Journal of Molecular Biology
Volume 332, Issue 3, 19 September 2003, Pages 657-674
Journal home page for Journal of Molecular Biology

Dynamical Properties of the MscL of Escherichia coli: A Normal Mode Analysis

https://doi.org/10.1016/S0022-2836(03)00851-9Get rights and content

Abstract

The mechanosensitive channel (MscL) is an integral membrane protein which gates in response to membrane tension. Physiological data have shown that the gating transition involves a very large change in the conformation, and that the open state of the channel forms a large non-specific pore with a high conductance. The Escherichia coli channel structure was first modeled by homology modeling, starting with the X-ray structure of the homologous from Mycobacterium tuberculosis. Then, the dynamical and conformational properties of the channel were explored, using normal mode analysis. Such an analysis was also performed with the different structures proposed recently by Sukharev and co-workers. Similar dynamical behaviors are observed, which are characteristic of the channel architecture, subtle differences being due to the different relative positioning of the structural elements. The ability of particular regions of the channel to deform is discussed with respect to the functional and structural properties, implied in the gating process. Our results show that the first step of the gating mechanism can be described with three low-frequency modes only. The movement associated to these modes is clearly an iris-like movement involving both tilt and twist rotation.

Introduction

Mechanosensitive channels are ubiquitous proteins, located in the inner membrane of bacteria.1 The presence of highly conserved homologues in various bacteria (Gram+ and Gram−) suggests that they play an important role in diminishing turgor pressure upon osmotic downshock,2 when bacteria are shifted from an high to a low osmolarity environment. Indeed, these channels have remarkable mechanical properties: they open in response to membrane strain, preventing cell lysis and the breakdown of the membrane. They act as “safety valves” for the bacteria cell. Small solutes (ions and water molecules) and much larger entities (proteins like thioredoxin-12 kD) can pass through the pore with no ionic selectivity.1., 3.

Three types of mechanosensitive channels have been characterized based on their physical properties (conductance, time of opening, pressure sensitivity): mechanosensitive channel of mini conductance “MscM”, small conductance “MscS” and large conductance “MscL” (C>2.5 nS). The MscL from bacteria Escherichia coli has been the focus of many electrophysiological, mutational and biophysical studies.4., 5. One of these studies proposed that the fully open state of the MscL consists of a large non-specific pore, with a large conductance, which deduced diameter is about 30–40 Å.6 However, the three-dimensional structure of the open form remains to be determined. Rees and co-workers7 have recently solved by X-ray crystallography the structure of a homologous mechanosensitive channel in its closed state, from Mycobacterium tuberculosis. This structure is consistent with the major structural properties expected for the MscL of E. coli. The channel consists of five identical subunits arranged around a central pore. Each subunit is composed of a transmembrane and a cytoplasmic domain. The transmembrane region is defined by two helices called TM1 and TM2. The TM1 (the “inner” helix) lines the permeation pathway, whereas the TM2 (the “outer” helix) is positioned on the outside of the channel, facing the lipids. The diameter at the entrance of the pore is approximately 35 Å, whereas, in its narrowest part, the minimum diameter is only 2 Å. Now, the main goal of structural studies is to elucidate the major conformational change occurring during the gating mechanism and to explain how tension of the surrounding lipid bilayer causes the channel opening.

On the basis of experimental data, two models for the gating mechanism and the plausible open structure were proposed. The first model (called the “10 helix pore” model) is based on the assumption that, in the open state, the pore is lined by the ten transmembrane helices of alternating TM1 and TM2 helices.7., 8., 9., 10. This model was elaborated on the basis of mutagenesis of amino acid residues located at the interface of TM1 facing the pore, which affected channel opening considerably. Some biophysical data are rather in conflict with the structural properties of the postulated open state of this model, namely: (i) the deduced diameter of the pore in the open state; (ii) the separate reconstruction of the two “domains” of the MscL, each domain containing either the TM1 or the TM2 helix. The TM2-domain has no electrophysiological activity, while the TM1-domain is able to form channels, but has no mechanosensitivity. These results suggest that the TM1 helices, and not the TM2 helices, participate in the structure of the open pore (A. Ghazi, personal communication).

In the second model proposed by Sukharev & Guy,11 the open state is obtained through a twist and a tilt of the TM1 helices with no major rearrangement of the TM2 helices. This model is nowadays the most studied and the more consistent with most recent experimental data (electrophysiological measurements, cross-linking with cysteine residues). The proposed gating mechanism is more complex than the first one, since a two-step process with different subconductance states is assumed. One main feature of this detailed model of the opening motion is the hypothesis that the first 12 residues, in each monomer, unresolved in the crystal structure, fold into an α-helix (called S1), the five S1 helices forming an Nterm S1 helical bundle, largely involved in the gating mechanism.11 In the early stage of the gating mechanism itself, the channel undergoes a conformational transition which implies a surface expansion of the protein in the bilayer with no significant increase in the size of the inner pore. This first, tension-dependent step leads to a closed-expanded structure (CE), which is still occluded by a ring formed by the amino acid side-chains of the S1 helix-bundle. The second step leads to the fully open state. Sukharev and co-workers also modeled a sequence of states along the opening pathway11 in order to describe as best as possible the different structural states and their conformational changes for the MscL of M. tuberculosis and E. coli.

In parallel with experiments devoted to the understanding of the gating mechanism, molecular modeling approaches are used in order to study protein conformational dynamics, mainly: (1) molecular dynamics (MD) simulations, with numerical integration of the classical Newton equations of motion; (2) targeted dynamics simulations; and (3) normal mode analysis. Due to the small value of the timestep required for numerical integration, i.e. 1–2 fs, simulations of protein conformational changes occurring on a ms timescale remains a challenge from a MD simulation perspective, especially if the simulation is performed for the case of a membrane protein in an explicit medium (lipids and water molecules), using classical parameters (T=300 K). Nevertheless, MD simulations of the M. tuberculosis MscL channel have been performed.12., 13. These studies have given useful insights about the regions of relative structural stability and instability in the structure. For the second method, the steered dynamics, two different structures are needed, one for each endpoint of the conformational pathway. Since the crystallographic structure of the open state is lacking, Ma and co-workers14 explored the transition pathway between the open and closed modeled structures recently proposed by Sukharev and co-workers. Very recently, an interesting work has been realized without any hypothesis about the different steps of the gating mechanism. In order to simulate the effect of the membrane tension, the authors have performed MD simulations, for the M. tuberculosis MscL, with different pressures in the presence of a membrane model. Their results propose a third possibility for the channel gating.15 Nevertheless, the authors admit that the results have been obtained with unrealistic pressure conditions. These conditions have the advantage to accelerate the conformational changes not currently accessible in the classical simulation time scale but they could provide important artifacts in the results obtained. Normal mode calculations provide an alternative to MD simulations for studying collective motions in macromolecules.16., 17., 18. Normal mode analysis is based on the diagonalization of the second derivatives of the mass-weighted energy matrix. The global motion of the system is then expressed as a superposition of collective variables, called the vibrational normal modes. High frequency modes are highly localized motions, of a few side-chains, of pairs of bonded atoms, etc., while the lowest frequency ones are collective motions of large groups of atoms, usually whole structural domains. These later modes mainly depend on the shape of the molecule, as shown for instance by Bahar and collaborators19 in a study of four structurally similar but functionally different proteins. Moreover, it was found that a handful of such collective motions, corresponding to a small subset of the lowest-frequency modes, often compare well with the conformational change observed upon ligand binding,20., 21., 22., 23. especially when the considered conformational change has an highly collective character.24., 25. Thus, normal mode analysis seems to be the best suited theoretical method for studying collective motions in proteins, in particular when a large modification of the structure is expected.

Classical normal mode analysis is limited by the energy minimization step, which can be very long for complex macromolecules, and by the diagonalization of the 3N dimensions Hessian matrix, where 3N is the number of degrees of freedom of the system. Recent developments permit to circumvent the second limiting factor using either an iterative diagonalization method26 or an approximation resting on the fact that low-frequency motions are still well described when amino acid residues are assumed to behave like rigid bodies,27., 28., 29. or both factors by means of a simplified potential30 and chain representation (only one particular per residue).31., 32., 33. Normal mode analysis were preformed on a set of proteins24., 31. using such a simplified method, and, in parallel, the same calculations were done using a standard semi-empirical potential. The results obtained with the approximate normal mode calculations are in agreement with those obtained with the classical method. These studies demonstrate that such approximations (force field and simplified model) provide a good description of domain motions, as described by the low frequency normal modes calculated with classical methods. Moreover, these results show that such a simplified approach is sufficient for studying backbone protein motions of large systems, as far as the low frequency modes are concerned.

The aim of the present study is to provide further insights, at the residue level, of the gating mechanism. To make good use of the numerous biological data related to the MscL of E. coli, the first step of the study consists in elaborating a structural model of this channel in its closed state (MscL of E. coli), based on the X-ray structure, by homology modeling. In a second stage of the study, the normal mode analysis method is applied to this model, as a predictive tool. Our work is based on the assumption that the zones that are easy to deform in the absence of a mechanical constraint are those involved in the response to membrane strain. The normal mode analysis is expected to provide new insights about the flexible and rigid parts of the protein. It could also anticipate the preferential conformational change occurring in the MscL during the gating mechanism. We use the method developed by Tirion30 to perform the normal mode calculations. On the basis of a homology modeling model that could be rather inaccurate at the atomic level, the use of such a simplified potential is the most appropriate.

Then, the normal mode calculated were compared to those of other closed structures of the MscL, the first one corresponding to the X-ray structure and the second one to the closed state of the postulated Sukharev & Guy's model (SG model). Finally, the hypothetical pathway proposed previously by Sukharev and co-workers is submitted to normal mode calculations too, and then, discussed within the frame of this theoretical approach. Three structures were considered along this pathway: the closed state, noted C, the first open state, noted CE, and the fully open state noted O. These structures correspond to the models numbered 1, 9 and 12, respectively, of the different models proposed by Sukharev and co-workers.11 Due to the lack of detailed knowledge about the open state, we use normal mode analysis both in order to try to predict the first step of the conformational pathway but also, as a way to evaluate the SG model.

Section snippets

E. coli MscL homology model

All the alignments tested between E. coli and M. tuberculosis shared about 36% of sequence identity and 73% of sequence similarity. In almost all the generated models, the transmembrane helices are correctly modeled, whereas the loop regions show much more diversity. The difficulty to model these zones comes from the presence of very different amino acid residues in the two sequences, which makes alignment difficult. For the cytoplasmic parts, we observe either interactions between the backbone

Discussion

Different structures of MscL have been examined, using the normal mode analysis. The main goal of our study was to try to understand how this channel can gate, and identify which parts of the protein could be involved in the conformational change. The main features that emerge from the present study concern different and essential points: the global dynamics of the system, in close relationship with the topology of the structure and its flexibility, the role of the Nterm and Cterm parts, and

Materials and Methods

Before carrying out the normal mode analysis, we need a stable structure of the molecule of interest (the MscL of E. coli), and thus, we must generate a three-dimensional model of this channel.

Acknowledgements

We thank S. Sukharev and H. R. Guy for providing the PDB files of the E. coli MscL opening mechanism.

References (65)

  • O. Keskin et al.

    Proteins with similar architecture exhibit similar large-scale dynamic behavior

    Biophys. J.

    (2000)
  • M. Delarue et al.

    Simplified normal mode analysis of conformational transitions in DNA-dependant polymerases: the elastic network model

    J. Mol. Biol.

    (2002)
  • D. Perahia et al.

    Computation of low-frequency normal modes in macromolecules: improvements of the method of diagonalization in a mixed basis and applications to hemoglobin

    Comput. Chem.

    (1995)
  • G. Li et al.

    A coarse-grained normal mode approach for macromolecules: an efficient implementation and application to ca(2+)-atpase

    Biophys. J.

    (2002)
  • J.U. Bowie

    Helix packing in membrane proteins

    J. Mol. Biol.

    (1997)
  • B. Ajouz et al.

    Contributions of the different extramembranous domains of the mechanosensitive ion channel Mscl to its response to membrane tension

    J. Biol. Chem.

    (2000)
  • F. Tama et al.

    The mechanism and pathway of pH induced swelling in cowpea chlorotic mottle virus

    J. Mol. Biol.

    (2002)
  • D.J. Barlow et al.

    Helix geometry in proteins

    J. Mol. Biol.

    (1988)
  • J.A. Maurer et al.

    Comparing and contrasting E. coli and M. tuberculosis mechanosensitive channels (MscL). New gain of function mutations in the loop region

    J. Biol. Chem.

    (2000)
  • I. Bahar et al.

    Collective motions in HIV-1 reverse transcriptase: examination of flexibility and enzyme function

    J. Mol. Biol.

    (1999)
  • A. Sali et al.

    Comparative protein modeling by satisfaction of spatial restraints

    J. Mol. Biol.

    (1993)
  • P. Koehl et al.

    Application of a self-consistent mean field theory to predict protein side-chains conformation and estimate their conformational entropy

    J. Mol. Biol.

    (1994)
  • H. Berendsen et al.

    GROMACS: a message-passing parallel molecular dynamics implementation

    Comp. Phys. Commun.

    (1995)
  • C. Chothia et al.

    Helix to helix packing in proteins

    J. Mol. Biol.

    (1981)
  • W. Humphrey et al.

    VMD—visual molecular dynamics

    J. Mol. Graph.

    (1996)
  • C. Berrier et al.

    Elongation factor tu and dnak are transferred from the cytoplasm to the periplasm of Escherichia coli during osmotic downshock presumably via the mechanosensitive channel MscL

    J. Bacteriol.

    (2000)
  • S.I. Sukharev et al.

    Mechanosensitive channels of Escherichia coli: the MscL gene, protein, and activities

    Annu. Rev. Physiol.

    (1997)
  • O.P. Hamill et al.

    Molecular basis of mechanotransduction in living cells

    Physiol. Rev.

    (2001)
  • G. Chang et al.

    Structure of the MscL homolog from Mycobacterium tuberculosis: a gated mechanosensitive channel

    Science

    (1998)
  • A.F. Batiza et al.

    Channel gate! Tension, leak and disclosure

    Structure

    (1999)
  • Y. Kong et al.

    Conformational pathways in the gating of Escherichia coli mechanosensitive channel

    Proc. Natl Acad. Sci. USA

    (2002)
  • N. Go et al.

    Dynamics of a small globular proteins in terms of low-frequency vibrational modes

    Proc. Natl Acad. Sci. USA

    (1983)
  • Cited by (94)

    View all citing articles on Scopus

    Supplementary data associated with this paper can be found at doi: 10.1016/S0022-2836(03)00851-9

    View full text