Bacterial glycosyltransferase-mediated cell-surface chemoenzymatic glycan modification

Abstract

Chemoenzymatic modification of cell-surface glycan structures has emerged as a complementary approach to metabolic oligosaccharide engineering. Here, we identify Pasteurella multocida α2-3-sialyltransferase M144D mutant, Photobacterium damsela α2-6-sialyltransferase, and Helicobacter mustelae α1-2-fucosyltransferase, as efficient tools for live-cell glycan modification. Combining these enzymes with Helicobacter pylori α1-3-fucosyltransferase, we develop a host-cell-based assay to probe glycan-mediated influenza A virus (IAV) infection including wild-type and mutant strains of H1N1 and H3N2 subtypes. At high NeuAcα2-6-Gal levels, the IAV-induced host-cell death is positively correlated with haemagglutinin (HA) binding affinity to NeuAcα2-6-Gal. Remarkably, an increment of host-cell-surface sialyl Lewis X (sLe^X) exacerbates the killing by several wild-type IAV strains and a previously engineered mutant HK68-MTA. Structural alignment of HAs from HK68 and HK68-MTA suggests formation of a putative hydrogen bond between Trp222 of HA-HK68-MTA and the C-4 hydroxyl group of the α1-3-linked fucose of sLe^X, which may account for the enhanced host cell killing of that mutant.

Fig. 1

Recombinant bacterial FTs and STs for live cell-surface glycan modification. a Specific positions on mammalian cell-surface LacNAc(Galβ1-4-GlcNAc)-containing glycans that can potentially be modified by fucosylation (α1-2- or α1-3-linked) and sialylation (α2-3- or α2-6-linked). Recombinant bacterial glycosyltransferases (FTs and STs) used in this study include Hm1,2FT, Hp1,3FT, Pm2,3ST-M144D, and Pd2,6ST. b Analysis of in vitro sialylation products by TLC. ++ indicates the final reaction system was further mixed with starting material LacNAc, and analyzed by TLC. c Analysis of in vitro fucosylation products by TLC. d, e Analysis of in vitro products generated by a combination of sialylation and fucosylation by TLC. sLe^X was formed by combining Hp1,3FT and Pm2,3ST-M144D (d). NeuAcα2-6-(Fucα1-2)-LacNAc was formed by combining Hm1,2FT and Pm2,3ST-M144D (e). f, g Analysis of newly formed glycan epitopes on the cell-surface of Lec2 CHO cells via chemoenzymatic glycan modification. Modified cells were stained by lectins and analyzed by flow cytometry. h, i Evaluation of the substrate tolerance of bacterial sialyltransferases. Unnatural sugar CMP-SiaNAz bearing the azide group were tested for STs. In figures f–i, error bars represent the standard deviation of three biological replicates. ** indicated Welch’s t-test P < 0.01. Source data for figures b–i are provided as a Source Data file

Fig. 2

One-step glycan labeling enabled by recombinant bacterial glycosyltransferases. The Pm2,3ST-M144D, Pd2,6ST, or Hp1,3ST-mediated incorporation of unnatural sugars conjugated to a fluorescent dye (Cy3) or an affinity tag (biotin), enabled a One-step cell-surface glycan labeling. a Nucleotides and analogs functionalized with biotin tag (CMP-SiaNAz-biotin and GDP-FucAz-biotin) or with Cy3 florescent dye (CMP-SiaNAz-Cy3 and GDP-FucAz-Cy3). b Direct STs-catalyzed conjugation of Cy3 (magenta) for imaging of live cell glycans. c Hp1,3FT-catalyzed conjugation of Cy3 (magenta) for imaging of live cell glycans. In b and c, cells were visualized by bright field images and DAPI staining (blue). Scale bar, 20 μm. d Time-dependence of activities of recombinant bacterial and human STs for cell-surface glycan labeling with CMP-SiaNAz-biotin. e Activity of Hp1,3FT using GDP-FucAz-biotin to conjugate biotin onto live cell-surface glycan directly. In d and e, error bars represent the standard deviation of three biological replicates. f, g Enzyme-assisted incorporation of biotin was mainly on N-linked glycans on CHO cells and CHO Lec2 cells, while CHO mutant Lec8 cells without LacNAc were not labeled. Protein loading was depicted by Coomassie blue staining or anti-tubulin western blot. Source data for figures d–g are provided as a Source Data file

Fig. 3

One-step recombinant bacterial STs-based labeling of glycans in tissue specimens. The embryonic frozen sections from E16 mouse were incubated with STs (Pm2,3ST-M144D or Pd2,6ST) or without STs, and CMP-SiaNAz-biotin, followed by Alexa Fluor 594-streptavidin conjugate staining. The resulting fluorescence (red) of different parts of the embryo was directly imaged using microscopy, including salivary glands region, lateral sections of spine, and anterior chest. The cells of frozen sections were stained with anti-actin (green) and DAPI (blue, nucleus). Scale bar, 1 mm

Fig. 4

A cell-based glycan array to assess HA–glycan interactions directly on live cells. a Profiling glycoforms of lung tissues obtained from healthy human donors. Lung tissue slides were stained with FITC-AAL, AF647-anti-CLA, Biotin-MAA, or Biotin-SNA conjugates to detect α1-3-fucosylation, sLeX epitopes, α2-3-linked, or α2-6-linked sialylation, respectively. b Major glycan epitopes presented on Lec2 cell-surface after chemoenzymatic glycan modification. CHO Lec2 cells were treated with glycosyltransferases indicted above and the corresponding nucleotide sugars. *indicates the potential modification site for the first-step glycan modification (black), and the second-step glycan modification (gray). c Relative binding affinity of HA from HK68 (H3N2) for glycan-modified Lec2 cells using the specified recombinant glycosyltransferases. In Fig. 4c, the error bars represent the standard deviation of six biological replicates. Source data for figure c are provided as a Source Data file

Fig. 5

Profiling IAV infection using glycocalyx-modified MDCK cells. a Modification of glycocalyx of MDCK cells using Pm2,3ST-M144D, Pd2,6ST, or Hp1,3ST and the corresponding donor substrate conjugated with biotin. Biotinylated cells were probed with Alexa Fluor 647-Streptavidin. b Modification of glycocalyx of MDCK cells using a combination of Pm2,3ST-M144D and Hp1,3ST. Newly generated sLe^X on the MDCK cell surface was confirmed by Alexa Fluor 647-anti-CLA conjugate staining. c Viability of Sia-edited MDCK cells or control cells upon infection by HK68. d Viability of Fuc-edited MDCK cells or control cells infected by HK68. e–i Viability of glycan (Sia or Fuc) edited MDCK cells or control cells upon infection by Aichi68 (e), Perth09 (f), WSN (g), PR8 (h), and SI06 viruses. j–l Viability of glycan edited MDCK cells or control cells upon infection by HK68, using analogs of CMP-Sia (j) or GDP-Fuc (k). Viability of Fuc-edited MDCK cells or control cells, at 10⁻⁴ virus dilution (l). In figures a and b, the error bars represent the standard deviation of three biological replicates. In c–l, the error bars represent the standard deviation of six biological replicates. Source data are provided as a Source Data file

Fig. 6

Profiling the structural constraints of IAV-HA for glycan binding. The activities of wild-type HK68 and its hemagglutinin-receptor-binding site mutants to infect Sia- or Fuc-edited host cells, were directly compared via host-cell killing. a–d Viability of Sia-edited MDCK cells or control cells upon infection by wild-type HK68 and its HA-RBS mutants, including HK68-MTA (a), HK68-LSS (b) and HK68-QAS (c). d Cell viability at 10⁻³ viral dilution. e–i Viability of Fuc-edited MDCK cells or control cells upon infection by wild-type HK68 and its HA-RBS mutants. Cell viability at 10⁻³ virus dilution (h), and at 10⁻⁴ virus dilution (i). j Structural alignment of HAs from HK68 and HK68-MTA. A minor shift of 220-loop backbone of HK68-MTA enables formation of a H-bond between C4 hydroxyl of α1-3-linked fucose of sLe^x and Nε1 of W222 (Fig. 6I), which is not observed between the HK68-WT HA and sLe^X. In Fig. 6A-I, the error bars represent the standard deviation of six biological replicates. Source data for figures a–i are provided as a Source Data file