B.A. Macher

The study of the expression of alpha-galactosyl epitopes on various mammalian cells is of particular interest, since as much as 1% of circulating IgG antibodies in humans interact with this carbohydrate residue. This natural antibody, designated "anti-Gal," was previously found to bind to terminal Gal alpha 1----3Gal beta 1----4GlcNAc-R on biochemically defined glycolipids (Galili, U., Macher, B. A., Buehler, J., and Shohet, S. B. (1985) J. Exp. Med. 162, 573-582; Galili, U., Buehler, J., Shohet, S. B., and Macher, B. A. (1987) J. Exp. Med. 165, 693-704). The expression of anti-Gal binding epitopes on nucleated cells from various mammalian species was studied by immunostaining with this antibody. The binding of anti-Gal to various cells was correlated with the binding of the lectin Bandeiraea (Griffonia) simplicifolia IB4 (BS lectin). The BS lectin also interacts with alpha-galactosyl residues and particularly with high affinity with Gal alpha 1----3Gal beta 1----4GlcNAc residues. We observed a striking evolutionary pattern in the expression of these epitopes on mammalian nucleated cells. Fibroblasts, epithelial cells, endothelial cells, smooth muscle cells, and lymphoid cells of nonprimate mammals, prosimians, and New World monkeys readily bound both anti-Gal and BS lectin. However, no such binding was detectable on cells of Old World monkeys, apes, and humans. Measurment of the binding of radiolabeled BS lectin to the various nucleated cells suggests that cells binding anti-Gal express 10(6) to 3.5 x 10(7) alpha-galactosyl epitopes, most of which, based on the anti-Gal specificity, seem to have the structure of Gal alpha 1----3Gal beta 1----4GlcNAc-R. The absence of these epitopes from human cells results from diminished activity of the enzyme alpha 1----3 galactosyltransferase, which catalyzes the following reaction. Gal beta 1----4GlcNAc-R + UDP-Gal(alpha 1----3-galactosyltransferase)----Gal alpha 1----3Gal beta 1----4GlcNAc-R + UDP This enzyme, which participates in the glycosylation of cell membrane glycoconjugates in nonprimate mammals, prosimians, and New World monkeys, appears to have been suppressed in Old World primates as a result of evolutionary events which occurred 20-30 million years ago. It is argued that an anomalous activity of this enzyme in man may result in initiation of autoimmune diseases because of the de novo expression of Gal alpha 1----3Gal beta 1----4GlcNAc-R epitopes recognized by anti-Gal.

Anti-Gal is a natural antibody, which constitutes as much as 1% of circulating IgG in humans and displays a distinct specificity for the structure Gal alpha 1----3Gal. This glycosidic structure has been found on various tissues of many nonprimate mammals. A comparative study of the occurrence of anti-Gal versus the expression of the Gal alpha 1----3Gal epitope was performed in primates, and a distinct evolutionary pattern was observed. Whereas anti-Gal was found to be present in Old World monkeys and apes in titers comparable to those in humans, its corresponding antigenic epitope is abundantly expressed on erythrocytes of New World monkeys. Immunostaining with anti-Gal of glycolipids from New World monkey erythrocytes indicated that the molecules to which anti-Gal binds are similar to those found in rabbit and bovine erythrocytes. These findings indicate that there is an evolutionary reciprocity between New World and Old World primates in the production of the Gal alpha 1----3Gal structure and the antibody that recognizes it. The expression of the Gal alpha 1----3Gal epitope was evolutionarily conserved in New World monkeys, but it was suppressed in ancestral lineages of Old World primates. The suppression of this epitope was accompanied by the production of anti-Gal. The observed in vivo binding of anti-Gal to human normal senescent and some pathologic erythrocytes implies that the Gal alpha 1----3Gal epitope is present in man in a cryptic form.

A natural IgG antibody (anti-Gal) with alpha-galactosyl binding specificity has been found in large amounts (0.5 - 1.0% of serum IgG) in all individuals tested. It has been purified by affinity chromatography on a column of melibiose-Sepharose. In addition to its affinity for normal and pathological senescent human red cells, the antibody readily interacts with rabbit red blood cell (RRBC) glycolipids with alpha-galactosyl terminal residues. Two types (glycosidic linkages of 1----3 vs. 1----4) of rabbit red cells glycolipids with terminal alpha-galactosyl residues were tested for antibody binding. The antibody specifically bound to glycolipids with Gal alpha 1----3 terminal residues, and treatment of these glycolipids with alpha-galactosidase abolished binding. Hemagglutination inhibition studies with oligosaccharides of known structure also showed that the antibody binds specifically to glycoconjugates with an alpha 1----3 terminal galactose residue. Anti-Gal did not bind to a human B-active glycolipid, indicating that fucose-linked alpha 1----2 to the penultimate galactose prevents anti-Gal binding. The anti-Gal specificity for RRBC glycolipids also paralleled that of the alpha-galactosyl-specific Bandeiraea simplicifolia lectin. The possible reasons for the occurrence of this unique antibody in human serum are discussed.

We have used the human Lewis blood group fucosyltransferase cDNA and cross-hybridization procedures to isolate a human gene that encodes a distinct fucosyltransferase. Its DNA sequence predicts a type II transmembrane protein whose sequence is identical to 133 of 231 amino acids at corresponding positions within the catalytic domain of the Lewis fucosyltransferase. When expressed by transfection in cultured cell lines, this gene determines expression of a fucosyltransferase capable of efficiently utilizing N-acetyllactosamine to form the Lewis x determinant (Gal beta 1----4[Fuc alpha 1----3]GlcNAc). By contrast, biochemical and flow cytometry analyses suggest that the enzyme cannot efficiently utilize the type II acceptor NeuNAc alpha 2----3Gal beta 1----4GlcNAc, to form the sialyl Lewis x determinant. In Chinese hamster ovary cells, however, the enzyme can determine expression of the alpha 2----3-sialylated, alpha 1----3-fucosylated structure known as VIM-2, a putative oligosaccharide ligand for ELAM-1. Cell adhesion assays using VIM-2-positive, sialyl Lewis x-negative transfected Chinese hamster ovary cells indicate that surface expression of the VIM-2 determinant is not sufficient to confer ELAM-1-dependent adhesive properties upon the cells. These results demonstrate that substantial structural similarities can exist between mammalian glycosyltransferases with closely related enzymatic properties, thus facilitating isolation of their cognate genes by cross-hybridization methods. The results further suggest that cell surface expression of the VIM-2 determinant is not necessarily sufficient to mediate ELAM-1-dependent cell adhesion.

We have used mouse mAbs, 3F11 and 06B4, that are specific for highly conserved epitopes of Neisseria gonorrhoeae lipooligosaccharides (LOS) to identify immunochemically similar structures on human erythrocytes. mAb 3F11 agglutinated erythrocytes from all randomly selected adult humans, while mAb 06B4 agglutinated only 80% of the same specimens. The antibodies had an activity with erythrocytes similar to human cold agglutinins in that agglutination occurred at 4 degrees C and decreased with increasing incubation temperature. Human infant erythrocytes were agglutinated less well, but enzymatic treatment of either infant or adult cells resulted in an increase in expression of the 3F11- and 06B4-defined epitopes. Both antibodies bound to a series of neutral glycosphingolipids from human erythrocytes and neutrophils that have a type 2 (Gal beta 1----4GlcNAc) or N-acetyllactosamine structure. Neither antibody bound to glycosphingolipids from human meconium, which have a type 1 (Gal beta 1----3GlcNAc) structure. The antibodies were unable to bind to N-acetyl-lactosamine glycosphingolipids with a nonreducing terminal sialic acid or a Gala1----3Gal disaccharide. Antibody binding also was blocked by the presence of fucose linked to the penultimate glucosamine residue of N-acetyllactosamine glycosphingolipids. Although both antibodies bound to linear and branched-chain N-acetyllactosamine glycosphingolipids, 3F11 had a higher affinity for branched structures than did 06B4. The activity of 3F11 with human adult and infant treated and untreated erythrocytes with N-acetyllactosamine glycosphingolipids, and with LOS was very similar, if not identical, in specificity to 1B2, an mAb prepared from mice inoculated with a linear N-acetyllactosamine glycosphingolipid.