The Evolutionary History of GlycosyltransferaseGenes

Carbohydrate structure is one of the important macro-biomolecules. Carbohydrate chains are synthesized through the coordinated action of glycosyltransferases, taking a step-by-step approach. The carbohydrate chains of glycolipids and glycoproteins at the cell surface change dramatically in a variety of biological phenomena, such as embryogenesis, development, differentiation, immunoresponses, and cancer metastasis. The Lewis x (Lex) epitope is one of such carbohydrate. Lex is defined as Galβ1,4 (Fuc αl,3) GlcNAc, which is synthesized by transferring a Fuc to the GlcNAc residue of the type 2 chain, Galβ1,4GlcNAc-R, with a α1,3-linkage. This fucose-transfer is catalyzed by α1,3FucT. So far, the human genes encoding five α1,3FucTs (hFucTIII, IV, V, VI and VII or FUT3, 4, 5, 6 and 7) have been cloned. CD15 is one of the differentiation markers of the cells, and the CD15- epitope has been determined as the Lex carbohydrate structure. Several immunohistochemical studies have detected CD15 antigens in certain neuronal cells and glial cells in the central nervous system (CNS) of humans and rodent. The expression of CD15 antigen in the CNS is developmentally regulated, and considered to play an important role in neuronal development. In our previous study, mFucTIX was identified as the most likely candidate for the enzyme synthesizing the Lex structure (CD15- epitope) in the mouse CNS. I therefore cloned a human fucosyltransferase gene, named hFucTIX, orthologous to the mouse FucTIX gene. I screened the human stomach cDNA library with a mFucTIX cDNA probe encompassing the full-length open reading frame (ORF), and obtained several clones encoding the hFucTIX gene. The deduced amino acid sequence of hFucTIX, consisting of 359 amino acid residues, indicated a type II membrane protein and was very highly conserved with mFucTIX. The Namalwa cells stably expressing the hFucTlX gene were established, used for flow cytometry analysis, and assaying of α1,3FucT activity. The hFucTIX transcripts were abundantly expressed in brain and stomach, and interestingly were detected in spleen and peripheral blood leukocytes. I also performed FISH analysis, and the hFucTIX gene was shown to be located in the 6q16, long arm of human chromosome 6. The phylogenetic tree indicated that the FucTIX first diverged in vertebrate evolution and the rate of nucleotide substitution indicates that FucTIX seems to be under strong selective constraint. To know the detailed phylogenetic relationship of this gene family, I obtained novel FucT gene sequences from chicken and xenopus cDNA or genomic DNA. Those were counterparts of FucTIX, named xFucTIX and cFucTIX, and I also found two novel FucTs, named XFTI and CFTII. Then I reconstructed the phylogenetic tree of the α1,3FucT gene family. XFucTIX and cFucTIX were clustered with the Fuc-TIX subfamily. XFTI and CFTII were clustered with the Fuc-TIV, and Fuc-TVII, respectively. These novel sequences also improved the bootstrap values of the phylogenetic tree. The dramatic changes of glycoconjugates observed during embryogenesis and the differentiation of cultured embryonal carcinoma (EC) cells suggest that cell surface glycoconjugates play a vital role in embryogenic development. A glycoconjugate having a Lex determinant (recognized by anti-SSEA- 1 antibody) was found to be maximally expressed at the morula stage and to decline greatly after that stage. Since compaction was inhibited by multivalent Lex oligosaccharide, this structure may play a role in this process, the very first overt morphogenic change during embryogenesis. I therefore studied fucosyltransferase mRNA levels in mouse early embryo using competitive PCR-methods. I determined that the FucTlX is responsible for the Lex synthesis in the mouse early embryo. The past few years have seen rapid advances in sequencing the genomic DNA of huunan, Caenorhabditis elegans, and so on. As a result, a large number of novel glycosyltransferase genes have been discovered from those genome sequences. How did they increase their family members during the genome evolution? In vertebrate genomes it is often found that the homologues of a group of genes often form another cluster on a different chromosome. It seems to be sure that the two genome duplication events occurred, one close to the origin of the vertebrates and the second close to the origin of the gnathostomes. As a result, the MHC paralogous regions were detected on chromosomes 1, 6, 9, and 19. Genes of the Hox cluster are also mapped to chromosomes 2, 7, 12, and 17. Therefore, glycosyltransferase genes may also show "homologous clusters" such as the MHC region and Hox cluster. If so, they might have arisen as the results of three separate duplication events like MHC and Hox clusters. I thus conducted molecular evolutionary analyses on 19 glycosyltransferase gene families. It is the first attempt that so many glycosyltransferase genes were analyzed through molecular evolutionary methods. (I) FucT: I constructed the phylogenic tree of α1,2FucT family. The topology of tree was not compatible with the established mammalian phylogenic tree. To confirm their phylogenetic relationship, I constructed phylogenetic network. As a result, I found that the gene conversion occurred in the α1,2FucT gene family. I also reconstructed the phylogenetic tree of α1,3FucTs including novel FucT sequence, xFucTIX, cFucTIX, XFTI, and CFTII. The α1,6FucT was distinct from other FucT, such as molecular weight, and the substrate specificity. (II) GalT: β1,3GalTs, and β1,4GalTs. The β1,3GalT family shares some motif with β1,3GnTs. The tree topology indicates the β3GalTs and β3GnTs diverged before Drosophila and vertebrate speciation, and they increased their family members by gene duplications. The phylogenetic tree of β1,4GalT family indicated that β1,4GalT increased gene numbers before divergence of vertebrates and invertebrates. (III) GlcNAcT includes the Mgat gene family. No sequence homology was found among these Mgats, in spite of their enzymatic similarity. (IV) β1,6GlcNAcT. This family contains two distinct groups based on their enzyme activity. One is C2GnT group, and the other is IGnT group. But they share their sequence homology, so I constructed the phylogenetic tree. As expected, IGnT first diverged from C2GnTs by gene duplication. (V) I performed the phylogenetic analysis of polypeptide: GalNAcT. This gene seems to the oldest glycosyltransferase within this study, for they diverged before Metazoans and Proteostome divergence. (VI) Sialyltransferases: The sialic acids are typically found at the outermost ends of N-glycans and O-glycans, and glycosphingolipids. Sialyltransferases were divided into three groups, depends on their enzyme activity and sequence homology. These phylogenetic analyses revealed that the glycosyltransferase genes increased their membres by gene duplication. I estimated the numbers of ancestral genes and duplication events. Although I failed to find simple and clear explanation between the chromosomal locations and the topology of phylogenetic trees. I found novel candidates of gene cluster region, 3, 11, 18, and 22. Finally, I calculated the numbers of synonymous (ds) and nonsynonymous (dN) nucleotide substitutions for each glycosyltransferase genes, and estimated the evolutionary rates. Comparison of evolutionary rates revealed that the glycosyltransferase tend to evolve slowly than other genes. FucTs indicated somewhat higher evolutionary rates than the others. However, FucTIX conspicuously showed a very slow evolutionary rate.