Efficient inhibition of O-glycan biosynthesis using the hexosamine analog Ac5GalNTGc

Abstract

There is a critical need to develop small-molecule inhibitors of mucin-type O-linked glycosylation. The best-known reagent currently is benzyl-GalNAc, but it is effective only at millimolar concentrations. This article demonstrates that Ac₅GalNTGc, a peracetylated C-2 sulfhydryl-substituted GalNAc, fulfills this unmet need. When added to cultured leukocytes, breast cells, and prostate cells, Ac₅GalNTGc increased cell-surface VVA binding by ∼10-fold, indicating truncation of O-glycan biosynthesis. Cytometry, mass spectrometry, and western blot analysis of HL-60 promyelocytes demonstrated that 50-80 μM Ac₅GalNTGc prevented elaboration of 30%-60% of the O-glycans beyond the Tn-antigen (GalNAcα1-Ser/Thr) stage. The effect of the compound on N-glycans and glycosphingolipids was small. Glycan inhibition induced by Ac₅GalNTGc resulted in 50%-80% reduction in leukocyte sialyl-Lewis X expression and L-/P-selectin-mediated rolling under flow conditions. Ac₅GalNTGc was pharmacologically active in mouse. It reduced neutrophil infiltration to sites of inflammation by ∼60%. Overall, Ac₅GalNTGc may find diverse applications as a potent inhibitor of O-glycosylation.

Keywords: O-glycan; cell adhesion; glycosylation; inflammation; inhibitor; mucin; neutrophil; selectin; sialyl-Lewis X; small molecule.

Figure 1. Peracetylated HexNAc analogs.

A. Structures of per-acetylated GalNAc (Ac₄GalNAc) and its analogs. B. Per-acetylated GlcNTGc (Ac₅GlcNTGc), a C-4 epimer of peracetylated GalNTGc. C. Schematic showing the possible mechanism of Ac₅GalNTGc action. Ac₅GalNTGc or its derivatives inhibit GalNAc-type O-glycan biosynthesis.

Figure 2. Effect of HexNAc analogs on cell-surface glycans.

A-D. 0.5×10⁶ HL60 cells/mL were cultured with 50μM peracetylated GalNAc or GlcNAc analogs for 40h (VC=vehicle control). Cell-surface carbohydrate structures were measured using flow cytometry. Antigens measured include: A. CD15/Lewis-X (mAb HI98), B. Sialyl Lewis-X (sLe^X) like antigen called CLA or Cutaneous Lymphocyte Antigen (mAb HECA-452), C. CD15s/sLe^X (mAb CSLEX-1) and D. the sialofucosylated epitope CD65s measured using mAb VIM-2. Flow cytometry histograms present data for isotype control (dashed-empty), VC (solid-empty) and Ac₅GalNTGc (black-filled) treated samples. Bar plots present results for all analogs. *P<0.05 with respect to other treatments, except as indicated in panel D. E-F. Same studies as panel AD, only Ac₅GalNTGc concentration was titrated from 0-200μM. ‡ P<0.05 with respect to 0μM Ac₅GalNTGc. Ac₅GalNTGc decreased mAb HECA-452 and CSLEX-1 binding, and upregulated Le^X expression.

Figure 3. Lectin binding to GalNTGc treated HL60s.

A-F. HL60s cultured with 80μM Ac₅GalNTGc, Ac₅GlcNTGc or VC for 40h were stained with labeled lectins, either before (left panels) or after (right panels) treatment with α2-3,6,8,9 Arthrobacter Ureafaciens neuraminidase: A. VVA (binds GalNAcα-Ser/Thr); B. PNA (Galβ1,3GalNAc); C. ECL (Galβ1,4GlcNAc); D. MAL-II (α(2,3)sialic acid); and E. PHA-L (complex N-glycans). Epitope bound by lectins is depicted in each panel using the Symbol Nomenclature For Glycans. Mean fluorescence intensity/MFI±SD are shown in individual panels for VC (top number, black), Ac₅GalNTGc (second, green), and Ac₅GlcNTGc (third, red). The fourth number (blue) depicts the MFI of CRISPR-Cas9 COSMC-KO cells that have truncated O-glycans (panel A) and MGAT1-KOs that have truncated N-glycans (panel E). * P<0.05 with respect to VC and Ac₅GalNTGc. F. Fluorescence images of Ac₅GalNTGc and Ac₄GalNAc treated HL-60s, stained with VVA-FITC. Ac₅GalNTGc increased VVA-lectin binding indicating the presence of truncated mucin-type O-glycans. Ac₅GalNTGc also reduced T-antigen formation.

Figure 4.. Effect of Ac₅GalNTGc on cell adhesion.

HL60 cells cultured with 80μM Ac₅GalNTGc for 40h were perfused over substrates composed of: A. recombinant L-selectin, B. CHO-P cell monolayer bearing P-selectin, or C. IL-1β stimulated HUVEC monolayers bearing E-selectin. Wall shear stress was 1 dyne/cm² in all cases. The density of rolling and adherent cells was quantified. Blocking antibodies used were against: PSGL-1 (αPSGL-1 mAb KPL-1), L-selectin (αL-sel DREG-56), P-selectin (αP-sel G1) and E-selectin (αE-sel HAE1f). D. HL60s were shear-mixed with TRAP-6 activated human platelets using a cone-plate viscometer at 650/s. Flow cytometry quantified % of HL-60s bound to at least one platelet. E. Western blots of HL60 cell lysates cultured with Ac₅GalNTGc, Ac₅GlcNTGc and Ac₄GalNAc were probed with anti-human PSGL-1 antibody (mAb TB5) and anti-CD43 mAb L60. Ac₅GalNTGc reduced cell adhesion to P- and L-selectin. It also reduced the molecular mass of PSGL-1 by truncating mucin-type O-glycans. * P<0.05 with respect to all other treatments, at indicated time. † P<0.05 with respect to other treatments excepts †’s are not different from each other.

Figure 5. Murine acute inflammation.

A. Mouse bone marrow cells (mBMC) and neutrophils (mPMN) were isolated from 10-12 week C57BL/6 mice, and cultured with 50μM HexNAc analogs or controls for 40h. Cells were analyzed using flow cytometry and used in a murine acute inflammation model. B.-E. Flow cytometry measured the binding of the following reagents to mouse neutrophils (CD11b+, Gr-1/Ly-6G/1A8+, F4/80−): B. VVA-FITC, C. Anti-mouse PSGL-1 mAb 2PH1, D. P-selectin-IgG and E. L-selectin-IgG. Ac₅GalNTGc increased VVA-lectin binding by 4-5 fold, and reduced L-/P-selectin IgG binding by 50–70% without affecting PSGL-1 expression. F. mBMCs cultured with 80μM Ac₅GalNTGc for 40h were mixed with VC at 1:1 ratio. In mix 1, Ac₅GalNTGc cells were labeled with CMTMR (Red) while VC was CMFDA (Green) labeled. Labels were swapped in Mix 2 (dot plot not shown). Mix 1 or 2 cells were tailvein injected into recipient mice following thioglycollate injection i.p. Red:green ratio of Gr-1+ cells in the peritoneal lavage and bone marrow was measured at 20h. Ac₅GalNTGc reduced neutrophil counts in peritoneum by 50% in both Mixes. G-I. Ac₅GalNTGc (100mg/kg/day) or VC was injected daily into mice for 4 days prior to induction of peritonitis. Murine neutrophil (CD11b+, Gr-1/Ly-6G/1A8+, F4/80−) counts in the peritoneum were quantified at 16h. Neutrophil counts in peritoneal lavage was reduced by 65% in Ac₅GalNTGc treated mice (panel H). VVA binding was augmented in peritoneal neutrophils (panel I). *P<0.05 with respect to all other treatments in each panel, except as indicated.

Figure 6. Effect of Ac₅GalNTGc on O- and N-glycan biosynthesis:

A. HL-60s were cultured with 80μM Ac₅GalNTGc or Ac₄GalNAc for 16h prior to addition of peracetylated GalNAc-O-Bn (100μM) for an additional 48h. [NOG]⁻ TKO HL-60s cultured with peracetylated GalNAc-O-Bn served as negative control as they lack C1GalT1-chaperone COSMC activity. GalNAc-O-Bn and related products from cell culture supernatant were purified at the study end point, permethylated and analyzed using LC-MS/MS. Product ion-current area-under-the-curve quantified products formed. GalNAc-O-Bn consumption was reduced upon Ac₅GalNTGc treatment and abolished in TKO HL60s. B. Cell lysates were prepared from HL60s cultured with 80μM Ac₅GalNAc, Ac₅GalNTGc or Ac₄GalNAc for 40h. 3.3mg/mL lysate was mixed with 0.5mM GalNAc-O-Bn (substrate) and 1mM UDP-Gal donor overnight. β1,3GalT was quantified in LC-MS/MS runs by measuring ion current AUC (area under the curve) for product (Gal(β1-3)GalNAc-O-Bn) vs. unreacted substrate (GalNAc-O-Bn). Results are presented after normalization with respect to vehicle control (100%). Product was not formed in the absence of lysate (negative control). C. MALDI-TOF MS profile of permethylated N-glycans for cells treated with 80μM Ac₅GalNTGc or vehicle for 40h. Putative structures are based on composition, tandem MS, and knowledge of biosynthetic pathways. All molecular ions are [M+Na]⁺. MALDI data are representative of duplicate runs. Ac₅GalNTGc reduced T-antigen formation and O-glycan extension, without a major effect on N-glycan biosynthesis. * P<0.05 with respect to GalNAc treatment for each structure. † Glycan products were not detected.

Figure 7. Direct incorporation of thiol into Ac₅GalNTGc treated HL60 wild-type and knockout cells.

HL60s were cultured with peracetylated HexNAc analogs or VC (80μM, 40-48h). A. Sugar nucleotide levels in cells determined using LC-MS/MS. Abundance are quantified based on area under MS¹ curve for individual species, normalized with respect to vehicle control which is set to 100 (*P<0.05) B. UDP-HexNAc and UDP-HexNTGc abundance in GalNTGc (80μM, 48h) treated cells. Abundance are presented relative to CMP-Neu5Gc internal standard ionization in all runs. UDP-HexNTGc may exist in both UDP-GalNTGc and UDP-GlcNTGc forms. UDP-HexNTGc was not detected in vehicle, GalNAc and GlcNTGc samples. C. Cells were reacted with fluorescein-5-maleimide/5-FM, in the presence (left axis) or absence (right axis) of 10mM TCEP. A similar increase in 5-FM binding was noted in the GalNTGc treated cells under both conditions, although the signal was brighter following TCEP mediated reduction. D. Fluorescence microscopy showing the reaction of maleimide with free sulfhydryl groups predominantly on the cell surface. E. HexNAc-analogs incorporation time course. GalNTGc incorporation peaked 24-48h after compound addition. *P<0.05 with respect to other treatments. F. HexNAc analogs were cultured with wild-type HL-60s and a panel of isogenic CRISPR/Cas9 HL60-KO cell lines containing truncated O-glycans ([O]⁻), N-glycans ([N]⁻), GSLs ([G]⁻), dual knockouts ([ON]⁻, [OG]⁻, [NG]⁻) and triple knockouts ([NOG]⁻). 1 mg/ml pronase was added to some of these cells for 2 h to cleave cell-surface glycoproteins prior to reanalysis using the flow cytometer. 5-FM was incorporated into N-glycans, O-glycans and GSLs. All treatments in the GalNTGc samples were statistically different with respect to each other, except those indicated by ‘n.s.’ (not significant).