TMEM65 regulates NCLX-dependent mitochondrial calcium efflux

Abstract

The balance between mitochondrial calcium (_mCa²⁺) uptake and efflux regulates ATP production, but if perturbed causes energy starvation or _mCa²⁺ overload and cell death. The mitochondrial sodium-calcium exchanger, NCLX, is a critical route of _mCa²⁺ efflux in excitable tissues, such as the heart and brain, and animal models support NCLX as a promising therapeutic target to limit pathogenic _mCa²⁺ overload. However, the mechanisms that regulate NCLX activity remain largely unknown. We used proximity biotinylation proteomic screening to identify the NCLX interactome and define novel regulators of NCLX function. Here, we discover the mitochondrial inner membrane protein, TMEM65, as an NCLX-proximal protein that potently enhances sodium (Na⁺)-dependent _mCa²⁺ efflux. Mechanistically, acute pharmacologic NCLX inhibition or genetic deletion of NCLX ablates the TMEM65-dependent increase in _mCa²⁺ efflux. Further, loss-of-function studies show that TMEM65 is required for Na⁺-dependent _mCa²⁺ efflux. Co-fractionation and in silico structural modeling of TMEM65 and NCLX suggest these two proteins exist in a common macromolecular complex in which TMEM65 directly stimulates NCLX function. In line with these findings, knockdown of Tmem65 in mice promotes _mCa²⁺ overload in the heart and skeletal muscle and impairs both cardiac and neuromuscular function. We further demonstrate that TMEM65 deletion causes excessive mitochondrial permeability transition, whereas TMEM65 overexpression protects against necrotic cell death during cellular Ca²⁺ stress. Collectively, our results show that loss of TMEM65 function in excitable tissue disrupts NCLX-dependent _mCa²⁺ efflux, causing pathogenic _mCa²⁺ overload, cell death and organ-level dysfunction, and that gain of TMEM65 function mitigates these effects. These findings demonstrate the essential role of TMEM65 in regulating NCLX-dependent _mCa²⁺ efflux and suggest modulation of TMEM65 as a novel strategy for the therapeutic control of _mCa²⁺ homeostasis.

Keywords: NCLX; TMEM65; calcium; mitochondria; sodium.

Extended Data Fig. 1:. Validation of mitochondrial localization of NCLX-BioID2-HA fusion protein.

Western blots showing progressive digestion of mitochondria from AC16 cardiomyocytes expressing NCLX-BioID2HA with increasing concentrations of trypsin. Arrow indicates ~25-kD BioID2-HA moiety. WCL, whole cell lysate; Cyto, cytosolic fraction. CypD, cyclophilin D. O.M.M., outer mitochondrial membrane; I.M.M., inner mitochondrial membrane.

Extended Data Fig. 2:. Effects of TMEM65 overexpression on _mCa²⁺ and cytosolic Ca²⁺ handling.

a-b, Densitometric quantifications of western blots shown in Fig. 2a. c, Western blots for pyruvate dehydrogenase (PDH) phosphorylation status in control and stable TMEM65-Myc-FLAG overexpression lines (TMEM65-OE). d, Densitometric quantification of PDH phosphorylation normalized to total PDH. (n = 3/group; **P<0.01, ****P<0.0001 by one-way ANOVA with Dunnett’s post-hoc test). e, _mCa²⁺ uptake rate for permeabilized cell experiments shown in Fig. 2b. (n = 4 for empty vector control, 3 for TMEM65-OE. **P<0.01 by one-way ANOVA with Sidak’s post-hoc test). f, Western blotting of AC16 cardiomyocytes transduced for 24 hours with adenovirus encoding human TMEM65. MOI, multiplicity of infection. Dashed line is to aid in visualizing separate conditions. g, Pooled traces for AC16 cardiomyocytes transduced with adenovirus encoding LacZ (Ad-LacZ) or human TMEM65 (Ad-TMEM65), showing extramitochondrial bath Ca²⁺ (Fura-FF). 4 million cells were permeabilized with digitonin (Dg) in the presence of thapsigargin (Tg), then administered a single 10µM Ca²⁺ bolus, with subsequent additions of the mitochondrial calcium uniporter inhibitor Ru360 and FCCP as indicated. (n = 3/group). Inset shows initial _mCa²⁺ efflux phase. _mCa²⁺ efflux rate over the first 25 seconds following addition of Ru360 (h); % of matrix Ca²⁺ effluxed post-Ru360 (i); and % net _mCa²⁺ uptake following 10µM Ca²⁺ bolus (j). (n = 3/genotype; *P< 0.05 by unpaired t-test). k, Western blotting of C2C12 myoblasts transduced for 24 hours with adenovirus encoding human TMEM65. l, Pooled traces for 2million mouse C2C12 skeletal myoblasts with acute transduction of Ad-LacZ or Ad-TMEM65, in the presence or absence of the NCLX inhibitor CGP-37157 (CGP), showing extramitochondrial bath Ca²⁺ (Fura-FF) as in (g) (n = 3/group). _mCa²⁺ efflux rate over the first 25 seconds following addition of Ru360 (m), % of matrix Ca²⁺ effluxed post-Ru360 (n); % net _mCa²⁺ uptake following the 10µM Ca²⁺ bolus (o), and initial _mCa²⁺ uptake rate for the first 25 seconds following the Ca²⁺ peak (p). (n = 3/group. ***P< 0.001, ****P<0.0001 vs. WT by two-way ANOVA with Sidak’s post-hoc test. All comparisons n.s. for panels o and p). Time to 50% decay (q) and half rise time (r) of KCl-evoked cytosolic Ca²⁺ transients as measured by Fluo-4 fluorescence in intact AC16 cardiomyocytes for experiment shown in Fig. 2j. (n= 9 cells/group. n.s. in all cases by unpaired t-test). s, peak amplitude of KCl-evoked cytosolic Ca²⁺ transients for experiment shown in Fig. 2j. (n = 9 cells/group. n.s. by Mann-Whitney test).

Extended Data Fig. 3:. Validation of NCLX-3xFLAG mouse and stages in modeling TMEM65-NCLX interaction.

a, Chromatogram of sequencing confirming proper in-frame insertion of DNA encoding the NCX-3xFLAG epitope tag prior to the stop codon of the endogenous mouse Nclx gene. b, Genotyping gel showing pups heterozygous (+/-) or homozygous (+/+) for the 3xFLAG knock-in. c, Western blots validating solubilization and specific detection of NCLX-3xFLAG protein (arrow) in isolated adult mouse heart mitochondria. Amido black staining for total protein is shown as a loading control. d, Absorbance at 280nm for gel filtration standards separated by size-exclusion chromatography. e, Standard curve of elution fraction vs. known molecular weight for gel filtration standards separated by size-exclusion chromatography. The equation shows the fit of the standards to a one-phase exponential decay. f, In silico molecular modeling showing interactions between individual TMEM65 transmembrane (TM) domains (red) and NCLX (blue).

Extended Data Fig. 4:. Phenotypic consequences of altered TMEM65 expression.

a, Representative images showing WGA labelling (green) and nuclear staining with DAPI (blue) in cross sections of the gastrocnemius muscle at 24 weeks of age. scr., scrambled. b, Total mass of the gastrocnemius + soleus + plantaris muscle, dissected as a unit, normalized to total body mass at 24 weeks of age. Open symbols represent male mice, filled symbols represent female mice. (n = 10 mice + scrambled shRNA, 9 mice + Tmem65 shRNA, *P<0.05 by unpaired t-test with Welch’s correction). Left ventricular end diastolic dimension (LVEDD) (c) and left ventricular end systolic dimension (LVESD) (d) as measured by echocardiography in mice at 6 weeks of age (n = 11 mice + scrambled shRNA, 7 mice + Tmem65 shRNA. *P<0.05 unpaired t-test). e, Pooled traces showing extramitochondrial bath Ca²⁺ (Fura-FF) and mitochondrial membrane potential (JC-1) in permeabilized WT and TMEM65^−/− AC16 cardiomyocytes in response to repeated additions of 5µM Ca²⁺ (arrows). Red arrows indicate the Ca²⁺ bolus typically triggering permeability transition in each genotype. (n = 3 replicates/genotype). f, Pooled traces showing extramitochondrial bath Ca²⁺ (Fura-FF) and mitochondrial membrane potential (JC-1) in permeabilized control or TMEM65-OE AC16 cardiomyocytes in response to repeated additions of 10µM Ca²⁺ (arrows). Red arrow indicates Ca²⁺ bolus triggering permeability transition in control cells. (n = 4 replicates for empty vector controls, 5 replicates for TMEM65-OE).

Fig. 1:. In vitro proximity biotinylation screen identifies TMEM65 as a member of the NCLX interactome.

a, Schematic of human NCLX-BioID2-HA fusion protein. IMS, intermembrane space. b, Western blotting of cellular fractions of AC16 cardiomyocytes transiently transfected with a plasmid encoding NCLX-BioID2-HA. MCU and VDAC served as mitochondrial markers. WCL, whole cell lysate; Cyto, cytosolic fraction, Mito, mitochondrial fraction. c, Expression of BioID2-HA and NCLX-BioID2-HA fusion protein and resulting cellular biotinylation. d, Summary of mass spectrometry results from proximity biotinylation screen in AC16 cardiomyocytes. Overlapping region of Venn diagram represents known or predicted mitochondrial proteins according to the IMPI, the biotinylation of which were enriched in NCLX-BioID2-HA samples compared to BioID2-HA negative controls. e, Immunofluorescence staining of AC16 cardiomyocytes transiently transfected with TMEM65-Myc-FLAG. Tom20 served as a mitochondrial marker. f, Western blots showing cellular fractionation and mitochondrial localization of exogenously expressed TMEM65-Myc-FLAG protein (arrows) in AC16 cardiomyocytes. g, Tmem65 mRNA expression in adult male (open symbols) and female (filled symbols) mouse tissues (n = 8 mice).

Fig. 2:. Gain of TMEM65 function enhances and loss of TMEM65 impairs sodium-dependent mitochondrial calcium efflux.

a, Western blotting of AC16 cardiomyocytes stably transduced with control empty vector or TMEM65-Myc-FLAG. Corresponding densitometric quantifications are shown in Extended Data Fig. 2 a–b. Line #20 (blue) was used in subsequent experiments. b, Pooled traces for AC16 cardiomyocytes with stable overexpression of TMEM65-Myc-FLAG (TMEM65-OE) or empty vector control, in the presence or absence of the NCLX inhibitor CGP-37157 (CGP), representing extramitochondrial bath Ca²⁺ (Fura-FF). 2 million cells were permeabilized with digitonin (Dg) in the presence of thapsigargin (Tg), then administered a single 10µM Ca²⁺ bolus, with subsequent additions of the mitochondrial calcium uniporter inhibitor Ru360 and FCCP as indicated. (n = 4 for control, 3 for TMEM65-OE). Inset shows initial _mCa²⁺ efflux phase. _mCa²⁺ efflux rate over the first 25 seconds following addition of Ru360 (c) and % net _mCa²⁺ uptake following the 10µM Ca²⁺ bolus (d). (n = 4 for control, 3 for TMEM65-OE. *P< 0.05, **P<0.01, ***P< 0.001, by one-way ANOVA with Sidak’s post-hoc test). e, Pooled traces showing KCl-evoked mitochondrial Ca²⁺ transients as measured by Mito-R-GECO fluorescence in intact AC16 cardiomyocytes. (n = 14 cells for empty vector control, 24 cells for TMEM65-OE). f, Time to 50% efflux for _mCa²⁺ transients (n = 11 cells for empty vector control, 24 cells for TMEM65-OE; **P< 0.01 by unpaired t-test with Welch’s correction). g, Area under the curve for _mCa²⁺ transients (n = 14 cells for empty vector control, 24 cells for TMEM65-OE; **P< 0.01 by Mann-Whitney test). h, Half rise time of _mCa²⁺ transients (n = 13 cells for empty vector control, 24 cells for TMEM65-OE, n.s. by Mann-Whitney test). i, Peak amplitude of _mCa²⁺ transients. (n = 14 cells for empty vector control, 24 cells for TMEM65-OE, n.s. by Mann-Whitney test). j, Pooled traces showing KCl-evoked cytosolic Ca²⁺ transients as measured by Fluo-4 fluorescence in intact AC16 cardiomyocytes. (n = 9 cells/group). k, Mean of pooled traces for plate reader _mCa²⁺ flux assays with permeabilized AC16 cells. Traces show extramitochondrial Ca²⁺ (Ca²⁺ Green 5N fluorescence), beginning with addition of the final of 5 consecutive 1-µM Ca²⁺ boluses, followed by stimulation of _mCa²⁺ efflux by the addition of 10mM NaCl. GCP, CGP-37157. Error bars were omitted for clarity. (n = 7/group). l, Quantification of Na⁺-stimulated and CGP-sensitive _mCa²⁺ efflux (n = 7/group, *P<0.05, ****P<0.001 by 1-way ANOVA with Tukey’s post-hoc test). m, Approach for CRISPR/Cas9-mediated disruption of TMEM65. n, Western blot confirming loss of TMEM65 protein expression in clonal AC16 cardiomyocyte lines following CRISPR/Cas9-mediated gene editing. o, Western blots for pyruvate dehydrogenase (PDH) phosphorylation status in wild-type (WT) and TMEM65^−/− AC16 cardiomyocytes. p, Densitometric quantification of PDH phosphorylation normalized to total PDH. (n = 3/group; **P<0.01 by unpaired t-test). q, Pooled traces for WT and TMEM65^−/− AC16 cardiomyocytes showing extramitochondrial bath Ca²⁺ (Fura-FF). 4 million cells were permeabilized with digitonin (Dg) in the presence of thapsigargin (Tg), then administered a single 10µM Ca²⁺ bolus, with subsequent additions of the mitochondrial calcium uniporter inhibitor Ru360 and FCCP as indicated. (n = 3/group). r, _mCa²⁺ efflux rate over the first 25 seconds following addition of Ru360. (n = 3/group, *P<0.05 by unpaired t-test with Welch’s correction). Percent of matrix Ca²⁺ effluxed post-Ru360 (s); % net _mCa²⁺ uptake following the 10µM Ca²⁺ bolus (t); and _mCa²⁺ uptake rate over the first 25 seconds following the Ca²⁺ peak (u). (n = 3 group. ***P< 0.001 by unpaired t-test). v, Mean of pooled traces for plate reader _mCa²⁺ flux assays with permeabilized AC16 cells as in (k). Error bars were omitted for clarity. (n = 5/group). w, Quantification of Na⁺-stimulated and CGP-sensitive _mCa²⁺ efflux (n = 5/group, *P<0.05 by 1-way ANOVA with Tukey’s post-hoc test).

Fig. 3:. TMEM65 and NCLX interact functionally and physically.

a, Pooled traces showing KCl-evoked _mCa²⁺ transients as measured by Mito-R-GECO fluorescence in intact Nclx^fl/fl and Nclx^−/− mouse embryonic fibroblasts after transduction with Ad-LacZ or Ad-TMEM65 (n = 16 cells for Nclx^fl/fl + Ad-LacZ; 21 for Nclx^−/− + Ad-LacZ; 23 for Nclx^fl/fl + Ad-TMEM65; 24 for Nclx^−/− + Ad-TMEM65). _mCa²⁺ efflux rate measured over the first 60 seconds following the peak of the mitochondrial Ca²⁺ transient (b), and area under the curve (c) for _mCa²⁺ transients shown in panel (a). (n = 16 cells for Nclx^fl/fl + Ad-LacZ; 21 for Nclx^−/− + Ad-LacZ; 23 for Nclx^fl/fl + Ad-TMEM65; 24 for Nclx^−/− + Ad-TMEM65; *P<0.05; **P<0.01; ****P<0.0001 by 2-way ANOVA with Sidak’s post hoc test). d, CRISPR-Cas9 knock-in strategy to insert a 3xFLAG epitope tag after the final sense codon of the mouse Slc8b1 gene. ssODN, single-stranded oligodeoxynucleotide. e, Western blotting showing molecular weight fraction distribution of NCLX-3xFLAG and TMEM65 after size-exclusion chromatography of isolated mouse heart mitochondria. Chromatogram and molecular weight standard curve for gel filtration standards of known molecular weight are shown in Extended Data Fig. 3d–e. f, Western blotting showing molecular weight fraction distribution of endogenous NCLX after size-exclusion chromatography of isolated WT and TMEM65^−/− AC16 cardiomyocyte mitochondria. g, quantification of NCLX signal intensity across fractions shown in panel (f), normalized to total NCLX signal intensity. (n = 3 / genotype; **P<0.01 by 2-way ANOVA with Sidak’s post-hoc test. h, In silico molecular modeling showing predicted interaction between NCLX (blue) and the transmembrane domains (TM1–3) of TMEM65 (red). Modeling places NCLX adjacent to the longest transmembrane helix of NCLX, depicted in ribbon and space-filling views. i, Superimposing the structure of full-length, mature mitochondrial TMEM65 onto the model shown in panel (h) predicts electrostatic interaction between the soluble domain of TMEM65 and the positively-charged regulatory region of NCLX’s longest transmembrane helix that includes R255 and R256.

Figure 4:. Loss of TMEM65 expression impairs striated muscle function and predisposes to _mCa²⁺-overload-induced cell death, while TMEM65 overexpression protects against Ca²⁺-induced cell death.

a, Western blots confirming reduction in TMEM65 protein expression in gastrocnemius muscle of 24-week-old mice injected as neonates with AAV9-Tmem65 shRNA. Dashed lines are to aid in visualizing separate conditions. b, Densitometric quantification of gastrocnemius muscle TMEM65 expression, normalized to total OXPHOS complexes (CI-CV). Male mice are shown in open symbols, female mice are shown in filled symbols. scr., scrambled. (n = 10 mice + scrambled shRNA, 9 mice + Tmem65 shRNA, **P<0.01 by Mann-Whitney test). c, Densitometric quantification of gastrocnemius PDH phosphorylation normalized to total PDH. (n = 10 mice + scrambled shRNA, 9 mice + Tmem65 shRNA, **P<0.01 by unpaired t-test). d, Representative images showing abnormal hindlimb positioning and/or clasping in Tmem65 knockdown (KD) mice at 14 weeks of age. scr., scrambled. e, Mean holding impulse representing performance of mice on 4-limb wire hang test at 24 weeks of age. (n = 10 mice + scrambled shRNA, 9 mice + Tmem65 shRNA. ***P<0.001 by Mann-Whitney test). Fiber cross-sectional-area (f), minimal Feret’s diameter (g), and % centrally nucleated fibers (h) for the gastrocnemius muscle. (n = 6 mice / group. **P<0.01 by unpaired t-test). i, Western blots confirming reduction in TMEM65 protein expression in the hearts of Tmem65 KD mice at 6 weeks of age. Dashed lines are to aid in visualizing separate conditions. Densitometric quantification of heart TMEM65 expression, normalized to total OXPHOS complexes (CI-CV) (j), and densitometric quantification of heart PDH phosphorylation normalized to total PDH (k) (n = 11 mice + scrambled shRNA, 7 mice + Tmem65 shRNA. **P<0.01, ****P<0.0001 by unpaired t-test). l, Left ventricular fractional shortening measured by echocardiography in mice at 6 weeks of age (n = 11 mice + scrambled shRNA, 7 mice + Tmem65 shRNA. *P<0.05 by unpaired t-test). m, Linear regression showing positive correlation between left ventricular fractional shortening and TMEM65 protein expression, as measured in mice with scrambled shRNA (n = 11) and mice with Tmem65 shRNA (n = 7). n, Number of Ca²⁺ boluses tolerated prior to spontaneous mitochondrial depolarization and release of matrix Ca²⁺ for the calcium retention capacity assay in permeabilized AC16 cells shown in Extended Data Fig. 4e (n = 3/genotype. *P<0.05 by unpaired t-test). o, TMRM staining intensity as a marker of mitochondrial membrane potential in cells incubated acutely with vehicle (veh.) or cyclosporine A (CsA). (n = 170 cells for WT + vehicle; 140 cells for TMEM65^−/− + vehicle; 225 cells for WT + CsA; 178 cells for TMEM65^−/− + CsA. ****P<0.0001 by two-way ANOVA with Sidak’s post-hoc test). p, Magnitude of drop in JC-1 fluorescence ratio upon addition of FCCP, corresponding to mitochondrial membrane potential after the final Ca²⁺ addition, for calcium retention capacity (CRC) assay in permeabilized AC16 cells shown in Extended Data Fig. 4f. (n = 4 replicates for empty vector control, 5 replicates for TMEM65-OE; **P<0.01 by unpaired t-test with Welch’s correction). q, Plasma membrane rupture as measured by Sytox green staining, as an indicator of cell death in response to 24-hour treatment with thapsigargin (Tg). (n = 56 for control + vehicle; 60/group for control + Tg, TMEM65-OE + vehicle, and TMEM65-OE + Tg; ****P<0.001 by 2-way ANOVA with Sidak’s post-hoc test). r, Plasma membrane rupture in response to 24-hour treatment with ionomycin (Iono.) (n =55 for control + vehicle; 60/group for control + Iono., TMEM65-OE + vehicle, and TMEM65-OE + Iono.; ****P<0.001 by 2-way ANOVA with Sidak’s post-hoc test).