Upstream stimulatory factor (USF) proteins induce human TGF-beta1 gene activation via the glucose-response element-1013/-1002 in mesangial cells: up-regulation of USF activity by the hexosamine biosynthetic pathway.

The hyperglycemia-enhanced flux through the hexosamine biosynthetic pathway (HBP) has been implicated in the up-regulated gene expression of transforming growth factor-beta1 (TGF-beta1) in mesangial cells, thus leading to mesangial matrix expansion and diabetic glomerulosclerosis. Since the -1013 to -1002 region of the TGF-beta1 promoter shows high homology to glucose-response elements (GlRE) formerly described in genes involved in glucose metabolism, we studied the function of the GlRE in the high glucose-induced TGF-beta1 gene activation in mesangial cells. We found that high glucose concentrations enhanced the nuclear amount of upstream stimulatory factors (USF) and their binding to this sequence. Fusion of the GlRE to the thymidine kinase promoter resulted in glucose responsiveness of this promoter construct. Overexpression of either USF-1 or USF-2 increased TGF-beta1 promoter activity 2-fold, which was prevented by mutation or deletion of the GlRE. The high glucose-induced activation of the GlRE is mediated by the HBP; increased flux through the HBP induced by high glucose concentrations, by glutamine, or by overexpression of the rate-limiting enzyme glutamine:fructose-6-phosphate aminotransferase (GFAT) particularly activated USF-2 expression. GFAT-overexpressing cells showed higher USF binding activity to the GlRE and enhanced promoter activation via the GlRE. Increasing O-GlcNAc modification of proteins by streptozotocin, thereby mimicking HBP activation, also resulted in increased mRNA and nuclear protein levels of USF-2, leading to enhanced DNA binding activity to the GlRE. USF proteins themselves were not found to be O-GlcNAc-modified. Thus, we have provided evidence for a new molecular mechanism linking high glucose-enhanced HBP activity with increased nuclear USF protein levels and DNA binding activity and with up-regulated TGF-beta1 promoter activity.

The adverse effects of hyperglycemia in human and experimental diabetic nephropathy have been linked to the enhanced expression and bioactivity of the prosclerotic cytokine transforming growth factor-␤1 (TGF-␤1) 1 (1)(2)(3). The hyperglycemiainduced TGF-␤1 stimulates the production of extracellular matrix proteins in mesangial cells and other renal cells (4 -7), thus leading to the thickening of glomerular and tubular basement membranes and the progressive expansion of the glomerular mesangium and the tubulointerstitium (8,9). Increased renal expression of TGF-␤1 has been found in experimental and human diabetes (2,3) and has also been demonstrated in high glucose-treated renal mesangial and tubular cells (5,6). The molecular mechanism of up-regulated human TGF-␤1 gene expression involves protein kinase C-and p38 MAPK-dependent pathways (10 -12) leading to AP-1 activation (13,14) and subsequently enhanced TGF-␤1 promoter activity via two adjacent AP-1 binding sites located at Ϫ418/Ϫ412 and Ϫ371/ Ϫ363, respectively (13).
Moreover, increased synthesis of amino sugars through the hexosamine biosynthetic pathway (HBP) has also been implicated in hyperglycemia-induced TGF-␤1 synthesis (15,16). Inhibition of the hexosamine pathway prevented the high glucose-induced TGF-␤1 synthesis and bioactivity and the enhanced expression of matrix proteins in mesangial cells (15). Furthermore, glucosamine, which promotes flux through the hexosamine pathway distal of the rate-limiting enzyme glutamine fructose-6-phosphate aminotransferase (GFAT) (17), activates the expression of TGF-␤1 and several matrix proteins (15). These effects were also observed by overexpression of GFAT in NIH 3T3 fibroblasts (18) and mesangial cells (19) in physiological glucose concentrations. The HBP has been implicated in increased promoter activity of genes under high glucose conditions (20 -23). The product glucosamine-6-phosphate (GlcN-6-P) is very rapidly further converted and activated to UDP-GlcNAc, the substrate of an O-GlcNAc transferase that modifies serine and threonine residues of cytosolic and nuclear proteins with a single monosaccharide O-GlcNAc group (20). This O-GlcNAc modification has been proven to be a regulator of protein activities, e.g. of the transcription factor Sp1 (16,22,24). However, the identification of the HBP-response elements in the TGF-␤1 promoter region and of the involved transcription factors remains open.
Examination of the human TGF-␤1 promoter region by computer analysis for transcription factor binding sites revealed that the region Ϫ1013 to Ϫ1002 (CACGTGGCGGCC) shows high homology to consensus motifs in the promoters of the liver pyruvate kinase (L-PK), the S14, and the fatty acid synthase gene, which refer to the glucose responsiveness of these genes, and are therefore called carbohydrate or glucose-response elements (GlRE) (25,26). Among the proteins that were reported to possess GlRE DNA binding activity are members of the basic helix-loop-helix leucine zipper family of transcription factors, the upstream stimulatory factors (USF), which bind to the E-box motifs in the GlRE (25,26).
The current studies were initiated to examine the function of the GlRE sequence in the TGF-␤1 promoter and to characterize the nuclear proteins that bind to this region. Furthermore, we investigated the hypothesis that the GlRE and GlRE DNAbinding proteins could be the link between the high glucoseinduced activation of the HBP and the resulting enhanced TGF-␤1 gene expression. We found enhanced binding of an USF-1 and -2 heterodimer to the GlRE in nuclear extracts obtained from high glucose-stimulated mesangial cells. Expression of either USF-1 or USF-2 induced TGF-␤1 promoter activity 2-fold with no additive effect in the presence of both USF proteins. Mutation or deletion of the GlRE prevented the USFinduced promoter activation almost completely. Increased flux through the HBP stimulated by high glucose or glutamine, by overexpression of GFAT, or enhancement of O-glycosylation by streptozotocin (STZ) predominantly activated the expression of USF-2, although USF proteins were not a direct target of O-GlcNAc modification. Thus, activation of the HBP, via elevated O-glycosylation activity, enhanced nuclear USF protein levels, and increased binding of USF proteins to the GlRE, up-regulates TGF-␤1 promoter activity.
The wild type human TGF-␤1 promoter construct Ϫ453/ϩ11 in pGL3b is described in Ref. 13. The human TGF-␤1 promoter construct pGL3-1065/ϩ11 was generated by insertion of nucleotides Ϫ1065 to ϩ11 of the human TGF-␤1 promoter into the KpnI and BglII sites of the vector pGL3b. The construct GlRE-TK was prepared using a oligonucleotide with BglII linkers containing base pairs Ϫ1022/Ϫ993 of the human TGF-␤1 gene including the GlRE. It was cloned into the BglII site of pGL3b-TK, which contains the herpes simplex virus thymidine kinase promoter fragment of pRL-TK ligated into the multiple cloning site of pGL3b (a scheme of the constructs is shown in Figs. 2A and 3B).
A full-length human USF-1 cDNA was generated by RT-PCR using human RNA as template with the primer 5Ј-CGGGATCCCGATGAAG-GGGCAGCAGAAAAC-3Ј and 5Ј-GGAATTCCCCATAGTTAGTTGCTG-TCATTC-3Ј containing BamHI and EcoRI linker and cloned into pcDNA3.1 (Invitrogen). The PCR product was verified by sequencing and identical to human USF-1 (GenBank TM accession number X55666). The plasmid pCMV-USF-2 containing full-length murine USF-2 is de-scribed in Ref. 29. Generation of a full-length human GFAT cDNA and cloning into the expression vector pcDNA3.1 was recently described (18,19).
Cell Culture-Mesangial cells isolated from porcine glomeruli were cultured and characterized as described previously (15). Cells were grown in RPMI 1640 with 15% fetal calf serum, 1 mM sodium pyruvate, 3 mM glutamine, non-essential amino acids, 100 units/ml penicillin, 100 g/ml streptomycin, 0.5 g/ml minocyclin, and 10 mM glucose. For experiments, RPMI 1640 containing 5.5 mM glucose and 3 mM glutamine was used, and fetal calf serum was substituted by 2% Ultroser. Mouse SV40 mesangial cells were cultured in Dulbecco's modified Eagle's medium containing 5.5 mM glucose, 10% fetal calf serum, 1 mM sodium pyruvate, 4 mM glutamine, non-essential amino acids, 100 units/ml penicillin, and 100 g/ml streptomycin. For equal osmolarity in the high glucose experiments, NaCl was added to control cells.
Transfection and Reporter Gene Assays-SV40 mesangial cells and porcine mesangial cells were transfected with FuGENE 6 according to the instructions of the supplier. One day prior to transfection, 0.2 ϫ 10 5 cells/well were seeded in 12-well plates with 1 ml of experimental medium. For one well, 1.5 l of FuGENE 6 and 0.5 g of total DNA were diluted in serum-free medium, and after 15 min, added to the cells. Glucose was added prior to transfection. Cells were harvested after 40 h. For normalization of transfection efficiencies, we cotransfected pRL-TK vector in porcine mesangial cells. Separate cell extracts of SV40 mesangial cells were obtained from triplicate wells to monitor the reproducibility of the transfection efficiency (replicates generally varied by Ͻ10% in luciferase activity).
Transfected cells were washed once with phosphate-buffered saline, incubated with 150 l of lysis buffer from the dual luciferase assay for 15 min, and harvested. Firefly and sea pansy luciferase activities were determined according to the instructions of the dual luciferase assay using Magic Lite analyzer from Ciba Corning (Fernwald, Germany). All transfection experiments were repeated at least three times.
Preparation of Nuclear Extracts-For preparation of nuclear extracts, 6.0 ϫ 10 6 porcine mesangial cells were seeded onto 15-cm culture dishes and cultured in 15 ml of experimental medium. Cells were incubated with 5.5 or 30 mM glucose or STZ for the indicated time points before harvesting. Nuclear proteins were prepared as described (30).
Western Blotting-Cellular extracts were separated by sodium dodecyl sulfate polyacrylamide (7.5%) gel electrophoresis. Proteins were transferred to nitrocellulose by semidry electroblotting (transfer buffer: 48 mM Tris, 39 mM glycine, 0.0375% sodium dodecyl sulfate, 20% (v/v) methanol). The nitrocellulose membranes were then blocked with NET buffer (150 mM NaCl, 50 mM Tris/HCl, pH 7.4, 5 mM EDTA, 0.05% Triton X-100, 0.25% gelatin) and incubated with the first antibody (diluted 1:1000 in NET) overnight at 4°C. After washing with NET buffer, the membranes were incubated with horseradish peroxidaseconjugated anti-rabbit or anti-goat IgG for 1 h at room temperature. Visualization of immunocomplexes was performed by enhanced chemiluminescence as described previously (13). India ink staining was performed as described in Ref. 31.
Electrophoretic Mobility Shift Assay-Synthetic oligonucleotides containing human TGF-␤1 sequence Ϫ1019 to Ϫ996 GlRE (CTGCCCCA-CGTGGCGGCCCCTGGG) or GlREmut (CTGCCCCATATGGCGGCC-CCTGGG), in which the mutated sequence is bold and underlined or containing a high affinity binding site for Sp1 (AGCCGGGGAGCCCG-CCCCCTTTCCCCCAGGGCTG) were end-labeled with [␣-32 P]dATP (3000 Ci/mM) and Klenow enzyme and were incubated with up to 6 g of nuclear protein in 20 l of 7 mM Hepes-KOH, pH 7.9, 100 mM KCl, 3.6 mM MgCl 2 , 10% glycerol on ice for 20 min. 0.05 mg/ml poly[d(I-C)] was added as unspecific competitor. The samples were run on a 5% nondenaturating polyacrylamide gel in a buffer containing 25 mM Tris-HCl, pH 8.0; 190 mM glycine, and 1 mM EDTA. Gels were dried and analyzed by autoradiography.
RT-PCR and Real-time Quantitative PCR Analysis-Reverse transcription of total RNA (1 g) was performed using random hexamers and the first strand cDNA synthesis kit for RT-PCR. Aliquots (2 l) of the reverse transcription reactions were then submitted in triplicate to online quantitative PCR with the Light Cycler system using the Fast-Start DNA-MasterSYBR Green I. The PCR reaction was optimized for Mg 2ϩ concentration and annealing temperature. Initial real-time amplifications were examined by agarose gel electrophoresis followed by ethidium bromide staining to verify that the primer pairs amplified a single product of the predicted size. Subsequent aliquots of the PCR reaction where checked by melting curve analysis as provided by the Light Cycler system. The porcine USF-1 primers (sense, 5Ј-ggtggaattctgtccaaagc-3Ј; antisense, 5Ј-gctggtagcagcagattcttg-3Ј) and USF-2 primers (sense, 5Ј-cagtaccagttccgcacaga-3Ј; antisense, 5Ј-acacccacctgggtcac-ag-3Ј) amplified a product of 161 and 164 bp, respectively. The mouse USF-1 primers (sense, 5Ј-agctgttgttaccacccagg-3Ј; antisense, 5Ј-tatgttgagccctccgtttc-3Ј) and USF-2 primers (sense, 5Ј-ctacagcagcacaacctgga-3Ј; antisense, 5Ј-tgggcaacgtatcaacagaa-3Ј) amplified a product of 202 and 206 bp, respectively. The PCR was performed in a volume of 20 l as follows: 2 l of FastStart DNA-MasterSYBR Green I, MgCl 2 3 mmol/ liter, and primers according to a primer concentration of 1 mol/liter. The instrument settings were as follows: denaturing was at 95°C for 10 min, cycling was performed by denaturing at 95°C for 15 s, annealing was at 68°C for 5 s, and elongation was performed for 8 s; the number of cycles was 45. Quantification was performed by online monitoring for identification of the exact time point at which the logarithmic linear phase was distinguishable from the background. Serially diluted samples obtained by PCR with the above mentioned primers from wild type rat mesangial cells were used as external standards in each run. The cycle numbers of the logarithmic linear phase were plotted against the logarithm of the concentration of the template DNA, and the concentration of cDNA in the different samples was calculated with the Light Cycler software (version 5.32).
Statistical Analysis-Results presented are derived from at least three independent experiments. Means Ϯ S.E. were calculated, and groups of data were compared using student's t test. Statistical significance was set at p Ͻ 0.05.

RESULTS
High Glucose Stimulates Binding of Nuclear Proteins USF-1 and USF-2 to the GlRE-To characterize the TGF-␤1 promoter region Ϫ1013/Ϫ1002 as a glucose-response element, the binding of nuclear proteins from normal and high glucose-treated porcine mesangial cells to this region was studied by electrophoretic mobility shift assay (EMSA). Incubation with 30 mM ambient high glucose for 15 h induced the DNA binding activity to the Ϫ1013/Ϫ1002-containing sequence, and this effect was enhanced after 40 h of high glucose stimulation and persisted for at least 72 h (Fig. 1A). The specificity of the binding was tested by using 30-fold molar excess of unlabeled GlRE oligonucleotide, which prevented the DNA binding completely, whereas 30-fold molar excess of the mutated GlRE does not (Fig. 1B, lane 2 and 3). Moreover, no binding to the radiolabeled mutated GlRE was detected (Fig. 1B, lane 4). To identify the transcription factors that bind to the TGF-␤1 GlRE sequence, antibodies against USF-1 and USF-2 were added to the binding reaction. With either antibody, the GlRE-protein complex was completely supershifted, leading to distinct new bands with lower mobility (Fig. 1C, lanes 3 and 4). In contrast, addition of an anti-Sp1 antibody did not affect the mobility of the GlREprotein complex (Fig. 1C, lane 2). Thus, USF-1 and USF-2 are identified as major components and proved to be essential for the nuclear DNA binding activity to the GlRE.
High Glucose Stimulates Promoter Activation via the GlRE-The functional activity of the TGF-␤1 promoter GlRE was studied as the ability of this promoter region to confer high glucose responsiveness to a different promoter. Fusion of the TGF-␤1 promoter region Ϫ1022/Ϫ993 to the thymidine kinase promoter ( Fig. 2A), which alone is not activated by high glucose concentrations in porcine mesangial cells (Fig. 2B, TK), led to a 1.5-fold induction of GlRE-TK promoter activity after 24 h of high glucose stimulation (Fig. 2B). Thus, the TGF-␤1 promoter region Ϫ1022/Ϫ993 is a functionally active and independent GlRE.
Expression of USF Proteins Enhances TGF-␤1 Promoter Activity-To investigate whether high glucose-induced binding of USF to the GlRE can enhance TGF-␤1 promoter activity, SV40 mesangial cells were cotransfected with expression vectors for USF-1 and/or USF-2 and TGF-␤1 promoter constructs in pGL3basic in 5.5 mM ambient glucose concentrations. In these experiments, we used SV40 mesangial cells since the transfection efficiency obtained with porcine mesangial cells, although suitable to perform luciferase reporter gene assays, is not sufficient to investigate the effect of transfected proteins. In trans-fected SV40 mesangial cells, expression levels of USF-1 and USF-2 were high as measured by Western blotting with specific antibodies (Fig. 3A), thus leading subsequently to enhanced binding to the GlRE (data not shown). The nuclear amount of the transcription factor c-Jun was essentially unchanged in USF-transfected cells (Fig. 3A). Promoter activity of the cotransfected Ϫ1065/ϩ11 TGF-␤1 promoter region (Fig. 3B) increased by either enhanced USF-1 or enhanced USF-2 levels 1.5-2.0-fold with no additive effect in the presence of elevated concentrations of both USF proteins (Fig. 3C). Mutation of the GlRE, which abrogates USF protein binding as demonstrated by EMSA (Fig. 1B), clearly reduces the effect of USF overex- pression on TGF-␤1 promoter activation (Fig. 3C). Remaining weak stimulation by USF-1 could possibly be mediated by a second putative USF binding site at position Ϫ877/Ϫ872. Accordingly, transfection of expression vectors for USF-1 and USF-2 showed no effect on a cotransfected Ϫ453/ϩ11 TGF-␤1 promoter construct lacking both USF binding sequences (Fig.  3C). These results demonstrate that elevated levels of USF-1 and USF-2 cause TGF-␤1 promoter activation via the GlRE.
Activation of the HBP Induce mRNA Expression of USF-2-Next, we investigated the underlying mechanism responsible for the high glucose-enhanced DNA binding of USF proteins to the GlRE. The nuclear amount of both USF-1 and USF-2 proteins was found to be enriched in porcine mesangial cells stimulated for 24 h with 30 mM glucose, whereas the level of the transcription factor c-Jun remained unchanged (Fig. 4A). To test the effect of high glucose concentrations on USF-1 and USF-2 mRNA expression levels in porcine mesangial cells, real-time PCR was performed. After a 24-h stimulation with 30 mM glucose, USF-2 mRNA levels were 1.7-fold elevated, whereas the effect on USF-1 mRNA expression was weaker (1.3-fold; Fig. 4B).
To assess whether the HBP mediates glucose-induced upregulation of both USF-1 and USF-2 mRNA, we incubated porcine mesangial cells with 2 mM glucosamine or 30 mM glucose plus 16 mM glutamine for 24 h. GFAT, the rate-limiting enzyme of the HBP, requires glutamine for conversion of fructose-6-phosphate to glucosamine-6-phosphate, whereas glucosamine enters the HBP downstream of GFAT. Furthermore, we treated cells for 24 h with 5 mM STZ, a GlcNAc analog, causing increased O-GlcNAc modification of proteins in beta cells (32) and also in the mesangial cells used in our experiments (data not shown). With all three substances, higher levels of USF-2 mRNA as compared with 30 mM glucose alone were obtained (Fig. 4C). This effect is specific for USF-2 since no activation of USF-1 mRNA expression was detected in the identical RNA samples (Fig. 4C).
Effect of Overexpression of GFAT on USF-2 Expression and GlRE Activity-Further support for the involvement of the HBP is provided when SV40 mesangial cells were transfected with GFAT expression vector, leading to high expression of GFAT protein as compared with control cells (Fig. 5A). We found an up-regulation of USF-2 mRNA levels in GFAToverexpressing cells, whereas again, no increase in USF-1 expression was detected (Fig. 5B). Moreover, only in the GFAT-overexpressing cells, glutamine (16 mM) enhanced USF-2, but not USF-1, mRNA expression as compared with corresponding basal levels in cells grown in ambient normal glucose concentrations (1.57 Ϯ 0.11, data not shown). Accordingly, the USF binding to the GlRE is enhanced by GFAT overexpression (Fig. 5C, lane 2 versus lane 1), identified by supershifted bands in the presence of anti-USF antibodies (Fig. 5C, lanes 3 and 4). Subsequently, the promoter activity of the GlRE-TK construct was found to be up-regulated in the GFAT-overexpressing cells (2.5-fold; Fig. 5D), whereas the promoter activities of the TK promoter alone or the construct containing mutated GlRE were comparably low (1.6-fold versus promoter activities found in control cells; Fig. 5D). This residual activation could be explained by HBP-enhanced transcriptional activity of the ubiquitous transcription factor Sp1, thereby affecting TK promoter activity. Thus, activation of USF-2 expression by high glucose mediated by the HBP appears to be the predominant mechanism leading to enhanced TGF-␤1 gene activation via GlRE. Overexpression of USF proteins increase TGF-␤1 promoter activity. SV40 mesangial cells were cotransfected with expression vectors coding for USF-1, -2, or empty expression vector (con) and TGF-␤1 promoter constructs in pGL3basic. A, Western blotting of nuclear extracts of transfected cells. Shown are representative immunoblots for USF-1 and USF-2. The blot was also probed for c-Jun, demonstrating equal protein loading. B, construct of the 5Ј-flanking region of the TGF-␤1 gene in pGL3basic, which contains the firefly (Photinus pyralis) luciferase coding region (LUC). Characterization of the AP-1 binding sites A and B has been described recently (13). C, effect of transfection of USF proteins on TGF-␤1 promoter region Ϫ1065/ϩ11, Ϫ1065/ϩ11mut containing mutated GlRE and Ϫ453/ϩ11 constructs. Cells were incubated with ambient 5.5 mM glucose, harvested 30 h after transfection, and assayed for luciferase activities. Luciferase activity measured in cells transfected with empty expression vector was defined as 100%. Data are means of at least three independent experiments.

Effect of Enhanced Nuclear O-GlcNAc Modification by STZ on USF-2-O-GlcNAc
modification of nuclear proteins is a common phenomenon, leading to enhanced activity of transcription factors, for example, by enhanced nuclear translocation or DNA binding activity (20,22,24,33). Since we found that treatment of mesangial cells with STZ resulted in elevated USF-2 mRNA expression (Fig. 4C), we studied the effect of STZ in more detail. First we investigated the effect of STZ on the nuclear amount of USF-2 protein. Incubation of mesangial cells with 5 mM STZ increased the nuclear USF-2 protein levels in a timedependent manner (Fig. 6A). Similarly, the DNA binding activity of the USF proteins to the GlRE was found to be activated after 6 and 24 h of STZ-treatment (Fig. 6B). In supershift experiments, the intensity of the USF-2-GlRE complex supershifted by the anti-USF-2 antibody was clearly increased after stimulation with STZ (Fig. 6B). Quite unexpectedly, addition of the anti-O-GlcNAc antibody RL-2 did not result in any supershifted bands (Fig. 6B). To demonstrate the effectiveness of the method to identify O-GlcNAc-modified DNA-binding proteins, we studied the DNA binding of the O-GlcNAc-modified transcription factor Sp1 (16,24,33) to a Sp1 consensus binding site after preincubation with the RL-2 antibody. In fact, the Sp1-DNA complex is clearly supershifted by the anti-O-GlcNAc antibodies, whereas again, USF binding activity was unaffected (Fig. 6C). To support the finding that the nuclear USF proteins that bind to the GlRE are not O-GlcNAc-modified, we incubated nuclear proteins with ␤-N-acetylglucosaminidase before EMSA to remove any potential O-GlcNAc groups on USF proteins. This treatment had no effect on the DNA binding activity of USF proteins to the GlRE of the TGF-␤1 promoter, whereas the DNA binding of Sp1 to the high affinity Sp1 binding site is completely prevented (Fig. 6D). Thus, it appears unlikely that USF-1 and -2 activities are regulated by direct O-GlcNAc modification and that these proteins are direct targets of O-GlcNAc-transferase. DISCUSSION In the present study, we describe a new link of the HBP to enhanced TGF-␤1 gene activation via a hitherto unknown mechanism involving USF transcription factors. USF proteins were found to be responsible for high glucose-induced TGF-␤1 promoter activation mediated by a GlRE sequence located at Ϫ1013 to Ϫ1002 of the TGF-␤1 promoter: (i) high glucose concentrations activate binding of a USF-1/USF-2 heterodimer to the GlRE, (ii) the nuclear amount of both proteins is elevated in high glucose-treated cells, (iii) high glucose concentrations induce USF-2 mRNA levels, and (iv) overexpression of USF-1 and USF-2 activates TGF-␤1 promoter activity via the GlRE. USF-1 and USF-2 have been shown to be involved as regulators of a high number of genes in numerous cellular processes. Gene knock-out mice and cell culture studies indicate that USFs are important for a kinetically normal activation of diet-dependent genes by glucose in the liver (26, 34 -37). However, several studies revealed that these transcription factors are not capable by themselves of conferring glucose responsiveness to hepatic and beta cell genes (38 -40). Moreover, numerous studies do not demonstrate transcriptional activation of USF by glucose (38 -40), e.g. there is evidence that USF is not the glucoseresponsive factor that stimulates S14 gene and liver pyruvate kinase expression (38,39). Thus, in the liver and beta cells, USF proteins appear to not be activated by high glucose concentrations.
In a recent report, glucose-induced USF protein accumulation and subsequently enhanced binding to osteopontin promoter sequences and up-regulation of osteopontin transcription have been demonstrated in vascular smooth muscle cells (41). In our study, we provide the first evidence for a mechanism of high glucose-enhanced USF activity and the increases in USF-2 expression: (i) activation of the HBP by high glucose or glutamine or (ii) increased flux through the HBP by elevated levels of the rate-limiting enzyme GFAT or (iii) by stimulation of O-GlcNAc modification by STZ resulted in increased USF-2 expression and elevated binding to the GlRE. The HBP produces UDP-GlcNAc, the substrate of an O-GlcNAc transferase, which modifies serine and threonine residues of cytosolic and nuclear proteins with a single monosaccharide O-GlcNAc group (20). STZ, an analog of GlcNAc, is known to inhibit ␤-N-acetylglucosaminidase (O-GlcNAcase), an enzyme that cleaves Glc-NAc residues from intracellular proteins (32,42). The resulting increases in O-GlcNAc modification of proteins modulate their function, e.g. the transcription factor Sp1 is activated (16,22,24,43). Thus, the possible mechanism for up-regulation of USF-2 mRNA levels via the HBP is enhanced O-GlcNAc modification of transcription factors responsible for USF-2 gene activation. Of note, the putative promoter region of the human USF-2 gene is GC-rich, suggesting several Sp1 binding sites (44). A further possibility for the elevated nuclear USF-1 and USF-2 protein levels is O-GlcNAc modification of the USF proteins themselves, thereby inducing their shuttle from cy-tosol to the nucleus. This has been demonstrated for the transcription factor Sp1 (24,43). However, using O-GlcNAc-specific antibodies, we did not find any evidence for O-glycosylation of nuclear USF proteins. Therefore, we conclude that the effect of an increased flux through the HBP on USF activity is to upregulate USF-2 expression. The mechanism responsible for the elevated USF-1 protein levels appears to be independent from a transcriptional regulation of USF-1 by the HBP. Moreover, we could not exclude from our data other, e.g. posttranslational, mechanisms leading to high glucose-induced activation of USF proteins.
With the present study, we obtained additional insights in the molecular mechanism of the hyperglycemia-induced upregulation of TGF-␤1 expression in mesangial cells. In a previous study, we described the involvement of two AP-1 binding sites in the activation of human TGF-␤1 promoter region Ϫ453/ ϩ11 by high glucose (13); here, we found that a glucose-response element homolog sequence located more distal in the human TGF-␤1 promoter is also activated by high glucose concentrations. To study the functional relevance of this GlRE sequence of the TGF-␤1 promoter, we investigated the activity of the GlRE when fused to an glucose-independent promoter. We chose this experimental approach since studying the high glucose response of the complete TGF-␤1 promoter construct Ϫ1065/ϩ11 revealed a similar, although weaker (1.5-fold, data not shown), promoter activation as compared with the previously published 2.0-fold induction of the shorter Ϫ453/ϩ11 TGF-␤1 promoter construct (13). An explanation for this unexpected result could be that the human TGF-␤1 promoter contains silencing elements between nucleotides Ϫ731 and Ϫ453 (45), which reduce basal promoter activity to 10% (45) and could cover a high glucose effect. Thus, the stimulatory effect of high glucose on the TGF-␤1 promoter regions Ϫ1065/ϩ11 and Ϫ453/ϩ11 appears similar without any additive effect of the high glucose-inducible AP-1 sites and the GlRE. Of note, a similar glucose-response element has been postulated to be involved in the high glucose-mediated up-regulation of the mouse TGF-␤1 promoter (46).
In conclusion, we have ascertained that the human TGF-␤1 gene contains a functional active GlRE, which is capable of inducing TGF-␤1 gene activation in mesangial cells after expo-FIG. 6. Involvement of O-GlcNAc modification in increased production and DNA binding activity of USF proteins. Porcine mesangial cells were incubated with 5 mM STZ as indicated. A, Western blotting of nuclear extracts. Shown is one representative immunoblot for USF-2 and india ink staining of a representative, strong band visible at 100 kDa to demonstrate equal protein loading. B, EMSAs were performed with nuclear extracts obtained from STZ-treated porcine mesangial cells with GlRE oligonucleotide. For supershift experiments, nuclear extracts were preincubated with anti-USF-2 antibody or anti-O-GlcNAc antibody. Supershifted USF-2-DNA complex is marked by the bracket. con, unstimulated cells. C, supershift experiment using anti-O-GlcNAc antibody and a high affinity Sp1 binding site or GlRE. Supershifted Sp1-DNA complex is marked by the bracket. Free DNA probe is not shown. D, prior to the protein DNA binding reaction, nuclear extracts were incubated with ␤-N-acetylglucosaminidase (hex) for 30 min at RT as indicated. Binding of Sp1 or USF proteins is marked by the arrow, and the free DNA probe is not shown. sure to high glucose concentrations. Thereby, we have provided evidence for a new molecular mechanism linking high glucoseenhanced HBP activity with the transcriptional activation of the TGF-␤1 gene via USF protein activation.