EMRE is a Matrix Ca²⁺ Sensor that Governs Gatekeeping of the Mitochondrial Ca²⁺ Uniporter

SUMMARY

The mitochondrial uniporter (MCU) is an ion channel that mediates Ca²⁺ uptake into the matrix to regulate metabolism, cell death and cytoplasmic Ca²⁺ signaling. Matrix Ca²⁺ concentration is similar to that in cytoplasm, despite an enormous driving force for entry, but the mechanisms that prevent mitochondrial Ca²⁺ overload are unclear. Here, we show that MCU channel activity is governed by matrix Ca²⁺ concentration through EMRE. Deletion or charge neutralization of its matrix-localized acidic carboxyl terminus abolishes matrix Ca²⁺ inhibition of MCU Ca²⁺ currents, resulting in MCU channel activation, enhanced mitochondrial Ca²⁺ uptake and constitutively elevated matrix Ca²⁺ concentration. EMRE-dependent regulation of MCU channel activity requires intermembrane space-localized MICU1, MICU2 and cytoplasmic Ca²⁺. Thus, mitochondria are protected from Ca²⁺ depletion and Ca²⁺ overload by a unique molecular complex that involves Ca²⁺ sensors on both sides of the inner mitochondrial membrane, coupled through EMRE.

INTRODUCTION

The mitochondrial calcium uniporter is a Ca²⁺-selective ion channel localized in the inner mitochondrial membrane (IMM) (Gunter et al., 1994; Kirichok et al., 2004) that mediates Ca²⁺ uptake into the mitochondrial matrix from the cytoplasm to regulate metabolism, cell death and cytoplasmic Ca²⁺ signaling. Under normal resting conditions the matrix free Ca²⁺ concentration is similar to that in the cytoplasm (Lukacs and Kapus, 1987; Nicholls, 2009), despite an enormous ~180 mV driving force for Ca²⁺ entry generated by proton pumping by the respiratory chain, suggesting that the Ca²⁺ uniporter possesses mechanisms to inactivate it under resting conditions to prevent mitochondrial Ca²⁺ overload. However, the nature of such mechanisms is unclear. The mitochondrial Ca²⁺ uniporter is a complex of proteins including the Ca²⁺ selective pore-forming subunit MCU and accessory proteins including MICU1, MICU2, MCUR1 and EMRE (De Stefani and Rizzuto, 2014; Foskett and Philipson, 2015; Kamer et al., 2014). Previously it was suggested that either MICU1 or MICU2 provided a so-called gatekeeping function that reduces MCU-mediated Ca²⁺ uptake in situ below a threshold value of 1–2 μM external free Ca²⁺ (the low cytoplasmic [Ca²⁺] regime) to prevent mitochondrial Ca²⁺ loading under basal conditions (Csordas et al., 2013; Mallilankaraman et al., 2012; Patron et al., 2014), most likely by reducing MCU single channel open probability. However, it is unclear if MICU proteins exert their effects from the matrix or inter-membrane space or if Ca²⁺ binding to their pairs of EF hands is required (Foskett and Madesh, 2014). Furthermore, their regulation of MCU-mediated Ca²⁺ uptake has not been examined by electrophysiological studies of the uniporter channel directly in its native membrane environment, so the molecular details of channel regulation by MICU1 and MICU2 remain unknown. The importance of understanding this regulatory mechanism is underscored by patients with loss of function mutations in MICU1, who lack inhibition of mitochondrial Ca²⁺ uptake under basal conditions and exhibit proximal myopathy, learning difficulties and a progressive extrapyramidal movement disorder (Logan et al., 2014).

RESULTS

We recorded uniporter Ca²⁺ currents (I_MiCa) using patch clamp electrophysiology of mitoplasts (Fieni et al., 2012; Kirichok et al., 2004; Vais et al., 2015) isolated from human embryonic kidney (HEK) cells. In the whole-mitoplast recording configuration with the pipette solution lacking Ca²⁺, ruthenium red (RuR, 200 nM)-sensitive Ca²⁺ currents were observed (Figure 1A) with densities and properties similar to those previously reported for I_MiCa with matrix [Ca²⁺] buffered either at zero or >10 μM (Fieni et al., 2012; Kirichok et al., 2004). Similar currents were nearly abolished in cells with MCU knocked down (Figure 1B), confirming their identity as uniporter currents. Unexpectedly, I_MiCa was markedly reduced when matrix [Ca²⁺] was raised from 0 into a range from 30 nM to ~400 nM (Figure 1C), resulting in a biphasic matrix [Ca²⁺] dependence with apparent inhibition constant of 60 ± 30 nM and Hill coefficient of 1.0 ± 0.2, and apparent recovery constant of 730 ± 15 nM and Hill coefficient of 3.1 ± 1.6, with peak inhibition of MCU currents by ~75% at ~400 nM (Figure 1D). Of note, resting matrix [Ca²⁺] is ~100–300 nM (Boyman et al., 2014; Brandenburger et al., 1996; Ivannikov and Macleod, 2013; Lukacs and Kapus, 1987; Nicholls, 2009; Palmer et al., 2006), suggesting that this inhibition of MCU activity may be related to the previously reported inhibition of MCU activity in the low Ca²⁺ regime attributed to MICU1 and MICU2.

Figure 1. MCU channel activity is modulated by a mechanism dependent upon matrix [Ca²⁺].

(A) MCU current density in various bath [Ca²⁺] (indicated in inset) with 0 Ca²⁺ (1.5 mM EGTA) in the pipette solution, in response to voltage ramps from −160 to 60 mV. Inhibition by ruthenium red (RuR, 200 nM) in 1 mM bath Ca²⁺ also shown.

(B) MCU current density measured from MCU-KD cells.

(C) Similar to panel (A), recorded with pipette solution containing 400 nM free Ca²⁺.

(D) Response of MCU Ca²⁺current density at V_m = −160 mV, with 1 mM Ca²⁺ in the bath, as function of free [Ca²⁺] in pipette (matrix) solution. Data fitted with a biphasic Hill equation (continuous line).

Using a proteinase sensitivity assay (Sato and Mihara, 2010), we determined that MICU1 and MICU2 are localized outside of the matrix, in the intermembrane space (IMS; Figure S1A), in agreement with other reports (Csordas et al., 2013; Kamer and Mootha, 2014; Lam et al., 2015; Petrungaro et al., 2015; Sancak et al., 2013). This suggested that matrix [Ca²⁺] regulation of I_MiCa is mediated by MCU itself or by another matrix-localized Ca²⁺ binding protein. Plasma membrane Ca²⁺ channels also have Ca²⁺ dependent channel inactivation, usually mediated by an associated protein (Christel and Lee, 2012). EMRE is a single pass 10 kD IMM protein that binds to MCU and is required for uniporter activity (Sancak et al., 2013), although its function is unknown. By protease sensitivity biochemistry, we determined that the carboxyl terminus of EMRE faces into the mitochondrial matrix (Figure 2A and Figure S1B). The carboxyl terminus contains a conserved acidic motif that is homologous to the Ca²⁺ binding “Ca²⁺ bowl” and “Ca²⁺ clasp” in Ca²⁺-activated large conductance (BK) K⁺ and bestrophin Cl⁻ channels, respectively (Figure S1C). We speculated that it similarly bound Ca²⁺ in EMRE. Deletion of the EMRE carboxyl terminus or charge neutralization of its seven terminal acidic residues (Figure S1C) was without effect on MICU1 or MICU2 localization (Figure S1D), EMRE topology (Figure S1D and S1E), expression of MCU, MICU1 or MICU2 (Figure S2C) or ability of EMRE to interact with MCU (Figure S3) but it abolished matrix [Ca²⁺] regulation of uniporter Ca²⁺ currents (Figure 2B–F). Loss of matrix [Ca²⁺] inhibition of MCU channel activity in mutant EMRE-expressing cells resulted in constitutively elevated matrix [Ca²⁺] (Figure 2G–L and Figure S4).

Figure 2. The carboxyl terminus of EMRE confers matrix [Ca²⁺] sensitivity to MCU activity.

(A) The carboxyl terminus of EMRE resides in the matrix. EMRE-rescue (V5-tagged EMRE over-expressed in EMRE-KO cells) was probed for its amino- (α-EMRE) and carboxyl (α-V5) termini. Carboxyl terminus is protected from protease similarly to matrix-localized proteins.

(B to D) MCU current densities in various bath [Ca²⁺], recorded from mitoplasts from EMRE KO cells expressing wild-type EMRE (B), EMRE-ΔC (C) or EMRE-QN6 (D), with pipette solution containing 400 nM free Ca²⁺. Color schemes as in Figure 1.

(E) Summary of experiments shown in B–D. Means and SEMs. Wild-type (Figure 1D) and EMRE-rescue data included for comparison.

(F) Response of MCU Ca²⁺current density at V_m = −160 mV, with 1 mM Ca²⁺ in the bath, as function of free [Ca²⁺] in pipette (matrix) solution for indicated cells, showing loss of matrix [Ca2+] regulation in mitoplasts expressing EMRE-QN6. Wild-type data fitted with a biphasic Hill equation (continuous line).

(G to K) Basal Rhod-2 fluorescence in cells stably expressing EMRE constructs.

(L) Rhod-2 fluorescence quantification (fluorescence arbitrary units (f.a.u.); Mean ± SEM, n = 3).

(M) Gatekeeping measured by mitochondrial Ca²⁺ uptake (average traces; n = 3) in permeabilized cells is abolished in cells expressing the EMRE-QN6 mutant, evidenced by MCU-mediated Ca²⁺ uptake following inhibition of the Na⁺/Ca²⁺ exchanger with CGP that reduces baseline bath [Ca²⁺] to much lower levels than wild-type (WT) cells, and their ability to rapidly take up Ca²⁺ following addition of 500 nM Ca²⁺ bolus. Tg: thapsigargin, 2 μM. Digitonin: 40 μg/ml. See text for details.

(N) Similar to (M), demonstrating lack of MCU activity in EMRE KO cells (no Ca²⁺ uptake following CGP addition or after addition of 10 μM Ca²⁺ bolus) and normal gatekeeping and Ca²⁺ uptake in WT and EMRE-rescue cells.

See also Figures S1 to S4.

To determine the relationship between these observations and previously reported regulation of MCU activity by MICU1 and MICU2, we examined MCU activity in intact mitochondria by simultaneous measurements of mitochondrial Ca²⁺ uptake and mitochondrial membrane potential in permeabilized HEK293 cells as previously described (Mallilankaraman et al., 2012). Addition of digitonin to permeabilize the plasma membrane and thapsigargin to inhibit SERCA activity slowly raised bath [Ca²⁺] due to leak of Ca²⁺ from the endoplasmic reticulum (ER). Addition of CGP37157 (CGP) to inhibit the IMM Na⁺/Ca²⁺ exchanger and reveal the rate of unidirectional mitochondrial Ca²⁺ uptake, reversed that trend for all cells with functional MCU (Figure 2M and 2N). In contrast, the slow rise in bath [Ca²⁺] was unaffected by CGP addition in EMRE-KO cells (Figure 2N), reflecting absence of mitochondrial Ca²⁺ uptake, as expected since EMRE KO abrogates MCU activity in HEK293 cells (Sancak et al., 2013). In response to addition of a 10 μM Ca²⁺ bolus, bath [Ca²⁺] was rapidly reduced by MCU-mediated Ca²⁺ uptake in WT cells, whereas it remained high with no reduction in EMRE-KO cells (Figure 2N), confirming that EMRE is critical for MCU Ca²⁺ channel function. Control experiments showed complete abrogation of I_MiCa in EMRE-KO cells (n = 6; data not shown). MCU-mediated Ca²⁺ uptake in response to addition of a 10 μM Ca²⁺ bolus was restored in EMRE-rescue cells (Figure 2N) and in cells rescued with carboxyl-terminus mutant EMRE (Figure 2M). Unidirectional Ca²⁺ uptake through MCU observed after CGP addition (Figure 2M and N) reduced bath [Ca²⁺] for WT cells to a significantly higher level than that for mutant EMRE cells, indicating that inhibition of MCU activity in the low cytoplasmic [Ca²⁺] regime that terminated Ca²⁺ uptake in WT cells was lost in cells expressing mutant EMRE. Even in the absence of Na⁺/Ca²⁺ exchanger activity in WT cells, no decrease in bath [Ca²⁺] was detected after the addition of a 0.5 μM Ca²⁺ bolus, (Figure 2N), whereas bath [Ca²⁺] was restored to the pre-Ca²⁺ addition level after the addition of 0.5 μM Ca²⁺ to carboxyl-terminus mutant EMRE cells (Figure 2N), suggesting also that normal inhibition of MCU activity in the low cytoplasmic [Ca²⁺] regime was impaired by EMRE carboxyl terminus charge neutralization or deletion. These results indicate that EMRE is a Ca²⁺ sensor that mediates matrix [Ca²⁺] regulation of uniporter activity.

It was originally reported that MICU1 conferred inhibition of MCU-mediated mitochondrial Ca²⁺ influx in the low cytoplasmic [Ca²⁺] regime (Mallilankaraman et al., 2012). It was subsequently suggested that MICU2 instead mediates this inhibition of MCU activity (Patron et al., 2014). Measurements of mitochondrial Ca²⁺ uptake in permeabilized cells confirmed that knockout of either MICU1 or MICU2 abrogated normal inhibition of MCU activity in the low cytoplasmic [Ca²⁺] regime (Figure 3A), resulting in constitutive matrix Ca²⁺ loading (Figure 3B–D and Figure S4). To explore the relationship between the EMRE-dependent matrix [Ca²⁺] regulation of I_MiCa and apparent inhibitory functions of MICU proteins, we recorded I_MiCa in cells with either MICU protein knocked down. Remarkably, knockdown or knockout of either MICU1 or MICU2 (Figure S2A and 2B) abolished matrix [Ca²⁺] regulation of I_MiCa (Figure 3E and 3F). I_MiCa regulation (Figure 3E and 3F) and constitutive mitochondrial Ca²⁺ loading (Figure 3B and 3C) were rescued by expression of the wild-type proteins (Figure 3F and 3H–L). These results indicate that matrix [Ca²⁺] regulation of I_MiCa is part of the so-called gatekeeping mechanism underlying previously reported inhibition of MCU activity in the low cytoplasmic [Ca²⁺] regime.

Figure 3. MICU1 and MICU2 are required for matrix [Ca²⁺] regulation of MCU.

(A) Average traces (n = 3) of bath [Ca²⁺] showing inhibition of gatekeeping in MICU1-KO and MICU2-KO cells as evidenced by MCU-mediated Ca²⁺ uptake that reduces baseline bath [Ca²⁺] to much lower levels than wild-type (WT) cells, and ability of KO cells to rapidly take up Ca²⁺ following addition of 500 nM Ca²⁺ bolus.

(B and C) Basal Rhod-2 fluorescence in MICU1-KO (B) and MICU2-KO (C) cells.

(D) Quantification of Rhod-2 fluorescence (mean ± SEM, n = 3).

(E and F) Representative MCU currents in various bath [Ca²⁺] from mitoplasts from MICU1-KD (E) or MICU1-rescue (F) cells, with pipette solution containing 400 nM free Ca²⁺. Color scheme as in Fig. 1.

(G and H) Representative MCU currents in various bath [Ca²⁺] recorded from MICU2-KO (G) or MICU2 rescue (H) mitoplasts, with pipette solution containing 400 nM free Ca²⁺. Color scheme as in Fig. 1.

(I) Summary of MCU current densities over range of matrix [Ca²⁺] in mitoplasts from wild-type (WT) and MICU1 and MICU2 KD/KO and rescue cells.

(J and K) Basal Rhod-2 fluorescence in MICU1 (J) and MICU2 (K) rescue cells.

(L) Quantification of Rhod-2 fluorescence (mean ± SEM, n = 3).

See also Figure S2 and S4.

It was suggested that MICU1- (Csordas et al., 2013) or MICU2- (Patron et al., 2014) mediated inhibition of MCU activity does not require Ca²⁺ binding to their EF hands. To explore the role of cytoplasmic Ca²⁺ in EMRE regulation of MCU activity, we removed Ca²⁺ from the bath and recorded MCU activity by measuring Na⁺ currents (Figure 4A–C) (Fieni et al., 2012; Kirichok et al., 2004). Normal matrix [Ca²⁺] (400 nM) inhibition of MCU activity was eliminated in the absence of bath Ca²⁺ (Figure 4A–C), suggesting that Ca²⁺ binding in the intermembrane space, perhaps to MICU1 and/or MICU2 or MCU itself, is required for EMRE-dependent matrix [Ca²⁺] regulation of I_MiCa. It has been suggested that MICU2 suppresses MCU activity (Patron et al., 2014). However, comparable Na⁺ currents were recorded in MICU2 KO cells (Figure 4C).

Figure 4. Matrix [Ca²⁺] regulation of MCU requires cytoplasmic Ca²⁺.

(A) Representative MCU-mediated Na⁺ currents (red trace) recorded in absence of bath Ca²⁺ with no free Ca²⁺ in pipette. Addition of 160 nM free Ca²⁺ to bath completely blocked the Na⁺ current (orange trace).

(B) Same as in (A), recorded with pipette solution containing 400 nM free Ca²⁺.

(C) Summary of MCU Na⁺ current densities in different cell types with pipette solutions with or without 400 nM free Ca²⁺.

(D) Model for MCU regulation, with Ca²⁺ sensors on both sides of the IMM: EMRE and an unknown protein (box associated with MCU in the cartoon) sense matrix [Ca²⁺] ([Ca²⁺]_m); MICU1 and/or MICU2, through their paired EF hands (red dots), represent intermembrane space (ims) cytoplasmic Ca²⁺ ([Ca²⁺]_cyt) sensor(s). Under normal resting conditions (middle conformation), channel open probability is strongly reduced (gatekeeping) when sensors on both sides of the IMM are liganded with Ca²⁺ (red dots with question marks signify Ca²⁺ binding sites possibly liganded). Channel inhibition is effective in the range of [Ca²⁺]_m = 50–800 nM, with maximal inhibition at [Ca²⁺]_m ~ 400 nM (from I_MiCa biphasic matrix [Ca²⁺] dependence with apparent inhibition constant of 60 nM and apparent recovery constant of 730 nM; Figure 1D) with peak inhibition of MCU currents by ~75% at ~400 nM (25% open probability indicated by dashed line in middle conformation)), and [Ca²⁺]_cyt between resting levels and ~ 2 μM (from permeabilized cell measurements here and in (Mallilankaraman et al., 2012). Physiological relief of MCU channel inhibition is achieved by agonist-induced elevated [Ca²⁺]_cyt that drives Ca²⁺ into the matrix, raising [Ca²⁺]_m to levels (>800 nM) that result in highly cooperative activation (Hill coefficient > 3) of mitochondrial Ca²⁺ influx by unknown mechanisms (right conformation, with the unknown matrix reactivation mechanism(s) depicted as a Ca²⁺ binding protein with multiple Ca²⁺ binding sites, and MICU proteins with EF hands fully liganded). MCU channel inhibition can also be relieved by very low [Ca²⁺]_m (left conformation) as a result of Ca²⁺ unbinding from EMRE, independent of the Ca²⁺ binding states of MICU proteins (See also Figure S4B).

DISCUSSION

Our results suggest a novel model employing Ca²⁺ sensors on both sides of the IMM in which the activity of the mitochondrial uniporter is tuned to provide optimal Ca²⁺ influx into the mitochondrial matrix under resting conditions. EMRE is a sensor of matrix [Ca²⁺] that governs the inhibition of uniporter activity under normal conditions, acting to prevent mitochondrial Ca²⁺ loading under basal conditions. This function requires conserved acidic residues in the matrix-localized carboxyl tail of EMRE as well as MICU1 and MICU2 and Ca²⁺ on the opposite side of the IMM (Figures 4D and S4B). Of note, EMRE mediates the interaction of MICU proteins with MCU (Sancak et al., 2013), suggesting a possible mechanism for EMRE and MICU proteins to have coupled functions. EMRE-dependent matrix [Ca²⁺] regulation of I_MiCa may also play a role in preventing mitochondrial Ca²⁺ depletion when cytoplasmic [Ca²⁺] becomes low or mitochondrial Ca²⁺ extrusion exceeds influx (Figure 4D and S4B). Under such conditions, Ca²⁺ may dissociate from EMRE and possibly MICU proteins, relieving inhibition of MCU and enabling uniporter activity to increase to preserve matrix [Ca²⁺] to support critical mitochondrial functions (Cardenas et al., 2010). Thus, mitochondria are protected both from Ca²⁺ depletion under low-Ca²⁺ conditions and from Ca²⁺ loading under normal resting conditions by a unique molecular complex that involves Ca²⁺ sensors on both sides of the IMM, governed by EMRE. Our results suggest a model in which agonist-generated cytoplasmic Ca²⁺ signals raise matrix [Ca²⁺] due to the non-zero open probability of MCU under basal conditions, to a level (> 1 μM; Figure 1D that relieves EMRE-mediated matrix [Ca²⁺] inhibition of uniporter activity (Figures 1D, 4D, and S4B), enabling agonist-activated cytoplasmic Ca²⁺ signals to be conveyed to the matrix to stimulate respiration. The mechanism of matrix [Ca²⁺]-mediated relief from inhibition is highly cooperative (apparent Hill coefficient ~3, Figure 1D) and may therefore account for observed cooperative Ca²⁺ activation of the uniporter (Bragadin et al., 1979; Gunter and Pfeiffer, 1990; Zoccarato and Nicholls, 1982) without the need to invoke amplification by Ca²⁺ binding to MICU1 (Csordas et al., 2013; Perocchi et al., 2010). Importantly, the nature of the mechanism(s) involved in matrix [Ca²⁺] activation of the uniporter remain to be determined (Figure 4D and S4B).

EXPERIMENTAL PROCEDURES

Cell lines and constructs

Wild-type HEK-293T, MICU1-KO, MICU2-KO and EMRE-KO cell lines were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin at 37 °C and 5% CO₂. All knockout (KO) cell lines were gifts from Dr. Vamsi Mootha. MICU1 knockdown (KD) cells were generated by lentiviral transduction of shRNAs targeting MICU1; transduced cells were selected and maintained in complete DMEM supplemented with puromycin (2 μg/ml), as described (Mallilankaraman et al., 2012). Full length MICU1, MICU2 and EMRE cDNAs were purchased from Origene Technologies USA. EMRE-ΔC-V5 and EMRE-QN6-V5 constructs were generated by a PCR-based cloning strategy. All constructs were sequence verified and stably transfected into either KD or KO backgrounds.

Uniporter electrophysiology

Mitoplast electrophysiology was performed as described (Fieni et al., 2012; Kirichok et al., 2004; Vais et al., 2015). For I_MiCa measurements, patch pipettes with tip diameters significantly less than 1 μm had resistances of 20 – 60 MΩ when filled with (in mM): TMA-OH, 130 mM; Hepes, 100; glutathione, 10; MgCl₂, 2; EGTA, 1.5; pH = 7.0 with D-gluconic acid; osmolarity 330–350 mOsm/kg. EGTA was also used to buffer different amounts of CaCl₂, for [Ca²⁺]_free ≤ 600 nM. However, EGTA was replaced by 1.5 mM of: dibromo-Bapta (for 600 nM < [Ca²⁺]_free ≤ 4 μM), HEDTA (for 4 μM ≤ [Ca²⁺]_free ≤ 30 μM), or NTA (for [Ca²⁺]_free = 100 μM). In all pipette solutions, [Ca²⁺]_free (< 100 μM) was confirmed by Ca²⁺-indicator dye fluorimetry. For Na⁺ current measurements, pipettes were filled with (in mM): Na-gluconate, 110; Hepes, 40; NaCl, 2; EGTA, 1; EDTA, 5; CaCl₂, see above; pH = 7.0 with NaOH; osmolarity 330–350 mOsm/kg. Mitoplasts were initially bathed in (in mM): KCl, 150; Hepes, 10; EGTA, 1; pH = 7.2, osmolarity 300 mOsm/Kg (“KCl-DVF” solution). Voltage pulses of 350 – 500 mV amplitude and 15 – 50 ms duration, delivered by the PClamp-10 (Molecular Devices) program, were used to obtain the “whole-mitoplast” configuration. Access resistance (30 – 90 MΩ) and mitoplast capacitance C_m (0.2 – 1 pF) were determined using the membrane test protocol of the PClamp-10 software. Because the volume of a mitoplast is over three orders of magnitude smaller than that of a cell, we assumed that matrix [Ca²⁺] _free rapidly equilibrated with pipette [Ca²⁺] _free. After the whole-mitoplast configuration was obtained, the KCl-DVF bath solution was exchanged with: Hepes-EGTA [(in mM): Hepes, 150; EGTA, 1.5; pH = 7.0 with Tris-base for baseline (control) measurements, followed by Hepes (no EGTA) solutions with 0.1; 0.3; 1 mM CaCl₂, successively. Finally, a Hepes-based solution with 1 mM CaCl₂ and 200 nM ruthenium red (RuR) was perfused into the bath to record the final baseline (= I_RuR) after block of MCU Ca²⁺ currents. For MCU Na⁺ current measurements, the Hepes-EGTA control solution was instead exchanged with (in mM): Na-gluconate 110; Hepes, 40; EGTA, 1; EDTA, 5; pH = 7.0, followed by the same solution supplemented with 4 mM CaCl₂ ([Ca²⁺]_f ≈ 180 nM, estimated with the MaxChelator software), to block the Na⁺ currents. Osmolarities of all bath solutions were 297–305 mOsm/Kg, adjusted with sucrose. The voltage protocol, delivered by the PClamp-10 software with a DigiData-1550 interface (Molecular Devices), consisted of stepping from V_m = 0 mV to -160 mV for 20 ms, followed by ramping to 80 mV (rate of 279 mV/s), dwelling at 80 mV for 20 ms and return to 0 mV. Currents were recorded using an Axopatch 200-B amplifier. Data were acquired at room temperature with a sampling rate of 50 kHz and anti-aliasing filtered at 1 kHz. Data analysis was performed with the PClamp-10 software. For quantitative comparisons, current densities defined as:

$$I_{\mathit{MCU}}/C_m=(I_{Ca}-I_{\mathit{RuR}})/C_m$$

with I_Ca and I_RuR measured at V_m = −160 mV in the presence of 1 mM Ca²⁺ in the bath, were used. Wild-type data corresponding to various concentrations of free Ca²⁺ in the pipette solution (=[Ca_f]) were fitted with the biphasic Hill equation:

$$I_{\mathit{MCU}}/C_m={(I_{\mathit{MCU}}/C_m)}_{max}\left(1-\frac{{[{Ca}_f]}^{H_{\mathit{inh}}}}{{[{Ca}_f]}^{H_{\mathit{inh}}}+{K_{\mathit{inh}}}^{H_{\mathit{inh}}}}\times \frac{{K_{\mathit{rec}}}^{H_{\mathit{rec}}}}{{[{Ca}_f]}^{H_{\mathit{rec}}}+{K_{\mathit{rec}}}^{H_{\mathit{rec}}}}\right),$$

where H_inh and H_rec are the Hill coefficients for inhibition and recovery from inhibition, respectively, while K_inh and K_rec are the correspondent mid-point pipette [Ca_f]. Least-square fitting was done with IGOR Pro software (WaveMetrics).

Analysis of uniporter gatekeeping by simultaneous measurement of mitochondrial Ca²⁺ uptake and mitochondrial membrane potential in permeabilized cells

Simultaneous measurement of mitochondrial Ca²⁺ uptake and mitochondrial membrane potential (ΔΨ_m) in permeabilized HEK293 cells was performed as described (Mallilankaraman et al., 2012), with experimental data converted to and represented as bath [Ca²⁺]. In brief, cells were trypsinized, counted and washed in a Ca²⁺-free buffer, centrifuged and transferred to an intracellular-like medium (in mM: 120 KCl, 10 NaCl, 1 KH₂PO₄, 20 HEPES-Tris, 2 succinate, pH 7.2, protease inhibitors, 2 μM thapsigargin, 40 μg/ml digitonin) in a spectrofluorometer. FuraFF (0.5 μM) was added at 0 s and JC-1 (800 nM) at 20 s to measure extra-[Ca²⁺]_m and ΔΨ_m at 37°C. Fluorescence was monitored in a temperature-controlled (37°C) multi-wavelength-excitation dual wavelength-emission spectrofluorometer (Delta RAM, Photon Technology International) using 490-nm excitation/535-nm emission for the monomer, 570-nm excitation/595-nm emission for the J-aggregate of JC1and 340-nm/380-nm for FuraFF. In this study, the size of the Ca²⁺ boluses added to stimulate subsequent mitochondrial Ca²⁺ uptake did not change ΔΨ_m.

Qualitative measurement of basal [Ca²⁺]_m

Qualitative measurement of basal [Ca²⁺]_m were determined as we previously described (Mallilankaraman et al., 2012). Briefly, cells (5×10⁵/well) grown on 0.2% gelatin-coated coverslips in six-well dishes were loaded with 2 μM rhod-2 AM for 50 min in extracellular medium containing 2 mM Ca²⁺. Coverslips were mounted in an open perfusion microincubator (PDMI-2; Harvard Apparatus) and imaged at 37 °C. All confocal images, obtained at 561 nm excitation using a 63× oil objective, were rapid 100 msec “snapshots” to avoid phototoxicity. All imaging used identical gain and exposure settings that enabled threshold detection of basal Rhod-2 fluorescence in control WT cells. Accordingly, the contribution of low-level diffuse background cytoplasmic Rhod-2 was essentially eliminated. Addition of the mitochondrial uncoupler CCCP (2–10 μM) was added to ensure that punctate fluorescence emanated from mitochondria (Figure S4). Images were analyzed using ImageJ software (NIH) as described (Mallilankaraman et al., 2012).

Topology analysis

Topology analyses of MICU1, MICU2 and EMRE were performed by proteinase K protection assay as described (Sato et al., 1998). Mitochondria were isolated by differential centrifugation as described (Sato et al., 1998). Immediately after isolation, the mitochondrial pellet was resuspended in MIM buffer (280 mM sucrose with 10 mM Hepes; pH 7.2). Mitochondria (40 μg per condition) were placed in MIM buffer containing varying concentrations of digitonin (0–2%) and constant concentration of Proteinase-K (100 μg/ml) for 15 min at room temperature. Samples without proteinase-K or with 1% Triton X-100 served as controls. Proteinase-K was inactivated by addition of 10 μM PMSF, and samples were subjected to Western or dot blotting.

Western blotting

Cells kept on ice were lysed with 1x RIPA lysis buffer (Millipore) supplemented with protease inhibitors (Complete, Roche). Membranes were blocked in 5% fat-free milk 1 hr at RT, incubated O/N at 4 °C with primary antibody, and then for 2 hr at RT with a secondary antibody conjugated to horseradish peroxidase (HRP). Chemiluminescence detection used ECL-plus reagent (Pierce). To ensure equal protein loading across gels, membranes were stripped (Restore western blot stripping buffer; Pierce) and re-probed with a loading control antibody. For dot blots, protein samples were spotted onto nitrocellulose membranes using a slot blot apparatus and immunoblotting was performed as described above. Antibodies for: MICU1 (ab102830, abcam; 1:200) and (HPA037480, Sigma; 1:300); MICU2 (ab101465, abcam 1:500) and (HPA045511; Sigma); MCU (HPA016480, Sigma; 1:500); EMRE (sc-86337; Santacruz) and (HPA032117, Sigma; 1:200); Tom20 (ab56783; abcam); Tim23 (611223; BD Biosciences); Tim44 (ab168649; abcam); Oxa1L (66128-1-Ig; proteintech); HSP60 (ab13532; abcam); anti-V5 (R960-25, Life technologies; 1:1000); anti-actin (A2228, Sigma 1:2000); anti-flag (TA 50011, Origene 1:500); anti-mouse IgG-HRP (7076S, Cell Signaling; 1:1000) and anti-rabbit IgG-HRP (7074S, Cell Signaling; 1:1000) were used in the study. All experiments were done in triplicate.

Co-immunoprecipitation

HEK293 cells (EMRE KO, EMRE KO + QN6 EMRE-V5 and EMRE KO + wild-type EMRE-V5) were grown to confluency in 150 mm plates. Cells were washed twice with PBS and lysed with 1 ml cold lysis buffer (50 mM HEPES-KOH pH 7.4, 150 mM NaCl, 0.2% n-dodecyl-β-D-maltoside, 0.5 mM EGTA, 0.3 mM CaCl₂, one Complete protease inhibitor tablet (Roche) per 30 ml lysis buffer). Cells were lysed at 4°C for 30 min and centrifuged at 16,000 × g. Cleared lysates were diluted to normalize for protein concentration. Appropriate volumes of lysates were added, corrected for expression levels based on western blotting and densitometric analysis. Anti-V5 Agarose (30 μl; Sigma #A 7345) was added to diluted lysates and rocked 16 hr at 4°C. Beads were washed with 1 ml PBS 3x and boiled in 90 μl 2x SDS sample buffer. 20 μl of each immunoprecipitate and 30 μg of total cell lysates were loaded on Tris-Glycine gels for Western Blotting. To maintain a constant range [Ca²⁺] in lysis and wash buffers, CaCl₂ and EGTA were added to buffers according to http://maxchelatorStanford.edu/CaEGTA-TS.htm. In addition, [Ca²⁺] in each preparation was measured with Fura-2. Solutions contained 150–450 nM free Ca²⁺.