Plasma Membrane Ca2+-ATPase Isoforms Composition Regulates Cellular pH Homeostasis in Differentiating PC12 Cells in a Manner Dependent on Cytosolic Ca2+ Elevations

Plasma membrane Ca2+-ATPase (PMCA) by extruding Ca2+ outside the cell, actively participates in the regulation of intracellular Ca2+ concentration. Acting as Ca2+/H+ counter-transporter, PMCA transports large quantities of protons which may affect organellar pH homeostasis. PMCA exists in four isoforms (PMCA1-4) but only PMCA2 and PMCA3, due to their unique localization and features, perform more specialized function. Using differentiated PC12 cells we assessed the role of PMCA2 and PMCA3 in the regulation of intracellular pH in steady-state conditions and during Ca2+ overload evoked by 59 mM KCl. We observed that manipulation in PMCA expression elevated pHmito and pHcyto but only in PMCA2-downregulated cells higher mitochondrial pH gradient (ΔpH) was found in steady-state conditions. Our data also demonstrated that PMCA2 or PMCA3 knock-down delayed Ca2+ clearance and partially attenuated cellular acidification during KCl-stimulated Ca2+ influx. Because SERCA and NCX modulated cellular pH response in neglectable manner, and all conditions used to inhibit PMCA prevented KCl-induced pH drop, we considered PMCA2 and PMCA3 as mainly responsible for transport of protons to intracellular milieu. In steady-state conditions, higher TMRE uptake in PMCA2-knockdown line was driven by plasma membrane potential (Ψp). Nonetheless, mitochondrial membrane potential (Ψm) in this line was dissipated during Ca2+ overload. Cyclosporin and bongkrekic acid prevented Ψm loss suggesting the involvement of Ca2+-driven opening of mitochondrial permeability transition pore as putative underlying mechanism. The findings presented here demonstrate a crucial role of PMCA2 and PMCA3 in regulation of cellular pH and indicate PMCA membrane composition important for preservation of electrochemical gradient.


Introduction
Neuronal differentiation is associated with spatially and temporary coordinated elevations in cytosolic Ca 2+ concentration -(Ca 2+ ) c -propagated due to Ca 2+ entry via plasma membrane and its release from internal stores [1,2]. These physiological and pathological Ca 2+ signals are modulated by the activity of mitochondria, which buffer (Ca 2+ ) c and regulate Ca 2+ -dependent activation or inhibition of several processes [3,4]. For example, mitochondrial control of Ca 2+ signal is crucial for regulation of both the cell membrane's voltage and, especially, for pH gradients driving ATP generation [5]. Mitochondria not only link Ca 2+ homeostasis to cell metabolism, but may also drive cell fate by controlling ATP/ADP ratio.
Acting as the energetic centers, they shape signaling pathways, control propagation of Ca 2+ waves and by providing ATP to calcium pumps boost calcium gradients [6]. Elevations of Ca 2+ in the mitochondrial matrix regulate voltage (DY m , negative inside) and pH (DpH, alkaline inside) components of electrochemical gradient. According to the chemiosmotic model, DY m and DpH are thermodynamically equivalent to power ATP synthesis [7]. Even though DpH constitutes only 20-30% of proton motive force, it is essential for electroneutral transport of ions and movement of metabolites into the matrix [8]. The electrical gradient establishes most of the potential difference. Together with DpH, it sets the driving force for ATP synthase, and for cytosolic Ca 2+ to enter the matrix [9]. Moderate elevations of Ca 2+ in the matrix activate dehydrogenases of Krebs cycle, modulate the activity of electron transport chain and stimulate the respiratory rate [6,10]. This may make mitochondrial membrane more negative. On the other hand, Ca 2+ overload may activate permeability transition pore (mPTP) formation allowing ions to leave the mitochondrion, thereby triggering cell death [9].
Mitochondrial Ca 2+ uptake in intact cells was observed at low cytosolic Ca 2+ concentrations ranging from 150 to 300 nM [11]. However, elevations in (Ca 2+ ) c stimulate matrix acidification and result in DpH drop what is suggested to decrease oxygen consumption [12]. The newest finding located plasma membrane calcium pump (PMCA) in the center for intracellular protons transport [13]. Because PMCA operates as Ca 2+ /H + countertransport with a 1:1 stoichiometry, the extrusion of Ca 2+ generates large quantities of protons that are transmitted to mitochondrial matrix leading to pH decrease [13]. Since Ca 2+ and protons have opposite effects on many cellular processes, the role of PMCA in the regulation of calcium homeostasis may be of fundamental importance for preservation of cellular energy.
PMCA exists in four isoforms PMCA1-4. Pumps 1 and 4 are ubiquitously distributed and perform a ''housekeeping'' role whereas the location of 2 and 3 isoforms is restricted to only some tissues where they perform more specialized functions [14][15][16]. Due to the abundance of PMCA2 and PMCA3 in the nervous system they are termed neuron-specific. During development their expression undergoes considerable changes reflecting the importance of the spatial organization of Ca 2+ extrusion systems for synaptic formation [17][18][19]. Moreover, the observation of mRNA distribution suggests that the expression of PMCA2 and PMCA3 is controlled by different mechanisms than the two other isoforms [20]. The studies on PMCA have made clear that unique PMCA2 properties distinguish it from other basic isoforms. It possesses the highest resting activity and calmodulin sensitivity, and represents more than 30-40% of the total pump protein in mature neurons [21]. Thus, PMCA2 is thought to be the principal ATPase that maintains Ca 2+ homeostasis following neural excitation. The existence of PMCA2 is expected to provide neuronal cells with higher sensitivity to even subtle (Ca 2+ ) c changes. This specificity of PMCA2, which is further highlighted by its interaction with specific partners [22], could explain why this isoform plays a predominant role in neuronal cells that have special Ca 2+ demands. The role of PMCA3 is much less understood. However, distribution, kinetic properties and scarce studies including our previous work on PC12 cells suggest that it should be also considered as an important Ca 2+ player in differentiation process.
To study the potential role of neuro-specific PMCA isoforms in regulation of cellular pH, we used differentiated PC12 lines with experimentally downregulated PMCA2 or PMCA3. Due to possessing of several features characteristic for sympathetic-like neurons [23], this cell line is an excellent model system to study neuronal processes. We found that PMCA2-or PMCA3-deficient cells maintained higher pH mito and pH cyto but only in PMCA2downregulated line increased DpH was observed in steady-state conditions. Also, we demonstrated that PMCA2 and PMCA3 were primarily responsible for Ca 2+ -dependent pH mito and pH cyto decreases and accompanying DpH drop during KCl stimulations. In PMCA2-downregulated cells, Ca 2+ overload led to dissipation of mitochondrial membrane potential, a phenomenon that was blocked by cyclosporin and bongkrekic acid suggesting the involvement of mitochondrial permeability transition pore. Our findings point out that neuro-specific PMCA isoforms are important regulators of cellular pH in steady-state conditions and may also shape Ca 2+ -dependent pH changes during depolarization events.

Reagents
All reagents, if not separately mentioned, were purchased from Sigma-Aldrich (Germany). The PC12 rat pheochromocytoma cell line was obtained from ATCC (USA) and from Sigma-Aldrich (Germany). RPMI 1640 medium was from PAA (Austria). Calf and horse sera were from BioChrom (UK The model of stable transfection pcDNA3.1(+) vectors carrying the antisense oligonucleotides directed to either PMCA2 or PMCA3 were used to establish a stably-transfected PC12 lines as described in [24]. Cells were cultured as described previously [25] and differentiated with 1 mM dibutyryl-cAMP for 48 h. All the results presented here were obtained following 2-day differentiation process. For pH measurements, mitoSypHer probe was transfected to undifferentiated antisense-carrying PC12 lines with TurboFect Transfection reagent and 2 days later cells were differentiated as described above. Routinely, the expression of PMCA2, PMCA3 and mitoSypHer was controlled by real-time PCR every two passages and no more than 6 passages were used. To increase the accuracy and maintain the reproducibility of the data we separately transfected two PC12 lines of different sources. The description of the lines as _2 (PMCA2-deficient line), _3 (PMCA3-deficient line) and C (mock transfected line) was adapted.

Transient transfection
PC12 cells transient transfection with antisense probes described in [26] and listed in Table 1 was conducted using TurboFect transfection reagent. In brief, three phosphothioate oligodeoxynucleotides antisense to translated regions of mRNA of either PMCA2 or PMCA3 were added in equimolar concentrations (4 mM) to a serum-free RPMI medium. Total concentration of oligodeoxynucleotides was kept at 12 mM during transfection. After 6 h, medium was replaced with complete RPMI medium and cells were allowed to recover for another 48 h. After recovery period transfection was repeated. Cells transfected with a mismatch oligonucleotide sequence (12 mM) was used as a control for antisense oligonucleotides transfection. Growth medium and reagents were changed in all culture flasks at the same time. Following second transfection, the efficiency of PMCA knockdown was assessed by Western blot determination of PMCA protein level. Cells were differentiated as described in The model of stable transfection section.

Viability assay
Cellular viability was assessed using WST-1 assay. 5610 3 cells were seeded in a 96-well plate and incubated with WST-1 solution in a 1:10 ratio for 4uC at 37uC. The absorbance of samples was spectrophotometrically measured at 450 nm.

pH measurements
For mitoSypHer/SNARF dual imaging, cells expressing mitoSypHer were adhered to poly-L-lysine coated coverslips and differentiated for 2 days. Then, the culture medium was changed into a buffer containing 20 mM HEPES, 131 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 10 mM glucose, 2.2 mM CaCl 2 , 10 mM NaHCO 3 and 1 mM KH 2 PO 4 at pH 7.4 and cells were loaded with 10 mM SNARF (with 0.01% pluronic acid) for 40 min at 37uC. Simultaneous pH mito /pH cyto measurements were performed in a thermostatic chamber at 37uC on TCS SP5 laser scanning confocal microscope equipped with DM6000 CFS system, DFC 360FX camera, HCX PL APO 636 objective and LAS AF Lite software (Leica). Fluorescence imaging was done with the tandem resonant scanner (16 kHz bidirectional, ,25 frames/s). SNARF and mitoSypHer were excited using argon laser low-intensity light (488 nm). The fluorescence emitted in the range of 500-530 nm was collected for mitoSypHer whereas fluorescence in two separate channels (620-765 nm and 560-600 nm) was collected for SNARF. At the end of each experiment, fluorescence changes were calibrated to absolute mitochondrial and cytosolic pH using nigericin (5 mg/ml) and monensin (5 mM) in pH 9.5-10.0 (20 mM N-methyl-D-glutamine), pH 8.0-9.0 (Tris), pH 7.0-7.5 (HEPES) or pH 5.5-6.5 (MES), as described in [13]. The calibration curve was fitted to sigmoidal equation using GraphPad Prism 5.01. The emission ratio (620 nm-765 nm)/(560 nm-600 nm) for SNARF was calculated in MetaFluor 6.3 (Universal Imaging) and processed in MS Excel. The measured bleedthrough between SNARF and mitoSypHer probe was less than 4%, as evaluated using online Fluorescence SpectraViewer software. Unless otherwise indicated, inhibitors were added 20 min before measurement. The role of NCX (Na + /Ca 2+ exchanger) in KCl-evoked pH changes was assessed in a loading buffer with 0-5 mM Na + . Ca 2+free solution contained 1 mM EDTA instead of 2.2 mM CaCl 2 .

Single-cell Ca 2+ imaging
Cells expressing mitoSypHer were adhered to poly-L-lysine coated slides and loaded with 10 mM Fura-2 for 1 h at 37uC. After several washes, cells were placed in a buffer composed as described in pH measurements section. When recording simultaneously with mitoSypHer, Fura-2 was alternately excited at 340 and 380 nm for 0.3 s through a 505 nm dichroic long pass filter and 535 nm emission filter. When recording only Fura-2, the dye was excited at 340 and 380 nm (0.3 s) through a 430 nm long way pass dichroic filter and a 510 nm bandpass emission filter. The mitoSypHer was alternately excited at 430 and 480 nm for 0.3 s through a 505 nm dichroic long pass filter and imaged with a 535 nm band pass emission filter. The contour of single cells was taken to define region of interest (ROI) from which the fluorescence was recorded. Background fluorescence was automatically subtracted from all measurements. Ratiometric images of pH mito and Ca 2+ were acquired using fluorescent Axiovert S100 TV inverted microscope (Carl Zeiss) equipped with a 406 Plan Neofluar objective and attached to a cooled CCD camera (Spectral Instruments Inc.).

Immunocytochemistry
Confocal microscopy was used to analyze SypHer mitochondrial targeting and mitochondrial mass. ,10 3 differentiated cells seeded on poly-L-lysine coated glass LabTek II chamber slides were fixed for 30 min with 3.8% paraformaldehyde and permeabilized with 0.1% Triton X-100 for 10 min at 4uC. Fixed cells were then blocked with 6% bovine serum albumin (BSA), overnight incubated with monoclonal anti-GFP antibody (1:100) at 4uC and probed with secondary antibodies conjugated to Alexa Fluor 488 (1:1000) for 2 h at room temperature. Next, mitochondria were stained for 15 min with MitoTracker Red 580 (40 nM). Images were taken on TCS S5 confocal laser scanning microscope with 636 objective (Leica). In a separate experiment, mitochondrial mass was determined with MitoTracker Green. In this method cells were first loaded with 150 nM MitoTracker Green FM for 30 min at 37uC and then fixed as described above. The average fluorescence intensity after background subtraction was measured with TCS S5 microscope accompanying software (Leica). Raw images were processes with CorelDraw Graphics Suite 11.

Total cell lysate preparation and Western blot analysis
Scraped cells were resuspended in RIPA buffer supplemented with 1 mM PMSF, 2 mM Na 3 VO 4 and protease inhibitor cocktail and lysed for 30 min on ice. Then, lysates were centrifuged at 10 0006 g for 5 min and supernatants were boiled for 5 min in Laemmli buffer. Total protein content was quantified using Bio-Rad Protein Assay. 20 mg of total cellular proteins were resolved on a 10% SDS-PAGE and transferred onto nitrocellulose membrane using semidry method. Membranes were first blocked with 6% BSA in TBS-T buffer (10 mM Tris-HCl, pH. 7.4, 150 mM NaCl, 0.05% Twwen-20) for 1 h at room temperature and then probed overnight at 4uC with primary antibodies against GFP (1:1000), PMCA2 (1:750), PMCA3 (1:750) and GAPDH (1:1000). Following several washes with TBS-T, membranes were incubated with appropriate secondary antibodies (1:5000) coupled with alkaline phosphatase at room temperature for 4 h. Bands were visualized using Sigma Fast BCIP/NBT used according to the manufacturer's instructions. Blots were scanned and the bands intensity was measured using GelDocEQ with Quantity One 1-D Analysis Software version 4.4.1 (Bio-Rad).

RNA isolation and PCR reactions
Total cellular RNA was extracted using Trizol reagent according to the procedure provided by the manufacturer. 1 mg of isolated RNA was subsequently used for cDNA synthesis with oligo(dT) primers in a 20 ml reaction mixture containing M-MLV reverse transcriptase. The cDNA templates were used to quantify gene expression level using Maxima SYBR Green Master Mix in the following conditions: 15 min at 95uC followed by 40 cycles at 95uC for 15 s, 60uC for 30 s and 72uC for 30 s. PCR reactions were performed in an AbiPrism 7000 sequence detection system (Applied Biosciences). For each PCR amplicon, a melting curve was run. The relative fold change after normalization to Gapdh expression was calculated using a comparative 2 2DDCt method [27].
Conventional PCR used to estimate the efficiency of mitoSy-pHer transfection was carried out using Paq5000 polymerase in the following conditions: 5 min at 95uC followed by 30 cycles at  95uC for 1 min, 50uC for 1 min, 72uC for 2 min with a final extension at 72uC for 10 min in T300 thermocycler (Biometra) using cDNA obtained as described above. PCR products after staining with ethidium bromide were analyzed under UV light in GelDocEQ system (Bio-Rad). The primers used in PCR reactions are listed in Table 2.

Determination of mitochondrial swelling
Mitochondria were isolated as described in [28]. The experiments were carried out at 30uC in a reaction medium containing 200 mM sucrose, 10 mM HEPES, pH 7.4, 10 mM EGTA, 5 mM KH 2 PO 4 , 2 mM rotenone (to inhibit electron backflow to complex I), 1 mg/ml oligomycin (to maintain constant ATP/ADP ratio) and mitochondria suspended at ,1 mg/ml. Before exposure to 10 mM CaCl 2 , mitochondria were energized with 5 mM succinate for 2 min. Cyclosporin (1 mM), bongkrekic acid (10 mM) or atractylate (20 mM) was added just prior to succinate. Swelling was assessed by changes in light scattering monitored spectrophotometrically at 520 nm under a continuous stirring of mitochondrial suspension.

Monitoring of mitochondrial and plasma membrane potential (DY m and DY p )
Mitochondrial membrane potential (DY m ) was measured with TMRE (tetra-methyl-rhodamine-ethyl ester), which accumulates in mitochondrial matrix according to the Nernst equation [29], whereas plasma membrane potential (DY p ) was measured with DiSBAC 2 (Bis-(1,3-diethylthiobarbituric acid)trimethine oxonol). For estimation, cells were loaded in a dark with 25 nM TMRE or 1 mM DiSBAC 2 for 30 min at 37uC in a buffer A containing 20 mM HEPES, pH 7.4, 2 mM CaCl 2 , 150 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 10 mM glucose and analyzed by FACScan Becton Dickinson flow cytometer using an accompanying software. Cells incubated with 0.1% DMSO, used as a solvent for TMRE, were monitored to record background fluorescence, which was later subtracted from the TMRE recordings. Changes in DY m were monitored in resting cells and at selected points following 59 mM KCl treatment (10 min after first KCl addition, recovery, 10 min after second KCl addition, recovery). The reliability of TMRE to be used for DY m measurement was confirmed by the pre-incubation with either 6 mM oligomycin or 1 mM FCCP (carbonyl cyanide p-trifluoromethoxyphenylhydrazone) for 10 min. The influence of DY p on mitochondrial TMRE uptake was assessed by 5 min preincubation of the cells with Ca 2+free buffer A containing 59 mM KCl before loading with 25 nM TMRE. Cyclosporin A (1 mM), bongkrekic acid (10 mM) or FK-506 (10 mM) were added to the culture medium 1 h before 25 nM TMRE loading.
TMRE fluorescence decay in single cells was assessed using TILL Photonics dual wavelength imaging system equipped with Polychrome IV monochromator (TILL Photonics GmbH). TMRE-loaded cells (25 nM) were illuminated at 535 nm through a 15 nm band-pass filter for 2 min and following 30 s depolarization with 1 mM FCCP. Fluorescence at 580 nm was recorded with equipped CCD camera (Spectral Instruments Inc.). Digital camera and monochromator were controlled by TILL Vision 4.0 imaging software, which was also used for data collection and processing. All procedures were performed at 37uC.

Statistical analysis
The data are shown as means 6 SEM of n separate experiments (n$3). Statistical analyses were done using STATISTICA 8.0 (StatSoft). Normally distributed data were analyzed with one-way ANOVA with Tukey's post-hoc test. In other cases, Kruskal-Wallis non-parametric ANOVA with post-hoc Dunn's test was applied. P-value ,0.05 was considered as statistically significant.

Results
The knockdown of PMCA2 or PMCA3 in stably transfected differentiated PC12 cells The expression of PMCA2 or PMCA3 mRNA was mostly abolished (,60% decrease) by an antisense mRNA targeted against it, but was not changed in mock-transfected cells (Fig. 1A). Similarly, PMCA2 or PMCA3 protein level was not affected by mock transfection but was decreased by ,50% following transfection with antisense-carrying vectors (Fig. 1B). Both PMCA2 or PMCA3 mRNA and proteins were normalized to endogenous GAPDH mRNA and protein levels, respectively. Unless otherwise stated, the experiments were performed using stably transfected lines.   to mitochondrial matrix, as demonstrated by enriched reactivity for an anti-GFP antibody and its high colocalization with Mito Tracker Red (Fig. S1B, Pearson coefficients: 0.8960.07 for C, n = 6; 0.9160.08 for _2, n = 8; 0.8660.04 for _3, n = 6). The transfection did not affect viability of the cells, which was in the range of 88-95% (data not shown). The mitoSypHer probe allows for dynamic pH measurement by monitoring the opposite changes in fluorescence at l ex = 430 and l ex = 485. To verify if PMCA2 or PMCA3 reduction could affect mitoSypHer spectral properties we performed in situ calibration, which showed a Hill slope of 0.9560.1, 0.9260.16 and 0.8960.09 for C, _2 and _3 lines, respectively (Fig. S1C). The switch from pH 7 to 10 resulted in impressive 18-fold increase in 485/430 ratio in all lines whereas in the pH range of 7-8 the observed rise was nearly 4-fold. The calibration curves of mitoSypHer in our differentiated PC12 lines closely matched those obtained in HeLa cells [13], demonstrating that mitoSypHer response was unaltered by PMCAs downregulation. Having validated the probe, we next performed simultaneous measurements of resting pH mito and pH cyto in single cells ( Fig. 2A) using mitoSypHer and cytosolic red-shifted fluorescent dye SNARF (5-(and 6)-carboxy-SNARF-1). The spectral properties of these indicators do not overlap allowing for efficient discrimination between pH changes between mitochondria and cytosol. Resting mitochondrial pH in _2 (7.7860.01) and _3 (7.6260.01) lines was notably higher in comparison to C (7.5360.02). Changes in pH cyto were in parallel to pH mito with the highest value noted in _2 (7.5660.02), followed by _3 (7.4960.02) and control cells (7.4160.03). In overall, the cytosolic pH was lower than mitochondrial in each line measured, consistent with chemiosmotic coupling hypothesis and experimental data [13]. As a consequence, pH gradient across the inner mitochondrial membrane (DpH = pH mito 2pH cyto ) was higher in _2 line.

PMCA2-and PMCA3-deficient cells maintain higher cytosolic and mitochondrial pH
PMCA2 and PMCA3 modulate the amplitude of K +evoked pH changes through regulation of intracellular Ca 2+ load To evaluate the effects of PMCA2 or PMCA3 reduction on pH changes during KCl-evoked Ca 2+ loads, PC12 lines expressing mitoSypHer were loaded with Fura-2 for simultaneous recording of (Ca 2+ ) c and pH mito . Depolarizing concentration of KCL was chosen because (i) it can mimic action potential-driven activation of voltage-dependent Ca 2+ channels (VDCCs) and Ca 2+ release from intracellular stores and (ii) PMCA is largely responsible for (Ca 2+ ) i restoration after such stimulation [30]. We observed, that cytosolic Ca 2+ elevations evoked by repetitive treatment with 59 mM KCl were in parallel with mitochondrial acidification (Fig. 3), however the magnitude of pH mito drop was PMCAsdependent. Interestingly, the degree of acidification in modified lines was inversely correlated with KCl-induced Ca 2+ load. While PMCA2-or PMCA3-reduction potentiated KCl-evoked (Ca 2+ ) c transients by 60618% and by 32613% in _2 and _3, respectively, during each stimulation the absolute pH mito response, in relation to control, was reduced by 5466% in _2 line and by 35611% in _3 line.
To follow DpH changes during KCl treatment we switched back to the concurrent recordings of pH mito and pH cyto (Fig. 4A). In all lines measured, the monophasic decay in pH mito typically exceeded those in the cytosol. The drop in pH cyto (Fig. 4B, i) and pH mito again (Fig. 4B, ii) was of smaller magnitude in _2 and _3 cells, but during each KCl stimulation DpH was undergoing more pronounced reductions in these cells than in control (Fig. 4, iii). The larger decrease in DpH was also observed during second stimulation. Upon KCl withdrawal, pH cyto returned to the resting  baseline, whereas pH mito typically surpassed its pre-stimulatory level. As a result, DpH was 0.1660.08, 0.2260.01, 0.2160.03 pH unit higher than in ''quiescent'' C, _2 and _3 cells, respectively, following recovery from the second stimulation (Fig. 4, iv).
PMCA2 and PMCA3 are the main sources of intracellular protons during (Ca 2+ ) c elevations generated by Ca 2+ entry through VDCCs To determine the contribution of particular Ca 2+ handling systems to cellular acidification, we next treated cells with thapsigargin (Tg) to inhibit sarco(endo)plasmic reticulum Ca 2+ -ATPase (SERCA) and with 2-APB (2-Aminoethoxydiphenyl borate), an inhibitor of store-operated calcium channels and IP 3 receptor. Under low extracellular Ca 2+ , Tg will deplete the ER and 2-APB will block its repletion through store-operated calcium entry (SOCE), when Ca 2+ will be restituted. As shown, in Fig. 5A re-addition of external Ca 2+ together with KCl showed a massive Ca 2+ influx and concomitant decrease in pH mito with the magnitude comparable to non-inhibitory conditions. Moreover, we observed that PMCAs downregulation slowed down Ca 2+ clearance following extracellular Ca 2+ removal, and the pH recovery was delayed until (Ca 2+ ) c was nearly at the resting level (Fig. 5A, insets). Thus, SOCE is not required for Ca 2+ -dependent mitochondrial acidification and SERCA is not a main producer of intracellular H + in our experimental model. Additional experiments with transiently transfected cells confirmed that PMCA2 or PMCA3 knockdown affected cellular pH response to KCl-induced (Ca 2+ ) c influx and similar profiles of (Ca 2+ ) c and DpH changes were observed in conditions with or without Tg (Fig. 6). This strengthen our conclusion regarding predominant role of neurospecific PMCA isoforms in the regulation of pH excursions in PC12 cells.
Because the stimulating effect of KCl results from membrane depolarization with subsequent opening of VDCCs [31], we next examined if the activation of these channels may represent a main source of Ca 2+ influx. Indeed, blockage of voltage-dependent calcium current by cadmium markedly reduced the amplitude of Ca 2+ transients and completely abolished subsequent intracellular pH changes (Fig. 5B).
Further, we attempted to verify NCX role by treating cells with Tg and replacing Na + with Li + in a buffer (Fig. 5C). The pH curves during first KCl stimulation matching those obtained under non-inhibitory conditions indicated neglectable NCX participation in observed pH changes. These conditions also allowed us to refine the activity of PMCA, which was directly proportional to the rate of (Ca 2+ ) c decrease upon the removal of extracellular Ca 2+ . La 3+ (5 mM), which is known to inhibit PMCA, added during second stimulation blocked Ca 2+ clearance and resulted in mitochondrial alkalinization even under low extracellular K + . We also inhibited PMCA by reducing the availability of H + to be exchanged with cytosolic Ca 2+ , by increasing extracellular pH up to 9 (Fig. 5D). This markedly delayed Ca 2+ recovery upon KCl removal, whereas pH restoration to 7.4 resulted in a rapid activation of (Ca 2+ ) c clearance, pointing out inhibition of PMCA under high extracellular pH. Also, alkaline pH completely attenuated KCl-induced pH mito decrease in all lines. Because all the conditions that inhibited PMCA every time decreased cellular acidification, PMCA may be considered as a main source of intracellular H + during KCL-evoked Ca 2+ loads. Thus, markedly decreased Ca 2+ -induced pH response in _2 and _3 lines can be attributed to diminished level of PMCA2 and PMCA3 isoforms.

Electron transport chain contributes to PMCA-dependent mitochondrial H + fluxes
Because PMCAs-dependent acidification of mitochondria during (Ca 2+ ) c transients was shown in this study to occur in parallel with cytosolic pH drop, we attempted to evaluate if the electron transport chain (ETC) may regulate cytosolic H + influx to the matrix. We first blocked ETC with rotenone (inhibitor of complex I) or KCN (inhibitor of complex IV). Application of inhibitors alone caused immediate decrease in pH mito in all lines, matching the pH mito response during first KCl stimulation before the inhibitors were added (compare first and second KCl stimulation in Fig. 7 A and B). Additionally, each of the inhibitors reduced KCl-evoked pH mito decrease to 4069% in C, 7465% in _2 and to 5967% in _3 of the value noted in these lines when no inhibitors were present. We then blocked ATP synthase by oligomycin (Fig. 7C). However, we did not observe expected pH mito increase over 5-min incubation period possibly due to maximal alkalization of mitochondria following first KCl stimulation. In each line, oligomycin exerted only moderate effect on the magnitude of Ca 2+ -dependent pH mito decrease.

Increased DpH coincided with elevated TMRE fluorescence in PMCA2 knock-down cells
Based on the results obtained from DpH imaging, we next measured mitochondrial membrane potential (DY m ), which is thought to reflect mitochondrial energization state. By using nonquenching concentration of TMRE (25 nM) we also determined whether PMCA2-or PMCA3-reduction may trigger depolarization during (Ca 2+ ) c transients. First, we observed increased TMRE fluorescence intensity in _2 and _3 cells in steady-state conditions in relation to control (Fig. 8A). Because TMRE uptake is sensitive to changes in either DY m and plasma membrane potential (DY p ), in parallel experiment we monitored DY p using DiSBAC 2 . Elevated DY p -related fluorescence observed in _2 and _3 lines indicated altered plasma membrane potential (Fig. 8A). To distinguish the relative contribution of DY m and DY p to observed TMRE fluorescence increase, we depolarized Y p with 59 mM KCl before loading with TMRE (Fig. 8B). Pretreatment with high K + exerted, however, only a small effect on TMRE suggesting that differences in signal intensity between lines were due to DY p . Additionally, TMRE uptake was not affected by increased mitochondrial biogenesis, as neither changes in expression of Tfam, Nrf-1 and Pgc-1a considered as mitochondrial biogenesis markers nor mitochondrially-encoded subunits I and III of cytochrome c-oxidase reflecting the copy number of mitochondrial DNA were detected (Fig. 8C). In addition, mitochondrial mass was unchanged in PMCA-deficient lines, as evaluated using Mitotracker Green FM probe (Fig. 8D).
To ensure that higher TMRE signal was not a result of dye release during loading and consequent unqenching, TMREloaded cells were treated with protonophore FCCP (1 mM) or oligomycin (6 mM). FCCP-induced depolarization resulted in massive decrease in TMRE signal in all lines coinciding with a slight increase in DY p . Application of oligomycin used to block protons re-entry into the matrix caused a small but significant DY m hyperpolarization notably higher in _2 and _3 lines without affecting DY p . The loss of punctuate TMRE signal, as a result of TMRE release during depolarization by FCCP, was also observed in individual cells (Fig. 9A). Moreover, FCCP evoked a significantly higher rise in (Ca 2+ ) c in _2 and _3 lines than in control (Fig. 9B), whereas the application of oligomycin did not change (Ca 2+ ) c (Fig. 9C). This demonstrates that basal state of mitochondrial Ca 2+ loading is increased in PMCA-deficient cells, particularly in _2 line.

KCl-evoked DY m depolarizations in PMCA2-deficient line are blocked by cyclosporin A or bongkrekic acid
To evaluate whether (Ca 2+ ) c elevations can affect DY m , TMRE fluorescence was measured at selected time points of KCl stimulation or recovery, at which DpH alterations were the most pronounced: 10 min after 1 st KCl stimulation, 10 min after KCl removal (1 st recovery phase), 10 min after 2 nd KCl stimulation and 10 min after 2 nd KCl removal (2 nd recovery phase). In control and _3 line, we observed only minor alterations in DY m during KCl treatment but these little depolarization events did not correlate with the amplitude of (Ca 2+ ) c transients, and were considered as insignificant. In contrast, in _2 line each KCl stimulation evoked DY m depolarization with subsequent decrease in TMRE fluorescence by 51618% in relation to resting level. (Fig. 10). We then treated cells with cyclosporin A (CsA), a potent inhibitor of mPTP, which fully rescued the reduced TMRE fluorescence. Because CsA is also a well-known inhibitor of calcineurin [32], we used bongkrekic acid (BA) which inhibits mitochondrial ATP/ADP translocase without affecting calcineurin activity. BA partially rescued the reduced TMRE fluorescence while an inhibitor of calcineurin (FK-506) was not able to preserve mitochondria from DY m loss during KCl-induced Ca 2+ loads.
To validate the effects of CsA or BA we induced mitochondrial swelling which correlates with decrease in light scattering. We found that addition of 10 mM Ca 2+ to the mitochondrial suspension induced a small but significant swelling in _2 line (Fig. S2) whereas atractylate added before Ca 2+ exposure resulted in extensive swelling in all lines. In both cases, swelling was fully prevented by 1 mM CsA or 10 mM BA, added before Ca 2+ . Therefore, our results indicate that Ca 2+ -driven DY m collapse in PMCA2-deficient line is mediated through CsA-sensitive mechanism.

Discussion
Despite the proposed predominant role of PMCA in Ca 2+dependent regulation of organellar pH, so far no reports have evaluated the contribution of particular PMCA isoforms to mitochondrial proton gradient. Moreover, no studies have been attempted to answer whether altered PMCA expression and concomitant disturbances in Ca 2+ signaling may affect intracellular pH. The study on deafwaddler mouse indicated that the reduction in PMCA2 expression by half may result in motor neuron dysfunctions and mediate neuronal death [33]. Therefore, to avoid dramatic compromise on cellular viability, we have obtained homogenous neuron-like PC12 stably transfected clones with nearly 50% decrease in PMCA2 or PMCA3 protein, with yet no visible symptoms of increased mortality. This allowed us to analyze if neuron-specific PMCA isoforms modulate Ca 2+ -driven intracellular pH changes.
The resting mitochondrial pH in our PC12 lines was lower in comparison to certain cell lines [34][35][36] but similar to the values reported in other [37,38]. Therefore, it seems that particular cell types maintain different resting pH to fulfill their specific functional requirements. The additional differences in intracellular pH were seen between our PC12 lines: the highest pH mito and pH cyto values were noted in _2 line, followed by _3 and control. Heterogeneous increase in basal pH mito was observed in HeLa cells and primary cultured neurons upon stimulation with Ca 2+mobilizing agents [34]. Elevated pH mito and pH cyto demonstrated in our PMCA-downregulated cells in steady-state conditions also  suggested a dependence on calcium level. Indeed, our previous study has shown that PMCA2 or PMCA3 reduction caused an increase in resting (Ca 2+ ) c [39]. Because PMCA transports large quantities of protons during Ca 2+ extrusion, parallel acidification of cytosol and mitochondria is expected if the activity of PMCA remains unaffected. Such a phenomenon has been demonstrated in cortical neurons stimulated with glutamate [12]. Based on PMCA/pH relationship, the extent of matrix alkalization in _2 and _3 lines may reflect reduced level (and activity) of neurospecific isoforms. We suppose that the knock-down of PMCA2 or PMCA3 which are counted as fast reacting, dramatically reduce the amount of H + entering cytosol leading to pH mito increase. The regulation of mitochondrial pH and function by cytosolic Ca 2+ transients requires the uptake of Ca 2+ to mitochondria and both Ca 2+ -dependent alkalization or acidification of matrix have been demonstrated [38,40,41]. The accumulation of Ca 2+ depends on DYm-driven electrochemical Ca 2+ gradient and the gradient of this ion between cytosol and mitochondria. Whether Ca 2+ uptake into mitochondria is through mitochondrial uniporter or Ca 2+ /H + exchanger, it should depolarize energized mitochondria (reviewed in [6]).
Tetramethylrhodamine probes have been widely used to monitor DY m [42,43]. However, TMRE uptake is also sensitive to DYp which may impact the amount of TMRE entering the cytoplasmic space, thereby affecting how much dye is available for mitochondria. Therefore, to resolve the potential contribution of DY p and DY m to increased TMRE fluorescence in _2 and _3 lines we depolarized plasma membrane with 59 mM K + before TMRE loading. This strategy was also used by Krohn et. al. [44] or by Perry et. al. [45]. Taking into account the Nernstian behavior of TMRE probe and the results presented in Fig. 8B, one may suggest that even in the presence of high K + , elevated TMRE fluorescence is almost entirely dependent on DY p . It is in agreement with DY p and/or DY m dependency of TMRE uptake. We confirmed the reliability of TMRE to measure membrane potential by using FCCP and oligomycin. It is known that lower FCCP concentrations will specifically collapse DY m , while high concentrations (.2.5 mM) will also significantly diminish DY p [43]. However, this effect is likely to be variable with cell type. In our study we applied 1 mM FCCP and observed slight hyperpolarization of DY p , similarly to the effect reported in [46]. The paradoxical FCCP-induced increase in DiSBAC 2 signal can also be due to the fact that plasma membrane potential is created not by proton pump so protonophore cannot short-circuit it. Instead, FCCP equilibrates pH across plasma membrane carrying positively charged protons from cytoplasm to outside medium thus generating higher membrane potential. Treatment with oligomycin caused a moderate increase in pH mito what is consistent with slight hyperpolarization shown in Fig. 8A. This indicates low rate of ATP turnover and state of mitochondria close to state 4.
In agreement with the prediction that mitochondrial Ca 2+ uptake may elevate DpH if Ca 2+ charge is compensated by protons moving through ETC, we found DpH to be increased in _2 line. The observed raise in matrix pH in steady-state conditions could then result from charge compensation by the respiratory chain. It is attractive to propose that mitochondrial Ca 2+ accumulation in _2 line will represent a major trigger coupling pH changes to the rate of ATP synthesis. Indeed, three matrix dehydrogenases activated by (Ca 2+ ) m increases [47][48][49] provide reducing equivalents to ETC without affecting matrix acidification. This may reflect increase in PMF when (Ca 2+ ) m responds to (Ca 2+ ) c elevations. Alternatively, elevation in PMF could result from the inhibition of pathways that dissipate H + gradient. It has been reported that mitochondrial Ca 2+ uptake may inhibit ATP synthase [50] consequently increasing PMF and reducing ATP level. However, our observations with FCCP and oligomycin as well higher ATP content detected in _2 line (yet unpublished) rather exclude ATP synthase inhibition as a mechanism of PMF increase. Therefore, the increases in pH mito may indicate higher capacity of mitochondria to produce ATP.
In our study we detected pronounced cellular acidification associated with (Ca 2+ ) c elevations, however with a PMCAsdependent magnitude. Bearing in mind that PMCA regulates the amount of protons entering cytosol, downregulation of fast responsive PMCA2 or PMCA3 isoforms may explain weaker pH response in _2 and _3, even despite potentiation of Ca 2+ influx in these lines during KCl stimulation. Different amplitudes of pH mito decreases between our lines could also reflect altered proton buffering capacity. Higher mitochondrial pH in _2 and _3 lines in steady-state conditions might affect DpH drop during KCl stimulation, as reduced buffering pH capacity of mitochondria in the alkaline pH was shown to underlie the loss of DpH upon treatment with Ca 2+ mobilizing agents [reviewed in 6]. Following successive stimulations, we observed the overshoot of DpH and a new resting (Ca 2+ ) c particularly visible in _2 line. This effect is most likely due to over-activation of mitochondrial matrix dehydrogenases by Ca 2+ transients but also indicate, that PMCAsdownregulated cells irreversibly lose a substantial part of Ca 2+ clearing potency. This is additionally supported by the observed KCl-evoked higher Ca 2+ influx in _2 and _3 lines. Our previous study demonstrated increased expression and concomitantly greater contribution of certain VDCCs to Ca 2+ influx in PMCAs-deficient lines [39]. Because colocalization of these channels and PMCA has been shown in specific types of neurons [51], we assume their functional relationship in the regulation of Ca 2+ influx in _2 and _3 cells. It is now apparent that mitochondria of some cell types can accumulate large amounts of Ca 2+ during membrane depolarization events [52,53]. Facing mitochondria to domains of high (Ca 2+ ) c allows direct mitochondrial Ca 2+ uptake following VDCCs activation and rapid uptake mode of the mitochondrial uniporter in response to extramitochondrial Ca 2+ bursts. Nonetheless, even under the disturbed Ca 2+ homeostasis and despite variations in the absolute cellular pH, all cell lines retained the ability to maintain positive matrix vs. cytosol gradient.
Our data show that the main function of protons transport during Ca 2+ load can be attributed to PMCA2, and to weaker extend also to PMCA3 because: (1) reduction of their level led to lower degree of mitochondrial acidification as less protons entered cytoplasm; (2) the acidification did not require Ca 2+ release from internal stores but was related to plasma membrane Ca 2+ influx through VDCCs; (3) all agents used to inhibit PMCA prevented KCl-induced pH drop and markedly delayed Ca 2+ clearance.
(Ca 2+ ) c elevations and subsequent uptake by mitochondria should result in DY m dissipation to restrict the availability of mitochondria to synthesize ATP. Decreases in DY m have been observed in isolated mitochondria exposed to Ca 2+ overload [54]. In intact cells transient depolarizations have been reported only in some cell types [55][56][57], but not in other [58,59]. The present study also found no detectable alterations in DY m in control and _3 lines, despite large pH mito and pH cyto drop during KCl-evoked (Ca 2+ ) c elevations. In agreement with the statement that mitochondrial Ca 2+ uptake must affect DY m , the depolarization events could be too faint to be detected in control and _3 lines. In turn, DY m depolarizations did occur in _2 cells in response to (Ca 2+ ) c elevations. We can assume it could be due to higher DY mdependent mitochondrial Ca 2+ uptake shown in this line, as DY m -driven elevation of mitochondrial Ca 2+ may itself dissipate DY m [60,61]. Ca 2+ influx through VDCCs resulting in DY m loss was shown in CA1 pyramidal cells in hippocampal slides [62]. It also seems possible that a rise in (Ca 2+ ) c and then in (Ca 2+ ) m may depolarize DY m through promotion of Ca 2+ cycling or by decreasing the ATP/ADP?P i ratio due to higher ATP consumption by Ca 2+ -dependent ATPases. This would in turn increase proton backflow to the mitochondrial matrix, depolarizing DY m and stimulating respiration. The net Ca 2+ accumulation may occur through the mitochondrial uniporter which activity in neural tissue is particularly high [63]. The entry of positively charged ions could then lower DY m allowing net H + extrusion by the ETC with the consequent increase in DpH. Another possible mechanism may involve Ca 2+ -dependent inhibition of ETC, as was demonstrated in mitochondria exposed to increasing Ca 2+ concentration [64][65][66]. However, at this stage we are unable to distinguish which portion of DY m changes during stimulation were due to collapsing of proton gradient or the exchange of charged molecules (e.g. Ca 2+ , P i , ADP).
Here, we report that DY m depolarization in PMCA2-deficient cells is mediated by the activation of CsA-sensitive mechanism. Studies from neuronal and non-neuronal cells suggest that during ion imbalance mitochondria depolarize, swell and release cytochrome c through CsA-sensitive Ca 2+ -activated mPTP opening [67][68][69][70]. In our model, mitochondrial Ca 2+ overload may lead to transient mPTP opening resulting in DY m collapse, outward Ca 2+ redistribution and matrix acidification. However, contrary to catastrophic nature of mPTP opening, our data demonstrate that DY m recovered upon KCl removal. This suggests that respiratory chain rebuilt the proton gradient and restored DY m , which may drive Ca 2+ re-uptake and its gradual accumulation in the matrix. Perhaps, only brief mPTP opening could be sufficient to trigger subsequent death in PMCA2-deficient cells. Because some studies have reported that neuronal mPTP is relatively CsA-insensitive [54,71], alternative mechanisms such as reactive oxygen species release or adenine nucleotide depletion should also be considered. Additionally, reduced mitochondrial H + concentration may by itself trigger mPTP opening, as an acidic pH was reported to block the opening of mPTP [72,73]. In line with it, more pronounced acidification observed in control and _3 cells may explain why DY m is not dissipated in these lines during KCl stimulation.
In summary, we showed that PMCA2 and PMCA3 are responsible for dynamic regulation of cellular pH. In steady-state conditions, concomitant elevation of (Ca 2+ ) c and higher Ymdependent accumulation of mitochondrial Ca 2+ , and/or decreased influx of cytosolic H + due to PMACA knock-down, may lead to mitochondrial alkalization. It is believable as the amount of H + entering cytosol in exchange for Ca 2+ seems to depend on the kinetic properties of PMCA isoforms. This could explain why pH response observed during (Ca 2+ ) c elevations was modulated in a manner dependent on isoform activity: the smallest response when PMCA2 was downregulated, which is regarded as the fastest reacting, followed by PMCA3 which is only slightly slower than PMCA2. However, during massive Ca 2+ loads, the potentiation in Ca 2+ influx observed in _2 line and, as a consequence, mitochondrial Ca 2+ overload may lead to DYm depolarization. Our data indicate that DYm collapse was triggered by CsAsensitive mechanism suggesting the involvement of mPTP opening as a possible underlying mechanism. Lack of signs for mPTP formation in _3 cells could indicate that the threshold Ca 2+ concentration required for Ca 2+ -dependent mPTP opening has not been achieved although an increased Ca 2+ influx during membrane depolarization was also observed in these cells. The overall data indicate that the relationship between mitochondria and PMCA is much more complex and intimate and exceeds far beyond a simple energetic connection. Our findings provide the evidence, that PMCA membrane composition might be of great importance for preservation of bioenergetic function of mitochondria. Therefore, changes in PMCA expression occurring i.e. in ageing brain or spinal cord injury [74,75] may profoundly affect cellular metabolic network and disturb mitochondrial function. In view of this, pathological alterations in PMCA expression, in particular PMCA2, may contribute to neurotransmission dysfunctions via a mechanism of mitochondrial depolarization. Undoubtedly, elucidating of the functional interplay between mitochondrial metabolism and neuronal function is of paramount importance for understanding of pathophysiology in various neurological diseases. Figure S1 In vitro characterization of mitoSypHer probe in differentiating PC12 cells. (A) The expression of mitoSypHer vector (i, SypHer) and the corresponding protein content (ii, anti-GFP) assessed using PCR or monoclonal anti-GFP antibodies, respectively. GAPDH was used as an internal control. Swelling was assessed by light absorbance at 520 nm in a suspension of mitochondria. The absorbance at time 0 (before Ca 2+ exposure) was taken as 100%.