Structural elucidation of the hexameric MmpS4-MmpL4 complex from Mycobacterium tuberculosis

  1. Jennifer C. Earp
  2. Nicolas P. Lichti
  3. Alisa A. Garaeva
  4. Virginia Meikle
  5. Michael Niederweis
  6. Markus A. Seeger

  • Curated by eLife

    eLife logo

    eLife Assessment

    This study provides valuable insights through the elucidation of the first full-length structure of the heterohexameric (MmpS4)₃-(MmpL4)₃ transporter complex from Mycobacterium tuberculosis, advancing understanding of its transport mechanism, linked to virulence and drug resistance. The structural analysis is solid and convincing, offering a clear framework for future mechanistic studies. Major strengths include a comprehensive structural characterization of the complex, though some conclusions require further validation.

Reviewed by

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

Log in to save this article

Abstract

Mycobacterium tuberculosis contains thirteen Mycobacterial membrane protein Large (MmpL) transporters, which belong to the family of secondary active RND transporters. MmpL4 and MmpL5, together with their operon partners MmpS4 and MmpS5, export the mycobacterial siderophore mycobactin and the last resort TB drug bedaquiline. Recently, we determined a structure of the MmpL4 monomer in complex with desferrated mycobactin, which lacked a functionally essential coiled-coil domain predicted to extend far into the periplasm. Here, we present a cryo-EM structure of the hexameric (MmpS4)3-(MmpL4)3 complex, which was enabled by rational disulfide cross-links based on AlphaFold predictions. We observed density for the coiled-coil domain, which protrudes into the periplasmic space at an angle of around 60° relative to the symmetry axis of the MmpL4 trimer. In the context of the hexameric complex, MmpL4’s conformation differs strikingly from the one observed for monomeric MmpL4, which includes formation of a large cavity in the periplasmic domain and rearrangements of conserved proton coupling residues at the transmembrane domain. Our work provides an experimental workflow to obtain single particle cryo-EM structures of labile multiprotein complexes by AlphaFold-informed stabilization of predicted protein interfaces.

Article activity feed

  1. eLife

    eLife Assessment

    This study provides valuable insights through the elucidation of the first full-length structure of the heterohexameric (MmpS4)₃-(MmpL4)₃ transporter complex from Mycobacterium tuberculosis, advancing understanding of its transport mechanism, linked to virulence and drug resistance. The structural analysis is solid and convincing, offering a clear framework for future mechanistic studies. Major strengths include a comprehensive structural characterization of the complex, though some conclusions require further validation.

  2. eLife

    Reviewer #1 (Public review):

    This manuscript adds to the recent, exciting developments in our understanding of the MmpL/S transporters from mycobacteria. This work provides solid support for the trimeric/hexameric arrangement of subunits in the complex, and reveals a possible pathway for substrate translocation.

    Overall, I think this manuscript is a solid body of work that adds to several recent studies from this team and others on the structure and mechanism of the MmpL/S transporter family, particularly MmpL4/S4. The combination of AF, disulfide engineering, and experimental structure is good, though it is a bit puzzling that the experimental structure based on disulfide stabilization of the AF prediction does not recapitulate key elements (MmpS periplasmic domain docking to MmpL, and altered CCD configuration).

    I have no major concerns about this manuscript; I could make only a few minor suggestions for clarification in a revision.

  3. eLife

    Reviewer #2 (Public review):

    Summary:

    The manuscript describes the structure of the Mycobacterium tuberculosis (MmpS4)3-(MmpL4)3 hetero-heximeric transporter complex. The structure was obtained by cryogenic electron microscopy using an engineered construct that cross-links MmpS4 to MmpL4 via a disulfide bond. The position of the disulfide bond was determined using an Alphafold2 model of the hetero-heximer. Although Alphafold2 predicts a symmetric hetero-heximer, the author found that the structure of the coiled-coil domain (CCD) is asymmetric, tilted at about 60˚ relative to the membrane domains, and only contains two of the three alpha helical hairpins, with the third being disordered.

    Strengths:

    The strategy of using Alphafold2 models to guide construct design for experimental structure determination is state-of-the-art, and this work provides a great example of its applications and limitations. I.e., the experimental structure does not fully recapitulate the prediction but provides unexpected results.

    The comparisons between the authors' structures and the previously published structures of the MmpL4 monomer and MmpL5 trimers strengthen the authors' findings.

    Weaknesses:

    A more detailed description of the current mechanistic hypothesis would strengthen the manuscript. The authors state that the two periplasmic domains "are expected to undergo rigid body movements that allow substrate transport through these periplasmic domains similar to the conformational changes observed in the E. coli multidrug efflux pump AcrB". A schematic of the proposed transport cycle, as a supplemental figure that shows the current hypothesis regarding transport, would be beneficial for understanding the previous structures and putting the current structure in context. Outside of "the mechanistic basis of how these conformational changes are coupled to protonation of the DY-pairs", what are the major controversies/open questions regarding the mechanism?

    The authors provide evidence that the cysteine-depleted S4L4 construct is functional, but do not show that the construct with the introduced disulfide bond #5 (D39C MmpS4 and S434C MmpL4) is also functional. Demonstrating this would allow the authors to better interpret their resulting structures.

    The analysis presented in Figure 5 and Supplementary Figure 7 seems to suggest that the authors are proposing that the CCD central cavity acts as a transport pathway for the transported substrate, but I am not sure that this hypothesis is explicitly stated. This makes the reasoning behind the analysis presented unclear. Clarity could be improved by stating that the hypothesis of direct transport of substrate through the CCD central channel is being examined using the structure prediction, and what the implications are for the structure solved with the incompletely formed CCD.

    Given that the results emphasize the flexibility of the CCD, the manuscript would be strengthened by 3D variability analysis either in cryoSPARC or using cryoDRGN (or both). This would allow the authors to better quantify the degree of motion in the CCD and how it may correlate to flexibility in other regions. Further 3D flex reconstruction in cryoSPARC may improve the map quality of the CCD.

  4. eLife

    Reviewer #3 (Public review):

    Summary:

    This manuscript by Earp et al reports cryoEM structures of the hexameric (MmpS4)₃-(MmpL4)₃ complex from Mycobacterium tuberculosis, which belongs to the RND family of transporters and is known to have a role in the export of siderophores and contribute to drug resistance. The experimental workflow showcased involves the design of disulfide pairs using distance constraints obtained from the AlphaFold predicted structure of the hexameric complex. One such disulfide pair was used to determine the ~3.0 Å structures. The structure reveals density for the previously unresolved coiled-coil domain (CCD), a tilted CCD arrangement, and a cavity within the periplasmic domain, which the authors assert is occupied by detergent. Comparison of this complex with the monomer structure of MmpL4 shows conformational variations interpreted to implicate different domains and conserved residues involved in proton coupling, which might be related to the transport mechanism. While the methodological aspects of the manuscript are solid, enthusiasm for the overall advance/significance is less so, with doubts about the relevance of the tilted CCD structure, considering disulfide trapping and an incomplete validation of the claim that the titled CCD represents a stable intermediate conformation. A clear, updated transport mechanism is largely missing from the manuscript.

    Strengths:

    Beautiful structures, AF prediction-experimental validation nexus that could be fine-tuned for different systems/difficult to target complexes.

    Weaknesses:

    Physiological relevance of the tilted CCD conformation. No clear mechanistic model for the transport. While the CCD may indeed be a stable intermediate, the fact that the rest of the trimeric arrangement is unaffected does not fully rule out disulfide trapping as a factor in promoting this. The findings would be strengthened if the same tilted conformation is seen using a different set of disulfides. The significance of the detergent molecule and the new cavity observed could also be better discussed in terms of an updated transport model.

  5. eLife

    Author response:

    We thank the three reviewers for their critical and in-depth assessment of our study. Below you find our comments to the public reviews and our revision plans.

    Public Reviews:

    Reviewer #1 (Public review):

    This manuscript adds to the recent, exciting developments in our understanding of the MmpL/S transporters from mycobacteria. This work provides solid support for the trimeric/hexameric arrangement of subunits in the complex, and reveals a possible pathway for substrate translocation.Overall, I think this manuscript is a solid body of work that adds to several recent studies from this team and others on the structure and mechanism of the MmpL/S transporter family, particularly MmpL4/S4. The combination of AF, disulfide engineering, and experimental structure is good, though it is a bit puzzling that the experimental structure based on disulfide stabilization of the AF prediction does not recapitulate key elements (MmpS periplasmic domain docking to MmpL, and altered CCD configuration).

    I have no major concerns about this manuscript; I could make only a few minor suggestions for clarification in a revision.

    We thank reviewer#1 for this positive assessment of our work. The deviation of the AF prediction from the experimental structure is , in our view, not puzzling. AF does not take the physical properties of proteins into account, but predicts structures based on strong sequence alignments. It therefore does not have “knowledge” about the general flexibility of domains such as the CCD, which is also observed in the corresponding MmpL5 structures, nor does it have knowledge about preferred conformational states. Rather than “failing” to confirm the AF predictions, our cryo-EM structure revealed an unexpected tilted conformation of the CCD. As we outline in comments below, the physiological relevance of the tilted CCD is unclear. Its flexibility might be required to interact with (still elusive) outer membrane protein components to form the fully assembled efflux machinery.

    Reviewer #2 (Public review):

    Summary:

    The manuscript describes the structure of the Mycobacterium tuberculosis (MmpS4)3-(MmpL4)3 hetero-heximeric transporter complex. The structure was obtained by cryogenic electron microscopy using an engineered construct that cross-links MmpS4 to MmpL4 via a disulfide bond. The position of the disulfide bond was determined using an Alphafold2 model of the hetero-heximer. Although Alphafold2 predicts a symmetric hetero-heximer, the author found that the structure of the coiled-coil domain (CCD) is asymmetric, tilted at about 60˚ relative to the membrane domains, and only contains two of the three alpha helical hairpins, with the third being disordered.

    Strengths:

    The strategy of using Alphafold2 models to guide construct design for experimental structure determination is state-of-the-art, and this work provides a great example of its applications and limitations. I.e., the experimental structure does not fully recapitulate the prediction but provides unexpected results.

    The comparisons between the authors' structures and the previously published structures of the MmpL4 monomer and MmpL5 trimers strengthen the authors' findings.

    We thank reviewer#2 for this positive assessment of our work and agree that it is interesting that the experimental structures do not fully agree with the AF predictions (see also comment to reviewer#1).

    Weaknesses:

    A more detailed description of the current mechanistic hypothesis would strengthen the manuscript. The authors state that the two periplasmic domains "are expected to undergo rigid body movements that allow substrate transport through these periplasmic domains similar to the conformational changes observed in the E. coli multidrug efflux pump AcrB". A schematic of the proposed transport cycle, as a supplemental figure that shows the current hypothesis regarding transport, would be beneficial for understanding the previous structures and putting the current structure in context. Outside of "the mechanistic basis of how these conformational changes are coupled to protonation of the DY-pairs", what are the major controversies/open questions regarding the mechanism?

    We thank the reviewer for this valuable comment. We will add a new figure with the model of the MmpL4 transport cycle based on our new data and discuss the proposed molecular transport mechanism in more detail in the main.

    The authors provide evidence that the cysteine-depleted S4L4 construct is functional, but do not show that the construct with the introduced disulfide bond #5 (D39C MmpS4 and S434C MmpL4) is also functional. Demonstrating this would allow the authors to better interpret their resulting structures.

    In the revised version, we will include additional data to assess the functional consequences of cross-linking.

    The analysis presented in Figure 5 and Supplementary Figure 7 seems to suggest that the authors are proposing that the CCD central cavity acts as a transport pathway for the transported substrate, but I am not sure that this hypothesis is explicitly stated. This makes the reasoning behind the analysis presented unclear. Clarity could be improved by stating that the hypothesis of direct transport of substrate through the CCD central channel is being examined using the structure prediction, and what the implications are for the structure solved with the incompletely formed CCD.

    We state clearly in the discussion that the channel through the CCD seems too narrow to let large molecules like mycobactin and bedaquiline pass:[AG1]

    Line 318ff: “ The channel radius of the MmpL4 CCD is very narrow with a minimum of 1.1 Å according to the AlphaFold3 predition (Fig. 5). This is much smaller than the smallest axis of a molecular model of mycobactin molecule of ?? nm as determined from a model of iron-free mycobactin. In addition, the cryo-EM structure of MSMEG_1382 revealed a constriction in the CCD channel [21]. Even though the methionine side chains lining the channel wall are considered to be flexible{Aledo, 2019 #69594}, large conformational changes of the α-helical hairpins relative to each other would be required to allow passage of molecules as large as mycobactin and bedaquiline. The AcrAB-TolC efflux machinery provides an example for such large conformational changes to enable transport of large molecules by iris-like opening and closing movements the outer membrane channel-tunnel TolC [33]. Similar helical twisting may widen the channel of the CCD. Alternatively, it is conceivable that the substrates of MmpL4/MmpL5 are transported along the CCD surface, potentially requiring further protein partners. It is interesting to note that siderophore secretion and drug efflux by MmpL4/MmpL5 systems involves at least two additional proteins, namely the periplasmic protein Rv0455, which was shown to be essential for mycobactin efflux [34] and an outer membrane channel, whose identity remains elusive. A complete molecular understanding of the transport mechanism through the MmpL4/MmpL5 systems hence requires the identification of the missing components and structural information about their interactions.”

    The channel radius of the MmpL4 CCD is very narrow (minimum of 1.1 Å) according to the AlphaFold3 prediction (Fig. 5), and the cryo-EM [AG2] [MN3] structure of MSMEG_1382 revealed a further constriction in the CCD channel [21]. We therefore consider direct substrate transport through the CCD central channel to be physically implausible for molecules of the size of mycobactin and bedaquiline. Even accounting for the flexibility of the methionine side chains lining the channel wall, the large conformational changes of the α-helical hairpins relative to each other would be required to accommodate such large substrates. While iris-like opening movements have been described for TolC in the AcrAB-TolC system [33], those movements widen an already substantially larger channel, and even such dramatic conformational changes would be insufficient to open a channel as narrow as that of the MmpL4 CCD to a diameter permissive for substrate passage. We instead favor a model in which substrates are transported along the outer surface of the CCD, potentially with the assistance of additional protein partners. This is consistent with the observation that MmpL4/MmpL5-mediated siderophore secretion and drug efflux involves at least two further proteins: the periplasmic protein Rv0455, shown to be essential for mycobactin efflux [34], and an as-yet-unidentified outer membrane channel. In this context, the overall flexibility of the CCD - illustrated here by the tilted, incompletely formed conformation - may reflect the conformational dynamics required for interaction with these partner proteins, rather than being directly involved in forming a transport conduit. A complete mechanistic understanding will require identification of the missing components and structural characterization of the fully assembled efflux machinery.

    We do not think that the incompletely formed CCD represents a conformation that is relevant for transport. But it is a demonstration of the overall flexibility of the CCD, which may be required to further open the channel in case the substrates are transported within the CCD tube. Further in-depth experiments will be needed to clarify this interesting question, which is beyond the scope of this paper.

    Given that the results emphasize the flexibility of the CCD, the manuscript would be strengthened by 3D variability analysis either in cryoSPARC or using cryoDRGN (or both). This would allow the authors to better quantify the degree of motion in the CCD and how it may correlate to flexibility in other regions. Further 3D flex reconstruction in cryoSPARC may improve the map quality of the CCD.

    This is a great suggestion. We will include a 3D variability analysisin the revised manuscript.

    Reviewer #3 (Public review):

    Summary:

    This manuscript by Earp et al reports cryoEM structures of the hexameric (MmpS4)3-(MmpL4) )3 complex from Mycobacterium tuberculosis, which belongs to the RND family of transporters and is known to have a role in the export of siderophores and contribute to drug resistance. The experimental workflow showcased involves the design of disulfide pairs using distance constraints obtained from the AlphaFold predicted structure of the hexameric complex. One such disulfide pair was used to determine the ~3.0 Å structures. The structure reveals density for the previously unresolved coiled-coil domain (CCD), a tilted CCD arrangement, and a cavity within the periplasmic domain, which the authors assert is occupied by detergent. Comparison of this complex with the monomer structure of MmpL4 shows conformational variations interpreted to implicate different domains and conserved residues involved in proton coupling, which might be related to the transport mechanism. While the methodological aspects of the manuscript are solid, enthusiasm for the overall advance/significance is less so, with doubts about the relevance of the tilted CCD structure, considering disulfide trapping and an incomplete validation of the claim that the titled CCD represents a stable intermediate conformation. A clear, updated transport mechanism is largely missing from the manuscript.

    We thank reviewer#3 for these useful comments, which we will address during the revision of the manuscript. In particular, we plan to include a scheme of an updated transport model.

    Strengths:

    Beautiful structures, AF prediction-experimental validation nexus that could be fine-tuned for different systems/difficult to target complexes.

    Weaknesses:

    Physiological relevance of the tilted CCD conformation. No clear mechanistic model for the transport. While the CCD may indeed be a stable intermediate, the fact that the rest of the trimeric arrangement is unaffected does not fully rule out disulfide trapping as a factor in promoting this. The findings would be strengthened if the same tilted conformation is seen using a different set of disulfides. The significance of the detergent molecule and the new cavity observed could also be better discussed in terms of an updated transport model.

    We believe that there was a misunderstanding about our interpretation of the tilted CCD. As a matter of fact, it must be a stable intermediate, otherwise no density would have been observed for it in the cryo-EM maps. Despite being a stable intermediate, it is indeed unlikely that it represents a conformational state that is relevant/required for transport. Firstly, only the upright, complete CCD can bridge the periplasm. because . Secondly, the structure was determined in detergent and lacks additional protein binder partners, which might stabilize the upright conformation of the CCD . It is also conceivable, as the reviewer pointed out, that disulfide cross-linking may have caused the tilt. However, as we wrote in the manuscript, we do not think that cross-linking caused this striking asymmetry of the CCD, because the three MmpL4 and MmpS4 chains are basically symmetrical in the C1-processed data (see also Figure 2E):

    Line 182 ff: “To assess whether there are asymmetries in other parts of the structure, we superimposed the individual protomers of the (MmpS4)3-(MmpL4)3 complex analyzed using C1 symmetry (Fig. 2E). Apart from the two resolved α-helical hairpins, the MmpL4 core domains and the resolved parts of MmpS4 differ by a RMSD of less than 0.6 Å and are therefore structurally identical considering the map resolution of around 3 Å. The fact that the core domains of MmpS4 and MmpL4 do not deviate between the protomers argues against the possibility that the cross-links established between them cause the (asymmetric) tilt of the CCD.”

    Regarding the DDM binding site, we will indeed include an updated transport model. That said, we wish to be cautious, because we lack experimental proof that MmpL4 can in fact transport DDM.

  6. Version published to 10.64898/2026.01.07.698164 on bioRxiv