Using CRISPR-Cas9 to quantify the contributions of O-glycans, N-glycans and Glycosphingolipids to human leukocyte-endothelium adhesion

Abstract

There is often interest in dissecting the relative contributions of the N-glycans, O-glycans and glycosphingolipids (GSLs) in regulating complex biological traits like cell signaling, adhesion, development and metastasis. To address this, we developed a CRISPR-Cas9 toolkit to selectively truncate each of these commonly expressed glycan-types. Here, O-glycan biosynthesis was truncated by knocking-out Core 1 β3Gal-T Specific Molecular Chaperone (COSMC), N-glycans by targeting the β1,2 GlcNAc-transferase (MGAT1) and GSLs by deleting UDP-glucose ceramide glucosyltransferase (UGCG). These reagents were applied to reveal the glycoconjugates regulating human myeloid cell adhesion to selectins under physiological shear-flow observed during inflammation. These functional studies show that leukocyte rolling on P- and L-selectin is ablated in cells lacking O-glycans, with N-glycan truncation also increasing cell rolling velocity on L-selectin. All three glycan families contributed to E-selectin dependent cell adhesion with N-glycans contributing to all aspects of the leukocyte adhesion cascade, O-glycans only being important during initial recruitment, and GSLs stabilizing slow cell rolling and the transition to firm arrest. Overall, the genome editing tools developed here may be broadly applied in studies of cellular glycosylation.

Figure 1. Schematic for the initiation of O-linked, lipid linked and N-linked glycosylation.

(a) Core-1 derived O-glycan biosynthesis begins with the formation of the T-antigen using the T-synthase C1GalT1 and its chaperone COSMC. (b) UGCG adds the first Glc to ceramide to form Glc-Cer glycolipids. (c) MGAT1 catalyzes the addition of GlcNAc to the Man-5 substrate. Knocking out these three key enzymes results in premature termination of specific types of glycoconjugates. All figures use the Consortium for Functional Glycomics symbolic nomenclature (shown in legends).

Figure 2. Generation of glycosyltransferase KO clones.

(a) Arrows represent the workflow for creating KOs. Arrows lead from original cell to KO. Some double and triple -KOs were generated by serial genome editing. Thus, genome edits made in one step remain in all subsequent editing steps. Solid and dashed lines lead to single (―), double (− −) or triple-KOs (― • ―). There is only one copy of the COSMC, MGAT1 and UGCG genes in HL-60s. (b) Flow cytometry measurement of VVA and L-PHA lectin binding to glycoT-KO clones. Knocking out O-glycans by targeting COSMC augments VVA-lectin binding. Knocking out MGAT1 reduces L-PHA binding. Additional lectin staining data are presented in the Supplemental Material section. (c) Sanger sequencing results for individual KO-clones is shown next to corresponding cytometry plots. In each case, wild-type sequence is shown on the first line with target gRNA underlined and PAM/NGG sequence in bold-italicized fonts. Lower line shows sequencing results for the individual KOs with dashes representing gene deletions. For simplicity, cells having truncated O-glycans, N-glycans and glycolipids are abbreviated [O]¯, [N]¯, [G]¯ respectively (see panel b). Double and triple KOs are also similarly defined.

Figure 3. Enzymatic activity of glycoT KO clones.

C1GalT1 (panel a), MGAT1 (panel b) and UGCG (panel c) enzyme activity was measured in WT and KO cell lines using the following acceptors: GalNAc-O-Benzyl for C1GalT1, man3-octyl for MGAT1 and fluorescent C6-NBD-ceramide for UGCG. Schematic representations of the individual reactions is presented above each TLC plate image captured using either phosphorimaging (a,b) or fluorography (c). In (a,b), the upper portion of the image presents independent control runs performed for each cell line in the absence of the acceptor substrate. This is not necessary in panel c, due to the use of a fluorescent substrate. Some lanes in panel a have minor background in both acceptor and ‘no acceptor’ controls. Knocking out specific genes abrogates corresponding enzyme activity, shown using red arrows to indicate missing reaction products.

Figure 4. Carbohydrate epitope expression on KO cell lines.

Flow cytometry measurements for the expression of the: (a) Cutaneous Lymphocyte Antigen (CLA) defined by mAb HECA-452, (b) CD15s/sLe^X/sialyl Lewis-X using mAb CSLEX-1 and (c) Lewis-X/CD15 using mAb HI98. MFI: mean fluorescence intensity. * P < 0.05 with respect to WT HL-60s. ‡ and ^† P < 0.05 with respect to all treatments except that bars marked by ‡ and † are not different from each other. ^#P < 0.05 with respect to all treatments.

Figure 5. GlycoT KO rolling on recombinant P- and L-selectin.

WT and KO HL-60s were perfused over (a,b) recombinant P-selectin or (c,d) L-selectin. Cell adhesion data are presented in left panels while cumulative rolling velocity plots appear on the right. Knocking out O-glycans (i.e. [O]⁻ cells) resulted in almost no interaction with either selectin. Knocking out N-glycans ([N]⁻ cells) augmented rolling velocity on L-selectin. *P < 0.05 with respect to all other cell types for rolling interactions. ^†P < 0.05 with respect to all other cell types for adherent cells. Dashed lines in panels b and d are used to indicate the median rolling velocity.

Figure 6. Cell adhesion to recombinant E-selectin-IgG under static and fluid shear conditions.

(a) Human E-selectin-IgG binding to WT and KO cells measured using flow cytometry (Mean Fluorescence Intensity). P2H3 is an anti-E-selectin blocking mAb. * and ^† P < 0.05 with respect to all other treatments except bars indicated by these symbols are not different from each other. Error bars are too small to be visible in some cases. (b) WT and KO cells interacting with immobilized E-selectin-IgG in microfluidic flow cells at a wall shear stress of 1 dyn/cm². Cells containing truncated N- and O-glycans (i.e. [ON]⁻ cells) were not captured from flow. * and ^† P < 0.05 for rolling and adherent cells respectively, with respect to WT cells. Statistics are not presented for anti-E-selectin blocking studies since cell adhesion was negligible in all cases. (c) Different cell types were captured under no-flow/static conditions for 2 min. prior to ramp-increase in wall shear stress starting with 1 dyn/cm² between 0–2 min. Detailed shear protocol is shown on right axis using dashed line. (d) Representative figure from experiment in panel c at 0 and 15 s. [NOG]⁻ cells were released immediately upon initiation of flow. [ON]⁻and WT cells continued to roll for longer times.

Figure 7. GlycoT KO cell rolling on E-selectin bearing stimulated HUVEC monolayers.

WT and KO HL-60s were perfused over IL-1β stimulated HUVECs at 1 dyn/cm². Both the density of rolling and adherent cells (panel a) and cell rolling velocity (panel b,c) were quantified. *P < 0.05 with respect to WT HL-60s for rolling cells. ^†P < 0.05 with respect to WT HL-60s for adherent cells. Statistical analysis results are not presented for anti-E-selectin blocking studies since cell adhesion is low in all cases. (d) Model for the contribution of different E-selectin ligands in the multistep leukocyte-endothelial cell adhesion cascade. Here, O- and N-glycans contribute to tethering. N-glycans regulate cell rolling velocity, with O-glycans potentially shielding some E-selectin binding. GSLs control slow rolling and the transition to firm arrest. The model does not account for biochemical processes regulating leukocyte activation.