By a tblastn search with ␤1,4-galactosyltransferases as query sequences, we found an expressed sequence tag that showed similarity in ␤1,4-glycosyltransferase motifs. The full-length complementary DNA was obtained by a method of 5-rapid amplification of complementary DNA ends. The predicted open reading frame encodes a typical type II membrane protein comprising 543 amino acids, the sequence of which was highly homologous to chondroitin sulfate N-acetylgalactosaminyltransferase (CSGalNAcT-1), and we designated this novel enzyme CSGalNAcT-2. CSGalNAcT-2 showed much stronger Nacetylgalactosaminyltransferase activity toward glucuronic acid of chondroitin poly-and oligosaccharides, and chondroitin sulfate poly-and oligosaccharides with a ␤1-4 linkage, i.e. elongation activity for chondroitin and chondroitin sulfate, but showed much weaker activity toward a tetrasaccharide of the glycosaminoglycan linkage structure (GlcA-Gal-Gal-Xyl-O-methoxyphenyl), i.e. initiation activity, than CSGalNAcT-1. Transfection of the CSGalNAcT-1 gene into Chinese hamster ovary cells yielded a change of glycosaminoglycan composition, i.e. the replacement of heparan sulfate on a syndecan-4/fibroblast growth factor-1 chimera protein by chondroitin sulfate, however, transfection of the CSGalNAcT-2 gene did not. The above results indicated that CSGalNAcT-1 is involved in the initiation of chondroitin sulfate synthesis, whereas CSGalNAcT-2 participates mainly in the elongation, not initiation. Quantitative real-time PCR analysis revealed that CSGalNAcT-2 tran-

We found a novel human gene (GenBank TM accession number AB037823, Kazusa DNA Research Institute KIAA1402) that possesses homology with chondroitin synthase. The full-length open reading frame consists of 772 amino acids and encodes a typical type II membrane protein. This enzyme had a domain containing ␤3-glycosyltransferase motifs, which might be a ␤3-glucuronyltransferase domain, but no domain with ␤4-glycosyltransferase motifs, although both are found in chondroitin synthase. The putative catalytic domain was expressed in COS-7 cells as a soluble enzyme. Its glucuronyltransferase activity was observed when chondroitin and chondroitin sulfate polysaccharides and oligosaccharides were used as acceptor substrates. However, it was not detected when dermatan sulfate, hyaluronan, heparan sulfate, heparin, N-acetylheparosan, lactosamine tetrasaccharide, and linkage tri-and tetrasaccharide acceptors were employed. The reaction product, which was speculated to exhibit a GlcA␤1-3GalNAc linkage structure at its non-reducing terminus, showed the following characteristics. 1) It was catabolized by ␤-glucuronidase. 2) It was an acceptor for Escherichia coli K4 chondroitin polymerase (K4 chondroitin polymerase). 3) The product of K4 chondroitin polymerase was cleaved by chondroitinase ACII. On the other hand, no N-acetylgalactosaminyltransferase activity was detected toward any acceptors. Quantitative real time PCR analysis revealed that its transcripts were highly expressed in the placenta, small intestine, and pancreas, although they were ubiquitously expressed in various tissues and cell lines. This enzyme could play a role in the synthesis of chondroitin sulfate as a glucuronyltransferase.

We found a novel glycosyltransferase gene having a hypothetical ␤1,4-galactosyltransferase motif (GenBank TM accession number AB081516) by a BLAST search and cloned its full-length open reading frame using the 5-rapid amplification of cDNA ends method. The truncated form was expressed in insect cells as a soluble enzyme. It transferred N-acetylgalactosamine, not galactose, to para-nitrophenyl-␤-glucuronic acid. The N-acetylgalactosamine-glucuronic acid linkage has been identified only in chondroitin sulfate; therefore, we examined its chondroitin elongation and initiation activities. N-Acetylgalactosaminyltransferase activity was observed toward chondroitin poly-and oligosaccharides, chondroitin sulfate oligosaccharides, and linkage tetrasaccharide (GlcA-Gal-Gal-Xyl-O-methoxyphenyl), and the chondroitin polysaccharide and linkage tetrasaccharide were better acceptor substrates than the others. Northern blot analysis and quantitative realtime PCR analysis revealed that its 4-kb transcripts were highly expressed in thyroid and placenta, although they were ubiquitously expressed in various tissues and cells. These results suggest that this enzyme has N-acetylgalactosaminyltransferase activity in both the elongation and initiation of chondroitin sulfate synthesis. Furthermore, we performed enzymatic synthesis of chondroitin pentasaccharide in vitro. In one tube reaction with four enzymes, ␤1,4-galactosyltransferase-VII, ␤1,3-galactosyltransferase-VI, glucuronyltransferase-I, and this enzyme, and a synthetic xylose-peptide acceptor, the structure GalNAc-GlcA-Gal-Gal-Xyl-peptide was constructed. This is the first report of a chondroitin pentasaccharide constructed with recombinant glycosyltransferases in vitro.

The degradation of type IX collagen, a minor collagen in cartilage, was examined by treatment with three different types of matrix metalloproteinases (MMPs) purified from the culture medium of rheumatoid synovial cells. Neither MMP-1 (collagenase) nor MMP-2 (so-called 'gelatinase') could digest type IX collagen, but MMP-3 (stromelysin) readily degraded it into smaller fragments. This suggests that MMP-3 may be responsible for the pathological degradation and/or normal turnover of type IX collagen.