Studies on the biosynthesis of ABH and Lewis epitopes on O-glycans

Sammanfattning: ABH and Lewis antigens are carbohydrate epitopes (glycans) found on a wide variety of cell surfaces. They are involved in many biological processes, e.g transplantation rejection, bacterial and viral adhesion, cell-cell communication and metastasis. The in vivo synthesis of glycans is mediated by glycosyltransferases (GTs). Common GTs for the formation of ABH and Lewis antigen is fucosyltransferases (FucTs), either alpha2- or alpha3/4-FucTs. Glycans on proteins are usually O- or Nlinked. On 0-glycans, ABH and Lewis epitopes are carried by different core structures. In this thesis we report on the glycan- and core chaindependence in vivo of FucTs and beta3galactosyltransferases (beta3GalTs) involved in ABH and type 1 chain Lewis antigen biosynthesis. To this end, P-selectin glycoprotein ligand-1 (PSGL-1) or CD43 (both O-linked type), or alpha1-acid glycoprotein (AGP; N-linked type) IgG fusion proteins were expressed in cells co-transfected with FucTs and the blood group A gene or beta3GalTs and different core enzymes. The knowledge gained was used to engineer various tools of diagnostic or therapeutic value. These include absorbers of anti blood group A-antibodies, and cells that can successfully be used in adhesion of Helicobacter pylori, which is a bacterium that causes gastric cancer and ulcers. Blood group A epitopes were synthesized on PSGL-1/mIgG in 293T, COS and CHOK1 cells and the epitope density was found to be highest in CHO co-expressing FUT2 and the A synthase. This PSGL1/mIgG was used for absorption of anti-blood group A antibodies in human blood group 0 serum. At least 80 times less A trisaccharides on PSGL- 1 /mIgG in comparison to a synthetically made absorbent were needed for the same level of antibody absorption. FUT1 and FUT2 were both found to direct alpha2-fucosylation on type 1 chains on both N- and 0-glycans. When comparisons were made for Galbeta4GIcNAc (type 2) and Galbeta3GaINAc (type 3 or core 1) preference on 0-glycans, FUT1 and FUT2 preferentially fucosylated type 2 and type 3, respectively. beta3Gal-T1, M and -T5 could synthesize type 1 chains on N-glycans, but only beta3GalT5 worked on 0glycans. The latter enzyme acted on both core 2 and core 3 0-glycans. Furthermore, the specificity of FUT3 and FUT5 in Le' and Le' synthesis was different, with FUT5 fucosylating H type 1 on core 2, but FUT3 fucosylating H type 1 almost only on core 3. We also found that the Lea-specific antibody T174 did not react with Le' on core 3. On CD43/IgG, FUT3, FUT5, FUT6 and FUT7 gave rise to SLex on core 2 0-glycans in CHO, whereas only FUT5 and FUT6 were effective in producing SLex on core 3, as detected with the monoclonal antibody CSLEX. PSGL- 1 did also support SLex production by FUT3, FUT5, FUT6 and FUT7 on core 2. In contrast, no SLex at all was detected on PSGL-1 carrying core 3 modifications. KM93 reacted well with SLex on core 2, but did not at all stain the same epitope on core 3 0-glycans, regardless of the fusion protein used. FUT3, FUT5, FUT6 and FUT7 were all able to produce SLex on AGP in CHO but not in COS and 293T. Taken together, our findings point to the importance of protein sequence and core saccharide structure for FucT activity and Lewis epitope biosynthesis, as well as for the binding of antibodies. Our results further provide a good framework for future investigations on the role of carbohydrates in metastasis, bacterial and cellular adhesion and the role of multivalency for inhibition of protein- carbohydrate interactions.