Structural insights into the mechanism of protein transport by the Type 9 Secretion System translocon

Abstract

Secretion systems are protein export machines that enable bacteria to exploit their environment through the release of protein effectors. The Type 9 Secretion System (T9SS) is responsible for protein export across the outer membrane (OM) of bacteria of the phylum Bacteroidota. Here we trap the T9SS of Flavobacterium johnsoniae in the process of substrate transport by disrupting the T9SS motor complex. Cryo-EM analysis of purified substrate-bound T9SS translocons reveals an extended translocon structure in which the previously described translocon core is augmented by a periplasmic structure incorporating the proteins SprE, PorD and a homologue of the canonical periplasmic chaperone Skp. Substrate proteins bind to the extracellular loops of a carrier protein within the translocon pore. As transport intermediates accumulate on the translocon when energetic input is removed, we deduce that release of the substrate-carrier protein complex from the translocon is the energy-requiring step in T9SS transport.

Fig. 1. Purification of a substrate-bound extended T9SS translocon complex.

a, Size exclusion chromatography profiles of Twin-strep-tagged SprA complexes purified by Streptactin-affinity chromatography from wild-type (wt; black) and ΔgldL mutant (red) strains. b, Top: Coomassie-stained sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS–PAGE) gels of the peak fractions from a. Band identities were assigned on the basis of peptide fingerprinting. SprAʹ and SprAʺ arise from proteolytic clipping in the loop between strands 24 and 25 of the SprA barrel. Proteins detected in all samples are labelled in black. Protein bands only visible in Peak I of the ΔgldL purification are labelled in red. Only trace levels of Plug (grey) were detected in Peak I. Similar data were obtained from two independent preparations. Bottom: the different classes of SprA complexes identified in peaks I (solid arrows) and II (dashed arrows) by cryo-EM. c, Cryo-EM volume of the substrate-bound Extended Translocon complex coloured by subunit. The detergent micelle and other unmodelled densities are shown in white at a lower contour level. The lower panel is cut through to show the position of the substrate CTD. d, Atomic model of the Extended Translocon shown in cartoon representation with the substrate CTD in spacefill representation and the modelled parts of the SprE glycans shown as atomic spheres. Source data

Fig. 2. Substrate binding to the Extended Translocon.

For clarity, only the core subunits of the Extended Translocon are shown in panels a, c and d. a, Cartoon representation with the front of the translocon cut away and the CTD shown as a green cartoon. b, The unmodelled EM volume overlaid on a cartoon representation of the Extended Translocon highlighting the detergent belt at the presumed location of the OM (dark grey volume), the unmodelled C-terminal domain of SprE (salmon volume) and the disordered density putatively identified as a portion of the trapped RemZ substrate (dark green volume). c, Equivalent slices through the cryo-EM volumes for the empty PorV complex (EMD-0133) (top) and the CTD-bound Extended Translocon (bottom) viewed from the periplasm. d, Horizontal slab view of SprA (blue) showing surfaces (yellow) in contact with the CTD (green ribbons). e, Interactions of the PorV loops with the CTD. f, The CTD surface in contact with the translocon is poorly conserved among type A CTDs except for a single Lys (position 1110 in RemZ) highlighted by a white *. Surface conservation was calculated using ChimeraX. g, Highly conserved RemZ Lys1110 forms a hydrogen bond with the main chain carbonyls of PorV Asn 87 and F115.

Fig. 3. Structural features involved in the assembly of the Extended Translocon.

a,b, The surfaces of SprE (a) and SkpA (b) that interact with other translocon components are highly conserved. Surface conservation was calculated using ConSurf. c, Closeup in ribbon representation of the region highlighted by a dashed oval in b showing the interaction between PorV and the ordered C-terminal helix of one copy of SkpA. Residues involved in the contact are labelled. d,e, The N terminus of SprA (d, blue) folds within the core of the SkpA chamber. Residues critical for formation of this fold, or that interact with SkpA, are highly conserved as assessed with ConSurf (e). f, Cartoon representation of SprE in rainbow colouring from the N (blue) to C terminus (red). The modelled portions of the glycans are shown as atomic spheres. g,h, PorD is bound to the Finger Region of SprE. Twin-strep–SprA (g) or SprE–Twin-strep (h) complexes were affinity purified from the indicated backgrounds where sprE(Δfingers) and sprE(Δcterm) are deletions of the structurally unresolved regions of the SprE Finger Region and C terminus respectively. The complexes were immunoblotted for SprE–Twin-strep (α-SprE), or PorD (α-PorD) or far western blotted by incubation with PorD followed by immunodetection of PorD (PorD + α-PorD). Similar data were obtained from two independent preparations. i, Inferred location of PorD (represented by an AlphaFold2 model) in the Extended Translocon complex. PorD interacts with the disordered portion of the SprE Finger Region (dashed line). Source data

Fig. 4. Functional importance of the Extended Translocon components.

a, Secretome analysis of culture supernatants. The samples were separated by SDS–PAGE and stained with Coomassie blue. b,e, Spreading (gliding) morphology on agar of Extended Translocon component mutants in the presence (b) or absence (e) of porV. Scale bars, 4 mm. c, Cell surface exposure of the adhesin SprB assessed by protease protection. Strains expressing a fusion between HaloTag and the final 448 amino acids of SprB (HaloTag-SprB₄₄₈), and containing a ΔporV mutation to reduce endogenous proteolysis, were incubated with Proteinase K and the detergent Triton X-100 as indicated. Reactions were stopped immediately (t₀) or after 15 min (t₁₅) and analysed by immunoblotting with anti-Halotag antibodies. d, Analysis of the secretion of a model substrate protein comprising a fusion between a signal sequence (SS), mCherry (mCh) and the T9SS-targeting C-terminal domain of RemA (CTD). SS-mCh is a control fusion protein lacking a CTD. Cells were grown to exponential or stationary phase as indicated, separated into cell and supernatant fractions and analysed by anti-mCherry immunoblotting. The successively processed forms of the fusion proteins are indicated to the right of the blots. f,g, Effects of removing Extended Translocon components on SprA complex composition. f, Size exclusion chromatography profiles of Twin-strep–SprA complexes purified by Streptactin-affinity chromatography from the indicated strains (wt and ΔgldL background profiles are from Fig. 1a). The inset shows a magnified view of Peak I and Peak I′. g, Coomassie-stained SDS–PAGE gels of the peak fractions from f. The composition of the ~100 kDa band (blue labels) was determined by peptide fingerprinting. The identities of other bands were assigned by analogy to the preparations in Fig. 2b and supported by whole-sample peptide fingerprinting (Supplementary Data 1). SprAʹ and SprAʺ indicate fragments of SprA that arise from proteolytic clipping. Similar data were obtained from two independent preparations. a–d, Similar data were obtained from three biological repeats. a,b,e–g, A twin-strep–sprA background was used (wt) except where indicated (untagged). Source data

Fig. 5. Protein–protein interactions during protein export by the T9SS translocon.

a,b, Single-molecule tracking of fluorophore-labelled HaloTag-SprA or SprE-HaloTag in live F. johnsoniae. Images were acquired by stroboscopy with a 50 ms frame rate and a one-frame-on-six-frames-off cycle. a, Representative trajectories in the indicated strains. The trajectories link the single-molecule localizations in successive frames. Trajectories are coloured as ‘immobile’ (cyan) or ‘mobile’ (red) according to the classification in b. Trajectories are 8.45 s in duration (25 frames). Scale bar, 1 μm. See also Supplementary Video 2. b, Distributions of the apparent diffusion coefficients (D*) of individual tracked molecules in the indicated backgrounds. Molecules with D* ≤ 0.0017 μm² s⁻¹ are classed as ‘immobile’ (blue) and molecules with D* above this value classed as ‘mobile’ (red). The number of tracks analysed (n) and the percentages of immobile and mobile molecules (blue:red) are given above each distribution. The protein synthesis inhibitor chloramphenicol was added to ‘+Cm’ cells. c, Updated model of the T9SS translocon mechanism for Type A substrates based on the results of this study. Substrates with a Type B CTD are anticipated to utilize an analogous cycle employing an alternative carrier protein. The Extended Translocon, shown here as composed of SprA, SprE and SkpA, is the physiologically relevant translocation complex. (1) PorV docks onto the lateral opening of SprA. (2) The CTD of a substrate protein binds to the loops of PorV that protrude into the SprA pore, assisted by interactions between the CTD and the SprA cap. (3) Energetic input forces PorV away from the translocon, pulling the PorV-bound substrate molecule through the lateral opening into the extracellular environment. d, Schematic of the energy chain that drives both T9SS protein transport and gliding motility. A motor complex transduces the energy of protons flowing down their electrochemical gradient (1) into rotary mechanical energy (2) which it transfers through a periplasm-spanning arm to a Hub complex at the OM (3). The Hub complex distributes this mechanical energy both to the Extended Translocon (4a) to extract carrier protein–substrate complexes (5a) and to a mobile track (4b) that propels carrier protein-anchored surface adhesins (5b).

Extended Data Fig. 1. Comparison of substrate−containing translocon complexes.

a, Graphical summary of the SprA complexes isolated from different genetic backgrounds. The bars in the graphs are coloured according to the structure cartoon key. The complexes and their relative abundances were determined by cryoEM analysis of the proteins in the indicated size−exclusions chromatography peaks. The identities of bound substrate molecule(s) are given where these can be assigned from the structural data. These substrate assignments are consistent with proteomics data (Extended Data Fig. 3a, Supplementary Data 1). RemZ (Fjoh_0803) is a homologue of the gliding adhesin RemA. NucA (Fjoh_4723) is a predicted nuclease. b, CryoEM volumes for substrate-containing translocon complexes isolated by in vivo trapping or in vitro reconstitution. The substrate-free PorV complex is shown for comparison. In each case the complex is shown in a cut-through of the protein viewed from within the plane of the membrane (left) and viewed from the periplasm (right). PorV, gold; SprA, blue; substrate CTD, green; PPI, detergent micelle, and unmodelled substrate regions, grey. Disordered substrate densities emerging from the SprA barrel are indicated with red arrows. The RemZ structures are those determined from the peak I′ fraction prepared from a ΔgldLΔsprE background.

Extended Data Fig. 2. Proteomics characterization of the substrates present in SprA complexes purified from a ΔgldL backround and cryoEM workflow for the Extended Translocon structure.

a, Peptide mass spectrometry of the substrate proteins most frequently associated with the SprA complexes isolated from a ΔgldL backround. The RemZ peptides are from direct sequencing of the ~110 kDa band of Peak I in Fig. 1b. The lack of coverage in the first part of the protein may reflect the paucity of trypsin sites in this region. The FspB and NucA peptides are from whole sample proteomics of Peak II in Fig. 1b. The peptide numbering is from the N terminus of the native precursor sequence. Regions of the proteins corresponding to the signal sequence are indicated by grey outline boxes and to the T9SS CTDs by red outline boxes. b, Image processing workflow for peak I of the SprA complexes purified from a ΔgldL background. c, Gold-standard Fourier Shell Correlation (FSC) plots calculated in Relion for the two focussed local refinements used to construct the composite Extended Translocon volume. d, The two focused local refinement volumes colored by local resolution estimates calculated in Relion. (i) The SprA-PorV-PPI-substrate volume is shown as both an intact (Left) and clipped (Right) surface. (ii) The SkpA-SprE volume.

Extended Data Fig. 3. CryoEM analysis resolves two comformers of the RemZ CTD bound to the T9SS translocon.

a, Image processing workflow for the ΔgldLΔsprE peak I′ sample depicting separation of the two conformers of the RemZ CTD. b, Gold standard Fourier Shell Correlation plots calculated in Relion for the two conformer volumes. c, Slab view in cartoon representation through the models for the full RemZ-bound translocon complex with both conformers of the CTD present. Proteins are colored by conformer (conformer 1/conformer 2): SprA, blue/purple; PorV, yellow/orange; RemZ CTD, dark green/light green; PPI, grey/grey). d, Comparison of the interactions between the two RemZ CTD conformers and SprA Q452. Experimental densities are shown to the right for the region indicated by the dashed box. e, Closeup of the bound CTD showing both conformations in cartoon representation (dark green/light green) and with SprA shown in a surface representation (blue) with the front face removed. f, Density and model for the strand of RemZ that is remodelled between the two conformers.

Extended Data Fig. 4. Workflow for the translocon complexes in peak II of the ΔgldL SprA purification.

a, Image processing workflow for the SprA complexes in peak II of the ΔgldL SprA purification. b, Gold-standard Fourier Shell Correlation (FSC) plots calculated in Relion for the NucA-translocon complex structure determined from peak II of the ΔgldL SprA purification. c, The refinement volume of the NucA-translocon complex shown as a clipped surface and colored by local resolution estimates calculated in Relion.

Extended Data Fig. 5. All CTDs studied bind in similar modes to the SprA/PorV complex.

a, Structure-informed alignment of the sequences of the F. johnsoniae Type A CTDs that are structurally characterized in complex with the T9SS translocon in this work. The most conserved residues are highlighted in red. The highly conserved and functionally important lysine is indicated (*). b, The highly conserved lysine in Type A CTDs exhibits distinct interactions with PorV in different substrate-translocon complexes. c, The CTDs of different Type A substrates exhibit similar modes of binding to the translocon. Aligning the substrate-translocon structures determined in this work on SprA (blue) reveals only small variations in the positions of the substrate CTD (green) and PorV loops (orange). d, Nature of the CTD-translocon interaction illustrated for two substrate-translocon complexes. The electrostatic surface of the CTD domains is shown together with atom (side chain) and cartoon (backbone) representations of all SprA (blue) and PorV (orange) residues within 4 Å of that surface. The interactions are primarily with hydrophobic regions of the CTD surface.

Extended Data Fig. 6. Characterization of SprE, SkpA, and PorV variants.

a, b, Functional analysis of the contact between the C-terminus of SkpA and the periplasmic face of PorV assessed by (a) deletion of the PorV periplasmic loop (residues 193-199, allele porV(Δloop)) or (b) deletion of the C-terminal SkpA helix (residues 203-218, allele skpA(Δhelix)). The PorV (a) or SkpA (b) proteins used in these experiments have C-terminal or N-terminal Twin-Strep epitope tags, respectively. T9SS export was assessed using a model substrate protein comprising a signal sequence (SS), mCherry (mCh), and the T9SS-targeting C-terminal domain of RemA (CTD). Cells and culture supernatants were separated and analysed by immunoblotting. The substrate protein was detected using mCherry antibodies and PorV-TwinStrep was detected by TwinStrep antibodies. PorV and mCherry in (a) were analysed from the same samples but on duplicate immunoblots, each with a GroEL loading control. Wt, wild type strain. *, mCh degradation product. c-i, Characterization of SprE variants. All sprE mutants were constructed in a sprE-twinstrep background (wt). A strain with untagged sprE (untagged) is included as a control. c-e, The Finger Region and C-terminal region of SprE are not required for T9SS function. Strains expressing SprE with either the Finger Region (residues E445 to T550) replaced with a (GS)₁₀ peptide (sprE(Δfinger)) or with the C-terminal tail region removed (residues F759 to P870, sprE(Δcterm)) were analysed. f-i, O-glycosylation is important for SprE stability. SprE variants with the indicated single amino acid substitutions of identified (structurally or [S538] biochemically) or predicted (T550 from a previously identified Bacteroidota O-glycosylation motif^,) glycosylated residues, or containing all eight substitutions (‘All’) were analyzed. c, f, Whole cell immunoblotting for SprE-TwinStrep variants. GroEL serves as a loading control. d, h, i, Secretome analysis of culture supernatants. The samples were separated by SDS-PAGE and stained with Coomassie Blue. e, g, Spreading (gliding) morphology of colonies on agar. Scale bar: 4 mm. j, The structure of the skpA-containing operon in F. johnsoniae and P. gingivalis. A sequence extension of F. johnsoniae skpA relative to P. gingivalis skpA is shown hatched. Attempts to delete skpB in F. johnsoniae and P. gingivalis have been unsuccessful (our observations and 27) suggesting that SkpB is an essential protein in these organisms. a-i, Similar results were observed in three biological repeats. Source data

Extended Data Fig. 7. Analysis of the function and expression levels of fusions to Extended Translocon components.

a-h, The protein of interest in each panel is epitope tagged with a TwinStrep peptide. The strain with the tagged protein was used as the wild type (wt) background for the mutant strains. A strain in which the protein of interest has not been epitope tagged is labelled ‘untagged’. i-k, Test strains have a HaloTag coding sequence fused to the end of the chromosomal sprE gene (sprE-halo allele). wt, wild type. a, b, i, k, TwinStrep epitope tagging of SprE or SkpA or HaloTag fusion to SprE does not affect Type 9 secretion or gliding function. a, i, Secretome analysis of culture supernatants. The samples were separated by SDS-PAGE and stained with Coomassie Blue. b, k, Spreading (gliding) morphology of colonies on agar. Scale bar: 4 mm. c-h, Analysis of the effects of removal of Extended Translocon components on the levels of other Extended Translocon components. Whole cell immunoblotting with anti-Strep antibodies for the presence of TwinStrep-tagged SprA (c,f), SprE (d,g), and SkpA (e,h) in the indicated genetic backgrounds. Immunoblotting with anti-GroEL antibodies serves as a loading control. k, Whole cell immunoblotting of the indicated strains with Halotag (α-Halo) and GroEL (α-GroEL) antibodies. GroEL serves as a loading control. Similar data were obtained for two (blots) or three (other analyses) biological repeats. Source data

Extended Data Fig. 8. In vitro reconstitution of substrate protein binding to the T9SS translocon.

a, Model substrate proteins are transported by the T9SS. Analysis of the export of model substrate proteins from wild-type F. johnsoniae cells (wt) or a strain lacking the T9SS translocon (ΔsprA). The substrate proteins consist of a tripartite fusion between a signal sequence (SS), mCherry (mCh), and the T9SS-targeting C-terminal domain of either RemA (CTD_RemA) or FspA (CTD_FspA). The two left hand lanes show experiments with a control fusion protein lacking a CTD (SS-mCh). Cells were grown in MM medium for 4 h, separated into cell (C) and supernatant (SN) fractions, and analyzed by anti-mCherry immunoblotting. The successively processed forms of the fusion proteins are indicated. Removal of the signal sequence marks transport across the inner membrane by the Sec apparatus and removal of the CTD marks transport across the outer membrane by the T9SS. Similar data were obtained for at least two biological repeats. b, In vitro reconstitution of RemA CTD binding to SprA complexes. Size exclusion chromatography profiles of affinity-purified TwinStrep-SprA complexes from wild type (wt, left panel) and ΔporV mutant (right panel) strains following incubation with a mCherry-CTD_RemA fusion protein. The elution profile of the total protein content is indicated in black (relative absorbance at 280 nm), and the elution profile of mCherry-CTD_RemA is indicated in red (fluorescence emission at 610 nm following excitation at 587 nm). The peak containing TwinStrep-SprA complexes is indicated. c,d Cartoon representation of the (c) mCherry-CTD_RemA and (d) mCherry-CTD_FspA complexes. The bound CTD is shown in surface representation (dark green) within the SprA barrel (blue). e, Overlay of cartoon representations of the CTDs and PorV loops from the translocon complexes containing the RemA and FspA CTDs. Source data

Extended Data Fig. 9. Cryo-EM processing workflows for in vitro reconstituted model substrate-translocon complexes.

a, b, (i) Image processing workflow. (ii) Gold-standard Fourier Shell Correlation (FSC) plot used for global resolution estimation as determined with Relion. (iii) Volume colored by local resolution estimate as determined within Relion. (iv) Closeup of model and volume for the bound CTD domain within the SprA barrel. a, mCherry-CTD_RemA-translocon complex. b, mCherry-CTD_FspA-translocon complex.

Extended Data Fig. 10. Workflows for the translocon structures not described in other Figures.

a-h Cryo-EM workflows for those SprA complexes described in Fig. 4 that are not shown in other Extended Data Figures. The background strains from which the complexes were purified and, where relevant, the size exclusion chromatography peak from which the sample was taken are as denoted.