MICS1 is the Ca2+/H+ antiporter of mammalian mitochondria

Mitochondrial Ca2+ ions are crucial regulators of bioenergetics, cell death pathways and cytosolic Ca2+ homeostasis. Mitochondrial Ca2+ content strictly depends 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 identified MICS1, the only mitochondrial member of the TMBIM family. Applying cell-based and cell-free biochemical assays, here we demonstrate that MICS1 is responsible for the Na+- and permeability transition pore-independent mitochondrial Ca2+ release and identify MICS1 as the long-sought mitochondrial CHE. 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.

Introduction expression of MICS1 in HEK293 MICS1 KO was able to restore Ca 2+ efflux ( Figure 5A). 181 Since MICS1 and LETM1 interact, and LETM1 was proposed as the mitochondrial CHE, we next 182 sought to address once again Ca 2+ fluxes in HEK293 LETM1 KD under the same conditions. The 183 presence or absence of LETM1 (Figure 4 ̶ figure supplement 1) did not alter Ca 2+ uptake 184 (Figure5 ̶ figure supplement 1) nor the Na + -independent Ca 2+ fluxes (Figure 5E-F). To assess 185 whether the permeability transition pore (PTP) contributes to the recorded Ca 2+ fluxes, we 186 repeated Ca 2+ uptake/efflux assays in presence of cyclosporin A (CsA), the PTP desensitizer 187 (Basso et al., 2008). MICS1WT displayed comparable Ca 2+ efflux as in the absence of CsA,188 confirming that Na + -independent Ca 2+ release was also independent of PTP flickering or 189 opening ( Figure 5G-H). Addition of CsA hardly altered the rate or magnitude of Ca 2+ release. 190 Since deletion of MICS1 or LETM1 reduces KHE activity, we asked whether increasing KHE 191 activity would restore Ca 2+ release in MICS1KO mitochondria. Therefore, we repeated the 192 previous experiment in the presence of nigericin, a highly selective ionophore catalyzing KHE,193 which did not restore Ca 2+ efflux ( Figure 5I-J). Thus, our results indicated that Na + -194 independent Ca 2+ efflux requires MICS1 but not LETM1 or LETM1-mediated KHE activity. 195

Thapsigargin-mobilized Ca 2+ induces PTP opening in MICS1KO cells 196
The similar vigorous Ca 2+ uptake by MICS1KO and -WT mitochondria but unequal Ca 2+ release, 197 unless alamethicin was used, raised the intriguing question of the fate of intramitochondrial 198 Ca 2+ . To exclude the ER as a Ca 2+ sink and deplete ER stores, we repeated Ca 2+ uptake/release 199 experiments using measurement media containing the SERCA pump inhibitor thapsigargin. 200 MICS1WT mitochondria behaved as in the absence of thapsigargin, with identical rapid Ca 2+ 201 influx, RR-induced Ca 2+ efflux and FCCP-induced release of total free matrix Ca 2+ (Figure 5K). 202 refractory to Ca 2+ efflux in absence of thapsigargin, released RR-induced Ca 2+ with rates 4-6 205 times higher than in MICS1WT. The levels of Ca 2+ efflux seemed saturated, as they almost 206 reached those of total Ca 2+ release after FCCP addition, which in presence of thapsigargin 207 were comparable to those of MICS1WT (Figure 5K-L). These drastic effects of thapsigargin on 208 mitochondrial RR-induced Ca 2+ efflux observed when MICS1 was deleted and NCLX inhibited, 209 suggested stimulation of the CHE or opening of the PTP, which could both be caused by 210 increased matrix Ca 2+ load. Consistent with PTP opening (Beghi and Giussani, 2018),  induced Ca 2+ release was accompanied by significant depolarization of MICS1KO but not 212 MICS1WT mitochondria as indicated by the membrane potential dye TMRM (Figure 5M-N). 213 To verify the PTP Ca 2+ -sensitivity and evaluate the total free Ca 2+ load tolerated by MICS1KO 214 mitochondria, we performed Ca 2+ retention capacity (CRC) assays. MICS1WT mitochondria 215 exposed to thapsigargin in presence of CGP37157 tolerated 5 Ca 2+ pulses, corresponding to 216 25 µM Ca 2+ before PTP opening ( Figure 5O). In contrast, MICS1KO#1 only tolerated 3 Ca 2+ 217 pulses, corresponding to 15 µM Ca 2+ ( Figure 5P). PTP desensitization with CsA increased the 218 to Ca 2+ -induced PTP opening when NCLX is inhibited (Luongo et al., 2017). Consistent with a 227 role of PTP opening in the large Ca 2+ release observed in MICS1KO mitochondria, addition of 228 CsA and ADP prevented excess Ca 2+ release from MICS1KO mitochondria (Figure 5Q-R). 229

Purified reconstituted MICS1 transports Ca 2+ 230
To assess the mechanism and selectivity of MICS1-dependent in cation transport we 231 produced purified MICS1 for reconstitution studies. Codon optimized hMICS1 cDNA ( Figure  232 6 ̶ figure supplement 1A) was cloned in pH6EX3 (Galluccio et al., 2013) and the recombinant 233 construct was used to transform E. coli Rosetta cells. During the exponential phase of growth 234 (OD ~ 0.8-1), the temperature was set to 37 °C and 0.4 mM IPTG was added to induce 235 synthesis of the protein. MICS1 was over-expressed in the insoluble fraction of the induced 236 cell lysate after 2 hours of IPTG induction (Figure 6 ̶ figure supplement 1B). The protein was 237 purified by Ni-chelating chromatography and reconstituted in proteoliposomes to assess in 238 vitro Ca 2+ transport activity assays using Calcium Green-5N as described Materials and 239 Methods and illustrated in Figure 6A. The incorporation of MICS1 in proteoliposomes was 240 verified by western blot analysis ( Figure 6B). As shown in Figure 6C-E, reconstituted MICS1 241 mediated Ca 2+ fluxes in a pH-dependent manner, with a maximum at pH 7.0 ( Figure 6D) and 242 inhibition of fluxes at pH 8.0 ( Figure 6E). To further investigate the involvement of H + in the 243 transport cycle, we measured H + flux using the pH sensitive dye pyranine ( Figure 6F). 244 Remarkably, alkalinization of the internal compartment of proteoliposomes detected by the 245 increase in pyranine fluorescence indicated a H + flux towards the external compartment 246 induced by Ca 2+ addition, i.e., concomitant to the inwardly directed Ca 2+ flux ( Figure 6A). 247

Discussion
The role and selectivity of LETM1 as an ion transporter/channel has not been unequivocally 250 To address the relatively low mitochondrial protein yield from mammalian cell cultures, we 261 developed a miniaturized proteomic approach that was validated with the MCU interactome 262 as a model. Among the most promising identified interactors of LETM1, we focused on MICS1. 263 Our study demonstrates that a complex containing LETM1 and MICS1 is involved in K + /H + 264 exchange in vivo, since decreased levels of both LETM1KD and MICS1KO led to a decrease of 265 K + transport. This effect is likely due either to reduced LETM1 levels, also in absence of MICS1,266 or to loss of protein interaction, a possibility that will be further explored in future analyses 267 of MICS1 and LETM1 mutations affecting their physical interaction. Comparison of the roles 268 of MICS1 and LETM1 in mitochondrial Ca 2+ efflux clearly showed that LETM1, unlike MICS1, is 269 not required for CHE activity. In contrast to LETM1, loss of MICS1 abrogated the function of 270 CHE, which was restored by re-expression of MICS1. Independent of any interaction partner 271 or protein complex, reconstituted MICS1 was able to transport Ca 2+ across proteoliposomes 272 in a pH-dependent manner and to drive Ca 2+ -dependent H + transport. Thus, based on the 273 consistency between cellular and cell-free activity of MICS1 in Na + -independent 274 mitochondrial Ca 2+ translocation, we have identified MICS1 as the long-sought mitochondrial 275 CHE. Interestingly, MICS1 does not belong to any mitochondrial carrier family. The The lack of a functional CHE also has severe implications on the permeability transition when 288 Na + dependent Ca 2+ efflux is concomitantly blocked, as revealed by thapsigargin-induced 289 hypersensitization of PTP opening. One reason that may explain why MICS1KO mitochondria 290 are so sensitive to the PTP opening is reduced levels of Sirt3, which is responsible for 291 deacetylation of CypD, a key PTP sensitizer (Sambri et al., 2020). The result is in accordance 292 with the modulatory effect of thapsigargin on shifting the ratio between bound and free Ca 2+ 293 towards free Ca 2+ (Korge and Weiss, 1999). The hypersensitivity of Ca 2+ -induced PTP opening 294 also correlates with the observed cristae reorganization, OPA1 cleavage pattern and OMA1 295 activation, which could explain increased predisposition to cell death in MICS1KO cells 296 exposed to thapsigargin. 297 In conclusion, the use of cell free and cell culture models has allowed us to demonstrate that 298 MICS1 is the mitochondrial CHE. In view of the established involvement of LETM1 in both KHE 299 and CHE activity, the identification of the LETM1 partner MICS1 is also a major step forward 300 in resolving current controversies on their relative role in mitochondrial Ca 2+

307
Reagents 308 All reagents used in this study were from Sigma Aldrich, unless otherwise indicated. strokes at 1600 rpm with a Yellowline OST basic homogenizer and mitochondria isolated by 377 differential centrifugation according to (Frezza et al., 2007). Protein complexes were reduced, alkylated and digested with trypsin as described 420 (Rudashevskaya et al., 2013). Peptides were desalted and concentrated by reversed-phase 421 tips (Rappsilber et al., 2007) and reconstituted in formic acid (5%) for LC-MS analysis. 422

Reversed-phase LC-MS data analysis and data filtering 423
All liquid chromatography mass spectrometry experiments were performed on an Aglient 424 1200 HPLC nanoflow system coupled to a linear trap quadrupole (LTQ) Orbitrap Velos mass 425 spectrometer (ThermoFisher Scientific). Raw data were matched to peptides and proteins 426 using Mascot and Phenyx, with a false discovery rate of 1% at the protein level. CRAPome and 427 SAINT analysis were applied to all AP-MS data. GFP pulldowns were used as controls together with publically-available CRAPome data that used similar sample preparation and MS 429 methods and instrumentation. Common contaminants and proteins with a frequency greater 430 than or equal to 0.1 in the CRAPome database were excluded. Proteins with a SAINT score 431 greater than 0.97 were identified as high confidence interactors. Over-expression, purification and reconstitution in proteoliposomes of MICS1 for Ca 2+ 509 transport assays 510

Expression of MICS1 protein 511
To produce the 6His-MICS1 recombinant protein, E. coli Rosetta cells (Novagen) were 512 transformed with the pH6EX3-hMICS1 construct. Selection of transformed colonies was 513 performed on LB-agar plates added with ampicillin (100 µg/mL) and chloramphenicol (34 514 µg/mL). A colony was inoculated and cultured overnight at 37 °C under rotary shaking (160 515 rpm). The day after, the culture was diluted 1:20 in fresh medium added with the specific antibiotics. When the optical density measured at OD600 nm wavelength was 0.8-1, different 517 IPTG concentrations (from 0.1 to 1 mM) were tested to induce protein expression except for 518 one aliquot, grown in absence of inducer (negative control). The cultures were continued for 519 up to 6 hours at 28 °C or 37 °C at 160 rpm. Every two hours, aliquots were collected and 520 centrifuged at 3000 ×g, and at 4 °C for 10 minutes; the pellets were stored at -20 °C. used to wash the column removing unbound proteins. In order to increase the purity of the 538 recovered MICS1, another washing step was performed using 3 mL of the same above-described buffer added with 10 mM imidazole. Finally, MICS1 was eluted in 5 fractions of 1 540 mL, using the same above-described buffer added with 50 mM imidazole. The purified protein 541 was eluted in a peak of 2.5 mL. The eluted protein was subjected to a buffer change for 542 imidazole and Na + removal, using a PD-10 column pre-conditioned with a desalt buffer 543 composed of Tris HCl pH 8.0 (20 mM), glycerol (10%), n-Dodecyl β-D-maltoside (0.1%) and 544 DTE (10 mM): 2.5 mL of the purified protein were loaded onto the PD10 column and collected 545 in 3.5 mL of desalt buffer. 546 Reconstitution in proteoliposomes of the purified hMICS1 547 The desalted hMICS1 was reconstituted by removing detergent from mixed micelles of 548 detergent, protein and phospholipids using the batch wise method previously described for

Cation transport measurements by spectrofluorometric assays 558
The Ca 2+ flux or the intraliposomal pH changes were monitored by measuring the 559 fluorescence emission of Calcium Green-5N or pyranine, respectively included inside the 560 proteoliposomes. After reconstitution, 600 µL of proteoliposomes was passed through a 561 differently indicated. Then, 200 µL proteoliposomes were diluted in 3 mL of the same buffer 563 and incubated for 10 min in the dark prior to measurements. To start the transport assay, 564 CaCl2 (7 mM) buffered at pH 7.0, except where differently indicated, was added to 565 proteoliposomes; the uptake of Ca 2+ or the efflux of H + was measured as an increase of 566 Calcium Green-5N or pyranine fluorescence, respectively. As a control, the same 567 measurements were performed using liposomes, i.e., vesicles without reconstituted hMICS1. 568 The measurements were performed in the fluorescence spectrometer (LS55) from Perkin 569 Elmer under rotatory stirring. The fluorescence was measured following time drive acquisition 570 protocol with λ excitation=506 nm and λ emission=532nm (slit 5/5) for Calcium Green-5N and 571 λ excitation=450 nm and λ emission=520nm (slit 5/5) for pyranine.