Key features of inhibitor binding to the human mitochondrial pyruvate carrier hetero-dimer

Abstract

Objective: The mitochondrial pyruvate carrier (MPC) has emerged as a promising drug target for metabolic disorders, including non-alcoholic steatohepatitis and diabetes, metabolically dependent cancers and neurodegenerative diseases. A range of structurally diverse small molecule inhibitors have been proposed, but the nature of their interaction with MPC is not understood, and the composition of the functional human MPC is still debated. The goal of this study was to characterise the human MPC protein in vitro, to understand the chemical features that determine binding of structurally diverse inhibitors and to develop novel higher affinity ones.

Methods: We recombinantly expressed and purified human MPC hetero-complexes and studied their composition, transport and inhibitor binding properties by establishing in vitro transport assays, high throughput thermostability shift assays and pharmacophore modeling.

Results: We determined that the functional unit of human MPC is a hetero-dimer. We compared all different classes of MPC inhibitors to find that three closely arranged hydrogen bond acceptors followed by an aromatic ring are shared characteristics of all inhibitors and represent the minimal requirement for high potency. We also demonstrated that high affinity binding is not attributed to covalent bond formation with MPC cysteines, as previously proposed. Following the basic pharmacophore properties, we identified 14 new inhibitors of MPC, one outperforming compound UK5099 by tenfold. Two are the commonly prescribed drugs entacapone and nitrofurantoin, suggesting an off-target mechanism associated with their adverse effects.

Conclusions: This work defines the composition of human MPC and the essential MPC inhibitor characteristics. In combination with the functional assays we describe, this new understanding will accelerate the development of clinically relevant MPC modulators.

Keywords: Inhibition; Mitochondria; Mitochondrial pyruvate carrier; Mitochondrial transport; Small molecules.

Graphical abstract

Figure 1

Pyruvate transport and ligand binding by human MPC. (A) Νickel affinity-purified MPC1L/MPC2 hetero-dimer (left), MPC1/MPC2 hetero-dimer (middle) and MPC2 protomer (right) analysed by SDS-PAGE and visualised by Coomassie Blue stain. MPC proteins are indicated by asterisks. (B) Time course of pyruvate homo-exchange at a ΔpH of 1.6 for the MPC1L/MPC2 hetero-dimer reconstituted into liposomes. (C-D) CPM thermostability shift assays for MPC hetero-dimers and individual MPC2 in the presence of absence of 100 μM MPC inhibitors. In (B), data represent the mean ± s.d. of six biological repeats, each performed with two technical replicates. In (C–D), data are representative of two biological repeats, each performed with three or two technical replicates, respectively.

Figure 2

Comparative analyses of ligand binding and pyruvate transport inhibition by structurally distinct small molecules. (A) Chemical structures of previously claimed MPC inhibitors. Color coding is consistent across all panels. (B) The compounds listed in (A) were tested for inhibition of pyruvate transport on MPC1L/MPC2 proteoliposomes at 100 μM, and the remaining transport activity was compared to that in their absence (CT). (C) Inhibition of [¹⁴C]-pyruvate homo-exchange by UK5099 (10–10,000 nM), zaprinast (10–10,000 nM), lonidamine (250–20,000 nM) and mitoglitazone (250–20,000 nM). (D) First derivative of MPC1L/MPC2 unfolding profiles with or without compounds (100 μM) via the CPM thermostability assay. (E) First derivative of MPC1L/MPC2 unfolding profiles (330 nm/350 nm ratio) with or without compounds (100 μM) via nano-DSF. In (B), data points represent the mean ± s.d. of two biological repeats performed in triplicate. In (C), data points represent the mean ± s.d of four (UK5099) or three (zaprinast, lonidamine, mitoglitazone) biological repeats, each performed in triplicate. Results in (D–E) represent characteristic experiments repeated independently. Averaged data from different biological repeats are summarised in Table S1.

Figure 3

Chemical evolution of MPC inhibitors. (A) Chemical structures selected using the three hydrogen bond acceptors of UK5099. The template used for virtual screening is highlighted in a grey box. The chemical groups added or changed in each compound are highlighted in yellow. (B) Chemical structures of compounds where the cyanide and carboxylic acid groups have been replaced in the background of the structure of compound 7. The template used for virtual screening is highlighted in a grey box. The chemical groups added or changed in each compound are highlighted in yellow. (C) First derivatives obtained from CPM thermostability shift assays for the compounds in (A) and (B). (D) The Tm values, calculated from the first derivative in thermostability shift assays, were plotted in a bar graph. The dotted line indicates the melting temperature of unliganded MPC. Color coding is as in (C). (E) Inhibition of [¹⁴C]-pyruvate homo-exchange by UK5099 (10–10,000 nM), compounds 2 (2.5–1,000 nM) and 7 (1–500 nM). In (C), data are representative of three biological repeats. Data in (D) have been calculated from three independent biological repeats (values included in Table S1). In (E), data points represent the mean ± s.d of three biological repeats, each performed in triplicate. The averaged IC₅₀-values were 52.6 ± 8.3 nM, 13.5 ± 4.1 nM and 5.4 ± 1.1 nM for UK5099, compound 2 and 7, respectively.

Figure 4

MPC cysteines do not form covalent interactions with inhibitors. (A) Cysteine-to-alanine replacement mutants purified in detergent in parallel with wild type (wt) human MPC1L/MPC2 via nickel affinity chromatography. (B) First derivative of protein unfolding profiles for wild type and cysteine-to-alanine replacement mutants via CPM thermostability assay. (C) CPM thermostability for wild-type and cysteine-to-alanine replacement mutants in the absence or presence of zaprinast, and mitoglitazone, which do not feature the activated double bond, or CHC, UK5099 and compound 7, each containing an activated double bond. In (B) data are representative of two biological repeats and three technical repeats. In (C), data represent the mean ± s.d of two biological repeats and three technical repeats.

Figure 5

Matching of an extended pharmacophore model predicts MPC inhibition. (A) Thermostability assays for nitrofurantoin and entacapone using nano-DSF. (B) Inhibition of [¹⁴C]-pyruvate homo-exchange by entacapone (100–10,000 nM) and nitrofurantoin (500–20,000 nM). (C) Pharmacophore features of compound 7 and comparison with UK5099. Red spheres indicate hydrogen acceptor features, yellow spheres indicate hydrophobic ring features, red star indicate negative ionizable features and bleu tori indicate aromatic ring features. (D) Fitting of entacapone in the extended pharmacophore. (E) Fitting of nitrofurantoin in the extended pharmacophore. (F-G) 3D and graphical representation of the extended pharmacophore with the identified features (Roman numberals) and their presence in the various compounds. The pharmacophore features are as follows: I and VII; aromatic ring (blue boxes), II and VIII; hydrophobic ring (yellow boxes), III, IV, V, IX and X; hydrogen bond acceptors (maroon boxes). In (A-B), data are representative of two biological repeats, each performed in triplicate.