Systematic Identification of MCU Modulators by Orthogonal Interspecies Chemical Screening

Abstract

The mitochondrial calcium uniporter complex is essential for calcium (Ca²⁺) uptake into mitochondria of all mammalian tissues, where it regulates bioenergetics, cell death, and Ca²⁺ signal transduction. Despite its involvement in several human diseases, we currently lack pharmacological agents for targeting uniporter activity. Here we introduce a high-throughput assay that selects for human MCU-specific small-molecule modulators in primary drug screens. Using isolated yeast mitochondria, reconstituted with human MCU, its essential regulator EMRE, and aequorin, and exploiting a D-lactate- and mannitol/sucrose-based bioenergetic shunt that greatly minimizes false-positive hits, we identify mitoxantrone out of more than 600 clinically approved drugs as a direct selective inhibitor of human MCU. We validate mitoxantrone in orthogonal mammalian cell-based assays, demonstrating that our screening approach is an effective and robust tool for MCU-specific drug discovery and, more generally, for the identification of compounds that target mitochondrial functions.

Keywords: MCU; bioenergetics; calcium; calcium signaling; drug discovery; drug screening; high-throughput screening; mitochondria; mitochondrial calcium uniporter.

Figure 1. A Yeast Bioenergetic Shunt to Reduce False-Positive Hits in MCU-Targeted Drug Discovery Screens

(A) Schematic representation of yeast mitochondrial energy production pathways in media with either mannitol/sucrose (MAS) and D-lactate or KCl and succinate. II, succinate dehydrogenase; III, coenzyme Q:cytochrome c-oxidoreductase; Q, coenzyme Q; DLD, D-lactate:cytochrome c oxidore-ductase; Cytc, cytochrome c; IV, cytochrome c oxidase. (B) Immunoblot analysis of cytoplasmic (Cyt) and mitochondrial (Mito) fractions isolated from yeast cells expressing human MCU, EMRE, and mt-AEQ. YME1, mitochondrial i-AAA protease; PKGa, protein kinase G alpha. (C) Representative traces of Ca²⁺-dependent, AEQ-based light kinetics in yeast mitochondria treated with either 0.1% DMSO or 10 μM Ru360 in the presence or absence of 15 μM ETH129 (ETH). RLU, relative luminescence units. Mean ± SEM; n = 4. (D) Effect of respiratory chain inhibitors (malonate, 10 mM; antimycin A, 1 μM; and KCN, 6 mM) and CCCP (6 μM) on Ca²⁺ uptake kinetics in yeast mitochondria energized with either succinate or D-lactate (10 mM) in MAS or KCl-based media. Averaged light kinetics (n = 3, brown) are quantified as ratio of light signal (L_drug) over maximal peak luminescence (Lmax_drug), normalized to DMSO (0.1%). (E) Quantification of Ca²⁺-dependent, AEQ-based light kinetics shown in (D). Mean ± SEM; n = 3; ***p < 0.001, one-way ANOVA.

Figure 2. Drug Discovery Screen in Reconstituted Yeast Mito-chondria

(A) General workflow of the yeast mitochondria-based drug discovery screen. CLZN, native coelenterazine; I_drug, inhibition score. (B) Drug screen in biological replicates (R1 and R2). Ca²⁺-dependent, AEQ-based light kinetics from reconstituted yeast mitochondria are shown for each compound, whereas averaged light kinetics are shown for positive (Ru360, n = 160) and negative (DMSO, n = 160) controls. (C) Reproducibility of the drug screen. Linear regression (solid orange line) is fitted to the inhibition score (I_drug) of each compound (dot). I_drug scores that deviate from the linear regression (gray dots) by two root mean square errors (dotted orange lines) are considered outliers. The correlation coefficient (R²) refers to data points upon the removal of outliers. (D) Correlation of I_drug scores for peak and uptake rate. (E) Performance of the screen based on Z′-factors. Highest and lowest values indicate third and first quartiles, while the thick line represents the median. Z′-factors that do not replicate between biological duplicates are shown as small circles. Median values > 0.5 indicate screening assays suitable for HTS. (F) Ranking of compounds based on I_drug scores.

Figure 3. Orthogonal Drug Screens in Permeabilized HeLa Cells and Reconstituted Yeast Mitochondria Validate Mitoxantrone as a Specific MCU Inhibitor

(A) General workflow of the permeabilized HeLa cell-based assay. CLZ n, coelenterazine derivative n; EC, extracellular-like solution; Tg, thapsigargin; Dig, digitonin; PM, plasma membrane; PMCCs, plasma membrane Ca²⁺ channels; SERCA, sarco/endoplasmic reticulum Ca²⁺-ATPase; IP3R, inositol trisphosphate receptor. See also Figures S1A–S1C. (B) Ca²⁺-dependent, AEQ-based light kinetics in digitonin-permeabilized HeLa cells upon the addition of Ca²⁺ in the presence and absence of Ru360 (5 μM), CGP-37157 (20 μM), or DMSO (0.2%). Mean ± SEM; n = 8. See also Figures S1D and S1E. (C) Performance of the drug screen based on Z′-factors. (D) Reproducibility of the drug screen. (E) Ranking of compounds based on I_drug scores. (F) Distribution of candidate drugs in therapeutic classes. (G) Validation of true-positive hits by orthogonal, inter-species screens. ETCIs, electron transport chain inhibitors. See also Figure S2. (H) Ca²⁺-dependent, AEQ-based light kinetics in yeast mitochondria treated with amiodarone. Mean ± SEM; n = 3. (I) Effect of amiodarone on Ca²⁺-dependent, AEQ-based light kinetics in yeast mitochondria pre-treated with either Ru360 or DMSO. Mean ± SEM; n = 3. (J) Dose response of mitoxantrone on Ca²⁺ uptake relative to DMSO and fitted with a Hill equation (continuous lines) to extract Michaelis constant (k_0.5) and Hill coefficient (n_H). Inset: mitoxantrone hydrochloride chemical structure. Mean ± SEM; n = 4. See also Figure S3.

Figure 4. MCU-Mediated Ca²⁺ Currents and Mitochondrial Bioenergetics in Response to Mitoxantrone

(A) Representative recordings of MCU current densities (I/C_m) in HEK293T cell mitoplasts in various bath [Ca²⁺]. (B) Representative time course of MCU current during exposure to and washout of 10 μM mitoxantrone in bath solution. Each point represents the amplitude of MCU current at −160 mV, sampled every 5 s. RuR, ruthenium red (200 nM). (C) Quantification of MCU Ca²⁺ current density (pA/pF) in 1 mM bath Ca²⁺ before (n = 5) and after (n = 4) the addition of 10 μM mitoxantrone. Mean ± SEM. (D) Quantification of time constant for inhibition (τ_on; n = 4) and recovery (τ_off; n = 3) of MCU Ca²⁺ current. Mean ± SEM. (E) Oxygen consumption rate (OCR) upon acute treatment of crude mouse liver mitochondria with either mitoxantrone or CCCP and then antimycin A. Mean ± SEM; n = 3. (F) Respiratory control ratio (RCR) of crude mouse liver mitochondria energized with succinate/rotenone (10 mM; 2 μM) and treated with DMSO (n = 4), mitoxantrone (n = 5), or oligomycin A (n = 3). ADP (4 mM), AntA, Antimycin A (4 μM). Mean ± SEM. (G) Effect of mitoxantrone on maximal respiration and spare respiratory capacity of permeabilized HeLa cells. Mean ± SEM; n = 3.

Figure 5. Effect of Mitoxantrone on Intracellular and Plasma Membrane Ion Channels

(A) Ca²⁺-dependent, AEQ-based light kinetics upon stimulation of intact HeLa cells with 100 μM histamine and after treatment with either vehicle (DMSO 0.2%) or different concentrations of mitoxantrone for 1 or 2 hr. (B and C) Quantification of peak luminescence (B) and rate of light emission (uptake rate) (C) for light kinetics in (A) after normalization to number of viable cells. Mean ± SEM; *p < 0.05, **p < 0.01, and ***p < 0.001, one-way ANOVA; n = 12. See also Figure S4. (D) Representative traces and quantification of mt-Ca²⁺ concentrations in HeLa cells stimulated with histamine (100 μM) in response to mitoxantrone (20 μM) or DMSO (0.2%) treatment for 2 hr. Mean ± SEM; n = 4; **p < 0.01, t test. Light emission was calibrated using the constant values previously published (Montero et al., 2000). (E) Representative traces and quantification of ER-Ca²⁺ concentrations in HeLa cells stimulated with histamine (100 μM) in the presence of 1 mM CaCl₂ and pre-treated with either 20 μM mitoxantrone (n = 4) or 0.2% DMSO (n = 3) for 2 hr. Mean ± SEM. (F) Representative traces and quantification of the amplitude and rate of released ER-Ca²⁺ upon stimulation of digitonin-permeabilized HeLa cells with IP₃ (0.2 μM) after mitoxantrone (20 μM) or DMSO (0.2%) treatment. Mean ± SEM; n = 3. (G) Effect of mitoxantrone on voltage-activated Ca²⁺ currents in non-differentiated SH-SY5Y cells. Mean ± SEM; n = 3. (H) Representative traces of store-operated Ca²⁺currents in Xenopus oocytes treated with mitoxantrone. (I) Time course of peak inward Ca²⁺ currents shown in (H) at −120 mV, normalized to maximum value. Mean ± SEM; n = 3. (J) Representative traces of Ca²⁺-activated Cl⁻ currents in Xenopus oocytes treated with mitoxantrone. (K) Time course of normalized peak Ca²⁺-activated Cl⁻ currents shown in (J). Mean ± SEM; n = 3. (L) Effect of mitoxantrone on voltage-activated Na⁺ and K⁺ currents in single mouse type II taste cells. (M) I–V relations of inward Na⁺ currents before (control), during exposure, and washout of the drug (Mitoxa, 10 μM). Mean ± SEM; n = 3. (N) I–V relations of outward K⁺ currents before (control), during (Mitoxa, 10 μM), and after exposure to mitoxantrone. Mean ± SEM; n = 3.

Figure 6. Structure-Activity Relationship Analysis of Mitoxantrone

Dose-response curves for anthracyclines (top) and structural analogs of mitoxantrone (bottom) in yeast mitochondria. The mt-Ca²⁺ uptake rates (μM.s⁻¹) relative to DMSO are fitted with a Hill equation (continuous lines) to extract the Michaelis constant (k_0.5). The quinizarin core is highlighted in blue. Mean ± SEM; n = 4. See also Figures S5 and S6.

Figure 7. Validation of Mitoxantrone-MCU Direct Interaction

(A) Representative recordings of MCU Ca²⁺ current densities (I/C_m) in HEK293T cell mitoplasts treated with 10 μM mitoxantrone in the pipette (matrix) solution. (B) Representative time course of MCU Ca²⁺ currents in the presence of mitoxantrone in the matrix and upon its addition to the bath solution. (C) Quantification of MCU current densities (pA/pF) in 1 mM bath Ca²⁺ after the addition of 10 μM mitoxantrone to the matrix (pipette; n = 6) and to the bath solution (n = 3). Mean ± SEM. (D) Average time constants for the inhibition (τ_on) of MCU current densities by the addition of mitoxantrone in the bath (n = 5) in the presence or absence of mitoxantrone in the matrix (pipette) (n = 4). (E) Flexible molecular docking analysis of C. elegans MCU pore domain and mitoxantrone. (F) Magnification of (E). (G) Prediction of molecular interactions between mitoxantrone and the MCU selectivity filter. (H–J) Dose-response curves for yeast mitochondria reconstituted with either wild-type human MCU or a D261A mutant and treated with RuR (H), mitoxantrone (I), or pixantrone (J). The mt-Ca²⁺ peak values (μM) relative to DMSO are fitted with a Hill equation (continuous lines) to extract the Michaelis constant (k_0.5). Mean ± SEM; n = 4. See also Figure S7.