Mitochondrial calcium uniporter regulator 1 (MCUR1) regulates the calcium threshold for the mitochondrial permeability transition

Significance

Cells injured by a variety of stressors feature a form of mitochondrial dysfunction termed the permeability transition. During this process, mitochondria swell and become disrupted, ultimately leading to cell death. In excitable cells such as cardiomyocytes or neurons, such injury is often triggered by calcium overload. By screening Drosophila cells, we have found a protein, mitochondrial calcium uniporter regulator 1 (MCUR1), that appears to regulate the amount of calcium required to induce the permeability transition. Modulating the function of this protein acutely may prove beneficial in limiting tissue damage during diseases that feature calcium overload.

Abstract

During the mitochondrial permeability transition, a large channel in the inner mitochondrial membrane opens, leading to the loss of multiple mitochondrial solutes and cell death. Key triggers include excessive reactive oxygen species and mitochondrial calcium overload, factors implicated in neuronal and cardiac pathophysiology. Examining the differential behavior of mitochondrial Ca²⁺ overload in Drosophila versus human cells allowed us to identify a gene, MCUR1, which, when expressed in Drosophila cells, conferred permeability transition sensitive to electrophoretic Ca²⁺ uptake. Conversely, inhibiting MCUR1 in mammalian cells increased the Ca²⁺ threshold for inducing permeability transition. The effect was specific to the permeability transition induced by Ca²⁺, and such resistance to overload translated into improved cell survival. Thus, MCUR1 expression regulates the Ca²⁺ threshold required for permeability transition.

The mitochondrial permeability transition (MPT) pore is large, and its opening collapses the mitochondrial membrane potential (ΔΨ), depleting the matrix of solutes <1.5 kDa. The osmotic imbalance swells and disrupts mitochondria, leading to cell death. The molecular structure of the MPT pore is unknown, although cyclophilin D [peptidyl-prolyl isomerase F (PPIF)], the ADP/ATP translocase, the F1-FO-ATP synthase, and spastic paraplegia 7 are key for its function (1–5).

Key triggers for the MPT include oxidative damage and Ca²⁺ overload. Reactive oxygen species attack a cysteine residue in mammalian PPIF (6, 7), but how Ca²⁺ overload activates the pore is unknown. Elimination of the known regulators typically inhibits the sensitivity of the MPT globally, not favoring any particular trigger (8–10). Because Ca²⁺ overload promotes cell death in excitable cells, targeting this pathway selectively may prove beneficial.

To discover novel regulators specific to mitochondrial Ca²⁺ overload, we studied MPT in Drosophila S2R+ cells, a system where screens have identified molecules involved in Ca²⁺ transport (11–13). We found that mitochondria within these cells were resistant to Ca²⁺ overload (14) but did possess an MPT. Moreover, we identified a mammalian gene, mitochondrial calcium uniporter regulator 1 (MCUR1), with no known Drosophila homolog, which is able to alter the MPT Ca²⁺ threshold. Inhibiting this gene confers resistance from cell death mediated by mitochondrial Ca²⁺ overload.

Results

As others have described (14), mitochondria isolated from Drosophila S2R+ cells are frequently damaged or defective. Therefore, we measured MPT-triggered release of the 622-Da fluorescent dye, calcein, from intact mitochondria (15). To obtain a mitochondria-specific signal, calcein-loaded cells were digitonin permeabilized, releasing cytoplasmic dye and leaving only the mitochondrial calcein. Repeated pulsing with 40 μM Ca²⁺ solution produced no calcein release (Fig. 1 A and B), although mitochondria depolarized after a few pulses, consistent with prior findings (14, 16). In these prior reports, this phenomenon was interpreted as revealing that, although the Drosophila possessed an MPT, its pore size was too small to release most solutes and lead to swelling. However, we were able to release larger solutes (calcein) by using 50 μM phenylarsine oxide (PAO), which triggers MPT independently of Ca²⁺ (17). These experiments suggest that Drosophila have an MPT response, but it is resistant to Ca²⁺ overload relative to mammalian mitochondria.

Fig. 1.

Drosophila MPT has a high Ca²⁺ threshold. (A and B) Exemplar permeabilized Drosophila S2R+ cell (A) and summary (B). Mitochondria depolarize (TMRM, 50 nM) but calcein is retained during 40 μM Ca²⁺ pulses (Ca²⁺). The 50 μM PAO releases calcein. Measurements were background subtracted and normalized to value before Ca²⁺ stimulation. Region of interest was over each whole cell, and an exemplar is shown in A (dashed circle). (Scale bar, 5 μm.) (n = 35 cells; error bars are SEM; and n are per condition throughout.) (C and D) Exemplar nonpermeabilized cells (C) and summary (D). Calcein exits mitochondria and redistributes throughout cells after 2 μM ionomycin/1 mM external Ca²⁺ (arrow, red trace). Redistribution abolished in 1.5 mM EGTA (green trace) (n > 35 cells). (E) Electron micrograph of S2R+ cell (Left), with magnified mitochondria within this cell (Right). Intact ultrastructure after control treatment [dimethyl sulfoxide (DMSO)]. (Scale bar, 500 nm.) (F) S2R+ cell treated with 2 μM ionomycin/1.5 mM Ca²⁺. (G) S2R+ cell treated with EGTA and ionomycin. (H) Quantification of mitochondrial area from EM images (n > 120, **P < 0.01). (I) Quantification of mitochondrial matrix density relative to surrounding cytoplasm (set to 0, n > 120, **P < 0.01). Cytoplasmic densities were not statistically different across treatments.

The lack of Ca²⁺-mediated MPT in Drosophila mitochondria could be explained by insufficient electrophoretic Ca²⁺ uptake or insensitivity to Ca²⁺. To distinguish these possibilities, we used the Ca²⁺ ionophore ionomycin, which we found induces much higher matrix Ca²⁺ than can be achieved by electrophoretic uptake. To image intact cells, we loaded calcein for >30 min and documented that extrusion of cytoplasmic dye leaves a predominantly mitochondrial signal (Fig. S1 A–H). Under these conditions, ionomycin/Ca²⁺ addition produced MPT, as calcein redistributed from mitochondria to a diffusely cytoplasmic pattern (Fig. 1 C and D). We confirmed that this MPT response was Ca²⁺ dependent. First, our media included 0.5 mM EDTA to chelate trace metals that may interfere with Ca²⁺-ionomycin transport. Second, binding the excess Ca²⁺ with EGTA abolished calcein redistribution (Fig. 1D). Next, we used transmission electron microscopy (EM) to examine mitochondrial morphology. Mitochondria in control cells retained normal morphology (Fig. 1E). Following ionomycin/Ca²⁺ treatment, cristae were disrupted and matrix contents lost, changes typical of MPT (Fig. 1F). Such changes were blocked by EGTA (Fig. 1G). To quantify the differences in mitochondria structure, we measured mitochondrial area and disruption in the EM images. For disruption, we measured the average matrix density value relative to the surrounding cytoplasm, which captures the main difference between normal mitochondria (matrices darker than cytoplasm) and disrupted ones (matrices that have lost their contents, and are closer in density to/lighter than cytoplasm), and accounts for differences in exposure across images. Using these measures, we found that, after ionomycin/Ca²⁺ treatment, S2R+ mitochondria were swollen and disrupted relative to controls (Fig. 1 H and I). Thus, Drosophila S2R+ cells possess an MPT response but require much higher Ca²⁺ loads than can be achieved electrophoretically.

Fig. S1.

Prolonged incubation with calcein-AM produces mitochondria-selective loading. (A–D) Drosophila cells loaded with 1.5 μM calcein-AM plus 1 mM cobalt (Co²⁺) chloride show a mitochondria-selective fluorescence profile (B) because Co²⁺ quenches cytoplasmic calcein. However, Ca²⁺ ionophores such as ionomycin bind Co²⁺ with high affinity and preclude this loading protocol for use in ionomycin-stimulated MPT experiments. (E–H) Prolonged loading of 1.5 μM calcein-AM (40 min) also yields a mitochondria-selective profile, as plasma membrane transporters remove cytoplasmic calcein, allowing the use of ionomycin or other Ca²⁺ ionophores. (C and G) TMRM selectively labels mitochondria. (D and H) Colocalization of calcein with TMRM in the merged images demonstrates mitochondria selective loading. (Scale bar, 5 μm.) (I) Electron micrograph of Kc167 cell (Left) with magnified individual mitochondria within this cell (Right) demonstrating intact ultrastructure after control treatment [dimethyl sulfoxide (DMSO)]. (Scale bar, 500 nm.) (J) Micrograph of Kc167+ cell treated with ionomycin/Ca²⁺. Low (Left)- and high (Right)-power images showing disrupted mitochondrial cristae and loss of matrix contents. (K and L) Quantification of mitochondrial area (K) and density (L) from EM images as in Fig. 1 H and I (n > 120, **P < 0.01, compared with DMSO control).

To show that Ca²⁺-activated MPT was not exclusive to S2R+ cells (hemocyte-like, late embryonic stage derived), we tested another Drosophila cell line (Kc167, plasmatocyte-like, dorsal closure stage derived). These Kc167 cells also underwent MPT (Fig. S1 I–L), although mitochondrial disruption was not as extensive as in S2R+ cells (Fig. S1L versus Fig. 1I).

To assess whether Drosophila MPT employs the same mechanisms as mammalian cells, we used RNAi or pharmacological inhibition of known MPT components, PPIF and ATP/ADP translocase. PPIF is the most studied MPT regulator, and a Drosophila homolog (Cyp-1) has been isolated in multiple proteomic studies of purified Drosophila mitochondria (18, 19) (Fig. S2A). Fluorescently tagged Cyp-1 localized to mitochondria in S2R+ or HeLa cells; and the predicted mitochondrial-targeting sequence of Cyp-1 was sufficient to drive mCherry-tagged human PPIF into S2R+ mitochondria (Fig. S2 B–E). Having established that Cyp-1 localizes to mitochondria, we found that inhibiting it led to a concordant reduction in MPT, whether induced by Ca²⁺ (Fig. 2 A and B) or PAO (Fig. S2F). Similarly, inhibiting the known Drosophila ATP/ADP translocase (sesB) also reduced MPT (Fig. 2 C and D, and Fig. S2F).

Fig. S2.

(A) Sequence alignment of human PPIF and Drosophila cyclophilin-1 using Clustal Omega. The putative mitochondrial-targeting sequence (MTS) (MitoProt II) is labeled in red, whereas green residues may also be part of the MTS based on homology to other PPIF homologs. (B) GFP-tagged Cyp-1 targets mitochondria in S2R+ cells. Live-cell imaging demonstrating Cyp-1-GFP fluorescence (Left), TMRM staining to identify mitochondria (100 nM, Middle), and merged image (Right) showing colocalization. (Scale bar, 5 µm.) (C) As in B, heterologously expressed human PPIF-GFP targeting S2R+ mitochondria. (D) Heterologously expressed, mCherry-tagged Drosophila Cyp-1 targets mitochondria in HeLa. (E) The MTS of Cyp-1 is sufficient to target human PPIF to mitochondria in S2R+ cells. Cells are expressing a construct encoding the Cyp-1 MTS attached N-terminal to human PPIF missing its own MTS, tagged with mCherry. (Scale bar, 5 µm.) (F) Inhibition of Cyp-1 or SesB blunts PAO-mediated MPT (n > 120 cells per condition, P < 0.01 for both Cyp-1 and SesB, compared with scrambled control). Protocol as in Fig. 2, except 100 µM PAO was used to induce MPT.

Fig. 2.

S2R+ MPT is regulated by Drosophila homologs of PPIF and ATP/ADP translocase. (A) Mitochondrial calcein redistribution in intact cells stimulated with ionomycin/Ca²⁺ (arrow) as in Fig. 1D. Cyclosporine A (CSA) (2 μM) inhibits PPIF, whereas cyclosporine H (2 µM) does not (n > 80 cells per condition, P < 0.001). (B) dsRNA targeting Drosophila Cyp-1 inhibits Ca²⁺-induced MPT (n > 80 cells per condition, P < 0.001). (C) Locking the Drosophila ADP/ATP translocase (SesB) in the m-conformation with 20 μM bongkrekic acid (BKA) inhibits Ca²⁺-induced MPT, whereas the c-conformation stabilizer carboxyatractyloside (CATR) (2 μM) does not (n > 50 cells, P < 0.001). (D) dsRNA targeting Drosophila SesB inhibits Ca²⁺-induced MPT (n > 50 cells, P < 0.001).

We considered three hypotheses for the result that wild-type S2R+ MPT resisted mitochondrial Ca²⁺ overload. First, S2R+ cells may have diminished mitochondrial Ca²⁺ uptake compared with other cells (Fig. S3A). Second, divergence of MPT components in Drosophila may lead to altered function. Finally, mitochondrial proteins absent in Drosophila but present in mammals may confer Ca²⁺ sensitivity to MPT. To test these hypotheses, we developed S2R+ cell lines stably expressing candidate proteins. Stable lines were not clonal, as attempts to isolate individual clones led to loss of cell adherence and failure to expand. For the first hypothesis, we expressed human MCU, the pore-forming subunit of the mitochondrial Ca²⁺ uniporter that mediates the mitochondrial Ca²⁺-selective current (20–23). For the second, we expressed human PPIF, because of the known MPT regulators, this homolog diverged most from its Drosophila counterpart. For the third, we identified candidates for S2R+ expression by screening for human nuclear-encoded mitochondrial genes (24) that had no homologs in the Drosophila genome, possessed transmembrane domains, were widely expressed, and interacted with PPIF (Fig. S3 B and C). We identified MCUR1, a potential component of the uniporter that may also function as a complex IV assembly factor (25, 26). In our S2R+ cell lines, human PPIF, MCU, and MCUR1 targeted mitochondria (Fig. 3A).

Fig. S3.

(A) Uniporter current density in S2R+ cells is less than in human HEK-293T cells. (Top) Ramp protocol. (Bottom) Exemplar traces from S2R+ and HEK-293T cells. S2R+ uniporter-mediated Ca²⁺ current (red) is ∼40% of the current seen in HEK-293T cells (black) (21) (n > 4 per condition, P < 0.01). (B) Our strategy for identification of novel regulators used bioinformatics for the first several steps, and biochemistry for the last step. (C) A coimmunoprecipitation screen for interaction with PPIF was used to further narrow the candidates identified after the third step in Fig. S3B. V5-epitope–tagged candidates were coexpressed with HA-tagged human PPIF in COS7 cells. Lysates were incubated with anti-HA conjugated beads, and proteins were resolved by SDS/PAGE and immunoblotted with anti-V5 antibodies. Although human MCUR1 overexpression was limited compared with other candidates, it was the only candidate that coprecipitated with human PPIF. (D) Simultaneous measurement of cytoplasmic and mitochondrial Ca²⁺ using R-GECO1 and mito-GEM-GECO1 (n = 40–80 cells per condition). Cells were stimulated with 5 µM thapsigargin (TG) at the indicated time (arrow), and Ca²⁺ transients were measured. (E) NADH autofluorescence at 405-nm excitation localizes to mitochondria in S2R+ cells. DIC, differential interference contrast image; Merge, merged image from NADH and MTO channels; MTO, MitoTracker Orange; NADH, NADH autofluorescence signal. (F) NADH autofluorescence measured in intact S2R+ cells. Store-activated Ca²⁺ uptake used to overload mitochondria. Cells in Ca²⁺-free media with 1 µM rotenone treated with 5 µM thapsigargin (2 min), and then stimulated with 10 mM extracellular Ca²⁺ (arrow). Only MCUR1 cells display Ca²⁺ overload-induced NADH release, with partial reversal by 1 µM cyclosporine A (CSA) (n > 400 cells, **P < 0.01 compared with control). (G) Stable expression of human MCUR1 in S2R+ cells confers Ca²⁺-induced release of nicotinamide adenine dinucleotides. Cells were digitonin-permeabilized and treated with 100 μM EGTA with or without 300 μM Ca²⁺. After centrifugation, cytosolic supernatants were assayed for NAD⁺ content, and the difference between Ca²⁺ and EGTA conditions (ΔNAD) was quantified (n = 5, *P < 0.05, **P < 0.01). (H) Ca²⁺-induced MPT induction in S2R+ cells does not lead to outer membrane permeabilization. Permeabilized S2R+ cell lines were incubated in high-KCl solutions with or without 200 μM Ca²⁺. After centrifugation, pellets and supernatants were resolved separately via SDS/PAGE to assay cytochrome c fractions released into the cytosol (supernatant) versus retained in the mitochondria (pellet). ATP5A and β-actin were used as mitochondrial and cytoplasmic loading controls, respectively. Only the pore-forming antibiotic alamethicin induced cytoplasmic cytochrome c release.

Fig. 3.

MCUR1 expression lowers the Ca²⁺ threshold for MPT activation in Drosophila S2R+ cells. (A) Mitochondrial targeting of stably expressed HA-tagged human PPIF (Top), MCU (Middle row), and MCUR1 (Bottom). Cells cotransfected with mitochondria-targeted GFP, stained with anti-HA (Left) and anti-GFP (Middle). (Right) Merge. (Scale bar, 5 µm.) (B) Exemplar traces, 2 × 10⁶ permeabilized S2R+ cells in a Ca²⁺-free suspension with 1 μM Oregon Green BAPTA-6F were stimulated with 10 μM Ca²⁺ (arrow). Traces were background subtracted. (Inset) Fraction of peak signal noted 30 s following stimulation (n = 3–4 independent trials, *P < 0.05). (C) Mitochondrial resting ΔΨ measured via the ratio of TMRM (50 nM, ΔΨ-dependent loading) to MitoTracker Green (100 nM, ΔΨ-independent) (n > 120 cells, **P < 0.01 compared with control). (D) Electron micrographs of permeabilized S2R+ cells (first and third column from Left) and individual mitochondria within each (second and fourth column). Cells treated with vehicle (No Ca²⁺) or 200 µM Ca²⁺ (+Ca²⁺). Only HsMCUR1 cells display substantial amounts of damage. (Scale bar, 500 nm.) (E) Fraction of total cells imaged for each condition that displayed abnormal mitochondria after Ca²⁺ overload as in D (n = 4 reviewers, for 22–54 cells per condition, *P < 0.05). (F) Fraction of total mitochondria that were abnormal in cells counted as having disrupted mitochondria in E (n > 90 mitochondria, for seven cells per condition, **P < 0.01). (G) Mitochondrial area as in Fig. 1H (n > 90, **P < 0.01). (H) Mitochondrial density as in Fig. 1I (n > 90, **P < 0.01).

We first established the Ca²⁺ handling properties of the cell lines. Basal [Ca²⁺]_mito, measured with the Ca²⁺ sensor mito-GEM-GECO (27), was greater in the MCU (2.0 ± 0.1 μM) and MCUR1 (1.1 ± 0.03 μM) lines, compared with baseline and PPIF lines (0.2 ± 0.02 μM, n > 170 cells for each condition). Stable S2R+ lines were more fragile than wild-type cells, and we were unable to isolate mitoplasts of suitably quality for electrophysiological assays of Ca²⁺ uptake. Instead, we monitored extramitochondrial Oregon Green BAPTA-6F fluorescence following a 10 μM Ca²⁺ pulse. Electrophoretic Ca²⁺ uptake was preserved in PPIF and MCUR1 lines but enhanced in the MCU line, as expected after increased uniporter expression (Fig. 3B, Fig. S3D, and Table S1). To test for changes in mitochondrial voltage gradients (ΔΨ), we measured the ratio of tetramethylrhodamine methyl ester (TMRM), which depends on ΔΨ, to load into mitochondria, to MitoTracker Green, which sequesters independently of ΔΨ (26), and found that MCU expression depolarizes mitochondria compared with control (Fig. 3C). Overexpression of either MCUR1 or PPIF did not significantly change ΔΨ relative to control.

Table S1.

Amplitudes and fits for mitochondrial Ca²⁺ imaging

Data are displayed as mean ± SEM. The red values indicate P < 0.05 compared with control. Peak amplitudes were calculated for each trace by subtracting the background value before the Ca²⁺ pulse, and averaged for each condition. For curve fitting, traces were normalized as follows:

$$F_{\mathrm{norm}}\left(t\right)=\frac{\left(F\left(t\right)-F_0\right)}{\left(F_{\max }-F_0\right)},$$

where F₀ was the baseline minimum before Ca²⁺ addition. Normalized traces were fit to a biexponential curve $F_{\mathrm{fit}}\left(t\right)=C_1{\exp }^{-t/{\tau }_1}+C_2{\exp }^{-t/{\tau }_2}$, subject to the constraint $C_1+C_2=1$ and nonnegative τ. Fits were performed by minimizing ${\displaystyle \sum }_t{\left[F_{\mathrm{norm}}\left(t\right)-F_{\mathrm{fit}}\left(t\right)\right]}^2$ using the GRG nonlinear method (default parameters) within the built-in Solver function in Microsoft Excel. Fits were checked by eye and, in rare instances, required rerunning the Solver function a second time. OGB 6F, Oregon Green BAPTA 6F. Overexpression of MCU in Drosophila S2R+ cells moves a greater fraction of mitochondria into the pool controlled by the fast time constant, whereas chronic inhibition of MCUR1 (shMCUR1-2) moves mitochondria into the pool with no uptake (very long time constant) and consequently reduces the mitochondrial Ca²⁺ amplitude.

We next queried susceptibility to Ca²⁺-induced MPT in these lines. To produce mitochondrial Ca²⁺ overload, digitonin-permeabilized cells were incubated in media containing 200 μM Ca²⁺, an amount sufficient to trigger mitochondrial depolarization in permeabilized cells (Fig. 1B). With electron micrography, we noted that only 3–14% of cells from control, MCU, or PPIF, had disrupted mitochondria, whereas in close to 30% of MCUR1 cells, Ca²⁺-overloaded mitochondria were damaged (Fig. 3 D and E). Although this near doubling of cells with aberrant mitochondria was substantial, this measure alone understated the difference between the Ca²⁺-treated MCUR1 line compared with the rest. For further quantitation, we focused on those cells with disrupted mitochondria. Within these cells, only 10–20% of mitochondria were altered in the control, MCU, or PPIF lines, whereas close to 60% of the mitochondria were abnormal in the MCUR1-expressing line (Fig. 3F), which led to obvious differences in mitochondrial area and density (Fig. 3 G and H). Of note, our S2R+ lines were nonclonal populations with variability in expression, so we speculate that particular MCUR1 clones might have had even higher fractions of cells with disrupted mitochondria.

Thus, increasing both Ca²⁺ uptake and baseline mitochondrial Ca²⁺, via MCU expression, was insufficient to endow substantial MPT to Drosophila cells. Similarly, making Drosophila MPT components more mammalian-like, via PPIF expression, was ineffective as well. Instead, our findings suggested that MCUR1 expression makes MPT more Ca²⁺-susceptible. We tested this hypothesis in two ways. First, we measured NADH, a mitochondria-localized fluorescent signal released during MPT (28, 29). The NADH signal located to mitochondria when measured as autofluorescence at 405-nm excitation (Fig. S3E) (30, 31). For mitochondrial overload, we raised intracellular Ca²⁺ levels to >1 μM with the store-operated current (32), which leads to mitochondrial Ca²⁺ uptake (13). Upon such stimulation, the NADH signal increased in the MCU line, likely from Ca²⁺-sensitive citric acid cycle enzymes (20, 33). However, significant NADH depletion—consistent with MPT—occurred only in MCUR1 cells (Fig. S3F). The effect was partly reversed with cyclosporine A. Next, we determined whether this depletion resulted from mitochondrial NADH release rather than mitochondrial oxidation. After Ca²⁺ overloading mitochondria in permeabilized cells, we separated cytosolic supernatants from the mitochondria-containing pellet. Because cytoplasmic NADH oxidizes (34), we measured cytoplasmic NAD⁺. Although Ca²⁺ lowered NAD⁺ levels in control, PPIF, and, to a lesser extent, MCU lines, it increased NAD⁺ only in MCUR1 cells (Fig. S3G). This would be expected if Ca²⁺-induced MPT caused a release of mitochondrial NADH. Therefore, in several independent assays, MCUR1 expression renders S2R+ MPT susceptible to Ca²⁺ overload.

In Drosophila, apoptosis requires mitochondrial function but not outer membrane permeabilization and cytochrome c release in most cases (35, 36). This may be partly due to resistance of the inner membrane to disruption, as we found here. Thus, we examined whether inner-membrane permeabilization induced outer-membrane rupture. We Ca²⁺ overloaded permeabilized cells, separating cytoplasmic supernatants from cellular pellets, and assayed cytochrome c content. Although cytochrome c could be released by addition of the pore-forming antibiotic alamethicin (14), no other condition proved successful (Fig. S3H).

Having induced Ca²⁺-mediated MPT in Drosophila, we turned to mammalian cells to further examine MCUR1 regulation. We acutely inhibited MCUR1 by transfecting HeLa cells with pooled siRNA (Fig. 4A). Compared with control, Ca²⁺ uptake, ΔΨ, and intramitochondrial pH were unaffected (Fig. 4 B–D, Fig. S4A, and Table S1), whereas basal and maximal oxygen consumption rates were reduced by such MCUR1 knockdown (Fig. 4E). Having established these baseline parameters of mitochondrial function, we next interrogated the Ca²⁺ sensitivity of MPT. In our first assay, we measured Ca²⁺-induced mitochondrial depolarization, indicative of MPT, by monitoring TMRM dequenching. In this protocol, loading cells with 20 µM TMRM leads to fluorescence quenching as ΔΨ sequesters the dye in mitochondria at high concentrations. Subsequent 30 µM Ca²⁺ pulses lead to transient depolarizations and release of the dye into bulk solution, causing increases in total fluorescence, until MPT triggers complete depolarization and a significant rise in the fluorescence signal. Applying this protocol, we found that MPT was triggered earlier in control (siScr) compared with MCUR1-inhibited (siMCUR1) cells, with an almost doubling in the number of pulses required to elicit depolarization [calcium retention capacity (CRC)] (Fig. 4F). The effect was dependent on Ca²⁺ overload, as MCUR1 knockdown did not inhibit PAO-mediated MPT (Fig. S4B). Moreover, the effect was independent of the assay used. With calcein imaging, a 100 μM Ca²⁺ pulse induced release in control cells, whereas siMCUR1 cells were resistant, with some induction noted after prolonged exposure to 500 μM Ca²⁺, despite intact PAO-mediated MPT (Fig. 4G). Finally, the effect was not due to siRNA off-target inhibition, as MCUR1 overexpression restored Ca²⁺ sensitivity.

Fig. 4.

MCUR1 inhibition increases the Ca²⁺ threshold for activation of the MPT. (A) RNAi efficacy measured using qPCR. Comparisons were between HeLa cells transfected with siRNA targeting MCUR1 (siMCUR1) and scrambled control, or between two clonal HeLa populations expressing shRNA targeting MCUR1 (shMCUR1-1 and -2) and a GFP control (shGFP). (B) Exemplar traces similar to Fig. 3B, except we used 5 × 10⁵ cells and measured extramitochondrial Ca²⁺ ([Ca²⁺]_ext, purple, left axis) with Calcium Green-5N and mitochondrial Ca²⁺ ([Ca²⁺]_mito, green, right axis) with rhod-2. (C) Fraction of peak Ca²⁺ signal left at 5 min (n = 4 trials). No difference between knockdown and scrambled control (siScr) noted. (D) TMRM (20 nM) to MitoTracker Green (50 nM) ratio as in Fig. 3C (n > 200 cells). No difference noted between siScr and siMCUR1. (E) Seahorse mitochondrial stress assay. Oxygen consumption rate (OCR) measured at baseline, and after sequential additions of 1 µM oligomycin (first arrow) for ATP production, 0.5 µM FCCP (second arrow) for maximal respiration, and 1 µM rotenone/1 µM antimycin A (third arrow) for nonmitochondrial respiration. To correct for any differences in cell numbers per well, cell counts were assayed using CyQuant after protocol completion, and OCR rates were normalized to relative CyQuant fluorescence. (n = 14–17, P < 0.01). (F) TMRM dequenching to monitor Ca²⁺-induced MPT. The 10⁶ permeabilized HeLa cells incubated in 20 μM TMRM were stimulated every 1.5 min with 30 μM Ca²⁺. Exemplar traces, siMCUR1- (red) and Scr-treated (blue) cells. The calcium retention capacity (CRC) is the number of additions before the mitochondria depolarize, visible as a sudden fluorescence rise (n = 6, P < 0.05). (G) Calcein imaging to monitor Ca²⁺-induced MPT. Control (orange), siMCUR1 (blue), or RNA-insensitive MCUR1 (black) expressing permeabilized HeLa cells imaged during 100 μM Ca²⁺, 500 μM Ca²⁺, and 50 μM PAO addition (arrows) (n = 50–130 cells, **P < 0.001). (H–M) Similar to B–G. (H) Ca²⁺ uptake blunted in the shMCUR1-2 but not shMCUR1-1 line. The baseline mitochondrial Ca²⁺ level for shMCUR1-2 was lower than the other lines (P < 0.05). (I) Fraction of peak Ca²⁺ left at 5 min (n = 14–16, P < 0.01). (J) No difference across groups (n > 140 cells). TMRM, 50 nM; MitoGreen, 100 nM. (K) Blunted respiration under all conditions for shMCUR1-2 (P < 0.01), with shMCUR1-1 showing an intermediate phenotype (n = 12–14). (L) CRC was elevated for shMCUR1-1 (n = 15–26, P < 0.01). (M) Calcein release was blunted for shMCUR1-1 (n = 95–233 cells, **P < 0.01).

Fig. S4.

(A) Mitochondrial pH measured using the genetically encoded, ratiometric fluorescent reporter mito-pHRed (51). No difference in ratio noted between siMCUR1 and siScr (n > 120 cells). (B) Change in ΔΨ measured in 10⁶ cells per trial by dequenching of TMRM (ΔTMRM) after MPT induction with 50 μM PAO (n = 4). No statistically significant difference was noted between conditions.

A potential discrepancy between our results and those published previously (25, 26) was the absence of a Ca²⁺ uptake phenotype after MCUR1 knockdown, which we hypothesized may reflect the choice of different cell lines or the length of knockdown induced by siRNA (acute) versus shRNA (chronic). To explore this further, we produced cell lines that stably expressed an shRNA targeting MCUR1, leading to clonal lines with intermediate (∼70%, shMCUR1-1) or more severe loss of MCUR1 (∼80%, shMCUR1-2) compared with control (shGFP) (Fig. 4A). We found that the more substantial chronic loss of MCUR1 led to diminished Ca²⁺ uptake, lower baseline [Ca²⁺]_mito, and a significant global deficit in respiration, whereas ΔΨ remained unchanged compared with control (shMCUR1-2, Fig. 4 H–J, and Table S1), similar to prior results (25). Turning to the line with moderate chronic MCUR1 reduction (shMCUR1-1), we found that Ca²⁺ uptake and ΔΨ were unchanged compared with control, and the deficit in respiration was intermediate between control and shMCUR1-2, consistent with the degree of genetic inhibition (Fig. 4 H–K). When we turned to assays of Ca²⁺-stimulated MPT (Fig. 4 L and M), we were surprised to find that, whereas the cell line with moderate MCUR1 reduction (shMCUR1-1) became resistant to MPT, the cell line with more substantial reduction (shMCUR1-2) lost that resistance. This normalization of resistance may be a consequence of MCUR1’s effects on cellular metabolism, as, in our hands, cell lines with more severe knockdown in MCUR1 (>85–90%) failed to expand beyond two to three splits and were unable to be tested further. Therefore, substantial acute or moderate chronic inhibition of MCUR1 leads to mitochondrial resistance to Ca²⁺ overload, whereas more substantial chronic inhibition apparently severely disrupts cellular metabolism, leading to cell death. These effects are separable, as we could maintain the MPT-resistant cells displaying moderate MCUR1 reduction (shMCUR1-1) for prolonged periods without disruption to Ca²⁺ uptake or ΔΨ.

Because our interests focused on Ca²⁺-induced MPT, we next explored potential mechanisms for MCUR1 regulation of this threshold. For protein biochemistry, we used a codon-optimized MCUR1 transcript, as this improved plasmid expression markedly compared with the native transcript. First, we confirmed that MCUR1 interacts with PPIF and Cyp-1 (Fig. S5 A and B). We found that MCUR1 was not itself a Ca²⁺ sensor, as the MCUR1–PPIF interaction was not Ca²⁺ regulated (Fig. S5C), and MCUR1 inhibition increased the MPT resistance to strontium (Sr²⁺) (Fig. S5D). This divalent enters mitochondria via the Ca²⁺ uniporter and disrupts ΔΨ via MPT (23, 37). Thus, mechanistically, inhibiting MCUR1 appears to decrease the sensitivity of MPT for divalents taken up by the uniporter, independent of ΔΨ.

Fig. S5.

(A) Domain structure of MCUR1, DUF1640, domain of unknown function 1640. CC, coiled-coiled; MTS, mitochondrial targeting sequence; TM, transmembrane segment. (B) Coimmunoprecipitation of MCUR1 with PPIF. HEK-293T cells cotransfected with MCUR1-HA and Flag-tagged human PPIF, Drosophila Cyp-1, or human superoxide dismutase 2 (SOD2). Protein was immunoprecipitated (IP) from lysates, resolved on SDS/PAGE, and Western blotted (WB) as indicated. MCUR1 appears as a larger band (mature protein without mitochondrial-targeting sequence) and a smaller C-terminal fragment. (C) The MCUR1–PPIF interaction is not Ca²⁺ dependent. HEK-293T cells coexpressing MCUR1-HA and PPIF-Flag were lysed in buffer containing varying Ca²⁺ concentrations and coimmunoprecipitated as indicated. (D) The strontium retention capacity (SRC) is also increased after MCUR1 knockdown (n = 8, P < 0.05). Protocol as in Fig. 4F except that 30 µM strontium was used instead of Ca²⁺. (E) Coimmunoprecipitation of MCUR1 with MCU. (F) Coimmunoprecipitation of PPIF with MCUR1 fragments. Peptides A–G are HA-tagged MCUR1 fragments with domain structure as represented above. (G) Coimmunoprecipitation of MCU with MCUR1 fragments. (H) mCherry-tagged MCUR1 fragments overexpressed in HeLa cells and subject to calcein imaging as in Fig. 4G.

One potential explanation for our findings is that MCUR1 approximates the uniporter and MPT complexes, exposing the MPT Ca²⁺ sensor to higher local divalent concentrations present near the matrix face of the uniporter pore. Sensing of a local Ca²⁺ signal may partly explain why the MPT Ca²⁺ threshold is independent of matrix Ca²⁺ levels, and rather depends on total Ca²⁺ uptake (38, 39). To test this hypothesis, we assayed for interactions between MCUR1 and MCU, the pore-forming uniporter subunit (25) (Fig. S5E). MCUR1 bound MCU, consistent with prior reports (25, 40), but did not interact with another MPT regulator, the ATP/ADP translocase (ANT). To determine what portion of MCUR1 is critical for interactions with the uniporter and MPT channels, we generated HA-tagged constructs encoding MCUR1 segments. When cotransfected with Flag-tagged PPIF, we found that only those MCUR1 constructs containing the amino-terminal portion of the domain of unknown function 1640 (DUF1640) were able to pull down PPIF (Fig. S5F). Conversely, when cotransfected with Flag-tagged MCU, we found that fragments containing the MCUR1 coiled-coil domain interacted with the uniporter complex (Fig. S5G). When the coiled-coil domain was expressed by itself, it aggregated with MCU in a high–molecular-weight complex. Thus, the nontransmembrane portion of DUF1640 appears critical for MCUR1 protein–protein interactions.

To examine the importance of such interactions, we tested MPT in cells expressing these MCUR1 fragments. Of seven mCherry-conjugated MCUR1 fragments, six successfully targeted mitochondria. Overexpression of the DUF1640 fragment alone produced no change compared with control during MPT (fragment C, Fig. S5H). This fragment interacts with both the MPT and uniporter complexes, and is tethered to the inner membrane by a carboxyl-terminal transmembrane domain. By eliminating this DUF1460 transmembrane domain, we generated a construct that, when expressed, made mitochondria more susceptible to MPT (fragment E, Fig. S5H). Conversely, eliminating the amino-terminal portion of DUF1640 before the coiled-coil domain resulted in peptides that interacted with the uniporter but not the MPT complex (fragments D and G, Fig. S5 G and H). Remarkably, expression of these fragments led to a dominant-negative effect, inhibiting MPT (Fig. S5H), particularly when the construct was sequestered to the membrane (fragment D versus G). Our results suggest that MCUR1 associates with only a fraction of uniporter and MPT complexes. First, there is limited enrichment of MCU following MCUR1 immunoprecipitation, and vice versa (Fig. S5E). Second, expression of an MCUR1 fragment able to associate with both the uniporter and MPT channels (fragment E, Fig. S5H) enhanced MPT, suggesting that this peptide accesses a pool of uniporter or MPT complexes unbound by native MCUR1.

Finally, we assessed whether inhibition of MCUR1 affected cell death induced by mitochondrial Ca²⁺ overload. For these assays, we used H9c2 rat cardiomyoblasts, because investigators have defined a cell death protocol produced by mitochondrial Ca²⁺ overload (41). We assayed cellular survival by flow cytometry. First, light-scattering properties for all counted cells defined populations of live and dead/injured cells (Fig. 5A). Second, we stained the cells with cytoplasmic calcein (which is retained in live cells) and nuclear ethidium homodimer-1 (which stains nuclear DNA in dead cells). Their respective fluorescence levels also defined populations of cells that were live (calcein⁺, Eth⁻), dead (calcein⁻, Eth⁺), or injured (calcein⁺, Eth⁺; membranes are damaged enough to allow Eth-1 staining but not enough to completely release calcein) (Fig. 5 B and C). At baseline, cellular survival was similar after acute depletion of MCUR1 (93% inhibition via qPCR) compared with scrambled control (Fig. S6). In a first protocol, we treated cells with hydrogen peroxide (H₂O₂), as subsequent cell death is enhanced by mitochondrial Ca²⁺ (22). We found that 0.5 mM H₂O₂ induced equivalent amounts of cell survival (Fig. 5 D and E), but inhibition of MCUR1 prevented cell death, maintaining a greater fraction of cells in an injured state (Fig. 5 E and F). Next, we turned to the mitochondrial Ca²⁺ overload-specific protocol (41). Here, addition of 40 μM C2-ceramide followed by mitochondrial uptake of Ca²⁺ released from internal stores leads to substantial cell death compared with untreated cells. Remarkably, acute MCUR1 depletion improved cell survival by approximately twofold and reduced cell death by threefold (Fig. 5 G–I).

Fig. 5.

MCUR1 knockdown inhibits cell death from mitochondrial Ca²⁺ overload. (A) Light-scatter analysis of control, untreated cells reveals populations of live (green) and dead/injured (purple) cells. The graph displays all cells analyzed per experiment. (B) Live/dead analysis. Calcein and ethidium homodimer-1 (Eth-1) fluorescence levels define populations of live, injured, dead, or unstained cells. Green and purple labeling are defined in A. Light-scatter and fluorescence assays show good overlap for live/dead/injured classification. The graph displays a subset of cells from A, gated on scatter to minimize the contribution of unstained cells. (C) Histogram analysis of calcein levels from cells shown in B, with the calcein⁺ gate (blue bars) defined as in B. (D) Analysis as in A for siScr or siMCUR1-treated cells after cell death induction with H₂O₂. Gating is identical for siScr and siMCUR1 throughout. Live fraction = (# of green cells) ÷ (# of green + # of purple cells). (E) Live/dead analysis as in B for H₂O₂-treated cells. (F) Histogram analysis of calcein levels for cells in E. The calcein⁺ fraction and mean total calcein fluorescence level are displayed. (G–I) Analyses as in D–F for siScr or siMCUR1-treated cells after mitochondrial Ca²⁺ overload and cell death induced by a protocol combining C2-ceramide, thapsigargin (TG), and caffeine.

Fig. S6.

Light-scatter properties and live/dead analysis of siScr and siMCUR1-treated H9c2 cells at baseline (no cell death induction). Cells were taken from a confluent dish 3 d after siRNA treatment. Analysis as in Fig. 5.

Discussion

Drosophila cells possess a permeability transition that is quite resistant to Ca²⁺ overload. Based on this differential phenotype, we identified a protein, MCUR1, which regulates the Ca²⁺ threshold for the MPT. When expressed in Drosophila S2R+ cells, MCUR1 markedly reduces the Ca²⁺ threshold for MPT. Conversely, genetic inhibition of MCUR1 made mammalian cells resistant to Ca²⁺ overload and protected them from consequent cell death.

MCUR1 was initially identified as an essential subunit of the mitochondrial Ca²⁺ uniporter (25), whereas a second report suggested that the MCUR1 helped assemble complex IV (26). The latter group suggested that MCUR1 inhibition blunts Ca²⁺ uptake by impairing OXPHOS activity and thus reducing ΔΨ. In more recent work, MCUR1 depletion blunts mitochondrial Ca²⁺ currents, even when the transmembrane voltage is clamped (42), suggesting a direct effect on the uniporter.

We did not see significant changes in ΔΨ after manipulating MCUR1 expression, consistent with the original report (25). In Drosophila cells, MCUR1 overexpression enhanced basal Ca²⁺ levels but did not enhance Ca²⁺ uptake rates. In mammalian cells, we did find deficits in respiration, and these appeared in a graded manner, correlated with the degree of MCUR1 depletion. The blunting of Ca²⁺ uptake follows this pattern, becoming evident after significant prolonged depletion. However, the resistance to MPT is only evident after acute or moderate MCUR1 reductions.

Our results lead to several conclusions. First, the differences between these studies may be partly due to the methodology (acute versus chronic inhibition) or cell types used. In fact, such variability in mitochondrial Ca²⁺ uptake when examining the same gene has been seen in studies focused on other components of the uniporter (43–47). Although the Ca²⁺ uptake results vary, we can conclude that MCUR1 is not essential for mitochondrial Ca²⁺ transport in all species, as Drosophila cells display intact Ca²⁺ uptake despite possessing no MCUR1 homolog.

Second, MCUR1 regulation of MPT is independent of its effects on Ca²⁺ uptake rates, as (i) Ca²⁺ uptake does not change after MCUR1 overexpression in S2R+ cells or after acute/moderate inhibition in mammalian cells, despite altered Ca²⁺ sensitivity of the MPT, and (ii) enhancing the Ca²⁺ uptake rate via MCU overexpression does not induce MPT in S2R+. Such independent roles for the same protein are consistent with the behavior of other MPT components, such as the ATP/ADP translocator or the F₁-F_O-ATP synthase.

Third, MCUR1 is not the MPT Ca²⁺ sensor itself. Instead, a potential mechanism is that MCUR1 bridges the uniporter and MPT complexes, although it is not established whether these interactions involve other scaffolding proteins. This bridging hypothesis fits with prior suggestions that MCUR1 may assemble inner membrane complexes (26). Other hypotheses for MCUR1 function, such as an effect on the MPT Ca²⁺ sensor, are not excluded by this model but are difficult to investigate, because MPT Ca²⁺ sensing and matrix Ca²⁺ buffering are poorly understood.

Our results suggest that mitochondrial Ca²⁺ overload can be regulated in a MCUR1-dependent manner, and that acute manipulation of intermediates between Ca²⁺ entry and the MPT may be beneficial in treating injury due to Ca²⁺ overload.

Experimental Procedures

Detailed procedures are available in SI Experimental Procedures.

Mitochondrial Calcein and ΔΨ Imaging.

Cells were grown for 2–3 d on coverslips. For MPT experiments, they were loaded with 1.5 µM calcein for 20–50 min, whereas for ΔΨ experiments, they were loaded for 10 min with TMRM (20–50 nM) and MitoGreen (50–100 nM). For ionomycin and ΔΨ experiments, the media was replaced with a modified Tyrode’s solution before imaging. For other experiments, we permeabilized cells for 1–2 min with digitonin, and replaced media with a high-KCl solution containing 0.5 mM EGTA. Experiments were performed on an epifluorescence microscope (Olympus). For S2R+ cells, measurements were taken over the whole cell, whereas for HeLa cells, measurements were taken over the region of densest mitochondrial staining (usually perinuclear).

Calcium and TMRM Imaging.

Cells were counted, washed with PBS, and incubated in a high-KCl solution containing digitonin on ice for 15 min. Cells were then spun down, washed, and resuspended in high-KCl media containing the appropriate Ca²⁺ indicator or TMRM. Aliquots were transferred to a 96-well plate, allowed to reach room temperature, and subsequently fluorescence imaged on a Flexstation 3 plate reader (Molecular Devices).

Electron Microscopy.

After Ca²⁺ or control treatment, cells were fixed and imaged by modifying a protocol described previously (48).

Cell Survival Imaging.

H9c2 cardiomyoblasts were treated with scrambled or MCUR1 siRNA for 3 d, and in some experiments, the procedure was repeated. Cells were treated with either 0.5 mM H₂O₂ for 6 h or 40 µM C2-ceramide, 15 mM caffeine, and 1 µM thapsigargin, as described previously (41).

Statistical Analyses.

Analysis was performed using Microsoft Excel and R. We rejected the null hypothesis for P < 0.05. For comparisons involving n > 15 per condition, we used Student’s t test. For comparisons involving n < 15, we used nonparametric methods, including the Kruskal–Wallis one-way ANOVA followed by Dunn’s test, the Wilcoxon signed-rank test, and the Wilcoxon ranked-sum test. For assays involving multiple conditions, we compared each test condition to control using a Bonferroni correction.

SI Experimental Procedures

Reagents and Antibodies.

All reagents were purchased from Sigma unless otherwise specified. The following antibodies were used: anti-cytochrome C (Abcam; ab13575), anti-β-actin (Abcam; ab8224), anti-ATP5A (Abcam; ab14748), anti-hemagglutinin (HA) (Abcam; ab99110), anti-HA agarose (Sigma; A2095), horseradish peroxidase (HRP)-conjugated anti-HA (Roche; 12013819001), FITC-conjugated anti-GFP (Novus Biologicals; NB100-1771), anti-FLAG affinity gel (Sigma; A2220), HRP-conjugated anti-FLAG (Sigma; A8592), and HRP-conjugated anti-V5 (Life Technologies; R961-25).

Cell Culture.

Drosophila S2R+ and Kc167 cells were obtained from the Drosophila RNAi Screening Center (DRSC) and cultured using Schneider’s media (Gibco) supplemented with 10% (vol/vol) heat-inactivated FBS (Gibco) at room temperature. HeLa and H9c2 cells were cultured in Dulbecco’s modified Eagle media supplemented with 10% FBS (Sigma) at 37 °C.

Constructs for Transfection and RNAi.

All human genes were cloned from HEK-293 cDNA using the HiFi Hotstart Kit (Kapa Biosystems) following the manufacturer’s instructions, except as indicated below. Constructs for expression in mammalian cells were ligated in-frame into a pCDNA6 plasmid (Life Technologies) where the 3′ end contained HA, V5, mCherry, GFP, or FLAG epitope tags followed by a stop codon. Constructs for expression in S2R+ cells were ligated in-frame into pAC5.1 (Life Technologies) where the 3′ end contained a HA, mCherry, or GFP tag. Human MCUR1 expressed poorly in mammalian cells despite trials of various tagged or untagged constructs, alternative promoters, different cell types (HEK-293, HeLa, or COS-7), or stable selection using antibiotic resistance genes encoded from either a separate promoter or following an internal ribosomal entry site on the same transcript as MCUR1. For this reason, we synthesized an alternate transcript encoding the MCUR1 amino acid sequence using codons optimized for human expression using an online tool (IDT) (Genbank nucleotide KT968833). The CMV-mito-GEM-GECO and CMV-R-GECO1 plasmids were a gift from Robert Campbell, University of Alberta, Edmonton, Alberta, Canada (Addgene 32461 and 32444). The human MCU-Flag plasmid was a gift from Vamsi Mootha, Massachusetts General Hospital, Boston. GW1-Mito-pHRed (49) was a gift from Gary Yellen, Harvard Medical School, Boston (Addgene plasmid 31474). All constructs were sequenced for accuracy following plasmid construction.

For Drosophila RNAi, we generated dsRNA using PCR with primer pairs obtained from the DRSC and subsequent in vitro transcription (MEGAscipt T7 kit; Life Technologies) following the manufacturer’s instructions. We used DRSC27284 for dsRNA targeting cyp-1 (CG9916) and DRSC29874 for dsRNA targeting sesB (CG16944), both of which gave knockdown efficiency >90% via qPCR. The control amplicon encoded a T7 promoter followed by a 325-bp fragment from mCherry (Table S2). PCR was performed using cDNA obtained from S2R+ cells. dsRNA was isolated using TRIZOL and the manufacturer’s instructions. For mammalian RNAi, we purchased small interfering RNA pools targeting human MCUR1 (SMARTpool M-010730-01; GE Healthcare Dharmacon), rat MCUR1 (SMARTpool M-090259-01), human MCU (SMARTpool M-15519-00), or nontargeting pool 1 (D-001206-13). Short hairpin RNA vectors targeting MCUR1 were purchased from Sigma (TRC 134711 and 135627).

Table S2.

Primer pairs used for dsRNA and qPCR

Primer pairs	dsRNA/qPCR
GTAATACGACTCACTATAGGCCTGTCCCCTCAGTTCATGT	mCherry T7+R primer
GTAATACGACTCACTATAGGCTTCAGCTTCAGCCTCTGCT	mCherry T7+S primer
TGAGCACGCGGCTTCGTCTG	Drosophila sesB qPCR
GGGCGACGGCAGTCTTGGAG	Drosophila sesB qPCR
AGCTTCGTCAGCGTGCAGTG	Drosophila Cyp-1 qPCR
TTGGGCACGACATCGGAGCG	Drosophila Cyp-1 qPCR
ACGGTACACCAGAGGATCGC	Human MCU qPCR
TGAGTGTGAACTGACAGCGTT	Human MCU qPCR
TGAGGTTGCTGGCCTCAAAA	Human MCUR1 qPCR
TCCCAGAGCTACTGTTAGGCA	Human MCUR1 qPCR
TGTTGCCATCAATGACCCCTT	Human GAPDH qPCR
CTCCACGACGTACTCAGCG	Human GAPDH qPCR
AATAGTGTCCCTGCATGCCC	Rat MCUR1 qPCR
AGGCGGTAAAATCCCAGAGC	Rat MCUR1 qPCR
TCTCTGCTCCTCCCTGTTCT	Rat GAPDH qPCR
GATGGTGATGGGTTTCCCGT	Rat GAPDH qPCR

For siRNA and shRNA experiments, HeLa cells were transfected using Lipofectamine RNAiMax (Life Technologies), following the manufacturer’s instructions. For siRNA experiments, the transfection was repeated in some trials, and experiments were carried out 72 h after transfection. For shRNA experiments, stable lines were generated by selecting with 2 µg/mL puromycin, and single clones were expanded.

Mitochondrial Calcein Imaging.

Cells seeded on to a 12-mm coverslip were incubated with 1.5 μM calcein-AM (Life Technologies) in a modified Tyrode’s solution (130 mM NaCl, 10 mM Hepes, 10 mM glucose, 2 mM MgCl₂, 2 mM CaCl₂, unless otherwise specified). For human cells, such incubations lasted 20–30 min at 37 °C. For S2R+ cells, incubations were for 20–50 min at 30 °C. During longer incubations (>30 min), cytoplasmic calcein was extruded from S2R+ cells, leaving a predominantly mitochondrial signal. In experiments requiring ionomycin application, the external media was supplemented with 0.5 mM EDTA to chelate trace non-Ca²⁺, non-Mg²⁺ divalents, as some of these (e.g., cobalt) bind ionomycin with higher affinity than Ca²⁺ and are capable of quenching calcein fluorescence. This manipulation was critical to distinguish true MPT from calcein quenching induced by ionomycin-divalent transport. In experiments requiring plasma membrane permeabilization, mammalian cells were incubated with 0.05–0.1 mg/mL digitonin in a high-KCl solution (125 mM KCl, 20 mM Hepes, 2 mM K₂HPO₄, 5 mM glutamate, 5 mM malate, 1 mM MgCl₂, pH 7.2 with KOH) with 0.5 mM EGTA, whereas for S2R+ cells the digitonin concentration was 0.1–0.2 mg/mL. Cells were imaged in either the Tyrode’s (intact cells) or the high-KCl solution (permeabilized). Free Ca²⁺ calculations were performed using MaxChelator (Stanford). Imaging was performed using a Lambda DG-4 arc lamp (Sutter), GFP filter set (Chroma), and Orca-ER CCD camera (Hamamatsu) at room temperature. Images were analyzed on Slidebook software (Intelligent Imaging Innovations). Brightness levels were slightly adjusted in the images to improve contrast. This adjustment was applied to each image as a whole and has not obscured or eliminated any particular feature of the image.

ΔΨ and Calcium Imaging.

Plate reader-based experiments were performed for both S2R+ and HeLa cells. HeLa cells were incubated with 1 µM rhod-2 AM for 30 min at 37 °C. For both S2R+ and HeLa, cells were washed with PBS containing 5 mM EDTA (PBS), and subsequently treated with 0.05% trypsin for 5 min (HeLa cells) or scraped off (S2R+). The cells were resuspended in ice-cold PBS solution, counted, and pelleted at 800 × g for 5 min. The pellet was resuspended in ice-cold high-KCl solution containing 0.5 mM EDTA and either 0.1 mg/mL (HeLa) or 0.2 mg/mL (S2R+) digitonin for 15 min. The cells were spun at 200 × g for 10 min and resuspended at a density of 5 × 10⁶ (S2R+) or 1 × 10⁷ cells/mL (HeLa) in ice-cold high-KCl solution containing 1 μM Oregon Green BAPTA 6F (Ca²⁺ imaging, S2R+ cells), 1 μM Calcium Green 5N (Ca²⁺ imaging, HeLa cells), or 20 μM TMRM (ΔΨ-based MPT imaging). After at least 5-min equilibration at room temperature, a 100-μL suspension was used for subsequent imaging in a FlexStation 3 (Molecular Devices). In each trial, we obtained data from paired control and test conditions processed identically to account for any differences due to incubation lengths. Excitation and emission for Ca²⁺ imaging were 488/510 nm, and for TMRM imaging, 546/590 nm.

For measurement of resting ΔΨ, cells grown on coverslips were incubated with 20–50 nM TMRM and 50–100 nM MitoTracker Green (Thermo Fisher) for 15 min, and subsequently imaged as in Mitochondrial Calcein Imaging above, using GFP or mCherry filter sets (Chroma).

For imaging using mito-GEM-GECO, cells were transfected 2 d before performing the assay, and seeded on to 12-mm coverslips. Cells were washed with PBS and imaged in modified Tyrode’s solution on an Olympus Fluoview 1000 confocal microscope. A fluorescence emission spectrum was taken from 420 to 570 nm using 405-nm laser excitation, and the 450/510-nm emission ratio was used for further calculations. Spectra were taken at baseline and, for control cells, after addition of 5 µM thapsigargin, 2 µM ionomycin, and 5 mM EGTA (R_min estimate), or 2 µM ionomycin and 10 mM CaCl₂ (R_max estimate). Ca²⁺ concentrations were calculated using the following:

$$\left[\mathrm{C}{\mathrm{a}}^{2+}\right]=K_{\mathrm{D}}\left(\frac{R-R_{\min }}{R_{\max }-R}\right)\left(\frac{S_{f510}}{S_{b510}}\right),$$

where R is the background-corrected 450- to 510-nm mito-GEM-GECO1 ratio, R_min is the ratio in Ca²⁺-free conditions, R_max is the ratio in saturated Ca²⁺, S_f510/S_b510 is the ratio of fluorescence at 510 nm under Ca²⁺-free to Ca²⁺-saturated conditions, and K_D is 340 nM (27, 50). For experiments combining R-GECO1 and mito-GEM-GECO1 imaging, we imaged the 450/510-nm emission ratio using 405-nm excitation, and red emission via a BA560IF filter using 543-nm excitation.

Electron Microscopy.

For ionomycin experiments, cells were washed in PBS and treated with DMSO alone or with 2 µM ionomycin in the presence of either 5 mM external Ca²⁺ or 5 mM EGTA for 5 min. For direct Ca²⁺ pulse experiments, cells were washed in PBS, permeabilized for 1 min in high-KCl solution containing 0.1 mg/mL digitonin and with or without 200 µM Ca²⁺. After permeabilization, cells were incubated for an additional 10 min in high-KCl solution with or without 200 µM Ca²⁺. After treatment as indicated, the cells were fixed with 2.5% glutaraldehyde, 1.25% paraformaldehyde, and 0.03% picric acid in 0.1 M sodium cacodylate buffer (pH 7.4). Subsequent procedures were carried out at the Electron Microscopy Core at Harvard Medical School. Cells were pelleted and fixation continued for at least 2 h at room temperature. The pellet was washed in 0.1 M cacodylate buffer and postfixed with 1% osmium tetroxide, 1.5% potassium ferrocyanide for 1 h, followed by triplicate H₂O washes. The cells were treated with 1% aqueous uranyl acetate for 1 h followed by two washes in water and subsequent dehydration in alcohol (10 min each: 50%, 70%, 90%; two washes of 10 min in 100%). The samples were then put in propylene oxide for 1 h and infiltrated overnight in a 1:1 mixture of propylene oxide and TAAB Epon (Marivac Canada). The following day, samples were embedded in TAAB Epon and polymerized at 60 °C for 48 h.

Ultrathin sections (∼60 nm) were cut on a Reichert Ultracut-S microtome, placed on copper grids, stained with lead citrate, and imaged with a JEOL 1200EX transmission electron microscope or a TecnaiG² Spirit BioTWIN. Images were recorded with an AMT 2k CCD camera.

Analysis of mitochondrial area and density was carried out on ImageJ. For stable lines, four blinded reviewers evaluated each image and scored the number of cells containing disrupted mitochondria.

NADH Imaging and NAD⁺ Quantitation.

For NADH imaging, S2R+ cells were seeded on to 12-mm coverslips. After 24–48 h, coverslips were washed and incubated in Ca²⁺-free PBS supplemented with 10 mM glucose and 1 μM rotenone. Cells were imaged on an Olympus Fluoview 1000 confocal microscope with 405-nm laser excitation and a 425- to 500-nm emission window. Cells were incubated in 5 µM thapsigargin for 3 min in Ca²⁺-free PBS. Images were taken every 30 s.

For the NAD⁺ assay, 2 million cells per condition were washed in Ca²⁺-free PBS, then permeabilized in a succinate–high-KCl solution (125 mM KCl, 20 mM Hepes, 1 mM MgCl₂ 2 mM K₂HPO₄, 5 mM succinate, 1 µM rotenone) with 2 mM EGTA and 0.2 mg/mL digitonin for 5 min at room temperature. The cells were centrifuged at 1,000 × g for 5 min, and each pellet was resuspended in 75 μL of the succinate–high-KCl solution containing 100 µM EGTA with or without 300 µM CaCl₂ for 5 min at room temperature. The suspension was centrifuged at 10,000 × g for 5 min, and the supernatant (cytoplasmic-released NAD⁺) and pellet (mitochondria-trapped NAD⁺) fractions were separated and processed for NAD⁺ using the Fluoro-NAD kit (Cell Technology), following the manufacturer’s instructions. Measurements were taken on a Victor³ microplate reader (PerkinElmer). Assays were normalized by separately measuring total protein content for 1 million cells using a BCA protocol (Thermo).

Cell Survival Imaging.

Approximately 5 × 10⁴ H9C2 cells were seeded into wells and transfected with 15 pmol of dsRNA targeting either rat MCUR1 or nontargeting pool 1 (GE Healthcare Dharmacon) using Lipofectamine RNAiMax (Life Technologies). In some trials, the transfection was repeated. Cell death induction treatments occurred 3 d later. In one protocol, cells were treated with 40 μM C2-ceramide for 4 h, followed by 10-min incubation in 20 mM KCl/5 mM CaCl₂-containing culture media. Cells were washed and then incubated in culture media supplemented with 40 μM C2-ceramide, 15 mM caffeine, and 1 μM thapsigargin for an additional 5 h (41). In the second protocol, cells were incubated with 0.5 mM H₂O₂ for 6 h. Cell death was analyzed using a LIVE/DEAD cytotoxicity assay (Thermo Fisher). Cells were loaded with 1 μM calcein-AM for 15 min at 37 °C and 4 µM ethidium homodimer-1 (Eth-1) for 5 min at room temperature. Subsequently, flow cytometry was performed on a BD LSRFortessa (BD Biosciences) using 488-nm excitation and 530-nm emission (calcein) or 561-nm excitation and 610-nm emission (Eth-1).

Immunocytochemistry.

Cells were grown on 12-mm coverslips, fixed with 4% paraformaldehyde in PBS for 15 min, and permeabilized with 0.5% Triton X-100 in PBS for 10 min. For blocking, primary, and secondary incubations, we used 10% goat serum and 0.1% Tween 20 (Bio-Rad) in PBS. The coverslips were imaged on an Olympus Fluoview 1000 confocal microscope using Alexa Fluor 488 and 532 filter sets. Brightness levels have been slightly adjusted in the images to improve contrast. This adjustment has been applied to each image as a whole and has not obscured or eliminated any particular feature of the image.

Cytochrome c Immunoblotting.

Eight million S2R+ cells per condition were washed in Ca²⁺-free PBS, and then permeabilized in high-KCl solution containing 0.2 mg/mL digitonin, protease inhibitors (Roche), and 1 mM EDTA at room temperature for 3–5 min. Cells to be treated with ionomycin or alamethicin were not permeabilized. Cells were pelleted at 800 × g for 5 min, and resuspended in 100 μL of high-KCl solution containing protease inhibitors and either 0.5 mM EDTA (no Ca²⁺ condition) or 200 µM Ca²⁺. For the alamethicin condition, this drug was added at 2 µM in PBS with 0.5 mM EDTA and protease inhibitors, whereas for ionomycin, this drug was added at 3 µM along with 10 mM Ca²⁺. Cells were treated for 1 h at room temperature. Alamethicin- or ionomycin-treated cells were Dounce-homogenized on ice. For all conditions, suspensions were subsequently spun at 12,000 × g for 10 min at 4 °C to pellet mitochondria. Proteins in the supernatant (cytoplasmic fraction) were precipitated with acetone, and resuspended in loading buffer [1× LDS buffer (Life Technologies) with 0.2 M DTT] for Western blotting. The remaining pellet (mitochondrial fraction) was resuspended directly into the loading buffer for Western blotting. Samples were heated to 85 °C for 10 min before electrophoresis and immunostaining using standard protocols.

Immunoprecipitation and Western Blotting.

For coimmunoprecipitation, COS-7 or HEK-293T cells were transfected 12–48 h before harvesting with 4–10 μg of plasmid using Lipofectamine 2000 (Life Technologies) following the manufacturer’s instructions. Particularly for MCUR1, cells were harvested within 12–24 h, as minimal expression was noted after 24 h. For experiments with PPIF overexpression, cells were suspended in Ca²⁺-free PBS, pelleted, and lysed in 0.2% sodium deoxycholate, 1% Triton X-100, 150 mM NaCl, 20 mM Hepes, 5 mM EDTA, and protease/phosphatase inhibitor mix (Thermo), pH 7.4 with NaOH. For other experiments, the lysis buffer did not include sodium deoxycholate. Lysates were passaged three to five times through a 27-gauge needle, and, for coimmunoprecipitation experiments, incubated with 25 μL of antibody-conjugated beads (12.5 μL of slurry) for 3 h. Beads were washed four times with lysis buffer and proteins eluted by heating to 85 °C for 10 min in 1× LDS buffer (Life Technologies) containing 0.2 M DTT.

Proteins were subject to immunoblotting using standard procedures. Membranes were imaged with a ImageQuant LAS 4000 (GE Healthcare Life Sciences). For protein quantitation, we used ImageJ (NIH).

Quantitative PCR.

Primers against Drosophila cyp-1 and sesB, human MCUR1 and MCU, and rat MCUR1 were designed using NCBI Primer-Blast (Table S2). Drosophila transcripts used for normalization were dTaf8 or dAct79b (51). For RNAi experiments, cells were transfected 72 h before the assay using Lipofectamine RNAiMax. Total RNA was purified using TriZOL reagent (Life Technologies) following the manufacturer’s instructions. We synthesized cDNA from 2 μg of RNA with SuperScript VILO (Life Technologies) and used 100 ng per well for qPCR. Reactions were carried out in triplicate using SYBR Green (Agilent) on an Applied Biosystems 7500 RT PCR system (Life Technologies). Results were quantified using the 2^−ΔΔCt method.

Seahorse-Based Oxygen Consumption Analysis.

Two to three days before analyses, 20,000–35,000 cells were seeded on XF analyzer plates pretreated with gelatin (0.1%, Embryomax; EMD Millipore). Oxygen consumption was measured on an XF-24 analyzer using a Mito Stress Test Kit (Seahorse Biosciences), following the manufacturer’s instructions. We used 1 µM oligomycin, 0.5 µM FCCP (determined as an optimal concentration in preliminary experiments), 1 µM rotenone, and 1 µM antimycin A. To account for differences in cell numbers due to variable cell growth, detachment, or seeding, and to allow comparison across repeated trials, after each experiment, relative cell density was measured using CyQuant fluorescence (Thermo Fisher). Oxygen consumption rates were then normalized by the fluorescence values.

Mitoplast Electrophysiology.

For S2R+ cells, we modified Kirichok’s protocol (23, 52). All solutions were kept ice-cold, and equipment was precooled. Cells grown to confluence in three 15-cm plates were washed with PBS, pelleted, and resuspended in a solution containing 250 mM sucrose, 5 mM Hepes, 1 mM EGTA, and 0.1% BSA. Cells were disrupted using a French Press (SLM-Aminco) at 800-psi internal pressure. The eluate was pelleted at 700 × g to remove debris. The supernatant was centrifuged at 8,500 × g to pellet a crude mitochondrial fraction. This pellet was resuspended in a solution containing 440 mM mannitol, 140 mM sucrose, 5 mM Hepes, and 1 mM EGTA for 10 min, and passed through the French Press again, this time at an internal pressure of 1,500–2,000 psi. The eluate was pelleted at 9,500 × g for 10 min. This mitoplast pellet was resuspended in a hypertonic solution containing 750 mM KCl, 100 mM Hepes, and 1 mM EGTA for storage. Aliquots were transferred to isotonic bath solution before recording. Whole-mitoplast electrophysiology was performed as previously described. We used 150 mM sodium gluconate, 10 mM Hepes, 2 mM EGTA, 1 mM MgCl₂ (pH 7.4 with NaOH; osmolarity 410–430 mOsmol with sucrose) for the pipette solution. Bath solution was initially 150 mM KCl, 10 mM Hepes, 1 mM EGTA (pH 7.4 with KOH), and was exchanged for a high-Ca²⁺ solution (100 mM calcium gluconate, 20 mM Hepes, 2 mM CaCl₂). Electrophysiology was performed using an Axopatch 200B amplifier (Molecular Devices).

Bioinformatic Screen.

We screened for novel MPT regulators by interrogating the Drosophila nonredundant protein sequence database with the MitoCarta compendium of human nuclear-encoded mitochondrial genes (24). Proteins with no Drosophila homolog were defined by BlastP expect value >10⁻³. We narrowed this list further by searching for candidates with expression across most tissues, using the MitoCarta annotation, as well as 1+ transmembrane domains, using TMHMM (version 2.0; CBS Prediction Servers).