TMBIM5 is the Ca2+ /H+ antiporter of mammalian mitochondria

Abstract

Mitochondrial Ca²⁺ ions are crucial regulators of bioenergetics and cell death pathways. Mitochondrial Ca²⁺ content and cytosolic Ca²⁺ homeostasis strictly depend on Ca²⁺ transporters. In recent decades, the major players responsible for mitochondrial Ca²⁺ uptake and release have been identified, except the mitochondrial Ca²⁺ /H⁺ exchanger (CHE). Originally identified as the mitochondrial K⁺ /H⁺ exchanger, LETM1 was also considered as a candidate for the mitochondrial CHE. Defining the mitochondrial interactome of LETM1, we identify TMBIM5/MICS1, the only mitochondrial member of the TMBIM family, and validate the physical interaction of TMBIM5 and LETM1. Cell-based and cell-free biochemical assays demonstrate the absence or greatly reduced Na⁺ -independent mitochondrial Ca²⁺ release in TMBIM5 knockout or pH-sensing site mutants, respectively, and pH-dependent Ca²⁺ transport by recombinant TMBIM5. Taken together, we demonstrate that TMBIM5, but not LETM1, is the long-sought mitochondrial CHE, involved in setting and regulating the mitochondrial proton gradient. This finding provides the final piece of the puzzle of mitochondrial Ca²⁺ transporters and opens the door to exploring its importance in health and disease, and to developing drugs modulating Ca²⁺ exchange.

Keywords: LETM1; TMBIM5 (MICS1); mitochondrial Ca2+-H+ exchanger; mitochondrial metabolism; permeability transition pore.

Figure 1. LETM1 and TMBIM5 interact

1. Scheme illustrating workflow for miniaturized AP‐MS experiments, left to right: whole cells or isolated mitochondria were lysed or solubilized, respectively. The cell/mitochondrial lysates were used for affinity purification (AP) using the STREP tag and tandem affinity purification (TAP) using STREP and HA tag found on the bait protein. Eluates of the AP and control experiments were reduced, alkylated, and digested by trypsin. Peptides are purified on a C18 stage tip and then run on an LTQ Orbitrap Velos. Protein identifications were made by internal tools using MASCOT and Phenyx and the removal of nonspecific interactors done using the CRAPome. Created with Biorender.com.

2. MCU was selected as a model protein, the functional complex consists of the five proteins above (MCU, MCUb, MICU1, MICU2, EMRE). Note that an additional tissue‐specific tertiary interaction partner (MICU3) is only expressed at very low levels in HEK293 cells (Diego De Stefani, personal communication). Illustration adapted from Sancak et al (2013).

3. All high‐confidence interaction partners of LETM1 are shown as nodes. Node color indicates SAINT score, a probability‐based measure of interaction confidence. See also Appendix Fig S1A–C.

4. Co‐immunoprecipitation of TMBIM5 and LETM1 protein in tandem in the left 3 panels. The input represents the mitochondrial crudely isolated from HEK293 cells and was used as input for the co‐IP, LETM1 was immunoprecipitated (left panel, IP: LETM1) using a LETM1 monoclonal antibody and Protein G magnetic beads (ProtG). ProtG beads alone were used as a negative control for binding, immunoprecipitates were immunoblotted (IB) for the indicated proteins to demonstrate interaction. 10% of the input was used for immunoblotting. Prohibitin (PHB) was used as a control to illustrate no nonspecific binding of inner mitochondrial membrane protein complexes. The middle and right panel of the co‐IPs illustrates the converse experiment, in the middle in TMBIM5WT and right TMBIMKO, using TMBIM5 as bait (right panel, IP: TMBIM5). The last two right panels show blots from BN–PAGE conducted in TMBIM5WT and KO.

Source data are available online for this figure.

Figure 2. TMBIM5KO causes mitochondrial matrix swelling and cristae disorganization

1. Western blot analysis of TMBIM5 in control and targeted HeLa and HEK293 clones.

2. Proliferation assay of HEK293 cells in the function of TMBIM5. Graph shows the mean ± SD of three individual counts, One‐way ANOVA with the Dunnett's multiple comparisons test performed against TMBIM5WT *P = 0.0155.

3. Live imaging of HEK293 TMBIM5WT and KO cells stained with MitoTracker Green FM. Scale bars: 10 μm.

4. Alteration of the mitochondrial ultrastructure shown by transmission electron microscopy, red arrow pointing to the dilated matrix. Wider mitochondria in the middle and right panel compared with controls, a middle panel showing the strongest phenotype of matrix width and cristae forms. Scale bars: 1 μm.

5. Isolated mitochondria from three independent replicates of HEK293 TMBIM5WT, and TMBIM5KO1 and KO2 were analyzed by immunoblotting using the indicated antibodies, HSP60 and TOM40 served as mitochondrial loading controls.

6. Densitometric analysis of the bands in (E) normalized to loading control, bar graph of three individual experiments (biological replicates), mean ± SD, one‐way ANOVA with the Bonferroni's multiple comparisons test performed against TMBIM5WT *P < 0.05, **P < 0.008, two‐way ANOVA with the Bonferroni's multiple comparisons test performed for the OPA1 statistics against TMBIM5WT, ***P = 0.0009, ****P < 0.0001.

Source data are available online for this figure.

Figure EV1. TMBIM5 and LETM1 are present in protein complexes of the same molecular weight

Immunoblotting analysis of blue native PAGE of isolated mitochondria from HeLa WT, TMBIM5KO, or LETM1KD cells, using the indicated antibody, arrows indicate TMBIM5 complexes.

Figure EV2. TMBIM5KD decreases LETM1 and mitochondrial bioenergetics

1. A

   Western blot analysis of LETM1 and TMBIM5 in HEK293 TMBIM5WT cells with a scramble shRNA and two different TMBIM5 knockdowns, HSP60 served as a loading control.

2. B

   Proliferation curve of TMBIM5WT scramble controls (scr) compared with TMBIM5KD cells (KD) over 4 days. Data are means ± SEM (scr, KD2 n = 3, KD1 n = 5) (biological replicates), at 96 h statistical analysis using an unpaired student's t‐test (*P < 0.05).

3. C–F

   Cellular bioenergetics of TMBIM5KD cells in various nutrient conditions. Oxygen consumption rate of WT cells with a scrambled control (WT) and TMBIM5KD cells (KD) grown in (C) 25 mM glucose, (E) 10 mM galactose for 24 h before measurement. Data are representative of at least three independent experiments (biological replicates). Shown are the mean data of triplicate measurements ± SEM. Inhibitors as indicated: (A) oligomycin (0.5 μM), (B, C) FCCP (0.2 μM each), (D) antimycin A/rotenone (0.5 μM). (D, F) Bar charts of XF experiment traces (C, E), data are means of multiple time points after experiment start or drug addition of at least three independent experiments ± SEM (biological replicates). Statistical analysis using an unpaired student's t‐test (**P < 0.01, ***P < 0.001).

Source data are available online for this figure.

Figure 3. TMBIM5 and LETM1 are involved in mitochondrial KHE activity

1. A–F

   KOAc‐induced swelling was measured in mitochondria from HEK293 TMBIM5WT, TMBIM5KO and TMBIM5KO cells stably re‐expressing TMBIM5WT (A, B), HeLa TMBIM5WT and TMBIM5KO (C, D) and HeLa LETM1 scramble (scr) and LETM1KD (E, F) cells. TMBIM5WT: black traces/bars, TMBIM5KO: red traces/bars, TMBIM5KO + TMBIM5WT: blue traces/bars, LETM1scr: black trace/bar, LETM1KD: green trace/bar. (B–E) Quantification of swelling amplitudes, data shown in (B, D) are the mean ± SD from three to five independent experiments (biological replicates). (B) HEK293 TMBIM5WT (100 ± 19.71) and HEK293 TMBIM5KO (48.08 ± 11.906). TMBIM5KO + TMBIM5WT (90.55 ± 12.93). (D) HeLa TMBIM5WT (100 ± 9.47) and HeLa TMBIM5KO (63.48 ± 8.60). Lower basal optical density indicates swollen matrix prior KOAc addition; Inhibition of KHE with quinine (WT gray bar, 18.14 ± 21.02; KO1 pink bar, 9.33 ± 28.17; TMBIM5KO + TMBIM5WT, light blue 15.79 ± 10.04. Statistical analysis: One‐Way ANOVA with Bonferroni correction (*P < 0.05, **P < 0.01) against HEK293 TMBIM5WT and an unpaired student's t‐test (**P < 0.01) against HeLaTMBIM5.

Source data are available online for this figure.

Figure 4. TMBIM5 controls Na⁺‐independent Ca²⁺ release

1. A–L

   Ca²⁺ uptake/release dynamics are shown as extramitochondrial Ca²⁺ changes of fluorescence intensities of Calcium Green 5N (Ca²⁺‐5N; 0.24 μM). Experiments conducted in permeabilized HEK293 TMBIM5WT, TMBIM5KO (A–H and K, L), TMBIM5KO + TMBIM5WT cells (A, B), WT overexpressing an empty vector (EV), or TMBIM5WT (OE) and TMBIM5KO overexpressing TMBIM5D325R (G, H), and HEK293 LETM1scr and LETM1KD cells (I, J). CGP37157 (2 μM) present in (A–L). Ca²⁺ (10 μM), RR (0.2 μM), and FCCP (2 μM) or alamethicin (2.5 μM) were added when indicated. Cyclosporin A (CsA) was added 2 min before measurements in (E, F and K, L), and nigericin 1 min before measurements (K, L). Quantification of Ca²⁺ release rates from three (D, H and F–L) and three to six (B, J) independent experiments (biological replicates; t: 300–920 s). Data are the mean ± SD and statistical analysis: One‐Way ANOVA with Bonferroni correction (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). See also Appendix Fig S3 for quantification of Ca²⁺ uptake.

Source data are available online for this figure.

Figure EV3. TMBIM5 mitochondrial Na⁺‐dependent Ca²⁺ release in HeLa cells

1. A–D

   Ca²⁺ uptake release assays were conducted in permeabilized HeLa TMBIM5WT or TMBIM5KO cells or TMBIM5KO cells expressing pcDNA3.1+TMBIM5 in presence of CGP37157 (2 μM) as described in Fig 4A–D applying a 10 μM (A) or 20 μM (C) Ca²⁺ pulse. In (C) Ca²⁺ uptake release was recorded in permeabilized HeLa TMBIM5WT or TMBIM5KO cells or TMBIM5KO re‐expressing TMBIM5 cells.

2. E, F

   Same experimental setting as in (C) but in presence of Tg (1 μM). Quantification of ≥ 3 independent experiments (biological replicates), data are the mean ± SD with an unpaired student's test (B, F) and one‐way ANOVA with Bonferroni correction (D), (*P < 0.05, ***P < 0.001, ****P < 0.0001, ns, not significant).

Source data are available online for this figure.

Figure EV4. TMBIM5 protein levels are not decreased by TMBIM5^D325R

Immunoblots of Isolated mitochondria from HEK293 TMBIM5WT and TMBIM5KO expressing TMBIM5^D325R or double mutant TMBIM5^D325R/D294R using the indicated antibodies, TOM40 served as mitochondrial loading control. Source data are available online for this figure.

Figure 5. TMBIM5KO reduces mitochondrial Ca²⁺ ([Ca²⁺]_mit) responses evoked by Ca²⁺‐mobilizing agonists

1. 4mtD3cpv ratio fluorescence changes (R/R0) evoked by 50 μM NaATP and 100 μM Carbachol (at time 60 seconds) in HEK293 cells depleted of NCLX and lacking or not TMBIM5.

2. Statistical evaluation of the [Ca²⁺]_mit response amplitude (left) and recovery rates (right). Data are mean ± SD of six separate recordings with 10–14 cells each in three independent experiments (biological replicates), *P < 0.05, unpaired two‐tailed Student's t‐test.

Source data are available online for this figure.

Figure EV5. TMBIM5KO induces PTP opening under Ca²⁺ overload

1. A–H

   Ca²⁺ uptake/release dynamics in presence of CGP37157 (2 μM) and Tg (1 μM) were monitored as in Fig 5. Ca²⁺ (10 μM), RR (0.2 μM), and FCCP (2 μM) were added when indicated. Membrane potential was recorded as the change in fluorescence intensities of TMRM (330 nM) (C, D) corresponding to the measurement of Ca²⁺ fluxes in (A, B). CsA was added 2 min before measurements in (E, F dotted lines). Quantification of Ca²⁺ release rates from three independent experiments (biological replicates) are shown as means ± SD (t: 300–920 s) and statistical analysis: One‐Way ANOVA with Bonferroni correction (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). Quantification of TMRM performed with an unpaired two‐sided t‐test (Welsh correction), *P < 0.05. (E, F) Calcium retention capacity (CRC) assays showing that the absence of TMBIM5 supersensitizes mitochondria to Ca²⁺‐induced PTP opening by Tg. See also Appendix Fig S6A and B for CRCs in absence of Tg. Permeabilized HEK293 TMBIM5WT (E) and TMBIM5KO1 (F) cells exposed or not to CsA were subjected to sequential Ca²⁺ bolus of 5 μM Ca²⁺, and fluorescence intensity was recorded.

Source data are available online for this figure.

Figure 6. TMBIM5 depletion alters mitochondrial pH and ∆pH

1. In situ calibration of mitoSypHer (left) and SNARF (right) in HEK293 cells. Data are the average of 120 cells from three independent experiments (biological replicates).

2. Average resting pH_mito (left) and pH_cyto (right) of WT and TMBIM5KO cells expressing the indicated shRNA, measured by high‐throughput ratio fluorescence imaging. Data are the mean ± SD of 20–60 individual cells in duplicates from three independent experiments (biological replicates). ****P < 0.0001; ***P < 0.0005; NS, not significant, one‐way ANOVA.

3. Averaged mitochondrial pH gradient (ΔpHm = pH_mito − pH_cyto) of the indicated cell lines in resting conditions. Data are the mean ± SD of 10–15 cells of 2–4 images, in duplicates from three independent experiments (biological replicates). **P < 0.01;*P < 0.05; NS, not significant, one‐way ANOVA.

4. Effect of Tg (1 μM for 10 min, right) on the ΔpHm of the indicated cell lines. Data are the mean ± SD of 10–15 cells of 2–4 images, in duplicates from three independent experiments (biological replicates). *P < 0.05 vs. WT Ctrl, NS, not significant, Student's T‐test.

Source data are available online for this figure.

Figure 7. TMBIM5 proteoliposomes mediate Ca²⁺ and Ca²⁺‐dependent H⁺ transport

1. A

   Western blot analysis of purified and reconstituted hTMBIM5 for evaluating the incorporation of hTMBIM5 into proteoliposomes prepared as described in Materials and Methods.

2. B

   Sketch illustrating the reconstitution of hTMBIM5 in proteoliposomes (red), and the empty liposomes (blue) prepared with Ca²⁺‐5N or pyranine for (C–F). Created with Biorender.com.

3. C–F

   Transport of Ca²⁺ by hTMBIM5 reconstituted in proteoliposomes containing 10 μM Ca²⁺‐5N at the pH indicated in the panels (C–E) or 20 μM pyranine at pH 7.0 (F). After reconstitution, the fluorescence measurement was started by diluting 200 μl proteoliposomes (red trace) up to 3 ml with transport buffer prepared as described in Materials and Methods at the indicated pH (C, E) or at pH 7.0 (D, F). After 100 s, as indicated by the arrow, 7 mM Ca²⁺ was added to the sample and fluorescence change was recorded. As a control, the same measurement was performed by diluting 200 μl liposomes (without incorporated protein, blue trace) up to 3 ml with the same transport buffer. See also Appendix Fig S7 for TMBIM5 optimization, induction, and structure overview.

4. G, H

   Transport of Ca²⁺ by hTMBIM5 reconstituted in proteoliposomes containing 10 μM Ca²⁺‐5N at pH 7.0. recorded as in (D). (G) and (H) left: sketch illustrating the assay. In (G), after reconstitution 200 μl proteoliposomes were diluted up to 3 ml with transport buffer prepared as described in Materials and Methods. Delta pH was generated adding Ca²⁺ at pH 6.0 (light green trace), 7.0 (red trace), or 8.0 (deep green trace). In (H), after reconstitution 200 μl proteoliposomes were diluted up to 3 ml with transport buffer prepared as described in Materials and Methods in the presence of 20 mM K⁺ and ethanol (red trace) or 10 μM nigericin (orange blue) to generate delta pH. Fluorescence intensity is indicated as Arbitrary Units (AU). Results are representative of three independent experiments (biological replicates).

Source data are available online for this figure.