The mitochondrial uniporter controls fight or flight heart rate increases

Heart rate increases are a fundamental adaptation to physiological stress, while inappropriate heart rate increases are resistant to current therapies. However, the metabolic mechanisms driving heart rate acceleration in cardiac pacemaker cells remain incompletely understood. The mitochondrial calcium uniporter (MCU) facilitates calcium entry into the mitochondrial matrix to stimulate metabolism. We developed mice with myocardial MCU inhibition by transgenic expression of a dominant negative (DN) MCU. Here we show that DN-MCU mice had normal resting heart rates but were incapable of physiological fight or flight heart rate acceleration. We found MCU function was essential for rapidly increasing mitochondrial calcium in pacemaker cells and that MCU enhanced oxidative phoshorylation was required to accelerate reloading of an intracellular calcium compartment prior to each heartbeat. Our findings show the MCU is necessary for complete physiological heart rate acceleration and suggest MCU inhibition could reduce inappropriate heart rate increases without affecting resting heart rate.


Introduction
Catecholamine agonists trigger physiological fight or flight increases in heart rate but the metabolic pathway(s) supplying ATP for increasing heart rate are incompletely understood 1,2 . Cardiac pacemaker cells drive heart rate acceleration, at least in part, by augmenting energy dependent flux of Ca 2+ through an intracellular, sarcoplasmic reticulum (SR), storage compartment 3 . SR Ca 2+ release triggers pacemaker cell membrane depolarization, leading to action potential initiation that triggers each heart beat 4 . Mitochondrial Ca 2+ entry through the mitochondrial Ca 2+ uniporter (MCU) can stimulate increased ATP production by enhancing activity of dehydrogenases in the mitochondrial matrix that supply NADH for electron transport 5,6 . The recent discovery of the gene encoding the MCU protein Ca 2+ permeation pore 7,8 allowed us to test the potential role of MCU-mediated mitochondrial Ca 2+ entry as a pathway for increasing ATP production to fuel heart rate increases. We developed new tools and approaches for studying the metabolic role of the MCU in cardiac pacing, including surgical gene transfer to pacemaker cells and transgenic mice with myocardial and pacemaker cell targeted expression of a dominant negative (DN) MCU with pore domain mutations that prevented rapid, MCU-mediated mitochondrial Ca 2+ entry. Here we show that Ca 2+ entry through the MCU is essential for telegraphing enhanced metabolic demand to pacemaker cell mitochondria and promoting oxidative phosphorylation. We found that isoproterenol (ISO) stimulates oxidative phosphorylation by the MCU pathway in cardiac pacemaker cells to fuel the activity of the sarcoplasmic-endoplasmic reticulum Ca 2+ ATPase (SERCA2a), which is required for reloading SR Ca 2+ stores and sustaining fight or flight heart rate increases. Inhibition of mitochondrial Ca 2+ entry prevented increased oxidative phosphorylation, enhanced SERCA2a activity and physiological rate responses in cardiac pacemaker cells exposed to ISO. Dialysis of cardiac pacemaker cells with exogenous ATP rescued the fight or flight response to ISO despite MCU inhibition but ATP dialysis was ineffective after SERCA2a inhibition, by expression of a super-inhibitory phospholamban (PLN) mutant or thapsigargin, identifying SERCA2a as a critical control point downstream of MCU for heart rate increases and a preferential sink for mitochondrially-sourced ATP. Isolated hearts from wild type mice with pacemaker-targeted DN MCU gene therapy were resistant to rate increases by ISO. We found selectively obtunded ISO triggered rate increases in isolated pacemaker cells, in excised Langendorff-perfused hearts from wild type mice with DN-MCU pacemaker targeted gene therapy and in vivo in DN-MCU transgenic mice. Furthermore, DN-MCU transgenic mice showed reduced heart rates in response to spontaneous activity compared to wild type littermate controls. In contrast to the profound loss of heart rate acceleration by MCU inhibition, unstimulated heart rates and autonomous pacemaker cell action potential firing were unaffected by loss of MCU mediated mitochondrial Ca 2+ entry. Our findings highlight a previously unrecognized subcellular mechanism for catecholamine-triggered heart rate increases and provide insight into a role for MCU-mediated mitochondrial Ca 2+ entry as a metabolic second messenger required for the physiological fight or flight stress response 9 . These results define the MCU as an essential activator of a metabolic pathway for heart rate control and suggest that MCU inhibition in cardiac pacemaker cells has therapeutic potential to selectively prevent excessive heart rates.

The MCU mediates rate increases in pacemaker cells
Isolated cardiac sinoatrial nodal (SAN) pacemaker cells spontaneously generate action potentials under basal conditions, in the absence of catecholamine stimulation. The rate of action potential initiation is increased with ISO, a catecholamine β adrenergic receptor agonist, in a concentration-dependent manner (Fig. 1a-d) 10 . We found that SAN cells dialyzed with Ru360 (5 μM), an MCU antagonist, had significantly reduced action potential frequency increases to ISO ( Fig. 1b and d) compared to SAN cells without Ru360 ( Fig. 1a  and d). The inhibitory effect of Ru360 on SAN cell action potential frequency responses to ISO was reversed by co-dialysis with ATP (4 mM), a concentration present in heart cells ( Fig. 1c and d) 11 . Co-dialysis with 1 and 2 mM ATP pipette solutions was inadequate to rescue ISO-mediated rate increases while 8 mM ATP did not result in greater rate responses than 4 mM ATP ( Supplementary Fig. 1a- Fig. 1a and b), indicating that basal heart rate was independent of an MCU pathway and that exogenous ATP did not affect Ru360independent cellular processes important for accelerating heart rate. We next corroborated our results with Ru360 dialysis by infecting SAN cells with adenovirus encoding a DN-MCU, containing pore domain charge reversal mutations that prevent MCU mediated mitochondrial Ca 2+ entry ( Fig. 1e-h) 7,8 . Both Ru360 and DN-MCU were similarly effective at preventing mitochondrial Ca 2+ entry ( Supplementary Fig. 2) and, like intracellular dialysis with Ru360, DN-MCU expression in SAN cells interfered with ISO induced rate increases ( Fig. 1f and h). The DN-MCU dependent loss of the ISO rate response was rescued by co-dialysis with ATP ( Fig. 1g and h), mirroring findings with MCU inhibition by Ru360 ( Fig. 1c and d). Cultured SAN cells, with and without DN-MCU expression, had slower basal spontaneous rates than freshly isolated counterparts (baseline in Fig. 1d and h) but nevertheless exhibited significant rate increases over baseline in response to ISO ( Fig. 1i and j). We considered the possibility that the mitochondrial Na + /Ca 2+ exchanger, a MCUindependent pathway for mitochondrial Ca 2+ efflux, could also contribute to SAN rate increases 12 by testing SAN responses to CGP-37157 (1 μM), a Na + /Ca 2+ exchanger antagonist. After CGP-37157 we measured a 1.2±6.0% increase in WT SAN cells (n=8) and 0.8±2.6% increase in DN-MCU SAN cells (n=3, P=0.88), suggesting that the mitochondrial Na + /Ca 2+ exchanger did not affect basal heart rate. After ISO (1 μM) we measured a 79.8±5.8% (n=4) increase from baseline rate in WT SAN cells and 32.0±11.1% increase from baseline (n=4) rate in DN-MCU SAN cells (P<0.01), similar to ISO mediated increases in the absence of CGP-37157 (Fig. 1i). These findings suggested that the mitochondrial Na + /Ca 2+ exchanger was unlikely to participate in SAN rate responses, consistent with other recent findings 13 . The rescue of physiological responses to ISO after MCU inhibition by ATP dialysis suggested that MCU activity was an upstream event required to supply adequate ATP for increasing heart rate.

MCU inhibition impairs heart rate acceleration
The selective control of pacemaker cell rate increases by MCU inhibition suggested that the MCU could be a novel target for controlling heart rate increases without slowing resting heart rates. In order to test this concept in hearts, we used a targeted gene painting approach to deliver adenovirus expressing DN-MCU or eGFP to the SAN in vivo 14 . One week after SAN gene painting, we verified SAN targeted gene expression ( Fig. 2a-b) and measured rate responses to ISO in excised, Langendorff-perfused hearts (Fig. 2c-f). We found DN-MCU SAN gene painting significantly and selectively reduced heart rate increases to ISO without affecting spontaneous heart rates in the absence of ISO. We next developed transgenic mice with myocardial DN-MCU expression (Fig. 3a, b and Supplementary Fig. 3) to test the role of MCU in heart rate acceleration in vivo. The DN-MCU mice had hearts with normal chamber size and function (Fig. 3c), detectable SAN DN-MCU expression (Fig.  3d), reduced ATP content (Fig. 3e) and complete loss of mitochondrial Ca 2+ uptake in myocardial cells ( Fig. 3f and g). DN-MCU hearts showed increased Mcu mRNA, as expected, and reduced mRNA expression for auxiliary MCU regulatory proteins ( Supplementary Fig. 4), possibly consistent with compensatory transcriptional reprogramming in response to loss of functional MCU. The reduced fight or flight response in DN-MCU hearts was selective for catecholamine agonist stimulation because DN-MCU and wild type SAN cells had similar rate responses to BayK 8644 (1 μM), a voltage-gated Ca 2+ channel agonist capable of accelerating heart rate independent of ISO. In the presence of BayK 8644 DN-MCU SAN cells (n=5) exhibited 69.9±10.9% rate increases over baseline, similar to published responses in wild type SAN cells 15 . During echocardiography measurements, lightly sedated DN-MCU mice exhibited resting heart rates similar to wild type littermate controls (Fig. 3c). Lightly sedated and restrained DN MCU and wild type mice during cutaneous ECG recording showed similar resting heart rates ( Fig. 3h-j), but reduced ISO stimulated heart rate increases ( Fig. 3j and k). DN-MCU mice had significantly prolonged P waves, PQ and PR intervals but similar QRS and QT interval durations compared to WT controls, suggesting that loss of MCU current slows atrial and atrioventricular conduction velocity without affecting conduction velocity in the distal conduction system or in ventricular myocardium (Supplementary Table 1). Finally, we measured heart rates in unrestrained and unsedated mice with surgically implanted ECG and activity telemeters. DN-MCU mice showed modest but significant reductions in basal heart rate compared to wild type controls and these differences were enhanced by activity ( Fig. 3l and m) and ISO ( Fig. 3n and o). Taken together, these data showed heart rate responses to spontaneous activity and ISO required MCU, while basal heart rates were independent of MCU, indicating MCU inhibition could selectively prevent heart rate increases in vivo.

MCU enables rapid refilling of SR Ca 2+ in pacemaker cells
Catecholamine stimulation increases heart rate by enhancing SAN cell membrane inward current and shortening the time between action potential firing 16 . Release of intracellular Ca 2+ from the SR provides the electrochemical driving force for the cell membrane Na + / Ca 2+ exchanger inward current (I NCX ) in SAN cells 17 . Elimination of SR Ca 2+ release by the toxin ryanodine or SERCA2a inhibition by thapsigargin ( Supplementary Fig. 5) significantly reduced the SAN response to ISO, findings that demonstrate the required connection between SR Ca 2+ release and physiological SAN cell acceleration 18 . To further test the apparent connection between MCU activity and SR Ca 2+ flux we performed confocal line scan measurements on SAN cells isolated from DN-MCU and wild type control mice (Fig. 4). DN-MCU SAN cells had significantly fewer diastolic Ca 2+ release events and reduced SR Ca 2+ content after ISO stimulation compared to wild type controls, suggesting that impaired fight or flight responses by MCU inhibition occurred because of reduced SR Ca 2+ flux during physiological stress. SERCA2a activity requires ATP to pump cytoplasmic Ca 2+ to the SR lumen 19 , so we measured the rate of decay of the cytoplasmic Ca 2+ concentration ([Ca 2+ ] cyto ) using a fluorescent indicator (Fura 2 AM, 0.1 μM) to test if SERCA2a was a sink for ATP produced by an Ru360-sensitive process. ISO significantly increased the rate of decline in [Ca 2+ ] cyto compared to SAN cells in control bath solution, reflecting enhanced activity of SERCA2a (Fig. 5a) 20 . Ru360 dialysis slowed the decline in [Ca 2+ ] cyto after ISO (Fig. 5b), while co-dialysis of ATP (4 mM) with Ru360 restored the rate of [Ca 2+ ] cyto decline to ISO-stimulated values present in the absence of Ru360 ( Fig. 5c and d). However, Ru360 did not slow the decline in [Ca 2+ ] cyto in the absence of ISO stimulation ( Fig. 5e), consistent with the lack of effect of Ru360 on SERCA2a activity or basal SAN cell action potential frequency. In contrast to the effect of ATP dialysis on SAN cells exposed to Ru360, ATP dialysis did not significantly increase the rate of [Ca 2+ ] cyto decline ( Fig. 5f-h) or SAN action potential frequency ( Fig. 5i) in SAN cells isolated from mice expressing a super inhibitory mutant form of phospholamban (N27A) 21 that constrains SERCA2a despite ISO stimulation. These results show that ISO increases SAN cell rates by actions that require MCU and SERCA2a. We considered the possibility MCU inhibition was somehow affecting the ability of ISO to enhance phospholamban phosphorylation, which reduces the inhibitory actions of phospholamban on SERCA2a 22 . We found that atrial tissues from DN-MCU and wild type littermates had similar increases in phospholamban phosphorylation after ISO ( Supplementary Fig. 6), suggesting that MCU inhibition did not interfere with SERCA2a activity by actions on phospholamban nor did DN-MCU expression promiscuously affect downstream signaling actions of ISO. We interpreted the rescue of ISO responses by exogenous ATP after elimination of MCU-mediated Ca 2+ entry but not after SERCA2a inhibition to suggest that MCU contributes to ATP synthesis targeted for SERCA2a consumption during physiological stress.

MCU is not a global effector of pacemaker currents
Because heart rate is responsive to multiple ionic currents 16 , we next asked if the MCU pathway affected the ATP and 3′-5′-cyclic adenosine monophosphate (cAMP) dependent cell membrane ion channel (HCN4) inward current (I f ) 23 an ionic current known to participate in SAN cell automaticity. We found that I f responses to ISO in isolated SAN cells were not reduced by Ru360 dialysis (Fig. 6a and b). The lack of effect of Ru360 on I f suggested that Ru360 did not reduce cAMP nor ATP availability globally in SAN cells below a threshold necessary to increase I f . To test this concept further, we measured the maximum diastolic cell membrane potential, which is primarily determined by activity of the Na + /K + ATPase 16 . MCU inhibition by Ru360 dialysis or by transgenic expression of DN-MCU did not affect the maximum diastolic membrane potential in isolated SAN cells (Fig. 6c). We also measured Ca V 1 L-type Ca 2+ current (I Ca ), an SAN cell membrane inward current enhanced by ISO through ATP-mediated phosphorylation 24 . Similar to our findings with I f , ISO-induced I Ca increases were not impaired by Ru360 ( Fig. 6d and e). These findings were consistent with a model where selective loss of heart rate acceleration after ISO by Ru360 was primarily or exclusively related to actions on SERCA2a.

MCU is required for ISO to increase NADH
Mitochondrial Ca 2+ entry increases oxidative phosphorylation by enhancing the activity of key mitochondrial dehydrogenases to provide NADH/NADPH reducing equivalents required for ATP synthesis 25 . This mechanism is activated when cellular Ca 2+ enters the inner mitochondrial membrane space from the cytosol through the MCU pathway 26 . We first asked if mitochondrial Ca 2+ entry was critical for oxidative phosphorylation-dependent ATP synthesis in SAN pacemaker cells. We infected cultured mouse SAN cells with adenovirus encoding mt-pericam 27 , a circularly permutated Ca 2+ -sensitive fluorescent protein, to measure mitochondrial Ca 2+ concentration ([Ca 2+ ] mito ). The mt-pericam expression was localized to mitochondria in adenovirus infected SAN pacemaker cells, based on co-localization with MitoTracker Orange (Fig. 7a). ISO caused an increase in [Ca 2+ ] mito (Fig. 7b) that was prevented by dialysis of Ru360 (Fig 3b-d). Mitochondrial Ca 2+ enhances ATP production by augmenting NADH, the primary electron donor for electron transport 28 . We next measured NADH fluorescence at baseline and after addition of ISO. ISO increased NADH fluorescence and this increase was prevented by Ru360 ( Fig. 8a- Supplementary Fig. 7a and b) and in SAN cells with transgenic DN-MCU expression (Supplementary Fig. 7c and d), suggesting that ISO enhancement of NADH required MCUmediated mitochondrial Ca 2+ entry. These data confirm the MCU-dependence of coupling between [Ca 2+ ] cyto and [Ca 2+ ] mito in SAN cells and show that the MCU provides critical metabolic support for SERCA2a activity and SR Ca 2+ loading. Together these data are consistent with a model where MCU-mediated enhancement of oxidative phosphorylation is a metabolic mechanism enabling the physiological fight or flight stress response in cardiac pacemaker cells (Fig. 8c).

Discussion
Our data provide new mechanistic understanding into the fight or flight response to physiological stress by showing that heart rate increases rely on the MCU in cardiac pacemaker cells. In contrast, basal rates do require SERCA activity but are independent of MCU, suggesting availability of MCU-independent ATP production is sufficient to sustain heart rates in the absence of extreme physiological stress. Oxidative phosphorylation is enhanced by [Ca 2+ ] mito , which is required to generate ATP that fuels SERCA2a activity under extreme physiological stress. Our findings show that Ca 2+ homeostatic mechanisms in pacemaker cells form the framework for fight or flight heart rate increases but do not exclude additional modulation of heart rate by other Ca 2+ sensitive signals 29,30 or by Ca 2+ independent ionic currents 16 . The MCU metabolic pathway appears optimized to generate heart rate increases during episodes of high energy demand that are signaled by catecholamines. Our data are consistent with earlier work showing mitochondrial Ca 2+ is required to optimize refilling of intracellular Ca 2+ stores by SERCA2a 31 and where ATP dialysis (3 mM) recovered Ca 2+ sequestration by intracellular stores after addition of mitochondrial toxins 32 . Our findings provide insight into recent work showing MCU knock out selectively impairs high workload activity in striated muscle 33 . However, our data show that basal pacemaker cell activity is uncoupled from MCU-dependent ATP production. Because the relationship between MCU and SERCA2a in pacemaker cells appears purposed to selectively enable heart rate acceleration, future therapies targeting MCU or SERCA2a in pacemaker cells could provide a means to fine tune heart rates by preventing excessive heart rates without reducing resting heart rates.

Methods
All experiments were carried out in accordance with the guidelines of Institutional Animal Care and Use Committee (PHS Animal Welfare Assurance, A3021-01).

NADH and mitochondrial calcium measurements
The autofluorescence of endogenous NADH, which derives primarily from mitochondria, 40, 41 was measured as described 41 . In brief, NADH was excited at 350 nm (AT350/50X, Chroma) and fluorescence was recorded at 460 nm (ET460/50m and T400LP, Chroma). We normalized NADH level with FCCP as 0%, Rotenone induced NADH change as 100%. The baseline level of NADH was 22%±3 of the rotenone-induced maximal value.
Mt-pericam, a mitochondrial matrix-targeted, circularly-permuted green fluorescent protein fused to calmodulin and its target peptide M13 27 . Pericam emission at 535 nm due to excitation at 415 nm reports changes in Ca 2+ 42 . We used Mt-pericam to measure Ca 2+ mito in isolated, cultured SAN cells. The mitochondrial Ca 2+ level was monitored in cells transiently expressing the Mt-pericam protein at excitation wavelengths of 415 nm, presented as 1-F/F 0 (Ca 2+ mito ) [42][43][44] , and the emission collected using a 535-nm band-pass filter.

Intracellular Ca 2+ transients
Cytosolic Ca 2+ levels were recorded from Fura-2-loaded cells, excited at wavelengths of 340 and 380 nm, and imaged with a 510 nm long-pass filter. Single isolated SAN cells were loaded with 0.1 μM Fura-2 AM for 20 minutes, and then perfused for 20 minutes to deesterfy the Fura-2 AM in normal Tyrode's solution. After placement on a recording chamber, the cells were perfused in normal Tyrode's solution at 36°C±0.5. Spindle-shaped, spontaneously beating cells were chosen for experiments. Action potential recording was performed simultaneously.

SAN gene painting
SAN painting was performed as previously described 14,45 . Briefly, Poloxamer 407 (Spectrum), trypsin (Sigma) and collagenase, type II (Worthington) mixture was made with 40% Poloxamer, 1% trypsin and 0.25% collagenase in PBS, and then added to equal volume of recombinant adenovirus expressing cDNA for the gene(s) of interest (DN-MCU-IRES GFP vs eGFP) in solution. This mixture was liquid in consistency at 4°C, but gelled at 37° C. Mice (6-7 weeks, females and males) were anesthetized using ketamine/xylazine (87.5/12.5 mg/kg respectively), intubated and ventilated. The junction of the superior vena cava and right atrium was visualized through a small incision in the 2nd intercostal space. The gel was applied to the posterior surface of the junction of the superior vena cava and right atrium with a fine brush. The intercostal muscles, pectoralis major and minor and the skin incision were closed using 6/0 silk and the mice were allowed to recover.

Ex vivo Langendorff-perfused heart rate measurements
ECG recording from Langendorff-perfused hearts was performed as described 29 . Briefly, excised hearts from 7-9 week female and male mice were rapidly mounted on a modified Langendorff apparatus (HSE-HA perfusion systems, Harvard Apparatus, Holliston, Mass) for retrograde aortic perfusion at a constant pressure of 80 mm Hg with oxygenated (95% O 2 , 5% CO 2 ) Krebs-Henseleit solution consisting of (mM) 25.0 NaHCO 3 , 118.5 NaCl, 4.0 KCl, 1.2 MgSO 4 , 1.2 NaH 2 PO 4 , 1.5 CaCl 2 , and 11.2 glucose, with pH equilibrated to 7.4. Each perfused heart was immersed in a water-jacketed bath and was maintained at 36°C. ECG measurements from the intact heart were continuously recorded with Ag + -AgCl electrodes, which were positioned around the heart in an approximate Einthoven configuration. After the heart was allowed to stabilize for 15 minutes, different concentrations of isoproterenol were added to the perfusate.

Calcium Green mitochondrial Ca 2+ uptake assays
We measured mitochondrial Ca 2+ uptake using permeabilized HEK cells as previously described 46,47 . Briefly, cells were grown in DMEM, with 10% FBS and 1% of Penicillin/ Streptomycin. At 80% confluence cells were infected with DN-MCU adenovirus at MOI 10. Cells were harvested after 24 h incubation and placed into a 96 well plate. Each well was loaded with 1 million cells in respiratory buffer containing 125.0 mM KCl, 2.0 mM K 2 HPO 4 , 20.0 mM HEPES, 5.0 mM glutamate, 5.0 mM malate, 0.005% saponin, and 5 μM thapsigargin. 5 μM Ca 2+ was injected at 3 minute intervals. Fluorescence was measured using a Tecan plate reader. Adult cardiac myocytes were isolated from 6-8 week old wild type or DN-MCU TG mice using a previously described isolation procedure 42 . 50 nM Blebbistatin was included in the myocyte buffer to prevent cellular contraction. Fluorescence intensity was measured from 50,000 cardiac myocytes/well. 100 μM Ca 2+ was injected at 3 min intervals for myocyte experiments.

DN-MCU overexpressing mice
The inter-membrane D 260 IME 263 amino acid motif of human MCU was mutated to the dominant-negative (DN) form 8 , QIMQ, by replacing the nucleotide sequence, gacatcatggag with cagatcatgcag using site-directed mutagenesis (agilent.com). The resulting DN-MCU DNA product was amplified by PCR with Phusion DNA polymerase using a forward primer containing a SalI restriction site (5′-ACC AAC GTC GAC ATG GCG GCC GCC GCA GGT AG-3′) and reverse primer containing a C-terminal myc epitope tag sequence and HindIII restriction site (5′-GAA CGC AAG CTT CCT ACA GGT CTT CTT CGC TAA TCA GTT TCT GTT CAT CTT TTT CAC CAA TTT GTC GGA G-3′). PCR products were digested and ligated into SalI and HindIII digested pBS-αMHC-script-hGH vector and positive clones confirmed by sequencing. Mouse embryonic stem cells were injected with the linearized DNA (digested with NotI) in the University of Iowa Transgenic Mouse Core Facility and implanted into pseudo-pregnant females to generate B6XSJL F1 mice. Insertion of the transgene into the mouse genome was confirmed by PCR analysis (not shown) using the forward primer, CCCACACCAGAAATGACAGACAGAT and reverse primer, AGAGGAGCAGCAGGAGCGATCTA, producing a product of 200 bases. Mice were backcrossed to F4 generation or greater into the CD1 background. Transgenic and control mice of either gender were sacrificed at the age of 2-3 months.

Western blots for detecting phospholamban
Heart lysates were prepared from flash-frozen mouse right atria from 6-8 week female and male mice. Mice were injected with ISO (0.4 mg/kg) 10 min before harvesting the right atrium and western blotting was performed with a SDS-PAGE electrophoresis system as described 29 . Briefly, 30 μg protein samples were size-fractionated on SDS-PAGE, and then transferred to PVDF membranes. The membranes were probed with anti-pSer16-PLN (1:5,000), anti-pThr17-PLN (1:5,000) (from Badrilla Ltd., Leeds, United Kingdom) at room temperature for 4 hours. Then membranes were incubated with Alexa-Fluor680-conjugated anti-mouse (Invitrogen Molecular Probes, Carlsbad, CA) and/or IR800Dyeconjugated antirabbit fluorescent secondary antibodies (Rockland Immunochemicals, Gilbertsville, PA), and scanned on an Odyssey infrared scanner (Li-Cor, Lincoln, NE). Integrated densities of protein bands were measured using ImageJ Data Acquisition Software (National Institute of Health, Bethesda, MD). Uncropped scans of all Western blot images are available in the SI.

Mouse surface electrocardiograms
Mouse surface electrocardiogram (ECG) tracings were acquired as described 48 . Prior to the ECG acquisition 6-8 week female and male mice were pre-anesthetized with 2% isoflurane in 1l oxygen/min (Isotec100 Series, Isoflurane Vaporizer, Harvard Apparatus, Holliston, USA). Mice were placed in a supine position on a heated ECG pad (Mousepad, THM 100, Indus Intruments, Webster, USA), and the limbs were attached to the pad electrodes using a tape to obtain ECG lead II. Anesthesia was maintained via facemask by continuous isoflurane ventilation as described above. The body temperature was continuously monitored using a rectal probe and sustained within 36-37°C. To obtain a baseline ECG, animals were allowed to rest for 5 min after being positioned on the pad. Thereafter Isoproterenol (10 μg/kg) was injected i.p. ECG acquisition was performed continuously using a multichannel amplifier and data acquisition system (Powerlab 16/30, AD Instruments, Colorado Springs, USA) converting the signal into digital for a further data analysis (Labchart Pro software, version 7, AD Instruments, Colorado Springs, USA).

Laser scanning confocal imaging of single SAN cells
Laser scanning confocal imaging of single SAN cells were performed as described 49 . Mouse SAN cells isolated from WT or DN-MCU mice were loaded with Fluo-4 AM (5 μM) for 20 min at room temperature. After 20 min of de-esterification, the cells were placed in a recording chamber and perfused with normal Tyrode solution (1.8 mM Ca 2+ ) at 36±1 °C (Temperature Controller, TC2BIP, Cell MicroControls). Spindle-shaped, active spontaneously beating cells were chosen for experiments. Confocal Ca 2+ imaging was performed in line-scan mode with a laser scanning confocal microscope (LSM 510, Carl Zeiss) equipped with a numerical aperture (NA) 1.35, 63X lenses. Images of spontaneously beating Ca 2+ transients and caffeine induced Ca 2+ transients (SR Ca 2+ contents) were acquired at a sampling rate of 1.93 ms per line along the longitudinal axis of the cells. SR Ca 2+ content was determined by measuring the amplitude of Ca 2+ release induced by local delivery of 20 mM caffeine. All digital images were processed with IDL 6.0 program (Research System Inc).

Electrocardiographic Telemetry
Surgical implantation of ECG telemeters was performed as described 29 . In brief, 8-10 week female and male mice were anesthetized with ketamine/xylazine (87.5/12.5 mg/kg), and an ECG transmitter (DSI model Ta10EA-F20) was implanted in the abdominal cavity. The leads were placed subcutaneously in a lead I configuration. Experiments were performed after a 5-day recovery period. ECG and activity were recorded for 72 hours from undisturbed mice at a sampling rate of 1 kHz. Heart rate and activity were calculated from serial 1-minute averages. On the fourth day there was a 1 hr baseline ECG recorded before ISO injection followed by an ISO injection and an additional 1 hour recording epoch. On the fifth day we repeated this protocol with another dose of ISO. The last dose of ISO was given on the sixth day. The heart rate in response to ISO was calculated from serial 10 seconds averages.

Echocardiography
Transthoracic echocardiography was performed as previously described 50 . Unanesthetized, sedated mice were used for echocardiography. A 30 MHz linear array transducer was applied to the chest to obtain cardiac images. The transducer was coupled to a Vevo 2100 imager (FUJIFILM Visual Sonics, Toronto, Canada). Images of the short and long axis were obtained with a frame rate of ~180-250 hertz. All image analysis was performed offline using Vevo 2100 analysis software (Version 1.5).

ATP Measurements
6-8 week old littermate mice were sacrificed. Atria were rapidly harvested and flash frozen in liquid nitrogen. Atria were homogenized, sonicated, and centrifuged at 10000 X g. The supernatants were collected and a luciferase assay (Invitrogen, ATP Determination Kit A22066) was used to detect ATP. Samples were measured in triplicate on a Femtomaster FB 12 luminometer (Zylux).

Quantitative RT PCR
RNA was isolated using Trizol reagent (Invitrogen) and purified using a RNeasy MinElute Cleanup Kit (Qiagen). RNA concentrations were measured using a Nanodrop 2000 Spectrophotometer (Thermo Scientific). iScript cDNA Synthesis Kit (Bio-Rad) was used to generate cDNA from RNA using oligo(dT) primers. Validated PCR primers (Bio-Rad) were used for qRT PCR on a StepOnePlus Real-Time PCR system (Applied Biosystems). Transcript levels were quantified by the ΔΔCt method.

Statistical analysis
Data are presented as mean±SEM, unless otherwise noted. Statistical analysis was performed either with 1-way ANOVA or an unpaired or paired Student's t test, as appropriate. The Holm-Sidak test was used for post-hoc comparisons after ANOVA. Analyses were performed with Sigmaplot or Sigmastat (Systat Software, Inc. San Jose, CA 95110 USA). The null hypothesis was rejected for a P<0.05.    c, Summaries of spontaneous diastolic calcium release events and SR calcium content before and after ISO. Scatter graph of diastolic calcium events before and after ISO (upper panel). Bar graph of diastolic calcium release events before and after ISO (middle panel). Bar graph of SR calcium content determined by caffeine-evoked calcium release before and after ISO (lower panel). *p<0.05, **p<0.01 vs control, #p<0.05 vs WT by Student's t-test, n=11-21 cells/group.  Fig. 1a-c. Ca 2+ transients were normalized to peak values for analysis of the decay phase. The decay phase of each tracing (marked by blue background), between the dashed lines marking 90-10% of peak values, shows rates of cytoplasmic Ca 2+ sequestration. Scale bars are 100 ms. d, Summary data for the cytoplasmic Ca 2+ decay slope from 90% to 10% of the peak amplitude before and after ISO. *p<0.05 versus all other groups. n=5-10/group, one way ANOVA. e, Superimposed (baseline) traces from panels ac show Ru360 or Ru360+ATP did not affect baseline Ca 2+ sequestration compared to control (−Ru360). Scale bar is 100 ms. f-i, The PLN super-inhibitor N27A mutant slowed SAN cell rates and cytoplasmic Ca 2+ sequestration by a mechanism that was resistant to rescue by ATP dialysis. f and g, Example tracings as in (a-c). Scale bars are 100 ms. h, Summary data for the slope of the cytoplasmic Ca 2+ decay, as in (d), NS between the groups. n=5-10/group, unpaired t test. i, Summary dose-response data for AP rate responses to ISO in SAN cells from N27A transgenic mice with and without ATP dialysis. NS for all comparisons, n=5-10/group, unpaired t test. Error bars indicate SEM. NS for I f current density in Ru360-dialyzed compared to SAN cells without Ru360 before and after ISO, unpaired t test. c, Maximum diastolic potential of spontaneous action potentials was not altered by MCU inhibition. n=12/group. d, Representative I Ca current recordings in response to voltage clamp commands (top panel) from an SAN cell without Ru360 (−Ru360, middle panel) and an Ru360-dialyzed SAN cell (+Ru360, lower panel) before (black trace) and after ISO (blue trace). e, Summary current-voltage relationship for I Ca from groups shown above. n=6-8/group. NS for I Ca current density in +Ru360 compared to −Ru360 cells before and after ISO, unpaired t test. Error bars indicate SEM.