Gene expression of the human plasma membrane-associated sialidase (NEU3), a key enzyme for ganglioside degradation, is relatively high in brain and is modulated in response to many cellular processes, including neuronal cell differentiation and tumorigenesis. We demonstrated previously that NEU3 is markedly up-regulated in various human cancers and showed that NEU3 transgenic mice developed a diabetic phenotype and were susceptible to azoxymethane-induced aberrant crypt foci in their colon tissues. These results suggest that appropriate control of NEU3 gene expression is required for homoeostasis of cellular functions. To gain insights into regulation mechanisms, we determined the gene structure and assessed transcription factor involvement. Oligo-capping analysis indicated the existence of alternative promoters for the NEU3 gene. Transcription started from two clusters of multiple TSSs (transcription start sites); one cluster is preferentially utilized in brain and another in other tissues and cells. Luciferase reporter assays showed further that the region neighbouring the two clusters has promoter activity in the human cell lines analysed. The promoter lacks TATA, but contains CCAAT and CAAC, elements, whose deletions led to a decrease in promoter activity. Electrophoretic mobility-shift assays and chromatin immunoprecipitation demonstrated binding of transcription factors Sp (specificity protein) 1 and Sp3 to the promoter region. Down-regulation of the factors by siRNAs (short interfering RNAs) increased transcription from brain-type TSSs and decreased transcription from other TSSs, suggesting a role for Sp1 and Sp3 in selection of the TSSs. These results indicate that NEU3 expression is diversely regulated by Sp1/Sp3 transcription factors binding to alternative promoters, which might account for multiple modulation of gene expression.
Gangliosides, sialic acid-containing glycosphingolipids, are components of cell-surface membranes which exert a wide variety of biological functions, alterations being associated with cell growth, differentiation and tumorigenesis . Furthermore, inherited defects in ganglioside catabolism are known to result in a lysosomal storage disease, gangliosidosis . Thus strict control of biosynthesis and degradation of gangliosides is prerequisite for homoeostasis of cell functions, although the control mechanisms remain largely unclear [3,4]. Recent work on gene regulation of enzymes involved in ganglioside metabolism suggest that transcriptional regulation may participate. Interestingly, some of the genes might undergo concerted regulation at the transcriptional level because of the presence of common promoter structures that lack TATA motifs and contain abundant GC elements and binding sites for Sp (specificity protein)/KLF (Krüppel-like factor) family transcription factors .
Plasma-membrane-associated sialidase Neu3 is one of four mammalian sialidases, and the human orthologue NEU3 shows strict preference for gangliosides as substrates . Our recent data indicated that NEU3 could modulate signal transduction in the vicinity of the plasma membrane  and that aberrant expression of the NEU3 gene could be causative of tumorigenesis  and diabetes mellitus . Up-regulation of gene expression in tumours was first found in human colon cancers and further investigation revealed suppressive effects on cell apoptosis . Consistent with these observations, mice overexpressing human NEU3 showed a high propensity for azoxymethane-induced ACF (aberrant crypt foci) formation in the colon . Down-regulation of NEU3 by siRNA (short interfering RNA)-mediated gene silencing induced apoptosis in cancer cells, but not in normal cells . In addition, NEU3 transgenic mice develop an insulin-resistant diabetic phenotype showing low glucose tolerance and enlarged pancreatic islets by 18–20 weeks .
We have reported levels of NEU3 gene expression in normal tissues and its chromosomal location previously , and other groups have reported genomic organization  and changes in NEU3 mRNA levels during cell differentiation and mitotic activation [15–17]. However, detailed information about the gene structure and control mechanism of the NEU3 gene is not available. In the present study, we identified cis-elements controlling its promoter and obtained evidence of regulation by Sp1 and Sp3 transcription factors that have attracted wide interest for their intimate involvement in growth control and tumorigenesis [18,19]. This work provides an important framework for further dissection of NEU3 gene regulation and a first step to unveiling the full significance of the NEU3 gene in biological processes, including cell differentiation and tumorigenesis.
Reagents and cell lines
Oligonucleotides were synthesized by Sigma–Aldrich or IDT with the sequences listed in Supplementary Table S1 (http://www.BiochemJ.org/bj/430/bj4300107add.htm). Antibodies against Sp1 (07-645), Sp3 (D-20) and Neo (ab33595) were purchased from Upstate Biotechnology, Santa Cruz Biotechnology and Abcam respectively. Human cell lines A172, IMR32, PC-3 and DLD-1 were obtained from the Human Science Research Resources Bank (Tokyo, Japan). HT-29, HCT116, and LNCap cells were from the A.T.C.C. (Manassas, VA, U.S.A.). HeLa, A549, MCF7 and NB-1 cells were from the Cell Resource Center for Biomedical Research (Tohoku University, Sendai, Japan). HEK (human embryonic kidney)-293 cells were from the RIKEN BioResource Center (Tsukuba, Japan). Cells were cultured in Dulbecco's modified Eagle's medium (Sigma–Aldrich) supplemented with 10% fetal bovine serum (Invitrogen) at 37 °C in a 5% CO2 atmosphere. NHEKs (normal human epidermal keratinocytes) were purchased from Kurabo (Osaka, Japan) and maintained according to the supplier's instructions.
Isolation of genomic clones
Genomic DNA fragments containing the human NEU3 gene were obtained by a combination of library screening and PCR. A cosmid library from human placental DNA (Clontech) was screened with the 32P-labelled EcoRI fragment of NEU3 cDNA . Eleven cosmid clones were isolated with the first screening, and the inserts were subcloned into pBluescript II SK+ (Stratagene), followed by DNA sequencing. Among them, eight clones had the same insert and were revealed to encompass exons 3–4 of the NEU3 gene. A DNA fragment encompassing exons 2–3 was obtained by PCR using primers GSP-01, a sense primer derived from the 5′-portion of NEU3 cDNA (K. Yamaguchi, T. Wada, Y. Shimada and T. Miyagi, unpublished work), and GSP-02, an antisense primer from nucleotides 377 to 348 of the cDNA . The PCR product (6 kb) obtained was subcloned into the SmaI site of pBluescript II SK+. To clone a DNA fragment containing exon 1 and the 5′-flanking region, the 5′-region of the PCR fragment subcloned was digested with BamHI and HindIII and used as a probe for screening of a human genomic λ phage library (Human Genome Center, Tokyo, Japan). Nine clones were obtained by first screening, and their inserts were excised by digestion with NotI followed by subcloning into the NotI site of pBluescript II SK+. Five clones were found to have the same insert of 11 kb and to contain a 5′-flanking region and exon 1 of the NEU3 gene. One of the clones, pN07, was used for construction of the reporter plasmids described below.
3′-RACE (rapid amplification of cDNA ends)
The 3′-UTR (untranslated region) of the NEU3 cDNA was obtained by the 3′-RACE method as described previously . Briefly, to prepare cDNA for RACE, 0.7 μg of poly(A)+ RNA from DLD-1 cells was reverse-transcribed with Superscript II (Invitrogen) using a (dT)17 adaptor primer. The cDNA generated was used for first PCR with the adaptor primer and a gene-specific primer, GSP-03. Nested PCR was performed with the adaptor primer and a nested primer, GSP-04. PCRs were carried out with LA-Taq polymerase (TaKaRa Bio) under the following conditions: initial denaturation at 95 °C for 5 min followed by 40 cycles of denaturation at 95 °C for 30 s, annealing at 55 °C for 30 s and elongation at 72 °C for 2.5 min. PCR products were blunted with T4 DNA polymerase, phosphorylated by T4 polynucleotide kinase (both TaKaRa Bio) and then subcloned into the SmaI site of pBluescript II SK+. The cloned fragments were sequenced with a BigDye Terminator Cycle Sequencing kit and a 3130 Gene Analyzer (Applied Biosystems).
Total RNAs of cultured cells were prepared with an RNeasy mini kit (Qiagen). Poly(A)+ RNAs from human liver, brain and colon were purchased from Clontech. Dephosphorylation, TAP (tobacco acid pyrophosphatase) treatment, and RNA-linker ligation of RNA were carried out by using a RACE kit (Ambion) following the manufacturer's recommendations. Reverse transcription of processed RNA and subsequent PCR were achieved using a PrimeScript cDNA synthesis kit (TaKaRa Bio) and LA-Taq polymerase respectively. Primers used for PCR were as follows; for first PCR, outer-primer (provided in the RACE kit) and OCP-01; for nested PCR, the inner-primer (provided in the kit) and OCP-02. For oligo-capping analysis of luciferase reporter constructs, primers OCP-luc1 and OCP-luc2 were used instead of OCP-01 and OCP-02 respectively. The amplified DNA fragments were gel-purified, digested with BamHI and NotI, and ligated to BamHI- and NotI-digested pBluescript II KS+ followed by transformation with Escherichia coli (XL-1 Blue) competent cells. Plasmid DNAs were prepared from randomly selected clones (approx. 50 clones per oligo-capping reaction) and used as templates for DNA sequencing.
Primer extension analysis was performed as described previously . Briefly, a 32P-labelled primer, PEP-01, was hybridized with 5 μg of poly(A)+ RNA prepared from HeLa cells in 80% formamide, 40 mM Pipes (pH 6.3), 1 mM EDTA and 400 mM NaCl at 50 °C for 12 h. After ethanol precipitation, the reaction mixture was incubated with 200 units of Superscript II (Invitrogen) under the conditions recommended by the manufacturer. Extended products were resolved on an 8% polyacrylamide, 8.3 M urea-denaturing gel and analysed with the FLA-3000 imaging system (FujiFilm).
Construction of reporter plasmids
Reporter plasmids for promoter activity assays were prepared as follows. After digestion of subclone pN07 with NcoI plus BamHI (6 kb), XbaI (2 kb) or EcoRI (1 kb), generated DNA fragments containing the 5′-flanking region of the NEU3 gene were blunted with T4 DNA polymerase (TaKaRa Bio) and inserted into the SmaI site of the promoter- and enhancer-less luciferase reporter vector PGV-B (Wako Pure Chemicals). Orientation of inserted fragments was confirmed by restriction digestion. To introduce deletion mutation in the minimal promoter region, the blunted NcoI/EcoRI fragment was subcloned into the HincII site of pBluescript II SK+ and subjected to PCR-based mutagenesis . Primers for each construct that is illustrated in Figure 4 were as follows; for ΔA, MP01 and MP02; for ΔB, MP10 and MP03; for ΔC, MP11 and MP04; for ΔD, MP12 and MP05; for ΔE, MP13 and MP06; for ΔF, MP14 and MP07; for ΔG, MP15 and MP08; and for ΔH, MP16 and MP09. Deletion (Figure 3) was performed with the primer sets of MP17 and MP18 for the CCAAT motif, and with MP19 and MP20 for CAAC motifs respectively. Mutation of Sp1-binding sites (Figure 7) was introduced sequentially by using primers MP21 and MP22 and then MP23 and MP24. PCR was achieved by using KOD polymerase (Toyobo) following the manufacturer's recommendations. After confirmation of DNA sequences, mutated fragments were excised by digestion with XhoI and SmaI, and inserted into PGV-B digested with XhoI and SmaI. All reporter plasmids were prepared for transfection using a Qiagen plasmid kit.
HCT116 cells and DLD-1 cells (0.8×105) were seeded into wells of 12-well plates 16 h before transfection. Transfection was carried out using Effectene (Qiagen) according to the manufacturer's instructions. For each well, 0.5 μg of reporter plasmid was transfected with 25 ng of pRL-TK vector (Promega) used for normalization. After 48 h, cells were harvested and assayed for luciferase activity, using the Dual-Luciferase Reporter Assay System (Promega). All experiments were performed independently at least three times.
EMSA (electrophoretic mobility-shift assay)
Nuclear extracts from the cells were prepared as described by Dignam et al. . For preparation of probes, oligonucleotides were end-labelled with polynucleotide kinase (Toyobo) and [γ-32P]ATP (PerkinElmer). Labelled oligonucleotides were purified with a NICK™ column (GE Healthcare) and annealed. Nucleotide sequences of probes and competitors are described in Supplementary Table S1 (only the upper strands are shown). Nuclear extract (1 μg) was pre-incubated on ice for 15 min in binding buffer [10 mM Hepes (pH 7.8), 50 mM KCl, 1 mM EDTA, 5 mM MgCl2, 10% glycerol, 5 mM dithiothreitol, 0.5 mM PMSF, 100 μg/ml BSA, 25 μg/ml poly(dI-dC)·(dI-dC) (GE Healthcare), 2 μg/ml aprotinin, 2 μg/ml leupeptin and 2 μg/ml pepstatin] with or without non-radioactive oligonucleotides (20-molar excess) or antibodies (2 μg) before the addition of labelled probe and incubation for 20 min at room temperature (25 °C). Samples were resolved in non-denaturing 0.5× TGE (Tris/glycine/EDTA)/4% polyacrylamide gels and processed with the FLA3000 imaging system.
ChIP (chromatin immunoprecipitation) assay
ChIP assays were carried out using an assay kit from Upstate Biotechnology. Cells were fixed by the addition of 1 ml of fixation buffer (50 mM Hepes, pH 8.0, 11.1% formaldehyde, 100 mM NaCl, 1 mM EDTA and 0.5 mM EGTA) to 10 ml of culture medium and incubation for 10 min at room temperature on a rocking platform. Fixation was stopped by the addition of glycine (0.125 M final concentration) followed by incubation for 10 min at room temperature. Fixed cells (5×106) were washed twice with ice-cold PBS, resuspended in 1 ml of the sonication buffer provided in the kit, and sonicated with Bioruptor (Cosmo Bio) to shear DNA into approx. 500-bp lengths. The lysate was clarified by brief centrifugation at 13400 g for 5min and the supernatant was diluted 10-fold with the dilution buffer provided in the kit. For one assay, 2 ml of diluted lysate was pre-cleared and subjected to immunoprecipitation, and DNA in the washed immunocomplex was recovered according to the manufacturer's instructions. PCR was carried out with LA-Taq using primers CAP-01 and CAP-02 (Supplementary Table S1).
siRNA-mediated gene silencing
siRNAs against Sp1 (used as a pool of #116546, #116547 and #143158) and Sp3 (a pool of #115336, #115337 and #115338) were purchased from Ambion. siRNAs against PURα (UAAACACGCCGUACUUGUUGGAGCC) and PURβ (UUGAAGGGUACGGUGAUGGCAUUGC) were purchased from Invitrogen. Transfection of siRNA was accomplished with Lipofectamine™ 2000 (Invitrogen) following the manufacturer's instructions. Total RNA was prepared from cells 24 h after transfection with an RNeasy mini kit and reverse transcribed with the aid of a PrimeScript kit using 1 μg of total RNA. Real-time PCR was performed on a LightCycler (Roche) with SYBR Green master mix (Qiagen). Primers used for PCR were as follows: evaluation for transcripts from TSS0, primers SSP-01 and SSP-03; for those from TSS1, SSP-02 and SSP-03; for 18S rRNA as a reference, SSP-06 and SSP-07.
Organization of the human
To explore the structure of the human NEU3 gene, genomic DNA clones were isolated from a human genomic library by screening with a NEU3 cDNA fragment as a probe. After DNA sequencing of the obtained genomic clones and comparing with a seuence of the cDNA cloned from a human brain cDNA library , the NEU3 gene was found to span approx. 22 kb and consisted of four exons. The third and the fourth exons encode the open reading frame for NEU3 (Figure 1A). All of the exon/intron boundary sequences adhere to the GT-AG rule for eukaryotic genes , as summarized in Table 1. The structure of the 3′-region of the NEU3 gene was determined by 3′-RACE with poly(A)+ RNA of DLD-1 cells and the primers listed in Supplementary Table S1. After two rounds of PCR, amplified products of 0.5 and 5 kb were obtained (results not shown). The sequence of the 0.5-kb fragment was identical with the 3′-UTR of the reported cDNA . Sequencing of the 5-kb fragment identified other distal polyadenylation signals, allowing transcripts with an elongated 3′-UTR. These results are consistent with our previous finding that transcripts of 2.5 and 7 kb can be detected by Northern blot analysis in various human tissues . The reported consensus motif (ATTTA) for destabilization of mRNA  was found to be present in the 3′-UTR, although its physiological significance for NEU3 gene expression has yet to be determined.
Structure of the human
|Exon .||.||Intron .|
|Number .||Size (bp) .||Splice donor .||Size (bp) .||Splice acceptor .|
|Exon .||.||Intron .|
|Number .||Size (bp) .||Splice donor .||Size (bp) .||Splice acceptor .|
Multiple TSSs (transcription start sites) of the
TSSs were determined using the oligo-capping method , utilizing the cap structure of mRNAs to determine the precise 5′-terminal nucleotides of mRNAs. The results using RNAs from human brain and HCT116 cells are shown in Figure 1(B). In the case of brain, transcription appeared to start mainly from multiple start sites distributed approx. 50 bp upstream of the 5′-end of the published cDNA sequence (indicated by arrows above the sequence in Figure 1B). On the other hand, in HCT116 cells, transcription appeared to start mainly from multiple sites in the first intron (indicated by arrows below the sequence in Figure 1B). A search of the ESTs (expressed sequence tags) of NEU3 deposited in dbEST revealed multiple TSSs in the region identified in the present study (see Supplementary Figure S1 at http://www.BiochemJ.org/bj/430/bj4300107add.htm), supporting the results obtained by the oligo-capping method. Several human tissues and cultured cell lines showed similar results as HCT116 cells (see Supplementary Figure S2 at http://www.BiochemJ.org/bj/430/bj4300107add.htm). To confirm the existence of the upstream minor TSSs, primer extension analysis was conducted by using poly(A)+ RNA of HeLa cells as a template and an upstream primer (Figure 1B), resulting in assignment of TSSs in a similar upstream region (indicated by arrowheads in Figure 1B). These results suggest that the NEU3 gene possesses alternative promoters [27,28] which are employed in a tissue-specific manner. For the present study, the TSSs used frequently in most cells and tissues are designated as TSS1 and those used mainly in brain are designated as TSS0. In addition, the most frequently used TSS in the TSS1 is numbered +1 (Figure 1B, ©).
Core promoter elements of the
The promoter region of the NEU3 gene was determined by luciferase reporter assay. To accomplish this, different lengths of the 5′-upstream region were prepared by digestion with the restriction enzymes shown in Figure 2, and inserted into a luciferase reporter vector. These reporter vectors were tested for their ability to promote transcription of the luciferase gene with transient transfection followed by a luciferase assay. All fragments prepared showed significant promoter activity in HCT116 cells (Figure 2), except when the orientation of cloned fragments was opposite with respect to the luciferase gene (results not shown). The reporter vector harbouring the EcoRI/NcoI fragment showed promoter activity comparable with that of the 5-kb XbaI/NcoI fragment, suggesting its minimal promoter activity. Although the EcoRI/NcoI and the ApaI/NcoI fragments showed similar promoter activities, we analysed further the former for the core promoter element of the NEU3 gene since it contains the CCAAT element that plays a critical role in transcriptional control of the gene as presented in Figure 3. At present, we do not know the exact reason that deletion of the EcoRI/ApaI fragment appeared to affect scarcely the overall promoter activity. However, it was suggested from the results in Figure 4(C) that repressive element might exist in the deleted region. Deletion of the ApaI/NcoI fragment yielded decreased promoter activity, presumably because of deficiency in the TSS1 region. Utilization of the TSS1 in luciferase reporter vectors was confirmed using the oligo-capping method (Figure 4B, wild-type). We also tested promoter activity of these constructs in other human cell lines including DLD-1, HT29 (colon cancer cell), HeLa (cervical carcinoma cell), IMR32 (neuroblastoma), A172 (glioblastoma) and HEK-293 (kidney) cells and obtained similar results, indicating that the EcoRI/NcoI fragment contains a minimal promoter region (results not shown). In the sequence of this EcoRI/NcoI fragment, there are two core promoter elements. One is a CCAAT motif 80 bp upstream of the TSS0 (double-underlined in Figure 1B) and another is a cluster of CAAC elements near the TSS1 (underlined in Figure 1B). The CCAAT motif is one of the well-characterized promoter elements recognized by several transcription factors including C/EBP (CCAAT/enhancer-binding protein)  and NF-Y (nuclear factor Y) [30,31]. The CAAC motif is also found in several genes [32–34] and is thought to function as a core promoter element, although transcription factor(s) for this motif have so far not been identified. Deletions in each of these two elements led to decreased promoter activity (Figure 3A), suggesting that both act as positive regulators of NEU3 gene expression. Since the NEU3 gene has alternative promoters, the question arose as to which promoter is affected by these elements. To address this, amounts of mRNA transcribed from TSS0 or TSS1 were measured by RT (reverse transcription)–PCR. After transfection of luciferase reporter vectors containing the deletion, total RNA was prepared from the transfected cells and reverse-transcribed followed by quantitative PCR using two different primer sets to distinguish utilization of each TSS as described in the legend of Figure 3. As shown in Figure 3(B), deletion of the CCAAT motif decreased transcription from both promoters, implying roles in positive regulation of TSS0 and TSS1 in common. In contrast, deletion of CAAC elements decreased transcription from TSS1, but scarcely affected that from TSS0, suggesting that the elements mainly promote transcription from TSS1. When both elements were deleted simultaneously, promoter activity decreased only to a level similar to that which resulted after each of the single deletion mutations (Figure 3A). This result implies that another undefined promoter element(s) drives the transcription of the NEU3 gene. There also exist other core promoter elements, an XCPE (X core promoter element) motif  110-bp upstream of TSS1 and a TATA-like motif 120 bp upstream of TSS0, but deletions of these regions did not affect the promoter activity as assessed using the luciferase assay (results not shown). The sequence of the EcoRI/NcoI fragment shows a high G+C content and high frequency of CG dinucleotides, satisfying the criteria of an CpG island [36,37], thought to be a target of gene regulation through methylation/demethylation on cytidines of CG dinucleotides. However, significant methylation on cytidines within the EcoRI/NcoI fragment has not been found in human tissues or cells so far examined (K. Koseki, K. Yamaguchi and T. Miyagi, unpublished work).
Transactivation activity of the 5′-flanking region of the human
Effects of deletion of core promoter elements on promoter activity
Promoter activity and TSS usage of serial deletion constructs
Determination of the promoter region for the
NEU3 gene expression
To characterize the promoter region more precisely, sequential deletion mutations were introduced into the EcoRI/NcoI fragment and tested for their effects on promoter activity. Each deleted region is approx. 100 bp in length except 50 bp for G region, which encompasses TSS1. As shown in Figure 4(A), deletion of the F or H region resulted in decreased promoter activity. The F region contains CAAC elements whose deletion led to decreased promoter activity (Figure 3A). In the H region, we did not find any known core promoter elements. Use of a luciferase reporter vector revealed that these two regions, F and H, had different effects on the selection of TSSs. As shown in Figure 4(B), the wild-type reporter vector preferentially utilized TSS1 and showed a similar distribution pattern of TSSs to that of the endogenous NEU3 gene in HCT116 cells (Supplementary Figure S2). The H-deleted reporter vector also utilized TSS1, although the distribution pattern was slightly different from those of the endogenous NEU3 gene and the wild-type reporter vector. Deletion of the F region resulted in a shift from TSS1 to TSS0, the latter not normally being utilized in HCT116 cells, suggesting suppression of transcription from TSS0 by the F region. These data suggest that the F region is likely to act as an activator of transcription from TSS1 through its CAAC motifs and, in contrast, as a suppressor for transcription from TSS0. Although deletion of the CCAAT motif led to a decrease in promoter activity (Figure 3A), deletion of the D region containing the motif appeared to have no effect on promoter activity (Figure 4A). This discrepancy cannot be explained at present, but deletion of the D region resulted in increased transcription from TSS1 (Figure 4C) and decreased transcription from TSS0. Oligo-capping analysis also indicated the preferential usage of TSS1 in the D-deleted reporter vector (Figure 4D). A repressive element(s) not yet identified may exist in the D region as well as the positive regulator CCAAT motif, so deletion of the D region might not cause a significant change in gross promoter activity assessed by the luciferase reporter assay. Although oligo-capping analysis shown in Figure 4(D) appeared to indicate no transcription initiation from TSS0 of D-deleted reporter vector, we could not exclude the possibility of a low level of TSS0 utilization because the RT–PCR experiment in Figure 4(C) suggested low but substantial initiation from TSS0. Therefore RT–PCR is considered to be more sensitive and suitable for measurement of a low level of transcripts than oligo-capping that counted approx. 50 tags for each experiment. Our previous results indicated that human NEU3 and mouse Neu3 genes show similar expression patterns in normal tissues [13,38]. The Comparative Viewer of the DBTSS revealed that the core promoter elements mentioned above also occurs in a predicted promoter region of mouse Neu3 gene, suggesting that the human and mouse genes transcription might be regulated in a similar manner (results not shown).
trans-factors affecting NEU3 gene expression
A search of databases (TFSEARCH, MatInspector) revealed putative binding motifs for several transcription factors in F and H regions as shown in Figure 5(A) and Supplementary Figure S3 (at http://www.BiochemJ.org/bj/430/bj4300107add.htm). To determine which transcription factors are involved in modulation of NEU3 gene expression, an EMSA was conducted using sets of labelled oligonucleotide probes covering the F or H regions and unlabelled competitors containing respective consensus binding motifs for transcription factors. As shown in Figure 5(B), probes F-1, -2 and -4 gave similar shifted bands (lanes 2, 13 and 23). They were specifically competed by oligonucleotides containing a GC box (lanes 6, 16 and 26) or a GT box (lanes 7, 17 and 27) which are consensus binding motifs of Sp/KLF family transcription factors [39,40]. In addition, mutations introduced in recognition motifs of the Sp/KLF transcription factors abolished competing activity (results not shown). These results suggest specific binding of the Sp/KLF transcription factor(s) to the F region of the human NEU3 gene. Interestingly, oligonucleotides F-1, -2 and -4 were able to compete with each other (lanes 9, 11, 19, 21, 29 and 30), suggesting that these probes might be recognized and bound by the same member(s) of the Sp/KLF family. As shown in Figure 5(A), putative binding sites for transcription factors other than Sp/KLF transcription factors were predicted in the F region, but unlabelled competitors containing each consensus binding motifs for these transcription factors (listed in Supplementary Table S1) failed to prevent the DNA–protein complex formation (Figure 5B, lanes 4, 5, 8, 15, 18, 25 and 28). Probe F-3 containing CAAC motifs did not give any shifted band in our system (results not shown). The Sp/KLF transcription factor family comprises 25 members that play common or distinctive physiological roles in transcription . To determine which member(s) of the Sp/KLF family binds to the F region, supershift assays were conducted using antibodies against Sp1 and Sp3, both of which are expressed in cells and tissues ubiquitously. As illustrated in Figure 5(C), anti-Sp1 and anti-Sp3 antibodies effectively supershifted the three complexes, indicating that the two factors are capable of recognizing and binding to the F region. Next, binding of Sp1 and Sp3 to the F region in vivo was confirmed by ChIP assays. Anti-Sp1 and anti-Sp3 antibodies specifically immunoprecipitated chromatin containing the F region, whereas species-matched control antibodies did not (Figure 5D). Involvement of other members of the Sp/KLF family, Sp4, KLF4, KLF5 and KLF6 were examined by siRNA-mediated gene silencing of these factors, but no effects on NEU3 gene expression were observed (results not shown).
Interaction of Sp1 and Sp3 with the F region of the human
NEU3 gene promoter in vitro and in vivo
In the same way, transcriptional factor(s) binding to the H region was investigated by EMSA (Supplementary Figure S3). Probes covering this region gave a similar pattern of shifted bands and competition was seen among the probes, suggesting that the same transcription factor(s) binds to this region. Although the bands were competed completely by consensus motifs for the GA-binding factors, PURα and PURβ, further experiments using siRNA-mediated knockdown of these factors failed to show involvement in modulation of NEU3 gene expression (results not shown).
Opposite effects of down-regulation of Sp1 and Sp3 on TSS0 and TSS1
Involvement of Sp1 and Sp3 in modulation of NEU3 gene expression was examined further by siRNA-mediated gene silencing. Pooled siRNAs against Sp1 or Sp3 were transiently transfected into HCT116 cells, which resulted in a more than 50% decrease in the levels of target gene transcripts (Figure 6A). In these treated cells, transcription from TSS0 (Figure 6B) increased, and this increase was enhanced by simultaneous knockdown of both factors. On the other hand, the knockdown decreased transcription from TSS1. Similar results were obtained from the knockdown experiments using DLD-1 (Figure 6C) and HeLa (Figure 6D) cells. To examine the involvement of Sp1/Sp3-binding motifs on the F region in the regulation of transcription from TSS0 or from TSS1, luciferase reporter vectors containing a mutation in the Sp1/Sp3-binding sites were tested for their transcription initiation sites. As shown in Figure 7, transcription from TSS1 decreased in the mutants, whereas that from TSS0 increased. These results, together with the result of the ChIP assay (Figure 6D), strongly suggest that Sp1 and Sp3 bound to the F region promote transcription from TSS1 and repress that from TSS0. The involvement of Sp1 and Sp3 in NEU3 gene expression was supported further by comparative evaluation of expression levels of NEU3 and these factors in human cell lines. As shown in Supplementary Figure S4 and Supplementary Table S2 (http://www.BiochemJ.org/bj/430/bj4300107add.htm), expression levels of NEU3 and that of Sp1 or Sp3 showed good correlations (P=0.0087, r=0.72, and P=0.0005, r=0.85, respectively), implying that Sp1 and Sp3 play a promoting role in NEU3 gene transcription. All of the results are illustrated schematically in Figure 8, indicating diverse regulation of NEU3 gene expression by Sp1/Sp3 transcription factors binding to the promoter.
Effects of Sp1- and Sp3-knockdown on human
NEU3 gene expression
Effects of Sp/KLF-binding sites on transcription initiation
Regulation of the human
Human plasma membrane-associated sialidase, NEU3, is a key enzyme in the degradation of gangliosides, for which it exhibits an especial substrate preference. It has been shown to control transmembrane signalling for many cellular processes, and we demonstrated previously marked up-regulation in various cancers, including colon, renal, ovarian and prostate lesions . Our further observations on NEU3 revealed that the sialidase activates molecules including EGFR (epidermal growth factor receptor), FAK (focal adhesion kinase), ILK (integrin-linked kinase), Shc, integrin β4 and Met , which often become up-regulated during tumorigenesis, and may thus contribute to accelerated development of malignant phenotypes in cancer cells. NEU3 involvement in tumorigenesis is also suggested by the high incidence of azoxymethane-induced ACF in colon mucosa of NEU3 transgenic mice . Moreover, NEU3 overexpression causes impaired glucose tolerance and hyperinsulinaemia together with overproduction of insulin in enlarged islets in the transgenic animals . Although the mechanisms underlying the different pathogenesis caused by NEU3 up-regulation in mice remain obscure, it is feasible that NEU3 brings about a signalling disturbance in cell apoptosis and in insulin responses, probably through cross-talk between signalling pathways. Supporting this idea, epidemiological reports [42,43] describing higher incidences of cancers in diabetic patients than in controls have suggested that these diseases might be closely related to each other in pathogenesis.
In the present study, we obtained strong evidence showing regulation of NEU3 gene expression by Sp1 and Sp3 transcription factors which were initially considered as constitutive activators of housekeeping genes and other TATA-less genes, but have been shown to play critical roles in regulating the transcription of genes involved in cell growth control and tumorigenesis [18,19,44]. They may even affect genes in response to extracellular signals such as insulin . Involvement of Sp1 and Sp3 is consistent with results of previous studies analysing NEU3 gene expression in tissues and cells. In normal human tissues, the NEU3 gene is expressed ubiquitously, but expression levels vary considerably [13,46], apparently influenced by cell growth  and differentiation [16,17,47]. In particular, up-regulation of NEU3 may be a cause for resistance to apoptosis and for abnormal growth of cancer cells and Sp1-mediated activation might be attributable for aberrant expression. In fact, Sp1 target genes are known to include cyclin D1, c-Myc, c-Jun, c-Fos, c-Src and receptor tyrosine kinases such as EGFR and PDGFR (platelet-derived growth factor receptor)  and up-regulation of Sp1 has been reported in human cancer tissues [48–50]. We propose that NEU3 is also an Sp1 target gene, in common with these genes essentially involved in cell growth and oncogenesis.
The present study also demonstrated the existence of alternative promoters in the NEU3 gene and repression of brain-type TSSs (TSS0) in other tissues and cells. It is likely that these alternative promoters might confer differences in regulation of NEU3 transcription in brain and other tissues. The exact physiological significance of alternative promoters has yet to be addressed, but is thought to allow complex transcriptional regulation. Besides, it may confer different 5′-UTRs on transcripts, which may permit different post-transcriptional control. Consistent with these possibilities, comprehensive analysis on TSSs of human genes has shown occurrence of alternative promoters preferentially in genes encoding signalling molecules or molecules specifically expressed in brain or testis, as they are regulated in response to various cell conditions or in a tissue-specific manner . Although we do not know at present why brain-type TSSs of the NEU3 gene are repressed in other tissues and cells, several genes expressed in nerves are repressed in other cells [51,52]. Disturbance of this repression results in abnormal development  or in diseases, including colon cancer [54,55], suggesting that some genes working in nerve cells should be repressed in other cells to prevent their aberrant expression causing developmental or growth abnormalities.
Our results indicate that Sp1 and Sp3 act as repressors for TSS0, whereas they act as activators for TSS1. Although it is totally unclear what makes this switch possible, several members of the Sp/KLF family have been shown to function as either activators or repressors through interaction with co-activators, co-repressors and other transcription factors. For instance, Sp1 activates genes through interaction with co-activators CRSP (cofactor required for Sp1), p300/CBP [CREB (cAMP-response-element-binding protein)-binding protein], or a component of basal transcription machinery, TAFII130 (TATA-box-associated factor II 130). Sp1 can also repress gene expression by recruiting HDACs (histone deacetylases) to the promoters [56–58]. How Sp/KLF transcription factors select their partners appears dependent on the cellular and promoter context . Detailed switching mechanisms of Sp1 and Sp3 in the case of the NEU3 gene are now under investigation.
Previously, we and other groups reported that overexpression of the NEU3 gene in cells led to change in other genes for gangliosides metabolism [12,59], implying concerted regulation at the transcriptional level. There have also been several reports of Sp/KLF-binding sites in genes encoding enzymes involved in the metabolism of gangliosides . Although few reports have proved the actual binding of factors to the genes [60,61], the concerted regulation of enzymes  by member(s) of the Sp/KLF transcription family is highly conceivable. Our previous results indicated that the four mammalian sialidase members exhibit different tissue distributions or varied activation during cell growth and differentiation . Despite such differences in sialidase gene expression, the genes might commonly be under the control of Sp/KLF family transcription factors. Our preliminary experiments showed that knockdown of Sp1 led to an increase in NEU1 gene expression, whereas knockdown of Sp3 resulted in a decrease in NEU2 gene expression (K. Koseki, K. Yamaguchi and T. Miyagi, unpublished work). Further investigations are now needed to test the hypothesis of a transcriptional network that regulates sialidases and other genes involved in ganglioside metabolism in relation to cell differentiation, cell growth and tumorigenesis.
aberrant crypt foci
epidermal growth factor receptor
electrophoretic mobility-shift assay
human embryonic kidney
normal human epidermal keratinocyte
rapid amplification of cDNA ends
short interfering RNA
transcription start site
Kazunori Yamaguchi contributed the most to this study by designing and performing the experimental work. Koichi Koseki, Momo Shiozaki and Yukiko Shimada participated in investigation of organization and structure of the gene. Tadashi Wada, who was the first author of the cDNA cloning work, contributed to determination of the promoter region of the gene. Taeko Miyagi directed and supervised the whole work.
We are grateful for Dr M. Tone and Dr Y. Tone (University of Pennsylvania, Philadelphia, PA, U.S.A.) for their helpful discussions. We also thank Dr S. Hashimoto (Tokyo University, Tokyo, Japan) for his kind advice on oligo-capping analysis.
This work has been supported in part by Core Research for Evolutional Science and Technology (CREST) of the Japan Science and Technology Agency and Grants-in Aid for Scientific Research on Priority Areas in Cancer from the Ministry of Education, Culture, Sports, Science and Technology of Japan.