J Biol Chem, Vol. 275, Issue 15, 11106-11113, April 14, 2000

Control of O-Glycan Branch Formation
MOLECULAR CLONING AND CHARACTERIZATION OF A NOVEL THYMUS-ASSOCIATED CORE 2 1,6-N-ACETYLGLUCOSAMINYLTRANSFERASE^*

Tilo Schwientek^§, Jiunn-Chern Yeh^¶, Steven B. Levery, Birgit Keck, Gerard Merkx^**, Ad Geurts van Kessel^**, Minoru Fukuda^¶, and Henrik Clausen

From the  School of Dentistry, University of Copenhagen, Nørre Allé 20, 2200 Copenhagen N, Denmark, the ^¶ Glycobiology Program, Cancer Research Center, The Burnham Institute, La Jolla, California 92037, the  University of Georgia, Complex Carbohydrate Research Center, Athens, Georgia 30602, and the ^** Department of Human Genetics, University Hospital Nijmegen, 6500 HB Nijmegen, The Netherlands

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Core 2 O-glycan branching catalyzed by UDP-N-acetyl--D-glucosamine: acceptor 1,6-N-acetylglucosaminyltransferases (6GlcNAc-Ts) is an important step in mucin-type biosynthesis. Core 2 complex-type O-glycans are involved in selectin-mediated adhesion events, and O-glycan branching appears to be highly regulated. Two homologous 6GlcNAc-Ts functioning in O-glycan branching have previously been characterized, and here we report a third homologous 6GlcNAc-T designated C2GnT3. C2GnT3 was identified by BLAST analysis of human genome survey sequences. The catalytic activity of C2GnT3 was evaluated by in vitro analysis of a secreted form of the protein expressed in insect cells. The results revealed exclusive core 2 6GlcNAc-T activity. The product formed with core 1-para-nitrophenyl was confirmed by ¹H NMR to be core 2-para-nitrophenyl. In vivo analysis of the function of C2GnT3 by coexpression of leukosialin (CD43) and a full coding construct of C2GnT3 in Chinese hamster ovary cells confirmed the core 2 activity and failed to reveal I activity. The C2GnT3 gene was located to 5q12, and the coding region was contained in a single exon. Northern analysis revealed selectively high levels of a 5.5-kilobase C2GnT3 transcript in thymus with only low levels in other organs. The unique expression pattern of C2GnT3 suggests that this enzyme serves a specific function different from other members of the 6GlcNAc-T gene family.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Control of mucin-type O-glycosylation involves an initiation step followed by a processing step. The initiation step is complex and is carried out by a large family of homologous UDP-GalNAc:polypeptide GalNAc-transferases (1). The polypeptide GalNAc-transferase isoforms have distinct enzymatic properties and are differentially expressed, thus presumably allowing for a high level of control in determining sites of O-glycan attachments in proteins. The processing step involves elongation, branching, and terminal modification of the O-glycans (2). It is apparent that essential steps in O-glycan elongation and branching are catalyzed by multiple glycosyltransferase isoforms from families of homologous glycosyltransferases. The main biosynthetic pathway utilized to synthesize complex-type O-glycans is through core 2 branching. Two UDP-GlcNAc:Gal1-3GalNAc 6GlcNAc-transferases, C2GnT1¹ and C2GnT2, have been identified and cloned thus far (3-5). Both function in vitro and in vivo as core 2 synthases, but C2GnT2 can also make the core 4 structure (4, 5). The existence of multiple core 2 and related 6GlcNAc-T² activities were predicted by analyses of enzyme activities in cells and organs (2, 6, 7). Elongation of core 2 O-glycans with poly-N-acetyllactosamine repeats (Gal1-4GlcNAc1-3)_n may be carried out by alternative action of multiple 4-galactosyltransferases and 3GlcNAc-Ts (8-11). In the case of 4-galactosyltransferases one isoform, 4Gal-T4, appears to have superior kinetic properties for catalysis of the core 2 O-glycan poly-N-acetyllactosamine elongation (9). Two 3GlcNAc-Ts elongate core 2 O-glycans, although comparative analyses have not been performed (10, 12). Although the catalytic functions of these glycosyltransferase isoforms are not completely understood, the available data support the hypothesis that individual enzyme isoforms display unique kinetic properties and specific catalytic activities with regard to glycan structure and types of glycoconjugate. This is further supported by the finding that glycosyltransferase isoforms have different tissue expression patterns.

Core 2 O-glycan branching is a key step in mucin-type O-glycosylation (2). Core 2 O-glycans such as sialyl-Le^x and associated structures serve as ligands for selectin- and galectin-mediated cell-cell adhesion events that play important roles in T-cell development (13), lymphocyte trafficking (14, 15), the inflammatory process (16, 17), and cancer metastasis (18). Some evidence indicates that a major control point for synthesis of these ligands is the core 2 branching event catalyzed by 6GlcNAc-T activities (15, 19). Marked changes in core 2 branching and 6GlcNAc-T activities are associated with T-cell maturation and malignant transformation (18, 20, 21). Multiple core 2 6GlcNAc-T isoforms exist; hence, it is important to identify and characterize these to understand the molecular genetic basis for their differential regulation and activity. The first core 2 6GlcNAc-T identified by transfection cloning, C2GnT1, is widely expressed, functions only in core 2 synthesis (3, 5), and resembles the leukocyte 6GlcNAc-T activity (also designated C2GnT-L) identified by Williams and Schachter (22). The second core 2 6GlcNAc-T, C2GnT2, was cloned by an EST cloning strategy and has a broader acceptor specificity (4, 5). C2GnT2 resembles the mucous 6GlcNAc-T activity (designated C2GnT-M) (23, 24) and is mainly expressed in mucous-secreting organs (4, 5). C2GnT1 was predicted to control synthesis of core 2 selectin ligands in leukocytes and lymphoid tissues; however, mice deficient in C2GnT1 exhibited partial reduction in selectin ligand production, and there were no significant changes in lymphocyte homing properties (17). One possible explanation for these results would be the existence of additional core 2 6GlcNAc-Ts. C2GnT2 does not appear to be a candidate for this gene as its expression pattern is restricted and mainly associated with mucous epithelia (4, 5, 17).

In the present report we describe the cloning and characterization of a third homologous core 2 6GlcNAc-T designated C2GnT3. C2GnT3 exhibits exclusive core 2 acceptor specificity, and it shows a unique tissue distribution with high expression found only in thymus. C2GnT3 is predicted to play an important role in T-cell development and lymphocyte homing.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Cloning and Sequencing of C2GnT3-- tBLASTn analysis with the coding sequence of human C2GnT2 (previously reported as C2/4GnT (4) and C2GnT-M (5)) was used to search the genome survey sequence data base at The National Center for Biotechnology Information. Two human genome survey sequences with similarity were identified. Clone CIT-HSP-2288B17 (GenBank^TM accession number AQ005888, obtained from Research Genetics Inc.) contained a novel partial open reading frame with significant sequence similarity to C2GnTs. The coding regions of human C2GnT1 and C2GnT2 were previously found to be organized in one exon (4, 25). The putative coding sequence identified in the genome survey sequence data base was incomplete but likely to be located in a single exon as well. The additional 3'-sequence of the open reading frame was therefore obtained by sequencing a genomic P1 clone (844/B1) obtained from a human foreskin genomic P1 library (DuPont Merck Pharmaceutical Co. Human Foreskin Fibroblast P1 Library) by screening with the primer pair TSHC96 (5'-GGTTTCACCGTCTCCAACATAT-3') and TSHC101 (5'-TCGTAAGGCACCTGATACCTTC-3'). An open reading frame of 1359 base pairs was represented in the clone and sequenced in full. Confirmatory sequencing was performed on a cDNA clone obtained by PCR (35 cycles at 95 °C for 10 s; 55 °C for 15 s, and 68 °C for 2 min 30 s) on reverse-transcribed human thymus poly(A)⁺ RNA (CLONTECH) with the sense primer TSHC99 (5'-CGAGGATCCAGAATGAAGATATTCAAATGTTA-3') and the antisense primer TSHC121 (5'-AGCGAATTCTTACTATCATGATGTGGTAGTG-3') (Fig. 1). The sequence of the 5' end of C2GnT3 mRNA, including the translational start site and 5'-UTR, was obtained by PCR (35 cycles at 95 °C for 15 s; 55 °C for 10 s, and 72 °C for 2 min 30 s) with the sense primer TSHC131 (5'-TGCGAATTCAATGCACTGTTCAAGGATG-3') and the antisense primer TSHC113 (5'-TGCGAATTCCTGTTCATAGATACCCGAACAG-3'). The sequence of the 3'-untranslated region including the translational stop site was obtained by PCR (35 cycles at 95 °C for 15 s; 55 °C for 10 s, and 72 °C for 2 min 30 s) with the sense primer TSHC115 (5'-TGCGAATTCAGCCCAGGATGTGTCTGATCTGC-3') and the antisense primer TSHC134 (5'-AGCGAATTCATTGGGTGGTTTTCTCAAGTG-3').

Expression of C2GnT3 in Insect Cells-- An expression construct designed to encode amino acid residues 39-453 of C2GnT3 (secreted form) was prepared by PCR using P1 DNA and the primer pair TSHC100 (5'-CGAGGATCCGCAAAAAGACATTTACTTGGTT-3') and TSHC121 with BamHI and EcoRI restriction sites, respectively (Fig. 1). The PCR product was cloned directionally between the BamHI and EcoRI site of pAcGP67A (PharMingen) and fully sequenced. Plasmid pAcGP67-C2GnT3-sol was cotransfected with Baculo-Gold^TM DNA (PharMingen) and recombinant baculovirus obtained after two successive amplifications in Sf9 cells, as described previously (26). Controls included pAcGP67-C2GnT2-sol and pAcGP67-C2GnT1-sol (4). The kinetic properties were determined with partially purified enzymes expressed in High Five^TM cells. Partial purification was performed by consecutive chromatography on Amberlite IRA-95, DEAE-Sephacryl, and SP-Sepharose essentially as described (27). Protein concentrations were determined using the Bio-Rad reagent with bovine serum albumin as standard protein.

Expression of C2GnT3 in CHO Cells-- A construct designed for expression of the full coding sequence of C2GnT3 in CHO cells was prepared by PCR using P1 DNA and the primer pair TSHC99 and TSHC121 with BamHI and EcoRI restriction sites, respectively (Fig. 1). The PCR product was cloned directionally between the BamHI and EcoRI site of pcDNA3 (Invitrogen), and the expression cassette was fully sequenced. CHO cells were transiently transfected with plasmids pcDNA3-C2GnT3-full, pcDNA3.1-C2GnT-M, pcDNAI-C2GnT-L, and pcDNAI-IGnT and selected for neomycin resistance. Evaluation of in vivo core 2 synthase activity was performed by cotransfection with pcDSR-leukosialin (28). Core 2 O-glycan-modified leukosialin (CD43) was detected by immunofluorescence staining with monoclonal antibody T305, and large I antigen was detected with anti-I serum (Ma) using procedures described previously (3, 5, 29).

Enzymatic Assays and Product Characterization-- Standard assays were performed with semi-purified C2GnT3 in 50-µl reaction mixtures containing 100 mM MES (pH 6.5), 0.1% Nonidet P-40, 150 µM UDP-[¹⁴C]GlcNAc (2,000 cpm/nmol) (Amersham Pharmacia Biotech), and the indicated concentrations of acceptor substrates (Sigma and Toronto Research Laboratories Ltd., see Table I for structures). Reaction products were quantified by chromatography on Dowex AG1-X8. Transfer of N-acetylglucosamine to glycoprotein acceptors was evaluated by acid precipitation and filtration through glass fiber filters as described (8). Complete glycosylation of core 1-para-nitrophenyl was performed in a reaction mixture consisting of 0.6 milliunits of C2GnT3 (specific activity determined with core 1-benzyl), 2 mg of core 1-para-nitrophenyl, 100 mM MES (pH 7.0), 5 mM EDTA, 4.6 µmol of UDP-GlcNAc, and 100 milliunits of alkaline phosphatase in a final volume of 200 µl. The glycosylation of core 1-para-nitrophenyl was monitored by thin layer chromatography and run for 11 h until completed. The reaction product was purified, deuterium exchanged, and dissolved in D₂O as described (4). One-dimensional ¹H NMR, two-dimensional ¹H-¹H-TOCSY, ¹H-detected, ¹³C-decoupled, phase-sensitive gradient ¹³C-¹H HSQC, and HMBC experiments were performed as described previously (Ref. 4 and references cited therein). One-dimensional reference ¹³C NMR spectra were acquired using direct detection on a Varian Unity Inova 500 MHz spectrometer. A 1-mg sample of core 1-para-nitrophenyl (Sigma) was prepared in similar fashion and analyzed under identical conditions for comparison. Chemical shifts are referenced to internal acetone (2.225 and 30.00 ppm for ¹H and ¹³C, respectively).

Expression Analysis-- A human RNA Master Blot^TM was obtained from CLONTECH. The cDNA fragment of soluble C2GnT3 (base pairs 115-1359) was used as a probe for hybridization. The probe was random primer-labeled using [-³²P]dATP and the Strip-EZ DNA labeling kit (Ambion). The membrane was probed overnight and washed according to the protocol of the manufacturer. A Northern blot of multiple human tissues (MTN II from CLONTECH) was probed as described above and washed as described previously (4).

Chromosomal Localization of C2GnT3 by Fluorescence in Situ Hybridization-- BAC DNA from the genomic clone CIT-HSP-2288B17 was used for fluorescence in situ hybridization on normal human lymphocyte metaphase chromosomes using procedures described previously (30). Evaluation of the chromosomal slides was performed as described (4).

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Identification and Cloning of C2GnT3-- A novel member of the 6GlcNAc-T gene family, designated C2GnT3, was identified by analysis of human genome survey sequences. The open reading frame of 1359 base pairs encodes a protein of 453 amino acids with four potential N-linked glycosylation sites. A type II domain structure with an N-terminal cytoplasmic domain of 11 residues, a transmembrane segment of 21 residues, and a stem region and catalytic domain of 421 residues was predicted by the TMpred-algorithm at the Swiss Institute for Experimental Cancer Research (Fig. 1). A single initiation codon preceding the putative transmembrane segment was identified according to the Kozak rule (31).

View larger version (82K): [in this window] [in a new window]   Fig. 1.   Nucleotide sequence and predicted amino acid sequence of human C2GnT3. The amino acid sequence is shown in single-letter codes. The hydrophobic segment representing the putative transmembrane domain is underlined with a double line (TMpred-algorithm). Four consensus motifs for N-glycosylation are indicated by asterisks. The location of the primers used for preparation of the expression constructs is indicated by single underlining.

Human O-Glycan 6GlcNAc-transferases Share Significant Sequence Similarities-- Fig. 2 shows a multiple amino acid sequence alignment (ClustalW) of C2GnT3, C2GnT2, C2GnT1, and IGnT. C2GnT3 shows a higher overall amino acid sequence identity to human C2GnT1 and C2GnT2 (42%) than to human IGnT (39%). High sequence similarity with several well conserved motifs can be found in three regions (A, B, and C) in the putative catalytic domains of the four human proteins as defined previously (29). The spacing of nine cysteine residues is conserved in all four 6GlcNAc-Ts (Fig. 2). There is one conserved potential N-linked glycosylation site located in the stem region of C2GnT3, C2GnT2, and C2GnT1. This N-linked glycosylation site was shown to be essential for the catalytic function of C2GnT1 (32).

View larger version (77K): [in this window] [in a new window]   Fig. 2.   Multiple amino acid sequence analysis (ClustalW) of human C2GnT3, C2GnT2, C2GnT1, and IGnT. Introduced gaps are shown as hyphens, and aligned identical residues are boxed (black for all sequences, dark gray for three sequences, and light gray for two sequences). The putative transmembrane domains are underlined with a single line. The positions of conserved cysteines are indicated by asterisks. One conserved N-glycosylation site is indicated by an open circle.

Chromosomal Localization and Genomic Organization-- Fluorescence in situ hybridization of the genomic clone CIT-HSP-2288B17 to human metaphase chromosomes located the C2GnT3 gene to 5q12 (Fig. 3). The coding region of C2GnT3 is contained in a single exon, similar to the genomic structure of the human C2GnT1 and C2GnT2 genes (4, 25).

View larger version (17K): [in this window] [in a new window]   Fig. 3.   Fluorescence in situ hybridization of C2GnT3 to metaphase chromosomes. The genomic C2GnT3 probe (BAC DNA from clone CIT-HSP-2288B17) specifically hybridizes to chromosome 5 band q12.

Kinetic Properties of Recombinant C2GnT3-- Transfection of Sf9 cells with pAcGP67-C2GnT3-sol resulted in marked increase in 6GlcNAc-T activity. Secreted C2GnT3 exhibited significant catalytic activity with disaccharide derivatives of O-linked core 1 (Gal1-3GalNAc1-R) as substrates (Table I). In contrast, no activity was detected with the core 3 substrate (GlcNAc1-3GalNAc1-R) indicating a lack of core 4 synthase activity. Similarly, no activity was detected with -D-GalNAc-1-para-nitrophenyl, suggesting that C2GnT3 does not synthesize core 6 (GlcNAc1-6GalNAc1-R). Several substrates were tested to detect I activity, and low activity was found with high concentrations of GlcNAc1-3Gal-methyl but not with lacto-N-neo-tetraose or para-lacto-N-hexaose, which indicated that C2GnT3 exhibits low in vitro distal IGnT activity. C2GnT3 activity was also evaluated with glycoprotein acceptors. There was high activity with asialoglycophorin A and asialofetuin and lower activity with bovine asialo-submaxillary mucin (Table II). No activity was detected with human ₁-acid glycoprotein, fetuin, IgG, and transferrin. C2GnT3 exhibited strict donor substrate specificity for UDP-GlcNAc and did not utilize UDP-galactose, UDP-N-acetylgalactosamine, UDP-glucose, or UDP-xylose with the acceptor substrates tested here (data not shown) (4). The in vitro activity of C2GnT3 was enhanced by several detergents including Nonidet P-40. In our hands recombinant secreted forms of C2GnT1 and C2GnT2 expressed in Sf9 cells are inactivated by detergents (4).

C2GnT3 Controls Core 2 Branching in Vivo-- C2GnT1 and C2GnT2 generate core 2 O-glycans in vivo upon heterologous expression in CHO cells (5). Likewise, parallel transfection of expression constructs encoding C2GnT3 and leukosialin (CD43) directed cell surface expression of CD43 glycoforms with core 2 branched oligosaccharides as revealed by immunoreactivity with monoclonal antibody T305, which defines a core 2 O-glycan on CD43 (33, 34) (Fig. 4). No immunoreactivity was detected with an antibody directed to I antigen structures, indicating C2GnT3 does not exhibit IGnT activity in vivo. The low activity found in vitro with GlcNAc1-3Gal1-Me (Table I) appears to be in conflict with this result, but in vitro evaluation of I branching activity is complicated by lack of suitable complex acceptor substrates. Furthermore, analysis of in vitro and in vivo I branching activity with C2GnT2 clearly revealed discrepancies. Thus, Yeh et al. (5) found very low in vitro activity of C2GnT2 with a tetrasaccharide substrate, and Schwientek et al. (4) found no I branching activity with simpler substrates. In contrast, in vivo analysis of I branching activity of C2GnT2 demonstrated a good I branching function similar to IGnT (Fig. 4 (5)).

View larger version (26K): [in this window] [in a new window]   Fig. 4.   Immunofluorescent staining of CHO cells transfected with C2GnTs and IGnT. Anti-I and T305 staining of CHO cells transfected with expression constructs for C2GnT1, C2GnT2, C2GnT3, and IGnT are shown. For staining with monoclonal antibody T305 cells were cotransfected with pcDSR-leukosialin. Cells were incubated with human anti-I serum or monoclonal antibody T305 followed by fluorescein isothiocyanate-conjugated secondary antibody.

Product Characterization by ¹H and ¹³C NMR Spectroscopy-- The product generated by C2GnT3 using Gal1-3GalNAc-1-para-nitrophenyl was characterized by NMR spectroscopy. Comparison of a one-dimensional ¹H NMR spectrum of the product (Fig. 5) with that of the substrate (data not shown) clearly showed an additional H-1 resonance (4.457 ppm) from a sugar residue linked in the -configuration (³J_1,2 = 7-9 Hz). Some resonances from a minor product (not unreacted substrate) were also observed (Fig. 5, asterisks). Complete assignments for all ¹H and ¹³C resonances of both major product and substrate were obtained from sequential TOCSY and HSQC experiments (Table III). With the exception of some H-6R and H-6S pairs (assigned by comparison of their ³J_5,6 coupling constants with those of the corresponding benzylglycosides (35)), all ¹H assignments agreed with those published previously for both Gal1-3(GlcNAc1-6)GalNAc1-1-para-nitrophenyl and Gal1-3GalNAc1-1-para-nitrophenyl (18). Linkage assignments in the product, previously made on the basis of interglycosidic nuclear Overhauser enhancements (18), were confirmed unambiguously by observation of appropriate interglycosidic H1C1O1Cx and C1O1CxHx correlations in an HMBC spectrum (data not shown). As observed previously for a biosynthetic core 4-para-nitrophenyl product (4), the newly formed GlcNAc16GalNAc linkage in the putative core 2-para-nitrophenyl product was clearly demonstrated by strong cross-peaks correlating the -GlcNAc H-1 at 4.457 ppm with -GalNAc C-6, and the corresponding -GlcNAc C-1 at 100.95 ppm with both -GalNAc H-6 resonances. The structure of the minor product is under investigation.

View larger version (24K): [in this window] [in a new window]   Fig. 5.   Sections of a one-dimensional 1H NMR spectrum of the C2GnT3 product, Gal1-3(GlcNAc1-6)GalNAc1-1-para-nitrophenyl, showing all nonexchangeable monosaccharide ring methine and exocyclic methylene resonances. Residue designations for Gal13 (3), GlcNAc16 (6), and GalNAc11 () are followed by proton designations (1-6). Resonances from an as yet undetermined alternate product are marked by asterisks.

Expression of C2GnT3 in Human Tissues and Cancer Cell Lines-- mRNA analyses with human mRNA from 50 adult normal organs revealed a highly selective expression pattern with high levels of a C2GnT3 transcript (approximately 5.5 kilobases) exclusively in thymus (Figs. 6 and 7). Low transcript levels were also found in pancreas, peripheral blood leukocytes, placenta, small intestine, and stomach, whereas expression in kidney, liver, spleen, lung, and lymph node was barely detectable. Expression of C2GnT3 in lymphoid cancer cells was determined in a panel of human cancer cell lines. The lymphoblastic leukemia cell line MOLT-4 expressed the C2GnT3 transcript, whereas it was not detected in HL-60, HeLaS3, K-562, Raji, SW480, A549, and G361 (data not shown). The size of the C2GnT3 transcript was similar to the largest of three transcripts of C2GnT1 (5.4 kilobases) (3). Multiple transcripts of C2GnT1 and -T2 are proposed to result from differential usage of polyadenylation signals, but this has not been confirmed (3).

View larger version (45K): [in this window] [in a new window]   Fig. 6.   Northern blot analysis of C2GnT3 in human tissues. A multiple human tissue Northern blot (MTN II from CLONTECH) was probed with a 32P-labeled probe corresponding to the soluble expression fragment of C2GnT3 (base pairs 115-1359).

View larger version (35K): [in this window] [in a new window]   Fig. 7.   mRNA dot blot analysis of C2GnT3 in human tissues. A, loading pattern for the human mRNA master blot (CLONTECH). Dots in row H contain 100 ng (H1-H7) or 500 ng (H8) of control DNA or RNA. B, autoradiogram of master blot expression analysis using a C2GnT3 probe as described in Fig. 6.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

At least three human 6GlcNAc-Ts are involved in O-glycan branching. The C2GnT3 presented in this report is similar to C2GnT1 in that it functions solely in core 2 O-glycan branching (3). In contrast, C2GnT2 has more diverse functionality in core 2 and core 4 O-glycan branching as well as in synthesis of I poly-N-acetyllactosamine structures (4, 5). The catalytic activities of all three enzymes have been studied in vitro with saccharide and glycoprotein substrates and in vivo in CHO cells using immunoprobing for product identification. Nevertheless, further studies are required to evaluate the fine detail of the apparent redundancy in core 2 and I branching activities. Differences may be envisioned that are related to type of glycoconjugate, peptide sequence around glycosylation sites, and final glycan structures. C2GnT3 uniquely exhibits a restricted expression pattern in normal human organs with high expression found only in thymus (Fig. 6). Core 2 O-glycans are important for T-cell development and lymphocyte trafficking (13-15), and given the present data showing that C2GnT3 is the dominant core 2 O-glycan synthase in thymus, C2GnT3 is likely to play a significant role for these processes. It is not possible to delineate fully individual roles of each member of the 6GlcNAc-T gene family from available data of mRNA expression or 6GlcNAc-T activities, mainly because the available information is from studies of whole organs. We postulate that there is an important role for C2GnT3 in T-cell biology, based on the finding that mice deficient in C2GnT1 only showed partial reduction in selectin ligands, and lymphocyte homing was essentially unaffected (17). On the other hand, mice deficient in C2GnT1 had no measurable core 2 GlcNAc-T activity in kidney (17). This is in partial agreement with the finding that Northern analyses indicate low expression levels of C2GnT2 (4, 5) and C2GnT3 (Fig. 7) in kidney. Availability of C2GnT3 is a significant step toward understanding the complex regulation of O-glycosylation and synthesis of selectin ligands in different organs.

C2GnT3 was identified by analysis of genome survey sequences. These sequences are short single reads similar to ESTs, but the origin is genomic DNA and may represent random sequences or sequences derived from ends of genomic clones. Genome survey sequence information is preliminary to the full genomic sequence. It is estimated that the EST sequence information available so far covers over 50% of all human genes, and genomic sequencing information is required to identify remaining genes. There may be a number of reasons why transcribed genes are not represented by ESTs. Rare messages or messages with highly restricted expression patterns may not be included, but also technical problems with the EST strategy may account for this. One major problem is that EST cDNA clones are size selected for average transcript sizes to ensure 5' sequence information from coding regions, and this may be in conflict with genes having long 3'-UTRs. Furthermore, the 3'-UTR may have secondary structures preventing cDNA synthesis. Several expressed glycosyltransferase genes, e.g. the histo-blood group ABO gene, have not been identified in the EST data bases,³ but the specific reason for this is not clarified. The C2GnT3 gene identified here was shown to produce an mRNA transcript of approximately 5.5 kilobases, mainly in the thymus (Fig. 6). The available 3'-UTR sequence information did not contain a consensus polyadenylation signal, and BLAST searches against the EST data base did not provide ESTs derived from this sequence. It is conceivable that further 3' sequence contains the polyadenylation signal.

The in vitro kinetic properties of C2GnT3 resembled those of C2GnT1 (4, 5, 28) and infer its exclusive function in core 2 O-glycan synthesis. C2GnT3 efficiently utilized glycoproteins carrying unsubstituted core 1 structures including asialo-glycophorin A and asialo-fetuin. Asialo-bovine submaxillary mucin, with only 6% fucosylated or unsubstituted core 1, was a poor substrate (36, 37). A similar pattern of activity toward glycoproteins was recently obtained with recombinant mouse C2GnT1 (38). Both C2GnT1 and C2GnT3 exhibited core 2 synthase activity in vivo with CD43, suggesting that both enzymes function with the same glycoprotein acceptor (33, 34). Glycosphingolipid substrates have not been tested with any of the human 6GlcNAc-Ts, mainly because of a lack of availability of these complex glycolipids. However, a kidney-associated form of murine C2GnT1 was shown to regulate expression of hybrid-type lacto-globoseries glycolipids (Gal1-3[GlcNAc1-6]GalNAc1-3Gal1-4Gal1-4Glc1-Cer) (39). To our knowledge 6GlcNAc branching has only been found in (neo)-lactoseries glycolipids in man.

Several previous studies described developmentally regulated core 2 GlcNAc-T activities during various cellular events (18, 20, 21, 34, 40), but these have only been correlated with specific 6GlcNAc-T gene expression in a few instances (13, 41). Nakamura et al. (42) found that enzyme activity and transcript levels of C2GnT1 are coordinately down-regulated during 12-O-tetradecanoylphorbol-13-acetate treatment of the human leukemia cell line KM3. Changes in C2GnT1 expression in the developing mouse embryo were shown by in situ hybridization (43). A recent study found a 100-fold difference in enzyme activity but unchanged levels of murine C2GnT1 mRNA upon induced retrodifferentiation of the embryonal cell lines PSA-5E and PYS-2 (44). Thus, the specific contribution of individual 6GlcNAc-T isoforms to core 2 and I branching activities in cells and tissues are still unclear.

Northern analysis indicated highly restricted expression of C2GnT3 mainly to the thymus (Fig. 6). C2GnT1 is variably expressed in most organs tested, and the highest levels are found in thyroid, spleen, and mucosal tissues (5). Expression of C2GnT1 in mucosal tissues resembles the expression pattern of C2GnT2 (4, 5). C2GnT3 is also weakly expressed in small intestine and stomach. C2GnT1 was found by Northern analysis to be weakly expressed in thymus (5), and in situ hybridization localized C2GnT1 expression to thymocytes (13). Thus, among the known core 2 O-glycan synthase genes, C2GnT3 appears to be the dominant core 2 synthase in thymus. One possibility is that C2GnT3 has important functions in synthesis of core 2 O-glycans on cortical thymocytes (13). Core 2 structures on thymocyte surface proteins are ligands for galectin-1, which participates in interactions between thymocytes and thymic epithelium (13). Both C2GnT1 and C2GnT3 are weakly expressed in peripheral blood leukocytes and lymph nodes; however, further studies are required to define specific cell types. C2GnT1 is highly expressed in spleen indicating strong expression in B-cells, which is in agreement with the finding that C2GnT1 expression correlates with O-glycan branch formation and sialyl-Le^x expression in the human pre-B lymphocytic leukemia cell line KM3 (42). In contrast, strong expression of C2GnT3 in thymus indicates association with T-cells. In situ hybridization and/or immunohistology with appropriate probes are required for further clarification.

The 6GlcNAc-T gene family consists of at least four glycosyltransferases with acceptor substrate specificity for galactose and/or N-acetylgalactosamine and one pseudogene (45). The genomic organizations of the three C2GnT genes are similar with the coding regions contained in a single exon. Introns in the 5'-UTR of C2GnT1 and C2GnT2 have been found (25).⁴ In contrast, the coding region of IGnT is contained in three exons (25). The genes are located at different chromosomal loci as follows: IGnT is at 6p24; C2GnT1 is at 9q21; C2GnT2 is at 15q22; and C2GnT3 is at 5q12 (4, 5, 29). The relatively high sequence similarity among the coding regions of the core 2 synthase genes and their similar genomic organizations indicate that these arose through gene duplication followed by sequence divergence. The more complex genomic organization of IGnT combined with lower sequence similarity to C2GnTs suggests that IGnT represents a distinct member of the 6GlcNAc-T family that diverged early in evolution. Surprisingly, a total of six homologues of the C2GnT gene family have been identified in the Caenorhabditis elegans genome, but functional characterizations of these genes have not been reported (46). The characteristics of the 6GlcNAc-T gene family resemble those of other homologous glycosyltransferase gene families in terms of genes and primary structures of the proteins. Moreover, there are differences in kinetic properties and expression patterns of individual isoforms of each family. Thus, it appears likely that gene duplication and divergence was influenced by functional requirements for synthesis of different glycoconjugates. Another possible influence was the need to obtain differential regulation of glycoconjugate synthesis in different cell types during various biological and developmental processes.

The biological roles of core 2 O-glycans are among the most extensively studied in the carbohydrate field. Understanding the regulatory mechanisms of core 2 O-glycosylation is of high interest. Identification and characterization of the core 2 synthase genes is a first step. The finding that C2GnT3 is a thymus-associated core 2 synthase is intriguing and holds promise for future studies of its role in the immune system.

	ACKNOWLEDGEMENT

We thank Dr. Michael A. Hollingsworth for helpful suggestions and critical reading of the manuscript.

	FOOTNOTES

^* This work was supported by The Danish Cancer Society, the Velux Foundation, the Danish Medical Research Council, funds from the EU Biotech 4th Framework, and National Institutes of Health Resource Center for Biomedical Complex Carbohydrates Grant 5 P41 RR05351.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The nucleotide sequence(s) reported in this paper has been submitted to the GenBank^TM/EMBL Data Bank with accession number(s) AF132035.

^§ To whom correspondence should be addressed: School of Dentistry, Nørre Alle 20, DK-2200 Copenhagen N, Denmark. Tel.: 45 35326519; Fax: 45 35326835; E-mail: tsc@odont.ku.dk.

¹ C2GnT1 (C2GnT, C2GnT-L), C2GnT2 (C2/4GnT, C2GnT-M), and IGnT represent human 6GlcNAc-transferases with GenBank^TM accession numbers M97347, AF038650/AF102542, and Z19550, respectively.

³ E. P. Bennett, T. Schwientek, and H. Clausen, unpublished observations.

⁴ T. Schwientek and H. Clausen, unpublished observations.

	ABBREVIATIONS

The abbreviations used are: 6GlcNAc-T, UDP-N-acetyl--D-glucosamine:Gal/GalNAc 1,6-N-acetylglucosaminyltransferase; 3GlcNAc-T, UDP-N-acetyl--D-glucosamine:Gal1,3-N-acetylglucosaminyltransferase; EST, expressed sequence tag; HMBC, heteronuclear multiple bond correlation; HSQC, heteronuclear single quantum correlation; MES, 4-morpholineethanesulfonic acid; PCR, polymerase chain reaction; TOCSY, total correlation spectroscopy; UTR, untranslated region; CHO, Chinese hamster ovary.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES