Molecular Cloning and Characterization of a Novel UDP-GlcNAc:GalNAc-peptide β1,3-N-Acetylglucosaminyltransferase (β3Gn-T6), an Enzyme Synthesizing the Core 3 Structure of O-Glycans*

The core 3 structure of theO-glycan, GlcNAcβ1–3GalNAcα1-serine/threonine, an important precursor in the biosynthesis of mucin-type glycoproteins, is synthesized by UDP-N-acetylglucosamine:GalNAc-peptide β1,3-N- acetylglucosaminyltransferase (β3Gn-T; core 3 synthase). The core 3 structure is restricted in its occurrence to mucins from specific tissues such as the stomach, small intestine, and colon. A partial sequence encoding a novel member of the human β3Gn-T family was found in one of the data bases. We cloned a complementary DNA of this gene and named it β3Gn-T6. The putative amino acid sequence of β3Gn-T6 retains the β3Gn-T motifs and is predicted to comprise a typical type II membrane protein. The soluble form of β3Gn-T6 expressed in insect cells showed β3Gn-T activity toward GalNAcα-p-nitrophenyl and GalNAcα1-serine/threonine. The β1,3-linkage between GlcNAc and GalNAc of the enzyme reaction product was confirmed by high performance liquid chromatography and NMR analyses. β3Gn-T6 effectively transferred a GlcNAc to the GalNAc residue on MUC1 mucin, resulting in the synthesis of a core 3 structure. Real time PCR analysis revealed that the β3Gn-T6 transcript was restricted in its distribution, mainly to the stomach, colon, and small intestine. We concluded that β3Gn-T6 is the most logical candidate for the core 3 synthase, which plays an important role in the synthesis of mucin-type O-glycans in digestive organs.

We expected the core 3 synthase to be a member of the ␤3GT family because it catalyzes the synthesis of GlcNAc␤1-3GalNAc. None of the ␤3GT, which we isolated, showed apparent activity to produce core 3. Therefore, we searched the data bases to find novel genes having the ␤3GT motifs, and cloned and characterized a sixth member of family (␤3Gn-T6). We identified ␤3Gn-T6 as the most logical candidate for the core 3 synthase.

EXPERIMENTAL PROCEDURES
Isolation of Human ␤3Gn-T6 cDNA-We performed a BLAST search of the expressed sequence tag data bases and identified two cDNAs (AW182889 and AW192172), homologous in amino acid sequence to the open reading frame (ORF) of ␤3Gn-T3 (25). On searching the human genomic DNA data base, we found that the two cDNAs mapped in the immediate vicinity of each other in a single contig (AP00752), having only a 26-bp gap. However, the 5Ј-region of this novel gene, which should be homologous to that of ␤3Gn-T3, was not present in the contig. To obtain this sequence, the 5Ј-rapid amplification of cDNA ends method was employed using a Marathon-Ready™ cDNA Amplification Kit (CLONTECH, Palo Alto, CA). Two reverse primers were designed for the first PCR, 5Ј-CTCCAGACACATGCCCATGTAGGC-3Ј, and for the nested PCR, 5Ј-GCCAGTCGAGCAAGTGCAGG-3Ј. A DNA fragment obtained by the 5Ј-rapid amplification of cDNA ends method was sequenced using an ABI PRISM BigDye™ Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems, Foster City, CA). The sequenced DNA fragment contained a part of the 5Ј-primer sequence of the cDNA (AW182889), a putative initiation codon, and a transmembrane domain. Finally, a cDNA encoding the full-length ORF of a novel family member was obtained by PCR using the Marathon-Ready™ cDNA of human stomach tissue (CLONTECH) as a template and subcloned into pDONR™201 vector (Invitrogen).

Acceptor substrate
Relative activity Biosynthetic pathways for the core structures of mucin type O-glycans. Cancer-associated antigens are shown in parentheses.
brane was probed with anti-FLAG M2 monoclonal antibody-conjugated peroxidase and stained with the Konica Immunostaining HRP-1000 (Konica, Tokyo, Japan). Each purified ␤3Gn-T was adjusted to the same amount according to the result of Western blotting.
Assay Conditions for Gn-T Activity-The basic reaction mixture for assaying Gn-T activity contained 50 mM sodium cacodylate buffer, pH 7.2, an appropriate concentration of UDP-GlcNAc, 10 mM MnCl 2 , 0.4% Triton CF-54, a suitable amount of acceptor substrate, and the purified enzyme. After incubation at 37°C for 16 h, the product was analyzed by various techniques.  Table I, were purchased from Calbiochem, Toronto Research Chemicals Inc. (Ontario, Canada), TaKaRa, and Sigma. Radioactive products were separated from the free UDP-[ 14 C]GlcNAc using a Sep-Pak Plus C 18 cartridge (Waters, Milford, MA) as described below. The cartridge was activated by washing with 10 ml of 100% methanol and then washed twice with 10 ml of water. The enzyme reaction was terminated by the addition of 200 l of water, and then the reaction mixture was applied to the equilibrated cartridge and washed twice with 10 ml of water. Elution of the radioactive product was achieved using 1 ml of 100% methanol. Eluted products were dried with a N 2 evaporator and dissolved in 20 l of 100% methanol. They were separated on a HPTLC plate (Merck) with a solvent system of chloroform, methanol, and 0.2% CaCl 2 (55:45:10, v/v/v). The radioactive intensities of the bands were measured with a FLA-3000 Imaging Analyzer (Fujifilm, Tokyo).

Assay of Gn-T Activity Using High Performance Thin Layer Chromatograpy (HPTLC)-The
Determination of a ␤1,3-Linkage in the Enzyme Reaction Product by HPLC and NMR-1 mM GalNAc␣-pNP was incubated with ␤3Gn-T6 in 20 l of a basic reaction mixture containing 2 mM UDP-GlcNAc to produce the reaction product. A 10-l aliquot of supernatant was subjected to HPLC on an ODS-80Ts QA column (4.6 ϫ 250 mm; Tosoh, Tokyo). The reaction products were eluted with 30 ml of 12% acetonitrile containing 0.1% trifluoroacetic acid and H 2 O at a flow rate of 1.0 ml/min at 50°C and monitored with an ultraviolet spectrophotometer (absorbance at 210 nm), SPD-10A VP (Shimadzu, Kyoto, Japan).
Glycosylation of GalNAc␣1-(N-trifluoracetyl)-L-threonine methyl ester was performed in 200 l of a basic reaction mixture containing 10 mM UDP-GlcNAc and 5 mM threonine derivative. The enzyme reaction product was separated by HPLC as described above, the deuterium exchanged by subsequent lyophilization from D 2 O and dissolved again in 0.5 ml of D 2 O for NMR analysis. One-dimensional 1 H and 13 C NMR as well as two-dimensional 1 H-1 H COSY, 1 H-1 H TOCSY, 1 H NOESY, and 1 H-13 C HMQC NMR experiments were performed at 298 K on a Bruker DMX-500, DMX-750, and JEOL ECP800 spectrometer. Chemical shifts are referenced to internal acetone (2.04 ppm for 1 H and 29.8 ppm for 13 C).
Transfer of GlcNAc by ␤3Gn-T6 on GalNAc-peptide and Native Glycoproteins-A fluorescein isothiocyanate (FITC)-labeled oligopeptide,  FITC-AHGVTSAPDTR, prepared commercially was purchased from Sawady Co., Ltd. (Tokyo). The peptide sequence was identical to that of Muc1aЈ used as acceptor substrate for polypeptide-GalNAc-Ts (30). In the present study, we cloned polypeptide-GalNAc-T6, one of the polypeptide-GalNAc-Ts, which had been reported previously (31), by reverse transcriptase PCR. We expressed it in the baculovirus expression system, as described in the previous section, to obtain a recombinant form. Muc1aЈ was glycosylated by use of GalNAc-T6, and the glycosylated peptide was isolated by HPLC (Waters 5C 18 -AR; 4.6 ϫ 250 mm; 1.0 ml/min; gradient 0 -50% acetonitrile in water containing 0.05% trifluoroacetic acid; fluorescence detection ex 492 nm, em 520 nm). The isolated fractions were subjected to MALDI-TOF mass spectrometry using Reflex TM III (Bruker Daltonics, Tsukuba, Japan) and protein sequencing analysis (Protein Sequencer PPSQ-23A, Shimadzu). Mass spectra were calibrated externally unless otherwise stated. The samples were dissolved in 0.1% trifluoroacetic acid to a concentration of 2 pmol/l and prepared by mixing 1 l of sample solution with 1 l of matrix solution (2,5-dihydroxybenzoic acid, 10 mg/ml in 30% acetonitrile containing 0.1% trifluoroacetic acid) directly on the target. The glycosylated peptide was determined to have a GalNAc residue and the sequence of FITC-AHGVT(-GalNAc) SAPDTR. This GalNAcglycosylated peptide was used as an acceptor for ␤3Gn-T6. Quantitative conversion of the glycopeptide was achieved after a 16-h incubation with ␤3Gn-T6 at 37°C in 750 l of a basic reaction mixture containing 5 mM UDP-GlcNAc and 170 nM acceptor.
The transfer of GlcNAc by ␤3Gn-T6 to glycoproteins was performed in 20 l of a basic reaction mixture containing 500 M UDP-GlcNAc, 24 mM (175 nCi) UDP-[ 14 C]GlcNAc, and 0.5 g of bovine submaxillary gland mucin (BSM), 5 g of fetuin, or 5 g of asialofetuin. After incubation at 37°C for 16 h, each 10 l of the reaction mixture was sub-jected to 10% SDS-PAGE. The radioactive intensities of the bands were measured with a FLA-3000 Imaging Analyzer (Fujifilm).
Quantitative Analysis of the ␤3Gn-T6 Transcript in Human Tissues by Real Time PCR-In the preliminary Northern blot analysis using commercially available human 12-Lane MNN™ Blot and human MTN ® Blot III membranes (CLONTECH), we could not detect any positive bands. Therefore, we employed the real time PCR method, as described in detail previously (32,33), for quantification of the ␤3Gn-T6 transcripts because the amount of the transcripts was found to be very small even in mRNAs derived from stomach and colon. Marathon Ready ® cDNAs of various human tissues were purchased from CLONTECH. Standard curves for the ␤3Gn-T6 cDNA and the human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA, as an endogenous control, were generated by serial dilution of a pDONR TM 201 vector DNA containing the ␤3Gn-T6 gene encoding the putative catalytic domain (amino acids 44 -384), and a pCR2.1(Invitrogen) DNA containing the GAPDH gene. The primer set and probe for ␤3Gn-T6 were as follows: the forward primer, 5Ј-CCTGCATGTACCGCGAGTT-3Ј; the reverse primer, 5Ј-CCACATGAGCAGCATCTCGT-3Ј; and the probe, 5Ј-TGCT-AGTGCACCGCT-3Ј. For GAPDH, we used Pre-Developed TaqMan ® Assay Reagents Endogenous Human GAPDH (Applied Biosystems). Primers, probes, and cDNAs were added to the TaqMan Universal PCR Master Mix (Applied Biosystems) which contained all reagents for PCR. The PCR conditions included 1 cycle at 50°C for 2 min, 1 cycle at 95°C for 10 min, 50 cycles at 95°C for 15 s, and 60°C for 1 min. PCR products were measured continuously with an ABI PRISM 7700 Sequence Detection System (Applied Biosystems). The relative amount of ␤3Gn-T6 transcript was normalized to the amount of GAPDH transcript in the same cDNA.

Isolation of a New ␤3Gn-T-We obtained a cDNA of a new member of the ␤3GT family as described under "Experimental
Procedures" and named it ␤3Gn-T6. An alignment of the amino acid sequences of five ␤3Gn-Ts made using ClustalW is shown in Fig. 2. The 1155-bp (384 amino acids) ORF of ␤3Gn-T6 encoded a typical type II membrane protein, the same as in other ␤3Gn-Ts, consisting of an N-terminal cytoplasmic domain of 12 residues, a transmembrane segment of 19 residues, and a stem region and catalytic domain of 353 residues (Fig. 2). ␤3Gn-T6 showed 41, 54, 42, and 35% identity to ␤3Gn-T2, ␤3Gn-T3, ␤3Gn-T4, and ␤3Gn-T5, respectively, and the sequence similarity was limited to the putative catalytic domains. The three motifs shared by the members of the ␤3GT family (24) were present in the predicted positions in the putative amino acid sequence of ␤3Gn-T6. The five cysteines residues were conserved in the five ␤3Gn-Ts, indicating that some would be essential for maintaining the tertiary structure. A triplet of aspartic acid residues, DDD, which may be a divalent cation binding site as proposed in a crystallization study of other glycosyltransferases (34), was also conserved. Three possible N-glycosylation sites were found in the primary sequence of ␤3Gn-T6. One of them was conserved in all ␤3Gn-Ts.
The ␤3Gn-T6 gene was found to be localized to a draft genome sequence (GenBank accession no. AP00752) which was  Shown is a 1 H NMR spectrum between 3.33 and 4.86 ppm of the reaction product purified from the ␤3Gn-T6 reaction mixture with GalNAc␣1-threonine by HPLC (data not shown), using an ECP800 spectrometer. mapped to 11q14 on the human chromosome, and its ORF was composed of a single exon.
Substrate Specificity of ␤3Gn-T6 -The calculated molecular mass of the FLAG-tagged recombinant ␤3Gn-T6, is 43 kDa; however, a major band of 46.6 kDa was observed, but no band was detected around 43 kDa on Western blot analysis. This result suggested that the recombinant protein would be glycosylated in insect cells. The glycosyltransferase activities of ␤3Gn-T6 were examined by a HPTLC analysis of Gn-T and Gal-T activities and HPLC analysis of GalNAc-T activity. No Gal-T or GalNAc-T activity toward any acceptor substrate was observed (data not shown). The substrate specificities of ␤3Gn-T6 are summarized in Table I. The activity toward GalNAc␣-pNP was the strongest. Thus, the activity of ␤3Gn-T6 for Gal-NAc␣-pNP is presented as 100%, and all other activities are given as relative values in Table I. ␤3Gn-T6 exhibited 15% activity toward GalNAc␤-pNP; 2.4% activity for Gal␤1-4Glc-NAc␣-pNP; and no activity toward the other acceptors. ␤3Gn-T6 preferred GalNAc␣-pNP over GalNAc␤-pNP as an acceptor. This result is consistent with that of a previous study (35). ␤3Gn-T6 exhibited no activity toward the ␣-GalNAc residue of blood group A and the core 6 substrates (substrates 3 and 22 in Table I). It also showed no activity toward the ␤-GalNAc residue of gangliosides, asialo-GM2, GM2, and lyso-GM2 (substrates 4, 5, and 6 in Table I).
The amounts of each of the five ␤3Gn-Ts were made equal for assaying the relative activity to transfer GlcNAc to GalNAc␣-pNP (Table II). The ␤3Gn-T6 activity is presented as 100%. All other activities are given as relative values. The Gn-T activities of ␤3Gn-T6 and -T2 were detectable, but those of the others were not. Although ␤3Gn-T2 transferred GlcNAc to GalNAc␣-pNP, its relative activity was only 2.4%. In our previous study (25), ␤3Gn-T2 showed the strongest activity for the transfer of GlcNAc to the ␤Gal residue of polylactosamine. In the present study, ␤3Gn-T2 exhibited the strongest activity to transfer GlcNAc to Gal␤1-4GlcNAc␣-pNP at more than 500 times that to GalNAc␣-pNP (data not shown).
Using equal amounts of ␤3Gn-T6 and -T2, HPLC analyses were carried out to measure Gn-T activity for Cy5-labeled GalNAc␣1-serine. A new product was created by ␤3Gn-T6 and was eluted before the acceptor substrate (data not shown). In contrast, ␤3Gn-T2 did not create any new products. ␤3Gn-T6 also transferred GlcNAc to Cy5-labeled GalNAc␣1-threonine to an equal or greater extent than to Cy5-labeled GalNAc␣1serine (data not shown).
Linkage Analysis of the Product Formed by ␤3Gn-T6 -Two linkages between GlcNAc and GalNAc are known to occur in GlcNAc:GalNAc␣1-serine/threonine in roots of mucin-type glycoproteins. They are GlcNAc␤1-3GalNAc (core 3) and GlcNAc␤1-6GalNAc (core 6). To estimate the linkage structure synthesized by ␤3Gn-T6, the reaction product of ␤3Gn-T6 obtained using GalNAc␣-pNP as an acceptor was subjected to HPLC analysis (Fig. 3). Two commercially available compounds, GlcNAc␤1-3GalNAc␣-pNP (core 3-pNP) and Glc-NAc␤1-6GalNAc␣-pNP (core 6-pNP), were used as standards as seen in Fig. 3A. The peak of enzyme product apparently shifted from the original peak of substrate (peak c in Fig. 3) to the position of core 3-pNP (peaks a and d in Fig. 3). This result suggested that ␤3Gn-T6 synthesizes the core 3 structure from GalNAc␣-pNP.
Characterization of the Core 3 Product by NMR Spectroscopy-To characterize the structure generated by ␤3Gn-T6 more precisely, 1 H NMR spectroscopy was performed.
The reaction product of ␤3Gn-T6 with UDP-GlcNAc and GalNAc␣1-(N-trifluoracetyl)-L-threonine methyl ester was isolated by HPLC, and the newly formed glycosidic linkage was characterized by 1 H and 13 C NMR spectroscopy. All 1 H resonances could be differentiated and assigned by their coupling constants in their one-dimensional 1 H NMR spectrum, as well as two-dimensional 1 H-1 H COSY, 1 H-1 H TOCSY, and 1 H-13 C HMQC NMR experiments. For the 13 C resonances, data were compared with those published previously for the corresponding pNP glycosides (36). Additional resonances showed the successful transfer of GlcNAc to the acceptor substrate. Among those additional peaks, the ␤-1 resonance (4.513 ppm) with a coupling constant (J 1,2 ϭ 8.4 Hz) showed a typical value for 1,2-trans biaxial coupling or ␤-gluco-configurated pyranosides (Fig. 4). Comparison of the 13 C NMR spectra of the GalNAcspecific resonances of the product with the spectrum of the acceptor substrate showed a shift of the ␣-3 resonance from 68.18 to 76.77 ppm. Thus the NMR spectra unambiguously confirmed the formation of a ␤-(133)-linkage and supported that the putative ORF encodes for a glycosyltransferase of the ␤3-family. We also performed a 1 H-NOESY NMR experiment (Fig. 5), as it allows one to monitor nuclear Overhauser effect or NOE contacts between 1 H nuclear spins, which are at close distance (Ångstrom range) but not necessarily chemically linked. Of special interest is the contact between different moieties of the molecules, for example that between ␣-3 and ␤-1, which allows us to predict that both sugar planes show the same orientation. Other O-glycans show a free rotation along the glycopeptide linkage and therefore show no specific contact between the peptide and sugar moieties. In contrast, Gal-NAc1␣-threonine can form a hydrogen bond between 2-acetamido of GalNAc and the carbonyl of threonine. This hydrogen bond results in a fixed conformation like in the DTR motif of MUC1 or AAT repeat of AFGP. Our hydrophilic protected threonine derivative displayed a similar fixed conformation, and we observed NOE contact between the threonine-␥ methyl group and protons ␣-1 and ␣-5 of GalNAc.
Core 3 Synthesizing Activity for GalNAc-peptide and O-Glycosylated Proteins-The FITC-labeled GalNAc-Muc1aЈ peptide was used as a substrate for assaying the ␤3Gn-T6 activity for core 3 synthesis. As seen in Fig. 6, ␤3Gn-T6 produced a new peak of product (peak c in Fig. 6A) which shifted to a position with a shorter retention time than that of the substrate (peak b in Fig. 6A). Three peaks of FITC-Muc1aЈ, FITC-GalNAc-Muc1aЈ, and the reaction product were isolated and subjected to measurements of molecular weight by MALDI-TOF mass spectrometry. The molecular weight of the product was determined to be 1877, which matched that expected of the FITClabeled GlcNAc-GalNAc-Muc1aЈ peptide.
To determine the activity of ␤3Gn-T6 to transfer GlcNAc to O-glycosylated protein, BSM, fetuin, and asialofetuin were used as acceptor substrates. As shown in Fig. 7, ␤3Gn-T6 effectively transferred GlcNAc to BSM and asialofetuin, but not to fetuin.
Distribution of ␤3Gn-T6 Transcripts in Human Tissues-The level of ␤3Gn-T6 transcripts expressed in various human tissues was determined by the real time PCR method. As summarized in Fig. 8, the expression level of the ␤3Gn-T6 transcripts was highest in the stomach, followed by the colon and small intestine. Skeletal muscle and testis expressed the ␤3Gn-T6 transcript at a relatively low level. The expression levels in the remaining tissues were very low or undetectable. DISCUSSION Core 3 synthase is an important enzyme in the synthesis of mucin-type O-glycans in digestive organs. The core 3 structure was first identified in O-glycans derived from stomach and colon, where core 3 synthase activity was also very strong (35,37). In the present study, we have cloned, expressed, and characterized a novel ␤3Gn-T, which can create the core 3 structure on GalNAc-peptide.
As demonstrated in the present study, ␤3Gn-T3, -T4, and -T5 are not the core 3 synthase. Although ␤3Gn-T2, which is involved in the initiation and elongation of poly-N-acetyllactosamine synthesis (25,26), exhibited weak activity for core 3 synthesis, its expression is ubiquitous, whereas the core 3 structure and core 3 synthase activity have been found in specific tissues. In addition, ␤3Gn-T2 did not transfer GlcNAc to GalNAc␣1-serine. Therefore, ␤3Gn-T2 is probably not responsible for the synthesis of core 3 in vivo. Furthermore, the synthase activity is detected strongly in normal colon tissue but is almost undetectable in colon cancer cell lines (35); however, the transcript level of ␤3Gn-T2 in Colo205 cells was higher than that in normal colon tissue (25). These findings also suggested that ␤3Gn-T2 is not the core 3 synthase. ␤3Gn-T6 transferred GlcNAc not only to GalNAc␣-pNP but also to GalNAc␤-pNP and Gal␤1-4GlcNAc␣-pNP. The activity in the latter two cases was very weak compared with that in the former, so it may not reflect the physiological activity of ␤3Gn-T6 in vivo. In addition to GalNAc␣-pNP, GalNAc␣1-serine and GalNAc␣1threonine were also good acceptors for ␤3Gn-T6.
The HPLC analysis (Fig. 3) strongly suggested that the reaction product of ␤3Gn-T6 is the core 3-pNP. We confirmed by NMR analysis that the reaction product derived from Gal-NAc␣1-threonine by ␤3Gn-T6 formed a ␤1,3-linkage. Finally, the core 3 synthesizing activity on proteins was confirmed using FITC-GalNAc-Muc1aЈ and native glycoproteins as acceptors. ␤3Gn-T6 effectively transferred GlcNAc to BSM and asialofetuin but not to fetuin. This suggested that it transferred GlcNAc to GalNAc of the Tn epitopes on asialofetuin but not to fetuin because the Tn epitopes are already masked by sialylation. This is of interest because it indicates that the synthesis of core 3 probably competes with that of sialyl-Tn (sTn). In a previous study, we cloned and characterized ST6GalNAc I (sTn synthase), a human sTn synthase (38). It effectively trans- ferred a sialic acid to the ␣-GalNAc residue of asialofetuin with an ␣1,6-linkage, resulting in the synthesis of sTn antigens (38). We will examine whether or not the two enzymes, ST6GalNAc I and ␤3Gn-T6, share or compete for the acceptor substrates, the Tn epitopes on mucins, to form sTn or core 3, respectively, in cells.
The expression of ␤3Gn-T6 transcripts was limited to specific tissues, such as the stomach, colon, and small intestine, consistent with the tissue distribution of core 3-containing structures and core 3 synthesizing activity reported by others (10,35). The expression level of this gene was especially high in the stomach, and moderate in the colon and small intestine. However, in stomach mucin, core 1-and core 2-based O-glycans appear to be the major components, whereas core 3-based Oglycans appear to be the major form in colonic mucin. The synthesis of the core 1 structure directed by core 1 ␤3Gal-T may compete with the synthesis of the core 3 structure directed by core 3 ␤3Gn-T. The expression level of ␤3Gn-T6 transcript does not necessarily correlate with the total amount of core 3-containing O-glycans. Very recently, the gene encoding the core 1 synthase, core 1 ␤3Gal-T, was cloned, and its expression was demonstrated in many tissues (39,40). The expression of core 1 ␤3Gal-T was somehow ubiquitous in the tissues examined; however, levels differed among the tissues, being relatively low in colon and small intestine. Although the stomach was not examined, we speculate that the expression level of core 1 ␤3Gal-T is much higher than that of core 3 ␤3Gn-T in stomach.
As seen in Fig. 1, certain core structures, namely Tn, sTn, and T (core 1) antigens, are known to be associated with cancer. Core 2 is also important for the expression of sialyl Lewis x (sLe x ) epitope, a well known cancer-associated antigen (41)(42)(43). Core 2 formation leads to the extension of the core 2 branch on which the sLe x epitope frequently occurs in cancer cells (44,45). Core 3 formation may compete not only with the core 1 expression but also against the expression of these cancerassociated antigens. We plan to perform transfection experiments with the ␤3Gn-T6 gene in cancer cells to determine whether this is true or not. All of the genes encoding the enzymes involved in the synthesis of Tn, sTn, T, and core 2 structures have already been cloned by us or others (31, 38, 40, 46 -50). It is now possible to perform experiments using these genes to analyze the competitive synthesis among antigens.
It has been reported that core 3 ␤3Gn-T activity was reduced to an undetectable level in colonic cancerous tissues (51) and is not detected in many colonic cancer cell lines (35). In our preliminary experiment, we measured the amount of ␤3Gn-T6 transcript in colonic cancer tissues and many colonic cancer cell lines and found a dramatic down-regulation of ␤3Gn-T6 transcription. These findings also support that ␤3Gn-T6 is the most logical candidate for the core 3 synthase. FIG. 8. Quantitative analysis of the ␤3Gn-T6 transcript in human tissues by real time PCR. Standard curves for ␤3Gn-T6 and GAPDH were generated by serial dilution of each plasmid DNA. The expression level of the ␤3Gn-T6 transcript was normalized to that of the GAPDH transcript, which was measured in the same cDNAs. Data were obtained from triplicate experiments and are indicated as the mean Ϯ S.D.