Migraine‐Associated Mutation in the Na,K‐ATPase Leads to Disturbances in Cardiac Metabolism and Reduced Cardiac Function

Abstract

Background

Mutations in ATP1A2 gene encoding the Na,K‐ATPase α₂ isoform are associated with familial hemiplegic migraine type 2. Migraine with aura is a known risk factor for heart disease. The Na,K‐ATPase is important for cardiac function, but its role for heart disease remains unknown. We hypothesized that ATP1A2 is a susceptibility gene for heart disease and aimed to assess the underlying disease mechanism.

Methods and Results

Mice heterozygous for the familial hemiplegic migraine type 2–associated G301R mutation in the Atp1a2 gene (α₂ ^+/G301R mice) and matching wild‐type controls were compared. Reduced expression of the Na,K‐ATPase α₂ isoform and increased expression of the α₁ isoform were observed in hearts from α₂ ^+/G301R mice (Western blot). Left ventricular dilation and reduced ejection fraction were shown in hearts from 8‐month‐old α₂ ^+/G301R mice (cardiac magnetic resonance imaging), and this was associated with reduced nocturnal blood pressure (radiotelemetry). Cardiac function and blood pressure of 3‐month‐old α₂ ^+/G301R mice were similar to wild‐type mice. Amplified Na,K‐ATPase–dependent Src kinase/Ras/Erk1/2 (p44/42 mitogen‐activated protein kinase) signaling was observed in hearts from 8‐month‐old α₂ ^+/G301R mice, and this was associated with mitochondrial uncoupling (respirometry), increased oxidative stress (malondialdehyde measurements), and a heart failure–associated metabolic shift (hyperpolarized magnetic resonance). Mitochondrial membrane potential (5,5´,6,6´‐tetrachloro‐1,1´,3,3´‐tetraethylbenzimidazolocarbocyanine iodide dye assay) and mitochondrial ultrastructure (transmission electron microscopy) were similar between the groups. Proteomics of heart tissue further suggested amplified Src/Ras/Erk1/2 signaling and increased oxidative stress and provided the molecular basis for systolic dysfunction in 8‐month‐old α₂ ^+/G301R mice.

Conclusions

Our findings suggest that ATP1A2 mutation leads to disturbed cardiac metabolism and reduced cardiac function mediated via Na,K‐ATPase–dependent reactive oxygen species signaling through the Src/Ras/Erk1/2 pathway.

Nonstandard Abbreviations and Acronyms

Erk1/2

p44/42 mitogen‐activated protein kinase

fhm2

familial hemiplegic migraine type 2

IPA

ingenuity pathway analysis

PLCγ

phospholipase Cγ

jc‐1

5,5´,6,6´‐tetrachloro‐1,1´,3,3´‐tetraethylbenzimidazolocarbocyanine iodide

ROS

reactive oxygen species

WT

wild type

Clinical Perspective

What Is New?

• Familial hemiplegic migraine‐associated mutation of the Na,K‐ATPase α₂ isoform, encoded by the ATP1A2 gene, was associated with mitochondrial uncoupling and high levels of oxidative stress in the heart.

• The disturbances in cardiac metabolism were associated with dilation of the left ventricle and reduced ejection fraction in elderly mice carrying the mutation.

• Our data indicated that imbalanced Na,K‐ATPase–dependent Src/Ras/Erk1/2 signaling in cardiomyocytes underlies mitochondrial uncoupling, leading to increased generation of reactive oxygen species.

What Are the Clinical Implications?

• Our finding of a link between a subtype of migraine and cardiac dysfunction provides important novel insight into the otherwise inexplicable comorbidity between migraine and cardiovascular disease.

• The ATP1A2 gene may be considered a risk gene for cardiovascular disease.

• Our study suggests that monitoring and special attention to cardiovascular health are warranted in patients carrying migraine‐associated ATP1A2 missense variants.

The Na,K‐ATPase is an ion pump essential for maintaining sodium and potassium ion gradients across the plasma membrane in all mammalian cells. In the cardiomyocyte membrane, several catalytic α subunits of the Na,K‐ATPase have distinct subcellular localizations ¹ , ² and are suggested to serve different cell functions. ³ , ⁴ , ⁵ It has been proposed that the α₂ isoform is localized mainly in transverse tubules at the junctions with the sarcoplasmic reticulum, whereas the quantitatively prevalent α₁ isoform is ubiquitously distributed in the plasma membrane of cardiomyocytes. ¹ , ² , ⁴ The α₁ isoform Na,K‐ATPase was proposed to serve mainly a housekeeping role for intracellular ion homeostasis, whereas the α₂ isoform has been suggested to serve more specific regulatory functions. ² Expressional changes in the Na,K‐ATPase have dramatic consequences for cardiac function. A lineal correlation between left ventricular ejection fraction and the Na,K‐ATPase content in the cardiac tissue has previously been shown. ⁶ Accordingly, reduction in the Na,K‐ATPase was reported in patients with heart failure ⁷ and dilated cardiomyopathy. ⁸ Moreover, experimental models suggest the importance of the α₂ isoform Na,K‐ATPase for cardiac pathology. ⁴ , ⁵ , ⁹ Thus, global heterozygote knockout of the Na,K‐ATPase α₂ isoform leads to cardiac hypercontractility attributable to increased intracellular Ca²⁺ transients. ¹⁰ It has been proposed that the α₂ isoform modulates spatially restricted intracellular Na⁺ concentrations ¹¹ and consequently intracellular Ca²⁺ concentrations via interaction with the Na,Ca‐exchanger. ³ , ⁴ , ¹² Intracellular Na⁺ concentrations, controlled by Na⁺ channels, Na⁺/Ca²⁺ exchange, and the Na,K‐ATPase, are vital for preserving the electrical and contractile activity of the heart. ¹³

In addition to controlling intracellular ion homeostasis, the Na,K‐ATPase α isoforms are proposed to be implicated in ion transport–independent protein kinase signaling cascades, ¹⁴ including the Na,K‐ATPase–Src kinase signal transduction. ¹⁵ Na,K‐ATPase inhibition leads to activation of Src kinase. Increased activation of Src kinase in cardiomyocytes leads to initiation of downstream signaling pathways (ie, Src/Ras/Erk1/2 [p44/42 mitogen‐activated protein kinase] and Src/phospholipase Cγ [PLCγ]/inositol trisphosphate‐receptor signaling). The Src/Ras/Erk1/2 pathway is associated with mitochondrial generation of reactive oxygen species (ROS), ¹⁶ whereas the Src/PLCγ/inositol trisphosphate‐receptor pathway is suggested to regulate intracellular Ca²⁺ homeostasis. These signaling pathways initiated from the cardiac Na,K‐ATPase have previously been implicated in changes of cardiac morphology and function, ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ including the Na,K‐ATPase/ROS signaling playing a key role in experimental uremic cardiomyopathy. ²¹ Inhibition of this signaling pathway ameliorates cardiomyopathic changes. ²² Although many studies suggest an extraordinary significance of the Na,K‐ATPase signaling, the specific signaling pathways involved in cardiac pathologies remain to be elucidated.

Mutation in the ATP1A2 gene encoding the Na,K‐ATPase α₂ isoform is associated with familial hemiplegic migraine type 2 (FHM2; OMIM No. 602481), a severe form of migraine with aura. ²³ Nationwide population‐based cohort studies have recently characterized migraine, particularly migraine with aura, as a significant risk factor for development of cardiovascular disease. ²⁴ We hypothesized that the FHM2‐associated mutation of the Na,K‐ATPase α₂ isoform leads to heart disease caused by mitochondrial dysfunction and oxidative stress as a consequence of changed protein kinase signaling pathways downstream from the Na,K‐ATPase and aimed to characterize the involved signaling.

We tested our hypothesis in heterozygous mice carrying the G301R mutation in the Atp1a2 gene ²⁵ (Figure S1), which is associated with a severe phenotype in patients with FHM2. ²⁶ The α₂ ^+/G301R mouse model was previously shown to phenocopy FHM2‐relevant disease traits (ie, neuropsychiatric ²⁵ and cerebrovascular ²⁷ , ²⁸ manifestations). We compared cardiac function in 3‐ and 8‐month‐old α₂ ^+/G301R mice with age‐matched wild‐type (WT) controls, addressing a progression of the pathology associated with decreased cardiac expression of the Na,K‐ATPase α₂ isoform. The functional analyses in vivo (ie, cardiac magnetic resonance imaging with and without hyperpolarization and radiotelemetry) and ex vivo (respirometry, membrane potential measurements, and transmission electron microscopy) were further strengthened by expressional and biochemical measurements as well as proteomics of heart samples. The results of this study provide mechanistic insight into how FHM2‐associated mutation in the Na,K‐ATPase α₂ isoform may lead to increased risk of heart failure and cardiomyopathy‐like changes.

Methods

The data that support the findings of this study are available from the corresponding author on reasonable request.

All procedures were performed according to the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes. The experiments were approved by and conducted with permission from the Animal Experiments Inspectorate of the Danish Ministry of Environment and Food (No. 2016‐15−0201‐00982). Animal experiments were reported in accordance with the Animal Research: Reporting in Vivo Experiments guidelines.

The α₂ ^+/G301R mice were generated and bred, as described previously. ²⁵ , ²⁷ Mice were housed under a 12:12 light/dark cycle, and food and water were provided ad libitum. Approximately 3‐ and 8‐month‐old heterozygous α₂ ^+/G301R mice (homozygous pups died immediately after birth) were used in the current study. A previous study found sex‐coupled differences in behavioural tests. ²⁵ Consequently, we included an equal number of males and females. However, as no sex‐coupled difference was observed, data from males and females were pooled.

Western Blot Protein Semiquantification in Left Ventricle Tissue

Mice were euthanized by cervical dislocation, and the hearts were dissected into ice‐cold physiological salt solution (in mmol/L: NaCl 115.8, KCl 2.82, KH₂PO₄ 1.18, MgSO₄ 1.2, NaHCO₃ 25, CaCl₂ 1.6, EDTA 0.03, and glucose 5.5, gassed with 5% CO₂ in air and adjusted to pH 7.4), blotted dry, and weighed. In the experiments where maximal capacity of ouabain‐dependent signaling was assessed, the dissected left ventricles were equilibrated for 1 hour in physiological salt solution gassed with 5% CO₂ in air at 37 °C, and then, incubated for 30 minutes with 1 mmol/L ouabain.

Left ventricles were snap frozen in liquid N₂ for later storage at −80 °C. Left ventricles were lysed in lysis buffer (in mmol/L: Tris‐HCl 10, sucrose 250, EDTA 1, EGTA 1, and Triton X‐100 2%, pH 7.4; and 1 tablet protease inhibitor per 10 mL) and centrifuged at 10 000g. Total protein concentrations in the supernatants were measured using BCA Protein Assay Kit (ThermoFisher Scientific, Waltham, MA). A total of 10 µg of total protein diluted in Laemmli sample buffer (Bio‐Rad, Hercules, CA) was loaded onto 4% to 20% precast polyacrylamide stain‐free gels (Criterion TGX Stain‐free precast gel; Bio‐Rad). Total protein load was detected on the stain‐free gels using UV light in imaging system (c600; Azur Biosystems, Dublin, CA).

The proteins were electrotransferred onto membranes that were then blocked by an incubation in 5% BSA and 5% nonfat dry milk in PBS with 0.5% vol/vol Tween 20. To detect PLCγ, Src, and Erk1/2, the membranes were divided before incubation with the antibodies into 2 parts; the upper part, above 75 kDa, was used to detect PLCγ and p‐PLCγ; and the lower part, below 75 kDa, was used to detect either Src and p‐Src or Erk1/2 and p‐Erk1/2, respectively. The membranes were then incubated overnight at 4 °C with the following antibodies:

• α₁ Isoform Na,K‐ATPase antibody (1:2000; No. NBP2‐61137H; Novus Biologicals Inc, Centennial, CO).

• Antibody against the α₂ isoform Na,K‐ATPase (1:2000; No. AB9094; Chemicon, Burlington, MA).

• Antibody against the Na,Ca‐exchanger‐1 (1:200; No. ANX‐011; Alomone Labs, Israel).

• Antibody against total cSrc (1:500; sc‐8056; Santa Cruz Biotechnology Inc, Dallas, TX) or antibody against cSrc phosphorylated at pY418 (1:200; No. 44660G; ThermoFisher Scientific).

• Antibody against total Erk1/2 (1:3000; No. CST4696; Cell Signaling Technology, Inc, Danvers, MA) or antibody against Erk1/2 phosphorylated at Thr202/Tyr204 (1:2000; No. CST9101; Cell Signaling Technology, Inc).

• Antibody against total PLCγ (1:3000; No. CST2822; Cell Signaling Technology, Inc) or antibody against PLCγ phosphorylated at Tyr783 (1:3000; No. CST2821; Cell Signaling Technology, Inc).

After intensive washing, the membranes were incubated with horseradish peroxidase–conjugated secondary antibody (1:4000; Dako, Denmark) for 1 hour in PBS with 0.5% vol/vol Tween 20. Excess antibody was removed by washing, and bound antibody was detected by an enhanced chemiluminescence kit (ECL, Amersham, UK). Detected protein was normalized using the ImageJ program (National Institutes of Health) as a ratio to total protein load measured in the membrane for the same probe, and expressed either as an absolute expression level or as a relative level of phosphorylated form over the total expression level of enzyme.

Mitochondrial Membrane Potential in Left Ventricles

Mice were euthanized by cervical dislocation, and the hearts were dissected into ice‐cold physiological salt solution, as described above. Left ventricles were sliced at the level of the papillary muscle into 160‐µm slices using vibratome (1200 seconds; Leica Biosystems, Germany). The slices were incubated in 5,5´,6,6´‐tetrachloro‐1,1´,3,3´‐tetraethylbenzimidazolocarbocyanine iodide (JC‐1) solution prepared in accordance with the manufacturer manual (JC‐1 mitochondria Staining Kit; catalog No. CS0390; Sigma‐Aldrich, St. Louis, MO) for 15 minutes at 4 °C. To access the specificity of fluorescence from JC‐1 aggregates in the polarized mitochondrial membrane, some left ventricle slices were incubated with JC‐1 solution supplied with 1 µmol/L valinomycin, a mitochondrial K⁺ ionophore that eliminate mitochondrial membrane potential.

JC‐1 has a potential‐sensitive shift in the emitted wavelength attributable to formation of red fluorescent J‐aggregates in the polarized mitochondrial membrane. ²⁹ JC‐1 in its monomeric cytoplasmic form emits green light, whereas JC‐1 in its aggregated form in the mitochondrial membrane emits red light. We placed JC‐1 loaded slices on the confocal microscope (Zeiss LSM 7 Pascal; Zeiss, Germany) and excited at 488 and 543 nm to visualize JC‐1 monomers and aggregates, respectively. Emitted light from JC‐1 monomer and aggregate was collected at 499 to 545 nm and 579 to 651 nm, respectively. Mitochondrial membrane potential was estimated as red (579–651 nm)/green (499–545 nm) fluorescence ratio. ²⁹

Cardiac Magnetic Resonance Imaging

Time‐resolved magnetic resonance imaging (CINE scanning) assessed cardiac contractility and ventricular geometry. Cardiac magnetic resonance imaging was performed using an Agilent 9.4‐T magnetic resonance imaging system (Santa Clara, CA) with a 40‐mm millipede coil. During the experiment, mice were anesthetized by continuous ventilation with 1.8% sevoflurane (AbbVie, North Chicago, IL). To synchronize data acquisition, an integrated subcutaneous 3‐electrode ECG was used in combination with a respiration‐sensing device (Small Animal Instruments, Inc, New York, NY). The body temperature was kept at 37°C with a warm‐air heating system connected to a rectal probe. ECG, respiration rate, and temperature were recorded in PC‐SAM32 software (Small Animal Instruments, Inc).

Image variables for short‐axis scans were as follows: flip angle, 15°; and 31 to 38 phases for 1 cardiac cycle, depending on heart rate. Field of view was 40×40 mm, and matrix was set to 192×192 pixels, resulting in pixel size of 0.21×0.21 mm². Eight slices with a thickness of 1.2 mm were acquired to cover the entire right and left ventricles. Two long‐axis images were obtained, 4‐chamber view and 2‐chamber view, respectively. From the short‐axis images, left ventricular myocardial mass was assessed by manual measurement of endocardial and epicardial borders on each slice. The short‐axis images were also used to quantify end‐diastolic and end‐systolic volume of both ventricles that was used to calculate ejection fraction, stroke volume, cardiac output, and cardiac index. The analysis was done blinded using custom‐made software.

Cardiac Magnetic Resonance With Hyperpolarization

Hyperpolarized [1‐¹³C]pyruvate magnetic resonance scans were performed in the same 9.4‐T preclinical magnetic resonance system equipped with a dual‐tuned 4‐mm ¹³C/¹H volume mouse coil (Rapid Biomedical, Rimpar, Germany). For each experiment, a sample of 127 mg of [1‐¹³C]pyruvic acid mixed with 15 mmol/L AH111501 was polarized in a SPINALIGNER (Polarize IVS, Copenhagen, Denmark) polarizer for >1.5 hours at 6.7 T at 1.3 K, to ensure a reproducible polarization of 50% on average, similar to previous reports in the SPINLAB polarizer. ³⁰ Each mouse received over 5 seconds an intravenous injection of 100 μL per 25 g body weight of 125 mmol/L hyperpolarized [1‐¹³C]‐pyruvate. A slice‐selective ¹³C spectroscopy‐free induction decay sequence with a repetition time of 1 second, flip angle of 10°, and slice thickness of 10 mm covering the entire heart was used. The data were filtered with a zero shifted sine‐bell function and a 15‐Hz exponential line broadening, Fourier transformed and displayed in magnitude mode. The respective peaks were manually integrated using iNMR 6.1.4b (Nucleomatica, Molfetta, Italy). The area under the curve metabolite signal for [1‐¹³C]lactate, [1‐¹³C]alanine, and ¹³C‐bicarbonate in the heart was normalized relative to the area under the curve [1‐¹³C]pyruvate signal as a measure of conversion from pyruvate to metabolite. ³¹

Telemetric Measurements of Blood Pressure and Heart Rate

Blood pressure was measured using radiotelemetry, as described previously. ³² Mice were anesthetized by a combination of ketamine (SC 33 mg/100 g; Ketaminolvet; Intervet International, Boxmeer, the Netherlands) and xylazine (SC 7.5 mg/100 g; Narcoxylvet; Intervet International) and placed on a thermostatically controlled warming platform to maintain body temperature at 37 °C. A midline incision through the shaved skin on the neck was made, and the mandibular glands were separated to access the carotid artery. The catheter of the radiotelemetry transmitter (PA‐C10 and HD‐X11; Data Sciences International, New Brighton, MN) was placed into the carotid artery, and the transmitter body was placed in a subcutaneous pocket. The skin incision was closed using 6‐0 nonabsorbable suture. Analgesia was given subcutaneously at the end of the operation and again 12 hours after and 24 hours after the operation (0.2 mL/kg, buprenorphine hydrochloride, Temgesic; Schering‐Plough Europe, Kenilworth, NJ). Mice were allowed to recover for at least 1 week before measurements were started. Telemetry signals were recorded at 256‐Hz frequency in 10‐second intervals each minute. Registration was performed with Dataquest A.R.T software 4.3 (Data Sciences International). Analyses were performed with Ponemah 8 (Data Sciences International). Arterial pressure was averaged at the time of minimal activity (between 12 am and 2 pm) and at the time of maximal activity (between 8 pm and 10 pm).

Electrophysiological Characterization of Isolated Heart

Electrical activity in the isolated heart was measured, as described previously. ³³ Mice were euthanized by cervical dislocation, and the hearts were rapidly excised. After dissection, the heart was immersed in an oxygenated physiological solution containing (in mmol/L): NaCl 130.0, KCl 5.6, NaH₂PO₄ 0.6, MgCl₂ 1.1, CaCl₂ 2, NaHCO₃ 20.0, and glucose 11.0, bubbled with carbogen (95% O₂ and 5% CO₂) with pH 7.4. The isolated hearts were pinned in the experimental chamber (3 mL) and superfused with physiological solution (10 mL/min; 37.5 °C). The hearts were cannulated through aorta and retrogradely perfused with constantly flowing solution of the same composition. The right auricle and right ventricular wall were opened to make endocardial surface within reach for microelectrode impalements.

Preparations were beating spontaneously in a stable rhythm generated by the sinoatrial node, throughout the experiment. After 30 minutes of equilibration in the perfusion chamber, transmembrane resting and action potentials were recorded from endocardial cardiomyocytes with sharp glass microelectrodes (30–45 MΩ) filled with mol/L KCl connected to a high input impedance amplifier model 1300 (A‐M Systems, Sequim, WA). The signal was digitized and analyzed using specific software (L‐card and DI‐Soft, Moscow, Russia; and Synaptosoft, Decatur, GA). The measurements were done in 4 distinct regions: right atrial posterior part of intercaval region that contains sinoatrial node pacemaker area, ³⁴ right atrial trabeculae, and right atrial and left ventricular wall.

Citrate Synthase Enzymatic Activity in Left Ventricle

Citrate synthase enzymatic activity in cardiac tissue homogenates was measured by spectrophotometry, as described previously, ³⁵ and the results are expressed as μmol/min per g tissue. The cardiac tissue was homogenized in 1.5 mL of 0.3 mmol/L K₂HPO₄ with 0.05% BSA (pH 7.7). A total of 15 µL of 10% Triton X‐100 was added, and samples were left on ice for 15 minutes. The homogenate was diluted 50 times in a solution containing (in mmol/L): 0.4 acetyl‐CoA, 0.6 oxaloacetate, 0.157 5,5′‐dithiobis‐(2‐nitrobenzoic acid), and 39 Tris·HCl (pH 8.0). The change of 5,5′‐dithiobis‐(2‐nitrobenzoic acid) to 5‐thiobis‐(2‐nitrobenzoic acid) at 37 °C was measured spectrophotometrically at 415 nm on an automatic analyser (Cobas 6000, C 501; Roche Diagnostics, Mannheim, Germany).

Respirometry of Left Ventricle

The 3‐ and 8‐month‐old mice were euthanized by cervical dislocation, and the left ventricle was quickly dissected and transferred to a biopsy preservation solution containing (in mmol/L): 10 Ca‐EGTA buffer, 10^‐4 free Ca²⁺, 20 imidazole, 20 taurine, 50 K‐MES, 0.5 dithiothreitol, 6.56 MgCl₂, 5.77 ATP, and 15 phosphocreatine, pH 7.1. The left ventricles were chemically permeabilized by incubation for 30 minutes in ice‐cold biopsy preservation solution buffer containing 50 μg/mL saponin. ³⁶ After washing twice for 10 minutes in an ice‐cold respiration medium (MiR05; in mmol/L: 110 sucrose, 60 K‐lactobionate, 0.5 EGTA, 0.1% BSA, 3 MgCl₂, 20 taurine, 10 KH₂PO₄, and 20 HEPES; pH 7.1), ventricles were blotted dry and weighed.

Mitochondrial respiratory capacity was measured with high‐resolution respirometry (Oxygraph‐2k; Oroboros Instruments, Innsbruck, Austria) with nonfatty substrates. ³⁶ All experiments were conducted at 37 °C in a hyperoxygenated environment (250–450 µmol/L) to preclude potential O₂ diffusion limitations. In the absence of adenylates, the basal respiration (ie, state 2 respiration supported by electron flow from complex I (glutamate and malate), was measured in the presence of 10 mmol/L glutamate and 2 mmol/L malate. The subsequent addition of 5 mmol/L ADP enables a coupled state 3 respiration supported by electron transfer from complex I (state 3 respiration with glutamate and malate). The whole data set was excluded, if the subsequent addition of 10 µmol/L cytochrome C (the test for integrity of the outer mitochondrial membrane) increased respiration by >10%. ³⁷ The maximal coupled state 3 respiration in complex I and II was assessed with 10 mmol/L succinate. Oligomycin (2 μg/mL; eliminates ATP synthesis) was used to estimate proton leak across the inner mitochondrial membrane (ie, state 4o of respiration). The residual O₂ consumption was measured in the presence of 0.5 µmol/L rotenone and 2.5 mmol/L antimycin A and serves as an indicator of uncoupled, nonmitochondrial respiration. ³⁷

The steady‐state respiratory rates were evaluated as an average oxygen consumption (O₂ in pmol/s) per milligram ventricle weight over the stable period of respiratory state using DatLab 6 software (Oroboros Instruments). The residual nonmitochondrial O₂ consumption was subtracted from all other values.

Lipid Peroxidation in Left Ventricle

Lipid peroxidation is commonly quantified by measuring the accumulating by‐products (eg, malondialdehyde), as described previously. ³⁸ Malondialdehyde is a common general product of nonenzymatic peroxidation of polyunsaturated fatty and arachidonic acids. The level of malondialdehyde was measured by thiobarbituric acid–reactive substances assay. Malondialdehyde in the sample was reacted with thiobarbituric acid to generate the malondialdehyde–thiobarbituric acid adduct. The malondialdehyde–thiobarbituric acid adduct was quantified colorimetrically at 532 nm. ³⁸

Relative Protein Quantification in Left Ventricle Tissue Using 10‐Plex Tandem Mass Tags

Left ventricles were dissected and lysed, as described above, for Western blot protocol. Proteins were isolated by acetone precipitation, redissolved in 0.2 mol/L triethylammonium bicarbonate, followed by reduction by dithiothreitol (5 mmol/L for 30 minutes at 50 °C), alkylation by iodoacetamide (15 mmol/L for 30 minutes at room temperature in the darkness), and proteolytic cleavage by the addition of trypsin in a 1:50 trypsin/protein ratio and overnight incubation at 37 °C. The resulting peptide samples were labeled with 2 sets of tandem mass tags from a 10‐plex tandem mass tags set using the mass tags 127N, 127C, 128N, 128C, 129N, and 129C. A pool of all samples was labeled with mass tag 131 and served as reference channel. Tagged peptide samples were mixed in 2 sets of tagged peptide mixture, which each were subsequently fractionated by hydrophilic interaction chromatography and analyzed by reversed phase nanoliquid chromatography tandem mass spectrometry, as previously described. ³⁹ In brief, the peptides were dissolved in hydrophilic interaction chromatography buffer B (90% [v/v] acetonitrile and 0.1% [v/v] trifluoroacetic acid) fractionated into 24 fractions using a Dionex UltiMate 3000 nano high‐performance liquid chromatograph using a 42‐minute linear gradient and a TSK gelamide‐80 hydrophilic interaction chromatography column. Each fraction was then analyzed by reversed phase nanoliquid chromatography tandem mass spectrometry using a Dionex UltiMate 3000 nano high‐performance liquid chromatograph coupled to a Q‐exactive Orbitrap mass spectrometer, as described. ³⁹ The resulting Q‐exactive raw data files were then processed and searched against the Uniprot mouse database using Proteome Discoverer version 2.1.0.81 (Thermo Scientific, Waltham, MA) integrated with the Sequest search engine and Mascot search engine (version 2.2.3) virtually, as previously described. ³⁹ The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE ⁴⁰ partner repository with the data set identifier PXD028952.

Ingenuity Pathway Analysis

Protein expression in the proteomics data from wild‐type and α₂ ^+/G301R hearts was compared with unpaired t‐test and uploaded into ingenuity pathway analysis (IPA) software (Qiagen, Redwood City, CA), where the differential protein expression was defined as P<0.05 (Data S1). First, gene ontology pathways were analyzed for enrichment of differentially expressed proteins. Differentially expressed proteins, which were considered relevant for the cardiovascular system by IPA, were included for further analysis (Data S2). The association of these proteins with cardiovascular disease and cardiac function was suggested by IPA (Data S3). A negative z‐score indicates suppression of that pathway, whereas a positive z‐score indicates enhancement.

Ultrastructural Analysis of Mitochondria

For ultrastructural analyses, three 8‐month‐old α₂ ^+/G301R and 3 WT mice were perfusion fixed with 2% glutaraldehyde in 0.1 mol/L cacodylate buffer, pH 7.4, through the left ventricle. Tissue pieces were subsequently postfixed for 1 hour in 1% OsO₄ in 0.1 mol/L cacodylate buffer, stained for 1 hour with 0.5% uranyl acetate in 0.05 mol/L maleate buffer, pH 6.0, dehydrated in graded alcohols, and embedded in Epon (TAAB resin 812; VWR ‐ Bie & Berntsen A/S, Soeborg, Denmark). Ultrathin sections of ≈70 nm were obtained with a Leica EM FC6 Cryoultramicrotome, collected on 100 mesh nickel grids and stained with uranyl acetate and lead citrate. Sections were examined using a JEOL JEM‐1400+ (JEOL, Freising, Germany) transmission electron microscope.

Histological Analysis of Left Ventricle Tissue

Eight‐month‐old mice were euthanized, and the left ventricles were dissected in PBS for histological analysis. Left ventricles were fixed in 4% formaldehyde, washed in PBS, embedded in paraffin, and sectioned in 5‐µm slices at the level of the papillary muscle. Sections were deparaffinized and processed following standard Masson trichrome staining protocol: 5 minutes in Weigert hematoxylin (1% hematoxylin in 99% ethanol), 5 minutes in picric acid (10 mL saturated picric acid solution in 40 mL 96% ethanol), 10 minutes in Biebrich scarlet‐acid fuchsin solution (300 mg Biebrich scarlet and 100 mg acid fuchsin in 40 mL distilled water and 160 mL 0.2% acetic acid), and 2 minutes in methyl blue solution (1.25 g methyl blue in 100 mL distilled water and 1 mL ethanol). The last 2 steps were followed by 10 minutes in 1% phosphomolybdic acid. Images were taken using Olympus VS120 slide scanner (Olympus, Tokyo, Japan) at ×20 magnification and analyzed using ImageJ software. Measurements were obtained from 31 to 42 cardiomyocytes from each mouse in long‐axis and cross‐sectional orientation. Measurements of cross‐sectional area and longitudinal diameter from each animal were averaged for analysis.

Structure Modeling of the Na,K‐ATPase α₂ Isoform

To model the Na,K‐ATPase from humans, we used the protein data bank entry 2zxe ⁴¹ as a template. The alignments were used to model human Na,K‐ATPase α₂ isoform in modeler. ⁴² For each template, a total of 100 models were generated. The model identified by the lowest discrete optimized protein energy score ⁴³ was used for further analysis.

Statistical Analysis

All data are summarized as the mean value±SEM of the sample group. Student t test, 1‐way ANOVA, or 2‐way ANOVA, followed by Bonferroni or Sidak correction for multiple comparisons, was used when appropriate to determine significant differences between means. Differential protein expression in the proteomics data set was analyzed with unpaired t test before upload to IPA software for further analysis. A probability (P) level of <0.05 was considered significant. In the IPA omics data analysis, P values for diseases or function annotations were calculated on the basis of the 1‐sided Fisher exact test. Statistical analyses of the functional data were performed with Microsoft Excel or GraphPad Prism software (version 8).

Results

Reduced Expression of the Na,K‐ATPase α₂ Isoform in Hearts of α₂ ^+/G301R Mice Was Associated With Increased Expression of the α₁ Isoform But Unchanged Na,Ca‐Exchanger Expression

Western blot semiquantification demonstrated an ≈25% reduction of the Na,K‐ATPase α₂ isoform in the hearts from 3‐ and 8‐month‐old α₂ ^+/G301R mice (Figure 1A and 1D). In both age groups of α₂ ^+/G301R mice, this was accompanied by ≈25% increased expression of the Na,K‐ATPase α₁ isoform (Figure 1B and 1E). No difference in the expression of Na,Ca‐exchanger 1 was detected between α₂ ^+/G301R and WT mice in both age groups (Figure 1C and 1F).

Figure 1. Changed expression of the Na,K‐ATPase α isoforms in the heart from α₂ ^+/G301R mice but similar expression of the Na,Ca‐exchanger‐1.

The hearts from 3‐ and 8‐month‐old α₂ ^+/G301R mice showed reduced expression of the Na,K‐ATPase α₂ isoform (A, n=7; D, n=6), increased expression of Na,K‐ATPase α₁ isoform (B, n=7; E, n=5 to 6), and no change in the expression of Na,Ca‐exchager‐1 (NCX1; C, n=7; F, n=12) in comparison with wild type (WT). The expression profile was the same in hearts of 3‐month‐old (A–C) and 8‐month‐old (D–F) mice. Upper part of each panel shows representative Western blot bands for the averaged data shown below. The representative bands are shown from the same membrane, as well as the molecular marker shown to the right. The images cropped to include molecular weights from 250 to 50 kDa, as indicated. Total protein load was detected with the stain‐free gel, and the corresponding representative bands were cropped to the same molecular weight range. Protein expression was compared using unpaired t‐test. *P<0.05 and **P<0.01.

Dilation of Ventricles and Impaired Ventricular Function in 8‐Month‐Old α₂ ^+/G301R Mice

There was no statistical difference in heart rate or respiration frequency between 3‐ and 8‐month‐old anaesthetized α₂ ^+/G301R and WT mice (Table S1). Cardiac magnetic resonance imaging (Figure 2A through 2H) showed similar left and right ventricular end‐diastolic and end‐systolic volumes in 3‐month‐old WT and α₂ ^+/G301R mice (Figure 2I). Left ventricular ejection fraction (EF; Figure 2J), stroke volume, cardiac output, and cardiac index (Table S1) were similar for 3‐month‐old mice of both genotypes.

Figure 2. Ventricular dilation and decreased ejection fraction in 8‐month‐old α₂ ^+/G301R mice.

Representative cardiac magnetic resonance end‐diastolic and end‐systolic images of the heart from an 8‐month‐old wild‐type (WT) (A and B) and α₂ ^+/G301R mouse (C and D) in the short axis. Representative cardiac magnetic resonance end‐diastolic and end‐systolic images of the same hearts in the long axis (E–H). The dotted red lines in (A–D) and the dotted green lines (E–H) show the orientation of the long axis and short axis, respectively. Bars (A–H) correspond to 5 mm. Left and right ventricular dimensions and ejection fraction were assessed from short‐axis images. No difference in ventricular dimensions (I) and ejection fraction (J) was found between 3‐month‐old α₂ ^+/G301R (n=6) and WT (n=6) mice. In contrast, both left and right ventricles in 8‐month‐old α₂ ^+/G301R mice (n=10) were dilated in diastole in comparison with WT mice (K; n=7). Moreover, 8‐month‐old α₂ ^+/G301R mice also showed increased end‐systolic left ventricular volume (K) and reduced ejection fraction (L). The age‐related changes in ventricular volume were only observed for α₂ ^+/G301R mice; the end‐systolic left and right ventricular volume and end‐diastolic right ventricular volume were increased from 3 to 8 months of age. *P<0.05 for comparison of WT and α₂ ^+/G301R mice of the same age group; ^† P<0.05 and ^†† P<0.01 for comparison of mice of the same genotype at 3 and 8 months of age (2‐way ANOVA with Sidak multiple comparisons test).

End‐diastolic volumes of left and right ventricles as well as end‐systolic volume of the left ventricle were increased in 8‐month‐old α₂ ^+/G301R mice (Figure 2K). Changes in ventricular volume from 3 to 8 months of age were only observed for α₂ ^+/G301R mice. The disarray was associated with decreased left ventricular EF in α₂ ^+/G301R mice (Figure 2L), whereas the difference in right ventricular EF did not achieve statistical significance (Table S1). Stroke volume, cardiac output, and cardiac index were similar between genotypes in 8‐month‐old mice (Table S1). No difference in left ventricular mass was seen between α₂ ^+/G301R and WT mice in both age groups (Table S1).

The 8‐Month‐Old α₂ ^+/G301R Mice Showed Reduced Nocturnal Blood Pressure

Both α₂ ^+/G301R and WT mice showed circadian rhythm variations in arterial blood pressure and heart rate that are consistent with murine nocturnal behavior of mice (Figure S2). When compared at the same time of the day, there was no difference in blood pressure and heart rate between 3‐month‐old α₂ ^+/G301R and WT mice (Figure 3A through 3C). However, 8‐month‐old α₂ ^+/G301R mice showed a reduction in systolic blood pressure compared with age‐matched WT mice (Figure 3D). This reduction was significant at nighttime, when mice are most active. The reduction in nighttime diastolic blood pressure did not achieve significance (P=0.077; Figure 3E). No difference in heart rate was observed between 8‐month‐old mice of the 2 genotypes (Figure 3F).

Figure 3. The 8‐month‐old α₂ ^+/G301R mice showed lower nocturnal blood pressure than wild‐type (WT) mice.

Averaged systolic (A and D) and diastolic (B and E) blood pressure and heart rate (C and F) of 3‐month‐old (A–C; n=6–9) and 8‐month‐old (D–F; n=7) α₂ ^+/G301R and WT mice. Blood pressure and heart rate were averaged over the period where night activity of mice was most prominent (from 8 pm to 10 pm) and the day activity was minimal (from 12 am to 2 pm). For circadian variations, see also Figure S2. BPM indicates beats per minute. **P<0.01 for comparison between WT and α₂ ^+/G301R mice of the same age; ^† P<0.05, ^†† P<0.01, and ^††† P<0.001 for comparison of mice of the same genotype at day and night (2‐way ANOVA with Sidak multiple comparisons test).

Cardiomyocytes From 8‐Month‐Old α₂ ^+/G301R Mice Exhibited Normal Electrophysiological Activity

To assess whether electrophysiological abnormalities may explain reduced ventricular function in 8‐month‐old α₂ ^+/G301R mice consequent to modified Na⁺ homeostasis, we recorded electrical activity in right atrial trabeculae, posterior part of intercaval region, right atrial wall, and right ventricular wall of isolated hearts. No differences in spontaneous heart rate, resting membrane potential, action potential amplitude, and duration were found between 8‐month‐old α₂ ^+/G301R and WT mice (Table S2 and Figure S3). This suggests that ion homeostasis and ion transport are not significantly affected in α₂ ^+/G301R cardiomyocytes.

Proton Leak Across the Inner Mitochondrial Membrane and Increased O₂ Consumption in Cardiomyocytes From 8‐Month‐Old α₂ ^+/G301R Mice

The Na,K‐ATPase is a major energy consumer, and its activity affects the cell energetics important for myocardial performance. ¹³ To test the impact of the G301R mutation, we assessed the O₂ consumption of the hearts from 3‐ and 8‐month‐old α₂ ^+/G301R and WT mice. The residual nonmitochondrial O₂ consumption was increased in hearts from 8‐month‐old α₂ ^+/G301R mice compared with WT mice (Figure 4A). In 3‐month‐old mice, nonmitochondrial O₂ consumption was similar between the genotypes (Figure 4A).

Figure 4. The hearts from 8‐month‐old α₂ ^+/G301R mice showed mitochondrial uncoupling.

Left ventricles from 3‐month‐old (n=5–6) and 8‐month‐old (n=10–11) α₂ ^+/G301R and wild‐type (WT) mice were compared for the following parameters: uncoupled, nonmitochondrial respiration (A), basal respiration (GM) (B), state 3 respiration in complex I (GM3) (C), state 3 respiration in complex I and II (GMS3) (D), state 4 respiration (4o) (E), and the outer mitochondrial membrane integrity (F), where open labels show O₂ consumption before and filled labels after cytochrome C incubation. Genotypes in each age group were compared using unpaired t test. *P<0.05 and ***P<0.001. See also representative traces in Figure S5. GM indicates glutamate and malate; and GM3, state 3 respiration with glutamate and malate.

There was no difference at state 2 respiration specific to complex I (glutamate and malate) between the groups (Figure 4B and Figure S5). The coupled state 3 respiration with glutamate and malate showed higher O₂ consumption in hearts from 8‐month‐old α₂ ^+/G301R than in WT mice (Figure 4C). Similarly, maximal coupled state 3 respiration in complex I and II was increased in hearts from 8‐month‐old α₂ ^+/G301R mice (Figure 4D). The state 4o respiration was also increased in the hearts from 8‐month‐old α₂ ^+/G301R mice in comparison with WT mice (Figure 4E), suggesting an increased proton leak across the inner mitochondrial membrane. ⁴⁴ In contrast, state 3 respiration with glutamate and malate, maximal coupled state 3 respiration in complex I and II, and 4o respiration were similar between genotypes in 3‐month‐old mice (Figure 4C through 4E).

All experimental groups had similar content of intact mitochondria. The citrate synthase level that serves as marker for mitochondrial content ⁴⁵ was similar in both genotypes and age groups (Figure S4). Cytochrome C incubation did not change respiration by >10% in all individual experiments (Figure 4F), except one excluded sample, suggesting intact integrity of the outer mitochondrial membrane.

Hearts From 8‐Month‐Old α₂ ^+/G301R Mice Showed Increased Oxidative Stress and Increased Lactate Production

We tested oxidative damage in tissue from the left ventricle by measuring malondialdehyde. No difference was detected in the hearts from 3‐month‐old WT and α₂ ^+/G301R mice, but hearts from 8‐month‐old α₂ ^+/G301R mice showed elevated malondialdehyde concentrations compared with age‐matched WT and 3‐month‐old α₂ ^+/G301R mice (Figure 5A).

Figure 5. Increased lipid peroxidation and lactate production in the hearts from 8‐month‐old α₂ ^+/G301R mice.

A, Lipid peroxidation was assessed as a concentration of malondialdehyde in the left ventricle. No difference in malondialdehyde concentration was seen between genotypes of 3‐month‐old mice (n=8–11), but hearts from 8‐month‐old α₂ ^+/G301R mice (n=21) showed elevated malondialdehyde concentrations in comparison with both hearts from age‐matched wild‐type (WT) mice (n=26) and 3‐month‐old α₂ ^+/G301R mice. B, Magnetic resonance hyperpolarization showed an increased pyruvate conversion ratio to lactate in the hearts from 8‐month‐old α₂ ^+/G301R mice (n=5) in comparison with age‐matched WT mice (n=4). No difference in conversion ratio to alanine and bicarbonate was seen. Malondialdehyde concentrations in the groups and at different ages were compared using 2‐way ANOVA with Sidak multiple comparisons test. The level of pyruvate metabolite products was compared between the groups using unpaired t‐test. *P<0.05 and **P<0.01.

Heart metabolism was assessed in vivo. Magnetic resonance hyperpolarization showed an increased lactate production in hearts from 8‐month‐old α_t ^+/G301R mice in comparison with age‐matched WT mice (Figure 5B). The levels of alanine and HCO₃ ^‐ were similar between the 2 genotypes (Figure 5B). This suggested a shift toward a heart failure–associated metabolism (ie, increased lactate/HCO₃ ^‐ ratio).

Increased Capacity of the Src/Ras/Erk1/2 Pathway Downstream From the Na,K‐ATPase in Hearts From 8‐Month‐Old α₂ ^+/G301R Mice

Total expression and phosphorylation levels of protein kinase signaling pathways downstream from the Na,K‐ATPase (ie, Src, Erk1/2 kinases, and PLCγ) were assessed in the hearts of 3‐ and 8‐month‐old mice. No difference in signaling molecule expression or phosphorylation was detected between 3‐month‐old α₂ ^+/G301R and WT mice (Figure S6).

The hearts from 8‐month‐old α₂ ^+/G301R mice had increased expression of total Src kinase but similar level of phosphorylated Src at rest (Figure 6A). Incubation with 1 mmol/L ouabain led to an increased level of phosphorylated Src in the hearts from 8‐month‐old α₂ ^+/G301R mice compared with WT mice (Figure 6D), suggesting larger capacity for Src phosphorylation in hearts from 8‐month‐old α₂ ^+/G301R mice (Figure 6F).

Figure 6. The hearts from 8‐month‐old α₂ ^+/G301R mice showed modified signaling of pathways downstream from the Na,K‐ATPase.

A semiquantification of expressional changes of total Src kinase (n=36) and the level of Src phosphorylation at Tyr418 (A; n=30); total Erk1/2 kinase (n=42) and its Thr202/Tyr204 phosphorylation level (B; n=36); and total PLCγ (n=42) and the level of its Tyr783 phosphorylation (C; n=18). When the hearts were incubated for 30 minutes with 1 mmol/L ouabain, the level of phosphorylated Src was significantly increased (A; n=6). The phosphorylation level of Erk1/2 after 1 mmol/L ouabain incubation tended to increase in α₂ ^+/G301R hearts in comparison with wild‐type (WT) mice, but this did not achieve significance (B; n=6; P=0.17). Representative images for Src (D), Erk1/2 (E), and PLCγ (F) semiquantification Western blot experiments that are averaged (A–C), respectively. Left images correspond to total protein of interest (Src, Erk1/2, and PLCγ, respectively). Images in the center correspond to phosphorylated protein of interest (p‐Src, p‐Erk1/2, and p‐PLCγ, respectively), as indicated. Molecular weight markers and total protein load in the membrane detected with stain‐free gels are shown to the right. All representative images were cropped to the size identified by molecular marker (D–F), respectively. All images are cropped to include at 3 molecular weight markers positioned both above and below the band. Before incubation with the antibodies, the membrane was divided into 2 parts: the upper part, above 75 kDa, was used to detect PLCγ and p‐PLCγ; and the lower part, below 75 kDa, was used to detect either Src and p‐Src or Erk1/2 and p‐Erk1/2, respectively. The expression was normalized to total protein load for the corresponding probe. An average level for the WT group was taken as 100%. Protein expression was compared between genotypes using unpaired t‐test. *P<0.05 and **P<0.01.

Total Erk1/2 expression was increased in hearts from α₂ ^+/G301R mice, but the level of phosphorylated Erk1/2 was decreased at rest (Figure 6B). The difference in the absolute level of phosphorylated Erk1/2 between α₂ ^+/G301R and WT mice did not achieve significance after ouabain incubation (Figure 6E). No difference in total PLCγ or its relative phosphorylation was seen between hearts from 8‐month‐old α₂ ^+/G301R and WT mice (Figure 6C). The hearts from α₂ ^+/G301R and WT mice incubated with ouabain showed similar phosphorylation levels of PLCγ (data not shown; n=6).

Similar Mitochondrial Membrane Potential in the Hearts From 8‐Month‐Old Mice of Both Genotypes

We assessed in situ mitochondrial health in the left ventricles from 8‐month‐old WT and α₂ ^+/G301R mice (Figure 7A). No difference in mitochondrial membrane potential was found between the groups (Figure 7B), suggesting similar mitochondrial polarization. Moreover, as early stages of apoptosis are characterized by mitochondrial disruption, these results suggest that apoptosis state is similar in α₂ ^+/G301R and WT hearts. Disruption of mitochondrial membrane potential with the K⁺ ionophore, valinomycin, equally reduced the fluorescence emission from JC‐1 aggregates in mitochondrial membrane in WT and α₂ ^+/G301R hearts and was without significant effect on cytosolic JC‐1 monomer emission, as expected (Figure 7A). This suggests similar number of mitochondria in both groups.

Figure 7. No difference in mitochondrial health between the hearts from 8‐month‐old wild‐type (WT) and α₂ ^+/G301R mice.

Fresh cross‐sectional myocardial slices were loaded with the fluorescent dye, 5,5´,6,6´‐tetrachloro‐1,1´,3,3´‐tetraethylbenzimidazolocarbocyanine iodide (JC‐1). When JC‐1 is in its monomeric cytoplasmic form, it emits green light; and in its aggregated form, which accumulates in mitochondria (M) in a potential‐dependent manner, it emits red light (A). The fluorescence ratio (red/green) reflects mitochondrial membrane potentials in the hearts from WT (n=6) and α₂ ^+/G301R mice (B; n=7). To test the signal specificity, some slices were pretreated with valinomycin, a potassium ionophore and an uncoupler of mitochondrial respiration that causes collapse of the mitochondrial membrane potential (WT, n=5; and α₂ ^+/G301R, n=6). Bars in A correspond to 50 μm. Fluorescence ratio in the 2 genotypes as well as the fluorescence ratio before and after valinomycin in each group were compared using 2‐way ANOVA with Sidak multiple comparisons test. **P<0.01 and ***P<0.001. Morphological analysis of mitochondria (WT, n=3; and α₂ ^+/G301R, n=3) were performed by transmission electron microscopy (C). Scale bars on high‐magnification micrographs: 1 μm; and on low‐magnification micrographs: 2 μm. Ns indicates no significant difference.

Similar Mitochondrial Ultrastructure and Cardiomyocyte Morphology in Both Genotypes

Mitochondrial ultrastructure was assessed in cross‐sections of left ventricle tissue from 8‐month‐old mice using transmission electron microscopy. Similar mitochondrial ultrastructure was observed in both genotypes (Figure 7C). Cardiac tissue morphology was assessed in left ventricle cross‐sections stained with Masson trichrome. There was no difference between genotypes in cross‐sectional area of cardiomyocytes from 8‐month‐old mice (Figure S7A and S7B). Also, cardiomyocyte diameter measured in the long axis was similar in both genotypes (Figure S7C and S7D). No infarctions were detected in any of the mice of both genotypes.

Proteomics Data Analysis Suggested the Molecular Basis for the Cardiac Phenotype in 8‐Month‐Old α₂ ^+/G301R Mice

Our proteomic data provided assessment of relative protein expression changes. Data are available via ProteomeXchange with identifier PXD028952. We identified 2888 mapped proteins, thereof 181 significantly upregulated and 22 downregulated analysis‐ready proteins in the hearts from α₂ ^+/G301R mice in comparison with WT mice (Data S1). A total of 145 of the proteins were associated with the cardiovascular system (Data S2). We identified several cardiovascular conditions that were correlated with the expressional pattern in the hearts from α₂ ^+/G301R mice. These were primarily conditions associated with systolic dysfunction (Data S3).

The reduced ejection fraction was associated with downregulation of several components of the troponin‐tropomyosin complex and cardiac α‐actin (Figure S8A). Together with the finding of upregulation of myosin light chain 7 and α‐actinin 2, these data suggested a reduced Ca²⁺ sensitivity of the contractile machinery in α₂ ^+/G301R cardiomyocytes (Figure S8A). No difference in the expression of Ca²⁺ transporting proteins was found between the hearts from α₂ ^+/G301R and WT mice (Figure S8B).

The proteomics data analysis confirmed that the expression of the Na,K‐ATPase α₁ isoform was increased, whereas the α₂ isoform was decreased, in the hearts from α₂ ^+/G301R mice (Figure S9A). On the basis of upregulation of protein phosphatases, mitogen‐activated protein kinase kinase 3, and Ras family proteins, the analysis suggested increased Src/Ras/Erk1/2 signaling in hearts from 8‐month‐old α₂ ^+/G301R mice (Figure S9B and S9C). Increased oxidative stress in the hearts from α₂ ^+/G301R mice was predicted on the basis of upregulation of enzymes involved in generation of ROS and regulation of the cellular oxidation‐reduction state (Figure S10).

Discussion

This study addressed the importance of migraine‐associated mutation of the Na,K‐ATPase α₂ isoform for cardiovascular function. In a mouse model for FHM2, we found that hearts from 8‐month‐old α₂ ^+/G301R mice exhibited dilation of left and right ventricles and reduced left ventricular EF but no left ventricular hypertrophy. The cardiac phenotype in 8‐month‐old α₂ ^+/G301R mice was associated with proton leak across the inner mitochondrial membrane, increased oxidative stress, and a heart failure–associated metabolic shift toward glycolysis accompanied by increased lactic acid production. Our data indicated that imbalanced Na,K‐ATPase–dependent Src/Ras/Erk1/2 signaling underlies mitochondrial dysfunction, leading to increased generation of ROS in 8‐month‐old α₂ ^+/G301R mice. Our finding of a link between FHM2‐associated mutation and cardiac dysfunction provides novel insight into the association between migraine and cardiovascular disease.

Oxidative Stress and Suppressed Cardiac Function in α₂ ^+/G301R Mice Were Associated With Altered Cardiac Metabolism

The hemodynamically or metabolically stressed heart often returns to a pattern of fetal metabolism, where it depends primarily on glucose and lactate for energy. ⁴⁶ This cardiac metabolic shift toward glycolysis is known as fetal programming and is a well‐known feature in the failing heart and in cardiomyopathy. ⁴⁷ We suggest that fetal programming in hearts from 8‐month‐old α₂ ^+/G301R mice is a consequence of chronically increased oxidative stress over many months. In support of this suggestion, our data demonstrated a similar production of HCO₃ ^‐ in α₂ ^+/G301R and WT mice. This may reflect that during the experiment, O₂ availability was sufficient to maintain pyruvate dehydrogenase flux. Maintained pyruvate dehydrogenase flux, increased glycolysis, and increased production of lactic acid in α₂ ^+/G301R mice suggest a metabolic shift toward fetal programming.

Cardiac fetal programming is closely associated with increased generation of ROS. ⁴⁷ ROS are produced continually in mitochondria as a by‐product of normal respiration through the one‐electron reduction of molecular oxygen. ⁴⁸ We found a markedly increased O₂ consumption in the hearts from 8‐month‐old α₂ ^+/G301R mice. Our results indicated that this is attributable to increased proton leak across the inner mitochondrial membrane and increased O₂ consumption by nonmitochondrial oxidation processes. Chronic heart failure was previously reported to reduce mitochondrial membrane potential, ⁴⁹ and this was suggested to contribute to cardiac remodeling. ⁵⁰ We demonstrated that mitochondria in hearts from 8‐month‐old α₂ ^+/G301R mice were able to maintain a hyperpolarized mitochondrial membrane potential similar to WT mice. Also, mitochondrial ultrastructure was similar between the 2 genotypes. Both citrate synthase assay and JC‐1 data suggested similar number of mitochondria. The dissociation between mitochondrial membrane potential generation and its use of oxygen for mitochondria‐dependent ATP synthesis indicates mitochondrial uncoupling in the hearts of 8‐month‐old α₂ ^+/G301R mice. Oxidative stress facilitates opening of permeability transition pores in the mitochondrial membrane, which on opening induce proton leak across the inner mitochondrial membrane, ⁵¹ , ⁵² as observed in 8‐month‐old α₂ ^+/G301R mice. This proton leak further increases oxidative stress. Thus, a positive feedback loop is established. In agreement with this finding, it has previously been demonstrated that the Na,K‐ATPase plays a central role in this oxidative stress–dependent positive feedback loop. ⁵³ , ⁵⁴ Moreover, the Na,K‐ATPase is suggested to contribute to oxidation‐reduction–dependent β‐adrenergic signaling ²⁰ (ie, we expect oxidative stress to increase during increased nocturnal activity, and we expect this increase to be larger in α₂ ^+/G301R mice). To summarize, we suggest that increased level of oxidative stress in the hearts from α₂ ^+/G301R mice originates from the metabolic shift as well as mitochondrial uncoupling (Figure 8).

Figure 8. Summary of metabolic changes in the heart with mutation of the Na,K‐ATPase α₂ isoform.

Migraine‐associated mutation led to reduced expression of the Na,K‐ATPase α₂ isoform that, in turn, increased the expression of the α₁ isoform. This was associated with unbalanced signaling of the Src/Ras/Erk1/2 pathway, leading to uncoupling of the electron transport chain. Mitochondrial uncoupling in combination with heart failure–associated metabolic shift toward fetal programming accompanied by increased level of lactic acid and H⁺ leak through the inner mitochondrial membrane led to oxidative stress. Oxidative stress further increases H⁺ leak across the inner mitochondrial membrane and thus establishes a positive feedback loop. This was associated with ventricular dilation and reduced cardiac performance. Pathological conditions are identified in red font, red arrows, and red text boxes. LDH indicates lactate dehydrogenase; PDH, pyruvate dehydrogenase; and TCA, tricarboxylic acid.

In heart failure, ROS are suggested to modify myofibrillar proteins, including cardiac α‐actin and tropomyosin, by oxidation, resulting in changed structure and function of the contractile machinery. ⁵⁵ , ⁵⁶ This is associated with release of troponin from the myocardium to the bloodstream, both in acute and chronic heart failure. ⁵⁷ In the present study, 8‐month‐old α₂ ^+/G301R mice that exhibited reduced EF and ventricular dilation showed reduced levels of cardiac α‐actin and components of the troponin‐tropomyosin complex in the cardiac tissue. We suggest that these changes leading to reduced Ca²⁺ sensitivity in the contractile machinery form the basis of systolic dysfunction in 8‐month‐old α₂ ^+/G301R mice and that they are caused by oxidative damage. Interestingly, this is not the first time an association between systolic dysfunction and changed expression of the Na,K‐ATPase has been observed. In humans, the concentration of the Na,K‐ATPase was shown to be reduced in the early stages of heart failure ⁷ and in dilated cardiomyopathy. ⁸ Although the causal relation was not addressed, the changed expression of Na,K‐ATPase in those studies is likely secondary to cardiac pathology. In contrast, we herein demonstrated that inherited changes in the expression of Na,K‐ATPase can cause cardiovascular disease.

Blood Pressure Changes in 8‐Month‐Old α₂ ^+/G301R Mice May Be a Result of Reduced Cardiac Function

We have previously reported unchanged ex vivo vascular tone of small peripheral arteries from α₂ ^+/G301R mice, suggesting unchanged total peripheral resistance. ²⁷ In the present study, 8‐month‐old α₂ ^+/G301R mice showed reduced nocturnal blood pressure. There may be several reasons for this inconsistency. Although cardiac output was similar in anaesthetized 8‐month‐old α₂ ^+/G301R and WT mice, reduced ability to increase cardiac output on physical activity in α₂ ^+/G301R may be responsible for reduced blood pressure. Consistent with this possibility, the difference in blood pressure between genotypes was observed only at night when mice are active. However, other explanations, such as reduced sympathetic tone in 8‐month‐old α₂ ^+/G301R mice, cannot be excluded and require further investigation. Nevertheless, the observed reduction in arterial blood pressure is highly unlikely to cause the cardiac phenotype in 8‐month‐old α₂ ^+/G301R mice.

Our findings in 3‐month‐old mice are in line with other reports on unchanged blood pressure on cardiac‐ and vascular smooth muscle cell–specific knockout of the Na,K‐ATPase α₂ isoform in mice. ⁵⁸ More important, these measurements were done with tail‐cuff method (ie, in stressed mice). Another telemetry study reported elevated blood pressure in mice with smooth muscle–specific knockout of the Na,K‐ATPase α₂ isoform, suggesting elevated total peripheral resistance. ⁵⁹ This is in contrast to our findings and may be a result of tissue‐specific knockout generation. To our knowledge, no study on aged mice with the Na,K‐ATPase expression manipulation is available yet.

Changed Expression of the Na,K‐ATPase α Isoforms in the Heart of α₂ ^+/G301R Mice Did Not Appear to Affect Electrical Activity of Cardiomyocytes

Previous expression analyses of the α₂ ^+/G301R mice showed a decrease in expression of the Na,K‐ATPase α₂ isoform in brain lysate ²⁵ and in the arterial wall. ²⁷ In the present study, we also found reduced expression of the α₂ isoform accompanied by increased expression of the α₁ isoform in the hearts from α₂ ^+/G301R mice compared with WT mice. The increased α₁ expression may be compensatory for the lack of electrogenic activity of the α₂ isoform. ⁴ , ¹⁰ In fact, similar membrane potentials in cardiomyocytes from the 2 genotypes were seen, and 3‐month‐old mice of both genotypes showed similar EF. However, increase of the cardiac α₁ isoform was observed in both 3‐ and 8‐month‐old α₂ ^+/G301R mice, suggesting that this cannot directly explain the decline in cardiac performance in 8‐month‐old α₂ ^+/G301R mice. Therefore, it is unlikely that the observed cardiac functional abnormalities are a result of changed ion transport.

It has previously been reported that heterozygous knockout of either α₁ or α₂ isoform of the Na,K‐ATPase affected the nontargeted α isoform differently. The α₂ isoform was increased in hearts from α₁ ^+/‐ mice, whereas the α₁ isoform expression was unchanged in hearts from α₂ ^+/– mice. ⁶⁰ On this background, the current finding of α₁ isoform upregulation in hearts from α₂ ^+/G301R mice is surprising. The expression of the Na,Ca‐exchanger and several other Ca²⁺ transporting proteins was similar in hearts from WT and α₂ ^+/G301R mice. Similar expression in both genotypes of the Na,Ca‐exchanger is in contrast to previous reports on reciprocal regulation of the expression of the Na,K‐ATPase α₂ isoform and the Na,Ca‐exchanger, ⁶¹ , ⁶² suggesting, at least at the expression level, that Na⁺ and Ca²⁺ exchange is not affected in α₂ ^+/G301R mice.

Changed Signaling of the Src/Ras/Erk1/2 Pathway Was Associated With Mitochondrial Uncoupling in α₂ ^+/G301R Mice

The cardiac Na,K‐ATPase is suggested to have a distinct function in oxidative stress amplification. ²⁰ , ⁵⁴ Thus, binding of cardiotonic steroids to the Na,K‐ATPase or its chemical modification by intracellular signaling activates Src kinase that, in turn, transactivates the epidermal growth factor receptor, which initiates Ras/Erk1/2 signaling. ¹⁶ This Src/Ras/Erk1/2 pathway in cardiomyocytes is suggested to increase generation of ROS, which is prevented by additive inhibition of mitochondrial complexes I and III. ¹⁶ The 8‐month‐old α₂ ^+/G301R mice showed increased signaling capacity of the Src/Ras/Erk1/2 cascade and amplification of this signaling after inhibition of the Na,K‐ATPase by ouabain. This larger capacity for activation suggests that under hemodynamic or metabolic stress, the hearts from 8‐month‐old α₂ ^+/G301R mice may experience increased activation of the Src/Ras/Erk1/2 pathway. Accordingly, proteomics data analysis predicted increased Src/Ras/Erk1/2 signaling in 8‐month‐old α₂ ^+/G301R mice. The increased total Src expression may be compensatory to the increased expression of the Na,K‐ATPase α₁ isoform, which is suggested to regulate the phosphorylation level of Src, ⁶³ and thus, keeping the level of activated Src within a normal range under resting conditions. This produces, however, a background for misbalance in the maximal activation capacity leading to disproportional Src/Ras/Erk1/2 signaling on Src activation.

The Na,K‐ATPase–dependent ROS pathway is suggested to be potentiated on carbonylation of the α₁ isoform directly by ROS. ⁵³ Moreover, in heart failure, ROS were shown to reversibly inhibit the Na,K‐ATPase through oxidative modification involving NADPH oxidase and glutathionylation. ⁶⁴ Notably, an upregulation in enzymes implicated in generation of ROS and regulation of cellular oxidation‐reduction state was detected in hearts from 8‐month‐old α₂ ^+/G301R mice. Thus, oxidative stress, increasing with age, might be amplified by the increased expression of the Na,K‐ATPase α₁ isoform in the hearts from α₂ ^+/G301R mice. Therefore, α₁ Na,K‐ATPase–ROS amplification through the Src/Ras/Erk1/2 pathway may be responsible for mitochondrial uncoupling and thereby attributable to ROS generation. ⁵⁴ Of note, the hearts from 3‐month‐old α₂ ^+/G301R and WT mice showed similar capacity and signaling of the Src/Ras/Erk1/2 pathway, which was associated with similar mitochondrial function and normal levels of oxidative stress in both genotypes. In contrast to the Src/Ras/Erk1/2 signaling pathway, we found no changes in PLCγ/inositol trisphosphate‐receptor signaling, which previously has been proposed to be associated with the Na,K‐ATPase. ¹⁴ , ¹⁷ This is in line with the functional data suggesting similar Ca²⁺ homeostasis in cardiomyocytes from mice of both genotypes.

This study has some limitations to be considered. First, the G301R mutation, that we have studied, is known to cause a severe FHM2 phenotype. ²⁶ However, this is 1 of >100 mutations in the ATP1A2 gene that are known to be associated with FHM2. These ATP1A2 mutations have been suggested to have different consequences for the α₂ isoform expression and membrane localization. ⁶⁵ Whether other ATP1A2 mutations also affect cardiac function remains to be studied. Second, the use of a mouse model to study the association between FHM2 and cardiovascular disease has inherent limitations in representing human beings. Thus, it remains to be investigated whether patients with FHM2 exhibit similar expression changes of Na,K‐ATPase α isoforms and a cardiac phenotype similar to what we observed in α₂ ^+/G301R mice. Finally, cardiac output, measured by magnetic resonance imaging, was assessed in anesthetized mice, which might affect the cardiovascular system. Direct radiotelemetric measurement of cardiac output would have been preferable, but this is not yet technically feasible.

In conclusion, our results suggest that the FHM2‐associated G301R mutation of the Na,K‐ATPase α₂ isoform leads to cardiac dysfunction and ventricular dilation accompanied by metabolic shift in cardiomyocytes. We provided functional evidence that this relates to mitochondrial uncoupling that was associated with oxidative stress. We suggest that these abnormalities are mediated via amplified Na,K‐ATPase–dependent ROS signaling through the Src/Ras/Erk1/2 pathway. These findings propose a novel mechanistic background for comorbidity between migraine and cardiovascular disease. ⁶⁶ , ⁶⁷ , ⁶⁸ , ⁶⁹ , ⁷⁰ , ⁷¹ , ⁷² , ⁷³ , ⁷⁴ , ⁷⁵ , ⁷⁶ , ⁷⁷ , ⁷⁸ , ⁷⁹ , ⁸⁰ , ⁸¹ , ⁸² , ⁸³ , ⁸⁴

Sources of Funding

This work was supported by the Independent Research Fund Denmark—Medical Sciences (8020‐00084B; 9149‐00003B) and the Novo Nordisk Foundation (NNF19OC0058460).

Disclosures

None.

Supporting information

Tables S1–S2

Figures S1–S10

References 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84

Data S1–S4