Excitation-contraction coupling, cardiomyocyte electrophysiology, and transcriptome profiles in two HFpEF murine models: Etiology and sex-dependent differences

Abstract

Heart failure (HF) with preserved ejection fraction (HFpEF) comprises heterogenous clinical phenotypes and variable co-morbidities. Recent two-hit translational animal models, including the hypertensive, nitrosative-stressed mice fed with high-fat diet and L-NAME (HFD+L-NAME) and the obese-diabetic leptin receptor-deficient db/db mice with excess aldosterone (db/db+Aldo), may phenocopy select subgroups of HFpEF. We systematically compared mechanisms of excitation-contraction coupling (ECC), electrophysiology, and gene transcription in these preclinical HFpEF models and between sexes, including morphometry, echocardiography, cellular electrophysiology, intracellular Ca²⁺ imaging, and RNA-sequencing. The multiorgan HFpEF phenotype showed key differences between the two models: db/db+Aldo mice were markedly obese, had severe hyperglycemia and hepatomegaly, whereas male HFD+L-NAME mice had more pronounced cardiac hypertrophy. Diastolic dysfunction (quantified as echocardiographic E/e’) was more severe in db/db+Aldo mice and worse in females, whereas females showed milder diastolic dysfunction in HFD+L-NAME. Marked proarrhythmic action potential (AP) changes (prolonged AP duration, increased short-term variability, and reduced alternans threshold) occurred in db/db+Aldo (in both sexes), while these AP changes were less severe in male HFD+L-NAME and absent in female HFD+L-NAME. In line with these findings, differential ionic current and Ca²⁺ handling changes occurred between these two HFpEF models and between sexes. RNA-sequencing revealed highly distinctive gene expression profiles between HFpEF models. We conclude that marked differences exist in cardiomyocyte ECC, electrophysiology, and gene expression between HFD+L-NAME and db/db+Aldo mice and between sexes. This indicates that a combination of translational HFpEF models that mimic select HFpEF sub-phenogroups are critical to better understand HFpEF mechanisms for therapeutic drug development.

NEW & NOTEWORTHY

Excitation-Ca²⁺ signaling-contraction coupling (ECC) mechanisms are fundamental to heart function. ECC mechanisms are differentially altered in murine models of heart failure with preserved ejection fraction (HFpEF) by dominant disease pathology and sex. Diastolic dysfunction is more pronounced in diabetic-obese HFpEF mice (worse in females) than in hypertensive-obese HFpEF mice (female sex is protective). Diabetic-obese mice and primarily hypertensive-obese HFpEF mice exhibit differential ECC alterations and largely distinctive transcription changes, providing mechanistic insights into HFpEF sub-phenogroups.

INTRODUCTION

Heart failure (HF) with preserved ejection fraction (HFpEF) account for at least half of all HF cases and is a major socio-economic burden (1). The incidence of HFpEF is rapidly growing due to the epidemic of disease-provoking co-morbidities, including diabetes mellitus (DM), obesity, hypertension, and lung and kidney diseases (2). Patients with HFpEF have poor quality of life, severe exercise constraints, and high mortality rates (5-yr survival is ~25–40%) with very limited effective therapeutics available (3). Profound deficits in the mechanistic understanding of HFpEF pathobiology remain, and this limits strategies for effective therapeutic interventions (4).

Excitation-contraction coupling (ECC) impairments and ion channel remodeling play key roles in contractile deficit and arrhythmias in HF (5, 6). In HF with reduced ejection fraction (HFrEF), prior research identified the molecular and cellular mechanisms underlying altered ECC and electrophysiology (7–9); however, their exact contributions to cardiac dysfunction in HFpEF are largely unknown (10). Patients with HFpEF (and especially with DM) are at higher risk for cardiac arrhythmias and sudden cardiac death (11, 12); however, the exact arrhythmia risk and mechanisms in HFpEF are unclear.

Preclinical animal models can inform precise molecular mechanisms, reveal new therapeutic targets, and test new drugs. However, many earlier preclinical HFpEF animal models did not capture the complex human HFpEF syndrome, providing limited depth and breadth of mechanistic understanding and therapeutic development (13). Recently, two- or multi-hit models have started to better mirror the human HFpEF syndrome, including cardiac hypertrophy, pulmonary congestion, exercise intolerance, and worsened diastolic function (14). HFpEF patients represent a very heterogenous population with various co-morbidities, so the one-size-fits-all strategy in animal models is unlikely to adequately translate research findings to clinics (13, 15, 16). Thus, we aimed to compare electrophysiology, ECC, and cardiomyocyte arrythmia mechanisms in two distinct translational HFpEF mouse models that combine metabolic and hemodynamic stresses: (a) high-fat diet (HFD) combined with the constitutive nitric oxide synthase (NOS) inhibitor L-Nitro-Arginine Methyl Ester (L-NAME) treatment (HFD+L-NAME) for 15 weeks (17), and (b) leptin-receptor deficient db/db mice combined with aldosterone (Aldo) infusion (db/db+Aldo) for 4 weeks (18). These distinct preclinical HFpEF models may phenocopy select subgroups of HFpEF patients and help to refine individualized therapeutics (16, 19). Importantly, opposite sex-differences have been reported in these HFpEF models, female sex is associated with worsened diastolic dysfunction in db/db+Aldo (20), but female sex is protective in HFD+L-NAME (21). A quantitative comparison of ECC and electrophysiology mechanisms between models and sexes in each will improve our mechanistic understanding of cardiomyocyte function in HFpEF and could reveal new therapeutic targets that may be applicable to select HFpEF sub-phenogroups (including by sex) or to the broader heterogenous HFpEF population.

MATERIALS AND METHODS

Ethical Approval

All animal handling procedures strictly adhered to the approved protocol #23175 of the Institutional Animal Care and Use Committee at University of California, Davis conforming to the NIH Guide for the Care and Use of Laboratory Animals (8^th edition, 2011).

Animal Procedures

Adult wild-type (WT, RRID:IMSR_JAX:000664) and Lepr^db/db (RRID:IMSR_JAX:000697) C57BL/6J mice were obtained from The Jackson Laboratory. Animals had unlimited access to sterilized water and food, and were housed in an enriched environment with standard bedding and nesting material, under a 12/12 h day-night cycle in a humidity (40–60%) and temperature (20–22°C) controlled facility.

HFpEF was studied in two contemporary two-hit models (Fig. 1A). First, WT mice (10-week-old) were fed with HFD (60 kcal% fat, D12492, Research Diets) and drinking water was supplemented with L-N^G-nitroarginine methyl ester (L-NAME, 0.5 g/L, Sigma-Aldrich) for 15 weeks (17, 22). Second, osmotic minipumps (Alzet, 2004) were implanted subcutaneously in 12-wk-old mice that delivered a continuous infusion of d-aldosterone (0.3 μg/h) for 4 weeks in db/db mice (18, 20). Age-matched sham control WT mice were implanted with minipumps filled with vehicle only (saline with 5% ethanol), and these mice were fed with a normal chow (Teklad, 2918). In another small cohort of control mice, we also used 25-wk-old wild-type mice fed with a normal chow; however, they showed no significant differences from the 16-wk-old adult sham control mice in any echocardiographic or morphometric parameters measured except for a 5% larger body weight in the chow control group (Table 1). Thus, we report data with the 16-wk-old sham control mice in the figures.

Figure 1.

Distinct morphometric and metabolic profiles in two HFpEF murine models. A: Translational murine models to study heart failure with preserved ejection fraction (HFpEF) combine metabolic and hemodynamic stimuli. Vehicle-treated adult wild-type sham control mice were compared with 2 HFpEF models: leptin receptor-deficient db/db mice with chronic aldosterone infusion (db/db+Aldo), and mice fed with high-fat diet (HFD) and L-NAME (an inhibitor of nitric oxide synthase, NOS) via drinking water (HFD+L-NAME). Male (M) and female (F) animals were enrolled in equal numbers to study sex-differences in HFpEF. B: Obesity. C: Increased blood glucose levels. D: Cardiac hypertrophy quantified as increased heart weight to tibial length ratio (HW/TL). E: Pulmonary oedema quantified as increased wet lung weight to dry lung weight ratio. F: Hepatomegaly. Welch ANOVA followed by Dunnett’s T3 multiple comparisons test except for lung and liver weights where Kruskal-Wallis ANOVA followed by Dunn’s multiple comparisons test was used. N=16 animals/experimental group except for female db/db+Aldo where N=15; and lung weight measurements where N=12 for each group. Values reported are means ± SD.

Table 1.

Morphometry and echocardiographic parameters in two healthy control groups of mice.

	Vehicle control		Normal chow
Sex	Male	Female	Male	Female
N (animals)	16	16	10	10
Age (week)	16	16	25	25
Body Weight (g)	31±2	23±1	33±2	25±2
Blood Glucose (mg/dL)	193±44	171±28	206±28	204±17
HW / TL (g/mm)	0.97±0.07	0.84±0.07	0.95±0.10	0.74±0.05
Ejection Fraction (%)	81±3	81±2	80±2	80±3
LV Mass (mg)	108±11	95±7	109±14	86±12
LVRI (LVM/LVIDd)	32±2	29±2	30±2	26±2
E/A	1.3±0.1	1.4±0.2	1.4±0.2	1.5±0.2
E/e’	29±2	28±2	28±1	27±4

Healthy wild-type mice implanted with osmotic minipumps filled with vehicle only or fed with normal chow were used as controls for HFpEF. Mice in the normal chow group were 9-wk older and had sligthly increased body weight but had similar cardiac functional parameters as vehicle control mice.

(HW/TL, heart weight to tibial length ratio; LV, left ventricle; LVM, LV mass; LVRI, LV remodeling index; LVIDd, left ventricular internal diameter at diastole; E/A, mitral E wave to A wave ratio; E/e’, mitral E wave to e’ wave ratio.)

Animal experiments were performed blinded, and we used block randomization with a block size of 4 animals (for each genotype, treatment, and sex), with 32 sham control, 32 db/db+Aldo, and 32 HFD+L-NAME mice included. Each treatment group included equal numbers of male and female animals. One female db/db+Aldo mouse died before the conclusion of the study and showed severe pulmonary and hepatic congestion. No animal was excluded from analysis.

Animals were injected with heparin (400 U/kg) and were subjected to general anesthesia by 2–5% isoflurane inhalation in 100% oxygen throughout the terminal surgical procedure. All animals were euthanized by surgical excision of the heart while in deep anesthesia. Enzymatic isolation of cardiomyocytes was performed as previously described (23). Briefly, upon confirmation of abolished pain reflexes, hearts were excised and rinsed in cold nominally Ca²⁺-free Minimal Essential Medium. The aorta was cannulated and retrograde perfused on a constant flow Langendorff apparatus at 37ºC with Ca²⁺-free normal Tyrode’s solution, gassed with 100% O₂. Then, the heart was perfused for 10–15 min with collagenase (type 2, Worthington) and protease (type XIV, Sigma-Aldrich) in Tyrode’s solution (with 10 μmol/L free [Ca²⁺]) to enzymatically isolate cardiomyocytes. Following digestion, the myocytes were gently triturated with a pipette, then filtered through a 200 μm nylon mesh and allowed to sediment for 10 min. The sedimentation was repeated three times using increasing [Ca²⁺] from 0.125 to 0.25 then 0.5 mmol/L. Finally, ventricular myocytes were kept in Tyrode’s solution (0.5 mmol/L [Ca²⁺]) at room temperature until use.

Echocardiography

Systolic and diastolic ventricular heart functions of mice were assessed by transthoracic echocardiography using the Vevo 2100 echocardiography system (FUJIFILM VisualSonics, Toronto, ON, Canada) equipped with a 40 MHz linear probe. Mice fur was removed by a depilatory the day before echocardiography recordings. During recordings, mice were anesthetized with isoflurane inhalation (1.5%), which was later individually adjusted (between 1 to 3%) to achieve a stable heart rate between 500 to 600 beats/min (to avoid depressed contractile function) when assessing systolic cardiac function, and between 350 to 450 beats/min (to avoid fusion of E and A waves) when assessing diastolic dysfunction in each animal. ECG monitoring was obtained using limb electrodes, and core temperature was carefully monitored and maintained at 37ºC during the entire procedure. Left ventricular (LV) M-mode echocardiography in parasternal short-axis view was performed for assessment of LV dimensions and systolic function. Pulsed wave Doppler and tissue Doppler images were acquired to assess LV diastolic function. At least three consecutive cardiac cycles were sampled for each measurement taken, and blinded analysis was performed off-line.

Myocyte Electrophysiology

Isolated single left ventricular murine cardiomyocytes were placed in a temperature-controlled perfusion chamber (Warner Instruments) mounted on a Leica DMI3000 B inverted microscope. Cells were bathed at 37°C (for 10 minutes before starting the experiments) and continuously perfused (2 mL/min) with Tyrode’s solution. Electrodes were fabricated from borosilicate glass (World Precision Instruments) having tip resistances of 2 to 2.5 MΩ when filled with an internal solution. Axopatch 200B amplifier (Axon Instruments) was used for recordings and the signals were digitized at 50 kHz by a Digidata 1322A A/D converter (Axon Instruments) under software control (pClamp10.4). Series resistance was typically 3 to 5 MΩ and it was compensated by ≥85%. Experiments were discarded when the series resistance was high or increased by ≥20% during the recordings. The pH of all applied solutions was regularly checked and carefully adjusted. All electrophysiology experiments were conducted at 37±0.2°C.

Action potentials (APs) were recorded in whole-cell I-clamp conditions where cells were stimulated using supra-threshold depolarizing pulses (with a duration of 2-ms) delivered via the patch pipette. The Tyrode’s solution for bath perfusion contained (in mmol/L): NaCl 140, KCl 4, CaCl₂ 1.8, MgCl₂ 1, HEPES 5, Na-HEPES 5, glucose 5.5; pH=7.40. Pipette solution contained (in mmol/L): K-aspartate 100, KCl 30, NaCl 8, Mg-ATP 5, phosphocreatine dipotassium salt 10, HEPES 10, EGTA 0.01, cAMP 0.002, and calmodulin 0.0001; pH=7.20 (with KOH). Using this solution, the intracellular Ca²⁺ transient and contraction of the cardiomyocyte were preserved. AP durations at 20, 50, and 90% repolarization (APD₂₀, APD₅₀, and APD₉₀, respectively) were used to characterize AP repolarization dynamics. Series of 50 consecutive APs were analyzed to estimate short-term variability (STV) of APD₉₀ according to the following formula: STV=Σ(│APD_n+1−APD_n│)/[(n_beats−1)×√2], where APD_n and APD_n+1 indicate the durations of the n^th and (n+1)^th APs, and n_beats denotes the total number of consecutive beats analyzed.

K⁺ currents were recorded in whole-cell V-clamp experiments using an internal solution containing (in mmol/L): K-Aspartate 100, KCl 20, NaCl 8, Mg-ATP 5, EGTA 10, CaCl₂ 4.1, HEPES 10, cAMP 0.002, phosphocreatine-K₂ 10, and calmodulin 0.0001, with pH=7.2 (free [Ca²⁺]_i=100 nmol/L, calculated using the WEBMAXC Extended version of the MaxChelator software, RRID:SCR_000459, https://somapp.ucdmc.ucdavis.edu/pharmacology/bers/maxchelator/webmaxc/webmaxcE.htm), and in the presence of Na⁺ and Ca²⁺ current inhibitors (10 µmol/L tetrodotoxin for I_Na, and 10 µmol/L nifedipine for I_Ca,L) in the perfusing Tyrode’s solution. Different K⁺ current components were separated using biexponential fitting (R²>0.9 in each case) to the decay of the voltage-gated outward K⁺ currents (I_Kv). Voltage-gated K⁺ currents were elicited using a 4.5 s-long test pulse to +60 mV from a holding potential of –80 mV with an inter-pulse interval of 5.5 s. I_K1 current traces were analyzed at the end of 500-ms test pulses to –140 mV (inward I_K1) and –40 mV (outward I_K1) from the holding potential of –80 mV.

I_Na,L was recorded using an internal solution containing (in mmol/L): CsCl 110, tetraethylammonium chloride 20, Mg-ATP 5, HEPES 10, phosphocreatine disodium salt 5, calmodulin 0.0001, EGTA 10, CaCl₂ 4.1 (free [Ca²⁺]=100 nmol/L), pH=7.20. Bath solution contained (in mmol/L): NaCl 140, CsCl 4, CaCl₂ 1.8, MgCl₂ 1, HEPES 5, Na-HEPES 5, glucose 5.5, 4-aminopyridine 5, nifedipine 0.01, pH=7.40. I_Na,L was measured at the final 10-ms of a 500-ms depolarizing pulse to −40 mV from a –120 mV holding potential (to maximize Na⁺ channel availability). I_Na,L could be inhibited by tetrodotoxin (TTX, 10 μmol/L), and the TTX-sensitive current amplitude was reported in each cell.

I_Ca,L was recorded using the same internal solution as listed above for I_Na,L measurements. The same bath solution as for I_Na,L measurements was used except that nifedipine was replaced with TTX (10 μmol/L). I_Ca,L was measured using 500 ms-long voltage steps to test potentials between –40 and +20 mV from a holding potential of –80 mV every 5 s with a 50 ms pre-step to –40 mV to inactivate Na⁺ channels. At the end of each experiment, I_Ca,L was inhibited using nifedipine (10 μmol/L).

All ionic currents were normalized to cell capacitance (i.e., current density), determined in each cell using short (10 ms) hyperpolarizing pulses from −10 mV to −20 mV.

Myocyte Ca²⁺ Imaging

Intracellular [Ca²⁺] transients (CaTs) and diastolic Ca²⁺ events (sparks) were measured in freshly isolated ventricular cardiomyocytes loaded with Fluo-4 AM (10 μmol/L, Invitrogen) and Pluronic F-127 (0.02%, Invitrogen). The dye was loaded for 30 minutes at room temperature followed by wash and de-esterification for 30 minutes. Fluo-4 was excited at 488 nm using an Argon laser, and emission was collected using a 500–530 nm bandpass filter. Images were recorded using confocal microscopy in line scan mode (Bio-Rad Radiance 2100) using a 40x objective and scanned at 6 ms/line. Intact cardiomyocytes were plated on laminin-coated coverslips and paced at 1 Hz and 2 Hz in a field stimulation chamber (Warner Instruments, Inc.). Myocytes were continuously perfused with Tyrode’s solution containing (in mmol/L): NaCl 140, KCl 4, CaCl₂ 1.8, MgCl₂ 1, HEPES 5, Na-HEPES 5, glucose 5.5; pH=7.40. After pacing was stopped, Ca²⁺ sparks were recorded for 1 min. Sarcoplasmic Ca²⁺ content was then assessed by local delivery of 10 mmol/L caffeine. Rate of Ca²⁺ removal via Na⁺/Ca²⁺ exchange (NCX) was determined by linear fitting of the initial 3 s of the caffeine-induced [Ca²⁺] transient decay. During analysis, the non-cellular background fluorescence was subtracted from all measured fluorescence values, and the initial baseline fluorescence (F₀) in myocytes quiescent for long time periods was determined in each cell. There was no statistical difference in F₀ values at rest between any experimental groups. That does not guarantee that resting [Ca²⁺]_i was unaltered among groups, but it allows assessment of the relative rises in end diastolic [Ca²⁺]_i between beats during regular pacing as F/F₀ (with each cell as its own resting F₀ control). Frequency of Ca²⁺ sparks was analyzed before the occurrence of a spontaneous Ca²⁺ wave (if any occurred in each cell).

RNA-sequencing

Total RNA was extracted from whole-heart lysates using TRIzol and the Qiagen RNeasy kit as previously described (24). RNA-sequencing was performed by BGI Genomics using paired-end 100 nucleotide sequencing with 40 million reads. Cleaned reads were aligned to reference genomes (mouse: GRCm38/mm10) using STAR (25) with de novo junction discovery and summarized at the gene level using featureCounts (RRID:SCR_012919) ((26). Differential expression analysis was completed using DESeq2 (RRID:SCR_000154) using the standard workflow (27). Batch correction for different sequencing runs between HFpEF models was performed using limma’s (RRID:SCR_010943) removeBatchEffect function using the standard workflow (28). Pathway and ontology enrichment analyses for all experiments were performed using gprofiler2 (29). Heatmaps were generated using the pheatmap package in R. Sequencing data have been deposited in GEO (accession number: GSE284354).

Statistics

Data are presented as Mean ± SD. Normality of the data was assessed by Shapiro-Wilk test and the equality of group variance was tested using Brown-Forsythe test. Statistical significance of differences was determined using ANOVA followed Šídák’s or Dunnett’s multiple comparisons test, when applicable. Data that were not normally distributed were analyzed by Kruskal-Wallis ANOVA with Dunn’s multiple comparisons test. Pairwise comparisons were made between the following groups (prespecified in experimental design): (1) male control vs. female control (2) male control vs. male db/db+Aldo, (3) male control vs. male HFD+L-NAME, (4) female control vs. female db/db+Aldo, (5) female control vs. female HFD+L-NAME, (6) male db/db+Aldo vs. female db/db+Aldo, and (7) male HFD+L-NAME vs. female HFD+L-NAME. Multiplicity adjusted P values were reported for each comparison. Blinded data acquisition and analysis have been performed for all in vivo measurements. Animals were grouped with no blinding but randomized in cellular experiments. Fully blinded analysis was not performed in cellular studies because the same person carried out the experiments and analysis. For proper allocation concealment, animals were recruited blinded based on sequential ear tag numbers randomly assigned by the animal housing staff. Male and female animals were used in equal numbers.

Data processing, analysis, and plotting have been performed using Clampfit 10 (Molecular Devices), ImageJ (RRID:SCR_003070) with SparkMaster plugin, GraphPad Prism 10 (RRID:SCR_002798), and Origin 2016 (OriginLab) software. Illustration in Fig. 1 was generated in part using Servier Medical Art, provided by Servier, licensed under a Creative Commons Attribution 3.0 unported license.

RESULTS

Distinct multiorgan HFpEF phenotypes between two HFpEF models and sexes

Two-hit models combining metabolic and hemodynamic stresses were used to study HFpEF in mice. First, we compared basic morphometric and metabolic parameters in db/db+Aldo and HFD+L-NAME HFpEF animals versus sham control wild-type mice and between sexes (Fig. 1A).

Mice in both HFpEF models were obese with larger weight gains in db/db+Aldo (Fig. 1B). Female animals have lower body weight in control and HFD+L-NAME; but female db/db+Aldo mice were just as obese as male db/db+Aldo mice (Fig. 1B). Blood glucose levels were markedly elevated in db/db+Aldo mice with a trend for even more elevated blood glucose in female db/db+Aldo, whereas only a slight increase in non-fasting glucose levels was seen in HFD+L-NAME mice (Fig. 1C). Cardiac hypertrophy, quantified as heart weight to tibial length ratio, was increased similarly in both db/db+Aldo and HFD+L-NAME, but with larger increases in male HFD+L-NAME and female db/db+Aldo (Fig. 1D). Pulmonary edema (wet lungs) was seen in both HFpEF models in line with common clinical HF symptoms, and the extent of congestion was larger in db/db+Aldo vs. HFD+L-NAME mice (Fig. 1E). Hepatomegaly was pronounced in db/db+Aldo mice but not in HFD+L-NAME mice (Fig. 1F). These data indicate important metabolic differences between the 2 HFpEF models and sexes affecting multiorgan HFpEF phenotype.

Cardiac hypertrophy and diastolic dysfunction severity differ between HFpEF models and sexes

Echocardiographic evaluation showed preserved ejection fraction (EF>50%) in both HFpEF models (Fig. 2). Concentric cardiac hypertrophy was evident from significantly increased left ventricular (LV) mass and LV remodeling index (LVRI, a ratio between LV mass and LV internal diameter) in both HFD+L-NAME and db/db+Aldo versus sham control mice (Fig. 2B). Cardiac hypertrophy was prominent in male HFD+L-NAME and female db/db+Aldo mice (Fig. 2B) paralleling the gravimetric analysis. Diastolic E/A ratio was more variable among HFpEF animals, it was reduced in some HFpEF male mice, while it was significantly increased in female HFpEF mice in both models indicating a restrictive filling pattern in female HFpEF (Fig. 2C). In contrast to E/A, diastolic E/e’ ratio was significantly increased in all HFpEF mice (Fig. 2C). The increase in E/e’ was larger in db/db+Aldo vs. HFD+L-NAME mice. E/e’ was significantly higher in female versus male db/db+Aldo mice, while E/e’ was slightly lower in female versus male HFD+L-NAME mice (Fig. 2C). Left atria were enlarged in all HFpEF mice, with smaller increases seen in female vs. male HFD+L-NAME mice, while male and female db/db+Aldo mice had similar atrial enlargement (Fig. 2C). These data indicate differential sex-dependent diastolic dysfunction in the two HFpEF models, i.e., female sex is protective in the HFD+L-NAME model, whereas females have worse diastolic dysfunction in the db/db+Aldo model.

Figure 2.

Comparison of echocardiographic parameters in two HFpEF murine models. A: Left ventricular (LV) M-mode, flow and tissue Doppler echocardiographic images in male (M) and female (F) mice. Vehicle-treated wild-type sham control mice were compared with 2 heart failure with preserved ejection fraction (HFpEF) models: db/db mice with chronic aldosterone infusion (db/db+Aldo), and wild-type mice with high-fat diet (HFD) and L-NAME treatment (LVAW, LV anterior wall; LVID, LV internal diameter; LVPW, LV posterior wall). B: Preserved ejection fraction (EF), increased LV mass and LV remodeling index (LVRI, a ratio between LV mass and LVID at diastole) indicating concentric cardiac hypertrophy in HFpEF. C: Diastolic dysfunction indices (E/A, E/e’) and enlarged left atria (LA) in HFpEF (E/A, ratio between mitral E wave and A wave; E/e’, ratio between mitral E wave and e’ wave). Welch ANOVA followed by Dunnett’s T3 multiple comparisons test except for E/A where Kruskal-Wallis ANOVA with Dunn’s multiple comparisons test was used. N=16 animals/experimental group, except for female db/db+Aldo, where N=15, and LA area where N=8 for each group. Values reported are means ± SD.

Action potential changes differ between HFpEF models and sexes

We studied excitation-contraction coupling mechanisms that may contribute to cardiomyocyte dysfunction. First, we measured action potentials (APs) in ventricular myocytes at various pacing frequencies to assess electrophysiological changes and arrhythmogenicity (Fig. 3). AP duration at 90% repolarization (APD₉₀) was markedly prolonged in both male and female db/db+Aldo myocytes and in male HFD+L-NAME myocytes vs. sham control (Fig. 3, A and B). However, APD₉₀ was unchanged in female HFD+L-NAME (Fig. 3). Further analysis revealed that the early AP repolarization (APD₂₀ and APD₅₀) was prolonged only in male db/db+Aldo myocytes (Fig. 3B). There was a trend for slightly more depolarized resting membrane potential and reduced maximal rate of AP upstroke velocity in both HFpEF models and sexes, although none were statistically significant (Fig. 3C).

Figure 3.

Action potential (AP) parameters in HFpEF cardiomyocytes. A: Representative APs in male (M) and female (F) control and HFpEF ventricular myocytes paced at 1 Hz. B: AP duration (APD) at 20%, 50% and 90% repolarization (APD₂₀, APD₅₀, and APD₉₀, respectively). C: Resting membrane potential (RMP), maximal upstroke velocity (dV/dt_max), and maximal phase 3 repolarization velocity (-dV/dt_max). n(cells)/N(animals) = 20/7 for control male, 25/8 for control female, 24/8 for db/db+Aldo male, 29/8 for db/db+Aldo female, 26/9 for HFD+L-NAME male, and 23/8 for HFD+L-NAME female. Welch ANOVA followed by Dunnett’s T3 multiple comparisons test. Values reported are means ± SD.

APD₉₀ prolongation was larger at lower pacing rates in HFpEF by nearly 50%, except in HFD+L-NAME females where there was little change (Fig. 4A). However, at rapid pacing rate of 10 Hz, significant APD₉₀ alternans was observed in db/db+Aldo cardiomyocytes, while only a subset of HFD+L-NAME myocytes exhibited APD₉₀ alternans (Fig. 4, B and D). A trend for larger APD₉₀ alternans was found in male mice in both HFpEF models (Fig. 4C). The short-term variability (STV) of APD₉₀, a marker for temporal lability of ventricular repolarization and arrhythmia susceptibility (30, 31), was significantly increased in both male and female db/db+Aldo and in male HFD+L-NAME but not in female HFD+L-NAME (Fig. 4, C and D). These data indicate that more pronounced arrhythmogenic AP remodeling occurs in db/db+Aldo vs. HFD+L-NAME mice, and male mice may be more susceptible for alternans than female HFpEF mice.

Figure 4.

Proarrhythmic action potential (AP) changes in HFpEF cardiomyocytes. A: Frequency-dependence of AP duration at 90% repolarization (APD₉₀) in control and HFpEF ventricular myocytes. B: APD₉₀ alternans (S, short; L, long) in HFpEF. C: Fifty consecutive APD₉₀ values used for quantifying short-term variability (STV) at 1 Hz pacing. D: Increased APD₉₀, STV, and APD₉₀ alternans magnitude in HFpEF. n(cells)/N(animals) at 1 Hz pacing = 20/7 for control male, 25/8 for control female, 24/8 for db/db+Aldo male, 29/8 for db/db+Aldo female, 26/9 for HFD+L-NAME male, and 23/8 for HFD+L-NAME female. n(cells)/N(animals) at 10 Hz pacing = 13/7 for control male, 16/8 for control female, 11/7 for db/db+Aldo male, 18/8 for db/db+Aldo female, 18/9 for HFD+L-NAME male, and 15/8 for HFD+L-NAME female. STV: Welch ANOVA followed by Dunnett’s T3 multiple comparisons test. Alternans occurrence: Chi-square test with Yates correction. APD₉₀ alternans magnitude: Kruskal-Wallis ANOVA with Dunn’s multiple comparisons test. Values reported are means ± SD.

Distinct K⁺ current changes between HFpEF models and sexes.

We performed voltage-clamp experiments to study ionic current changes in HFpEF, which may be responsible for the observed changes in AP repolarization. The cell capacitance was significantly increased, indicating cardiomyocyte hypertrophy in both db/db+Aldo and HFD+L-NAME HFpEF models (Fig. 5), consistent with the ventricular hypertrophy (Figs. 1D and 2B). To characterize electrophysiological remodeling, we report net ionic current magnitudes and then each current was normalized to that myocyte’s capacitance (i.e., current density). First, we measured the major K⁺ currents in murine ventricular cardiomyocytes (Fig. 6A and 7A).

Figure 5.

Increased cell capacitance of HFpEF ventricular cardiomyocytes. n(cells)/N(animals) for cell capacitance = 66/8 for control male, 67/8 for control female, 67/8 for db/db+Aldo male, 73/8 for db/db+Aldo female, 70/9 for HFD+L-NAME male, and 78/8 for HFD+L-NAME female. M, male; F, female. Kruskal-Wallis ANOVA with Dunn’s multiple comparisons test. Values reported are means ± SD.

Figure 6.

Inward rectifier K⁺ current (I_K1) changes in HFpEF cardiomyocytes. A: Representative I_K1 traces at −140 mV in male (M) and female (F) control and HFpEF ventricular myocytes. B: I_K1 net current at −140 mV (inward I_K1) and −40 mV (outward I_K1) in HFpEF. C: I_K1 density (I_K1 magnitude divided by cell capacitance in each cell). n(cells)/N(animals) for I_K1 = 20/6 for control male, 21/6 for control female, 23/6 for db/db+Aldo male, 22/6 for db/db+Aldo female, 27/9 for HFD+L-NAME male, and 28/8 for HFD+L-NAME female. Kruskal-Wallis ANOVA with Dunn’s multiple comparisons test. Values reported are means ± SD.

Figure 7.

Voltage-gated K⁺ current (I_Kv) changes in HFpEF cardiomyocytes. A: Representative I_Kv traces male (M) and female (F) control and HFpEF ventricular myocytes. B: Transient outward K⁺ current (I_to), slowly inactivating K⁺ current (I_K,slow), and sustained K⁺ current (I_sus) were separated by biexponential fitting to I_Kv traces. C: I_to, I_K,slow, and I_sus densities (current magnitude divided by cell capacitance in each cell). n(cells)/N(animals) = 15/5 for control male, 20/6 for control female, 17/6 for db/db+Aldo male, 18/6 for db/db+Aldo female, 24/9 for HFD+L-NAME male, and 22/8 for HFD+L-NAME female. Welch ANOVA with Dunnett’s T3 multiple comparisons test. Values reported are means ± SD.

The inward rectifier K⁺ current (I_K1) magnitude was slightly reduced only in female db/db+Aldo myocytes when measured in the inward (at −140 mV) direction, but it was unchanged at the physiologically relevant outward (at −40 mV) direction (Fig. 6B). However, I_K1 density was significantly reduced in both the inward and outward directions in db/db+Aldo cardiomyocytes with no sex-difference (Fig. 6C). I_K1 density was reduced also in male HFD+L-NAME cardiomyocytes; however, it was unchanged in female HFD+L-NAME cardiomyocytes (Fig. 6C). Thus, the functional expression of I_K1 fails to keep pace with the increased myocyte size in HFpEF, except in female HFD+L-NAME.

The voltage-gated K⁺ currents (I_Kv; Fig. 7A) were measured using a long (4.5 s) voltage step, and their components were separated by biexponential fitting. The fast-inactivating transient outward K⁺ current (I_to) magnitude was reduced only in female db/db+Aldo myocytes (Fig. 7B). The slowly inactivating K⁺ current (I_K,slow) and non-inactivating sustained K⁺ current (I_sus) magnitudes were unchanged in both HFpEF models (Fig. 7B). However, I_K,slow was larger and I_sus was smaller in female vs. male HFD+L-NAME (Fig. 7B). When normalized to cell capacitance, I_to density was markedly reduced in both male and female db/db+Aldo myocytes, whereas only a trend for I_to reduction was seen in male HFD+L-NAME and no change in female HFD+L-NAME myocytes (Fig. 7C). There was a trend for reduced I_K,slow density in male and female db/db+Aldo and in male HFD+L-NAME myocytes (Fig. 7C). I_sus density was only reduced in male db/db+Aldo (Fig. 7C). These data suggest that the net K⁺ current magnitudes are relatively preserved, and only the density of select K⁺ currents is reduced in HFpEF suggesting a mismatch between channel expression and cardiomyocyte hypertrophy. The reductions in I_K1, I_to, and I_sus densities could contribute to APD prolongation in HFpEF, especially in db/db+Aldo. However, in male HFD+L-NAME mice remodeling in other ion channels may have a dominant impact on APD prolongation.

Distinct L-type Ca²⁺ current and late Na⁺ current changes between HFpEF models and sexes.

We then measured the major inward currents through voltage-gated ion channels, which occur during AP repolarization, the L-type Ca²⁺ current (I_Ca,L) and the late Na⁺ current (I_Na,L) in HFpEF cardiomyocytes (Fig. 8 and 9). I_Ca,L was differentially remodeled in the two HFpEF models and between sexes (Fig. 8). I_Ca,L magnitude and density were both significantly increased in male HFD+L-NAME cardiomyocytes with a leftward shift in activation voltage-dependence (Fig. 8, A, B, and C). I_Ca,L density was also slightly increased in female HFD+L-NAME, but this increase was significantly smaller than in male HFD+L-NAME (Fig. 8C). In contrast with this, I_Ca,L density was significantly reduced in male db/db+Aldo cardiomyocytes and unchanged in female db/db+Aldo cardiomyocytes (Fig. 8C).

Figure 8.

Distinct L-type Ca²⁺ current changes in two HFpEF mouse models. A: Representative L-type Ca²⁺ current (I_Ca,L) traces at 0 mV in male (M) and female (F) control and HFpEF ventricular myocytes. B: Current-voltage (I-V) relationships of I_Ca,L in control and HFpEF. C: Net I_Ca,L magnitude and I_Ca,L density (I_Ca,L magnitude divided by cell capacitance in each cell). I_Ca,L density was reduced in male db/db+Aldo cardiomyocytes in contrast with an increasing I_Ca,L densities in male and female HFD+L-NAME cardiomyocytes. n(cells)/N(animals)= 25/7 for control male, 21/8 for control female, 19/8 for db/db+Aldo male, 21/8 for db/db+Aldo female, 10/5 for HFD+L-NAME male, and 20/8 for HFD+L-NAME female. ANOVA with Šídák’s multiple comparisons test. Values reported are means ± SD.

Figure 9.

Late Na⁺ current enhancement in HFpEF cardiomyocytes. A: Representative late Na⁺ current (I_Na,L) traces at −40 mV in male (M) and female (F) control and HFpEF ventricular myocytes. I_Na,L was markedly upregulated in HFpEF except for female HFD+L-NAME cardiomyocytes. n(cells)/N(animals)= 8/4 for control male, 10/5 for control female, 10/6 for db/db+Aldo male, 10/6 for db/db+Aldo female, 9/4 for HFD+L-NAME male, and 16/7 for HFD+L-NAME female. Welch ANOVA with Dunnett’s T3 multiple comparisons test. Values reported are means ± SD.

I_Na,L magnitude and density were markedly enhanced in both HFpEF models in male cardiomyocytes, whereas I_Na,L was less increased in female db/db+Aldo and unchanged in female HFD+L-NAME myocytes (Fig. 9, A and B). These data indicate key differences in I_Ca,L and I_Na,L between the two HFpEF models, and changes in I_Ca,L and I_Na,L occurred predominantly in male hearts.

Distinct Ca²⁺ handling alterations between HFpEF models and sexes

Impairments in AP profiles and I_Ca,L are frequently associated with altered intracellular Ca²⁺ transient (CaT) in failing cardiomyocytes. We measured intracellular Ca²⁺ signals in Fluo-4AM loaded cardiomyocytes paced at 1 Hz and 2 Hz (Fig. 10). The CaT amplitude and the sarcoplasmic reticulum (SR) Ca²⁺ content, assessed by rapid caffeine (10 mmol/L) application, were unchanged in both HFpEF models and sexes (Fig. 10). The CaT decay was significantly slower in male db/db+Aldo, unchanged in female db/db+Aldo, and there was a slight trend for faster CaT decay in HFD+L-NAME at 1 Hz pacing (Fig. 10, A and C). At 2 Hz pacing, CaT decay was faster in all experimental groups (vs. 1 Hz pacing), but it was still significantly prolonged in male db/db+Aldo (Fig. 10, A and D). Consistent with the uniquely slower [Ca²⁺]_i decline in the male db/db+Aldo, that was also the only group to show a significant higher end-diastolic [Ca²⁺]_i vs. control (e.g., 1 Hz exemplar in Fig. 10A, with mean F/F₀ increased to 2.09±0.38 vs. 1.75±0.40 in control, where 1.0 is the resting cell level, P=0.021). The decay of the caffeine-induced transient was slowed only in male HFD+L-NAME, indicating reduced NCX function (Fig. 10, B and D).

Figure 10.

Intracellular Ca²⁺ handling differences between two HFpEF murine models. A: Representative intracellular Ca²⁺ signals in male (M) and female (F) control and HFpEF ventricular myocytes paced at 1 Hz. B: Representative caffeine-induced Ca²⁺ transients (Caff Ts) to assess sarcoplasmic reticulum (SR) Ca²⁺ load. C: Intracellular Ca²⁺ transient (CaT) amplitude and CaT decay tau at 1 Hz pacing, and SR Ca²⁺ content. D: CaT amplitude and decay tau at 2 Hz pacing, and the decay slope of the Caff T. n(cells)/N(animals) for CaT at 1 Hz = 19/8 for control male, 19/7 for control female, 17/8 for db/db+Aldo male, 19/7 for db/db+Aldo female, 20/8 for HFD+L-NAME male, and 51/8 for HFD+L-NAME female. n(cells)/N(animals) for CaT at 2 Hz = 19/8 for control male, 19/7 for control female, 15/8 for db/db+Aldo male, 14/5 for db/db+Aldo female, 20/8 for HFD+L-NAME male, and 40/7 for HFD+L-NAME female. n(cells)/N(animals) for Caff T = 19/8 for control male, 19/7 for control female, 17/8 for db/db+Aldo male, 19/7 for db/db+Aldo female, 15/8 for HFD+L-NAME male, and 36/7 for HFD+L-NAME female. Welch ANOVA with Dunnett’s T3 multiple comparisons test. Values reported are means ± SD.

There was no statistical difference in diastolic Ca²⁺ spark rate, assessed during a 1-min period after 30 paced beats, in the two HFpEF models; however, there was a trend for more Ca²⁺ sparks in male db/db+Aldo and in female HFD+L-NAME cardiomyocytes (Fig. 11). There were also more cells that showed spontaneous Ca²⁺ waves in female HFD+L-NAME; however, the Ca²⁺ wave frequency was low and there was only a trend for an increase in HFD+L-NAME (Fig. 11).

Figure 11.

Spontaneous sarcoplasmic Ca²⁺ release in HFpEF cardiomyocytes. A: Representative Ca²⁺ spark images in male (M) and female (F) control and HFpEF ventricular myocytes. B: Frequency of diastolic Ca²⁺ sparks and spontaneous Ca²⁺ waves in HFpEF. n(cells)/N(animals) for Ca²⁺ sparks = 19/8 for control male, 13/7 for control female, 11/7 for db/db+Aldo male, 12/5 for db/db+Aldo female, 17/8 for HFD+L-NAME male, and 36/7 for HFD+L-NAME female. For Ca²⁺ sparks rate, Kruskal-Wallis ANOVA with Dunn’s multiple comparisons test, and for Ca²⁺ wave occurrence, Chi-square test with Yates correction were used. Values reported are means ± SD.

Distinct transcriptome profiles in the two HFpEF models

Ion channels and Ca²⁺ handling proteins are importantly regulated at the transcription level, and distinct cardiac transcriptional profiles have been reported in human HFpEF sub-phenogroups and in HFrEF (32). We performed bulk RNA-sequencing in whole-heart lysates in the two HFpEF models and their respective controls. Principal component analysis showed distinct clustering of experimental animals in the two HFpEF models and by sex (Fig. 12A). In general, gene expression changes were more pronounced in the db/db+Aldo (larger changes in females) vs. HFD+L-NAME (larger changes in males), mirroring in vivo and cellular functional impairments. There were 3315 differentially expressed genes (DEGs) in the two HFpEF models overall (Fig. 12, B and C) with more DEGs found in the db/db+Aldo model owing to higher sample power and lower inter-sample variability. Only 151 DEGs were shared in the two HFpEF models (Fig. 12B). Among these 151 DEGs, 109 changed in the same direction (75 upregulated and 34 downregulated in both HFpEF models) and 42 genes changed in the opposite directions in the two HFpEF models (Fig. 12B). Analysis of Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) pathways revealed that genes associated with glucose and carbohydrate homeostasis and insulin resistance (e.g., Slc2a4 that encodes the glucose transporter type 4, GLUT4) were downregulated in both HFpEF models, while genes associated with lipid and carboxylic acid metabolism and peroxisome proliferator–activated receptor (PPAR) signaling were upregulated in both HFpEF models (Fig. 13). Periostin (Postn), an activated fibroblast marker and key regulator (33), was also upregulated in both male and female db/db+Aldo and in male HFD+L-NAME but not in female HFD+L-NAME. Interestingly, genes associated with SR compartment and intracellular transport were upregulated in db/db+Aldo but downregulated in HFD+L-NAME hearts (Fig. 13). We then analyzed the expression of key ion channel and Ca²⁺ handling genes in HFpEF. The L-type Ca²⁺ channel α_1C and β₂ subunits, Cacna1c (Ca_V1.2) and Cacnb2 were both downregulated and a critical regulator Rrad (inhibitory small G-protein Rad) was upregulated in db/db+Aldo (Fig. 12D), in line with reduced I_Ca,L (Fig. 8C). The I_to subunit Kcnip2 (KChIP2) was downregulated in both HFpEF models (Fig. 12D), in line with I_to reduction (Fig. 7C). The SR Ca²⁺ ATPase Atp2a2 (SERCA2) was downregulated in db/db+Aldo (Fig. 13D) in line with slower SR Ca²⁺ reuptake in db/db+Aldo myocytes (Fig. 10D). Additional Ca²⁺ handling proteins, calsequestrin 2 (Casq2) and calmodulin 3 (Calm3) genes were upregulated, whereas junctophilin 2 (Jph2) and striated muscle preferentially expressed protein kinase (Speg) were downregulated in db/db+Aldo (Fig. 12D). Kcna5 (K_V1.5) was downregulated in male HFD+L-NAME but upregulated in female HFD+L-NAME hearts mirroring the sex-dependent changes in I_K,slow in HFD+L-NAME (Fig. 7C). These gene expression changes are in line with impairments in cardiomyocyte ECC and electrophysiology, and highlight marked transcriptomic differences between the two cardiometabolic HFpEF models providing mechanistic basis for our functional findings.

Figure 12.

Distinct transcriptome profiles in two HFpEF murine models. A: Principal component (PC) analysis of whole-heart transcriptomes reveals distinct clusters in vehicle-treated wild-type sham control mice (WT+Vehicle, n=7), normal chow control (Norm. chow, n=4) and in HFpEF models: db/db mice with chronic aldosterone infusion (db/db+Aldo, n=8), and high-fat diet with L-NAME treatment (HFD+L-NAME, n=4). B: Venn diagram of differentially expressed genes (5% false discovery rate threshold) for the upregulated and downregulated genes in db/db+Aldo and HFD+L-NAME versus their respective controls. C: Normalized gene expression of all genes differentially expressed in at least one HFpEF model color-coded by condition and sex. Genes are clustered by hierarchical clustering using Euclidean distance. D: Normalized gene expression for select differentially expressed genes encoding Ca²⁺ handling proteins, ion channels, and their auxiliary proteins in HFpEF models versus their respective controls.

Figure 13.

Downregulated and upregulated biological processes in the two HFpEF murine models. Enrichment of Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways and Gene Ontology (GO) biological processes based on genes downregulated or upregulated in db/db mice with chronic aldosterone infusion (db/db+Aldo), and high-fat diet with L-NAME treatment (HFD+L-NAME). Color of dot represents Benjamini-Hochberg–corrected P value. Size of dot is proportional to the number of differentially expressed genes in that category.

DISCUSSION

Preclinical animal models of HFpEF have been extensively used to study disease mechanisms (4, 13–15). Initial one-hit models of HFpEF only mimicked one critical aspect of HFpEF pathophysiology but these models frequently fall short of capturing the multiorgan HFpEF phenotype, lead to mild congestive heart failure, and frequently transition to HFrEF, which is rarely seen in human HFpEF (4, 13). Multi-hit models that combined hemodynamic and metabolic stress could mimic several cardiac and extracardiac effects of HFpEF. Both small and large animal multi-hit HFpEF models have been developed, including HFD+L-NAME mice (17), obese ZSF-1 rats with pulmonary hypertension induced by SU5416, an inhibitor of vascular endothelial growth factor receptor (34), western diet-fed pigs with aortic banding (35) or excess mineralocorticoid, 11-deoxycorticosterone acetate (36). Among these models, the HFD+L-NAME mouse model has become popular for translational investigations in cardiometabolic HFpEF. The HFD+L-NAME model also increases inducible nitric oxide synthase (iNOS; that is constitutively active) causing reduced expression of an unfolded protein response effector XBP1, leading to dysfunction. However, the phenotype is less severe in HFD+L-NAME females (21), contrasting with human patients where women comprise the majority of HFpEF cases (2). Our recent db/db+Aldo HFpEF model complements these multi-hit models by exhibiting an overt obese-diabetic phenotype (Fig. 1) with severe diastolic dysfunction (Fig. 2), and markedly elevated plasma BNP levels (18). Notably, db/db+Aldo mice recapitulate key sex-differences in human HFpEF, including worse diastolic dysfunction in females (Fig. 2) but higher arrhythmia susceptibility in males (20). Here we quantitatively compare cardiomyocyte electrophysiology, ECC, and gene transcription in these two contemporary murine models of cardiometabolic HFpEF (Figs. 3–13). Such mechanistic knowledge is timely and relevant as the dominant clinical phenotype of HFpEF patients has shifted from the classic predominantly hypertensive disease to a severely obese, cardiometabolic, multiorgan disease (1). Recent deep phenotyping studies combining clinical data with cardiac omics data (transcriptomics, proteomics, metabolomics) and laboratory measurements identified distinct HFpEF patient subgroups (32, 37, 38). These studies showed that predominantly diabetic and obese HFpEF patients have markedly increased mortality risk and exhibit a distinct omics profile from other HFpEF subgroups and that is also different from obese or diabetic controls without HFpEF.

Arrhythmia risk is increased in patients with HFpEF, particularly in DM (12). Ion channel remodeling, Ca²⁺ handling impairments, inflammation, and fibrosis can all contribute to arrhythmias in HFpEF. Increased susceptibility to ventricular arrhythmias, QT and APD prolongation, reduced I_to and I_K1, and elevated I_Ca,L have previously been reported in Dahl salt-sensitive HFpEF male rats fed a high-salt (8% NaCl) diet (39, 40). Arrhythmogenic AP remodeling, including APD prolongation, increased APD-STV, and higher susceptibility for alternans was found in db/db+Aldo (in both sexes), and in male HFD+L-NAME, whereas female sex was protective against cardiomyocyte proarrhythmogenic changes in HFD+L-NAME (Figs. 3 and 4) in line with the less severe diastolic dysfunction (E/e’, Fig. 2).

Downregulation of I_to and I_K1 has been frequently reported in cardiac hypertrophy, HF, and DM (6, 41–43). I_to and I_K1 densities were both significantly reduced in db/db+Aldo (in both sexes), and trends for slight reductions in these currents were seen in male HFD+L-NAME, whereas they were unaltered in female HFD+L-NAME (Figs. 6 and 7). Net I_K1 and I_to magnitudes were only reduced in db/db+Aldo but not in male HFD+L-NAME, indicating that the K⁺ channel functional expression was not significantly reduced but it did not keep up with cellular hypertrophy in HFD+L-NAME (Figs. 5–7). In line with this, the I_to subunit Kcnip2 was downregulated in both murine HFpEF hearts (Fig. 12) and also in human HFpEF ventricles (32). Kcnj12 (K_ir2.2) and Kcnj4 (K_ir.2.3) but not Kcnj2 (K_ir2.1) channels are downregulated in human HFpEF, especially in a HFpEF subgroup characterized by female sex, high BMI, and stimulated immune pathways (32), which may be better mirrored by the db/db+Aldo mouse model. In contrast with K⁺ currents, both net I_Ca,L magnitude and I_Ca,L density were significantly increased in male HFD+L-NAME (Fig. 8). However, I_Ca,L density but not I_Ca,L magnitude was reduced in male db/db+Aldo, indicating that the functional expression of I_Ca,L did not follow cellular hypertrophy in male db/db+Aldo (Figs. 5 and 8). Cacna1c (Ca_V1.2) expression was unchanged human HFpEF, but in the obese female HFpEF subgroup a trend for reduced expression was seen (32). Moreover, expression of Rrad (Rad), an I_Ca,L inhibitory small G protein, was markedly increased both in human HFpEF (32) and db/db+Aldo mice (Fig. 12), which could contribute to reduced I_Ca,L. I_Na,L, which is a tiny current in healthy cardiomyocytes, was markedly increased in male and female db/db+Aldo, and in male HFD+L-NAME; but it was unchanged in female HFD+L-NAME both at the level of total current magnitude and density (Fig. 9). The expression of Scn4b (β₄ subunit of Na⁺ channels) was increased in both our mouse HFpEF models (Fig. 12) and in human HFpEF (32), and Scn4b has been associated with an increased I_Na,L (44). Thus, a shift in balance between depolarizing and repolarizing currents underlies AP remodeling in both HFpEF models. However, downregulation of repolarizing K⁺ currents dominated in db/db+Aldo, whereas the increased depolarizing currents (I_Ca,L and I_Na,L) caused arrhythmogenic AP changes in HFD+L-NAME male mice. These important results should be considered in future antiarrhythmic approaches in specific animal models representing different HFpEF sub-phenogroups.

ECC has been studied previously in HFpEF mice with conflicting results. Studies reported that CaT amplitude was either unchanged (45) or increased (46, 47), and CaT decay rate was either slower (46) or faster (45, 47) in HFD+L-NAME myocytes. We found that CaT amplitude and CaT decay time were both unchanged in HFD+L-NAME myocytes at 1 Hz and 2 Hz pacing (Fig. 10). These differences in the reported outcomes may be explained by the variations in the HFD+L-NAME model used, including different mouse strains (most frequently C57BL/6N and C57BL/6J, while the latter may exhibit a less severe HFpEF phenotype), duration of treatment (5–15 weeks), dose of L-NAME (0.5–1 g/L), age of animals (ranges from 6–8-wk to 2-yr-old mice at the start of the treatment), and presence of additional stressors (e.g., angiotensin II or high fructose diet). However, we found significantly increased I_Ca,L (Fig. 8) and reduced NCX function (Fig. 10) in male HFD+L-NAME myocytes.

Cardiomyocytes isolated from primarily hypertensive HFpEF models (e.g., Dahl salt-sensitive rats) showed stimulated baseline Ca²⁺ handling with increased diastolic and peak [Ca²⁺], increased I_Ca,L, and unchanged SR Ca²⁺ reuptake (40). Increased mechanical afterload has been shown to increase I_Ca,L and CaT amplitude, and not alter or slightly increased SR Ca²⁺ uptake in cardiomyocytes embedded in viscoelastic hydrogel (cell-in-gel system (48, 49)) and in early (5 days after) transverse aortic constriction (TAC) (50), both dependent on CaMKII activation. In contrast with primarily hypertensive heart disease models, the CaT amplitude is unchanged (or slightly reduced at high pacing rates), and CaT decay is prolonged in diabetic HFpEF models, e.g., in ZSF1 rats (51), in line with our data in db/db+Aldo mice (Fig. 10). In fact, SERCA2 gene (Atp2a2) was downregulated in db/db+Aldo (Fig. 12) and in human HFpEF, although less so than in HFrEF (32). Moreover, diabetic hyperglycemia can prolong CaT by the enhanced inhibitory effect of the O-GlcNAc-modified phospholamban on SERCA2 (23). Diabetic hyperglycemia can also induce CaMKII activation via O-GlcNAcylation and enhance diastolic Ca²⁺ sparks (23). In line with this, Ca²⁺ spark rate was markedly increased in myocytes from a modified HFD+L-NAME model combined with a high fructose diet (47). Interestingly, Ca²⁺ spark rate was unchanged in diabetic db/db+Aldo myocytes (Fig. 11). Of note, we studied myocytes bathed in Tyrode’s solution with normal glucose concentration, and this could partially reverse the high glucose-induced activation of CaMKII with consequent RyR2 phosphorylation and Ca²⁺ leak as previously demonstrated (23, 52). Ca²⁺ spark frequency was unaltered also in HFD+L-NAME here and in a previous study (45); however, we found an increased spontaneous Ca²⁺ wave occurrence in HFD+L-NAME, ever more so in female HFpEF myocytes (Fig. 11). Thus, differential therapeutic approaches, e.g., increasing SR Ca²⁺ reuptake vs. inhibiting RyR2 leak merit further investigations in select HFpEF sub-phenogroups and may be different between sexes.

Diastolic dysfunction is mediated by multiple mechanisms in HFpEF, including mechanical load effects, altered relaxation rate through impairments in Ca²⁺ handling and crossbridge detachment kinetics, and altered muscle stiffness from cardiac fibrosis, titin isoform expression changes, and posttranslational modifications of myofilament proteins, titin and microtubules (53). We recently reported that Ca²⁺ handling impairments contribute to diastolic dysfunction in male db/db+Aldo mice; whereas in female db/db+Aldo mice (in which CaT was unaltered) myofilament and titin changes and cardiac fibrosis, including increased expression of stiff N2B titin isoform and profibrotic marker periostin, titin PEVK phosphorylation, and reduced TnI phosphorylation, led to severe diastolic dysfunction (20). Increased fibrosis has been shown in obese-diabetic db/db mice (54) and in HFD+L-NAME (17, 55). Recent research showed that enhanced expression of Angiopoietin-like 4 (Angptl4) in HFD+L-NAME can be an important marker specific for activated fibroblasts in HFpEF (55), and Angptl4 was also upregulated in db/db+Aldo hearts (in addition to periostin) and multiple fibrosis genes (e.g., Col1a1, Col1a2, Col5a1, Col5a2, Col6a1, and Timp1). Increased tubulin detyrosination was also reported in diabetic-hypertensive-obese ZSF1 male rats, and inhibition of vasohibin carboxypeptidases that detyrosinate tubulin improved diastolic dysfunction (56). Impaired mitochondrial function is also associated with diastolic dysfunction, and reduced expression of Acsl6 (Acyl-CoA Synthetase Long Chain Family Member 6) correlated with diastolic dysfunction in female HFD+L-NAME mice (57). Another study also found decreased respiratory function, increased reactive oxygen species (ROS) production, and blunted mitophagy; however, these mechanisms were worse in male HFD+L-NAME hearts (58). Thus, future studies are required to compare mitochondrial dysfunction and redox homeostasis in the two HFpEF models and between sexes.

CONCLUSIONS

HFpEF animal models and mechanisms, including alterations in ECC, ionic currents, and transcriptomics, as we have done here, are heterogenous similar to the human HFpEF population. Combined use of translational multi-hit small and large animal models of HFpEF with different disease etiology and sex-differences could advance mechanistic understanding and reveal therapeutic targets in HFpEF sub-phenogroups and reveal if these targets are shared across different disease phenotypes in the heterogenous HFpEF population.

LIMITATIONS AND FUTURE DIRECTIONS

Diastolic Ca²⁺ handling impairments may contribute to diastolic dysfunction in HFpEF, but our measurements here could be made more quantitative by using ratiometric Ca²⁺ indicators and careful in situ calibration thereof. Subcellular distribution of Ca release units and synchrony of Ca release may also change in HFpEF (40, 51). Myofilament alterations, inflammation, fibrosis, and metabolic changes could significantly contribute to HFpEF phenotype, and these mechanisms require further investigation. HFpEF susceptibility in animals should also be tested at various treatment durations and age, especially in post-menopause that represent the majority of human HFpEF patients.

DATA AVAILABILITY

Source data for this study are openly available at https://doi.org/10.5061/dryad.76hdr7t8r.