Structural and Functional Characterization of the Type Three Secretion System (T3SS) Needle of Pseudomonas aeruginosa

Abstract

The type three secretion system (T3SS) is a macromolecular protein nano-syringe used by different bacterial pathogens to inject effectors into host cells. The extracellular part of the syringe is a needle-like filament formed by the polymerization of a 9-kDa protein whose structure and proper localization on the bacterial surface are key determinants for efficient toxin injection. Here, we combined in vivo, in vitro, and in silico approaches to characterize the Pseudomonas aeruginosa T3SS needle and its major component PscF. Using a combination of mutagenesis, phenotypic analyses, immunofluorescence, proteolysis, mass spectrometry, atomic force microscopy, electron microscopy, and molecular modeling, we propose a model of the P. aeruginosa needle that exposes the N-terminal region of each PscF monomer toward the outside of the filament, while the core of the fiber is formed by the C-terminal helix. Among mutations introduced into the needle protein PscF, D76A, and P47A/Q54A caused a defect in the assembly of the needle on the bacterial surface, although the double mutant was still cytotoxic on macrophages in a T3SS-dependent manner and formed filamentous structures in vitro. These results suggest that the T3SS needle of P. aeruginosa displays an architecture that is similar to that of other bacterial needles studied to date and highlight the fact that small, targeted perturbations in needle assembly can inhibit T3SS function. Therefore, the T3SS needle represents an excellent drug target for small molecules acting as virulence blockers that could disrupt pathogenesis of a broad range of bacteria.

Keywords: Pseudomonas aeruginosa; T3SS needle; immunofluorescence microscopy; mutagenesis; structure; type III secretion system; virulence.

FIGURE 1

Sequence alignment of needle proteins from different bacteria. PscF from Pseudomonas aeruginosa, YscF from Yersinia sp., MxiH from Shigella sp., PrgI from Salmonella sp. The residues mutated in this study are indicated with a green asterisk. Numbers refer to the PscF sequence. Secondary structure elements from the PrgI structure are indicated underneath the sequence alignment. The figure was generated with ESPript, using the new ENDscript server (Robert and Gouet, 2014). Conserved and similar residues are shown in red and blue boxes, respectively.

FIGURE 2

T3SS activity in Pseudomonas PscF mutants. (A) The cytotoxicity of P. aeruginosa strains toward macrophage cells was measured by monitoring LDH release after 2 and 3 h of infection using a multiplicity of infection (MOI) of 5. The PscF deletion mutant (ΔF), D76A, and P47A/Q54A strains show a higher statistical difference at both 2 and 3 h of infection compared to wild-type complemented strain (ΔF/Fwt). LDH measurements were corrected with values corresponding to cells that were not infected; we considered as 100% of cytotoxicity a 1% Triton X-100 treated well. Pairwise differences relative to ΔF/Fwt based on the Tukey test are indicated: ns, non-significant, ^∗∗∗P < 0.001, ^∗P < 0.05. Cytotoxicity of original CHA strain is also reported. (B) Western blot of the supernatant (secreted proteins) and total bacteria (expression) fractions after centrifugation were developed with anti-PopB and anti-PcrV antibodies. The expression of PscF variants was confirmed by loading total bacterial extracts on SDS- 15% PAGE. The Western blot was developed using anti-PscF antibodies obtained using a monomeric 6His-PscF (see section “Materials and Methods”). The D76A strain secretes almost no PopB or PcrV while P47A/Q54A secretes less of both translocators as compared to other mutants and the wild-type strain. LasB was used as loading control for the secreted protein, and DsbA as loading control for bacterial expression and as a lysis control for the supernatant (data not shown).

FIGURE 3

Characterization of PscF wild-type and mutant filaments. (A) Negative-staining electron microscopy images of purified PscF wild-type (left panel), PscF P47A/Q54A (middle panel) and PscF D76A (right panel) visualized at the same protein concentration (0.1 mg/ml). Scale bar: 100 nm. (B) Diameter distribution of PscF filaments measured on EM images shown in (A). The diamonds represent the average diameter. Wild-type PscF has an average diameter of 4.72 ± 1.02 nm (n = 1390) while PscF P47A/Q54A filaments are thinner with average diameters of 4.15 ± 0.97 nm (n = 986). Statistical analysis: ^∗∗∗P < 0.001. (C) Trypsin digestion profiles of purified wild-type PscF and P47A/Q54A filaments were analyzed by 16.5% Tris-Tricine and LC/ESI Mass spectrometry. Both proteins are stable with a major cleavage-band at 6.5 kDa (one asterisk) corresponding to fragments 37–93. Only for the wild-type protein is visible a second lower band of 4.5 kDa (residues 57–93) (two asterisks). A scheme of trypsin digestion kinetics determined by MS analysis is reported below, with the corresponding identified fragments. The gray box represents the 6His tag, black lines show the position of D76, P47, and Q54 residues mutated in this study. All trypsin-cleavable sites (lysines, K and arginines, R) are shown with their relative position on the top scheme.

FIGURE 4

Localization of PscF-needles on the surface of Pseudomonas aeruginosa. (A) P. aeruginosa CHA strain isolated on cystic fibrosis patient depleted for the pscF gene (ΔpscF) and complemented with pIApG-pscF constructs were grown in T3SS-inducing conditions. PscF needles were visualized by immuno fluorescence on fixed bacteria using anti-PscF (in green) or anti-PcrV (in cyan). SYTO24 was used to visualize bacteria (in red). As a negative controls we used P. aeruginosa ΔF and ΔV, while ΔF/Fwt and ΔV/Vwt were used as a positive control. Only few PscF and PcrV spots were visible on P. aeruginosa strains carrying a PscF-D76A or a PscF-P47A/Q54A. (B) MicrobeJ reconstitution of PscF distribution around the bacteria. n = number of spots in the image.

FIGURE 5

Presence of PscF and PcrV on Pseudomonas aeruginosa PscF-mutant strains. (A) Images of immunolabeled fixed bacteria in Figure 4 were analyzed with MicrobeJ to detect and associate each PscF or PcrV spot to only one bacterium. Frequency was considered as the percentage of bacteria with at least one PscF and PcrV spot on the total counted bacteria. About 30–40% of cells had needles with the PcrV tip protein on their surface, with the exception of the D76A strain with only 6.9 ± 1.6% (PscF) and 3.2 ± 1.2% (PcrV), and the P47A/Q54A strain with 8.9 ± 2.4% (PscF) and 5.6 ± 2.5% (PcrV) of cells with spots. (B) Distribution of the number of PscF spots/needles per bacterial cell. Between 630 and 2300 individual bacteria were counted. P. aeruginosa wild-type and mutant strains had between 1 to 3 PscF spots per cell, except for D76A (in pink) and P47A/Q54A (in green) that presented only one or two spots per cell. Overall comparisons using the Kruskal–Wallis’ test indicates significant differences between classes (P < 0.001). Pairwise differences relative to wild-type based on Dunn’s post hoc test are shown: ^∗P<0.05.

FIGURE 6

Needle reconstruction for PscF wild-type and mutants. (A) The PrgI Solid-state NMR structure was used as a model for the construction of PscF in the Forward and Reverse orientations. Mutants D76A and P74A/Q54A were constructed on the PscF-forward model. The exterior view (left) and lumen view (right) of the T3SS needle models are colored by surface electrostatic potentials, at pH 7, using the APBS plugin for PyMol. (B) Distribution of conserved residues distribution on PrgI and PscF needle models in Forward and Reverse orientations. (C) Hydrophobicity distribution of the different models shown in (B).

FIGURE 7

Model of the PscF P47A/Q54A and PscF D76 assembled filament. (A) Ribbon representation of PscF monomer carrying the two mutations: D76A and P47A/Q54A. (B) Electrostatic surface representation of the two mutant reconstituted filaments with a channel view. (C) Top view of a PscF-forward wild-type model, colors show residues implicated in binding of the T3SS-inhibitor phenoxyacetamide (Bowlin et al., 2014). (D) Comparison between reverse and forward PscF and PscF-P47A/Q54A filaments. PscF-forward model (with the N-termini exposed toward the outside), with Lys36, Lys29, and Lys21 solvent-exposed, Lys56 and Lys59 less accessible, and Lys84, Arg72, and Arg75 completely buried. PscF-reverse model (with the C-termini exposed toward the outside), with solvent accessible residues: Lys84, Arg75, Lys59, and Lys36. The PscF-P47A/Q54A forward model showing accessibility of the same residues as compared to the PscF-forward wild-type model except for the Lys56 and Lys36 that are slightly less or more trypsin-accessible, respectively (indicated by two arrows). All needle models are represented as surfaces with one monomer of PscF represented as a cyan ribbon. All trypsin-cleavable sites are shown as sticks.