(531e) Engineering of Sialyltransferase into Glycan Binding Protein Using a Novel Mammalian Cell Surface Display Platform

Glycosylation is an essential post-translational modification that regulates a variety of biological processes. An expanded toolkit of molecular probes is needed to recognize cellular glycan structures in a stereo-specific manner, as this can yield novel biomarker of cell differentiation and disease progression. Currently available lectins have shallow binding pockets and commonly exhibit broad binding specificity or only recognize terminal epitopes. Anti-carbohydrate antibodies are difficult to produce due to their low immunogenicity to glycan antigens. To address these limitations, we tested the hypothesis that mammalian glycosyltransferases can be converted into glycan binding proteins with high specificity due to their relatively deep binding pocket and selective substrate binding properties. A fusion protein was created with human IgG1 Fc N-terminally linked to pig ST3Gal1 (abbreviated PS1). This protein exhibited enzyme activity but only weak binding to O-glycans. Introduction of an H302A mutation in this construct resulted in loss of enzymatic activity but strong binding preference for sialoglycans in multiple cell types. By performing a broad CRISPR-Cas9 based library screen, studies with a panel of isogenic glycogene knockouts, and glycan microarray studies, we determined that H302A specifically binds α(2-3)sialylated core-2 O-glycans. Additionally, its binding specificity was distinct from other known sialic acid binding lectins. To expand the repertoire of lectins, a novel mammalian cell surface display platform was developed, and this was used to screen for additional H302A variants. Using a rationally designed mutant library with 1680 candidates, we identified a triple mutant of PS1, sCore2, containing H302A, A312I and F313S mutations that displayed better binding properties than H302A for sialyl-core 2 O-glycans. In spectral flow cytometry studies that analyzed glycan profiles of 35 immune cell types, sCore2 exhibited strong binding for neutrophils, basophils, monocytes, and terminally differentiated NK- and T-cells, suggesting that sialyl-core 2 O-glycan expression is abundant in terminal effector cell types. In human tissue microarrays, sCore2 stained normal spleen and breast cancer tissue, confirming the reagent utility in paraffin embedded sections. In summary, we present a streamlined approach to generate glycan binding proteins based on the known substrate specificity of the starting glycosyltransferases.