Figure 1.
Structures of individual domains from the Salmonella SPI-1 basal body.
(A) Schematic representation of the proteins composing the basal body. The domain organisation for PrgH (green) and InvG (orange) is shown on the left, and their proposed localization in the EM map of the injectisome is shown on the right. (B) Ribbon representation of the structures of monomeric InvG22–178, PrgH170–392, and PrgH11–120, with their localization in the basal body shown. The boundaries of residues that were observable in the density are indicated.
Table 1.
Data collection and refinement statistics for the structures of PrgH11–120, PrgH170–392 and InvG22–178.
Figure 2.
Illustration of the 2-step ring structure modelling approach for the PrgH periplasmic domain.
(A) In the first step, fixed-backbone symmetric docking was performed starting from the structure of the monomer. The resulting two clusters of ring arrangements cl1 and cl2 differ in the orientation of the C-terminal domain (B). More sampling was then performed, focusing in the vicinity of the two candidate structures, using perturbation symmetric docking runs (see methods) (C). At this step, the backbone is allowed to move to allow for more efficient energy discrimination. While Cluster 2 quickly diverges from the starting-point seed structure (C-cl2), Cluster 1 runs sink into a lower-energy funnel (C-cl1) (starting structures for the perturbation runs are indicated with vertical dashed lines). The final, low-energy ensemble (indicated below the horizontal dashed line in (B-cl1)) shows highly converged features in terms of the backbone conformation and side-chain packing along the interface (D). Only two adjacent subunits of the 24mer complex are shown for simplicity. The RMSDs are computed for backbone atoms of the entire modelled 24mer complex in (A), while in (C) RMSD values are reported for a single dimeric interface relative to the lowest-energy sampled model. The EM map EMD-1875 is used to restrain both steps of docking calculations, although in the last flexible-backbone stage the weights are reduced to one-half relative to the first, rigid-backbone step.
Figure 3.
Models of the PrgH and InvG rings.
(A) Ribbon representation of the lowest-energy models obtained for PrgH cytoplasmic domain (bottom, dark green), PrgH periplasmic domain (centre, light green) and InvG periplasmic domain (top, orange). (B) Fit of the PrgH (green) and InvG (orange) ring models in the SPI-1 T3SS EM map (EMD-1875) [11]. A close view for the fit of each individual subunit in the EM map density is shown on the right, illustrating the good fit of each subunit within the density.
Figure 4.
Experimental validation of the ring models.
(A) The interface between two subunits for each of the three models is shown. Selected residues involved in contacts are highlighted, which were mutated for secretion assays. (B) Coomassie-stained SDS-page gel of proteins secreted by S. Typhimurium strains with a chromosomal deletion of the genes coding for PrgH or InvG, complemented with plasmids containing the corresponding genes, or mutants. The flagellin protein FliC is used as a loading control. The effector protein SipA is secreted when the bacteria express wild-type proteins, but mutation of Gln 97 to leucine in InvG, Gly 322 to tyrosine in PrgH periplasmic domain, or both Leu 20 and Leu 87 to alanine in PrgH cytoplasmic domain show impaired secretion. (C) Electron micrographs of purified T3SS injectisome particles for various InvG or PrgH mutants. Deletion of the cytoplasmic domain of PrgH leads to the formation of injectisome particles that lack the needle (bottom panel, red arrows). Mutation of Gly 322 to tyrosine in the PrgH periplasmic domain (middle panel) leads to the observation of a mix of assembled basal body (red arrow) and individual rings which could consist of either the inner- or outer-membrane components (blue arrows). Finally, mutation of Gln 97 to leucine in InvG severely disrupts the assembly of the T3SS, with only few single rings visible (top panel, red arrows). The white bar indicates 100 nm.
Figure 5.
Molecular basis for the SPI-1 basal body assembly.
(A) Overlaid view of the interface formed by the RBMs of adjacent molecules for InvG (orange) and PrgH (green for domain 1, blue for domain 2) in the ring models. In all three cases, the orientation of the domains is similar, with the two helices stacking against the three-stranded sheet. (B) Electrostatic surface representation of individual subunits from the PrgHperiplasmic (bottom) and InvG (top) ring models. The charged surfaces forming the basis of the interaction between RBMs are indicated in grey boxes. (C) Electrostatic surface representation of the InvG ring model (top) and PrgH periplasmic domain ring model (bottom). The negative charge at the bottom of the InvG ring, as well as the positive charge on the top of the PrgH ring, is shown, suggesting a charge-driven mechanism for the assembly of the Salmonella SPI-1 T3SS basal body. The positively charged interior of the InvG N0 ring is also visible.
Figure 6.
Crystal contacts reflect the oligomerization interface.
Overlay of the crystal contacts (grey structures) and the oligomeric interface in the reported ring models, for the cytoplasmic domain of PrgH (A), periplasmic domain of PrgH (B) and periplasmic domain of InvG (C). In all three cases, the crystallographic contacts are formed with an interface that is similar to the interface in the ring model, although the curvature imposed by the oligomeric state differs, leading to some differences in specific side-chain interactions.