Patch-clamp technique to study mitochondrial membrane biophysics

Regulated ion flux across biological membranes is essential for maintaining many fundamental physiological processes such as electrochemical gradient of ions, membrane excitability, pH homeostasis, transport of metabolic substrates, and several energy transduction processes (e.g., ATP production in mitochondria; Hille, 1992). A high-resolution electrophysiological analysis of these fluxes by patch clamp is feasible, provided there is a net transfer of an electrical charge across the membrane (i.e., ion transport is electrogenic) and protein mediating the ion flux is expressed in sufficient amounts. These patch-clamp investigations have shed light not only on the mechanism of ion transport but also helped understand the structure–function relationship of many membrane proteins. Though challenging, patch-clamp technique has been successfully applied directly to several subcellular organelles like endolysosomes, nucleus, mitochondria, plant vacuoles, and chloroplast (Sorgato et al., 1987; Hedrich and Kurkdjian, 1988; Schönknecht et al., 1988; Mazzanti et al., 1990; Kirichok et al., 2004; Saito et al., 2007). In this regard, mitochondrial membranes present a unique case because there are two of them, the outer mitochondrial membrane (OMM) and the inner mitochondrial membrane (IMM), each with its own distinct biophysical properties.

Besides efficient ATP production, mitochondria are crucial for the synthesis of important metabolic precursors for lipids, proteins, DNA, and RNA (Spinelli and Haigis, 2018). Considered to be a relic from an α-proteobacteria ancestor, the IMM likely originated from the bacteria, while the OMM is believed to have been derived from the eukaryotic host (Sagan, 1967). Separated by an intermembrane space (IMS), these membranes are rich in numerous ion channels and carriers involved in the flux of ions, metabolites, lipids, and other signaling factors (Fig. 1 A). The OMM acts as a sieve, housing numerous β-barreled proteins, including well-known porins or voltage-dependent anion channels (VDAC) that are porous for most hydrophilic molecules (up to ∼5 kD, Fig. 1 A; Colombini, 1979; Mannella, 1998; Ujwal et al., 2008). VDAC has been characterized electrophysiologically by patch-clamp recordings of reconstituted proteoliposomes or planar lipid bilayer (often 1,2-didiphytanoyl-snglycero-3-phosphocholine and cholesterol-based). These recordings demonstrated that although VDAC has a predominant anion-selective open state (in the range ∼600–700 pS, tested using ATP or Cl⁻), a few moderately cation-selective (K⁺ or Ca²⁺) substates are also present (Blachly-Dyson et al., 1993; Gincel et al., 2001; Najbauer et al., 2021). These substates, generally referred to as the “closed” or “partially closed” states, exhibit lower overall conductance and are almost impermeable to ATP. Pavlov et al. (2005) also reported 600–680 pS “open” cation-selective state of the VDAC, although this state occurs at low frequency. Other β-barreled proteins, such as translocator of outer membrane 40, sorting assembly machinery proteins, and mitochondrial import 1 complex, function as protein translocators but may also conduct cations as shown by planar lipid bilayer recordings (Krüger et al., 2017).

In contrast to OMM, the IMM has a large surface area (estimated to be 3–10 times larger than the OMM) due to extensive folding of the membrane termed cristae. Relative to other membranes, the IMM contains high levels of a signature lipid called cardiolipin (Fig. 1 B, red color). Cardiolipin is a diphosphatidylglycerol molecule with four acyl chains (these acyl chains are primarily linoleic acid in the heart mitochondria) that is essential for the optimal functioning of many IMM proteins (Falabella et al., 2021). The IMM is highly selective and its tight permeability is crucial for maintaining the transmembrane electrochemical gradient of H⁺ (ΔΨ = approximately −160–180 mV and ΔpH = ∼0.5–0.8 pH units [that corresponds to ∼30–50 mV]), also called protonmotive force. The protonmotive force contributes to a total membrane potential of ∼210 mV depending on the metabolic state of the mitochondria. This protonmotive force is generated by the proton-pumping complexes of the electron transport chain and is coupled to ATP synthesis by ATP-Synthase or thermogenesis mediated by uncoupling protein 1 (UCP1) in brown fat. It is also utilized by many ion channels and carriers for the active or passive flux of ions across the IMM.

The selective permeability of IMM for Ca²⁺ was identified during the same time period as the formulation of the chemiosmotic theory (Deluca and Engstrom, 1961; Carafoli et al., 1964). Energized mitochondria in suspension or in intact cells sequester substantial Ca²⁺ when extramitochondrial [Ca²⁺] is elevated either via a bolus of Ca²⁺ or during cytosolic transients, respectively. The matrix Ca²⁺ uptake in turn controls the rate of ATP production, buffers cytosolic Ca²⁺, and maintains redox homeostasis (McCormack et al., 1990; Glancy and Balaban, 2012; Finkel et al., 2015; Wescott et al., 2019). Although several different pathways have been suggested to mediate mitochondrial Ca²⁺ influx and efflux (Gunter and Pfeiffer, 1990; Hamilton et al., 2018; Fig. 1 A), the mechanism of an acute, electrogenic uptake of cytosolic Ca²⁺ by a tetrameric ion channel, mitochondrial calcium uniporter complex (MCUcx, Fig. 1 B), is perhaps the only one that is most firmly established.

In 2004, using direct whole-mitoplast recordings, the Clapham group (Kirichok et al., 2004) conclusively demonstrated that mitochondrial calcium uniporter is an inwardly rectifying, ruthenium red-sensitive, and highly selective Ca²⁺ channel. This seminal study not only comprehensively characterized the biophysical properties of the uniporter but also established the whole-mitoplast patch-clamp technique on a solid foundation, which has since been widely adopted by many as the gold standard method for studying mitochondrial ion channels and transporters. Previous studies were confined primarily to single-channel recordings of easily resolvable large-conductance channels, frequently encountered in excised-patches of the IMM (Kinnally et al., 1989; Petronilli et al., 1989; Antonenko et al., 1991; Inoue et al., 1991). Few attempts to perform whole-mitoplast recordings in the past (Sorgato et al., 1987; Klitsch and Siemen, 1991; Borecký et al., 1997) had limited success due to the small size, fragility of the mitoplasts (absence of any internal skeleton), and perhaps the inconsistent quality of the mitoplast preparation. Kirichok et al. (2004) overcame these barriers by optimizing the isolation procedure, mitoplast visualization, and break-in protocol for gaining access to the matrix while maintaining seal integrity. The Kirichok group has since refined the whole-mitoplast patch clamp, enabling them to record several other conductances of the IMM with exceptional precision and gain insights into the molecular mechanisms that govern mitochondrial bioenergetics (Fedorenko et al., 2012; Fieni et al., 2012; Bertholet et al., 2019; Garg et al., 2021; Bertholet et al., 2022).

Within a decade of the first whole-mitoplast patch-clamp recording of the Ca²⁺ uniporter, the groups of Mootha and Rizzuto successfully cloned several proteins that makeup the core uniporter complex, including the pore-forming MCU subunit, an essential accessory subunit Essential MCU Regulator, a dominant negative subunit MCUb, and several EF-hand domain-containing proteins termed MICU1-3 (Mitochondrial Calcium Uptake 1-3; Perocchi et al., 2010; Baughman et al., 2011; De Stefani et al., 2011; Chaudhuri et al., 2013; Plovanich et al., 2013; Raffaello et al., 2013; Sancak et al., 2013). Although not an integral component of the core MCUcx complex, other proteins such as SLC25A23 and MCUR1 (Mitochondrial Calcium Uniporter Regulator 1) modulate MCUcx-mediated Ca²⁺ currents through mechanisms that are currently unresolved (Hoffman et al., 2014; Vais et al., 2015). Recent structural characterization (Fan et al., 2020; Wang et al., 2020a; Wang et al., 2020b; Zhuo et al., 2021) informs that MCUcx is indeed a macromolecular protein complex, often appearing as a “Supercomplex” (a dimer of two MCUcx pores, just like ATP synthase; Blum et al., 2019), as also predicted by the Blue Native PAGE in earlier studies (Baughman et al., 2011; Sancak et al., 2013). Despite the availability of high-resolution structures of MCUcx, many aspects of its structure–function relationship, gating, and the precise role of accessory subunits are yet to be fully understood, and most of these are only amenable to patch-clamp investigations.

In contrast to MCUcx, the identity and physiological significance of other mitochondrial Ca²⁺ flux mechanisms (Fig. 1 A), as well as whether the flux is electrogenic or not, are still the subjects of active debate. LETM1 (leucine zipper-EF-hand–containing transmembrane protein 1) was proposed to be the Ca²⁺/H⁺ exchanger mediating pH-driven rapid Ca²⁺ influx at low, nM cytosolic [Ca²⁺] (followed by slower uptake mediated by a uniporter mechanism at [Ca²⁺] > 1 µM), while cytosolic acidification led to Ca²⁺ efflux in permeabilized cells (Jiang et al., 2009). This was later supported by studies on purified LETM1, reconstituted in proteoliposomes by Tsai et al. (2014). These experiments showed that LETM1 is an electroneutral Ca²⁺/H⁺ antiporter with a K_m of ∼25 µM for Ca²⁺ turnover. However, other groups have argued against the proposal that Ca²⁺/H⁺ exchange is the primary function of LETM1. Using optical methods in isolated mitochondria and permeabilized cells, it was suggested that LETM1 primarily mediates K⁺/H⁺ transport across the IMM and plays an important role in mitochondrial volume homeostasis (Nowikovsky and Bernardi, 2014; Austin et al., 2017). Recently, two groups (Austin et al., 2022; Patron et al., 2022) simultaneously proposed a LETM1-interacting mitochondrial protein, MICS1 (GHITM), as the Ca^2+/H⁺ exchanger, which is functional when reconstituted in liposomes and shows channel-like activity in planar lipid bilayer (Patron et al., 2022). Similarly, in some tissues like brain and heart, a Na⁺-dependent Ca²⁺ efflux mechanism plays a predominant role (Gunter and Pfeiffer, 1990). Using fluorescence-based mitochondrial Ca²⁺ uptake assays, this Ca²⁺ efflux is proposed to be mediated by NCLX, a Na⁺- or Li⁺-dependent Ca²⁺ exchanger (SLC8B1; Palty et al., 2010; Boyman et al., 2013). Of interest, a few studies noted substantial extramitochondrial localization of the NCLX protein and a disconnect between the expression and function in native mouse tissues (Cai and Lytton, 2004; Han et al., 2015; Rysted et al., 2021).

In addition, single-channel recordings of excised or mitoplast-attached patches have identified a macrochannel (∼1–1.3 nS in 150 mM KCl) that is gated by Ca²⁺, voltage, pH, and the redox state of the mitochondria (Kinnally et al., 1989; Petronilli et al., 1989; Carraro and Bernardi, 2023). Other conductances between 30 pS and 1 nS were also identified, some of which may represent substates of the macrochannel. The macrochannel is often correlated with mPTP (a pathological increase in IMM permeability), first described by Haworth and Hunter (1979), using mitochondrial swelling or light scattering assay. Although several candidate genes, such as ADP/ATP carrier (AAC) and ATP synthase, have been suggested, there is currently no consensus due to the lack of direct conclusive evidence (Bauer and Murphy, 2020; Carraro and Bernardi, 2023).

Since the unitary conductance of carriers is at least three orders of magnitude smaller than most ion channels (Hille, 1992), direct electrical analysis of electron transport chain complexes or most of the mitochondrial carriers has not been reported. Of the 53 members of the mitochondrial solute carriers (SLC25) family, only UCP1- and AAC-mediated currents have been recorded with patch clamp so far, likely due to extremely high density of these proteins in the IMM (∼10% of total mitochondrial protein). Whole-mitoplast patch clamp was successfully applied to show that UCP1 is a fatty acid anion/H⁺ symporter, essentially operating as an H⁺ uniporter in the presence of long-chain fatty acid (FA; where the substrate itself, i.e., a tightly-bound FA, translocates an H⁺ across the membrane via an alternating access mechanism; Fedorenko et al., 2012; Kunji et al., 2020). Similarly, AAC-mediated electrogenic exchange of Mg²⁺-free ADP³⁻ with ATP⁴⁻ was reliably recorded by patch clamp (Bertholet et al., 2019). In addition to its canonical role of ADP/ATP exchange, it was also shown that AAC mediates H⁺ currents in the presence of FA (Bertholet et al., 2019) and thus plays an important thermogenic role in non-adipose tissues. Recently, the same group reported that many of the mitochondrial uncouplers or protonophores like 2,4-dinitrophenol and cyanide-4-(trifluoromethoxy) phenylhydrazone also mediate H⁺ current across the IMM via their selective action on the AAC and UCP1 (Bertholet et al., 2022).

Several groups have reported cation channels in the IMM, with varying K⁺ conductances often recorded in the single-channel mode (reviewed in Checchetto et al., 2021 and Kulawiak and Szewczyk, 2022). Inoue et al. (1991) discovered a small-conductance (∼10 pS in 100 mM cytosolic-side [K⁺] and 33.3 mM matrix-side [K⁺]), ATP-sensitive K⁺ channel in excised inside-out patches from mouse liver mitoplasts. Three additional ATP-insensitive non-selective large conductances (between 50 and 200 pS) were also reported (Inoue et al., 1991). These mitoplasts were generated by subjecting them to digitonin treatment and hypotonic swelling to remove the OMM and then fusing them by incubation at low pH and elevated temperatures in the presence of Ca²⁺. Whether such treatments result in activation of certain dormant conductances, changes in IMM biophysical properties, or fusion with non-IMM vesicles is not clear. In the absence of exact molecular correlates (and in most cases selective pharmacological modulators) for these single-channel conductances, it is challenging to relate them to the whole-mitoplast currents or their physiological role in the intact mitochondria or the cell. Using single-channel patch-clamp or lipid-bilayer recordings, a diverse set of genes have been implicated in mediating mitochondrial K⁺ conductance(s) e.g., KCNA3 (Szabò et al., 2005), KCNMA1 (Singh et al., 2013), CCDC51 (Paggio et al., 2019), and KCNJ1 (Laskowski et al., 2019). Teasing out their exact role in mitochondrial physiology, tissue-specific expression and density at the IMM, mechanism of dual targeting to the plasma membrane and mitochondria, and IMM-specific biophysical properties is still a work in progress. Nonetheless, specific K⁺ and anion conductances have been identified via the whole-mitoplast patch clamp, although their molecular identity still remains a mystery (Sorgato et al., 1987; Klitsch and Siemen, 1991; Borecký et al., 1997; Fieni et al., 2012).

In addition to mitochondrial conductances discussed above, many organelles that closely interact with mitochondria also harbor specific ion channels that in turn regulate mitochondrial ion flux (Fig. 1 A). The ER for instance, serves as a Ca²⁺ reservoir, with the sarco/ER Ca²⁺ ATPase pump actively transporting Ca²⁺ from the cytosol to the ER. Close proximity of mitochondria to ryanodine receptors on junctional sarcoplasmic reticulum in cardiomyocytes enables brief exposure to high local [Ca²⁺] during each heartbeat. At mitochondria–ER membrane contact sites, Ca²⁺ transfer to mitochondrial surface is mediated via inositol 1,4,5-trisphosphate channels (Giacomello and Pellegrini, 2016). Similarly, lysosomal TRPML1 channels are proposed to mediate Ca²⁺ transfer at mitochondria–lysosome contact sites (Wong et al., 2018). Although not an issue for whole-mitoplast patch-clamp studies, it is often challenging to get rid of these associated membranes and get a relatively pure mitochondrial fraction for bioenergetic assays.

Here, we discuss several aspects of whole-mitoplast patch-clamp methodology as applicable to direct investigation of mitochondrial membrane biophysics using two prototypical conductances mediated by MCUcx and UCP1. We also discuss the technical limitations and advantages of mitochondrial patch clamp compared with other methodologies.

All animal experiments were performed according to the Institutional Animal Care and Use Committee approved protocols and adhered to National Institutes of Health standards. Mice (C57BL6, 6–8 wk old) were euthanized by CO₂ asphyxiation followed by cervical dislocation. After trimming the adrenal glands, both kidneys were extracted and then placed in 5 ml ice-cold Initial medium. The rest of the steps were performed at 4°C. The tissue was minced into pieces using fine scissors and washed twice with Initial medium to remove the excess blood. The minced tissue was resuspended in 5 ml ice-cold Initial medium and homogenized using a glass grinder (Potter-Elvehjem homogenizer, 10 ml capacity, Wheaton) with slow strokes (7–8 times) of a Teflon pestle rotating at 280 rpm. The homogenate suspension was centrifuged at 700 × g for 7 min. The supernatant, containing mitochondria, was collected in a separate tube. The remaining pellet was resuspended in 5 ml of fresh Initial medium and the suspension was homogenized and centrifuged again at 700 × g for 7 min. The supernatant from two rounds of homogenization and centrifugation was combined. Mitochondria were collected by centrifugation of the supernatant at 3,200 × g for 10 min. The pellet containing mitochondria was stored on ice for further use. Similarly, respective tissues (brown fat, heart, liver, etc.) were isolated from the mice and placed in the ice-cold Initial medium. The next steps in mitochondria isolation from these tissues are the same as mentioned above.

We routinely use HEK293T cells or MEFs that are deficient in Dynamin-1–like protein (DRP1-KO; Garg et al., 2021) for the experiments. Cells were maintained in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin at 37°C in a humidified atmosphere (5% CO₂). 2 d before the experiment, 1 × 10⁶ cells were plated in three 150-mm dishes. On the day of the experiment, the cells were rinsed with 10 ml of ice-cold Mg²⁺-free and Ca²⁺-free PBS. All the steps were performed at 4°C. Further, 10 ml PBS was added and the cells were scraped with a gentle blade cell scraper (Sarstedt 83.1831). Cells were spun down at 700 × g for 5 min and the supernatant was discarded. Cells were resuspended in 5 ml of Initial medium and transferred to an ice-chilled 10 ml Wheaton glass homogenizer (Teflon pestle). Cells were homogenized on ice with 5–6 slow strokes at 280 rpm and centrifuged at 700 × g for 7 min. The supernatant was stored in a separate tube while the cell pellet was resuspended in fresh 5 ml Initial medium. After one to two more homogenization and centrifugation steps (helps to eliminate nuclei-mitochondria aggregates and ruptures remaining cells to increase yield the of mitochondria), supernatants were combined and centrifuged at 3,200 × g for 10 min to pellet mitochondria.

For the mitoplast preparation, the isolated mitochondria pellet was resuspended in ∼4 ml of mannitol hypertonic solution. After 15 min incubation, mitochondria were subjected to mechanical disruption of OMM using Thermo Electron French press (FA-078A) with mini cell (FA-003 20) at ∼120 psi (gauge reading, calculated internal pressure in mini-cell with 3/8-inch piston ∼1,500 psi) to rupture the outer membrane as shown in Fig. 2 A. French press procedure is preferred over other procedures for the preparation of mitoplast because we can isolate mitoplasts with fully preserved integrity, including the matrix and cristae (Decker and Greenawalt, 1977). The mitochondrial suspension is loaded into a 4 ml prechilled mini-pressure cell (3/8-inch piston diameter) of the French press. Following 1 min of sustained pressure at 105 psi (gauge reading), the mini-cell valve was slowly opened and the mitoplast suspension was collected at the rate of one drop per second in a prechilled tube. Mitoplasts were pelleted at 3,900 × g for 15 min and the pellet was resuspended for storage in 0.5–1 ml of a KCl storage solution. All the steps were performed at 4°C, and mitoplasts were stored on ice for up to ∼4–6 h. Immediately before the electrophysiological recording experiments, 20–50 μl of the mitoplast suspension was added to 500 μl of KCl bath solution and plated on 5-mm coverslips pretreated with 0.1% gelatin (Cat.# ES-006-B; Millipore) to reduce mitoplast adhesion.

Thick-walled borosilicate glass capillaries with filament (BF150-86-7.5, Sutter Instruments) were used to prepare the recording pipettes for patch-clamp experiment. On the day of recording, pipettes with small tips were pulled using a micropipette puller (P-1000, Sutter Instruments) and polished using a microforge (MF 830; Narishige). The pulled recording pipettes were stored in a closed container to prevent the accumulation of any dust particles, which could interfere with seal formation. Polished pipettes were used on the same day.

Our patch-clamp rig consists of a vibration isolation table (Newport or TMC), a custom-made Faraday cage, an inverted microscope (IX73; Olympus), an amplifier (Axopatch 200B, Molecular Devices) with headstage (CV 203BU, Molecular Devices), digitizer (Digidata 1550, Molecular Devices), small bath volume (50 μl) recording chamber (RC-24E, Warner Instruments) that permits fast solution changes, gravity-driven perfusion system, data acquisition and analysis software (pClamp 11 software), and operated by the computer running Windows 10. The microscope has Olympus UPlan super apochromatic 60× water immersion objective (numerical aperture 1.20), with a correction collar between 0.13 and 0.21 mm (that enables clear visualization of outer and inner lobes of mitoplasts) and air-based 40×, 20× objectives capable of differential interface contrast (DIC) imaging and equipped with fluorescence light source and filter sets for GFP and mCherry. The electrode maneuvering is enabled by a Sutter micromanipulator system that consists of an MPC-200 controller, an MPC-225 mechanical manipulator, and ROE-200 (Sutter Instruments).

Mitoplasts are generally 2–6 µm in size and thus require a high-resolution optical setting in the microscope for visualization. During patching, a 5-mm coverslip previously plated with mitoplast is placed in a 50-μl recording chamber. For patching, an eight-shaped mitoplast (light pink colored lobe corresponding to IMM and green colored lobe is OMM; Fig. 2, A and C) will form a quick and tight seal. The recording pipette was filled with Na-gluconate or tetramethylammonium hydroxide (TMA) pipette solution and attached to the head stage. During the period just prior to seal formation, positive pressure was applied to the pipette to keep it free of debris. The pipette is bought into close proximity (on the top) to the selected mitoplast using the coarse setting of the micromanipulator, while a finer setting is used for the final approach when the pipette is moved downward until it touches the edge of the mitoplast inner membrane from directly above. A gentle suction or negative pressure with a syringe is applied for a high-resistance Gigaohm seal (>1–3 GΩ; Fig. 2 C and Fig. 4 E). After compensating for pipette capacitance (Fig. 4 E), we rupture the patch to attain the whole-mitoplast configuration (Fig. 2 D and Fig. 4 F). We find the most effective technique to get access to the matrix is to apply break-in protocol, a sharp pulse of suction or gradual ramping of pressure, 5–15 ms of 400–600 mV pulse (using the high-voltage range of the amplifier), or a combination of suction and voltage pulses. We observed that different preparations (different tissue or cell lines) might require different break-in protocols or optimizations individually. A quick increase in the capacitive transient accompanying the test voltage step indicates a successful break-in (Fig. 2 D and Fig. 4 F). Seal formation can also be checked by applying the ramp protocol in KCl when an outward current primarily due to Cl⁻ inward movement at positive potentials can be clearly discerned.

KCl (Cat# 409316; Sigma-Aldrich), HEPES (Cat# H7523; Sigma-Aldrich), EGTA (Cat# 03780; Sigma-Aldrich), Trizma base (Cat# T1503; Sigma-Aldrich), sodium D-gluconate (Cat# G9005; Sigma-Aldrich), TMA (Cat# 331635; Sigma-Aldrich), EmbryoMax 0.1% gelatin solution (Cat# ES-006-B; EMD Millipore), D-gluconic acid (Cat# G1951; Sigma-Aldrich), D-mannitol (Cat# M4125; Sigma-Aldrich), sucrose (Cat# S7903; Sigma-Aldrich), guanosine 5′-diphosphate Tris salt (GDP, Cat# G7252; Sigma-Aldrich), Oleic acid (Cat# O1008; Sigma-Aldrich), and methyl-β-cyclodextrin (Cat# C4555; Sigma-Aldrich).

Analytical grade reagents were used for the preparation of the various solutions. All buffers were prepared using deionized water and stored at 4°C.

260 mM sucrose, 20 mM HEPES, 1 mM EGTA, and 0.1% BSA (pH adjusted to 7.2 with Trizma base).

140 mM sucrose, 440 mM D-mannitol, 5 mM HEPES, and 1 mM EGTA (pH adjusted to 7.2 with Trizma base).

750 mM KCl, 100 mM HEPES, and 1 mM EGTA (pH adjusted to 7.2 with Trizma base).

150 mM KCl, 10 mM HEPES, and 1 mM EGTA (pH adjusted to 7.0 with Trizma base).

150 mM HEPES, 80 mM sucrose, and 1 mM EGTA (pH 7.0 with Trizma base, tonicity ∼300 mmol/kg with sucrose).

150 mM HEPES, 80 mM sucrose, and 100 µM–5 mM [Ca²⁺] (pH 7.0 with Trizma base, tonicity ∼300 mmol/kg with sucrose).

110 mM Na-gluconate, 40 mM HEPES, 10 mM EGTA, and 2 mM MgCl₂ (pH 7.0 with NaOH) tonicity ∼350 mmol/kg.

130 mM TMA, 1.5 mM EGTA, 2 mM magnesium chloride, and 100 mM HEPES (pH adjusted to 7.2 with D-gluconic acid, tonicity adjusted to ∼367 mmol/kg with sucrose).

130 mM TMA, 1 mM EGTA, 2 mM magnesium chloride, 100 mM HEPES, (pH adjusted to 7.0 with D-gluconic acid, tonicity adjusted to ∼360 mmol/kg with sucrose).

Ion flux across mitochondrial membranes is traditionally determined by using ion-sensitive electrodes or optical methods like spectrophotometric and fluorescence microscopy of isolated mitochondrial suspensions, permeabilized, or intact cells. However, these methods are semiquantitative at best as there is no control over crucial parameters like matrix pH, ion gradients across the membranes, and changes in ΔΨ. Electrophysiological recordings can be performed on reconstituted proteoliposomes or artificial planar lipid bilayers, although concerns remain regarding the native composition and conformation of the membrane lipids and protein complexes, and the risk of contaminating proteins. Thus, mitochondrial patch-clamp remains the most reliable, direct, and quantitative way to characterize ion channel function at high resolution and under native conditions.

Sorgato et al. (1987) made the first successful attempt to patch IMM in the whole-mitoplast and mitoplast-attached mode by recording a 108 pS anion channel in symmetrical, i.e., same bath and pipette KCl solutions. They used an osmotic swelling-shrinking-sonication procedure to disrupt the OMM and get access to the IMM. Kirichok et al. (2004) used the similar method without sonication to prepare mitoplasts and demonstrated that MCUcx is a highly selective Ca²⁺ channel, not just a non-specific IMM permeability induced by Ca²⁺. Later on, Fieni et al. (2012) used French press to mechanically break the OMM and isolate mitoplasts from several mouse tissues and observed improved success rate for reliable and stable whole-mitoplast recordings. Mitoplasts prepared by French press not only retain the structural and functional integrity for ∼4 h (Decker and Greenawalt, 1977) but also retain the cristae structure or “condensed” configuration while in the hypertonic medium (Decker and Greenawalt, 1977).

We routinely use French press to subfractionate mitochondria (Garg and Kirichok, 2019; Garg et al., 2021). This not only avoids the use of detergents at any step but also provides higher yield of consistent quality mitoplast preparation each time in comparison to mitoplasts prepared by hypoosmotic shock. In this method, mitochondria are first suspended in a D-mannitol/sucrose-based hyperosmotic medium that helps in separation of IMM and OMM, thus increasing the intermembrane distance. This suspension is then forced to pass through a needle valve in the French press mini-cell under controlled pressure (gauge reading ∼120 psi, calculated internal pressure in mini-cell with 3/8-inch piston ∼1,500 psi, Fig. 2 A). The loose OMM is disrupted as the suspended mitochondria experiences shear stress and decompression while coming out of the narrow valve (one can optimize the pressure for different purposes and different tissue samples). The primary goal is to apply just enough pressure to disrupt a portion of the OMM, giving space for the IMM to protrude out once medium is changed back to the isotonic conditions. These mitoplasts when observed with DIC optics using a 60× water immersion objective appear eight-lobed, an optically dense lobe corresponding to OMM, and a less dense lobe corresponding to IMM (Fig. 2, C and D). Both phase contrast microscopy (traditionally used in patch-clamp rigs) and DIC microscopy can be used to visualize transparent specimens, such as cells and subcellular organelles. However, DIC generally provides better axial resolution when there is only a slight difference in the refractive index in different parts of the specimen, such as the two lobes of a mitoplast. In a phase contrast microscope, the two lobes of the mitoplast are difficult to distinguish, and abrupt changes in refractive index at the edges of the specimen can introduce distortions in the image, resulting in a “halo” around the edges (see Fig. 1 C in Sorgato et al., 1987). In contrast, DIC detects edges as a gradual shift in refractive index, resulting in images with high contrast and axial resolution, ultimately resulting in a pseudo three-dimensional image of the mitoplast lobes (Fig. 2, C and D). Additionally, the use of a phase plate in phase contrast microscopy limits the effective numerical aperture of the microscope’s condenser and objective lenses due to the masking effect, which is not the case in DIC microscopy.

Fig. 3 shows the electron microscopy images of the isolated mitochondria (Fig. 3 A) and mitoplasts (at different stages of swelling, Fig. 3 B) from the mouse heart, prepared sequentially by differential centrifugation and French press, and then suspended in isotonic KCl bath solution. Usually, the remnants of the OMM remain stuck to the side of mitoplasts after disruption (Fig. 2 D and Fig. 3 B, blue arrow), appearing as a “cap” if the IMM is fully swollen or as the denser lobe in an eight-shaped structure (Fig. 2 C and Fig. 3 B, red arrow). Mitoplasts prepared above are kept in a KCl storage medium to retain condensed configuration for ∼4–5 h. Approximately 15 min before the experiment, a small aliquot is diluted with an isotonic KCl bath solution and plated onto a coverslip. The plated mitoplasts are further subjected to patch clamp to record different ion conductances of the IMM.

To make mitoplast patch pipettes, we use thick-walled borosilicate glass capillaries (with an outer diameter of 1.5 mm and an inner diameter of 0.86 mm) with a filament. Borosilicate glass is preferred due to its low dielectric constant, low dissipation factor, and good relaxation properties that results in low noise. Additionally, borosilicate glass maintains a consistent inner diameter and outer diameter ratio over the length of a taper when pulled. Using a filamented capillary is a must when creating high-resistance electrodes with small tips. The filament annealed to the inner wall of the capillary allows for easy backfilling of the solution and reduces the chance of introducing air bubbles into the pipettes. The optimal pipette shape and resistance are crucial factors in successful formation of Giga seals with the IMM. Our pipettes typically have a long, gradual taper, and a final tip size of ≤1 μm (with a resistance of ∼20–40 MΩ). Fig. 2 B depicts the pipette tip used for whole-mitoplast recordings and compares it to the pipette used for whole-cell patch clamp. These pipettes are carefully scanned and fire-polished using a microforge to remove debris and any small imperfections, ensuring a smooth tip.

Depending on the ion conductance of interest, appropriate solution composition and voltage-clamp protocols can be selected. Pipettes are filled with solutions having tonicity ∼350 mmol/kg (∼40–50 units higher than the bath solution) and are loaded into the pipette holder. As the pipette is dipped in the KCl bath solution, the pipette offset is set to zero (Fig. 4 C). We typically apply repetitive ±10 mV test pulses at 33 Hz from a holding potential of 0 mV to monitor the seal formation and track pipette resistance, seal resistance, and access resistance. Under these conditions, pipettes typically had resistances in the range of 20–40 MΩ and the access resistance is 40–80 MΩ.

Using a micromanipulator, the pipette is brought closer to a floating mitoplast in the bath, as shown in Fig. 4 C. Eight-shaped mitoplasts with clear membranes are preferred compared to fully enlarged ones as the success rate for stable recordings is higher, likely because the IMM is still fresh while coming out of the OMM. The current traces obtained during different steps of attaining whole-mitoplast mode are shown in Fig. 4, C–F. When the recording pipette makes a gigaohm seal with mitoplast (mitoplast-attached mode), the current diminishes very quickly and only the stray (or pipette) capacitance transients are visible (Fig. 4 D). This stray capacitance transient should be compensated to the maximum extent possible (Fig. 4 E) for reliable measurement of membrane capacitance. From this configuration, the whole-mitoplast mode can be established upon rupturing the IMM to get access to the matrix by applying short-duration high-voltage pulses (break-in protocol, see Materials and methods) that eventually result in the appearance of clear capacitance transient (Fig. 4 F). The membrane capacitance of mitoplasts isolated from mice heart or kidney ranges from 0.2 to 1.2 pF, and for cells (HEK293T or MEF mitoplast) the range is 0.3–0.8 pF. Once high-quality seals are established, current recordings can be performed by perfusing different test solutions into the bath chamber.

We first present a methodology for reliable recording of whole-mitoplast MCUcx currents (I_MCU); however, essentially the same protocol can be applied with little modifications (primarily in solutions) to record other IMM conductances of interest, as discussed later for UCP1-mediated H⁺ currents.

For recording I_MCU, pipettes are filled with Na⁺ gluconate or TMA-based pipette solution. Immediately after achieving the whole-mitoplast mode and measurement of membrane capacitance in the KCl bath solution, different bath solutions are perfused sequentially, and currents are recorded using ramp or step voltage protocols. All recordings are performed under continuous perfusion of bath solution. The voltage ramp from −160 mV to +80 mV (potential on the matrix-side relative to cytosolic-side of the IMM) covers almost all the physiological V_m across the IMM (ΔΨ approximately −160 to −180 mV, Fig. 5 B). In whole-mitoplast mode, inward currents are currents flowing into the matrix and are negative, while those flowing out of the matrix are positive. Baseline currents are always measured by perfusing HEPES/EGTA solution that corresponds to nominally Ca²⁺-free state, followed by perfusion of bath solution containing different [Ca²⁺] (Fig. 5). Of note, the bath solution containing Ca²⁺ (100 µM to 5 mM) only has HEPES and Tris as the major ions, it does not have EGTA or any other chelators. The reported [Ca²⁺] refers to total Ca²⁺ only. With the exception of mitoplasts from brown fat that manifest UCP1-mediated H⁺ currents (described below), the inward baseline currents in HEPES/EGTA at −160 mV are typically minimal (<10 pA/pF), which is the best indicator of a tight and reliable seal. These baseline currents are subtracted from the I_MCU elicited by different [Ca²⁺]. Although the origin of these baseline currents is unclear, they have little effect on I_MCU density when currents are large, but if not subtracted, they can significantly alter current density quantification, especially when I_MCU is small, such as in the heart (Fig. 5 G).

Physiologically, it is unlikely that ΔΨ will go more positive than 0 mV. However, currents in the positive V_m range (0 mV to +80 mV) provide a good estimate for any cation or anion currents (thus helping in the biophysical analysis of the conductance under consideration) or test leak currents across the IMM (for evaluation of the integrity of the seal). With Na⁺ ions in the pipette and EGTA in the bath, there is a large outward, ruthenium red-sensitive, MCU-mediated Na⁺ current in WT-MEF at +80 mV, which is absent when the pipette does not contain any Na⁺ (i.e., contains TMA-based solution, Fig. 5, B and C, indicated by an arrow) or when MCU is knocked out (Fig. 5 E). Importantly, this outward Na⁺ current gradually diminishes as soon as Ca²⁺ is perfused and disappears completely in ≥100 µM [Ca²⁺] (Fig. 5 D) due to increased dwell time for Ca²⁺ in the MCU pore. This corresponds to the high Ca²⁺ or divalent selectivity of the MCUcx, the inward MCU-mediated Na⁺ current being highly Ca²⁺ sensitive (K_d < 2 nM) in comparison to the outward Na⁺ current (Kirichok et al., 2004; Garg et al., 2021). In our experience, success rate is higher with Na⁺-gluconate in comparison with the TMA-based pipette solution. When recording MCUcx currents, the outward Na⁺ current in the presence of HEPES/EGTA solution also provides a rough estimate of the MCU expression level.

Application of 100 µM, 1, and 5 mM [Ca²⁺] gradually induced a large amplitude Ca²⁺ current (I_MCU) in mitoplasts from MEFs (Fig. 5 D). These MEFs are deficient in Drp1 (a mitochondrial fission factor) and thus provide higher yield of larger mitoplasts, which significantly enhances the success rate for stable current recordings. In our recent work (Garg et al., 2021), we extensively validated these Drp1-KO MEFs for exploring MCUcx physiology using patch-clamp and mitochondrial Ca²⁺ uptake assays in intact cells and isolated mitochondria. We also showed that I_MCU density is similar irrespective of the presence or absence of Drp1 (Garg et al., 2021). As reported before (Garg et al., 2021), even 10 µM [Ca²⁺] can induce a significant and measurable I_MCU. I_MCU develops rapidly in mM [Ca²⁺] but might take a while to stabilize at low-µM [Ca²⁺], which is clearly detectable when mitoplasts are larger and MCU protein density is high, e.g., in cultured cell lines. Absence of a similar Ca²⁺ current in MCU-KO (Fig. 5 E) implies that MCU mediates the primary and potentially the only electrogenic pathway for Ca²⁺ flux across IMM under our recording conditions.

Fig. 5 A compares the I_MCU density at −160 mV in mitoplasts from MEFs and HEK293T cells at 1 mM [Ca²⁺]. I_MCU density is typically similar in the cell lines we have tested so far, MEFs (388 pA/pF ± 23, n = 16), HEK293T (382 pA/pF ± 30, n = 6; Fig. 5 A), and COS7 (∼400 pA/pF; Kirichok et al., 2004; Garg et al., 2021). In contrast to the effect of cations from the cytosolic side (Kirichok et al., 2004; Garg et al., 2021), changes in matrix-side (pipette solution) levels of Mg²⁺, EGTA, and EDTA, or substitution of permeant Na⁺ with non-permeant TMA does not appear to have a major effect on inward I_MCU density (Fig. 5, A–C). The outward Na⁺ current varies with different pipette solution compositions (Fig. 5, B and C), but the inward Ca²⁺ current remains similar (Fig. 5, A–C). Besides the voltage ramps, step voltage protocols can be employed to record I_MCU peak and steady-state currents (Fig. 5 F) at different voltages. Although I_MCU does not show any Ca²⁺-dependent inactivation (Garg et al., 2021), there is a small and highly variable inactivation visible at high negative potentials with high [Ca²⁺] (Kirichok et al., 2004).

Exemplary traces in Fig. 5, G and H, show that mitoplasts from mouse native tissues, especially the heart, exhibit very low I_MCU density, while kidney mitoplasts have substantial I_MCU although still lower than the cell lines. Accordingly, the outward MCU-mediated Na⁺ current is smaller in the heart in comparison to the kidney mitoplast (Fig. 5, G and H, marked by an arrow). Fieni et al. (2012) investigated a number of mouse tissues in this regard and found that although I_MCU in different tissues have broadly similar biophysical properties, the skeletal muscle and brown fat express the highest I_MCU expression compared with any other tissue. At the molecular level, these differences could be due to differences in MCUcx protein expression or subunit stoichiometry. At the physiological level, the reason for differences in I_MCU density in different tissues could potentially relate to differences in the individual bioenergetic requirements of a tissue, total mitochondrial content per volume of the cell type, frequency of intracellular [Ca²⁺] elevations, or sensitivity of different tissues to detrimental effects of mitochondrial Ca²⁺ overload (Fieni et al., 2012; Williams et al., 2013; Paillard et al., 2017). Teasing out the exact physiological role of MCUcx and its stoichiometry in different tissues or pathological conditions is an area of active investigation.

One important distinction while recording UCP1-mediated H⁺ currents (I_H+) from brown fat mitoplasts (Fig. 6) is that pipettes are filled with TMA-based buffered solution (pH 7.0), while the bath is being perfused with HEPES/EGTA bath solution (pH 7.0). Both solutions are designed in such a way that they only contain salts with large impermeable anions and cations (Fedorenko et al., 2012).

Similar to the procedure described above for MCUcx, a whole-mitoplast seal is formed in the KCl bath. After measuring membrane capacitance, the bath is immediately perfused with HEPES/EGTA solution and the same voltage ramp protocol (−160 mV to +80 mV over 850 ms, 5 s interval) is applied. The inward I_H+ observed after break-in (in the absence of any externally applied ligand, long-chain FA) will stabilize over tens of seconds and is mediated by endogenous FA generated by phospholipase(s) in the IMM (Fig. 6, Control). These endogenous FA in the IMM can be extracted by the application of 0.5% BSA (FA-free) or 10 mM methyl-β-cyclodextrin (MβCD), leading to dissipation of the I_H+. Thus, UCP1 has no constitutive activity on its own, UCP1-bound long-chain FA is required to translocate H⁺ across the membrane (Fedorenko et al., 2012).

However, incomplete extraction of FA by these agents may not entirely inhibit the UCP1-mediated I_H+. High concentration of MβCD (10 mM) robustly inhibits the current especially when applied on both sides of the IMM. As an alternative, or perhaps an easier way to get an accurate baseline, 1 mM GDP (a canonical UCP1 inhibitor, Fig. 6) or other Mg²⁺-free purine nucleotides (ATP or ADP) are typically applied from the cytosolic side (bath solution) that leads to complete inhibition of I_H+. For precise quantification or comparison of UCP1 functional expression across multiple groups, the endogenous FA are first extracted by the application of 10 mM MβCD to dissipate the endogenous FA-activated I_H+. Subsequent application of 10 mM MβCD in combination with a defined concentration of externally applied FA (typically 0.2–2 mM oleic acid) reactivates the I_H+, which can be precisely quantified (Fedorenko et al., 2012; Bertholet and Kirichok, 2020).

One of the primary challenges in applying patch clamp either to cell membranes or subcellular membranes is to closely match the ionic composition of the solutions with the complex physiological milieu. Nevertheless, a simple solution composition (with minimal ionic components) is essential for correct interpretation of the conductance of interest. Selecting the optimal composition of pipette and bath solution for recording a particular ion conductance of the IMM is a major decision one must make before starting the experiment. Along similar lines, it is sometimes necessary to use extraphysiological ionic concentrations to isolate, detect, and biophysically characterize a particular conductance. MCUcx single-channel currents are minuscule and exhibit multiple subconductance states, which can only be discerned at high mM [Ca²⁺], typically in the range of ∼2.6–5.2 pS at a holding potential of −160 mV and symmetrical 105 mM [Ca²⁺] (Kirichok et al., 2004; Garg et al., 2021). Correspondingly, high µM (≥10 µM) or mM [Ca²⁺] are required to record I_MCU in comparison to ≤1 µM [Ca²⁺] encountered in intact cells during cytosolic transients (Garg and Kirichok, 2019; Garg et al., 2021). Notably, [Ca²⁺] might elevate in tens of µM range physiologically at hotspots near mitochondria especially when close to ER (Rizzuto et al., 1998).

Second, matrix solution (i.e., pipette solution) composition cannot be changed while recording the IMM currents in the whole-mitoplast mode. However, macropatch recordings can be performed in case solutions with different ionic compositions need to be tested sequentially on the matrix side (Kirichok et al., 2004).

Third, physiologically, the pH just outside the IMM is slightly acidic ∼6.8–7.0, while the mitochondrial matrix is alkaline ∼7.5–8.0 (Santo-Domingo and Demaurex, 2012). This gradient is crucial for mitochondrial physiology as it is responsible for maintaining the protonmotive force, redox balance, and the activity of many transporters that rely primarily on pH gradient or rely on both pH gradient and voltage gradient across the IMM, such as Ca²⁺/H⁺ exchangers, K⁺/H⁺ exchanger, AAC, and UCP1. An H⁺ gradient is not required for patch clamp experiments since IMM can be clamped to the desired potential using the amplifier. However, to characterize any new conductance, one should initially start with a symmetrical pH 7.0 on both sides of the IMM, as IMM is stable near pH 7.0, and then test the effect of different pH or gradients for further biophysical characterization. For instance, UCP1 in brown fat mitoplasts mediates a significant FA-activated I_H+ even when pH is symmetrical (pH 7.0) across the IMM, requiring only a voltage gradient to manifest (Fig. 6). An enhanced proton gradient with a more alkaline matrix might enhance the amplitude of UCP1-mediated or AAC1-mediated I_H+ due to increased driving force. However, pH has a complex biophysical relationship with these FA-activated H⁺ conductances as pH influences both the UCP1 activity (H⁺ dissociation from the bound FA) and the activity of the IMM-associated phospholipases. The UCP1-mediated I_H+ is maximal in symmetrical alkaline pH ∼8 and is markedly reduced at symmetrical acidic pH 5.0 (Fedorenko et al., 2012).

On the other hand, for analyzing an ion channel like MCUcx that relies primarily on the electrochemical gradient of Ca²⁺ across the IMM, a pH 7.0 in patch clamp experiments is appropriate. Although the effect of varying matrix pH on MCUcx has not been examined, it can be conveniently assessed by employing pipette solutions with different pH. This is not feasible in conventional optical or bioenergetic assays, where a change in matrix or cytosolic pH can markedly affect the quantification of MCUcx activity since many other factors such as ΔΨ, K_d for the Ca²⁺ dye and its fluorescence, Ca²⁺ buffering, or transport of other ions like PO₄³⁻, etc., would be affected.

Fourth, at the subcellular level, extensive inter-organelle connections exist, like mitochondrial networks with other organelles, specifically ER or lysosomes (Fig. 1 A; Rizzuto et al., 1998; Elbaz and Schuldiner, 2011; Rowland and Voeltz, 2012). The preparation of mitoplasts involves breakage of interorganelle connections as well as connections between IMM and OMM. For recording IMM conductances, there is always a risk that relevant protein–protein interactions in the matrix, between IMM and OMM proteins, or in the IMS might get disrupted. For example, under conditions of high Ca²⁺ and oxidative stress, mitochondrial permeability transition may occur, triggered by the activation of CypD, which is a matrix peptidyl prolyl cis-trans isomerase (Tanveer et al., 1996). This activation was proposed to result in the modification of an IMM resident protein(s) and a physical association between several membrane proteins of the IMM and OMM, leading to equilibration of solutes ≤1.5 kD between the matrix and the cytosol (Haworth and Hunter, 1979; Bauer and Murphy, 2020). Another example is related to the binding of an IMS localized protein, MICU1, to the MCUcx. It has been argued that a varying amount of MICU1 is lost while preparing mitoplasts. Phillips et al. (2019) performed co-immunoprecipitation (Co-IP) experiments in the presence of low (50 mM) and high (500 mM) NaCl and observed a greatly reduced MCU/MICU1 ratio. The detrimental effects of detergent and high salt concentrations for protein–protein interactions are well known. In our own experiments, we isolated mitoplast by French press method and incubated them in high (750 mM) KCl for 30 min before pelleting and subjecting them to Co-IP lysis buffer. This Co-IP procedure showed that MICU1 binding to MCU in the mitoplasts and isolated mitochondria is similar (Garg et al., 2021). However, one caveat with the Co-IP experiments or even Blue-Native PAGE (that shows the molecular weight of the whole uniporter complex) is that one cannot completely rule out whether MICU1 was bound to the membrane or the uniporter complex before isolation, as the possibility remains that MICU1 can dissociate in the presence of the detergent or associate back with the MCU/EMRE pore during incubation in the lysis buffer. Regardless, complementary assays using isolated intact mitochondria and intact cells should also be performed to understand the physiological role of the respective conductance (with due consideration to other ion flux mechanisms).

Fifth, direct application of patch clamp to record transporters of the OMM is challenging, and most studies are primarily confined to recording of single-channel activity of VDAC reconstituted in liposomes or in direct OMM patches (Kinnally et al., 1987; Pavlov et al., 2001). In our hands, OMM often does not stick well with the glass pipettes in comparison to the fast, tight seal formation with the IMM, which could be related to different lipid composition of the two membranes and/or higher leak of the OMM due to the presence of high density of porins (Tedeschi et al., 1987). We plan to optimize the conditions for patching OMM in the future.

Lastly, the technique is laborious and a low success rate (compared with patch-clamp recordings of plasma membrane channels) leads to a slow rate of data acquisition. Perhaps, there is a scope to develop automated mitochondrial patch-clamp technologies just like the whole-cell automated systems currently used for recording ion channels of the plasma membrane (Priest et al., 2007).

Exploration of the mitochondrial ion channels and carriers using patch clamp is one of the really exciting and promising fields in membrane biophysics. Many of the limitations noted above also turn out to be some of the greatest strengths of the patch-clamp technique. Often, it is difficult to interpret the results from experiments with isolated mitochondria or intact cells. The existence of multiple but well-coordinated processes of influx, efflux, and dynamic buffering or cotransport of other ions (e.g., for Ca²⁺ and PO₄³⁻), the extensive contacts mitochondria form with other organelles especially ER and lysosomes, and the lack of availability of specific pharmacological modulators/blockers necessary to isolate the conductance of interest, are just some of the many challenges that researchers face when studying mitochondrial ion channels in intact organelles or cells. Direct mitochondrial patch clamp overcomes these barriers. The high temporal (<1 ms) and amplitude resolution (<1 pA) enables clear interpretation of structure–function relationship studies. It is the only method that can measure the ion selectivity of a channel. The exact molecular correlates for many of the IMM conductances (e.g., K⁺, anions, large conductances corresponding to mPTP, etc.) are still unclear, which is a rich opportunity for future research by patch-clamp technique.

Jeanne M. Nerbonne served as editor.

Schematic in Fig. 4 is created with BioRender.com. We would like to thank University of Maryland School of Medicine’s and School of Dentistry’s Electron Microscopy Core, Baltimore, MD, USA, for assistance with TEM imaging. We thank Dr. Yuriy Kirichok and Dr. Carmen A. Mannella for their comments and critical reading of the manuscript.

This work is supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases (Training Program in Muscle Biology Grant T32 AR007592-27 to D.M. Nguyen) and funds from the University of Maryland School of Medicine (to V. Garg).

Author contributions: A. Kumari and V. Garg designed and performed the experiments, analyzed the data, and wrote the paper. D.M. Nguyen discussed and contributed to the writing of the paper.