Bacterial strains and plasmids used in this study were listed in Supplementary Tables S1–S3. DNAs for nleB and sseK1/2/3 genes were inserted into pCS2-EGFP and pCS2-3Flag vectors for mammalian expression, and inserted into pET28a vectors for protein expression in E. coli. For complementation in EPEC, DNA for NleB and NleB mutants was ligated into the pTRC99A vector under the trc promoter. For complementation in S. Typhimurium strain, DNAs for SesK1, SseK2, and SseK3, together with their upstream promoter regions, were inserted into pET28a vector. Human cDNAs for TRADD, TRADD DD, TNFR1 DD, FADD, and RIPK1 DD were amplified from a HeLa cDNA library as previously described (Li et al., 2013). All point mutations were generated by QuickChange site-directed mutagenesis kit (Stratagene). All plasmids were verified by DNA sequencing.

Antibodies for Arg-GlcNAc (ab195033) and DnaK (8E2/2) were purchased from Abcam. Antibodies for Flag M2 (F2426) and α-tubulin (T5186) were Sigma products. Horse radish peroxidase (HRP)-conjugated goat anti-mouse IgG (NA931V) and HRP-conjugated goat anti-rabbit IgG (NA934) were purchased from GE Healthcare. Cell culture products were from Invitrogen, all other reagents used in this study were Sigma-Aldrich products unless specially noted.

293T cells and HeLa cells obtained from the American Type Culture Collection (ATCC) were grown in DMEM (GIBCO) medium supplemented with 10% FBS (GIBCO), 2 mM L-glutamine (GIBCO), 100 U/ml penicillin, and 100 mg/ml streptomycin (GIBCO). Vigofect and jetPRIME were used for transient transfection following the respective manufacturer's instructions.

Cells were fixed with 4% paraformaldehyde for 10 min at room temperature, then treated with 0.2% Triton X-100 for 15 min. After that, cells were blocked with 2% BSA for 30 min, followed by the incubation with the indicated primary antibody and secondary antibody. All image data shown are representative of at least three randomly selected fields. The immunoprecipitation assay with Flag M2 beads was performed as the manufacturer's instruction.

Recombinant protein expression was induced in E. coli BL21 (DE3) strain (Novagen) at 22°C for 16 h with 0.4 mM isopropyl-b-D-thiogalactopyranoside (IPTG) after the absorbance of 600 nm reaching 0.8. Affinity purification of LFn-NleB/SseK1/SseK2/SseK3 and His-NleB/SseK1/SseK2/SseK3 proteins was performed using Ni-NTA agarose (Qiagen) following the manufacturer's instructions.

CD spectra in the “far UV” region (185–260 nm) was used to determine the NleB/NleB RA, SseK1/SseK1 RA, and SseK3/SseK3 RA proteins secondary structure. The CD spectroscopy is Chirascan (Applied photophysics), and the operating conditions for the spectrum were set as: spectral bandwidth is 1 nm, step size is 0.5 nm, pathlength is 1 mm, the sample concentration is 0.22 mg/ml, solvent is 10 mM sodium phosphate pH 7.4, time-per-point is 3 s, spectral scan temperature is 25°C, and total N₂ flow is 5.0 l/min. The spectra measurement data was analyzed by the program Spectrum Manager 2, and the CD spectrum of NleB/NleB RA, SseK1/SseK1 RA, and SseK3/SseK3 RA, plotted as CD (mdeg) against wavelength (nm) is shown in figures.

HeLa cells were seeded at a concentration of 2 × 10⁴ per well in 96-well plates day before infection. For bacterial infection, a single colony in 0.5 ml LB was incubated overnight at 37°C. EPEC strains were then diluted by 1:40 in DMEM supplemented with 1 mM IPTG and cultured in the presence of 5% CO₂ at 37°C for an additional 4 h. For S. Typhimurium SL1344 infection, the bacterial cultures were diluted by 1:33 in LB (without antibiotics) and cultured for an additional 3 h. Infection assays were performed at a multiplicity of infection (MOI) of 200 in the presence of 1 mM IPTG for 2 h for EPEC, or at MOI of 100 for 30 min for Salmonella, with a centrifugation at 800 g for 10 min at room temperature to promote infection. At the end time point of infection, cells were washed four times with PBS and the extra bacteria were killed with 200 μg/ml gentamicin for EPEC, or with 100 μg/ml gentamycin for Salmonella. One-hour CHX pretreatment was used to sensitize TRAIL and TNF stimulation of cell death. Fifteen hours later, CellTiter-Glo® Luminescent Cell Viability Assay kit (Promega) was used to detect the cell survival.

The recombinant proteins of NleB, SseK1, SseK2, SseK3, and their mutants were loaded onto a C4 capillary column (MAbPacTM RP, 4 μm, 2.1 × 50 mm, Thermo Scientific, USA), and eluted by a Dionex Ultimate 3000 HPLC system with the following solvent gradient: 5–100% B in 10 min (A:0.1% formic acid; B: 0.1% formic acid/80% acetonitrile isopropanol). The eluted proteins were sprayed into a Q Exactive Plus mass spectrometer equipped with a Heated Electrospray Ionization (HESI-II) Probe. The protein charge envelop was averaged across the corresponding protein elution peaks, and de-convoluted into non-charged forms by the Thermo Scientific Protein Deconvolution program.

To determine the exact auto-GlcNAcylation sites on NleB, SseK1, SseK2, and SseK3, these purified proteins were separated by SDS-PAGE, and subjected to in-gel trypsin digestion. The final peptide samples were analyzed by the Q Exactive Plus mass spectrometer equipped with nanoflow reversed-phase liquid chromatography (EASY nLC 1200, Thermo Scientific). EASY nLC 1200 was fitted with a Thermo Scientific Acclaim Pepmap nano-trap column (C18, 5 μm, 100 Å, 100 μm × 2 cm) and a Thermo Scientific EASY-Spray column (Pepmap RSLC, C18, 2 μm, 100 Å, 50 μm × 15 cm), and run at 300 nl/min with the following mobile phases (A: 0.1% formic acid; B: 80% CH3CN/0.1% formic acid). The liquid chromatography separation was carried out with the following gradient: 0~8% B for 3 min, 8~28% B for 42 min, 28~38% B for 5 min, 38~100% B for 10 min. Eluted peptides were electro sprayed directly into the mass spectrometer for MS and MS/MS analyses in a data-dependent acquisition mode. One full MS scan (m/z 350–1500) was acquired, then immediately the 10 ions with the highest intensity were selected for MS/MS analyses. Dynamic exclusion was set with repeat duration of 24 s and exclusion duration of 12 s.

The MS raw data were processed by Proteome Discoverer (Thermo Scientific) and searched against NleB/SseKs protein database downloaded from UniProt. N-Acetylhexosamine addition to arginine (arginine-GlcNAcylation) set as the variable modifications. The precursor mass tolerance was set at 10 ppm and the fragment mass tolerance was set at 0.02 Da. Maximum missed cleavage was set at 2. Both peptide and protein assignments were filtered to achieve a false discovery rate (FDR) < 1%.

Statistical analysis was performed using Student's T-test to compare two experimental groups, and one-way analysis of variance (ANOVA) was used to compare the differences between multiple groups. Significant differences are marked with ^*P < 0.05, ^**P < 0.01, and ^***P < 0.001; n.s, not significant. All results are graphed as means ± SD for triplicate samples.

Type III secretion system effectors NleB/SseKs harbor arginine GlcNAc transferase activity and display distinct differences in host substrate specificity (Gao et al., 2013; Li et al., 2013; Pearson et al., 2013; Yang et al., 2015; El Qaidi et al., 2017; Gunster et al., 2017; Scott et al., 2017; Ding et al., 2019; Newson et al., 2019). In order to study the subcellular localization of NleB and SseK proteins, HeLa cells were transfected with the ectopic expression plasmids, pCS2-GFP-NleB and pCS2-GFP-SseK1/2/3. GFP-NleB and GFP-SseK1 were diffusely distributed in the cytoplasm. Accordingly, the arginine GlyNAcylation catalyzed by NleB and SseK1 did not show obvious subcellular localization (Figure 1A). GFP-SseK2 and GFP-SseK3 were found to be co-localized with the Golgi marker GM130. The arginine GlcNAcylation was not detected in GFP-SseK2-transfected cells due to its weak GlcNAc transferase activity. Interestingly, the arginine GlcNAcylation catalyzed by GFP-Ssek3 was co-localized in the host Golgi network (Figure 1A). Considering NleB, SseK1, and SseK3 showed related subcellular localization and modification patterns, we asked if they can modify themselves. To determine whether NleB/SseKs could GlcNAcylate themselves, we purified NleB/SseKs and their dead enzymatic mutants in E. coli. These effectors could GlcNAcylate themselves when they were expressed as recombinant proteins in prokaryotic systems (Figures 1B–D). Western blot results showed that NleB, SseK1, and SseK3 were auto-GlcNAcylated, whereas their corresponding dead enzymatic mutants were not. SseK2 was barely able to be auto-GlcNAcylated due to weak enzymatic activity (Figure 1B). Mass spectrometry analysis determined the percentage of 203-Da increase in the total molecular weight of NleB, SseK1, and SseK3 (Figures 1C,D). NleB showed the strongest auto-modification percentage, that was around 15%. SseK2 exhibited little mass change, which was consistent with the weak signal detected by the anti-Arg-GlcNAc antibody (Pan et al., 2014). Recombinant SseK1 and SseK3 showed the auto-modification percentages were 3 and 6%, respectively (Figures 1B–D). Importantly, SseK1 and SseK3 expressing and secreting during Salmonella infection showed strong auto-modification. The auto-modification ratio of SseK1 and SseK3 during infection was to the same degree as that of the recombinant purified NleB, which showed a 15% auto-modification ratio (Supplementary Figure 1, Figure 1C). Furthermore, to identify the specific auto-GlcNAcylation sites of NleB and SseKs, we performed the liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis. A total of 10 arginine residues in three proteins were detected as modification sites. They are Arg13, Arg53, Arg159, and Arg293 in NleB, Arg30, Arg158, and Arg339 in SseK1, as well as Arg153, Arg 305, and Arg335 in SseK3 (Figure 1E, Supplementary Figures 2–4). The modification sites were shown in structures of NleB, SseK1, and SseK3 (Supplementary Figures 5A–C). These findings were similar to a recent report that Arg153, Arg184, Arg305, and Arg335 were the auto-Arg-GlcNAcylation sites of SseK3 in an in vitro reconstitution system (Newson et al., 2019).

In order to validate the modification sites, mutants of single mutation for each modification site were generated. Single mutations did not abolish the auto-modification. However, mutation of R159 in NleB, R158 in SseK1, and R335 in SseK3 showed decreased auto-modification comparing with their corresponding wild-type proteins and other single residue mutants (Supplementary Figures 5D–F). Then we replaced the arginine residues by alanine residues to generate NleB (4RA) (NleB_{Arg13/53/159/293Ala}), SseK1 (3RA) (SseK1_{Arg30/158/339Ala}), and SseK3 (4RA) (SseK3_{Arg153/184/305/335Ala}) mutants. First, we purified recombinant proteins, NleB (4RA), SseK1 (3RA), and SseK3 (4RA). Mass spectrometry analyses showed that the total molecular weight of these mutants exhibited no 203-Da increase comparing with NleB, SseK1, and SseK3, respectively (Figure 2A). The properties of NleB (4RA), SseK1 (3RA), and SseK3 (4RA) proteins were similar to that of wild type proteins, which were evaluated by ion-exchange, gel-filtration chromatography and circular dichroism (Supplementary Figures 6, 7). Additionally, we verified the auto-GlcNAcylation in a eukaryotic system. 3Flag-NleB, 3Flag-SseK1, 3Flag-SseK3, and their site-directed RA mutants were ectopically expressed in 293T cells. Auto-arginine-GlcNAcylation was measured with anti-Arg-GlcNAc antibody in western blot assay following immunoprecipitation with anti-Flag antibody-conjugated beads. The results showed that the NleB (4RA), SseK1 (3RA), and SseK3 (4RA) mutants completely abolished the auto-arginine-GlcNAcylation signals (Figures 2B–D), indicating that the Arg13/53/159/293 in NleB, Arg30/158/339 in SseK1, and Arg153/184/305/335 in SseK3 were the bona fide modification sites.

NleB is an inverting glycosyltransferase toward the conserved arginine of TRADD (Arg235), FADD (Arg117), and RIPK1 (Arg603) (Li et al., 2013; Ding et al., 2019). One recent proteomics study have shown that TRADD and TNFR1 are the preferential substrates of SseK1 and SseK3, respectively, during Salmonella infection (Newson et al., 2019; Xue et al., 2019). We found that NleB 4RA mutant showed decreased binding ability with TRADD compared to NleB WT (Figure 3A). To verify the biological function of the auto-arginine-GlcNAcylation, we tested whether the auto-modification deficient mutants of NleB, SseK1, and SseK3 could still GlcNAcylate their host targets during infection. Arg-GlcNAc transferase deficient EPEC and Salmonella strains were constructed and then expressed with NleB, NleB (4RA), SseK1, SseK1 (3RA), SseK3, or SseK3 (4RA) as indicated (Figures 3B–F). NleB delivered by EPEC could efficiently modify TRADD, FADD, and RIPK1 DD. However, the NleB (4RA) mutant lost the ability to modify these targets, suggesting that the auto-modification sites Arg13/53/159/293 in NleB were essential for the enzymatic activity of NleB (Figures 3B–D). Consistently, SseK1 and SseK3 delivered by S. Typhimurium strain could modify TRADD DD and TNFR1 DD, respectively (Figures 3E,F). Whereas, SseK1 (3RA) mutant decreased the enzymatic activity toward TRADD DD, and the SseK3 (4RA) mutant abolished the GlcNAcylation activity toward TNFR1 DD (Figures 3E,F). Further, we evaluated the glycosylation of TRADD DD by NleB single mutants during EPEC infection in order to verify which auto-modification site is important for modification of the host substrate. The NleB mutants R53A and R159A showed decreased glycosylation ability to the host target TRADD DD. Whereas, the mutation of R13A and R293A did not affect the glycosylation on TRADD DD (Supplementary Figure 5G). Taken together, mutation of the auto-modification sites of NleB and SseKs abolished or attenuated their modification on host target proteins. For NleB, the mutation of R53 and R159 severely disrupted its enzymatic activity toward TRADD DD.

SseK1 and SseK3 inhibit cell death during Salmonella infection in macrophages (Gunster et al., 2017). The DDs of TRADD, FADD, and RIPK1 can be GlcNAcylated by NleB, resulting in the inhibition of death receptor signaling and nuclear factor-κB (NF-κB) signaling (Li et al., 2013). To assess the biological importance of the auto-arginine-GlcNAcylation, we evaluated the inhibition effect of NleB (4RA), SseK1 (3RA), and SseK3 (4RA) on certain death receptor signaling pathways. First, we determined that the variant forms of NleB (4RA), SseK1 (3RA), and SseK3 (4RA) were translocated into cells at a similar level to that of the WT proteins during infection (Figures 4A–C). Then, HeLa cells were infected with the indicated bacterial strains and stimulated with TNF or TRAIL. NleB delivered by EPEC inhibited both TNF-induced or TRAIL-induced cell death (Figures 4D,E). SseK1 or SseK3 delivered by Salmonella was sufficient to inhibit TNF-induced cell death (Figure 4F). However, the auto-modification deficient mutants, NleB (4RA), SseK1 (3RA), and SseK3 (4RA), could not inhibit the death receptor signaling pathways (Figures 4D–F). Therefore, these data suggest that the auto-arginine-GlcNAcylation of NleB, SseK1, and SseK3 is crucial for their biological activity during infection.