Antibodies used were as follows: LETM1 (H00003954-M03, Abnova, Taipei, Taiwan), GAPDH (sc25778, Santa Cruz Biotechnology, Dallas, TX). Secondary HRP conjugated AffiniPure antibodies were from Jackson Immunoresearch (West Grove, PA): goat anti-mouse (115-035-008) and goat anti-rabbit (111-035-008).

All reagents were from Sigma Aldrich (Vienna, Austria) unless otherwise indicated. Thapsigargin was from Tocris (Bristol, UK) and puromycin from Invivogen (San Diego, CA). All buffers as formulated were completed with MilliQ (Billerica, MA) ddH₂O.

Primers are listed in the Table S1. All primers were from Microsynth (Balgach, Switzerland). All shRNA constructs were obtained from Origene Technologies (Rockville, MD). Product numbers for control and constructs by gene are: LETM1 #1 TR311758A) and #2 (TR311758D). Sequences of shRNA constructs and targeted exons can be found in Table S2. pcDNA3 mtRFP was obtained as a generous gift from Prof. Tullio Pozzan (University of Padova, Italy).

HeLa cells were a gift from E. Tomasich (Medical University of Vienna, Austria). All cells were maintained in DMEM (Gibco) supplemented with 10% (v/v) FBS (Gibco) and 1X penicillin/streptomycin/glutamine (Gibco). Cells were regularly screened for mycoplasma (MycoAlert Lonza kit).

Cells were transfected with Turbofect (Invitrogen, Paisley, UK) according to manufacturer's instructions; 48 h post transfection the media was changed to growth media with puromycin (2 μg/mL). After a resistant population of cells was established, cells were maintained in growth medium containing puromycin (2 μg/mL).

Cells were washed with PBS (Gibco) then harvested using lysis buffer (150 mM NaCl, 50 mM Tris (pH 8), 1% IGEPAL C360 and one tablet of protease inhibitors with EDTA (Roche) per 10 mL). Cell lysates were vortexed for 30 min at 4°C and then cleared by centrifugation at 15,000 g for 15 min, 4°C. Proteins were quantified by Bradford assay (Bio-Rad, Hercules, CA) using BSA as standard. Protein lysates were separated by SDS-PAGE and blotted onto 0.45 μm PVDF membranes (ThermoFisher Scientific, Rockford, IL). Membranes were immunoblotted as indicated and fat-free milk was used as blocking medium. Blots were quantified using the BioRad Image Lab (v5.2.1) software.

Total RNA was isolated from cells using TRIzol reagent (Invitrogen) according to the manufacturer's instructions. Total RNA was DNase I (Invitrogen) digested before reverse transcription using the High-Capacity cDNA Reverse Transcription Kit (AB Biosystems, Foster City, CA). Diluted cDNA was then used for quantitative PCR using the GoTaq qPCR master mix (Promega, Madison, WI) and the reaction was run on the StepOne Plus (AB Biosystems). Analysis was performed using the StepOne Plus software (v2.3) and Expression Suite (v1.0.3).

Cell counting, harvesting and permeabilization were performed as in Wilfinger et al. (2016), with only minor modification. Permeabilized cells were studied in either intracellular like buffer I (ICL buffer I) (130 mM KCl, 10 mM Tris-MOPS (pH 7.4), 10 μM EGTA-Tris, 1 mM KPi, 5 mM malate, 5 mM sodium glutamate, 1 μM Ca²⁺ green 5N) or intracellular like buffer II (ICL buffer II) (130 mM KCl, 10 mM Tris-MOPS (pH 7.4), 10 μM EGTA-Tris, 1 mM KPi, 5 mM malate, 5 mM glutamic acid, 1 μM Ca²⁺ green 5N).Importantly, these buffers are identical, with the exception that ICL I contains sodium glutamate and ICL II glutamic acid. The pH of the ICL buffers was verified prior to measurements. Sequential additions of 10 μM Ca²⁺ (as CaCl₂), 0.2 μM ruthenium red (RR) or 1.2 μM ruthenium 360 (R360), 10 mM Na⁺ (as sodium acetate, NaOAc) and 2 μM FCCP were as indicated in Figure 4. Ca²⁺ uptake/release measurements were performed at 25°C on a LS55 spectrofluorometer (Perkin Elmer) with the following parameters, λ_ex = 505 nm, λ_em = 535 nm, slit width: 2.5 nm.

Isolated mitochondria (Frezza et al., 2007) were deenergized by incubation with 5 μM antimycin A for 10 min at RT (25°C). Mitochondria were pelleted and resuspended in isotonic isolation buffer (10 mM Tris-MOPS (pH 7.4), 1 mM EGTA-Tris (pH 7.4), 200 mM sucrose) at a concentration of 1 mg/mL and placed in a cuvette. In the case of KOAc-induced swelling, mitochondria were additionally depleted of Mg²⁺ immediately prior to measurement by incubation with 1 μM A23187 and 1 mM EDTA to eliminate Mg²⁺, the blocker of the KHE. To measure KOAc- or NaOAc-induced passive swelling, KOAc (as described in Nowikovsky et al., 2004: 55 mM KOAc, 5 mM TES, 0.1 mM EDTA) or NaOAc (55 mM NaOAc, 5 mM TES, 0.1 mM EDTA) buffer, respectively, was added and the optical density changes were immediately recorded at 25°C using an UltroSpec 1100 Pro spectrophotometer (Biochrom, Cambridge, UK). Swelling rate was quantified by fitting the raw swelling data for one-component exponential decay as shown, and the k value was used to quantify the rate of the process.

Cells (2 × 10⁸) were harvested and permeabilized as described in Wilfinger et al. (2016). Rhodamine123 (Molecular Probes) was used as ΔΨ probe at a concentration of 0.15 μM in ICL buffer II. Mitochondria were energized with glutamate (5 mM) and malate (5 mM), maximal mitochondrial depolarization was achieved by the addition of FCCP (1 μM). Measurements were performed at 37°C on a LS55 spectrofluorometer with the following parameters: λ_ex = 503 nm, λ_em = 523 nm, slit width: Ex-3.5 nm, Em-2.5 nm.

The potassium probe mitoPOP was used to monitor the mitochondrial K⁺ concentration. The probe was tested for its pH and K⁺ sensitivity as follows: pH sensitivity was measured by dissolving 10 μM mitoPOP in a solution containing 150 mM KCl and 25 mM MES (pH 6.5), HEPES (pH 7 and 7.4) or TRIS (pH 8 and 8.5). Fluorescence was recorded in a microplate reader (Perkin Elmer Envision) using 485/10 nm and 535/30 nm bandpass filters for excitation and emission, respectively. Raw fluorescence values were blank subtracted, normalized on neutral pH and plotted against pH values. In parallel, K⁺ sensitivity was tested by dissolving 10 μM mitoPOP in a solution containing 25 mM HEPES (pH 7) and 0, 150, or 300 mM KCl at 25°C. Raw fluorescence values (measured as described above) were blank subtracted, normalized on K⁺-free medium and plotted against [K⁺].

Cells (30,000/well) were loaded with mitoPOP (1 μM) for 45 min at 37°C under cell culture conditions in modified Krebs-Ringer buffer (135 mM NaCl, 5 mM KCl, 1mM MgSO₄, 0.4 mM K₂HPO₄, 20 mM HEPES, 5.5 mM glucose, 1 mM CaCl₂, pH = 7.4). Cells were washed once with Krebs-Ringer buffer and then incubated for 5 min in 140 mM Tris-KCl (pH 7.4). Thereafter increasing K⁺ or sucrose concentrations were added as indicated for the in situ response curve. Measurements were performed on a TriStar plate reader (λ_ex:485 nm, λ_em:535 nm) at 25°C. Cells were subsequently counted by Trypan blue exclusion to normalize fluorescence readings to cell number. As a double control, to directly normalize the results to the amount of mitochondria independently of potentially variable cell numbers, cells plated in duplicates were analogously loaded with MitoTracker green (50 nM, Molecular Probes) at 37°C for 30 min. The staining was washed once, the fluorescence intensity quantified as above in KCl buffer and used to normalize mitoPOP fluorescence values.

Scrambled shRNA-transfected (control) HeLa cells were seeded onto an Ibidi μ-slide (Martinsried, Germany) and transfected as previously described with pcDNA3 mtRFP and stained with mitoPOP for the last 45 min prior to visualization. The cells were visualized and images acquired on an LSM780 confocal microscope. Microscope, objective and aperture specifications may be found in Wilfinger et al. (2016).

All statistical analyses were performed using GraphPad (La Jolla, CA) Prism v6 for Windows. Similarly, all graphs were generated with GraphPad Prism. One way ANOVA with Dunnett's posttest was performed in each case to assess significance. Individual p-values are indicated in the figure legends.

To address the role of LETM1 in mitochondrial K⁺ and Ca²⁺ release, we stably transfected HeLa cells with shRNAs targeted against LETM1, or randomly scrambled constructs. HeLa cell lines with LETM1 knockdown of varying efficiency were obtained. We observed that the most efficient viable knockdown had a 50% reduction in gene expression when compared to scramble controls (shSCR). This was achieved with shRNA LETM1 #2, while for another construct, shRNA LETM1 #1, no significant reduction was observed (Figure 1C). However, surprisingly, at the protein level both knockdown cell lines showed decreased LETM1 expression, with shLETM1 #2 proving more efficient than shLETM1 #1 (Figures 1A,B). As mentioned before, WHS patients have only one copy of LETM1, hence a reduction of 50% is clinically relevant (Hart et al., 2014). We further asked if reduced LETM1 expression altered the expression of known mitochondrial Ca²⁺ transporters, the mitochondrial Ca²⁺ uniporter (MCU) or NCLX (SLC8B1), and found both genes had comparable transcription levels regardless of LETM1 expression (Figure 1C). To further exclude that LETM1 KD may lead to loss of mitochondrial DNA and reduced mitochondrial mass, we compared the expression of the mitochondrially-encoded genes COX1 and ATP6 to that of a nuclear gene (18S) in our shLETM1 and control cells, and found no differences (Figure 1D).

So far, LETM1-dependent KHE activity has been assessed in S. cerevisiae, a yeast species that lacks mitochondrial Ca²⁺ transport systems. Using light scattering-based assays, yeast mitochondria devoid of LETM1 were shown to be refractory to KOAc-induced swelling, indicating reduced KHE activity and matrix K⁺ overload (Nowikovsky et al., 2004). To date, light scattering monitoring of passive swelling represents the most reliable way to evaluate KHE activity (Nakashima and Garlid, 1982; Garlid et al., 1986). Incubation of isolated non-respiring mitochondria in KOAc results in a rapid diffusion of acetic acid into the matrix where it deprotonates. The dissociation of H⁺ from AcO⁻ leads to the acidification of the matrix catalysing the uptake of K⁺ in exchange for H⁺. The matrix accumulation of K⁺ acetate induces matrix swelling which is seen as decreased of scattered light or optical density. Here, we applied the light scattering approach to isolated mitochondria from HeLa cells, and monitored for the first time the changes in optical density (OD) of suspensions of control and LETM1 KD mitochondria (Figure 2).

To reveal changes in the activity of the mitochondrial KHE we used hypotonic KOAc buffer, and deenergized mitochondria depleted of Mg²⁺, a known inhibitor of KHE activity. Incubation of isolated non-respiring mitochondria in KOAc leads to the acidification of the matrix, catalysing the uptake of K⁺ in exchange for H⁺. The matrix accumulation of K⁺ induces matrix swelling which is seen as decreased scattered light or optical density. In general, while a rapid and high amplitude swelling is expected from mitochondria with functional KHE, a reduction of KHE activity will be reflected by lesser swelling. We observed that shLETM1 #1 and #2 mitochondria had slower and decreased swelling as compared to shSCR controls, as reflected by a reduced and slower decrease of scattered light. The decrease of OD was inhibited by quinine, the inhibitor of KHE, confirming that the changes in OD were due to KHE activity (Figures 2A,B). Moreover, the matrix volume of shLETM1 mitochondria was increased as compared to that of shSCR mitochondria, as indicated by a lower OD from the beginning of recording, suggesting a higher content of osmolytes (Figure 2A). The mitochondrial Na⁺/H⁺ exchange activity was assayed in a similar swelling assay. In this case, Mg²⁺ depletion was not carried out and Na⁺ fluxes could not be inhibited by quinine (Bernardi, 1999). We obtained qualitatively identical results: both shLETM1 knockdown mitochondria displayed decreased swelling (Figures 2C,D). The data collected here using well-established and reliable methods, establish that LETM1 downregulation results in a significant decrease of proton exchange for each monovalent cation, K⁺ or Na⁺.

Reduced KHE activity should result in accumulation of K⁺ in the matrix of energized mitochondria. To directly assess whether K⁺ indeed accumulated in LETM1 KD mitochondria, we employed a mitochondria-targeted fluorescent K⁺ probe (mitoPOP) which accumulates in the mitochondrial matrix due to the latter's negative electrical potential (Mattarei et al., unpublished results). The specific localization of the probe was confirmed by co-localization with mtRFP in shSCR control cells (Figures 3A–C). A titration of mitoPOP fluorescence as a function of [K⁺] (Figure 3D) and pH (Figure 3E) demonstrated the sensitivity of the probe for K⁺ changes in the desired concentration range and its insensitivity to pH (Figures 3D,E). In order to assess K⁺ differences in LETM1 KD cells, we preliminarily confirmed that the membrane potential was unaffected by LETM1 KD (Figure 3F) as previously reported (Doonan et al., 2014). Significant changes of the in situ mitochondrial transmembrane potential upon increasing additions of KCl up to 400 mM were excluded in experiments employing TMRM (Figure S1A). To demonstrate a response to K⁺ in situ, we then added KCl to intact shSCR mitoPOP-loaded HeLa cells. This did generate a response curve (Figure 3G). In comparison, adding sucrose to intact shSCR cells at increasing concentration had no effect on the mitoPOP fluorescence (Figure 3H and Figure S1B), thus excluding increasing osmotic pressure and confirming increasing K⁺ concentration as the reason for the observed fluorescence changes. Lastly, to assess if matrix K⁺ accumulation occurs in LETM1 KD cells, KD and control cells were exposed to a solution of 140 mM KCl after mitoPOP loading (Figures 3I,J). A 1.6-fold increase in fluorescence was observed in shLETM1 #2 cells in comparison to shSCR control cells, while shLETM1 #1 cells showed a more modest 1.3-fold increase. Together these results confirm that reduced KHE activity leads to matrix accumulation of K⁺ in LETM1-depleted cells.

Using our shLETM1 cells we performed mitochondrial Ca²⁺ uptake and release experiments similar in setup to those of Doonan et al. (2014) (Figure 4A). One pulse of Ca²⁺ was added to permeabilized LETM1 KD or control cells suspended in ICL buffer I containing the Ca²⁺ indicator Ca²⁺ 5N Green (Figure 4A, first arrow). Next, RR, an inhibitor of MCU, was added to block Ca²⁺ uptake and induce H⁺- and Na⁺-dependent Ca²⁺ release. As expected, a slow Ca²⁺ extrusion was observed (Figure 4A, second arrow). This was followed by a single pulse of the uncoupler FCCP to release sequestered mitochondrial Ca²⁺ (Doonan et al., 2014) (Figure 4A, third/fourth arrow). Consistent with the findings of Doonan et al., shLETM1 #2 cells had reduced mitochondrial Ca²⁺ efflux (Figures 4A–D), in a fashion that was dependent on LETM1 expression. The assays conducted at 25°C or 37°C showed comparable results. Based on the hypothesis that LETM1 embodies the KHE, which also transports Na⁺ (Nowikovsky et al., 2004; Froschauer et al., 2005), we next assessed Ca²⁺ transport in Na⁺-free buffer (ICL buffer II) (Figure 4E). In the absence of Na⁺, no change in Ca²⁺ efflux between cells expressing LETM1 at wild-type or knockdown levels was observed (Figures 4E,F). Next, the Na⁺-induced Ca²⁺ release was initiated by the addition of a Na⁺ pulse (Figure 4E, third arrow). As illustrated in Figure 4G, shLETM1 #2 showed reduced Na⁺-induced Ca²⁺ release. A comparison of the results obtained in the presence or absence of Na⁺ suggests that the effect of LETM1 on Ca²⁺ efflux depends on Na⁺ (Figures 4B,D,G vs. Figure 4F). For further validation, we subjected LETM1 control and knockdown cells to Ca²⁺ uptake/release assays in presence of the NCLX inhibitor CGP37157 using the Na⁺ replete buffer. This allows monitoring mitochondrial Ca²⁺ release that is not due to NCLX. As shown in Figures 4H–K. we found similar Na⁺-independent Ca²⁺ mitochondrial release activities in presence or absence of LETM1. The results were consistent at 25°C (4H, I) and 37°C (4J, K). To additionally evaluate the role of the KHE in mitochondrial Ca²⁺ fluxes using a pharmacological approach, we resuspended shSCR cells in Na⁺-replete buffer, and quantified Na⁺-dependent Ca²⁺ fluxes in control cells in the absence or presence of the KHE inhibitor quinine. As shown in Figures 4L,M, inhibition of the KHE severely decreased the Na⁺-dependent Ca²⁺ efflux in comparison to vehicle controls. These data support the hypothesis that the deregulation of mitochondrial monovalent cation homeostasis induced by KHE dysfunction reduces Ca²⁺ efflux.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.