor protein (46, 49 -51). Disruptions of O-GlcNAc modification of some of these proteins may contribute to certain neurodegenerative disorders. For example, the OGT gene itself maps on the X chromosome to the same locus as X-linked Parkinson dystonia (6). The microtubule-associated protein, Tau, which is a major component of neurofibrillary tangles in Alzheimer's diseased brains, is multiply O-GlcNAc-modified in normal brains (49). In contrast, Tau is hyperphosphorylated in association with Alzheimer's disease (52). The ␤-amyloid precursor protein, which gives rise to the Alzheimer's disease-associated neurotoxic ␤-amyloid peptide upon proteolysis, is also O-GlcNAc-modified (51). We have hypothesized that some neurodegenerative diseases might result directly from decreased glucose metabolism by aging neurons, which results in a gradual reduction in O-GlcNAc modification of key cytosolic and nuclear proteins. This gradual failure to add O-GlcNAc could subsequently lead to the abnormal phosphorylation of many proteins, but the abundant cytoskeletal proteins would manifest the effect first.
O-GlcNAc and Cancer-As noted above, O-GlcNAc is present on many transcription regulatory proteins. Mutations of many of these proteins contribute to the oncogenic phenotype. Some of these mutations may exert their effects in part by disruption of O-Glc-NAc-mediated regulation of these proteins. For example, the c-myc oncogene is glycosylated in its transactivation domain at Thr-58, which is also the mutational hotspot found in a large percentage of Burkitt's lymphomas from human patients. Interestingly, this glycosylation site is also an important phosphorylation site that regulates c-myc transcriptional activity (53). Clearly the many sitedirected mutagenesis studies that were intended to elucidate the roles of O-phosphate on this and other proteins, in fact, cannot distinguish between the biological importance of O-phosphate and O-GlcNAc at these sites of reciprocal modification. There also appears to be a reciprocal relationship between the O-phosphorylation and O-GlcNAc modification of the tumor-associated SV-40 Large T antigen (54). The tumor suppressor, p53, which is the most commonly mutated gene in a wide range of human cancers, is also O-GlcNAc-modified (55). There is preliminary evidence that the O-GlcNAc on p53 regulates its binding to DNA. Many essential regulators of cellular function are subject to complex phosphoregulation pathways (56). O-GlcNAc modification adds another level of regulation, which could allow for exquisite control of cell regulatory mechanisms. Disruptions of either of these post-translational modifications may interfere with critical control mechanisms, leading to the transformed phenotype.
O-GlcNAc and Diabetes-Recent evidence points to a link between O-GlcNAc misregulation and diabetes. Several studies have clearly established that the conversion of glucose into glucosamine, which is catalyzed by glutamine:fructose-6-phosphate amidotransferase (GFAT, EC 2.6.1.16), is essential for the development of insulin resistance (type 2 diabetes) (57). The action of GFAT to produce glucosamine provides the key metabolic precursor to UDP-GlcNAc, the donor for O-GlcNAc modification. GFAT-deficient cells cannot be made insulin-resistant without a source of glucosamine. Conversely, overexpression of GFAT leads to hyperinsulinemia and insulin resistance (58). Exposure of cells or whole animals to millimolar concentrations of glucosamine invariably results in the development of insulin resistance (59,60). It is not clear at this point whether elevated glucosamine contributes directly to O-Glc-NAc misregulation or subsequently to diabetes, but a recent study showed that insulin and glucosamine infusions increase the amount of O-GlcNAc in skeletal muscle proteins in vivo (61). Streptozotocin, a structural analog of glucosamine, selectively destroys the ␤-cells of the pancreas and induces diabetes in a single relatively small dose (50 mg/kg of body weight). Streptozotocin is a weak inhibitor of O-GlcNAcase and elevates O-GlcNAc levels in the pancreas (62)(63)(64), suggesting that its mode of action may involve disruption of O-GlcNAc-mediated processes. Another recent study found that glucosamine may induce insulin resistance by preventing the transport of Glut4 vesicles from the cytoskeleton to the plasma membrane (65). Direct evidence for a role of O-GlcNAc in diabetes is still lacking. Nevertheless, given the key role of glucosamine metabolism in the disease, the effect of glucosamine on O-GlcNAc levels, and the presence of O-GlcNAc on many essential regulatory proteins, it is plausible that O-GlcNAc modification is directly involved in the development of insulin resistance. One model that incorporates the existing data reasons that high glucose or glucosamine leads to elevated cellular UDP-GlcNAc concentrations, resulting in hyper-O-GlcNAc modification of proteins. The elevation of O-GlcNAc on certain vesicle-associated proteins such as Glut4 could prevent the phosphorylation of sites required to release the vesicles from the cytoskeleton, thus preventing their subsequent transport to the plasma membrane. The hyper-O-Glc-NAc modification of key transcription factors or signaling molecules involved in the response to insulin could also contribute to insulin resistance. Current work in several laboratories is actively considering these and other possibilities for determining the mechanisms of glucosamine-induced diabetes.

Perspectives
We have been aware of the O-GlcNAc modification of nuclear and cytosolic proteins for over 16 years, yet we are only beginning to understand the functions of this saccharide. This dilatory progression is partly because of the inherent difficulty in detecting the modification using conventional biochemical tools, the lack of analytical methods, the enzymatic and chemical lability of the O-GlcNAc linkage, and the small number of laboratories that have focused their efforts in this area of research. The recent development of monoclonal antibodies capable of detecting O-GlcNAc, the availability of potent O-GlcNAcase inhibitors, as well as advances in mass spectrometry (66) will greatly facilitate analysis of O-GlcNAc on low abundance proteins. It is clear that O-GlcNAc modification plays a role in many of the most fundamental cellular events, including transcription, translation, nuclear transport, and cytoskeletal assembly. Aberrations in the control of O-GlcNAc modification in any of these systems is likely to contribute to any of a number of disease states. As in the case for protein phosphorylation (67,68), it will require the combined efforts of many investigators in each of these areas to elucidate the function of O-GlcNAc in these critical processes.