Na+ controls hypoxic signalling by the mitochondrial respiratory chain

Abstract

All metazoans depend on the consumption of O₂ by the mitochondrial oxidative phosphorylation system (OXPHOS) to produce energy. In addition, the OXPHOS uses O₂ to produce reactive oxygen species that can drive cell adaptations^1-4, a phenomenon that occurs in hypoxia^4-8 and whose precise mechanism remains unknown. Ca²⁺ is the best known ion that acts as a second messenger⁹, yet the role ascribed to Na⁺ is to serve as a mere mediator of membrane potential¹⁰. Here we show that Na⁺ acts as a second messenger that regulates OXPHOS function and the production of reactive oxygen species by modulating the fluidity of the inner mitochondrial membrane. A conformational shift in mitochondrial complex I during acute hypoxia¹¹ drives acidification of the matrix and the release of free Ca²⁺ from calcium phosphate (CaP) precipitates. The concomitant activation of the mitochondrial Na⁺/Ca²⁺ exchanger promotes the import of Na⁺ into the matrix. Na⁺ interacts with phospholipids, reducing inner mitochondrial membrane fluidity and the mobility of free ubiquinone between complex II and complex III, but not inside supercomplexes. As a consequence, superoxide is produced at complex III. The inhibition of Na⁺ import through the Na⁺/Ca²⁺ exchanger is sufficient to block this pathway, preventing adaptation to hypoxia. These results reveal that Na⁺ controls OXPHOS function and redox signalling through an unexpected interaction with phospholipids, with profound consequences for cellular metabolism.

Extended Data Fig. 1 |. Interference of Slc8b1 affects NCLX function, but inhibition of the NCLX has no effect on general mitochondrial function.

a–c, Assessment of the effect of different siRNAs against Slc8b1 mRNA (siNCLX) on mitochondrial Ca²⁺ influx and efflux rates in BAECs transfected with the mitochondria-directed Ca²⁺ reporter protein Cepia 2mt (siSCR n = 4, siNCLX1 n = 5, siNCLX2 n = 4, siNCLX3 n = 3, siNCLX4 n = 5), after addition of histamine; mean traces plotted as fluorescence change relative to initial fluorescence (F/F₀) (a); mitochondrial Ca²⁺ peak amplitude (calculated from the highest value of fluorescence minus the first basal value of fluorescence, from panel a) (b); percentage of mitochondrial Ca²⁺ efflux (calculated from the highest value of fluorescence minus the lowest value of fluorescence after histamine application, relative to percental siSCR values, from panel a) (c). d, Assessment of NCLX protein amount after interference in whole-cell extracts from BAECs, by western blot (Representative image of three independent experiments). e–g, Mitochondrial Ca²⁺ influx and efflux rates in WT, NCLX KO, KO+pNCLX or WT+dnNCLX MEFs transfected with Cepia 2mt (n = 4, except WT n = 5 and KO+pNCLX n = 3) after addition of histamine, as in a–c. h–j, Assessment of NCLX protein amount in MEFs mitochondrial extracts by western blot (representative images of n = 4 in h and n = 2 in i and j). k, Mitochondrial membrane potential in BAECs measured with 20 nM TMRM in non-quenching mode (n = 3). l, Oxygen consumption rate (OCR) in BAECs with NCLX inhibition by 10 μM CGP-37157 (upper panel, n = 4), or NCLX interference with siNCLX1 (lower panel, n = 3). All data are represented as mean ± s.e.m. One-way ANOVA with Tukey’s test (f, g and k for the siNCLX inset) and two-tailed Student’s t-test (b, c, k for the CGP inset). For gel source data, see Supplementary Fig. 1.

Extended Data Fig. 2 |. Hypoxia activates Na⁺/Ca²⁺ exchange through the NCLX.

Cytosolic Ca²⁺ (Ca²⁺_i) or Na⁺ (Na⁺_i) were measured by live imaging fluorescence microscopy with Fluo-4 AM or CoroNa Green AM, respectively, in normoxia or acute hypoxia (2% O₂). a–d, BAECs not treated (No treat) or transfected with siSCR or siNCLX. e–h, BAECs treated or not with the NCLX inhibitor CGP-37157 (10 μM). a, c, e, g, time-course traces; b, d, f, h, slopes. All data are represented as mean ± s.e.m. of n = 3 independent experiments. Two-tailed Student’s t-test (b, d, f, h): * P < 0.05, ** P < 0.01, *** P < 0.001, n.s. not significant.

Extended Data Fig. 3 |. Inhibition of the NCLX prevents the increase in ROS production triggered by hypoxia.

a–c, Superoxide detection by fluorescence microscopy after incubation with DHE in 10-min time windows in normoxia (Nx) or hypoxia (1% O₂); AA = antimycin A. BAECs treated or not with 10 μM CGP-37157, images of one experiment (a) and mean intensity of three independent experiments (b). HUVECs treated or not with 10 μM CGP-37157, mean intensity of three independent experiments (c). d, Detection of H₂O₂ by live confocal microscopy in CytoHyPer-transfected BAECs either not treated or treated with 10 μM CGP-37157 in normoxia (Nx) or acute hypoxia (1% O₂, Hp); representative images and time-course traces as mean of four independent experiments. e, Detection of ROS by live fluorescence microscopy with DCFDA in normoxia or acute hypoxia (2% O₂); time-course traces and slopes, mean of three independent experiments. f, g, Superoxide detection by fluorescence microscopy with DHE in normoxia (Nx) or 10 min hypoxia (1% O₂) or normoxia with AA in primary (f) or immortalized (g) HUVECs; mean of three independent experiments. All data are represented as mean ± s.e.m. Two-tailed Student’s t-test for pairwise comparisons (d, e) and one-way ANOVA with Tukey’s test for multiple comparisons (b, c, f, g): n.s. not significant, * P < 0.05, **P < 0.01, *** P < 0.001. In b, c, statistical comparisons shown only for Nx vs 0–10 groups.

Extended Data Fig. 4 |. NCLX activation in acute hypoxia depends on mitochondrial CI.

a, b, Assessment of interference of subunits of CI (NDUFS4) or CIII (RISP) by western blot in whole-cell extracts from BAECs. Representative images of two independent experiments. c, d, Cytosolic Ca²⁺ (Ca²⁺_i; c) or Na⁺ (Na⁺_i; d) measured by live imaging confocal microscopy with Fluo-4 AM or CoroNa Green AM, respectively, in normoxia or acute hypoxia (1% O₂); time-course traces and slopes, n = 4 (c), n = 3 (d). e, f, Effect of OXPHOS inhibitors on cytosolic Ca²⁺ (e; n = 3) or Na⁺ (f; n = 6) measured as in c and d. All data are represented as mean ± s.e.m. Two-tailed Student’s t-test or Mann–Whitney U test (c; siSCR). n.s. not significant, * P < 0.05, **P < 0.01. For gel source data, see Supplementary Fig. 1.

Extended Data Fig. 5 |. Acute hypoxia promotes A/D transition in CI and matrix acidification independently of NCLX activity.

a–c, ND3·Cys39 exposure, which reflects the D conformation of CI, measured as the ratio between TMR signal (Cys39 labelling) and Sypro Ruby staining (total protein for the ND3 band, identified by mass spectrometry). Thermal deactivation is used as a positive control of CI D state. Scheme of the technique (a); BAECs (b; n = 5 and n = 4 for de-activated samples) or HepG2 (c; n = 2) exposed to normoxia (Nx), 5 min of hypoxia (1% O₂, H5), normoxia with CGP-37157 (NxCGP), 5 min of hypoxia with CGP-37157 (H5CGP). d, Complex I reactivation rate measured in the presence of Mg²⁺ in isolated mitochondrial membranes from BAECs subjected to normoxia, 10 min of hypoxia (1% O₂) and NCLX inhibition with CGP-37157 (n = 4). e–g, Mitochondrial matrix pH measured using calibrated mitosypHer in WT or CI KO cybrid cell lines (e; n = 8), MEFs preincubated or not with rotenone (f; n = 4) or NCLX WT or KO MEFs (g; n = 8). h, Mitochondrial matrix acidification using mitosypHer in BAECs in normoxia or acute hypoxia (1% O₂) treated or not with NCLX inhibitor CGP-37157; time-course traces and slopes (n = 4). i, j, Effect of the CI inhibitor rotenone on cytosolic Ca²⁺ (i) or cytosolic Na⁺ ( j) measured by live confocal microscopy with Fluo-4 AM or CoroNa Green AM, respectively, in BAECs in normoxia or acute hypoxia (1% O₂), n = 3. All data are represented as mean ± s.e.m. Two-tailed Student’s t-test (b, d, h, i, j): n.s. not significant, * P < 0.05, **P < 0.01, *** P < 0.001.

Extended Data Fig. 6 |. Mitochondrial matrix acidification promotes mitochondrial Na⁺/Ca²⁺ exchange via the NCLX.

a, TEM-EDX determination of calcium element (Ca) versus carbon (C), oxygen (O) and lead (Pb) content in regions of mitochondria with (empty circles) or without (filled circles) electron dense spots (n = 7). b, Frequency of CaP precipitates per mitochondrion in BAECs, seen by TEM during normoxia (741 mitochondria), 10 min of hypoxia (1% O₂; 619 mitochondria) or 30 min with 1 μM FCCP (393 mitochondria), three independent experiments. c, Total Ca²⁺ content of mitochondria extracted from mouse adult fibroblasts (MAFs) which had been treated for 10 min with 1 μM FCCP or 1 μM rotenone, measured in a hypotonic buffer at pH 6.8 (n = 4). d, Western blot showing MCU and Fp70 in CRISPR Control, MCU KO2 and MCU KO3 cells (representative image of two independent experiments). e, Mitochondrial Ca²⁺ measured by live cell confocal microscopy in MCU WT or KO human breast cancer cells transfected with Cepia2mt (fluorescence signal relative to starting signal –F/F0–, representative traces of n = 5 for WT, n = 9 for MCU KO2 and n = 5 for KO3, independent experiments). f, Total Ca²⁺ content of mitochondria extracted from MAFs which had been subjected to normoxia or 10, 30 or 60 min of hypoxia (1% O₂), measured in a hypotonic buffer at pH 6.8 (n = 5). g, Ca²⁺ content of mitochondria extracted from MAFs which had been subjected to 10 min normoxia or 10 min hypoxia (1% O₂) in a medium without Ca²⁺, measured in a hypotonic buffer at pH 6.8 (total mitochondrial Ca²⁺) or pH 7.5 (soluble mitochondrial Ca²⁺; n = 4). h, Cytosolic Ca²⁺ measured by live cell confocal microscopy in MEFs transfected with cyto-GEM-GECO in the absence of Ca²⁺ in the incubation medium (n = 4). i, Endoplasmic reticulum Ca²⁺ measured by live cell confocal microscopy in MEFs transfected with erGAP3 in the presence or absence of Ca²⁺ in the incubation medium (n = 8). j, Superoxide detection with DHE in MCU WT or MCU KO immortalized breast cancer cells in normoxia (Nx) or after 10 min of hypoxia (1% O₂; Hp; n = 4). All data except e are represented as mean ± s.e.m. One-way ANOVA with Tukey’s test (c and f) and two-tailed Student’s t-test (a, b, g and j): n.s. not significant, * P < 0.05, ** P < 0.01. Two-tailed Student’s t-test (CRISPR Control Hp vs MCU KOs Hp in j): ^& P < 0.05, ^&& P < 0.01. For gel source data, see Supplementary Fig. 1.

Extended Data Fig. 7 |. Mitochondrial Na⁺ import decreases OXPHOS and produces ROS.

a, Effect of NaCl and/or CaCl₂ additions on H₂O₂ production detected with Amplex Red in isolated rat heart mitochondria (500 μg) respiring after addition of glutamate/malate (GM) in KCl-EGTA buffer. Representative traces of five independent experiments. b–e, Effect of NCLX activation by 10 mM NaCl/0.1 mM CaCl₂ on glutamate/malate- (b, d) or TMPD-based (c, e) oxygen consumption rate (OCR) in isolated coupled mitochondria from BAECs (b, c; n = 5) or MEFs (d, e; n = 6). f, g, Effect of 0.1 mM CaCl₂ (f; n = 5) or NaCl (g; n = 4) additions on CII+III activity in isolated mitochondrial membranes from BAECs. h, i, Cytosolic Ca²⁺ (h) or cytosolic Na⁺ (i) measured by live cell confocal microscopy of BAECs transfected with cyto-GEM-GECO (h) or treated with ANG2-AM (i), either not treated (No treat) or treated with 10 μM DMM during normoxia and hypoxia (1% O₂; n = 4). j, Superoxide detection by fluorescence microscopy after incubation with DHE in 10-min time windows in non-treated BAECs (No treat) or treated with 10 μM DMM during normoxia (Nx) or hypoxia (1% O₂; n = 3, except AA n = 2). k, l, Effect of 10 mM NaCl addition on succinate dehydrogenase activity (k) or ubiquinone 2-cytochrome c activity in the presence of n-Dodecyl β-D-maltoside (DDM; l) from BAECs mitochondrial membranes (n = 4). m, n, Effect of hypoxia (1% O₂) in mitochondrial Na⁺ content measured with SBFI in BAECs transfected with siSCR or siNCLX (m) or treated with CGP-37157 (n) (n = 4). o, Effect of hypoxia (1% O₂) in mitochondrial Na⁺ content in WT and KO MEFs, measured with CoroNa Green adapted for mitochondrial loading and live fluorescence (n = 10 MEFs WT and n = 12 MEFs KO). p, Representative images of three independent experiments showing colocalization of CoroNa Green-AM and TMRM signals after application of a long incubation protocol for CoroNa Green AM. All data are represented as mean ± s.e.m. Two-tailed Student’s t-test (f, k, l) and one-way ANOVA with Tukey’s test for multiple comparisons (b–e, h–j, m–o): n.s. not significant, * P < 0.05, ** P < 0.01, *** P < 0.001. Student’s t-test (No treat vs DMM; h, i): n.s. not significant, ^& P < 0.05. j, statistical comparisons shown only for Nx vs 0–10 groups. Pearson correlation coefficient in g: R = −0.9753.

Extended Data Fig. 8 |. Mitochondrial Na⁺ import decreases IMM fluidity.

a, FRAP of BAECs expressing mitoRFP in normoxia or hypoxia for 20 min (1% O₂), with or without CGP-37157 (n = 15 for Nx, n = 16 for Hp, n = 13 for NxCGP and n = 10 for HpCGP, obtained from four independent experiments). b, NAO quench FRAP signal of BAECs exposed to normoxia or hypoxia for 15 min (1% O₂; n = 5), with or without CGP-37157. c, CoQ₁₀ redox state of HUVECs subjected to normoxia (Nx), 10 min of hypoxia (Hp 10 min; 1% O₂), or normoxia with antimycin A (AA), n = 4. d, e, Anisotropy of phosphatydilcholine (PC) or PC:cardiolipin (PC:CL) liposomes treated with increasing concentrations of NaCl measured by TMA-DPH (d; n = 3) or DPH (e; n = 4) fluorescence. f–i, ESR spectra of 5-, 12- and 16-Doxyl PC in DOPC liposomes. 5-Doxyl PC exhibited an increased restricted motion (broadening of the hyperfine splitting, 2A_max) as a function of the NaCl concentration whereas the correlation time for 12- and 16-Doxyl PC remained unchanged. Hyperfine splitting (2A_max) of 5-Doxyl PC measured by ESR in PC liposomes treated with increasing concentrations of NaCl (n = 2) (f). Chemical structures of 5-, 12- and 16-Doxyl PC (g). ESR of 5-, 12- and 16-Doxyl PC in PC liposomes (h) and rotational times (i), τ_c, of 12- and 16-Doxyl PC in PC liposomes treated with increasing concentrations of NaCl as measured by ESR (n = 2). j, IR spectroscopy absorption spectra of PC liposomes treated or not with 16 mM NaCl (n = 2). Data are represented as mean ± s.e.m., except mean ± s.d. in d–f and i. Two-tailed Student’s t-test (c) and one-way ANOVA with Tukey’s test (a, b, d, e): n.s. not significant, * P < 0.05, ** P < 0.01, *** P < 0.001.

Extended Data Fig. 9 |. Mitochondrial Na⁺ influx inhibition abolishes hypoxic pulmonary vasoconstriction.

a, Effectiveness of NCLX silencing in mice pulmonary artery smooth muscle cells (PASMCs) measured by quantitative RT–PCR analysis, n = 3. b, Superoxide detection by fluorescence microscopy after incubation with DHE in PASMCs treated with siSCR or siNCLX in normoxia (Nx) or hypoxia (Hp; 1% O₂), n = 4 (except AA, n = 2). c, d, Representative traces (c) and average values (d) of hypoxic pulmonary vasoconstriction (HPV) measured in rat pulmonary arteries in the absence of pretone (precontraction). Each artery was exposed twice to hypoxia, and the second hypoxic challenge was performed in the absence or the presence of 30 μM CGP-37157, n = 7 (except peak CGP, n = 8; steady state CGP, n = 9). All data are represented as mean ± s.e.m. One-group t-test vs. 1 (a), two-tailed Student’s t-test (a, d) and one-way ANOVA with Tukey’s test (b): * P < 0.05, ** P < 0.01, *** P < 0.001. In b, statistical comparisons shown only for Nx vs 0–10 groups.

Extended Data Fig. 10 |. Scheme of the molecular pathway driving ROS production in acute hypoxia.

a, Complex I undergoes the A/D transition, this leads to mitochondrial matrix acidification and free Ca²⁺ release from the CaP precipitates. b, The increase in mitochondrial matrix free Ca²⁺ activates NCLX which introduces Na⁺ into the mitochondrial matrix. c, Na⁺ interacts with phospholipids in the inner leaflet of the inner mitochondrial membrane, decreasing diffusion of CoQH₂, uncoupling CoQ cycle only in free complex III and promoting superoxide production at the Qo site of complex III. d, Mitochondrial ROS in hypoxia activate downstream targets. In addition, the entry of extracellular Ca²⁺ promotes a positive feed-back loop that further increases mitochondrial Na⁺ import and ROS production in hypoxia.

Fig. 1 |. Hypoxia activates Na⁺/Ca²⁺ exchange and enhances ROS production through NCLX.

a–e, Ca²⁺_cyto (Cyto-GEM-GECO) or Na⁺_cyto (Asante NaTRIUM-2 AM (ANG2-AM)) measured by live-cell confocal microscopy in normoxia and after induction of hypoxia (1% O₂) in the following: BAECs transfected with scrambled siRNA (siSCR), siNCLX1 or siNCLX3 (a, b); untreated BAECs (No treat) or treated with CGP-37157 (CGP; c); and wild-type (WT), NCLX KO (KO), NCLX KO + pNCLX or NCLX WT + dnNCLX MEFs (d, e). f–i, ROS production measured by live-cell confocal microscopy (g, i) or fixed-cell fluorescence microscopy (f, h) in normoxia (Nx) or hypoxia (Hp; 1% O₂). Superoxide detection after incubation with dihydroethidium (DHE) in 10-min time windows in non-treated (Control), siSCR-treated or siNCLX-treated BAECs (f). AA, antimycin A. Detection of H₂O₂ with CytoHyPer in siSCR-treated or siNCLX-treated BAECs (g). Same as f in WT or NCLX KO immortalized MEFs (h). Same as g in WT, NCLX KO, KO + pNCLX or WT + dnNCLX immortalized MEFs (i). In a–e, time-course traces of three (a, b), six (c) or four (d, e) independent experiments, in f, h, the mean intensity of three independent experiments, in g, time-course traces of four independent experiments, and in i, the slopes of nine independent experiments (except for NCLX KO, n = 7; WT + dnNCLX, n = 10) are shown. All data are represented as mean ± s.e.m. One-way analysis of variance (ANOVA) with Tukey’s test for multiple comparisons (f, h, i) and two-tailed Student’s t-test (last normoxia time versus hypoxia times; a–e, g): *P < 0.05, ^**P < 0.01, ^***P < 0.001. Two-tailed Student’s t-test (last normoxia time versus hypoxia times in KO + pNCLX; d, e): ^&P < 0.05, ^&&P < 0.01. In f, h, statistical comparisons are shown only for normoxia versus 0–10 groups. AU, arbitrary units; NS, not significant.

Fig. 2 |. Mitochondrial CaP precipitates are the source of mitochondrial Ca²⁺ raise and of NCLX activation.

a, Ca²⁺_mito (Cepia 2mt) measured by live-cell confocal microscopy in BAECs transfected with siSCR or siNCLX during normoxia and hypoxia (1% O₂; n = 4). b, Same as a in WT or MCU KO human breast cancer cells during normoxia and hypoxia (1% O₂; n = 3). c, Ca²⁺_cyto (Cyto-GEM-GECO) measured by live-cell confocal microscopy in MCU WT or KO human breast cancer cells subjected to normoxia and hypoxia (1% O₂; n = 6). d, Representative transmission electron microscopy images and the mean area of mitochondrial CaP precipitates during normoxia (64 precipitates), 10-min hypoxia (1% O₂; 70 precipitates) or 30-min treatment with 1 μM FCCP (40 precipitates) in three independent experiments. The yellow asterisks mark the mitochondrial CaP precipitates. The box plots show the median and the first and third quartiles. e, Ca²⁺ content of mitochondria extracted from MEFs, which had been subjected to 10 min of normoxia or hypoxia (1% O₂), measured in a hypotonic buffer at pH 6.8 (total mitochondrial Ca²⁺) or pH 7.5 (soluble mitochondrial Ca²⁺; n = 8). f, Same as a in BAECs treated or not treated with 1 μM rotenone during normoxia and acute hypoxia (1% O₂; untreated n = 6 and rotenone n = 5). g, Same as a in WT or CI KO cybrids during normoxia and acute hypoxia (1% O₂; n = 8). All data are represented as mean ± s.e.m., except in d. One-way ANOVA with Tukey’s test for multiple comparisons (b, d) and two-tailed Student’s t-test (last normoxia time versus hypoxia times in a; normoxia versus hypoxia in e): *P < 0.05, ^**P < 0.01, ^***P < 0.001. Two-tailed Student’s t-test (siSCR versus siNCLX in a): ^&P < 0.05, ^&&P < 0.01, ^&&&P < 0.001.

Fig. 3 |. NCLX governs superoxide production and OXPHOS function in hypoxia through Na⁺-dependent alteration of IMM fluidity.

a, b, Superoxide detection with DHE with siSCR and siRISP (a), and untreated (No treat) or 1 μM myxothiazol-treated (Myxo) BAECs (n = 3) (b). c, d, Effect of NCLX activation on succinate-based respiration in BAECs (c; n = 5) or WT and NCLX KO (KO) MEFs (d; n = 6). e, f, Succinate-cytochrome c (CytC) activity in mitochondrial membranes from BAECs with or without NaCl (e; n = 4) and from hypoxic or normoxic WT or NCLX KO MEFs (f; 1% O₂; n = 3). g, ROS production on succinate-CytC activity in BAEC mitochondrial membranes with or without NaCl (n = 10; AA n = 9). The box plots show the median, the first and third quartiles, and 5th and 95th percentiles of the data. h, NADH-CytC activity in BAEC mitochondrial membranes with or without NaCl (n = 3). i, G3PD-CytC activity in BAEC mitochondrial membranes with or without NaCl (n = 3). j, k, Effect of hypoxia (1% O₂) on Na⁺_mito content in WT and NCLX KO MEFs, measured with CoroNa red by life-cell confocal miscroscopy ( j; n = 8, except KO + pNCLX and WT + dnNCLX n = 4), or with SBFI fluorescence after mitochondrial isolation (k; n = 5). l, m, FRAP of siSCR or siNCLX BAECs (l; n = 16) and WT, NLCX KO, KO + pNCLX or WT + dnNCLX MEFs (m; n = 14; except WT Hp and KO pNCLX Hp n = 15) expressing mitoRFP in normoxia or hypoxia (20 min; 1% O₂). n, NAO FRAP quench signal of siSCR or siNCLX BAECs exposed to normoxia or hypoxia (15 min; 1% O₂; n = 5). o, Anisotropy of PC liposomes with increasing concentrations of NaCl by diphenyl-1,3,5-hexatriene (DPH) fluorescence (mean ± s.d.; n = 2). p, IR absorption spectra of the carbonyl group of PC with or without NaCl (n = 2). All data are represented as mean ± s.e.m., except in g and o. Two-tailed Student’s t-test (a, e, h–j) and one-way ANOVA with Tukey’s test (b–d, f, g, k–o): NS, *P < 0.05, ^**P < 0.01, ^***P < 0.001. Two-tailed Student’s t-test ( j): ^&P < 0.05. In a, b, statistics are shown only for normoxia versus 0–10 groups.