INTRODUCTION

Polysialic acid (PSA)¹ is a linear homopolymer of 2,8-sialic acid residues mainly associated with the neural cell adhesion molecule (N-CAM) in mammalian cells, and is implicated in the reduction of N-CAM adhesion through its large negative charge (1). The expression of PSA on N-CAM is developmentally regulated in the central and peripheral nervous systems (2, 3, 4, 5), i.e. PSA on N-CAM is more abundant in the fetal than in the adult stage. Recent data imply important functions of PSA in the pathfinding and targeting on innervation of axons, migration of neuronal cells and tumor cells, and spatial learning and memory (6, 7, 8). PSA on N-CAM was shown to be synthesized through the action of specific sialyltransferase(s) (9, 10), but the mechanisms underlying the regulation of PSA synthesis and PSA expression remain poorly understood.

Recently, two cDNAs encoding 2,8-sialyltransferases named ST8Sia II/STX (11, 12, 13) and ST8Sia IV/PST-1 (14, 15, 16) were cloned. Mouse ST8Sia II (mST8Sia II) exhibits overall amino acid sequence identity of 56% with mouse ST8Sia IV (mST8Sia IV). Both sialyltransferases can synthesize PSA on 2,3-linked sialic acids on N-CAM without any initiators (16, 17, 18, 19, 20). Northern blot analysis indicated that expression of the mST8Sia II gene was restricted to the brain and testis, whereas the mST8Sia IV gene was expressed strongly in lung and heart rather than brain (13, 16). The expression of the mST8Sia II gene in the brain was strictly regulated during developmental stages (13), i.e. an mST8Sia II transcript could be detected in brain on embryonic day 14, which peaked on embryonic day 20 and then decreased progressively to an almost undetectable level by 2 weeks after birth. The expression of the mST8Sia IV gene was also higher in fetal than adult brain but was less regulated during brain development compared with that of the mST8Sia II gene (16). To elucidate the mechanisms underlying of their tissue- and development-specific expression, it is important to know the structures and activities of their promoters.

We showed recently that expression of the mST8Sia II gene increased in parallel with the increased activity of PSA synthase and the quantity of PSA N-CAM during neuronal differentiation of mouse teratocarcinoma P19 cells, which provides us with an in vitro model system for studying its promoter activity in the neural differentiation processes (21). In this paper, we describe the genomic structure and promoter sequence of the mST8Sia II gene and determination of the 5-flanking region responsible for the cell type-specific promoter activity by means of transfection experiments using differentiated P19 cells.

EXPERIMENTAL PROCEDURES

A mouse genomic cosmid library was constructed and screened as described previously (22). The locations of the exons of the mST8Sia II gene were determined by PCR (GeneAmp XL PCR kit; Perkin-Elmer) with specific oligonucleotide primers or by hybridizing radiolabeled sialyltransferase cDNA to the same blots.

The 3-untranslated region of the mST8Sia II cDNA was isolated from a 3-day-old mouse brain cDNA library (23) by PCR using primers O1-3A (5-GGTATACCCCTAGAACTATATAGCCCTGC-3; nucleotides +2999 to +3024) and O1-3C (5-CTCAGGGTCACCTCCAGAACCTGAC-3; complementary to nucleotides +3420 to +3396).

PCR Amplification of the 5 cDNA End (RACE)

mRNAs from newborn mouse brain were extracted by the guanidinium isothiocyanate method and purified with Oligotex-dT30 (Takara-Shuzo, Japan). Amplification of the 5 end of mST8Sia II cDNA was performed essentially according to the procedure of Frothman et al. (24). cDNA was synthesized by reverse transcription (Superscript II; Life Technologies, Inc.) of 5 µg of mouse brain poly(A) RNA using a primer O1-N6, 5-TTAGATTTGCTATGTAAGCTGTT-3, which is complementary to nucleotides +327 to +303 of the mST8Sia II gene. The excess primers and deoxynucleotides were removed by passage of the cDNA through a MicroSpin S-400 column (Pharmacia Biotech Inc.). The cDNA was A-tailed with 0.6 unit of terminal deoxynucleotidyltransferase (Boehringer Mannheim), using 0.05 mM dATP. Two consecutive PCRs were performed with two nested sets of primers. For pair 1, the forward primer was NotI-(dT)₁₈ (Pharmacia), and the reverse primer was O1-N6. For pair 2, the forward primer was as above but without the T-tail, 5- AACTGGAAGAATTCGCGGCCGCAGGAA-3, and the reverse primer was O1-EX3 (5-CTTCGATCGCTGAGATGTCTGCGAAGATGAGG-3; complementary to nucleotides +255 to +224 of the mST8Sia II gene). The cDNA was amplified for 35 cycles of a step program (94 °C, 40 s; 55 °C, 40 s; 72 °C, 60 s). The amplification products were subcloned into pUC119 and then sequenced.

The reverse primer O1-EX3, which is complementary to a portion of the first exon of the mST8Sia II gene, was end labeled with [-³²P]ATP using T4 polynucleotide kinase. The ³²P-labeled O1-EX3 was hybridized with 5 µg of poly(A) RNA prepared from mouse brain or 5 µg of yeast tRNA in a 10-µl hybridization solution (250 mM KCl, 10 mM Tris-HCl, pH 8.3, and 1 mM EDTA) for 1.5 h at 37 °C after denaturation of the mixture for 1 h at 60 °C. Following hybridization, 60 µl of reverse transcriptase buffer and 200 units of Superscript II (Life Technologies, Inc.) were added (final buffer concentrations: 75 mM KCl, 20 mM Tris-HCl, pH 8.3, 10 mM dithiothreitol, 10 mM MgCl₂, and 0.25 mM dNTPs), followed by incubation for 1 h at 45 °C. The protected products were separated on a 6% sequencing gel along with a dideoxy chain termination sequencing reaction of pO1-22E4.8, using O1-EX3 as the primer.

pO1-22E4.8 was constructed by subcloning a 4.8-kb EcoRI fragment from COS O1-22 into the pUC119 plasmid. This fragment contains the 1.7-kb 5-flanking region, all of the first exon, and part of the first intron of the mST8Sia II gene. A ³²P-labeled S1 probe complementary to nucleotides 393 to +255 was generated using a GeneAmp XL PCR kit, with pO1-22E4.8 as the template and O1-EX3 as the primer. The extension products were digested with SmaI and denatured at 100 °C for 3 min, and then the 648-nucleotide ³²P-labeled S1 probe was separated from the template by electrophoresis on a 6% sequencing gel. The probe was isolated by soaking gel pieces in the elution buffer (300 mM sodium acetate, pH 5.0, and 10 mM EDTA) at room temperature for 8 h, followed by ethanol precipitation.

The S1 probe was hybridized with 5 µg of mouse brain poly(A) RNA or 5 µg of yeast tRNA in 10 µl of hybridization buffer (40 mM PIPES, pH 6.4, 1 mM EDTA, 0.4 M NaCl, and 50% formamide) overnight at 50 °C, after denaturation of the mixture for 10 min at 80 °C. Then 200 µl of S1 mapping buffer (280 mM NaCl, 50 mM sodium acetate, pH 5, 4.5 mM ZnSO₄, 10 mg/ml salmon sperm DNA, and 1000 units/ml S1 nuclease) was added to each sample, and the mixtures were incubated for 1 h at 37 °C. Digestion was stopped by the addition of 1 µl of 0.5 M EDTA, and the nucleic acids were recovered by ethanol precipitation. The protected products were separated on a 6% sequencing gel along with a dideoxy chain termination sequencing reaction of pO1-22E4.8, using O1-EX3 as the primer.

Recombinant Sp1 was purchased from Promega. Plasmid pO1-FT was constructed by subcloning the PCR product using pO1-22E4.8, as the template, the O1-FTA primer (5-AGCAAAGCTGTCAAACTGCGCCTGGAGCCCAG-3; mST8Sia II coding strand, nucleotides 124 to 93) and O1-FTB primer (5-TCCACGCGCACGAGGGACACACACCCTGCGCT-3; complementary to the mST8Sia II coding strand, nucleotides +94 to +63) into pBluescript SK+. ³²P-Labeled DNA fragments were prepared by PCR using ³²P-labeled O1-FTA and cold O1-FTB, with pO1-FT as the template. The binding reaction was carried out for 30 min at 25 °C in 50 µl of a mixture comprising 25 mM HEPES-KOH, pH 7.9, 0.5 mM EDTA, 50 mM KCl, 10% glycerol, 1 mM phenylmethanesulfonyl fluoride, 1 mM dithiothreitol, 0.3 µg of poly(dI-dC), 10 kcpm of ³²P-labeled DNA fragment, and 1 footprinting unit of recombinant Sp1. Five µl of freshly diluted DNase I (50 µg/ml) was added to the mixture for 60 s at 25 °C, and then the reaction was stopped by the addition of 100 µl of 1% sodium dodecyl sulfate containing 20 mM EDTA and 200 mM NaCl. After phenol-chloroform extraction and ethanol precipitation, the reacted products were separated on a 6% sequencing gel along with a dideoxy chain termination sequencing reaction of pFT-O1, using O1-FTA as the primer.

To obtain various lengths of the 5-flanking region of the mST8Sia II gene, PCRs were performed with Tth DNA polymerase (GeneAmp XL PCR kit) for 25 cycles in a DNA thermal cycler (Perkin-Elmer). PCR was carried out using pO1-22E4.8 as the template, and as primer, O1-ATGNco (5-GCAGCTTGGGGTCGGTGCCTCCGG-3; complementary to the mST8Sia II coding strand, nucleotides +177 to +149 and including an NcoI linker) and a reverse sequencing primer. The amplified fragment was digested with KpnI within the pUC119 vector polylinker, blunt ended, and then digested with NcoI within the O1-ATGNco primer. For construction of pBO1-EN1.8, the resultant DNA fragment was subcloned into SmaI-NcoI-digested pPicaGene-Basic II (pPGBII; Toyo-ink, Japan). pBO1-BN0.8 and pBO1-SN0.45 were generated by subcloning the BamHI-NcoI- and SmaI-NcoI-digested PCR products into pPGBII, respectively. Series of deletion plasmids were constructed by subcloning the restriction enzyme-digested PCR products. The primers and template plasmids used were pBO1-XN0.15, carrying a 0.15-kb XhoI-NcoI fragment amplified by using the primer set of O1-150X/O1-ATGNco; and pBO1-XN0.35, carrying a 0.35-kb XhoI-NcoI fragment amplified by using the primer set of O1-350X/O1-ATGNco; the sequences of the primers were O1-150X, 5-AGCGGGCCGCGCCAGAGCAAC-3 (nucleotides 16 to +9, including an XhoI linker) and O1-350X, 5-TGGCCGTTAGTGGGAGGA-3 (nucleotides 165 to 142, including an XhoI linker). Plasmid pO1-22E4.8 was used as the template. Plasmids pBO1-NhN3.5, pBO1-RN5.5, and pBO1-EN9.6 were constructed as follows. An 8-kb EcoRI fragment from COS O1-22 (from 9600 to 1646 in Fig. 5) was subcloned into the EcoRI site of pBluescript SK+, and the resultant plasmid was then double-digested with either set of restriction enzymes, NheI-EcoRI (3400 to 1645) or EcoRV-EcoRI (5400 to 1645), or only EcoRI (9600 to 1645). Each digested DNA fragment was then subcloned into pBO1-EN1.8 to obtain plasmids pBO1-EN9.6, pBO1-RN5.5, and pBO1-NhN3.5, respectively.

NIH 3T3 and undifferentiated P19 cells were seeded at 5 × 10⁴ cells/60-mm-diameter dish in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum 24 h prior to transfection, respectively. For the differentiation of P19 cells into neuronal cells, the cells were seeded into and aggregated in bacteriological grade dishes in the presence of 1 µM retinoic acid at the cell density of 1 × 10⁵/ml. After 3 days, the aggregates were trypsinized, and approximately 1 × 10⁵ cells/60-mm diameter dish (tissue culture-grade dishes) were plated in Dulbecco's modified Eagle's medium, 10% fetal calf serum 24 h prior to transfection.

The luciferase plasmid (5 µg) used as the reporter and the pSR-Gal plasmid (0.5 µg) used as an insertional control for transfection efficiency were transfected into the cells by means of Lipofectamine (Life Technologies, Inc.). After 48-h transfection, the cells were washed three times with phosphate-buffered saline and then lysed with cell lysis buffer (PG-50; Toyo-ink). Luciferase activity was measured using a PicaGene Luciferase assay system (Toyo-ink) and a Luminescencer AB-2000 (ATTO, Japan). Light activity measurements were performed in quadruplicate, averaged, and then normalized as to -galactosidase activity to correct for the transfection efficiency. -Galactosidase activity was measured using a Luminescent -Galactosidase Detection Kit (Clontech).

RESULTS

Screening of an NIH 3T3 cell cosmid library for mST8Sia II cDNA resulted in the isolation of four independent clones. A restriction map of the approximately 100-kb region containing the mST8Sia II gene is shown in Fig. 1.

We sequenced the exons, to determine their exact sizes, as well as the sequences of the intron/exon junctions (Table I). The sequences of all the intron-exon splice junctions conformed to the GT-AG rule (25). Introns were present only in the protein coding region of mST8Sia II. All of the splicing junctions of the mST8Sia II gene occurred after the second nucleotide of the amino acid codon. The mST8Sia II cDNA was divided into six exons, ranging from 63 to 4,340 bp, with intron sizes of about 4-39 kb and spanning approximately 80 kb of genomic DNA (Fig. 1). Exon 1 contained the entire 5-untranslated region and the beginning of the coding region to amino acid residue 32. This exon contained a cytoplasmic domain, a hydrophobic signal anchor sequence, and part of a stem domain. Exons 4-6 encoded the putative active domain of the enzyme, and exon 6 contained a large 3-untranslated region. We reported previously mST8Sia II cDNA sequences lacking the whole 3-untranslated region. Therefore, to determine the 3 end of the 5-kb mST8Sia II transcript, the mouse brain cDNA library was screened by PCR using primers distributed along the 3 part of the gene. Sequence analyses of the PCR products revealed that the size of the transcribed RNA was 5,350 bp, thus it included a large 3-untranslated region of 4055 bp. Poly(A) addition had occurred 18 nucleotides downstream of the sequence at the T residue of the polyadenylation signal (AATAAA).

Table I. Exon/intron junctions of the mST8Sia II gene Nucleotide sequences at the intron (lowercase letters) and exon (uppercase letters) junctions are shown. The derived amino acid sequence is displayed below the nucleotide sequence. Exons are numbered from the 5 end as described in Fig. 1. Exon and intron sizes are indicated in base pairs. Numbering starts with position +1 at the adenosine of the initiator methionine. Exon Splice donor Intron Splice acceptor bp kbp 32 34 1 265 GAA GAA ATC GGgtaaatagccgccc 39  tttcctttgtcttgcagG AAT TCT Glu Glu Ile Asn Ser 53 55 2 63 AAA TCT AAT AGgtttgtaaattaga 8  tttttacccccttccagA GCT GAA Lys Ser Asn Ala Glu 96 98 3 129 CTG AGG ATC AGgtattggtcggtca 6  ttatttggttttcacagG AAG CAA Lys Arg Ile Lys Gln 182 184 4 258 TTT GTC ATA AGgtaaccatccacca 4  tgctttttattcctcagG TGC AAC Phe Val Ile Cys Asn 280 282 5 294 GCA GTC CGA GGgtgagtagctctca 16  ttttatttcttctccagA TAT TGG Ala Val Arg Tyr Trp

The transcription start site was determined by S1 nuclease mapping and primer extension with RNA recovered from 1-day mouse brain, in which the mST8Sia II gene was highly expressed (Fig. 2). A single stranded ³²P-labeled S1 probe, corresponding to nucleotides 393 to +205 of the mST8Sia II gene, was hybridized with poly(A) RNA, followed by S1 nuclease digestion. The protected fragments were analyzed on a 6% sequencing gel. Yeast tRNA was used as a control to protection specificity. The end points of the protected fragments were determined by comparison with a sequence ladder derived from the same genomic DNA template (pO1-22E4.8) and the original primer, O1-EX3 (nucleotides +255 to +224, which is complementary to the mST8Sia II mRNA), used for synthesizing the S1 probe. The end point was determined to be at cytosine (+1), which corresponded to a position 167 nucleotides upstream from the initiation codon, ATG (Fig. 2A). Control experiments involving tRNA showed no such band. Primer extension with the same primer (O1-EX3) as used for S1 analysis resulted in extension products corresponding to the same cytosine (+1) (Fig. 2B). Consistent with the results of primer extension and S1 nuclease experiments, a predominant initiation site was found 167 nucleotides upstream of ATG, which gave the only comigrating product in both S1 nuclease protection and primer extension experiments. Moreover, we performed 5 RACE-PCR on newborn mouse brain poly(A) RNA to identify the 5 end of the ST8Sia II gene, and the longest RACE-PCR products corresponded to the transcription initiation site determined in the S1 nuclease protection and primer extension experiments.

Analysis of the 5-flanking Region

Analysis of the sequence 2.5-kb upstream of the transcription initiation site revealed that the mST8Sia II gene promoter has no typical TATA or CAAT boxes (Fig. 3). As shown in Fig. 3, the sequence of the putative promoter region of the mST8Sia II gene was embedded in a GC-rich domain. The GC content of the sequence between immediately upstream of the transcription initiation site and the translational initiation site (nucleotides 175 to +168) was 74%. A pyrimidine-rich region (nucleotides 500 to 453) and a purine-rich region (nucleotides 83 to 11) were found in the 5-flanking region. The TATA- and CAAT-less mST8Sia II gene promoter contains three Sp1 binding sequences, (G/T)GGGCGG(G/A)(G/A)(C/A), at positions 34 (matching 9-10), 80 (matching 9-10), and 170 (matching 6-10), and an inverted Sp1 binding site at position 573 (matching 8-10). To determine whether or not the proximal Sp1 binding sites upstream of the transcriptional initiation site are functional, we performed DNase I footprinting analysis with Sp1. As shown in Fig. 4, the Sp1 binding sites at positions 34 and 80 were both protected.

Nucleotide sequence of the 5-flanking region of the mST8Sia II gene.

To determine mST8Sia II gene promoter activities, we used mouse embryonal carcinoma P19 cells in which the mST8Sia II gene dramatically increases during neural differentiation on retinoic acid treatment (21).

To determine whether or not the 5-flanking sequence of the mST8Sia II gene contains a cell type-specific promoter, we constructed a reporter plasmid, pBO1-EN9.6, containing a 9.6-kb 5-flanking sequence of the mST8Sia II gene (9600 to +168) fused to the promoterless luciferase gene in pPGBII (Fig. 5). The construct was assayed for promoter activity by transient transfection into undifferentiated and differentiated P19 cells, and fibroblast NIH 3T3 cells (not expressing the ST8Sia II gene). As a negative control, pPGBII was transfected into parallel cultures of each cell line. The luciferase activity due to each luciferase reporter plasmid was normalized as to the -galactosidase activity by cotransfecting an internal control plasmid, pSR-Gal, carrying a -galactosidase gene under the control of the SR promoter. The pBO1-EN9.6 construct showed the highest level of promoter activity when expressed in the differentiated P19 cells, which was 4-fold higher than that in undifferentiated P19 cells. No promoter activity was detected in NIH 3T3 cells. These results suggest that the 9.6-kb 5-flanking sequence contains sufficient information to direct cell type-specific expression of the mST8Sia II gene.

A series of reporter plasmids containing progressive 5 deletions between nucleotides 9600 and 9 was constructed, and the plasmids were examined for their promoter activities in each cell type to map the regions that regulate cell type-specific gene expression (Fig. 5). Sequential deletions of the region between nucleotides 9600 and 5400 had little effect on luciferase activity. However, further deletions from nucleotides 5400 to 3400 increased the promoter activity in differentiated P19 cells, suggesting the presence of a negative regulatory element in this region. Deletions up to nucleotide 659 had no effect on the activity in differentiated P19 cells but increased the activity in undifferentiated P19 cells. Deletions of the sequence from positions 659 to 293 increased the luciferase gene expression in differentiated P19 cells but caused a decrease to one-half in undifferentiated P19 cells. The region between 659 and 293 contained an inverted Sp1 binding site and a pyrimidine-rich region (Fig. 3). Moreover, further deletions to nucleotide 158 (pBO1-XN0.35) caused the promoter activity to decrease to approximately half of that of pBO1-SN0.45 in both differentiated and undifferentiated P19 cells, suggesting the presence of a positive regulatory element in this region (nucleotides 158 to 293), which contains one Sp1 binding site. Extension of the 5 deletion to nucleotide 9 (pBO1-XN0.15) drastically reduced the promoter activity to the level seen with the promoterless control vector pPGBII in all cells, indicating that the sequence between 158 and 9 is necessary for the minimal promoter activity. Two Sp1 binding sites, which were revealed to be functional on footprinting analysis (Fig. 4), and a purine-rich region were found in the region between 9 and 158. Interestingly, the promoter activities due to the pBO1-XN0.35 and pBO1-SN0.45 constructs were approximately 10-fold higher in differentiated P19 cells than undifferentiated P19 cells. On the other hand, all constructs gave hardly detectable activity in NIH 3T3 cells, which do not express the mST8Sia II gene. These results suggest that the 325-bp sequence upstream from the translational initiation codon (nucleotides 158 to +167) is necessary for the minimum and cell type-specific promoter activity.

DISCUSSION

In this study, we determined the entire genomic organization of the mST8Sia II gene, characterized the functional promoter activity for the 5-flanking region in transient transfection assays, and investigated the expression of the gene in mouse teratocarcinoma P19 cells. We demonstrated that the minimal promoter region of the mST8Sia II gene is sufficient to express cell type-specific promoter activity, which is correlated with mST8Sia II gene expression in the cells.

Previously, the genomic structures of five other sialyltransferase genes have been reported, i.e. rat galactoside 2,6-sialyltransferase (rST6Gal I) (26, 27), human galactoside 2,3-sialyltransferase (hST3Gal I) (28), human Gal1,3GalNAc/Gal1,4GlcNAc 2,3-sialyltransferase (hST3Gal IV) (29), mouse N-acetylgalactosamide 2,6-sialyltransferase (mST6GalNAc II) (30), and mouse sialoside 2,8-sialyltransferase (mST8Sia III) (22). The genomic structure of the mST8Sia II gene was most similar to that of the mST8Sia III gene, as follows (Fig. 6). The putative active domains of mST8Sia II and mST8Sia III were encoded by only three and two exons, respectively, whereas those of the other genes consisted of at least five exons. In particular, the positions of the two exon-intron boundaries of the mST8Sia II gene (introns 3 and 5) and the mST8Sia III gene (introns 2 and 3) were identical, and all of their splice junctions occurred after the second nucleotide of the amino acid codon. Although in mST8Sia III sialyl motif L, a highly conserved putative nucleotide sugar binding domain (31) is encoded by one exon, the corresponding motif in mST8Sia II is encoded by discrete exons, like those observed in other genes. The similarity of the genomic organization and amino acid sequences between mouse ST8Sia II and ST8Sia III suggests that these genes constitute a subgroup distinct from other sialyltransferase genes.

The results of primer extension and S1 protection analyses showed that mST8Sia II gene has a single transcription initiation site. In addition, we demonstrated that the 5-flanking sequence of the mST8Sia II gene contained a functional promoter that was highly active in neural differentiated P19 cells, which strongly express the mST8Sia II gene. Deletion analysis clearly indicated that the minimal promoter, the region contained within the 158 bp upstream of the transcription start site (pBO1-XN0.35), exhibited substantial tissue specificity (Fig. 5). In differentiated P19 cells, an approximately 10-fold increase in promoter activity was observed with the pBO1-XN0.35 construct compared with that in undifferentiated P19 cells. We showed recently that the expression of the mST8Sia II gene in neural differentiated P19 cells increased dramatically to a level approximately 10-20 times higher than that in undifferentiated cells (21). The increase in minimal promoter activity on neural differentiation is correlated with the gene expression of mST8Sia II in P19 cells. Moreover, the promoter activity was negligible in NIH 3T3 cells, in which the mST8Sia II gene was not expressed. Therefore, the minimal promoter region may possess the ability to initiate transcription in a cell-specific fashion.

The proximal region 158 bp upstream from the transcription initiation site was characterized as a minimal promoter that contained two functional Sp1 binding sites but lacked consensus TATA and CAAT boxes. In addition, the 5-untranslated region of the mST8Sia II gene contained a large GC-rich domain with the characteristics of a CpG island. These structural features are usually associated with housekeeping gene promoters. However, recent studies have shown that many tissue-specific gene promoters, including neural cell-specific promoters, include the structural features of housekeeping gene-like promoters. Neuron-specific TATA-less promoters have been shown to have either a single transcription initiation site, as observed in the synapsin I gene (32), synapsin II gene (33), nerve growth factor receptor gene (34), and several olfactory neuron-specific genes (35), or to have multiple initiation sites, such as those found in the genes encoding synaptophysin, the type II sodium channel (36), the D_1A dopamine receptor (37), and brain-specific aldolase C (38). Therefore, the mST8Sia II promoter may belong to the former group, i.e. GC-rich and TATA-less promoters with a single transcription initiation site.

Two Sp1 binding sites were found in the minimal promoter region, and DNase I protection assays with the purified Sp1 protein revealed that these two sites were functional (Fig. 4). Moreover, two additional Sp1 binding sites were present in the regulatory regions (158 to 293 and 293 to 659). It was reported previously that the distantly located site functions synergistically with the promoter-proximal site to activate transcription strongly in vivo, and that this synergism is a direct consequence of interactions between remote and local Sp1s (39). Deletion of the sequence from positions 158 to 293 of mST8Sia II reduced the promoter activity, suggesting that Sp1 bound to the site at 170 probably interacts with another Sp1 in the minimal promoter region. The expression of Sp1 is ubiquitous, but it was reported previously that Sp1 levels appeared to be highest in developing hematopoietic cells, fetal cells, and spermatides in the mouse (40). The regulated expression of the mST8Sia II gene in fetal and newborn brain and testis seemed to correspond to that of Sp1. Although Sp1 was found to be expressed in NIH 3T3 and undifferentiated and differentiated P19 cells (data not shown), the expression level of the reporter gene fused to the minimal promoter sequence of ST8Sia II (158/+168) was very low in undifferentiated P19 cells and hardly detectable in 3T3 cells. These results suggest that some cell-specific transcription factor is involved in differentiated P19 cells and that it might be required for interaction with Sp1 in the minimal promoter elements, which allows activation of the mST8Sia II gene promoter. Further studies involving identification of the regulatory elements and their binding factors will facilitate elucidation of the tissue-specific and developmental regulation of ST8Sia II gene expression.