The S. Typhi effector StoD is an E3/E4 ubiquitin ligase which binds K48- and K63-linked diubiquitin

Abstract

Salmonella enterica (e.g., serovars Typhi and Typhimurium) relies on translocation of effectors via type III secretion systems (T3SS). Specialization of typhoidal serovars is thought to be mediated via pseudogenesis. Here, we show that the Salmonella Typhi STY1076/t1865 protein, named StoD, a homologue of the enteropathogenic Escherichia coli/enterohemorrhagic E. coli/Citrobacter rodentium NleG, is a T3SS effector. The StoD C terminus (StoD-C) is a U-box E3 ubiquitin ligase, capable of autoubiquitination in the presence of multiple E2s. The crystal structure of the StoD N terminus (StoD-N) at 2.5 Å resolution revealed a ubiquitin-like fold. In HeLa cells expressing StoD, ubiquitin is redistributed into puncta that colocalize with StoD. Binding assays showed that StoD-N and StoD-C bind the same exposed surface of the β-sheet of ubiquitin, suggesting that StoD could simultaneously interact with two ubiquitin molecules. Consistently, StoD interacted with both K63- (K_D = 5.6 ± 1 μM) and K48-linked diubiquitin (K_D = 15 ± 4 μM). Accordingly, we report the first S. Typhi-specific T3SS effector. We suggest that StoD recognizes and ubiquitinates pre-ubiquitinated targets, thus subverting intracellular signaling by functioning as an E4 enzyme.

Figure 1.. StoD is a member of the NleG family of effector proteins.

(A) A diagrammatic representation of the genomic localization of stoD within the S. Typhi Ty2 genome. Colours indicate different gene functions: phage genes (yellow), stoD (green), and miscellaneous genes (light blue). (B) The evolutionary history of the NleG family members from EHEC, EPEC, C. rodentium, S. bongori, S. Typhi, and S. Paratyphi B. (C) Secretion assay of 4HA-tagged StoD from WT and ΔprgH S. Typhimurium; SipD and empty pWSK29-Spec vector (EV) were used as positive and negative controls, respectively. DnaK was used as a lysis and loading control. An anti-HA antibody was used to detect HA-tagged StoD. SipD and DnaK were detected using respective antibodies. The blot is representative of two repeats. (D) HeLa cell translocation of StoD-TEM1 and SopD-TEM1 fusions from WT or ΔprgHΔssaV S. Typhimurium; empty pWSK29-Spec vector (EV) was used as a control. Graph shows mean + SEM. Translocation of each protein was compared between the WT and ΔprgHΔssaV genetic backgrounds using a Multiple t test with the Holm-Sidak correction for multiple comparisons (****P < 0.0001). Graph represents an average of three independent repeats.

Figure S1.. Amino acid sequence alignments of NleG-like proteins.

Shaded residues are those conserved within the RING/U-box domain, and residues with an asterisk indicate residues that have been shown to be involved in E2 binding. Sequences were obtained from the Kyoto Encyclopedia of Genes and Genomes, alignments performed using Clustal Omega, and formatted using Strap.

Figure S2.. StoD is not required for S. Typhi invasion or replication.

(A, B) S. Typhi Ty2 WT, ΔinvA, and ΔstoD mutants were grown under (A) static conditions overnight or (B) shaking subculture conditions. Invasion was assessed after infection of HeLa cells. (C) Replication of WT, ΔssaV, and ΔstoD after infection of THP-1 cells. Graph shows mean + SD. Mutant Ty2 strains were compared with WT in each infection using a Kruskal–Wallis test with a Dunn’s multiple comparisons test (**P < 0.01 and ***P < 0.001). Each figure represents the average of three independent repeats.

Figure S3.. StoD is an E3 ubiquitin ligase.

Recombinant 6His-StoD and biotinylated ubiquitin were used in autoubiquitination assays with different E2 ubiquitin–conjugating enzymes. Western blots were visualised using streptavidin conjugated to HRP.

Figure 2.. StoD is an E3 ubiquitin ligase.

(A) Both StoD and StoD_L167A have an E3 ubiquitination activity in the presence of ATP (upper panel shows an anti-Ub-FK2 antibody blot). Western blotting using anti-His tag antibodies shows autoubiquitination of StoD (lower panel). (B) Autoubiquitination assay using StoD, StoD_P204K, StoD-N, StoD-C, or StoD-C_P204K visualised with anti-Ub-FK2 antibody. Only StoD and StoD-C exhibit an E3 ubiquitin ligase activity. Image is representative of two independent repeats. (C) Model for the interaction between StoD-C (grey surface) and UBE2E1 (cyan cartoon) based on the CHIP U-box/UBE2D2 structure (23) (PDB ID 2OXQ) constructed using Superpose (68). A SCWRL homology model (22) for StoD, constructed using the sequence alignment in Fig S1 and solution structure of reduced NleG2.3 (12) (PDB ID 2KKX), was superimposed with CHIP U-box (RMSD 2.07 Å over 55 residues). The UBE2E1 structure (69) (PDB ID 3BZH) was superimposed with UBE2D2 (RMSD 0.61 Å over 149 residues). These superimpositions are shown in Fig S6C. CSPs from titration of 100 μM ¹⁵N-StoD-C [134–233] with 100 μM UBE2E1 are mapped onto the surface of StoD-C: peak disappearances due to line broadening are shown in red, whereas peak shifts greater than 0.05 ppm are shown in orange. P204 is shown in blue.

Figure S4.. Complete backbone assignments of StoD-N and StoD-C.

(A, B) ¹H, ¹⁵N-HSQC spectra of ¹⁵N/¹³C–labelled (A) 545 μM StoD-N [1–101] and (B) 575 μM StoD-C [134–233] acquired in 25 mM NaPi, pH 7.0, with backbone amide assignments for all non-proline residues shown. The zoomed inset corresponds to the region of the spectra indicated by the dashed box.

Figure S5.. Titration of StoD-N with UBE2E1.

Overlay of ¹H, ¹⁵N-HSQC spectra of 100 μM ¹⁵N-labelled StoD-N [1-101] before (black) and after (cyan) addition of 335 μM UBE2E1 acquired in 20 mM Tris–HCl, pH 7.5, and 150 mM NaCl.

Figure S6.. Titration of StoD-C variants with UBE2E1.

(A, B) Overlay of ¹H, ¹⁵N-HSQC spectra of 100 μM ¹⁵N-labelled (A) StoD-C [134–233] or (B) StoD-C_L167A before (black) and after (cyan) addition of 100 μM UBE2E1 acquired in 20 mM Tris–HCl, pH 7.5, 150 mM NaCl, and 1 mM TCEP. * denotes the peak for L167. (C) Model for the interaction between StoD-C (grey cartoon) and UBE2E1 (cyan cartoon) from Fig 2C superimposed with the CHIP U-box/UBE2D2 structure (black cartoon) (PDB ID 2OXQ). (D) Model for the interaction between StoD-C (grey surface) and UBE2E1 (cyan cartoon) from Fig 2C, showing L167 within the interaction interface in red.

Figure 3.. The StoD N terminus is a ubiquitin-like domain.

(A) Views of the 2.5 Å crystal structure of the StoD-N [1–101] shown as a cartoon representation. Chain A is shown coloured from the N terminus (blue) to the C terminus (red). Residues visible from the N-terminal tag are coloured black. (B) Superimposition of StoD-N [1–101] chain A (grey) with human ubiquitin (orange) (28) (PDB ID 1UBQ) using Superpose (68). The RMSD is 3.40 Å over 38 residues.

Figure S7.. The final model and electron density difference map for StoD-N.

The final StoD-N model is shown as a stick representation with carbon, nitrogen, and oxygen shown in green, blue, and red, respectively. The corresponding 2F_O-F_C map is shown as a mesh representation contoured at 1.08σ. Visualisation in Coot (1).

Figure S8.. StoD-N crystallised as a tetramer.

(A, B) The tetramer found within the asymmetric unit of the StoD-N crystals is shown as a (A) cartoon and (B) surface representation. Native residues for StoD-N within chain A (green), B (cyan), C (magenta), and D (yellow) are distinguished, whereas non-native residues corresponding to the N-terminal tag are shown in black for all chains.

Figure S9.. StoD is a monomer in solution.

(A–C) Analysis of the oligomeric state of (A) full-length StoD, (B) StoD-N [1–101], and (C) StoD-C [134–233] by SEC with in-line MALS. The left panel provides the complete SEC elution profiles for the dilution series indicated. The right panel shows the molecular weight at any given point across the major elution peak. For each construct, the observed average molecular weight (Obs) does not vary with concentration or elution volume and is in close agreement with the expected molecular weight (Ex) of the monomer.

Figure S10.. StoD-N is a rigid domain.

(A) Chain A of StoD-N is coloured according to the C^α B-factor (Ų). (B) Superimposition of chain B (cyan; RMSD 0.24 Å over 107 residues), chain C (magenta; RMSD 0.38 Å over 107 residues), and chain D (yellow; RMSD 0.39 Å over 104 residues) with chain A (green) of the StoD-N tetramer shown in Fig S8 using Superpose (2).

Figure S11.. Structural homologues of StoD-N.

(A) Superimposition of StoD-N chain A (grey) with the solution structure of the NleG5-1 N-terminal domain (magenta) (PDB ID 5VGC) using Superpose (2). The RMSD is 1.99 Å over 76 residues. (B) Superimposition of StoD-N chain A (grey) with the solution structure of the SH2 domain of Grb2 (green) (3) (PDB ID 1QG1) using Superpose (2). The RMSD is 2.63 Å over 45 residues. The Shc-derived peptide from the overlaid structure is shown as a red ribbon, with the Tyr(P) side-chain shown as spheres. (C) Overlay of ¹H, ¹⁵N-HSQC spectra of 100 μM ¹⁵N-labelled StoD-N [1-101] before (black) and after (green) addition of 8.75 mM buffered Tyr(P) acquired in 20 mM Tris–HCl, pH 7.5, 150 mM NaCl.

Figure 4.. StoD forms puncta upon ectopic expression which colocalize with cellular ubiquitin.

(A) Immunofluorescence of HeLa cells transfected with HA-StoD reveals formation of discrete puncta. (B, C) Colocalization of transfected StoD and StoD-N with ubiquitin; no colocalization was seen in cell transfected with StoD-C or StoD_P204K and the mCherry negative control. (B) DNA and actin were visualised using Hoechst 33258 and Phalloidin-iFluor 647, respectively. StoD-HA, StoD_P204K-HA, and HA-mCherry were visualised using an anti-HA antibody, whereas ubiquitin was visualised using an anti-Ub-FK2 antibody. Scale bar, 5 μm. Images representative of at least two independent repeats. (C) Percentage of transfected cells where colocalization of ubiquitin with either StoD or StoD-N is observed. (D) Direct Y2H assay in AH109 cotransformed with either empty pGBKT7 (EV) or ubiquitin and StoD derivatives. Cotransformants were plated on control DDO plates and QDO plates to assess protein–protein interactions. StoD, StoD_P204K, StoD-N (aa 1–133), and StoD-C (aa 134–223) interacted with ubiquitin. No interaction was seen in cotransformants expressing StoD-C_P204K and ubiquitin. Image is representative of three independent repeats.

Figure S12.. StoD and StoD-N form puncta in transfected HeLa cells.

StoD and StoD-N, but not StoD-C or the mCherry control, form puncta in transfected HeLa cells. DNA and actin were visualised using Hoechst 33258 and Phalloidin-iFluor 647, respectively. StoD, StoD-N, StoD-C, and mCherry were visualised using an anti-HA antibody. Scale bar, 10 μm. Images representative of two independent repeats.

Figure S13.. StoD does not colocalize with vesicular proteins.

Immunofluorescence images of StoD, or mCherry as a control, 24 h after transfection of HeLa cells with pRK5 and GFP constructs. DNA was visualised using Hoechst 33258. (A–C) StoD was visualised using an anti-HA antibody, whereas (A) LC3, (B) Rab11a, and (C) Vamp3 were visualised using an anti–GFP antibody. Scale bar, 5 μm. Images representative of two independent repeats.

Figure S14.. StoD does not colocalize with ubiquitin-like proteins.

Immunofluorescence images of HA-mCherry or StoD-HA in HeLa cells 24 h post-transfection with pRK5 constructs. DNA was visualised using Hoechst 33258. (A–C) StoD-HA and HA-mCherry were visualised using an anti-HA antibody, whereas (A) NEDD8, (B) SUMO-1, or (C) SUMO-2/3 were visualised using respective antibodies. Scale bar, 5 μm. Images representative of two independent repeats.

Figure S15.. NleG–ubiquitin interactions.

Direct Y2H assay in AH109 cotransformed with either empty pGBKT7 or ubiquitin and NleG7 or NleG8 derivatives. Cotransformants were plated on DDO and QDO plates to assess growth and protein–protein interactions, respectively. Full-length NleG7 and NleG8 interact with ubiquitin, whereas Nle7_P177K, NleG7-N, or NleG8-N did not bind ubiquitin. Image is representative of three independent repeats.

Figure 5.. StoD has two UBDs.

(A) MST measured for a titration of 61 nM–2 mM ubiquitin with 40 nM fluorescently labelled StoD. 20% LED power and 40% laser power and data from the thermophoresis contribution alone were used. The normalized fluorescence signal was taken relative to that of the fully bound state and shown as an average of four independent dilution series. The data were fitted with a four parameter logistic (4PL) fit, yielding a Hill coefficient of 1.67 ± 0.23. (B) Fluorescence intensity measured for a titration of 16 nM–500 μM full-length StoD with 40 nM fluorescently labelled ubiquitin_G76C. The fluorescence signal was taken relative to that of the fully bound state and shown as an average of three independent dilution series. The data were fitted with a 4PL fit, yielding a Hill coefficient of 0.89 ± 0.06. (C, D) CSPs from titration of 100 μM ¹⁵N-StoD-N [1–101] or ¹⁵N-StoD-C [134–233] with 100 μM ubiquitin mapped onto the surface of the (C) StoD-N [1–101] crystal structure or (D) StoD-C [134–233] model shown in Fig 2C, respectively. Cartoon and surface representations of the same view are shown for clarity for each model. Peak disappearances due to line broadening are shown in red, peak shifts greater than 0.1 ppm are shown in orange, and those between 0.05 and 0.1 ppm are shown in yellow.

Figure S16.. StoD-N and StoD-C interact with ubiquitin with K_D ∼100 μM.

(A, B) MST measured for a titration of 61 nM–2 mM ubiquitin with 40 nM fluorescently labelled (A) StoD-N [1–101] and (B) StoD-C [134–233]. 20% LED power and 20% laser power and data from the combined thermophoresis and T-jump contributions were used. A representative example of a trace is shown where the normalized fluorescence signal is given relative to that predicted for the fully bound state.

Figure S17.. Titration of StoD-N and StoD-C with ubiquitin.

(A, B) Overlay of ¹H, ¹⁵N-HSQC spectra of 100 μM ¹⁵N-labelled (A) StoD-N [1–101] and (B) StoD-C [134–233] before (black) and after (orange) addition of 100 μM human ubiquitin acquired in 20 mM Tris–HCl, pH 7.5, 150 mM NaCl, and 1 mM TCEP.

Figure S18.. StoD-C can form a ternary complex with UBE2E1 and ubiquitin.

(A) Model for the interaction between StoD-C (grey surface) and UBE2E1 (cyan cartoon) conjugated to human ubiquitin (orange cartoon), adopting the “closed” conformation observed in available E2–Ub/RING-E3 complex structures. The position of ubiquitin is extracted from the RNF4/UBE2D1–Ub crystal structure (4) (PDB ID 4AP4) after superimposition of UBE2D1 chain B (RMSD 0.82 Å over 150 residues) with UBE2E1 from the model in Fig 2C using Superpose (2). (B) Model for the interaction between StoD-C (grey surface) and UBE2E1 (cyan cartoon) conjugated to human ubiquitin, adopting “open” conformations of uncomplexed E2–Ub. The first position of ubiquitin (orange cartoon) is extracted from the UBE2D1–Ub crystal structure (5) (PDB ID 3UGB) after superimposition of UBE2D1 (RMSD 0.85 Å over 145 residues) with UBE2E1 from the model in Fig 2C using Superpose (2). The second ubiquitin (wheat cartoon) has been manually positioned to show that the conjugated ubiquitin cannot access the ubiquitin-binding surface of StoD-C. (A, B) In both (A) and (B), CSPs from titration of 100 μM ¹⁵N-StoD-C [134–233] with 100 μM ubiquitin (Figs 5D and S13B) are mapped onto the surface of StoD-C: peak disappearances due to line broadening are shown in red, peak shifts greater than 0.1 ppm are shown in orange, and those between 0.05 and 0.1 ppm are shown in yellow. The side chain of the catalytic cysteine of UBE2E1 is shown in magenta as spheres. (C) Overlay of ¹H, ¹⁵N-HSQC spectra of 100 μM ¹⁵N-labelled StoD-C [134–233] before (black) and after addition of 100 μM UBE2E1 (cyan), then after further addition of 100 μM human ubiquitin (orange), all acquired in 20 mM Tris–HCl, pH 7.5, 150 mM NaCl, and 1 mM TCEP. * denotes examples of characteristic CSPs for UBE2E1 (cyan; Fig S6A) and ubiquitin (orange; Fig S17B), which indicate that StoD-C can simultaneously interact with UBE2E1 and ubiquitin.

Figure S19.. Human ubiquitin does not interact directly with UBE2E1.

(A) ¹H, ¹⁵N-HSQC spectrum of ¹⁵N-labelled 100 μM human ubiquitin acquired in 20 mM Tris, pH 7.5, and 150 mM NaCl with backbone amide assignments for all non-proline residues shown. The zoomed inset corresponds to the region of the spectra indicated by the dashed box. Backbone ¹H and ¹⁵N assignments were obtained from BMRB entries 68 (6) and 2,573 (7), respectively. (B) Overlay of ¹H, ¹⁵N-HSQC spectra of 100 μM ¹⁵N-labelled ubiquitin before (black) and after addition of 100 μM UBE2E1 (cyan) acquired in 20 mM Tris–HCl, pH 7.5, 150 mM NaCl, and 1 mM TCEP.

Figure S20.. Titration of human ubiquitin with StoD variants.

(A–C) Overlay of ¹H, ¹⁵N-HSQC spectra of 100 μM ¹⁵N-labelled ubiquitin before (black) and after addition of 100 μM (A) StoD-N [1–101] (red), (B) StoD-C [134–233] (gold), or (C) StoD-FL [1–233] (purple) all acquired in 20 mM Tris–HCl, pH 7.5, 150 mM NaCl, and 1 mM TCEP.

Figure 6.. StoD preferentially binds to diubiquitin.

(A) CSPs from titration of 100 μM ¹⁵N-ubiquitin with StoD-N [1–101], StoD-C [134–233], or StoD-FL [1–233] mapped onto the surface of human ubiquitin (28) (PDB ID 1UBQ). Cartoon and surface representations of the same view are shown for clarity for each model. Peak disappearances due to line broadening are shown in red, peak shifts greater than 0.1 ppm are shown in orange, and those between 0.05 and 0.1 ppm are shown in yellow. (B) Fluorescence intensity measured for a titration of 16 nM–500 μM StoD with 40 nM fluorescently labelled K48-linked or K63-linked diubiquitin. The fluorescence signal was taken relative to that of the fully bound state and shown as an average of three independent dilution series. The data were fitted with a 4PL fit, yielding Hill coefficients of 0.88 ± 0.09 (K48Ub₂) and 0.77 ± 0.05 (K63Ub₂). (C, D) CSPs shown in (A) for StoD are mapped onto the surface of (C) K48-linked (33) (PDB ID 1AAR) and (D) K63-linked (34) (PDB ID 2JF5) diubiquitin.

Figure S21.. StoD-N and StoD-C are connected by a flexible linker in the full-length protein.

(A) The sequence of StoD showing the StoD-N [1–101] (red) and StoD-C [134–233] (gold) constructs used for our structural studies. * denotes residues predicted to be disordered by the RONN algorithm (8) and are largely confined to the interdomain linker region. (B) Overlay of ¹H, ¹⁵N-HSQC spectra of 100 μM ¹⁵N-labelled StoD-FL [1–233] (black), ¹⁵N-labelled StoD-N [1–101] (red), and ¹⁵N-labelled StoD-C [134–233] (gold) all acquired in 20 mM Tris–HCl, pH 7.5, 150 mM NaCl, and 1 mM TCEP.

Figure S22.. Controlled synthesis of diubiquitin.

(A) Overlay of SEC elution profile for K48-linked (green) and K63-linked (red) diubiquitin following its synthesis from proximally blocked ubiquitin_G76C and distally blocked ubiquitin_K48R or ubiquitin_K63R, respectively. Peaks corresponding to diubiquitin (Ub₂) and unreacted monoubiquitin (Ub) are indicated. (B) Analysis of the main eluted species in (A) by 15% SDS–PAGE and Coomassie staining confirms the formation and isolation of diubiquitin.