Sp1 and Sp3 Mediate Constitutive Transcription of the Human Hyaluronan Synthase 2 Gene*

The linear glycosaminoglycan hyaluronan (HA) is synthesized at the plasma membrane by the HA synthase (HAS) enzymes HAS1, -2, and -3 and performs multiple functions as part of the vertebrate extracellular matrix. Up-regulation of HA synthesis in the renal corticointerstitium, and the resultant extracellular matrix expansion, is a common feature of renal fibrosis. However, the regulation of expression of these HAS isoforms at transcriptional and translational levels is poorly understood. We have recently described the genomic structures of the human HAS genes, thereby identifying putative promoter regions for each isoform. Further analysis of the HAS2 gene identified the transcription initiation site and showed that region F3, comprising the proximal 121 bp of promoter sequence, mediated full constitutive transcription. In the present study, we have analyzed this region in the human renal proximal tubular epithelial cell line HK-2. Electrophoretic mobility shift and promoter assay data demonstrated that transcription factors Sp1 and Sp3 bound to three sites immediately upstream of the HAS2 transcription initiation site and that mutation of the consensus recognition sequences within these sites ablated their transcriptional response. Furthermore, subsequent knockdown of Sp1 or Sp3 using small interfering RNAs decreased constitutive HAS2 mRNA synthesis. In contrast, significant binding of HK-2 nuclear proteins by putative upstream NF-Y, CCAAT, and NF-κB recognition sites was not observed. The identification of Sp1 and Sp3 as principal mediators of HAS2 constitutive transcription augments recent findings identifying upstream promoter elements and provides further insights into the mechanism of HAS2 transcriptional activation.

The linear glycosaminoglycan hyaluronan (HA) is synthesized at the plasma membrane by the HA synthase (HAS) enzymes HAS1, -2, and -3 and performs multiple functions as part of the vertebrate extracellular matrix. Up-regulation of HA synthesis in the renal corticointerstitium, and the resultant extracellular matrix expansion, is a common feature of renal fibrosis. However, the regulation of expression of these HAS isoforms at transcriptional and translational levels is poorly understood. We have recently described the genomic structures of the human HAS genes, thereby identifying putative promoter regions for each isoform. Further analysis of the HAS2 gene identified the transcription initiation site and showed that region F3, comprising the proximal 121 bp of promoter sequence, mediated full constitutive transcription. In the present study, we have analyzed this region in the human renal proximal tubular epithelial cell line HK-2. Electrophoretic mobility shift and promoter assay data demonstrated that transcription factors Sp1 and Sp3 bound to three sites immediately upstream of the HAS2 transcription initiation site and that mutation of the consensus recognition sequences within these sites ablated their transcriptional response. Furthermore, subsequent knockdown of Sp1 or Sp3 using small interfering RNAs decreased constitutive HAS2 mRNA synthesis. In contrast, significant binding of HK-2 nuclear proteins by putative upstream NF-Y, CCAAT, and NF-B recognition sites was not observed. The identification of Sp1 and Sp3 as principal mediators of HAS2 constitutive transcription augments recent findings identifying upstream promoter elements and provides further insights into the mechanism of HAS2 transcriptional activation.
Hyaluronan (HA) 3 is a linear non-sulfated glycosaminoglycan found commonly in the vertebrate extracellular matrix and which has a variety of functions during and following development (1)(2)(3)(4)(5)(6). HA is synthesized by the HA synthase (HAS) enzymes that are encoded by the cor-responding multigene family HAS1, -2, -3a, and -3b (7-10), and its importance in the extracellular matrix is underlined by the expanding range of pathological contexts in which modified or aberrant HA metabolism appears to play a role. These include malignancy, osteoarthritis, and pulmonary and vascular disorders, along with other immune and inflammatory diseases (11)(12)(13)(14)(15)(16)(17)(18)(19). HA has also been implicated in regenerative processes such as wound healing (e.g. Ref. 6 and Refs. 20 -22) and as a key mediator of the immune process (19).
Under homeostasis in the healthy kidney, the expression of HA in the cortical interstitium is low, with high levels found only in the renal papilla. Following acute ischemic injury, interstitial inflammation, or during progressive renal fibrosis, however, there is greatly increased peritubular expression of both HA and the cell surface HA receptor CD44 in the cortex (23)(24)(25)(26)(27)(28). This suggests that alterations in HA synthesis and turnover may be involved either in the maintenance of homeostasis or in the development of pathological events, but the role of HA remains unclear. To date, HAS2 appears to be the most frequently induced isoform in the context of clinical nephrology, and increased HAS2 transcription, together with raised levels of HA synthesis, have been reported in renal human proximal tubular epithelial HK-2 cells cultured in vitro in glucose concentrations similar to those found in diabetic nephropathy (29). HAS2 up-regulation has also been reported in the inflammation that occurs commonly as a consequence of peritoneal dialysis (20) and in autoimmune renal injury (30).
Despite the multifunctional role of HA in this wide variety of physiological and pathological processes, comparatively little is known about the transcriptional activation of the human HAS genes. Recent reports provide evidence for a natural antisense for HAS2 (31) and for enhanced transcriptional activation by STAT3 and retinoic acid (RA) response elements (RAREs (32)). However, details of the factors governing constitutive HAS2 activation have yet to be established.
To investigate the transcriptional regulation of the human HAS genes, we first deduced their genomic structures and thus identified putative proximal promoter regions for each isoform (33). Using 5Ј-rapid amplification of cDNA ends analysis and promoter activity data from luciferase assays, we then extended the exon 1 sequence of HAS2 by 130 nucleotides, repositioning the HAS2 transcription initiation site (TIS (34)). In addition, we found that luciferase reporter construct F3, containing an insert spanning a 121-bp region upstream of the HAS2 TIS, mediated full constitutive transcriptional activity (34). Analysis of the HAS2 proximal promoter region in silico (35) identified a range of putative upstream transcription factor-binding sites (TFBSs). These included Sp1, NF-Y, and CCAAT sites within the F3 region, and an NF-B motif further upstream.
In the present study, we have analyzed these TFBSs in HK-2 cells. Electrophoretic mobility shift assays (EMSAs) demonstrated binding of Sp1 and Sp3 to three oligonucleotide probes spanning 63 bp immediately upstream of the HAS2 TIS. Mutation of key residues in consensus Sp1/Sp3 recognition sites within these regions ablated the ability of the F3 promoter region to drive transcription of the luciferase gene. Small interfering RNA (siRNA) knockdown of mRNAs for either Sp1 or Sp3 resulted in a significant decrease in HAS2 transcription levels. However, although the stimulation of these cells with interleukin (IL)-1␤ has been reported to result in up-regulation of HAS2 mRNA via NF-B (29), this effect was not mediated by the putative site for this transcription factor in the proximal promoter. In addition, putative recognition sites for transcription factors NF-Y and CCAAT failed to bind HK-2 nuclear proteins.

MATERIALS AND METHODS
Cell Culture-Culture of the neuroblastoma cell line TE671 and the renal proximal tubular epithelial cell line HK-2 cells was as described previously (34). Recombinant IL-1␤ was obtained from R&D Systems Europe Ltd. (Abingdon, Oxfordshire, UK), SN50 was from Merck Biosciences Ltd. (Nottingham, Nottinghamshire, UK), and both were used in accordance with guidance from the manufacturer.
Preparation of Nuclear Extracts-Nuclear extracts were prepared from TE671 and HK-2 cultures in 75-cm 2 flasks (BD Biosciences, Cowley, Oxfordshire, UK). At 70% confluence, the growth medium was replaced with 10 ml of Optimem (Invitrogen Ltd., Paisley, Renfrewshire, UK), which was replaced 24 h later with 10 ml of low nutrient growth medium (Invitrogen). After a further 24 h, this was removed, and nuclear proteins were extracted by standard means and stored at Ϫ70°C, with the Bradford dye binding assay used to quantify nuclear protein extract concentrations at 595 nm (36). Where appropriate, cells were stimulated with 10 ng/ml IL-1␤ for 30 min prior to nuclear protein extraction.
EMSA-Complementary single-stranded oligonucleotides (Invitrogen) with 5Ј-CGA overhangs were annealed and Klenow-filled using [ 32 P]dCTP (Amersham Biosciences UK Ltd., Chalfont St. Giles, Buckinghamshire, UK) with the MegaPrime DNA labeling kit (Amersham Biosciences), and the labeled probes were purified using ProbeQuant G-50 micro columns (Amersham Biosciences) according to the manufacturer's instructions. The locations of probe regions, mutant and consensus probe sequences, luciferase reporter construct inserts, and PCR primers are illustrated in Fig. 1; oligonucleotide sequences are given in Table 1.
EMSAs were performed as we have described previously, using 1 ng of labeled probe, 5 g of nuclear protein, and a 100-fold excess of unlabeled probes in competition analyses (36). These included consensus probes for Sp1, NF-B, and NF-Y (Table 1, sequence data from Santa Cruz Biotechnology Inc. Santa Cruz, CA) and an oligonucleotide con-FIGURE 1. Locations of EMSA probes, consensus TFBS motifs, and wild-type and mutated HAS2 promoter constructs. The 5Ј terminus of HAS2 sequence AJ604570 (34) is shown as nucleotide ϩ1. The six EMSA probes used in the present study are labeled, and their sequences shown in bold color and in uppercase, except where described below. Probes 1-5 are contiguous between positions Ϫ2 and Ϫ121 of the F3 region (34). Selected putative TFBSs identified by in silico analysis (35) are highlighted as described below and designated as sense strand (s-s) or antisense strand (a-s) matches. The a-sNF-B site within the Probe 6 sequence is shown in lowercase, and the overlapping s-sSTAT motif (see also Ref. 6) is underlined; the s-sNF-Y motif within the Probe 4 sequence is shown in lowercase, with the overlapping a-sCCAAT underlined; and the a-sSp1 motif in the Probe 3 sequence and overlapping a-sSp1 motifs in the Probe 2 sequence are shown in lowercase. The s-s PCR primer sequences for the amplification of inserts for vectors F3 and F6 (34) are labeled, and their common a-s primer-binding site is labeled R1; each is shown in uppercase. The F3 sequence is identical to that of Probe 5 and also represents the 5Ј terminus of the inserts of constructs HAS2-Sp1 and mut-HAS2-Sp1; the region of HAS2 exon 1 spanned by these inserts is shown in bold and lowercase. taining the STAT-binding consensus sequence SIE-m67 as an unrelated control (37). Mutant probes mut-1-4 were designed to alter key residues in the putative binding sites for Sp1 and NF-Y (38 -40). EMSA and NF-B supershift reaction products (see below) were separated by electrophoresis through 6% non-denaturing polyacrylamide gel at 150 V for 4 h at 4°C and dried prior to autoradiography at Ϫ80°C. For supershift assays, 10 g of Sp1-, Sp3-, or NF-B-subunit-specific antibodies (Santa Cruz Biotechnology (36,41)) was incubated with 10 g of nuclear protein for 20 min at room temperature prior to the addition of radiolabeled probe by means we have detailed (36,41). Where appropriate, binding of these polyclonal, rabbit-reared antibodies was compared with preimmune rabbit serum as reported previously (42). Electrophoresis of supershift reaction products with Sp-specific antibodies was carried out through 4% non-denaturing polyacrylamide gel at 150 V for 7 h at 4°C, with autoradiography as described above.
Synthesis of Promoter Construct Inserts-The mutant probe mut-HAS2-Sp1 was designed to alter key residues in the putative binding sites for Sp1 identified in the present study ( Fig. 1) (38). In addition, on the basis of our EMSA data, we included a site within the Probe 1 sequence that was identified by visual inspection and not in silico, as discussed below. Two oligonucleotides spanning the Probes 1-5 sequences ( Fig. 1) were thus designed with overlapping, central, complimentary portions and obtained from Sigma (Gillingham, Dorset, UK); their sequences are given in Table 1. A total of 1 g of each oligonucleotide was diluted in 5 l of H 2 O, mixed together, and annealed in a 100-l reaction containing 10 l of 10ϫ Buffer 3 (New England Biolabs, Hitchin, Hertfordshire, UK) and 80 l of H 2 O. After 10 min at 100°C, the reaction mixture was allowed to cool to room temperature. The 65-bp 5Ј-overhangs either side of the central complimentary region of the annealed product were then filled using Klenow and the Megaprime kit according to the manufacturer's instructions. Finally, the product was purified as described above.
This purified template was then amplified by PCR using the HAS2-Sp1 primers shown in Table 1. In common with inserts for our previously created HAS2 constructs (34), the sense and antisense strand primers bore KpnI and HindIII tails, respectively, for cloning into a modified pGL-3 luciferase reporter vector (43). A wild-type promoter fragment (HAS2-Sp1) of the same length as mut-HAS2-Sp1 was generated using a genomic DNA template (44). PCR was carried out in a reaction volume of 15 l comprising 6 l of template, 1ϫ PCR buffer containing 1.5 mM MgCl 2 (Applied Biosystems, Warrington, Cheshire, UK), 5 mM dNTPs, 2.5 units of Amplitaq Gold Taq polymerase (Applied Biosystems), and 1 M each primer. Touchdown thermocycling followed a denaturation step of 94°C for 5 min and comprised an initial cycle of 30 s at 94°C, 30 s at 65°C, and 30 s at 72°C; thereafter the annealing temperature was reduced by 0.5°C per cycle for the next 10 cycles, remaining at 60°C for the concluding 25 cycles. Products were then sized using flat-bed agarose gel electrophoresis and visualized with ethidium bromide.
Promoter Activity Analysis-The above PCR-amplified HAS2 promoter fragments were purified using the QIAquick gel extraction kit (Qiagen Ltd., Crawley, West Sussex, UK), digested (restriction endonucleases from New England Biolabs), ligated into our modified pGL-3 luciferase reporter vector (43), and sequenced to ensure fidelity of amplification using vector-specific primers RV (5Ј-CTAGCAAAAT-AGGCTGTCCC-3Ј) and GL-2 (5Ј-CTTTATGTTTTTGGCGTCT-TCC-3Ј). Transient transfection into HK-2 cells cultured in 6-well plates (BD Biosciences) was carried out using FuGENE 6 transfection reagent (Roche Diagnostics Ltd., Lewes, East Sussex, UK) in accordance with the manufacturer's advice. The ability of each HAS2 promoter fragment to drive transcription of the luciferase gene was tested using the Dual Luciferase reporter assay kit (Promega UK Ltd., Southampton, Hampshire, UK) according to the manufacturer's instructions.
The regions of the HAS2 promoter inserted in our previously described luciferase vectors F3 and F6 (34), and which were used in our second series of luciferase analyses in the present study, are illustrated in Fig. 1. As a positive control, a construct containing a 181-bp fragment of the IL-8 promoter (45) was used that we knew to be IL-1␤-responsive via NF-B in human peritoneal mesothelial cells. 4 We have described the growth medium to maximize luciferase output elsewhere (34).
siRNA Gene Knockdown-Annealed oligonucleotides for the siRNA knockdown of Sp1 (catalog number 143158) and Sp3 (catalog number 115337) and scrambled negative control transfection (catalog number 4611) were purchased from Ambion (Huntingdon, Cambridgeshire, UK) and used in accordance with the manufacturer's guidelines. Transfection of siRNAs into HK-2 cells cultured in 12-well plates (BD Biosciences) was carried out according to the manufacturer's instructions. Each siRNA was used at a final concentration of 30 nM, and transfection took place over 48 h prior to further analysis.
Reverse Transcription and Quantitative PCR (qPCR)-Extraction of total RNA and first-strand cDNA synthesis by reverse transcription were carried out as described previously (34). TaqMan gene expression assays for HAS2 (Hs00193435_m1), Sp1 (Hs00412720_m1), Sp3 (Hs01595812_mH), and 18 S rRNA (4310893E) were purchased from Applied Biosystems and used as recommended by the manufacturer.
For each reaction, 1 l of cDNA was added to 24 l of a qPCR multiplex reaction mix for one target gene plus 18 S rRNA. Samples were assayed in quadruplicate in ABI Prism 96-well optical reaction plates (Applied Biosystems) in an ABI Prism 7000 sequence detection system (Applied Biosystems). Default cycling parameters comprised an initial cycle of 50°C for 1 min, one cycle of 95°C for 10 min, and 40 cycles of both 15 s at 95°C and 1 min at 60°C. Output data were analyzed using ABI PRISM 7000 sequence detection system software (Applied Biosystems).
Sequence Data Base Analysis-HAS2 locus sequences were retrieved from the genome browser at the UCSC Genome Bioinformatics Site and analyzed for the presence of putative TFBSs using the updated MatInspector program from the Genomatix suite (35). Selected putative TFBSs are shown in Fig. 1.
Statistical Analysis of Promoter Activity Assay Data-Data were calculated as the ratio of the fluorescence values for each HAS2 promoter luciferase reporter construct to the corresponding value for the cotransfected Renilla vector, and p Ͻ 0.05 was considered statistically significant. Where appropriate, statistical analysis was performed using the Wilcoxon signed rank test from SPSS 11.5 for Windows (SPSS Inc., Chicago, IL).
Statistical Analysis of qPCR Data-Fold changes in expression were calculated using 2 Ϫ(⌬Ct1Ϫ⌬Ct2) , where ⌬Ct represents the difference between amplification threshold for each target gene and 18 S rRNA. Values for p were calculated by analysis of variance using Microsoft Excel and were considered significant below 0.05.

RESULTS
EMSA Analysis of the F3 Region Immediately Upstream of the HAS2 TIS-Polyacrylamide gel electrophoresis of contiguous EMSA Probes 1-5 (detailed in Fig. 1) is shown in Fig. 2A. The EMSA profiles seen after incubation with HK-2 nuclear extracts, and labeled a, were characteristic of binding by the transcription factors Sp1 and Sp3 (36,46,47). This profile, comprising an upper doublet and lower singlet bands, was strongest in lane 1, with reproducible but less pronounced binding in lanes 2 and 3. Probe 4, which contained putative NF-Y and CCAAT motifs did not bind nuclear proteins specifically, whereas Probe 5 showed evidence of weak protein binding, labeled b. Weak nonspecific binding, labeled c, was seen with Probes 1-5. These profiles were very similar to those obtained using nuclear extracts from TE671 cells (data not shown), which we have used previously in luciferase analysis of the HAS2 promoter (33,34). No further comparisons between TE671 and HK-2 cells were carried out. Fig. 2B shows the results of EMSA analysis using Probe 1 in competition with an excess of selected unlabeled probes. HK-2 nuclear protein binding to radiolabeled Probe 1 alone resulted in the same profile as seen in Fig. 2A, labeled d in Fig. 2B. The addition of excess unlabeled Probe 1 resulted in the ablation of this binding. This finding was replicated with the addition of excess unlabeled consensus Sp1 probe, providing evidence that Probe 1 binding was specific for this consensus sequence, which also binds the related protein Sp3 (36,46,47). Excess unlabeled unrelated STAT-binding probe SIE-m67 probe, or mutated Probe 1, did not interrupt binding of nuclear proteins. These data were replicated with Probes 2 and 3 with proportional intensity to the binding shown in Fig. 2A (data not shown). Intermittent band e was not ablated by competition with either unlabeled Probe 1 or the consensus Sp1 probe and was assumed to be nonspecific.
The results of the corresponding analysis for Probe 4 are displayed in Fig. 2C. A comparison of Fig. 2, A and C, illustrated both transitory (band f ) and constant (bands c and g) products of nonspecific binding. Despite the fact that in silico analysis for TFBSs routinely identifies one or more putative sites per 10 bp of input sequence (33), the weak protein interaction to Probe 5 was not supported by strong evidence of TFBSs in this region. Probes 4 and 5 were thus not included in further analyses.
Supershift Analysis for Sp1 and Sp3 Using Probes 1-3- Fig. 3 shows supershift EMSA data for Probes 1-3 using Sp1-and Sp3-specific antibodies. In the absence of antibodies, protein binding patterns characteristic for transcription factors Sp1 and Sp3 were again observed. Bands a and b shown in Fig. 3 represented the upper doublet band from Fig. 2A resolved by electrophoresis through the lower percentage polyacrylamide gel, with band c below (36,46,47). The upper Sp1-specific band a was supershifted following incubation with Sp1-specific antibody, whereas the Sp3-specific bands b and c were similarly displaced by the Sp3-specific antibody. Each probe showed variation in binding intensity consistent with that seen in EMSA analysis in Fig. 2A. Indeed, for purposes of clarity, a composite gel is provided for Fig. 3 in which the Probe 1 result was exposed for 2 days, but the results for Probes 2 and 3 were exposed for 5 days. In combination with the findings from Fig. 2, these data demonstrated that Probes 1-3 interacted directly with Sp1 and Sp3 proteins and that the magnitude of binding was in the order Probe 1 Ͼ Probe 3 Ͼ Probe 2.
Promoter Activity Analysis of HAS2 Constructs- Fig. 1 shows that the insert HAS2-Sp1 and the corresponding mutated mut-HAS2-Sp1 spanned positions Ϫ121 to ϩ77 of the HAS2 promoter, thus including the promoter sequence from the F3 construct that mediated full constitutive transcriptional activity in our previous luciferase analysis (34). Following the demonstration of specific Sp1-and Sp3-protein interactions with Probe sequences 1-3, we carried out luciferase analysis on constructs in which the key residues within these probe sequences corresponding to the Sp1 core consensus sequence (38) were mutated, but the sequences for Probes 4 and 5 were unaltered. The results in Fig. 4 illustrated a pronounced difference in the luciferase activity of the wildtype construct HAS2-Sp1 when compared with the empty pGL-3 vector and the mut-HAS2-Sp1 vector. These data further emphasized the importance of the interaction of transcription factors Sp1 and Sp3 with the HAS2 proximal promoter and suggested that the Probe 4 and Probe 5 sequences did little to mediate constitutive HAS2 transcription in the absence of functional sequences for Probes 1-3.
Effect of siRNA Knockdown of Sp1 and Sp3 on HAS2 Expression-The ability of siRNAs specific to Sp1 and Sp3 to knock down their respective mRNAs was investigated in HK-2 cells, and the results of these experiments are shown in Fig. 5, A and B, respectively. Relative expression of each target mRNA was evaluated by qPCR 48 h following siRNA transfection. In each case, a reduction of the target gene mRNA level by Ͼ70%, in comparison with scrambled negative control siRNA, was  Table 1). Bands labeled a-g are discussed under "Results." obtained. Fig. 5C shows the effect of Sp1 and Sp3 knockdown on the transcription of HAS2 in HK-2 cells transfected with siRNA for 48 h prior to RNA extraction. Knockdown of Sp1 mRNA decreased HAS2 transcription by Ͼ55%, whereas reduction of Sp3 mRNA synthesis resulted in a decrease of almost 75% in HAS2 mRNA levels.
EMSA and Supershift Analysis of Probe 6 and NF-B Consensus Probes- Fig. 6A shows that IL-1␤ stimulation lead to the accumulation of proteins in HK-2 nuclei that bound strongly to the NFB consensus probe. At both 10-s and 30-min time points, this up-regulation was inhibited by the addition of SN50, a cell-permeable peptide inhibitor of NF-B (48). In contrast, the modest binding of these proteins to the HAS2-specific Probe 6 sequence seen in Fig. 6B did not appear to be up-regulated following IL-1␤ treatment, and SN50 had little effect on this binding. Competition and supershift EMSA data using antibodies to the various NF-B subunits (Fig. 6C) demonstrated the presence of NF-B subunits p50 and p65 in these HK-2 nuclear extracts.
Promoter Activity Analysis of HAS2 and IL-8 Constructs-In addition to the data in Fig. 6, we carried out promoter activity analysis on HAS2 luciferase promoter constructs from our previous study (34). The empty pGL-3 vector shown in Fig. 7 was compared with the corresponding vector containing the HAS2 TIS and sufficient upstream sequence to elicit peak luciferase response (34) in the absence (F3) and presence (F6) of the Probe 6 sequence spanning the putative NF-B site detailed in Fig.  1. The data in Fig. 7 showed that IL-1␤ addition to HK-2 cells transfected with constructs F3 and F6 did not result in a change in luciferase output. In contrast, a highly significant increase in luciferase activity was seen using the positive control IL-8 construct (45) shown previously to be to IL-1␤-responsive via NF-B. 4  Luciferase activity is expressed as the ratio of luciferase activity of test pGL-3 constructs to that of co-transfected Renilla constructs. Test constructs are labeled: pGL-3, empty luciferase reporter vector; F3 and F6, HAS2 promoter constructs detailed in the legend for Fig. 1 (34); IL-8, IL-8 promoter construct described under "Results" (44). White bars designate unstimulated HK-2 cells, and shaded bars signify cells stimulated with 10 ng/ml IL-1␤ for 30 min. Data are shown as means Ϯ S.E. of the mean from one representative experiment of three (n ϭ 3). . Promoter activity of wild-type and mutated HAS2 constructs spanning Probes 1-5. Luciferase activity is expressed as the ratio of luciferase activity of test pGL-3 constructs to that of co-transfected Renilla constructs. Test constructs are labeled: pGL-3, empty luciferase reporter vector; mut-HAS2-Sp1, HAS2 proximal promoter with key residues in three putative Sp1 sites altered by point mutation; HAS2-Sp1, wild-type sequence of mut-HAS2-Sp1 (as detailed in Table 1). Data are shown as means Ϯ S.E. of the mean from one representative experiment of three (n ϭ 3).

DISCUSSION
HA has been implicated in a diverse range of extracellular matrixmediated processes including cell adhesion, migration, and proliferation. To date, however, comparatively little is known of the regulation of expression of the human HAS enzymes that synthesize HA at the cell membrane and via which this multifunctional molecule is extruded to the extracellular matrix. We have described the genomic structures of the human HAS genes and thereby identified putative upstream sequences that elicit their transcriptional regulation (33). We then identified the HAS2 TIS and the adjacent promoter region and showed consistent output in quiescent cells of a nested set of luciferase reporter vectors spanning 1 kb of upstream sequence (34). The F3 construct contained the smallest promoter insert that mediated this full constitutive transcriptional activity, spanning 121 bp upstream of the HAS2 TIS (34), and thus formed the starting point of the present investigation.
We began with an in silico analysis of the sequence immediately upstream of the HAS2 TIS using updated software (35). As shown in Fig.  1, among the putative TFBSs identified in the F3 region were two Sp1binding sites, one NF-Y and one CCAAT motif. In addition, further upstream at position Ϫ241 to Ϫ256, a putative NF-B-binding site was found.
The binding of proteins from HK-2 nuclear extracts to radiolabeled Probes 1-3 directly upstream from the HAS2 TIS gave typical EMSA profiles showing commonly occurring Sp1 and multiple Sp3 isoforms (36,46,47). The specificity of protein binding was demonstrated for Probes 1-3 and was exemplified using Probe 1, as shown in Fig. 2B. These data demonstrated that protein binding was removed on the addition of excess unlabeled Probe 1 and with excess unlabeled Sp1 consensus probe but not with unrelated or mutated controls. Co-reactivity between HAS2-specific sequence Probe 1 and the consensus Sp1 motif was thus observed. Interestingly, as shown in Fig. 1, our in silico analysis revealed putative Sp1 sites for Probes 2 and 3 but not for the Probe 1 sequence. Nevertheless, visual inspection of the Probe 1 sequence revealed the motif CCCTCCCC, the sense strand complement of which, GGGGCGGG, corresponds to the Sp1 core motif in which a central substitution of an adenine for the single cytosine residue is not uncommon (38) and matches part of the Sp1 site identified in Probe 3. This informed our mutant luciferase reporter construct design, as discussed below.
The supershift analysis shown in Fig. 3 confirmed the specific binding of Sp1 and Sp3 to Probes 1-3. The magnitude of protein binding was in the order Probe 1 Ͼ Probe 3 Ͼ Probe 2, and this ranking may be of significance in the control of HAS2 constitutive transcription. Both of these Sp factors are expressed ubiquitously in mammalian cells, where Sp1 activates the transcription of a number of genes by binding to GC elements of the type present upstream of the HAS2 orthologues in human, murine, and equine genomes (34,46,47,49,50). Sp3 may act to stimulate or repress transcription and can inhibit the Sp1-dependent transcription of promoters containing adjacent recognition sites, the slower migrating Sp3-DNA complexes formed with these multiple sites having greater stability than either monomeric Sp3-DNA complexes or multimeric Sp1-DNA complexes (46, 47, 49 -51). Although the Sp1:Sp3 ratio may influence the activity of these proteins in a cell-specific manner (46,47,49,50), Fig. 3 showed no obvious differences in quiescent HK-2 cells for Probes 1-3. However, changes in the cellular environment and response to external stimuli have the potential to alter the respective levels of these transcription factors and modulate their function (46,47,49,50).
These data are augmented by recent findings describing functional upstream elements in the HAS2 promoter, which may interact with Sp1 and/or Sp3 (32). Using chromatin immunoprecipitation, the STAT element in the proximal promoter, which overlaps the putative NF-B site in Probe 6 ( Fig. 1), was shown to respond to stimulation of keratinocytes with epidermal growth factor by binding STAT3 (32). Cooperative function to facilitate gene expression has been reported for STAT3 with both Sp1 (52) and Sp3 (53), and direct STAT3-Sp1 interaction has also been posited (54). In addition, RAREs located ϳ1.5 and 2.8 kb from the HAS2 TIS were found to bind RA receptor (RAR)␥ following the addition of all-trans RA; retinoic X receptor (RXR), STAT3, and pSTAT3 were also isolated (32). The current paradigm maintains that RARs only activate transcription following dimerization with RXRs (55) and RAR: RXR/Sp1 interactions are well documented (56). However, the modulation of the interaction between Sp1 and the GC-rich DNA motifs to which it binds by RAR via ternary complex formation has been described (57)(58)(59). Indeed, a growing body of evidence that RA and other nuclear receptors are capable of such interaction (60, 61) might provide a further mechanism for the up-regulation of HAS2 transcription by all-trans-RA. Interestingly, up-regulation of RARs, RXR, and Sp1 in the nuclei of articular chondrocytes followed HA 6 oligosaccharide treatment (62).
In contrast to the results for Probes 1-3, Probe 4 did not bind proteins from HK-2 nuclear extracts, suggesting that this site was not transcriptionally active in quiescent HK-2 cells. On the basis of comprehensive analyses of CCAAT box and NF-Y-binding CCAAT boxes, the sense strand NF-Y site was unlikely to be functional due to the presence of multiple thymine residues toward the 3Ј-end of the motif (38 -40). Similarly, the putative CCAAT site in antisense orientation possessed the core sequence CCATT, also making this sequence less likely to activate transcription (38 -40).
EMSA analysis using Probe 5 resulted in only weak protein binding, as shown in Fig. 2A. Software analysis to predict putative TFBSs is a useful starting point for gene regulation analysis but cannot provide a functional understanding (35) and generates large quantities of data (33). It was thus unusual to find that the Probe 5 sequence contained a single putative TFBS, for the promyelocytic leukemia zinc finger motif, which occurred only once more in the first 3 kb of the HAS2 promoter. Promyelocytic leukemia, which may be treated using all-trans RA, is characterized by a fusion of the RAR␣ gene to alternative partners, including promyelocytic leukemia zinc finger, and HA receptor CD44 has been postulated as a potential therapeutic target (63). However, in view of the inconclusive EMSA findings, Probe 5 was not analyzed further. This decision was supported by the results of the luciferase analysis in Fig. 4, which demonstrated that point mutation at key residues in consensus recognition sites for Sp1 and Sp3 in the Probes 1, 2, and 3 sequences resulted in an almost total ablation of the ability of the promoter region of the F3 fragment to drive transcription of the luciferase gene. This provided strong evidence that constitutive transcriptional activity of HAS2 was mediated by Sp1 and Sp3.
To examine this assertion further, siRNA knockdown of Sp1 and Sp3 was carried out in HK-2 cells as shown in Fig. 5, A and B, respectively. The effect of this down-regulation on HAS2 transcription was then examined by qPCR. Fig. 5C shows that knockdown of Sp1 or Sp3 significantly reduced the amount of detectable HAS2 mRNA. These data suggested that both Sp1 and Sp3 are required as co-activators of HAS2 constitutive transcription and that a reduction in the mRNA levels of either was sufficient to reduce HAS2 mRNA synthesis.
In Fig. 6A, the NF-B consensus probe showed pronounced binding with HK-2 nuclear proteins in response to IL-1␤-stimulation, which was most obvious after 30 min, and which was inhibited by SN50. In contrast, HAS2-specific Probe 6, located upstream of the F3 promoter region and which contained a putative NF-B site and an overlapping STAT site, showed low levels of protein binding that were not affected by the presence of SN50 (Fig. 6B). These data were supported by the luciferase analysis of Fig. 7.
Previous work has shown that IL-1␤ increases HAS2 mRNA expression in HK-2 cells in an NF-B-dependent manner (29). Furthermore, work on a variety of cell types has shown HAS2 up-regulation in response to IL-1␤ (e.g. Refs. 16, 62, 64, and 65). Since this response in HK-2 cells was not mediated by the putative NF-B recognition site assayed in the present study, analysis of upstream sequences for active NF-B sites will be of great interest. Despite the fact that none were identified in silico in the first 3 kb upstream of the HAS2 TIS (35), it may be relevant that this analysis was carried out using default software settings. Fig. 6C illustrated that NF-B subunits p50 and p65 were present in the nuclei of IL-1␤-stimulated HK-2 cells. This suggested signaling via the classical NF-B pathway, which mediates innate immunity and via which these subunits interact with a host of immune response genes (66). The role of p50/p65 in the up-regulation of HA synthesis and HAS1 transcription following the addition of tumor necrosis factor-␣ to MRC-5 fibroblasts has been described (67). In contrast, recent findings on human intestinal mesenchymal cells has shown that IL-1␤ induces HAS2 transcription and increased HA synthesis via p38 and extracellular signal-regulated kinase (ERK) mitogen-activated protein (MAP) kinase pathways, and the authors have suggested that multiple signaling pathways combine to this end (65). The recent description of the natural antisense transcript HA synthase 2 antisense (HASNT) (31) reinforces the concept of multifactorial transcriptional regulation at the HAS2 locus.
In summary, we have identified transcription factors Sp1 and Sp3 as principal mediators of constitutive HAS2 transcription in the renal proximal tubular cell line HK-2. Three adjacent recognition sites upstream of HAS2 exon 1 were demonstrated to be functional, with the most active immediately proximal to the TIS. We have also shown that putative NF-Y and CCAAT and NF-B motifs identified within 300 bp of the HAS2 TIS were not functional in these cells. The interaction of the constitutively active Sp factors with the recently identified STAT3 site in the HAS2 proximal promoter and upstream RAREs (32), together with other factors still awaiting discovery, will be the subject of further studies.