Modulation of O-linked N-acetylglucosamine levels on nuclear and cytoplasmic proteins in vivo using the peptide O-GlcNAc-beta-N-acetylglucosaminidase inhibitor O-(2-acetamido-2-deoxy-D-glucopyranosylidene)amino-N-phenylcarbamate.

O-Linked N-acetylglucosamine (O-GlcNAc) is a ubiquitous and abundant post-translational modification found on nuclear and cytoplasmic proteins and is thought to be a dynamically regulated modification much like phosphorylation. In this study we have demonstrated that O-(2-acetamido-2-deoxy-D-glucopyranosylidene)amino-N-phenylcarbama te (PUGNAc), a potent in vitro inhibitor of the enzyme responsible for the removal of O-GlcNAc from proteins (peptide O-GlcNAc-beta-N-acetylglucosaminidase), can be used to increase O-GlcNAc levels on nuclear and cytoplasmic proteins in vivo. Overall, PUGNAc caused approximately a 2-fold increase in O-GlcNAc levels in the human colon cancer cells, HT29, although the effects on individual proteins varied. The increase appeared to be the result of the direct inhibition of the peptide O-GlcNAc-beta-N-acetylglucosaminidase since neither the O-GlcNAc transferase nor UDP-GlcNAc levels were affected by the treatment. O-GlcNAc levels in other cell lines tested (NIH 3T3, CV-1, and HeLa) were also affected by PUGNAc, although the effects on HeLa cells were minimal. At the concentrations tested, PUGNAc was non-toxic and had no affect on the growth rate of any of the cell lines examined. Interestingly, we demonstrated that an increase in O-GlcNAc levels on the transcription factor Sp1 resulted in a reciprocal decrease in its level of phosphorylation, supporting the hypothesis that O-GlcNAc competes with phosphate on some proteins. These studies demonstrate that PUGNAc is an effective inhibitor of O-GlcNAc turnover within cells and can be used to selectively alter the extent of O-GlcNAc on cellular proteins.

O-Linked N-acetylglucosamine (O-GlcNAc) 1 is a major form of post-translational modification found on nuclear and cytoplasmic proteins (1)(2)(3). It consists of the monosaccharide Nacetylglucosamine linked to the hydroxyl group of either serine or threonine. Well over 50 proteins have been identified to date with the modification (1-3), including RNA polymerase II (4), several RNA polymerase II transcription factors (5), nuclear pore proteins (6), the tumor suppressor protein p53 (7), and c-Myc (8). Sites of glycosylation have been identified on a number of O-GlcNAc-modified proteins, and although a strict consensus sequence cannot be discerned, most sites contain a proline residue amino-terminal to the modified serine or threonine (1). About half of the sites contain the motif PX(S/T), where X is valine, serine, or threonine and (S/T) is the modified serine or threonine.
Accumulating evidence suggests that O-GlcNAc is a regulated modification much like phosphorylation (1)(2)(3). In several instances it has been shown that the sugar turns over more rapidly than the protein it modifies (9,10), implying that O-GlcNAc is dynamically added to and removed from proteins. In addition, conditions of growth stimulation (11) and of growth arrest (12) have been demonstrated to alter the level of O-GlcNAc in cells. Enzymes for the addition (O-GlcNAc transferase) (13)(14)(15)(16) and removal (O-GlcNAcase) (17) of the sugar are known to exist in the cytoplasm of most eukaryotic cells. Thus, a system capable of dynamic regulated addition and removal of the sugar analogous to a kinase/phosphatase system is present in cells.
A major function of O-GlcNAc may be to compete with phosphorylation for sites on proteins (1)(2)(3). Since GlcNAc is neutral, it is possible that it would have different effects on protein function than a strongly negatively charged phosphate group. Thus, competition between GlcNAc and phosphate for similar sites would add an extra level of control to signal transduction cascades in cells. Alternatively, O-GlcNAc or phosphate could affect protein activity in a similar manner, although the modifications would be the result of different signaling cascades. In either case, only one modification could occur at an individual site at any one time. Evidence indicating that this type of competition occurs has been obtained by demonstration that the sites of glycosylation on both RNA polymerase II (4) and c-Myc (8) coincide with known phosphorylation sites. Even so, direct modulation of the level of phosphate on a protein by changing the level of glycosylation on the same protein has not yet been demonstrated in vivo.
Selective inhibitors of protein kinases and phosphatases have been used extensively to modulate the level of phosphate on proteins as an approach to determine the functional implications of phosphorylation. By analogy to the phosphorylation system, specific inhibitors of either the O-GlcNAcase or O-GlcNAc transferase would be valuable tools in the study of the function of the O-GlcNAc modification. In the present study, we sought to address two goals. First, we wanted to examine the feasibility of using a known in vitro inhibitor of the O-GlcNA-case for modulation of O-GlcNAc levels on proteins in vivo. Second, if such modulation was feasible, we wanted to test the hypothesis that O-GlcNAc and phosphate compete on proteins by determining whether alterations in the level of O-GlcNAc on a given protein would result in changes in the level of phosphorylation of the same protein. In these studies, we have utilized the potent in vitro inhibitor of O-GlcNAcase, PUGNAc (17), to address these goals.

EXPERIMENTAL PROCEDURES
Materials-Bovine milk galactosyltransferase, Protein G-agarose, disaccharides (galactose-␤1,3-N-acetylgalactosamine, galactose-␤1,3-N-acetylglucosamine, and galactose-␤1,4-N-acetylglucosamine), and ovalbumin were from Sigma. The alditol derivatives of the disaccharides were prepared by reduction with sodium borohydride as described (13). Peptide N-glycosidase F (PNGase F) was purified from the culture filtrate of Flavobacterium menigisepticum as described (18 Cell Culture and PUGNAc Treatments-All cells were maintained in a humidified incubator with 5% CO 2 in Dulbecco's modified Eagle's medium (low glucose) containing 10% fetal bovine serum and penicillin/ streptomycin (complete Dulbecco's modified Eagle's medium). PUGNAc treatments were performed by replacing the media on subconfluent plates (50 -75% confluent) with fresh complete Dulbecco's modified Eagle's medium containing the indicated concentration of PUGNAc. Stock solutions of PUGNAc were made in Milli Q water and filtersterilized prior to use. At the appropriate time points, the cells were washed three times with 10 mM Tris-HCl, pH 7.5, 0.15 M NaCl (Trisbuffered saline). Cells were lysed for 10 min on ice by addition of cold radioimmune precipitation assay buffer (1 ml/100-mm dish, 50 mM Tris-HCl, pH 8.0, 0.15 M NaCl, 0.1% SDS, 0.5% sodium deoxycholate, 1% Nonidet P-40) containing protease inhibitor mixtures 1 and 2 (Pic 1 and Pic 2) (19). The cells were scraped off of the dish, and aggregates were broken by passage of the lysates through a 25-gauge needle several times. Insoluble material was removed by centrifugation at 10,000 ϫ g for 5 min at 4°C.  (20) in the presence of increasing concentrations of PUGNAc. HT29 cell lysates for the enzyme assays were made by lysing subconfluent (80 -90%) plates (100-mm dishes) of HT29 cells in 0.5 ml of 20 mM Tris-HCl, pH 7.5, 20% glycerol (w/v), 0.5% Nonidet P-40, containing Pic 1 and 2. The cells were lysed for 15 min on ice, and the insoluble material was removed by centrifugation (10,000 ϫ g, 10 min, 4°C). Samples for O-GlcNAc transferase assays were desalted on 1-ml Sephadex G-50 columns prior to assay as described (20). The O-GlcNAcase activity was assayed in the presence of 50 mM GalNAc at pH 6.4 (optimal conditions for the O-GlcNAcase) to inhibit lysosomal hexosaminidase activities (17).

Metabolic
UDP-GlcNAc levels were determined by high pressure liquid chromatographic analysis of perchloric acid extracts of HT29 cells as described (20). Duplicate plates (100 mm) of subconfluent (80 -90%) HT29 cells were treated for 24 h with 0, 20, or 40 M PUGNAc. The cells were removed from the plates using trypsin/EDTA, counted, and extracted with perchloric acid as described (20). The chromatographic procedure used to separate the nucleotides does not resolve UDP-GlcNAc from UDP-GalNAc, but we have previously shown that the ratio of UDP-GlcNAc to UDP-GalNAc in HT29 cells is approximately 2:1, as it is in most cells (20,21). Mammalian cells contain an epimerase which maintains this ratio. Thus, the amount of UDP-GlcNAc was calculated from the values obtained for uridine diphospho-N-acetylhexosamine (UDP-GlcNAc plus UDP-GalNAc) using this ratio.
Other-Galactosyltransferase labeling after denaturing proteins with 1% SDS was performed as described previously (20). Immunoprecipitations, Western blots, and carbohydrate analysis (digestion with PNGase F, alkali-induced ␤-elimination, size fractionation, and high pH anion exchange chromatography) were all performed as described previously (20). Densitometric scans of Western blots and fluorographs were performed on a Bio-Rad model GS-670 Imaging Densitometer. All gels were 10% SDS-polyacrylamide gels as described by Laemmli (22). Protein was estimated using the BCA assay (Pierce).

PUGNAc Increases O-GlcNAc Levels on Proteins in HT29
Cells-To examine the effects of PUGNAc on O-GlcNAc levels in cells, we treated HT29 cells with varying concentrations of PUGNAc overnight (Fig. 1A) and with 40 M PUGNAc for varying lengths of time (Fig. 1B). The cells were lysed at the indicated times, and the level of terminal GlcNAc moieties per unit of protein was determined using galactosyltransferase and UDP-[ 3 H]galactose. Both a concentration-and time-dependent increase in the level of terminal GlcNAc moieties resulted from PUGNAc treatment (Fig. 1, A and B). The maximal increase appeared at 40 M PUGNAc. No further increase in GlcNAc levels was seen in cells treated with higher PUGNAc concentrations (up to 400 M, data not shown). Increases could be seen after 2 h of treatment and seemed to plateau after approximately 8 h. The increased level of terminal GlcNAc moieties could be maintained for several days without changing media ( Fig. 1B), indicating that the inhibitor was not being significantly metabolized by the cells. Interestingly, the level of terminal GlcNAc in the absence of PUGNAc was already substantial, indicating that the O-GlcNAc levels in untreated cells are constitutively high. Finally, the effects of the inhibitor could be reversed fairly rapidly (Fig. 1C) by simply replacing the inhibitor-containing media with inhibitor-free media.
Since galactosyltransferase will modify GlcNAc moieties found on both N-glycans as well as O-GlcNAc (19), we needed to determine which of these forms of glycosylation were being affected by the PUGNAc treatment. Samples treated with or without 40 M PUGNAc for 24 h were galactosyltransferaselabeled as above and subjected to digestion with PNGase F. The PNGase F-released material (N-glycans) was separated from the PNGase F-resistant material (O-glycans, see below), and both were quantified ( Fig. 2A). As we have seen before (20), the majority of the radioactivity appeared to be O-linked (i.e. was resistant to PNGase F digestion), indicating that most terminal GlcNAc moieties in HT29 cells are O-linked. The O-linked material showed approximately a 1.5-to 2-fold increase in terminal GlcNAc moieties following PUGNAc treatment. An increase in terminal GlcNAcs on N-glycans was also observed, although it was not as statistically significant. To confirm that the PNGase F-resistant fraction consisted of O-linked sugars, it was subjected to alkali-induced ␤-elimination (Fig. 2B). In both cases, essentially all of the radiolabel was released from the proteins, indicating that it was O-linked to the protein. Size fractionation of the released O-glycans (Fig. 2C) showed that all of the radiolabel was in the form of a disaccharide. Finally, to confirm that the disaccharide was the result of O-GlcNAc, high pH anion exchange chromatography analysis (Fig. 2D) demonstrated that the disaccharide was Gal␤1,4GlcNAcitol, the expected product of ␤-eliminated, galactosyltransferaselabeled O-GlcNAc. Thus, the majority of the increase in terminal GlcNAc moieties caused by PUGNAc appeared to be due to an increase in the total O-GlcNAc in the cell. In HT29 cells, the increase was approximately 2-fold.
We next examined whether the increase in O-GlcNAc levels caused by PUGNAc was a result of changes in the selective glycosylation of a subset of proteins or if the changes were more general. Galactosyltransferase-labeled samples from HT29 cells treated with increasing concentrations of PUGNAc were digested with PNGase F and subjected to SDS-PAGE (Fig. 3A).
An increase in [ 3 H]galactose labeling was seen on a number of proteins. Since we had demonstrated that essentially all of the [ 3 H]galactose remaining after PNGase F treatment was O-GlcNAc ( Fig. 2), we concluded that PUGNAc induces a generalized increase in O-GlcNAc levels on many proteins. This suggests that the effects of PUGNAc are specific for the enzyme responsible for deglycosylation of most O-GlcNAc-modified proteins. Densitometric scans (Fig. 3B) of individual bands from the fluorograph in Fig. 3A showed that a similar increase in O-GlcNAc levels was seen on all of the proteins analyzed, although the extent of the change varied slightly from protein to protein (from approximately 1.5-to 2.5-fold increase in intensity). Consistent with the data shown in Fig. 1A, the majority of the increase in O-GlcNAc was seen between 0 and 20 M PUGNAc, with a smaller increase occurring between 20 and 40 M. Thus, PUGNAc appears to induce a generalized increase in O-GlcNAc levels on numerous proteins in HT29 cells, although the extent of the increase is somewhat protein-dependent.
Effect of PUGNAc on the O-GlcNAc Glycosylation Machinery-Although PUGNAc is known to be a competitive inhibitor of the O-GlcNAcase in vitro, we wanted to determine whether the change in O-GlcNAc levels we had observed in HT29 cells was in fact due to inhibition of this enzyme. There are three components that determine the level of O-GlcNAc in cells as follows: the O-GlcNAc transferase, O-GlcNAcase, and UDP-GlcNAc levels (1,20). We assayed the effect of PUGNAc on each of these components. As expected, PUGNAc has a dramatic effect on the activity of the O-GlcNAcase in HT29 cells (Fig.  4A). The concentration dependence of inhibition correlates very well with the concentration required to achieve maximal effects in cells (40 -50 M, see Fig. 1A). In contrast, PUGNAc had no effect on the O-GlcNAc transferase activity from HT29 cells. In addition, treatment of HT29 cells with 20 or 40 M PUGNAc for 24 h (concentrations which increase O-GlcNAc levels, Fig. 1A) caused no increase in UDP-GlcNAc levels (Fig. 4B). If anything, a slight decrease was observed. Consistent with these findings, we have shown that overnight treatment of HT29 cells with media containing 5 mM glucosamine (which results in more than a 10-fold increase in UDP-GlcNAc levels) has no effect on the amount of terminal GlcNAc in the cells. 2 Thus, increasing the concentration of UDP-GlcNAc is not a mechanism for increasing O-GlcNAc levels in HT29 cells. These results strongly suggest that the PUGNAc-induced increase in O-GlcNAc levels is due to a direct inhibition of the O-GlcNAcase within these cells.
Effects of PUGNAc on Other Cell Lines-We have also examined the effect of PUGNAc on O-GlcNAc levels in several other cell lines (NIH-3T3, HeLa, and CV-1). PUGNAc caused an increase in terminal GlcNAc in all of the cells examined, although the level of increase seen in HeLa cells was significantly less than that seen in the others (Fig. 5A). PNGase F treatment of the galactosyltransferase-labeled samples from each of the different cell types showed that the majority of the PUGNAcinduced increase was due to O-GlcNAc (data not shown). Therefore, PUGNAc appears to increase the level of O-GlcNAc on proteins in all cell lines examined, although the level of increase is cell line-dependent. PUGNAc did not appear to have any toxic effects on the cells tested, so we sought to determine whether long term exposure to PUGNAc affected cell growth. Each of the cell lines examined above was grown in the presence or absence of 40 M PUGNAc for up to a week (Fig. 5B). shown). During this time, the presence of PUGNAc did not affect the growth rate of any of the cell lines tested (Fig. 5B). Both the HT29 and NIH-3T3 cells became confluent and stopped growing during the last 2 days of the experiment, although the presence or absence of PUGNAc had no effect.
Effects of PUGNAc on Phosphorylation State of Proteins-One of the major hypotheses describing a potential function for O-GlcNAc is that it may compete with phosphate for individual hydroxyl groups (1-3). Now that we could reproducibly increase O-GlcNAc levels in cells, we could directly test whether this competition occurs in vivo. HT29 cells were treated with or without 40 M PUGNAc (conditions known to approximately double O-GlcNAc levels in these cells, see Figs. 1 or 5) and metabolically labeled in vivo with [ 32 P]orthophosphate. Proteins from crude lysates of these cells were separated by SDS-PAGE, and the radiolabeled proteins were analyzed on a Phos-phorImager. Fig. 6A shows that the presence or absence of PUGNAc had no noticeable effect on the intensity of phosphorylation of any proteins that could be detected in these crude samples. Thus, PUGNAc does not cause dramatic changes in the overall phosphorylation state of major proteins in the cell. We next investigated the effects of PUGNAc on the phosphorylation of Sp1, a protein known to be modified with both phosphate (23) and O-GlcNAc (5). Sp1 was immunoprecipitated from [ 32 P]phosphate-labeled HT29 cells that had been treated with and without 40 M PUGNAc (Fig. 6B, 32

O-GlcNAcase in vivo, resulting in a general increase in O-
GlcNAc levels on numerous proteins.
In addition to demonstrating that PUGNAc can be used to increase O-GlcNAc levels on proteins in cells, we have used the inhibitor to test the hypothesis that protein modifications by O-GlcNAc and phosphate are reciprocally related to one another. PUGNAc had no effect on the overall phosphorylation pattern in HT29 cells, indicating that the major phosphoproteins in these cells are not affected by changes in O-GlcNAc levels, perhaps because many of these proteins are not modified with O-GlcNAc. These results also suggest that PUGNAc does not have any global detrimental effect on the protein phosphorylation machinery. In contrast, when we looked specifically at a protein known to be modified with both O-GlcNAc and phosphate (Sp1 (5, 23)), we observed that the increase in O-GlcNAc caused by PUGNAc was accompanied by a decrease in phosphorylation. The decrease in phosphorylation following the PUGNAc treatment was smaller than the corresponding increase in O-GlcNAc levels, indicating that some phosphorylation sites on Sp1 may be unaffected by changes in O-GlcNAc levels. It is likely that there is not complete overlap between the glycosylation and phosphorylation sites. Neither the sites of glycosylation nor phosphorylation on Sp1 are known, so we cannot conclusively state that O-GlcNAc and phosphate compete for identical sites, although at least three Class I O-GlcNAc sites (1) are discernible in the sequence (24). These sites could also be modified by proline-directed or mitogenactivated protein kinases. Nonetheless, these results suggest that glycosylation and phosphorylation are in a reciprocal relationship on this particular protein. To our knowledge, this is the first demonstration of altering the level of phosphorylation by increasing the level of O-GlcNAc on a protein in vivo, and it serves as an important step toward understanding the role of O-GlcNAc on nuclear and cytoplasmic proteins in cellular function.
Although the majority of the PUGNAc-induced increase in terminal GlcNAc moieties in cells was due to O-GlcNAc, a small increase in terminal GlcNAc on N-glycans also occurred. Since PUGNAc is a known inhibitor of lysosomal hexosaminidases (25), this is most likely the result of inhibition of lysosomal hexosaminidases and accumulation of N-glycans bearing terminal GlcNAc moieties. Alternatively, PUGNAc could be indirectly affecting the biosynthesis of N-glycans by some unknown mechanism. The fact that PUGNAc had little or no effect on UDP-GlcNAc pools suggests that the effect is not being mediated through alterations in the nucleotide sugar pools in the cell.
The inhibition of the O-GlcNAcase in HT29 cells resulted in a increase in O-GlcNAc levels on essentially all O-GlcNAcmodified proteins (see Fig. 3). This suggests that the O-GlcNAcase which has been purified and characterized (17)  enzymes exist, PUGNAc may have similar inhibitory effects on these activities as well. Interestingly, the extent of PUGNAcinduced changes on different proteins varied somewhat. The increase in O-GlcNAc on Sp1 was greater than that seen on the proteins analyzed in Fig. 3, suggesting that the rate of O-GlcNAc turnover may differ from protein to protein. Similar increases in O-GlcNAc levels were seen in HT29, NIH-3T3 cells, and CV-1 cells, although much smaller changes were seen in HeLa cells. The reason for differences in the effect of PUG-NAc on the HeLa cells is not clear. It may be the result of lower rates of O-GlcNAc turnover in these cells. If the cells express low levels of O-GlcNAcase then they will not be greatly affected by PUGNAc. Tissue searches of the O-GlcNAcase enzyme activity demonstrated variability from tissue to tissue (17), suggesting that the expression levels may vary. Alternatively, PUGNAc may be efficiently excluded from HeLa cells, preventing significant intracellular accumulation of the inhibitor.
Despite the fact that removal of O-GlcNAc from proteins was inhibited, the growth rate of all of the cells examined was unaffected by continued exposure to PUGNAc. This suggests that turnover of O-GlcNAc is not essential for cell growth. Effective inhibitors of the O-GlcNAc transferase will need to be developed before it can be determined how the lack of O-GlcNAc on proteins affects growth. It is somewhat surprising that there is no growth effect, especially in light of the finding that changes in O-GlcNAc also affect phosphorylation of proteins like Sp1. The results suggest that the presence of O-GlcNAc does not inhibit growth but may stimulate growth. Since several sites of O-GlcNAc modification resemble mitogenactivated protein kinase sites, it is possible that the presence of O-GlcNAc or phosphate on a protein at these sites may have a similar growth stimulatory effect. Thus, instead of competition at individual sites, there may be sharing of sites by different systems that are responding to different sets of upstream signals, both resulting in cell growth. Further work comparing the effects of O-GlcNAc to the effects of phosphate on a protein's function when the modifications occur at the same site needs to be done before such conclusions can be drawn.
Our results on the lack of growth effects are somewhat in conflict with a preliminary report by Snow and co-workers (2,26). They demonstrated that expression of galactosyltransferase in the cytoplasm of Chinese hamster ovary cells resulted in rapid cell death. The galactosyltransferase put a galactose "cap" on O-GlcNAc moieties in the cytoplasm. They concluded that the galactose prevented turnover of O-GlcNAc, and thus, turnover of O-GlcNAc must be essential for cell viability (2,26). In view of our data, an alternative interpretation may be that unmodified GlcNAc moieties are essential for cell viability, possibly to allow interaction with specific GlcNAc receptors. Other workers (27) have shown the existence of GlcNAc-binding proteins in nuclear and cytoplasmic compartments. Treatment with PUGNAc causes an accumulation of terminal Glc-NAcs and may have no effect on the binding of these GlcNAcbinding proteins. Capping with galactose would "hide" the O-GlcNAcs and prevent these essential interactions resulting in cell death. Alternatively, the galactose capping could cause detrimental conformational changes or protein-protein interactions that result in cell death.
The use of PUGNAc to selectively alter the level of glycosylation of proteins in cells is a valuable new tool for the study of the role of O-GlcNAc on individual proteins. There is a great deal of interest in identifying specific inhibitors of the O-Glc-NAc transferase, which will be extremely useful for lowering O-GlcNAc levels in cells for similar types of studies. In combination, these types of inhibitors will be invaluable for assessing the role of O-GlcNAc on individual proteins. Others have shown that O-GlcNAc levels can be reduced by lowering UDP-GlcNAc levels in cells (28). The treatments required to lower UDP-GlcNAc levels are somewhat drastic (either glucose starvation or treatment with nonspecific amidotransferase inhibitors such as 6-diazo-5-oxonorleucine). These treatments will affect all forms of glycosylation containing GlcNAc or GalNAc (N-glycans, mucin-type O-glycans, glycosylphosphatidylinositol anchors, glycosaminoglycans, glycosphingolipids, and O-GlcNAc) and must be interpreted with caution. Nonetheless, intriguing results have been generated with this approach. Recently, it was shown that the transcriptional activation of certain genes (28) as well as the degradation rate of Sp1 (29) could be dramatically affected by reduced O-GlcNAc levels. As inhibitors such as PUGNAc are used more widely, further insights into the function of O-GlcNAc modification of proteins will be possible.