Calnexin as a metabolic nexus
Ca2+ flow from the ER to mitochondria promotes ATP production through oxidative phosphorylation. Gutiérrez et al. investigated the role of calnexin, an ER chaperone that interacts with SERCA, the ATPase that pumps Ca2+ into the ER, in mitochondrial bioenergetics. The authors found that calnexin maintained SERCA in a redox state that was optimal for activity. Mitochondria were closer to the ER in cells without calnexin than in cells with calnexin. This enabled calnexin knockout cells to partially rescue Ca2+ influx into mitochondria and to perform limited oxidative phosphorylation that was supplemented with increased glycolysis. These data suggest that calnexin positions mitochondria to regulate Ca2+ flow from the ER and respiration.
Abstract
Chaperones in the endoplasmic reticulum (ER) control the flux of Ca2+ ions into mitochondria, thereby increasing or decreasing the energetic output of the oxidative phosphorylation pathway. An example is the abundant ER lectin calnexin, which interacts with sarco/endoplasmic reticulum Ca2+ ATPase (SERCA). We found that calnexin stimulated the ATPase activity of SERCA by maintaining its redox state. This function enabled calnexin to control how much ER Ca2+ was available for mitochondria, a key determinant for mitochondrial bioenergetics. Calnexin-deficient cells compensated for the loss of this function by partially shifting energy generation to the glycolytic pathway. These cells also showed closer apposition between the ER and mitochondria. Calnexin therefore controls the cellular energy balance between oxidative phosphorylation and glycolysis.
INTRODUCTION
Interorganellar contacts between the endoplasmic reticulum (ER) and mitochondria were first described in liver tissue recovering from prolonged nutrient deprivation (1–3). These ER-mitochondria contacts undergo dynamic changes and become most prominent immediately upon the removal of nutrition (4, 5). These findings suggest a mechanistic link between these structures and the control of cellular metabolism (6). In mammalian cells, ER-mitochondria Ca2+ signaling is central to this function (7). Ca2+ signaling requires the tethering of the ER to mitochondria, often through multimeric protein complexes (8) such as the inositol 1,4,5-trisphosphate receptor (IP3R)/voltage-dependent anion channel (VDAC) complex (9) or the ER-mitochondria encounter structure (ERMES). This multimeric protein complex (8) is represented in mammalian cells by the paralog of its Mmm1 component (10), the mammalian PDZ domain containing protein 8 (PDZD8) (11). However, the molecular machinery allowing ER-mitochondria contacts to react to metabolic and stress insults remains incompletely understood.
An intriguing, universal link exists between the enzymes that mediate oxidative protein folding and that control mitochondrial bioenergetics (12). Over the years, this link has been solidified through the discovery of numerous chaperones and folding assistants that cofractionate with mitochondria-associated membranes (MAMs) (13). These proteins include Ero1α (14) and ERp44 (15), two oxidoreductases that compete to bind to IP3Rs. In addition, the chaperone calnexin and the disulfide isomerase TMX1 have been found in MAMs in unbiased proteome studies (16, 17). Both calnexin (18) and TMX1 interact with sarco/endoplasmic reticulum Ca2+ adenosine triphosphatase (ATPase) (SERCA) at MAMs (19). Through this interaction, TMX1 decreases ER Ca2+ pumping, thus directing Ca2+ flux toward mitochondria. In the absence of TMX1, mitochondria move closer to the plasma membrane to compensate for their Ca2+ needs in the presence of hyperactive SERCA (20). Typically, however, SERCA activity positively aligns with mitochondrial metabolism, as shown by the activation of oxidative phosphorylation (OXPHOS) observed upon inhibition or knockdown of another SERCA inhibitor, Bcl-2 (21, 22). Likewise, the knockout of the SERCA activator p53 results in increased sulfenylation of SERCA (23) but decreased mitochondrial OXPHOS (24, 25). Although results from our laboratory have implicated the chaperone calnexin in the activation of SERCA (18), our data seemed to contradict an earlier study showing that overexpression of calnexin reduces cytosolic Ca2+ waves (26), which is associated with reduced SERCA activity (27). In the present manuscript, we identified the mechanistic causes and consequences of these observations and demonstrated a critical role for calnexin in controlling mitochondrial positioning and metabolism.
RESULTS
Calnexin activates SERCA by maintaining its redox state
We previously proposed that calnexin activates SERCA (18), apparently contradicting previous observations that had been interpreted as calnexin acting as an inhibitor of SERCA (26). However, neither study took into account the complex interplay between ER and mitochondria Ca2+ storage and buffering and membrane contact formation between the two organelles. SERCA activity is highest when luminal and cytosolic cysteines are partially oxidized (28, 29). Although hyperoxidized SERCA is inactive, a baseline level of SERCA oxidation is necessary for its activity (30). Thus, we determined the SERCA2b redox state in calnexin wild-type and knockout mouse embryonic fibroblasts (MEFs) in the presence of a cysteine-recognizing biotinylated compound. We performed a biotinylated iodoacetamide (BIAM) switch assay on Myc immunoprecipitates from wild-type and knockout MEFs transfected with Myc-tagged SERCA2b (31, 32). Under control conditions, SERCA2b was partially oxidized in wild-type MEFs, as determined from baseline signals established with dithiothreitol (DTT) and diamide treatment (Fig. 1A). In contrast, SERCA2b from calnexin knockout MEFs was fully reduced (Fig. 1A). Similarly, SERCA2b lost its partial oxidation in HeLa cells transfected with calnexin-directed small interfering RNA (siRNA) (siCNX) (fig. S1A). This finding suggested that calnexin was required to keep SERCA2b active. Consistent with this hypothesis, we detected a 35% reduction in SERCA-derived ATPase activity in cellular membranes from calnexin knockout MEFs as compared to those from wild-type MEFs (Fig. 1B). In siCNX-transfected HeLa cells, this reduction was not significant (fig. S1, B and C). In contrast, calnexin knockout cells stably reconstituted with wild-type calnexin (fig. S1, D and E) showed increased SERCA activity.
(A) Analysis of SERCA2b for the exposure of reduced cysteines, as detected with biotinylated iodoacetamide (BIAM). Cells were analyzed by Western blot for the Myc and BIAM signals. The BIAM signal from knockout MEFs was normalized to the signals from wild-type MEFs (n = 4 biological replicates for each group, *P = 0.001 by one-sample t test). (B) Light membranes were biochemically isolated from calnexin wild-type and knockout MEFs and cellular SERCA ATPase activity was determined in the presence of the SERCA inhibitor thapsigargin (n = 7 biological replicates for each group, *P = 0.04 by two-sample t test). SERCA ATPase activity was normalized to protein concentration in light membranes in calnexin wild-type and knockout MEFs (n = 7 biological replicates for each group, *P = 0.04 by one-sample t test). (C) Mathematical modeling of cytoplasmic Ca2+ waves, taking into account activating calnexin modification of SERCA, as predicted in wild-type and calnexin knockout cells. (D) ER Ca2+ was assayed upon tert-BuBHQ-mediated Ca2+ release from the ER in wild-type and calnexin knockout MEFs transfected with ER-GECO. The drop of the fluorescent signal was recorded and average drops (n = 12 biological replicates for each group, ***P = 0.000006 by one-sample t test) and depletion slopes (n = 13 biological replicates for each group, *P = 0.01 by one-sample t test) were quantified. (E) Cytoplasmic Ca2+ was quantified by fluorimeter in wild-type and knockout MEFs loaded with Fura-2. ATP (600 μM) was added to open IP3Rs and induce Ca2+ efflux from the ER, as indicated. Cytoplasmic Ca2+ clearance was determined by the decay slope (n = 6 biological replicates for each group, **P = 0.002 by one-sample t test). The peak of cytoplasmic Ca2+ was also determined (n = 6, P = 0.31 by one-sample t test).
To determine the relevance of these findings to cytosolic Ca2+ levels, we first used our mathematical Ca2+ signaling model (33). We set the Hill function as activating to be consistent with our results suggesting that calnexin stimulated the ATPase activity of SERCA. We entered calnexin at a molarity of 0.05 and 0 μM into our model, corresponding to wild-type and knockout MEFs, respectively. Moreover, we also entered an ideal resting ER-mitochondria distance of 30 nm (34). The simulation results (Fig. 1C) showed that calnexin slowed down the frequency of cytosolic Ca2+ waves by 56%, consistent with published observations (26) leading to overall reduced availability of Ca2+ in the cytoplasm. Running our mathematical model with an increased ER-mitochondria distance of 150% again showed that calnexin slowed down the frequency of cytosolic Ca2+ waves by 32%, whereas entering calnexin as a SERCA inhibitor (by setting the Hill function as negative) had the opposite effect and further decreased cytosolic Ca2+ waves (fig. S1F).
To further test whether calnexin activated SERCA, we determined how calnexin affected Ca2+ availability within the ER and the cytoplasm. We used the ER-targeted Ca2+ probe ER-RGECO to measure the total amount of Ca2+ in the ER in cells treated with the SERCA inhibitor tert-BuBHQ (TBHQ) and upon removal of extracellular Ca2+ to completely deplete ER stores (Fig. 1D). In calnexin knockout MEFs, the ER-RGECO signal was reduced by 50% and the rate of the Ca2+ drop was decreased, thus suggesting that these cells not only have less ER Ca2+ content but also retain ER Ca2+ better (Fig. 1D, right). Consistent with the ATPase results (fig. S1C), the ER Ca2+ content did not differ substantially between control and siCNX-transfected HeLa cells (fig. S1G). In contrast, the reconstitution of calnexin knockout MEFs with wild-type calnexin partially restored ER Ca2+ content (fig. S1H).
Use of the translocon inhibitor anisomycin abolished the difference in leak between wild-type and knockout MEFs, suggesting that calnexin promoted Ca2+ leakage from the ER through the translocon (fig. S2A). The reduction in ER Ca2+ in calnexin knockout MEFs could be due to decreased store-operated Ca2+ entry (SOCE), which we tested by adding Ca2+ to the medium of previously Ca2+-starved cells while inhibiting SERCA. Calnexin knockout MEFs took up more Ca2+ under this condition (fig. S2B), suggesting that calnexin may inhibit SOCE but ruling out decreased SOCE as an explanation for the reduced Ca2+ content in calnexin knockout MEFs. Last, we measured the speed at which Ca2+ was cleared from the cytosol after a controlled release of Ca2+ through IP3R opening using the cytosolic Ca2+ indicator dye Fura-2 (Fig. 1E). Calnexin knockout MEFs exhibited >50% reduction of the cytosolic Ca2+ clearance speed compared to wild-type MEFs, corroborating reduced SERCA activity in the absence of calnexin (Fig. 1E, left). The amplitude of the Ca2+ signal in the cytoplasm was unchanged (Fig. 1E, right). Together, our data showed that calnexin determines the ER Ca2+ content predominantly by maintaining the SERCA redox state and increases SERCA-mediated Ca2+ uptake from the cytoplasm. Within the cytoplasm, calnexin reduced the availability of Ca2+ by increasing its clearance.
Calnexin influences mitochondria positioning
The spatial arrangement of mitochondria depends on the amounts of cytosolic Ca2+, which at high levels arrest mitochondria by blocking mitochondrial motor proteins (35, 36), thus explaining why the ER-mitochondria distance depends on cytosolic Ca2+ levels (37, 38). Because calnexin knockout delayed Ca2+ clearance, thus potentially increasing Ca2+ levels in the proximity of the ER, we hypothesized that the absence of calnexin could reduce ER-mitochondria distance. We first modeled the effects of calnexin on ER-mitochondria distance based on the inhibitory effect of cytosolic Ca2+ for mitochondria movement (37, 38). In our model, we fixed r0, the minimum distance derived from spatial constraints between the ER and the inner mitochondrial membrane (IMM) at a constant value in the range from 5 to 35 nm. Next, we set rc as the maximum modulatory distance of 35 nm. We then corrected rc to take into account the repelling force of low levels of [Ca2+]cyt that are more frequent if waves are inhibited. This effect increases the actually expected distances (r = r0 + rc) to at least 25 to 30 nm (Fig. 2A). From published results (26) and our measurements and modeling (Fig. 1, C and D), we expected rc to be lower in calnexin knockout cells, leading to reduced distance in this case. At an r0 of 20 nm, which is close to the minimum physically possible ER-IMM distance (34), the predicted effective distance r was 29 nm for calnexin knockout MEFs and 42 nm for wild-type MEFs (Fig. 2A). This 31% decrease in the distance between ER and mitochondria in calnexin knockout MEFs would be consistent with the 50% slower Ca2+ clearance in calnexin knockout MEFs (Fig. 1D) that would be expected to increase cytosolic Ca2+ puffs and oscillations (Fig. 1C).
(A) Mathematical modeling of ER-mitochondria distance, taking into account calnexin expression (WT = 0.05 μM calnexin; KO = no calnexin). Depending on r0, the minimum distance between the two organelles, rc was determined as a function of cytosolic [Ca2+]. r corresponds to the sum of the two and is the postulated actual distance between the ER and mitochondria. (B) Representative electron microscopy images of calnexin wild-type and knockout MEFs. Mitochondria are labeled in red and ER is labeled in green. Scale bar, 200 nm. (C) Average ER-mitochondria distance as determined from electron micrographs (P = 0.07 by two-sample Wilcoxon test; n = 41 MERCs for wild-type MEFs, n = 51 MERCs for knockout MEFs). (D) Average ER-mitochondria contact length as determined from electron micrographs (P = 0.15 by two-sample Wilcoxon test; n = 41 MERCs for wild-type MEFs, n = 51 MERCs for knockout MEFs). (E) Coefficient of ER-mitochondria contact length divided by ER-mitochondria distance (*P = 0.03 by two-sample Wilcoxon test; n = 41 MERCs for wild-type MEFs, n = 51 MERCs for knockout MEFs). (F) Split of distance-sorted ER-mitochondria contacts into super-tight contacts (<12 nm), tight contacts (12 to 24 nm), and loose contacts (24 to 50 nm), derived from (B) to (E). (G) SPLICSshort signals were determined by quantifying signal dots per cell (Control: MEF CNX WT n = 11 cells, MEF CNX KO n = 23 cells, P = 0.81 by two-sample Wilcoxon test; NAC: MEF CNX WT n = 9 cells, MEF CNX KO n = 19 cells, P = 0.26 by two-sample Wilcoxon test). Scale bar, 20 μm. (H) SPLICSlong signals were determined by quantifying signal dots per cell (Control: MEF CNX WT n = 8 cells, MEF CNX KO n = 9 cells, *P = 0.01 by two-sample Wilcoxon test; NAC: MEF CNX WT n = 8 cells, MEF CNX KO n = 10 cells, *P = 0.04 by two-sample Wilcoxon test). Scale bar, 20 μm.
Next, to corroborate these in silico results in cells, we determined the ER-mitochondria average distance in calnexin wild-type and knockout MEFs on electron micrographs (Fig. 2B). The observed average distance between the ER and the outer mitochondrial membrane was 20 nm in calnexin knockout MEFs and 26 nm in calnexin wild-type MEFs, or about 30% wider (Fig. 2C), which correlated well with the predictions of our mathematical model. Moreover, we also detected increased mitochondria-ER contact (MERC) length in calnexin knockout MEFs (Fig. 2D), which led to an overall significantly increased MAM coefficient in calnexin knockout MEFs over calnexin wild-type MEFs (Fig. 2E). The analysis of the frequency of the different MERC types, classified as super tight (0 to 12 nm), tight (12 to 24 nm), and loose (24 to 50 nm), showed that the absence of calnexin promoted an increase in the frequency of super tight MERCs and a reduction in that of loose MERCs (Fig. 2F). We also analyzed calnexin knockout MEFs with a split-green fluorescent protein–based contact site sensor (SPLICS) system that can quantify tight (SPLICSshort) and loose contacts (SPLICSlong) but not super-tight MERCs (39). This system showed no difference in tight MERCs (Fig. 2G), but in contrast to our electron micrograph analysis, we found more signals for loose MERCs in calnexin knockout MEFs (Fig. 2H). This slightly different result could derive from the larger sample size of the SPLICS assay. Together, our results demonstrate in silico and in cells that the activation of SERCA by calnexin controlled Ca2+ clearance on the cytosolic face of the ER and apposition between the ER and mitochondria.
Calnexin controls mitochondrial Ca2+ responses
Our results so far indicated that calnexin increased ER Ca2+ content, cytosolic Ca2+ clearance, and ER-mitochondria apposition. Each of these properties might alter availability of Ca2+ at mitochondria, but the final mitochondrial Ca2+ and membrane potential readouts might depend on the combination of effects or the effects might cancel each other out. To distinguish between these two possibilities, we used the sensitive mitochondria-targeted Ca2+ probe mito-RGECO, which allows a wider detection range of Ca2+ spikes in mitochondria than the high-affinity Rhod2 probe (40). IP3R-mediated Ca2+ release from the ER and uptake into mitochondria did not result in significantly increased Ca2+ transfer in either cell type (fig. S2C). However, application of the protonophore carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP) revealed that the mitochondrial free Ca2+ content was 25% higher in wild-type MEFs than in calnexin knockout MEFs (Fig. 3A). This free mitochondrial Ca2+ content decreased to a greater extent upon removal of extracellular Ca2+ in calnexin wild-type MEFs compared with knockout MEFs (Fig. 3B), suggesting that calnexin is important to retain Ca2+ not just within the ER (Fig. 1D) but also within mitochondria of wild-type MEFs.
(A) Wild-type and calnexin knockout MEFs were transfected with Mito-GECO and imaged by fluorescence microscopy to measure total mitochondrial Ca2+. As indicated, 10 μM FCCP was added to disable mitochondrial Ca2+ transport (n = 4 biological replicates for each group, *P = 0.05 by one-sample t test). (B) Wild-type and calnexin knockout MEFs were transfected with Mito-GECO. As indicated, Ca2+ was removed from the growth medium and cells were imaged by fluorescence microscopy to measure total mitochondrial Ca2+ (n = 3 biological replicates for each group, P = 0.07 by one-sample t test). (C) Wild-type and calnexin knockout MEFs were loaded for 30 min with 40 nM TMRM. Mitochondrial potential was measured and, as indicated, 10 μM FCCP was added to uncouple mitochondria. The average FCCP-induced decrease in the TMRM fluorescent signal was quantified (n = 3 biological replicates for each group, *P = 0.17 by one-sample t test). (D) Wild-type and calnexin knockout MEFs were loaded with TMRM as in (C). Mitochondrial potential was measured and, as indicated, 10 μM thapsigargin was added to release Ca2+ from the ER. The average thapsigargin-induced decrease in the TMRM fluorescent signal was quantified (n = 3 biological replicates for each group, *P = 0.04 by one-sample t test).
To further investigate a potential role of calnexin for mitochondria, we next assayed the Ca2+ responsiveness of the mitochondrial membrane potential (ΔΨ) to Ca2+. Although ΔΨ did not substantially differ between calnexin wild-type and knockout MEFs (Fig. 3C), its responsiveness to artificial, thapsigargin-triggered ER-mitochondria Ca2+ flux was increased threefold in calnexin knockout MEFs (Fig. 3D). Together, these results indicated that mitochondria in calnexin knockout MEFs were depleted of Ca2+. Potentially as an adaptation, they were more responsive to ER-derived Ca2+ if pharmacologically provided with it.
Calnexin regulates mitochondria metabolism
Our results indicated that calnexin could activate mitochondria (because of the increase in mitochondria Ca2+ content; Fig. 3, A and B) or inhibit them (because of the increased ER-mitochondria distance; Fig. 2, E and H) or the two effects might cancel each other out (as seen with the IP3R-mediated Ca2+ transfer to mitochondria; fig. S1C). We sought to distinguish between these possibilities. First, we measured total cellular adenosine 5′-triphosphate (ATP) levels, which showed a small reduction in calnexin knockout MEFs (Fig. 4A). However, this value is derived from a combination of mitochondrial and glycolytic energy production. We therefore tested whether there was a connection between calnexin and OXPHOS or glycolysis. We found that wild-type MEFs, but not calnexin knockout MEFs, showed a robust reduction in ATP upon oligomycin treatment (Fig. 4B). In contrast, 2-deoxyglucose affected both wild-type and calnexin knockout MEFs to a similar extent (Fig. 4B). Because oligomycin blocks OXPHOS, but 2-deoxyglucose interferes with both OXPHOS and glycolysis (41), our results suggested that calnexin deficiency still allowed for respiration but led to a greater relative reliance on glycolysis. Overall, these data suggest that calnexin maintains normal OXPHOS.
(A) Total cellular ATP content was measured in wild-type and calnexin knockout MEFs (n = 35 biological replicates for each group, *P = 0.04 by two-sample Wilcoxon test). (B) After incubation with 1 μM oligomycin for 4 hours (in high glucose DMEM) or with 25 mM 2-deoxyglucose (2-DG, in DMEM lacking glucose) for 4 hours, total ATP content was measured in wild-type and calnexin knockout MEFs (oligomycin n = 6 sets of cells for each group, *P = 0.02 by two-sample t test; 2-DG n = 4 sets of cells for each group, P = 0.34 by two-sample Wilcoxon test). (C) Calnexin wild-type and knockout MEFs were immunoblotted for the indicated glycolytic and mitochondrial proteins. Hexokinase 1 (n = 5 sets of cells for each group, *P = 0.04 by one-sample t test), mitochondrial complex 2 (n = 3 sets of cells for each group, P = 0.5 by two-sample Wilcoxon test), MCU (n = 8 sets of cells for each group, P = 0.4 by two-sample Wilcoxon test), and mitofusin-2 (n = 3 sets of cells for each group, P = 0.7 by one-sample t test). n ≥ 2 sets of cells for all other proteins. (D) Wild-type and calnexin knockout MEFs were loaded with 1 μM Mitotracker Green for 30 min and assayed for the total mitochondrial area per cell (P = 0.9 by two-sample t test with equal variance), for the number of mitochondria (P = 0.7 by two-sample t test with equal variance), and for the mean area of individual mitochondrial particles (P = 0.5 by two-sample t test with equal variance). n = 5 sets of cells for wild type and n = 6 sets of cells for calnexin knockout MEFs. (E) Measurement of the mitochondrial oxygen consumption rates in wild-type and calnexin knockout MEFs. Basal respiration, leak, ATP turnover, and ETC capacity (pmol O2/s/CS activity) were determined and plotted (n = 10 sets of cells for each group, P = 0.0007 by two-sample Wilcoxon test for Basal, P = 0.06 by two-sample Wilcoxon test for Leak, ***P = 0.009 by two-sample Wilcoxon test for ATP turnover, and ***P = 0.00001 by two-sample Wilcoxon test for ETC capacity). (F) Quantification of citrate synthase activity in wild-type and calnexin knockout MEFs (n = 10 sets of cells for each group, ***P = 0.00001 by two-sample Wilcoxon test). (G) Mitochondrial ATP within wild-type and calnexin knockout MEF cells were assessed using mtAT1.03 signals. Quantification shown on the right (basal mitochondrial ATP: n = 10 sets of cells for each group, P = 0.3 by two-sample t test with equal variance. Δ ratio: n = 11 set of cells, P = 0.05 by two-sample Wilcoxon test). (H) Analysis of protein lysates for AMPK activity, as assessed by the phospho-AMPK signal. Tubulin was used as a loading control (n = 4 sets of cells for each group, **P = 0.006 by one-sample t test). (I) Quantification of apoptosis as assessed by Annexin V and PI in calnexin wild-type and knockout MEFs incubated for 24 hours with oligomycin (1 μM) or rotenone (1 μM) (Control n = 3 sets of cells for each group, P = 0.07 by two-sample Wilcoxon test; oligomycin n = 3 sets of cells for each group, **P = 0.005 by two-sample Wilcoxon test; rotenone n = 3 sets of cells for each group, P = 0.07 by two-sample Wilcoxon test).
We also measured the abundance of select proteins that control glycolysis. There was a 50% increase in the amount of hexokinase 1 in calnexin knockout MEFs, consistent with increased levels of glycolysis in calnexin knockout cells (Fig. 4C). In contrast, we did not detect differing amounts of mitochondrial proteins between genotypes (Fig. 4C). Likewise, we did not detect differences in mitochondria mass, number, or size (Fig. 4D). In contrast, intact calnexin knockout MEFs showed reduced basal respiration resulting in a 45% reduction in electron transport chain capacity compared to wild-type cells upon normalization to citrate synthase activity (Fig. 4E). This phenotype was reproduced in HeLa cells transfected with calnexin RNA interference (RNAi) compared to cells transfected with scramble RNAi (fig. S2D). Mitochondrial enzymatic activity measurements showed a 38% increase in citrate synthase activity in calnexin knockout MEFs under experimental conditions that supplied mitochondria with external Ca2+ (Fig. 4F). This phenotype prompted us to measure intramitochondrial ATP levels. Although mitochondria in calnexin knockout MEFs did not have different baseline ATP levels, these mitochondria had a threefold increased ability to rescue their ATP content upon an OXPHOS block (Fig. 4G). We also measured glucose consumption and lactate secretion of calnexin wild-type and knockout MEFs, which showed that calnexin deficiency did not significantly alter the glycolytic flux (fig. S2E). This metabolic signature indicated that calnexin knockout MEFs can compensate for the decrease in OXPHOS-derived energy generation with increased glycolysis and improved Krebs cycle capacity.
To analyze the importance of these findings, we measured the activity of the key metabolic sensor adenosine 5′-monophosphate–activated protein kinase (AMPK). Consistent with their altered energy balance, calnexin knockout MEFs showed a 50% increased activity of AMPK (Fig. 4H). Last, we hypothesized that blocking OXPHOS would be less toxic to calnexin knockout MEFs because they rely more on glycolysis than OXPHOS. Treatment with oligomycin and rotenone induced apoptosis to a lesser extent in calnexin knockout MEFs, confirming a reduced reliance on mitochondrial OXPHOS in these cells (Fig. 4I). Together, our results demonstrated that the role of the ER chaperone calnexin in mitochondrial bioenergetics is to maintain mitochondrial OXPHOS. In the absence of calnexin, cells compensate by increasing Krebs cycle capacity and glycolysis while becoming less vulnerable to OXPHOS disruption.
Cells adapt to the absence of calnexin by boosting Ca2+ flux across MAMs
We next examined whether calnexin knockout also altered MAM composition and function. First, we measured the expression of select MAM-regulatory proteins, focusing specifically on IP3R1 and VDAC1, which control Ca2+ flux from the ER to mitochondria. Calnexin knockout MEFs expressed about 50% more IP3R1 and VDAC1 (Fig. 5A), reflecting their maintained ability to accept ER Ca2+ signals from the ER, despite depleted ER Ca2+ stores (fig. S2C), and consistent with their increased ER-mitochondria apposition. In contrast, the abundance of the ER proteins protein kinase RNA-like ER kinase (PERK), SERCA2b, ERp57, and Nox4 and the autophagy adaptor p62 did not differ between genotypes (Fig. 5A). However, an acute knockdown of calnexin led to small decreases in the amounts of SERCA2b and Nox4 (fig. S3A). Similarly, MAMs of calnexin knockout MEFs had a relatively small increase of the mitofusin-2 tether but otherwise did not show substantial changes (fig. S3B), suggesting that MAM composition was not affected in a major way by the absence of calnexin.
(A) Calnexin wild-type and knockout MEFs were immunoblotted for VDAC1 (n = 6 sets of cells for each group, *P = 0.04 by one-sample t test), IP3R1 (n = 8 sets of cells for each group, *P = 0.02 by one-sample t test), IP3R3 (n = 6 sets of cells for each group, P = 0.65 by one-sample t test), and TMX1 (n = 4 sets of cells for each group, **P = 0.004 by one-sample t test). Tubulin served as a loading control. n ≥ 2 sets of cells for all other proteins. (B) Measurement of the basal mitochondrial oxygen consumption rates in wild-type and calnexin knockout MEFs upon incubation with 10 μM BAPTA-AM for 1 hour (Control n = 5 sets of cells for each group, *P = 0.02 by two-sample t test with equal variance; BAPTA n = 5 sets of cells for each group, P = 0.07 by two-sample t test with equal variance).
To further investigate the basis of the altered metabolism of calnexin knockout MEFs, we examined whether these cells relied on altered Ca2+ handling. Chelation of intracellular Ca2+ by 1,2-Bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid tetrakis(acetoxymethyl ester) (BAPTA)–AM reduced basal oxygen consumption rate (OCR) in wild-type cells by more than 50%, but only by about 35% in knockout MEFs (Fig. 5B), thus eliminating the significant difference between the OCRs of the two cell lines (Fig. 4E). This result suggests that the key metabolic differences in calnexin knockout MEFs stem from altered Ca2+ flux at the MAM originating from a reduction of both ER Ca2+ stores and SERCA activity.
Oxidizing conditions are necessary for the metabolic roles of calnexin
Changes in SERCA activity could be due to different protein amounts or different activity. Our findings so far indicated that calnexin knockout or knockdown decreased redox-dependent posttranslational SERCA2b modifications that determine its activity to a greater extent (Fig. 1A and fig. S1A) than its total abundance (Fig. 5A and fig. S3A). We therefore sought to identify the source of SERCA2b-modifying reactive oxygen species (ROS). Calnexin knockout MEFs had lower cytosolic ROS as previously published (42) (Fig. 6A), but similar amounts of mitochondrial ROS to wild-type MEFs (Fig. 6A). Therefore, we surmised that ROS from outside the mitochondria trigger activating SERCA2b oxidation.
(A) Measurement of the cellular ROS production with flow cytometry to detect MitoSox (left) and CellROX (right) in wild-type and calnexin knockout MEFs (n = 7 sets of cells for each group, P = 0.2 by one-sample t test for MitoSOX, n = 3 sets of cells for each group, **P = 0.008 by one-sample t test for CellROX). (B) Analysis of SERCA2b for the exposure of reduced cysteines, as detected with BIAM, in the presence of 5 mM NAC for 4 hours. Cells were analyzed by Western blot for Myc and BIAM signals. DTT-treated signals represent fully reduced SERCA2b (n = 2 independent experiments). (C) ER Ca2+ measurements upon tert-BuBHQ-mediated Ca2+ release from the ER in wild-type and calnexin knockout MEFs in the presence of 5 mM NAC for 4 hours (n = 12 biological replicates for each group, ***P = 0.000005 by one-sample t test for Control; n = 6 biological replicates for each group, P = 0.4 by two-sample t test with equal variance for NAC). (D) ER Ca2+ depletion slope upon tert-BuBHQ-mediated Ca2+ release from the ER in wild-type and calnexin knockout MEFs in the presence of 5 mM NAC for 4 hours (n = 13 biological replicates for each group, *P = 0.01 by one-sample t test for Control; n = 6 biological replicates for each group, P = 0.6 by two-sample t test with equal variance for NAC). (E) Measurement of the basal mitochondrial oxygen consumption rates in wild-type and calnexin knockout MEFs upon incubation with 5 mM NAC for 4 hours (n = 10 biological replicates for each group, **P = 0.005 by two-sample Wilcoxon test for Control, P = 0.1 by two-sample t test with equal variance for NAC). (F) Quantification of citrate synthase activity in wild-type and calnexin knockout MEFs (n = 10 biological replicates for each group, ***P = 0.000001 by two-sample Wilcoxon test for Control; n = 5, **P = 0.001 by two-sample t test with equal variance for NAC).
Thus, we sought to pharmacologically equalize the cytosolic redox status between the two cell types. Treatment with N-acetylcysteine (NAC), which does not affect the unfolded protein response (43, 44), resulted in fully reduced SERCA2b in cells of both genotypes (Fig. 6B), as expected. Analysis of ER Ca2+ loading through SERCA showed that NAC reduced ER Ca2+ content and Ca2+ release in calnexin wild-type MEFs, but not knockout MEFs (Fig. 6, C and D), thus indicating that NAC eliminated the Ca2+ storage difference between the two genotypes. Consistent with this finding and with the role of ER Ca2+ filling for mitochondrial OXPHOS, calnexin wild-type MEFs, but not knockout MEFs, exhibited markedly reduced respiration upon incubation with NAC (Fig. 6E), similar to treatment with BAPTA-AM (Fig. 5B). During the relatively short time frame of 4 hours, NAC did not correct the differences in the activity of the Krebs cycle enzyme citrate synthase (Fig. 6F). Similarly, NAC reduced basal OCR more in control HeLa cells, than in siCNX-transfected cells, and did not correct citrate synthase activity (fig. S4, A and B). Last, NAC also reduced the difference in the SPLICSlong signal without affecting the SPLICSshort signal in calnexin wild-type and knockout MEFs (Fig. 2, G and H).
Next, we sought to investigate the source of oxidative stress in wild-type MEFs by using inhibitors for two ER ROS producers, Ero1 (EN460) and Nox1/4 (GKT137831) (45, 46). Both EN460 and GKT137831 decreased the SERCA2b oxidation state (fig. S5A) and partially reduced the ATPase activity of SERCA2b in wild-type MEFs, thus decreasing the difference in ATPase activity (fig. S5B) and ER Ca2+ content (fig. S5C) between wild-type and knockout MEFs. These results suggested that a combined ER-derived ROS source was responsible for the correct SERCA2b redox and activity and that the role of calnexin for mitochondrial Ca2+ and metabolism is not connected to its role as a chaperone. Rather, our results suggest that calnexin maintains ER Ca2+ levels and mitochondria metabolism by maintaining properly oxidized SERCA.
DISCUSSION
An important function of MAMs is to establish a dedicated gateway for Ca2+ between the ER and mitochondria (47). ER-derived Ca2+ controls the activity of mitochondrial dehydrogenases (48) and hence the progression of the Krebs cycle (49). The flux of Ca2+ ions between the ER and mitochondria controls mitochondrial metabolism, as shown by its disruption in cells lacking all versions of IP3R, whose respiration is about one-third of their wild-type counterparts (50). Accordingly, these cells also activate signaling through AMPK (50). Understanding the proteins that control ER-mitochondria Ca2+ transfer is therefore expected to yield insight into the control of cellular metabolism. Proteomic studies have identified ER chaperones and folding assistants as prominent MAM proteins (16, 17). However, few mechanistic connections between ER chaperones, including calnexin, and mitochondrial respiration have been made.
Calnexin localizes in close proximity to the translocon, where it regulates protein folding (51), but it is also found on MAMs and is part of the SERCA2b interactome (18, 26). Because the inhibition of SERCA decreases the IP3R-mediated transfer of Ca2+ to mitochondria (52), we hypothesized that by controlling SERCA activity, calnexin controls the flux of Ca2+ between the ER and mitochondria and thus should have metabolic functions (Fig. 7). Calnexin knockout MEFs show about 30% less respiration, based on a reduction of SERCA cysteine oxidation by 54% (Fig. 1A) that led to a reduction of the SERCA activity by 32%, characteristics that were reproduced in HeLa cells transfected with calnexin RNAi and attenuated upon calnexin reconstitution. As an adaptation, calnexin knockout MEFs showed increased AMPK activation and increased MAMs, which coincided with an increased ability to use glucose within mitochondria upon OXPHOS inhibition (Fig. 4, G to I). Although calnexin knockout MEFs showed higher levels of the SERCA inhibitor TMX1 (Fig. 5A), this was not the case with HeLa cells transfected with calnexin RNAi (fig. S3A), despite a similar Ca2+ phenotype, thus ruling out the idea that the effects were due to altered TMX1 levels.
The calnexin-mediated activation of SERCA through the maintenance of its redox state by Nox4 and other ER ROS sources allows for proper Ca2+ filling of the ER. A properly filled ER is necessary to provide mitochondria with sufficient Ca2+ necessary for the Krebs cycle and OXPHOS. In the absence of calnexin, cells compensate for the reduced Ca2+ flux with closer ER-mitochondria apposition. This compensation is associated with the binding of the reductase TMX1 to SERCA as previously published (19).
Overall, therefore, although TMX1 inactivates SERCA and promotes OXPHOS (19), in the case of calnexin, SERCA activation is functionally linked to mitochondrial OXPHOS. In contrast, uncoupling protein 3 inhibits both SERCA and OXPHOS (53). These discrepant findings suggest that mere control of ER Ca2+ handling is not sufficient to assign the role that a MAM protein could play in mitochondria metabolism. For calnexin, the phenotype we observed was a combination of overall reduced OXPHOS and an increase in Krebs cycle activity, the latter potentially an adaptation (see also below). A reason for the complexity of the phenotype in calnexin knockout MEFs could be that the amounts of Ca2+ available within mitochondria depend on at least three factors: ER Ca2+ content, the distance and tethering between the ER and mitochondria, and the Ca2+ uptake and release capabilities of the ER and mitochondria. Cross-talk between these factors could complicate the metabolic readout and compensatory mechanisms may nullify effects. For instance, SOCE could interfere with ER-mitochondria Ca2+ cross-talk. Another possibility is a functional overlap between Ca2+ handling and ER-mitochondria tethering. These two mechanisms can enter a positive feedback loop, because, on the one hand, IP3Rs not only release ER Ca2+ but also are part of a tethering complex (9) and, on the other hand, increased release of ER Ca2+ into the cytoplasm inhibits mitochondria movement (37, 38). It appears that the main role of calnexin is to maintain ER Ca2+ content, enabling the ER to maintain IP3R-mediated Ca2+ cross-talk that is critical for OXPHOS (50). At the same time, its activity to promote cytoplasmic Ca2+ clearance reduces ER-mitochondria apposition. This role could also underlie the increased ΔΨ responsiveness to Ca2+ in calnexin knockout MEFs. Moreover, we observed increased Ca2+ leak in the presence of calnexin. This could further alter Ca2+ cross-talk and could also mitigate the effects we describe in our in silico model of ER-mitochondria apposition. The observed differences in ER-mitochondria apposition were substantive between calnexin wild-type and knockout MEFs.
In addition to Ca2+ availability from the ER side of the MAM, mitochondrial Ca2+ uptake and release can also determine the extent of mitochondrial ATP production through interference with Ca2+ handling at the mitochondrial membranes. For instance, cells lacking the mitochondrial Ca2+ uniporter (MCU) cannot ramp up energy production during stress (54). In contrast, a deletion of the mitochondrial Na+/Ca2+ exchanger results in increased production of superoxide and ultimately apoptosis (55).
Based on our findings, the manipulation of Ca2+ handling at the level of the ER is an important avenue to control the balance between OXPHOS and glycolysis. Calnexin executes such a function primarily through control of SERCA redox but might also control SERCA amounts through a chaperoning function. The relative importance and potential connection of these two functions could depend on the cell system or baseline redox conditions used, which could explain earlier findings (26). Moreover, the calnexin-mediated control of Nox4 and Ero1α/β activities could be especially critical in migrating endothelial cells, where Nox4 is upstream of SERCA oxidation upon incubation with vascular endothelial growth factor (56). Consistent with both findings, the antioxidant NAC fully reduced SERCA and eliminated redox control of OXPHOS in a calnexin-dependent manner (Fig. 6E). Other SERCA redox regulatory proteins exist, such as p53, which interacts with SERCA on ER-mitochondria contact sites and activates it by controlling its oxidation. In contrast to the oxidation-preserving role of calnexin, p53 prevents hyperoxidation of SERCA. Nevertheless, the consequences of this redox function are similar, because p53 thus also acts to preserve ER-mitochondria Ca2+ flux and, as a consequence, mitochondrial apoptosis induction (23) and OXPHOS (57).
Thus, our results presented in this study suggest that calnexin could play a role for tumor cell metabolism, as suggested by our experiments in HeLa cells. Because high amounts of calnexin have been found in lung cancer and have been implicated as a poor prognostic marker in colorectal cancer (58, 59), we propose that calnexin would increase cancer cell viability by maintaining OXPHOS in these scenarios, as has been shown for IP3Rs (60), a possibility that warrants future research. Calnexin has also been implicated in neurodegeneration, where its deficiency leads to neuronal dysmyelination but blocks neuroinflammation (61). Because inflammation could depend on the formation of functional MAMs (62), the described imbalance of this structure and the absence of mitochondrial ROS in calnexin knockout cells could provide an explanation.
MATERIALS AND METHODS
Antibodies and reagents
Basic chemicals were from Sigma-Aldrich (Oakville, ON). The following chemicals were purchased as indicated: reduced form of nicotinamide adenine dinucleotide (NADH), ATP, tert-BuBHQ, thapsigargin, Fura-2, Fluo-8, FCCP, tetramethylrhodamine methyl ester (TMRM), 2,7-dichlorofluoroscein diacetate (DCF), BAPTA-AM, MitoSOX, CellROX, Oligofectamine, MitoTracker Green and Red (Thermo Fisher Scientific, Waltham, MA), oligomycin (Sigma-Aldrich, Oakville, ON), EN460 (Sigma-Aldrich, Oakville, ON), GKT137648 (Cayman, Ann Arbor, MI), and Percoll (GE Healthcare, Piscataway, NJ). Antibodies were purchased as indicated: against hexokinase 1 (Cell Signaling, Danvers, MA), hexokinase 2 (Cell Signaling, Danvers, MA), glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Cell Signaling, Danvers, MA), protein disulfide isomerase (PDI) (Pierce/Fisher, Rockford, IL), pyruvate dehydrogenase (PDH) (Cell Signaling, Danvers, MA), pyruvate kinase isozyme M2 (PKM2) (Cell Signaling, Danvers, MA), glucose transporter 1 (GLUT1) (GeneTex, Irvine, CA), GLUT4 (Abcam, Cambridge, MA), mitochondrial complex 2 (Cell Signaling, Danvers, MA), MCU (Sigma-Aldrich, Oakville, ON), mitofusin-2 (Sigma-Aldrich, Oakville, ON), VDAC1 (Abcam, Cambridge, MA), IP3R1 (Thermo Fisher Scientific, Waltham, MA), PERK (Cell Signaling, Danvers, MA), p-PERK (Cell Signaling, Danvers, MA), ERp57 (StressMarq, Victoria, BC, Canada), NOX4 (Abcam, Cambridge, MA), AMPK (EMD Millipore, Billerica, MA), p-AMPK (EMD Millipore, Billerica, MA), SERCA2b (EMD Millipore, Billerica, MA), p62 (BD Biosciences, Franklin Lakes, NJ), actin (Thermo Fisher Scientific, Waltham, MA), tubulin (Sigma-Aldrich, Oakville, ON), IP3R2 (Alomone, Jerusalem, Israel), IP3R3 (BD Biosciences, Franklin Lakes, NJ), and FACL4 (Abcam, Cambridge, MA). The affinity-purified rabbit TMX1 antiserum was generated by 21st Century Biochemicals (Marlborough, MA) using a peptide corresponding to amino acids 251 to 271 (AESKEGTNKDFPQNAIRQRSL). The rabbit calnexin antibody has been previously described (63). HeLa cells were from ECACC (Porton Down, UK).
Mathematical modeling
A mathematical model that considers the Ca2+ cross-talk within the ER-mitochondria microdomain was used to develop a modified model to account for the effect of calnexin on cytosolic Ca2+ oscillations and ER-mitochondria distance (33). First, the positive regulatory role of calnexin on the SERCA pump activity was described by an activating Hill function
Here, r is the actual distance between ER and IMM, r0 is the minimal ER-IMM distance, and rc is the maximal modulatory distance (set at 35 nm), which we multiply with a factor that accounts for changes of [Ca2+]Cyt during waves on the ER-IMM distance. KDOC is the half-saturation constant with KDOC = 0.35 μM (36).
Light membrane fractionation
Light membranes were separated as described previously (63). MEF cells were grown to 80% confluency on 10-cm dishes and treated as described in each experiment. Then, cells were washed twice with phosphate-buffered saline (PBS), harvested in 600 μl of cold homogenization buffer [250 mM sucrose, 10 mM Hepes-OH (pH 7.4), 1 mM EDTA, and 1 mM EGTA] + protease inhibitors (cOmplete, Roche), and homogenized by passing them seven times through a 26½ G needle. The resulting cell lysates were centrifuged at 800g for 10 min to remove nuclei and unbroken cells. The heavy membrane fraction was separated by centrifuging the postnuclear lysates at 10,000g for 10 min. The light membrane fraction was obtained by centrifuging the remaining supernatant at 60,000g for 1 hour and resuspending the pellet in 300 μl of 50 mM Hepes (pH 7.4), 125 mM NaCl, and 1 mM MgCl2.
MAM fractionation
MEFs were grown in fifteen 20-cm dishes, scraped with 5 ml of homogenization buffer [0.25 M sucrose, 10 mM Hepes-NaOH (pH 7.4), 1 mM EDTA, and 1 mM EGTA, containing cOmplete protease inhibitor; Roche, Mississauga, ON], centrifuged at 300g for 5 min, and resuspended in 5 ml of homogenization buffer. Then, cells were homogenized with seven passages through a ball-bearing homogenizer (Isobiotec, Heidelberg, Germany, ball clearance 18 μM). The resulting homogenate was centrifuged for 10 min at 600g to remove unbroken cells and nuclei. Postnuclear supernatant was centrifuged for 10 min at 8500g to yield the crude mitochondria fraction. The supernatant was then centrifuged for 60 min at 60,000g to separate the cytosolic fraction and microsome fraction. The previously isolated crude mitochondria fraction was resuspended in 1 ml of homogenization buffer and layered on top of 7.9 ml of 18% v/v Percoll in homogenization buffer and centrifuged for 30 min at 30,000g. Purified MAM and pure mitochondria fraction bands were extracted from the gradient using a syringe. The band containing the MAM fraction was centrifuged for 1 hour at 95,000g to yield the MAM fraction. The band containing the pure mitochondria fraction was centrifuged for 10 min at 10,000g to yield the pure mitochondria fraction. The cytosolic fraction was precipitated overnight with acetone and centrifuged for 20 min at 16,000g. All fractions were resuspended in loading buffer [60 mM tris-HCl (pH 6.8), 2% SDS, 10% glycerol, 10% β-mercaptoethanol, and 0.004% bromophenol blue] and equal proportional amounts were analyzed by Western blotting.
ATPase assay
The ATPase assay was carried out using an enzyme coupled reaction in which the regeneration of ATP is coupled to the oxidation of NADH, as previously described (64). Activity was detected as a decrease in NADH absorbance at 340 nm, which was measured every 1 min for 120 min using a microplate reader (Synergy, BioTek) and the path-length correction function. All reactions were carried out in triplicate in a final volume of 100 μl using a 96-well plate. The decrease in NADH absorbance at 340 nm was converted to micromoles of ATP using Beer’s law and then expressed as a function of time. The final conditions for all the reactions were 25 mM Hepes (pH 7.2), 125 mM NaCl, 5 mM MgCl2, 0.1 mM CaCl2, 1 mM DTT, 0.6 mM NADH, 2 mM ATP, 1 mM phosphoenol pyruvate, 2.5 μl of pyruvate kinase/lactate dehydrogenase (Sigma-Aldrich), and 0.02% dimethyl sulfoxide. Identical reactions were treated with 10 μM thapsigargin and subtracted from untreated reactions to identify specific SERCA activity. Protein concentration of light membrane samples was quantified using the Pierce BCA Protein Assay Kit and 1 μg of protein was added to each well. Reactions were started by adding the regenerating system consisting of DTT, NADH, ATP, phosphoenol pyruvate, and pyruvate kinase/lactate dehydrogenase to the resuspended light membranes. Fit lines were calculated with the following equation: Y = ((Bmax * X)/(Kapp + X)) + X0. SERCA ATPase activity is shown as the percentage of the ATPase activity inhibited by thapsigargin in the light membranes and as ATPase activity per microgram of protein in the light membranes.
Cell lysates and Western blotting
MEFs were grown to 80% confluency on 60-mm dishes, washed twice with PBS, and harvested in 300 μl of cold CHAPS buffer [10 mM tris (pH 7.4), NaCl 150 mM, 1 mM EDTA, and 1% CHAPS] + protease inhibitors (cOmplete, Roche) + phosphatase inhibitors (PhosSTOP, Roche). After centrifuging at 800g for 10 min to remove nuclei and unbroken cells, the lysate was mixed with loading buffer [60 mM tris-HCl (pH 6.8), 2% SDS, 10% glycerol, 10% β-mercaptoethanol, and 0.004% bromophenol blue]. Equal amounts of protein per sample were loaded and separated by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) (10% acrylamide gels) and transferred to nitrocellulose membranes. Membranes were blocked with blocking buffer [10 mM tris (pH 8), 150 mM NaCl, 0.05% Triton X-100, and 2% bovine serum albumin (BSA)] and incubated with primary antibodies overnight. The next day, membranes were incubated with secondary antibodies and proteins were detected using a LI-COR imaging system (Biosciences). Band intensity was quantified using the Fiji (65) distribution of ImageJ 1.52b (66).
Microscopy and image acquisition
Microscopy experiments were performed on an FV1000 laser-scanning confocal microscope (Olympus) using a 60× objective [XLUMPLANFL, numerical aperture (NA) 1.0; Olympus] or a 20× objective (XLUMPLANFL, NA 1.0; Olympus), equipped with a perfusion system consisting of a peristaltic pump (for the FV1000 system, Watson-Marlow Alitea-AB; Sin-Can). Live-cell images were taken with a 559-nm laser excitation with a 575- to 675-nm bandpass emission filter for MitoTracker Red, ER R-GECO (LAR-ER-GECO), mitochondrial R-GECO, and TMRM fluorescent probes, and a 473-nm laser excitation with a 490- to 540-nm bandpass emission filter for SPLICS-short, SPLICS-long, FLUO8, and MitoTracker Green using a PL-A686 6.6 megapixel camera (Capture SE software, Pixelink). The temperature in the recording chamber was kept at 25° to 27°C (TC-324B, Harvard Apparatus). Cells were incubated in tissue culture medium supplemented with drugs as indicated. Fluorochromes were used as indicated. Images were acquired with Olympus FluoView software. Image analysis and mitochondrial particle quantification were done with the Fiji (65) distribution of ImageJ 1.52b (66) using the Time Series Analyzer v3.0 plugin for measurements over time.
ER and mitochondrial Ca2+ measurements
To measure ER and mitochondrial Ca2+, we used the low-affinity ER R-GECO (LAR-ER-GECO) and the mitochondrial R-GECO fluorescent Ca2+ indicators (40). To express these Ca2+ indicators, MEFs were transfected with plasmids containing the corresponding fluorescent probe using Nucleofector 2b Device (Lonza, AAB-1001) and the MEF 1 Nucleofector Kit (Amaxa, catalog number VPD-1004). Following the manufacturer’s instructions, 2 × 106 MEFs were transfected with 5 μg of plasmid containing the corresponding probe using the A-023 program. Transfected cells were seeded in poly-l-lysine–coated coverslips (12 mm) in 24-well plates (300,000 cells per well). The next day, coverslips were transferred to the microscope and perfused with Hanks’ balanced salt solution (HBSS) with Ca2+ and Mg2+ (HBSS/Ca/Mg) at 5 ml/min. Live-cell images were taken every 3 s with a 559-nm laser excitation and a 575- to 675-nm bandpass emission filter. After establishment of a 30-s baseline, the perfusion medium was changed to HBSS/Ca/Mg + ATP 100 μM for 5 min or HBSS without Ca2+ + TBHQ 60 μM for 10 min, and the experiment was continued for 5 to 10 min using HBSS/Ca/Mg or HBSS without Ca2+, respectively. Independent experiments consisted of 10 to 20 cells.
SOCE measurements
Cells were seeded in poly-l-lysine–coated coverslips (12 mm) in 24-well plates (50,000 cells per well). Then, the next day, cells were incubated with FLUO8 1 μM for 30 min. The coverslips were transferred to the microscope and perfused with HBSS with Ca2+ and Mg2+ (Thermo Fisher Scientific, catalog number 14025) (HBSS/Ca/Mg) at 5 ml/min. Live-cell images were taken every 5 s with a 473-nm laser excitation with a 490- to 540-nm bandpass emission filter. After establishment of a 30-s baseline, the perfusion medium was changed to HBSS without Ca2+ but containing 10 μM thapsigargin. After 10 min, the perfusion medium was changed to HBSS/Ca/Mg for 10 min. Independent experiments consisted of 20 to 30 cells.
Mitochondrial membrane potential
Cells were seeded in poly-l-lysine–coated coverslips (12 mm) in 24-well plates (50,000 cells per well). The next day, cells were incubated with 40 nM TMRM for 30 min. Then, coverslips were transferred to the microscope and perfused with HBSS with Ca2+ and Mg2+ (Thermo Fisher Scientific, catalog number 14025) (HBSS/Ca/Mg) at 5 ml/min. Live-cell images were taken every 5 s with a 559-nm laser excitation and a 575- to 675-nm bandpass emission filter. After establishment of a 30-s baseline, the perfusion medium was changed to HBSS/Ca/Mg + 10 μM FCCP for 10 min. Independent experiments consisted of 20 to 30 cells.
Cytoplasmic measurements of Ca2+ release
To measure ER Ca2+ release into the cytosol, we used the fluorescent Ca2+ probe Fura-2. MEFs were seeded in 60-mm dishes at a concentration of 2 × 106 cells per dish. The next day, cells were washed with HBSS buffer with Ca2+ and Mg2+ (Thermo Fisher Scientific, catalog number 14025) and 0.1% BSA (HBSS/Ca/Mg/BSA) and then incubated with 1.5 ml of 1 μM Fura-2 in HBSS/Ca/Mg/BSA for 30 min at room temperature in the dark. Next, we replaced the loading medium with HBSS/Ca/Mg/BSA and incubated it for a further 15 min in the dark. After trypsinization and quenching with 1 ml of Dulbecco’s modified Eagle’s medium (DMEM)/10% fetal bovine serum (FBS), cells were washed twice with 1 ml of HBSS/Ca/Mg/BSA. Cells were then resuspended in HBSS/Ca/Mg/BSA and transferred into a cuvette containing a stirring bar, and 505-nm emissions were measured with 340/380 nm excitation on an 814 photomultiplier detection system (PTI), upon establishment of a 200-s baseline. Drugs (50 μM histamine or 10 μM thapsigargin) were added as 100 μl of a 200× stock solution prepared in HBSS/Ca/Mg/BSA. The Ca2+ response was determined as the relative ratiometric signal obtained from the baseline and after adding drugs. For the decay calculation, we fit a line into the data derived from the peak until 30 s later.
Electron microscopy
Cell monolayers were fixed for 20 min using 2% paraformaldehyde and 2% glutaraldehyde in 100 mM sodium cacodylate buffer at pH 7.4. Cells were scraped from the plates and pelleted. Secondary fixation was in osmium tetroxide 1% followed by quick rinses in water and staining in 1% aqueous solution of uranyl acetate. Dehydration was in increasing concentrations of ethanol and was followed by incubation in propylene oxide at room temperature. Pellets were infused with Embed 812 and blocks were hardened at 60°C for a minimum of 48 hours. We used an Ultracut E, Reichert-Jung for sectioning and imaged the samples using a charge-coupled device (CCD) camera (iTEM, Olympus Soft Imaging Solutions) mounted on a Philips 410 TEM. On the images, we identified ER tubules in close proximity to mitochondria and determined their distance in nanometers. We recorded apposed ER tubules per mitochondrion (distance <50 nm) and quantified their distance and length.
SPLICS
MEF cells were transfected with plasmids containing the SPLICS-long or SPLICS-short probes using Nucleofector 2b Device (Lonza, AAB-1001) and the MEF 1 Nucleofector Kit (Amaxa, catalog number VPD-1004). Following the manufacturer’s instructions, 2 × 106 MEFs were transfected with 5 μg of plasmid containing the corresponding probe using the A-023 program. Transfected cells were seeded in poly-l-lysine–coated coverslips (12 mm) in 24-well plates (300,000 cells per well). Forty-eight hours later, cells were treated with NAC 5 mM for 4 hours, stained with MitoTracker Red 1 μM for 30 min at 37°C, and fixed with 4% paraformaldehyde for 20 min at room temperature. Cells were imaged using a laser scanning confocal microscope with a 60× objective (XLUMPLANFL, NA 1.0; Olympus). Images were convolved, filtered using a Gaussian Blur filter, and thresholded (39). SPLICS particles were quantified using ImageJ.
Total ATP determination
Cellular ATP content was determined using the ATP Determination kit (Molecular Probes, A22066) based on a luciferase enzymatic reaction. MEFs were seeded in six-well plates at a concentration of 500,000 cells per well. The next day, cells were treated as described in each experiment. Then, cells were trypsinized and 600,000 viable cells were centrifuged and resuspended in CHAPS buffer [10 mM tris (pH 7.4), NaCl 150 mM, 1 mM EDTA, and 1% CHAPS] + protease inhibitors (cOmplete, Roche). Luminescence was measured using Lumat LB 9507 (Berthold Technologies) at 560 nm.
Oxygen consumption rates
The respiratory capacity of the mitochondrial electron transport chain was measured in intact cells using high-resolution respirometry (Oxygraph-2K; Oroboros), as previously described (67). MEFs were seeded in 60-mm dishes at a concentration of 2 × 106 cells per dish. The next day, cells were trypsinized, centrifuged at 100g for 5 min, and resuspended in culture media at a cell density of 106 cells/ml. Basic parameters of mitochondrial function were assessed at 37°C and expressed as pmol O2/s/citrate synthase activity in a single protocol designed to assess basal respiration, ATP turnover respiration (after addition of the ATP synthase inhibitor oligomycin), and maximal respiration (or electron transfer system capacity, after addition of the protonophoric uncoupler carbonyl cyanide-4-[trifluoromethoxy]phenylhydrazone). All respiratory rates were corrected for nonmitochondrial respiration (2.5 μM antimycin-inhibited rate) (68).
Citrate synthase activity
The activity of the mitochondrial enzyme citrate synthase was quantified by measuring the formation of thionitrobenzoate from 5,5-dithiobis-2-nitrobenzoate after the addition of 0.3 mM acetyl-coenzyme A and 0.5 mM oxaloacetate. The production on thionitrobenzoate was measured spectrophotometrically at 412 nm for 2 min.
Mitochondrial area
Cells were seeded in poly-l-lysine–coated coverslips (12 mm) in 24-well plates (50,000 cells per well). The next day, cells were incubated with 1 μM MitoTracker Green for 30 min at 37°C. Coverslips were perfused with HBSS with Ca2+ and Mg2+ (HBSS/Ca/Mg). Live cells were imaged using a laser scanning confocal microscope with a 60× objective (XLUMPLANFL, NA 1.0; Olympus). A background threshold was subtracted from images, and mitochondrial particles were quantified using ImageJ.
Metabolomics
Cells were seeded in 10-cm dishes (2 × 106 cells per dish). The next day, culture medium was replaced with 7 ml of fresh culture medium with the indicated inhibitors. After 24 hours, culture medium was filtered with 0.2-μm filters and mixed with the dextran sulfate sodium (DSS) reference solution. Nuclear magnetic resonance (NMR) spectroscopy was performed at the NANUC National High Field NMR Facility using a 2.2-K pumped 800-MHz Oxford magnet updated to a Bruker Neo-Advance IV console equipped with a 5-mm cryoprobe. One-dimensional NMR spectra were collected at 25°C using the standard noesypr1d pulse sequence with a transmitter presaturation delay of 990 ms for water suppression, a 100-ms mixing time, 8 steady-state scans, and a spectral width of 12 ppm. The total acquisition time was 4 s per transient, with 128 transients acquired per sample. The raw data were processed using the Chenomx Processor module from the Chenomx NMR suite. The free induction decays were apodized with an exponential window function corresponding to a line broadening of 0.1 Hz, zero-filled to 128K complex points, Fourier-transformed, phased, and baseline-corrected for further analysis. Reference deconvolution using the methyl peak of DSS as the reference peak was used to correct for line shapes. Metabolites were quantified using the Chenomx NMR Suite software (version 8.4) application developed by Chenomx Inc. (Edmonton, AB, Canada).
Mitochondrial ATP measurements
Before fluorescence microscopic experiments, cells were equilibrated for at least 30 min in a buffer composed of 2 mM CaCl2, 135 mM NaCl, 1 mM MgCl2, 5 mM KCl, 10 mM Hepes, 2.6 mM NaHCO3, 0.44 mM KH2PO4, 0.34 mM Na2HPO4, 1× amino acids, 1× vitamins, 10 mM glucose, and 2 mM l-glutamine with a pH of 7.45. During microscopic experiments, cells were perfused using buffers composed of 2 mM CaCl2, 138 mM NaCl, 1 mM MgCl, 5 mM KCl, and 10 mM Hepes with or without 10 mM glucose, with or without 3 μM Oligomycin-A. Buffers were exchanged using a gravity-based perfusion system (NGFI GmbH, Graz, Austria).
Mitochondrial ATP dynamics of MEFs were assessed using mtAT1.03. MEFs were analyzed at an OLYMPUS IX73 inverted microscope (OLYMPUS, Vienna, Austria) using a 40× objective (UApo N 340, 40×/1.35 Oil, ∞/0.17/FN22, OLYMPUS). Illumination was performed using an OMICRON LedHUB High-Power light-emitting diode (LED) light engine, equipped with a 455-nm LED light source (OMICRON electronics, Vienna, Austria), and a 430-nm excitation filter (AHF Analysentechnik, Tübingen, Germany). Images were captured at a binning of 4 using a Retiga R1 CCD camera (Teledyne QImaging, Surrey, Canada), and emissions were separated using an optical beam splitter (DV2, Photometrics, Arizona, USA). Device control and image acquisition was performed using VisiView image acquisition and control software (Visitron Systems, Puchheim, Germany).
ROS measurements
Mitochondrial ROS levels were measured using MitoSOX (Molecular Probes, M36008), and cellular ROS levels were measured using CellROX (Molecular Probes, C10443). MEFs were seeded in six-well plates at a concentration of 500,000 cells per well. The next day, MEFs were treated with 5 μM MitoSOX for 10 min at 37°C or 5 μM CellROX for 30 min. After trypsinization and quenching with 1 ml of DMEM/10% FBS, cells were centrifuged at 100g for 5 min and resuspended in HBSS buffer with Ca2+ and Mg2+ (Thermo Fisher Scientific, catalog number 14025) and 0.1% BSA (HBSS/Ca/Mg/BSA). MitoSOX fluorescence signal (excitation 510/emission 580 nm) and CellROX (excitation 545/emission 565 nm) were measured using flow cytometry (LSRFortessa, BD).
SERCA redox state
Cells were transfected with Myc-SERCA for 24 hours before incubation with 10 mM DTT for 10 min, 5 mM diamide for 5 min, or no treatment, respectively, at 37°C. Cells were rinsed with warm PBS twice, lysed with BIAM (Thermo Fisher Scientific, 21334) reaction buffer [50 mM tris-HCl (pH 8), 50 mM NaCl, 5 mM EDTA, 1% Triton X-100, 0.1% IGEPAL CA-630, 200 μM BIAM, and Roche EDTA-free protease inhibitors], and centrifuged at 10,000g for 5 min. Samples were incubated at 25°C for 90 min in the dark to allow biotinylation of reduced proteins. Unreacted BIAM was removed by Bio-Spin P-6 columns (Bio-Rad, 732-6221) with buffer exchanged to CHAPS buffer [1% CHAPS, 10 mM tris-HCl (pH 7.4), 150 mM NaCl, 1 mM EDTA, and Roche EDTA-free protease inhibitors]. Immunoprecipitation of Myc-SERCA was performed by rocking the samples with mouse c-Myc antibody (Thermo Scientific, 13-2500) at 4°C overnight, followed by 1 hour of rocking with Protein G Sepharose beads (GE Healthcare, 17-0618-01). Beads were washed four times with CHAPS buffer before boiling with 2× sample loading buffer containing 50 mM DTT. Samples were subjected to SDS-PAGE. Biotinylated SERCA and Myc-SERCA were visualized by immunoblotting using goat biotin antibody (Sigma-Aldrich B3640) and rabbit Myc antibody (Millipore 06-549), respectively.
CNX knockdown
HeLa cells were transfected with siCtrl or siCNX (Stealth Select RNAi, CANX HSS188690, control Stealth siRNA, Thermo Fisher Scientific) using Oligofectamine (Thermo Fisher Scientific). Cells were seeded in six-well plates (1 × 106 cells per well). The next day, cells were transfected following the manufacturer’s instructions. Experiments were performed 48 hours after transfection.
Data analysis
Data were analyzed using LibreOffice Calc 6.0 and R 3.5 (with RStudio 1.1). Data normality was determined with Shapiro-Wilk test and visual examination of QQ plots. Equality of variances between samples was determined with Levene’s test. Normally distributed data were analyzed with t test, and data with non-normal distribution were analyzed with Wilcoxon test. One-sample t test and one-sample Wilcoxon test were used to compare the differences between one sample and a normalized control (for example, to 100%). Two-sample t test and two-sample Wilcoxon test were used to compare two samples, with the corresponding adjustment depending on the equality of variances. The Holm adjustment method was used to compare the differences between more than two samples.
SUPPLEMENTARY MATERIALS
stke.sciencemag.org/cgi/content/full/13/638/eaax6660/DC1
Fig. S1. Additional controls and Ca2+ modeling.
Fig. S2. Additional ER and mitochondria Ca2+ and metabolic measurements.
Fig. S3. Additional biochemical analyses.
Fig. S4. NAC modulation of metabolic characteristics in HeLa cells.
Fig. S5. Analysis of ROS sources.
This is an article distributed under the terms of the Science Journals Default License.
