A High-Throughput Screening Identifies MICU1 Targeting Compounds.

Mitochondrial Ca2+ uptake depends on the mitochondrial calcium uniporter (MCU) complex, a highly selective channel of the inner mitochondrial membrane (IMM). Here, we screen a library of 44,000 non-proprietary compounds for their ability to modulate mitochondrial Ca2+ uptake. Two of them, named MCU-i4 and MCU-i11, are confirmed to reliably decrease mitochondrial Ca2+ influx. Docking simulations reveal that these molecules directly bind a specific cleft in MICU1, a key element of the MCU complex that controls channel gating. Accordingly, in MICU1-silenced or deleted cells, the inhibitory effect of the two compounds is lost. Moreover, MCU-i4 and MCU-i11 fail to inhibit mitochondrial Ca2+ uptake in cells expressing a MICU1 mutated in the critical amino acids that forge the predicted binding cleft. Finally, these compounds are tested ex vivo, revealing a primary role for mitochondrial Ca2+ uptake in muscle growth. Overall, MCU-i4 and MCU-i11 represent leading molecules for the development of MICU1-targeting drugs.

Introduction

Mitochondrial Ca2+ uptake plays multiple roles impinging on cell homeostasis. Specifically, Ca2+ clearance by mitochondria contributes to the maintenance of physiological cytosolic [Ca2+] required for Ca2+-dependent functions. Upon increased energy demand, mitochondrial Ca2+ uptake positively regulates the activity of three tricarboxylic acid (TCA) cycle enzymes. However, in specific pathological conditions, excessive mitochondrial Ca2+ accumulation occurs, triggering the opening of the mitochondrial permeability transition pore (mPTP) and the release of pro-apoptotic cofactors (Rizzuto et al., 2012).

Mitochondrial Ca2+ uptake occurs in response to physiological stimuli, which trigger the release of Ca2+ from intracellular stores, mainly the endoplasmic reticulum (ER) and sarcoplasmic reticulum (SR). The proximity of the ER to mitochondria in specific sites of close juxtaposition permits the generation of high [Ca2+] microdomains at the mouth of Ca2+ entry (Rizzuto et al., 1993). Mitochondrial Ca2+ uptake relies on the activity of a highly selective channel, the mitochondrial calcium uniporter (MCU) (Baughman et al., 2011, De Stefani et al., 2011), composed of pore-forming subunits (i.e., MCU, MCUb, and Essential MCU Regulator [EMRE]) and regulatory proteins (i.e., MICU1, MICU2, and MICU3) (De Stefani et al., 2016). Several studies have reported the structure of the MCU alone (Baradaran et al., 2018, Fan et al., 2018, Nguyen et al., 2018, Yoo et al., 2018) or in combination with EMRE (Wang et al., 2019), disclosing the tetrameric structure of the MCU channel and the stoichiometry of MCU subunits to EMRE as one to one.

The rate of mitochondrial Ca2+ entry follows a sigmoidal curve depending on cytosolic [Ca2+]. Accordingly, Ca2+ entry is slow at low cytosolic [Ca2+], although it becomes rapid when cytosolic [Ca2+] increases. This behavior has a dual function. On one hand, at low cytosolic [Ca2+], it prevents continuous and sustained Ca2+ entry that eventually would cause Ca2+ overload. On the other, it warrants prompt Ca2+ uptake in response to agonist-induced Ca2+-store release stimuli. Controlled mitochondrial Ca2+ entry is ensured by the activity of the MICU family of MCU interactors. Although MICU3 expression is mainly confined to the brain (Patron et al., 2019), MICU1 and MICU2 are ubiquitously expressed, although with different stoichiometry in different tissues (Paillard et al., 2017). MICU1 and MICU2 form homo- or heterodimers according to their relative abundance, ensuring channel gating at low cytosolic [Ca2+]. In addition, at high cytosolic [Ca2+], MICU1 is a cooperative activator of MCU in many systems (Csordás et al., 2013, Mallilankaraman et al., 2012, Perocchi et al., 2010). The crystal structures of MICU1 (Wang et al., 2014), of MICU2 (Kamer et al., 2019, Xing et al., 2019), and of MICU3 (Xing et al., 2019) allowed the identification of common and distinct motifs and permitted to propose models for the gating mechanism of MICU protein family.

The identification of genetic variants linked to diseases together with the analysis of experimental models have uncovered the pathophysiological role of the MCU complex. These studies underline the importance of mitochondrial Ca2+ accumulation for organ physiology and the deleterious consequences of its dysregulation (Mammucari et al., 2018) and highlight the need for pharmacological targeting of the MCU complex. Among the druggable targets, the MCU complex regulators are likely the most promising due to their involvement in known diseases and to their rapid turnover. In particular, the essential role of MICU1 is demonstrated by the severe phenotype of MICU1−/− mouse models (Antony et al., 2016, Liu et al., 2016) and the diseases affecting patients carrying loss-of-function mutations in its coding sequence. In detail, MICU1 mutations have been reported in patients affected by proximal myopathy, learning difficulties, and extrapyramidal movement disorder (Logan et al., 2014). Other patients carrying a homozygous deletion of MICU1 exon 1 experience fatigue, lethargy, and weakness (Lewis-Smith et al., 2016). Finally, a third founder mutation predicted to cause loss of function of MICU1 was identified in additional families (Musa et al., 2019). Concerning MICU2, a homozygous truncating mutation was found to fully segregate with a severe neurodevelopmental disorder affecting consanguineous patients (Shamseldin et al., 2017).

Thus, the identification of lead compounds targeting the MCU complex is of utmost importance for the pharmacological modulation of this channel and for the study of the MCU function in cells and tissues in which genetic modulation is challenging. To this aim, cell-permeable MCU inhibitors have been recently discovered, either by screening small-molecule compounds libraries or by synthesis of Ru360 derivatives. All of them decrease mitochondrial Ca2+ uptake without affecting cytosolic Ca2+ dynamics or mitochondrial membrane potential. The antineoplastic compound mitoxantrone is a direct inhibitor of MCU, independently of its antineoplastic properties (Arduino et al., 2017). DS16570511 was selected from a small-molecule library for its ability of reducing mitochondrial Ca2+ influx in intact cells and in isolated mitochondria and to inhibit mitochondrial Ca2+ overload in isolated perfused heart (Kon et al., 2017). Finally, a ruthenium complex that targets the MCU at the N-terminal domain, Ru265, was synthetized and shown to be cell permeable and more potent than the widely used Ru360 (Woods et al., 2019). Nevertheless, pharmacological targeting of the MCU complex regulators is still missing.

Here, we report the screening of a 44,000 non-proprietary compound library and subsequent hit validation. From this screen, we selected two molecules that decreased mitochondrial Ca2+ uptake, and accordingly, we named them MCU-i4 and MCU-i11, respectively. Direct binding of these molecules to MICU1 was essential for their activity. Both compounds reduced mitochondrial Ca2+ uptake in isolated myofibers from mouse flexor digitorum brevis (FDB) muscles and, in agreement with the role of MCU complex in skeletal muscle trophism (Debattisti et al., 2019, Gherardi et al., 2019, Mammucari et al., 2015), prevented myotube growth.

Thus, these molecules represent the first lead MICU1-targeting compounds for experimental and disease conditions in which a reduction in mitochondrial Ca2+ uptake is desirable.

Results

A High-Throughput Screening Identifies MCU Modulators

To identify active compounds able to modulate MCU activity, we set up a high-throughput screening assay based on aequorin, a holoenzyme that emits photons upon Ca2+ binding due to an irreversible reaction. Expression of recombinant aequorin in cells allows the precise measurement of [Ca2+] in specific compartments where aequorin is targeted (see STAR Methods for further details). Thus, by measuring light emission of HeLa cells transfected with mitochondria-targeted aequorin (mtAeq) after cellular stimulation with histamine, an InsP3-generating agonist, we were able to identify compounds that modulate mitochondrial Ca2+ uptake. Using this cell-based assay, we screened a library of 44,000 small-molecule compounds. The screening flow is depicted in Figure 1A and five criteria were used to select negative and positive modulators. (1) The decrease or increase in mitochondrial Ca2+ uptake in intact HeLa cells was to be more than 30% or 50%, respectively (Figure 1B). (2) The compounds should have no effect on cytosolic Ca2+ responses (to exclude molecules with an effect on intracellular stores). (3) The compounds should have no effect on mitochondrial membrane potential, because it represents the main driving force for organelle Ca2+ accumulation. (4) The compounds were to be effective also in permeabilized HeLa cells, to further exclude the contribution of mitochondria-independent events. (5) The activity was to be reliably observed also with newly resynthesized compounds.

Figure 1

A High-Throughput Screening Identifies MCU Modulators

(A) Screening flow chart to identify MCU modulators.

(B) Activity distribution of 44K compounds in the primary mitochondrial Ca2+ uptake screen. The aequorin fluorescence signals for each compound were normalized to the activity of an inhibition control (−100% full inhibition) and of a neutral control (stimulation by histamine; 0% inhibition), which were included in each 384-well screening plate.

(C) Summary table of primary and confirmation screen results.

(D) Chemical structure and name of the two selected compounds.

We found two compounds that were able to negatively modulate mitochondrial Ca2+ uptake meeting all of these criteria (Figures 1C and 1D), and we named these compounds MCU-i4 and MCU-i11 (i.e., mitochondrial calcium uptake inhibitors), respectively.

MCU-i4 and MCU-i11 Negatively Modulate MCU Activity in a Small-Scale Validation Assay

Although the high-throughput screening allowed a fast selection of a small number of promising hits based on luminescence data, further analyses were aimed at quantifying mitochondrial [Ca2+] upon treatment with the compounds (Figure 2). In detail, by means of a well-established calibration algorithm (Tosatto et al., 2017), we calculated [Ca2+] based on the light emission. In control cells, histamine-induced mitochondrial Ca2+ peak reached 50–70 μM. When cells were incubated with MCU-i4 and MCU-i11 soon before measurements, mitochondrial [Ca2+] peak was limited to about 30 μM (Figure 2A). Next, we wished to validate, by small-scale quantification, the luminescence data of cytosolic [Ca2+]. In histamine-stimulated aequorin-transfected cells, cytosolic Ca2+ peaks reached about 2.5 μM and the selected compounds had no effect compared to untreated cells (Figure 2B).

Figure 2

MCU-i4 and MCU-i11 Negatively Modulate MCU Activity in a Small-Scale Validation Assay

(A) Agonist-induced mitochondrial calcium uptake in intact HeLa cells. Cells were stimulated with 100 μM histamine and treated with 10 μM of each compound in 0.1% DMSO. Left: mean [Ca2+]mt peaks are shown. Right: representative traces of mitochondrial calcium uptake are shown.

(B) Agonist-induced cytosolic calcium transients in intact HeLa cells. Cells were stimulated with 100 μM histamine and treated with 10 μM of each compound. Left: mean cytosolic calcium peaks are shown. Right: representative traces of cytosolic calcium transients are shown.

(C) Mitochondrial calcium uptake in permeabilized HeLa cells. A buffer mimicking the cytosolic ionic composition (IB) supplemented with either 100 μM EGTA (IB/EGTA) or 3 μM [Ca2+] was used. HeLa cells were permeabilized by a 1-min perfusion with 100 μM digitonin (in IB/EGTA) during luminescence measurements. Left: mean mitochondrial [Ca2+] speed is shown. Right: representative traces of mitochondrial calcium uptake are shown.

(D) Mitochondrial calcium uptake in intact HeLa cells upon long-term compound incubation.

(E) Agonist-induced mitochondrial calcium uptake in intact MEFs. Cells were stimulated with 100 μM ATP and treated with 10 μM compound.

(F) Agonist-induced mitochondrial calcium uptake in MDA-MB-231 cells upon treatment with 10 μM of each compound. Cells were stimulated with 100 μM ATP.

(G) Agonist-induced mitochondrial calcium uptake in HEK293T cells upon treatment with 10 μM of each compound. Cells were stimulated with 100 μM ATP.

(H) Δψ measurements in HeLa cells upon treatment with 10 μM of each compound at different time points.

Data are presented as mean ± SD. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; one-way ANOVA except Kruskal-Wallis for (C).

See also Figure S1.

A further validation of the high-throughput screening regarded the rate of mitochondrial Ca2+ uptake in permeabilized cells. For this purpose, we measured mitochondrial Ca2+ uptake speed in cells permeabilized with digitonin and then perfused with 3 μM Ca2+ in the presence of the active molecules (Figure 2C). MCU-i4 and MCU-i11 decreased mitochondrial Ca2+ uptake speed. These results confirm that the selected compounds impinge directly on the MCU complex activity rather than on Ca2+ release from intracellular stores or on cytosolic [Ca2+].

The experiments conducted so far took into account only the effects of acute incubation with the specific molecules (1 min). We wished to determine the effects on mitochondrial Ca2+ uptake of longer incubation time. For this purpose, we measured mitochondrial Ca2+ uptake in intact HeLa cells treated with the selected molecules for 90 min (Figure 2D). MCU-i4 and MCU-i11 modulated mitochondrial Ca2+ uptake similarly to 1 min of incubation, i.e., exerting a negative modulation, indicating that the primary effect of these compounds is a sudden and direct modulation of MCU complex activity.

Next, we wished to determine whether the molecules were effective in different cell lines. To this aim, we measured mitochondrial Ca2+ uptake in mouse embryonic fibroblasts (MEFs), MDA-MB-231, and HEK293T cell lines (Figures 2E–2G) upon treatment with 10 μM compound. Both MCU-i4 and MCU-i11 reduced mitochondrial Ca2+ uptake in all tested cell lines, indicating the existence of a conserved mechanism.

Moreover, to confirm the high-throughput mitochondrial membrane potential (Δψ) measurements, we performed single-cell measurements in cells treated with the selected compounds. Although at short incubation time, MCU-i4 and MCU-i11 did not affect Δψ, at longer incubation time (i.e., 15 min), MCU-i4 exerted a negative effect on Δψ (Figure 2H).

Importantly, MCU-i4 and MCU-i11 had no effect on cell viability (Figures S1A and S1B). Finally, we assessed resting mitochondrial [Ca2+], detecting no difference in compound-treated cells versus controls (Figure S1C).

In conclusion, according to the above results, MCU-i4 and MCU-i11 are bona fide negative modulators of the MCU complex.

MICU1 Is Required for the Activity of MCU-i4 and MCU-i11

MICU family of MCU interactors comprises regulatory proteins that are responsible for the fine-tuned regulation of mitochondrial Ca2+ handling. In detail, MICU1 and MICU2 form a heterodimer that controls both the threshold and the cooperativity of mitochondrial Ca2+ uptake (De Stefani et al., 2016). Thus, we wondered whether MICU1 or MICU2 were essential for compound activity. Accordingly, we silenced either MICU1 or MICU2 in HeLa cells and we measured mitochondrial Ca2+ uptake upon histamine stimulation in the presence of either compound (Figure 3A). As previously reported (Csordás et al., 2013, Patron et al., 2014), silencing of MICU1 or MICU2 causes the loss of MCU gatekeeping, thus resulting in increased mitochondrial Ca2+ uptake when aequorin is used as Ca2+ probe. Most importantly, MICU1 silencing blunted the inhibitory effect of MCU-i4 and MCU-i11 on mitochondrial Ca2+ entry, although in MICU2-silenced cells, either compound was still able to inhibit mitochondrial Ca2+ uptake. These data indicate that MICU1 is required for MCU-i4 and MCU-i11 activity although MICU2 is dispensable. We further analyzed the role of MICU1 and demonstrated that HeLa cells infected with shMICU1 lentiviral particles (Marchi et al., 2019) were also resistant to inhibition of mitochondrial Ca2+ uptake (Figures 3B and S2A).

Figure 3

MICU1 Is Required for MCU-i4 and MCU-i11 Activity

(A) Agonist-induced mitochondrial calcium uptake peaks in intact HeLa cells in which MICU1 or MICU2 were transiently silenced. Cells were stimulated with 100 μM histamine and treated with 10 μM of either compound.

(B) Mitochondrial calcium uptake peaks in control or shMICU1 HeLa stable clones. Cell were treated with 10 μM compound and stimulated with 100 μM histamine.

(C) Agonist-induced mitochondrial calcium uptake peaks in intact Micu1−/− MEFs upon expression of Micu1 or mock plasmids. Cells were treated with 10 μM compound and stimulated with 100 μM ATP.

(D) Agonist-induced mitochondrial calcium uptake peaks in intact MICU1−/− HEK293T cells upon expression of Micu1 or mock plasmids. Cells were treated with 10 μM compound and stimulated with 100 μM ATP.

Data are presented as mean ± SD. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; two-way ANOVA.

See also Figure S2.

Similarly to MICU1-silenced HeLa cells, both MCU-i4 and MCU-i11 had no effect on mitochondrial Ca2+ uptake in Micu1−/− MEFs (Antony et al., 2016; Figures 3C and S2B) and MICU1−/− HEK293T cells (Sancak et al., 2013; Figures 3D and S2C). Additionally, rescuing Micu1 expression in Micu1−/− MEFs and HEK293T cells restored the ability of MCU-i4 and MCU-i11 to inhibit mitochondrial Ca2+ uptake (Figures 3C, 3D, S2D, and S4B). We wondered whether MCU-i4 and MCU-i11 exerted their activity by interfering with Ca2+ binding to the regulatory EF-hand domains of Micu1. Interestingly, the compounds efficiently decreased mitochondrial Ca2+ uptake in HeLa cells and in Micu1−/− MEFs expressing a Micu1 EF-hand mutant (Micu1EFmut) (Patron et al., 2014; Figures S2E and S2F), indicating that they function independently of Ca2+ binding to the EF-hands.

Altogether, these data demonstrate that MICU1 is required for the activity of MCU-i4 and MCU-i11.

MCU-i4 and MCU-i11 Bind MICU1

We hypothesized that MCU-i4 and MCU-i11 may interact directly with MICU1. To verify this hypothesis, we took advantage of the surface plasmon resonance (SPR) technique. Purified recombinant Micu1, Micu2, and the dimeric Micu1-Micu2 proteins were immobilized on different channels of a SPR sensor chip surface. The compounds were injected at different concentrations over the chip surface, and the SPR signal was monitored. MCU-i4 and MCU-i11 showed a concentration-dependent interaction with Micu1 and Micu1-Micu2 complex reflected by an increase of SPR signal during the association phase (Figures S3A–S3D). Using an equilibrium fit of the extracted sensorgram report points, the following KD values were determined: (1) MCU-i4/Micu1 and MCU-i4/Micu1-Micu2 8.4 and 18 μM, respectively, and (2) MCU-i11/Micu1 and MCU-i11/Micu1-Micu2 2.9 and 2.7 μM, respectively. On the contrary, neither compound interacted specifically with Micu2 protein alone (not shown). In agreement with the calculated KDs, dose-response measurements indicated that MCU-i4 and MCU-i11 effectively inhibited mitochondrial Ca2+ uptake when used at 5 μM, although 1 μM was insufficient to exert the inhibitory effect (Figure S3E). These results indicated that MCU-i4 and MCU-i11 directly bind to MICU1, and their effect on mitochondrial Ca2+ uptake is due to this interaction.

We next examined the structural determinant of the interaction between the selected compounds and Micu1. For this purpose, a model of Micu1 was prepared from the available crystal structure in its calcium-free hexameric assembly (PDB: 4NSC). The docking simulations were performed on the monomer and in the hexameric assembly, considering that the putative binding sites could be located in each monomer but also at the interface between different monomers. Both simulations were in agreement in locating a suitable binding site in the cleft separating the N-lobe and the C-lobe of Micu1 (Figure S4A). Interestingly, this cavity is sufficiently large to accommodate both compounds, providing polar and hydrophobic residues and forming anchoring spots of a potential binding pocket. The most favorable poses of MCU-i4 and MCU-i11 according to the docking score (−79.20 and −87.56 kcal/mol, respectively) are reported in Figures 4A and 4B. MCU-i11 pose shows a good steric complementarity with the cavity and establishes a hydrogen bond between its carbonyl in position 4 of the tetrahydropyridothieno[2,3-d]pyrimidine moiety and the side chain of Gln306. Also, Gln302 sidechain is faced with the same carbonyl group and may further stabilize the tricyclic scaffold. In addition, the benzyl moiety is accommodated in a hydrophobic pocket surrounded by Leu443, Met444, and Pro221. The phenacetin portion is partially exposed to the solvent and oriented toward Arg223 and Pro221, which is involved in hydrophobic contact with phenacetin aromatic moiety. Interestingly, MCU-i4 besides the different scaffold establishes a similar hydrogen bond between Gln306 and the ethyl ester group. Again, the Gln302 is closely oriented, with the carbonyl moiety further stabilizing the ester. The hydrophobic pocket formed by Leu 443, Met444, and Pro221 here is occupied by the N,N-diethylaniline group, which extends the ethyl substituents until Leu310 and Phe370, establishing further hydrophobic contacts. Lastly, the quinoline portion interacts with Pro221 and Met444.

Figure 4

MCU-i4 and MCU-i11 Bind MICU1

(A and B) Predicted binding mode of MCU-i4 (A) and MCU-i11 (B). The pose is reported in the upper representation: the protein surface, as well as the ribbon, is reported in gray color. The ligands MCU-i4 and MCU-i11 are reported in magenta and cyan, respectively. The atomic distances of crucial interactions are reported in green. The three Micu1 residues selected for mutagenesis are shown in gray. Below each pose, the ligand interaction diagram reports the residues forming the pocket and the interactions observed.

(C) Agonist-induced mitochondrial calcium uptake peaks in intact Micu1−/− MEFs transfected with Micu1 or Micu1Q302A, Q306A, L443A. Cells were stimulated with 100 μM ATP and treated with 10 μM of MCU-i4.

(D) Agonist-induced mitochondrial calcium uptake peaks in intact Micu1−/− MEFs transfected with Micu1 or Micu1Q302A, Q306A, L443A. Cells were stimulated with 100 μM ATP and treated with 10 μM of MCU-i11.

Data are presented as mean ± SD. ∗∗p < 0.01; ∗∗∗p < 0.001; two-way ANOVA.

See also Figures S3 and S4 and Table S1.

In order to validate the prediction obtained by in silico studies, three potential key residues for binding were mutated: Gln302; Gln306; and Leu443. Gln306 mediated a hydrogen bond formation with both compounds. Gln302 did not clearly contribute to an interaction with the low-molecular-weight ligands in the in silico model, but its close proximity to Gln306 could eventually rescue such interaction and surely has an electrostatic role. Leu443 is a relevant residue in the hydrophobic pocket for both compounds. These findings were further verified by an in silico alanine scan mutating each amino acid by an alanine residue in the potential Micu1 binding pocket. The effect of alanine mutations was monitored by predicting their influence on the ligand binding (Table S1). We performed mitochondrial Ca2+ measurements in Micu1−/− MEFs expressing a Micu1 protein in which the three key residues in the predicted binding pocket were mutated (Micu1Q302A, Q306A, L443A). As reported above, MCU-i4 and MCU-i11 significantly reduced mitochondrial Ca2+ uptake in Micu1−/− MEFs rescued with wild-type (WT) Micu1 (Figures 4C and 4D). Conversely, this effect was lost when Micu1−/− cells were rescued with Micu1Q302A, Q306A, L443A mutant (Figures 4C, 4D, and S4B). Importantly, these mutations did not interfere with Micu1 function. Indeed, reintroduction of either WT or mutant Micu1 in Micu1−/− MEFs restored normal resting mitochondrial [Ca2+], indicating that the mutations of residues located in the binding cleft had no major impact on the physiological function of MICU1 (Figure S4C).

MCU-i4 and MCU-i11 Reduce Mitochondrial Ca2+ Uptake in Skeletal Muscle Fibers and Impair Muscle Cell Growth

Finally, we wondered whether the negative MCU complex modulators exerted similar effects in an ex vivo system of skeletal muscle. To this aim, we measured mitochondrial Ca2+ uptake in flexor digitorum brevis (FDB) myofibers freshly isolated from adult mice and previously transfected in vivo with the mitochondria-targeted Ca2+ probe 4mt-GCaMP6f (Figure 5A). After having assessed unaltered resting mitochondrial [Ca2+] upon acute compound addition (Figure 5B), caffeine was added to the cultured myofibers to trigger Ca2+ release from the SR store. Both MCU-i4 and MCU-i11 inhibited mitochondrial Ca2+ uptake in myofibers (Figure 5C). Next, given that genetic inhibition of mitochondrial Ca2+ uptake has been shown to decrease muscle mass (Debattisti et al., 2019, Gherardi et al., 2019, Mammucari et al., 2015), we used these two compounds in a well-established in vitro model mimicking myotube growth. For this purpose, we treated C2C12 myotubes with MCU-i4 and MCU-i11, respectively. Both compounds decreased myotube width (Figure 5D). These data prove the efficacy of MCU-i4 and MCU-i11 ex vivo and indicate that pharmacological modulation of mitochondrial Ca2+ uptake interferes with an essential biological process.

Figure 5

MCU-i4 and MCU-i11 Reduce Mitochondrial Ca2+ Uptake in Skeletal Muscle Fibers and Impair Muscle Cell Growth

(A) Representative scheme of the experimental design.

(B) Resting mitochondrial Ca2+ levels of single isolated FDB fibers treated with either compound.

(C) Mitochondrial Ca2+ uptake in single isolated FDB fibers transfected with 4mtGCaMP6f. Fibers were treated with 10 μM of MCU-i4 and MCU-i11, respectively. 6 min later, cells were stimulated with 40 mM caffeine. Left: mean mt Ca2+ peaks are shown. Right: representative traces of mitochondrial calcium uptake are shown.

(D) Left: representative scheme of the experimental design. Right: measurements of myotubes width upon compound treatment are shown.

Data are presented as mean ± SD. ∗p ˂ 0.05; ∗∗p < 0.01; ∗∗∗p ˂ 0.001; one-way ANOVA.

Discussion

We report here the identification of low-molecular-weight compounds targeting the MCU complex. To our knowledge, although a certain number of MCU-targeting ligands have been reported (Arduino et al., 2017, Kon et al., 2017, Woods et al., 2019), no negative modulators of mitochondrial Ca2+ uptake with MICU1-binding properties have been identified so far. Notably, the two negative modulators exert similar effects on different cell lines, both of human and of mouse origin, and on adult mouse myofibers. However, in light of its mild depolarizing effect, caution should be placed in the use of MCU-i4, especially in medium- and long-term experiments. The fact that MICU1 is the direct target of MCU-i4 and MCU-i11 is demonstrated both by functional and by structural evidence. Functionally, the two compounds require MICU1 to decrease mitochondrial Ca2+ uptake. Acute or long-term silencing of MICU1 as well as genetic MICU1 deletion blunt the compound effect. This is specific for MICU1, because MICU2 was dispensable for compound activity, as demonstrated by MICU2 silencing experiments. Structurally, MCU-i4 and MCU-i11 bound purified MICU1 and MICU1-MICU2 complex, but not MICU2 alone, with an affinity constant compatible with the compound concentration required for activity. Finally, docking prediction based on the MICU1 structure (Wang et al., 2014) allowed the identification of critical amino acids necessary for compound activity. The binding pocket resides in a highly conserved region spanning the N-lobe and the C-lobe. One of the two EF-hands, EF1, is located on the N-lobe sided by two helices (named NH3 and NH4), whose relative orientation is perturbed upon Ca2+ binding. It is interesting to note that the NH3 (residues 222–229) strictly connects EF1 with the compound binding site. Nonetheless, mutations of the Micu1 EF-hand domains did not impair compound activity. Whether the binding alters the conformational state of this specific MICU1 region, thus hampering the interaction of MICU1 with other complex components, is an intriguing possibility.

Previous studies have highlighted the multiple roles played by MICU1 at different [Ca2+]. Ca2+ binding to MICU1 EF-hand domains determines conformational modifications of the whole uniporter complex that translate into different responses (Patron et al., 2014). Thus, the outcomes of genetic MICU1 targeting are dependent on the [Ca2+] sensed by the complex. In addition, MICU1 silencing or deletion exerts different effects, depending on the cell type, most likely because of differences in complex composition, in the role played by MICU1 in determining complex stability, or in the peculiarity of the different shapes of Ca2+ increases triggered by agonist stimulation. MCU-i4 and MCU-i11 decrease agonist-induced InsP3-dependent mitochondrial Ca2+ uptake in a MICU1-dependent manner, but they do not alter resting mitochondrial [Ca2+], indicating that the gatekeeping role of MICU1-MICU2 complex is maintained. Thus, differently from known MCU inhibitors, MICU1 interactors allow a fine-tuning modulation of mitochondrial Ca2+ uptake.

We show here that treatment with MCU-i4 and MCU-i11 inhibits myotube growth, in line with the role of MCU complex on skeletal muscle mass and function (Debattisti et al., 2019, Gherardi et al., 2019, Mammucari et al., 2015, Zampieri et al., 2016). In various other different settings, these compounds may serve as tools to modulate mitochondrial Ca2+-entry-dependent events. Similarly to MCU deletion (Gherardi et al., 2019), negative modulators of Ca2+ uptake may modulate mitochondrial metabolism and substrate preference in the skeletal muscle. Mitochondrial Ca2+ overload may be prevented by compound treatment, thus offering additional strategies to study Ca2+ dependent cell death in disease states, including cancer, heart failure, muscular dystrophy, and amyotrophic lateral sclerosis (ALS). Pharmacological inhibition of MICU1 activity can be exploited to modulate cell motility (Mallilankaraman et al., 2012, Prudent et al., 2016, Tosatto et al., 2016, Xu and Chisholm, 2014). Finally, the Ca2+ buffering activity exerted by mitochondria in specific cell types, including neurons and pancreatic acinar cells, can be attenuated by compound treatment (Rizzuto et al., 2012).

STAR★Methods

Key Resources Table

Lead Contact and Materials Availability

All unique/stable reagents generated in this study are available from the Lead Contact without restriction. Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Cristina Mammucari (cristina.mammucari@unipd.it).

Experimental Model and Subject Details

Mice

All animal experiments were approved by the internal Animal Welfare Body and the Italian Ministry of Health and performed in accordance with the Italian law D. L.vo n_26/2014. Adult 2-4 month old CD1 male mice (Charles River) were housed in conventional cages regularly supplied with environmental enrichment, water bottles, and ad libitum food. Daily oversight and care was provided by specialized veterinarians, trained technicians, and animal husbandry staff.

Cell lines

For the high-throughput screening HeLa cells were maintained in growth medium consisting of MEM NEAA, no glutamine (Life Technologies) supplemented with 10% fetal bovine serum (FBS) (Life Technologies), 100 U/mL Penicillin-Streptomycin (Life Technologies) and 2 mM L-Glutamine (Life Technologies) at 37°C and 5% CO2 incubator.

For the small-scale validation assays, HeLa, MEFs, HEK293T and C2C12 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Life Technologies). MDA-MB-231 cells were cultured in DMEM-F12 (Life Technologies). Both media were supplemented with 10% FBS, 100 U/mL penicillin and 100 mg/mL streptomycin. Cells were maintained at 37°C and 5% CO2 incubator.

Primary cell cultures

Adult male CD1 mouse myofibers were cultured in DMEM with HEPES (Thermo Fisher Scientific), supplemented with 10% FBS, 100 U/mL penicillin, 100 μg/mL streptomycin. Myofibers were maintained in culture at 37°C with 5% CO2.

Method Details

High-throughput screening

Plasmids

The plasmid targeting aequorin in the cytosol (cytAeq) was previously described (Brini, 2008). The plasmid targeting aequorin to mitochondria (mtAeq) was generated by subcloning aequorin between the BssHII (in the mitochondria targeting signal) - XhoI restriction sites in the pCMV/myc/mito vector (Thermo Fisher). The myc signal is not translated into a protein because of a stop codon at the end of aequorin.

Calcium measurements

For measurements of mitochondrial and cytoplasmic Ca2+ levels in intact cells, HeLa cells were transiently transfected with pCMV vector expressing cytoplasmic- or mitochondria-targeted aequorin (cytAeq and mtAeq, respectively). Cells were seeded onto T175 tissue culture flasks at a density of 0.7x106 cells/flask in growth medium. Three days later, cells were transfected with mtAeq or cytAeq with lipofectamine 3000 (Invitrogen) for 4 h according to manufacturer’s instructions. Cells were harvested and seeded into 384 well tissue culture plates (Falcon) at a density of 8000 cells/well. 24 h post seeding, cells were treated with 5 μM coelenterazine H (Regis Technology) for 90 min. Compounds were then injected at 10 μM concentration in DMSO for 10 min followed by 10 μM histamine (Sigma). Mitochondrial and cytoplasmic calcium was measured using FDSS 7000 uCell (Hamamatsu).

Mitochondrial membrane potential measurements

HeLa cells were plated onto 384 well plate at a density of 6000 cells/well in growth medium. 24 h later, the mitochondrial membrane potential was assessed using the Mitochondrial Membrane Potential Indicator Kit (m-MPI; Codex Biosolutions) according to manufacturer’s instructions.

Active compounds

MCU-i4 [3-Quinolinecarboxylic acid, 4-[[4-(diethylamino)phenyl]amino]-6-methyl-, ethyl ester], SciFinder ID 371924-24-2, and MCU-i11 [Pyrido[4’,3′:4,5]thieno[2,3-d]pyrimidine-1(2H)-acetamide, 7-acetyl-N-(4-ethoxyphenyl)-3,4,5,6,7,8-hexahydro-2,4-dioxo-3-(phenylmethyl)-], SciFinder ID 902903-59-7, were purchased from AKos GmbH. (cat # of MCU-i4 AKOS001675514; cat # of MCU-i11 AKOS001868841).

Plasmids and oligonucleotides

Plasmids encoding Micu1-HA, Micu1EFmut, and mtGCaMP6f were previously described (Patron et al., 2014, Tosatto et al., 2016). pcDNA3.1 plasmids were used as control in all overexpression experiments (referred to as the control condition).

To generate Micu1Q302A,Q306A,L443A pcDNA3.1-Micu1-HA was mutagenized using the following primers:

5′- TTAGAACGTCATGCGCAAGTTTGCGCGCAAATTCCAGGAAGTTTTTGATGGTCAG −3′

5′- CTCCAGGCCTCTCATCGCCCGCTGCTTCATGATG −3′

The QuikChange XL Site-Directed Mutagenesis Kit (Agilent) was used following manufacturer’s instructions.

To silence MICU1 and MICU2 the following siRNA sequences were designed:

siRNA-MICU1: 5′- GGCUAAAGUGGAGCUCUCA −3′

siRNA-MICU2: 5′- GAUAUAAUAGUAUGGCAAU −3′

A non-targeting siRNA was used as a control in all silencing experiments. siRNA-MICU1, siRNA-MICU2, and the non-targeting siRNA (siRNA-scrambled, MISSION siRNA Universal Negative Control #1, cat no. SIC001) were purchased from Sigma-Aldrich.

To stably silence MICU1, HeLa cells were infected with shMICU1 lentiviral particles as reported (Marchi et al., 2019).

In vitro transfection of cell lines

HeLa and HEK293T cells were transfected with a standard Ca2+ phosphate procedure as already described (De Stefani et al., 2011). MEFs were transfected with LipofectamineTM 3000 transfection reagent (Invitrogen). MDA-MB-231 cells were transfected with LipofectamineTM 2000 transfection reagent (Invitrogen).

Cell viability measurements

Cell viability was measured by cytofluorometry. Cells were incubated with 10 μM compounds for 10 h. In specific experiments, 40 μM ceramide was also added. Apoptotic and necrotic cells were respectively identified by FITC-Annexin V (Roche) and propidium iodide (Sigma) labeling for 15 min at 4°C. Cell populations were analyzed by FACS (FACS Canto II, BD BioSciences). Data were processed using the BD Vista software.

Western blotting and antibodies

To monitor protein levels, protein extracts were prepared in RIPA buffer (125 mM NaCl, 25 mM Tris-Cl pH7.4, 1 mM EGTA-Tris pH 7.4, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS and Complete EDTA-free protease inhibitor mixture-Roche-). 40 μg of total proteins were loaded, according to BCA quantification. Proteins were separated by SDS-PAGE electrophoresis, in commercial 4%–12% acrylamide gels (Thermo Fisher Scientific) and transferred onto nitrocellulose membranes (Thermo Fisher Scientific) by wet electrophoretic transfer. Blots were blocked 1 h at RT with 5% non-fat dry milk (Bio-Rad) in TBS-tween (0.5M Tris, 1.5M NaCl, 0.01% Tween) solution and incubated at 4°C with primary antibodies. Secondary antibodies were incubated 1 h at RT. The following antibodies were used: anti-MCU (1:1000, Sigma-Aldrich), anti-MICU1 (1:1000 Novus Biologicals), anti-MICU2 (1:1000 Sigma-Aldrich), anti-GRP75 (1:5000, Santa Cruz Biotechnology) and anti-TOM20 (1:20000, Santa Cruz Biotechnology). Secondary HRP-conjugated antibodies were purchased from Bio-Rad and used at 1:5000 dilution.

C2C12 differentiation assay

Growing medium of 90% confluent C2C12 myoblasts was replaced with DMEM containing 2%-heat inactivated horse serum (HS) (differentiation medium). Five days later, cells were incubated for 24 h with MCU-i4 and MCU-i11 (10 μM) or with DMSO alone (0.1% v/v). At the end of the experiment, myotubes images were taken and myotubes width was measured by means of ImageJ software.

In vivo DNA transfection of mouse skeletal muscle and myofibers isolation

Hyaluronidase solution (2 mg/ml) (Sigma-Aldrich) was injected in the hind limb footpad of anesthetized animals. 30 min later, 20 μg of plasmid DNA in 20 μl of 0.9% NaCl solution was injected. One gold-plated acupuncture needle was placed under the skin at heel, and a second one at the base of the toes, oriented parallel to each other and perpendicular to the longitudinal axis of the foot and connected to a BTX Electroporation System (Harvard apparatus). 20 pulses, 20 ms each, 1 s of interval were applied to yield an electric field of 100 V. Single fibers cultures were carried out 7–10 days later.

FDB fibers were isolated 7-10 days after in vivo transfection. Muscles were digested in collagenase A (4 mg/ml) (Roche) dissolved in Tyrode’s salt solution (pH 7.4) (Sigma-Aldrich) containing 10% FBS. Single fibers were isolated, plated on laminin-coated glass coverslips and cultured as reported in the “primary cell cultures” paragraph.

Ca2+ measurements

For measurements of cytosolic Ca2+ transients and mitochondrial Ca2+ uptake, cells grown on 12-mm round glass coverslips were transfected with cytAeq plasmids or with low-affinity mitochondrial (mtAeq-mut) plasmids, as previously described (Tosatto et al., 2017). 24 h later, cells were incubated with 5 μM coelenterazine in Krebs-Ringer modified buffer (KRB) (125 mM NaCl, 5 mM KCl, 1 mM MgCl2, 20 mM HEPES, 1 mM MgSO4, 0.4 mM KH2PO4, 1 mM CaCl2, 5.5 mM glucose, pH 7.4) supplemented with 1 mM CaCl2 at 37°C. 2h later, coverslips were transferred to the perfusion chamber, and Ca2+ transients were evoked by agonist treatment. Specifically, HeLa cells were stimulated with 100 μM histamine, while MEFs, MDA-MB-231 and HEK293T cells were stimulated with 100 μM ATP. 30 s before agonist addition 10 μM compounds were added and maintained during stimulation. At the end of the experiment cells were lysed with 100 μM digitonin in a hypotonic Ca2+-rich solution (10 mM CaCl2 in H2O), thus discharging the remaining aequorin pool. The light signal was collected and calibrated into [Ca2+] values by an algorithm based on the Ca2+ response curve of aequorin at physiological conditions of pH, [Mg2+] and ionic strength, as previously described (Tosatto et al., 2017).

Alternatively, measurements were carried out on an EnVision plate reader (PerkinElmer) equipped with a two-injector units. Cells were transfected as described above in 6-well plates and then replated into 96-well plates (1:6 dilution) the day before the experiment. After reconstitution with 5 μM coelenterazine, cells were placed in 70 μl of KRB containing 10 μM compounds or 0.1% DMSO and luminescence from each well was measured for 1 min. During the experiment, 100 μM histamine was first injected to activate calcium transients, then a hypotonic, Ca2+-rich, digitonin-containing solution was added to discharge the remaining aequorin pool.

In the experiments with permeabilized cells, a buffer mimicking the cytosolic ionic composition (IB) was employed: 130 mM KCl, 10 mM NaCl, 2 mM K2HPO4, 5 mM succinic acid, 5 mM malic acid, 1 mM MgCl, 20 mM HEPES, and 1 mM pyruvate [pH 7] at 37°C. IB was supplemented with either 100 mM EGTA (IB/EGTA) or a 2 mM EGTA-buffered [Ca2+] of the indicated concentration (IB/Ca2+). HeLa cells were permeabilized by a 1 min perfusion with 100 μM digitonin (added to IB/EGTA) during luminescence measurements. Mitochondrial Ca2+ uptake speed was calculated as the first derivative by using the SLOPE Excel function and smoothed for three time points. The higher value reached during Ca2+ addition represents the maximal Ca2+ uptake speed. Calculated [Ca2+] free was confirmed fluorimetrically with the Fura2 free acid form.

For measurements of resting mitochondrial [Ca2+], HeLa cells were grown on 24-mm coverslips and transfected with plasmids encoding 4mtGCaMP6f. 24 h later, coverslips were placed in 1 mL of KRB and imaging was performed on Zeiss Axiovert 200 microscope with PlanFluar 40 × /1.3 N.A. objective. Excitation was performed with a DeltaRAM V high-speed monochromator (Photon Technology International) equipped with a 75 W xenon arc lamp. The system is controlled by MetaFluor 7.5 (Molecular Devices) and was assembled by Crisel Instruments. Cells were thus alternatively illuminated at 474 and 410 nm, and fluorescence was collected through a 515/30-nm band-pass filter (Semrock). Exposure time was set to 200 ms at 474 nm and to 400 ms at 410 nm, in order to account for the low quantum yield at the latter wavelength. At least 5 fields were collected per coverslip, and each field was acquired for 30 s (1 frame/s). Analysis was performed with the Fiji distribution of ImageJ (Schindelin et al., 2012). Both images were background corrected frame by frame by subtracting mean pixel values of a cell-free region of interest. Data are presented as the mean of the averaged ratio of all time points.

For Ca2+ measurements in adult isolated myofibers, FDB muscles were electroporated with plasmids encoding 4mtGCaMP6f. One day after single fiber isolation, real-time imaging was performed. During the experiments, myofibers were maintained in KRB at RT, in the presence of 75 μM N-benzyl-P-toluenesulfonamide (BTS, Tocris) to avoid the fiber contraction. 3 min after the addition of the active compound, 40 mM caffeine (Sigma-Aldrich) was added when indicated to elicit Ca2+ release from intracellular stores. Experiments were performed on a Zeiss Axiovert 200 microscope equipped with a × 40/1.3 N.A. PlanFluor objective. Excitation was performed with a DeltaRAM V high-speed monochromator (Photon Technology International) equipped with a 75 W xenon arc lamp. Images were captured with a high sensitivity Evolve 512 Delta EMCCD (Photometrics). The system is controlled by MetaMorph 7.5 (Molecular Devices) and was assembled by Crisel Instruments. 4mtGCaMP6f was excited every second at 490/20 and 403/20 nm band-pass excitation filters and images were collected through a dual band emission filter (520/40 and 630/60). Exposure time was set to 50 ms. Acquisition was performed at binning 1 with 200 of EM gain. Changes in Ca2+ levels (490/403 nm fluorescence ratio) were expressed as R/R0, where R is the ratio at time t and R0 is the ratio at the beginning of the experiment. Mitochondrial [Ca2+] peak was expressed as (R-R0)/R0 and normalized for the control value. Analysis was performed with the Fiji distribution of ImageJ (Schindelin et al., 2012). Images were background corrected frame by frame by subtracting the mean pixel value of a cell-free region of interest.

Mitochondrial membrane potential (Δψ) measurements

Cells were cultured on coverslips, incubated with 20 nM tetramethyl rhodamine methyl ester dye (TMRM) (Life Technologies) in KRB for 20 min at 37°C. Each compound (10 μM) was added 2 min after the first acquisition. The probe was excited at 560 nm, and the emission light was recorded in the 590–650 nm range; 10 μM CCCP (carbonyl cyanide-p-trifluoromethoxyphenylhydrazone) was added at the end of the experiment to completely collapse the Δψ. Data are expressed as difference of TMRM fluorescence before and after CCCP depolarization.

In vitro expression and purification of Micu proteins

Expression and purification of Micu1 and Micu2 in E. coli

For the expression of Micu1 and Micu2, Mus musculus cDNAs were cloned into a pETite vector (Lucigen), suitable for heterologous expression in E. coli, in frame with a 6xhis-tag coding sequence at the N terminus (Vecellio Reane et al., 2016). E. coli BL21(DE3) cells were transformed with the recombinant plasmids and positive clones selected by antibiotic resistance. Transformed cells were grown overnight in selective LB medium and then subcultured the following day in fresh medium. The expression of the protein was induced by adding 1 mM isopropyl-thiogalactopyranoside (IPTG) and incubating the cells at 22°C overnight. For protein purification, cells were harvested by centrifugation, resuspended in lysis buffer (25 mM HEPES-HCl, pH 7.4, 500 mM NaCl, and protease inhibitors 1 mM PMSF) and lysed using one shot cell disrupter (Constant System Ltd.) at 1.35 kBar. The supernatant fractions were isolated from the cells debris by centrifugation (20,000 x g, 30 min, 4°C). Both proteins were purified to homogeneity by combining an affinity chromatography (HisTrap HP, GE Healthcare) and a size exclusion chromatography (Superdex 200 10/300 GL, GE healthcare).

Baculovirus/insect cells system and protein purification

For the expression of the Micu1-Micu2 complex, Mus musculus Micu1 and Micu2 coding sequences were cloned in the pFH vector suitable for expression in insect cell (Vecellio Reane et al., 2016). The baculovirus production and insect cells infection were performed as previously described (Fitzgerald et al., 2006). Briefly, the expression cassette was recombined into a baculovirus genome (EmBacVSVG) using the Bac-to-Bac Baculovirus Expression System (ThermoFisher), and the resulting bacmid was transfected into Spodoptera frugiperda (Sf9) insect cells to produce recombinant baculoviruses. Sf9 cells were cultured in Sf-900 III SFM medium (GIBCO) at 27°C, and the virus was amplified twice to support large-scale protein expression. Expi293 cells (GIBCO) grown in suspension at 37°C in a serum-free Expi293 medium (GIBCO), were infected with Micu1-Micu2 baculoviruses (15% v/v) after reaching a density of 4 × 106/ml. 24 h after infection, the culture was supplemented with 4 mM valproic acid to increase protein expression, and the growth was continued at 30°C for additional 48 h. The cell pellet was resuspended in lysis buffer (Tris HCl pH 8, 200 mM NaCl, 10% glycerol, 10 mM imidazole, 1 mM DTT and 1 mM PMSF) and lysed in a one shot disrupter at 0.35 kBar. A clarified crude extract was then obtained by centrifugation (20000xg, 40 min, 4°C) and purified by affinity chromatography (HisTrap HP, GE Healthcare). The affinity purified fractions were subjected to size exclusion chromatography with a Superdex 200 10/300 GL column, equilibrated in 25 mM HEPES-HCl pH 7.4, 500 mM NaCl.

SPR assay

For SPR analysis, a Biacore T200 (GE Healthcare) system was used. Purified recombinant proteins Micu1, Micu2 and Micu1-Micu2 complex were immobilized on TRIS-Ni-NTA sensor chip (XanTec) that was previously activated with NiSO4. Following immobilization levels were used for the experiments: Micu1 6600 RU, Micu2 3800 RU and Micu1-Micu2 complex 3100 RU. As reference a flow cell without immobilized protein was used. Binding analysis was carried out in a running buffer consisting of 50 mM HEPES, pH 7.4, 500 mM NaCl, 0.005% Tween 20 applying a flow rate of 50 μl/min. Compounds solutions were prepared as serial dilutions in 100% DMSO and diluted 100 fold in running buffer with a final concentration of 0.4 to 100 μM compound and 1% DMSO. For binding experiments the compound dilution series were injected with increasing concentrations. After each compound dilution serie, 25 μM MCU-i11 was injected to monitor binding efficiency of SPR chip surface during the experiment. To show specificity of interaction unrelated compounds showing no binding to Micu proteins were tested. Each sensorgram (time-course of the surface plasmon resonance signal) was double-referenced by substracting the binding signal of control flow cell and 1% DMSO control (0 μM compound). High performance injection parameters were used: the contact time 60 s followed by a 180 s dissociation phase. The SPR sensorgram data were analyzed using the BIAevaluation software (GE Healthcare), the curves were fitted with a steady-state affinity model.

Molecular Modeling studies

The 3D-model of Mus musculus Micu1 (UNIPROT entry: Q8VCX5) was obtained from the experimentally solved structure of Ca2+‐free Micu1 in its hexameric form (PDB: 4NSC; Resolution: 3.2 Å) (Wang et al., 2014). Two different models were generated: the monomeric form and the hexameric assembly. Since the six units of the X-ray structure of the hexamer cover different portion of the Micu1 sequence, the most complete chain (chain C) was selected to obtain the wider range in the Micu1 model (107-469). The Protein Preparation tool of MOE 2018 suite (https://www.chemcomp.com) was used to process the X-ray structure to add hydrogen atoms, incorporating missing atoms in 17 side chains, and modeling three missing loops respectively of length of 5, 7, and 16 residues (I137-A143; F176-E184; H260-A277). Protein partial charges were assigned using Amber2014 force field (Maier et al., 2015). A similar strategy was adopted to model the hexameric model. The identified MCU-i4 and MCU-i11 were built and their partial charge calculated after semi-empirical (PM6) energy minimization using MOE 2018 (https://www.chemcomp.com) (Stewart, 2007). Molecular docking studies were performed with Plants1.2 coupled to chemPLP scoring function (Korb et al., 2009) on both the hexamer and in the monomer defining as binding site a sphere placed on the model center of the mass and using a radius of 65 Å and 21 Å respectively. For each run 100 output conformations were generated and analyzed by visual inspection. In silico mutagenesis studies were performed using the Protein Design tool implemented in MOE 2018 suite (https://www.chemcomp.com).

Quantification and Statistical Analysis

Statistical details of experiments can be found in the figure legends. All results are representative of at least 3 independent experiments unless otherwise specified and are presented as mean ± SD. Significance was calculated by ANOVA (one-way or two-way) or by Kruskal-Wallis test. All statistical tests were run with SigmaPLot.

Data and Code Availability

This study did not generate datasets or code.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Micu1 (CBARA1)	Novus Biologicals	Cat# NBP1-86663; RRID:AB_11021790
Micu2 (EFHA1)	Sigma-Aldrich	Cat# HPA045511; RRID:AB_10959894
MCU (CCDC109A)	Sigma-Aldrich	Cat# HPA016480; RRID:AB_2071893
GRP 75 (D-9)	Santa Cruz Biotechnology	sc-133137; RRID:AB_2120468
Tom20 (FL-145)	Santa Cruz Biotechnology	sc-11415; RRID:AB_2207533
Bacterial and Virus Strains
E.coli BL21 (DE3)	New England BioLabs	Cat#C2530
E.Coli DH10MultiBac	Geneva Biotech	N/A
Chemicals, Peptides, and Recombinant Proteins
caffeine	Sigma-Aldrich	CAS #: 58-08-2
Histamine dihydrochloride	Sigma-Aldrich	H7250
ATP	Sigma-Aldrich	CAS #: 34369-07-8
Coelenterazine n	Biotium	10115
digitonin	Sigma-Aldrich	CAS #: 11024-24-1
TMRM	thermo Fisher Scientific	T668
MCU-i4 3-Quinolinecarboxylic acid, 4-[[4-(diethylamino)phenyl]amino]-6-methyl-, ethyl ester	AKos GmbH	AKOS001675514; SciFinder ID 371924-24-2
MCU-i11 Pyrido[4’,3′:4,5]thieno[2,3-d]pyrimidine-1(2H)-acetamide, 7-acetyl-N-(4-ethoxyphenyl)-3,4,5,6,7,8-hexahydro-2,4-dioxo-3-(phenylmethyl)-	AKos GmbH	AKOS001868841; SciFinder ID 902903-59-7
Coelenterazine H	Regis Technology	CAS #: 50909-86-9
Critical Commercial Assays
Mitochondrial Membrane Potential Indicator Kit (m-MPI)	Codex biosolutions	CB-80600-010
Experimental Models: Cell Lines
MICU1−/− MEFs	Gyorgy Hajnoczky, Thomas Jefferson University, Philadelphia	Antony et al., 2016
HEK293T	ECACC	Cat# 12022001, RRID:CVCL_0063
MDA-MB-231	NCI-DTP	Cat# MDA-MB-231, RRID:CVCL_0062
HeLa	ATCC	Cat# CCL-2, RRID:CVCL_0030
C2C12	ATCC	CRL-1772; RRID:CVCL_0188
Experimental Models: Organisms/Strains
CD-1 IGS mouse	Charles River Laboratories	Crl:CD1(ICR)
Oligonucleotides
SI-CONTROL	Sigma-Aldrich	MISSION siRNA Universal Negative Control #1 cat. no. SI001
si-MICU1 GGCUAAAGUGGAGCUCUCA	Sigma-Aldrich	N/A
si-MICU2 GAUAUAAUAGUAUGGCAAU	Sigma-Aldrich	N/A
Recombinant DNA
pcDNA3.1-mtGCaMP6f	available upon request	Mammucari et al., 2015
Micu1Q302A,Q306A,L443A	this paper	N/A
mtAeq	available upon request	Rizzuto et al., 1992
mtAeq-mut	available upon request	Rizzuto et al., 1992
pETite-Micu1	available upon request	Vecellio Reane et al., 2016
pETite-Micu2	available upon request	Vecellio Reane et al., 2016
pFH-Micu1-Micu2	S.Pasqualato and S.Monzani (European Institute of Oncology, Milan, Italy)	Vecellio Reane et al., 2016
cytAeq	available upon request	Brini et al., 1995
pcDNA3.1-Micu1-HA	available upon request	Patron et al., 2014
Micu1EFmut	available upon request	Patron et al., 2014
shMICU1		Marchi et al., 2019
Software and Algorithms
MOE 2018 Suite		https://www.chemcomp.com
Plants	Korb et al., 2009	https://uni-tuebingen.de/fakultaeten/mathematisch-naturwissenschaftliche-fakultaet/fachbereiche/pharmazie-und-biochemie/pharmazie/pharmazeutische-chemie/pd-dr-t-exner/research/plants/
Sigma Plot 12.0	N/A	https://systatsoftware.com/products/sigmaplot/sigmaplot-statistical-analysis/
Fiji	ImageJ	Schindelin et al., 2012