Structure and dynamics of a mycobacterial type VII secretion system

Abstract

Mycobacterium tuberculosis is the cause of one of the most important infectious diseases in humans, which leads to 1.4 million deaths every year¹. Specialized protein transport systems-known as type VII secretion systems (T7SSs)-are central to the virulence of this pathogen, and are also crucial for nutrient and metabolite transport across the mycobacterial cell envelope^2,3. Here we present the structure of an intact T7SS inner-membrane complex of M. tuberculosis. We show how the 2.32-MDa ESX-5 assembly, which contains 165 transmembrane helices, is restructured and stabilized as a trimer of dimers by the MycP₅ protease. A trimer of MycP₅ caps a central periplasmic dome-like chamber that is formed by three EccB₅ dimers, with the proteolytic sites of MycP₅ facing towards the cavity. This chamber suggests a central secretion and processing conduit. Complexes without MycP₅ show disruption of the EccB₅ periplasmic assembly and increased flexibility, which highlights the importance of MycP₅ for complex integrity. Beneath the EccB₅-MycP₅ chamber, dimers of the EccC₅ ATPase assemble into three bundles of four transmembrane helices each, which together seal the potential central secretion channel. Individual cytoplasmic EccC₅ domains adopt two distinctive conformations that probably reflect different secretion states. Our work suggests a previously undescribed mechanism of protein transport and provides a structural scaffold to aid in the development of drugs against this major human pathogen.

Fig. 1. Cryo-EM structure of the intact ESX-5 inner-membrane complex of M. tuberculosis.

a, Genetic organization of the esx-5 locus of M. tuberculosis H37Rv, which was cloned and expressed in M. smegmatis MC²155. b–e, Cryo-EM density of the intact ESX-5 inner-membrane complex of M. tuberculosis, zoned and coloured for every individual component. Components are inner EccB₅ (dark green), outer EccB₅ (light green), EccC₅ (blue), inner EccD₅ (beige), outer EccD₅ (orange), EccE₅ (purple) and MycP₅ (red). The full complex is 28.5 nm in width and 20 nm in height, and has an absolute stoichiometry of 6:6:12:6:3 for EccB₅:EccC₅:EccD₅:EccE₅:MycP₅. b–e, Side (b), top (c) and bottom (d) views and a top cross-section (e) of the complex at the membrane level, highlighting the arrangement of the 165-TMH region. Inset, top cross-section of an extracted dimeric unit. f, Single dimer viewed from the centre of the intact complex, highlighting the central EccC₅ TMH bundle and the position of MycP₅ with its active site directed towards the inside of the periplasmic cavity. g, Ribbon model of the M. tuberculosis ESX-5 assembly.

Fig. 2. MycP₅ drives EccB₅ hexamerization and stabilization of the membrane complex.

a, Transparent assembly of intact M. tuberculosis ESX-5, with EccB₅ and MycP₅ coloured as in Fig. 1. b, Complete structure of monomeric M. tuberculosis EccB₅, highlighting its overall fold and domains. c, Top and bottom view of the EccB₅–MycP₅ periplasmic assembly with one unit (EccB₅ dimer and MycP₅ monomer) as ribbon model, highlighting the active site of MycP₅ in yellow. CD, central domain; TM, TMH. d, EccB₅ dimerization site, highlighting the C-terminus of outer EccB₅ that is wrapped around the R1 and R4 domains of the adjacent inner EccB₅ monomer, the conserved GIPGAP motif of EccB₅ (in yellow) and the interactions of the EccB₅ dimer with loop 2 and the linker connection of MycP₅. e, Transparent map of M. tuberculosis ESX-5 without copurified MycP₅, with EccB₅ highlighted in dark green. The high flexibility of EccB₅ and the overall heterogeneity of the membrane complex in the absence of MycP₅ is indicated by curved lines. f, EccB₅–MycP₅ interaction surface, highlighting the three buried tryptophans. g, Angle variation range between protomers of the MycP₅-bound (+) and two unbound (−) states (I and II). Intra, between two protomers within a dimer: 62.7°, 62.4° and 62.4° (MycP₅-bound); 61.8°, 61.3° and 61.5° (unbound, I); 61.4°, 62.3° and 60.2° (unbound, II). Inter, between two protomers of adjacent dimers: 57.5°, 57.3° and 57.5° (MycP₅-bound); 58.5°, 58.3° and 58.3° (unbound, I); and 57.9°, 58.3° and 59.6° (unbound, II).

Fig. 3. A basket formed by the EccB₅ TMHs holds three four-TMH bundles of EccC₅.

a, Angled view from the outside of the complex, showing the TMH and N terminus of an outer EccB₅ interacting with a pocket formed by TMH8, TMH10 and TMH11 of inner EccD₅ from the adjacent barrel. b, Side cross-section through the EccB₅ basket that contains the EccC₅ TMH bundles. Light blue densities depict the three copies of EccC₅ TMH2 that form the central pyramid. Two TMHs of EccC₅ were removed for clarity. Sizes indicate the inner diameters of the EccB₅ basket. c, Side cross-section through an EccB₅ basket, showing that the EccC₅ TMH bundle does not interact with outer EccB₅ from its own dimer, but instead forms lipid-mediated interactions with the outer EccB₅ TMH of the adjacent dimer. Lipids are shown in gold. d, Top view of the central EccB₅ basket and the EccC₅ TMH bundles. Dashed line marks the TMHs that belong to one dimeric unit. e, As in d, highlighting the lipid-rich environment. In the central area that surrounds the EccC₅ pyramid, lipids are not clearly distinguishable (which suggests fluidity in this area). f, g, Surface model displaying the hydrophobicity of an EccC₅ TMH bundle (f) and the EccB₅ basket (g). Hydrophilic amino acids are shown in turquoise, and hydrophobic residues are shown in sepia.

Fig. 4. EccC₅ adopts an extended and a contracted conformation.

a, Side cross-section of density maps, showing the extended and contracted conformation of EccC₅. The periplasmic and cytoplasmic chambers formed by EccB₅–MycP₅ and by EccC₅ upon closing are highlighted. Homology models of the three C-terminal NBDs of EccC₅ are fitted in the cytosolic densities. Cytosolic bridge components are coloured as in Fig. 1. b, Model of the intact T7SS inner-membrane complex, highlighting the two conformations of EccC₅.

Extended Data Fig. 1. Purification of the M. tuberculosis ESX-5 membrane complex.

a, Genetic organization of the esx-5 locus of M. tuberculosis H37Rv, which has been cloned and expressed in M. smegmatis MC²155. b, BN-PAGE and western blot analysis using an anti-EccB₅ antibody of DDM-solubilized membranes from M. smegmatis MC²155 expressing M. xenopi or M. tuberculosis ESX-5 (ESX-5_mxen or ESX-5_mtb, respectively). Experiment was reproduced three times. c, d, Coomassie-stained SDS–PAGE (c) and BN-PAGE (d) of Strep- and SEC-purified ESX-5_mtb membrane complexes. e, Negative-stain electron microscopy analysis of ESX-5_mtb membrane complexes shown in c and d. Experiments in c–e were replicated three times. f, BN-PAGE and Coomassie staining of Strep-purified ESX-5_mtb complexes without nucleotides (−) or in the presence of nucleotides ATP, ADP or the transition-state analogue ADP–AlF₃. Upon purification, in the presence of either nucleotide, the higher-molecular-weight species of the membrane complex becomes more prominent. g, SDS–PAGE and Coomassie staining of the same samples as in f, showing a similar SDS–PAGE protein pattern between the four conditions. Experiment in f, g was performed three times. h, i, Coomassie-stained SDS–PAGE (h) and BN-PAGE (i) of Strep- and SEC-purified ESX-5_mtb membrane complexes in the presence of ADP–AlF₃. j, Negative-stain electron microscopy of the same sample as in h, i, showing improved sample homogeneity as compared to purifications in the absence of nucleotides, as shown in e. Experiment shown in h–j was performed twice.

Extended Data Fig. 2. Cryo-EM data collection and single-particle reconstruction procedure.

a, b, This figure relates to the initial Talos-Arctica-collected dataset (a) and the first higher-resolution Titan-Krios-collected dataset (b).

Extended Data Fig. 3. Cryo-EM data collection and single-particle reconstruction procedure.

This figure relates to the second high-resolution Titan Krios collected dataset.

Extended Data Fig. 4. Single-particle reconstructions of the ESX-5_mtb membrane complex.

a–c, Angular distribution plots, local-resolution estimations and Fourier shell correlation (FSC) plots of the C1 reconstruction of the entire MycP₅-bound ESX-5_mtb membrane complex (a), and C1 reconstructions of the two heterogeneous MycP₅-unbound ESX-5_mtb membrane complexes (b, c). d, e, Local-resolution estimation and FSC plot for the C1-refined periplasmic map (d) and the map of the cytosolic bridge (e). f, Examples of cryo-EM densities and corresponding models.

Extended Data Fig. 5. Top cross-sections through the intact ESX-5_mtb membrane complex.

a, MycP₅ trimer top view, highlighting the pore formed at the periplasmic side. b, Section through the periplasmic assembly at the EccB₅–MycP₅ interface, showing the position of the protease domain sitting on top of inner EccB₅. c, Section through the periplasmic assembly at the EccB₅ dimer interface level, highlighting the MycP₅ linker connection to the TMH. d, Top view of the six membrane protomers with the closed EccC₅ TMH pyramid at the centre. e, Top cross-section through the six membrane protomers, highlighting 153 of the 165 TMHs. At the central area towards the cytosol, the three-EccC₅ TMH pyramid opens up in a manner similar to an iris. MycP₅, the protease domain of which interacts with the protomer containing inner EccB₅, interacts with the outer EccD₅ barrel of the adjacent protomer at the membrane level. At the membrane level, the angle between protomers within a dimer and between adjacent protomers of different dimers differs by only 0.5°. f, Top section displaying the region below the inner leaflet of the inner membrane, highlighting a further opening of the EccC₅ gated pore and the lower part of the EccB₅ basket, formed by EccB₅ N termini. g, At the cytosolic level, the angle between protomers differs to that at the membrane level. As such, the angle between protomers within a dimer grows to 65.3°, while the angle between adjacent protomers of different dimers decreases to 54.7°. The change in angles between the membrane and cytosolic regions of protomers is caused by MycP₅ binding, which induces a slight tilting to the protomers that it binds via inner EccD₅. h, Section through the lower region of the cytosolic bridge, containing the DUF domain of EccC₅ and the cytosolic domain of inner EccD₅. i, Same view as in h, but overlaid with the EccC₅ extended state, highlighting the radial extension of the EccC₅ NBD1, NBD2 and NBD3 almost parallel to the inner membrane. j, Same view as in h, but then overlaid with the EccC₅ contracted stated.

Extended Data Fig. 6. Hexameric EccB₅ adopts a triangular conformation in the periplasm.

a, b, Side (a) and top (b) view of an intact ESX-5_mtb assembly in which inner EccB₅ and outer EccB₅ are coloured as in Fig. 1 and the rest of the components are transparent. c, A V-shaped EccB₃ dimer (PDB 6SGY) was fitted into the M. smegmatis ESX-3 dimer cryo-EM density (EMDB EMD-20820) together with the corresponding dimeric ESX-3 model for the membrane and cytosolic domains (PDB 6UMM) using the Chimera fit in map tool. This composite dimer model was subsequently trimerized, on the basis of our full ESX-5 map reconstructions. The clashing of EccB₃ periplasmic domains between the dimers, towards the central area, in this hybrid model are highlighted in red. d, Upon MycP₅ binding to the assembly, the periplasmic EccB₅ dimer is rotated by 52°, avoiding the clashes observed in c. Angles were measured by aligning the hybrid model and the ESX-5_mtb model at the membrane level. Subsequently, centres of mass were defined for the combined R1 domains of each EccB₃ and EccB₅ dimer (at the base of the dimer) and for every R2 and R3 EccB monomer (toward the tips of the EccB dimer). Planes defined by these three points were generated for both EccB₃ and EccB₅ dimers and angles were measured between these two planes. EccB₃ dimer is shown as a ribbon model and EccB₅ model is shown as zoned density. e, Compared to the V-shaped EccB₃ dimer (ribbon model), the angle between the two EccB₅ monomers (zoned density) grows by 48° upon MycP₅ binding. EccB dimer angles were calculated by measuring the angle between the centres of mass of the R2 and R3 domains of each EccB protomer in relation to the centre of mass of both R1 domains. f, Side views of the TMH region of inner EccD₅, depicted as a ribbon model, and the TMH and N terminus of an interacting outer EccB₅, depicted as zoned density. An array of lipids found in the EccD₅ barrel (but also surrounding this inner EccD₅–EccB₅ interaction site) are depicted in gold. g, Bottom view of the lower cytosolic area of the EccB₅ basket, formed by EccB₅ N termini (residues 10–48) and depicted with the interacting pocket formed by TMH10, TMH11 and TMH8 (not shown for clarity) of inner EccD₅ of the adjacent protomer. The EccB₅ N terminus is also buttressed in this position by a short helix (residues 119–130) of outer EccD₅, which connects outer EccD₅ TMHs with its cytosolic domain, and also by part of the inner EccD₅ loop (residues 307–315) that subsequently folds along the stalk and DUF domain of EccC₅. h, Same map as in g, but viewed from the top. k, Superposition of inner EccB₅ and outer EccB₅, highlighting conformational differences between the two, which are the result of the interaction with MycP₅.

Extended Data Fig. 7. MycP₅ caps a periplasmic cavity with its active site directed towards the lumen.

a, Side and top view of an intact ESX-5_mtb assembly with MycP₅ coloured as in Fig. 1 and the rest of the components transparent. Insets, side and top views of the periplasmic assembly with EccB₅ in white, the MycP₅ density shown in transparent red and loop 5 of MycP₅ depicted in solid red at a higher threshold, to highlight it capping the periplasmic pore. Loop 5 folds along the protease domain, towards the pore formed by the MycP₅ trimer. At higher thresholds, loop 5 caps this pore. b, Top and side view of a dimer of EccD₅ barrels, of which one barrel (left) binds via inner EccD₅ to the MycP₅ TMH. c, Top or bottom view of MycP₅ trimers depicted in grey with the loops that are involved in MycP₅–MycP₅ interactions depicted in different colours. d, Side and bottom views showing the MycP₅–MycP₅ interactions mediated by the same domains depicted in the same colours as in e.

Extended Data Fig. 8. MycP₅ drives hexamerization of periplasmic EccB₅ and complex stability.

a, Cryo-EM density map of a MycP₅-free ESX-5_mtb membrane complex, zoned and coloured as in Fig. 1. In the absence of MycP₅, the periplasmic domains of EccB₅ display high flexibility. The rest of the membrane complex displays increased heterogeneity when compared to the MycP₅-bound map. b, Map of difference created by subtracting the MycP₅-free map from the MycP₅-bound map. c, Overlay of a and b. d, MycP₅-bound map in red and the two MycP₅-free maps in blue and green. e, A model of the MycP₅-bound map, in which MycP₅ and residues 84–504 of EccB₅ were removed, was fitted into the models of the two MycP₅-free maps, as described in Methods. Models were aligned at one EccD₅ barrel (dark dotted circle), revealing substantial variations and shifts between the three maps. Top inset shows that there is consistent variation between all three maps at the membrane level (EccD₅ barrel). Middle inset shows variations between maps at cytosolic level (EccB₅ N-terminal helix, residues 20–38). Bottom inset highlights inner EccD₅ from the EccD₅ barrel that was used for the alignment, showing that overall protomer structure does not change in the absence of MycP₅. f, Dimers from every individual map, colour-coded the same as in d (different shades), were extracted and aligned to each other on one EccD₅ barrel (left) as in e. Insets from these alignments, derived from both protomers, show that all three dimers of the MycP₅-bound map show little to no variation, whereas the two MycP₅-free maps show a higher degree of heterogeneity between dimers.

Extended Data Fig. 9. MycP₅ creates more interaction points between protomers and dimers.

a, Transparent surface model of a MycP₅-free map with one EccB₅ dimer at the periplasmic side, highlighting the interfaces of the protomers from a dimer. In the absence of MycP₅, the two protomers from within one dimer exhibit two interactions: an EccB₅–EccB₅ interaction between their periplasmic domains, and a cytosolic one between EccB₅–inner EccD₅. The dimer further contacts the two immediate protomers of adjacent dimers through the mentioned EccB₅–EccD₅ cytosolic interaction. b, Transparent surface model of the MycP₅-bound map, highlighting the interface of protomers from a dimer. On top of the mentioned contacts, in the presence of MycP₅, protomers from a dimer interact with each other at the periplasmic side through inner EccB₅–MycP₅–outer EccB₅, while MycP₅ further anchors the periplasmic assembly to the stable EccD₅ raft through EccB₅–MycP₅–inner EccD₅. MycP₅ also guides dimer–dimer interactions. By stabilizing the three EccB₅ dimers in the triangle assembly, MycP₅ promotes inner EccB₅–outer EccB₅ interactions between opposing protomers from adjacent dimers. Additionally, MycP₅ promotes dimer–dimer contacts through MycP₅–MycP₅ interactions in the periplasm. Colour-coded legend applies to a, b. c, d, Inside (c) and outside (d) view of a dimer containing EccB₅, MycP₅ and the TMHs of EccD₅. For purposes of clarity, one EccD₅–EccB₅ protomer is coloured in blue, and the second protomer is in green and the MycP₅ in red. Interactions between protomers of a dimer are highlighted and colour-coded as in a, b. e, Top view of a surface model missing the periplasmic domains of EccB₅ and MycP₅. The planes of protomers in which MycP₅ binds inner EccD₅ are tilted by about 5° compared to the MycP₅-unbound ones.

Extended Data Fig. 10. Six lipid-filled EccD₅ barrels form a central raft.

a, Side and top view of an intact ESX-5_mtb assembly, in which inner EccD₅ and outer EccD₅ are coloured as in Fig. 1 and the rest of the components are transparent. b, Top view of the membrane region of the ESX-5_mtb model overlaid with observed lipids, coloured in bright yellow. c, Same view as b, but showing only the lipids. d, Same image as c, but rotated 90° to show a side view, highlighting a bilayer-like structure. e, Top view of an EccD₅ barrel with observed lipids bound to inner EccD₅. f, Side view of an inner EccD₅ monomer displayed as zoned density and overlaid with observed lipids.

Extended Data Fig. 11. Three four-TMH-bundles of EccC₅ gate a central pore.

a, Side and top view of an intact ESX-5_mtb assembly, in which EccC₅ is coloured in alternating light and dark blue and the rest of the components are transparent. b, Extracted dimeric EccC₅ and the TMHs and N termini of EccB₅ from the same dimer. The 90° inset rotation shows that the TMH of outer EccB₅ is not contacted by the TMHs of EccC₅. c, Top membrane cross-section through a local-resolution map of a C1 full-complex reconstruction, displaying decreased resolution of the central space occupied by the TMHs of EccC₅, compared to the surrounding EccB₅ basket and TMHs of inner EccD₅. d, Top view of the full membrane complex with the EccC₅ TMH pyramid in light blue and the periplasmic EccB₅–MycP₅ in the same colours as in Fig. 1. The EccC₅ TMH pyramid aligns with the periplasmic cavity and the MycP₅-formed pore. MycP₅ top part is partially sectioned, for clarity. Inset showing a 90° rotation side cross-section of the same map. e, Ribbon model highlighting the structural features of the cytosolic bridge.

Extended Data Fig. 12. EccC₅ adopts two separate conformations.

a, Extended conformation in which an EccC₅ NBD1–NBD2–NBD3 model is fitted to highlight the overall position of these domains with respect to the rest of the membrane complex. b, As in a, but for the contracted conformation.