Prohibitin controls MRS2-Mediated mitochondrial Mg²⁺ uptake to Shape bioenergetics

SUMMARY

Mitochondrial magnesium (_mMg²⁺) is essential for cellular metabolism and bioenergetics, yet the mechanisms governing its transport remain poorly understood. Although MRS2 constitutes the pore of the mMg²⁺ channel, the molecular machinery regulating its function is unknown. Here, unbiased proteomics identified the prohibitin (PHB) complex as a prominent MRS2-interacting partner. Integrated biochemical and functional analyses demonstrate that the conserved coiled-coil domain mediates MRS2 homo-oligomerization, whereas the C-terminal region of MRS2 interacts with PHB1 to promote channel activity. Quantitative calibration of mitochondria-targeted MagFRET sensors revealed maximal _mMg²⁺ uptake (~15 mM), which was markedly reduced in Phb1-deficient hepatocytes. Complementary loss- and gain-of-function studies establish PHB1 as a positive regulator of MRS2-mediated _mMg²⁺ uptake without affecting MCU-dependent Ca²⁺ transport. In vivo, hepatic Phb1 deletion attenuated _mMg²⁺ uptake and enhanced cellular bioenergetics. These findings identify PHB1 as an activator of the MRS2, advancing our understanding of _mMg²⁺ uptake machinery and its role in metabolic regulation.

In brief

Mitochondrial Mg²⁺ transport is essential for cellular metabolism but remains poorly understood. Maity, Venkatesan et al. identify PHB1 as a positive regulator of the MRS2 Mg²⁺ channel and define the MRS2 complex architecture. Quantitative mitochondrial Mg²⁺ biosensors reveal high-capacity Mg²⁺ uptake, linking mitochondrial Mg²⁺ homeostasis to bioenergetic control.

Graphical Abstract

INTRODUCTION

Magnesium (Mg²⁺) is an essential divalent cation required for ATP hydrolysis, ribosomal function, and cellular signaling. Intracellular Mg²⁺ (_iMg²⁺) homeostasis is tightly regulated, and its disruption has been implicated in hypertension, obesity, muscular dysfunction, immunodeficiency, cardiovascular disease, Parkinson’s disease, and cancer.^1–3 While most _iMg²⁺ is complexed with ATP, ribosomes, and polyphosphates, free Mg²⁺ is stored in the endoplasmic reticulum (ER).^4–6 Upon lactate stimulation, Mg²⁺ is released from the ER and mobilized into the mitochondria, modulating bioenergetics, signaling pathways, and metabolic adaptation.^5,7 Within mitochondria, Mg²⁺ primarily functions as a critical metabolic cofactor rather than a classical second messenger. Mg²⁺ directly regulates key mitochondrial enzymes, including pyruvate dehydrogenase (PDH), and thereby influences oxidative metabolism and ATP production. Consistent with this, perturbations in mitochondrial Mg²⁺ (_mMg²⁺) homeostasis have been shown to alter oxygen consumption rate (OCR), lipid metabolism, and HIF-1α signaling, supporting the central role of Mg²⁺ in metabolic regulation.^5,7 Unlike calcium (Ca²⁺), where rapid mitochondrial fluxes act as well-established signaling events, _mMg²⁺ has not been established as a canonical second messenger with defined oscillatory dynamics.

Although Mg²⁺ and Ca²⁺ are both essential divalent cations that influence mitochondrial metabolism, their transport mechanism and intracellular regulation are mechanistically distinct, particularly within the mitochondria. Mg²⁺, as a cofactor for ATP, is critical in energy metabolism, while Ca²⁺ regulates enzyme activity, signaling pathways, and mitochondrial function.^8,9 The steep negative mitochondrial membrane potential (ΔΨ_m ~−180 mV) drives the inward movement of both ions.^5,10–12 Despite this shared dependence on ΔΨ_m, their uptake machinery is distinct. Mitochondrial Ca²⁺ (_mCa²⁺) uptake is mediated by the mitochondrial calcium uniporter (MCU), which is tightly regulated by MICU1, MICU2, EMRE, MCUR1, and MCUb.^13–18 MCUR1, an integral membrane protein, serves as a scaffold factor that modulates MCU-mediated _mCa²⁺ uptake.^19,20 In contrast, mitochondrial RNA splicing 2 (MRS2) has been identified as a _mMg²⁺ selective channel. However, the molecular machinery regulating MRS2 channel activity remains largely unknown.

Interestingly, the pentameric prokaryotic Mg²⁺ channel CorA revealed an unexpected mechanism of Mg²⁺ transport regulation.^21–23 Specifically, the 5-fold symmetry was broken by asymmetric conformational changes in the cytoplasmic domain, with one subunit displaced away from the central axis while another pivoting toward it. The asymmetric opening was demonstrated both in detergent and in nanodiscs. These open, active structures revealed the atomic basis for Mg²⁺ conductance, promoted by an ensemble of asymmetric open states that allowed ion uptake, contrasting the closed, Mg²⁺-bound inactive state.²¹ Remarkably, MRS2, a _mMg²⁺ channel, also forms a pentameric structure such as CorA, though it has notable differences in protein sequence. While the overall structural fold between the two channels shows limited sequence homology, both share key features, including a Gly-Met-Asn (GMN) motif and two transmembrane helices that contribute to ion transport. Studies using AlphaFold modeling and cryo-electron microscopy (Cryo-EM) suggest that MRS2 adopts a pentameric configuration similar to CorA, despite differences in sequence homology.^24–26 The structural mechanisms controlling the opening and closing of the MRS2 channel remain under investigation, with recent studies offering conflicting findings on its gating dynamics.^25–28 Conversely, the _mCa²⁺ selective MCU channel forms a tetrameric architecture with symmetric closed and open configurations when interacting with gatekeeper proteins (MICUs/EMRE).^29–35 Nevertheless, the opening and closing of the MCU and MRS2 are also modulated by the ΔΨ_m and the cytosolic elevation of these ions. Although recent experimental structural elucidation revealed the pentameric nature of MRS2, whether it forms a heteromeric complex, similar to MCU, and the mechanism of channel activity are yet to be assessed.

In this study, we aimed to identify MRS2-interacting proteins that are essential for its channel assembly and functional regulation. Using an unbiased proteomic approach, we identified prohibitin 1 (PHB1), PHB2, and Stomatin-like protein 2 (SLP2) as key MRS2-associated proteins. PHB1 is an evolutionarily conserved and ubiquitously expressed protein involved in various cellular processes, including plasma membrane signaling, mitochondrial chaperoning, and nuclear transcriptional regulation. Within the inner mitochondrial membrane (IMM), PHB1, PHB2, and SLP2 form a scaffold complex that stabilizes respiratory complexes, maintains cristae structure, and preserves mitochondrial integrity.^36–38 Our biophysical and mutagenesis studies reveal that PHB1 directly interacts with the C-terminal domain of MRS2, forming a hetero-oligomeric complex. Systematic characterization using mitochondria-targeted MagFRET (MagFRET_mito) sensors demonstrated maximal matrix Mg²⁺ accumulation (of ~15 mM) mediated by MRS2, which was markedly blunted in Phb1-deficient hepatocytes. Furthermore, functional analyses demonstrate that PHB1 acts as a positive regulator of the MRS2 channel, modulating _mMg²⁺ uptake and bioenergetics.

RESULTS

Identification of prohibitin proteins as members of the MRS2 mitochondrial Mg²⁺ uptake complex

Despite the established role of MRS2 as the principal _mMg²⁺ uptake channel in the IMM, the molecular architecture, and regulatory components of the MRS2 channel complex remain elusive. Ion channel activity is precisely regulated through specific molecular interactions with auxiliary scaffold and regulatory proteins that modulate channel activity, stability, and assembly.^29,39–45 Therefore, defining the MRS2 interactome is essential to understanding the mechanisms governing _mMg²⁺ homeostasis. To identify the binding partners of MRS2 and characterize the MRS2 channel holo-complex, we employed an unbiased proteomic approach using liquid chromatography with tandem mass spectrometry (LC-MS/MS) (Figure 1A). A FLAG-tagged human full-length MRS2 protein, which is 44 kDa (mature form) in size, was ectopically expressed in HEK293T cells, and total cell lysates were immunoprecipitated with an anti-FLAG antibody. The immunoprecipitation was confirmed by Coomassie staining (Figure 1B) and western blot analysis (Figure 1C). The immunopurified samples were subjected to LC-MS/MS analysis, which identified several mitochondrial proteins as putative binding partners of MRS2 (Figures 1D; Table S1). Among the most enriched interactors were PHB1, PHB2, and Stomatin-like protein 2 (SLP2), key IMM scaffold proteins involved in the organization of mitochondrial protein complexes.⁴⁶

Figure 1. Unbiased proteomics identifies prohibitin as a binding partner of MRS2.

(A) Schematic representation of the experimental workflow for identifying MRS2 binding partners. FLAG-tagged human full-length MRS2 protein was ectopically expressed in HEK293T cells, followed by immunoprecipitation (IP) using an anti-FLAG antibody. The immunoprecipitated samples were confirmed by SDS-PAGE and Coomassie staining prior to LC-MS/MS analysis.

(B) Coomassie brilliant blue (CBB) staining of a gel shows total lysates and FLAG IP eluates. Protein was ectopically expressed in HEK293T cells. Lane 1: Molecular weight markers (MW), lane 2: Total lysate from MRS2-FLAG-expressing cells, lanes 3 and 4: IP eluate from empty vector-transfected cells (UT). Lanes 5 and 6: IP eluate from MRS2-FLAG-expressing cells, lane 7: empty, lane 8: total lysate from MRS2-FLAG-expressing cells, lane 9: MW.

(C) Western blot analysis of immunoprecipitated MRS2-FLAG probed with anti-FLAG antibody. Lane −1 input, lane −2 empty, and lanes 3 and 4: MRS2-FLAG eluates.

(D) LC-MS/MS profile of MRS2-FLAG immunoprecipitants. The table shows the mitochondrial proteins detected in the MRS2-FLAG eluates. PHB1, PHB2, and SLP-2 are the top candidates based on abundance.

(E) Western blot analysis of immunoprecipitated MRS2-FLAG and interacting proteins. Input and FLAG IP eluates were probed with antibodies against FLAG, PHB1, SLP-2, and VDAC2. Top image: FLAG IP confirmation; second image: PHB1 interaction; third image: SLP-2 interaction; bottom image: lack of interaction with VDAC2.

(F) Fast protein liquid chromatography (FPLC) profiling of MRS2-FLAG complexes. HEK293T detergent-solubilized cell lysates expressing MRS2-FLAG were subjected to SEC. Fractions were collected and analyzed by western blot using anti-FLAG and anti-PHB1 antibodies. The elution profile indicates the oligomerization state of MRS2 and its interaction with PHB1. The single asterisk indicates precursor MRS2 with a mitochondrial signal peptide, and double asterisks indicate mature MRS2 after cleavage of the signal peptide.

(G) Quantitative densitometric analysis of MRS2-FLAG and endogenous PHB1 intensity in SEC fractions. The graph shows the band intensity of MRS2-FLAG and PHB1 across different fractions, highlighting their co-elution and complex formation.

To validate the interaction between MRS2 and PHB1/2 or SLP2, independent immunopurified materials from HEK293T cells expressing MRS2-FLAG protein were probed with PHB1, SLP2, and Voltage-Dependent Anion Channel 2 (VDAC2) antibodies (Figure 1E). The western blot analysis validated the interaction of MRS2 with PHB1 (Figure 1E, second image) and SLP2 (Figure 1E, third image). However, VDAC2 was not detected in the MRS2-FLAG immunoprecipitants (Figure 1E, bottom image), suggesting that the interaction between VDAC2 and MRS2 in LC-MS/MS samples is likely nonspecific. The binding of MRS2 to ATP synthase subunits is likely indirect, as the PHB complex is integral to the assembly and stability of mitochondrial respiratory complexes and ATP synthase.^47,48 Additionally, although protein phosphatase 1B is a cytosolic Mg²⁺-dependent protein, its potential indirect binding to MRS2 suggests a likely interaction, considering the use of total cell lysate for identifying MRS2 binding partners and MRS2’s role in Mg²⁺ transport.⁴⁹ In contrast, heat shock protein 90 (HSP90) and Heat Shock Protein 60 (HSP60) were not considered as proteins of interest because these are ubiquitous molecular chaperones that play a pivotal role in the regulation of folding and aggregation of newly synthesized proteins rather than specific complex formation.^50,51

PHB1 forms ring-like structures comprising multiple PHB1 and PHB2 subunits within the IMM.^52,53 SLP2 is also an IMM-localized scaffold protein that stabilizes several mitochondrial proteins, including the PHB complex, via direct interaction with PHB2.⁵⁴ Ion channels usually exist as hetero-oligomeric complexes with multiple associated proteins.^14,29,41,55 After confirming the interaction of PHB1 and MRS2, we utilized fast protein liquid chromatography (FPLC) to examine whether PHB1 is a component of the MRS2 channel complex. First, to estimate the molecular weight of the MRS2 heterooligomeric complex formed by MRS2 and its associated proteins, total cell lysates were analyzed with size exclusion chromatography (SEC). HEK293T cells were transfected with MRS2-FLAG alone or co-transfected with MRS2-FLAG and PHB1-HA plasmids. Western blot analysis of the chromatography fractions collected from MRS2-FLAG-expressed lysates revealed the oligomerization of MRS2 with an apparent molecular mass starting from 158 kDa to more than 670 kDa (Figure 1F, top image). When we probed the same fractions with an anti-PHB1 antibody to check the elution pattern of endogenous PHB1, western blot analysis demonstrated that PHB1 co-eluted with MRS2 only in the higher-order oligomeric fractions (Figure 1F, bottom image; Figure 1G), indicating that PHB1 participates in the MRS2 heteromeric complex. For further clarification of the MRS2-PHB1 complex, SEC fractions collected from MRS2-FLAG and PHB1-HA co-expressed HEK293T cell lysates were probed with FLAG or HA antibody (Figures S1A and S1B). Further, to determine whether MRS2 forms heteromeric complexes, we performed size-exclusion fractionation followed by immunoblot analysis of PHB1-HA in wild-type and Mrs2 knockout (KO) cells (Figures S1C and S1D). In wild-type cells, PHB1-HA was distributed broadly across higher-molecular-weight fractions, with peak enrichment observed in fractions corresponding to ~670 kDa complexes. In contrast, deletion of Mrs2 shifted the PHB1-HA distribution toward lower-molecular-weight fractions, with maximal enrichment detected in fractions corresponding to smaller complexes. Quantitative analysis of normalized PHB1-HA intensity across fractions demonstrated a rightward shift in the elution profile in Mrs2 KO cells compared with wild-type controls (Figure S1D), indicating disruption of higher-order PHB1-containing assemblies in the absence of MRS2. These data indicate that PHB1 and MRS2 likely form complexes within mitochondria.

The MRS2 C-terminal region mediates PHB1 interaction and complex formation

To determine whether MRS2 forms a native protein complex with PHB1, we performed reciprocal endogenous co-immunoprecipitation analysis in HEK293T cells. Immunoprecipitation of PHB1 highly enriched endogenous MRS2, as evidenced by a distinct MRS2 band absent in IgG controls (Figure 2A). Input fractions confirmed comparable basal expression of MRS2 and PHB1 across samples. Reciprocally, immunoprecipitation of MRS2 selectively recovered endogenous PHB1, confirming the bidirectional nature of the interaction. (Figure 2B). These data demonstrate that MRS2 and PHB1 reside within a stable endogenous complex under physiological conditions. Having observed a reciprocal endogenous association between MRS2 and PHB1, we next sought to determine whether the interaction is mediated directly or through intermediary mitochondrial proteins. To address this question, recombinant human MRS2 (Full-length human MRS2 lacking the mitochondrial targeting sequence (MTS); residues 58–443) and full-length human PHB1 (residues 1–272) were independently purified and refolded to homogeneity, as confirmed by Coomassie-stained SDS-PAGE (Figure S2A). Using these purified preparations, reciprocal co-immunoprecipitation experiments in a defined cell-free system demonstrated selective interaction between MRS2 and PHB1 and vice versa, whereas IgG controls showed no detectable nonspecific binding (Figures 2C and 2D). These findings demonstrate that MRS2 and PHB1 engage in direct physical interaction, independent of additional mitochondrial factors.

Figure 2. PHB1 directly binds MRS2 through its C-terminal regulatory domain.

(A) Endogenous co-immunoprecipitation (IP) performed using anti-PHB1 antibody in whole-cell lysates followed by immunoblotting (WB) with antibodies against PHB1 and MRS2. Input, IgG control, and IP fractions are shown. Arrows indicate PHB1 and MRS2 bands.

(B) Reciprocal endogenous co-immunoprecipitation performed using anti-MRS2 antibody followed by immunoblotting with antibodies against MRS2 and PHB1. Input, IgG control, and IP fractions are shown. Arrows indicate MRS2 and PHB1 bands.

(C) Co-immunoprecipitation analysis using purified proteins. Recombinant MRS2 and PHB1 proteins were subjected to immunoprecipitation using anti-PHB1 antibody, followed by immunoblotting with antibodies against PHB1 and MRS2. Input, IgG control, and IP fractions are shown. Arrows indicate MRS2 and PHB1 bands.

(D) Purified MRS2 and PHB1 proteins were subjected to reciprocal immunoprecipitation using anti-MRS2 antibody, followed by immunoblotting with antibodies against MRS2 and PHB1. Input, IgG control, and IP fractions are shown. Arrows indicate MRS2 and PHB1 bands.

(E) Schematic diagram of wild-type (WT) and truncated human MRS2-FLAG constructs. Full-length human MRS2 protein consists of a mitochondrial targeting sequence (MTS), a large soluble domain, and two transmembrane domains (TM1 and TM2), connected by the conserved GMN motif loop. Sequential deletions (Δ1-Δ7) were generated by removing 30–40 amino acid segments.

(F) Western blot analysis of WT or truncated MRS2-FLAG constructs co-expressed with PHB1-HA in HEK293T cells. Total lysate (Input) was probed with anti-FLAG antibody (top left image) or anti-HA antibody (bottom left image) to confirm expression of constructs. FLAG immunoprecipitants were probed with anti-FLAG antibody (top right image) or anti-HA antibody (bottom right image) to detect the interaction between truncated MRS2 constructs and PHB1. The absence of the PHB1-HA band in the Δ7 construct (bottom right image, red dashed circle) indicates that the C-terminus of MRS2 is essential for PHB1 interaction.

(G) Solution NMR demonstrates a direct interaction between the MRS2 Δ7 region and human PHB1. The ¹H-¹⁵N-HSQC of recombinantly expressed and isolated ¹⁵N-MRS2_W394–443 shows ~54 backbone ¹H(¹⁵N) cross-peaks, clustered in the central ¹H region of the spectrum (blue cross-peaks). An ¹H-¹⁵N-HSQC acquired using a mixed ¹⁵N-MRS2_W394–443 and PHB1 sample (magenta cross-peaks) shows a loss in peak intensity for >30% of the amides (green arrows), indicating a direct interaction between the proteins. All experiments were acquired with proteins at a final concentration of 1.0 mg/mL in 10% (v/v) D₂O and 90% (v/v) buffer (20 mM Tris, 150 mM NaCl, 1 mM DTT, 100 mM CHAPS, pH 8), 15°C and 600 MHz.

(H) Left: Cartoon representation of a single MRS2 monomer (PDB ID: 8TUL), colored according to the truncation mutants: Δ1 (red), Δ2 (green), Δ3 (yellow), Δ4 (orange), Δ5 (blue), and Δ6 (pink). Δ7 is not shown because residues 401–443 are unresolved in the cryo-EM structure. Right: AlphaFold3-predicted model of the full-length, mature MRS2 pentamer (residues 71–443), colored by pLDDT confidence score, with the structurally unresolved C-terminal region indicated as Δ7.

(I) Schematic representation shows the disruption of the WT MRS2 and PHB1 interaction in the absence of the C-terminal region (Δ7; 410–443) of MRS2. IMS: Intermembrane space; IMM: inner mitochondrial membrane.

We next sought to define the region of MRS2 required for PHB1 association. A series of truncated human MRS2-FLAG constructs (Δ1-Δ7) were generated by sequentially deleting 30–40 amino acid segments of full-length MRS2 without altering transmembrane domains, the essential GMN motif loop, and the MTS. The schematic diagram represents the full-length (WT) MRS2 and its truncated constructs (Δ1 to Δ7), highlighting the specific deleted regions (Figure 2E). To identify the PHB1 interaction site of MRS2, we ectopically expressed these MRS2-FLAG mutants individually or in combination with the PHB1-HA plasmid in HEK239T cells, and the cell lysates were immunoblotted with anti-FLAG (Figure 2F, top left) or anti-HA antibody (Figure 2F, bottom left) to confirm the expression of these constructs. The cell lysates were immunoprecipitated with an anti-FLAG antibody, and the eluents were probed for FLAG (Figure 2F, top right) or HA (Figure 2F, bottom right) to determine whether these MRS2 truncated constructs interact with PHB1-HA. Western blot of immunoprecipitants showed that the HA band is absent only in the eluent from the lysates co-expressing PHB1-HA and MRS2- Δ7-FLAG. These results indicate that the C-terminus (residues 410–443) of MRS2 is the determinant of PHB1 binding (Figure 2F). Although IP exhibited reduced MRS2 recovery, the Δ3-FLAG mutant retained strong PHB1 association, suggesting that the Δ3 MRS2 region may be required to maintain the proper conformational architecture necessary for PHB1 stoichiometry and functional MRS2 channel activity mediated through the Δ7 C-terminal region.

Further, to determine whether the C-terminus of MRS2 directly interacts with PHB1, we expressed and purified the full C-terminal domain of human MRS2 using an E. coli system. The full C-terminal domain construct (i.e., MRS2 residues 394–443) was designed based on the UniProt-assigned topology (UniProt: Q9HD23), where residues 394–443 are predicted to reside in the matrix. We similarly expressed and purified full-length human PHB1 using an E. coli system and applied solution nuclear magnetic resonance (NMR) spectroscopy to assess the interaction. The C-terminal domain of human MRS2 (CTD; residues Trp394–443) was successfully purified and resolved as a discrete band migrating at the expected molecular weight on Coomassie-stained SDS-PAGE, with minimal contaminating species (Figure S2B, top). In parallel, recombinant human 6×His-tagged PHB1 (residues 1–272) was purified with a dominant band at the predicted molecular mass (~30 kDa) across elution fractions (Figure S2B, bottom). The ¹H-¹⁵N-HSQC spectrum of uniformly ¹⁵N-labeled MRS2_W394–443 revealed ~54 ¹H(¹⁵N) amide cross-peaks (excluding the side chain amides), close to the expected 50 amides for the construct (Figure 2G). Most of the MRS2_W394–443 amides clustered in the ~8.0–8.5 ¹H ppm region, indicative of an unstructured and/or dynamic domain conformation. Upon mixing unlabeled PHB1 (¹H-¹⁵N-HSQC invisible) with the ¹⁵N-labeled MRS2_W394–443 (¹H-¹⁵N-HSQC visible), >30% of the amides showed severe broadening, indicative of direct interaction between the proteins (Figure 2G). The amide peak broadening suggests only weak binding that may have been exacerbated by the 10 mM CHAPS, included to maximize the number of observable MRS2_W394–443 amides. Nevertheless, the solution NMR data from isolated PHB1 and MRS2_W394–443 confirm direct interaction between the proteins. The strictly localized, residue-specific line broadening and small but discrete chemical shift perturbations observed upon PHB1 addition suggest a highly specific binding interface rather than non-specific aggregation. Note that our biochemical analysis indicates that residues 394–409 of MRS2 are not necessary for PHB1 binding, as deletion of MRS2 residues 410–443 is sufficient to disrupt the interaction.

To develop a structural model of PHB1-mediated regulation of the _mMg²⁺ channel MRS2, we combined the cryo-EM MRS2 structure and a high-confidence structural model of the pentameric nature of MRS2 generated by AlphaFold3 (Figure 2H). The left image illustrates a monomeric unit of the MRS2 complex with the six predicted helical transmembrane domains (Δ1-Δ6) distinctly colored (Figure 2H).²⁵ However, the C-terminal MRS2 Δ7 (401–443 residues) was unresolved by the structural analysis. Since MRS2 functions as a Mg²⁺ channel, and we previously showed that matrix Mg²⁺ could inhibit MRS2 activity,²⁴ we next examined the influence of Mg²⁺ ions on the stability of the MRS2-PHB1 complex. HEK293T cells were co-transfected with full-length MRS2-FLAG and PHB1-HA, and the cell lysate was immunoprecipitated with an anti-FLAG antibody in the presence of increasing concentrations of Mg²⁺ (0–20 mM). The eluents were probed with anti-HA antibody to determine the complex stability (Figure S2C and S2D). The western blot analysis revealed that no significant change was observed in the abundance of the HA band of immunoprecipitants (Figure S2B), indicating that the MRS2-PHB1 complex formation is Mg²⁺ independent. This schematic emphasizes that the PHB complex is essential for MRS2 channel complex formation (Figure 2I). These observations support a model in which PHB1 associates with the unresolved MRS2 C-terminus to regulate channel complex organization independently of Mg²⁺ binding.

PHB1 regulates MRS2 mediated mitochondrial Mg²⁺ uptake

To investigate the role of PHB1 in MRS2-mediated _mMg²⁺ uptake, we conducted a series of loss- and gain-of-function studies in different cell types using knockdown (KD), KO, and overexpression of PHB1. We first employed siRNA-mediated KD to determine the role of PHB1 in _mMg²⁺ uptake in mouse embryonic fibroblasts (MEFs) using our established confocal microscopy-based _mMg²⁺ uptake assay.⁵ MEFs were loaded with Mag Green-AM, and _mMg²⁺ uptake was monitored after a lactate pulse. Lactate stimulation induces _iMg²⁺ dynamics characterized by rapid efflux from the ER, followed by mitochondrial uptake through the MRS2 channel. Western blot analysis confirmed efficient KD of PHB1 in MEFs transfected with siRNA against Phb1, compared to scrambled (Scr) siRNA (Figures 3A and 3B). Decreased Mag-Green fluorescence intensity was observed at the basal state and after a 5 mM lactate pulse in PHB1 KD MEFs compared to controls (Figure 3C). Quantification of the basal fluorescence and lactate-derived change in fluorescence amplitude revealed significantly lower basal Mg²⁺ levels and reduced _mMg²⁺ uptake in MEFs with suppressed PHB1 abundance (Figures 3D and 3E), endorsing the regulatory role of PHB1 in MRS2-mediated _mMg²⁺ uptake.

Figure 3. Loss of PHB1 abrogates MRS2-mediated _mMg²⁺ uptake.

(A) Western blot analysis shows Phb1 knockdown (KD) in mouse embryonic fibroblasts (MEFs). MEFs were transfected with scrambled (Scr) siRNA or siPHB1 for 72 h. Total cell lysate was probed with the indicated antibodies.

(B) Quantification of Phb1 KD efficiency. Relative expression of PHB1 normalized to β-actin is shown. Data are presented as mean ± SEM (n = 3). ***p < 0.001 compared to Scr siRNA.

(C) Representative trace of Mag-Green fluorescence in MEFs transfected with Scr siRNA or siPHB1. Cells were stimulated with 5 mM lactate to induce _mMg²⁺ uptake as shown. The increase in fluorescence after a lactate pulse indicates Mg²⁺ uptake by mitochondria.

(D and E) Quantification of Mag-Green fluorescence changes in response to lactate stimulation.

(D) The basal _mMg²⁺ and (E) the amplitude of the Mg²⁺ uptake are shown. Data are presented as mean ± SEM (n = 3). ****p < 0.0001 compared to Scr siRNA.

(F) The scheme represents the in vitro Phb1 KO model. Western blot analysis of PHB1 protein levels in hepatocytes isolated from Phb1^fl/fl and Phb1^fl/fl + Ad-Cre mice. β-actin was used as a loading control.

(G) Representative image of mitochondrial MagFRET sensor (Ad-MagFRET-1_mito) in hepatocytes isolated from Phb1^fl/fl.

(H and I) Measurement of _mMg²⁺ uptake using Ad-MagFRET-1_mito in hepatocytes.

(H) Representative traces of relative FRET signal (ΔFRET SE) indicating _mMg²⁺ uptake in response to lactate stimulation in Phb1^fl/fl and Ad-Cre-transduced Phb1^fl/fl hepatocytes.

(I) Quantification of _mMg²⁺ uptake mean ± SEM. ***p < 0.001 compared to Phb1^fl/fl.

(J) Genotyping image depicts confirmation of liver-specific Phb1 KO.

(K) Representative traces of relative FRET signal (ΔFRET SE) indicating _mMg²⁺ uptake in response to lactate stimulation in Phb1^fl/fl and Phb1^ΔHep hepatocytes. The bar graph represents the quantification of _mMg²⁺ uptake. ***p < 0.001 compared to Phb1^fl/fl.

(L) Western analysis of PHB1, MRS2 and MCU protein abundance in liver tissues harvested from Phb1^fl/fl and Phb1^−/− mice (n = 4).

(M) Densitometric analysis of protein from (L). All data shown as mean ± SEM; ****p < 0.0001, ***p < 0.001, and **p < 0.01, n.s. = not significant.

To validate the role of PHB1 in MRS2 channel activity, we developed an in vitro Phb1^−/− model using Phb1^fl/fl mouse hepatocytes. These hepatocytes were transduced with Cre-expressing adenovirus (Ad-Cre) for 72 h and were then examined for PHB1 protein expression and _mMg²⁺ uptake. Western blot analysis showed that PHB1 protein is mostly abolished in Ad-Cre-transduced Phb1^fl/fl hepatocytes compared to Phb1^fl/fl controls (Figure 3F). To monitor _mMg²⁺ uptake, we used our previously described mitochondrial MagFRET-1 (Ad-MagFRET-1_mito) assay.⁵ After confirmation of the mitochondrial localization of the Ad-MagFRET-1_mito construct (Figure 3G), hepatocytes were stimulated with 5 mM lactate to induce the mobilization of _iMg²⁺, followed by mitochondrial uptake. Quantification of the fluorescence resonance energy transfer (FRET) signal revealed a significant decrease in _mMg²⁺ uptake in Phb1 KO hepatocytes (Figures 3H and 3I), confirming that PHB1 regulates MRS2-mediated _mMg²⁺ uptake. To further confirm these studies, _mMg²⁺ uptake was measured in permeabilized hepatocytes. Magnesium Green Penta potassium (Mag-Green PP) salt was loaded in permeabilized hepatocytes, and _mMg²⁺ uptake was monitored after a 1 mM of Mg²⁺ pulse (Figure S3A). The bath Mg²⁺ clearance showed a significant decrease in _mMg²⁺ uptake after the ablation of PHB1 (Figures S3B and S3C), further validating that PHB1 is an MRS2 regulator.

Next, we generated conditional Phb1^−/− mice to ensure that PHB1 is an MRS2 channel regulator. Liver-specific Phb1^−/− (Phb1^ΔHep) mice were generated by crossing Phb1^fl/fl with Alb-Cre transgenic mice. Genotyping confirmed the successful KO of Phb1 by identifying the floxed and Alb-Cre alleles in the same mouse (Figure 3J). _mMg²⁺ uptake was determined using the MagFRET-1_mito-based assay in intact hepatocytes and the Mag-Green dye-based assay in permeabilized hepatocytes. The Phb1^fl/fl hepatocytes exhibited normal _mMg²⁺ uptake, whereas negligible uptake was observed in Phb1 KO hepatocytes (Figures 3K and S3D–S3F), implicating PHB1 as an MRS2 channel regulator. Since the ablation of PHB1 restricted MRS2-mediated _mMg²⁺ uptake, overexpression of PHB1 was employed as a gain-of-function study. HA-tagged PHB1 plasmid was overexpressed in Phb1^fl/fl hepatocytes (Figure S3G). _mMg²⁺ uptake was monitored in the permeabilized hepatocytes with a bolus Mg²⁺ pulse in the presence of Mag-Green dye. As expected, PHB1 overexpression further enhanced significant _mMg²⁺ uptake (Figures S3G–S3I), demonstrating that PHB1 is a positive regulator of the MRS2 channel.

Calibration of mitochondrial Mg²⁺ dynamics reveals PHB1-Dependent regulation of MRS2-Mediated Mg²⁺ uptake

Since PHB1 regulates MRS2-mediated Mg²⁺ uptake, we next asked whether mitochondrial matrix Mg²⁺ sequestration exhibits properties analogous to MCU-dependent Ca²⁺ uptake and high-capacity _mCa²⁺ retention.^{10,13–15,56} To quantitatively define the dynamic range and regulatory architecture of _mMg²⁺ uptake, we employed a panel of MagFRET_mito sensors spanning distinct Mg²⁺ affinities and calibrated their responses under controlled ionic conditions in permeabilized primary murine hepatocytes (Figure 4A). Primary hepatocytes were transfected with the MagFRET_mito sensors and briefly permeabilized with digitonin (40 μg/mL) containing ICM after 72 h of transfection. Permeabilized cells bathed in ICM containing 5 mM succinate were challenged with various concentrations of MgCl₂. Each MagFRET_mito variant exhibited a rapid, concentration-dependent increase in FRET following MgCl₂ addition, with response kinetics and saturation levels aligning with the sensors’ intrinsic dissociation constants (Figure 4B). MagFRET-1_mito (Kd ~0.15 mM) responded rapidly to sub-millimolar MgCl₂ concentrations, reaching saturation near 2 mM. While MagFRET-8_mito (Kd ~0.89 mM) displayed intermediate sensitivity, with half-maximal responses in the low millimolar range and saturation at about 5 mM. In contrast, the lower-affinity sensors MagFRET-5_mito (Kd ~7.4 mM) and MagFRET-6_mito (Kd ~15 mM) required substantially higher MgCl₂ concentrations (≥5 mM) to elicit detectable responses, with saturation occurring near 15 mM (Figure 4C). These findings indicate that mitochondria possess a remarkably high Mg²⁺ sequestration capacity spanning a broad concentration range. To further validate _mMg²⁺ calibration and define the specificity of MRS2-mediated _mMg²⁺ uptake, we examined _mMg²⁺ uptake in permeabilized Mrs2^−/− hepatocytes transfected with MagFRET-8_mito sensor alone or in human MRS2 reconstituted Mrs2^−/− hepatocytes. Primary hepatocytes isolated from Mrs2^−/− mice exhibited no detectable Mg²⁺ uptake following MgCl₂ addition, while re-constitution of human WT MRS2 in the Mrs2^−/− background restored Mg²⁺ uptake (Figure 4D). This validates both the functionality of the MagFRET_mito platform and the specificity of the assay for MRS2-dependent transport.

Figure 4. Calibration of mitochondria-targeted MagFRET sensors reveals MRS2-PHB1-mediated mitochondrial Mg²⁺ sequestration following ER Mg²⁺ release.

(A) A schematic illustrates the experimental workflow and principle of _mMg²⁺ calibration. Murine hepatocytes were isolated and transfected with mitochondrial-targeted MagFRET (MagFRET_mito) sensor variants with distinct dissociation constants (MagFRET-1_mito, Kd ~0.15 mM; MagFRET-8_mito, Kd ~0.89 mM; MagFRET-5_mito, Kd ~7.4 mM; MagFRET-6_mito, Kd ~15 mM), enabling the detection of Mg²⁺ uptake across a wide dynamic range.

(B) Normalized mean traces (top row) and corresponding titration curves (bottom row) of MagFRETmito in permeabilized murine hepatocytes. Following EDTA-mediated _mMg²⁺ depletion, cells were exposed to dose-dependent MgCl₂ concentrations (0.15–20 mM) in the presence of 5 mM succinate. MagFRET-1_mito (Kd ~0.15 mM, orange), MagFRET-8_mito (Kd ~0.89 mM, red), MagFRET-5_mito (Kd ~7.4 mM, blue), and MagFRET-6_mito (Kd ~15 mM, green) exhibited differential sensitivities.

(C) Dose-response relationships were fit using nonlinear regression to define sensor performance. Data are presented as mean ± SEM from n = 3 independent experiments.

(D) Representative mean traces depict MagFRET-8_mito changes in permeabilized murine hepatocytes of Mrs2^−/− and Mrs2^−/− + WT MRS2 reconstituted cells following sequential addition of EDTA (150 μM) and different concentrations of MgCl₂ in the presence of 5 mM succinate. WT MRS2 reconstitution restored robust Mg²⁺-dependent FRET responses, whereas Mrs2^−/− cells showed nominal changes. Quantification of MagFRET-8_mito amplitude shows dose-dependent _mMg²⁺ uptake in Mrs2^−/− + WT MRS2 cells (1–10 mM MgCl₂), confirming MRS2-dependent Mg²⁺ transport activity. Data represent mean ± SEM from n = 3 independent experiments.

(E) Representative mean traces depict MagFRET-8_mito changes in permeabilized murine hepatocytes of Phb1^−/− and Phb1^−/− + WT PHB1 reconstituted cells following sequential addition of EDTA (150 μM) and different concentrations of MgCl₂ in the presence of 5 mM succinate. WT PHB1 reconstitution restored robust Mg²⁺-dependent FRET responses, whereas Phb1^−/− cells showed nominal changes. Quantification of MagFRET-8_mito amplitude shows dose-dependent mMg²⁺ uptake in Phb1^−/− + WT MRS2 cells (1–10 mM MgCl₂), confirming MRS2-dependent Mg²⁺ transport activity. Data represent mean ± SEM from n = 3 independent experiments.

(F) Quantification of _mMg²⁺ uptake measured using MagFRET_mito (ΔMagFRET_mito) in Mrs2^−/− and Phb1^−/− hepatocytes. Statistical significance was determined using one-way ANOVA followed by multiple-comparison testing. n.s., not significant; **p < 0.01 and ****p < 0.0001.

We next assessed the contribution of PHB1 to MRS2-mediated Mg²⁺ uptake in Phb1^−/− permeabilized hepatocytes. Loss of Phb1 markedly suppressed _mMg²⁺ uptake following MgCl₂ addition (Figure 4E). Reconstitution of WT PHB1 restored Mg²⁺ uptake to levels comparable to those observed following MRS2 reconstitution in Mrs2^−/− hepatocytes, supporting a critical role for PHB1 in facilitating MRS2 channel activity (Figure 4E). Notably, compared with Mrs2^−/− hepatocytes, Phb1^−/− hepatocytes exhibited lesser but detectable Mg²⁺ uptake response at higher extra-mitochondrial MgCl₂ concentrations (5–10 mM) (Figure 4F), suggesting the regulatory role of PHB1 in MRS2-mediated _mMg²⁺ uptake. However, the relatively modest dynamic range observed under these conditions likely reflects partial saturation of the MagFRET-8_mito sensor at elevated Mg²⁺ concentrations. Collectively, these data indicate the MagFRET_mito platform as a sensitive and quantitative tool for measuring _mMg²⁺ uptake and establish MRS2 as the dominant route for _mMg²⁺ uptake, with PHB1 functioning as a positive regulator of channel activity.

The coiled-coil domain coordinates MRS2 oligomerization, PHB1 association, and mitochondrial Mg²⁺ channel activity

Most ion channels form a pore through oligomerization of the core protein to permeate ions across the membrane.^{21,30,32,33,35,57–60} Recent studies established that MRS2 forms a pentamer with a central pore for ion permeation.^25–28 We next set out to identify the critical region(s) of MRS2 participating in homo-oligomeric interactions. To achieve this, we generated a full-length MRS2 with V5 tagged (MRS2-V5) plasmid in addition to MRS2-FLAG and its mutant constructs (Figure 2A). HEK293T cells were transfected with full-length MRS2-FLAG or mutant MRS2-FLAG constructs (Δ1-Δ7) individually or co-transfected with MRS2-V5. After 48 h of transfection, total cell lysate was immunoblotted for FLAG (Figure 5A, top left) or V5 (Figure 5A, bottom left) to check the expression of the plasmids. The cell lysates were immunoprecipitated with anti-FLAG antibody and immunoblotted using antibodies against FLAG (Figure 5A, top right) or V5 (Figure 5A, bottom right) to identify the non-interactive mutant constructs. Western blot analysis showed that mature MRS2 mutant constructs Δ5 and Δ6 do not co-IP with full-length MRS2-V5 (Figure 5A, bottom right), indicating that amino acids present in the Δ5 and Δ6 region are involved in binding. However, observed precursor bands in immunoprecipitants indicate that full-length MRS2 precursor protein still interacts with the mutant Δ5 and Δ6. To clarify these interactions, we performed reciprocal immunoprecipitation experiments in a mitochondrial fraction using full-length MRS2, Δ5, and Δ6 constructs. Full-length MRS2-FLAG, Δ5, and Δ6 constructs were transfected individually or with MRS2-V5 in HEK293T cells, and the mitochondrial fraction was immunoprecipitated with antibodies against FLAG (Figure 5B, middle image) or V5 (Figure 5B, right image). FLAG IP results showed that full-length MRS2-V5 precursors still interact with both mutant constructs in the mitochondrial fractions. However, reciprocal IP analysis demonstrated no interaction between full-length MRS2 and Δ5 or Δ6 mutant protein in either precursor or mature forms. Thus, the observed precursor MRS2-V5 band in MRS2-FLAG immunoprecipitants is most likely due to the nonspecific interaction of FLAG magnetic beads with the signal sequence of precursor MRS2-V5. Intriguingly, neither the Δ5 nor the Δ6 region of MRS2 interacts with the full-length MRS2 protein.

Figure 5. Coiled-coil regions are necessary for MRS2 oligomerization and channel assembly.

(A) Immunoprecipitation analysis of full-length and truncated MRS2-FLAG constructs co-expressed with MRS2-V5 in HEK293T cells. Total cell lysate (input) was probed with anti-FLAG antibody (top left image) or anti-V5 antibody (bottom left image) to confirm expression of all constructs. FLAG IP samples were probed with anti-FLAG antibody (top right image) to confirm immunoprecipitation or anti-V5 antibody (bottom right image) to determine the interaction region for homo-oligomerization. The absence of a lower V5 band (mitochondrial targeted, purple dashed circle) in Δ5 and Δ6 constructs indicates that these regions are crucial for MRS2 homo-oligomerization.

(B) Reciprocal immunoprecipitation (IP) analysis in mitochondrial fractions of HEK293T cells co-expressing full-length MRS2 and Δ5 or Δ6 constructs. Total mitochondrial lysate was probed with anti-FLAG or anti-V5 antibodies (left image) to confirm the expression of the constructs. Eluates from FLAG IP (middle image) or V5 IP (right image) were probed with anti-FLAG or anti-V5 antibodies. MRS2-Δ5 and Δ6 are not detected when IP using V5 as the bait (purple dashed circle).

(C) Representative traces of _mMg²⁺ uptake using Ad-MagFRET-1_mito sensor in Mrs2^−/− hepatocytes expressing wild-type (WT) MRS2, ^Δ5MRS2, ^Δ6MRS2, or ^Δ7MRS2 constructs with stimulation of 5 mM lactate.

(D) Quantitative analysis represents the amplitude of the Mg²⁺ uptake (ΔMagFRET-1_mito SE). Data are presented as mean ± SEM (n = 3). ****p < 0.0001 compared to Mrs2^−/− n.s., not significant compared to Mrs2 ^−/−.

(E) Representative traces of Mag-Green fluorescence in primary murine hepatocytes lacking Mrs2 (Mrs2^−/−) and hepatocytes expressing (WT MRS2-RFP), Δ5MRS2-RFP, Δ6MRS2-RFP, or Δ7MRS2-RFP constructs. Cells were stimulated with 5 mM lactate to induce _mMg²⁺ uptake. The decrease in fluorescence indicates Mg²⁺ uptake by mitochondria. The traces show that the expression of WT MRS2 restores Mg²⁺ uptake, while Δ6MRS2 and Δ7MRS2 mutations fail to rescue the Mg²⁺ uptake deficit.

(F) Quantification of _mMg²⁺ uptake amplitude in primary murine hepatocytes from the experimental groups shown in (A). Data are presented as mean ± SEM (n = 3). The WT MRS2-RFP construct significantly restores Mg²⁺ uptake compared to the control, while Δ5MRS2-RFP, Δ6MRS2-RFP and Δ7MRS2-RFP constructs do not show significant uptake, indicating the importance of the deleted regions in the MRS2 function.

(G) Schematic model illustratesthe disruption of homo-oligomerization of MRS2 in the absence of amino acids in the Δ5 and Δ6 regions. IMS: Intermembrane space; IMM: inner mitochondrial membrane.

After confirming the homo-oligomerization of MRS2, we focused on determining the oligomerization order. The mitochondrial fraction of the MRS2-FLAG-expressed HEK293T cells was immunoprecipitated using anti-FLAG antibody in RIPA buffer containing 1% SDS to disrupt the interaction between MRS2 and its binding partners. The western blots of the resulting immunopurified MRS2-FLAG protein indicate that MRS2 forms a pentamer, which is consistent with structural studies (Figure S4A). For further validation, we performed crosslinking experiments in the MRS2-FLAG-expressing intact cells and immunopurified MRS2-FLAG proteins. For intracellular crosslinking experiments, MRS2-FLAG-expressing HEK293T cells were incubated with cell-permeable crosslinkers, paraformaldehyde (PFA), or disuccinimidyl glutarate (DSG), and the mitochondrial fraction was immunoprecipitated with anti-FLAG antibody. The western blots of the immunoprecipitants showed intracellular oligomerization of MRS2 in the presence of both crosslinkers (Figure S4B). Laddering and smearing above 100 kDa could be due to crosslinking of the homo-oligomeric MRS2 with additional interacting proteins, such as PHBs and SLP2. To distinguish the homo-oligomeric nature of MRS2 above dimers, we applied universal crosslinker bis-sulfosuccinimidyl suberate (BS3) to immunopurified MRS2-FLAG protein. BS3 crosslinking data demonstrate a clear MRS2 dimerization at 0.05 mM and formation of the higher-order oligomers, including tri-to pentameric MRS2, from 0.25 to 1.0 mM (Figure S4C). These results confirm the previous observation that MRS2 forms a pentamer with a central pore for ion permeation.^25–28

To determine the apparent molecular weight of the physiological MRS2 complex, we performed SEC analysis for the mitochondrial fraction of the MRS2-FLAG-expressing HEK293T cells. The mitochondrial lysate was prepared in an SDS-free IP lysis buffer optimized for preserving protein-protein interactions to maintain the integrity of the MRS2 supercomplex. Like whole-cell lysate, MRS2-FLAG protein was detected in the early SEC fractions of the mitochondrial lysate (Figures S4D and S4E). We did not see any monomeric MRS2-FLAG band around SEC fraction number 17, which corresponds to 50 kDa (Figure S4D. The MRS2-FLAG bands appeared at 158 kDa, corresponding to fraction number 12 in the absence of PHB1 as homo-oligomeric complexes. To demonstrate the spatial distribution of MRS2 and PHB1 in situ, we utilized stochastic optical reconstruction microscopy (STORM), a super-resolution imaging technique. MRS2-FLAG and PHB1-HA were co-transfected in COS-7 cells and visualized with Alexa Fluor 488 and 647-labeled secondary antibodies, respectively. The STORM images revealed close proximity between MRS2 and PHB1 in numerous IMM regions. The PHB1-HA signal was predominantly oriented in the opposite direction of the MRS2-FLAG signal. Notably, the PHB1-HA signal was predominantly positioned adjacent to, and oriented opposite the MRS2-FLAG signal, consistent with their expected topological organization within the IMM (Figure S4F). Quantitative assessment of the fluorescence intensity profiles along selected mitochondrial segments demonstrated partial overlap between the MRS2 and PHB1 signals, as expected given that the FLAG-tagged C-terminus of MRS2 and PHB1 are exposed to opposite sides of the IMM. Consistent with this, confocal imaging showed high Manders’ overlap coefficients (M1 = 0.873 and M2 = 0.926), whereas STORM analysis yielded lower coefficients (M1 = 0.132 and M2 = 0.342), reflecting the higher spatial resolution of STORM and the distinct topological localization of the two proteins. Importantly, line profile analysis demonstrated that fluorescence peaks corresponding to MRS2 often coincided with those of PHB1, providing further evidence to support their interaction. (Figure S4G). To further determine whether PHB1 forms a heteromeric complex with MRS2, we performed Blue Native PAGE (BN-PAGE) using cells expressing MRS2-V5 or PHB1-HA alone or co-expressing both MRS2-V5 and PHB1-HA (Figure S4H). Immunoprecipitation of MRS2-V5 revealed the formation of a complex migrating predominantly between 480 and 720 kDa. While the co-expression of MRS2 and PHB1 resulted in a discrete high-molecular-weight complex at ~720 kDa, suggesting a heteromeric complex with PHB1. These findings support a model in which PHB1 positively regulates MRS2-mediated _mitoMg²⁺ uptake (Figure S4H).

Because coiled-coil domains mediate protein-protein interactions and functional channel formation,^61,62 we next investigated whether the MRS2-deleted regions are necessary for channel activity. To determine the role of coiled-coil (Δ5), (Δ6) and C-terminal PHB1 binding (Δ7) regions of MRS2, we generated mRFP-tagged MRS2 Δ5, Δ6 and Δ7 plasmids. Primary murine Mrs2^−/− hepatocytes were overexpressed with Δ5-MRS2-mRFP, Δ6-MRS2-mRFP, Δ7-MRS2-mRFP or WT-MRS2, followed by transduction with Ad-MagFRET-1_mito constructs. After 48 h post-transduction with the FRET construct, we measured the Ad-MagFRET-1_mito signals after stimulation with 5 mM lactate. The mitochondrial Ad-MagFRET-1_mito signal was almost absent upon lactate stimulation in Mrs2^−/− hepatocytes. As expected, the _mMg²⁺ uptake persists with the reconstitution of WT-MRS2 (Figures 5C and 5D). However, no significant FRET change was observed in Δ6-MRS2 or Δ7-MRS2 expressed in Mrs2^−/− hepatocytes, indicating that these regions are essential for MRS2 activity. To validate these, Δ5-MRS2-RFP, Δ6-MRS2-RFP, Δ7-MRS2-RFP, or WT-MRS2 overexpressed in WT hepatocytes were loaded with Mag-green-AM, and _mMg²⁺ uptake was monitored after a 5 mM lactate pulse. Overexpression of WT-MRS2 did not increase further _mMg²⁺ uptake but rather increased _mMg²⁺ uptake rates. In contrast, Δ5-MRS2-RFP, Δ6-MRS2-RRP, or Δ7-MRS2-RFP overexpressed hepatocytes exhibit reduced _mMg²⁺ uptake compared to WT hepatocytes with the expression of the control vector (Figures 5E and 5F), confirming that coiled-coil and PHB1 interacting domains regulate MRS2 activity. Altogether, these results indicate that the coiled-coil domain is essential for MRS2 channel formation, and PHB1 regulates MRS2 channel activity as a channel component. We investigated the potential coiled-coil motif in MRS2. The structure-predicting InterPRO and HMMER tools indicate that MRS2 has one coiled-coil domain comprising amino acids from 289 to 309, which is present in the Δ6 region of MRS2. Collectively, these data indicate that the MRS2 coiled-coil domain and C-terminal region are crucial for oligomerization and channel formation (Figure 5G).

PHB1 regulates MRS2-mediated _mMg²⁺ but not MCU-mediated _mCa²⁺ uptake

Although MRS2 and MCU mediate Mg²⁺ and Ca²⁺ uptake across the IMM, respectively, our previous study using the Mrs2^−/− mouse model revealed that MCU-mediated Ca²⁺ uptake occurs independently of MRS2 activity.⁷ However, PHB1 is recognized as a mitochondrial chaperone and scaffold protein involved in the regulation of mitochondrial biogenesis, metabolism, and bioenergetics.⁶³ Given the relationship between mitochondrial bioenergetics and _mCa²⁺ dynamics,⁶⁴ we next determined whether PHB1 deletion impacts MCU activity. To address this, mitochondrial fractions isolated from HEK293T were subjected to immunoprecipitation using antibodies specific for MCU and PHB1, separately (Figure 6A). Reciprocal immunoprecipitation analyses revealed no detectable interaction between MCU and PHB1, suggesting that PHB1 selectively associates with the MRS2 channel complex. Consistent with this observation, MCU protein abundance was not altered in Phb1 KO hepatocytes (Figure 3L and M). We next examined _mCa²⁺ uptake in AAV8-TBG-iCre-induced Phb1 KO hepatocytes. Phb1^−/− hepatocytes were isolated and assessed for _mCa²⁺ uptake. Western blot analysis revealed a significant reduction in Phb1 protein abundance (Figure S5A). FCCP-induced _mCa²⁺ release in permeabilized hepatocytes revealed no significant differences in basal _mCa²⁺ between control and Phb1 KO hepatocytes (Figures 6B and 6C). To assess MCU-dependent _mCa²⁺ uptake and retention capacity, permeabilized hepatocytes were challenged with sequential Ca²⁺ pulses (20 μM) followed by FCCP addition. No significant differences in _mCa²⁺ kinetics or retention capacity were observed between control and Phb1^−/− hepatocytes (Figures 6D–6F). In contrast, basal [Mg²⁺] and _mMg²⁺ uptake were significantly reduced with ablation of PHB1 (Figures S5B–S5F). These results demonstrate that PHB1 selectively regulates MRS2-mediated _mMg²⁺ uptake without affecting MCU-mediated Ca²⁺ transport.

Figure 6. MRS2-PHB1 heteromeric complex is specific and without any effect on MCU activity.

(A) Mitochondrial lysate from HEK293T cells was immunoprecipitated with PHB1 or mitochondrial calcium uniporter (MCU). Total lysate (input) and immunoprecipitants were probed with anti-MCU and anti-PHB1 antibodies.

(B) Representative traces of _mCa²⁺ release in hepatocytes isolated from Phb1^fl/fl and Phb1^fl/fl + AAV8-TBG-iCre mice.

(C) Quantification of the basal _mCa²⁺ concentration ([Ca²⁺]_out) in hepatocytes from Phb1^fl/fl and Phb1^fl/fl + AAV8-TBG-iCre mice. Data are presented as mean ± SEM (n = 3). n.s. = not significant.

(D and E) Number of Ca²⁺ pulses cleared by mitochondria in hepatocytes from Phb1^fl/fl and Phb1^fl/fl + AAV8-TBG-iCre mice. Data are presented as mean ± SEM (n = 3). n.s. = not significant.

(F) Rate of _mCa²⁺ uptake (1/τ) in hepatocytes from Phb1^fl/fl and Phb1^fl/fl + AAV8-TBG-iCre mice. Data are presented as mean ± SEM (n = 3). n.s. = not significant.

PHB1 regulates MRS2-dependent mitochondrial respiration independent of mitochondrial morphology

Having demonstrated that PHB1 interacts with MRS2 and regulates its channel function, we next determined whether PHB1 deletion recapitulates the mitochondrial phenotypes associated with MRS2 deficiency. The mitochondrial morphology was analyzed in hepatocytes isolated from Phb1^fl/fl and liver-specific Phb1^−/− (Phb1^ΔHep) mice stained with TMRE. Representative confocal images revealed significant changes in mitochondrial morphology in Phb1^−/− hepatocytes, including reduced mitochondrial area, length, and perimeter compared with control hepatocytes (Figures 7A–7D).^65,66 Notably, these morphological changes contrasted with those previously following MRS2 deficiency, where loss of MRS2 promoted mitochondrial elongation.⁵

Figure 7. Loss of PHB1 alters mitochondrial phenotype and bioenergetics.

(A) Representative confocal microscopy images of mitochondria in hepatocytes from Phb1^fl/fl and Phb1 ^fl/fl mice stained with TMRE. White box indicates the region of zoom in the right image. Quantification of TMRE in Phb1^fl/fl and Phb1^ΔHep hepatocytes. Mean ± SEM; n = 3–5 cells per condition, 25 mitochondria measured per cell. n = 3 independent experiments. n.s. = not significant.

(B–D) Quantification of mitochondrial morphology parameters including (B) length, (C) area and (D) perimeter in hepatocytes from Phb1^fl/fl and Phb1 ^fl/fl mice. Data are presented as mean ± SEM (n = 3). ***p < 0.001 compared to Phb1^fl/fl.

(E) Mitochondrial oxygen consumption rate (OCR) in hepatocytes from Phb1^fl/fl and liver-specific Phb1^−/− mice. The OCR was measured under basal conditions and following the sequential addition of oligomycin, FCCP, and antimycin A/rotenone (Rot+Anti-A). Data are presented as mean ± SEM (n = 3). ***p < 0.001 compared to Phb1^fl/fl.

(F) Mitochondrial OCR in Phb1^fl/fl hepatocytes with or without PHB1-HA overexpression. The graphs represent basal, maximal respiration, and proton leak. Data are presented as mean ± SEM (n = 3). ***p < 0.001 compared to Phb1^fl/fl.

(G) OCR measurements in WT, WT + PHB1 overexpression (PHB1 OE), Mrs2 KO, and Mrs2 KO + PHB1 OE cells assessed by Seahorse flux analysis and normalized to total protein content. Sequential injections of oligomycin, FCCP, and Anti-A+Rot are indicated.

(H) Basal respiration, maximal respiration, and proton leak were quantified from OCR traces. Data represent individual wells from independent experiments in each group. All data are shown as mean ± SEM; ****p < 0.0001, ***p < 0.001, **p < 0.01, and *p < 0.05, n.s. = not significant. Table S1. Mass spectrometry data for MRS2-FLAG immunoprecipitants.

Next, we determined whether PHB1 deletion recapitulates the mitochondrial respiratory phenotype associated with MRS2 deficiency. Mitochondrial OCR was measured in Phb^fl/fl and Phb1^−/− hepatocytes. Remarkably, Phb1^−/− hepatocytes exhibited significantly increased basal and maximal OCR compared to Phb^fl/fl hepatocytes (Figure 7E), resembling the elevated respiratory phenotype previously observed in Mrs2^−/− hepatocytes.⁵ Because PHB1 enhances MRS2-mediated _mMg²⁺ uptake, we next examined whether PHB1 overexpression reciprocally alters mitochondrial respiration. Seahorse analysis revealed that PHB1-overexpressing hepatocytes displayed significantly reduced basal and maximal OCR compared with control hepatocytes (Figure 7F), indicating that PHB1 negatively regulates mitochondrial respiration through the MRS2 pathway. To determine whether PHB1-mediated suppression of mitochondrial respiration requires functional MRS2, we next performed Seahorse flux analysis in WT and Mrs2^−/− cells with or without PHB1 overexpression. OCR traces revealed that PHB1 overexpression markedly suppressed mitochondrial respiration in WT cells across all phases of the assay (Figure 7G). Quantitative analysis confirmed significant reductions in basal respiration, FCCP-stimulated maximal respiration, and proton leak compared to WT controls (Figure 7H), indicating a broad attenuation of mitochondrial respiratory capacity. In contrast, MRS2 deficiency altered the respiratory profile, with Mrs2^−/− cells exhibiting reduced basal respiration but relatively preserved maximal respiration and proton leak (Figures 7G and 7H). Notably, PHB1 overexpression in the Mrs2^−/− background did not further impact mitochondrial respiration. OCR traces remained comparable between Mrs2^−/− and Mrs2^−/− + PHB1 OE cells (Figure 7G), and quantification revealed no significant differences in basal respiration, maximal respiration, or proton leak (Figure 7H). These findings demonstrate that PHB1-mediated regulation of mitochondrial respiration requires functional MRS2. Since PHB1 deletion enhanced OCR (Figure 7E), we next assessed the lipid droplet content in both control and AAV-TBG-iCre hepatocytes. A significant reduction of lipid droplets in AAV-TBG-iCre hepatocytes reciprocally correlates with increased OCR in Phb1^−/− hepatocytes (Figure S5G). Consistent with these observations, both OCR and extracellular acidification rate (ECAR) were increased in AAV-TBG-iCre hepatocytes, whereas PHB1 reconstitution suppressed both parameters (Figures S5H and S5I), supporting a role for PHB1 in regulating MRS2-dependent _mMg²⁺ uptake and metabolic remodeling. Because altered mitochondrial respiration is frequently associated with mitochondrial ROS generation,⁶⁷ we next measured mitochondrial superoxide in Phb1-deficient hepatocytes, which exhibited increased MitoSOX fluorescence relative to controls (Figure S5J), further supporting a role for PHB1 in regulating MRS2-mediated _mMg²⁺ uptake and mitochondrial bioenergetics.

DISCUSSION

In this study, we identified PHB1, PHB2, and SLP2 as key mitochondrial proteins interacting with MRS2, functioning as regulatory components that control MRS2-mediated Mg²⁺ uptake in mitochondria. PHB1 protein positively regulates the MRS2 channel complex rather than as a pore-forming subunit. This organizational framework parallels the MCU complex, in which auxiliary scaffold and regulatory proteins are required for proper channel activity despite not directly constituting the ion-conducting pore. Our findings, therefore, suggest that _mMg²⁺ uptake, similar to _mCa²⁺ uptake, is mediated by a multi-component regulatory complex rather than by MRS2 alone. Our findings demonstrate that MRS2 assembles into higher-order pentameric oligomers through coiled-coil domain interactions, with PHB1 playing an essential role by binding to the C-terminus of MRS2 and contributing to the formation of an MRS2 channel complex. The co-elution of PHB1 with higher-order MRS2 oligomers, together with the direct interaction observed using purified proteins, strongly supports the existence of a native MRS2-PHB1 complex within the IMM. Our study defines a remarkably high _mMg²⁺ sequestration capacity, with MRS2-mediated matrix accumulation reaching ~15 mM and requiring PHB1-dependent channel regulation. Loss- and gain-of-function studies further revealed that MRS2-PHB1-mediated _mMg²⁺ uptake controls mitochondrial bioenergetics, without affecting _mCa²⁺ dynamics. These findings suggest a degree of specificity in free Mg²⁺ regulation, with PHB1 acting as a key molecular mediator of _mMg²⁺ transport and bioenergetics remodeling. Our findings further reveal that changes in _mMg²⁺ occur on a distinct temporal scale and are tightly coupled with glycolytic metabolite accumulation, particularly lactate, and the associated Mg²⁺-dependent signaling events. Therefore, in this study, we define _mMg²⁺ “dynamics” as regulated alterations in _mMg²⁺ concentration linked to metabolic state and signaling output, rather than rapid ion transients.

Human MRS2 shares significant homology with the yeast _mMg²⁺ channel Mrs2p, and it is evolutionarily related to the bacterial CorA channel.^1,68,69 Notably, CorA is the only member of the MRS2 family with a well-characterized structural description.²¹ Although the basic structural features of MRS2 have been described in prior studies, much less is known about its interacting partners and regulatory mechanisms.^24–28 Given the fact that many mitochondrial functions are mediated by large, multi-subunit protein complexes, we hypothesized that the _mMg²⁺ uptake channel could form a heteromeric complex with MRS2 acting as the core subunit, regulated by auxiliary proteins. Upon confirming that MRS2 serves as an essential _mMg²⁺ uptake machinery, we focused on the identification of interacting partners through unbiased proteomic analysis. Using techniques such as LC-MS/MS and SEC analysis of immunopurified MRS2-FLAG, we identified PHB1, PHB2, and SLP2 as key mitochondrial interactors (Figure 1).

PHB1, originally recognized as a potential tumor suppressor,⁷⁰ and PHB2, which binds to IgM antigen receptors,⁷¹ are known to form a ring-like structure in the IMM.^47,53,72,73 These two proteins show high sequence identity (47.3%) and structural conservation (Figure S6A). While SLP2 shares minimal sequence identity with PHBs (<20%), structural predictions suggest that it might play a similar role in the formation of the mitochondrial complex. It is proposed that PHB1 forms cage-like structures anchored to the IMM. Recent cryo-electron tomography studies support this hypothesis by showing that PHB1/2 form hetero-11-mer cage-like structures, with the soluble domains facing the intermembrane space.⁵² This topology contrasts with MRS2’s structure^25,74 (Figure S6D), indicating that PHB1 and PHB2 might form larger complexes or adopt a different membrane topology (Figure S6 E). Furthermore, it is possible that PHBs assist in the maturation or insertion of MRS2 into the IMM, as they interact with other mitochondrial proteins.⁴⁷ Additionally, SLP2’s interaction with Mfn2⁷⁵ suggests that the PHB-SLP2 complex plays a key role in regulating mitochondrial health by acting as a membrane-bound molecular chaperone.⁴⁸

Dysregulation of PHB has been implicated in aging, metabolic disorders, and various diseases,^76,77 often linked to Mg²⁺ imbalance and mitochondrial dysfunction. Given PHB1’s critical regulatory role in MRS2, we examined its effect on _mMg²⁺ uptake using both loss- and gain-of-function studies (Figures 3 and S5). KD of PHB1 in MEFs via siRNA resulted in reduced _mMg²⁺ uptake (Figures 3A–3E), which supports the notion that PHB1 is involved in regulating MRS2-mediated Mg²⁺ uptake. Further, in vitro (Figures 3F–3I, S3A–S3C) and in vivo (Figures 3J and K, S3D–S3F) KO models corroborated this role. Overexpression of PHB1 enhanced _mMg²⁺ uptake (Figures 3L–3N), further reinforcing its positive regulation of MRS2. The loss of MRS2 caused mitochondrial matrix Ca²⁺ overload by increasing the activity of the MCU.⁷⁸ Notably, PHB1 is essential for mitochondrial integrity and biogenesis, prompting us to investigate whether ΔΨ_m and MCU activity are also dampened in Phb1 KO hepatocytes. Hepatocellular carcinoma has been observed in liver-specific Phb1 KO mice.⁷⁹ However, in our mouse model, PHB1 deletion in the liver did not result in any detectable adverse phenotypes until 24 weeks of age. Here, we observed no significant changes in ΔΨ_m and _mCa²⁺ uptake, implying that PHB1 specifically regulates _mMg²⁺ transport (Figure 6 and S6).

Ion channels, such as MRS2, typically form functional pores through homo-oligomerization of core proteins.^21,33,35 Our SEC analysis and crosslinking experiments confirm that MRS2 forms pentamers for channel formation (Figure 4) through coiled-coil domain interactions (Figure 3). The stoichiometry of this pentameric assembly is crucial for the channel’s functionality, as the oligomerization of core proteins is typically required for the formation of a functional ion-conducting pore. Deletion of the coiled-coil domain or the C-terminus of MRS2 compromised _mMg²⁺ uptake (Figure 4 and S4), highlighting the importance of homo-oligomerization and hetero-oligomerization with PHB1 for function. Mitochondrial bioenergetics are intricately regulated by the dynamics of _mCa²⁺ and _mMg²⁺. Recent studies revealed that inhibition of MRS2-mediated _mMg²⁺ uptake led to elevated OCR and elongated mitochondria in Mrs2 ^−/− hepatocytes.^5,7,78,80 PHBs are also known to influence mitochondrial morphology, with PHB loss causing mitochondrial fragmentation in C. elegans muscle cells, MEFs, and HeLa cells.^72,81,82 In Phb1 KO hepatocytes, we observed fragmented mitochondria (Figures 7A–7D); however, changes in mitochondrial morphology did not significantly affect MCU-mediated _mCa²⁺ uptake (Figure 6D), while increased OCR in Phb1 KO cells (Figure 7E) suggests the direct regulation of MRS2-mediated _mMg²⁺ uptake by PHB1, which is critical for maintaining mitochondrial bioenergetics. The co-localization and clustering of MRS2 and PHB1 within the mitochondrial membrane suggest a highly coordinated role in facilitating Mg²⁺ transport across the IMM. Collectively, PHBs are evolutionarily conserved, ubiquitously expressed proteins primarily localized in mitochondria. The mitochondrial PHB complex, composed of PHB1 and PHB2 subunits, forms a ring-like structure at the IMM, playing a crucial role in mitochondrial biogenesis, function, inflammation, cancer, cell death, and replicative senescence.^{47,48,52,53,83} Although their precise biochemical function remains unclear, studies across species from yeast to mammals have illuminated their importance in mitochondrial biogenesis and metabolism. Our findings contribute to this body of research, highlighting a novel role for the PHB complex in ion channel regulation, specifically in _mMg²⁺ transport via the MRS2 machinery.

In conclusion, this study identifies PHB1 as a positive regulator of MRS2-mediated _mMg²⁺ uptake and mitochondrial bioenergetics. Our _mMg²⁺ calibration studies indicate that mitochondria function not only as energy-sensing buffers but also as major sequestration sites for Ca²⁺ and Mg²⁺, thereby preventing excessive accumulation of cytosolic divalent cations (Figure 4).^5,10,15,84 Mitochondrial divalent cations likely form a positive feedback circuit that coordinates ATP production with fluctuations in cellular energetic demand. Our findings contribute to the growing body of knowledge on _mMg²⁺ homeostasis in mitochondria and provide new insights into how the regulation of ion transport impacts mitochondrial function. The thermodynamic properties of mitochondrial bioenergetics are intricately linked to ion channel activity, where factors such as the proton gradient (ΔpH), ΔΨ_m, and pH play a crucial role in ATP production and maintaining mitochondrial function. Mg²⁺ is a key regulator of these processes by stabilizing ATP and enzymes involved in oxidative phosphorylation. Understanding the molecular mechanisms governing _mMg²⁺ dynamics will provide important insights into disorders associated with Mg²⁺ dysregulation, potentially offering therapeutic avenues for diseases linked to mitochondrial dysfunction.

Limitations of the study

Although this study found the C-terminal region of MRS2 to be essential for PHB1 interaction, the detailed binding mode, its heteromeric complex structure, and the resulting conformation change of MRS2 during closed and open states remain for future studies. The pentameric mitochondrial MRS2 channel mediates Mg²⁺ transport, yet the molecular determinants governing its higher-order assembly and interaction with PHB1 remain incompletely understood. Our biochemical data predict that the conserved coiled-coil region is necessary for the funnel-like structure of MRS2. Nevertheless, additional structural and biophysical studies are required to define the organization, stoichiometry, and assembly dynamics of the PHB1-MRS2 complex. The super-resolution STORM data exhibit the juxtaposition of MRS2 and PHB1 in the mitochondrial inner membrane. Since the deletion of Phb1 did not alter MRS2 protein abundance, systematic studies are needed to comprehensively determine if, and how, PHB1 regulates MRS2-mediated Mg²⁺ uptake. Further investigation is needed to clarify whether the PHB1-MRS2 interaction is dependent on membrane lipids such as cardiolipin for its proper channel activity.

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Muniswamy Madesh (muniswamy@uthscsa.edu).

Materials availability

All unique resources generated in this study are available from the lead contact upon reasonable request with a completed materials transfer agreement.

Data and code availability

• The mass spectrometry dataset generated and analyzed during this study is available in the MassIVE repository as a partner with ProteomeXchange and can be accessed by: MassIVE dataset identifier MSV000102618, ProteomeXchange dataset identifier PXD081706.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Cell lines and cell culture

HEK-293T (ATCC# CRL-3216) and COS-7 (ATCC# CRL-1651) cells were cultured in high-glucose DMEM media (Hyclone, SH30022.01) supplemented with 10% (v/v) FBS, and 1% (v/v) antibiotic-antimycotic solution. All cells were maintained at 37°C with 5% CO₂ in a humidified CO₂ incubator. For further experiments, cells were plated onto gelatin-coated 22-mm round glass coverslips placed in 6-well plates or 10-cm petri dishes as indicated for each experiment. All human cell lines were verified by STR DNA profiling, and PCR assays were performed to detect mycoplasma contamination.

Animals

Phb1^fl/fl mice were originally obtained from Arianne Theiss (University of Colorado Denver).⁸⁵ All mice were housed and maintained in the UT Health San Antonio animal facility under standard husbandry conditions. All animal experiments were approved by the UT Health San Antonio Institutional Animal Care and Use Committee (IACUC; Protocol Nos. 20180075AR and 20180043AB) and performed in accordance with institutional guidelines. Hepatocyte-specific Phb1 knockouts (Phb1^Δhep) were generated by crossing Phb1^fl/fl mice with hepatocyte-specific Cre recombinase transgenic mice (B6.Cg-Tg(Alb-cre)21Mgn/J, obtained from The Jackson Laboratory, USA). Male and female mice aged 8–12 weeks, including WT, Phb1^fl/fl & Mrs2^−/−) mice were used for this study. Additionally, hepatic deletion of Phb1 in adult mice deletion was generated by tail vein injection of the 1 × 10¹¹ viral particles of adenovirus-associated virus 8- iCre (AAV8-iCre) under the control of the TBG promoter twice between a 7-day interval. Mice were allowed to recover for one week following the final injection prior to hepatocyte isolation. Isolated hepatocytes were used to assess knockout efficiency and to measure mitochondrial Ca²⁺ (_mCa²⁺) and Mg²⁺ (_mMg²⁺) uptake, oxygen consumption rate (OCR), extracellular acidification rate (ECAR), lipid accumulation by BODIPY staining, and mitochondrial reactive oxygen species using MitoSOX Red fluorescence.

METHOD DETAILS

Isolation and culture of primary murine hepatocytes

Primary hepatocytes were isolated from 8 to 12-weeks-old mice using a two-step perfusion technique.⁵ Briefly, the mouse liver was sequentially perfused through the portal vein with perfusion medium – I (Ca²⁺ and Mg²⁺-free HBSS; GIBCO #14174, supplemented with 35 mM HEPES and 0.75 mM EGTA) and digestion medium-II (DMEM, GIBCO #12320) containing collagenase D (380 μg/mL, Worthington). After perfusion, the liver was gently dissected and dissociated in isolation media (DMEM; Gibco #12320 supplemented with 10% (v/v) FBS; Hyclone # SH30396.03). The isolated hepatocytes were passed through a 100 μm cell strainer followed by centrifugation (500×g, 4°C, 5 min) to pellet parenchymal hepatocytes. After being washed three times with DMEM containing 10% (v/v) FBS, the pellet was resuspended in Williams E media (Sigma W4128) containing 10% (v/v) FBS, 1% (v/v) antibiotic-antimycotic solution (GIBCO #15240-062), non-essential amino acids (NEAA) and 200 mM L-glutamine. The resuspended cells were plated on in-house collagen-coated 22-mm glass coverslips for confocal microscopy or 96-well plate for OCR measurement or collagen-pre-coated dishes (Corning BioCoat #catalog354400, 354450) for all other experiments. After 4–6 h, when hepatocytes were adhered to the surface, the media was replaced with fresh growth media.

Mouse embryonic fibroblast (MEFs) isolation

Mouse embryonic fibroblasts (MEFs) were isolated from wild-type (WT) pregnant female mice at embryonic day (E8.75–10.5). Uterine horns were individually separated, and each embryo was placed into a Petri dish and rinsed with 70% ethanol. The heads and other red organs were dissected, and the tissue was finely minced using a sterile razor blade. Subsequently, the minced tissue was incubated with 0.25% trypsin for 20 min at 37°C in a shaking water bath. 1 volume of freshly prepared MEFs medium was added to inactivate the trypsin. Finally, after centrifugation (500 ×g, 5 min), the cell pellet was carefully resuspended in pre-warmed MEFs medium.

Plasmid and siRNA transfection

Mouse embryonic fibroblasts (MEFs) cells were transfected with scrambled (Scr) siRNA or PHB1 siRNA (20 pmol) using the RNAiMAX (Invitrogen) reagent according to the manufacturer’s instructions. After 72 h of transfection, knockdown efficiency was validated by western blots, and the mitochondrial magnesium (_mMg²⁺) uptake was measured using Mag-Green assay.⁵ For plasmid-based experiments, cells were transfected with MRS2 or PHB1 or both constructs using Lipofectamine-3000 (Invitrogen) according to the manufacturer’s instructions. After 72 h post-transfection, ectopic expression of the constructs was confirmed with western blot or confocal imaging, as indicated.

Mass spectrometry analysis

HEK-293T cells were transiently transfected with the WT-MRS2-FLAG plasmid. After 48 h, both transfected and untransfected cells were harvested and lysed using IP lysis buffer (ThermoFisher). The total cell lysate was subjected to immunoprecipitation (IP) using Anti-FLAG M2 magnetic beads (Sigma), and the elution was performed using 5% SDS. The eluants were then separated on a 4–12% Bis-Tris polyacrylamide gel (ThermoFisher) and subsequently subjected to western blot analysis and Coomassie staining (ThermoFisher) to confirm the expression and abundance of MRS2-FLAG. The protein-containing region of each gel lane was excised, and within each individual excision, the proteins were reduced in situ using TCEP (tris(2-carboxyethyl)phosphine), followed by alkylation in the dark with iodoacetamide. Subsequently, the samples were treated with trypsin (Promega) to generate digests. The resulting digests were subjected to analysis using capillary HPLC-electrospray ionization tandem mass spectrometry on a Thermo Fisher LTQ Orbitrap Velos Pro mass spectrometer equipped with a New Objective Digital PicoView 550 NanoESI source. On-line HPLC separation was achieved using an RSLC NANO HPLC system (Thermo Scientific/Dionex). The column used was a PicoFrit (New Objective; 75 μm i.d.), which was packed in-house to a length of 15 cm with C18 adsorbent (Vydac; 218MS 5 μm, 300 Å). The mobile phase consisted of two components: mobile phase A, composed of 0.5% acetic acid (HAc) and 0.005% trifluoroacetic acid (TFA) in water, and mobile phase B, composed of 90% acetonitrile, 0.5% HAc, 0.005% TFA, and 9.5% water. The gradient for separation was set at 3%–42% mobile phase B over 120 min, with a flow rate of 0.4 μL/min. In the mass spectrometry analysis, precursor ions were acquired in the orbitrap over a range of m/z 300 to m/z 2000 in centroid mode, achieving a resolution of 60,000 at m/z 400. Data-dependent collision-induced dissociation spectra of the six most intense ions from each precursor scan were obtained simultaneously in the linear trap. Parameters for this process included an MS2 isolation width of 3, rejection of unassigned charge states, normalized collision energy set at 30, dynamic exclusion with a repeat count of 1, and dynamic exclusion repeat duration of 30 s. For spectrum searches, the Mascot software (version 2.6.2; Matrix Science) was utilized. The search involved using a human subset of the UniProt database [Uniprot_Human 20181204 (95,936 sequences; 38,067,061 residues)] in combination with databases containing common contaminants, antibodies used for immunoprecipitation, and recombinant protein constructs, including MRS2-FLAG. The peptide tolerance was set at 20 ppm, and the fragment tolerance was set at 0.8 Da. Cysteine carbamidomethylation was designated as a fixed modification, while methionine oxidation, deamidation of glutamine and asparagine were considered variable modifications. Trypsin was specified as the proteolytic enzyme, and one missed cleavage was allowed. The identification and characterization of the identified proteins were carried out using X! Tandem [CYCLONE (2010.12.01.1)]. Cross-correlation with the Mascot results and determination of protein and peptide identity probabilities were performed using Scaffold software (version 4.9.0; Proteome Software). In Scaffold, the acceptance thresholds for peptide and protein assignments were set at 95% and 99%, respectively, with a minimum requirement of two peptides for protein identification.

Size exclusion chromatographic analysis (SEC)

Gel filtration was performed using fast protein liquid chromatography (AKTA Pure FPLC; GE Healthcare) using a Superdex 200 increase 10/300 GL column equilibrated with DPBS (Gibco 14190). Column calibration was carried out with a gel filtration protein standards kit (Bio-Rad Cat #1511901). The whole cell lysate or mitochondrial fraction was prepared using cell lysis buffer (50 mM Tris-HCl, pH 8, 150 mM NaCl, 0.5% de-oxycholic acid, 1% NP-40, 10% glycerol, Halt protease and phosphatase inhibitor cocktail; ThermoFisher). Lysates were centrifuged at 10,000 × g for 15 min, and the supernatant was loaded onto the column and eluted at a flow rate of 0.5 mL/min 0.5 mL fractions were collected and concentrated using 10 kDa molecular weight cutoff (MWCO) centrifugal column. The concentrated fractions were probed with antibodies as indicated in the figures.

Expression and purification of full-length human MRS2 and PHB1 for co-immunoprecipitation experiments

Full-length human MRS2 (residues 58–443), excluding the mitochondrial targeting sequence, was subcloned out of a BSD vector into pET-28a (Novagen) using the NdeI and XhoI restriction sites, expressed and purified as previously described²⁴ with the following modifications. The 6×His-MRS2_58–443 expression in BL21 (DE3) E. coli cells was induced with 0.4 mM IPTG and allowed to proceed at 37°C for 4 h. Protein was purified under denaturing conditions using HisPur (Thermo-Fisher Scientific) nickel-nitrilotriacetic acid (Ni-NTA) beads as per the manufacturer’s guidelines. Briefly, the lysis and wash buffers contained 6 M guanidine hydrochloride (Gdn-HCl), 20 mM Tris (pH 8.8), 7.5 mM β-mercaptoethanol, while the elution buffer was identical to the wash buffer but supplemented with 325 mM imidazole. Finally, the Gdn-HCl was removed by extensive dialysis in 20 mM Tris (pH 8.0), 150 mM NaCl, 10 mM CHAPS and 1 mM dithiothreitol (DTT) buffer using a 3,500 Da MWCO membrane. For co-IP experiments, full-length human PHB1 (residues 1–272) (NCBI: NP_002625.1), cloned into pET-28a as described above, was expressed in E. coli, and the 6×His-PHB1_1–272 was purified using Ni-NTA beads, precisely as described for the 6×His-MRS2_58–443 protein. Purity and intactness of both 6×His-MRS2_58–443 and 6×His-PHB1_1–272 were assessed post-dialysis/refolding by Coomassie blue-stained SDS-PAGE (Figures S2A and S2B).

Cloning, expression and purification of uniformly ¹⁵N-labeled MRS2_W394–443

The cDNA region encoding human MRS2 residues 394–443 (NCBI: NP_065713.1) was cloned into pET28a using PCR and NheI and XhoI restriction sites. Tryptophan was integrated into the construct upstream of residue 394 by PCR-mediated site-directed mutagenesis, yielding MRS2_W394–443, to facilitate robust UV₂₈₀ absorbance and Coomassie staining. MRS2_W394–443 was expressed in BL21 (DE3) Escherichia coli cells cultured at 37°C in M9 minimal medium with ¹⁵NH₄Cl as the sole nitrogen source. Expression was induced at an optical density (600 nm) of ~0.6 with 0.3 mM isopropyl β-d-1-thiogalactopyranoside (IPTG), and expression was allowed to proceed for 12 h at 24.5°C. Proteins were purified under denaturing conditions using HisPur nickel-nitrilotriacetic acid agarose beads as per the manufacturer’s guidelines (Thermo Scientific). The protein was refolded by dialysis in 20 mM Tris, 150 mM NaCl, 1 mM DTT, pH 8.0, 4°C. The amino (N)-terminal hexahistidine (6×His) tag was cleaved with ~1 Unit of bovine thrombin (Millipore) per 1 mg of protein. Size-exclusion chromatography (SEC) was used as the final purification step using 20 mM Tris, 150 mM NaCl, 1 mM DTT, pH- 8.0.

Cloning, expression and purification of PHB1

The cDNA encoding human PHB1 residues 1–272 (NCBI: NP_002625.1) was cloned into pET-28a using PCR and NdeI and XhoI restriction sites. Overnight protein expression at 37°C from the pET-28a-PHB1 vector was done using BL21 (DE3) E. coli cells cultured in LB, induced with 0.4 mM IPTG. Protein was purified under native conditions using HisPur (Thermo Scientific) nickel-nitrilotriacetic acid beads as per the manufacturer’s instructions. After dialysis in 20 mM Tris (pH 8.0), 150 mM NaCl, and 1 mM DTT, the 6×His tag was cleaved with ~2 U of bovine thrombin (Millipore) per 1 mg of protein. A final SEC step through an S200 10/300 GL column was performed.

Solution NMR assessment of ¹⁵N-MRS2_W394–443:PHB1 interactions

¹H-¹⁵N-heteronuclear single quantum coherence (HSQC) spectra were acquired on Bruker Avance NEO 600 MHz spectrometer equipped with a TXI S3 ²H, ¹H, ¹⁵N, ¹³C room temperature probe. Spectra were acquired at 15°C using 48 scans, 2048 points in the ¹H dimension and 64 increments in the ¹⁵N dimension. ¹H and ¹⁵N sweep widths were set to 8,196 and 1,800 Hz, respectively. After acquiring the ¹H-¹⁵N-HSQC of MRS2_W394–443 in the free state at 1.0 mg/mL, an ¹H-¹⁵N-HSQC of MRS2 _W394–443:PHB1 each at 1.0 mg/mL was acquired. Buffers in all samples were 20 mM Tris (pH 8.0), 150 mM NaCl, 1 mM DTT, 10 mM CHAPS.

Calibration of mitochondria-targeted MagFRET sensors

Primary murine hepatocytes were isolated from (WT, Mrs2^−/− and Phb1^−/− mice) and seeded onto 25 mm collagen-coated glass coverslips. After overnight culture, cells were transiently transfected with 1 μg of organelle-targeted MagFRET plasmids using Lipofectamine 3000 (Thermo-Fisher). MagFRET sensors were fused to the Cox8α-sequence, (MagFRET-1_mito (Kd ~0.15 mM), MagFRET-8_mito (Kd ~0.89 mM), MagFRET-5_mito (Kd ~7.4 mM), and MagFRET-6_mito (Kd ~15 mM)). For _mMg²⁺ uptake, hepatocytes were permeabilized in intracellular medium (ICM - mM, 120 KCl, 10 NaCl, 1 KH₂PO4, 20 HEPES-Tris, pH - 7) containing digitonin (40 μg/mL, 2 min at 37°C), washed, and res uspended in fresh ICM lacking digitonin for live imaging. To measure _mMg²⁺ uptake was measured after the suspension of permeabilized hepatocytes in 5 mM succinate-containing ICM. A bolus of indicated concentrations of MgCl₂ was applied.^4,6 Confocal FRET imaging was performed at 37°C on a Leica TCS SP8 equipped with a 63× oil-immersion objective and an FRET-SE detection module. Sequential line-by-line acquisition was used. After a 30 s baseline, Mg²⁺ was acutely depleted using ICM containing 150 μM EDTA (5 mM succinate for mitochondrial assays), followed by stepwise additions of MgCl₂ to generate in situ calibration curves. Once low steady-state signals were reached, stepwise additions of MgCl₂ were applied to establish calibration curves. FRET ratios were extracted using Leica FRET-SE software optimized for live-cell quantitative FRET analysis. We set the laser/detectors for the cerulean (donor) at 458 nm excitation and the emission signal was collected at 460–490 nm for cross-talk correction. For the acceptor, the citrine was excited at 488 nm, and the emission signal was collected at 510–550 nm for cross-talk correction. For the FRET signal, Cerulean was excited at 458 nm, and Citrine (acceptor) and the FRET signal were collected at 510–550 nm. The following parameters were used for FRET-SE calculation. The FRET channel is denoted as B and its signal is derived from the acceptor channel. Here $\mathrm{\beta }$ refers to a donor cross-talk; $\mathrm{\beta }=\mathrm{B}/\mathrm{A}$, which generates a donor-only reference correction factor. $\gamma$ refers to and corrects for acceptor cross-excitation; $\mathrm{\gamma }=\mathrm{b}/\mathrm{C}$. After these corrections, we calculated apparent FRET efficiency (${\mathrm{E}}_{\mathrm{A}}$) by using the following equation: ${\mathrm{E}}_{\mathrm{A}}=\mathrm{\beta }-\mathrm{A}3\times \mathrm{\beta }–\mathrm{C}\times \mathrm{\gamma }/\mathrm{C}$. FRET data were normalized to basal signals to compare the response between groups.

Chemical cross-linking assay

Intracellular crosslinking experiments were carried out in HEK293T cells using paraformaldehyde (PFA) and disuccinimidyl glutarate (DSG). Briefly, HEK293T cells were transfected with WT-MRS2-FLAG. 48 h post-transfection, cells were trypsinized and washed three times with PBS. Cells were resuspended in PBS before treatment with or without different concentrations of PFA or DSG for 30 min followed by incubation with 50 mM Tris-HCl buffer (pH-8) for 15 min to stop the crosslinking. Mitochondrial fractions were isolated using a mitochondrial isolation kit (Abcam#110170) and lysed with RIPA buffer (Abcam #156034). The crosslinking of MRS2 was assessed by western blot analysis of the mitochondrial lysate. BS3 was used to perform crosslinking of the immunopurified MRS2-FLAG protein. Briefly, the Mitochondrial fraction from HEK293T cells expressing WT-MRS2-FLAG was lysed with RIPA buffer and immunoprecipitated using FLAG-M2 agarose magnetic beads. After three washes, the beads were resuspended in PBS and incubated with or without different cross-linking reagents at various concentrations of BS3 (0.005, 0.05, 0.1, 0.25, 0.5, 1 mM) for 30 min at room temperature. The crosslinking reaction was terminated by adding 50 mM (final concentration) Tris-HCl buffer, pH 8 for 15 min at room temperature. After three washes, immunopurified crosslinking MRS2-FLAG was eluted with 1× LDS Sample Buffer (ThermoFisher #NP0007) containing 2-mercaptoethanol and subjected to western blot analysis.

Immunoprecipitation and immunoblotting

Whole-Cell or mitochondrial lysates were prepared using IP lysis or RIPA buffer supplemented with protease and phosphatase inhibitor cocktail (ThermoFisher # 78445) as indicated. Protein concentrations were determined using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, 23225). Immunoprecipitation was performed using agarose magnetic beads tagged with the specified antibodies mentioned in the figures. Equal amounts of protein or immunoprecipitants or crosslinking material were separated on 4%–12% Bis-Tris polyacrylamide gel (ThermoFisher #WG1402BOX), transferred to a PVDF membrane (ThermoFisher #IB24001), and probed with corresponding antibodies as specified. Following incubation with primary and secondary antibodies, the membranes were developed on X-ray film. Densitometric analysis of the scanned film images was performed using ImageJ software to quantify protein band intensity.

Blue Native PAGE electrophoresis

Blue Native PAGE (BN-PAGE) was performed using the NativePAGE Novex Bis-Tris Gel System (Thermo Fisher Scientific) according to the manufacturer’s instructions with minor modifications. All buffers were prepared fresh and pre-chilled before use. Protein samples were maintained under native conditions throughout sample preparation and electrophoresis. Samples were not heated or boiled and were kept on ice during preparation. Final salt concentrations were maintained below 50 mM to preserve native protein complex integrity. For sample preparation, protein lysates were mixed with 4× NativePAGE sample buffer to achieve a final 1× concentration. For detergent-solubilized membrane protein samples, 5% G-250 sample additive was included immediately before loading. Samples were adjusted to the desired final volume with deionized water and mixed gently without heating. Electrophoresis was performed at 4°C using a two-step voltage protocol consisting of 150 V for 60 min followed by 250 V for 30–90 min until optimal protein complex separation was achieved. Following electrophoresis, gels were either processed for Coomassie staining or transferred for immunoblot analysis.

Measurement of _mMg²⁺ using Ad-MagFRET-1_mito in hepatocytes

The Ad-MagFRET-1_mito was performed as described previously.⁵ Primary murine hepatocytes were plated on 25-mm collagen-coated glass coverslips. After 6 h of plating, cells were transduced with Ad-MagFRET-1_mito adenovirus at a multiplicity of infection (MOI-10) and incubated for 48 h. For live cell imaging, cells were excited using the 405 nm laser line, and the emitted light was collected using the hybrid detector HyD of a Leica SP8 confocal microscope. The measurements were performed at 37°C within a temperature-controlled chamber. The Cerulean channel, spanning 460–490 nm, and the Citrine channel, ranging from 510 to 550 nm, were used to detect the emissions resulting from the fluorescence energy transfer (FRET) process, as established in prior studies.

Measurement of MRS2-mediated Mg²⁺ uptake, MCU-mediated Ca²⁺ uptake, and basal [Mg²⁺]_m or [Ca²⁺]_m

Primary hepatocytes were rinsed with Ca²⁺ and Mg²⁺ free DPBS at pH 7.4. An equivalent number of hepatocytes (approximately 4×10⁶) were resuspended and rendered permeable using 40 μg/mL digitonin in 1.5 mL of intracellular medium (ICM; 120 mM KCl, 10 mM NaCl, 1 mM KH₂PO₄, and 20 mM HEPES-Tris at pH 7.2) supplemented with 5 mM succinate and 1.5 mM K⁺/ATP (for Mg²⁺ uptake) or Mg²⁺/ATP (for Ca²⁺ uptake). The clearance of extramitochondrial Mg²⁺ or Ca²⁺ ([Mg²⁺]_out or [Ca²⁺]_out) was recorded as an indicator of _mMg²⁺ or _mCa²⁺ uptake ([Mg²⁺]_m or [Ca²⁺]_m). To assess basal Mg²⁺ levels, 1 μM Mag-Green PP dye was introduced into the permeabilized cells. The dye was excited at 505 nm, and the emission intensity was recorded at 535 nm. After a 450-s background acquisition, cells were pulsed with 1 mM bolus MgCl₂ to quantify _mMg²⁺ uptake. For basal Ca²⁺ measurements, Fura-2FF dye (1 μM) was used. This dye was subjected to excitation at two wavelengths, 340 nm and 380 nm, with emission measured at 510 nm. The [Ca²⁺]_out was expressed as the excitation ratio (340 nm excitation/380 nm emission) of Fura2-FF/FA fluorescence. To evaluate _mCa²⁺ uptake and retention capacity, permeabilized hepatocytes were subjected to multiple 20 μM CaCl₂ pulses, followed by the addition of the mitochondrial uncoupler, FCCP (10 μM). The fluorescence measurements were performed using a multi-wavelength excitation dual-wavelength emission spectrofluorometer (Delta RAM, PTI, HORIBA) at a constant temperature of 37°C with continuous stirring.

Seahorse mito stress assay

The mitochondrial stress tests were conducted using a Seahorse XF96 Extracellular Flux analyzer (Agilent Technologies) by measuring the oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) of each well. Primary murine hepatocytes (4 × 10⁵ cells/well) were plated onto in-house collagen-coated 96-well Seahorse XF Cell Culture Microplates (Agilent #101085-004). On the following day, the cultured media was substituted with XF Cell Mito Stress Test Kit (Agilent #103015-100) assay media supplemented with glucose, glutamine, and pyruvate according to the cultured media composition and HEPES to maintain pH at 7.4. The cells were then incubated in a non-CO₂ incubator for 1 h at 37°C. The OCR and ECAR were measured using an XF96 extracellular flux analyzer at 37°C after calibration with Seahorse XF Calibrant solution (Seahorse Bioscience, Agilent) in a CO₂-free incubator overnight, following the manufacturer’s protocol. Throughout the measurement, primary hepatocytes were sequentially exposed to oligomycin (2 μM), FCCP (5 μM), and a mixture of antimycin A (1 μM) and rotenone (1 μM) at the designated time points. Data was collected using the Agilent Seahorse Wave 2.6.1 Desktop software and then analyzed using GraphPad Prism version 9 to assess key mitochondrial parameters, including basal respiration, ATP-linked respiration, and maximal respiratory capacity.

Measurement of mitochondrial Mg²⁺ uptake

Primary murine hepatocytes were grown on 25-mm collagen-coated glass coverslips overnight and loaded with 2.5 mM of the magnesium indicator Mag Green-AM (ex/em 488/510 nm) and 1 mM of the mitochondrial indicator Mito-Tracker Deep Red FM (ex/em 644/665 nm) in growth media for 30 min at 37°C (5% CO₂). Cells were washed and imaged within a temperature-controlled environmental chamber (set at 37°C, unless otherwise indicated) using a Leica TCS SP8 live cell imaging system. After 30 s of baseline recording, 5 mM lactate was added. Images were obtained every 3 s using a 63× oil objective and quantified using Leica Application Suite X.

BODIPY staining

Hepatocytes isolated from Phb1^fl/fl and Phb1^fl/fl AAV-TBG iCre mice were grown on 25-mm collagen-coated glass coverslips overnight. Cells were loaded with lipid indicator BODIPY (ex/em 493/503 nm, 1mg/mL) and mitochondrial membrane potential indicator tetramethylrhodamine, ethyl ester (TMRE, ex/em 556/610 nm, 100 nM) in serum-free conditions for 30 min (37°C, 5% CO₂). The stained cells were washed and imaged using a Leica SP8 Confocal microscope coupled with a temperature-controlled environmental chamber. The acquired images were quantified using the Leica Application Suite X software.

Mitochondrial reactive oxygen species measurement

Hepatocytes isolated from Phb1^fl/fl and Phb1^fl/fl AAV-TBG-iCre mice were grown on 25-mm collagen-coated glass coverslips overnight. Cells were loaded with the mitochondrial superoxide sensitive fluorophore MitoSOX red (ex/em 556/610 nm, 5 mM) for 30 min at 37°C. The stained cells were washed and imaged using a Leica SP8 Confocal microscope (Manheim, Germany) equipped with a temperature-controlled environment chamber. The images were acquired and analyzed using the Leica Application Suite X software.

Sample preparation for STORM microscopy

COS-7 cells were cultured on gelatin-coated glass coverslips and co-transfected with 1 μg of MRS2-FLAG and PHB1-HA plasmids. After 48 h, the cells were washed with DPBS and then fixed using 3.0% PFA supplemented with 0.1% glutaraldehyde for 10 min at room temperature without agitation. The samples were reduced with freshly prepared 0.1% NaBH₄ for 7 min at room temperature. Following three DPBS washes, the cells were permeabilized by incubating with 0.2% Triton X-100 for 15 min at room temperature with gentle shaking. After three washes, the cells were incubated with blocking solution (10% normal goat serum in DPBS containing 0.05% Triton X-100) for 90 min at room temperature. Subsequently, the cells were incubated with primary antibodies (1:3000) for 60 min, followed by Alexa Fluor 488 (for MRS2-FLAG) and Alexa Fluor 647 (for PHB1-HA) secondary antibodies (1:1000) for 30 min at room temperature. Cells were washed and stored in an STORM imaging buffer (540 mM glucose, 7.6 μM Glucose Oxidase, 3.1 μM Catalase, 10 mM NaCl, 20 mM Cysteamine) as described previously.⁹³ Long-term hypoxic conditions were maintained with an OKO Labs stage top incubator flooded with a constant flow of nitrogen gas. Images were acquired on an Andor Dragonfly microscope using Borealis total internal reflection fluorescence (TIRF) excitation with a 63× Leica TIRF objective (numerical aperture, 1.47 NA), an astigmatic lens group, and an Andor Zyla PLUS 4.2 sCMOS camera. Alexa Fluor 647 was acquired first using a 638 nm laser, 698/77 band-pass filter, and 20,000 frames, followed by Alexa Fluor 488 using a 488 nm laser, a 521/38 band-pass filter, and 20,000 frames. Data was processed using Picasso software suite⁸⁹ by conducting three-dimensional calibration with tetraspeck beads, identifying and fitting single molecules, undrifting with a redundant cross-correlation algorithm, and rendering a 100 nm slice to remove out of focus localization events for analysis.

In silico protein structure prediction

To predict the structure of the PHB1 oligomer, Alphafold-Multimer (v2.3.2) and AlphaFold3 were used.^87,88 Various copy numbers (12, 14, 16, 18 & 20) of PHB1 sequences were input with max template date set on 2024–04-30. After structure prediction. Models with highest pLDDT score were analyzed. For the PHB1/PHB2 11-mer, 6 copies of PHB1 and 5 copies of PHB2 were entered into AlphaFold2. For the MRS2/PHB1/PHB2 heterocomplex, 5 copies of MRS2, 6 copies of PHB1, and 5 copies of PHB2 were entered into the AlphaFold3 web interface. UCSF ChimeraX⁹² was used for visualization of the models. ChimeraX and Biorender were used to generate figures in the manuscript. Sequence alignment was conducted with ClustalΩ in the MPI Bioinformatics Toolkit,^90,94 and Espript 3.0⁹¹ was utilized to generate the sequence alignment figure.

QUANTIFICATION AND STATISTICAL ANALYSIS

All data are presented as mean ± SEM unless stated otherwise. Statistical analysis was conducted using a Two-tailed Student’s un-paired t test for one or multiple samples and one-way ANOVA with Tukey’s multiple comparisons or regression/correlation analysis when required. Statistically significant was defined as p < 0.05. All experiments were conducted at least three times unless specified otherwise. GraphPad Prism was used for statistical testing. GraphPad Prism version 9, Bio-render and Canvas 11 were used to generate graphs and plot the figures.

Supplementary Material

SUPPLEMENTAL INFORMATION

Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2026.117873.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Monoclonal ANTI-FLAG M2	Sigma Aldrich, MO, USA	Cat# F3165;RRID: AB_259529
TOM20	Cell signaling	Cat# 42406S; RRID: AB_2687663
β-Actin	Santa Cruz Biotechnology	Cat# SC47778; RRID: AB_626632
MRS2	Thermo Fisher Scientific	Cat# PA5-53616;RRID: AB_2644211
PHB1	Cell Signaling	Cat# 2426S ;RRID: AB_823689
VDAC2	Cell Signaling	Cat# 9412S ;RRID: AB_10828727
SLP2	Cell Signaling	Cat# 89199T
V5 - Tag	Cell Signaling	Cat# D3H8Q; RRID: AB_2687461
HA - Tag	Cell Signaling	Cat# C29F4; RRID: AB_1549585
MCU	Cell Signaling	Cat# D2Z3B; RRID: AB_2721812
Rabbit IgG	Cell Signaling	Cat# 3900S; RRID: AB_1550038
Alexa Fluor 488	Molecular Probes, Invitrogen	Cat # A-11029; RRID: AB_2534088
Alexa Fluor 647	Molecular Probes, Invitrogen	Cat # A-20006
α-Mouse Secondary (HRP-conjugated)	Cell Signaling	Cat# 7076; RRID: AB_330924
α-Rabbit Secondary (HRP-conjugated)	Cell Signaling	Cat# 7074; RRID: AB_2099233
Bacterial and Virus Strains
Ad-Mito-MagFRET-1	Vector Biolabs, PA, USA	N/A
Ad-Cre	Vector Biolabs, PA, USA	N/A
AAV8-TGB-iCre	Vector Biolabs, PA, USA	N/A
BL21-CodonPlus Competent Cells (Escherichia coli)	Agilent Technologies, CA, USA	Cat# 230280
Chemicals, Peptides, and Recombinant Proteins
Halt™ Protease and Phosphatase Inhibitor Cocktail	Thermo Fisher Scientific	Cat# 78430
Restore™ PLUS Western Blot Stripping Buffer	Thermo Fisher Scientific	Cat# 46430
NuPAGE™ MOPS SDS Running Buffer (20×)	Thermo Fisher Scientific	Cat# NP0001
NativePAGE™ Bis-Tris Mini Protein Gels	Thermo Fisher Scientific	Cat#BN1002BOX
NativePAGE™ Running Buffer Kit	Thermo Fisher Scientific	Cat#BN2007
NativePAGE™ Sample Buffer	Thermo Fisher Scientific	Cat#BN2003
NativeMark™ Unstained Protein Standard	Thermo Fisher Scientific	Cat# LC0725
Coomassie dye	Novex	LC6065
SuperSignal™ West Pico PLUS Chemiluminescent Substrate	Thermo Fisher Scientific	Cat# 34578
Bis-Tris Polyacrylamide Gel (4%-12%)	Thermo Fisher Scientific	Cat# WG1402BOX
Dulbecco’s Phosphate Buffered Saline (DBPS)	Gibco	Cat#14190144
IP lysis buffer	Thermo Fisher Scientific	Cat #87788
Dynabeads protein G	Thermo Fisher Scientific	Cat #10003D
HisPur™ Ni-NTA Magnetic Beads	Thermo Fisher Scientific	Cat #88831
CHAPS Detergent	Thermo Fisher Scientific	Cat#28299
TaqMan™ Fast Advanced Master Mix	Thermo Fisher Scientific	Cat# 4444557
Imidazole	Thermo Fisher Scientific	Cat#A10221.36
Guanidine hydrochloride solution, 6M	Millipore Sigma	Cat#SRE0066-100ML
L-Lactic Acid	Sigma Aldrich	Cat# L1750
Magnesium Green, AM	Invitrogen	Cat#M3735
Magnesium Green Pentapotassium Salt	Sigma Aldrich	Cat# 2115
Opti-MEM Reduced Serum Medium, no phenol red	Thermo Fisher Scientific	Cat#11058021
BODIPY 493/503	Sigma Aldrich	Cat# P2256
MitoSOXRed	Sigma Aldrich	Cat# 10127647001
Mito Tracker Deep Red	Invitrogen	Cat#M22426
FURA-FF Pentapotassium Salt	Sigma Aldrich	Cat# L7022
Magnesium Chloride	Sigma Aldrich	Cat# M1028
Adenosine -5-triphosphate magnesium	Sigma	Cat#A9187-1G
Adenosine -5-triphosphate potassium	Sigma	Cat#A8937
Calcium Chloride	Sigma Aldrich	Cat# 2115
Normal Goat Serum	Invitrogen	Cat# 10000C
Dihydrorhodamine 123	Thermo Fisher Scientific	Cat# D632
Tetramethylrhodamine, Ethyl Ester, Perchlorate (TMRE)	Thermo Fisher Scientific	Cat# T669
Fetal Bovine Serum, Characterized, Heat Inactivated	Hyclone (GE Life Sciences) UT, USA	Cat# SH30396.03
Rotenone	Sigma-Aldrich	Cat# 557368
Antimycin A	Sigma Aldrich	Cat# A8674
Antibiotic-Antimycotic, 100×	Gibco (ThermoFisher Scientific)	Cat# 15240
Collagenase (Type 4)	Worthington Biochemical Corp., NJ, USA	Cat# LS004188
Nonidet P40 (NP40)	Thermo Fisher Scientific	Cat# 11-102-0016
Digitonin	Sigma Aldrich	Cat# D141
Triton X-100	Sigma Aldrich	Cat# 93443
Glucose	Sigma-Aldrich	Cat# G8644
L-Glutamine, 200mM	Gibco (ThermoFisher Scientific)	Cat# 250030
Paraformaldehyde (PFA)	Thermo Fisher Scientific	Cat#28908
Disuccinimidyl glutarate (DSG)	Thermo Fisher Scientific	Cat#20593
Bis (sulfosuccinimidyl)suberate (BS3)	Thermo Fisher Scientific	Cat#21580
Glutaraldehyde	Electron Microscopy Sciences	Cat#16000
Sodium borohydride	EM science	Cat# SX0355-1
MEM Non-Essential Amino Acids (100×)	Gibco (Thermo Fisher Scientific)	Cat# 11140-050
Sodium Pyruvate, 100mM	Gibco (Thermo Fisher Scientific)	Cat# 11360
Glycerol	Sigma	Cat#56-81-5
Succinate	Sigma Aldrich	Cat# S2378
Isopropyl β-D-1-thiogalactopyranoside, IPTG, Isopropyl β-D-thiogalactoside (IPTG)	Sigma	Cat#I6758-1G
HEPES (1M)	Gibco (Thermo Fisher Scientific)	Cat# 15630
Non-fat dry skimmed milk	Lab Scientific, MA, USA	Cat# M-0841
Dithiothreitol (DTT)	Thermo Fisher Scientific	Cat#R0861
Lipofectamine 3000 Transfection Reagent	Thermo Fisher Scientific	Cat# L3000001
Lipofectamine RNAi MAX Transfection Reagent	Thermo Fisher Scientific	Cat# 13778075
Sodium Chloride	Sigma-Aldrich	Cat# S9888
Potassium Chloride	Sigma-Aldrich	Cat# P3911
HBSS	Gibco	Cat#14-175-095
Oligomycin	Sigma-Aldrich	Cat# 75351
FCCP	Sigma-Aldrich	Cat# C2920
Tris 1M buffer solution	Thermo Fisher Scientific	Cat# J61429
Acetic Acid (Glacial)	Sigma-Aldrich	Cat# 695092
Collagen 1, Rat Tail	ThermoFisher Scientific	Cat# A10483
Coomassie G-250	ThermoFisher Scientific	Cat#LC6065
10× PBS	Thermo Fisher Scientific	Cat#70011044
TBS-Tween– 20 buffer	Thermo Fisher Scientific	Cat#28360
MES SDS running buffer	Invitrogen	Cat#B0002
1× LDS Sample Buffer	Thermo Fisher	Cat#NP0007
Agilent Seahorse XF DMEM Medium pH – 7.4	Agilent Technologies	Cat# 103575-100
Agilent Seahorse XF calibrant media	Agilent Technologies	Cat# 100840-000
Seahorse XF Pyruvate solution	Agilent Technologies	Cat# 103578- 100
Seahorse XF Glucose solution	Agilent technologies	Cat# 103577-100
Seahorse XF Glutamine solution	Agilent technologies	Cat# 103579-100
Critical Commercial Assays
Pierce™ BCA Protein Assay Kit	ThermoFisher Scientific	Cat# 23225
Mitochondrial isolation kit	Abcam	Cat##110170
RNeasy Plus Universal Mini Kit	QIAGEN, MD, USA	Cat# 73404
iBlot PVDF Regular Stacks	Invitrogen, ThermoFisher Scientific, USA	Cat# IB24001
Seahorse XF Cell Mito Stress Kit	Agilent Technologies	Cat# 103015-100
Deposited data
MassIVE dataset identifier ProteomeXchange dataset identifier	This paper	MSV000102618, PXD081706.
Experimental Models: Cell Lines
Human: HEK 293T cell Line	ATCC	Cat# CRL-3216
Monkey COS-7 Cell Line	ATCC	Cat# CRL-1651
Mouse: Primary Hepatocytes	This Study	N/A
Mouse: Mouse Embryonic Fibroblasts Cells (MEFs)	This Study	N/A
Experimental Models: Organisms/Strains
C57BL/6J Mice	The Jackson Laboratory, USA	C57BL/6J
Phbfl/fl Mice	He et al.85	N/A
Mrs2−/− (Mrs2 KO) Mice	Daw et al.5	N/A
Oligonucleotides
Scr siRNA Pool	Horizon Discovery (PerkinElmer)	Cat# D-001820-01-05
siPHB1	Santa Cruz Biotechnology	Cat#37630
Recombinant DNA
H.s.MRS2 - C-FLAG, pcDNA3.1 (+)	Genescript	This study
H.s.MRS2 Δ80-110-FLAG, pcDNA3.1 (+) – Δ1	Genescript	This study
H.s.MRS2Δ120-150-FLAG, pcDNA3.1 (+)– Δ2	Genescript	This study
H.s.MRS2 Δ160-190-FLAG, pcDNA3.1 (+)– Δ3	Genescript	This study
H.s.MRS2Δ210-240-FLAG, pcDNA3.1 (+)– Δ4	Genescript	This study
H.s.MRS2 Δ242-270-FLAG, pcDNA3.1 (+)– Δ5	Genescript	This study
H.s.MRS2Δ290-330-FLAG, pcDNA3.1 (+)– Δ6	Genescript	This study
H.s.MRS2Δ410-443-FLAG, pcDNA3.1 (+)– Δ7	Genescript	This study
H.s.MRS2 - C-mRFP, pcDNA3.1 (+)	Genescript	This study
H.s.MRS2 Δ242-270-mRFP, pcDNA3.1 (+)– Δ5	Genescript	This study
H.s.MRS2Δ290-330-mRFP, pcDNA3.1 (+)– Δ6	Genescript	This study
H.s.MRS2Δ410-443-mRFP, pcDNA3.1 (+)– Δ7	Genescript	This study
H.s.MRS2 - C-FLAG, pcDNA3.1 (+)	Genescript	This study
H.s.MRS2 Δ80-110- V5, pcDNA3.1 (+)	Genescript	This study
H.s.MRS2Δ120-150- V5, pcDNA3.1 (+)	Genescript	This study
H.s.MRS2 Δ160-190- V5, pcDNA3.1 (+)	Genescript	This study
H.s.MRS2Δ210-240- V5, pcDNA3.1 (+)	Genescript	This study
H.s.MRS2 Δ242-270- V5, pcDNA3.1 (+)	Genescript	This study
H.s.MRS2Δ290-330- V5, pcDNA3.1 (+)	Genescript	This study
H.s.MRS2Δ410-443- V5, pcDNA3.1 (+)	Genescript	This study
H.s PHB1-HA - C-FLAG, pcDNA3.1 (+)	Genescript	This study
Software and Algorithms
Canvas 11.0	ACD Systems	https://www.canvasgfx.com/en/support/canvas-11/
GraphPad Prism version 9	GraphPad Software	https://www.graphpad.com/scientific-software/prism
Leica Application Suite X (LAS X)	Leica Microsystems Inc., IL, USA	https://www.leica-microsystems.com/products/microscopesoftware/p/leica-las-x-ls/
ImageJ	Schindelin et al.86	https://imagej.nih.gov/ij/
Illustration Toolkit	Pathway Motifolio, MD, USA	https://www.motifolio.com
Biorender	Toronto, Canada	https://www.biorender.com
Alphafold-Multimer (v2.3.2) and AlphaFold3	Abramson et al.87; Jumper et al.88	https://alphafold.ebi.ac.uk/
Picasso software suite 3.9.141.303	Schnitzbaueret al.89
MPI Bioinformatics Toolkit	Gableret al.90	https://toolkit.tuebingen.mpg.de/
Espript 3.0	Robert and Gouet,91	https://espript.ibcp.fr/ESPript/
ChimeraX v1.8	Goddardet al.92	https://www.cgl.ucsf.edu/chimerax/
Mascot software	version 2.6.2; Matrix Science	https://www.matrixscience.com/mascot_support_v2_6.html
Scaffold software	version 4.9.0	https://www.proteomesoftware.com/

Highlights.

• Prohibitin 1 is identified as an activator of mitochondrial Mg²⁺ channel MRS2

• Mitochondria accumulate Mg²⁺ in a manner analogous to Ca²⁺ uptake

• Genetically encoded Mg²⁺ biosensors quantify mitochondrial Mg²⁺ uptake up to ~15 mM

• PHB1-dependent MRS2 activity regulates mitochondrial Mg²⁺ uptake and cellular bioenergetics