Cells and cell culture conditions

L1210, BaF3, DT40, Fl5.12, and primary T cells were cultured in RPMI that contained 10% FBS (Gibco), 11 mM glucose, 2 mM glutamine, 1 mM Na pyruvate, 20 mM HEPES, and antibiotic/antimycotic. In addition, Fl5.12 culture media was supplemented with 10 ng/ml IL-3 (R&D Systems), DT40 culture media was supplemented with 3% chicken serum (Sigma-Aldrich) and primary T-cell culture media was supplemented with 10 mM 2-mercaptoethanol, 100 U/ml IL-2 (R&D Systems) and 2 mg/ml anti-human CD28 (BioLegend). S-HeLa cells were grown in DMEM that was supplemented with 10% FBS, 1 mM sodium pyruvate, and antibiotic/antimycotic. L1210 cells were also grown in modified RPMI when comparing growth conditions (Supplementary Fig. 3a, b). For these experiments, we used RPMI containing 4% FBS and either high (25 mM) glucose, low (4 mM) glucose, or 10 mM galactose with no glucose.

Isolation of naive CD3+ and CD8+ primary T cells was carried out from an unpurified buffy coat (Research Blood Components) from which PBMCs were isolated using Ficoll-Paque Plus density gradient (GE). After isolation of the PBMCs, the cells were subjected to red blood cell lysis using ACK lysis buffer (Thermo Fisher Scientific). The PBMCs were then washed three times, and T cells were isolated using Naive CD3⁺ or CD8⁺ T Cell Isolation Kit (Miltenyi Biotec), according to kit instructions. The cells were then activated by culturing the cells on an anti-CD3 coated cell culture plate in the media detailed above. T cells were used for experiments approximately 30 h after activation. The activation of T cells was validated by monitoring cell volumes and counts using coulter counter (Beckman Coulter).

Within all SMR experiments, the culture media used was identical to that described above, apart from any indicated fluorescent probe and/or chemical inhibitor addition. All cell culture reagents were obtained from Invitrogen, unless otherwise stated. L1210 cells we obtained from ATCC (Cat# CCL-219), BaF3 cells were obtained from RIKEN BioResource Center (Cat# RCB4476), DT40 cells, which harbor a CDK1as⁶³, were a gracious gift from K. Samejima and B. Earnshaw from University of Edinburgh, Fl5.12 cells were a gracious gift from M. Vander Heiden from Massachusetts Institute of Technology, and S-HeLa cells were a gracious gift from K. Elias from Brigham Women’s Hospital. All cell lines were tested to be free of mycoplasma.

Generation of reporter cell lines

The FUCCI cell cycle marker expressing L1210 cells were generated in a previous study²⁹. Cells that stably express an ATP reporter (A-Team), reactive oxygen species reporter (roGFP2-Orp1), glutathione redox potential reporter (Grx1-roGFP2) or mitochondria localized RFP (mito-RFP) were generated using lentiviral vectors obtained from AddGene^{64,65,66,67,68}. pEIGW roGFP2-Orp1 was a gift from Tobias Dick (Addgene plasmid #64993; http://n2t.net/addgene:64993; RRID:Addgene_64993), pEIGW Grx1-roGFP2 was a gift from Tobias Dick (Addgene plasmid # 64990; http://n2t.net/addgene:64990; RRID:Addgene_64990), ATeam1.03-nD/nA/pcDNA3 was a gift from Takeharu Nagai (Addgene plasmid # 51958; http://n2t.net/addgene:51958; RRID:Addgene_51958), pclbw-mitoTagRFP was a gift from David Chan (Addgene plasmid # 58425; http://n2t.net/addgene:58425; RRID:Addgene_58425). Lentiviruses for the genetically encoded sensors were produced by calcium phosphate transfection of plasmids into HEK-293T cells. For each sensor, the corresponding viral vector was co-transfected with separate plasmids expressing VSV-G and Gag/Pol. Following a media change after an overnight transfection, virus aliquots were harvested at 48 h and 72 h post transfection and filtered through 0.45-µm filters. Lentiviruses containing Ca²⁺ reporter (GCaMP3)⁶⁹ vectors were obtained from Kerafast (cat# FCT188). All metabolic sensors were expressed under strong promoters, but FUCCI cell cycle reporter was expressed under the weak endogenous Geminin promoter.

Lentiviruses were transfected into L1210 and BaF3 cells using spinoculation. Approximately 1.5 × 10⁵ cells were mixed with 10 mg/ml polybrene (EMD Millipore) and the lentiviruses. This mixture was centrifuged at 800 g for 60 min at 25 °C, after which the cells were moved to normal culture media for 12 h. This was repeated four times. Twenty-four hours later, the selection process was started with puromycin or neomycin, depending on the transfected vector. Following 5 days of selection, a subpopulation with high fluorescence reporter expression was sorted out using BD FACS Aria.

Suspended microchannel resonator (SMR) setup

The SMR devices were fabricated at CEA-LETI, France. The exact dimension and geometry of the device can be found in ref. ²³. The SMR devices were actuated the second vibration mode by a piezo-ceramic placed underneath the chip. The SMR devices were operated in a closed-loop, where the output motion of the cantilever is amplified, delayed, and fed back to the piezo-ceramic to drive the cantilever. The signal delay was chosen so that the amplitude of the cantilever motion was maximized. The amplification was set to minimize the resonant frequency error, but not in the limit where it distorts the linearization of the resonant frequency. A digital platform was implemented to track the change in resonant frequency (frequency of the closed-loop signal) over time. This closed-loop operation configuration results in measurement bandwidth ~1500 Hz, wide enough to capture the frequency modulation during the cell transit event, which is typically set to ~150–200 ms. The resonance frequency changes were analyzed and converted to buoyant mass as detailed in refs. ^23,28. The serial SMR experiments (Fig. 3e) were carried out as detailed previously^24,41,70.

Optical detection setup for SMR

The optical setup used in this study is similar to our previous setup²³, except a few additions/modifications in optical components to achieve simultaneous measurements of TMRE and FUCCI signals. First, we implemented a “full multiband configuration”, which is composed of a multiband exciter filter (Semrock, FF01- 387/485/559/649-25), a multiband emission filter (Semrock, FF01- 440/521/607/694/809-25), and a multiband dichroic beam splitter (Semrock, FF 408/504/581/667/762-Di01-25×36). This filter set cube was mounted in our microscope (Nikon). Second, a dichroic beam splitter (Semrock, FF580-FDi01-25×36) was mounted beyond the image plane to separate the multi-color fluorescence output signal into single-color fluorescent signals (i.e., FUCCI and TMRE signals). The split fluorescent signals were measured by photomultiplier tubes (Hamamatsu, H10722-20) mounted on top of the microscope with single band-pass filters (Semrock, FF01-520/35 for FUCCI, Semrock, FF01-607/35 for TMRE, respectively). Third, to minimize phototoxicity from long-term measurements, the excitation light source was kept on for <500 ms for each measurement, during which the cell was flown through the excitation light path. The cell was directly exposed to the light for ~50 ms during each measurement. This fluorescence measurement was carried out approximately every 2 min (in conjugation with every second buoyant mass measurement) (Supplementary Fig. 1a, b). The area of light exposure was limited 60 × 60 µm, and the area of emission collection was limited to 40 × 60 µm. We did not observe changes to single-cell growth rates when measuring TMRE signal with this setup. All parts of the optical detection setup (including objective lens: 50×/0.55-NA Nikon-CFI, LU Plan ELWD WD 10.1 mm; rectangular slit for emission area control: Thorlab; light source: Lumencor, Spectra X Light Engine) were identical to those used before²³. The optical path for simultaneous two-color measurements is similar to what has been shown in the previous work²⁹.

System operation and error quantification

Buoyant mass and fluorescence signals were continuously measured using previously explained hydrodynamic trapping approach^23,24,28,29. The hydrodynamic trapping, SMR measurements, and optical operation were controlled using LabVIEW 2012 software and are explained in more detail in Supplementary Fig. 1a, b.

We quantified the error in optical measurement by repeatedly measuring fluorescently labeled fixed cells. We then carried out linear fitting to the data and calculated the average deviation of each measurement from a linear fitting. The signal-to-noise ratio displayed in Supplementary Fig. 1c, d was calculated by dividing the mean fluorescence intensity of each cell by the standard deviation of the measurements from the fitted line. Using this approach, we estimated our optical measurements to have an error (CV) of approximately 2% for mitotic cells. See Supplementary Fig. 1c, d for details. The error in buoyant mass measurement was quantified previously (by repeatedly measuring a 10-µm polystyrene bead) to be <0.1 pg, which corresponds to an error of <0.2%²⁴.

Detecting cell cycle transitions within the SMR

Three cell cycle transitions were assigned for each cell in the SMR data. First, the G2/M transition was detected using the node deviation signal (Supplementary Fig. 5c), which is an acoustic, stiffness-dependent signal detected by the SMR²³. This timing was also validated by measurements of mitotic cell swelling (Fig. 1e and Supplementary Fig. 5b), which start immediately following mitotic entry^26,27. Single-cell density was measured as detailed in ref. ²⁶. Second, the metaphase–anaphase transition was detected using the node deviation signal (Supplementary Fig. 5c) and also using the rapid change in buoyant mass accumulation rates²⁴. The timing of metaphase–anaphase transition was further validated using the FUCCI cell cycle sensor expressing cells (Fig. 1f, g and Supplementary Fig. 5d). The imaging of FUCCI cells (Fig. 1f) was carried out using IncuCyte. Third, the abscission of the daughter cells was assigned using the sudden ~50% reduction in buoyant mass (Fig. 1a). In addition, we assigned anaphase to the last 15 min (Fig. 4e), as based on our previous quantifications of L1210 cell-elongation duration²³.

Membrane potential and mitochondrial content measurements

For all cells, mitochondrial membrane potential was examined using TMRE (Invitrogen) in 10 nM (non-quenching) concentration, with the exception that DT40 cells and primary T cells were measured using 20 nM TMRE. TMRE was detected on the TRICT channel of our SMR setup (see above). L1210 cells at a confluency of ~300.000 cells/ml were stained with 10 nM TMRE for 30 min, after which cells were loaded into the SMR, where the cells were grown in the presence of 10 nM TMRE. Mitochondrial membrane potential was also measured using 10 μM (quenching) concentrations of the Rhod123 (Invitrogen). Rhod123 was detected on the FITC channel of our SMR setup (see above). Cells were stained for 45 min with 10 μM Rhod123, washed twice with media and loaded into the SMR. No Rhod123 was present in the media within the SMR. Note that Rhod123 slowly diffuses out of the cell and this excludes long-term experiments. In order to observe mitosis before Rhod123 diffused out of the cell, we carried out Rho123 experiments by loading only large G2 cells into the SMR. For both TMRE and Rhod123, the quenching and non-quenching states were validated by staining L1210 cells as described above and quantifying fluorescence levels after 30 min of treatment with 1 µM FCCP or oligomycin using BD Biosciences flow cytometer LSR II HTS with excitation lasers at 488 nm and 561 nm, and emission filters at 530/30 and 585/15.

When analyzing the error of mitochondrial fluorescence measurements, L1210 cells were stained for 30 min with MitoTracker Red CMXRos (Invitrogen), after which cells were washed with PBS and fixed in 4% PFA for 10 min. The cells were then washed, and the mitochondrial signal of each cell was repeatedly measured in SMR using the TRICT channel of our SMR setup (see above).

Mitochondrial content was examined using the MitoTracker Green probe (Invitrogen) in 50 nM concentration. MitoTracker Green was detected on the FITC channel of our SMR setup (see above). As with TMRE, cells were stained for 30 min after which cells were loaded into the SMR. MitoTracker Green was present in the culture media within SMR. Note that in these concentrations, the presence of TMRE and MitoTracker Green did not affect cell growth rate.

Plasma membrane potential was examined using Bis-(1,3-dibutylbarbituric acid) trimethine oxonol (DiBAC4(3)) probe (Invitrogen) in 1 μM concentration. DiBAC4(3) was detected on the FITC channel of our SMR setup (see above). Cells were stained for 45 min, after which cells were loaded into the SMR. Within the SMR, cells were grown in the presence of DiBAC4(3).

When using a flow cytometer to measure live cell mitochondrial or plasma membrane potential, TMRE and DiBAC4(3) were used at 10 nM and 1 µM concentrations, respectively, and cells were analyzed in the presence of the fluorescence probes. When using a flow cytometer to measure mitochondrial membrane potential changes between G2 and mitosis, L1210 cells were stained for 30 min with 100 nM MitoTracker Red CMXRos (Invitrogen), after which cells were washed with PBS and fixed in 4% PFA for 10 min. The fixed cells were stained for DNA and for p-Histone H3 or Cyclin B as detailed below (“cell cycle synchronizations”) and in ref. ²⁴. The cell cycle marker and MitoTracker Red CMXRos labeling were quantified using BD Biosciences flow cytometer LSR II HTS. See Supplementary Fig. 5e, f, for example, FACS plots with gating.

Mitochondrial imaging and image analysis

For all imaging experiments, L1210 cells were grown under identical conditions to those used in SMR experiments. The cells were stained with PicoGreen DNA dye (Thermo Fischer Scientific) for 1 h under normal growth conditions. Note that PicoGreen stains both nuclear and mitochondrial DNA. When imaging TMRE-stained mitochondria, cells were stained for an additional 30 min with 20 nM TMRE. Then, the cells were washed twice with culture media and plated on coverslips coated with 0.1% poly-L-lysine. For TMRE imaging, 20 nM TMRE was maintained in the media during plating and imaging. Imaging was carried out at 37 °C and started 30 min after plating. Imaging was carried out using a DeltaVision wide-field deconvolution microscope with standard filters (FITC, TRITC) and ×100 oil-immersion objective. We utilized immersion oil with a refractive index of 1.516 (Cargille Laboratories). For each cell, we collected ~75 z layers, each 0.2-µm apart. The image xy resolution was ~0.1 µm, and z resolution 0.2 µm. Ten rounds of image deconvolution were carried out using SoftWoRx 7.0.0 software.

Images were analyzed using ImageJ (version 2.0.0-rc-69/1.52p). Each cell was cropped out into its own image stack, and the top and bottom z layers which did not contain the cell were removed. The TRITC channel with mito-RPF signal was converted to 8-bit grayscale image and then converted to a binary image based on a user-defined threshold value. We extracted the pixel intensity for each image stack and observed that mitotic cells displayed lower total intensities. To avoid bias in mitochondrial network morphology analysis, the threshold value used for conversion to the binary image was based on a visual impression for reproducing the mitochondrial network. To extract the three-dimensional mitochondrial network across the z stack from the binary images, the Skeletonize (2D/3D) and Analyze Skeleton (2D/3D) functions of ImageJ were used. Out of every image stack, the total length of the three-dimensional mitochondrial network along with the average length and the total number of mitochondrial structures were extracted.

Chemical perturbations

Chemical treatments within the SMR were carried out by adding the chemical of interest to the culture media within the SMR. Thus, cells were exposed to the chemical when loading the cells into SMR, which in a typical experiment was approximately 2 h prior to mitosis. Where separately indicated (Fig. 2f, g), chemicals were injected into the SMR culture media during the experiment. All SMR experiments with chemical treatments were stopped following mitotic exit so that each replicate with chemical inhibitors represents a completely independent experiment.

Unless otherwise indicated, chemical perturbations were done using the following chemical concentrations: 1 µM oligomycin, 1 µM FCCP, 1 µM TPB (Sigma-Aldrich), 5 µM EIPA (Sigma-Aldrich), 1 µg/ml nocodazole (Sigma-Aldrich), 5 µM STLC (Sigma-Aldrich), 15 µM proTAME, 1 µM RO-3306, 400 nM BMS-265246, 100 nM okadaic acid, 2 µM rotenone, 2 µM antimycin A, and 2 mM thymidine. All chemicals were diluted in DMSO, except okadaic acid, which was diluted in EtOH. Unless otherwise indicated, all chemical inhibitors were obtained from Cayman Chemicals. For cell cycle targeting chemicals, treatment efficiency was validated by examining the cell cycle profile.

Validating CDK1 activity levels

CDK1 activity was measured using two approaches, western blotting and flow cytometry, both based on labeling proteins using MPM2 antibody (monoclonal mouse antibody, EMD Millipore, #05-368). L1210 cells were first treated with chemicals indicated in Fig. 2b for 30 min, after which cells were washed with PBS and immediately lysed in XT sample buffer (Bio-Rad). Samples were sonicated to breakdown DNA, mixed with XT reducing agent (Bio-Rad), and loaded on a 4–12% Bis-Tris Criterion XT precast gel (Bio-Rad). Proteins were separated by running the gels in MES buffer (Rio-Rad) for 45 min at 180 V. The proteins were then transferred on a nitrocellulose membrane by blotting for 2 h at 40 V in Tris/Glycine buffer (Bio-Rad). The membrane was blocked by incubating with 5% BSA in TBST for 30 min in RT. CDK1 target protein phosphorylation was detected using MPM2 antibody (used at 1 μg/ml) and β-tubulin III was used as a loading control (polyclonal rabbit antibody, Sigma-Aldrich, #T2200) (used at 0.4 μg/ml). MPM2 and β-tubulin antibodies were labeled using fluorescent secondary antibodies (IRDye 680RD goat anti-rabbit secondary antibody, LI-COR, #925-68071; and IRDye 800CW donkey anti-mouse secondary antibody, LI-COR, #925-32212). The primary antibodies were incubated o/n in +4 °C, followed by three washes with TBST and a 1-h incubation with secondary antibody. Following TBST washes of the secondary antibody, the membrane was imaged using LI-COR Odyssey detection system. For flow cytometry-based quantifications, cells were treated with chemicals as for western blotting, washed with PBS, fixed in 4% PFA for 10 min and permeabilized with 0.5% Triton X-100 for 10 min. The cells were then washed with PBS and blocked with 5% BSA in PBS for 30 min. This was followed by MPM2 antibody labeling (used at 10 μg/ml) o/n in +4 °C. Following PBS washing, the MPM2 antibody was then labeled using Alexa Fluor 488-conjugated anti-mouse secondary antibody (Cell Signaling Technology, #4408) for 1 h at RT, after which the cells were washed three times with PBS. The MPM2 labeling was quantified using BD Biosciences flow cytometer LSR II HTS.

Cell cycle synchronizations

In L1210 and BaF3 cells, the cell cycle was synchronized to G2 using a double-thymidine block followed by G2 arrest. Briefly, cells were treated with 2 mM thymidine for 15 h, washed twice with PBS, cultured for 6 h in normal culture media, retreated with 2 mM thymidine for 6 h, washed twice with PBS, returned to normal culture conditions for 3 h and treated with 5 μM RO-3306 for 7 h. At this point, ~80–90% of cells were arrested in G2, and upon removal of RO-3306 part of the cells enter mitosis immediately²⁴.

The cell cycle status of L1210 cells was analyzed as follows: The cells were washed with PBS, fixed in 4% PFA for 10 min, washed with PBS, permeabilized with 0.5% Triton X-100 for 10 min, washed with PBS and blocked with 5% BSA in PBS for 30 min. The cells were then stained with p-Histone H3 (S10) antibody (D2C8, conjugated to Alexa 488, Cell Signaling Technology, #3465 S) in a PBS solution containing 5% BSA o/n at +4 °C. The p-Histone H3 antibody was used in the concentration recommended by the supplier. The following day the cells were washed with PBS and stained with 1:2000 dilution of NuclearMask Blue (#H10325, Thermo Fisher Scientific) for 30 min in RT. Finally, the cells were washed three times with PBS, mixed into PBS supplemented with 1% BSA, and put on ice until FACS analysis. The relative proportions of mitotic (4 N DNA content, p-Histone positive), G2 cells (4 N DNA content, p-Histone negative) and G1&S cell (<4 N DNA content, p-Histone negative) was quantified using BD Biosciences flow cytometer LSR II HTS with excitation lasers at 355, 488, and 561 nm, and emission filters at 450/50, 530/30, and 585/15. For quantification of cell cycle status in BaF3 and DT40 cells, we used propidium iodide (PI) staining of DNA in cells fixed with ice-cold 70% EtOH o/n. DNA content was quantified using BD Biosciences flow cytometer LSR II HTS. See (Supplementary Fig. 9a, b) for representative FACS plots with gating.

Oxygen-consumption measurements

Oxygen consumption was measured using the XF24 Seahorse Extracellular Flux Analyzer. The Seahorse cell culture plates were coated with poly-L-lysine, and ~80,000 L1210 cells were loaded into each well. The experiment was started 3 h after plating the cells. The oxygen-consumption measurements were carried out in normal cell culture media (see above) to maximize comparability to the conditions within SMR. Oxygen consumption was measured four times with 6-min intervals, after which cells were treated with 1 μM oligomycin and, four measurements later, with 2 μM rotenone and 2 μM antimycin A. After the experiment, cells in each well were counted, and this cell count was used to normalize the data.

When measuring oxygen consumption in cells arrested in G2 and mitosis, G2-arrested cells were washed and moved back into 5 µM RO-3306 containing media (to maintain G2 arrest) or into 5 µM STLC containing media (to achieve prometaphase arrest)²⁴. The cells were then plated on to the Seahorse cell culture plates, and 2 h later the oxygen-consumption measurement was carried out as described above. During the oxygen-consumption measurements, parallel cultures were processed for FACS-based cell cycle analysis as described above. The relative proportions of G2 and mitotic cells in the populations were used to normalize the oxygen-consumption data to account for incomplete cell cycle synchrony.

Lactate efflux rate measurements

Lactate efflux rate was measured from cells arrested in G2 and from cells released in free mitosis by quantifying lactate amounts in media using CG-MS. Cells were first synchronized to G2 using a double-thymidine block followed by a RO-3306-mediated G2 arrest, as described above. To release cells from G2 arrest, cells were spun down 500g at 37 °C for 5 minutes and washed once in media lacking RO-3306. Cells were spun down once more at 600 × g for 1 min and resuspended in RPMI with 11 mM glucose, 2 mM glutamine, 1 mM U-¹³C-pyruvate, 10% dialyzed FBS, and penicillin/streptomycin with 10⁶ cells/ml. One flask of cells of each biological replicate (separate synchronization) was immediately spiked with 5 μM RO-3306 to maintain G2 arrest. Media samples were collected at 19, 32, and 45 minutes after release from G2 arrest by spinning the sample at 500 × g for 3 minutes at 4 °C to pellet cells. The supernatants were transferred to a new tube and stored at −20 °C. Parallel cell cultures were used for cell counting and for FACS-based cell cycle analysis as described above.

For GC-MS, 150 μl of cell-free media was extracted in 800 μl of methanol containing 1.25 μM ¹³C1-sodium lactate to serve as an internal standard. Samples were spun down at 15,000 × g for 10 min at 4 °C and 400 μl of the supernatant was transferred to a new tube for derivatization. Samples were dried down under nitrogen gas. Dried samples were derivatized with 16 μl of methoxamine (MOX) reagent (Thermo Fisher, #TS-45950) and 20 μl of N-tert-butyldimethylsilyl-N-methyltrifluoroacetamide with 1% tert-butyldimethylchlorosilane (Sigma-Aldrich, #375934). Following derivatization, samples were analyzed using a DB-35MS column (30 m × 0.25 mm i.d. × 0.25 μm, Agilent J&W Scientific) in an Agilent 7890 gas chromatograph (GC) coupled to an Agilent 5975C mass spectrometer (MS). Data were analyzed and corrected for natural isotope abundance using in-house algorithms. The lactate efflux rate was obtained by a linear fit across the three timepoints for each independent replicate. Cell counts and the relative proportions of G2 and mitotic cells in the populations were used to normalize the data to account for incomplete cell cycle synchrony.

ATP- and ADP-level measurements

For ATP- and ADP-level measurements, cells were synchronized and processed identically to that used for lactate efflux rate analysis above. Thirty minutes after releasing the cells from G2 arrest, 100 μl of media containing 10⁶ cells/ml were collected and flash-frozen. Intracellular ATP and ADP measurements were collected on 10 μl of thawed media in technical duplicates using an enzymatic, Luciferase-based assay according to the manufacturer’s instructions (Sigma-Aldrich, #MAK135). Measurement linearity was established using an external ATP standard. Cell counts and the relative proportions of G2 and mitotic cells in the populations were used to normalize the data to account for incomplete cell cycle synchrony.

Data analysis

All details of the electrical circuit model and data analysis for the model can be found in (Supplementary Notes 3–8). The modeling and SMR & fluorescence data processing were carried out using custom MATLAB codes. Quantifications of TMRE increase during mitosis was carried out by normalizing the TMRE signal of each data point to the median signal a 3 h time period prior to daughter cell abscission, after which the highest TMRE value during mitosis was used for quantification.

Statistics

In population-level experiments, such as cell cycle analyses or oxygen-consumption measurements, each replicate represents an independent culture. In control SMR experiments, cell progenies were monitored for various durations, so that each independent experiment can have one or more mitotic events. By contrast, all chemical treatments within the SMR were carried out so that only one mitotic event was analyzed for each experiment. Details of replicate numbers can be found in figures and figure legends. The experiments used for modeling were repeated over 30 times. Importantly, as some data were excluded from the modeling, as cells did not reach a steady TMRE baseline prior to cell division. For full details of data exclusion, please see (Supplementary Note 5). Statistical significances are indicated in each figure and the statistical details can be found in the figure legends. For statistical analysis of the electrical circuit model results, please see (Supplementary Note 8). All statistical analyzes were carried out using MATLAB or Origin Pro 2019.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.