Perturbations in O-linked β-N-Acetylglucosamine Protein Modification Cause Severe Defects in Mitotic Progression and Cytokinesis*

The dynamic modification of nuclear and cytoplasmic proteins with O-linked β-N-acetylglucosamine (O-GlcNAc) is a regulatory post-translational modification that is rapidly responsive to morphogens, hormones, nutrients, and cellular stress. Here we show that O-GlcNAc is an important regulator of the cell cycle. Increased O-GlcNAc (pharmacologically or genetically) results in growth defects linked to delays in G2/M progression, altered mitotic phosphorylation, and cyclin expression. Overexpression of O-GlcNAcase, the enzyme that removes O-GlcNAc, induces a mitotic exit phenotype accompanied by a delay in mitotic phosphorylation, altered cyclin expression, and pronounced disruption in nuclear organization. Overexpression of the O-GlcNAc transferase, the enzyme that adds O-GlcNAc, results in a polyploid phenotype with faulty cytokinesis. Notably, O-GlcNAc transferase is concentrated at the mitotic spindle and midbody at M phase. These data suggest that dynamic O-GlcNAc processing is a pivotal regulatory component of the cell cycle, controlling cell cycle progression by regulating mitotic phosphorylation, cyclin expression, and cell division.

well understood. Although phosphorylation is the molecular mechanism generally associated with the regulation of cell cycle proteins, another potential regulator that has not been studied in this context is the abundant post-translational modification O-GlcNAc 2 (8).
O-GlcNAc is a ubiquitous post-translational modification in which a single ␤-N-acetylglucosamine molecule is O-linked to serine or threonine residues on cytoplasmic and nuclear proteins (9). O-GlcNAc is thought to act as a modulator of protein function in a manner analogous to protein phosphorylation; the addition of O-GlcNAc to the protein backbone is dynamic and responds to morphogens (10), cellular stress (11), and changes in glucose metabolism (12,13). O-GlcNAc transferase (OGT) (14 -16) adds and O-GlcNAcase removes O-GlcNAc (17,18) in a dynamic manner at sites on the protein backbone similar to those modified by protein kinases and is reciprocal with phosphorylation on some well studied proteins, such as RNA polymerase II (19), estrogen receptor-␤ (20), and the c-Myc proto-oncogene product (21). These data suggest a mechanism by which O-GlcNAc modulates cellular function by competing with phosphorylation.
Accumulating evidence suggests O-GlcNAc may function as a global regulator of cell growth and division. Deletion of OGT in mouse embryonic fibroblasts is associated with delayed growth, increased expression of the cyclin inhibitor p27, and death (22,23). A reduction in O-GlcNAc levels, the result of lowering UDP-GlcNAc levels (substrate for OGT) to 5% of normal, results in cell growth defects (24). Studies in Xenopus demonstrated maturation defects in oocytes microinjected with ␤-galactosyltransferase, which caps terminal O-GlcNAc residues and prevents O-GlcNAc removal (25). Finally, incubation of Xenopus oocytes with the O-GlcNAcase inhibitor PUGNAc (O-2-acetamide-2-deoxy-D-glucopyranosylideneamino-N-phenylcarbamate) (26,27) altered progression of oocytes through progesterone-mediated maturation (28).
In this study we show that global O-GlcNAc protein modification is regulated in a cell cycle-dependent manner, and this dynamic regulation is essential for cell cycle progression. Disruption of O-GlcNAc cycling through pharmacological manipulation or overexpression of OGT or O-GlcNAcase results in major cell cycle defects. Furthermore, we show that the enzymatic processing of O-GlcNAc is critical for proper M-phase progression, cytokinesis, and mitotic protein phosphorylation. These data suggest O-GlcNAc plays a key role in the coordinate regulation of protein function required for correct cell cycle progression and cell division.
Cell Culture and Synchronization-Unless otherwise noted, most cell lines were grown in Dulbecco's modified Eagle's media (5 mM glucose, Invitrogen) with 10% (v/v) fetal bovine serum (Cell Gro) and 1% (v/v) penicillin-streptomycin (Invitrogen). For double thymidine block synchronization, 1 day post-plating (2.0 or 4.0 ϫ 10 5 cells/10-cm dish) cells were incubated in complete media containing 2 mM thymidine. Cells were returned to normal media for 12 h then placed back into thymidine media for an additional 12 h (at this time 10 M DON or 100 M PUG-NAc was added). Cells were then released into complete media (with or without DON/PUGNAc) and harvested at various time points. Alternatively, HeLa cells were incubated in nocodazole (400 ng/ml) overnight (12 h) before harvesting. 3T3-L1 cells were grown to confluence in Dulbecco's modified Eagle's medium (5 mM glucose, Invitrogen) supplemented with 10% (v/v) calf serum and 1% (v/v) penicillin-streptomycin (Invitrogen). Two days post-confluence, cells were released into mitotic clonal expansion by the addition of 0.5 mM 3-isobutyl-1-methylxanthine, 0.25 M dexamethasone, and 10 g/ml insulin (30) in Dulbecco's modified Eagle's medium (20 mM glucose, Invitrogen) supplemented with 10% (v/v) fetal bovine serum (Cell Gro) and 1% (v/v) penicillin-streptomycin (Invitrogen) according to previously published procedures (30).
Adenoviral Constructs and Infection-O-GlcNAcase was subcloned from pcDNA3.1HisA-O-GlcNAcase (NotI/XbaI; Gao et al. (17)) into pShuttle (NheI/XbaI, BD Biosciences Clontech). O-GlcNAcase was subsequently subcloned from pShuttle-O-GlcNAcase into Adeno-X viral DNA (BD Biosciences Clontech), and the resulting adenovirus was propagated and amplified in HEK293A cells according to manufacturer's instructions. Adenovirus was purified from crude viral lysate by cesium chloride density gradient centrifugation. Adenoviral titer was determined by plaque assay according to manufacturer's instructions. OGT adenovirus was a generous gift from Dillmann (University of California at San Diego) (31). The GFP adenovirus was purchased from the Baylor College of Medicine Vector Development Laboratory (vector. bcm.tmc.edu). HeLa cells were infected at the first thymidine block with O-GlcNAcase, OGT, or control GFP adenoviruses at a multiplicity of infection ϭ 50 (vO-GlcNAcase or vOGT). At the second thymidine release, the cells had been infected for 48 h.
Analysis of Proteins-Cells were washed and lysed as described previously (11). Protein concentrations were determined using the Bio-Rad protein dye reagent. Protein extracts (20 g) were separated by SDS-PAGE on criterion Tris-glycine gels (Bio-Rad), transferred to polyvinylidene difluoride membrane (Millipore), and blocked with 3% bovine serum albumin (w/v) (Sigma) in Tris-buffered saline, and detected with antibodies listed above. Some blots were stripped at room temperature for 1 h in 200 mM glycine (pH 2.7) and reprobed using different antibodies. YY1 and Sp1 were immunoprecipitated from radioimmune precipitation assay buffer lysates (Tris-buffered saline with 1% Np-40, 1% sodium deoxycholate, 0.1% SDS) and blotted as described above.
Cell Growth Assay-Cells (1.0 ϫ 10 4 cells/2 cm well) were then incubated with PUGNAc (100 M in Milli-Q water, Millipore) every 12 h. The alamarBlue proliferative assay was performed as per the manufacturer's instructions (BIOSOURCE Inc.), and the absorbance was read at 630 nm.
Confocal Microscopy-HeLa cells were plated onto Lab-Tek II chamber slides (Nalge Nunc) coated with poly-D-lysine (Sigma). At the time of harvest, cells were washed in PBS containing 1 mM Mg 2ϩ , then fixed in PBS/Mg 2ϩ containing 3% (w/v) paraformaldehyde (Sigma). Next, cells were washed 2 ϫ 10 min in PBS/Mg 2ϩ containing 100 mM glycine (pH 7.4), permeabilized in PBS/Mg 2ϩ containing 1% (v/v) Triton-X100 (Sigma) for 10 min, and washed as before. The cells were blocked in 3% (w/v) bovine serum albumin Tris-buffered saline with .05% Tween 20 and incubated with various antibodies overnight at 4°C. After washing 3 ϫ 10 min in buffered saline with .05% Tween 20 and secondary treatment for 1 h at room temperature, cells were washed again and then incubated for 1 min in PBS/Mg 2ϩ with 0.1% (v/v) Triton-X100 and propidium iodide (1 g/ml). Cells were washed as before and mounted. Fluorescent images were obtained using a Zeiss Ultra View 2 confocal microscope at the Johns Hopkins University microscope core facility.
Flow Cytometry and Thymidine Incorporation-HeLa cells were synchronized and released as described above. Cells were harvested after trypsin (Invitrogen) digestion and washed in PBS. The cells were resuspended in 1 ml of ice-cold PBS then fixed with ice-cold 70% ethanol (v/v) added dropwise while vortexing. The cells were stored until further use at Ϫ20°C. Fixed cells were washed with ice-cold PBS and resuspended in PBS containing 0.1% (v/v) Nonidet P-40, 1 mg/ml RNase A, and 0.2% (w/v) propidium iodide. Next, cells were examined for cell cycle stage at the Johns Hopkins University Flow Cytometry core facility using a BD Biosciences FACScaliber. Data were analyzed using Cell Quest. [ 3 H]Thymidine incorporation was performed according to the method described by Tang and Lane (32).

Dynamic O-GlcNAc Protein Modification Is Required for Cell Growth-To
test the hypothesis that dynamic O-GlcNAc protein modification is a control mechanism for cellular growth, we raised global levels of O-GlcNAc using the O-GlcNAcase inhibitor PUGNAc and determined the growth rate of multiple cell lines. Cell number was measured daily; differences between control and PUGNAc-treated cells were evident at 4 days and pronounced at 5 days (Fig.  1A). In almost every cell line studied, elevating O-GlcNAc levels reduced the cells growth rate (Fig. 1A). However, logarithmically growing 3T3-L1 cells showed only a slight inhibition of growth upon chronic PUGNAc treatment. This is most likely due to the low permeability of PUGNAc in undifferentiated 3T3-L1 cells.
O-GlcNAc-and Proline-directed Phosphorylation Are Reciprocal at M Phase-Next, we determined if O-GlcNAc levels were regulated in a cell cycle-dependent manner; HeLa cells were mitotically arrested at the metaphase-anaphase transition with nocodazole (400 ng/ml, 12 h) (33). O-GlcNAc-modified proteins in total cell extract were determined by immunoblot with an O-GlcNAc-specific antibody (29). O-GlcNAc staining was decreased compared with asynchronous control (Fig. 1B, right panel). Although the overall O-GlcNAc levels decrease in these cells and O-GlcNAc staining on a few proteins completely disappears (see blue arrows), an increase in O-GlcNAc staining is seen on several proteins (see red arrows).
We next determined the level of proline-directed phosphorylation using the antibody MPM-2, whose staining is known to increases during M phase (34). As expected the M-phase-arrested cells (nocodazoletreated) show a dramatic increase in the levels of MPM-2 reactivity (Fig.  1B, left panel). Sensitive Coomassie Blue staining of gels indicates that the polypeptide levels of most of the O-GlcNAc-modified or M-phasephosphorylated proteins does not change (data not shown). To determine whether O-GlcNAc and phosphorylation were reciprocal, we measured MPM-2 reactivity in cells treated with PUGNAc. Because PUGNAc inhibits O-GlcNAcase, the enzyme responsible for removing the sugar modification, we expected the increase in overall O-GlcNAc levels in both asynchronous and nocodazole-treated cells. Although PUGNAc increased the extent of 110.6 staining on many proteins in nocodazole-treated cells, PUGNAc treatment did not alter the distribution of O-GlcNAc-modified proteins in these cells. Proline-directed phosphorylation is not perturbed. Although no effect of PUGNAc was seen on proline-directed phosphorylation, we cannot rule out changes on specific proteins not resolved at the one-dimensional level.
To confirm that O-GlcNAc levels change on individual low abundance proteins and not the levels of the polypeptides themselves, we looked at two regulatory proteins already known to be O-GlcNAc-modified, Sp1 (35) and YY1 ( Fig. 1C) (36,37). When YY1 is immunoprecipitated from asynchronous and nocodazole-treated HeLa cells, the levels of O-GlcNAc on this transcription factor change ( Fig. 1C). Nocodazole treatment causes an increase in 110.6 staining in a higher molecular weight band of YY1 and a decrease in 110.6 staining in a lower molecular weight band. When YY1 is immunoprecipitated from PUGNAc/nocodazole-synchronized cells, the extent of O-GlcNAc modification significantly increases, but the electrophoretic pattern is similar to untreated cells. The protein expression of YY1 is constant, and the 110.6 staining is completely abolished when 100 mM N-acetylglucosamine is added to the primary antibody mixture (data not shown). The Sp1 also shows an increase in glycosylation at M phase with no change in the protein expression (data not shown).
Interestingly, when HeLa cells are treated with PUGNAc, the levels of the O-GlcNAc-processing enzymes change (  lower levels. DON reduces O-GlcNAc levels due to its ability to inhibit glutamine fructose-6-amidotransferase, the rate-controlling enzyme of the hexosamine biosynthetic pathway, which produces the substrate for OGT (38,11).
Cells with elevated O-GlcNAc (PUGNAc treatment) consistently demonstrate a delay in G 2 /M progression ( Fig. 2A). In control cells almost 50% of the population had progressed through M phase and returned to G 1 phase 10 h after release. In contrast, only 20% of PUG-NAc-treated cells had returned to G 1 . Conversely, cells with decreased O-GlcNAc (DON) progress through the cell cycle at an accelerated rate compared with control (Fig. 2B). DON-treated cells moved through S phase faster, and by 14 h post-release 81% of the cells had returned to G 1

O-GlcNAc Is a Dynamic Regulator of the Cell Cycle
phase compared with only 53% in controls. Glucosamine is a downstream metabolite of glutamine fructose-6-amidotransferase and should ablate the effect of DON. When 10 mM glucosamine is added with DON, cells progress at rates similar to control or PUGNAc-treated cells (Supplemental Fig. 1).
To determine whether DON treatment accelerated S-phase progression, DNA synthesis was measured in synchronized HeLa cells by [ 3 H]thymidine incorporation (Fig. 2C). DON-treated cells reached maximal thymidine incorporation at 4 h, whereas control peaked at 5 h, and PUGNAc-treated cells peaked at 6 h.
Next, we repeated these experiments in an independent model of cell cycle synchronization that does not relay on chemical block. 3T3-L1 preadipocytes cells when grown to confluency become growth-quies-cent. These cells are hormonally induced to differentiate into mature adipocytes. Induction of differentiation is first characterized by two rounds of synchronized cell cycle progression called mitotic clonal expansion (39). Quiescent preadipocytes cells were treated with 100 M PUGNAc or 25 M DON 12 h before induction of mitotic clonal expansion and again at induction. Cells with elevated O-GlcNAc (PUGNActreated) consistently showed significant delays in cell cycle progression compared with non-treated controls (Fig. 3A). The delay corresponded with the results seen earlier in the HeLa cells (Fig. 2).
Cells treated with DON failed to escape the G 0 /G 1 phase of the cell cycle. The addition of DON to 3T3-L1 cells at the time of mitotic clonal expansion induction or 12 h post-release in late G 1 failed to escape G 1 . However, DON added at 16 h (start of S phase) post-release did not interfere with 3T3-L1 cell growth (data not shown). This is consistent with HeLa cells, which show defects in G 1 to S phase transition when treated with DON (data not shown).
Similar to HeLa cells, PUGNAc treatment altered the expression of O-GlcNAcase and OGT (Fig. 3B). Uninfected (data not shown) and GFP-infected cells exit G 1 /S and reach G 2 /M phase between 9 and 12 h, and by 15 h a large majority of the cells had returned to G 1 phase (Fig. 4A). In the vO-GlcNAcase-infected cells, a significant defect in M phase progression was observed (Fig. 4A). At 15 h, when more than 65% of control cells had returned to G 1 phase, 55% of vO-GlcNAcase-infected cells were still in M phase. This result was repeatedly seen in at least five independent experiments. Interestingly, a slight decrease in the length of S phase was observed in vO-GlcNAcase-infected cells, but the difference to GFP cells was not significant. Similar to vO-GlcNAcase, vOGT-overexpressing cells exhibit a prolonged M phase, but these cells also display severe aneuploidy (Fig.  4B). Upon release from the G 1 /S block, ϳ20% of the cells showed signs of aneuploidy, and the aneuploid nature of the cells increased upwards of 30% as the cells progressed though S and M phase. Uninfected and GPF-expressing cells showed only a small population of aneuploid cells (Ͻ2%).

Overexpression of O-GlcNAcase and O-GlcNAc Transferase Disrupt
Mitotic Phosphorylation-Next, the effect of overexpressing OGT and O-GlcNAcase on O-glycosylation and mitotic phosphorylation was assessed. Overall levels of O-GlcNAc decrease in vO-GlcNAcase-infected cells ϳ10 -15% (Fig. 5A, left panel), whereas overexpressed vOGT causes a slight increase in O-GlcNAc levels (Fig. 5B, left panel) and a substantial increase on an unidentified 55 kDa protein appears.
Proline-directed phosphorylation, as determined by MPM2 staining, is reduced, and maximal phosphorylation lags compared with GFP cells when vO-GlcNAcase is overexpressed (Fig. 5A, right panel). The red box in the right panel of Fig. 5A is a longer exposure showing the dramatic difference in proline-directed phosphorylation after vO-GlcNAcase treatment in the 45-75-kDa range. When vOGT is overexpressed, cells never reach maximal mitotic phosphorylation; however, a 70-kDa band appears phosphorylated throughout the time course and is not present in the uninfected and GFP lanes (Fig. 5B, right panel). At G 1 /S, vO-GlcNAcase-and vOGT-infected cells are in a higher percentage of M phase than GFP (Fig. 4, A and B, at time 0) as judged by flow cytometry; hence, the level of proline-directed phosphorylation is higher (Fig.  5, A, red panel, and B). Both mitotic kinases cyclin-dependent kinase 1 and 2 were assayed for activity (Supplemental Fig. 2) from vO-Glc-NAcase-infected cells, and no dramatic changes in the level of activity during the cell cycle is found, suggesting overexpression is not affecting in vitro cyclin-dependent kinase activity.
The expression of the key G 1 /S phase regulator protein pRb is unaffected by viral infection (Fig. 6). This was not surprising since the levels of this protein remain constant during the cell cycle. The pRb protein is regulated by phosphorylation on threonine 826 (ppRb-T826) at late G 1 by a cyclin-dependent kinase 4-cyclin D complex and disrupts its binding to E2F family proteins (40,41). The vO-GlcNAcase-infected cells show delayed and decreased pRb phosphorylation when compared with uninfected/GFP-infected cells (Fig. 6). Interestingly, vOGT-infected cells reach normal levels of phosphorylation at ppRB-T826, but staining fails to decline during the time course.
Cyclin Expression Is Altered after Modulation of O-GlcNAc Levels-Next, expressionlevelsofseveralcyclinsweremeasured.InGFP-infectedcells,cyclins A and B show peak expression during M phase (9-12 h post-release) and a subsequent decline post-M phase (Fig. 7A), consistent with published data (1). In both vO-GlcNAcase-and vOGT-infected cells, protein levels of cyclins A and B repeatedly fail to decline post-M phase (Fig. 7A). Interestingly, expression of cyclins A and B was prolonged in 3T3-L1 cells treated with PUGNAc (Fig.  7B).AsimilarpatternofcyclinexpressionwasseeninHeLacellsafterPUGNAc treatment (Supplemental Fig. 3).
In GFP and uninfected cells, cyclin E expression is maximal at release, declines throughout M phase, and increases again as cells enter G 1 . In the vO-GlcNAcase-infected cells, cyclin E protein levels decline at a slightly faster rate than control/GFP cells upon thymidine release; however, as the cells reentered G 1 , little cyclin E expression is seen. Cyclin E expression is severely depressed in vOGT-overexpressing cells (Fig. 7A). The G 1 cyclin D in both sets of infected cells never fully reaches maximal expression compared with GFP controls. Normally, the expression of the cyclin-dependent kinase inhibitor p21 increases in cells as they enter G 1 , but p21 fails to express in both vO-GlcNAcase-and OGT-overexpressing cells (Fig. 7A). Cyclins are regulated by cellular localization (2), but the viral treatments appear to have little effect on cytolocalization of the cyclins (Supplemental Fig. 4).
Increased O-GlcNAc Transferase Levels Causes Abnormal Cytokinesis and Aneuploidy-We further explored the M phase defects using confocal microscopy to determine the cytolocalization of O-GlcNAcase and OGT at G 1 /S and M phase and what effects overexpression of these enzymes have on ␣-tubulin and DNA morphology. Synchronized HeLa    SEPTEMBER 23, 2005 • VOLUME 280 • NUMBER 38 cells were stained for DNA (red), ␣-tubulin (green), and either O-Glc-NAcase or OGT (blue). At G 1 /S, control cells predominantly exhibited cytoplasmic staining for O-GlcNAcase and nuclear/perinuclear staining for OGT (Fig. 8A). At M phase, O-GlcNAcase appears ubiquitous throughout the cell, but the enzyme is mostly excluded from the nascent nuclear membranes of the daughter cells (Fig. 8B). Two different antibodies against O-GlcNAcase and OGT were used, and each antibody set gave the same result (Supplemental Fig. 5). Additionally, the preimmune sera for each antibody had little to no staining (Supplemental Fig. 5).

O-GlcNAc Is a Dynamic Regulator of the Cell Cycle
When vO-GlcNAcase is expressed, the cell morphology dramatically changes. First, the DNA appears more condensed compared with controls, and the ␣-tubulin network is dramatically perturbed. The cells tend to round up, and the ␣-tubulin forms a meshwork around the nucleus (Fig. 8A, second column, see arrow). At least 20% of the infected cells consistently show this phenotype. Previous work suggested that ␣-tubulin is modified by O-GlcNAc (42); therefore, vO-GlcNAcase overexpression may alter O-GlcNAc levels on tubulin and disrupt proper function.
In uninfected cells, OGT staining appears ubiquitous throughout the cell during interphase, but OGT localization is concentrated with the mitotic spindle during M phase (Fig. 9). At prophase, OGT becomes concentrated during the nascent formation of the mitotic spindle. A clearly defined increase in localization with the spindle is seen during metaphase. As the daughter chromosomes appear to pull apart during anaphase, OGT is found concentrated at the central spindle assembly. As the cleavage furrow forms during cytokinesis, OGT becomes concentrated at the midbody.
HeLa cells expressing vOGT exhibited multiple phenotypes at G 1 /S. Approximately 15-20% of the cells are aneuploid, whereas other cells still have visible midbodies (Fig. 8A, far right column, see the arrow). A small percentage of cells round up like the vO-GlcNAcase-infected cells, but no visible condensation of the DNA is present nor does the ␣-tubulin network appear concentrated in a concentric ring around the nuclear membrane.
Co-localization of OGT with ␣-tubulin was observed by the presence of a light green color around the spindle and later within the midbody (Fig. 8B, column 1, see the arrow). When vOGT is overexpressed, a large population of cells (ϳ20%) repeatedly shows abnormal midbodies. Increased midbody staining is seen with even more OGT concentrated in this region; the ␣-tubulin polymers at the midbody appeared larger, and midbodies fail to separate and break off. The actin ring at the midbody cleavage furrow is intact, but the cells still fail to cytokinese properly (Supplemental Fig. 6).

DISCUSSION
Cell growth and division are the result of carefully coordinated events, which are regulated by protein expression, protein degradation, and protein phosphorylation. In this study we show that dynamic intracellular glycosylation, O-GlcNAc, also plays a role in cell cycle progression and cell division. The salient findings of this paper include the following: 1) Elevating or reducing O-GlcNAc levels results in changes in cell cycle length; 2) pharmacological modulation of global O-Gl-cNAc levels induces changes in the expression of OGT and O-Glc-NAcase; 3) overexpression of vOGT or vO-GlcNAcase disrupts mitotic phosphorylation and the proper, timed expression of cyclin proteins; 4) OGT localizes to the mitotic spindle and midbody during cell division; 5) overexpression of OGT results in aneuploidy, due to defective cytokinesis.
Dynamic Glycosylation and Phosphorylation Regulate Cell Cycle Progression-Pharmacological increase (PUGNAc) in O-GlcNAc causes cell growth delays in multiple cell lines. Previously, PUGNAc was reported to have no effect on growth rates (27). In this study cells were treated with 100 M PUGNAc every 12 h, whereas it was administered every 48 h in the previous study. The effect on growth may result in changes to cell cycle length. Flow cytometry analysis on PUGNActreated HeLa and 3T3-L1 cells support this observation (Figs. 2 and 3). These data concurred with previous reports demonstrating that increased UDP-GlcNAc levels slowed cell growth and differentiation in human colon cancer cells (43). A good example of this model was seen in Xenopus laevis oocytes (44). Several nuclear pore proteins are modified by O-GlcNAc (45), and these proteins are phosphorylated as oocytes progress from S to M phase (44). These proteins are mutually modified by O-GlcNAc and O-phosphate (44), but when O-GlcNAc proteins are capped with galactose by microinjecting galatosyltransferase, severe defects in M-to S-phase transition occurs (25). Furthermore, when oocytes are preincubated in PUGNAc before hormone-induced maturation, nuclear envelope breakdown is significantly delayed (28).
Like proline-directed phosphorylation, the O-GlcNAc protein modification is regulated in a cell cycle-dependent manner. Global O-Glc-NAc levels after nocodazole synchronization decreased during M phase, when proline-directed phosphorylation, a hallmark of mitosis, peaks. However, an increase in glycosylation is seen on some protein bands. Previous work did not see changes in glycosylation after nocodazole synchronization (33); however, our study employed the 110.6 antibody that is highly specific for the O-GlcNAc modification.
In support of this, we looked at the glycosylation and expression status of Sp1 and YY1. YY1 is ubiquitously expressed transcription factor that is heavily modified by post-translational modifications including phosphorylation, acetylation, and O-GlcNAc (36, 37, and 46). O-GlcNAc levels on YY1 appear dynamically responsive in a cell cycle-dependent manor, possibly influencing specific protein-protein interactions and gene transcription. When YY1 is O-GlcNAc-modified, it is unable to associate with pRb-promoting increased binding of YY1 to specific promoters (36). Maximal binding of YY1 to pRb is found at G 0 /G 1 (36). In this study we observed an increase in YY1 O-GlcNAc modification at M phase (nocodazole synchronization), which is when pRb is hyperphosphorylated and unable to interact with promoters. Additional support for dynamic cycling of the sugar modification can be found in COPII protein cycling. COPII proteins (specifically Sec24p) involved in endoplasmic reticulum-Golgi transport are phosphorylated at M phase, disrupting membrane binding; conversely, during interphase the protein is modified by O-GlcNAc and is functional (47). Of course, not all O-GlcNAc modified proteins demonstrate a decrease in O-GlcNAc at M phase. Previously, glycosylation of keratin 18 increased after nocodazole synchronization, whereas keratin 8 demonstrated an increase in phosphorylation after the same treatment (48,33).
Lowering O-GlcNAc levels with DON leads to defects in G 0 /G 1 progression. Although 3T3-L1 cells show G 0 /G 1 arrest, HeLa cells show defects in G 1 progression after DON treatment. These data are supported in OGT knockouts of fibroblast. Cells show delays in cell cycle progression and eventually become quiescent (22). Together, these data support a model in which low nutrients would result in low O-GlcNAc levels, and this would induce G 0 /G 1 cell cycle arrest until nutrients are available again. Cells are most sensitive to low nutrient levels before G 1 release since most of G 1 is spent preparing the cells for replication and division (2).
Consistent with dynamic modulation of O-GlcNAc levels regulating cell cycle progression, once past the G 1 checkpoint, lowering O-GlcNAc  SEPTEMBER 23, 2005 • VOLUME 280 • NUMBER 38 levels with DON accelerates S phase. A potential explanation for accelerated DNA synthesis is that O-GlcNAc-modified proteins are concentrated at condensed chromatin compared with transcriptionally active regions (49). As DNA is replicated, transcriptionally active euchromatin is replicated before the more condensed heterochromatin (50). O-Glc-NAc on chromatin proteins might promote condensation of DNA. Histone deacetylase 1, which deacetylates histones and promotes histone-DNA interaction, is modified by O-GlcNAc, and found in a transcriptional repressor complex with mSin3A and OGT (51). Additionally, O-GlcNAcase is a histone acetyltransferase capable of relaxing histone structure (52,53). Together these data suggests that OGT and O-GlcNAcase may affect DNA synthesis by altering chromosomal structure.

O-GlcNAc Is a Dynamic Regulator of the Cell Cycle
Even though DON is commonly used to lower O-GlcNAc levels in cells (11,28,38), it is a glutamine analogue and could potentially have side effects. DON-mediated reductions in the DNA synthesis and growth of neoplastic cell lines (54) are attributed to inhibition of de novo purine and pyrimidine biosynthesis (55). To exclude this possibility, our experiments were performed in media supplemented with 4 mM gluta-mine. Additionally, accelerated DNA synthesis is ablated upon incubation of double thymidine-blocked HeLa cells with glucosamine, which would circumvent DON inactivation of glutamine fructose-6-amidotransferase, suggesting the affects seen with DON can be partially attributed to changes in O-GlcNAc levels (Supplemental Fig. 1).
O-GlcNAc Levels Control the Expression of O-GlcNAcase and OGT-Cells maintain a steady state level of O-GlcNAc by controlling the protein expression of the O-GlcNAc-processing enzymes. PUGNAc-treated HeLa and 3T3-L1 cells caused a decrease in OGT levels and an increase in O-GlcNAcase levels, whereas DON treatment in 3T3-L1 cells caused a reciprocal change. In a cellular system finely tuned to react to changes in the environment, this feedback regulation either at the transcriptional or translational level is not surprising. Loss of the ability of the cells to maintain O-GlcNAc levels could lead to growth quiescence and apoptosis or contribute to the etiology of numerous metabolic diseases such as cancer, diabetes, and Alzheimer disease (9). For example, elevated O-GlcNAcase activity and decreased O-GlcNAc levels were found in human breast cancer tissue (56). Tau, a major component of paired-helical filamentous fibers in Alzheimer patients, is modified with O-GlcNAc. A significant loss in O-glycosylation is seen in post-mortem brains of Alzheimer FIGURE 9. OGT localization during M phase. A, confocal microscopy was performed on doublethymidine-blocked HeLa cells 10 h after serum release. Al-28 staining for OGT is the first panel; the middle panel is DNA-stained with propidium iodide, and the third panel is the merge. Prophase cells demonstrate OGT staining at the nascent formation of the mitotic spindle. At metaphase, OGT localization is increased at the mitotic spindle. During anaphase, OGT is found within the central spindle assembly, and during cytokinesis OGT is localized to the midbody. B, the AL-28 pre-immune staining is shown as a negative control. disease patients, suggesting a loss of O-glycosylation concomitant with increased phosphorylation as the disease progresses (57,58).
O-GlcNAcase and O-GlcNAc Transferase Control M-phase Phosphorylation-Overexpression of both vO-GlcNAcase and vOGT influences mitotic phosphorylation. The delay in maximal M phase phosphorylation after vO-GlcNAcase infection mirrors the flow cytometry data in which a large population of cells was delayed in exiting M phase compared with control. Similarly, the higher levels of mitotic phosphorylation after vOGT infection agree with the flow cytometry data showing that a large population of cells is in M phase after G 1 release.
Phosphorylation of pRb, a specific protein involved in cell cycle regulation, is disrupted upon overexpression of vO-GlcNAcase and vOGT. Phosphorylation of pRb mirrored the proline-directed phosphorylation differences seen upon overexpression. This difference might be due to changes in specific protein-protein interactions between cell cycle proteins and vO-GlcNAcase/vOGT. These data strongly suggest that alterations in the expression of O-GlcNAcase or OGT lead to a mitotic exit phenotype. This phenotype could be caused by alterations in the expression of the cyclins.
O-GlcNAc as a Mediator of Cyclin Levels-As cells move through the cell cycle, the periodic expression of cyclins activates specific cyclin-dependent kinases, which in turn phosphorylate specific cell cycle substrates (2). The overexpression of both O-GlcNAcase and OGT disrupts proper cyclin periodicity. As HeLa cells are released at G 1 /S, the S-phase cyclin, cyclin E, is expressed. At release, vO-GlcNAcase-expressing cells have similar levels of cyclin E to control, but these levels decrease quicker than uninfected or control GFP cells. This corroborates the previous data in which decreased O-GlcNAc levels after DON treatment accelerated S phase. However, vO-GlcNAcase-expressing cells display reduced levels of cyclin E compared with control at the later G 1 time points, suggesting errors in M phase release. The expression of vOGT severely disrupts cyclin E expression. These cells express low levels of cyclin E at the G 1 release points and fail to re-express the cyclin at later time points, strengthening the argument for mitotic exit errors in these cells. The mitotic cyclins A and B show prolonged expression after PUGNAc treatment or vO-GlcNAcase and vOGT infection (Figs. 3 and 6). Again, these data strongly suggest delays or errors in proper M-phase progression in these cells. Furthermore, the disruption in cyclin expression could potentially lead to the aberrant phosphorylation seen in vO-GlcNAcase-and vOGT-expressing cells.
Control of the cyclin oscillator is through timed destruction of the appropriate cyclin by the proteasome (2), several components of which are modified by O-GlcNAc (59). Increased O-GlcNAc protein modification correlates with reduced proteasome function (60). In the case of vO-GlcNAcase overexpression, the prolonged expression of the cyclins is likely not due to the improper function of the proteasome since decreased O-GlcNAc does not inhibit the proteasome (60). However, this cannot be ruled out for vOGT overexpression and PUGNAc treatment.
Overexpression of OGT Interferes with Proper Cytokinesis-One potential mechanism for the increased number of cells in M-phase after vOGT overexpression is due to the inability of these cells to properly cytokinese. Notably, OGT co-localized with the central spindle assembly at M phase. Many MPM2-reactive epitopes are localized to the centromere along with several kinases such as POLO-like kinase and Aurora B (61,62). The presence of OGT at the spindle assembly suggests that OGT acts in concert with mitotic kinases to delicately regulate cell division. OGT could dynamically add O-GlcNAc to specific proteins or act as a scaffold through its N-terminal TPR domain (63) to bring assembly proteins together. Several proline-directed phosphatases localize to the spindle assembly, such as CDC-14 (64), and control segregation of chromosomes by interacting with microtubule motor proteins (65). OGT is known to interact with protein phosphatase 1␤ and 1␥ (66); therefore, OGT could act as an assembly protein recruiting phosphatases to the spindle assembly complex. Disruption of the interactions of OGT with components of the spindle assembly by overexpression of vOGT could lead to the increased number of cells in M phase. This is supported by the aberrant cellular morphology seen during cytokinesis.
Deciding to divide and control the timing of these events is key to proper cellular function. O-GlcNAc, a metabolic nutrient and stress sensor, provides the cell with a mechanism to integrate multiple cellular signals to discriminatingly regulate cell cycle events. Coordinated regulation of the O-GlcNAc-processing enzymes would finely tune cells to environmental signals, allowing for proper cell cycle progression and controlled cytokinesis. Failure to properly regulate O-GlcNAc could then lead to cell cycle errors and diseases such as cancer and diabetes.