Structure of the Legionella Virulence Factor, SidC Reveals a Unique PI(4)P-Specific Binding Domain Essential for Its Targeting to the Bacterial Phagosome

Abstract

The opportunistic intracellular pathogen Legionella pneumophila is the causative agent of Legionnaires' disease. L. pneumophila delivers nearly 300 effector proteins into host cells for the establishment of a replication-permissive compartment known as the Legionella-containing vacuole (LCV). SidC and its paralog SdcA are two effectors that have been shown to anchor on the LCV via binding to phosphatidylinositol-4-phosphate [PI(4)P] to facilitate the recruitment of ER proteins to the LCV. We recently reported that the N-terminal SNL (SidC N-terminal E3 Ligase) domain of SidC is a ubiquitin E3 ligase, and its activity is required for the recruitment of ER proteins to the LCV. Here we report the crystal structure of SidC (1-871). The structure reveals that SidC contains four domains that are packed into an arch-like shape. The P4C domain (PI(4)P binding of SidC) comprises a four α-helix bundle and covers the ubiquitin ligase catalytic site of the SNL domain. Strikingly, a pocket with characteristic positive electrostatic potentials is formed at one end of this bundle. Liposome binding assays of the P4C domain further identified the determinants of phosphoinositide recognition and membrane interaction. Interestingly, we also found that binding with PI(4)P stimulates the E3 ligase activity, presumably due to a conformational switch induced by PI(4)P from a closed form to an open active form. Mutations of key residues involved in PI(4)P binding significantly reduced the association of SidC with the LCV and abolished its activity in the recruitment of ER proteins and ubiquitin signals, highlighting that PI(4)P-mediated targeting of SidC is critical to its function in the remodeling of the bacterial phagosome membrane. Finally, a GFP-fusion with the P4C domain was demonstrated to be specifically localized to PI(4)P-enriched compartments in mammalian cells. This domain shows the potential to be developed into a sensitive and accurate PI(4)P probe in living cells.

Fig 1. Crystal structure of SidC871 (aa. 1–871).

(A) Schematic diagram of the domain structure of SidC. SidC contains an N-terminal SNL domain, an insertion domain (INS), a C-terminal PI(4)P-binding domain (P4C), and a C-terminal domain with unknown function (named CTD, colored in brown). (B) Ribbon diagram of the overall structure of SidC. (C) Overall structure of SidC with the SNL and INS domains shown in surface and the P4C and CTD in ribbon. The ubiquitin ligase active site in the SNL domain is colored in red. (C) and (B) have the same orientation. (D) A 90⁰ rotated view of (B). (E) A 90⁰ rotated view of (C). Color scheme from (A) to (E): the SNL domain is in blue; the INS domain is in green; the P4C domain is in pink; and the CTD domain is in brown.

Fig 2. Conformational dynamics of the SNL and INS domains.

(A) Stereo view of the Cα trace of the SNL (blue) and INS (green) domains of SidC in overlay with our previously reported SNL-INS domain structure (light brown; PDB ID: 4TRH). The three areas that have major conformational changes are labeled with I, II, and III, respectively. The INS domain in SidC871 is bent by about 30⁰ relative to the SNL domain. (B) Zoom-in view of the conformational changes at the catalytic site. C46 is shifted away from H444 and D446 in SidC871 (blue) compared with the SidC542 structure (brown). (C) Zoom-in view of the conformational change at the non-conserved loop (residue 59–66). (D) The SNL domain (SidC542) forms a stable complex with UbcH7~Ub. SDS gel of the size exclusion chromatography fractions from the sample containing SNL domain with ubiquitin-charged UbcH7. (E) The INS domain is involved in the binding of the SNL domain with UbcH7~Ub. SDS-gel of the size exclusion chromatography fractions from the sample containing the SNLΔINS domain with ubiquitin-charged UbcH7. UbcH7~Ub did not co-migrate with the SNLΔINS domain.

Fig 3. The P4C domain of SidC specifically binds with PI(4)P.

(A) Ribbon diagram of the P4C domain. (B) Molecular surface of the P4C domain. The surface is colored based on electrostatic potential with positively charged region in blue (+5 kcal/electron) and negatively charged surface in red (-5 kcal/electron). (C) Back view of the surface model of the P4C domain. (D) Liposome floatation assays. Input and float samples were analyzed by SDS-PAGE and immunoblotted with anti-GFP antibodies. Recombinant GFP-P4C showed selective binding to PI(4)P-positive liposomes but not the liposomes with other components. (E) Quantification of liposome floatation assays from three independent experiments. Error bars represent standard deviation. (F) Fluorescent images of liposome binding by GFP-P4C. Only the liposomes containing PI(4)P showed strong binding of GFP-P4C. Scale bar = 10 μm (G) Quantification of liposome binding of GFP-P4C. GFP fluorescent signals were normalized to red Dil dye signals on the same liposome and averaged on three randomly picked liposomes. Error bars represent standard deviation. ** p < 0.01; *** p < 0.001.

Fig 4. Determinants of PI(4)P recognition and membrane targeting by the P4C domain.

(A) Ribbon diagram of the P4C domain. Residues that play a role in membrane interaction are highlighted in sticks. R652 and R638 form a pocket for the binding of the PI(4)P headgroup. The L1 (W642, W643, and F644) and L2 (W704 and F705) loops are colored in green, and form the membrane interacting motif (MIM). (B) Liposome floatation assay for P4C mutants. The R652Q and the L1/L2 (W642S/W643S/F644S/W704S/F705S) mutants completely abolished PI(4)P- liposome binding. (C) Quantification of liposome floatation assays of P4C mutants averaged from three independent assays. (D) Fluorescent images of liposome binding by GFP-P4C mutants. Mutations of cationic residues in the PI(4)P binding pocket and hydrophobic residues at the two membrane insertion loops significantly reduce the binding to PI(4)P-containing liposome. Scale bar = 10 μm. (E) Quantification of liposome binding of GFP-P4C mutants. GFP fluorescent signals were normalized to red Dil dye signals on the same liposome and averaged on three randomly picked liposomes. Error bars represent standard deviation. ** p < 0.01; *** p < 0.001.

Fig 5. Intracellular localization of fluorescent protein fusions of the P4C domain from SidC.

(A) GFP-tagged wild type P4C domain localized to the perinuclear region and plasma membrane, while this localization is altered in PI(4)P-binding defective P4C mutants in N2A cells. The nucleus was stained with DAPI. Wild type P4C showed both plasma membrane and perinuclear localization. The R638Q, L1, and L2 mutants had a more diffuse localization while the R652Q and the L1/L2 mutants were completely cytosolic. (B) Quantification of the intracellular localization of GFP-P4C represented by the percentage of the fluorescence intensities at the plasma membrane (PM), perinuclear region (PNR), and other areas of the cell. Error bars represent standard deviation. The measurements were averages of three randomly selected cells. (C) Confocal images of localizations of the plasma membrane marker mCherry-PLCδ-PH with GFP-tagged PI(4)P probes in N2A cells. (D) Confocal images of colocalizations of the Golgi marker DsRed-GalT with GFP-tagged PI(4)P probes in N2A cells. Scale bar = 10 μm in all images. (E) Quantification of the intracellular localization of PI(4)P probes. Error bars represent standard deviation. ** p < 0.01; *** p < 0.001.

Fig 6. The interface between the P4C domain and the SNL domain.

(A) and (B) Two orthogonal views of the interface between the P4C and SNL domains. Hydrophobic residues at the interface are shown in sticks. The P4C domain is colored in pink and the SNL domain in blue. (C) SidC743 C46A does not form a stable complex with ubiquitin-charged UbcH7, as indicated by SDS-PAGE gel analysis of fractions from size exclusion chromatography experiment. (D) SidC743 C46A/L629R forms a stable complex with UbcH7~Ub as demonstrated by the co-migration of SidC743 C46A/L629R with UbcH7~Ub on the size exclusion column.

Fig 7. PI(4)P stimulates the ubiquitin E3 ligase activity of SidC.

(A) In vitro ubiquitin ligase activity assay with SidC542 in the absence of liposomes and in the presence of liposomes containing PC/PS or PC/PS/PI(4)P. The reactions were stopped at the indicated time points and the samples were analyzed by SDS-PAGE. The decreased intensity of ubiquitin bands indicates the consumption of free ubiquitin during the ligase reaction. (B) Percentage of free ubiquitin left in the reaction at each indicated time points averaged from three independent experiments. (C) and (D) In vitro ubiquitin ligase activity assay with SidC743. The ubiquitin ligase activity is enhanced in the presence of PI(4)P. (E) and (F) In vitro ubiquitin ligase activity assay with SidC743 L629R. This mutant has higher ubiquitin ligase activity even in the absence of PI(4)P, presumably due to the open conformation caused by this mutation. (G)-(H) and (I)-(J) In vitro ubiquitin ligase activity assay with SidC743 R652Q andSidC743 L1/L2. No stimulation of the ubiquitin ligase activity by PI(4)P was observed with these two PI(4)P-binding defective mutants.

Fig 8. PI(4)P-binding by the P4C domain is essential for the recruitment of ubiquitinated species to the LCV.

(A) Immuno-fluorescent staining of ubiquitinated species on the LCV. U937 Cells were infected with indicated L. pneumophila strains at an MOI of 1 for 2hrs and samples were fixed prior to immunostaining with antibodies against Legionella or m-ubiquitin (FK1). Strains: WT: L. pneumophila Philadelphia-1 strain Lp02; dotA: the Dot/Icm type IV secretion system defective strain Lp03; ΔsidC-sdcA: the SidC and SdcA double deletion mutant of the Lp02 strain; ΔsidC-sdcA(pSidC), ΔsidC-sdcA(pSidC R652Q), ΔsidC-sdcA(pSidC L1/L2), and ΔsidC-sdcA(pSidC L1/L2/R652Q): ΔsidC-sdcA strain complemented with a plasmid expressing either wild type or PI(4)P binding-defective mutants. (B) Percentage of cells containing ubiquitin positive LCVs counted from three independent experiments (at least 150 vacuoles were scored in each experiment). ** p < 0.01; *** p < 0.001.

Fig 9. A schematic model of SidC functions at the LCV surface.

SidC/SdcA is translocated into the cytosol through the Dot/Icm apparatus. SidC anchors to the LCV membrane through the binding to PI(4)P by its P4C domain. The binding of PI(4)P allows the P4C domain to move away from the SNL domain and exposes the ubiquitin ligase active site. The activated SidC/SdcA ubiquitinates unknown factors, presumably proteins residing on the LCV that determine the identity of the bacterial phagosome.