Salmonella Effectors SseK1 and SseK3 Target Death Domain Proteins in the TNF and TRAIL Signaling Pathways

Abstract

Strains of Salmonella utilize two distinct type three secretion systems to deliver effector proteins directly into host cells. The Salmonella effectors SseK1 and SseK3 are arginine glycosyltransferases that modify mammalian death domain containing proteins with N-acetyl glucosamine (GlcNAc) when overexpressed ectopically or as recombinant protein fusions. Here, we combined Arg-GlcNAc glycopeptide immunoprecipitation and mass spectrometry to identify host proteins GlcNAcylated by endogenous levels of SseK1 and SseK3 during Salmonella infection. We observed that SseK1 modified the mammalian signaling protein TRADD, but not FADD as previously reported. Overexpression of SseK1 greatly broadened substrate specificity, whereas ectopic co-expression of SseK1 and TRADD increased the range of modified arginine residues within the death domain of TRADD. In contrast, endogenous levels of SseK3 resulted in modification of the death domains of receptors of the mammalian TNF superfamily, TNFR1 and TRAILR, at residues Arg³⁷⁶ and Arg²⁹³ respectively. Structural studies on SseK3 showed that the enzyme displays a classic GT-A glycosyltransferase fold and binds UDP-GlcNAc in a narrow and deep cleft with the GlcNAc facing the surface. Together our data suggest that salmonellae carrying sseK1 and sseK3 employ the glycosyltransferase effectors to antagonise different components of death receptor signaling.

Keywords: Bacteria; Cell death*; Glycosylation; Infectious disease; Inflammation; Inflammatory response; Salmonella; SseK; Virulence; death receptor signaling; glycosyltransferase.

Graphical abstract

Fig. 1.

Immunoblot of RAW264.7 cells infected with derivatives of S. Typhimurium SL1344. Wild type S. Typhimurium, a triple Δssek123 mutant and triple mutant complemented with plasmids encoding one of HA-tagged SseK1, SseK2 or SseK3, or with catalytically-inactive effector derivatives were used to infect RAW264.7 cells for 20 h, as indicated (A and B). Overexpression of the effectors was induced during host cell infection by the addition of 1 mm IPTG. RAW264.7 cells were lysed, and proteins detected by immunoblot with anti-ArgGlcNAc, anti-HA, and anti-β-actin antibodies as indicated. Representative immunoblot of at least three independent experiments.

Fig. 2.

Enrichment of peptides Arg-GlcNAcylated by SseK1 derived from Salmonella-infected RAW264.7 cells. A, Label-free quantification of Arg-GlcNAc peptides immunoprecipitated from RAW264.7 cells infected with S. Typhimurium ΔsseK123 complemented with either SseK1-HA or SseK1E_255A-HA. Arg-GlcNAcylated peptides are presented as a volcano plot depicting mean ion intensity peptide ratios of SseK1-HA versus SseK1E_255A-HA plotted against logarithmic t test p values from biological triplicate experiments. Arg-GlcNAcylated peptides with corresponding t test p values below 0.001 are annotated by protein name, with human peptides shaded blue and bacterial peptides shaded red. B, Manually curated EThcD spectra showing glycosylation of Arg²⁴³ within the death domain of mouse TRADD, Andromeda score 249.32. Within MS/MS spectra NL denote neutral loss associated ions. C, Parallel reaction monitoring of ArgGlcNAc peptide immunoprecipitated from RAW264.7 cells infected with S. Typhimurium ΔsseK123 or S. Typhimurium ΔsseK23. Arginine-glycosylated peptides are presented as a volcano plot depicting mean log2 ion intensity peptide ratios of ΔsseK123 versus ΔsseK23 plotted against logarithmic t test p values from biological triplicate experiments. Arg-GlcNAcylated peptides are annotated by protein name and shaded blue. D, Manually curated EThcD spectra showing glycosylation of Arg²³³ within the death domain of mouse TRADD, Andromeda score 71.45. Within MS/MS spectra NL denote neutral loss associated ions.

Fig. 3.

Mutagenesis of putative SseK1 glycosylation sites of TRADD. A, Immunoblot showing Arg-GlcNAcylation of ectopically expressed Flag-hTRADD or Flag-hTRADD mutants in HEK293T cells co-transfected with pEGFP-SseK1. Cells were harvested for immunoblotting and detected with anti-ArgGlcNAc, anti-GFP, and anti-Flag antibodies. Antibodies to β-actin were used as a loading control. Representative immunoblot of at least three independent experiments. B, Manually curated EThcD spectra of Arg-GlcNAcylated Flag-hTRADD enriched by anti-Flag immunoprecipitation following ectopic expression in HEK293T cells and co-transfection with pEGFP-SseK1. Various observed sites of Arg-GlcNAcylation are highlighted in red, and presented alongside corresponding M/Z values and observed Andromeda scores. Within MS/MS spectra NL denote neutral loss associated ions.

Fig. 4.

Identifying substrates of SseK3 by Arg-GlcNAc peptide enrichment. A, Label-free quantification of Arg-GlcNAc peptide immunoprecipitated from RAW264.7 cells infected with S. Typhimurium ΔsseK123 or ΔsseK12. Arginine-glycosylated peptides are presented as a volcano plot depicting mean log2 ion intensity peptide ratios of ΔsseK123 versus ΔsseK12 plotted against logarithmic t test p values from biological triplicate experiments. Arg-GlcNAcylated peptides are annotated by gene name and shaded blue. B, Heat map showing observed ion intensity of Arg-GlcNAcylated peptides between biological triplicates. C, Partial sequence alignment showing observed GlcNAcylated arginine residue is conserved between identified substrates. D, Manually curated HCD spectra of arginine glycosylated TNFRSF10B/TNFR1 (upper) and TNFRSF1A/TRAILR (lower). Observed sites of Arg-GlcNAcylation are highlighted in red, and presented alongside corresponding m/z values and observed Andromeda scores.

Fig. 5.

In vitro validation of host substrate modifications by SseK3. A, Immunoblot of inputs and immunoprecipitates (IP) of anti-Flag immunoprecipitations performed on lysates of HEK293T cells co-transfected with pFlag-hTRAILR2_DD and pEGFP-SseK3 or pEGFP-SseK3_E258A. Proteins were detected with anti-Arg-GlcNAc, anti-GFP and anti-Flag antibodies as indicated. Antibodies to β-actin were used as a loading control. Representative immunoblot of at least three independent experiments. B, LC-MS analysis of tryptic digest derived from co-incubation of recombinant His-hTRAILR2_DD and GST-SseK3 in the presence of UDP-GlcNAc. C, LC-MS analysis of tryptic digest fractions derived from co-incubation of recombinant His-hTRAILR2_DD and GST-SseK3 with no sugar donor. D, HCD fragmentation of recombinant His-hTRAILR2_DD incubated with GST-SseK3 and UDP-GlcNAc, Andromeda score 167.28. E, Immunoblot of recombinant His-hTRAILR2_DD and GST-SseK3 following co-incubation at 37 °C for 5 h. Proteins were detected with anti-Arg-GlcNAc, anti-GST, and anti-His antibodies as indicated. Arrow indicates Arg-GlcNAcylated His-hTRAILR2_DD. Representative immunoblot of at least three independent experiments.

Fig. 6.

In vitro binding studies of SseK3 and Arg-GlcNAcylation of human TNFR1. A, Immunoblot of input and immunoprecipitate (IP) of anti-Flag immunoprecipitations performed on lysates of HEK293T cells co-transfected with pFlag-hTNFR1_DD and pEGFP-SseK3 or pEGFP-SseK3_E258A. Proteins were detected with anti-Arg-GlcNAc, anti-GFP and anti-Flag antibodies as indicated. β-actin detection was used as a loading control. Representative immunoblot of at least three independent experiments. B, EThcD fragmentation of Flag-hTNFR1_DD enriched from HEK293T cells by anti-Flag immunoprecipitation following co-transfection with pEGFP-SseK3, peptide confirmed by manual annotation. C, S. cerevisiae Y2HGold co-transformed with pGBKT7-SseK3 and pGADT7-hTRAILR2 DD or pGADT7-hTNFR1 DD and plated onto selective media to select for plasmid carriage (DDO) or to select for protein-protein interactions (QDO). S. cerevisiae Y2HGold co-transformed with pGBKT7-NleB1 and pGADT7-FADD DD was used as a positive control for protein-protein interactions. Self-activation by the bait or prey fusion proteins was discounted by co-transformation of S. cerevisiae Y2HGold with pGADT7 and pGBKT7-SseK3 or co-transformation with pGBKT7 and pGADT7-hTNFR1 DD or pGADT7-hTRAILR2 DD.

Fig. 7.

SseK3 structure. A, Cartoon representation of the SseK3·UDP complex. The structure of SseK3 is shown with the bound UDP drawn as sticks. Secondary structures are numbered. SseK3 displays a GT-A fold and contains an α-helical insertion (marked by a dashed line) The secondary structure elements are rainbow colored, from blue at the N terminus to red at the C terminus. B, Topology diagram of SseK3 with colors of secondary structures matching those in panel A. C, Coordination of UDP in the E258Q active site. Residues in SseK3 are shown in green and the UDP moiety is shown in cyan. Hydrogen bonds are shown as black dashed lines. Arg59 is directed toward the uracil whereas Arg55 is mobile and assumes different conformations in various structures. D, Coordination of the Mg²⁺ ion and the hydrogen bonds to the diphosphate in the E258Q active site. The Mg²⁺ ion is shown as a cyan sphere. E, Superposition of SseK3 (red) and GT44 family Clostridium difficile toxin A (TcdA) glucosyltransferase domain (PDB code 3SRZ, blue). Toxin A domain is larger than SseK3 and the segments that have no correspondence in SseK3 are painted gray. Two segments of SseK3 without correspondence in toxin A are painted pink. F, The UDP-GlcNAc and arginine are modeled into the active site of SseK3. SseK3 is represents as a solvent accessible surface colored by the electrostatic potential (red - negative, blue - positive). UDP-GlcNAc was taken from structure 3SRZ and placed in SseK3 based on the position of UDP in the SseK3 crystal structure. Arginine was positioned in the long arm of the groove and can be easily accommodated. In this position, the arginine could accept GlcNAc from UDP-GlcNAc. The UDP moiety occupies the short arm of the l-shaped groove and an acceptor arginine is modeled into the long arm. G, The superposition of UDP and GlcNAc from SseK3(E258Q) (white carbons) with UDP-Glc from toxin A glucosyltransferase domain (pink carbons). The figures were prepared with PyMOL software (Schrodinger Inc, Cambridge, MA).