Cork-in-bottle mechanism of inhibitor binding to mammalian complex I

Abstract

Mitochondrial complex I (NADH:ubiquinone oxidoreductase), a major contributor of free energy for oxidative phosphorylation, is increasingly recognized as a promising drug target for ischemia-reperfusion injury, metabolic disorders, and various cancers. Several pharmacologically relevant but structurally unrelated small molecules have been identified as specific complex I inhibitors, but their modes of action remain unclear. Here, we present a 3.0-Å resolution cryo-electron microscopy structure of mammalian complex I inhibited by a derivative of IACS-010759, which is currently in clinical development against cancers reliant on oxidative phosphorylation, revealing its unique cork-in-bottle mechanism of inhibition. We combine structural and kinetic analyses to deconvolute cross-species differences in inhibition and identify the structural motif of a "chain" of aromatic rings as a characteristic that promotes inhibition. Our findings provide insights into the importance of π-stacking residues for inhibitor binding in the long substrate-binding channel in complex I and a guide for future biorational drug design.

Fig. 1. IACS-2858 is a stronger inhibitor of mouse complex I than IACS-010759.

(A) Chemical structures of IACS-010759 and IACS-2858. (B) Experimentally measured NADH:decylubiquinone (dQ) oxidoreduction rates of purified mouse complex I (average ± SEM), technical replicates (n ≥ 3) plotted against a titration of IACS-010759 (red) or IACS-2858 (black) concentrations. Hill slopes were constrained to −1. IC₅₀ values are reported with the SE.

Fig. 2. IACS-2858 bound in the structure of mouse complex I.

(A) A single IACS-2858 molecule (blue) binds at the entrance of the ubiquinone-binding site. The iron-sulfur clusters (red and yellow) and the ubiquinone-binding site are indicated, and charged residues in the proton-transfer domain are marked in purple. MM, mitochondrial matrix; IMM, inner mitochondrial membrane; IMS, intermembrane space. (B) Electrostatic potential density of IACS-2858, presented from two viewpoints and plotted using UCSF ChimeraX with contour level 0.05. (C) IACS-2858 (blue) in the ubiquinone-binding cavity formed by the surrounding subunits (left) and the expected position of ubiquinone-10 (magenta; alternating isoprenoid units in purple) (17) modeled into the cavity (right). The interior surface cavity in green was identified using CASTp (57).

Fig. 3. IACS-2858 interacts with residues at the entrance of the ubiquinone-binding site.

(A) Hydrogen bonding (blue dashed line) and π-stacking (yellow dashed line) and polar interactions (purple dashed line) made between IACS-2858 (sky blue) and residues (white) surrounding the (a) sulfonylmethyl (black box), (b) piperidine (blue box), (c) phenyl (green box), and (d) 2-pyridone (red box) moieties of IACS-2858. Insets a to d have been marginally rotated to give the clearest field of view. (B) Van der Waal’s and hydrophobic contacts made by residues within 3.5 Å of IACS-2858. Residues from ND1, NDUFS2, and NDUFS7 are labeled in black, yellow-orange, and pink, respectively. PC, phosphatidylcholine.

Fig. 4. IACS-2858 binds to mammalian, yeast, and bacterial complexes I with different affinities.

Experimentally measured NADH oxidation rates (average ± SEM, n ≥ 3) in (A) membranes or (B) purified complex I plotted against inhibitor concentration. IC₅₀ values for each inhibitor and species pair are indicated on the right with the SE. Hill slopes were either fixed to −1 (solid lines) or not constrained (dashed lines). (C) An overlay of the structures of wild-type Y. lipolytica respiratory complex I (PDB: 6YJ4) (25) and the IACS-2858–bound mouse complex I showing residues within 3.5 Å of IACS-2858 that are not conserved. (D) Details of an overlay of the IACS-2858–bound complex I model with a hypothetical model for P. denitrificans complex I, created by mutating the mouse model in PyMOL. a to d show specific regions where differences in residues affect key interactions or result in likely steric clashes. Inhibitor interactions determined in the IACS-2858–inhibited model are indicated as in Fig. 3. The models (except for M. musculus in white) are colored by species as in (A) and (B). Residues are labeled for M. musculus and mutations to them noted in color accordingly.

Fig. 5. IACS-010759 and IACS-2858 inhibit RET more potently than FET.

Experimentally measured FET (black) and RET (blue) rates (NADH oxidation rate and succinate oxidation–driven NAD⁺ reduction rates, respectively) of bovine SMPs (average ± SEM, n ≥ 3) are plotted against a titration of IACS-010759 (left) or IACS-2858 (right) concentrations. IC₅₀ values are reported with the SE.

Fig. 6. Competitive model for inhibition of complex I in proteoliposomes by IACS-2858.

(A) Scheme for a simple competitive mode of inhibition. Rapid reoxidation of ubiquinol by AOX (k₄) prevents appreciable levels accumulating. Experimentally measured rates (average ± SEM, n ≥ 3) are shown in (B) K_M and (C) IC₅₀ plots, alongside the best-fit predictions from the models (see table S3 for parameters). Plots of trends in (D) K_M,app, (E) V_max,app, (F) IC_50,app, and (G) Hill slope were produced by using the Michaelis-Menten equation or the standard dose-effect relationship (see Materials and Methods) to fit the individual datasets shown in (B) and (C). Values from the experimental data are given as black points (error bars are 95% confidence intervals), and values from the output data from the models are shown as black lines.