Research Article

Mechanistic insight from the crystal structure of mitochondrial complex I

See allHide authors and affiliations

Science  02 Jan 2015:
Vol. 347, Issue 6217, pp. 44-49
DOI: 10.1126/science.1259859

Energy conversion in complex 1

ATP, the energy source of the cell, is synthesized by a protein residing in the mitochondrial inner membrane. The synthesis is driven by a proton gradient generated by redox reactions that transfer electrons between a series of enzymes in the membrane. The largest complex in this electron transfer chain is the 1-MD complex 1. It couples electron transfer from NADH to ubiquinone to the translocation of four protons. Zickermann et al. report the crystal structure of a complex comprising the 14 central subunits and the largest accessory subunit of mitochondrial complex 1 from a yeast-genetic model at 3.6 Å resolution. The structure identifies four potential proton translocation pathways and gives insight into how energy from the redox reactions is transmitted to drive proton pumping.

Science, this issue p. 44

Abstract

Proton-pumping complex I of the mitochondrial respiratory chain is among the largest and most complicated membrane protein complexes. The enzyme contributes substantially to oxidative energy conversion in eukaryotic cells. Its malfunctions are implicated in many hereditary and degenerative disorders. We report the x-ray structure of mitochondrial complex I at a resolution of 3.6 to 3.9 angstroms, describing in detail the central subunits that execute the bioenergetic function. A continuous axis of basic and acidic residues running centrally through the membrane arm connects the ubiquinone reduction site in the hydrophilic arm to four putative proton-pumping units. The binding position for a substrate analogous inhibitor and blockage of the predicted ubiquinone binding site provide a model for the “deactive” form of the enzyme. The proposed transition into the active form is based on a concerted structural rearrangement at the ubiquinone reduction site, providing support for a two-state stabilization-change mechanism of proton pumping.

With a molecular mass of ~1 MD, proton-pumping NADH:ubiquinone oxidoreductase (complex I) is the largest membrane protein complex of the mitochondrial respiratory chain (1, 2). Complex I couples electron transfer from NADH (reduced nicotinamide adenine dinucleotide) to ubiquinone to the translocation of four protons from the mitochondrial matrix to the intermembrane space (IMS), thereby providing much of the proton-motive force for adenosine triphosphate synthesis. Dysfunction of the enzyme is the most frequent cause of mitochondrial disorders (3) and has been implicated in numerous neurodegenerative conditions (4). Complex I is also a major source of deleterious reactive oxygen species (5). Mitochondrial complex I consists of 14 central subunits and a large number of accessory subunits. Specific inhibitors have been essential for unraveling structure-function relationships within complex I and are a focus of biomedical research, as some of them can induce Parkinson’s disease (6). The central subunits of complex I harboring the bioenergetic core functions are conserved from bacteria to humans. Complex I from bacteria and from mitochondria of Yarrowia lipolytica, a yeast genetic model for the study of eukaryotic complex I (7), were previously analyzed by x-ray crystallography (810). Yet the catalytic mechanism has remained enigmatic. Here, we present the x-ray structure of all central subunits and the largest accessory subunit of mitochondrial complex I from Y. lipolytica, providing insight into the molecular basis of redox-driven proton pumping.

Overall structure

Complex I was purified and crystallized as described (8). Optimized cryoprotection and data collection improved x-ray diffraction in the best direction to 3.6 Å. Phases were obtained using single isomorphous replacement with anomalous scattering from highly redundant data of heavy-atom derivatives, FeS clusters, and selenomethionine. Phases were combined, averaged, and extended to the resolution of the native data set collected at 5 K. Model building with the 3.8 Å experimental electron density map was aided by the large content of α helices, close to 100 selenomethionine positions, and homology models based on the bacterial complexes (10, 11). Secondary structure constraints were applied for refinement. Anisotropy correction counteracted the lower order of the crystals in the membrane plane, and the structure was refined at 3.9 Å × 3.9 Å × 3.6 Å (table S1 and fig. S1). The structure contains the majority of residues for central subunits and the largest accessory subunit (table S2). Side chains are well resolved in tightly packed subunits with high helix content, and between 50 and 89% of the residues in these subunits could be assigned. The end of the peripheral hydrophilic domain is less ordered (fig. S1 and table S1).

Mitochondrial complex I has a slightly opened L-shape with an angle of ~120° between the two arms (Fig. 1). The matrix arm protrudes into the organelle interior and is oriented perpendicular to the membrane plane and the membrane arm. The 14 central subunits forming the core of the complex are surrounded by the accessory subunits comprising 44% of the total mass. Complex I is 250 Å long and 190 Å high, with a width of 70 Å for the membrane arm and 120 Å for the matrix arm. The complex is organized in four functional modules (1, 8). The distal half of the matrix arm, comprising the central 75-kDa, 51-kDa, and 24-kDa subunits, forms the N module that oxidizes NADH (nomenclature as for bovine complex I is used throughout the text). The proximal half of the peripheral arm, which comprises the central 49-kDa, 30-kDa, PSST, and TYKY subunits, is the Q module that reduces ubiquinone and docks onto the membrane arm. A chain of eight canonical FeS clusters (2) runs over a distance of ~100 Å through the matrix arm (Fig. 1C). The edge-to-edge distances for seven of these clusters are <14 Å, allowing fast electron tunneling (12). This arrangement establishes an electron transfer path from the [4Fe-4S] cluster of the 51-kDa subunit (cluster N3) to the [4Fe-4S] cluster of PSST (cluster N2). Cluster N2 reduces ubiquinone and resides ~30 Å above the membrane plane (8). The slightly bent membrane arm features 82 transmembrane helices (TMHs), with 64 of them contributed by central subunits (Fig. 2A). The proximal pump module (PP) comprises central subunits ND1, ND2, ND3, ND4L, and ND6, whereas the distal pump module (PD) contains central subunits ND4 and ND5 (Fig. 2A).

Fig. 1 Overall structure.

(A and B) View from peripheral arm (A) and rotated 90° (B). N module: red, 75-kDa; yellow, 51-kDa; orange, 24-kDa. Q module: green, 49-kDa; violet, 30-kDa; blue, PSST; cyan, TYKY. PD module: dark blue, ND5; cyan, ND4. PP module: lilac, ND2; red, ND4L; orange, ND6; yellow, ND3; pink, ND1. Accessory subunits are depicted in gray. (C) Arrangement of 4Fe-4S(4Fe) and 2Fe-2S(2Fe) clusters in the peripheral arm. The coordinating subunits are denoted in brackets. Center-to-center and edge-to-edge (in brackets) distances are in angstroms.

Fig. 2 Membrane arm.

(A) Central transmembrane subunits and TMHs of accessory subunits seen from the matrix side (colors as in Fig. 1). Solid circles, discontinuous helices in ND1, ND2, ND4, and ND5; broken circle, TMH3ND6 containing a π bulge. (B) Discontinuous helices and side chains of residues of the central axis. Inset shows arrays of polar and titratable residues of ND5. Residues of the central axis of protonable residues are underlined. (C) Discontinuous TMH7 and TMH12 after global superposition of ND2, ND4, and ND5 seen along the long axis of the membrane arm; δ+ and δ are partial charges imposed by helix dipoles. Amino acid abbreviations: D, Asp; E, Glu; H, His; K, Lys; N, Asn; Q, Gln; R, Arg; S, Ser; T, Thr; W, Trp; Y, Tyr.

Membrane arm and proton-pumping modules

The largest membrane-embedded subunits, ND5, ND4, and ND2 (Fig. 2A), share a structurally highly similar core of 14 TMHs (fig. S2) with two repeats of five TMHs (A, TMH4–8; B, TMH9–13) in inverted topology (fig. S2). Each repeat features a discontinuous helix (TMH7a/b, TMH12a/b). Such helices are hallmarks of ion-translocating membrane proteins (1315). Indeed, ND5, ND4, and ND2 are homologous to the Mrp Na+/H+ antiporter family, which suggests a role in proton pumping (8, 11, 16). ND5 has a C-terminal extension with a lateral helix, >60 Å long, lining ND5, ND4, and ND2 on the concave side of the arm close to the matrix side (Fig. 2A), thus bridging the PP and PD modules. The C terminus of ND5 is anchored to ND2 via a V-shaped arrangement of TMH16 and TMH17. The lateral helix was previously identified in the low-resolution analysis of mitochondrial complex I from Y. lipolytica (8) and is also present in the bacterial (9, 11) and bovine complexes (17), in which its C terminus is anchored by one TMH only. Whether this prominent structural element is involved in energy transduction for proton pumping, as proposed previously (8, 9), remains controversial (2).

Adjacent to ND2 are three small central subunits, ND4L, ND6, and ND3, which form two layers of straight helices crossing the membrane domain (Fig. 2A and fig. S3A). Notably, TMH3ND6 features a π bulge near the membrane center. Remote with respect to TMH1–3ND6, TMH4ND6 is laterally associated with the membrane arm and lines TMH16ND5. On the intermembrane-space side, a long loop links THM4ND6 to a short surface helix connecting to TMH5ND6, which is surrounded by helices of ND2, ND4L, and ND3. In the bacterial complex (9, 11), TMH4ND6 sits closer to the core helices of ND6 at a position occupied in the mitochondrial complex by a single helix of an accessory subunit, tentatively identified as NUJM (17). TMH1ND3 is anchored remotely in a surface cleft of ND1. ND1 at the proximal end of the membrane arm comprises eight TMHs and noticeable loop regions, including short surface helices (Fig. 2A and fig. S3B). The fold of TMH2–6 is similar to one repeat of the antiporter-like subunits. This agrees with homology predictions (18) and was noted in the bacterial complex (10). Notably, TMH5 is discontinuous in mitochondrial complex I.

The interior of the membrane arm is rich in polar and protonable residues (i.e., residues that may take up or release protons) constituting a remarkable hydrophilic central axis across all subunits (Fig. 2B). Toward the matrix and IMS, each of the antiporter-like subunits has typical arrays of titratable and polar residues constituting possible proton uptake and release pathways (Fig. 2B and fig. S4). For ND5 (Fig. 2B), Asp397 at the membrane center is the first in a series of polar residues (Tyr392, Glu401, Thr403, and Ser402) arranged along TMH12b to the IMS. Neighboring Lys396 connects this path to the central polar network with major contributions of protonable residues of TMH11 (His331, His335, and Lys339). The likely entry point to this network is His251 of TMH8, to which protons could be provided from the matrix along the tip of TMH7b, TMH8, and TMH10 (Glu241, Ser247, Thr244, and Lys302). At the center of discontinuous TMH7, Lys226 (TMH7a) is sandwiched between the indole side chains of Trp143 and Trp235 and connects on one side to His251 (through Ser227 and Thr254) and on the other side to TMH12ND4 via protonable residues of TMH5ND5 and TMH6ND5 (Glu144, Arg175, Asp178). A similar architecture is repeated in ND4 and ND2 (fig. S4). Whereas the central network and the proposed proton exit routes are consistent with suggestions for the bacterial complex, the proposed routes for proton entry differ as they were assigned between TMH5 and TMH7 in bacteria (10, 19).

In summary, all three antiporter-like subunits carry the structural signature of a proton-pumping unit with two structurally distinct discontinuous helices. Consistently, TMH7a/b contributes a central Lys residue that is next to a Glu of TMH5, and TMH12a/b is part of a central polar network connecting the proton exit route along TMH12b with the likely proton entry point at loops TMH7b–8 and TMH9–10 (Fig. 2B). The dipoles of the discontinuous helices add to the polarity of the central axis, because the C termini of TMH7a and TMH12a point to the center of the membrane arm, thereby introducing evenly spaced negative partial charges along the arm. In contrast, the positive polarity of the N termini of TMH7b and TMH12b is directed away from the central axis toward the periphery of the membrane arm (Fig. 2, B and C).

In line with a pump stoichiometry of 4H+/2e (20), a fourth potential proton translocation pathway could be anticipated in the small NDs or ND1. From the neighboring residues Glu30 and Glu66 of ND4L and Glu131 of ND2 continuing the central axis, a row of polar residues toward the IMS at the interface of ND2 and ND4L could serve as a proton exit (Fig. 2B and fig. S4). Directly above, a series of polar residues constitutes a potential fourth proton entry at the interface initially also proposed to be present in bacterial complex I (19). The central axis extends further toward Glu196 at the center of discontinuous TMH5ND1, along Asp67 and Glu69 of ND3 and Glu147 of ND1. The π bulge of TMH3ND6 provides an additional polar contribution. From Glu196 onward, a row of acidic residues of ND1 (Asp203, Glu206, Glu208, and Glu210) continues toward the peripheral arm. A similar arrangement was described in complex I from T. thermophilus (10) and was interpreted as a possible fourth pathway for proton uptake, replacing the one proposed earlier (19). However, the structure of mitochondrial complex I suggests that the central axis of protonable residues is more likely to play a critical role in energy transmission.

Peripheral arm

The peripheral arm comprises the N module extending into the matrix and the Q module docking it onto the membrane arm. The central subunits of the Q module (the 49-kDa, PSST, TYKY, and 30-kDa subunits) are related to Ni-Fe hydrogenases (1, 21). The 49-kDa subunit comprises two three-stranded antiparallel β sheets and several α helices. The longest helices form a prominent four-helix bundle inclined toward the membrane surface (Figs. 1 and 3). PSST, on the other side of the Q module, contains a central four-stranded parallel β sheet surrounded by five helices and harbors the [4Fe-4S] cluster N2. The N-terminal helix α1PSST protrudes toward the membrane surface (Figs. 1 and 3B). TYKY provides contact to the N module. Its ferredoxin-type fold holds two [4Fe-4S] clusters (Fig. 3). Its N-terminal helix α1 docks onto ND1 and contacts the four-helix bundle of the 49-kDa subunit. The latter is also the docking site for the 30-kDa subunit, with a central five-stranded β sheet laterally surrounded by four α helices. In the N module, the 75-kDa subunit resides on the same side as PSST and coordinates one [2Fe-2S] and two [4Fe-4S] clusters (Figs. 1 and 3B). The 51-kDa subunit located above the 49-kDa subunit holds one [4Fe-4S] cluster and contains the binding sites for FMN and NADH in a Rossman-fold domain. The 24-kDa subunit coordinates the [2Fe-2S] cluster detached from the electron transfer chain leading to cluster N2. As noted previously (8), the Q module is rotated outward about 3° relative to the N module as compared to bacterial complex I (10).

Fig. 3 Interface between peripheral and membrane arm.

(A and B) Two views of the interface between central subunits of the Q module and the PP module (colors as in Fig. 1) and accessory subunit NUEM (dark green). Dashed arrow indicates putative access for ubiquinone.

The largest accessory subunit, NUEM—an ortholog of the mammalian 39-kDa subunit—flanks the Q module and docks onto PSST (fig. S5). It belongs to the family of short-chain dehydrogenases (SDRs). A characteristic Rossman fold typically found in all SDRs (22) is present in the N-terminal domain, with a central seven-stranded β sheet surrounded by five α helices (Fig. 3 and fig. S5). NADH phosphate (NADPH) binding to Y. lipolytica complex I was shown experimentally (23). The helices of the C-terminal domain characteristic for the subfamily of extended SDRs are in contact with the membrane arm, consistent with the reported cross-linking to ND3 (24). Assembly defects of Y. lipolytica complex I in deletion mutants (23) suggest a structural role, but the function of NUEM remains elusive. Some additional accessory subunits were provisionally localized in the Y. lipolytica complex I structure (fig. S6) on the basis of a model of supernumerary subunits of bovine complex I recently obtained by cryo–electron microscopy (17).

Interface between peripheral and membrane arm and ubiquinone access

Structural elements of the ND1, ND3, 49-kDa, PSST, and TYKY subunits form the interface between the membrane and the peripheral arm (Fig. 3). The matrix surface of ND1 provides the main contact area with the 49-kDa subunit and PSST. Helix α1PSST contacts surface helix α1-2ND1. On the opposite side, helix α1TYKY protrudes into a groove between the four-helix bundle of the 49-kDa subunit and ND1. TMH1ND3 and TMH2ND3 are in contact with loop TMH1–2ND1 including helix α1-2ND1 and with loop β23 of the N-terminal β sheet of the 49-kDa subunit, respectively. At a central point of the interface, contact between the two arms of complex I involves loop TMH5–6ND1 (Glu206 to Gly221) (Fig. 4). This loop is rich in acidic residues, protrudes into a cleft between PSST and 49-kDa subunit, and is in contact with 49-kDa loop β1-β2 (Pro89 to Leu98) of the N-terminal β sheet, where loop TMH1–2ND3 also approaches.

Fig. 4 Access path of quinone to the active site.

(A) Side view. A cavity ~30 Å in length (beige) connects the active site below cluster N2, with the matrix bilayer leaflet permitting substrate access. A constricted opening (*) is located between the V-shaped arrangement of TMH1ND1 and TMH6ND1 and below the amphipathic helix α1-2ND1 (colors as in Fig. 1). (B) View from PSST, which has been removed for clarity.

Because cluster N2 resides well above the membrane plane, an access path allowing the head group of the hydrophobic substrate ubiquinone to reach its electron donor is needed (25). This path is provided by a quinone exchange cavity crossing the interface region. Its opening lies between TMH1ND1, TMH6ND1, and amphipathic helix α1-2ND1 positioned at the periphery of the matrix bilayer leaflet. The cavity extends ~30 Å toward the tip of loop β1-β249-kDa (Fig. 4). The small triangular-shaped entry pore is ~7 Å wide and has a hydrophobic surface. The side chain of Ala54 of α1-2ND1 points into the pore opening. In humans, substitution of the corresponding Ala52 of ND1 with the larger Thr is among the most prevalent mutations leading to Leber’s hereditary optic neuropathy and interferes with ubiquinone reduction (26). Furthermore, substitution of Trp77 in PSST (lining the inner side of the narrow entry passage) by Glu abolishes ubiquinone reductase activity (27). The surface and immediate vicinity of the cavity exhibit a bipartite distribution of charged residues. Toward the surface of the complex, basic residues are present at the C-terminal end of TMH1ND1 (Arg27, Lys28, Arg36, and Arg37), TMH7ND1, and loop TMH7–8ND1, some of which may interact with phospholipids. Deeper into the pocket toward the protein interior, the cavity is lined by acidic residues of TMH5ND1 and adjacent loop TMH5–6ND1 (Asp203, Glu206, Glu208, and Glu210) (Fig. 4).

Ubiquinone and inhibitor binding sites

To define the substrate and inhibitor binding sites within the Q module, we cocrystallized complex I with brominated derivatives of 2-decyl-4-quinazolinylamine (DQA; Fig. 5). DQA is a potent class I/A type inhibitor competitive to piericidin A (28) and shown to qualify as a true ubiquinone analog (29). Brominating the quinazoline scaffold (QA-1) increased the half-maximal inhibitory concentration (IC50) from 17 nM for DQA to 320 nM, and replacing the side chain with a brominated phenylethylamine moiety (QA-2) hardly affected binding (IC50 = 23 nM) (fig. S7). Anomalous Fourier electron density maps (Fig. 5) positioned the bromine atoms close to each other in the vicinity of His95 of the 49-kDa subunit at the deepest point of the quinone exchange cavity (Fig. 5B). For both derivatives, the geometrical constraints imposed by anomalous signals and structure are consistent with an identical toxophore position; a quinazoline ring was modeled into the structure to visualize its likely position. The planar aromatic ring system stacks between the tip of loop β1249-kDa and Met91 of PSST (Fig. 5B), placing the ring bromine at the quinazoline ring and the bromine in the tail moiety 14 Å and 16 Å away, respectively, from cluster N2. Supporting the notion that the position of the substrate analogs also reflects a binding site for ubiquinone, mutations of His91, His95, Val97, Leu98, and Arg99 in loop β1249-kDa as well as substitution of Met91 in PSST by Lys or Glu cause drastic reduction of enzymatic activity (21, 27, 30, 31), whereas substitution of the same Met91 by Cys showed a marked resistance against DQA and rotenone. Further residues (Ser192, Met188, and Phe203 of the 49-kDa subunit and Val88 of PSST) with a role in ubiquinone or inhibitor binding (27) are located nearby (Fig. 5).

Fig. 5 Binding site of the ubiquinone analogous inhibitor DQA.

(A) Constitutions of DQA and bromo-substituted derivatives QA-1 and QA-2. (B) Stereo view of inhibitor binding pocket (colors as in Fig. 1). Single peaks in the bromine anomalous Fourier electron density maps are shown (purple, QA-1, 3.8σ; red, QA-2, 4σ; superimposition of electron density maps from two separate experiments on structure). Orange, quinazoline ring modeled into the site.

Some distinct structural features in mitochondrial complex I can explain a known functional difference to the bacterial enzyme: The quinone exchange cavity does not provide access to several other residues of the 49-kDa subunit (Asp143, Tyr144, Val145, Asp458, and Val460) for which biochemical evidence obtained with Y. lipolytica suggested a direct interaction with ubiquinone and inhibitors (29, 30). Indeed, binding of ubiquinone and piericidin A next to these residues and closer to cluster N2 was modeled into the structure of the bacterial enzyme (10), revealing a marked difference (fig. S8) between the otherwise similar (table S3) core structures of Y. lipolytica and T. thermophilus complex I. Evidently, in the Y. lipolytica structure, access deeper into the protein is blocked by the tip of loop β1249-kDa bringing His95 close to Tyr144, whereas in the bacterial enzyme, the ubiquinone head group was located between these two residues (10). The different position of loop β1249-kDa is accompanied by significantly different orientations of the adjacent acidic loop TMH5–6ND1 and loop TMH1–2ND3.

This observation provides a straightforward explanation for the reversible A/D transition (32) occurring only in complex I of some eukaryotes, including Y. lipolytica (33), and discussed as a protective mechanism against excessive oxygen radical formation (34). Preparations of Y. lipolytica complex I are always in the so-called deactive (D) form, which slowly reverts into the active (A) form upon addition of substrates (33). Consistent with the very low catalytic activity of the D form, the structure of Y. lipolytica complex I shows the ubiquinone binding site at markedly greater distance from cluster N2, as compared to the always-active structure of the bacterial enzyme, in which ubiquinone was positioned much closer to its electron donor, allowing for efficient electron transfer (10). Supporting this interpretation, loop TMH1–2ND3 undergoes conformational changes during the A/D transition (35). Considering the remarkably high sequence conservation within the structural elements involved (fig. S9) (10, 30) and because binding of ubiquinone next to cluster N2 is also supported by exhaustive mutagenesis data for the mitochondrial enzyme (21, 27, 30, 31), we propose that in the A form, the ubiquinone binding pocket of Y. lipolytica complex I adopts a conformation similar to the bacterial enzyme that has no D form. We suggest that the interface region of complex I can switch between two distinct conformational states, thereby shifting the ubiquinone binding site. This switch involves a concerted movement of loops from three subunits (49-kDa, ND1, and ND3).

Mechanistic implications

The central question concerning the mechanism of energy conversion of complex I is how the redox energy released exclusively in the peripheral arm is transmitted to the proton pump modules of the membrane arm (Fig. 6). Several lines of evidence indicate that ubiquinone reduction plays a pivotal role in this process. The hypothetical two-state stabilization-change mechanism (36) proposes that stabilization of negatively charged quinone intermediates drives a conformational change, thereby transmitting energy to the membrane arm to drive proton pumping. It postulates that complex I switches between the E state, in which ubiquinone can be reduced by cluster N2, and the P state, in which it is moved away from its electron donor. This description is reminiscent of the A and D state conformations discussed above. A displacement of the acidic loop TMH5–6ND1 at the start of the chain of titratable residues reaching through the membrane arm seems ideally suited to transmit an “electrostatic pulse” (37). We therefore hypothesize that an orchestrated movement of the three loops associated with the A/D transition could also reflect the critical energy-converting steps during catalytic turnover (Fig. 6). Such a mechanism would imply that the E state corresponds essentially to the A form, whereas the P state would resemble the D form, with the notable exception that the D form cannot revert rapidly and spontaneously to the E/A state, because this is prevented by a yet unidentified structural feature not present in the bacterial enzyme. Because the A/D transition is observed only with eukaryotic enzymes, it is tempting to speculate that stabilization of the D state may involve nearby accessory subunits such as NUEM, which has been shown to take part in the associated conformational changes (38).

Fig. 6 Hypothetical two-state stabilization change mechanism.

Electrons are transferred (red arrow) via a chain of iron-sulfur clusters from NADH to ubiquinone (Q). Within three antiporter-like subunits (violet frame) and the other ND subunits, a pattern of titratable residues defines a central axis in the membrane connected to the IMS and matrix side by putative proton translocation pathways (dark blue). Loop TMH5–6ND1 (red), loop β1249-kDa (green), and the tip of loop TMH1–2ND3 (yellow) line the ubiquinone exchange cavity. During turnover, these loops perform a coordinated rearrangement resulting in a shift of the ubiquinone binding site and movement of the cluster of negative charges in loop TMH5–6ND1, which may trigger an electrostatic pulse toward the membrane arm. Stabilization of the anionic species in the site leads to transition from E state (left) to P state (right), driving a stroke of proton pumping. The idling enzyme can convert reversibly from the active A form into the deactive D form with a structure similar to the P state.

Notably, this series of events could be triggered not only by stabilization of ubisemiquinone, but also of the ubiquinol anion resulting from the second reduction step (36). Although so far experimental evidence for such a second pump stroke is missing, partitioning of the free energy change by making use of both electron transfer steps to ubiquinone seems to make thermodynamic and mechanistic sense, given that complex I can operate in reverse as a proton gradient–driven NAD+ reductase.

The described mechanistic principle—charge-induced conformational changes that result in secondary electrostatic polarization of charged residues—may also be important to drive the individual proton-pumping sites of complex I. Defined localized conformational changes should ensure controlled vectorial charge translocation, whereas energy transfer between the sites could occur by electrostatic coupling. Indeed, recent large-scale molecular dynamics simulations suggested that long-range energy transmission in complex I is executed through charge-induced protonation changes of key residues (39).

Supplementary Materials

www.sciencemag.org/content/347/6217/44/suppl/DC1

Materials and Methods

Figs. S1 to S11

Tables S1 to S3

References (4058)

References and Notes

  1. Acknowledgments: Supported by the German Research Foundation (CRC 746 to C.H.; ZI 552/3-1 to V.Z.) and the Excellence Initiative of the German Federal and State Governments (EXC 115 to H.S., U.B., and V.Z.; EXC 294 BIOSS to C.H.). We thank the Swiss Light Source and European Synchrotron Radiation Facility for beamline access and staff support during visits, and A. Duchene and G. Beyer for excellent technical assistance. Coordinates and structure factors are deposited in the Protein Data Bank with accession code 4wz7.
View Abstract

Stay Connected to Science

Navigate This Article