Specific Protein-1 Is a Universal Regulator of UDP-glucose Dehydrogenase Expression ITS POSITIVE INVOLVEMENT IN TRANSFORMING GROWTH FACTOR- (cid:1) SIGNALING AND INHIBITION IN HYPOXIA*

UDP-glucose dehydrogenase (UGDH) is a key enzyme of the unique pathway for the synthesis of UDP-glucur-onate, the substrate for the numerous glucuronosyl transferases, which act on the synthesis of glycosamin-oglycans and glucuronidation reaction of xeno- and en-dobiotics. Using the bacterial artificial chromosome ap-proach, we have cloned and characterized the human UGDH promoter. The core promoter of (cid:1) 644 nucleotides conferred reporter gene activity in transient transfection assay of a variety of cell types, including MRC5 fibroblasts and the HepG2 hepatoma cell line. The minimal promoter of (cid:1) 100 nucleotides contains a functional inverted TATA box. No consensus CAAT sequence was found up to (cid:1) 2133 nucleotides. The expression of UGDH was up- and down-regulated by transforming growth factor (TGF)- (cid:2) and hypoxia, respectively. TGF- (cid:2) enhanced the activity of all the deletion constructs, except the minimal promoter.

animals, it constitutes the unique pathway for glucuronate formation (1). Because glucuronate is a component of glycosaminoglycans, the mutation inactivation of UGDH (sugarless) abolishes glycosaminoglycan assembly and, consequently, abolishes GAG-dependent growth factor signaling. For example, such a mutation was shown to induce cardiac valve malformation in zebrafish (2) or the "wingless" and "no white eyes" phenotypes in Drosophila (3,4). The primary structure of the mammalian enzyme was obtained from the protein sequence of bovine UGDH (5). The human gene, recently cloned in our laboratory (6), was assigned to chromosome 4p15.1 by radiation hybrid mapping (7). It contains 12 exons, extends over 26 kb, and has one major transcription start site (6). The structure of the enzyme is well conserved between the species and phyla. The cloned mammalian proteins from different species showed overall 97% identity. The human sequence has 27% identity with the Escherichia coli ortholog, with 100 and 60% identity of the NAD ϩ binding and the catalytic sites, respectively (8).
Glycosaminoglycan chains of proteoglycans and hyaluronan are ubiquitous components of extracellular matrix and pericellular spaces. There is a growing body of information on the implication of GAGs in cell behavior, including signal transduction, cell proliferation, spreading, migration, and cancer growth and metastasis (9 -11). GAG synthesis is influenced by cytokines and growth factors. TGF-␤ is the most potent stimulator of proteoglycan and glycosaminoglycan synthesis, including that of hyaluronan. Its action, however, depends on the cell type (12). The synthesis of GAGs is also modulated by oxygen cell status. Hypoxic endothelial cells and lung fibroblasts enhanced heparan sulfate/chondroitin sulfate ratio, which led to an increase of basic fibroblast growth factor reactivity on the cell surface (13,14). It was also shown that the level of intracellular UDP-glucuronate could influence GAG synthesis (1).
The knowledge of the cytokine regulation of UGDH expression is scanty. The human UGDH was shown to be an early response gene after interleukin-1␤ treatment of ocular fibroblasts (15), as well as an early androgen response gene in breast cancer (16). It was also up-regulated by fetal serum (8,17).
TGF-␤ signaling in the cells is initiated through receptor-dependent Ser/Thr kinase phosphorylation of SMAD3 (18), which forms a heterodimer complex with SMAD4 (19). The complex is translocated into the nucleus where it activates its target genes (20). SMAD proteins, however, cooperate with other transcription factors to activate the promoters of different genes. In several systems, Sp1 is a powerful intermediate of the TGF-␤ signaling cascade. For example, Sp1 drives TGF-␤ activation of promoters such as the collagen ␣1(I) (21) and ␣2(I) chains (22,23), plasminogen activator inhibitor-1 (24), SMAD7 (25), TGF-␤ receptor II, and TGF-␤ itself (26). The involvement of Sp1 in the TGF-␤-mediated induction of different genes is mediated by its interaction with the canonical GC-rich sequences, implicating the formation of a complex with phosphorylated SMAD3. The methylation inhibition of Sp1 expression (27,28), or mutation of GC-rich binding sequences (24,25,29,30), completely suppressed TGF-␤ action, indicating similar roles of SMAD and Sp1 systems in signaling cascades. Several genes are activated through Sp1 after TGF-␤ treatment and export proteins, and it was proposed that Sp1 is a major regulator of the expression of extracellular proteins (31). In TATAless promoters, Sp1 can also drive the basal activity of the gene (32).
In the present study, we characterized the regulation of UGDH expression. The expression of its mRNA was studied in different cell strains of human origin. We cloned the human UGDH promoter and defined its minimal sequence and basal activity. We showed that the core promoter of 650 bp is activated under TGF-␤ treatment through numerous Sp1 sites. On the other hand, in HepG2 cells, which harbored a high constitutive level of UGDH activity, we demonstrated that hypoxic conditions led to the down-regulation of this enzyme by nuclear depletion/inactivation of the same Sp1 protein. We thus proposed that Sp1 transcription factor may be a universal modulator of UGDH expression and, by modulating the enzyme level, may control the UDP-glucuronate level in different metabolic pathways, including GAG synthesis.
Before TGF-␤ stimulation, confluent MRC5 cells were preincubated for 18 h in a medium containing 0.5% FCS, and the stimulation was performed as indicated in the figure legends. For studies involving hypoxic conditions, HepG2 cells were cultured to confluence, then the medium was replaced with fresh medium supplemented with 10% FCS. The cells were purged with 100% N 2 for 15 min and then incubated with 0.1-20% O 2 /5% CO 2 /balance N 2 ) for 24 h. Oxygen concentration was monitored with an Oxypocket oxymeter (Bioblock Scientific, Illkirch, France).
Northern Blot-Total RNA from different cell lines were extracted by the guanidine thiocyanate method and separated on 1.1% denaturing gel electrophoresis. After capillary transfer and UV cross-linking, the membranes were hybridized at 60°C with random priming-labeled 496-bp insert containing a part of human open reading frame UGDH (8). The membranes were washed and exposed to X-Omat-AR film in an intensifier screen containing cassette at Ϫ80°C for various periods. After stripping, membranes were rehybridized with full-length cDNA of glyceraldehyde-3-phosphate dehydrogenase (GAPDH, a generous gift of Prof. P. Fort, Montpellier, France) (34) or with the acidic ribosomal phosphoprotein 36B4 cDNA probe (35).
Promoter Cloning and Plasmid Constructions-The DNA from the BAC RP11-472B18 (BACPAC Resource Center at the Children's Hos-pital, Oakland Research Institute, Oakland, CA) was digested with EcoRI restriction enzyme, and the fragments were separated on a 0.8% agarose gel followed by a Southern blot. The blot was hybridized with a 165-bp probe containing the 5Ј portion of UGDH cDNA (6). A 3-kb fragment, which gave a weak hybridization band with the probe, was excised from the gel, subcloned into a pBluescript plasmid, and fully sequenced. The resulting sequence was analyzed with the Transfac data base (36) and MatInspector algorithms (37), using the Transcription Element Search System (available at www.cbil.upenn.edu/tess).

FIG. 1. Northern blot analysis of UGDH expression in a panel of cultured human cells.
A, total RNA was isolated from confluent cells, separated on denaturing gel, transferred onto a membrane, and sequentially hybridized with radiolabeled cDNA probes for human UGDH (upper row) and a human full-length GAPDH probe (middle row) under high stringency conditions. The bottom row shows a 28 S ribosomal subunit stained with BET. B, autoradiographic signals were quantified by densitometric analysis, and the UGDH mRNA/GAPDH mRNA ratio is indicated. The names of the cell lines are indicated above the blot and below the graph.
Transient Transfections and Luciferase Assay-UGDH promoter-luciferase reporter gene constructs (3 and 1 g for MRC5 and HepG2 cells, respectively) were co-transfected into MRC5 cells or HepG2 cells with 25 ng of pRL-TK vector, containing the herpes simplex virus thymidine kinase promoter upstream of Renilla luciferase gene using the Fugene-6 transient transfection kit according to the manufacturer's protocol. After 24 h of incubation in the appropriate media, MRC5 and HepG2 cells were harvested and lysates were prepared. Firefly luciferase activities were assayed with a dual luciferase reporter assay system (Promega) in a Lumi-One luminometer (Bioscan, Washington, D. C.) and normalized to Renilla luciferase activity.
Nuclear Extracts and Electrophoretic Mobility Shift Assay-Nuclear extracts were prepared from confluent flasks of TGF-␤-treated or untreated MRC5 or from HepG2 cells incubated under normal aerobic conditions or under 1% O 2 (see hypoxia experiments above), as described by Dignan et al. (38).
Nuclear proteins (7.5 g/assay) were incubated at room temperature, in a reaction solution containing 4% glycerol, 1 mM MgCl 2 , 0.5 mM EDTA, 0.5 mM dithiothreitol, 50 mM NaCl, 10 mM Tris-HCl (pH 7.5), and 50 g of poly(dI-dC)⅐poly(dI-dC). For competition experiments, a 100ϫ molar excess of non-labeled competitor oligonucleotides was added, as indicated in the figures legends. After 15 min, the 32 P-labeled duplex oligonucleotide (1.75 fmol) was added, and the incubation was followed by another 30 min at room temperature, in a total volume of 20 l.
DNA-protein complexes were separated on 6% non-denaturing polyacrylamide gels in Tris borate/EDTA buffer, pH 8.0, at room temperature and 120 V for 4 h. The gels were dried, and complexes were revealed by autoradiography.
For supershift assay, the nuclear extracts were preincubated with 1 l of antiserum (0.2 mg/ml) at room temperature for 1 h before analysis 2Ϫ . After 24-h incubation with different concentrations of TGF-␤, 50 l of medium was spotted on Whatman 3M paper. Low molecular mass radioactivity was dialyzed out into 1% cetylpyridinium chloride solution (59). The radioactivity incorporated into proteoglycan was counted by liquid scintillation. B, dose-dependent effect of TGF-␤ on the expression of human UGDH. MRC5 cells were cultured as described under "Experimental Procedures" and treated with TGF-␤ for 24 h. Total RNA was extracted from cells, and Northern blot analyses were performed to detect UGDH transcripts. Membranes were subsequently rehybridized with a GAPDH probe. A representative autoradiogram is shown along with normalized data in the form of a column graph. C, time course effect of TGF-␤ on the expression of human UGDH. Cultures were incubated with medium supplemented with TGF-␤ (10 ng/ml) for 0, 4, 8, 12, or 24 h. C24 indicates that cells were incubated for 24 h without TGF-␤ treatment. D, cycloheximide increases UGDH mRNA levels in MRC5 cells. Cultured cells were treated with medium alone or supplemented with TGF-␤ (10 ng/ml), or with cycloheximide (10 g/ml), or both, as indicated, for 8 h. Total RNA was extracted from cultures and analyzed by Northern blot. Results show the mean of triplicate experiments Ϯ S.E. by EMSA as described above. Human anti-Sp1 monoclonal antibody was obtained from Santa Cruz Biotechnology (SC-420) and from Geneka. Human anti-Sp3 polyclonal immunoserum was from Santa Cruz Biotechnology (SC-644). The Sp1 anti-sera specifically detect the presence of the Sp1 transcription factor, whereas the Sp3 anti-serum lightly cross-reacted with the Sp1 transcription factor (verified by Western blot, data not shown). Nonspecific IgGs were incubated with the nuclear extracts as described and used as negative control. Nuclear extract from HeLa cells (Geneka) were used as positive control.

RESULTS
Basal Expression of UGDH mRNA, Effects of TGF-␤, and Hypoxia-It was previously shown that UGDH is ubiquitously expressed in all analyzed tissues (8,15). We used a panel of different human cells in culture to establish the basal expression of the enzyme. All the tested cells expressed a double transcript of UGDH mRNA of 2.8 and 3.4 kb (Fig. 1), with a slightly different proportion of those two transcripts. The lowest mRNA expression was observed in umbilical vein endothelial cells and in monocytes, whereas HepG2 hepatoma cells displayed the highest level. The mRNA abundance suggests that the transformation of these epithelial cells of hepatic origin had little or no influence on UGDH expression. Skin fibroblasts and MRC5 embryonic pulmonary fibroblasts expressed an intermediate level of UGDH mRNA. In dermal fibroblasts, we observed some variability of the expression, depending on the age of donor, biopsy site, and cell passage number. Consequently, we decided to use MRC5 fibroblasts, which give more constant results as a model of low UGDH expression, to analyze the TGF-␤ stimulatory effects, and the HepG2 hepatoma cells as a model of high UGDH expression, to analyze the hypoxia inhibitory effects.
Because TGF-␤ is a major growth factor able to influence GAG synthesis, we studied its effects on UGDH expression by MRC-5 cells. TGF-␤ increased the incorporation of [ 35 S]sulfate into GAG fraction in a dose-dependent manner ( Fig. 2A). In parallel, TGF-␤ enhanced the steady-state level of UGDH mRNA. Both transcripts were equally affected (Fig. 2B). The effect was time-dependent, with pronounced expression after 8 h of stimulation ( Fig 2C). The expression was maximal after 12-h stimulation and persisted for 24 h. Cycloheximide had an additive effect (Fig. 2D), indicating that the stimulation did not need new protein synthesis and that UGDH is a direct response gene after TGF-␤ treatment. The late response of UGDH suggests, however, that the mechanism of stimulation may be complex and involve the recruitment of different transcription factors.
Hypoxia is another factor capable of influencing GAG synthesis (13). For that reason, we studied the effect of oxygen status on the UGDH mRNA expression in HepG2 cells. Twenty-four hours of hypoxia inhibited the steady-state UGDH mRNA level in a dose-dependent manner (Fig. 3A). Hypoxia also reduced the incorporation of [ 35 S]sulfate into the GAG in HepG2 cells cultures (Fig. 3B). The effect was similar in MRC5 cells (not shown).
Characterization of UGDH Promoter-To investigate the mechanism of the regulation of UGDH expression by TGF-␤ and hypoxia, the 3-kb genomic fragment, including the exon 1 was cloned from a BAC containing a part of the human chromosome 4p (6). From this construction, a subcloned fragment of 2183 bp, limited by EcoRI and SacI, was fully sequenced (Fig.  4). It contained 50 bases of the previously determined 5Ј untranslated region. A detailed computer analysis using the Transfac (36) and MatInspector (37) algorithms revealed a high probability score for the presence of the promoter. The inverted TATA box was located 22 bases upstream from the transcription start site, and no consensus CAAT sequence was found. The proximal fragment of 374 bp contained seven consensus sequences for Sp1 transcription factor and one consensus sequence for hypoxia-inducible factor-1 (39).
To ascertain the transcriptional promoter activity, a panel of constructions containing different lengths of the 5Ј-flanking region coupled to firefly luciferase reporter gene was generated (Fig. 5A). The transient transfection of the constructs drove the luciferase activity resulting from the promoter activity in both MRC5 and HepG2 cells (Fig. 5B). The luciferase level was always higher in HepG2 cells compared with MRC5 fibroblasts, even for the shortest constructs. Increasing the length of the promoter up to Ϫ644 bp enhanced the basal expression of the reporter gene. The longest construct, containing 2133 bp of the promoter, displayed about 30% of the activity compared with the Ϫ644-bp construct, suggesting the presence of inhibitory elements. The shortest construct (Ϫ100 bp) showed low, but measurable, promoter activity. This sequence contained only one Sp1 consensus site and an inverted TATA box. Mutation of this box decreased the reporter gene activity to the background level (Fig. 5C), indicating that Ϫ100 bp may be defined as a minimal promoter for the UGDH gene and that the inverted TATA box is necessary for its activity.
Influence of TGF-␤ and Hypoxia on Promoter Activity-The constructs containing different lengths of the UGDH promoter were transiently transfected to MRC5 cells and HepG2 cells to study the mechanisms of TGF-␤ and hypoxia effects, respectively. TGF-␤ (10 ng/ml) had no effect on minimal and Ϫ2133-bp promoters but enhanced the activity of all the constructs up to Ϫ644 bp (Fig. 6A). The strongest stimulation was observed for the constructs of Ϫ374 and Ϫ644 bp. These fragments contain six and eleven additional Sp1 consensus sequences, respectively (Fig. 4). When the plasmids were transiently transfected into the HepG2 cells maintained in 1% O 2 , the activity of the constructs Ϫ165 and Ϫ196 bp were enhanced about 2-fold in comparison to the cells cultured in 20% O 2 . The constructs containing Ϫ374, Ϫ644, and Ϫ2133 bp were strongly inhibited in hypoxic conditions suggesting that the hypoxia inhibitory site was located between Ϫ249 and Ϫ374 bp. This portion of the promoter contains three Sp1-binding sequences.
Implication of SP1 in UGDH Promoter Activity-Recently, it was shown that bisanthracycline (WP-631) acts as a selective inhibitor of Sp1 transcription factor (40). We used this compound to further study the implication of Sp1 in the activity of UGDH promoter. WP-631 at a concentration of 1 M slightly inhibited basal UGDH mRNA level in MRC5 fibroblasts and completely abolished the stimulatory effect of TGF-␤ (Fig. 7A). In MRC5 fibroblasts, WP-631 did not influence the activity of Ϫ100-bp and Ϫ165-bp constructs, indicating that the Sp1 cisacting elements of these sequences are not involved in the regulation of UGDH promoter. In keeping with the mRNA results, WP-631 inhibited all the longer promoter constructs and suppressed the stimulatory effect of TGF-␤ (Fig. 7B).
More intense inhibition was obtained when WP-631 was added to the cultures of HepG2 cells. At the concentration of 1 M, WP-631 significantly inhibited UGDH mRNA level (Fig.  8A). Moreover, in this cell line, WP-631 inhibited all the constructs activity in a dose-dependent manner (Fig. 8B).
The role of Sp1 was further investigated by EMSA studies with nuclear extracts, using a consensus Sp1 sequence DNA fragment. The nuclear extracts of MRC5 cells and HepG2 cells incubated with labeled probe showed on the gel three delayed complexes (Fig. 9A, first lane).
Using a 100ϫ excess of non-labeled Sp1 fragment (specific competition) (Fig. 9A, third lane), we observed a disappearance of the bands corresponding to Sp1 and Sp3, showing that the lowest one corresponded to nonspecific fixation, the weak one to FIG. 4. Nucleotide sequence of the 5-flanking region of the human UGDH gene. The 5Ј-flanking region of the human UGDH gene was isolated from BAC clone RP11-472B18 and sequenced. The transcription start site is indicated by the broken arrow. The left numbering is relative to the transcription start site, whereas the right numbering starts at the first nucleotide. Multiple putative binding sites for transcription factors are underlined. Putative Sp1 response elements and the HIF-1 element are underlined and in boldface. An inverted TATA box, located 25 bp upstream of the transcription start site, is shown in a gray box. Three Sp1 sequences studied in detail are double-underlined. the binding of SP3 factor (41), and the highest, which was also the more labeled, to SP1. In contrast to the specific competition experiment, a 100-fold molar excess of the AP1 consensus sequence had no effect on the intensity of the fixation (Fig. 9A,  fourth lane). TGF-␤ enhanced the fixation of the DNA consensus binding to Sp1 and Sp3 (Fig. 9A, second lane). The nuclear extracts obtained from HepG2 cells incubated in the presence of 1% O 2 showed weaker fixation of Sp1 factor as compared with normoxic cells (Fig. 9B).
Results from Fig. 6A showed that the main TGF-␤ enhancement effect was obtained with the Ϫ374 bp, construct (1.6-and 4.1-fold stimulation for the Ϫ249-bp and Ϫ374-bp constructs, respectively). The fragment of promoter Ϫ374/Ϫ249 contains three putative Sp1 binding sites (Fig. 4). To identify which potential Sp1 binding site from this fragment is involved in the TGF-␤ stimulation, we used three oligonucleotide probes from this region and a commercially available Sp1 consensus oligonucleotide as a control of migration in EMSA experiments. Incubation of nuclear extracts from control and TGF-␤-treated MRC5 cells with the different probes revealed an increase of Sp1-DNA binding under TGF-␤ treatment with Sp1 sequences located at Ϫ253 and Ϫ276 bp but not with the Sp1-312 probe (Fig. 10). The main binding effect was obtained with the Sp1-276 probe (Fig. 10, lane 6). The specificity of binding was checked using 100-fold molar excess of unlabelled doublestrand wild type consensus Sp1 sequence or site-directed Sp1 oligonucleotide mutant. The non-labeled wild type probe, but not the mutant probe, suppressed Sp1-and Sp3-shifted bands (Fig. 10B), indicating the specificity of binding. These results were further confirmed by supershift experiments (Fig. 11). By using both the commercial and the Sp1-276 probes, we demonstrated that the increased Sp1 binding activity present in TGF-␤-treated MRC5 cells was actually due to the Sp1 transcription factor. Two supershifted bands were observed when using the commercial anti-Sp1 antibodies. It is not clear at present whether these doublings correspond to partial fixation of another transcription factor or the formation of higher order complexes. Specificity of the Sp1-supershifted band was confirmed using (i) control nuclear extract from HeLa cells (Fig.  11) and with an anti-Sp1 antiserum from another supplier and (ii) supershift experiments with the anti-Sp3 antiserum (data not shown). Anti-Sp3 antiserum slightly cross-reacted with the under basal conditions. Each deletion construct was transiently transfected, together with PRL-TK plasmid containing the Renilla luciferase reporter gene, into the MRC5 or HepG2 cells. Twenty-four hours later, the firefly luciferase activity was measured and normalized to Renilla luciferase activity. The 100% activity was arbitrarily chosen for the Ϫ644-bp construct in HepG2 cells. Data are the means of at least three independent experiments performed in duplicates, with at least two different plasmid preparations. C, identification of the inverted TATA box element required for basal UGDH promoter activity in HepG2 cells. Sequences for the sense strand for wild type (WT) and substitution mutant are shown inside the box. The substitution mutant was generated by using the QuikChange site-directed mutagenesis kit (Stratagene) using the Ϫ100-bp construct WT as template. The WT and mutated constructs were transfected into HepG2 cells, and the resulting luciferase activity was measured. Sp1 transcription factor and was probably directed against the DNA-binding site, leading to a decrease in the complexes formation as shown by other authors (42). All these results strongly suggest that Sp1 is a universal modulator of UGDH promoter activity and that, according to its concentration/phosphorylation status and supply to the nucleus, it might increase or decrease UGDH mRNA expression. DISCUSSION UDP-glucuronate is an intermediate metabolite and the substrate for numerous glycosyltransferases, including those implicated in sulfated GAGs and hyaluronan synthesis. The syn-thesis of UDP-glucuronate is catalyzed by the unique pathway of UGDH in animals, and it was shown that mutation inactivation of the enzyme impairs embryonic growth by lack of GAGs. There is a body of evidence that the enzyme has a regulatory role in the cell. First, the relative concentration of UDP-glucuronate in the mesenchymal cells was estimated between 10 Ϫ4 and 10 Ϫ5 M (43, 44), a value similar to the K m of the downstream acting glycosyltransferases. Second, the expression of UGDH mRNA is higher in the cells and tissues with enhanced UDP-glucuronate metabolism as, for example, liver, or kidney (Ref. 8 and Fig. 1). Third, UGDH is a direct response gene as demonstrated by time-course experiments (Fig. 2C) FIG. 7. Sp1 binding is necessary for UGDH mRNA expression in MRC5 cells. A, total RNA from cells incubated for 24 h with or without TGF-␤ (10 ng/ml) and without or with WP-631, an inhibitor of Sp1 binding, was subjected to Northern blot analysis. The blot was probed with the UGDH cDNA fragment, then stripped and reprobed with GAPDH probe. Ethidium bromide staining for 28 S rRNA was used as a control for loading. A representative autoradiogram is shown along with normalized data in the form of a column graph. B, effects of WP-631 on UGDH promoter activity. Each promoter was transiently transfected into MRC5 cells, and luciferase activity was measured after 24 h of incubation in the absence or presence of TGF-␤ (10 ng/ml) or WP-631 (1 M), or both. Firefly luciferase activity was normalized to Renilla luciferase activity and expressed as a percentage of the Ϫ644-bp construct. Results are shown as the mean Ϯ S.E. of three independent experiments. and by the cycloheximide-enhanced response to TGF-␤ stimulation (Fig. 2D). Similar results were previously reported with interleukin-1 (15).
To investigate the molecular regulation of UGDH, we cloned its 5Ј-upstream region that drove the promoter activity of luciferase reporter gene in transient transfection assays. Present work is the first report of the characterization of UGDH promoter. The minimal promoter (Fig. 5) contains an inverted TATA box, as it was shown for several mammalian genes (45)(46)(47)(48), and no CAAT box was detected in the promoter up to Ϫ2133 bp. The core promoter (Fig. 4) contains numerous GC-rich sequences, characteristic of TATA-less and CAAT-less promoters of housekeeping genes (49). In such genes, Sp1 consensus sites were shown to take the function of TATA boxes for polymerase II assembly (50). Here, the inverted TATA box of UGDH promoter drove the basal activity of the reporter gene (Fig. 5), a feature characteristic of "atypical" housekeeping genes that show a low level of basal expression but are regulated by different extracellular factors, including cytokines. This feature of UGDH promoter is distinct from other sugar or alcohol dehydrogenases (51), including the enzymes of the glycolytic pathway (52,53). The UGDH core promoter contains twelve Sp1 consensus sequences. The successive deletion analysis of different fragments showed a decreased activity of the reporter gene proportional to the promoter length, up to Ϫ644 bp. The full cloned sequence of Ϫ2133 bp was much less active, indicating the presence of putative inhibitor sequence. The core promoter of about 650 bp was responsible for the increased activity in HepG2 cells compared with MRC5 fibroblasts, indicating that the major transcription regulatory cis-elements are located in this region from these two cell types. The increase of the promoter activity correlated with its length and its downregulation by bisanthracycline suggest that several of the Sp1 sites are functional in defining HepG2 cell activity.
The Sp1 transcription factor can act as an initiator for assembling polymerase II complex (50,54), but often it acts as an enhancer/modulator for different kinds of signals (55). Expression of UGDH mRNA is up-regulated by serum (8,17). TGF-␤ is one of the major factors in the serum that enhances GAG synthesis. TGF-␤ increased UGDH mRNA abundance rapidly FIG. 8. Bisanthracycline (WP-631) acts as a mimetic of hypoxia on UGDH expression in HepG2 cells. A, total RNA extracted from cells incubated for 24 h with or without WP-631 (1 M) was analyzed by Northern blot. The blot was probed with UGDH, then stripped and reprobed with GAPDH. Ethidium bromide staining for 28 S rRNA was used as a control for loading. A representative autoradiogram is shown along with normalized data in the form of a column graph. B, effects of WP-631 on UGDH promoter activity. Each promoter construct was transiently transfected into HepG2 cells and luciferase activity was measured after 24 h of incubation in the absence or presence of increasing concentration of WP-631 (0, 10, and 100 nM or 1 M). Firefly luciferase activity was normalized to Renilla luciferase activity and expressed as a percentage of the Ϫ644-bp construct.
FIG. 9. Electrophoretic mobility shift assay of the effect of TGF-␤ and hypoxia on Sp1 DNA binding. MRC5 cells were incubated with TGF-␤ (10 ng/ml), and HepG2 cells were exposed to hypoxia (1% O 2 ) for 24 h. Cells were harvested, then nuclear extraction and gel mobility shift assays were performed as described under "Experimental procedures." A, TGF-␤ enhanced specifically Sp1 and Sp3 binding to DNA. Binding with nuclear extracts (3 g of protein was used for each assay) from untreated MRC5 cells (lane 1) were compared with nuclear extracts from cells treated with 10 ng/ml TGF-␤ (lane 2). The following two lanes represent binding with nuclear extracts from untreated MRC5 cells preincubated with a 100-fold molar excess of unlabeled Sp1 consensus sequence (lane 3) or AP1 consensus sequence (lane 4). B, hypoxia-decreased Sp1 binding. Nuclear extracts from untreated HepG2 cells (lane 1) were compared with nuclear extracts from HepG2 cells subject to 1% O 2 for 24 h. and in a dose-dependent manner, with a sustained effect after 24-h treatment. Cycloheximide had an additive effect, indicating the direct action of TGF-␤. In mammals, the signaling of TGF-␤ is transduced via phosphorylated SMAD proteins, which bind to the target genes by its Smad binding element (56). However, this binding is of low affinity and always requires the interaction with additional transcription factors to become functional. There is a body of evidence showing that Sp1 cooperates in TGF-␤ signal transduction. Here, we showed that the promoter of human UGDH does not contain any consensus Smad binding element (Fig. 4) but includes many Sp1 sites in its sequence. On the other hand, all the constructs except that of Ϫ100 bp showed enhanced reporter gene expression after TGF-␤ treatment of MRC5 cells (Fig. 6). The most efficient were the constructs containing Ϫ374 and Ϫ644 bp of the UGDH promoter. These constructs have three and eight additional Sp1 binding motifs, respectively, compared with the Ϫ249-bp construct. Moreover, the specific Sp1 inhibitor, WP-631, completely suppressed the TGF-␤ stimulation of UGDH mRNA expression and blocked the activity of all the constructs in HepG2 cells. There was a specific fixation of nuclear proteins on Sp1 consensus sequence oligonucleotide as revealed by EMSA and supershift assays. Accordingly, we show that two Sp1 sites located at Ϫ253 and Ϫ276 bp upstream from the transcription start site are involved in the enhanced binding after TGF-␤ stimulation, the Ϫ276 bp being the most active (Fig. 10). All these results strongly support the hypothesis that TGF-␤ acts on the human UGDH promoter through numerous Sp1 sites. The differences in the susceptibility of MRC5 cells and HepG2 cells to WP-631 treatment are correlated with the differences in their basal expression of UGDH mRNA. The higher susceptibility of HepG2 cells to WP-631 inhibition may be ascribed for enhanced metabolism and enhanced growth of this hepatocarcinoma cell line. Either the pool of Sp1 factor might be used to stimulate multiple gene expression, or the chromatin might be less protected and more attractive to Sp1.
Another factor influencing UGDH expression is hypoxia. In the human body, an acute hypoxic state may be acquired in rapidly growing tumor tissue with poor vasculature and in wound healing, when the blood vessels are damaged, or, in physiological state, in working muscular tissue. The hypoxia is created when the oxygen/energy consumption surpass its delivery. Hypoxia impairs GAG synthesis with preferential inhibition of chondroitin/dermatan sulfate production (57). In hypoxic conditions, the cell metabolism, including energy production, is switched to anaerobic phosphorylation, which produces a large quantity of lactate and acidifies the cell environment. It was shown that, in hypoxic conditions, at least 9 among the 11 glycolytic enzymes are up-regulated (58). The reaction catalyzed by UGDH consumes 2 mol of NAD ϩ for each mole of UDP-glucuronate synthesized. In hypoxic conditions, this NAD ϩ is no more available for the oxidative phosphorylation pathway. It is therefore conceivable that UGDH synthesis may be down-regulated when oxygen supply is low. The inhibition is relatively rapid, and the dose-dependent decrease of UGDH mRNA abundance is observed after 24-h exposure (Fig.  3). The down-regulation of the reporter gene was observed in transient transfection experiments with the constructs containing the Ϫ374-bp and Ϫ644-bp core promoters (Fig. 6). The inhibition of UGDH promoter activity was, at least in part, mediated by the Sp1 factor in HepG2 cells. Indeed, WP-631 treatment mimics hypoxic conditions, and the EMSA assay showed a decreased capacity for Sp1 fixation on its consensus sequence in the nucleus from hypoxic cells. All these results indicate that the Sp1 transcription factor can play a determinant role in the promoter activity of UGDH. Sp1 can either increase the UGDH expression after TGF-␤ stimulation or inhibits its expression in hypoxic conditions. Accordingly, the enzyme concentration could control the UDP-glucuronate supply for glycosaminoglycan synthesis and glucuronoconjugate synthesis.