Demonstration of beat-to-beat, on-demand ATP synthesis in ventricular myocytes reveals sex-specific mitochondrial and cytosolic dynamics

Abstract

The energetic demands on ventricular myocytes imposed by the transport of ions and cross-bridge cycling are well known, yet the spatiotemporal dynamics of ATP supply and demand remain poorly understood. Here, using confocal microscopy and genetically encoded fluorescent sensors targeted to mitochondria and cytosol, we visualized beat-to-beat ATP dynamics in ventricular myocytes from male and female mice. These probes showed fluctuations in mitochondrial ATP levels with each contraction, revealing two distinct, spatially localized waveforms – ATP ‘gain’ and ATP ‘dip’ – representing transient increases or decreases in matrix ATP levels, respectively. These waveforms were tightly phase-locked to intracellular Ca²⁺ transients and organized into energetic microdomains. Inhibition of the mitochondrial Ca²⁺ uniporter or the adenine nucleotide translocase attenuated these ATP transients. Although female myocytes exhibited larger mitochondrial ATP transients than their male counterparts, their mitochondrial volume was lower. Female myocytes also exhibited tighter coupling between the sarcoplasmic reticulum and mitochondria and showed a higher density of mitofusin 2 and ATP synthase catalytic α-subunit per unit volume, suggesting more efficient ATP production. Cytosolic ATP transients mirrored mitochondrial waveforms and domain structure in both male and female myocytes. During faster pacing, diastolic cytosolic ATP rose more rapidly in female myocytes, whereas beat-locked ATP transients increased in both sexes but proportionally more in males than in females. These findings demonstrate that ATP is synthesized on a beat-to-beat basis in a modular, microdomain-specific manner. We propose that male myocytes rely on greater mitochondrial mass for energetic scaling, whereas female cells employ architectural precision to optimize ATP delivery.

Graphical Abstract

Beat-locked mitochondrial ATP transients reveal modular, sex-specific bioenergetic control during excitation–contraction coupling. A, each action potential activates L-type Ca_V1.2 channels, producing a Ca²⁺ influx that triggers ryanodine receptors (RyR2) and elicits SR Ca²⁺ release. B, the cytosolic Ca²⁺ signal is decoded by mitochondria into spatially distinct ‘high-gain’ and ‘low-gain’ regions, shaped by the extent of sarcoplasmic reticulum (SR)–mitochondrial tethering via mitofusin 2 (Mfn2), yielding heterogeneous mitochondrial activation rather than a uniform, cell-wide metabolic response. C, mitochondria generate rhythmic, phase-locked ATP transients within discrete microdomains, such that ATP increases and ATP dips can coexist within the same cell. Ca²⁺ entry through the outer membrane [via the voltage-dependent anion channel (VDAC)] and into the matrix [via the mitochondrial Ca²⁺ uniporter (MCU)] stimulates oxidative phosphorylation, increasing ATP production; ATP is exported by adenine nucleotide translocase (ANT) to create local cytosolic ‘supply bursts’ aligned with beat-to-beat demand (‘paycheck-to-paycheck’ energetics). Female myocytes show a higher prevalence of tightly coupled, high-gain ATP-producing microdomains, whereas male myocytes display a shifted balance toward lower-gain regions, consistent with sex-dependent SR–mitochondrial coupling and ATP microdomain patterning.

Introduction

The function of the heart is to pump blood through systemic and pulmonary circulations, delivering oxygen and nutrients to tissues and removing metabolic waste. To accomplish this, the heart contracts rhythmically – about 100,000 times per day in humans and 1 million times in mice – through a sequence of electrical and mechanical events known as the cardiac cycle. Each cycle begins with the generation of an action potential by pacemaker cells in the sinoatrial node and culminates in ventricular contraction and relaxation – tightly regulated processes that ensure adequate cardiac output under varying physiological demands.

The conversion of electrical excitation into mechanical force is mediated by a process termed excitation–contraction (EC) coupling (Eisner et al., 2017). In ventricular myocytes, membrane depolarization triggers the opening of clusters of L-type Ca²⁺ channels (Ca_V1.2) in transverse tubules (Dixon et al., 2012, 2015; Spooner et al., 2025; Westhoff et al., 2024), allowing a small influx of Ca²⁺ into the cytosol (Santana et al., 1996; Wang et al., 2001). This local Ca²⁺ signal activates ryanodine receptors in the adjacent junctional sarcoplasmic reticulum (SR) through a Ca²⁺-induced Ca²⁺-release mechanism (Fabiato, 1983). The resulting ‘Ca²⁺ sparks’ raise intracellular cytosolic Ca²⁺ concentration ([Ca²⁺]_i) and initiate actomyosin cross-bridge cycling and myocyte contraction (Cannell et al., 1995; Cheng et al., 1993; López-López et al., 1995). This is followed by relaxation as Ca²⁺ is re-sequestered into the SR by the sarcoplasmic/endoplasmic reticulum Ca²⁺-ATPase (SERCA) pump and extruded from the cell by the Na⁺/Ca²⁺ exchanger, allowing the cycle to repeat (Balke et al., 1994; Li et al., 1998).

ATP fuels SERCA-driven Ca²⁺ reuptake, ion transport (via pumps and exchangers) and actomyosin cross-bridge cycling, with ~95% of this ATP supplied by mitochondrial oxidative metabolism and the remaining ~5% by glycolysis (Lopaschuk et al., 2021; Saddik & Lopaschuk, 1991; Wisneski et al., 1990; Zhou & Tian, 2018). Classic work has established coupling between Ca²⁺ handling and ATP production, showing that Ca²⁺ entering the mitochondrial matrix via the mitochondrial Ca²⁺ uniporter (MCU) activates three Ca²⁺-sensitive dehydrogenases – pyruvate, isocitrate and 2-oxoglutarate dehydrogenase; this boosts NADH production and drives proton flux through ATP5A-containing F₁F_o-ATP synthase (Denton & McCormack, 1990; Denton et al., 1980; Garbincius & Elrod, 2022; Kirichok et al., 2004; McCormack et al., 1990). In this process, cytosolic ADP and P_i must access the matrix and newly synthesized ATP must be exported, a transfer mediated by voltage-dependent anion channels (VDACs) in the outer mitochondrial membrane and adenine nucleotide translocase (ANT) in the inner membrane. Accordingly, mitochondrial efficiency is shaped not only by enzymatic capacity but also by membrane transport, organelle architecture and proximity to SR Ca²⁺-release sites. The tethering protein mitofusin-2 (Mfn2), located on the outer mitochondrial membrane, mediates physical and functional coupling between mitochondria and the junctional SR, facilitating rapid Ca²⁺ transfer and enhancing oxidative phosphorylation (Chen et al., 2012; Dorn et al., 2015). Mitochondrial ATP production is supplemented by glycolysis and is further supported by creatine and adenylate kinase reactions, which draw on phosphocreatine and ADP pools to regenerate ATP. These pathways act in concert to maintain ATP supply across varying workloads. Despite the substantial energetic cost of each contraction cycle, prevailing models have assumed that intracellular ATP remains effectively constant due to the high throughput of mitochondrial ATP production, the rapid kinetics of phosphotransfer reactions and the supplementary contribution of glycolysis (Chen et al., 1998; Murphy & Steenbergen, 2008).

High-resolution imaging of intracellular ATP concentration ([ATP]_i) in beating ventricular myocytes has now upended this view. In Rhana et al. (2024), we showed that cytosolic ATP fluctuates with each ventricular beat and that these oscillations disappear when oxidative phosphorylation is inhibited, revealing the limited buffering capacity of phosphotransfer and the dominant role of mitochondria in sustaining EC coupling. These findings raise two questions that are addressed in the present study: (i) How tightly are SR Ca²⁺ release and mitochondrial ATP synthesis temporally linked during each beat? (ii) Given emerging evidence for sex-specific differences in mitochondrial content and function (Cao et al., 2022), how do the kinetics and amplitudes of cytosolic and mitochondrial ATP transients differ between male and female cardiomyocytes?

In this study, we address these questions by directly monitoring cytosolic and mitochondrial ATP dynamics in beating male and female ventricular myocytes using newly developed expressible indicators, revealing a previously unrecognized, modular and beat-synchronized mode of ATP production. By comparing mitochondrial and cytosolic ATP fluctuations within the same cell type under physiological pacing conditions, we test the hypothesis that mitochondrial ATP, like its cytosolic counterpart, varies beat to beat in response to EC coupling. Building on evidence that mitochondrial volume, gene expression and SR–mitochondrial coupling differ between male and female hearts (Cao et al., 2022; Clements et al., 2023), we further examine how stimulation frequency and biological sex shape these dynamics.

Our results reveal that cytosolic and mitochondrial ATP oscillate in synchrony with the cardiac cycle and show that these oscillations are modulated by stimulation frequency and differ according to sex. Female myocytes exhibit a faster rise in diastolic cytosolic ATP during increased pacing that coincides with stronger SR–mitochondrial coupling and higher expression of Mfn2 and ATP5A. In contrast, mitochondrial volume is greater in male myocytes, and the amplitude of beat-to-beat ATP transients in these cells is selectively increased under load. Together, these findings support a model in which mitochondria participate dynamically in the beat-to-beat regulation of cardiac energetics and suggest sex-specific strategies – mass-based scaling versus architectural precision – for matching energy supply to contractile demand.

Methods and materials

Ethical approval

This study adhered to the guidelines outlined in the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals. The Institutional Animal Use and Care Committee (IACUC) at the University of California Davis approved the animal use and protocols described (IACUC #24065), which align with ARRIVE guidelines.

Animal source and housing

Adult male and female C57BL/6J mice (8–12 weeks old) were obtained from The Jackson Laboratory (Bar Harbor, ME, USA). Animals were housed in standard cages on a 50:50 h light/dark cycle with unrestricted access to food and water ad libitum.

Infection of mice with adeno-associated viruses

Adeno-associated virus serotype 9 (AAV9) particles carrying either the cytosolic iATP reporter iATPSnFR^1.0 (cyto-iATP) (Lobas et al., 2019) or a mitochondria-targeted iATP reporter iATPSnFR^2.0 (mito-iATP) (Marvin et al., 2024) were prepared at 4 × 10¹² viral genome copies per millilitre (vg/mL). Mice were induced with 5% isoflurane in oxygen (0.8–1.2 L/min) and maintained at 2% via a nose cone. Depth of anaesthesia was confirmed before the procedure and monitored throughout the procedure by the absence of the pedal withdrawal response to a firm toe pinch. For retro-orbital intravenous delivery, 100 μL of viral suspension was administered using a sterile insulin-type micro syringe and one drop of ophthalmic lubricant was applied after the injection. Post-procedure checks were done after 1 h and at least once daily for 24–48 h.

Isolation of mouse ventricular myocytes

Mice were killed 1–2 weeks after AAV infection and cardiomyocyte isolation was performed as previously described (Rossow et al., 2009; Shioya, 2007). Briefly, mice were heparinized (1000 UI/kg, i.p.) and a lethal dose of pentobarbital (100 mg/kg, i.p.) was administered. Hearts were rapidly excised and plunged into cold cell isolation buffer (130 mm NaCl, 5 mm KCl, 0.5 mm MgCl₂, 0.33 mm NaH₂PO₄, 25 mm Hepes, 22 mm glucose and 3 mm C₃H₃NaO₃, pH 7.4 adjusted using NaOH) supplemented with 150 μm EGTA. After aortic cannulation, an enzyme-containing cell isolation buffer (0.04 mg/mL protease type 2 collagenase, supplemented with 50 μm CaCl₂) at 37 °C was retrogradely perfused for 6–9 min until flow increased. Ventricles were removed and minced for further digestion in fresh enzyme solution. The suspension was filtered and Ca²⁺ restored stepwise. Isolated ventricular myocytes were maintained at room temperature (~24°C) in Tyrode solution (140 mm NaCl, 5.4 mm KCl, 1.8 mm CaCl₂, 1 mm MgCl₂, 5 mm Hepes and 5.5 mm glucose, pH 7.4 adjusted using NaOH) and used within 5 h of isolation. To assess the role of other mitochondrial substrates in ATP dynamics, we perfused myocytes with a ‘physiological mix’ of Tyrode containing 0.1 mm C₃H₃NaO₃, 1 mm C₃H₅NaO₃, 2 mm l-carnitine, 0.3 mm palmitate (BSA complex solution).

Field stimulation

Action potentials were evoked by field stimulation using two platinum wires separated by 0.5 cm, positioned at the bottom of the perfusion chamber. Square, 4 ms voltage pulses were generated using a Grass stimulator (AstroMed Inc., West Warwick, RI, USA). Pulse amplitudes were 10–40 V and delivered at frequencies ranging from 1 to 2 Hz.

Confocal imaging of action potential-evoked [Ca²⁺]_i, [ATP]_i and [ATP]_mito signals

Cytosolic Ca²⁺ was imaged by loading ventricular cardiomyocytes expressing the cyto-iATP or mito-iATP sensor with 10 μm Rhod-3-AM (Thermo Fisher, Waltham, MA, USA) following the manufacturer’s guidelines. After loading with Rhod-3, a drop of the cell suspension was transferred to a temperature-controlled (37°C) perfusion chamber on a microscope stage and allowed to settle onto the coverslip for 5 min. Cells were then perfused with Tyrode solution and field stimulated. Diastolic and systolic [Ca²⁺]_i and [ATP]_i or [ATP]_mito signals were simultaneously recorded using an inverted Olympus FV3000 confocal microscope (Olympus, Tokyo, Japan) operating in 2D or line-scan mode. Fluorescent indicators were excited with solid-state lasers emitting at 488 nm (cyto-iATP and mito-iATP) or 561 nm (Rhod-3) through a 60 oil-immersion lens (PlanApo) with a numerical aperture (NA) of 1.40. × During analysis, background was subtracted from all confocal images, and fluorescence signals were expressed as F/F₀, where F is the fluorescence intensity at a given time point, and F₀ is the mean baseline fluorescence. Cyto-iATP fluorescence values were converted to concentration units using the pseudo-ratiometric method (Cheng et al., 1993). A dissociation constant (K_d) of 1460 μm and a resting level of [ATP]_i of 457 μm was used in these calculations (Rhana et al., 2024).

Ratio-metric mito-iATP/HaloTag recordings

Isolated myocytes co-expressing the mito-iATP sensor and HaloTag were incubated for 30 min at 37°C with the HaloTag ligand JFX554 (Promega, Madison, WI, USA; #HT1030) at a final concentration of 0.1 μm. After rapid solution exchange, cells were transferred to pre-warmed (37°C) Tyrode solution and placed in a perfusion chamber on the microscope stage. Myocytes were allowed to settle onto the coverslip for 5 min before imaging. Images were acquired as described above (‘Confocal imaging‘ section) using solid-state laser excitation at 488 nm (mito-iATP) and 561 nm (HaloTag).

Super-resolution radial fluctuation imaging

Freshly isolated ventricular myocytes expressing cyto-iATP or mito-iATP sensors were incubated for 30 min at 37°C with 250 nm MitoTracker Deep Red (Molecular Probes, Carlsbad, CA, USA). Two-dimensional super-resolution radial fluctuation (SRRF) images were acquired using Fusion software on an Andor Dragonfly 200 spinning-disc confocal platform (Andor Technologies, Belfast, UK) coupled to an inverted Leica DMi* microscope fitted with a 60× oil-immersion objective (NA = 1.40) and Andor iXon EMCCD camera. Co-localization analysis was performed in IMARIS using the Colocalization module. Two fluorescence channels (mito-iATP vs. MitoTracker or cyto-iATP vs. MitoTracker) were analysed pixel-by-pixel within a user-defined region of interest (ROI) to exclude background and non-cellular areas. Manders’ overlap coefficient (M1) was calculated as the fraction of the iATP sensor signal that overlapped with the MitoTracker signal, providing an intensity-weighted measure of spatial overlap that is less sensitive than correlation-based metrics to differences in absolute signal amplitude. Consistent with Marvin et al. (2024), the mito-iATP sensor exhibited near-complete overlap with the mitochondrial dye MitoTracker (M1 = 0.94 ± 0.02 in female and 0.98 ± 0.02 in male myocytes), whereas, as shown by Rhana et al. (2024), cyto-iATP was diffusely expressed throughout myocytes and showed substantially lower overlap with MitoTracker (M1 = 0.59 ± 0.13 in female and 0.55 ± 0.12 in male myocytes). Because mitochondria occupy a large fraction of myocyte volume, a non-targeted cytosolic fluorophore will exhibit non-zero M1; thus, the key discriminator is the near-unity M1 for mito-iATP versus the markedly reduced M1 for cyto-iATP.

VDAC1–SERCA2 colocalization analysis

Immunofluorescence labelling was performed on freshly dissociated ventricular myocytes (wild-type, non-AAV-infected adult C57BL/6J mice). Myocytes were left to adhere for 1 h at room temperature prior to fixation, in coverslips coated with poly-l-lysine and laminin. Myocytes were fixed with 4% formaldehyde diluted in phosphate-buffered saline (PBS) (Fisher Scientific, Houston, TX, USA) for 15 min at room temperature, washed, and incubated with 50 mm glycine for 10 min to reduce aldehydes. Cells were then incubated in blocking buffer made of 3% (w/v) bovine serum albumin and 0.25% Triton X-100 in PBS, followed by incubation with mouse anti-SERCA2 (#MA3–919, Invitrogen, Carlsbad, CA, USA; 1:100) and mouse anti-VDAC1 (#ab14734, Abcam, Cambridge, MA, USA; 1:100) diluted in blocking buffer overnight at 4°C. Myocytes were washed, incubated at room temperature for 1 h with Alexa Fluor 488-conjugated goat anti-mouse IgG2b (#A21141, Invitrogen; 1:1000) and Alexa Fluor 647-conjugated goat anti-mouse IgG2a (#A21241, Invitrogen; 1:1000) diluted in blocking buffer followed by washes in PBS. All washes were performed with PBS three times for 5 min. Coverslips were mounted onto microscope slides in Vectashield mounting medium (Vector Labs, Burlingame, CA, USA) and sealed with clear nail polish. Images were collected on a Dragonfly 200 spinning disk confocal (Andor), coupled to a DMi* Leica microscope (Leica, Wetzlar, Germany) equipped with a 60 × oil-immersion objective (NA = 1.40) and acquired using an Andor iXon EMCCD camera. Two-dimensional wild-field images were collected via Fusion software. Using ImageJ, images were background-subtracted, and each channel was subject to thresholding to generate binary masks and image math was performed to visualize and quantify the SERCA2–VDAC1 colocalization.

Mitochondrial volume analysis

Ventricular myocytes from male and female mice were incubated with 250 nm MitoTracker Deep Red (Thermo Fisher Scientific) for 30 min at 37°C. Cells were allowed to adhere to poly-l-lysine/laminin-coated coverslips for 1 h at room temperature, then fixed with 4% formaldehyde diluted in PBS for 15 min. Coverslips were mounted using VectaShield mounting medium and sealed with clear nail polish. Confocal z-stacks were acquired on an Olympus FV3000 confocal microscope using a 60 ×oil-immersion objective (NA = −1.40), with a z-step of 0.5 μm. After importing three-dimensional (3D) image stacks into IMARIS 10 (Andor), mitochondria were segmented based on MitoTracker fluorescence intensity using the Surfaces tool and a consistent thresholding approach. Total mitochondrial volume was quantified for each cell and normalized to cytosolic volume to obtain the mitochondrial volume fraction.

Western blots

Female and male ventricles were collected and homogenized in ice-cold RIPA buffer supplemented with a cocktail of protease inhibitors (Pierce Protease Inhibitor Mini Tablets; Thermo Scientific). Proteins (60 μg) were separated by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by semi-dry transfer onto a PVDF membrane (Merck, USA). Membranes were washed in Tris-buffered saline containing 0.1% Tween-20 (TBS-T), blocked in 5% bovine serum albumin (BSA) TBS-T for 1 h at room temperature, and incubated with rabbit anti-Mfn2 (#9482S, Cell Signaling, Danvers, MA, USA; 1:500), mouse anti-ATP5A1 (#459240, Invitrogen; 1:1000) and mouse anti-VDAC1 (clone N152B/23, RRID 2877354; UC Davis/NIH Neuromab Facility, USA; 1:200) diluted in blocking buffer. Membranes were then rinsed with TBS-T and incubated with mouse IgG 800CW or rabbit IgG 680CW secondary antibodies (1:15,000; Li-Cor, Lincoln, NE, USA) for 1 h at room temperature. Fluorescence signals were detected using an Odyssey Infrared Imager (Li-Cor). Blots were digitized, and bands were quantified using ImageJ. Protein levels were expressed as the ratio of optical densities of specific bands to VDAC1.

Statistics

Data are presented as mean ± SD, from n cells, sites or transients, and N animals. Normality was assessed for all data sets (Shapiro–Wilk test), and all passed this test (P > 0.05). Accordingly, only parametric statistics were implemented. Hierarchical statistics (nested t tests) and paired Student’s t tests were used throughout the paper, as indicated in the figure legends. Exact P values are provided in the text and/or figures, with P < 0.05 considered statistically significant. All data points are shown in the figures, with light symbols representing individual cells and dark symbols individual animals.

Results

Mitochondrial ATP transients exhibit two beat-locked waveforms with a sex-dependent spatial distribution

Mitochondrial ATP was imaged using a mitochondrial matrix-targeted intracellular ATP (iATP) sensor termed mito-iATP, a version of the low-affinity A95A/A119L variant (apparent K_d ≈ 0.5 mm) of the high-dynamic-range reporter, iATPSnFR2, adult mouse ventricular myocytes expressing mito-iATP (Fig. 1A) showed that mito-iATP fluorescence formed a finely striated reticulum that colocalized with MitoTracker Deep Red, with no detectable cytosolic signal in either male or female myocytes. A magnified 10 × 10 μm view (Fig. 1A insets) illustrates the pixel-level concordance between the two channels. A quantitative colocalization analysis yielded Manders’ coefficients of 0.94 ± 0.02 for female and 0.98 ± 0.02 for male myocytes, indicating precise targeting throughout the mitochondrial lattice. As shown in Fig. 1B, brief bath application of 1 μm FCCP, an uncoupler of oxidative phosphorylation, decreased mito-iATP fluorescence from a normalized F/F₀ value of 1.00 to 0.75 ± 0.17 in female (P = 0.0004) and 0.78 ± 0.14 in male mice (P = 0.0002), emonstrating that mito-iATP reliably reports dynamic changes in matrix ATP concentration. No difference was observed in the FCCP-induced minimum fluorescence between female and male myocytes (P = 0.7748).

Figure 1. Targeting and validation of the mitochondrial ATP sensor, mito-iATP, in adult mouse ventricular myocytes.

A, live-cell, wide-field confocal images of myocytes expressing mito-iATP (green) and Mitotracker Deep Red (red). The merged panel and 10 × 10 μm inset demonstrate pixel-level concordance between channels. Manders colocalization coefficient: female, 0.94 ± 0.02 (N = 3 mice, n = 26 cells); male, 0.98 ±0.02 (N = 3 mice, n = 33 cells). Data are presented as mean ± SD. Scale bars: 10 μm. B, time course of normalized mito-iATP fluorescence (F/F₀) after bath application of 1 μM FCCP (female: N = 3 mice, n = 10 cells; male: N = 2 mice, n = 11 cells).

To test whether [ATP] mito fluctuates on a beat-to-beat basis, we loaded myocytes expressing mito-iATP with the red-shifted Ca²⁺ indicator Rhod-3 and applied field stimulation at 1 Hz. We chose 1 Hz as a standard baseline pacing rate for isolated adult ventricular myocytes because it provides stable Ca²⁺ cycling and preserves cell health while allowing accurate temporal alignment of Ca²⁺ and ATP signals with minimal rundown. Although mouse hearts beat faster in vivo, the core Ca²⁺-dependent control of mitochondrial metabolism is conserved; thus, 1 Hz provides a well-controlled reference condition for mechanistic experiments, with frequency-dependent behaviour examined separately at higher pacing rates.

Representative line-scan images (Fig. 2A, B) captured simultaneous [Ca²⁺]_i (upper trace, red) and matrix ATP (lower trace, green) responses during individual action potentials. Each stimulus elicited a phasic [ATP]_mito fluctuation that was tightly phase-locked to the [Ca²⁺]_I transient. Two reproducible waveform types were identified. In the first (Mode 1), [ATP]_mito rose rapidly and then returned to baseline (Fig. 2A). In the second (Mode 2), [ATP]_mito dipped transiently and recovered with similar kinetics (Fig. 2B). These opposite-polarity responses coexisted within single myocytes, but were spatially segregated, revealing a patchwork of energetic microdomains. Importantly, the amplitude of whole-cell [Ca²⁺]_i transients was similar in cells displaying either Mode 1 or Mode 2 fluctuations across both sexes (Fig. 2C) (2.40 ± 0.53 and 2.16 ± 0.5 F/F₀ for Mode 1 sites in female and male, respectively – P = 0.2289; and 2.59 ± 0.56 and 2.58 ±1.07 F/F₀ for Mode 2 sites in female and male, respectively – P = 0.8580). The average [ATP]_mito amplitudes of Mode 1 sites (1.53 ± 0.27 F/F₀ in females and 1.46 ± 0.20 F/F₀ in males, P = 0.4124) and Mode 2 sites (0.71 ± 0.08 F/F₀ in females and 0.70 ± 0.09 F/F₀ in males, P = 0.5883) were similar across sexes (Fig. 2D).

Figure 2. Beat-locked mitochondrial ATP transients adopt two discrete waveforms (Modes 1 and 2).

A and B, representative line-scan images of [Ca²⁺]_i (top) and [ATP]_mito (bottom) from female and male myocytes with Mode 1 (A) or Mode 2 (B) ATP dynamics. Red and green traces above each line-scan show [Ca²⁺]_i and [ATP]_mito time courses, respectively. Scale bars: 20 μm. C–E, scatter plots of the amplitudes of [Ca²⁺]_i (C) and [ATP]_mito (D) transients, and spatial spread of [ATP]_mito (E) in female (magenta) and male (blue) myocytes (female: N = 3 mice, n = 27 and 17 cells of Mode 1 and 2 sites, respectively; male: N = 3 mice, n = 21 and 15 cells of Mode 1 and 2 sites, respectively). F, representative regions of interest in Mode 1 [ATP]_mito microdomains in female (top) and male (bottom) myocytes. Green trace (left) is the spatial profile of the [ATP]_mito signal oriented along the dotted line. Red, yellow and orange traces are normalized fluorescence intensity of the local [ATP]_mito microdomain. G and H, scatter plots of the coefficient of variation (CV) of Mode 1 [ATP]_mito sites (G) and peak amplitude for the [ATP]_mito microdomain (H) (female: N = 3 mice, n = 18 cells; male: N = 3 mice, n = 12 cells). Data are presented as mean ± SD. All significant values are provided from a nested t test. Light symbols: individual cells; dark symbols: individual animals.

Quantification of over 30 individual mito-iATP sites revealed a modest, but consistent, sex difference in waveform prevalence. In females, 43% of cells exhibited Mode 1 exclusively, 7% exhibited Mode 2 exclusively and 50% showed both modes; in males, the corresponding proportions were 38, 8 and 54%. Despite this difference in spatial distribution, the average length of local transients was indistinguishable between sexes: Mode 1 sites averaged around 25 μm (23 ± 9 μm in females and 26 ± 7 μm in males, P = 0.4279), whereas Mode 2 sites averaged 18 ± 8 μm in both sexes (P = 0.8578) (Fig. 2E). Given that a single mitochondrion spans ~ 1.5–2 μm, these measurements suggest that each narrow line-scan region captures synchronous ATP fluctuations across strings of adjacent ~ 10–18 interfibrillar mitochondria (Hom & Sheu, 2009; Moreland, 1962), confirming the microdomain nature of these events.

Despite comparable amplitudes, the spatial patterning of [ATP]_mito elevations along interfibrillar mitochondrial strings was more heterogeneous in females (Fig. 2F). To quantify this, we calculated the coefficient of variation (CV = standard deviation/mean) of mito-iATP signals along the transient peak. The CV was significantly greater in female (0.14 ± 0.05) than male (0.09 ± 0.03) myocytes (P = 0.0381) (Fig. 2G), suggesting greater spatial variability in mitochondrial ATP synthesis. Consistent with this, ATP ‘hotspots’ in cardiomyocytes reached higher peak values in females (1.91 ± 0.46 F/F₀) than males (1.51 ± 0.18 F/F₀; P = 0.0410), even though the average amplitude per site remained equivalent (Fig. 2H).

Because standard Tyrode solution contains glucose as the only exogenous carbon source, we investigated whether supplementing the bath with a more physiological mixture of metabolic substrates would alter beat-locked ATP dynamics (Fig. 3). To approximate the metabolic milieu of arterial blood, we perfused myocytes with a ‘physiological mix’ containing glucose together with a cocktail of long-chain fatty acids (palmitate) and additional oxidizable substrates (lactate, pyruvate and l-carnitine), with BSA serving as a carrier for lipid species. Under these conditions, the amplitude of mito-iATP transients in both Mode 1 (ATP-gain) and Mode 2 (ATP-dip) microdomains were indistinguishable from those recorded in Tyrode solution (Fig. 3B) (Mode 1: 1.62 ± 0.23 F/F₀ in female, P = 0.2625 vs. control; and 1.62 ± 0.18 F/F₀ in male, P =0.1003 vs. control; Mode 2: 0.66 ± 0.09 F/F₀ in female, P = 0.1597 vs. control; and 0.64 ± 0.09 F/F₀ in male, P = 0.2510 vs. control). Likewise, [Ca²⁺]_i transient amplitudes were similar in Tyrode and physiological mix (Fig. 3C) (Mode 1: 2.41 ± 0.31 F/F₀ in female, P = 0.9548 vs. control; and 2.36 ± 0.45 F/F₀ in male, P = 0.1255 vs. control; Mode 2: 2.41 ± 0.35 F/F₀ in female, P = 0.3068 vs. control; and 2.33 ± 0.56 F/F₀ in male, P = 0.7855 vs. control). Thus, ATP transients are preserved when myocytes are supplied with a physiological substrate mix, indicating that these signals are not an artefact of perfusion with a simplified glucose-only Tyrode solution.

Figure 3. Effect of mixed substrates on mitochondrial ATP and [Ca²⁺]_i transient measurements.

A, representative time course from line scans of [Ca²⁺]_i (top) and [ATP]_mito (bottom) in female and male myocytes with Mode 1 (left) or Mode 2 (right) ATP dynamics using a new physiological substrate mix containing pyruvate, lactate, carnitine and palmitate. B and C, scatter plots of the amplitudes of [ATP]_mito (B) and [Ca²⁺]_i (C) transients in female (magenta) and male (blue) myocytes. Data are presented as mean ± SD (female: N = 3 mice, n = 26 and 17 cells of Mode 1 and 2 sites, respectively; male: N = 3 mice, n = 25 and 24 cells of Mode 1 and 2 sites, respectively). All significant values are provided from a nested t test. Light symbols: individual cells; dark symbols: individual animals.

To assess sex-dependent differences in the kinetics of mitochondrial ATP transients, we quantified time-to-peak and time-to-decay for Mode 1 and Mode 2 responses (Appendix Fig. A1A, B). In Mode 1 sites, [ATP]_mito rose faster in females (80 ± 23 ms) than in males (97 ± 25 ms; P = 0.0453), while Mode 2 rise times were similar between sexes (females: 107 ± 43 ms; males: 106 ± 45 ms; P = 0.9662) (Fig. A1C). Time to 90% decay was also comparable across sex and mode (Fig. A1D) (Mode 1: 123 ± 51 and 135 ± 56 ms in female and male, respectively, P = 0.4820;Mode 2: 144 ± 65 and 121±60ms in female andmale, respectively, P=0.4025). The time-to-peak and time-to-decay values for [Ca2+]_I transients were similar inMode 1 andMode 2 sites across both sexes (Fig. A1E, F) (time to peak: Mode 1 – 59 ± 9 and 63 ± 14 ms in female and male, respectively, P = 0.2201; Mode 2 – 61 ± 12 and 60 ± 11 ms in female and male, respectively, P = 0.8721) (Time to decay: Mode 1 – 228 ± 70 and 226 ± 50 ms in female and male, respectively, P = 0.9676; Mode 2 – 233 ± 62 and 209±43ms in female andmale, respectively, P=0.4244). It is important to note that, while [Ca2+]_i and [ATP]_mito transients are shown for temporal alignment, a direct comparison of their kinetics should be interpreted with caution, given the markedly faster response properties of Rhod-3 compared with those of the EGFP-based mito-iATP sensor.

To exclude motion artefacts and validate the pseudo-ratiometric approach, we labelled the HaloTag domain of mito-iATP with the red-shifted JFX554 HaloTag ligand and simultaneously imaged both channels during field stimulation. Robust beat-locked Mode 1 and Mode 2 [ATP]_mito transients were evident in the mito-iATP channel, whereas the JFX554 HaloTag signal exhibited no detectable beat-synchronous changes in line-scan recordings (Fig. 4A, C). Across regions exhibiting ATP gains or dips, the mito-iATP ratio signal scaled linearly with the underlying non-ratiometric mito-iATP fluorescence (Fig. 4B,D), indicating that the ratio faithfully reports changes in matrix ATP rather than movement of mitochondria within the confocal volume. These data argue against motion or volume changes as the source of the observed [ATP]_mito waveforms and support the use of mito-iATP ratios to quantify beat-locked microdomain ATP dynamics.

Figure 4. Ratiometric validation confirms that beat-to-beat mitochondrial iATP transients are independent of motion artifacts.

A and C, representative time course from line scans of [ATP]_mito (green), reference fluorophore JFX554 HaloTag (red), and ratiometric signal (orange) from myocytes with Mode 1 (A) and Mode 2 (C) ATP dynamics. All signals are shown as normalized fluorescence (F/F₀). B and D, relationship between [ATP]_mito transient amplitude and ratiometric transient amplitude for Mode 1 (B) and Mode 2 (D) events. Points indicate individual transients, and line shows the fitted relationship with the coefficient of determination (R²) reported on-graph. Data are presented as mean ± SD (female: N = 2 mice n = 6 cells/14 transients and 5 cells/18 transients of Mode 1 and 2 sites, respectively; male: N = 2 mice, n = 4 cells/14 transients and 3 cells/7 transients of Mode 1 and 2 sites, respectively).

Together, these findings demonstrate that millisecond-scale ATP transients can be visualized within adult ventricular mitochondria and that they adopt two discrete, spatially confined waveforms. While peak amplitudes are similar, the relative distribution, spatial heterogeneity and kinetics of these events differ between male and female myocytes, raising the possibility that sex-specific energetic strategies reflect both mitochondrial abundance and subcellular organization.

MCU and ANT are required for beat-locked mitochondrial ATP transients

To determine whether mitochondrial Ca²⁺ uptake through the MCU is required for the beat-locked [ATP]_mito signals, ventricular myocytes expressing mito-iATP and loaded with Rhod-3 were exposed to 10 μm Ru360, a selective MCU blocker (Fig. 5A, B, E, F). Under control conditions, each action potential evoked robust Mode 1 or Mode 2 [ATP]_mito transients (Mode 1: 1.60 ± 0.13 and 1.59 ± 0.21 F/F₀; Mode 2: 0.68 ± 0.05 and 0.69 ± 0.05 F/F₀; female and male, respectively) that were tightly phase-locked to the [Ca2+]i transient (Mode 1: 3.03 ± 0.69 and 2.73 ± 0.69 F/F₀; Mode 2: 2.96 ± 0.53 and 2.65 ± 0.48 F/F₀; female and male, respectively), as described above. Exposure to Ru360 decreased these beat-coupled [ATP]_mito oscillations in both modes, markedly reducing local ATP fluctuations to near-baseline noise (Fig. 5C, G) (Mode 1: 1.22±0.18 F/F₀ in females, P = 0.0013 vs. control; and 1.24 ± 0.11 F/F₀ in males, P = 0.0031 vs. control; Mode 2: 0.79 ± 0.05 F/F₀ in female, P = 0.0082 vs. control; and 0.78 ± 0.05 F/F₀ in males, P = 0.0122 vs. control), while simultaneously recorded [Ca²⁺]i transients were maintained (Fig. 5D, H) (Mode 1-associated Ca²⁺ signal: 2.06 ± 0.32 F/F₀ in females, P = 0.0041 vs. control; and 2.04 ± 0.57 F/F₀ in males, P = 0.0381 vs. control; Mode 2-associated Ca²⁺ signal: 1.97 ± 0.46 F/F₀ in females, P = 0.0093 vs. control; and 2.07 ± 0.53 F/F₀ in males, P = 0.0459 vs. control). Thus, MCU-mediated Ca²⁺ entry is critical for generating the phasic mitochondrial ATP responses during EC coupling.

Figure 5. Beat-to-beat metabolic signalling requires mitochondrial Ca²⁺ uptake and ANT-mediated ATP export.

A and B, representative time course from line scans of [Ca²⁺]_i (top) and [ATP]_mito (bottom) in female (A) and male (B) myocytes with Mode 1 ATP dynamics in control cells and in cells exposed to 10 μM BKA or Ru360 for 30 min. C and D, scatter plots of the amplitudes of [ATP]_mito (C) and [Ca²⁺]_i (D) transients in female (magenta) and male (blue) myocytes with Mode 1 ATP dynamics under control and treated conditions (female: N = 3 mice, n = 22, 20 and 22 cells; male: N = 3 mice, n = 21, 23 and 22 cells; data from control, BKA and Ru360 conditions, respectively). E and F, representative time course from line scans of [Ca²⁺]_i (top) and [ATP]_mito (bottom) in female (E) and male (F) myocytes with Mode 2 ATP dynamics in control cells and in cells exposed to 10 μM BKA or Ru360 for 30 min. G and H, scatter plots of the amplitudes of [ATP]_mito (G) and [Ca²⁺]_i (H) transients in female (magenta) and male (blue) myocytes with Mode 2 ATP dynamics under control and treated conditions (female: N = 3 mice, n = 15, 7 and 9 cells; male: N = 3 mice, n = 10, 10 and 18 cells; data from control, BKA and Ru360 conditions, respectively). Data are presented as mean ± SD. All significant values are provided from a nested t test. Light symbols: individual cells; dark symbols: individual animals.

We next tested whether export of matrix ATP via the ANT is necessary to sustain beat-locked ATP signals in the mitochondrial matrix (Fig. 5A, B, E, F). Inhibition of ANT with bongkrekic acid (BKA; 10 μm) reduced the beat-synchronized [ATP]_mito transients (Fig. 5C, G) (Mode 1: 1.16 ± 0.13 F/F₀ in females, P = 0.0005 vs. control; and 1.19 ± 0.13 F/F₀ in males, P = 0.0020 vs. control; Mode 2: 0.75 ± 0.04 F/F₀ in females, P = 0.0310 vs. control; and 0.76 ± 0.05 F/F₀ in males, P = 0.0360 vs. control) maintaining [Ca²⁺]i transients (Fig. 5D, H) (Mode 1: 2.20 ± 0.46 F/F₀ in females, P = 0.0108 vs. control; and 2.05 ± 0.47 F/F₀ in males, P = 0.0368 vs. control; Mode 2: 2.04 ± 0.34 F/F₀ in females, P = 0.0133 vs. control; and 1.99 ± 0.45 F/F₀ in males, P = 0.0328 vs. control), indicating that continued nucleotide exchange across the inner mitochondrial membrane is required for the mito-iATP signal. Together, these interventions demonstrate that beat-to-beat ATP transients in mitochondrial matrix require MCU-dependent mitochondrial Ca²⁺ uptake to drive oxidative phosphorylation and ANT-mediated nucleotide exchange to transmit this oscillatory ATP output to the cytosol.

Female myocytes trade mitochondrial mass for tighter SR–mitochondrial coupling and higher Mfn2/ATP5A density

We postulated that the smaller local ATP signals and lower prevalence of ATP-gain microdomains observed in male myocytes might stem from sex differences in both mitochondrial content and the spatial relationship between mitochondria and sites of SR Ca²⁺ release. Voxel segmentation of 3D confocal stacks from Mitotracker-labelled ventricular myocytes revealed that mitochondria occupied 6391 ± 1700 μm³ of cytosolic volume in females versus 8192 ± 1264 μm³ in males (n = 30 cells per sex; P = 0.0307), indicating that male myocytes devote a larger fraction of their cytosol to mitochondrial volume (Fig. 6A, B).

Figure 6. Female myocytes harbour a smaller mitochondrial volume but exhibit tighter SR–mitochondrial coupling.

A, 3D reconstructions of the mitochondrial volume in a female (left) and male (right) ventricular myocyte. Mitochondria were volumetrically rendered from deconvolved confocal z-stacks after staining with MitoTracker Deep Red. B, scatter plot of the mitochondrial volume in female (magenta) and male (blue) myocytes (female and male: N = 3 mice, 30 cells). Significant values are provided from a nested t test. C, representative immunoblots of the mitochondrial outer-membrane tether protein mitofusin 2 (Mfn2, 80 kDa), of the α subunit of the catalytic core of mitochondrial ATP synthase protein (ATP5A, 54 kDa) and of the voltage-dependent anion channel 1 (VDAC1, 30 kDa) in ventricular lysates from five female (♀ WT 1–5) and five male (♂ WT 1–5) mice. D and E, densitometric quantification of ATP5A (D) and Mfn2 (E), normalized to VDAC1. Significant values are provided from a Student’s t test. All data are presented as mean ± SD. Light symbols: individual cells; dark symbols: individual animals.

To explore the molecular correlates of this architectural divergence, we quantified protein levels of Mfn2, a key tether that links mitochondria to the SR and facilitates Ca²⁺-dependent stimulation of oxidative phosphorylation (Chen et al., 2012; Dorn et al., 2015; Rhana et al., 2024), and ATP5A, a core subunit of the mitochondrial ATP synthase complex. Immunoblotting of ventricular homogenates showed that levels of ATP5A (1.36 ± 0.09 in female, 1.19 ± 0.08 in male; P = 0.0052) and Mfn2 (0.80 ± 0.03 in female, 0.70 ± 0.10 in male; P = 0.0351) were higher in females than males when normalized per mitochondrial density (Fig. 6C–E).

We next investigated whether this inferred molecular enrichment is accompanied by tighter physical coupling between mitochondria and SR Ca²⁺ release sites (Fig. 7). Ventricular myocytes were co-immunolabelled with antibodies against the outer mitochondrial membrane protein VDAC and the SR Ca²⁺ pump SERCA, and confocal images from each channel were independently thresholded to generate binary masks (Fig. 7A). These binary images were then multiplied pixel-by-pixel so that only voxels positive for both VDAC and SERCA were set to 1; in the resulting overlap maps, colocalized regions appear as discrete black puncta, whose total area was quantified as an index of SR–mitochondrial apposition. This analysis revealed a significantly larger VDAC–SERCA overlap area in female than male myocytes (Fig. 7B) (59 ± 7% in female vs. 47 ± 9% in male, P = 0.0324), indicating that a greater fraction of mitochondrial surface resides in close proximity to SR Ca²⁺ handling machinery in females.

Figure 7. SERCA2–VDAC1 colocalization reveals sex-dependent SR–mitochondria proximity.

A, representative confocal immunofluorescence images from female and male ventricular myocytes labelled for the mitochondrial outer membrane marker VDAC1 (green) and the sarcoplasmic reticulum Ca²⁺-ATPase SERCA2 (red). Merged images and 10 × 10 μm insets are shown alongside a binary colocalization map that shows pixels in which SERCA2 and VDAC1 are completely overlapped. Scale bars: 10 μm. B, scatter plot of SERCA2–VDAC1 percentage colocalization in female (magenta) and male (blue) myocytes. Data are presented as mean ± SD (female: N = 3 mice, 33 cells; male: N = 3 mice, 51 cells). All significant values are provided from a nested t test. Light symbols: individual cells; dark symbols: individual animals.

Together with the mito-iATP imaging, these data support a model in which female myocytes trade mitochondrial mass for more densely equipped and more tightly SR-coupled mitochondria, favouring rapid, spatially confined ATP surges in response to local Ca²⁺ fluxes, whereas male myocytes rely more heavily on increased mitochondrial volume to scale ATP output. This structural and molecular arrangement dovetails with our prior finding that beat-locked increases in mitochondrial Ca²⁺ are tightly associated with cytosolic iATP transients in ventricular myocytes (Rhana et al., 2024), suggesting that enhanced SR–mitochondrial tethering and higher Mfn2/ATP5A density in females sharpen this Ca²⁺-to-ATP coupling axis.

Cytosolic ATP transients mirror mitochondrial waveforms and exhibit a sex-dependent domain organization

To assess cytosolic ATP dynamics, we used iATPSnFR1, the genetically encoded cytosolic fluorescent ATP sensor developed by Lobas et al. (2019) (Fig. 8). The construct, here termed cyto-iATP, was packaged into an AAV9 vector and expressed in adult ventricular myocytes from male and female mice. Confocal imaging showed a uniform, diffuse fluorescence pattern throughout the cytosol in male or female myocytes (Fig. 8A,B).

Figure 8. Cytosolic ATP transients mirror mitochondrial modes and show sex-dependent domain organization.

A and B, confocal image of an exemplar ventricular myocyte expressing cytosolic iATP sensor in female (A) and male (B) mice. Scale bar: 10 μm. C and D, representative line scan images of [Ca²⁺]_i (top) and [ATP]_i (bottom) in female and male myocytes with Mode 1 (C) or Mode 2 (D) ATP dynamics. Red and green traces above each line-scan show the time course of [Ca²⁺]_i and [ATP]_i transients, respectively. Scale bars: 20 μm. E and F, scatter plots of global and local amplitudes of [ATP]_i transients in Mode 1 (E) and Mode 2 (F) sites in female (magenta) and male (blue) myocytes. G and H, scatter plots of global and local amplitudes of [Ca²⁺]_i transients in Mode 1 (G) and Mode 2 (H) sites in female (magenta) and male (blue) myocytes. Data are presented as mean ± SD [female: N = 4 mice, n = 24 cells/35 sites and 4 cells/16 sites of Mode 1 and 2 (global/local), respectively; male: N = 3 mice, n = 20 cells/26 sites and 5 cells/17 sites of Mode 1 and 2 (global/local), respectively]. All significant values are provided from a nested t test. Light symbols: individual cells; dark symbols: individual animals.

To determine whether [ATP]_i exhibits beat-to-beat fluctuations during EC coupling, we recorded simultaneous line-scan images of cyto-iATP and Rhod-3 fluorescence in cells subject to field stimulation (1 Hz). As with mitochondrial ATP, [ATP]_i underwent rhythmic phasic changes that were tightly phase-locked to the [Ca²⁺]_i transient. Two distinct waveform types – Mode 1 and Mode 2 – were identified. In Mode 1 (Fig. 8C), [ATP]_i rose rapidly after the Ca² transient and returned to baseline before the next beat. In Mode 2 (Fig. 8D), [ATP]_i dipped transiently and recovered with similar timing. Both modes were present within single cells and segregated spatially, revealing a cytosolic energetic microdomain structure analogous to that observed in the mitochondrial matrix.

Quantification of local and global [ATP]_i and [Ca²⁺]_i transients revealed striking parallels to the mitochondrial data (Fig. 8E–H). Mode 1 and Mode 2 events were detected in myocytes of both sexes, and their spatial distribution followed a similar sex-dependent pattern: female myocytes exhibited a higher proportion of ATP-gain (Mode 1) domains (58% in females and 40% in males), whereas male myocytes had more ATP-dip (Mode 2) regions (10% in females and 13% in males). Local transient amplitudes did not differ significantly between sexes for either mode (Fig. 8E,F) (Mode 1: 1.25 ± 0.11 F/F₀ in female and 1.27 ± 0.15 F/F₀ in male, P = 0.6054; Mode 2: 0.77 ± 0.09 F/F₀ in female and 0.79 ± 0.09 F/F₀ in male, P = 0.5789), indicating that the intrinsic dynamics of each waveform are comparable. However, global F/F₀ signals were significantly larger in male than female myocytes for both Mode 1 (1.09 ± 0.05 vs. 1.13 ± 0.06 F/F₀; P = 0.0432) and Mode 2 (0.96 ± 0.02 vs. 0.91 ± 0.03 F/F₀; P = 0.0435), consistent with a greater fraction of the cytosol participating in eachmode inmale cells. Thus, the amplitude of [Ca²⁺]i transients was similar in cells displaying either Mode 1 (Global: 2.23 ± 0.57 F/F₀ in female and 2.30 ± 0.72 F/F₀ in male, P = 0.7489; Local: 2.15 ± 0.58 F/F₀ in female and 2.38 ± 0.75 F/F₀ in male, P = 0.2335) or Mode 2 fluctuations (Fig. 8G, H) (Global: 1.96 ± 0.37 F/F₀ in female and 2.17 ± 0.59 F/F₀ in male, P = 0.5469; Local: 2.08 ± 0.53 F/F₀ in female and 2.49 ± 0.88 F/F₀ in male, P = 0.2746) across both sexes.

To evaluate sex- andmode-dependent differences in the kinetics of [ATP]_i transients, we measured the time to peak and decay in Mode 1 and Mode 2 regions of male and female myocytes (Fig. A2). Unlike [ATP]_mito transients, [ATP]_i transients in Mode 1 sites showed no significant difference in the time to 90% peak between sexes (87 ms for both, P = 0.9962). However, time to 90% peak [ATP]_I at Mode 2 sites differed between female (75 ± 20 ms) and male (107 ± 48 ms) myocytes (P = 0.0448). As was the case for [ATP]_mito transients, we observed no difference in time-to-decay of [ATP]i transients across sexes andmodes or the kinetics of whole-cell [Ca²⁺]_i transients.

Cytosolic ATP transients persist under a mixed oxidative substrate supply

To test whether beat-locked cytosolic ATP dynamics are preserved under more physiological oxidative substrates, we recorded [Ca²⁺]_i and [ATP]_i in field-stimulated ventricular myocytes (1 Hz) perfused with a substrate mix containing pyruvate, lactate, carnitine and palmitate. Representative line-scan time courses show that Mode 1 cells maintained beat-synchronous [ATP]_i increases aligned with [Ca²⁺]_i transients, whereas Mode 2 cells retained beat-locked [ATP]_i decreases (ATP ‘dips’) despite robust [Ca²⁺]_i transients, in both female and male myocytes (Fig. 9A). As with mito-iATP, the amplitude of [ATP]_i transients in both Mode 1 and Mode 2 ATP dynamics were indistinguishable from those recorded in Tyrode solution (Fig. 9B) (Mode 1: 1.30 ± 0.13 F/F₀ in female, P = 0.1649 vs. control; and 1.30 ± 0.12 F/F₀ in male, P = 0.5957 vs. control; Mode 2: 0.76 ± 0.10 F/F₀ in female, P = 0.7108; and 0.79 ± 0.08 F/F₀ in male, P = 0.9373 vs. control).

Figure 9. Effect of mixed substrates on cytosolic ATP and [Ca²⁺]_i transient measurements.

A, representative time course from line scans of [Ca²⁺]_i (top) and [ATP]_i (bottom) in female and male myocytes with Mode 1 (left) or Mode 2 (right) ATP dynamics using a new physiological substrate mix containing pyruvate, lactate, carnitine and palmitate. B and C, scatter plots of the amplitudes of [ATP]_i (B) and [Ca²⁺]_i (C) transients in female (magenta) and male (blue) myocytes. Data are presented as mean ± SD (female: N = 3 mice, n = 30 and 21 cells of Mode 1 and 2 sites, respectively; male: N = 3 mice, n = 27 and 22 cells of Mode 1 and 2 sites, respectively). All significant values are provided from a nested t test. Light symbols: individual cells; dark symbols: individual animals.

In parallel, [Ca²⁺]_i transient amplitudes were comparable across sexes and modes (Fig. 9C) (Mode 1: 2.24 ± 0.41 F/F₀ in female, P = 0.5033 vs. control; and 2.19 ± 0.40 F/F₀ in male, P = 0.3075 vs. control; Mode 2: 2.15 ± 0.45 F/F₀ in female, P = 0.6893; and 2.18 ± 0.46 F/F₀ in male, P = 0.3474 vs. control), indicating that the eservation of ATP transients in this substrate condition is not attributable to gross differences in [Ca²⁺]_i transient magnitude. These data show that cytosolic ATP transients are preserved under pyruvate/lactate/carnitine/palmitate and are qualitatively similar to those observed in Tyrode solution, supporting the view that Mode 1 and 2 ATP dynamics exist across substrate environments.

Cross-bridge inhibition selectively attenuates Mode 2 cytosolic ATP dips while preserving Mode 1 gains

Because cross-bridge cycling is a major ATP sink during each beat, we next investigated whether contractile activity contributes to the Mode 2 cytosolic ATP phenotype. We therefore applied 10 μm blebbistatin to inhibit myosin II ATPase and uncouple shortening from the Ca²⁺ transient (Fig. 10). Blebbistatin decreased contraction by 74 ± 16% (P < 0.0001). Representative line-scan recordings show beat-locked [ATP]_i transients in both female and male myocytes exhibiting Mode 1 gains or Mode 2 dips even in the presence of blebbistatin (Fig. 10A, B).

Figure 10. Blebbistatin suppresses contraction while preserving beat-to-beat cytosolic ATP and [Ca²⁺]_i transients.

A and B, representative time course from line scans of [ATP]_i (green) and [Ca²⁺]_i (red) in female and male myocytes with Mode 1 (A) or Mode 2 (B) ATP dynamics in control condition and in the presence of 10 μM blebbistatin (BB) for 30 min. C–F, scatter plots of the amplitude of [ATP]_i (C, D) and [Ca²⁺]_i (E, F) transients of Mode 1 and 2 ATP dynamics under control and treated conditions in female (magenta) and male (blue) myocytes. Data are presented as mean ± SD (female: N = 3 mice, n = 13 and 6 cells of Mode 1 and 2 sites, respectively; male: N = 3 mice, n = 12 and 5 cells of Mode 1 and 2 sites, respectively). All significant values are provided from a paired t test. Light symbols: individual cells; dark symbols: individual animals.

Notably, while the amplitude of Mode 1 [ATP]_i transients was largely preserved (1.26 ± 0.12 vs. 1.24 ± 0.14 F/F₀ in female, P=0.3276; and 1.22±0.09 vs. 1.23±0.09 F/F₀ in male, P = 0.2310), Mode 2 [ATP]_i dips were significantly attenuated in both females (0.83 ± 0.09 vs. 0.92 ± 0.14 F/F₀, P = 0.0179) and males (0.82 ± 0.04 vs. 1.05 ± 0.04 F/F₀, P = 0.0005) (Fig. 10C,D). The [Ca²⁺]_i transient amplitude showed no evident change with blebbistatin in either mode or sex (Fig. 10E, F). These results, consistent with our prior observations in ventricular myocytes (Rhana et al., 2024), support the view that Mode 2 reflects local ATP depletion driven by contractile ATP utilization, whereas Mode 1 gains persist when cross-bridge ATP demand is reduced.

Increasing workload elevates [ATP]_i to a greater extent in male than female myocytes

To determine whether the ~ 20% lower mitochondrial volume in female ventricular myocytes slows cytosolic ATP replenishment during increased workload, we monitored diastolic [ATP]_i in cells paced at varying frequencies (Fig. 11). Upon switching from quiescence (0 Hz) to 2 Hz stimulation, diastolic [ATP]_i rose and stabilized at a higher set point that persisted throughout pacing (Fig. 11A–C). The new diastolic [ATP]_i values (in F/F₀ units) were 1.22 ± 0.11 in female and 1.25 ± 0.07 in male myocytes (P = 0.8967) (Fig. 11D). Notably, further increasing pacing to 3 Hz produced no additional rise (P = 0.9644), suggesting a physiological ceiling to workload-induced ATP enhancement.

Figure 11. Diastolic cytosolic ATP rises with increased pacing frequency.

A, representative confocal images of female (♀, left) and male (♂, right) ventricular myocytes expressing cyto-iATP. Fluorescence is shown before (1 min) and after (6 min) 2 Hz field stimulation. Scale bars, 10 μm. B and C, representative time course of [ATP]_i in normalized fluorescence (F/F₀) (B) and in micromolar units (C) during the 6 min experiment in female (magenta) and (male) myocytes. The grey box indicates the 2 Hz stimulation period (1–5 min). F/F₀ data were converted to intracellular ATP concentration (μM) using a pseudo-ratiometric method. D, scatter plot of diastolic [ATP]_i before and after 2 Hz stimulation as normalized fluorescence (F/F₀). E, scatter plot of diastolic [ATP]_i at the end of 2 Hz stimulation as concentration (μM). F, scatter plot of the time constant for the rise in [ATP]_i during pacing. Data are expressed as means ± SD (female: N = 4 mice, n = 16 cells; male: N = 5 mice, n = 10 cells). All significant values are provided from a nested t test. Light symbols: individual cells; dark symbols: individual animals.

Because the relationship between absolute cytosolic [ATP]_i and cyto-iATP fluorescence is non-linear (Lobas et al., 2019; Rhana et al., 2024), we converted F/F₀ values to estimated ATP concentrations using experimentally determined diastolic cytosolic [ATP]. We measured diastolic [ATP]_i in female myocytes (471 ± 81 μm) and found it to be similar to that in malemyocytes (457 ± 199 μm; P = 0.6169) (Rhana et al., 2024). Although the 2 Hz-evoked mean diastolic [ATP]_i trended lower in female (603 ± 69 μm) than male (642 ± 63 μm) myocytes, this difference was not statistically significant (P = 0.5303), indicating comparable resting ATP levels between sexes (Fig. 11E).

We then fitted the time course of the diastolic [ATP]_i rise (in μm) using a single-exponential function. Kinetic fits were applied to the estimated [ATP]_i rather than F/F₀ to avoid non-linearity-induced distortions in estimating time constants. This analysis revealed that female myocytes reached their new steady-state ATP level with a time constant (τ) of 3.62 ± 0.77 min, whereas male myocytes required ~23% longer (τ = 4.72 ± 0.94 min; P = 0.0405) (Fig. 11F). This sex difference supports the hypothesis that tighter SR–mitochondrial coupling in female myocytes enables more rapid activation of oxidative phosphorylation, while weaker coupling in male myocytes introduces a delay.

Together, these data suggest that the diastolic component of the cytosolic ATP response to increased workload is governed by intrinsic kinetic programmes modulated by both mitochondrial reserve and SR–mitochondrial coupling. Female myocytes achieve faster ATP upregulation despite reduced mitochondrial content, whereas male myocytes rely on greater mitochondrial mass but exhibit slower metabolic adaptation.

Differential modulation of beat-to-beat [ATP]_i transients by pacing frequency in male and female myocytes

To test how acute workload modulates cytosolic. ATP oscillations, we imaged cyto-iATP signals in ventricular myocytespacedfirst at1Hz and thenat2Hz (Fig. 12A, B). We selected 2Hz because diastolic [ATP]_i rises in response to this – but not higher – frequencies, as we mentioned in Fig. 11, making it the highest rate that reliably elevates energetic demand while preserving cellular integrity. Because diastolic [ATP]_i is higher at 2 Hz, beat-evoked F/F₀ values were converted to concentration units using sex-specific baselines (male, 642 μm; female, 603 μm).

Figure 12. Increased workload amplifies cytosolic ATP transients in male but not female myocytes.

A and B, representative time course of [ATP]_i from line scans of female (magenta) and male (blue) myocytes with Mode 1 (A) and Mode 2 (B) ATP dynamics field-stimulated at 1 Hz (left) or 2 Hz (right). Double-headed arrows indicate the net shift in [ΔATP]_i consumption or production (ATP) after increasing workload (2 Hz). C and D, scatter plots of the amplitudes of [ATP]_i transients of female (magenta) and male (blue) myocytes with Mode 1 (C) and Mode 2 (D) ATP dynamics under 1–2 Hz stimulation. E and F, scatter plots of ATP transients of female (magenta) and male (blue) myocytes with Mode 1 (E) and Mode 2 (F) ΔATP dynamics under 1–2 Hz stimulation. G and H, scatter plots of the amplitudes of [Ca²⁺]_i transients of female (magenta) and male (blue) myocytes with Mode 1 (G) and Mode 2 (H) ATP dynamics under 1–2 Hz stimulation. Data are presented as mean ± SD (female: N = 3 mice, n = 15 and 8 cells of Mode 1 and 2 sites, respectively; male: N = 3 mice, n = 16 and 15 cells of Mode 1 and 2 sites, respectively). All significant values are provided from a paired t test. Light symbols: individual cells; dark symbols: individual animals.

Pacing at 2 Hz increased the amplitude of Mode 1 ATP gains in both sexes (Fig. 12C) (females, 548 ± 31 → 722 ± 57 μm; males, 587 ± 58 → 811 ± 111 μm; each P < 0.0001), with males showing the larger absolute rise. Expressed as Δ[ATP]_i, values in male myocytes again exceeded those in female myocytes at both 1 Hz (130 ± 58 vs. 91 ± 31 μm; P = 0.0140) and 2 Hz (173 ± 111 vs. 114 ± 57 μm; P = 0.0350) (Fig. 12E).

Mode 2 regions, which exhibit ATP dips, behaved similarly at 1 Hz in both sexes. Raising the rate to 2 Hz shifted the dips to higher absolute concentrations (Fig. 12D) (females: 380±48→542±67 μm, P=0.0003; males: 383 ± 51 → 459 ± 73 μm; P = 0.0004), yet the depth of the dip (Δ[ATP]_i) became markedly larger in males (−180 ± 73 μm) than in females (−66 ± 67 μm; P = 0.0009) (Fig. 12F).

Notably, the accompanying [Ca2+]i transients increased from1 to 2 Hz in both sexes inMode 1 (Fig. 12G) (female: 2.17 ± 0.43 → 2.74 ± 0.74 F/F0, male: 1.91 ± 0.32 → 2.34 ± 0.51 F/F0, each P < 0.0001) and Mode 2 regions (Fig. 12H) (female: 2.29 ± 0.76 → 2.82 ± 1.04 F/F0, P = 0.0019; male: 2.06 ± 0.39 → 2.59 ± 0.60 F/F0, P < 0.0001), indicating that the frequency-dependent amplification of ATP gains and deepening of ATP dips occurs in the context of a larger Ca²⁺ signal. Taken together with our evidence of greater mitochondrial volume in males and tighter SR–mitochondrial coupling in females, these data support a model in which female myocytes meet heightened demand through architectural precision and efficient local production, whereas male myocytes rely on a mass-based scaling strategy that permits larger beat-coupled ATP swings but may approach capacity sooner. Thus, beat-synchronized ATP synthesis is modular and plastic, and its quantitative tuning differs between male and female cardiomyocytes.

Discussion

In this study, we demonstrate that ATP is not continuously produced in adult ventricular myocytes but instead is generated in a beat-to-beat fashion through an on-demand process that is tightly coupled to EC coupling. We directly visualize rhythmic mitochondrial ATP transients with millisecond precision and show that they are phase-locked to each action potential and propagate to the cytosol. These transients appear as spatially confined microdomains – some showing ATP gains, others ATP dips – with similar amplitudes but different regional prevalence in male and female myocytes. Thus, ATP supply is organized into discrete, temporally synchronized bursts that match the periodicity of cellular demand, such that ventricular myocytes ‘live paycheck-to-paycheck’, producing just enough ATP during each beat to fuel contraction.

By resolving ATP within the mitochondrial matrix, we extend our prior findings of cytosolic ATP oscillations during EC coupling (Rhana et al., 2024) and identify their origin as upstream pulses of mitochondrial ATP production. The coexistence of gain and dip domains within the same cell, together with their regional segregation, underscores that mitochondrial output is both phasic and spatially modular. Pharmacological dissection reinforces a causal chain from Ca²⁺ release to matrix ATP pulses and cytosolic ATP changes: mitochondrial uncoupling with FCCP abolished mito-iATP fluctuations and reduced matrix ATP levels, establishing dependence on intact oxidative phosphorylation, whereas the MCU blocker Ru360 markedly attenuated beat-locked mito-iATP signals while preserving cytosolic Ca²⁺ transients, indicating that Ca²⁺ entry through the MCU is necessary to drive these ATP pulses. Blocking the ANT with BKA reduced mito-iATP transients to near baseline while preserving Ca²⁺ cycling, demonstrating that beat-locked ATP synthesis and export through ANT are obligate steps in propagating mitochondrial ATP to the cytosol. Together with our earlier observation that Mode 1 cytosolic ATP gains are associated with increases in matrix Ca²⁺, whereas Mode 2 dips accompany Ca²⁺ loss (Rhana et al., 2024), these data support a unified mechanism in which SR-to-mitochondria Ca²⁺ transfer gates oxidative phosphorylation, ANT couples synthesis to export and the resulting ATP pulses are read out as cytosolic ATP microdomains.

Mitochondrial bioenergetics and Ca²⁺-ATP coupling are highly conserved across mammalian species, and the fundamental processes we interrogate – mitochondrial Ca²⁺ uptake, tricarboxylic acid (TCA) cycle activation and ATP synthesis – operate via the same molecular machinery in mouse and human cardiomyocytes. Our use of 1 Hz stimulation allowed us to study these conserved mechanisms under conditions directly relevant to human resting heart rates, while taking advantage of the genetic tractability of the mouse for mechanistic perturbation. Thus, although the experiments were performed in mouse ventricular myocytes, the underlying principles of beat-resolved mitochondrial ATP supply are likely to generalize to human myocardium.

An interesting aspect of our data is that mito-iATP signals – presumably arising from intrafibrillar mitochondrial ‘strings’ – extend over comparable distances in male and female myocytes (~30 μm), implying that each string contains roughly the same number of individual mitochondria (~15–20, assuming 1.5–2.0 μm length per mitochondrion in adult cells). Yet female myocytes have a smaller overall mitochondrial volume, and therefore fewer such strings per cell. On this reduced network, we find a higher per-mitochondrion density of the tethering protein Mfn2, the catalytic α-subunit of the F₁F_o-ATP synthase (ATP5A), and, importantly, greater local overlap between VDAC and SERCA immunosignals, quantified by pixel-wise multiplication of binarized VDAC and SERCA masks and measurement of the area of overlap. These structural data indicate that, in females, each mitochondrion is more likely to be anchored close to SR Ca²⁺ release sites and endowed with higher ATP synthase content, providing two complementary advantages: tighter Ca²⁺ microdomains and higher catalytic capacity at each favoured organelle.

This dual enrichment is consistent with the pronounced ATP ‘hot spots’ we observe in female cells – regions with larger [ATP]_mito transients where Ca²⁺ delivery (reflected by increased VDAC–SERCA colocalization and higher Mfn2) coincides with elevated ATP5A – alongside neighbouring regions with fewer tethers and lower ATP5A. In contrast, male myocytes show a larger mitochondrial footprint with lower ATP5A per mitochondrion and less VDAC–SERCA overlap, corresponding to more spatially homogeneous increases in [ATP]_mito and suggesting energetic scaling via abundance rather than microdomain precision. Prior work suggests sex-dependent differences in mitochondrial morphology/cristae organization and lower mitochondrial matrix Ca²⁺ uptake/accumulation in females (Arieli et al., 2004; Clements et al., 2023; Khalifa et al., 2017). Together, these features support the view that architectural precision, coupled with enhanced local catalytic capacity, allows female myocytes to meet localized energy demands efficiently. This sex-specific balance between mitochondrial abundance and per-organelle capacity may contribute to resilience to acute energetic stress (Ostadal & Ostadal, 2014).

Consistent with this microdomain view, recent systems-genetics work by Cao et al. (2022) uncovered a sex-biased gradient in global mitochondrial supply and performance. Across >100 inbred mouse strains and two human heart cohorts, their study reported that female ventricles carry ~ 10–20% fewer mitochondrial genomes and display coordinated down-regulation of nuclear- and mitochondrial DNA-encoded OXPHOS transcripts, and that mitochondria isolated from female hearts exhibit lower state-3 and maximal respiration and reduced complex II/IV flux. Moreover, gonadectomy plus hormone-replacement experiments showed that mitochondrial biogenesis is supported by androgens and restrained by oestrogens, and overexpression of the male-biased fatty-acyl–CoA ligase ACSL6 improved both mitochondrial respiration and diastolic function, particularly in females. Viewed alongside our data, these findings suggest that female hearts rely on a lean but precisely deployed mitochondrial reserve that suffices at baseline yet may have limited headroom under sustained high workloads, whereas male hearts buffer demand with greater mitochondrial mass and more globally distributed OXPHOS capacity.

This architectural divergence has clear functional implications under stress. When pacing was raised from 1 to 2 Hz, mitochondrial ATP output reached the cytosol within one beat and preserved the microdomain pattern, but the way each sex accommodated the increased workload differed. Mode 1 (ATP-gain) amplitudes increased in both sexes, with larger absolute gains in males emerging only at higher frequency, pointing to recruitment of their larger mitochondrial pool. In Mode 2 (ATP-dip) regions, trough [ATP]_i rose in both sexes, but male myocytes exhibited deeper negative excursions, whereas female myocytes showed shallower dips despite preserved gains. These observations suggest that male myocytes meet demand primarily by scaling flux through their greater mitochondrial mass, tolerating larger beat-coupled swings in ATP at the cost of smaller net gains under high consumption, whereas female myocytes rely on tighter SR–mitochondrial coupling and high-capacity microdomains to limit the depth of local ATP deficits and match production to use within each hotspot.

Temporal analysis revealed additional sex-specific distinctions. Most female myocytes re-established diastolic ATP levels within ~ 3.7 min after the onset of higher workload, consistent with rapid oxidative mobilization in pre-engaged, highly wired mitochondrial domains. Male myocytes exhibited a slower (~4.5 min) but more gradual increase in ATP, aligning with recruitment of a larger, more heterogeneous mitochondrial pool that requires time to build up matrix NADH and activate latent oxidative phosphorylation capacity. We interpret the fast kinetic phase as activation of mitochondria already positioned near SR release sites and operating close to maximal dehydrogenase activity, and the slower phase as reflecting mobilization of more distant or less well-tethered domains that depend on cumulative Ca²⁺ and redox signals. This refinement of the classical concept of ‘mitochondrial reserve capacity’ (Bertero & Maack, 2018) links reserve not only to enzyme excess but to spatial architecture and Ca²⁺ access, integrating structural sex differences (Cao et al., 2022) with functional disparities in ATP dynamics.

More broadly, our findings reframe cardiac energetics as modular and dynamic rather than homogeneous. Instead of a single, well-mixed ATP pool, mitochondrial output is organized into spatially restricted domains that deliver ATP in synchrony with each heartbeat, providing local control over supply and demand for ion transport, Ca²⁺ handling and contraction. Biological sex remodels this topology through differences in mitochondrial abundance, tether density and SR contact geometry, thereby shaping how the heart adapts to stress and matches energy supply with demand at the level of individual microdomains.

These principles resonate with observations in other excitable cells. In neurons, presynaptic mitochondria positioned near active zones supply ATP in proportion to firing demand, and disruption of this local coupling impairs synaptic transmission and plasticity (Rangaraju et al., 2014). In sinoatrial node pacemaker cells, earlier work showed that Ca²⁺-driven cAMP/PKA signalling scales mitochondrial ATP production to match changes in spontaneous firing on a seconds-to-minutes timescale (Yaniv et al., 2013), and our recent study in intact mouse sinoatrial node and isolated myocytes now demonstrates that ATP supply in this tissue is also organized into beat-locked cytosolic and mitochondrial ATP microdomains with high- and low-gain phenotypes, mapping a Ca²⁺-timed energetic hierarchy across the node (Munoz et al., 2025). Taken together with the present ventricular data, the convergence of spatially targeted mitochondria, activity-dependent Ca²⁺ entry and localized, phasic ATP delivery suggests that modular, time-locked ATP supply is a conserved strategy for sustaining the function of excitable cells.

Although we did not perform a full calibration of absolute matrix [ATP], the robust mito-iATP oscillations we observe have important implications for mitochondrial ATP levels. Based on the apparent K_d of 0.5 mm for iATPSnFR2 (Marvin et al., 2024), fluorescence changes of the magnitude we detect are most consistent with free mitochondrial [ATP] in the low-millimolar range (~1–5 mm), rather than the ~ 7.5–8 mm values inferred from biochemical or NMR measurements of total ATP. Taken together with our prior finding that cytosolic [ATP]i is <1 mm (Rhana et al., 2024), this yields an apparent paradox: if neither compartment harbours very high free ATP, how do we reconcile these values with the ~8 mm total cellular ATP reported by classic biochemical and NMR studies (e.g. Schwenke et al. 1981)? A plausible resolution, aligned with recent conceptual work by Eisner and Murphy (2024), is that most cellular ATP is bound to proteins or sequestered in compartments that are largely invisible to genetically encoded sensors, which report free nucleotide. In this revised view, both matrix and cytosol operate with relatively low free ATP, explaining why inhibition of oxidative phosphorylation does not lead to the large cytosolic ATP increases predicted by earlier models and why modest fractional changes in ATP can exert substantial control over ATP-dependent enzymes and channels. Motion artefacts are unlikely to account for our signals, as a JFX554 HaloTag reference channel showed no beat-locked changes while the mito-iATP ratio tracked linearly with non-ratiometric mito-iATP fluorescence, indicating that the ratio readout faithfully reports changes in matrix ATP rather than movement of mitochondria within the confocal volume.

Our findings also sharpen the link between cardiac energetics and coronary blood flow, placing functional hyperaemia at the centre of beat-resolved ATP homeostasis. Classic work reviewed by Feigl (1983) established that coronary flow is dynamically regulated to match myocardial oxygen demand, with local metabolic signals acting in tandem with myogenic and neural inputs to tune vascular resistance. Recent studies extend this framework by highlighting microvascular rarefaction and altered capillary–myocyte coupling as key determinants of contractile reserve and arrhythmia susceptibility (Manning et al., 2024, 2025). Within this context, the rapid, beat-synchronized nature of mitochondrial ATP synthesis that we describe implies that continuous delivery of oxygen and substrates through the coronary microcirculation is essential not just for long-term metabolic integrity but for preserving the energetic fidelity of each heartbeat. Even brief disruptions of blood flow or impaired functional hyperaemia could compromise the mitochondrial response to [Ca²⁺]_i transients, degrading contractile performance and electrical stability within a few cycles. Thus, perfusion and metabolism emerge as moment-to-moment co-regulators of cardiac function, tightly integrated through beat-locked ATP microdomains.

The capillary-mitochondria-ion channel (CMIC) axis proposed by Santana and Earley (2026) provides a useful lens for interpreting these findings across scales: microvascular delivery sets boundary conditions for oxidative metabolism, mitochondria generate ATP in a structured network, and ATP-dependent membrane/SR effectors convert local ATP availability into excitability and Ca²⁺ handling. While we did not directly manipulate the capillary network here, by resolving beat-locked matrix ATP microdomains and their propagation to cytosol – together with MCU and ANT dependence – we establish the mitochondrial ‘middle link’ of CMIC on the timescale of individual heartbeats. This framing predicts that limitations in oxygen/substrate delivery or capillary–myocyte coupling will be expressed as reduced fidelity of beat-synchronized ATP supply, with rapid consequences for electrical and mechanical stability. In addition, our male–female comparisons suggest that CMIC tuning can be achieved by distinct design strategies: greater mitochondrial mass in males versus tighter SR–mitochondrial wiring and higher per-organelle catalytic capacity in females, producing different frequency-dependent scaling of diastolic versus beat-locked ATP components under workload stress.

In summary, ATP supply in adult ventricular myocytes is shaped by a network of beat-coupled, spatially organized production units. Mitochondria discharge ATP in phasic bursts that are aligned with each action potential, exported via ANT and propagated to the cytosol to support rhythmic contractile demands. Sex-specific differences in mitochondrial mass, tethering and SR contact geometry tune this architecture, with male myocytes utilizing mitochondrial abundance and female myocytes achieving stability through precision wiring and higher per-organelle capacity. By providing the first direct visualization of beat-to-beat mitochondrial ATP transients in adult heart cells, our work compels a revision of traditional bioenergetic models – from static, well-mixed ATP pools to dynamic, microdomain-based supply – and suggests that cardiac metabolic resilience is governed as much by spatial design and Ca²⁺ access as by absolute mitochondrial capacity.

Supplementary Material

Additional supporting information can be found online in the Supporting Information section at the end of the HTML view of the article. Supporting information files available:

Key points.

• It is known that each heartbeat requires precise ATP delivery to fuel ion transport and cross-bridge cycling, but the timing and spatial organization of ATP production in heart cells has been unclear.

• Using advanced imaging and genetically encoded sensors, we visualized beat-to-beat ATP fluctuations in the mitochondria and cytosol of individual male and female mouse ventricular myocytes.

• Mitochondrial ATP levels rose or fell with each beat in spatially confined regions, forming ATP ‘gain’ or ‘dip’ microdomains that were synchronized with Ca²⁺ transients.

• At higher firing rates, beat-locked, diastolic ATP transients rose more quickly in female myocytes, but were larger in male myocytes, highlighting distinct sex-specific strategies for matching energy supply to contractile demand.

• Ventricular myocytes ‘live paycheck-to-paycheck’, producing just enough ATP on demand to fuel each beat. Male and female myocytes adopt distinct strategies to meet this demand: male myocytes scale output through greater mitochondrial mass, while female myocytes achieve energetic precision via enhanced sarcoplasmic reticulum–mitochondrial coupling.

Biography

Paula Rhana is an AHA Postdoctoral Fellow in Physiology & Membrane Biology, UC Davis. Trained in Biomedicine (BSc, FUMEC, Brazil) and Biochemistry & Immunology (MSc, PhD, UFMG, Brazil), Dr Rhana discovered how voltage-gated Na and Ca channels drive breast tumour progression. Upon joining the Santana lab in 2022, she led the first effort to identify and characterize beat-to-beat mitochondrial ATP transients in adult cardiomyocytes using combined imaging and electrophysiological approaches. Her current research centres on mapping sex-specific metabolic microdomains that tune cardiac excitability, with the ultimate goal of translating these energetic insights and fluorescent–sensor innovations into new strategies for preventing and treating heart failure.

Appendix

Figure A1.

A and B, representative time course from line scan images of [ATP]_mito (left) and [Ca²⁺]_i (right) in female (magenta) and male (blue) myocytes with Mode 1 (A) or Mode 2 (B) ATP dynamics. C and D, scatter plots for time to 90% of the [ATP]_mito transient peak (C) and decay (D) in Mode 1 and Mode 2 regions of female (magenta) and male (blue) myocytes. E and F, scatter plots for time to 90% of the [Ca²⁺]_i transient peak (E) and decay (F) in Mode 1 and Mode 2 regions of female (magenta) and male (blue) myocytes. Data are presented as mean ± SD (female: N = 3 mice, n = 28 and 17 cells of Mode 1 and 2 sites, respectively; male: N = 3 mice, n = 24 and 16 cells of Mode 1 and 2 sites, respectively). All significant values are provided from a nested t test. Light symbols: individual cells; dark symbols: individual animals.

Figure A2.

A and B, representative time course from line scan images of [ATP]_i (left) and [Ca²⁺]_i (right) in female (magenta) and male (blue) myocytes with Mode 1 (A) or Mode 2 (B) ATP dynamics. C and D, scatter plots for time to 90% of the [ATP]_i transient peak (C) and decay (D) in Mode 1 and Mode 2 regions of female (magenta) and male (blue) myocytes. E and F, scatter plots for time to 90% of the [Ca²⁺]_i transient peak (E) and decay (F) in Mode 1 and Mode 2 regions of female (magenta) and male (blue) myocytes. Data are presented as mean ± SD (female: N = 4 mice, n = 34 and 15 cells of Mode 1 and 2 sites, respectively; male: N = 3 mice, n = 24 and 18 cells of Mode 1 and 2 sites, respectively). All significant values are provided from a nested t test. Light symbols: individual cells; dark symbols: individual animals.

Data availability statement

The raw data files supporting all findings presented in this paper are available from the corresponding author.