Metabolic precision labeling enables selective probing of O-linked N-acetylgalactosamine glycosylation

Abstract

Protein glycosylation events that happen early in the secretory pathway are often dysregulated during tumorigenesis. These events can be probed, in principle, by monosaccharides with bioorthogonal tags that would ideally be specific for distinct glycan subtypes. However, metabolic interconversion into other monosaccharides drastically reduces such specificity in the living cell. Here, we use a structure-based design process to develop the monosaccharide probe N-(S)-azidopropionylgalactosamine (GalNAzMe) that is specific for cancer-relevant Ser/Thr(O)-linked N-acetylgalactosamine (GalNAc) glycosylation. By virtue of a branched N-acylamide side chain, GalNAzMe is not interconverted by epimerization to the corresponding N-acetylglucosamine analog by the epimerase N-acetylgalactosamine-4-epimerase (GALE) like conventional GalNAc-based probes. GalNAzMe enters O-GalNAc glycosylation but does not enter other major cell surface glycan types including Asn(N)-linked glycans. We transfect cells with the engineered pyrophosphorylase mut-AGX1 to biosynthesize the nucleotide-sugar donor uridine diphosphate (UDP)-GalNAzMe from a sugar-1-phosphate precursor. Tagged with a bioorthogonal azide group, GalNAzMe serves as an O-glycan-specific reporter in superresolution microscopy, chemical glycoproteomics, a genome-wide CRISPR-knockout (CRISPR-KO) screen, and imaging of intestinal organoids. Additional ectopic expression of an engineered glycosyltransferase, "bump-and-hole" (BH)-GalNAc-T2, boosts labeling in a programmable fashion by increasing incorporation of GalNAzMe into the cell surface glycoproteome. Alleviating the need for GALE-KO cells in metabolic labeling experiments, GalNAzMe is a precision tool that allows a detailed view into the biology of a major type of cancer-relevant protein glycosylation.

Keywords: bioorthogonal; glycosylation; glycosyltransferase; mucin.

Fig. 1.

Design of an O-GalNAc–specific metabolic labeling reagent. (A) Rationale of probe design. UDP-GalNAc analogs that are not epimerized to the corresponding UDP-GlcNAc derivatives are O-GalNAc specific by design. Derivatives are delivered to the living cell by virtue of per-acetylated or phosphotriester-caged precursors. Compounds with a sterically congested diversification may be resistant to GALE-mediated epimerization but are accepted by GalNAc-Ts. Inset shows UDP-GlcNAc and UDP-GalNAc binding by GalNAc-T2 and GALE, respectively. (B) In vitro epimerization as assessed by ion-pair HPLC. Retention times of UDP-GalNAc analogs (yellow) and UDP-GlcNAc analogs (blue) are highlighted based on retention times of standards or epimerization reactions with 50-fold higher GALE concentration (SI Appendix, Fig. S1B). Arrowhead depicts epimerization of compound 3. Numbers are percentage epimerization as assessed by peak integration as means ± SD of three independent replicates or not detected (n.d.). Traces depict relative intensity of absorbance at 260 nm. Data are from one representative of three independent experiments and were reproduced using lysates of WT cells as a source of GALE or GALE-KO cells as a negative control in two independent replicates (SI Appendix, Fig. S1B). ATP, adenosine triphosphate; PDB, protein database identifier.

Fig. 2.

GalNAzMe can be used to label the cell surface glycoproteome. (A) Biosynthesis of UDP-GalNAzMe by mut-AGX1. HEK293T cells were transiently transfected with plasmids encoding for different AGX1 constructs or left nontransfected. Cells were fed with 200 μM compound 11 or Ac₄GalNAz, and cell lysates were analyzed by HPAEC-PAD. Inset, active site of WT-AGX1. (B) Cell surface labeling workflow using either CuAAC or SPAAC. (C) Dose dependence of GalNAzMe labeling by K-562 cells stably expressing WT-AGX1 or mut-AGX1, as assessed by flow cytometry. Data are mean ± SD from three independent replicates. (D) Cell surface mucin labeling by GalNAzMe and GalNAz. K-562 cells stably expressing WT-AGX1 or mut-AGX1 were fed with DMSO, 3 μM Ac₄GalNAz, or 100 μM compound 11 and treated with CF680-alkyne as outlined in B. Cells were optionally treated with 50 nM StcE before the click reaction. Data are from one representative of two independent experiments. (E) Cells were treated with either StcE or Vibrio cholerae sialidase and then treated with MB^TM 488-DIBAC as outlined in B, and glycosylation was assessed by flow cytometry. Data are mean + SD of three independent experiments. GAPDH, glyceraldehyde-3-phosphate dehydrogenase; FLAG, DYKDDDDK epitope tag; n.t., non-transfected; PDB, protein database identifier.

Fig. 3.

UDP-GalNAzMe is not epimerized and labeled a subset of the UDP-GalNAz–modified glycoproteome. (A) Schematic of the pathways probed herein. Both GalNAc-1-phosphate analog 11 and Ac₄GalNAz are precursors for O-GalNAc glycosylation. (B) UDP-GalNAzMe is not epimerized in the living cell, while UDP-GalNAz and UDP-GlcNAz are epimerized. K-562 GALE-KO and control cells stably transfected with mut-AGX1 were treated with 200 μM compound 11, Ac₄GalNAz, DMSO, or Ac₄GlcNAz, and UDP-sugar biosynthesis was assessed by HPAEC-PAD. (C) GALE-KO cells were treated with 100 μM compound 11, 10 μM Ac₄GalNAz, or 10 μM Ac₄GlcNAz and supplemented with GalNAc or GlcNAc in the indicated concentrations. Cell surface labeling was assessed by flow cytometry after SPAAC using MB^TM 488-DIBAC, and fluorescence intensity was normalized to DMSO-treated cells. Data are mean + SD from three independent experiments. (D) K-562 cells stably expressing WT- or mut-AGX1 were fed with DMSO, 100 µM compound 11, 3 µM Ac₄GalNAz, or 8 µM Ac₄GlcNAz and subjected to PEG mass tagging. K-562 cells stably expressing WT- or mut-AGX1 and GFP::CD47 were fed with DMSO, 100 µM compound 11, 3 µM Ac₄GalNAz, or 8 µM Ac₄GlcNAz and subjected to PEG mass tagging. (E) Cells were fed with compounds as in D, live cells were treated with CF680-alkyne under CuAAC conditions, and proteins in cell lysates were visualized by in-gel fluorescence. Ac₄ManNAz (0.5 μM) was used as a positive control. GAPDH, glyceraldehyde-3-phosphate dehydrogenase; FLAG, DYKDDDDK epitope tag; MFI, mean fluorescence intensity; C, control-sgRNA.

Fig. 4.

GalNAzMe is a reporter for the biology of O-GalNAc glycosylation. (A) GalNAzMe as a reporter in MS-based glycoproteomics of the HepG2 secretome. Exemplary mass spectra from GalNAzMe-containing glycopeptides. (B) GalNAzMe as a reporter for superresolution microscopy using K562 cells for labeling with GalNAzMe or GalNAz and CuAAC with Alexa Fluor 647 alkyne as a visualization strategy. (Scale bar, 10 µm.). Insets, whole cell images. (C) GalNAzMe as a reporter for a genome-wide CRISPR-KO screen in K-562 cells stably transduced with Cas9 and mut-AGX1 followed by feeding with Ac₄GalNAz or compound 11, labeled by MB^TM 488-DIBAC, and subjected to FACS to sort the bottom 15% fluorescent cells and sequence sgRNAs. Effects on selected glycogenes are shown—color depicts the relative phenotype (positive/red: enriched in the low-fluorescence population; negative/blue: depleted in the low-fluorescence bottom population), while asterisks depict false discovery rate (FDR) as a measure of statistical significance from two independent experiments. ETD, electron-transfer dissociation; HCD, higher-energy collisional dissociation; ns, nonsignificant. *FDR 5%; **FDR 2%.

Fig. 5.

An engineered BH-T2 double mutant enhances GalNAzMe labeling. (A) The principle of BH engineering using engineered GalNAc-T2 (BH-T2) that preferentially accommodates UDP-GalNAzMe. (B) In vitro glycosylation using WT- or BH-T2 as enzyme sources. UDP-GalNAz 2 and UDP-GalNAzMe 5 were used as substrates, and UDP-GalNAc 1 was used as a competitor at different concentrations. Azide-labeled glycoproteins were visualized as in Fig. 2B. Data are from one representative of two independent replicates. (C) Live cell surface glycosylation by K-562 cells stably transfected with mut-AGX1 and WT- or BH-T2 and fed with DMSO, 50 μM compound 11, or 3 μM Ac₄GalNAz. Data are from one representative of two independent replicates. (D) Glycosylation in intestinal organoids transfected with mut-AGX1 and BH-T2. Organoids were fed with 50 µM compound 11 or 1.5 µM Ac₄GalNAz, fixed, and treated with biotin alkyne under CuAAC conditions followed by streptavidin Alexa Fluor 647 staining. Data are from one representative of two independent experiments and shown as grayscale images for each channel and a color merge image of all three channels. (Scale bar, 100 µm.) Insets, magnifications. DAPI, 4′,6-diamidino-2-phenylindole.