The redox driven Na⁺-pumping mechanism in Vibrio cholerae NADH-quinone oxidoreductase relies on dynamic conformational changes

Abstract

The Na⁺-pumping NADH-quinone oxidoreductase (Na⁺-NQR) is a key respiratory enzyme in many marine and pathogenic bacteria that couples electron transfer to Na⁺-pumping across the membrane. Earlier X-ray and cryo-electron microscopy structures of Na⁺-NQR from Vibrio cholerae suggested that the subunits harboring redox cofactors undergo conformational changes during catalytic turnover. However, these proposed rearrangements have not yet been confirmed. Here, we have identified at least five distinct conformational states of Na⁺-NQR using: mutants that lack specific cofactors, specific inhibitors or low-sodium conditions. Molecular dynamics simulations based on these structural insights indicate that 2Fe-2S reduction in NqrD/E plays a crucial role in triggering Na⁺ translocation by driving structural rearrangements in the NqrD/E subunits, which subsequently influence NqrC and NqrF positioning. This study provides structural insights into the mechanism of Na⁺ translocation coupled to electron transfer in Na⁺-NQR.

The Na⁺ -pumping NADH-quinone oxidoreductase is a redox-driven sodium pump often found in pathogenic bacteria. Here, the authors demonstrate how enzyme structural changes efficiently couple electron transfer to Na⁺ translocation.

Introduction

The Na⁺-pumping NADH-quinone oxidoreductase (Na⁺-NQR) is the first enzyme in the respiratory chain of numerous marine and pathogenic bacteria, including Vibrio cholerae, Neisseria gonorrhoeae, and Haemophilus influenzae. Na⁺-NQR catalyzes the transfer of electrons from NADH to quinone, coupled to the translocation of Na⁺, generating an electrochemical Na⁺ gradient across the inner membrane. This gradient powers essential energy-dependent processes such as ATP synthesis, rotation of the flagellar motor, transport of nutrients, and operation of multi-drug transporters^1–3. As originally revealed by biochemical and biophysical studies, Na⁺-NQR is an integral membrane complex composed of six subunits (NqrA-F) encoded by the nqr operon. It contains five spectroscopically visible redox cofactors (FAD, 2Fe-2S, two FMNs, and riboflavin) as well as one spectroscopically invisible 2Fe-2S cluster and has a total molecular mass of ~200 kDa. Na⁺-NQR is exclusively found in prokaryotes and is structurally unrelated to the mitochondrial H⁺-pumping NADH-quinone oxidoreductase (respiratory complex I), making it a promising target for highly selective antibiotics^3,4.

X-ray crystallographic^5,6 and single-particle cryo-EM analyses^6,7 have revealed the entire architecture of Na⁺-NQR (Fig. 1a). The cytoplasmic domain of the NqrF subunit includes a Rossmann nucleotide binding motif, where NADH binds, and contains one non-covalently bound FAD (FAD^NqrF) that is the initial acceptor of electrons, and one 2Fe-2S cluster (2Fe-2S^NqrF)^8,9. The membrane-embedded NqrD and NqrE subunits form an inverted repeat structure that coordinates a second 2Fe-2S (2Fe-2S^NqrD/E)^2,10–12, originally assigned as a single iron atom in the initial X-ray crystallographic model⁵ and subsequently identified as a 2Fe-2S cluster by cryo-EM⁷, although it still lacks any spectroscopic or kinetic signature. The periplasmic NqrC has one covalently bound FMN (FMN^NqrC) in its hydrophilic region^13,14. The NqrB subunit is an integral membrane protein with a covalently bound FMN (FMN^NqrB) and a distinct riboflavin cofactor (RBF^NqrB) that is the final donor of electrons to the ubiquinone (UQ) substrate^15,16. The hydrophilic NqrA subunit is attached to the membrane through a tight interaction with the protruding part of the N-terminal region of NqrB. This cytoplasmic interfacial area has a significant influence on the binding of UQ^17,18. These structural studies, along with earlier spectroscopic^19–23, kinetic^12,24, and mutagenetic investigations^8–10,13,14, have established a consensus electron transfer pathway: NADH → FAD^NqrF → 2Fe-2S^NqrF → 2Fe-2S^NqrD/E → FMN^NqrC → FMN^NqrB → RBF^NqrB → UQ.

Fig. 1. Architecture of Na⁺-NQR.

a Location of redox cofactors and electron transfer steps in the overall architecture of Na⁺-NQR (WT_ox: pdb: 7XK3). Left hand panel: simplified structure of Na⁺-NQR showing the locations of redox cofactors and depicting electron transfer steps as arrows. Distances consistent with electron transfer are colored in gray, whereas long distances in this structural conformation, inconsistent with electron transfer (2Fe-2S^NqrF → 2Fe-2S^NqrD/E, 2Fe-2S^NqrD/E → FMN^NqrC), are colored in red. Middle and right-hand panels: the Na⁺-NQR assembly from front and side views, respectively. Concentric partial circles in red around NqrF indicate flexibility of this region. b The “up”, “middle”, and “down” conformational states: The left panel shows the FNR domain containing the FAD cofactor, ferredoxin-like domain containing 2Fe-2S, and TMH1 of NqrF (side view). The remaining panels show the density fit view of the structure of NqrF in each conformational state illustrating the pivotal motion of NqrF in WT_red/ - Na⁺. To compare WT_ox and all states in WT_red/- Na⁺, the overlay image of the three structures are also represented in Supplementary Fig. 13. c Changes in distances between cofactors that accompany these conformational changes. The changes in distance between 2Fe-2S and FAD in NqrF, WT_red/ - Na⁺. For comparison of WT_ox with all states in WT_red/- Na⁺, the relative locations of WT_ox are overlapped as gray silhouette. Black continuous arrows indicate feasible distances for electron transfer, and red dashed arrows indicate distances that are not consistent with electron transfer.

However, the distances between two of the donor-acceptor pairs in two of these electron transfer steps: 2Fe-2S^NqrF → 2Fe-2S^NqrD/E, and 2Fe-2S^NqrD/E → FMN^NqrC, shown as red arrows in Fig. 1a are too large (26-33 Å, edge-to-edge) for physiologically relevant electron transfer^7,25. This indicates that the subunits containing these cofactors must undergo significant conformational rearrangements to reduce these spatial gaps during catalytic turnover^7,20. Nevertheless, there is as yet no structural evidence to support these proposed rearrangements. In the case of the 2Fe-2S^NqrF → 2Fe-2S^NqrD/E electron transfer step, our previous cryo-EM study of wild-type Na⁺-NQR suggested that the cytosolic domain of the NqrF subunit is highly flexible, but the inferred range of motion is still not large enough to account for the electron transfer ⁷. In the case of 2Fe-2S^NqrD/E → FMN^NqrC, an apparent discrepancy between X-ray and cryo-EM results may actually indicate a crucial conformational change. In the X-ray crystallographic structures of V. cholerae Na⁺-NQR^5,6, the periplasmic domain of NqrC containing FMN^NqrC is positioned near the NqrD/E subunits, whereas the cryo-EM structure places this domain near NqrB. These results suggest that repositioning of the NqrC subunit during turnover could allow FMN^NqrC to shuttle electrons from 2Fe-2S^NqrD/E to FMN^NqrB. However, there is no evidence to link this apparent rearrangement of NqrC to different intermediates in the catalytic mechanism.

Another enigma involves in the pathway(s) for Na⁺ translocation in the enzyme. Based on the biochemical characterization of conserved acidic residues within the TMHs, Juárez et al. identified critical acidic residues necessary for Na⁺ translocation (e.g., NqrB-D397, NqrD-D133, and NqrE-E95)²⁶. Hau et al. observed two bound Na⁺ ions on the periplasmic side of NqrB in their cryo-EM map and proposed a Na⁺ channel within NqrB⁶. More recently, Kumar et al. reported a cryo-EM structure of the “Rhodobacter nitrogen fixing” protein (RNF) complex, an evolutionary ancestor of Na⁺-NQR²⁷. Combined with atomic molecular dynamics (MD) simulations, they showed that the reduction of a 2Fe-2S cluster coordinated by the symmetric RnfA/E subunits (homologs of NqrD/E), induces capture of Na⁺ and subsequently triggers an inward/outward transition of the subunits. This model aligns with the alternating-access mechanism proposed by Bogachev and colleagues for Na⁺-NQR¹². Despite numerous attempts to elucidate the mechanism of the redox-mediated Na⁺ translocation in Na⁺-NQR, no consensus has been reached. To address this, Na⁺-NQR structures of different reaction states (or different reaction conditions) are needed.

In this study, we combine cryo-EM and atomic level MD simulations to show how each subunit undergoes conformational rearrangements that facilitate electron transfer within the enzyme and how these rearrangements drive Na⁺ translocation. Our synergistic approach demonstrates that the reduction of the 2Fe-2S^NqrD/E triggers Na⁺ translocation across the membrane by coordinating with structural rearrangements in the NqrD/E subunits, which subsequently influence positioning of NqrC and NqrF. This study provides structural insights into the mechanism of Na⁺ translocation coupled to electron transfer in Na⁺-NQR.

Results

Strategy for the preparation of cryo-EM grids

As described in the Introduction, structural studies on Na⁺-NQR to date suggest that redox driven Na⁺-translocation likely requires structural rearrangements in three different parts of the enzyme: the cytoplasmic domain of NqrF, the membrane embedded helices of NqrD/E, and the periplasmic region of NqrC, in order to modulate the distances between key pairs of redox cofactors: 2Fe-2S^NqrF to 2Fe-2S^NqrD/E and 2Fe-2S^NqrD/E to FMN^NqrC7. To capture snapshots of Na⁺-NQR from V. cholerae in other putative conformational states, cryo-EM grids were prepared under nine different conditions that alter the catalytic cycle of the enzyme (Table 1): wild-type enzyme reduced by NADH in the presence (1) or absence (2) of Na⁺ (WT_red, WT_red/-Na⁺, respectively), (3) wild-type reduced by NADH in the presence of the inhibitor korormicin A (WT_red+KR), (4) wild-type reduced by NADH in the presence of the inhibitor aurachin D-42 (WT_red+AD-42), (5) NqrB-G141A reduced by NADH in the presence of korormicin A (NqrB-G141A_red+KR), NqrB-T236Y as isolated (6) or reduced (7) by NADH (NqrB-T236Y_ox, NqrB-T236Y_red, respectively), and NqrC-T225Y as isolated (8) or reduced (9) by NADH (NqrC-T225Y_ox, NqrC-T225Y_red, respectively).

Table 1.

The list of samples for cryo-EM analysis

Sample name (PDB ID)	variant	NADH	inhibitor	Na+	trait	TMHNqrCDEF conformationc
WTred (9UUU)	wild-type	+	−	+	−	inward-open
WTred/ -Na+a “up” state (9UD6) “middle” state (9UD8) “down” state (9UD9)	wild-type	+	−	−	−	inward-open (all states)
WTred + KR (9UD5)	wild-type	+	KR	+	−	inward-open
WTred + AD-42 (9UDG)	wild-type	+	AD-42	+	−	inward-open
NqrB-G141Ared + KRb “stable” state (9UDA) “shifted” state (9UDF)	NqrB-G141A	+	KR	+	KR resistant	inward-open (“stable” state) outward-open (“shifted” state)
NqrB-T236Yox (9UD3)	NqrB-T236Y	−	−	+	FMNNqrB missing	inward-open
NqrB-T236Yred (9UD4)	NqrB-T236Y	+	−	+	FMNNqrB missing	inward-open
NqrC-T225Yox (9U5G)	NqrC-T225Y	−	−	+	FMNNqrC missing	outward-open
NqrC-T225Yred (9UD2)	NqrC-T225Y	+	−	+	FMNNqrC missing	outward-open

^aParticles from the WT_red/ -Na⁺ were classified into “up”, “middle”, or “down” states based on the positions of ferredoxin-like domain in the NqrF subunit.

^bParticles from the NqrB-G141A_red + KR were classified into “stable” or “shifted” states based on the position of the hydrophilic region of the NqrC subunit.

^cThe TMH^NqrCDEF bundle adopts inward-open or outward-open conformations depending on the redox state of 2Fe-2S^NqrD/E.

The NqrB-T236Y and NqrC-T225Y mutation results in the loss of covalently bound FMN^NqrB and FMN^NqrC, respectively, disrupting electron flow within the enzyme²⁰. Korormicin A binds to the enzyme and allosterically inhibits the final electron transfer step in which ubiquinone becomes reduced^7,28. The NqrB-G141A mutant exhibits significant resistance to korormicin A, though the inhibitor still binds to the enzyme²⁸. These three mutations all interfere with the redox cycle of the enzyme and may thus be expected to influence the distributions in its conformational states. After the 3D reconstruction of particle images of the nine samples, each dataset was processed using the following analytical tools in the cryosparc software²⁹ toolbox to isolate intermediate states in molecular motion: 3D variability analysis (3DV)³⁰, and 3DFlex (3DF)³¹ (Supplementary Figs. 1–10, and Supplementary Table 1).

Structure of Na⁺-NQR reduced by NADH

We determined the structure of wild-type enzyme reduced by 5 mM NADH (WT_red) at a resolution of 3.01 Å (Supplementary Fig. 1). As all the spectroscopically visible cofactors in Na⁺-NQR are reported to be fully reduced under these conditions²⁰, this structure is thought to represent the fully reduced structure of the enzyme. The overall structure is very close to that of the enzyme as isolated (WT_ox, PDB ID: 7XK3), with a single, well-resolved state observed for the NqrC subunit (Supplementary Fig. 11a). In the structure of WT_red, NADH is bound in a long flat density around the Rossmann nucleotide binding motif of NqrF⁸ (Supplementary Fig. 12). Focusing on the membrane domain, the TMH bundle formed by NqrF, NqrD/E, and NqrC (TMH^NqrCDEF bundle, Supplementary Fig. 11a–i) has a structure consistent with an “inward-open” state. Notably, TMHs 3-4 of NqrE on the cytoplasmic side have shifted slightly toward 2Fe-2S^NqrD/E compared to the WT_ox structure (Supplementary Fig. 11a).

NqrF is a cytoplasmic subunit anchored to the membrane by a single transmembrane helix (TMH1^NqrF). It is also tightly bound to the neighboring hydrophilic NqrA subunit through electrostatic interactions. The hydrophilic part of NqrF is highly flexible and comprises two distinct domains: a ferredoxin-NADPH reductase (FNR)-like domain (K132–G408) containing the FAD cofactor, and a ferredoxin-like domain (G33–K131) harboring the 2Fe-2S^NqrF as its cofactor^8,9 (Fig. 1b).

In the currently obtained cryo-EM structures, including that of fully reduced state (WT_red), the NqrF subunit displays high conformational flexibility, adopting multiple distinct states. Nevertheless, the extent of flexibility does not extend into the membrane region as would be necessary for electron transfer. The distance between 2Fe-2S^NqrF and the 2Fe-2S^NqrD/E, even at their closest proximity, is 30 Å (edge to edge), which is inconsistent with catalytic turnover rates. Therefore, during the catalytic cycle, NqrF likely undergoes a pivotal motion of its ferredoxin-like domain between FAD and 2Fe-2S^NqrD/E, while the TMH1^NqrF remains firmly anchored to the membrane. However, the electron transfer in NqrF is too fast (below milliseconds) to capture the snapshot in motion²¹.

Resolution of three conformations of the cytoplasmic NqrF subunit

To capture such rapid motions of NqrF, Na⁺-NQR was reduced by 5 mM NADH in the absence of Na⁺ (WT_red/-Na⁺). In these conditions, the rate constant for FMN^NqrC reduction is slowed approximately eight fold²⁰, which may allow the identification of previously unobserved conformational states of NqrF.

Following several rounds of heterogeneous refinement without any solvent mask, three distinct conformational states of NqrF were identified. These states, named as “up”, “middle”, and “down”, were distinguished based on the position of the ferredoxin-like domain (Fig. 1b, c, and Supplementary Figs. 2 and 13). Particles belonging to each class were then subjected to non-uniform refinement. The obtained maps which contain continuous density corresponding NqrF were further refined using 3DF with a solvent mask covering entire protein complex (Supplementary Fig. 10).

The “up” state comprises over 87% of the particle population in the dataset, providing a high-resolution map at 2.65 Å, where the distance between FAD^NqrF and 2Fe-2S^NqrF has increased to 15.7 Å as the ferredoxin-like domain is rotated compared to the enzyme as isolated (WT_ox). The “down” state, reconstructed from the smallest population of the particles (5%), gave a final density map at 3.11 Å. Remarkably, in the “down” state, the ferredoxin-like domain fits into the pocket formed by the NqrD/E subunits, whereas this pocket is exposed to the cytoplasm in the other two states. Consequently, the distance between 2Fe-2S^NqrF and 2Fe-2S^NqrD/E is reduced to 16.6 Å, consistent with electron transfer between them, while the distance between FAD^NqrF and 2Fe-2S^NqrF is increased to 31.0 Å (Fig. 1c). Although the resolution of the “middle” state is relatively low (3.75 Å), the density map is sufficient to position the ferredoxin-like domain in an intermediate location between the “down” and “up” states (Fig. 1b, c). Among the three conformational states identified, only the “down” state provides a structural alignment compatible with electron transfer between 2Fe-2S^NqrF and 2Fe-2S^NqrD/E.

Notably, in all three states, the structures of the membrane domains, including TMH^NqrF, remain almost identical to those in WT_red, with the TMH^NqrCDEF bundle in an “inward-open” state (Fig. 2 and Supplementary Fig. 11b, c). This shows that the reduction of the NqrF cofactors allows the ferredoxin-like domain harboring 2Fe-2S^NqrF to bridge transport of electrons from FAD^NqrF to 2Fe-2S^NqrD/E, without altering the conformation of the TMH^NqrCDEF bundle.

Fig. 2. Conformation of the transmembrane helices of NqrCDEF as the catalytic cycle progresses from four different views.

Four steps in the catalytic cycle are represented by the following structures: WT_ox, “down” state of WT_red/ - Na⁺, “shifted” state, and “stable” state of NqrB-G141A_red + KR. Gray images show the structure of the previous state; red arrows show the direction of movement from the previous state to the current state. To aid orientation, the top views in (a–d) show the overall architecture of the enzyme, indicating the viewing directions for (a–d). Dashed lines correspond to the planes used for sectional views in (c). a Cytoplasmic view. b Periplasmic view. c Sectional side views of NqrCDEF, sliced perpendicular to the membrane. d Tube models from front view. The critical amino acid residues to coordinate Na⁺ in the inward and outward gates in the MD simulation are shown as stick models. Supplementary Videos 2,3, and 4 show how the arrangement of the helices morphs during the catalytic cycle from the complete view of the enzyme, cytoplasmic, and periplasmic views.

The conformational shift of the NqrC subunit between NqrB and NqrD/E

Korormicin A (KR) and aurachin D-42 (AD-42) are specific inhibitors of Na⁺-NQR that can serve as tools for capturing key reaction intermediates of the enzyme^32–34. However, wild-type enzyme with korormicin A or aurachin D-42 bound (WT_red + KR, WT_red + AD-42), treated with NADH, did not show any distinct reaction intermediates (Supplementary Figs. 3, 4, 11h, i and 14), as the inhibitor completely terminates UQ reduction at the final step of the catalytic cycle, mimicking WT_red. To decelerate the catalytic cycle of Na⁺-NQR, we prepared a cryo-EM grid of the NADH-treated NqrB-G141A mutant with korormicin A bound (NqrB-G141A_red + KR). This alanine substitution excludes the alkyl side chain of korormicin A from the binding cavity (Supplementary Fig. 15), resulting in a decrease in the binding affinity (~160-fold increase in IC₅₀ values) although the inhibitor molecule still binds to the mutated enzyme^28,35,36. This distinct combination of mutation and inhibitor could potentially influence the catalytic cycle of the enzyme, and lead to the identification of multiple conformational states of the enzyme.

The image particles were processed through iterative 3DV with a mask covering the entire Na⁺-NQR complex. This approach identified two distinct states of Na⁺-NQR, one in which the hydrophilic domain of the NqrC subunit approaches the NqrD/E and one in which it approaches NqrB, likely reflecting its motion during catalytic turnover (Supplementary Fig. 5, and Supplementary Video 1). These continuous density maps were further refined by 3DF (Supplementary Fig. 10b) to reconstitute two conformational states: “stable” and “shifted” with particle populations of 88 and 12%, respectively (Fig. 3).

Fig. 3. The shift of NqrC: Transition of NqrC from “stable” to “shifted” conformation.

a Comparison of two states in NqrB-G141A_red + KR sample, complete structure from three different views; “stable” state (gray) and “shifted” state (color). In the “stable” state (88% of particles) the hydrophilic region of NqrC is located close to NqrB; in the “shifted” state (12% of particles) NqrC is close to NqrD/E (Supplementary Fig. 5). b NqrC is shown alone. Left, “stable” state from NqrB-G141A_red + KR; center, “shifted” state from NqrB-G141A_red + KR; right, NqrC from NqrC-T225Y_red. For comparison, the gray silhouettes in the center and right-hand images show the “stable” state. The red arrows indicate how specific amino acids (L33 in the joint of TMH1 and hydrophilic domain, G213 in the hydrophilic domain, P174 in the membrane domain, and T225 (light green) forming a covalent bond with FMN) shifted from “stable” states. The loop of Glu169-Gly177 is colored in red.

In the “stable” state (Fig. 3b, and Supplementary Fig. 11j), resolved at 2.61 Å, the NqrC subunit has moved closer to NqrB, consistent with rapid electron transfer between covalently bound FMN^NqrC and FMN^NqrB (7.8 Å). This structure aligns with that of the enzyme treated with NADH (WT_red). In contrast, the “shifted” state, reconstructed at 2.93 Å with improved local alignment of particle images, exhibits a weaker density for the hydrophilic domain of NqrC, suggesting increased flexibility in this region. This flexibility makes full de novo modeling challenging; however, a structural model with rigid body fitting indicates that the hydrophilic domain of NqrC subunit has shifted toward the NqrD/E subunits.

In the shifted state (Fig. 3b, and Supplementary Fig. 11k), while FMN^NqrC remains distant from 2Fe-2S^NqrD/E (~27 Å), the loop of Glu169-Gly177^NqrC, which includes Thr173 and Leu176 that stabilize the isoalloxazine ring of this FMN⁹, is embedded in the pocket formed by the NqrD/E subunits. The existence of these two distinct conformations shows that the periplasmic domain of NqrC, which houses FMN^NqrC, switches its position between NqrB and NqrD/E in a way that allows it to mediate electron transfer from 2Fe-2S^NqrD/E to FMN^NqrB (Figs. 1a and 3). Notably, the TMH^NqrCDEF bundle adopts the “inward-open” conformation in the “stable” state but transitions to the “outward-open” conformation in the “shifted” state (Fig. 2). This inward-to-outward transition of the bundle appears to be linked to the relocation of the periplasmic domain of NqrC from the NqrB side to the NqrD/E side.

We note that the distance between 2Fe-2S^NqrD/E and FMN^NqrC in the shifted state (~27 Å) would need to be reduced by about half for efficient electron transfer. This suggests the existence of an additional intermediate in which the hydrophilic domain of NqrC shifts considerably closer to NqrD/E. As discussed later, the structure of the FMN^NqrC-lacking mutant may represent a pseudo-structure corresponding to such an intermediate.

The rearrangement of NqrC is coordinated with the inward/outward transition of the TMH^NqrCDEF bundle

To understand the factors that could cause NqrC to move from the NqrB side to the NqrD/E side, we analyzed the structure of mutants lacking either FMN^NqrB (NqrB-T236Y) or FMN^NqrC (NqrC-T225Y), both of which are covalently bound to threonine residues. In the mutant lacking FMN^NqrB (Supplementary Fig. 16a), both “as isolated” and “NADH-treated” enzymes (NqrB-T236Y_ox and NqrB-T236Y_red) showed “inward-open” structures with the NqrC subunits positioned near NqrB at resolutions of 3.83 and 3.31 Å, respectively (Supplementary Figs. 6, 7 and 11d, e). These structures closely resemble WT_ox and WT_red (Supplementary Fig. 17), suggesting that this mutant adopts a similar conformation in which NqrC has shifted fully from the NqrD/E side to the NqrB side. In contrast, the mutant lacking FMN^NqrC (Supplementary Fig. 16b) show identical outward-open structures with NqrC shifted toward NqrD/E, irrespective of NADH treatment (NqrC-T225Y_ox and NqrC-T225Y_red) with resolutions of 2.66 and 2.59 Å, respectively (Supplementary Figs. 8, 9 and 11f, g). In these conformations, as in the “shifted” state in NqrB-G141A_red + KR, the periplasmic domain of NqrC is positioned closest to the dimetric NqrD/E subunits (Fig. 3b and Supplementary Fig. 16c). These findings, including the structure of the korormicin A-bound NqrB-G141A mutant (NqrB-G141A_red + KR), suggest that the rearrangement of NqrC is coupled to the inward-to-outward transition of the NqrD/E subunits, which bind 2Fe-2S^NqrD/E.

Finally, we consider the factor that drives inward-to-outward transition of NqrD/E subunits. In the structure of Na⁺-NQR reduced by NADH in the absence of Na⁺ (WT_red/-Na⁺), all observed NqrF states (up, middle, and down) corresponded to the inward-open state of the TMH^NqrCDEF bundle (Fig. 2, Supplementary Fig. 11a–c), with NqrC in the “stable” state. This suggests that the position of 2Fe-2S^NqrF does not influence the membrane domain structure. By contrast, the outward-open state of the TMH^NqrCDEF bundle was always associated to the “shifted” state of NqrC. Considering that 2Fe-2S^NqrD/E must be reduced in order to transfer an electron to FMN^NqrC when NqrC shifted toward NqrD/E also in the wild-type enzyme, it is plausible that, the redox state of 2Fe-2S^NqrD/E may drive the inward-to-outward transition of the TMH^NqrCDEF bundle. This transition likely triggers the relocation of the periplasmic domain of NqrC, thereby facilitating the reduction of FMN^NqrC.

Shift in local resolution map reflects redox driven conformational changes of Na⁺-NQR

During single-particle analysis, we noticed that some of the redox-induced rearrangements of Na⁺-NQR subunits correlate with shifts in the local resolution map. A notable example of this is the rearrangement of NqrF during electron transfer from FAD to 2Fe-2S^NqrD/E (Fig. 1b, c). In the “down” state, where 2Fe-2S^NqrF is positioned close enough to 2Fe-2S^NqrD/E to allow facile electron transfer, the ferredoxin-like domain of NqrF shows lower resolution, while the FNR-like domain containing FAD remains highly resolved (Supplementary Fig. 10a). This suggests that reduction of 2Fe-2S^NqrF increases the flexibility of the ferredoxin-like domain, thereby raising the probability of physiological electron transfer to 2Fe-2S^NqrD/E, while the FNR-like domain remains stable, preventing electron backflow. In addition to the increased flexibility of the ferredoxin-like domain upon reduction, it is also possible that the binding of this domain to the NqrD/E pocket becomes entropically more favorable when 2Fe-2S^NqrF is reduced.

Interestingly, this observation is even more pronounced in the structure of the NqrB-G141A mutant with bound korormicin A (NqrB-G141A_red + KR), which exhibits two distinct states of the TMH^NqrCDEF bundle. In the “stable” state, where the bundle adopts an “inward-open” conformation, the cytoplasmic domain of NqrF appears blurred (flexible), whereas the periplasmic domain of NqrC exhibits sharp density (rigid). Conversely, in the “shifted” state, where the bundle adopts an “outward-open” conformation, NqrF displays sharp density (rigid), while NqrC appears blurred (flexible) (Supplementary Figs. 5 and 10b).

A similar pattern appears in the NqrC-T225Y mutant, which lacks FMN^NqrC. This mutant adopts the “outward-open” conformation of its TMH^NqrCDEF bundle, similar to the “shifted” state, and shows the highest resolution for NqrF, indicating that it remains in a single, stable “up” conformation (Supplementary Figs. 8, 9). Together, these results suggest that the “inward-to-outward” transition alternates the flexibility of NqrF and NqrC, coordinating the probability of physiological electron transfer from 2Fe-2S^NqrF to FMN^NqrC via 2Fe-2S^NqrD/E. The shifting resolution patterns highlight how structural flexibility and stability work together to control electron flow.

A proposed mechanism for Na⁺ translocation

Recently, Müller and coworkers published a cryo-EM structure of the RNF complex from A. woodii²⁷. In this structure, the RnfA/E subunits, which coordinate a 2Fe-2S cluster, are homologous to the NqrD/E subunits of Na⁺-NQR. Long-range molecular dynamics (MD) simulations predicted that the dimeric RnfA/E subunits can adopt two distinct conformations—inward- and outward-open—depending on the redox state of the 2Fe-2S cluster. This rearrangement appears to be crucial for both Na⁺ translocation and electron transfer within the enzyme. Their report supports the hypothesis that, in Na⁺-NQR, the redox state of the 2Fe-2S^NqrD/E not only drives the inward-to-outward transition of TMHs^NqrCDEF bundle but also facilitates Na⁺ translocation.

We have performed MD simulations beginning from the cryo-EM structure with an inward-open state of TMHs^NqrCDEF bundle, which provide some insight into the mechanism of Na⁺ uptake from the cytoplasm (Supplementary Table 2). Spatial density analysis of MD simulation results showed that Na⁺ ions and water can access the cytoplasmic pocket near 2Fe-2S^NqrD/E in the wild-type Na⁺-NQR with reduced 2Fe-2S^NqrF and 2Fe-2S^NqrD/E (Fig. 4a, b). The reduction of 2Fe-2S^NqrD/E is linked to formation of a Na⁺ binding site on the cytoplasmic side. The negative charge introduced by the reduction of 2Fe-2S^NqrD/E should be neutralized by Na⁺ to achieve a stable occluded conformation and further allow the conformational change to the outward-open state. In contrast, Na⁺ ions could not fully access the cytoplasmic pocket with the oxidized 2Fe-2S^NqrD/E (Supplementary Fig. 18). Pairs of hydrophobic residues, that could function as gates for cation movement, are found above and below the Na⁺ binding site: Leu26^NqrD-Leu115^NqrE and Leu107^NqrD-Leu23^NqrE (Fig. 4f). These residues control the access of Na⁺ ions and water and are conserved between NQR and RNF (Supplementary Fig. 19).

Fig. 4. Na⁺ translocation pathway from MD simulation.

a, b The accessibilities of Na⁺ and water are shown in the structure of wild-type of Na⁺-NQR with reduced 2Fe-2S^NqrF and 2Fe-2S^NqrD/E. a The spatial density of Na⁺ is shown in red; the relative iso-density surface of 3.5 to the bulk value is shown. b The spatial density of water is shown in blue; the relative iso-density surface of 0.3 to the bulk value is shown. c–e A first conventional MD (cMD1) using the inward-open model containing reduced 2Fe-2S^NqrF and 2Fe-2S^NqrD/E based on WT_ox structure is followed by the targeted MD (TMD) inducing the inward-to-outward transition of TMHs bundle and the second conventional MD (cMD2) containing oxidized 2Fe-2Ss and reduced FMN^NqrC. c The inward gate and outward gate size are defined by the closest heavy-atom distances between Leu26^NqrD and Leu115^NqrE (orange), and between Leu107^NqrD and Leu23^NqrE (blue), respectively. The transitions of these gate sizes are shown during the simulation. d The z coordinates (perpendicular to the membrane surface) of the Na⁺ being translocated and the 2Fe-2S^NqrD/E are shown together with the position of the membrane surface. e Number of water molecules coordinated to the Na⁺ as a function of time in the simulation. f, g Representative snapshots during Na⁺ translocation in MD simulation. The amino acid residues that coordinate Na⁺ ion in the inward and outward gates are indicated in orange and blue stick models, respectively. Na⁺ bound in the inward-open and outward-open conformations are shown in (f) and (g), respectively. The release of Na⁺ to the periplasm is shown in (h).

Because the spontaneous inward-to-outward transition of the TMH^NqrCDEF bundle was not observed within the time scale of simulations, we employed targeted MD (TMD) simulation to induce this conformational transition³⁷. During the TMD simulation, an external force was applied to make the conformation of NqrD/E move toward the outward-open state. During the TMD and subsequent equilibrium simulations, the inside gate closed while the outside gate opened (Fig. 4c, f, g, and Supplementary Fig. 20). The conformational transition shifted the bound Na⁺ to the outward side of the binding site relative to 2Fe-2S^NqrD/E (Fig. 4d, g). With the oxidation of 2Fe-2S^NqrD/E, the bound Na⁺ was eventually released to the outer side of the membrane (Fig. 4h). Na⁺ translocation was observed in two additional independent trajectories following the TMD simulation (Supplementary Fig. 21). It is noteworthy that the Na⁺ ion was partially hydrated during the translocation with two or more coordinated water molecules (Fig. 4e). This may be due to the lack of acidic residues in the binding site that would strongly bind the ion and dehydrate it as in some transporter proteins³⁸. Instead, the binding site is made up of backbone atoms of Thr110^NqrD, Cys112^NqrD, Val118^NqrE, and Cys120^NqrE (Fig. 4f–h and Supplementary Fig. 22).

In the MD simulations, although Na⁺ remained hydrated with at least two water molecules during translocation, no continuous water connection spanning from the cytoplasmic to the periplasmic side was observed. Instead, spatial gaps in water were consistently present in both the cytoplasmic-open and periplasmic-open states (Supplementary Fig. 23). These discontinuities suggest that proton translocation or leakage through a Grotthuss-type water wire across the membrane is unlikely.

Discussion

The overall reaction mechanism of Na⁺-NQR

Since the first publication of a structural model of Na⁺-NQR, it has been clear from the spatial arrangement of the redox cofactors that structural rearrangements would have to play a critical role in the enzyme and that understanding these conformational changes would be essential to elucidating the mechanism of redox-driven Na⁺ pumping. In this study, we have identified at least five distinct conformational states of Na⁺-NQR, which may reflect key reaction steps of the enzyme. Based on the multiple conformations of Na⁺-NQR revealed by the current cryo-EM work and previous studies⁷, together with MD simulations, we propose an overall reaction mechanism for the enzyme, which progresses through six key steps (Fig. 5 and Supplementary Fig. 24).

Fig. 5. The model of conformational changes.

a The cascade of conformational changes. One electron flow is shown as black arrows. Step 1: NADH binding, Step 2: Reduction of 2Fe-2S^NqrD/E and Na⁺ uptake, Step 3: Transition to outward-open conformation and Na⁺ release, Step 4: Return to inward-open conformation, Step 5: UQ binding, Step 6: UQ release. Since UQ requires a two-electron reduction, steps 5 or 6 need to go back steps 2 or 3. Concentric partial circles in red and blue around NqrF and NqrC, respectively, indicate flexibility of the corresponding region. b The detailed model of essential TMHs for conformational changes in step 3–4.

Step 1 (NADH binding): In the absence of NADH, all cofactors are fully oxidized except for RBF, which exists as a stable neutral semiquinone¹⁵. This initial state I is characterized by the dimeric NqrD/E subunits adopting an inward-open conformation. The hydrophilic domain of NqrF remains flexible and does not take on a distinct structure, but its range of motion does not allow it to come close to the NqrD/E subunits. When NADH binds to the NqrF subunit, it donates two electrons. One of these electrons is transferred to the 2Fe-2S^NqrF, resulting in the designation of this state as II.

Step 2 (Reduction of 2Fe-2S^NqrD/E and Na⁺ uptake): In state II, the reduction of 2Fe-2S^NqrF enhances the flexibility of the ferredoxin-like domain of NqrF. This highly flexible domain then approaches the cytoplasmic pocket formed by the dimeric NqrD/E. As a result of this rearrangement, electron transfer occurs from 2Fe-2S^NqrF to 2Fe-2S^NqrD/E. The negative charge on the reduced 2Fe-2S^NqrD/E allows for the accommodation of hydrated Na⁺ between the inward and outward gates, which are composed of conserved hydrophobic amino acids. This state is designated as state III.

Step 3 (Transition to outward-open conformation and Na⁺ release): In this step, the NqrD/E subunits undergo a conformational change to the outward-open state upon the reduction of the 2Fe-2S^NqrD/E complex and the binding of Na⁺. This transition exposes the bound Na⁺ to the periplasmic side and increases the flexibility of the hydrophilic domain of NqrC. This flexibility allows NqrC to move closer to NqrD/E. In contrast, the flexibility of NqrF decreases, which helps prevent electron backflow into the 2Fe-2S^NqrF complex. This Na⁺- and redox-dependent repositioning of NqrC facilitates electron transfer from the 2Fe-2S^NqrD/E to FMN^NqrC. At the same time, the bound Na⁺ loses its affinity and is released into the periplasm as 2Fe-2S^NqrD/E is reoxidized. This state is referred to as state IV.

Step 4 (Return to inward-open conformation): In state IV, the reoxidation of 2Fe-2S^NqrD/E causes a conformational relaxation of the NqrD/E subunits, allowing them to return to the inward-open conformation. This transition enables NqrC to move back toward the NqrB side. This movement aids in the transfer of electrons from FMN^NqrC to RBF^NqrB through FMN^NqrB, resulting in the generation of state V. It is important to note that: (i) the reduction of UQ is a two-electron process, while (ii) FMNs and RBF undergo single-electron redox reactions. Consequently, a similar series of reactions (from steps 3 to 4) occurs for the transfer of a second electron, which is coupled with the translocation of a second Na⁺.

Step 5 (UQ binding): In this step, the final electron acceptor, ubiquinone (UQ), binds to the enzyme in state V, resulting in the UQ-bound enzyme, referred to as state VI. Through two sequential redox cycles, from 2Fe-2S^NqrF to RBF^NqrB, UQ accepts two electrons, which leads to the formation of ubiquinol (UQH₂).

Step 6 (UQ release): UQ is finally released from the enzyme in state VI, allowing the enzyme to return to its initial oxidized state I. Although the structure of Na⁺-NQR with bound UQ has not yet been identified, photoaffinity labeling experiments suggest that the UQ head-ring is positioned within a pocket in the NqrA subunit, ~20 Å above the membrane surface³⁶. It is important to note that Hau et al. reported a structure of UQ₂-bound Na⁺-NQR, in which UQ₂ was modeled inside the inhibitor binding pocket located in the N-terminal region of NqrB⁶. However, the density map of the UQ₂ molecule within the pocket is too weak to definitively model the UQ₂ structure. Therefore, it is premature to conclude that UQs accept electrons from RBF in the inhibitor binding pocket of NqrB.

Na⁺-translocating pathway

The mechanism of redox-driven Na⁺ translocation by Na⁺-NQR remains to be fully elucidated. Based on the X-ray crystallographic and cryo-EM models, Fritz and colleagues proposed that cytoplasmic Na⁺ is translocated to the periplasm via a pathway within NqrB, where it is coordinated by residues such as Ile371, Arg372, Pro376, and Tyr378^5,6. Our MD simulations also identified Na⁺ binding to periplasmic residues of NqrB, such as Thr236 and Ser239 (Fig. 4a, b, Supplementary Figs. 18, 22, and 25, and Supplementary Table 3). However, this Na⁺ was consistently observed across all enzyme states identified in this study, indicating that NqrB is unlikely to be directly involved in redox-driven Na⁺ translocation across the membrane.

Although we were unable to identify any continuous ion translocation pathway spanning the transmembrane region of NqrB, our cryo-EM and MD analyses demonstrate that the inward/outward transition of the NqrD/E subunits is driven by redox reactions and is tightly coupled to Na⁺ translocation from the cytoplasm to the periplasm. These results lead to conclusions markedly different from those of previous Na⁺-NQR studies^5,6, which attributed the Na⁺ pumping function to NqrB. Consistent with the recent study on the A. woodii RNF complex²⁷, our findings strongly suggest that Na⁺ translocation is mediated exclusively by the NqrD/E subunits, rather than by NqrB.

The redox-dependent inward-to-outward transition of the TMH^NqrCDEF bundle was not observed in our MD simulations. During structural modeling of Na⁺-NQR in the inward- and outward-open states, we noticed that the conformation of the NqrD/E subunits appears to undergo a conformational change centered on the proline at position of 114 in TMH4 of NqrE—a highly conserved residue located near the 2Fe-2S^NqrD/E and surrounded by hydrophobic residues (Fig. 5b, and Supplementary Fig. 19). A change in the redox state of 2Fe-2S^NqrD/E may be linked to subtle local rearrangement in adjacent amino acid residues by electrostatic interactions, and these rearrangements could be amplified by conserved proline, leading to larger conformational shifts (Supplementary Fig. 19d). A similar proline-mediated switching mechanism has been proposed in the structure of voltage-gated potassium (Kv) chanel³⁹.

Finally, we address how Na⁺ is translocated against the sodium motive force (Na⁺ gradient across the membrane). Our findings indicate that the “inward-to-outward” transition of the TMH^NqrCDEF bundle alternately regulates the flexibility of the NqrF and NqrC subunits, thereby modulating the probability of physiological electron transfer from 2Fe-2S^NqrF to FMN^NqrC via 2Fe-2S^NqrD/E. This mechanism likely essential for the redox-driven Na⁺ translocation—namely, that electron does not flow forward without Na⁺ being expelled to the periplasm. For instance, in the outward-open conformation with periplasmic Na⁺ bound, electron backflow from 2Fe-2S^NqrD/E to 2Fe-2S^NqrF would be necessary to trigger the transition to the inward-open conformation, that would allow the bound Na⁺ to move back toward the cytoplasm. Nevertheless, as NqrF is firmly stabilized in this state, such electron backflow may not take place.

This property of Na⁺-NQR highlights its functional difference from the RNF complex, which shares the general principle of redox-driven Na⁺ pumping. Unlike Na⁺-NQR, the RNF complex can also operate Na⁺ translocation in the reverse direction, driven by NADH oxidation²⁷. These functional differences are primarily reflected in the RnfB and NqrF subunits, each donating one electron to the 2Fe-2S in RnfA/E and NqrD/E to pump Na⁺, respectively. RnfB accepts a single electron from ferredoxin and transfers it through a series of Fe-S clusters, whereas NqrF receives two electrons from NADH. The two-electron nature of NADH oxidation by NqrF may serve as an electron-gating mechanism that regulates Na⁺ translocation. Such mechanistic divergence between RnfB and NqrF—both donating electrons to the Na⁺-translocating RnfA/E and NqrD/E subunits, respectively— could have defined the directionality of Na⁺ transport in the evolutionary ancestor of Na⁺-NQR.

In summary, the present structural analyses of Na⁺-NQR in multiple distinct conformations, combined with MD simulations, provide critical insights into the coupling mechanism between the electron transfer and Na⁺ translocation. These findings illustrate how structural rearrangements of individual subunits can control Na⁺ translocation and electron transfer, to achieve coupling between the two processes. This study establishes a valuable framework for further mechanistic studies of Na⁺-NQR and antibiotics development targeting the enzyme.

Methods

Na⁺-NQR expression and purification

Expression of Vibrio cholerae Na⁺-NQR from a plasmid, in the parent organism, construction of site-specific mutants, and cultivation of V. cholerae strains have been described previously⁷. Na⁺-NQR was purified as previously described^7,8. The membrane pellets were solubilized in a buffer containing 50 mM NaPi, 300 mM NaCl, 5.0 mM imidazole, and protease inhibitor cocktail (Sigma-Aldrich), 0.4% (w/v) n-dodecyl- β-D-maltoside (DDM), pH 8.0, with stirring for 1 h at 4 °C. The mixture was clarified by ultracentrifugation (100,000 × g), and the supernatant was mixed with 20 mL of Ni-NTA resin (Qiagen) that had been equilibrated with binding buffer (50 mM NaPi, 300 mM NaCl, 5.0 mM imidazole, and 0.05% DDM). After gentle agitation for 1 h at 4 °C, the suspension was poured into an empty plastic column. The column was washed with 4 volumes of the binding buffer, followed by 5 volumes of a wash buffer (same as binding buffer but containing 10 mM imidazole). The protein was eluted from the column using an elution buffer (same as binding buffer but containing 100 mM imidazole). The resulting Na⁺-NQR preparation was applied to a DEAE anion exchange column (HiPrep DEAE FF 16/10, connected to an ÄKTA system, Cytiva) equilibrated with buffer A (50 mM Tris/HCl, 1.0 mM EDTA, 5% (v/v) glycerol, and 0.05% (w/v) DDM, pH 8.0). The protein was eluted with a linear gradient of buffer B (buffer A containing 2.0 M NaCl), and the fractions containing pure Na⁺-NQR were pooled and concentrated to 15–20 mg of protein/mL. Subunit assembly was confirmed by SDS-PAGE⁴⁰ and enzyme activity was checked by the inhibitors-sensitive NADH-UQ₁ oxidoreductase activity measurements at 282 nm (ε = 14.5 mM⁻¹ cm⁻¹)². To remove glycerol from samples, the target protein was further separated by size-exclusion chromatography (Superose^TM 6 increase 10/300 gel filtration column, Cytiva) with an elution buffer containing 50 mM Tris-HCl, 100 mM NaCl, 1.0 mM EDTA, 0.05% DDM (pH 8.0).

Cryo-EM acquisition

To decrease DDM concentrations, immediately prior to cryo-EM grid preparation, all of the purified Na⁺-NQR samples were washed with the same elution buffer without detergent (50 mM Tris-HCl, 100 mM NaCl, 1.0 mM EDTA, pH 8.0). To obtain the reduced form of Na⁺-NQR, Na⁺-NQR solution was mixed with an NADH solution to final concentration of 5 mM just before application onto a grid. 2.7 µL of Na⁺-NQR solution at the concentration of 10 mg/mL was applied onto a glow-charged Quantifoil Cu R1.2/1.3. The grid was blotted by Vitrobot IV and vitrified with liquid ethane. Cryo-EM movies were acquired with a Titan Krios electron microscope (Thermo Fisher, USA) equipped with K3 BioQuantum camera (Gatan, United States). Data collection was automatically controlled by SerialEM software at a nominal magnification of 81,000 and the pixel size is 0.88 Å/pix. The total electron dose and frame rate were 60 electron/Ų and 0.1 s, respectively.

Cryo-EM data processing

The details of image processing steps in each condition are shown in Supplementary Figs. 1–9. All of seven datasets (WT_red, WT_red/ -Na⁺, WT_red + KR, WT_red + AD-42, NqrB-G141A_red + KR, NqrB-T236Y_ox, NqrB-T236Y_red, NqrC-T225Y_ox, and NqrC-T225Y_red) were processed by cryoSPARC v3.3.1 or v4.0. 6086, 13,302, 4881, 5,096, 13,316, 4440, 8212, 13,302, and 5086 movies were used for data analyses, respectively. After motion correction, CTF estimation, and manually curate exposures, the machine-learning-based particle picking by Topaz software extracted 1,391,844, 2,497,690, 2,157,492, 1,072,343, 2,378,419, 1,026,158, 2,496,421, 3,653,368, and 1,384,33 particles from the datasets, respectively. After particle selections with Heterogeneous Refinement and Ab-initio Reconstruction at the class similarity of 0.6, selected better particles were used in Non-uniform Refinement with the optimizing per-particle defocus option. The datasets having weak density regions were classified with 3DV or 3DC, and the components having stronger density were refined by Local refinement, and/or Non-uniform Refinement. Finally, all datasets provided density maps at 3.07- (WT_red,), 2.65- (“up” state of WT_red/ -Na⁺), 3.75- (“middle” state of WT_red/ -Na⁺), 3.11- (“down” state of WT_red/ -Na⁺), 2.90- (WT_red + KR), 2.74- (WT_red + AD-42), 2.61- (“stable” state of NqrB-G141A_red + KR), 2.93- (“shifted” state of NqrB-G141A_red + KR), 3.83- (NqrB-T236Y_ox), 3.31- (NqrB-T236Y_red), 2.66- (NqrC-T225Y_ox), and 2.59- (NqrC-T225Y_red) Å resolution, respectively. These resolutions were estimated based on the gold standard criteria (FSC = 0.143). Furthermore, to capture the overall structure of flexible NqrF subunits, non-uniform refined models (WT_red/ -Na⁺, and NqrB-G141A_red + KR) were reconstructed with 3DFlex Refinement using a mask covering all subunits. The Flex maps provided by 3DFlex reconstruction are referenced for model building.

Model building and refinement

The atomic models were built based on our previous cryo-EM structures of Na⁺-NQR (PDB ID: 7XK3)⁷. The initial models were fitted into the density maps with the Chain Refine and Space Refine Zone functions of COOT software⁴¹. The parts containing weak densities were adjusted with rigid body fitting and morphing. The Flex map output from 3DFlex Reconstruction was used to model the heterogeneous conformations with flexibility. Using the phenix_real_space_refine program⁴², the corrected models were refined with secondary structure and Ramachandran restraints. These results were manually validated with COOT. The iteration of model refinement provided the correct models following the geometry. The model validation was evaluated with MolProbity⁴³. To avoid over-fitting, the final models were validated with Fourier shell correlation computed from the final density maps. The model statistics are shown in Supplementary Table 1. Images of structures were generated with UCSF chimeraX software⁴⁴. UCSF ChimeraX was used to compute the electrostatic potential, with default parameters and the Coulombic command with a range of −10 to 10 kcal/(mol·e). Rigid-body fitting was used to build the atomic models into the weaker density maps. The initial atomic model was positioned in the map without any modifications to its backbone or side chains. This fitting was performed with UCSF ChimeraX.

Molecular dynamics simulation

The all-atom MD simulations of the Na⁺-NQR complex were performed using the cryo-EM structure (PDB ID: 7XK3) and the crystal structure (PDB ID: 8ACW) as the initial structures. For modeling of missing residues at the N-termini (for NqrB, NqrC, and NqrD) and C-termini (for NqrB and NqrC), GalaxyFill plug in CHARMM-GUI⁴⁵ was used. The pK_a of titratable residues was evaluated using PROPKA⁴⁶, and the protonation state was determined at pH 8, which is the same pH at which the experiments were conducted. The cofactors (FAD^NqrF, 2Fe–2S^NqrF, 2Fe–2S^NqrD/E, FMN^NqrC, FMN^NqrB, and RBF^NqrB) were retained as in the PDB structures. FAD^NqrF is not covalently attached to NqrF. 2Fe-2S^NqrF is covalently attached to NqrF-Cys70, NqrF-Cys76, NqrF-Cys79, and NqrF-Cys111. The 2Fe-2S^NqrD/E is covalently attached to NqrD-Cys29, NqrD-Cys112, NqrE-Cys26, and NqrE-Cys120. FMN^NqrC is covalently attached to NqrC-Thr225, while FMN^NqrB is attached to NqrB-Thr236. RBF^NqrB is not covalently attached to NqrB.

The Na⁺-NQR structure was embedded in a 150 × 150 Å POPC membrane and solvated with TIP3P water and 150 mM sodium and chloride ions with the Membrane Builder plugin in CHARMM-GUI^47,48. The CHARMM36 force field⁴⁹ was used except for the cofactors whose parameters were adapted from the works of Kaila group^27,50,51. The total number of atoms is ~380,000.

The systems were energy minimized with the restraint of non–hydrogen POPC, protein, and cofactor atoms for 10,000 steps. Then, they were equilibrated in three stages: (1) 100 ps equilibration at T = 150 K and no pressure control with the restraint of non–hydrogen POPC, protein, and cofactor atoms, (2) 500 ps equilibration at T = 310 K and 1 bar pressure with the restraint of non–hydrogen protein and cofactor atoms, (3) 500 ps equilibration at T = 310 K and 1 bar pressure with the restraint of protein–Cα and non–hydrogen cofactor atoms. The restraint force constant was 1.0 kcal mol⁻¹ Å⁻² for each atom. The production runs were performed at T = 310 K and 1 bar pressure without restraints. Long-range electrostatic interactions were calculated by the particle mesh Ewald method⁵² with a direct space cut off of 12 Å. Langevin dynamics with 1 ps⁻¹ damping coefficient was used for temperature control at T = 310 K, and the Nosé–Hoover Langevin piston was used for pressure control at 1 bar^53–55. The integration timestep was set to 2 fs by applying the SHAKE method⁵⁶ to bonds involving hydrogen. Targeted MD (TMD) simulations were also performed using an outward-open conformation of NqrD/E (shifted state of NqrB-G141A_red + KR) as the reference. The collective variable of the TMD simulations was a root-mean-square deviation (RMSD) to the reference structure using all Cα atoms of NqrD/E. The force constant for TMD force was set to 500 kcal mol⁻¹ Å⁻² and applied to the Cα atoms of NqrD and NqrE. All MD simulations were performed using NAMD 2.12 /2.14/3.0⁵⁷. VMD⁵⁸, UCSF Chimera⁵⁹, UCSF ChimeraX⁴⁴ and MDAnalysis⁶⁰ have been used for the analysis. To explicitly direct readers to information regarding reproducibility of the MD simulations, MD simulation checklist was included as Supplementary Table 4.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

M.I.-F., J.K., T.M., B.B., H.M., and M.M. designed the research; M.I.-F., T.M., and B.B. expressed and purified Na⁺-NQR; M.I.-F., J.K., and T.M. performed structural determination and built the atomic models; T.S. and K.O. performed MD simulation; M.I.-F., T.S., J.K., T.M., K.O., T.K., B.B., H.M. and M.M. analyzed data; K.J. and M.M. directed the project and wrote the paper with M.I.-F., T.S., T.M., K.O., B.B. and H.M.

Peer review

Peer review information

Nature Communications thanks Leonid Sazanov and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The cryo-EM maps have been deposited in the EMDB under accession codes, 64518 (WT_red), 63872 (NqrC-T225Y_ox), 64059 (NqrC-T225Y_red), 64060 (NqrB-T236Y_ox), 64061 (NqrB-T236Y_red), 64063 (WT_red/ -Na⁺, “up” state), 64064 (WT_red/ -Na⁺, “middle” state), 64065 (WT_red/ -Na⁺, “down” state), 64062 (WT_red + KR), 64069 (WT_red + AD-42), 64066 (NqrB-G141A_red + KR, “stable” state), and 64068 (NqrB-G141A_red + KR, “shifted” state). The atomic models have been deposited in the Protein Data Bank under accession codes, 9UUU (WT_red), 9U5G (NqrC-T225Y_ox), 9UD2 (NqrC-T225Y_red), 9UD3 (NqrB-T236Y_ox), 9UD4 (NqrB-T236Y_red), 9UD6 (WT_red/ -Na⁺, “up” state), 9UD8 (WT_red/ -Na⁺, “middle” state), 9UD9 (WT_red/ -Na⁺, “down” state), 9UD5 (WT_red + KR), 9UDG (WT_red + AD-42), 9UDA (NqrB-G141A_red + KR, “stable” state), and 9UDF (NqrB-G141A_red + KR, “shifted” state). The initial model of Na⁺-NQR with and without bound inhibitors for model building is accessible in PDB under accession number 7XK3 (oxidized form of Na⁺-NQR: WT_ox⁷), 7XK6 (oxidized form of Na⁺-NQR with bound AD-42: WT_ox + AD-42⁷), and 7XK7 (oxidized form of Na⁺-NQR with bound KR: WT_ox + KR⁷). The data that support the findings of this study are available from the corresponding author upon request.

Code availability

The MD-simulation files are available on GitHub at [https://github.com/OkazakiLab/nqr_mdfiles] and are also available on Zenodo under [10.5281/zenodo.18333351]⁶¹.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-69182-w.