Characterization of the Second Type of Human β-Galactoside α2,6-Sialyltransferase (ST6Gal II), Which Sialylates Galβ1,4GlcNAc Structures on Oligosaccharides Preferentially GENOMIC ANALYSIS OF HUMAN SIALYLTRANSFERASE GENES

A novel member of the human β-galactoside α2,6-sialyltransferase (ST6Gal) family, designated ST6Gal II, was identified by BLAST analysis of expressed sequence tags and genomic sequences. The sequence of ST6Gal II encoded a protein of 529 amino acids, and it showed 48.9% amino acid sequence identity with human ST6Gal I. Recombinant ST6Gal II exhibited α2,6-sialyltransferase activity toward oligosaccharides that have the Galβ1,4GlcNAc sequence at the nonreducing end of their carbohydrate groups, but it exhibited relatively low and no activities toward some glycoproteins and glycolipids, respectively. It is concluded that ST6Gal II is an oligosaccharide-specific enzyme compared with ST6Gal I, which exhibits broad substrate specificities, and is mainly involved in the synthesis of sialyloligosaccharides. The expression of the ST6Gal II gene was significantly detected by reverse transcription PCR in small intestine, colon, and fetal brain, whereas the ST6Gal I gene was ubiquitously expressed, and its expression levels were much higher than those of the ST6Gal II gene. The ST6Gal I gene was also expressed in all tumors examined, but no expression was observed for the ST6Gal II gene in these tumors. The ST6Gal II gene is located on chromosome 2 (2q11.2-q12.1), and it spans over 85 kb of human genomic DNA consisting of at least eight exons and shares a similar genomic structure with the ST6Gal I gene. In this paper, we have shown that ST6Gal I, which has been known as the sole member of the ST6Gal family, also has the counterpart enzyme (ST6Gal II) like other sialyltransferases.

Cell surface carbohydrate chains on glycoproteins and glycolipids are considered to play important roles in a variety of biological phenomena, such as cell-cell communication, cellsubstrate interaction, adhesion, and protein targeting. Among the carbohydrate chain components, sialic acids (Sia) 1 are negatively charged acidic sugars and usually occur at the terminal ends of carbohydrate chains. Therefore, sialic acids function as key determinants of carbohydrate structures. The sialylglycoconjugates of glycoproteins and glycolipids also vary according to the tissue and cell type and are subject to change during development and oncogenesis (1). For the synthesis of sialylglycoconjugates, a family of glycosyltransferases called sialyltransferases catalyzes the transfer of a sialic acid from CMP-Sia to an acceptor carbohydrate. All mammalian sialyltransferases characterized to date have type II transmembrane topology and contain highly conserved motifs called sialyl motifs L (long), S (short), and VS (very short) (2)(3)(4). Sialyl motif L is characterized by a 45-60-amino acid region in the center of the protein, and it has been shown to be involved in the binding of a donor substrate, CMP-Sia (5). Sialyl motif S is located in the COOH-terminal region and consists of a 20 -30-amino acid stretch. It has been shown to be involved in the binding of both the donor and acceptor substrates (6). Sialyl motif VS is also located in the COOH-terminal region, within which one glutamic acid residue is always found separated by four amino acid residues from a highly conserved histidine residue. This motif is thought to be involved in the catalytic process (4,7). Based on the high sequence conservation of sialyl motifs L and S, PCR-based cloning of sialyltransferase cDNAs has been performed extensively (reviewed in Refs. 8 and 9). In addition, some sialyltransferase cDNAs have been cloned efficiently using the sequence information derived from the expressed sequence tag database (10 -14). So far, cDNA cloning of 19 members of the mammalian sialyltransferase family has been performed, and they have been grouped into four families according to the carbohydrate linkages they synthesize: ␤-galactoside ␣2,3-sialyltransferase (ST3Gal I-VI), ␤-galactoside ␣2,6sialyltransferase (ST6Gal I), GalNAc ␣2,6-sialyltransferase (ST6GalNAc I-VI), and ␣2,8-sialyltransferase (ST8Sia I-VI) (14).
Among them, ST6Gal I 2 is the sole member of the ␣2,6-sialyltransferase family that synthesizes the Sia␣2, 6Gal␤1,4GlcNAc structure. This structure is found mainly in N-linked glycans, but it has also been found in some O-glycans, glycosphingolipids, and sialyloligosaccharides. It has also been known as the ligand for the B cell-specific lectin, CD22/Siglec-2 (16 -18), which is important for B cell function (19 -21). Investigation of ST6Gal I knock-out mice showed that they are viable but are deficient in cell surface Sia␣2,6Gal␤1,4GlcNAc structures and exhibit hallmarks of severe immunosuppression (22). These mice displayed reduced serum IgM levels, impaired B cell proliferation in response to IgM and CD40 cross-linking, and attenuated antibody production to T-independent and Tdependent antigens. Deficiency of ST6Gal I was further found to alter phosphotyrosine accumulation during signal transduction from the B lymphocyte antigen receptor. These data reveal that ST6Gal I and its corresponding product of the Sia␣2,6Gal␤1,4GlcNAc structures are essential in promoting B lymphocyte activation and immune function. Although ST6Gal I knock-out mice lost ST6Gal activity to the background level, there remained a possibility that other ST6Gal enzymes exist that produce low levels of Sia␣2,6Gal␤1,4GlcNAc structure or exhibit different substrate specificity from that of ST6Gal I. Recent progress of the human genome project enables us to detect another ST6Gal gene. We report here cloning and expression of the second type of human ␤-galactoside ␣2,6-sialyltransferase (ST6Gal II) that has preference for synthesizing sialyloligosaccharides.
Isolation of ST6Gal II cDNA-Human expressed sequence tags (EST; GenBank TM accession numbers BE612797, BE613250, and BF038052) and human genomic sequences (GenBank TM accession numbers AC005040, AC016994, and AC108049) with similarity to human ST6Gal I were identified using the tBLASTn algorithm against the dbEST and high throughput genomic sequence databases at the National Center for Biotechnology Information, respectively. The EST clones were obtained from the Image Consortium. To obtain the entire coding region, reverse transcription (RT)-PCR was performed using primer sets, 5Ј-TTGGAATTCTCATCATGATGTCCATGTGC-3Ј (nucleotides 1491-1519 in Fig. 1A, the synthetic EcoRI site being underlined) and 5Ј-ACTTTGAGTACAACAGTAGTACC-3Ј (complementary to nucleotides 1797-1819) and 5Ј-CAGATTCTGACCAACCCCAG-3Ј (nucleotides 1214 -1233) and 5Ј-CAAGAATTCCAATGAAACCAGAAGATG-GTG-3Ј (complementary to nucleotides 1473-1502, the synthetic EcoRI site being underlined), with the first strand cDNA of human colon (human multiple tissue cDNA panels; Clontech) as a template. The above PCRs were performed as follows: 94°C for 60 s, 45 cycles of 94°C for 60 s, 50°C for 60 s, 72°C for 90 s, and 72°C for 10 min. The PCR products and EST clones were cloned into pBluescript II SK(ϩ) vector and combined. The nucleotide sequence was convinced by the dideoxy termination method using an ABI PRISM 377 DNA sequencer (Applied Biosystems, Foster City, CA).
Construction of Expression Vectors-We constructed an expression vector encoding the soluble ST6Gal II. The XhoI site was introduced at the nucleotide position 268 by PCR-based site-directed mutagenesis using a primer, 5Ј-TCATCTACTTCACCTCGAGCAACCCCGCTG-3. Then the XhoI fragment encoding a truncated form of ST6Gal II (lacking the first 32 amino acids of the coding region) was prepared and subcloned into the XhoI site of the expression vector pcDSA. The resulting plasmid, designated pcDSA-ST6Gal II, encodes a soluble fusion protein consisting of the IgM signal peptide, the Staphylococcus aureus protein A IgG-binding domain, and the truncated form of ST6Gal II. The expression vector of the soluble ST6Gal II (short form) was constructed as described above except using the cDNA encoding short form of ST6Gal II as a template for PCR-based site-directed mutagenesis and was designated pcDSA-ST6Gal II/short.
The expression vector of the soluble ST6Gal I was constructed as follows. The DNA fragment encoding the whole coding region of human ST6Gal I was amplified by PCR using primers 5Ј-TTATGATTCACAC-CAACCTGAAG-3Ј (nucleotides 309 -331 of GenBank TM accession number NM_003032) and 5Ј-GCCTGTGCTTAGCAGTGAATG-3Ј (complementary to nucleotides 1519 -1539) with human liver cDNA as a template and cloned into pBluescript II SK(ϩ) vector. Then the 1.1-kb EcoRI fragment encoding a truncated form of ST6Gal I (lacking the first 43 amino acids of the coding region) was prepared and subcloned into the EcoRI site of pcDSA, which was designated pcDSA-ST6Gal I.
We also constructed an expression vector containing the whole coding region of ST6Gal II. The 1.8-kb HindIII fragment containing the whole coding region of ST6Gal II was prepared from the cloned cDNA and subcloned into the HindIII site of the expression vector pRc/CMV, which was designated pRc/CMV-ST6Gal II.
Preparation of Soluble Sialyltransferases-For production of soluble forms of sialyltransferases, COS-7 cells were transfected with the above pcDSA vectors using LipofectAMINE TM reagent (Invitrogen) and cultured as described previously (23). The protein A-fused sialyltransferases expressed in the medium was adsorbed to IgG-Sepharose gel (Amersham Biosciences) and used as the enzyme source.
Sialyltransferase Assays and Product Characterization-Sialyltransferase assays were performed as described previously (25,26). In brief, enzyme activity was measured in 50 mM MES buffer (pH 6.0), 1 mM MgCl 2 , 1 mM CaCl 2 , 0.5% Triton CF-54, 100 M CMP-[ 14 C]NeuAc, an acceptor substrate, and an enzyme preparation in a total volume of 10 l. As acceptor substrates, 10 g of glycoproteins, 5 g of glycolipids, or 10 g of oligosaccharides were used. The enzyme reaction was performed at 37°C for 3-20 h. For glycoproteins, the reaction was terminated by the addition of SDS-PAGE loading buffer, and the reaction mixtures were subjected directly to SDS-PAGE. For glycolipids, the reaction mixtures were applied to a Sep-Pak Vac C 18 column (100 mg; Waters, Milford, MA), and purified glycolipids were subjected to high performance thin-layer chromatography (HPTLC) (Silica-Gel 60; Merck) with a solvent system of chloroform, methanol, and 0.02% CaCl 2 (55: 45:10). For oligosaccharides, the reaction mixtures were directly subjected to HPTLC with a solvent system of 1-propanol, aqueous ammonia, and water (6:1:2.5). The radioactive materials were visualized and quantified with a Fuji BAS2000 radioimage analyzer. The intensity of the radioactivity was converted into moles using the radioactivities of various amounts of CMP-[ 14 C]NeuAc (12.0 GBq/mmol, 925 kBq/ml) as standards. Quantification was performed within the linear range of the standard radioactivity.
For kinetic analysis, the reaction was performed as described above except using various concentrations of acceptor substrates. Under these conditions, the product formation from the individual acceptor substrates was linear up to 4 h. Kinetic parameters were determined by Lineweaver-Burk plots.

Analysis of ST6Gal I and II Gene Expression in Various Human
Tissues and Tumors-Relative expression levels of ST6Gal I and II mRNAs were estimated by RT-PCR using human multiple tissue cDNA panels (Clontech) as templates. For the analysis of ST6Gal II gene expression, ST6Gal II-specific primers 5Ј-AGACGTCATTTTGGTGGC-CTGGG-3Ј (nucleotides 1264 -1286) and 5Ј-TTAAGAGTGTGGAAT-GACTGG-3Ј (complementary to nucleotides 1745-1765) were used. For the analysis of ST6Gal I gene expression, ST6Gal I-specific primers 5Ј-TTATGATTCACACCAACCTGAAG-3Ј (nucleotides 309 -331 of Gen-Bank TM accession number NM_003032) and 5Ј-CTTTGTACTTGT-TCATGCTTAGG-3Ј (complementary to nucleotides 658 -680) were used. As a control, glyceraldehyde 3-phosphate dehydrogenase (G3PDH) gene expression was also measured using G3PDH-specific primers 5Ј-GGATCCACCACAGTCCATGCCATCAC-3Ј and 5Ј-AAGCTTTCCACCACCCTGTTGCTGTA-3Ј (27). PCRs were performed as follows: 94°C for 60 s, 40 cycles of 94°C for 60 s, 50°C for 60 s, and 72°C for 90 s for ST6Gal I and II genes and 25 cycles for G3PDH gene and 72°C for 10 min. The PCR products were electrophoresed on a 2% agarose gel, stained with ethidium bromide, and then visualized under UV light.

RESULTS
Cloning and Nucleotide Sequencing of a New Sialyltransferase cDNA-Using the human expressed sequence tag and high throughput genomic sequence databases, we found some sequences (GenBank TM accession numbers BE612797, BE613250, and BF038052 (EST clones) and AC005040, AC016994, and AC108049 (genomic sequences)) with similarity to human ST6Gal I. These sequences were distinct from those of the sialyltransferases cloned previously, suggesting that these clones encode a novel member of the sialyltransferase family. The above EST clones were obtained from the Image Consortium, but these clones did not contain the expected entire coding sequence of the novel sialyltransferase. As these EST clones were considered to lack the DNA sequence encoding the COOH-terminal region of the new sialyltransferase, we performed RT-PCR to obtain it with the human colon first strand cDNA as a template. However, we could not amplify the target sequence as a single DNA fragment at first, so we amplified it as two DNA fragments (nucleotides 1214 -1502 and 1491-1819 in Fig. 1A), and each fragment was cloned into pBluescript II SK(ϩ) vector. Then these fragments were combined using a synthetic EcoRI site, which does not change encoded amino acids. It should be noted that later we could amplify the corresponding region as a single fragment by RT-PCR with another primer set (see Fig. 5), and this fragment encoded the same amino acid sequence with the above clone. From the plasmid containing the combined DNA fragment, the 0.5-kb AatII-XhoI fragment was prepared, and this was ligated into the AatII-XhoI sites of the plasmid containing the EST clone BE612797, and the DNA fragment encoding the new sialyltransferase having the expected amino acid sequence was obtained.
The nucleotide sequence of the putative new sialyltransferase cDNA and its deduced amino acid sequence are shown in Fig. 1A. The predicted protein consists of 529 amino acids with a calculated molecular mass of 60,157 Da and with four potential N-linked glycosylation sites. The position of the initiation codon was estimated according to the Kozak consensus sequence (28). Hydropathy analysis (29) indicated one prominent hydrophobic sequence of 19 amino acids in length in the NH 2terminal region, predicting that the protein has type II transmembrane topology characteristic of many other glycosyltransferases cloned to date (Fig. 1B). Comparison of the deduced amino acid sequence with those of other human sialyltransferases showed significant sequence identity in two regions, sialyl motifs L (41.7-65.9%) and S (26.1-56.5%). The overall amino acid sequence of the predicted protein showed the highest sequence identity with ST6Gal I (48.9%) (Fig. 2). These results strongly suggest that the predicted protein belongs to the sialyltransferase family, especially the ST6Gal-family. Thus, we tentatively designated the new sialyltransferase as ST6Gal II. It is striking that the stem region of ST6Gal II, which is located between the transmembrane domain and the active domain, is very long like mouse, chicken, and human ST6GalNAc I (30 -32).
We also found by extensive database searches that ST6Gal II is related closely to KIAA1877 proteins (GenBank TM accession numbers AB058780 and XM_038616), whose cDNA clones were isolated as one of unidentified human genes by a sequencing project (33). The exon 1 of the AB058780 clone is different from that of the ST6Gal II gene (Fig. 1A). The start codon of the AB058780 clone is not identified, but this clone has essentially the same amino acid sequence with ST6Gal II except it has five additional amino acid residues in its NH 2 -terminal region. The XM_038616 clone has been reported as a protein consisting of 463 amino acids that shares the COOH-terminal 234 amino acids with ST6Gal II at first. The differences in the NH 2terminal region were caused by frame shifts of the coding region. However, this sequence was updated recently, and it was shown that the XM_038616 clone has the same amino acid sequence with ST6Gal II. In addition, there is a splicing variant of ST6Gal II that has short different amino acid sequence in the COOH-terminal region and lacks most of the sialyl motif S (GenBank TM accession numbers BC008680 and BE613250; see Fig. 1A, Short form).
Sialyltransferase Activity of the Newly Cloned Enzyme-To facilitate determination of the enzymatic activity of the new sialyltransferase, we constructed the expression plasmid pcDSA-ST6Gal II, which allows expression of ST6Gal II lacking the transmembrane domain as a secretable protein fused with the IgG-binding domain of S. aureus protein A. The plasmid was then transfected into COS-7 cells, and the protein A-fused ST6Gal II expressed in the medium was adsorbed to IgG-Sepharose resin, which was used as the enzyme source. For comparative analysis, the protein A-fused ST6Gal II short form and ST6Gal I were also prepared. As shown in Table I  bohydrate groups. The apparent K m values of ST6Gal II for Gal␤1,4GlcNAc and lacto-N-neotetraose were estimated to be 0.71 and 0.48 mM, respectively, which were significantly lower than that of ST6Gal I for Gal␤1,4GlcNAc (2-10 mM) (34). However, ST6Gal II did not exhibit activity toward oligosaccharides such as Gal␤1,3GalNAc, Gal␤1,3GlcNAc, lactose, and lacto-Ntetraose, all of which do not contain Gal␤1,4GlcNAc structure at the nonreducing end of their carbohydrate groups. ST6Gal II also exhibited relatively low activity toward some glycoproteins, which are considered to have Gal␤1,4GlcNAc structure at the nonreducing end of their carbohydrate groups. However, ST6Gal II did not exhibit activity toward glycolipids examined in this study, including paragloboside, which has Gal␤1,4GlcNAc structure at the nonreducing end of its carbohydrate group. We also examined the enzymatic activity of the short form of ST6Gal II lacking most of the sialyl motif S (Fig.   FIG. 2. Sequence comparison of human ST6Gal I and II. The conserved amino acid residues are boxed. Sialyl motifs L and S are double underlined and dashed underlined, respectively. The conserved His and Glu residues in sialyl motif VS are marked with asterisks.

TABLE I Acceptor substrate specificity of human ST6Gal I and II
Various acceptor substrates were incubated in the standard assay mixture using soluble sialyltransferase fused with protein A as an enzyme source. Each substrate was used at the concentration of 0.5 mg/ml for glycolipids and 1 mg/ml for glycoproteins and oligosaccharides. Relative rates are calculated as a percentage of the incorporation obtained with Gal␤1, 4GlcNAc. R represents the remainder of the N-linked oligosaccharide chain. *, 1.03 pmol/h/ml medium; **, 8.14 pmol/h/ml medium. 1A), but it exhibited no sialyltransferase activity (Fig. 3).
On the other hand, ST6Gal I exhibited more or less activity toward oligosaccharides Gal␤1,4GlcNAc, lacto-N-neotetraose, lacto-N-tetraose, Gal␤1,3GlcNAc, and lactose (see Fig. 3 and Table I). ST6Gal I also exhibited relatively high activity toward some glycoproteins. In addition, although low, ST6Gal I exhibited activity toward a glycolipid, paragloboside. These indicate that the substrate specificity of ST6Gal II is more narrow than that of ST6Gal I.
Linkage Specificity of ST6Gal II-The linkages of the incorporated sialic acids were also examined. Gal␤1,4GlcNAc was sialylated by ST6Gal II, and it was treated with linkage-specific exosialidases (Fig. 4A). The incorporated [ 14 C]NeuAc was resistant to treatment with ␣2,3-specific exosialidase (NANase I), but it was digested by ␣2,3and ␣2,6-specific exosialidase (NANase II). The sialylated product generated in this experiment comigrated with 6Ј-sialyl-N-acetyllactosamine on TLC (data not shown), and it was also resistant to treatment with ␤-galactosidase (Fig. 4B), suggesting that incorporated [ 14 C]NeuAc binds to galactose, but not N-acetylglucosamine, through ␣2,6-linkage. Moreover, these resulting patterns of ST6Gal II were virtually identical to those of ST6Gal I (Fig. 4,  A and B). When asialofetuin was sialylated by ST6Gal II, it was detected by Sambucus nigra lectin blotting (data not shown), which recognizes Sia␣2,6Gal␤1,4GlcNAc structure. These results indicated that ST6Gal II transfers sialic acid on to galactose of Gal␤1,4GlcNAc structure through ␣2,6-linkage, and the cloned sialyltransferase certainly belongs to the ST6Gal family.
Expression of the ST6Gal II Gene in Human Tissues and Tumors-The expression levels of the ST6Gal II gene in various tissues were too low to be detected by Northern blotting. Thus we performed semiquantitative RT-PCR to examine the expression of the ST6Gal II gene in various tissues and tumors (Fig. 5). The expression of the ST6Gal II gene was significantly detected by RT-PCR in small intestine, colon, and fetal brain. But the expression levels of the ST6Gal II gene were very low in other tissues examined when compared with those of the ST6Gal I gene. It should be also noted that the ST6Gal I gene was expressed in all tumors examined, but no expression was observed for the ST6Gal II gene in these tumors. This suggests that the elevation of Sia␣2,6Gal␤1,4GlcNAc structures in these tumors can be attributed to the ST6Gal I activity but not the ST6Gal II activity.
Genomic Organization of the ST6Gal II Gene-To know the genetic and evolutional relation of the ST6Gal II gene with other sialyltransferase genes, we analyzed the genomic organization of the ST6Gal II gene by database search. The genomic sequences containing the ST6Gal II gene (GenBank TM accession numbers AC005040, AC016994, and AC108049) were obtained and analyzed by the BLAST search of the human genome database using ST6Gal II-related cDNA sequences (GenBank TM accession numbers AB059555, BE613250, AB058780, and XM_038616) as queries. A schematic representation of the most probable genomic structure of the ST6Gal II gene is shown in Fig. 6. We found that the ST6Gal II gene is located on chromosome 2 (2q11.2-q12.1), and it spans over 85 kb of human genomic DNA consisting of at least eight exons (see Table II and Fig. 6A). The existence of exons 1a and 1b encoding different 5Ј-untranslated regions was suggested by sequence analysis of some ST6Gal II-related clones. The COOH-terminal region of the active form of ST6Gal II is encoded by exon 6b, whereas that of the short inactive form is encoded by exon 6a. The sequences of the splice junctions of the ST6Gal II gene obey the GT-AG rule (35) (Fig. 6B). Some amino acid residues in the exon/intron boundaries of the ST6Gal I and II genes are highly conserved (Fig. 6B). In our previous study, we found that codons for Arg in the sialyl motif L are highly conserved as a splice junction among many mouse sialyltransferase genes. Exceptionally, the codon for Asp in the sialyl motif L is a splice junction of the mouse ST6Gal I gene (36) (Fig.   FIG. 3. Incorporation of sialic acids into various oligosaccharides by human ST6Gal I and II. The various oligosaccharides (10 g/lane) indicated were incubated with ST6Gal I or II, and the resulting oligosaccharides were analyzed by HPTLC with a solvent system of 1-propanol, aqueous ammonia, and water (6:1:2.5). 7). We also found that the codon for Asp in the sialyl motif L is a splice junction of the human ST6Gal II gene. In addition, the split patterns of the coding sequences for sialyl motif S of the ST6Gal I and II genes are different from the ST3Gal I and II genes and ST6GalNAc I and II genes (Fig. 7). Comparison of the exon/intron boundaries and exon sizes suggests that the ST6Gal I and II genes have a similar genomic structure (37) (see Fig. 6 and Fig. 8).

DISCUSSION
So far, ST6Gal I has been known as the sole member of the ␤-galactoside ␣2,6-sialyltransferase family for more than ten years. However, the existence of other ␤-galactoside ␣2,6-sialyltransferases that have different substrate specificities or preferences from ST6Gal I has been expected (34). With the progress of the human genome project, the extensive database search enabled us to detect the second type of ␤-galactoside ␣2,6-sialyltransferase (ST6Gal II).
As shown in this study, ST6Gal II exhibited activity toward oligosaccharides containing Gal␤1,4GlcNAc structure at the nonreducing end of their carbohydrate groups, but it exhibited weak or no activity toward glycoproteins and glycolipids, respectively. We also examined the in vivo activity of ST6Gal II by transfecting the expression vector of full-length ST6Gal II cDNA (pRc/CMV-ST6Gal II) into several kinds of cultured cells. However, significant changes were not observed in the sialylation pattern of glycoproteins in these cells analyzed by S. nigra lectin blotting (data not shown). Together with the in vitro substrate preference of ST6Gal II, it is most likely that in vivo substrates of ST6Gal II are oligosaccharides containing Gal␤1,4GlcNAc structure at the nonreducing end, although there remains a possibility that some glycoproteins and/or glycolipids may be sialylated specifically by ST6Gal II. The ST6Gal I knock-out mice exhibited great loss of cell surface Sia␣2,6Gal␤1,4GlcNAc structures and hallmarks of severe immunosuppression (22). These indicate that ST6Gal II are not involved in the production of cell surface Sia␣2, 6Gal␤1,4GlcNAc structures and cannot compensate for the ST6Gal I activity in immune system (the existence of mouse ST6Gal II has been suggested by some EST clones; GenBank TM accession numbers BB552328, BB633550, BB651169, and BB666153). Therefore, it can be said that ST6Gal II is an oligosaccharide-specific enzyme compared with ST6Gal I, which exhibits broad substrate specificities toward glycoproteins, glycolipids, and oligosaccharides. Although the main substrate of ST6Gal I in vivo has been considered as glycoproteins, it is also likely that ST6Gal I is significantly involved in the synthesis of sialyloligosaccharides in some tissues. Our in vitro analysis showed that ST6Gal I can sialylate not only Gal␤1,4GlcNAc and lacto-N-neotetraose but also Gal␤1, 3GlcNAc, lactose and lacto-N-tetraose. On the other hand, ST6Gal II cannot sialylate Gal␤1,3GlcNAc, lactose and lacto-N-tetraose (see Fig. 3 and Table I). This suggests that some kinds of sialyloligosaccharides are produced by ST6Gal I only.
The biological importance of sialyloligosaccharides produced by ST6Gal II is unclear at present, but expression of the ST6Gal II gene seems to be regulated developmentally or tissue-specifically (Fig. 5), suggesting that sialyloligosaccharides produced by ST6Gal II may play important roles in various biological phenomena. Sialyloligosaccharides are considered to play important roles in physiological functions in infancy, such as growth and development (38). Therefore, it may be possible that sialyloligosaccharides produced by ST6Gal II in the fetal brain are involved in brain development or function.
It has been reported that some sialyloligosaccharides in human milk have growth-promoting effects on bifidobacteria and lactobacilli present in the intestinal flora (38). The predomi-   (38). Fluid accumulation of cholera toxin-induced diarrhea in rabbit intestine was also clearly reduced by the presence of sialyllactose (38). Many of the Sia-binding pathogens exhibit a preference for the ␣2,3-sialyl linkage (39), but it is considered that compounds containing an ␣2,6-sialyl linkage may act as decoys or smoke screens to foil potential pathogens (40). Therefore, it may be possible that sialyloligosaccharides produced by ST6Gal II in small intestine and colon contribute to the maintenance of the intestinal flora and protection against enteric infections.
So far, genomic structures and chromosomal localization of 18 human sialyltransferase genes have been analyzed (36,41,42). We have also performed an extensive database search to obtain more detailed information on genomic organization of 20 human sialyltransferase genes in this study. The results are summarized in Fig. 8 and Table II. Genomic structural analysis of the ST6Gal II gene revealed that this gene has a similar genomic structure with the ST6Gal I gene, suggesting that these genes share a common ancestral gene. The split pattern of the coding sequences for sialyl motifs L and S of these genes are different from other sialyltransferase genes (Fig. 7), also suggesting that the ST6Gal I and II genes may have evolved independently or differently from the most ancestral sialyltransferase gene. Besides the ST6Gal I and II genes, there are several sets of sialyltransferase genes that encode similar enzymes and share similar genomic structures. Among them, ST6GalNAc I and II genes, ST6GalNAc III and V genes, and ST6GalNAc IV and VI genes are located close to each other on chromosomes 17, 1, and 9, respectively (Table II), suggesting that each gene pair is closely related from an evolutional standpoint. Probably each gene pair has arisen from a common ancestral gene by tandem duplication. It should be noted that the genome sizes of each gene pair are also similar to each other (Fig. 8). On the other hand, other pairs of similar sialyltransferase genes, such as the ST6Gal I and II genes, are not located on the same chromosome. This suggests that these genes have arisen from a common ancestral gene by gene duplication and subsequently dispersed in the human genome by translocation. We found by database search that besides the functional sialyltransferase genes, there are significant amounts of sialyltransferase gene-related nonfunctional DNA sequences, such as pseudo genes and partial fragments of sialyltransferase genes, in the human genome. The existence of these remnants of sialyltransferase genes also suggests that dynamic events of the human genome have contributed to the evolution of sialyltransferase genes.
The human ST6Gal I gene is expressed ubiquitously, and its expression levels are much higher than those of the ST6Gal II gene (Fig. 5). It has been known that the expression of the ST6Gal I gene is regulated by physically distinct multiple promoters in a tissue-and stage-specific manner (37,43). In most cases, resultant ST6Gal I transcripts differ in the 5Јuntranslated regions but encode the same protein. Multiple mRNA isoforms that differ only in the 5Ј-untranslated regions have been also identified in human ST3Gal IV-VI (44 -46). These transcripts are produced from a single gene locus by a combination of alternative splicing and alternative promoter usage in a tissue-and stage-specific manner. We found by database search and genomic structural analysis of the ST6Gal II gene that there are some isoforms of the ST6Gal II mRNA that differ in the 5Ј-untranslated region and/or the regions encoding the COOH terminus of the protein and the 3Ј-untranslated region. These suggest that the ST6Gal II transcripts are also produced by a combination of alternative splicing and alternative promoter usage in a tissue-and stage-specific manner. Although the expression levels of the ST6Gal II gene are relatively low, above transcriptional regulation may contribute to the specific expression of the ST6Gal II gene. It should be noted that besides the transcripts encoding the active form of ST6Gal II, there are other transcripts encoding the short inactive form of ST6Gal II lacking most of the sialyl motif S. At present, we do not know the biological importance and function of the short form of ST6Gal II. However, it may be possible that the short form of ST6Gal II acts like a lectin and is involved in some interaction events, because it should be able to bind sialic acids through the sialyl motif L. The detailed analysis of transcriptional regulation of the ST6Gal II gene will help elucidate biological significance of each transcript.
The mammalian sialyltransferase family is supposed to consist of more than 20 sialyltransferases. It is interesting that all the members of so-far cloned sialyltransferases have the counterpart with similar enzymatic properties and genomic structure. The biological significance of these multiple genes is unclear at present. One interpretation is that they may be important for fine control of the expression of sialylglycoconjugates, resulting in a variety of developmental stage-specific and tissue-specific glycosylation patterns. All the members of the sialyltransferase family will be identified by the genome project in the near future. Characterization of each sialyltransferase and analysis of the transcriptional regulation of each gene will help elucidate the biological significance of each sialyltransferase and the sialylglycoconjugates they produce.