HEK293T, HEK293T MCU-KO (generous gift from Vamsi Mootha, Harvard Medical School, Boston, MA), HEK293T MCU-rescue, immortalized fibroblasts, transformed fibroblasts and MCU-KO transformed fibroblasts were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Mediatech, MT10013CM) supplemented with 10% fetal bovine serum (FBS, HyClone, SH30071.03) and 1% antibiotic-antimycotic (anti-anti, Invitrogen, 15240062), and incubated in a humidified incubator at 37°C with 95% air/5% CO₂. MCU-KO negative control (NC) transformed fibroblasts and MCU-rescue transformed fibroblasts were cultured in DMEM (10% FBS, 1% anti-anti) and 150 μg/ml Hygromycin B (Mediatech, MT30-240-CR), and incubated in a humidified incubator at 37°C with 95% air/5% CO₂.

Cells in culture were washed with 1X Dulbecco’s phosphate-buffered saline (DPBS, Mediatech, MT21-031-CM), detached with 0.25% trypsin (Invitrogen, 15090046) and resuspended in DMEM (10% FBS, 1% anti-anti). The cell suspension was washed twice with 1X DPBS and lysed with RIPA buffer [50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1% NP-40, 0.25% deoxycholic acid, 1 mM EDTA] supplemented with 200 μM phenylmethylsulfonyl fluoride (PMSF) and protease inhibitor cocktail (Roche, 11697498001). For cell lysis, samples were placed in a tube rotator for 30 min at 4°C. Lysates were centrifuged for 10 min. Protein concentration was determined using the Pierce BCA Protein Assay kit (Thermo Scientific, 23227). Samples were prepared for loading with 4X LICOR loading buffer (LICOR, 928-40004) and β-mercaptoethanol. Samples were boiled at 100°C for 5 min. Gel electrophoresis was performed in 4%–12% Bis-Tris gels (NuPAGE, NP0322) and MOPS running buffer (Novex, NP0001). Transfer used nitrocellulose membranes (Sigma, RPN303D) in 20% MeOH Tris/Glycine buffer for 1 h at 100 V. Immunoblotted membranes were blocked with TBS Odyssey blocking buffer (LICOR, 927-50100) for 1 h and then incubated in primary antibodies overnight at 4°C. Membranes were washed 3 times with 0.1% Tween-20 (Bio-Rad, 1706531) TBS (TBS-T) for 5 min and incubated with IRDye secondary antibodies for 1 h in the dark. Blotted membranes were washed 3 times with TBS-T for 5 min and imaged using the Odyssey CLx system. Relative levels of MCU, PDH, pPDH, Tim23, and HSP60 were normalized to tubulin expression detected on the same blots. The antibodies used were: anti-Tubulin (1:5,000, Invitrogen, 322600), anti-MCU (D2z3B) (1:5,000, Cell signaling, 14997s), anti-Pyruvate Dehydrogenase E1-alpha subunit (1:5,000, Abcam, ab110334), anti-phospho-PDHE1-A type I (Ser293) (1:5,000, Millipore, ABS204), anti-Hsp60 (1:5,000, Abcam, ab46798), anti-Tim23 (1:5,000, BD Biosciences, 611223), IRDye 680RD goat anti-mouse (1:10,000, LICOR, 925-68070), IRDye 680RD goat anti-rabbit (1:10,000, LICOR, 926-32211), IRDye 800CW goat anti-mouse (1:10,000, LICOR, 926-68070), and IRDye 800CW goat anti-rabbit (1:10,000, LICOR, 925-3211).

Cells in culture were washed with 1X DPBS, detached and resuspended in DMEM (10% FBS, 1% anti-anti). The cell suspension was incubated for 10 min in DMEM (10% FBS, 1% anti-anti) in a humidified incubator at 37°C with 95% air/5% CO₂, and then centrifuged. The pellet was resuspended in Ca²⁺-free extracellular-like buffer (ECM: 20 mM HEPES-NaOH, 120 mM NaCl, 5 mM KCl, 1 mM KH₂PO₄, 0.2 mM MgCl₂, 0.1 mM EGTA, pH 7.4) made using dH₂O treated with BT Chelex^® 100 resin (Bio-Rad, 143-2832) and incubated for 10 min in a humidified incubator at 37°C with 95% air and 5% CO₂. The cell suspension was centrifuged and resuspended in Ca²⁺-free intracellular-like buffer (ICM: 20 mM HEPES-NaOH, 10 mM NaCl, 120 mM KCl, 1 mM KH₂PO₄, 5 mM succinate, pH 7.5) made using dH₂O treated with BT Chelex^® 100 resin. Fluorescence was monitored in a fluorimeter with multiwavelength excitation and emission (Delta RAM, PTI) at a constant temperature of 37°C. Fura-FF (AAT Bioquest, 21028, Kd = 5.5 μM) excited at 340 nm and 380 nm was monitored at 535 nm emission. TMRE (Molecular Probes, T669) was excited at 560 nm and emission monitored at 595 nm. Addition of reagents during fluorometric measurements was performed according to the following timeline: T = 0, ICM-cell suspension; T = 25 s, 1 μM Fura-FF and 10 nM TMRE; T = 50 s, 0.004% digitonin; T = 100 s to permeabilize the plasma cell membrane, 2 μM thapsigargin (Sigma, T9033) to inhibit cytosolic Ca²⁺ clearance by the ER; T = 200 s, 10 μM CGP37157 (Tocris, 1114) to inhibit Ca²⁺ extrusion from mitochondria; T = 400 s, 3–5 μM CaCl₂; T = 600 s, 2 μM CCCP; T = 700 s, 1 mM EGTA; and T = 750 s, 1 mM CaCl₂ (Sigma-Aldrich, 21115). To determine cytosolic (bath) Ca²⁺ concentration ([Ca²⁺]_cyt) based on the ratiometric calibration of Fura-FF, we used the following equation: C a 2 + c y t = R − R min R max − R * S f 2 S b 2 * K d where R is the ratio of Fura-FF fluorescence at 340/380 nm excitation ;   R min is the R measured with [Ca²⁺ ] = 0; R max is the R measured at saturating [Ca²⁺]; S f 2 is fluorescence of Fura-FF excited at 380 nm at [Ca²⁺ ] = 0 . S b 2 is fluorescence of Fura-FF excited at 380 nm at saturating [Ca²⁺]; K d is the dissociation constant of Fura-FF K d = 5.5 μM), which was experimentally validated using a set Ca²⁺ calibration buffers (World Precision Instruments, Calbuf-2). Rates of mitochondrial Ca²⁺ uptake were determined by fitting a single exponential from the initial peak after Ca²⁺ addition (T = 400 s) until a steady state was reached (T = 550 s). To determine the mitochondrial membrane potential (Δψ_m), fluorescence of TMRE was normalized to the maximal value obtained after the addition of CCCP. Then, normalized values between T = 150 s to T = 200 s were averaged to obtain the reported Δψ_m.

Primary mouse fibroblasts were isolated from the skin of 1-week old homozygous Mcu^fl/fl pups (The Jackson Laboratory, stock #029817). Pups were euthanized by decapitation. The skin was washed three times with ice-cold 1X DPBS (1% anti-anti) and sliced into 1-cm pieces using a clean razor blade. Tissue was placed in pre-warmed 500 μg/ml thermolysin solution (Sigma, T7902) prepared in HEPES buffer (Sigma, H3375) and incubated for 2 h at 37°C with constant agitation. The tissue was transferred to a pre-warmed 0.125 U/mL collagenase solution (Collagenase Type 2, Worthington Biochemical, LS004176) prepared in DMEM (10% FBS, 1% anti-anti) and incubated for 2 h at 37°C with constant agitation. Digested tissue was forcefully ground using a 10 ml syringe plunger and centrifuged. The fibroblast-containing pellet was plated in DMEM (10% FBS, 1% anti-anti). To validate the identity of the isolated primary fibroblast population, the cells were sorted by fluorescence-activated cell sorting (FACS) for the surface marker platelet-derived growth factor receptor alpha (PDGFRA), using antibodies anti-CD140a (PDGFRA) Monoclonal Antibody (APA5) (1:100, Thermo Fisher, 14-1401-81) and goat anti-Rat IgG H&L (Alexa Fluor^® 488) (1:200, Abcam, 150157), and emission filter 530 nm. HEK293T and human fibroblasts were used as negative and positive controls, respectively.

Plasmids for the expression of pRRL-TERT, constitutively-active pRRL-CDK4^R24C, dominant-negative pRRL-TP53^R248W, and pRRL-HRas^G12V were generated and kindly donated by Dr. Todd W. Ridky. Lentivirus was produced in HEK293T cells. A day before transfection, HEK293T cells were seeded in 15-cm dishes, aiming for 50% confluency the next day. 24 h after seeding, cells were co-transfected with 22.5 μg of lentiviral plasmid, 16.9 μg of the packaging plasmid psPAX2 (Addgene, 12260) and 5.6 μg of the envelope plasmid VSV-G (Addgene, 14888) using Lipofectamine 3000 (Thermo Fisher, L3000015). After 16 h, 10 mM Na-butyrate was added to the culture. Lentivirus-containing media was collected after 48 and 72 h of transfection and filtered through a 45 μm filter.

Primary fibroblasts isolated from 1-week old Mcu^fl/fl mouse pups were immortalized by sequential transduction with TERT and CDK4^R24C lentiviruses. Immortalized cells were kept in culture as immortalized cells or transduced with p53^R248W and HRas^G12V lentiviruses for oncogenic transformation. Genomic elimination of MCU (MCU-KO) in transformed fibroblasts was achieved by transient expression of Cre recombinase by transfection with the plasmid pLM-CMV-R-Cre (Addgene, 27546), which codes for mCherry-Cre recombinase, using Lipofectamine 3000. The cells were sorted by FACS for red mCherry fluorescence. Collected cells were diluted and seeded at one cell/well in 96-well plates. Colonies were grown and verified by Western blot for MCU-KO. Rescue of MCU (MCU-rescue) was achieved by transfection of MCU-KO transformed fibroblasts with the plasmid pCMV3-MCU-FLAG (Sino Biological, MG5A1846-CF), which codes for MCU, using Lipofectamine 3000. Generation of MCU-KO NC transformed fibroblasts was achieved by transfection of MCU-KO transformed fibroblasts with the plasmid pCMV3-C-FLAG (Sino Biological, CV012), using Lipofectamine 3000. Transfection of MCU-KO NC and MCU-rescue was followed by Hygromycin B selection and isolation of single clones in 96-well plates. Colonies were grown and verified by Western blot for MCU expression.

All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Pennsylvania (protocol #806559). Tumor xenografts of primary mouse fibroblasts were performed in 4–8 week old male outbred athymic nude mice J:Nu (Jackson Laboratory, stock #007850). Tumor xenografts of HEK293T were performed in 4–8 week old male immunodeficient mice (NOD.Cg-Prkdc^scid /J) (Jackson Laboratory, stock # 001303). Mice were housed in pathogen-free conditions in a 12 h light/12 h dark cycle with food and water ad libitum. The day of surgery, fibroblasts were washed with 1X DPBS, trypsinized and resuspended in DMEM (10% FBS, 1% anti-anti). For immortalized vs. transformed xenograft experiments, cells were resuspended at a density of 10 × 10⁶ cells/mL. For HEK293T xenograft experiments, cells were resuspended at a density of 10 × 10⁶ cells/mL. For transformed vs. transformed MCU-KO xenograft experiments, cells were resuspended at a density of 40 × 10⁶ cells/mL. Mice were anesthetized using 1.5%–4% isoflurane in an induction chamber. Just before injection, cells were mixed with Matrigel (Fisher Scientific, 354234) in equal parts. Injection was done into the subcutaneous space of the mouse flanks. Each mouse received one injection into the subcutaneous space of one flank and another of immortalized or MCU-KO fibroblasts into the opposite flank. For HEK293T xenografts, each mouse received one injection of HEK293T-WT cells into one flank and another of HEK293T MCU-KO or HEK293T MCU-rescue into the opposite flank. After injections, tumor formation and progression were monitored for 3–4 weeks and measured with a caliper. Tumor volume was determined using the following equation: T u m o r   v o l u m e   m m 3 = D l o n g D s h o r t 2 ÷ 2 where D l o n g is the longest diameter and D s h o r t is the shortest diameter of the tumor. At the end of the experiment, the mice were euthanized by CO₂ exposure followed by decapitation. Tumors were excised and formalin fixed for further processing. In brief, tumors were submerged in zinc formalin solution for 24 h at 4°C and then transferred into 70% ethanol. Tumors were embedded in paraffin (Fisher Scientific, T56-5), cut into 5-μm sections, and mounted by the Molecular Pathology and Imaging Core of the Perelman School of Medicine at the University of Pennsylvania.

Immunostaining was performed by the Skin Histology and Characterization Core of the Perelman School of Medicine at the University of Pennsylvania. Tumor sections were imaged at ×20 using a Keyence B2-X710 microscope at eight different regions of heterogeneous tissue excluding necrotic areas, the centers and tumor edges. Each region was imaged at 20 continuous field of views that were stitched to create a single image. Regions were imaged with three different channels: blue (DAPI), red (Ki-67), and green (TUNEL) and saved as a composite. Images were imported into the software GNU Image Manipulation Program (GIMP, www.gimp.org) and the fluorescent signal of each channel was extracted as a greyscale component. Then, images were imported into the software FIJI (Schindelin et al., 2012). A signal index was determined by summing the pixel counts of the fluorescence signals as previously described (Billings et al., 2015; Asrani et al., 2019). In brief, greyscale components were transformed into binary images where the number of black pixels, representative of fluorescence, were quantified to obtain a DAPI, Ki-67, and TUNEL index. Using the histogram analysis tool in FIJI, each pixel was cataloged as background or signal and then summed to obtain total number of pixels, index of background, and index of signal. As a complementary approach to quantify individual Ki-67⁺ cells in tumor sections, each field of view of a region imaged with the red channel was individually processed in FIJI. First, the background was subtracted, then the image was converted into a binary component. Finally, each nucleus was represented as a single particle and counted to obtain the total number of Ki-67⁺ cells.

Tumor-section slides were processed for hematoxylin and eosin (H&E) staining by the Molecular Pathology and Imaging Core of the Perelman School of Medicine at the University of Pennsylvania. In brief, tumor sections were deparaffinized in xylene (Azer, ES609), rehydrated in ethanol (Azer, ES631) and washed with deionized water. Then, tissue was stained with hematoxylin (Leica Biosystems, 3801540) followed by eosin (Leica Biosystems, 3801600). Tissue was dehydrated with alcohol and mounted. Each slide was manually scored under the supervision of a pathologist (Dr. John T. Seykora). Necrosis analysis was done by measuring normal tissue area and necrotic area in the whole section using an eyepiece graticule with calibrated grids (25 mm). For scoring mitotic figures and giant multinucleated cells, each slide was examined at ×40 at 10 different regions of heterogeneous tissue excluding necrotic areas, the centers and tumor edges. Representative images of necrotic and normal tumor tissue, mitotic figures, and giant multinucleated cells were taken at ×20 using a Keyence B2-X710 microscope.

Cells in culture were washed with 1X DPBS, trypsinized and resuspended in tumorsphere media, composed of DMEM/F12 medium (Sigma-Aldrich, D8437) supplemented with 0.4% bovine serum albumin (BSA, Life Technologies, Invitrogen, 15561020), 1% anti-anti, 20 ng/ml epidermal growth factor (Sigma-Aldrich, E5036), 10 ng/ml basic fibroblast growth factor (Sigma-Aldrich, F0291), 5 µg/ml insulin (Life Technologies, Invitrogen, A11429IJ) and 1X B27 supplement (Life Technologies, Invitrogen, 17504-044), as previously described (Johnson et al., 2013; Weissenrieder et al., 2020). Briefly, cells were plated at a density of 200 cells/well in 200 µL media in low adhesion plates, generated by applying Aggrewell solution (Stemcell Technologies, 7010) for 5 min, spinning down for 5 min at 1,300 rpm and then rinsing gently with basal media (DMEM/F12, 0.4% BSA, 1% anti-anti). To reduce evaporation, outside wells were filled with sterile DPBS and not used as experimental wells. Quantification of spheroids, defined as rounded aggregates of cells with a smooth surface and poor cell-to-cell definition, was performed after 7 days of incubation.

Cells in culture were washed with 1X DPBS, trypsinized and resuspended in DMEM (10% FBS, 1% anti-anti). Cells were plated at a density of 15,000 cells/well at day 0. Quantification of the number of cells per well was performed for 4 days-post-seeding without continuous passage. For the proliferation assay in low nutrient conditions, cells were plated at a density of 15,000 cells/well at day 0 in high glucose (25 mM) and glutamine (4.5 mM) media. Then, at day post-seeding one media was changed to low glucose (4.5 mM) and glutamine (0.75 mM) media. Accordingly, quantification of the number of cells per well was performed for 4 days post-seeding without continuous passage.

Cells in culture were washed with 1X DPBS, trypsinized and resuspended in DMEM (10% FBS, 1% anti-anti). Cells were plated at a density of 15,000 cells/well atop of 10 mg/ml Matrigel coated Transwells (Corning, 3464) in serum-free DMEM, while the bottom well contained 10% FBS DMEM. After 24 h incubation, cells at the bottom of the membrane were fixed with 4% PFA (Electron Microscopy Sciences, 15713) and stained with 1 μg/μL Hoechst 33342 (Invitrogen, H3570). Quantification was done by analyzing three different fields of views of each well, imaged at ×20, and quantified automatically by the software FIJI.

Cells in culture were washed with 1X DPBS, trypsinized and resuspended in DMEM (10% FBS, 1% anti-anti). Cell suspension was centrifuged at 1,000 rpm for 5 min. The pellet was washed with 1X DPBS and centrifuged at 1,000 rpm for 5 min. The supernatant was aspirated and the cell pellet was resuspended and incubated in binding buffer with fluorochrome [488]-conjugated Annexin V (Thermo Fisher Scientific, R37174), according to manufacturer’s instructions. After 15-min incubation, 4 mM DAPI (Abcam, ab228549) was added to the suspension and cells were analyzed by FACS with emission filters 530 and 440 nm for Annexin V/DAPI detection. Data were plotted and analyzed using FCS Express software. Debris was excluded by a first gate set in forward-area and side-area (FSC-A vs. SSC-A) plots. Doublets were excluded by a second level of gating in forward-area and forward-height (FSC-A vs. FSC-H) plots. Cells within gates were plotted in DAPI vs. Annexin V plots. Based on negative and 1 μM staurosporine-treated controls for each FACS plot, gated cells were divided into four quadrants: lower left (healthy cells), lower right (early apoptotic cells), upper right (late apoptotic/dead cells), and upper left (necrotic cells). Final values represent percent of gated cells in each quadrant.

Cells in culture were washed with 1X DPBS, trypsinized and resuspended in DMEM (10% FBS, 1% anti-anti), plated at a density of 30,000 cells/well and incubated for ∼16 h. Media was changed to DMEM base assay medium (Sigma, D5030) supplemented with 1 mM Na-pyruvate, 2 mM glutamine, and 10 mM glucose. Baseline OCR was measured three times before the addition of 2 μM oligomycin, followed by 0.5 μM FCCP, and finally 0.5 μM rotenone/antimycin A+ 1 μg/μL Hoechst 33342 (Invitrogen, H3570) (Seahorse XF Cell Mito Stress Test Kit, Agilent, 103015) according to the manufacturer’s instructions. To measure changes in OCR following acute agonist stimulation, baseline OCR determination was followed by the addition of 500 μM ATP. At the end of each assay, the center portion of each well was imaged at ×20 to detect nuclei counterstained with Hoechst. Nuclei were automatically counted using FIJI and used to normalize individual OCR/well.

Cells in culture were washed with 1X DPBS, trypsinized and resuspended in DMEM (10% FBS, 1% anti-anti), plated at a density of 35,000 cells/well and incubated for ∼16 h. Media was changed to DMEM base assay medium (Sigma, D5030) supplemented with 1 mM glutamine. Baseline ECAR was measured three times before the addition of 10 μM glucose, followed by 1 μM oligomycin, and finally 50 μM 2-deoxy-glucose (2-DG) + 1 μg/μL Hoechst (Seahorse XF Glycolysis Stress Test Kit, Agilent, 103020) according to the manufacturer’s instructions. At the end of each assay, the center portion of each well was imaged at ×20 to detect nuclei counterstained with Hoechst. Nuclei were automatically counted using FIJI and used to normalize individual ECAR/well.

Cells in culture were washed with 1X DPBS, trypsinized, resuspended in DMEM (10% FBS, 1% anti-anti) and plated at a density of 30,000 cells/well. After ∼16-h incubation, cells were incubated with XTT (Millipore Sigma, 11465015001) according to the manufacturer’s instructions. Reduction of XTT was quantified by formazan absorption at 490 nm and normalized to reference wavelength absorbance at 690 nm. XTT absorbance was detected every 30 min for 4 h, during which readout is linear, allowing for the best detection of differences between cell lines.

Cells in culture were washed with 1X DPBS, trypsinized, resuspended in DMEM (10% FBS, 1% anti-anti), and plated at a density of 30,000 cells/well. After ∼16 h incubation, cells were incubated with 25 μM 2′,7′-dichlorodihydrofluorescein diacetate (DCFDA) (Sigma-Aldrich, D6883) and 1 μg/μL Hoechst. DCFDA oxidation was detected at 535 nm and normalized to Hoechst.

Cells in culture were washed with 1X DPBS, trypsinized, resuspended in DMEM (10% FBS, 1% anti-anti), and seeded onto coverslips. After ∼16 h, the cells were transfected with Lipofectamine 3000 and 2.5 μg of the plasmid pCMV CEPIA2mt (Addgene, 58218) or CEPIA3mt (gift from M. Trebak, Addgene, 58219). Cells were imaged 48 h after transfection. In brief, coverslips were transferred to a perfusion chamber and perfused with extracellular-like solution (135 mM NaCl, 5.9 mM KCl, 1.2 mM MgCl₂, 11.6 mM HEPES, 10 mM glucose, 1.5 mM CaCl₂). Fluorescence of CEPIA2/3 mt was monitored using a Nikon Eclipse microscope at 20x at the excitation/emission wavelengths 488/500–550 nm (Suzuki et al., 2014). For each measurement, three images of the same field of view were taken 60 s apart. To calibrate CEPIA2mt signals, cells were perfused with 1 μM ionomycin (Invitrogen, 124222) followed by a 0-Ca²⁺ Tyrode’s solution (135 mM NaCl, 5.9 mM KCl, 1.2 MgCl₂, 1 mM EGTA, 11.6 mM HEPES) to obtain maximum and minimum intensity of CEPIA2mt. Data analysis was performed using VisiView software (Visitron Systems GmbH). For each field of view, a region of interest (ROI) was delineated around the edges of each cell to obtain basal, maximum, and minimum fluorescence of CEPIA2mt. To determine the basal mitochondrial Ca²⁺ concentration ([Ca²⁺]_mit), we used the following equation: C a 2 + m i t = F − F min F max − F * K d where F is CEPIA2mt fluorescence, F min is F at 0-Ca²⁺, F max is F at saturating Ca²⁺, and K d is the dissociation constant of CEPIA2mt ( K d = 0.16 μM) (Suzuki et al., 2014).

Cells were washed with 1X DPBS, trypsinized, resuspended in DMEM (10% FBS, 1% anti-anti), and seeded onto coverslips. After 2 days, cells were incubated with 2 μM Fura-2 AM (Thermo Fisher Scientific, F1225) for 30 min at room temperature. Imaging was performed with a Nikon Eclipse Ti-U microscope using a ×20/0.75 objective. Fura-2 AM ( K d = 140 nM) was excited at 340 nm and 380 nm and monitored at 535 nm emission. Before imaging, cells were perfused with 1.5 mM Ca²⁺ Tyrode’s solution (135 mM NaCl, 5.9 mM KCl, 1.2 MgCl₂, 1.5 mM CaCl₂, 11.6 mM HEPES) for 5 min. Baseline recording was done for 1 min while perfusing with 1.5 mM Ca²⁺ Tyrode’s solution. Then, 2 μM ATP was added into perfusion chamber and agonist-induced responses were recorded for 300 s. Recording was terminated after washing the cells with 0-Ca²⁺ Tyrode’s solution (135 mM NaCl, 5.9 mM KCl, 1.2 MgCl₂, 1 mM EGTA, 11.6 mM HEPES) for 100 s. Raw traces for each cell were obtained using VisiView. For analysis, background fluorescence recorded from a cell-free space in the field of view of each coverslip was subtracted from each image, and the background-subtracted signal was normalized to the averaged baseline value (0—50 s) to obtain the ratio over time (R/R₀). Responses were classified as oscillating, single peak, and sustained using a series of statistical parameters applied from the end of the initial peak to the end of the ATP stimulation. Those traces in which there was no agonist-induced response, the Fura-2 AM signal was significantly elevated or too low (20%), and where the responses were qualitatively identified as complex or mixed (19%), were not included in the analysis.

Activated lovastatin was prepared from its inactive lactone prodrug form (Millipore Sigma, PHR1285) as described (Keyomarsi et al., 1991). Cells were washed with 1X DPBS, trypsinized, resuspended in DMEM (10% FBS, 1% anti-anti), and seeded onto 6-well plates. Two days post-seeding, media of all samples, except asynchronous controls, was replaced with 15 μM lovastatin media. After 24 h incubation, lovastatin media was replaced with DMEM (10% FBS, 1% anti-anti). At this moment, asynchronous control and lovastatin-unreleased controls were collected and fixed. The rest of the samples were collected after 24 h of cell cycle release from lovastatin synchronization. For sample collection and fixation, media was collected, the cells were washed with 1X DPBS, trypsinized, and resuspended with the collected media. The cell suspension was centrifuged at 1,000 rpm for 5 min. The supernatant was aspirated and the cell pellet was resuspended in ice-cold 1XDPBS. The cells were then centrifuged at 1,000 rpm for 5 min. The supernatant was aspirated and the cells were resuspended in 100 μL of 1X DPBS. Finally, fixation buffer (80% EtOH in dH₂O) was added in a drop-wise manner while cells were vortexed. For FACS analysis, cells were stained with 4 mM DAPI (Abcam, ab228549) and analyzed at 440 nm emission. Data were plotted and analyzed using FCS Express. Debris was excluded by a first gate set in FSC-A vs. SSC-A plots. Doublets were excluded by a second level of gating in FSC-A vs. FSC-H plots and verified in a DAPI-W vs. DAPI-A plot. Gated cells were plotted in DAPI vs. cell count histogram. Final values represent the percent of gated cells in each phase (G1, S, G2) of the cell cycle, which were determined using the Multicycle AV DNA analysis software in FCS. Unsynchronized- and lovastatin-synchronized samples were used as controls to validate the G1, S, and G2 peaks identified using Multicycle software for cell cycle analysis.

Cells in culture were washed with 1X DPBS, detached with 0.25% trypsin, and resuspended in DMEM (10% FBS, 1% anti-anti). Cells were plated at a density of 50,000 cells/well in 6-well plates in a total of 2 mL of culture media. Wells with no cells were kept as controls for normalization of metabolite concentrations in media. After 48 h post-seeding, media was removed, wells were washed with 1X DPBS, and fresh media was added to each well. Samples of culture media were collected 24 h post-media change and stored at −80°C until analysis. For normalization purposes, number of cells per well was determined manually using a hemacytometer. Quantification of glucose, lactate, glutamine, and glutamate concentrations (nM) in samples of cell culture supernatant was determined enzymatically with a bioanalyzer (YSI2950, YSI Incorporated, Yellow Springs, OH, United States). Rate of metabolite consumption ( v c ), rate of metabolite production ( v p ), cell number area under the curve ( A ), and doubling time ( d ) where calculated with the following equations: v c = V ( x m e d i u m   c o n t r o l − x f i n a l ) A v p = V ( x f i n a l − x m e d i u m   c o n t r o l ) A A = N T * d ln ⁡ 2   1 − 2 − T d d = T log 2 log Q 2 Q 1 where V is volume of cell culture media, x is metabolite concentration, A is cell number area under the curve, N T is final cell count, d is doubling time, T is time of experiment, Q 1 is initial cell number, and Q 2 is final cell number.

Cells in culture were washed with 1X DPBS, trypsinized and resuspended in DMEM (10% FBS, 1% anti-anti). Cells were plated at a density of 52,000 cells/well in 6-well plates. Tracing of labeled nutrients was performed 2 days post-seeding. For tracing of glucose carbons, no-glucose and no-glutamine media was supplemented with 5 mM of ¹³C₆ d-Glucose (Cambridge Isotope Laboratories, Inc., CLM-1396) and 4 mM unlabeled glutamine. Tracing of glutamine was performed by supplementing no-glucose and no-glutamine media with 4 mM ¹³C₅ l-glutamine (Cambridge Isotope Laboratories, Inc., CLM-1822-H) and 5 mM unlabeled glucose. Unlabeled controls were maintained on unlabeled low glucose (5 mM) and glutamine (4 mM) media. For metabolite extraction, after 5-h incubation with the isotope tracers, media was removed, and cells were rinsed with 1X DPBS. Then, 500 µL of pre-chilled (on dry ice) analytical-grade 80% methanol (MeOH, Thermo Fisher, AA22909K2): 20% water (Invitrogen, 10977015) (volume/volume) was added to the well. Cell extracts were scraped and transferred to microcentrifuge tubes, and the well was rinsed with another 500 µL of 80% MeOH that was combined with the first round of extract. Samples were vortexed and incubated on dry ice for 15 min, then centrifuged at >15,000 rpm at 4°C for 15 min. Supernatant containing the extracted metabolites was transferred to a new tube and stored at −80°C until further processing. On the day of analysis, samples were dried using a vacuum concentrator (Savant SpeedVac SPD130, Thermo Fisher Scientific), and dried metabolite pellets were resuspended in 60 µL of 60:40 (v/v) acetonitrile: water, vortexed, and centrifuged at >13,000 rpm at 4°C for 15 min. The supernatant was transferred to glass vials with polypropylene inserts for analysis by liquid chromatography-mass spectrometry (LC-MS). The instrument autosampler was maintained at 4°C, and the sample injection volume was 2.5 µL. Samples were analyzed by hydrophilic interaction chromatography coupled to a quadrupole-orbitrap mass spectrometer (Q Exactive, Thermo Fisher Scientific) via electrospray ionization. The liquid chromatography system (Vanquish Flex UHPLC with binary pump VF-P10 and split sampler VF-A10, Thermo Fisher Scientific) used a BEH amide column (ACQUITY Premier BEH Amide VanGuard FIT column, 2.1 mm × 100 mm, 1.7 µm particle size, Waters Corporation #186009508) for separation. The column was held at 35°C, and the flow rate was 300 µL/min with a gradient of solvent A (20 mM ammonium acetate, 20 mM ammonium hydroxide in 95:5 water: acetonitrile (v/v), pH 9.5) and solvent B (acetonitrile). The gradient was 95% B to 40% B from 0 to 9 min, hold 40% B for 2 min, reverse to 95% B in 0.6 min, and hold at 95% B until 20 total min. Samples were directed to the mass spectrometer for 0.25–16 min. The mass spectrometer was operated in negative ion mode with an automatic gain control (AGC) target of 1E6, maximum inject time of 100 min, scan range of 55-825 m/z, and 140,000 resolution. Electrospray ionization source settings included spray voltage of 3 kV and auxiliary gas heater temperature of 350°C with capillary temperature of 325°C. Raw LC-MS data were converted to mzXML file format using “MSConvert” (Chambers et al., 2012; Adusumilli and Mallick, 2017), and data were analyzed using El-MAVEN software (Melamud et al., 2010; Clasquin et al., 2012). Chemical standards for glucose, glutamine, and TCA cycle intermediates were used to validate metabolite identification. Natural ¹³C abundance correction was performed using AccuCor (Su et al., 2017).

To investigate the roles of MCU in tumorigenesis, we developed in vitro immortalization and transformation models using a set of transgenes to immortalize and then to transform primary fibroblasts isolated from Mcu^fl/fl mice (Figure 1A). Isolated primary fibroblasts were transduced to overexpress TERT, the catalytic subunit of telomerase, and with constitutively-active CDK4 (CDK4^R24C) (Figure 1A). In combination, these genes efficiently immortalize cells without conferring tumor-forming ability or genomic instability (Sasaki et al., 2009). For malignant transformation, the immortalized fibroblasts were transduced with dominant-negative p53 (p53^R248W) and HRas (HRas^G12V) (Figure 1A).

Subcutaneous tumors were produced by injecting immortalized and transformed fibroblasts under the skin of immunocompromised nude (J:Nu) mice. After 3 weeks post-injection, immortalized fibroblasts (n = 5) failed to form tumors whereas all injections (n = 4) of transformed fibroblasts resulted in tumors (Figure 1B). Thus, transformed fibroblasts have a malignant tumorigenic potential while immortalization serves as syngeneic control for a non-malignant phenotype.

Oncogenic mutations including HRas^G12V and p53^R248W can induce remodeling of numerous networks to support cancer progression, including mitochondrial biogenesis and turnover (Yao et al., 2019). Mitochondrial HSP60 and Tim23 expression levels were increased ∼1.5-fold in the transformed vs. immortalized fibroblasts (Supplementary Figure S1A), indicating that oncogenic transformation in vitro enhanced mitochondrial biomass (Supplementary Figure S1B, C). Similarly, MCU protein expression (Figure 1C) was higher by 1.5-fold in transformed vs. immortalized fibroblasts (Figure 1D). To determine whether increased MCU expression was a transformation-associated process, we quantified MCU in three different primary fibroblast preparations (Fibroblasts WT) before any genetic manipulation (Supplementary Figure S1D). WT fibroblasts expressed lower levels of MCU than transformed ones (Supplementary Figure S1E). These results provide in vitro evidence that oncogenic transformation is associated with up-regulation of MCU. To establish the relationship between MCU expression and rates of mitochondrial Ca²⁺ uptake, cells were suspended in an intracellular-like bath solution containing the cell-impermeable Ca²⁺ indicator Fura-FF to monitor bath [Ca²⁺] ([Ca²⁺]_cyt) and digitonin to solubilize the plasma membrane and expose mitochondria to the bath solution. ER Ca²⁺ uptake and mitochondrial Ca²⁺ extrusion were inhibited by addition of thapsigargin and CGP37157, respectively. Ca²⁺ clearance by mitochondrial Ca²⁺ uptake was measured by changes in [Ca²⁺]_cyt after addition of a 3–5 μM bolus of Ca²⁺ to the bath solution. Enhanced MCU expression in transformed fibroblasts was associated with significantly faster rates of mitochondrial Ca²⁺ uptake in permeabilized cells compared with both untransduced WT fibroblasts (Supplementary Figures S1F, G) and immortalized fibroblasts (Figures 1E, F). Enhanced Ca²⁺ uptake observed following transformation was not caused by altered mitochondrial membrane potential (ΔΨ_m) (not shown, but see Figure 5C). Transformation was also associated with altered [Ca²⁺]_mit kinetics in response to ligand-induced intracellular Ca²⁺ release (Supplementary Figure S1H). PDH is activated by dephosphorylation by the mitochondrial matrix Ca²⁺-sensitive PDH phosphatase (PDP) (Denton, 2009). Stimulation of PDH by Ca²⁺ is known to be important for the regulation of mitochondrial metabolism and cancer progression (Denton, 2009; Pan et al., 2013; Luongo et al., 2015; Chakraborty et al., 2017; Anwar et al., 2021). Thus, we examined phosphorylation of PDH by Western blot (Figure 1G). Phospho-PDH (pPDH) was ∼4-fold lower (and almost undetectable) in the transformed cells compared with the immortalized fibroblasts (Figure 1H), suggesting that malignant transformation in vitro results in decreased pPDH associated with more efficient ER-to-mitochondria transfer of Ca²⁺ (Cardenas et al., 2010; Baughman et al., 2011; De Stefani et al., 2011; Hall et al., 2014; Tosatto et al., 2016; Ren et al., 2017; Yu et al., 2017; Zhao et al., 2019; Liu et al., 2020). The oxygen consumption rates (OCR) under basal and stimulated conditions were quantified to determine if oncogenic transformation resulted in alterations of mitochondrial respiration (Figure 1I). Surprisingly, cell transformation was associated with only a slight, insignificant enhancement of basal respiration (Supplementary Figure S1I), although uncoupled maximal respiration was significantly increased (Supplementary Figures S1J, K). More importantly, acute stimulation of mitochondrial respiration by ATP activation of metabotropic purinergic receptors promoted a rapid and significant increase of OCR in transformed, but not immortalized fibroblasts (Figure 1J). These data suggest that an increase in transformation-associated MCU expression has no effect on basal mitochondrial respiration, whereas it promotes Ca²⁺-dependent mitochondrial respiration during acute stimulation.

To further establish the relationship between MCU expression and rates of mitochondrial Ca²⁺ uptake, we utilized HEK293T cells with MCU genetically deleted (MCU-KO). In addition, these cells were used to stably express human MCU to create an isogenic HEK293T MCU-rescue line (Supplementary Figure S1L). Rescue of MCU resulted in a ∼2-fold higher expression compared with WT HEK293T levels (Supplementary Figure S1M). Mitochondrial Ca²⁺ uptake in permeabilized cell suspensions was absent in MCU-KO cells whereas it was restored in cells re-expressing MCU (Supplementary Figure S1N). Of note, the rate of Ca²⁺ uptake was ∼2-fold faster in the MCU-rescue cells (Supplementary Figure S1O). There were no differences between ΔΨ_m of WT, MCU-KO and MCU-rescue cells (Supplementary Figure S1P), indicating that faster rates in MCU-rescue compared to WT cells were not due to differences in ΔΨ_m. These results suggest that observed higher levels of MCU expression in tumor cells are associated with enhanced rates of Ca²⁺ uptake in permeabilized cells and altered kinetic responses in intact cells.

We developed two models to more directly explore the roles of MCU in tumorigenesis in vivo. In the first, we generated tumor xenografts in immunodeficient NOD SCID mice injected with HEK293T-WT cells in one flank and either HEK293T MCU-KO or MCU-rescue cells in the other. When examined 4 weeks post-injection, all 3 cell lines had formed tumors. Compared with WT tumors, those formed by MCU-KO cells were considerably smaller (Supplementary Figure S2A): WT tumors had an average volume of ∼900 mm³ whereas MCU-KO tumors were ∼100 mm³ (Supplementary Figure S2B). Notably, the sizes of tumors generated by MCU-rescue cells were similar to those generated by WT cells (Supplementary Figure S2B). These results suggest that MCU is dispensable for tumor formation, whereas it plays an important role in tumor growth.

In the second model, we induced tumor xenografts in immunocompromised nude (J:Nu) mice using transformed fibroblasts. We genetically eliminated MCU (MCU-KO) by expressing mCherry-Cre recombinase in the Mcu^fl/fl-transformed fibroblasts and selecting mCherry-positive cells. Genetic deletion of MCU in clonal lines was validated by Western blot (Figure 2A) and by mitochondrial Ca²⁺ uptake assays (Figure 2B). Both cell lines formed tumors (Supplementary Figure S2C), but MCU-KO tumors were significantly (>60%) smaller (Figure 2C).

To explore the mechanisms underlying reduced tumor size associated with lack of MCU, we quantified cell death and the proliferation index of fibroblast tumor xenografts. Paraffin-embedded tumor slices were stained with DAPI, immunolabeled for Ki-67 as a marker of cell proliferation, and TUNEL (terminal deoxynucleotidyl transferase-mediated dUTP Nick-end labeled)-stained as a marker of apoptotic cell death (Supplementary Figure S2D). Notably, cell death was not enhanced in tumors formed by MCU-KO transformed fibroblasts compared with those formed by MCU-expressing transformed fibroblasts (Figure 2D). In contrast, the proliferation index of MCU-KO tumors was markedly lower than in the tumors of transformed fibroblasts (Figure 2E). To validate this, we analyzed small tumor-tissue sections and counted individual Ki-67⁺ cells (Supplementary Figure S2E). The proliferation potential of transformed tumor fibroblasts was strikingly decreased by elimination of MCU (Supplementary Figure S2F). These results suggest that MCU-KO does not eliminate the tumorigenic potential of transformed fibroblasts, whereas it markedly slows tumor growth primarily by strongly reducing cell proliferation with lesser enhancement of cell death.

To further understand the mechanism by which deletion of MCU impedes tumor growth, we performed histological analyses of tumors (Supplementary Figure S2G), examining characteristics associated with tumor growth patterns, including amount of normal and necrotic tumor tissue (Supplementary Figure S2H) and mitotic activity (Supplementary Figure S2I). Necrosis in tumors often indicates aggressiveness associated with high cell density due to rapid cell division with low availability of nutrients and anoxic conditions. In agreement, the smaller tumors formed by MCU-KO transformed fibroblasts were associated with significantly reduced size of necrotic areas (Figure 2F). The number of mitotic figures was considerably lower in MCU-KO tumors (Figure 2G), in agreement with the Ki-67 analysis. Some cells with morphological characteristics associated with senescence, namely markedly large cell size and/or polyploidy, were recognized as giant multinucleated cells (Supplementary Figure S2J). Of interest, a substantially higher number of multinucleated giant cells was observed in MCU-KO tumors (Supplementary Figure S2K).

The results from the in vivo transformed fibroblast tumor model suggests that MCU is required for cell proliferation to support tumorigenesis. We undertook a series of in vitro experiments to evaluate the effects of MCU depletion on cell proliferation, as well as other cellular phenotypes that drive cancer cell malignancy, including inhibition of cell death, cell division, sphere formation and matrix invasion.

The transformed fibroblasts proliferated significantly faster than the immortalized fibroblasts, as expected (Figure 3A). Notably, proliferation of the transformed fibroblasts was only moderately reduced by genetic deletion of MCU in both clonal MCU-KO lines examined (Figure 3A). MCU-KO mice have no apparent physiological phenotypes until they are physically stressed (Kwong et al., 2015; Luongo et al., 2015). Therefore, we examined proliferation of transformed fibroblasts under low-nutrient conditions, a stress imposed in the tumor microenvironment. Fibroblasts were initially seeded in high glucose (25 mM) and glutamine (4.5 mM) media, and after 24 h the media was changed to one with low glucose (4.5 mM) and glutamine (0.75 mM) that was sufficient to allow fibroblasts to proliferate and remain viable for a period of 4 days post-seeding. Under these conditions, genetic deletion of MCU much more strongly decreased proliferation of transformed fibroblasts (Figure 3B). To confirm the requirement of MCU expression for optimal cell proliferation, we generated MCU-rescue cell lines, which were validated by Western Blot (Supplementary Figures S3A, B) and by Ca²⁺ uptake assays (Supplementary Figures S3C, D). Proliferation was enhanced in both rescue cell lines in both nutrient-rich (Supplementary Figure S3E) as well as in nutrient-poor (Supplementary Figure S3F) conditions, confirming that MCU expression correlates with cell-proliferative capacity and ruling out the possibility that decreased proliferation of MCU-KO cells was due to non-specific effects of Cre expression.

To determine the mechanisms of reduced proliferation of transformed cells lacking MCU, we examined cell death and cell-cycle progression. Apoptotic and non-apoptotic cell death were assessed by annexin-V/DAPI staining and fluorescence-activated cell sorting (FACS) (Figure 3C). Populations of live, dead, apoptotic, and necrotic cells were established based on viable and 1 μM staurosporine-treated controls (Supplementary Figure S3G). There were no differences in the percent of either non-viable (early/late apoptotic and necrotic cells) or healthy cells between the immortalized and transformed populations (Supplementary Figure S3H). Genetic deletion of MCU diminished the percentage of healthy cells and moderately increased the number of early-apoptotic cells (Figure 3D). Thus, the major effect of MCU-KO in transformed cells was to decrease cell proliferation. Notably, these in vitro results are highly consistent with the conclusions reached in the in vivo model.

To understand the basis for reduced proliferation observed in vivo and in vitro, we examined the effects of MCU deletion on cell-cycle progression. Fibroblasts were synchronized in the G1 phase using lovastatin (Keyomarsi et al., 1991), and then examined by DAPI-FACS 24 h after release from synchronization. Unsynchronized and lovastatin-synchronized samples provided controls (Supplementary Figure S3I). At 24 h after release from synchronization, a significant fraction of the transformed cells progressed through mitosis into the G2 phase of the cell cycle (Figures 3E, F). In contrast, both MCU-KO clonal cell lines contained a significantly lower percent of cells in G1 phase and a much higher percentage in S phase (Figures 3E, F). Accumulation of MCU-KO cells in the S phase suggests that MCU is important for cell proliferation by promoting progression through the cell cycle (Cardenas et al., 2016; Koval et al., 2019; Zhao and Pan, 2021).

The ability to form clonally-derived spheres on non-adherent substrates is a stem cell-like capacity related to metastatic tumor initiation and progression (Uchida et al., 2010). In a sphere-formation assay, the transformed fibroblasts, but not the immortalized cells, readily formed spheres (>30 spheres per well) (Figure 4A), consistent with their differential abilities to form tumors in vivo (Figure 1B). In contrast, MCU-KO transformed fibroblasts formed <5 spheres per well (Figure 4A). These results suggest that MCU plays an important role in both cell proliferation as well as the capacity to self-renew.

To explore the role of MCU in cell chemotaxis and extracellular matrix invasion, processes associated with cancer metastasis in which Ca²⁺ signaling has been proposed to play a role (Tang et al., 2015; Tosatto et al., 2016), we employed a transwell-invasion assay. Cells were seeded atop a porous Matrigel-coated membrane and examined after 24 h (Supplementary Figure S4A). The number of transformed fibroblasts that penetrated through the coated membrane was ∼3-fold greater than the number of immortalized fibroblasts that traversed it (Figure 4B). Importantly, genetic deletion of MCU in transformed fibroblasts markedly reduced the number of invading cells by ∼50% (Figure 4B).

Together, these results validate our transformed fibroblasts as a model that recapitulates many phenotypes associated with malignancies, including enhanced proliferation, ability to form clonally-derived spheres, and cell migration and tissue invasion. Importantly, elimination of MCU strongly suppressed these in vitro phenotypes, suggesting that it could play an important role in carcinogenesis, as observed in our in vivo models.

To investigate mechanisms by which elimination of MCU affects cancer-progression phenotypes, we examined several mitochondrial functions. First, we examined total dehydrogenase activity using the XTT assay. Oncogenic transformation significantly increased dehydrogenase activity, but genetic deletion of MCU was without effect (Figure 5A). Similarly, ROS production was decreased after transformation but was not changed by deletion of MCU (Figure 5B). ΔΨ_m was not significantly different after transformation or in MCU-KO fibroblasts (Figure 5C). In addition, mitochondrial matrix [Ca²⁺] ([Ca²⁺]_mit) was also not different after transformation or in MCU-KO cells (Figure 5D). Together these findings indicate that reduced malignancy of MCU-KO transformed fibroblasts is not associated with reduced mitochondrial function.

To further explore the role of MCU in cellular bioenergetics, we quantified OCRs (Figure 5E). Unexpectedly, neither basal nor maximal respiration of transformed fibroblasts was affected by genetic deletion (Figure 5F) or rescue (Supplementary Figures S5A, B) of MCU. Spare respiratory capacity, proton leak, coupling efficiency, ATP production, and non-mitochondrial OCR were also not altered by transformation, MCU-KO or MCU-rescue (data not shown). Nevertheless, diminished phosphorylation of PDH associated with transformation (Figure 2G) was strongly suppressed by genetic deletion of MCU (Figures 5G, H) and enhanced by MCU-rescue (Supplementary Figures S5C, D). Although genetic deletion of MCU did not affect OCR, it significantly increased glycolysis (Figures 5I, J) and glycolytic capacity (Supplementary Figure S5E). The glycolytic reserve of fibroblasts was increased after transformation and remained unchanged after MCU-KO (Supplementary Figure S5F). No changes were observed in non-glycolytic acidification rates after transformation or deletion of MCU (data not shown). Thus, the major effects of MCU deletion on cellular bioenergetics in vitro was suppression of PDH phosphorylation and upregulation of glycolysis, while mitochondrial respiration remained unaffected.

Altered metabolism is a fundamental characteristic of cancer cells (Hanahan and Weinberg, 2011). Cancerous cells metabolize large quantities of glucose into lactate to support rapid energy and biomass production for cellular growth and proliferation (Hatzivassiliou et al., 2005; DeBerardinis et al., 2008a; Deberardinis et al., 2008b; Lunt and Vander Heiden, 2011; Ganapathy-Kanniappan and Geschwind, 2013; Liberti and Locasale, 2016). Indeed, the observed increase in ECAR in MCU-KO cells was associated with increased glucose uptake (Figure 6A) and lactate production (Figure 6B), effects that were attenuated by MCU re-expression (Figures 6A, B). When the rate of lactate production was normalized to glucose uptake rate, there were no significant difference between cell lines (Figure 6C), suggesting that glucose is largely metabolized to lactate through glycolysis in MCU-KO cells. To determine if increased glucose consumption in MCU-KO cells alters cellular metabolism, we performed metabolic tracing with ¹³C₆ d-Glucose (Figure 6D). Fractional labeling all metabolites included in our analysis is shown in Supplementary Table S1. The network depicted in Figure 6D represents metabolic routes, labeling pattern of ¹³C₆ d-Glucose-derived metabolites, and fractional labeling of isotopologues of interest. Glucose entry into the TCA cycle via acetyl-CoA and PDH leads to labeling of TCA cycle intermediates with two ¹³C, represented by m+2. Alternatively, entry of glucose-derived ¹³C into the TCA cycle via pyruvate carboxylation by pyruvate carboxylase (PC) or malic enzyme (ME) results in the incorporation of three ¹³C in TCA-cycle intermediates, represented by m+3. Labeling of lactate in aerobic glycolysis (m+3) was significantly increased in MCU-KO as compared to transformed cells (Figure 6D). Notably, labeling of m+3 serine and m+2 glycine was also increased in transformed MCU-KO cells. This phenotype was reversed by MCU-rescue (Figure 6D), suggesting that deletion of MCU promotes the diversion of glucose-derived carbons into biosynthetic pathways, such as for purine biosynthesis (Vanhove et al., 2019). In all transformed cell lines, m+2 citrate was ∼30% of the labeling from glucose, while labeling of m+2 fumarate and malate was ∼5–10%, suggesting exit of citrate from the TCA cycle to fuel biosynthetic reactions such fatty-acid synthesis. Interestingly, m+2 glutamate, fumarate, malate, and aspartate levels were not different between transformed and MCU-KO cells, although rescue of MCU enhanced their labeling beyond those observed in the wild-type transformed cells (Figure 6D). This result is consistent with the observed lack of effect of MCU deletion on basal OCR. Notably, elevated levels of m+3 fumarate, malate, and aspartate indicate that entry of pyruvate through alternative pathways, likely PC, was significantly enhanced by MCU-KO (Figure 6D). These results suggest a striking metabolic adaptation in MCU-KO cells that promotes serine biosynthesis and TCA cycle activity.

Many cancer cells shift their substrate preference to fuel flux through the TCA cycle (Reitzer et al., 1979). Glutamine provides another key carbon source for the TCA cycle through anaplerosis (DeBerardinis et al., 2007; Wise et al., 2008). We therefore evaluated the contribution of glutamine to the TCA cycle. The schematic in Figure 6H represents the labeling pattern of ¹³C₅ l-Glutamine-derived TCA-cycle intermediates, γ-aminobutyric acid (GABA), and fractional labeling of isotopologues of interest. Isotopic labeling with ¹³C is represented by green circles, while naturally-occurring ¹²C is depicted as black circles. Glutamine anaplerosis by oxidation of α -KG results in a m+4 mass increase of TCA-cycle intermediates. On the other hand, partial reverse flow of the TCA cycle for reductive carboxylation of α -KG by enzymatic activity of Ca²⁺-independent IDH2 leads to formation of m+5 citrate-isocitrate. Fractional labeling of all metabolites included in our analysis is shown in Supplementary Table S2. Genetic deletion of MCU was associated with an increased glutamine uptake (Figure 6E) and its conversion to glutamate (Figure 6F). With glutamate production normalized to glutamine uptake rate, there was no difference between cell lines (Figure 6G), suggesting that MCU deletion was primarily associated with increased nutrient uptake rather than involvement of alternative glutamine metabolic pathways. In agreement, ¹³C₅-glutamine tracing revealed increased labeling of m+5 glutamate, m+4 aspartate, m+4 citrate-isocitrate, and m+4 malate in MCU-KO vs. WT transformed cells, which was reversed by MCU re-expression (Figure 6H). We also found a significant increase in the diversion of glutamine-derived carbons into the GABA shunt in MCU-KO cells and significant decrease with MCU-rescue (Hoang et al., 2021). The GABA shunt can serve as a reservoir for TCA cycle anaplerosis by promoting entry of glutamine-derived carbons through succinate and by passing α -KGDH (Figure 6H). Taken together, these data suggest that in the absence of MCU, glucose metabolism via glycolysis is enhanced, and cells rely more on glutamine to maintain TCA-cycle integrity.

Because mitochondria can play an important role in buffering changes of [Ca²⁺]_cyt, and alterations in [Ca²⁺]_cyt may regulate cell-biological functions of tumor cells, we also examined InsP₃R-mediated Ca²⁺ signaling in transformed fibroblasts and two clones with MCU deleted. Exposure of cells to ATP elicited three types of responses: a single transient spike, oscillations, and a sustained rise (Figure 7A). In transformed fibroblasts, 32% of cells responded with a sustained elevation of [Ca²⁺]_cyt, 47% displayed [Ca²⁺]_cyt oscillations, and 21% responded with a single [Ca²⁺]_cyt spike (Figure 7B). In contrast, a sustained elevation was rarely observed in transformed fibroblasts lacking MCU, with cells responding with either single spikes (∼41%) or oscillations (∼56%) (Figure 7B). The amplitude of the first [Ca²⁺]_cyt peak was elevated in the cells lacking MCU (Figure 7C). Among the oscillatory responses (Figure 7D), those in the cells with MCU deleted were of lower frequency compared with those of transformed fibroblasts (Figure 7E). Taken together, these results suggest that [Ca²⁺]_cyt signaling is altered in transformed cells by deletion of MCU.