Interferon-stimulated gene GALNT2 restricts respiratory virus infections

Abstract

The innate immune response involves interferons (IFNs), antiviral cytokines that upregulate numerous IFN-stimulated genes, many of which have uncharacterized functions and mechanisms. Here we performed transcriptomic profiling of lung tissues from wild-type and IFNAR^-/- mice infected with SARS-CoV-2 and single-cell RNA sequencing of bronchoalveolar lavage fluid and peripheral blood mononuclear cells from patients with COVID-19. We identified O-GalNAc transferase 2 (GALNT2), an N-acetylgalactosaminyltransferase, as an antiviral IFN-stimulated gene restricting the replication of multiple coronaviruses and influenza A viruses in vitro and in vivo, contributing to viral clearance and reducing disease severity. Mechanistically, GALNT2-dependent O-linked glycosylation may regulate viral glycoprotein proteolytic processing and impair viral growth by blocking virus-cell fusion. In addition, we found that serine residues at 810/813 in the viral spike protein undergo O-glycosylation and function as the primary genetic determinants of sensitivity or evasion towards GALNT2. Human genetic data analysis revealed that individuals with GALNT2 variants that lost antiviral function had elevated risk of hospitalization following SARS-CoV-2 infection. This study establishes GALNT2 as an antiviral factor against some respiratory virus infections.

Fig. 1. GALNT2 deficiency associated with increased and prolonged SARS-CoV-2 infection.

a, A schematic overview of the scRNA-seq design. The composition and characteristics of PBMC and BALF patient samples included in the analysis are shown in Extended Data Fig. 2a–c. b, The uniform manifold approximation and projection (UMAP) of the cells from the BALF of HCs (n = 28), patients with moderate COVID-19 (n = 5) and patients with severe COVID-19 (n = 59). Each dot represents a cell, and different cell types are shown in different colours. Canonical markers for different cell types, including T, NK, B and ciliated cells in BALF are shown in dot plots. c, Comparison of GALNT2 expression in pseudo-bulk (left panel) and ciliated cells (right panel) of BALF samples using scRNA-seq. The box plots show the median (centre line) and the 25th and 75th percentiles (bounds of the box), and the whiskers extend to the smallest and largest values within 1.5 times the interquartile range from the box. Each dot represents an individual, and the y-axis represents the average expression level of GALNT2 across all cells or specifically in ciliated cells from that individual. Pearson correlation coefficients and P values by two-tailed test are indicated. The P values were calculated by two-tailed Wilcoxon signed-rank test. The pseudo-bulk analysis provides an overall expression trend across all cells, while the cell-type-specific analysis highlights distinct regulation in ciliated cells. d, Comparison of GALNT2 dynamic expression in PBMC samples using scRNA-seq. Each open dot represents an individual, and the y-axis represents the average expression level of GALNT2 in a specific cell type from that individual. A solid dot represents the average expression. The sampling time points are denoted as T1 (initial infection period), T2 (conversion phase) and T3 (recovery, after negative testing for SARS-CoV-2 virus). Samples from the same individual are connected by dashed lines. The P values were calculated by two-tailed Student’s t-test. e,f, Analysis of genetic variation at the GALNT2 gene locus between SARS-CoV-2-infected patients and control populations using data from the RGC COVID-19 results (e). The forest plot shows the risk of COVID-19 associated with GALNT2 SNP rs76000797 (f). P values were calculated by two-tailed Firth logistic regression test, with appropriate covariates and genomic control. Error bars indicate the 95% confidence interval. See Methods for details. Source data

Fig. 2. GALNT2 is an ISG with broad antiviral activity.

a, Huh7 cells were transfected with plasmids encoding GALNT2 (EV and eGFP as the controls). At 24 h post-transfection, cells were infected with CoVs including SARS-CoV-2 WT and Omicron BA.5 at an MOI of 0.01, HCoV-OC43 at an MOI of 0.05 and HCoV-229E at an MOI of 0.001. The viral titres were detected at 24, 48 and 72 hours post infection (h.p.i.) by FFA. Data are means ± s.d. from n = 3 independent experiments. P values are from two-tailed Student’s t-test. b, A549 cells were transfected with plasmids encoding GALNT2 (EV and eGFP as the controls). At 24 h post-transfection, cells were infected with IAVs at an MOI of 0.01, including H1N1 (PR8 and CA04), H3N2 (HK68) and H9N2 (G1_PB2_E627K). Viral titres were determined in MDCK cells at indicated time points by FFA. Data are means ± s.d. from n = 3 independent biological repeats. P values are from two-tailed Student’s t-test. c, Huh7 (Ctrl), Huh7 GALNT2^−/− or Huh7 STAT1^−/− cells were infected with SARS-CoV-2 (MOI = 0.05, 24 h), HCoV-OC43 (MOI = 0.05, 48 h) and HCoV-229E (MOI = 0.001, 24 h). Viral infection efficiency was detected by IFA. A549 (Ctrl) and A549 GALNT2^−/− cells were infected with H1N1 pdmCA04 (MOI = 0.03, 24 h), H3N2 HK68 (MOI = 0.01, 24 h) and H9N2 G1_PB2_E627K (MOI = 0.01, 24 h). Viral infection efficiency was detected by IFA. Reconstitution of GALNT2 in the knockout cells by transient transfection and infection efficiency was determined by IFA. Data are means ± s.d. from n = 3 or 6 independent biological repeats. P values were from two-tailed Student’s t-test. d, Huh7 cells infected with WT SARS-CoV-2 (MOI = 0.1) for 0 h (mock), 24 h and 48 h. GALNT2 expression levels were determined by western blot analysis. e, THP-1 cells were treated with 1 μg ml⁻¹ IFNα2, IFNβ or IFNλ. At 24 h post-treatment, GALNT2 expressions in cell lysates were analysed by western blot. f, Western blot analysis of the mouse GALNT2 and STAT1 expressions in SARS-CoV-2 WT and IFNAR^−/− mouse lungs at 4 d.p.i. Western blots (d–f) were representative of three independent biological repeats. NS, not significant. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001. Source data

Fig. 3. GALNT2 protects mice from SARS-CoV-2 and IAV infection in vivo.

a, WT (n = 12) and GALNT2^−/− (n = 10) C57BL/6 mice were infected with 350 FFU of PR8 H1N1 and monitored for 14 days for weight changes and survival. Two independent experiments were performed. Viral titres (n = 5) in the lungs collected at 3 d.p.i. were determined in MDCK cells by FFA. b, WT (n = 6) and GALNT2^−/− (n = 5) C57BL/6 mice were infected with 5 × 10³ FFU of SARS-CoV-2. Viral titres in the lungs collected at 2 d.p.i. were determined in Vero E6 cells by FFA. c, Groups of C57BL/6 mice were transduced with either 3.3 × 10¹¹ genome copies (GC) of AAV9-EV (n = 11) or AAV9-GALNT2 (n = 10) intranasally. The mice were infected with 1,000 FFU of H1N1 PR8 14 days post-transduction and were monitored for 14 days for weight changes and survival. Two independent experiments were performed. Viral titres in the lungs collected at 3 d.p.i. (n = 6) and 5 d.p.i. (n = 5 or 6) were determined in MDCK cells by FFA. d, Groups of C57BL/6 mice were intranasally transduced with 3.3 × 10¹¹ GC of AAV9-EV or AAV9-GALNT2 (n = 7). The mice were infected with 1 × 10⁵ FFU of SARS-CoV-2 (WT) 14 days post-transduction. Weight changes were monitored for 14 days. Viral titres (n = 6) in the lungs collected at 2 d.p.i. and 4 d.p.i. were determined in Vero E6 cells by FFA. e, Groups of IFNAR^−/− (n = 5) or STAT1^−/− (n = 6) mice were transduced intranasally with 3.3 × 10¹¹ GC of AAV9-EV or AAV9-GALNT2. The mice were infected with 1 × 10⁵ FFU of SARS-CoV-2 (WT) 14 days post-transduction. Viral titres in the lungs collected at 2 d.p.i. were determined in Vero E6 cells by FFA. Data are mean ± s.d. P values were from one-tailed Student’s t-test in a, b and d and two-tailed Student’s t-test in c and e. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. Source data

Fig. 4. GALNT2 catalyses the O-linked glycosylation of SARS-CoV-2 spike and IAV HA.

a, MALDI-TOF spectrum analysis of GalNAcylation reactions catalysed by purified GALNT2 (GalNAc-T2) on the synthetic multibasic peptide of the SARS-CoV-2 spike protein (S2′ (TMPRSS2): 800–819, S1/S2 (furin): 674–693). An increase of 203 Da corresponding to the addition of one GalNAc residue was indicated in the site of O-glycosylation by GALNT2. b,c, ETD-MS² spectrum analysis of the O-GalNAcylated peptide from the GALNT2 reaction on the S810/S813-O-GalNAcylated peptide. The mass of c- and z- fragment ions unambiguously assigned the GalNAc modification to S810 (b) and S813 (c) (labelled with a yellow square above the peptide sequence). The GalNAc residues are denoted as yellow squares according to the Consortium for Functional Glycomics standard. d, Full-length SARS-CoV-2 spike was co-transfected in HEK293T cells with GALNT2 and analysed with LC–MS/MS as described above. e, A schematic showing the GALNT2-dependent O-glycosylated sites on the WT SARS-CoV-2 spike. NTD, N-terminal domain; RBD, receptor binding domain; HR1, heptad repeat 1; CH, central helices; CD, connector domain; TM, transmembrane. f, A schematic showing the biotin-based pulldown of an O-glycosylated spike from cell lysates using helix pomatia agglutinin (HPA). g, Immunoblots of spike obtained from HPA pulldown of the HEK293T cell lysates. Western blots (WB) were representative of three independent biological repeats. Quantitative data from the representative western blot are shown below. h, Immunoblots of spike obtained from HPA pulldown of the Calu3 cell lysates overexpressing 100 ng or 400 ng EV or GALNT2, infected with authentic WT SARS-CoV-2 (MOI = 0.1) for 24 h. Western blots were representative of three independent biological repeats. Quantitative data from representative western blot are shown below. i, MALDI-TOF analysis of GalNAcylation reactions catalysed by purified GALNT2 on the synthetic multibasic peptide of H1N1 HA proteins. j, Immunoblots of HA protein, obtained from HPA pulldown of the HEK293T cell lysates co-expressing HA and EV or GALNT2. Western blots were representative of three independent biological repeats. Quantitative data from a representative western blot are shown below. k, Immunoblots of HA protein, obtained from HPA pulldown of the Calu3 cell lysates overexpressing 100 ng EV or GALNT2, infected with PR8 H1N1 (MOI = 0.1) for 24 h. Western blots were representative of three independent biological repeats. Quantitative data from representative western blot are shown below. Source data

Fig. 5. GALNT2 inhibits proteolytic processing of viral glycoproteins and blocks virus–cell fusion.

a–c, Immunoblots of the SARS-CoV-2 full-length S, cleaved S2′, derived from HEK293T (a) and HEK293T-TMPRSS2 (b) cell lysates, co-transfected with plasmids encoding a spike (S) of SARS-CoV-2 or ΔRRAR spike mutant (c), ACE2, empty vector (EV) or GALNT2 for 24 h. Western blots were representative of three independent biological repeats. Quantitative data from a representative western blot are shown below. d, A schematic showing the Cre-loxP-based fusion assay. Fluorescent images of the mCherry reporter expression and cell nuclei (Hoechst33342), captured from spike and Cre co-expressing HEK293T cells in the presence of EV or GALNT2, co-cultured with control or ACE2 cells with stop-mCherry for 24 h. Images were representative of 3 individual repeats. Scale bars, 50 μm. Cell–cell fusion was quantified as RLUs, derived from co-cultured HEK293T cell lysates using a stop-luciferase reporter. Data are means ± s.d. from n = 14 repeats. P values were from two-tailed Student’s t-test. e, Immunoblots of the SARS-CoV-2 full-length S and S2 collected from the supernatant and lysates of HEK293T cells SARS-CoV-2 pseudovirus preparation. The spike was co-expressed with EV or GALNT2 for 24 h, followed by infection of VSV pseudovirus for 24 h. Supernatant and cell lysates were collected and detected by western blots. Western blots were representative of three independent biological repeats. f, Luciferase activity detected from HEK293T-AEC2 cells infected with SARS-CoV-2 pseudovirus, prepared from EV or GALNT2-expressing cells. Data are means ± s.d. from n = 4 independent biological repeats. P values from two-tailed Student’s t-test. g, Immunoblots of S and S2 on SARS-CoV-2 VLP supernatants, prepared from HEK293T cells co-expressing spike, envelope (E), membrane (M), nucleocapsid (N) and combined with EV, GALNT2 for 72 h. Western blots were representative of three independent biological repeats. h, A schematic showing the quantification of SARS-CoV-2 VLP entry into HEK293T-ACE2 cells overexpressing the LgBiT NanoLuc reporter. RLUs were derived from HEK293T-ACE2 LgBiT cells infected with SARS-CoV-2 VLPs for 8 h. Data are means ± s.d. from n = 3 individual repeats. P values were from two-tailed Student’s t-test. i, Immunoblots showing full-length PR8 HA and cleaved HA1 and HA2 expression in HEK293T cells co-transfected without or with TMPRSS2 and GALNT2 for 24 h. Blots are representative of three individual repeats. The red arrow indicates the cleaved HA1 and HA2. j,k, GALNT2 inhibits the fusogenic activity of the HA protein from the avian influenza virus strain H5N6 (j) and H7N9 (k) with polybasic cleavage sites. Fluorescent images of the GFP reporter expression captured from HA and GFP co-expressing HEK293T cells in the presence of EV or GALNT2 for 24 h. Images were representative of 3 individual repeats. Scale bars, 50 μm. Individual syncytium derived from HA-mediated cell–cell fusion was quantified as relative fluorescent dots. Data are means ± s.d. from n = 20 repeats. P values from two-tailed Student’s t-test. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001. Source data

Fig. 6. The primary targets of GALNT2 against SARS-CoV-2 infection.

a, The temporal dynamics of SARS-CoV-2 S813 site mutation frequency (https://outbreak.info/situation-reports?muts=S:S813I&pango). b, A schematic showing the generation of SARS-CoV-2 recombinant viruses and confirmed by next generation sequencing (NGS). c, The recombinant viruses were amplified in Vero E6 cells, and titres were determined by PFA. Images were representative of viral plaque forming for recombinant viruses. Data are means ± s.d. from (n = 50 for each virus) the diameter of plaques. P values are from two-tailed Student’s t-test. d,e, Vero E6 cells (d) and Huh7 cells (e) were infected with SARS-CoV-2 WT and recombinant viruses at an MOI of 0.01. Viral titres were detected at indicated time points by PFA. Data are means ± s.d. from n = 3 or 4 independent experiments. P values were from two-tailed Student’s t-test. f, Huh7 cells were transfected with plasmids encoding GALNT2 (EV as the control). At 24 h post-transfection, cells were infected with SARS-CoV-2 WT and the recombinant viruses. Viral titres were detected at indicated time points by PFA. Data are means ± s.d. from n = 3 independent experiments. P values were from two-tailed Student’s t-test. g, Groups of C57BL/6 mice (n = 5) were intranasally transduced with 3.3 × 10¹¹ GC of AAV9-EV or AAV9-GALNT2. Mice were infected with 5 × 10⁴ plaque-forming unit (PFU) of SARS-CoV-2 (WT, S810/813A, S810/813I) 14 days post-transduction. Viral titres in the lungs collected at 2 d.p.i. and 4 d.p.i. were determined in Vero E6 cells by PFA. Data are presented as means ± s.d. h, MALDI-TOF analysis of GalNAcylation reactions catalysed by purified GALNT2 on the synthetic multibasic peptide of SARS-CoV-2 spike proteins (WT, S810/813A and S810/813I). Reactions were performed and analysed as described in Methods. An increase of 203 Da corresponds to an addition of one GalNAc residue observed with GALNT2 in WT spike protein. i, Streptavidin pulldown of HPA–biotin mixed with SARS-CoV-2 pseudoviral particles incorporated with SARS-CoV-2 WT or S810/813A spikes. Supernatants containing the SARS-CoV-2 pseudoviral particles derived from EV or GALNT2-expressing HEK293T cells were collected and verified for SARS-CoV-2 spike incorporation and assembly of VSV-M. Normalized SARS-CoV-2 pseudovirus supernatants were gently mixed with 5 μg ml⁻¹ biotin–HPA overnight at 4 °C. Pulldown samples attached on the streptavidin magnetic beads were washed three times in NP-40 lysis buffer before boiling in the 2× Laemmli loading buffer. Protein samples were then separated by SDS-PAGE and detected by immunoblotting of the full-length spike and S2. Western blots were representative of three independent biological repeats. Quantitative data from representative western blot are shown below. j, Fluorescent images of HEK293T control or HEK293T-ACE2 cells carrying the stop-mCherry reporter, co-cultured with cells expressing spike (SARS-CoV-2 WT or S810/813I mutant) co-expressed with EV or GALNT2. Images are representative of three individual repeats. Scale bars, 20 μm. All data are mean ± s.d. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001. Panel b created with BioRender.com. Source data

Extended Data Fig. 1. Impaired type I IFN response lead to increased and prolonged SARS-CoV-2 infection in vivo.

a, A schematic showing the mouse infection model of SARS-CoV-2 by exogenous delivery of human ACE2 with a replication-deficient adenovirus (Ad5-hACE2). b, IFNAR-/- (n = 3) and WT (n = 3) were intranasally transduced with 2.5×108 FFU of Ad5-hACE2. Five days post transduction of Ad5-hACE2, mice were infected with 1×105 PFU of SARS-CoV-2 (WT). The lungs were harvested at 2 days post infection and performed mRNA-seq. The transcriptional profiling of homogenized lungs are expressed as normalized read counts to WT mice. c, The genes showed in (b) were homo sapiens codon-optimized and synthetized de novo into pcDNA3.1 vector. Huh7 cells were transfected with plasmids encoding these genes (EV and eGFP as the controls). At 24 h post transfection, cells were infected with SARS-CoV-2 (WT, MOI = 0.05) for 48 h. Viral infection efficiency was determined by IFA. Data were normalized to EV. Data are means ± s.d. from n = 3 independent experiments. d, Huh7 cells were transfected with plasmid encoding GALNT2 (EV as the control), and GALNT2 protein expression level was detected at 24 h post transfection by western blots. Representative IFA images that GALNT2 inhibits SARS-CoV-2 infection in Huh7 cells. SARS-CoV-2 N was stained using indirect fluorescent antibody (green), nuclei are stained with DAPI (blue). Western blots and IFA images were representative of three independent biological repeats. e, GALNT2 and STAT1 protein levels of WT and IFNAR-/- mice (n = 3) at 2, 4, and 6 days post infection were determined by western blots. f, Related to (e), comparison viral titres and GALNT2 protein levels of WT mice at 2, 4, and 6 d.p.i. n = 3 mice per group. GALNT2 protein levels were normalized to that in WT mice (4 d.p.i.). g, Related to (e), comparison viral titres and GALNT2 protein levels of IFNAR-/- mice at 2, 4 and 6 d.p.i. n = 3 mice per group. GALNT2 protein levels were normalized to that in WT mice (4 d.p.i.). h, Huh7 cells were transfected with plasmids encoding GALNT2-HA (EV as the control) for 36 h. Cells were then fixed with 4% paraformaldehyde and stained using Anti-HA-tag antibodies conjugated with Allophycocyanin (APC). Cells were analyzed by IFA. Scale bars are indicative of 20 μm. i, Huh7 and HEK293T cells were transfected with plasmids encoding EV, eGFP and GALNT2-HA for 24 h. GALNT2 expression levels were detected by flow cytometry and data analysis was performed using the FlowJo 10.4 (LCC). Data are means ± s.d. from n = 3 independent experiments. j, Huh7 cells were transfected with plasmids encoding GALNT2 (EV and eGFP as the controls. At 24 h post transfection, the cell viability was measured by CCK-8 and normalized to Huh7 WT cells. Data are means ± s.d. from n = 12 independent experiments. k, Huh7 cells were transfected with plasmids encoding GALNT2 (EV and eGFP as the controls. At 24 h post transfection, cells were seeded in 24-well plates at 1×104 cells/well. Cell growth is counted at indicated time points. Data are means ± s.d. from n = 3 independent experiments. P values from two-tailed Student’s t-test. Source data

Extended Data Fig. 2. Single-cell transcriptional profiling of BALF and PBMCs from HDs and SARS-CoV-2 infected patients.

a, Canonical markers for different cell types in BALF were shown in UMAP and dot plots. b, The UMAP plots were split by different studies. c, Comparison of GALNT2 expression in BALF samples using scRNA-seq. Each dot represents an individual, and the y-axis represents the average expression level of GALNT2 in a specific cell type from that individual. d, Comparison of GALNT2 expression in Myeloid, T and B, ciliated and secretory cells of BALF samples using scRNA-seq. Each dot represents an individual, and the y-axis represents the average expression level of GALNT2 across all cells or specifically in ciliated cells from that individual. The box plots show the median (center line), the 25th and 75th percentiles (bounds of the box), and the whiskers extend to the smallest and largest values within 1.5 times the interquartile range from the box. The p values were calculated by two-tailed Wilcoxon signed-rank test. ((ns): no significant, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001). e, Protein expression of GALNT2 in the PBMC samples at conversion phase (T2) from mild and severe patients were detected by western blots. Western blots were representative of three independent biological repeats. Quantitative data from representative western blot is shown. Data are means ± s.d. from n = 3 independent experiments. P values from two-tailed Student’s t-test. ((ns): no significant, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001). Source data

Extended Data Fig. 3. Expression and regulation of GALNT2 and classical ISGs across tissues and COVID-19 severity groups.

a, RNA expression patterns of OAS1, MX1, and GALNT2 across human tissues (data from the Human Protein Atlas; https://www.proteinatlas.org), visualized as z-score normalized heatmaps (red: higher expression, blue: lower expression). Scatter plots below show pairwise expression correlations between genes, with locally estimated scatterplot smoothing (LOESS) fits (blue line) and 95% confidence intervals (gray shading). Pearson correlation coefficients and p-values by two-tailed test are indicated. b, Protein-level expression of OAS1, MX1, and GALNT2 across human tissues, from the Human Protein Atlas. Expression levels are categorized as not detected(grey), low (light blue), medium (blue), or high (dark blue) by HPA database. c, Left: Comparison of OAS1 expression in ciliated epithelial cells from bronchoalveolar lavage fluid (BALF) across COVID-19 severity groups based on scRNA-seq data. The box plots show the median (center line), the 25th and 75th percentiles (bounds of the box), and the whiskers extend to the smallest and largest values within 1.5 times the interquartile range from the box. Each dot represents an individual; the y-axis shows average OAS1 expression per individual. Right: Longitudinal OAS1 expression in peripheral blood mononuclear cell (PBMC) subsets (B cells, T cells, and monocytes) across healthy controls (HC) and patients with mild or severe COVID-19. Open dots represent individual-level averages; solid dots represent group means. Time points: T1 (acute phase), T2 (conversion), T3 (recovery; see Methods for definitions). Samples from the same individual are connected by dashed lines. P values were calculated using two-tailed Wilcoxon signed-rank test (left) or two-tailed Student’s t-test (right). d, Comparison of MX1, SOCS1, STAT1, and USP18 expression in ciliated BALF cells across clinical severity groups. Each dot represents one individual; the y-axis shows average gene expression. The box plots show the median (center line), the 25th and 75th percentiles (bounds of the box), and the whiskers extend to the smallest and largest values within 1.5 times the interquartile range from the box. P values by two-tailed Wilcoxon signed-rank test. e, Summary of differential protein expression for the indicated genes across tissues in SARS-CoV-2 infected patients versus controls, based on published proteomics data65. Red indicates significant upregulation (FDR < 0.05); gray, not significant. ((ns): no significant, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001). Source data

Extended Data Fig. 4. GALNT2 associated with the severity or susceptibility of SARS-CoV-2 and IAV infections.

a, GALNT2 association results in influenza A virus H7N9 genome-wide association studies (GWAS). Mx1 gene (left panel) as the control. Red plots represent the SNPs associated with virus infection. b, GALNT2 association results in influenza A virus H1N1 genome-wide association studies (GWAS). Mx1 gene (left panel) as the control. Red plots represent the SNPs associated with virus infection. P values from (a) and (b) were calculated by two-tailed Firth logistic regression test, with appropriate covariates and genomic control. c, Analysis of genetic variation at the GALNT2 gene locus between SARS-CoV-2 infected patients and control populations using data from Host Genetics Initiative (https://www.covid19hg.org/). Population frequency information for the SNP was sourced from https://www.ncbi.nlm.nih.gov/snp/rs2273970. P values were calculated by two-tailed Firth logistic regression test, with appropriate covariates and genomic control. d, Huh7 cells were transfected with plasmids encoding GALNT2 (WT or four mutants), and cells then infected with SARS-CoV-2 (MOI = 0.5) and viral infection efficiency after 24 h.p.i. by IFA. Data are means ± s.d. from n = 3 or 5 independent experiments. P values from two-tailed Student’s t-test. ((ns): no significant, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001). e, Huh7 cells were transfected with plasmids encoding GALNT2 (WT or V554M), and cells then infected with H1N1 PR8 (MOI = 0.01) and viral infection efficiency after 24 h.p.i. by IFA. Data are means ± s.d. from n = 3 independent experiments. P values from two-tailed Student’s t-test. ((ns): no significant, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001). f, GALNT2 protein expression levels in Huh7 OE-GALNT2 (GALNT2-stable cells) and Huh7 GALNT2-/- cells were detected by western blot. Western blots were representative of three independent biological repeats. g, GALNT2 protein expression levels in A549 OE-GALNT2 (GALNT2-stable cells) and A549 GALNT2-/- cells were detected by western blot. Western blots were representative of three independent biological repeats. h, STAT1 protein expression levels in Huh7 STAT1-/- and A549 STAT1-/- cells were detected by western blot. Western blots were representative of three independent biological repeats. i, Huh7 WT, GALNT2-/- and reconstitution of GALNT2 in the knockout cells by transient transfection (eGFP as the control) were infected with HCoV-OC43 at MOI 0.05. Supernatants were harvested at 24, 48 and 72 h.p.i. and viral titres were determined in HRT-18 cells by FFA. Data are means ± s.d. from n = 3 independent experiments. j, Huh7 WT, GALNT2-/- and reconstitution of GALNT2 in the knockout cells by transient transfection (eGFP as the control) were infected with HCoV-229E at MOI 0.001. Supernatants were harvested at 24, 48 and 72 h.p.i. and viral titres were determined in Huh7 cells by FFA. Data are means ± s.d. from n = 3 independent experiments. k, A549 WT (Ctrl) and A549 GALNT2-/- cells were infected with IAVs at MOI of 0.01, including H1N1 (PR8 and CA04) and H3N2 (HK68). Supernatants were harvested at indicated time points and viral titres were determined on MDCK cells by FFA. Data are means ± s.d. from n = 3 independent experiments. Data are means ± s.d. from n = 3 independent experiments. P values from two-tailed Student’s t-test. (ns): no significant, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001). Source data

Extended Data Fig. 5. GALNT2 is an antiviral ISG.

a-b, AAV9-EV and AAV9-GALNT2 transduced Calu3 cells. At 48 h post transduction, cells were infected with SARS-CoV-2 WT at MOI 0.01, supernatants were harvested at 48 h.p.i. and viral titres were determined in Vero E6 cells by FFA. Images of viral foci were captured by ELISPOT. P values from two-tailed Student’s t-test. c-d, Foci forming sizes (n = 20) were measured and analyzed. e, Quantative PCR analysis of GALNT2 mRNA in Huh7 cells after SARS-CoV-2 infection at MOI of 0.5. Data are means ± s.d. from n = 3 independent experiments. f, THP-1 cells were infected HCoV-OC43 at MOI of 1. The mRNA expression levels of GALNT2 was analyzed by qPCR at indicated time points. Data are means ± s.d. from n = 3 independent experiments. g, Huh7 cells were infected with HCoV-OC43 for 6 (n = 3) or 24 h (n = 3), respectively. GALNT2 and phosphorylated STAT1 (pSTAT1) protein expression levels were detected by western blot. h, A549 cells were infected with PR8 H1N1 at MOI of 0.01. Cell lysates were harvested at 6 and 24 h.p.i., respectively. GALNT2 and pSTAT1 protein expression levels were detected by western blot. Western blots were representative of three independent biological repeats. i. Huh7 STAT1-/- and Huh7 cells (Ctrl) were treated with 5 μg/mL of IFN-α2 for 24 h. The mRNA levels of GALNT2, STAT1 and OAS1 were detected by qPCR. Data are means ± s.d. from n = 3 independent experiments. j, Huh7, Huh7 GALNT2-/- and Huh7 STAT1-/- cells were treated with 100 ng/mL of IFN-β for 24 h. Then cells were infected with PR8 H1N1 at MOI of 0.01. Viral infection efficiency was detected the by IFA. Data are means ± s.d. from n = 6 independent experiments. k, Schema depicting the presence and position of GAS and ISRE elements in the promoter region (2 kb upstream from the transcription initiation site) of human GALNT2 gene. All Western blots were representative of three independent biological repeats. Data are mean ± s.d. P values from two-tailed Student’s t-test. ((ns): no significant, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001). Figure created with BioRender.com. Source data

Extended Data Fig. 6. Increased lethality and pathogenicity of H1N1 PR8 in GALNT2-/- mice.

a, Using CRISPR/Cas9-mediated gene editing technology, two single guide RNAs (sgRNAs) were designed targeting the second exon and the region between the second and third exons of the GALNT2 gene. b, Sequencing analysis revealed the complete knockout of the second exon of GALNT2. c, WT (n = 5) and GALNT2-/- (n = 4) C57BL/6 mice were intranasally treated by 2 μg IFN-α2. Lungs were harvested at 3 d.p.i. GALNT2 protein expression levels were determined by western blot. Western blots were representative of three independent biological repeats. d, WT (n = 5) and GALNT2-/- C57BL/6 (n = 5) mice were intranasally infected with 350 FFU of PR8 H1N1. Lungs were harvested at 3 d.p.i. e, Quantitative PCR analysis of Il-1β mRNA in the lungs from (d). f, HE staining of the infected lungs of WT (n = 5) or GALNT2-/- (n = 5) C57BL/6 mice infected with PR8 H1N1 at 3 d.p.i. The scale bars are indicative of 100μm and 1000μm respectively. Data are mean ± s.d. P values from two-tailed Student’s t-test in (b), (d). ((ns): no significant, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001). Source data

Extended Data Fig. 7. Overexpression of GALNT2 alleviates pneumonia induced by IAV and SARS-CoV-2 infection.

a, Groups of C57BL/6 mice (n = 2) were either transduced with 3.3×1011 GC of AAV9-EV or AAV9-hGALNT2 intranasally. Lungs were harvested at 14 d.p.i. GALNT2 protein expression levels were monitored by western blot. Western blots were representative of three independent biological repeats. b-c, AAV9-EV or AAV9-GALNT2 transduced C57BL/6 mice (n = 5 or 6) infected with 1000 FFU PR8 H1N1. Three days post infection, GALNT2 expression levels (b) and PR8 M (c) gene mRNA in the lungs were analyzed by qPCR. Two of five mice in AAV9-GALNT2 group could not be detected by RT-qPCR (RT-qPCR Cq values above 38 were considered as negative). d, AAV9-EV or AAV9-GALNT2 transduced C57BL/6 mice infected with 1000 FFU PR8 H1N1. Three days post infection, lungs were harvested. Representative images of infected lungs were shown. e, HE staining of the lungs from (d). The scale bars are indicative of 1000μm and 500μm respectively. f, Quantitative PCR analysis of Il-1β mRNA in the lungs from (b). g-h, Groups of C57BL/6 mice were either transduced with 3.3×1011 GC of AAV9-EV or AAV9-GALNT2 (n = 5 or 6) intranasally and subsequently transduced with 2.5×108 FFU of Ad5-hACE2 after 9 days post AAV transduction. Five days after secondary transduction, mice were infected with 1×105 FFU of SARS-CoV-2 (WT). GALNT2 and SARS-CoV-2 N genes mRNA expression levels in the lungs were determined by qPCR. RT-qPCR Cq values above 38 were considered as negative. i, Representative images of infected lungs from (g-h). j, HE staining of the lungs of from (g-h). The scale bars are indicative of 100μm and 1000μm respectively. k, Quantitative PCR analysis of Il-1β mRNA in the lungs from (g-h). Data are mean ± s.d. P values from two-tailed Student’s t-test. ((ns): no significant, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001). Source data

Extended Data Fig. 8. Overexpression of GALNT2 in A549 has a minimal impact on host genes.

a, Scatter plot showing relative up- and down-regulated genes, identified by RNA-seq analysis of GALNT2-overexpressing A549 cells. GALNT2 expression was validated by western blots (Extended Data Fig. 4g). b, Proteomics: Volcano plot illustrates GALNT2-regulated proteins from the Stable Isotope Labeling By Amino Acids In Cell Culture (SILAC) experiment. Greater than 2-fold change in abundance upon GALNT2 overexpression and lower than 0.05 adjusted p value were set as the threshold criteria for GALNT2-regulated proteins (see method). c, Glycoproteomics: Volcano plot illustrates GALNT2-regulated O-glycosylation sites from the SILAC experiment. Greater than 2-fold increase in abundance upon GALNT2 overexpression and lower than 0.05 adjusted p value were set as the threshold criteria for GALNT2-regulated sites (see method). Pearson correlation coefficients and p-values by two-tailed test are indicated.

Extended Data Fig. 9. GALNT2-dependent O-linked glycosylation at S810/813 of spike.

a, Soluble GalNAc-T1, T2, T3, T4, T5, T7, T11 and T18 were overexpressed in FreeStyleTM HEK293T cells and purified by Ni-NTA chromatography. All proteins were analyzed by Western blots. Western blots were representative of three independent biological repeats. b, MALDI-TOF analysis of GalNAcylation reactions catalyzed by purified GalNAc-Ts on the synthetic spike S2’ (TMPRSS2) peptides. Reactions were performed and analyzed as described in the Methods. An increase of 203 Da corresponds to addition of one GalNAc residue was only detected O-glycosylation by GALNT2. c, Soluble GALNT2 (WT and four mutations) were overexpressed in FreeStyleTM HEK293T cells and purified by Ni-NTA chromatography. All proteins were analyzed by Western blots. Western blots were representative of three independent biological repeats. d, MALDI-TOF analysis of GalNAcylation reactions catalyzed by purified GALNT2 on the synthetic spike S2’ (TMPRSS2) peptides. Reactions were performed (3 h) and analyzed as described in the Methods. An increase of 203 Da corresponds to addition of one GalNAc residue was obvious detected O-glycosylation by WT GALNT2, but an apparent decline in four mutants. Source data

Extended Data Fig. 10. The S810/813I of spike evades GALNT2 antiviral function in a SARS-CoV-2 replicon system.

a, Constructing a reverse genetic complementary SARS-CoV-2 replicon system. The N gene is deleted from the viral genome and replaced with a GFP reporter gene. Through in vitro ligation and transcription, a SARS-CoV-2 RNA genome lacking the N gene but carrying the GFP reporter gene is obtained. Then RNA genome is electroporated into Caco2 cells stably expressing the viral N protein (Caco2-N, packaging cell line), allowing the production of SARS-CoV-2-NdeI replicon. Here we generated 2 replicons, comprising SARS-CoV-2 WT and S810/813I spike mutations. b, Fluorescent and bright field images of SARS-CoV-2 replicons, including WT or S810/813I. Images were representative of three independent biological repeats. Scale bars are indicative of 200 μm. c, Caco2-N cells were infected by WT SARS-CoV-2 or S810/813I replicon at MOI of 0.01 for 24 or 48 h in presence of EV or GALNT2. Viral titres were determined by TCID50. Data are means ± s.d. from n = 3 independent experiments. P values from two-tailed Student’s t-test. d, Caco2-N cells were infected with SARS-CoV-2_Ndel-WT and SARS-CoV-2_Ndel-S810/813I viruses at MOI of 0.01 for 48 h in presence of EV or GALNT2. Caco2-N cells were infected with SARS-CoV-2_Ndel-WT and SARS-CoV-2_Ndel-S810/813I viruses at MOI of 0.01 for 48 h in presence of EV or GALNT2. The GFP-positive cells were quantified of syncytia among total cells in Caco2-N cells. Data are means ± s.d. from n = 3 independent experiments. P values from two-tailed Student’s t-test. Images were representative of three independent biological repeats. Scale bars are indicative of 200 μm. e, A schematic shows the S810/813I and S810/813 A of SARS-CoV-2 recombinant viruses evaded GALNT2 antiviral activity, while reducing the viral fitness due to reduced utilization of TMPRSS2 for viral entry. Source data