Down-regulation of Sp1 Activity through Modulation of O-Glycosylation by Treatment with a Low Glucose Mimetic, 2-Deoxyglucose*

2-Deoxyglucose (2-DG), a nonmetabolizable glucose analogue, blocks glycolysis at the phosphohexose isomerase step and has been frequently used as a glucose starvation mimetic in studies of a wide variety of physiological dysfuctions. However, the effect of 2-DG on protein glycosylation and related signal pathways has not been investigated in depth. In HeLa, an HPV18-positive cervical carcinoma line, 2-DG treatment down-regulates human papillomavirus early gene transcription. This down-regulation was also achieved by low glucose supply or hypoxia, suggesting that this is a response commonly modulated by cellular glucose or energy level. We investigated how 2-DG and low glucose affect transcriptional activity. Human papillomavirus gene transcription was only marginally affected by the inhibition of ATP synthesis or the supplementation of pyruvate to 2-DG-treated cells, suggesting that poor ATP generation is involved only to a limited extent. 2-DG treatment also inhibited activation of p21 WAF1 promoter, which is controlled by p53 and/or Sp1. In a reporter assay using p21 WAF1 promoter constructs, 2-DG exerted a strong inhibitory effect on Sp1 activity. DNA binding activity of Sp1 in 2-DG-treated HeLa cells was intact, whereas it was severely impaired in cells incubated in a low glucose medium or in hypoxic condition. Unexpectedly, Sp1 was heavily modified with GlcNAc in 2-DG-treated cells, which is at least partially attributed to the inhibitory effect of 2-DG on N-acetyl-β-d-glucosaminidase activity. Our results suggest that 2-DG, like low glucose or hypoxic condition, down-regulates Sp1 activity, but through hyper-GlcNAcylation instead of hypo-GlcNAcylation.

For most tumor cells and some rapidly proliferating cells, poor blood supply would impose an environmental stress such as nutritional starvation and hypoxia, which evidently leads to apoptotic or necrotic death of cells (1). Glucose deprivationinduced ATP depletion stimulates the mitochondrial death pathway cascade, which involves the loss of mitochondrial membrane potential and cytochrome c release (2)(3)(4) or causes necrotic cell death (5). Also, in glucose-deprived cells, the level of prooxidants, such as superoxide and hydrogen peroxide, increases probably as a result of the metabolic shift to oxidative phosphorylation, and this triggers signal transduction pathways involving the activation of c-Jun N-terminal kinase/mitogen-activated protein kinase that leads to apoptotic cell death (6,7). Both hypoxia and hypoglycemia induce HIF-1␣ in most cells in an effort to restore oxygen homeostatsis by stimulating glycolysis, erythropoiesis, and angiogenesis but also induce G 1 arrest (8) and apoptosis (9) in a manner dependent on HIF-1␣ and p53 in most cell types. In addition, because glucose is a key source for sugar moieties in glycoproteins, glucose deprivation also blocks the completion of protein modification and thereby affects various aspects of cell physiology and biochemical activities mediated by glycoproteins. Furthermore, accumulation in the endoplasmic reticulum of proteins misfolded because of poor glycosylation (unfolded protein response or glucose-regulated stress) is known to induce apoptotic cell death and socalled glucose-regulated proteins such as Grp78 and Grp94, probably as a cellular defense response to endoplasmic reticulum stress (reviewed in Ref. 10).
A variety of glycoprotein linkages have been discovered to form between 13 different sugar molecules and 8 amino acids on proteins (reviewed in Ref. 11). Among them, three types of protein glycosylations are most widely distributed and well characterized. N-Linked glycosylation is transfer of oligosaccharides to asparagine residues en bloc in endoplasmic reticulum, and O-linked glycosylation is transfer of a small number of glucose, galactose, or N-acetylgalactosamine to serine or threonine residues in succession during transport through Golgi network. Another form of O-linked protein glycosylation occurs in a variety of proteins including nuclear pore complex proteins, cytoskeletal proteins, and transcription factors (reviewed in Refs. 12 and 13). These proteins are modified by a single monosaccharide, GlcNAc, at serine or threonine residues. This O-GlcNAc modification (O-GlcNAcylation) is catalyzed by UDP-N-acetylglucosamine:peptide N-acetylglucosaminyl transferase (O-GlcNAc transferase; OGT) 1 (14). UDP-GlcNAc, the substrate of this enzyme, is synthesized from glucose via the hexosamine biosynthetic pathway (15). The availability of UDP-GlcNAc correlates with the glycosylation levels of many intracellular proteins and the transcriptional levels of some genes (16). Because many transcription factors are modified by O-GlcNAc (17), it is likely that O-GlcNAcylation of certain transcription factors could regulate gene expression in response to glucose flux.
In numerous studies, interference with glycolysis or conditions for nutritional deprivation or energy depletion have been simulated in in vitro culture by utilizing 2-deoxyglucose (2-DG) (Refs. 18 -22, for example). 2-DG, a nonmetabolizable glucose analogue, is known to block glycolysis by inhibiting phosphohexose isomerase (23,24). 2-DG also induces glucose-regulated proteins, as does low glucose stress (25,26). 2-DG has been shown to effectively block the growth of tumor cells in animal models (27,28) and of a variety of human tumors cells alone or in combination with hypoxia or other tumor therapy tools (29 -31). 2-DG has also been utilized as a mimetic of calorie restriction in an animal study (32). Despite such an extensive utilization as an effective low glucose mimetic in a wide variety of physiological situations, however, the effect of 2-DG on protein glycosylation and the related signal pathway has not been closely investigated nor significantly considered in previous studies.
In a previous study (33), human papillomavirus (HPV) early gene transcription was shown to be suppressed in HeLa cervical carcinoma cells and its derivative cells by the treatment of 2-DG, and it was postulated that high glucose availability is indispensable for both high glycolysis rate of cervical cancer cells and the maintenance of the carcinogenic state by continued expression of HPV oncoproteins. However, the molecular mechanism by which 2-DG treatment caused down-regulation of HPV gene expression was not investigated. There might be several different routes through which 2-DG causes repression of HPV gene expression: inhibition of glycolysis and ATP production, interference in protein glycosylation and the resulting endoplasmic reticulum stress, and attenuation of activities of certain transcription factors through alterations in its O-Glc-NAcylation level. Transcription of HPV early genes is controlled by a region of HPV DNA called the upstream regulatory region (URR), which contains numerous sequence elements that interact with cellular and viral transcription factors (see Ref. 34 for a recent review).
Using HPV URR as a model, we investigated how 2-DG and hypoxia affect gene transcription. In our study, not only HPV URR but also p21 WAF1 promoter activities were severely suppressed by 2-DG treatment. Through investigation on the possible regulatory events associated with the glucose-linked modulation of HPV URR and p21 WAF1 transcription, we found that Sp1 activity is inhibited and that its glycosylation status is severely altered by the treatment of 2-DG. Our results suggest that Sp1, through O-GlcNAcylation, functions as a molecular switch that modulates both host and viral gene transcription in response to cellular nutritional status. Our study also suggests that hypoxia and low glucose also affect Sp1 activity but through mechanisms different from that of 2-DG.

EXPERIMENTAL PROCEDURES
Cell Culture-Cervical carcinoma cell lines positive for HPV DNA, HeLa (HPV 18), and CaSki (HPV 16); a non-small cell lung carcinoma cell line, H460; and a rat cell line, PC12, were all incubated in normal Dulbecco's modified Eagle's medium (DMEM; Invitrogen) containing 25 mM glucose. All of the media were supplemented with 10% fetal bovine serum. Hypoxic incubation condition was established with an ambient oxygen concentration of 0.01% (using a controlled incubator with CO 2 /O 2 monitoring and CO 2 /N 2 gas sources). For 2-DG treatment, the cells were incubated in DMEM supplemented with 45 mM 2-DG (Sigma). For low glucose treatment, the cells were incubated in DMEM (without glucose; Invitrogen) supplemented with 1 mM glucose. To measure cellular ATP levels, 2 ϫ 10 5 HeLa cells were collected, washed in phosphate-buffered saline, lysed, and applied to an ATP Somatic cell assay kit (Sigma) following the manufacturer's instructions. The luminescence emitted from the ATP-dependent luciferase reaction was monitored in a luminometer and used as the relative ATP level in test cells.
Transient Transfection and Reporter Assay-HeLa cells in 6-well plates were transfected with 2 g/well of pGL2-0.3 plasmids (kindly provided by Dr. J. Y. Lee, Hallym University, Chuncheon, Korea) by using GeneCarry Transfection reagent (GoodGene, Seoul, Korea). 24 h after transfection, the cells were treated with 45 mM 2-DG for 24 h and lysed. The extracts were applied to luciferase activity by using a luciferase assay kit (Promega). The luciferase activities were normalized based on protein concentrations.
Electrophoretic Mobility Shift Assay-The cells were resuspended in buffer A (10 mM Hepes, pH 7.8, 1.5 mM MgCl 2 , 10 mM KCl, 1 mM dithiothreitol, 1 mM PMSF, 10 g/ml leupeptin) for 25 min. After centrifugation, the cells were incubated in 0.1% Triton X-100 for 5 min. The cell pellets were incubated in nuclear protein isolation buffer B (20 mM Hepes, pH 7.8, 25% glycerol, 420 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 1 mM dithiothreitol, 1 mM PMSF, 10 g/ml leupeptin) on ice for 30 min. In binding reactions, radiolabeled probe for Sp1 binding from HPV18 URR (AGGGAGTAACCGAAAACGGT) and 10 g of nuclear extracts were incubated in binding buffer (4 mM Tris-Cl, pH 7.8, 12 mM Hepes, pH 7.9, 60 mM KCl, 30 mM NaCl, 0.1 mM EDTA, 1 mM ZnCl 2 , 1 g of poly(dI-dC), 10% glycerol) for 20 min at 22°C. In supershift experiments, nuclear extract was incubated with 2 g of anti-Sp1 antibody for 30 min at 22°C prior to the addition of the probe. In competition experiments, a 300-fold higher amount of unlabeled probe was incubated with binding mixture containing radiolabeled probe. The reaction products were resolved in 5% nondenaturing polyacrylamide gel in 0.25ϫ TBE buffer at 4°C. The gel was dried and applied to autoradiography. A known Sp1-binding sequence in HPV18 URR (5Ј-AGGGAGTAACCGAAAACGGT-3Ј) was used for the DNA probe. The single-stranded oligomers were annealed by incubation in 80°C for 5 min followed by cooling down slowly to room temperature and labeled by using a Klenow fragment of DNA polymerase and [ 32 P]dGTP (Amersham Biosciences).
Immunoprecipitation-Nuclear extracts were prepared as described under "Electrophoretic Mobility Shift Assay." Equal amounts of protein were incubated overnight at 4°C with anti-Sp1 antibody, followed by overnight incubation with protein A-Sepharose (Amersham Biosciences). The beads were washed three times in nuclear protein isolation buffer B (20 mM Hepes, pH 7.8, 25% glycerol, 420 mM NaCl, 1 mM PMSF), boiled, and spun down. The supernatants were applied to SDS-PAGE followed by Western blotting. Quantitation of the band signals on film (from three independent experiments) was carried out by Gel-Pro analyzer version 3.1 (Media Cybernetics).
In Vitro O-GlcNAcase Assay-PC12 cells were collected in 50 mM Tris-HCl, pH 7.5, containing 150 mM NaCl, 10% glycerol, 2 mM EDTA, 1 mM Na 3 VO 4 , and protease inhibitors. After homogenization by using 26-gauge needle, the samples were centrifuged at 20,000 ϫ g for 30 min at 4°C. The supernatant was collected as O-GlcNAcase-rich cytosols. The cell lysate was incubated first with each chemical (STZ, 2-DG, and glucose) for 30 min at room temperature, and then assay buffer (50 mM sodium cacodylate, pH 6.5, 50 mM N-acetyl galactosamine, 2 mM pnitrophenyl-N-acetyl-D-glucosaminide) was added and incubated for 1 h at 37°C. At the end of the incubation, the reaction was stopped by the addition of 0.5 M sodium carbonate. Color development was measured spectophotometrically at 400 nm. To compare O-GlcNAcase activity levels in 2-DG or mock-treated HeLa cells, the cells were lysed in 1/10 volume of the buffer, and 40-g protein samples were applied to the assay as described above.

Repression of HPV Early Gene Expression by 2-DG Treatment, Hypoxia, or Low Glucose Stress in Cervical Carcinoma
Cells-Previously, 2-DG treatment was reported to down-regulate HPV early gene expression in HPV18-positive HeLa cells (33). The results of RT-PCR and Western blotting showed that 45 mM 2-DG caused a marked reduction in the levels of HPV16 E6/E7 transcripts and proteins in HPV16-positive CaSki cells as well (Fig. 1, A and B). This suggests that 2-DG acts through modulation on a common factor that plays a critical role in activation of URR of a variety of HPV types. Fig. 1 also shows that HPV gene expression in HeLa cells is down-regulated by hypoxia or low glucose supply (1 mM) as well ( Fig. 1, C and D), suggesting that HPV transcription may be modulated by the cellular energy or glucose level. During a 24-h incubation period, approximately one-fourth of cells in the culture treated with 2-DG or low glucose underwent either apoptotic or necrotic death, and the remaining cells were viable but growtharrested. Meanwhile, hypoxic cells remained growth-arrested during first 24 h but underwent extensive apoptosis afterward (data not shown).
Effect of Inhibition in ATP Generation on HPV Early Gene Expression-In 2-DG-treated cells, the ATP level is severely reduced (33). Whether the reduced ATP level exerts an inhibitory effect on HPV URR was determined by testing whether a direct inhibition of ATP generation down-regulates HPV UPP. HeLa cells were treated with a complex II inhibitor, thenoyltrifluoroacetone (TTFA; 100 M) (5), or a cytochrome c oxidase inhibitor, sodium azide (NaN 3 ; 1 mM) (36). TTFA treatment caused a 30% reduction of cellular ATP level (Fig. 2B). 2-DG treatment also caused a 33% reduction. Despite this similar degree of repression in ATP levels, there was no significant reduction in HPV E7 protein level in TTFA-treated cells, whereas it was almost extinct in 2-DG-treated cells ( Fig. 2A, lane 4 versus lane 2), and in NaN 3 -treated cells, where cellular ATP synthesis was almost abolished (7.3% of untreated cells), the HPV E7 level was reduced but still maintained by nearly 40% of that in untreated cells (lane 3). (This is a significantly high level considering a global decrease in cellular protein levels upon NaN 3 treatment.) These results suggest that a decrease in ATP synthesis may cause some reduction in HPV gene transcription, but its effect is only marginal. This was confirmed by an RT-PCR experiment, in which the HPV E6 mRNA level in TTFA-treated cells was only moderately reduced (data not shown). Therefore, the effect of 2-DG on HPV gene expression may in a small part be attributed to a low ATP level but for the most part originates from mechanisms not related to ATP status.
Effect of 2-DG on p21 WAF1 Promoter Activity-2-DG-induced inhibition of E6/E7 expression caused up-regulation of p53 (Ref. 32 and Fig. 3, A and B). This is most likely through increased stability of p53 resulting from depletion of E6 protein, which induces ubiquitin-mediated proteolysis of p53 (37). Meanwhile, the level of E2F1 protein and its activity as determined by c-Myc expression level were reduced upon 2-DG treatment (Fig. 3A), most likely because of the depletion of E7, which has been shown to activate E2F1 either directly (38) or through inactivation of Rb (39). Therefore, the status of these factors is as expected in HeLa cells where E6/E7 proteins were depleted. Similar changes have been induced in HeLa cells where E6/E7 genes were repressed by the expression of BPV E2, which blocks transcription of HPV early promoter (Refs. 40 and 41 and Fig. 3C). Depletion of E6 and E7 in HeLa cells transduced with BPV E2 lead to an increase in the steady state level of p53 followed by an induction of p21 WAF1, a p53 response gene (Fig. 3C, lane 2). In contrast, the p21 WAF1 was not induced in 2-DG-treated cells (Fig. 3, A and B). This is not because p53 is transcriptionally inactive because MDM2 and Bax1, other p53 response genes, were up-regulated in parallel to the increase of p53 level (Fig. 3B). These results suggest that certain element(s) present in the upstream of p21 WAF1 gene is not properly responding to p53. This assumption was further supported by the results in Fig. 3 (D and E). Treatment of sodium butyrate, an inhibitor of histone deacetylase (42), induces p21 WAF1 independently of p53 (43). This indeed is the case in HeLa cells as shown in Fig. 3D (lane 3), but treatment of cells with 2-DG reduced this effect of butyrate (lane 4). Therefore, 2-DG represses a p21 WAF1 promoter activity that is independent of p53. In a separate experiment, doxorubicin treatment of H460 cells, a lung carcinoma line with a wild-type p53, induced p53 and accordingly p21 WAF1 (Fig. 3E, lane 3). However, 2-DG treatment abolished the effect of p53 on p21 WAF1 expression without affecting the level of p53 (lane 4). Therefore, 2-DG represses p53-dependent p21 WAF1 promoter activity as well. Overall, these results suggest that a transcription factor that functions either by itself or in cooperation with p53 in activation of p21 WAF1 promoter is affected by 2-DG treatment.
Sp1 Activity Is Repressed by 2-DG Treatment-In addition to p53, Sp1 plays a critical role in activation of p21 WAF1 promoter (44). In fact, the effect of sodium butyrate on p21 WAF1 transcription is through modulation of the availability of Sp1 to its binding sites within the proximal region of the promoter (43). Furthermore, there are sequence elements in the promoter proximal region of HPV URR that bind with Sp1 and regulate HPV early gene transcription in Sp1-dependent manner (45,46). We tested whether Sp1 and the Sp1 sites in p21 WAF1 upstream respond to 2-DG by transfecting HeLa cells with a reporter construct derived from the upstream region of the p21 WAF1 promoter. The plasmid lacks a distal region of p21 WAF1 upstream that contains p53-binding sites but has the proximal region that contains six putative Sp1 sites (Fig.  4A). As predicted, butyrate caused a strong activation of this promoter (Fig. 4B). However, treatment of 2-DG almost abolished the butyrate-mediated increase in the reporter activity (Fig. 4B). This confirms the previous Western blotting results demonstrating that Sp1 activity on p21 WAF1 transcription is affected by 2-DG treatment.
Effect of 2-DG, Low Glucose, and Hypoxia on DNA Binding Activity of Sp1-To determine whether Sp1 activity is indeed attenuated in the cells treated with 2-DG and to find the  4) for 20 h. The cells were collected and applied to Western blotting for HPV18 E7. Erk-1/2 protein bands served as a loading control. B, 2 ϫ 10 5 HeLa cells were treated as in A and lysed to measure cellular ATP content. The values of luminescence measured from three independently treated samples were averaged and listed in the second row. In the third row, relative luminescence compared with that of cells treated with vehicle only was expressed as a relative ATP level. a denotes that these values are not statistically different from each other through one-way analysis of variance test.

2-Deoxyglucose Down-regulates Sp1 Activity
underlying reasons, the specific DNA binding activity of Sp1 was tested by electrophoretic mobility shift assay using 20-mer double-stranded oligonucleotides that contain Sp1-binding sequences in HPV 18 URR. As shown in Fig. 5, nuclear extract from untreated HeLa cells had an activity that specifically interacted with the probe DNA (lanes 1-3). Incubation of cells in hypoxic conditions nearly abolished the DNA binding activity of Sp1 (lanes 8), suggesting that hypoxia-induced downregulation of HPV gene expression as shown in Fig. 1C is through the loss of DNA binding activity of Sp1. Incubation of cells in low glucose conditions (1 mM) caused a moderate reduction in DNA binding activity (lane 7). In contrast, 2-DG had no apparent effect on the DNA binding of Sp1 (lanes 4 and 6). Therefore, different from low glucose and hypoxia, 2-DG does not affect the DNA binding activity of Sp1.
Effect of 2-DG, Low Glucose, and Hypoxia on O-linked Glycosylation of Sp1-Sp1 is one of the nuclear proteins that are glycosylated at serine/threonine residues with N-acetylglucosamine (O-GlcNAcylation) (47). It has been suggested that Sp1 activity is regulated by a fine balance between O-GlcNAcylation and phosphorylation at overlapping residues especially at the N terminus (48). Therefore, we assumed that 2-DG might attenuate Sp1 activity by inducing hypo-GlcNAcylation through inhibition of UDP-GlcNAc synthesis and supply. To determine whether this is the case, the O-GlcNAcylation status of Sp1 was examined by using RL2, an antibody that recognizes O-GlcNAc moiety linked to serine residues (49). After treatment of 2-DG, nuclear Sp1 proteins were immunoprecipitated and subjected to Western blotting for RL2 or Sp1 (Fig. 6A). The Sp1 protein level itself was not significantly affected by 2-DG treatment. However, to our surprise, Sp1 O-GlcNAcylation was dramatically increased in 2 h after 2-DG treatment (4.2-fold), peaked at 4 h post-treatment (5.7-fold), and was maintained over 24 h (Fig. 6D). In this time course study, a decrease in E7 protein level was detected immediately following the increase of Sp1 O-GlcNAcylation (from 2 to 4 h). Meanwhile, O-Glc-NAcylation status of Sp1 in cells incubated in hypoxic condition remained rather constant (Fig. 6B). Interestingly, levels of E7 decreased at later time points, being prominent only after 16 h in incubation at hypoxic conditions, clearly demonstrating a difference in kinetics between the effects exerted by 2-DG and hypoxic treatments. In cells incubated in low glucose medium, both O-GlcNAcylation status and the steady state level of Sp1 protein slowly decreased, possibly because of a slow decrease in the cellular glucose level (Fig. 6, C and D). These decreases in both the O-GlcNAcylation level and the steady state level of Sp1 reflect the previous finding that reduced O-GlcNAcylation  causes proteosome-mediated degradation of Sp1 protein in glucose-starved cells (50).  (Fig. 7B). The cell lysates were preincubated with either STZ or 2-DG for 30 min before an artificial substrate p-nitrophenyl N-acetyl-␤-D-glucosamine was added. The effect on the O-GlcNAcase was determined by measuring the degree of GlcNAc removal through a colorimetric method. 50 mM 2-DG caused a reduction in the enzyme activity by ϳ40%, and 100 mM 2-DG caused a 60% decrease in activity. This demonstrates that 2-DG indeed has an inhibitory effect on O-GlcNAcase, albeit one less potent than STZ. In addition, the relative effectiveness of 2-DG as an O-GlcNAcase inhibitor and the relative effectiveness of 2-DG as a protein O-GlcNAcylation stimulator shown in Fig. 7A are quite similar to each other, indicating that the increase in protein O-Glc-NAcylation by 2-DG treatment is in large part due to its inhibition on O-GlcNAcase activity. Importantly, this suggests that the hyper-GlcNAcylated state of Sp1 in the 2-DG-treated cells can be at least in part explained by the inhibition of deGlc-NAcylation by 2-DG.

Inhibitory Effect of 2-DG on O-GlcNAcase-The finding that 2-DG causes an increase in O-GlcNAcylation level in
Effect of STZ on the Status of Sp1 O-GlcNAcylation and Activity-Another important question raised is whether this O-GlcNAcylation of Sp1 is indeed responsible for the decrease of Sp1 activity and transcription of HPV early gene and p21 WAF1. It has been reported that O-GlcNAcylation of the Sp1 activation domain down-regulates its transcriptional activity (54). Along with our previous results, this suggests that the down-regulation of HPV URR and p21 WAF1 promoter in 2-DG-treated cells might be attributed to Sp1 hyperglycosylation. Based on the finding that STZ treatment, like 2-DG, causes protein hyper-GlcNAcylation (Fig. 7A), we attempted to check whether Sp1 is hyper-GlcNAcyated in the presence of STZ and whether its transcriptional activity is affected by STZ treatment. Therefore, HeLa cells were incubated in the presence of 10 mM STZ, and its effects on Sp1 O-GlcNAcylation, p21 WAF1 promoter activity, and HPV E7 expression were determined. As expected, STZ treatment caused an increase in the level of Sp1 O-GlcNAcylation without affecting the steady state level of the protein (Fig. 8A), and the extent of O-GlcNAcylation is quite similar to that caused by 2- DG (lane 2 versus lanes 3  and 4). Furthermore, E7 expression and p53-mediated expression of p21 WAF1 were both down-regulated by STZ treatment as well (Fig. 8A). Treatment of H460 cells with STZ caused an increase in p53 protein level and a slight increase of p21 WAF1 (Fig. 8B, lane 3), whereas 2-DG treatment did not (lane 2). However, STZ suppressed the p53-mediated activation of p21 WAF1 promoter in the cells treated with doxorubicin (lane 6), as did 2-DG (lane 5). Overall, these results suggest that STZ causes Sp1 hyper-GlcNAcylation and thereby represses its activity. The quite similar outcomes of the 2-DG and STZ treatments, i.e. hyper-GlcNAcylation of Sp1 and down-regulation of the promoters that is dependent on Sp1, strongly indicate that 2-DG and STZ both inhibit Sp1 activity by causing its hyper-GlcNAcylation.
Effect of 2-DG on Sp1 Phosphorylation Status in Vivo-It has been suggested that O-GlcNAcylation and phosphorylation occur on identical serine/threonine residues in certain proteins known to be O-GlcNAcylated (55,56), and in Sp1, O-GlcNAcylation and phosphorylation have opposite effects on its activity (48). Whether Sp1 hyper-GlcNAcylation in the cells treated with 2-DG is accompanied by its hypophosphorylation was determined using an antibody that recognizes phosphorylated threonine residues. (Phosphorylation status at serine residues could not be determined because no antibody specific for phosphorylated serine worked in Western blotting.) As shown in Fig. 9, the level of Sp1 phosphorylation at threonine residues decreased as its O-GlcNAcylation level increased, and these two changes appeared quite reciprocal. This and the previous results suggest that, first, Sp1 proteins in the 2-DG-treated cells are in hyper-GlcNAcylated state, and second, this high O-GlcNAc level functionally blocks phosphorylation of Sp1. Furthermore, our results strongly support the notion that Sp1 is in inactive state because of hyper-GlcNAcylation in cells treated with 2-DG. DISCUSSION Our results indicate that 2-DG causes repression of certain promoters by inducing hyper-O-GlcNAcylation of Sp1. Because Sp1 is a ubiquitous transcription factor regulating TATA-less housekeeping genes, a variety of cellular genes are expected to be affected by 2-DG. The 2-DG-induced blocks to glycolysis and energy depletion appear to be minimally involved in this transcriptional dysfunction. Incubation of cells in low glucose or hypoxic condition also affects the activity of Sp1 (this study and Ref. 57). Thus, when 2-DG or low glucose treatment to cells or animals is applied as a mimetic of cellular energy deprivation, glycolysis block, or an unfolded protein response, the effect on Sp1 activity should be considered. Our results also demonstrate that the p21 WAF1 promoter activity is modulated by Sp1 O-GlcNAcylation. Previous DNA transfection experiments using the Sp1-less Drosophila cell line SL2, reported a severe reduction in constitutive and p53-inducible p21 promoter activity by the mutations in the Sp1 binding site (58), a result suggesting that Sp1 must play crucial roles in the activity of p53 on the p21 WAF1 promoter. Our results show, in a more appropriate cellular context, that attenuation of Sp1 activity causes a severe reduction in p53-inducible p21 WAF1 promoter activity.
Although both 2-DG and low glucose cause down-regulation of Sp1 activity, they modulate Sp1 through different mechanisms. 2-DG treatment down-regulates Sp1 activity without alleviating its DNA binding activity, whereas low glucose causes attenuation of DNA binding activity of Sp1 (this study and Ref. 56). This difference may originate from the fact that 2-DG treatment causes hyper-GlcNAcylation of Sp1, whereas low glucose treatment causes its hypo-GlcNAcylation. Both hyper-and hypo-GlcNAcylation impair the activity of Sp1, but by different mechanisms. Using peptides for Sp1 activation domain, Kudlow and co-workers (54,59) proposed that O-Glc-NAcylation of Sp1 activation domain blocks its homomultimerization or interactions with TAF II 130 and thereby interferes with Sp1-driven transcription. Sp1 contains at least 9 serine and threonine sites that are O-GlcNAcylated (47), and the activation domain spans two serine/threonine-rich domains  (60). Although the location of the O-GlcNAcylation in Sp1 molecules in cells treated with 2-DG was not determined, O-Glc-NAcylation is likely to occur at the sites in the activation domain, resulting in Sp1 inactivation. Meanwhile, under low glucose conditions, Sp1 is affected in two different ways. First, Sp1 was reported to be in hyperphosphorylated state in cells under glucose starvation (61), and hyperphosphorylated Sp1 was shown to have a defect in DNA binding (62). Second, hypo-GlcNAcylated Sp1 was shown to undergo rapid proteolysis mediated by a 20 S proteosome system (50).
Why is Sp1 hyper-GlcNAcylated in the presence of 2-DG? One plausible model is that Sp1 deGlcNAcylation (removal of GlcNAc residues from protein) is inhibited by 2-DG. This possibility is strongly supported by our finding that 2-DG inhibits O-GlcNAcase in vitro (Fig. 7B). Actually, not only Sp1, but also overall protein O-GlcNAcylation increased in cells treated with 2-DG as well as those treated with STZ (Fig. 7A). This inhibitory effect of 2-DG would shift the equilibrium in favor of O-GlcNAcylation of cellular target proteins and therefore contribute to an increase of Sp1 O-GlcNAcylation.
However, the O-GlcNAcase inhibitory effect of 2-DG appeared not very high, although in vitro activity may not be reflected in vivo, and the quite dramatic increase in O-Glc-NAcylations in cells suggests that some additional mechanisms may also be working. Sp1 hyperglycosylation could be caused by the activation of the hexosamine synthesis pathway through an increased cellular pool of glucose caused by a block in glycolysis. The large quantity of unused glucose may be converted through the hexosamine synthesis pathway to GlcNAc and further to UDP-GlcNAc (63). As a consequence of high substrate supply, hyperglycosylation might have occurred. However, the possibility of hexosamine synthesis pathway activation is not strongly supported theoretically or experimentally. First, 2-DG inhibits phosphohexose isomerase (23,24), which generates fructose 6-phosphate, the precursor of glucosamine 6-phosphate, and therefore GlcNAc production would be inhibited as well in 2-DG-treated cells. Second, supplementation of 20 mM glucosamine did not increase O-GlcNAcylation level of Sp1 in HeLa cells, 2 and treatment of 6-diazo-5-oxo-Lnorleucine, which blocks the hexosamine pathway by inhibiting glutamine:fructose-6-phosphate aminotransferase (50), had no apparent effect on Sp1 O-GlcNAcylation level in HeLa cells (data not shown), which is in accordance to the previous finding that glucose availability is not a limiting factor for O-GlcNAcylation in HeLa cells (50). Sp1 hyperglycosylation may also be induced through up-regulation of OGT or downregulation of O-GlcNAcase at the level of gene expression and/or enzyme activity. The results of our RT-PCR experiments, in which the levels of mRNA of both genes did not appear to be altered during the 2-24-h period after 2-DG treatment, 2 rule out the possibilities of 2-DG-induced up-regulation of OGT gene or down-regulation of O-GlcNAcase gene expression. Also, the O-GlcNAcase activity in HeLa cell extracts measured in vitro did not decrease upon 2-DG treatment, suggesting that 2-DG does not cause alteration of O-GlcNAcase enzyme status. 3 Meanwhile, other possibilities such as direct stimulation of enzyme activities of OGT or other enzymes involved in the production of UDP-GlcNAc by 2-DG have not been examined in this study. Therefore, OGT activity and the UDP-GlcNAc level could be determined and compared to help sort out these possibilities.
So far, PUGNAc (64) and streptozotocin (STZ) (50) are two known inhibitors of O-GlcNAcase enzyme. Based on our study, 2-DG is also an inhibitor of O-GlcNAcase. In addition, in this study, STZ treatment resulted in an increase in the level of Sp1 O-GlcNAcylation to a level comparable with that caused by 2-DG. Down-regulation of HPV early gene transcription occurred as well. Furthermore, both 2-DG and STZ abolished p53-mediated activation of p21 WAF1 promoter. The similar effects of 2-DG and STZ on Sp1 O-GlcNAcylation and transcriptional activity strongly support our hypothesis that the inhibitory effect of 2-DG is through hyper-GlcNAcylation. Our study also showed for the first time that STZ causes inhibition of Sp1 activity again through hyper-GlcNAcylation. Although 2-DG and STZ caused identical changes on Sp1 O-GlcNAcylation and its activity, there may be some differences in their target molecules and action mechanisms. STZ treatment increased p53 protein level in H460 cells, whereas 2-DG did not (Fig. 8B). Currently, we do not know how STZ up-regulates p53. STZ may alter p53 stability either by directly modulating O-GlcNAcylation and phosphorylation or through activating the p53-stimulatory pathway(s). Meanwhile, 2-DG not only inhibits O-Glc-NAcase but also is expected to increase the cellular pool of glucose through the block in glycolysis, whereas STZ does not.
Down-regulation of HPV early gene transcription by 2-DG treatment was first observed by Maehama et al. (33). In this study, c-Myc transcription increased upon 2-DG treatment. c-Myc induction was reportedly detected as early as in 2 h after 2-DG treatment. Meanwhile, in our study, the c-Myc protein level decreased gradually in the course of 24 h. Although it is a possibility that c-Myc expression may be induced by an immediate effect of 2-DG or 2-DG-induced [Ca 2ϩ ] fluctuation (33), our results certainly reflect the cellular event that follows depletion of E7: a decrease of c-Myc transcription that is modulated by E2F1 (65), which is in turn modulated by E7. E7 activates E2F1 gene expression either directly (50) or through inactivation of Rb (51).
Sp1 plays a critical role in the transcription of a number of glycolytic enzymes (55,66,67). Thus, it is conceivable that Sp1 and O-GlcNAc function as glucose sensors that link the metabolic activities of a cell to its nutritional condition. Other than HPV, regulatory elements of many viral genome such as SV40 early promoter (68), herpes simplex virus immediate early promoter (69), and human immunodeficiency virus long terminal repeat (70) contain Sp1-binding sequences and are also regulated by Sp1, possibly reflecting that Sp1 is a ubiquitous transcription factor. Why is Sp1 involved in expression of genes essential for virus production? Viruses are obligatory intracellular parasites, and therefore their reproduction is totally dependent on the state of host metabolism. The result of this study provides one example of the possibility that viral replication is governed by the metabolic state of the host, and this is through the transcription factors of the host that senses its nutritional state. HPV E6 and E7 and SV40 T antigen proteins play essential roles in uncontrolled cell proliferation through recruitment of the machinery of the host for S phase entry and inactivation of checkpoint control functions. For the viruses, which depend on the machinery of the host for their own replication, it may be more advantageous if the virus stays dormant when the host cell is in an environment that is not only less supportive for ATP production but also in danger of being induced to apoptosis. Therefore, Sp1 functions as a molecular sensor not only for the host cell but also for the virus in that it enables the virus to adjust its replication potential to the status of the nutrition supply and energy metabolism of the host.
In this study, although not extensively studied, hypoxia also was shown to down-regulate Sp1 activity. Hypoxia has previously been shown to regulate Sp1-dependent promoters by modulation of the DNA binding activity of Sp1 (56). Indeed, a loss of DNA binding activity of Sp1 proteins was observed in cells incubated in hypoxic conditions in our study. In addition, unlike the situation of 2-DG treatment, in cells under hypoxia, Sp1 glycosylation status was not affected, and HPV repression occurred in later time points. Clearly, hypoxia and 2-DG treatment cause down-regulation of Sp1 activity in different mechanisms. A detailed study on hypoxia-mediated Sp1 regulation would greatly help understanding of the mechanism involved in the DNA binding of Sp1.
Overall, the results of our study suggest that, unlike low glucose supply, 2-DG and STZ treatment causes down-regulation of Sp1 activity through hyperglycosylation. In addition, it is also suggested that 2-DG and STZ can be used as an efficient way to enhance protein O-GlcNAcylation. Further clarification of how 2-DG induces hyper-GlcNAcylation of Sp1 will not only help elucidation of how Sp1 activity is regulated but will also enhance the usefulness of this approach and provide another valuable tool in the field of glycobiology and the study of diabetes mellitus.