Ethical Approval

All animal experiments were approved by the Norwegian Food Safety Authority (FOTS IDs 14692 and 28736) and followed the Guide for the Care and Use of Laboratory Animals (NIH Publication No. 85-23, Revised 2011) (22). We also adhered to the Animal Research: Reporting In Vivo Experiments (ARRIVE) guidelines (23).

The R403Q HCM Mouse Model

The R403Q mouse model of HCM has been previously described (3, 20, 21). In brief, we used male mice heterozygous for the R403Q mutation on α-myosin heavy chain 6 (αMHC^R403Q/+) and wild-type (WT) littermate controls (αMHC^+/+) on the 129S6/SvEvTac background. Female mice were not used. In the initial description of the model (21), females did not develop a consistent phenotype, and the induction of left ventricular hypertrophy (LVH) by CsA has not been described for female R403Q mice. The following references to mice in this article, therefore, refer to male mice. The HCM phenotype induction in R403Q mice, but not in WT controls, is well described (3, 20). In this article, we denote R403Q mice who received CsA as “HCM mice” to be separated from “R403Q mice” who were left unexposed to CsA. Hence, all mice in this study were genotype positive for HCM, except WTs.

We administered the CsA orally through the feed, which was served ad libitum. CsA-mixed feed was prepared by soaking 40 g of standard mouse chow in a mix of 10 mL of sesame oil (ACROS Organics) and 0.025 mL of CsA (100 mg/mL Sandimmun Neoral Mixture, Novartis) for 24 h. Controls received 40 g of chow soaked in 10 mL of sesame oil for 24 h. Feed was replaced once per week for 3 wk.

Data in this manuscript are from four mouse cohorts. Cohort 1 (n = 24 mice: 12 WTs and 12 R403Q) was used to show that we could induce the HCM phenotype in R403Q mice but not in WTs (Supplemental Fig. S1; all Supplemental material is available at https://doi.org/10.5281/zenodo.7760177). Cohorts 2–4 were used in exercise experiments: cohort 2 had n = 22 mice: 11 exercised R403Q and 11 sedentary R403Q; cohort 3 had n = 24 mice: 12 exercised R403Q and 12 sedentary R403Q; cohort 4 had n = 10 mice: 5 exercised WTs and 5 sedentary WTs. In total, 80 mice were used for this work. We have noted in each figure legend which cohort the data were collected from.

Treadmill Exercise Protocol

Mice, at 6 wk old, were adapted to the treadmill (40 cm long, 6-lane treadmill with 25% incline; Columbus Instruments) 10 min/day for 3 days before the start of the high-intensity interval training (HIIT) protocol. Each HIIT session started with a 10-min warm up with incrementally increasing speed, followed by five intervals consisting of 8 min at high intensity (starting at 90% of max speed achieved during the baseline exercise test) and 2 min of active rest (60% of baseline exercise test max speed), resulting in a total of 1 h of exercise per session. The mice exercised 6 days/wk. After every third exercise day, the high intensity and active rest speeds were increased by 0.6 and 0.4 m/min, respectively.

The treadmill exercise protocol lasted for a total of 6 wk (Fig. 1A). Induction of the HCM phenotype was initiated after 3 wk in both exercised and sedentary R403Q mice (n = 23/group) and continued for the last 3 wk of the protocol. This protocol was designed to mimic physical activity initiated before the development of a hypertrophic phenotype and continued during early disease progression. We performed cardiac imaging [echocardiography or magnetic resonance imaging (MRI)] and an exercise test before the start of the protocol, after 3 wk (before starting phenotype induction), and at the end of the 6-wk protocol.

Echocardiography

We performed echocardiography for the characterization of LV function and morphology before and after any experimental interventions (i.e., phenotype induction and activity). An experienced mouse echocardiographer blinded regarding group performed all examinations, as previously described (24). Anesthesia was induced with 3% and maintained with 1.7% isoflurane with O₂ as the carrier gas. Hair was removed from the thorax before the mouse was placed on a heating pad to maintain body temperature throughout the scanning. Transthoracic B-mode and M-mode examinations were conducted using a Vevo 3100 system (FUJIFILM VisualSonics, Toronto, ON, Canada). We collected aortic valve, LV long and short axis B-mode recordings, mid-ventricular and left atrial M-mode recordings. Tissue Doppler recordings in circumferential LV yielded e′, whereas Doppler recordings of mitral inflow provided E, A, and E deceleration time, as well as pulmonary valve peak velocity, and aortic valve velocity time integral (AoV VTI). Fractional shortening was calculated as (LVID;d – LVID;s)/LVID;d, i.e., the relative change of the LV inner diameter from diastole to systole. Images were analyzed with the VevoLab software (FUJIFILM VisualSonics, Toronto, ON, Canada), by an investigator blinded with respect to group.

Magnetic Resonance Imaging

Researchers blinded with regard to group performed cardiac MRI with a 9.4-T horizontal 21-cm bore size scanner (Agilent Technologies) with a 35 mm quadrature coil (Rapid Biomedical, GE). Mice were anesthetized using isoflurane (maintenance concentration 1.5–2.5%), with O₂ as a carrier gas). ViscoTears (Théa) was applied to avoid corneal drying. Body temperature, breathing rate, and ECG signals were continuously monitored with a small animal monitoring and gating system (Model 1030, SA Instruments). Respiration and ECG signal gating was used for cine and tissue phase mapping (TPM) recordings. Body temperature was maintained around 37°C using hot air that was automatically adjusted based on body temperature. A stack of cine short-axis slices (parameters in Table 1) covering the LV was collected using compressed sensing as previously described (25, 26). Black-blood TPM recordings (5 short-axis slices of the left ventricle) and phase contrast MRI (PC-MRI) of mitral valve blood flow (3 short-axis slices) were collected (27), as well as three to six slices of ECG-gated T1 mapping short-axis recordings of the LV. Total duration of the examination was ∼1.5 h and the recording parameters are listed in Table 1. Cardiac morphology (LV mass) and global function [LV end diastolic volume (EDV), cardiac output (CO), and stroke volume (SV)] were calculated from cine recordings using the freely available software Segment (28). In-house developed MATLAB codes were used for analysis of TPM recordings of myocardial function (SRe: peak circumferential strain rate) (29), T1 recordings, and PC-MRI of blood flow (E).

Characterization of Arrhythmia Propensity in Isolated Hearts

Eight mice from each group (HIIT and sedentary HCM mice) were anesthetized using 4.5% isoflurane (Abbott) with O₂ as the carrier gas. During anesthesia, each mouse received 100 IU heparin by intraperitoneal (ip) injection. After 5 min the heart was excised and quickly placed in ice cold HEPES Tyrodes (HT) buffer containing (in mM): HEPES 5, NaCl 140, KCl 5.4, MgCl₂ 0.5, glucose anhydrate 5.5, NaH₂PO₄·H₂O 0.4, CaCl₂ 1, at pH 7.4. The heart was then mounted on a Langendorff apparatus by cannulation of the aorta, secured with a nylon suture, and retrogradely perfused for 10 min with 37°C, oxygenated HT solution at a constant pressure of 70 mmHg. Following the 10-min stabilization period, the hearts were burst paced at 8 Hz with platinum electrodes for 20 s every minute for 5 min. During the last 3 min, β-adrenoceptor stimulation with 100 nM isoprenaline (NAF, Norway) was added. ECGs were recorded with electrodes from a telemetry transmitter (Data Sciences International, Minneapolis, MN) placed on the base and apex of the LV side of the heart. At the end of the protocol, the hearts were cut down, and the LV quickly snap frozen in liquid nitrogen. Premature ventricular complexes (PVCs) and episodes of ventricular tachycardia (VT) were counted manually from ECG recordings using the Ponemah software (Data Sciences International, Minneapolis, MN). PVCs were defined as premature beats with a wider QRS complex, compared with regular sinus beats. VTs were defined as ≥4 consecutive ventricular beats according to the Lambeth convention (30). The investigator was blinded with respect to group during experiments and arrhythmia analysis.

Gene Expression Analyses

Total RNA was isolated from snap-frozen LVs using the RNeasy Mini Kit (Cat. No. 74106, Qiagen Nordic, Norway) according to the manufacturer’s protocol. Reverse transcription and cDNA generation from isolated RNA was performed using iScript cDNA Synthesis Kit (Cat. No. 1708891, Bio-Rad). Droplet digital PCR (ddPCR) was performed using QX200 AutoDG Droplet Digital PCR system (Bio-Rad) according to the manufacturer’s protocol. Data were analyzed using the QuantaSoft Software (v.1.7.4; Bio-Rad). The full list of TaqMan Gene Expression Assays used for ddPCR is listed in Supplemental Table S1. Gene expression was normalized to the expression of 60S ribosomal protein L32 (Rpl32). The investigator was blinded with respect to group for gene expression analysis.

Amplification and detection of relevant inflammatory genes was performed on a custom-order qPCR array (Thermo Fisher Scientific), using the QuantStudio 3 Real-Time PCR System (Applied Biosystems, CA). The list of genes measured by the array is presented in Supplemental Table S2.

Collagen Abundance

Total collagen was determined by hydroxyproline content [collagen (µg)/wet weight (mg)] measured by high-performance liquid chromatography (HPLC) (n = 11 or 12/group) as previously described (31), or Masson’s trichrome (Polysciences, Warrington, PA, Cat. No. 25088-1) staining of short axis, mid ventricular LV cryosections (n = 11/group) with a protocol modified for cryosections: seven µm thick sections were fixed in 10% formalin (NBF) (1 h, RT), followed by fixing in Bouin’s solution (1 h, 35°C), before gently washing in running tap water (5 min) to remove the picric acid. Sections were then stained in Weiget’s Iron Hematoxylin Working Solution (1:1 mix of Weigert’s hematoxylin A and Weigert’s hematoxylin B) (10 min), washed in running tap water (5 min) and rinsed in distilled water (20 quick dips). Sections were then stained with in Biebrich scarlet-acid fuchsin solution (5 min), rinsed in distilled water (3 changes), and transferred to phosphotungstic/phosphomolybdic acid (12 min). Slides were drained and transferred to Aniline Blue (5 min), rinsed in distilled water (5 changes), and transferred to 1% acetic acid (3 min) and rinsed in distilled water. Finally, sections were dehydrated in 95 and 100% ethanol (2 min each), cleared in Histo-Clear (2 min) and mounted with Vectamount (xylene-free, nonaqueous). Whole heart short axis section images were captured on Axioscan Z1 (Carl Zeiss, Germany) with a ×20 objective. Fibrosis was quantified using the open access QuPath software (v.0.3.1, University of Edinburgh, UK) with a trained random tree pixel classifier (32) and is reported as percent area of LV that was covered by fibrosis in each section. The operator was blinded with regard to group identity.

Lysate Preparation for Whole Tissue Lysate

Total protein lysates were prepared by adding 30 mg of frozen LV tissue in 700 µL PH lysis buffer (PBS with 1% Triton and 0.1% Tween20) supplemented with one cOmplete, Mini, EDTA-free Protease Inhibitor Cocktail tablet (Sigma-Aldrich) and one PhosSTOP phosphatase inhibitor tablet (Sigma-Aldrich) per 10 mL of buffer. The tissue was homogenized using a metal bead and a tissue lyser (30 Hz, 100 s), before transferring the supernatant to fresh tubes and placing them on ice (30 min). Tubes were then centrifuged (14,000 rpm, 10 min, at 4°C), before removing and aliquoting the supernatant. Aliquots were stored at −80°C. The investigator was blinded to group identity for lysate preparation and subsequent experiments and analyses.

Compartmental Protein Extraction

Cytoplasmic and nuclear protein fractions were extracted using a compartmental protein extraction kit (Millipore) according to manufacturer’s instructions, except using 30 mg of tissue in 250 µL of buffer C, and adding 60 µL each of buffers N and M.

Immunoblotting

Protein concentrations were measured using the Pierce Micro BCA Protein Assay Kit (23235, Thermo Scientific) according to the manufacturer’s instructions. Lysates and a 0.2 mg/mL of bovine serum albumin standard were thawed on ice. Albumin standards were diluted to 10, 20, 30, 40, and 60%, and ran in triplets, while samples were diluted to 1% and ran in duplicates. Working reagent was prepared by mixing 50:48:2 parts of reagents A:B:C in a reagent reservoir. Standard (100 µL) and sample (100 µL) were pipetted into a 96-well microplate, before working reagent (100 µL) was added to each well using a multichannel pipette. The plate was covered with sealing film, incubated (1 h, 60°C) before absorbance was measured at 562 nm using a Hidex plate reader. For immunoblotting, protein (30 µg) was loaded onto 4–15% gradient Tris·HCl gels and separated by SDS-PAGE. Proteins were transferred to a PVDF membrane using a Trans-Blot Turbo system (Bio-Rad) and blocked in 5% nonfat dry milk in Tris-buffered saline with 0.1% Tween 20 (TBST) for 1 h at room temperature. Membranes were incubated with anti-fibronectin (ab2413, Abcam) and anti-GAPDH (sc-32233, Santa Cruz Biotechnology) overnight (4°C, in 5% nonfat milk TBST) and washed (1 × 15 min and 2 × 5 min in TBST), before incubation with secondary antibody anti-rabbit IgG-HRP (GE Healthcare, Oslo, Norway). Washing was repeated, and blots were then developed using enhanced chemiluminescence (ECL plus; GE Healthcare) and imaged with Azure C600. The images were captured and quantified with AzureSpot software (Azure Biosystems).

Statistical Analyses

Statistical analysis was performed in GraphPad Prism 9 (GraphPad). Student’s t test was used for all single comparisons, with Welch’s correction when groups displayed significantly different variances. If distributions failed normality tests, the Mann–Whitney test was used. For comparisons with multiple time points, we used two-way ANOVA or mixed model if any data points were missing. Data are presented as means ± SD. Initial sample size calculations suggested the need for n = 8–10. Because of larger SDs than expected, post hoc power calculations of tests that reached significance varied between 57 and 82%.

Exercise Training Increased Exercise Capacity in HCM Mice

As expected from previous studies, HCM mice had a clear phenotype induced by CsA (Supplemental Fig. S1), which allowed us to study the effects of exercise training initiated before the development of the hypertrophic phenotype. The exercise training protocol started 3 wk before CsA administration was initiated and continued for 3 more wk in parallel with CsA exposure (Fig. 1A). Exercise capacity, measured as running distance during the test and speed at V̇o_2peak, was increased in the exercised group compared with sedentary mice at both the 3- and 6-wk time points (Fig. 1, B and C). V̇o_2peak was not different in the exercised and sedentary HCM mice (Fig. 1D). During the 6-wk protocol, body weight increased in both exercised and sedentary HCM mice, but the total weight gain was less in exercised compared with sedentary mice (Fig. 1E). Increase in exercise capacity was similar to WT mice undergoing the same protocol (Supplemental Table S3).

Exercise Training Attenuated Left Atrial Dilatation in HCM Mice

Dilatation of the left atrium (LA) was attenuated in the exercised HCM mice, as documented by MRI and echocardiography (examples, Fig. 2, A, B, and E, and Supplemental Fig. S2). Indeed, the LA diameter (Fig. 2C) and area (Fig. 2D) were smaller in exercised HCM mice at the 6-wk time point, measured by MRI. By echocardiography, the LA diameter after 6 wk of exercise training was comparable with that in sedentary mice at the 3-wk time point, i.e., before phenotype induction (Fig. 2F, and Supplemental Table S4). We did not observe a difference in fractional shortening, LV posterior wall thickness, LV inner diameter, LV septum thickness, E/e′, E/A, or heart rate between exercised and sedentary HCM mice, but there was a trend toward a lower E deceleration time in the exercised group (Supplemental Table S4).

Using MRI, we observed no difference in E/SRe, but there was a trend toward higher E in exercised HCM mice at the 6-wk time point. We found no difference in LV, EDV, or stroke volume between sedentary and exercised mice (Supplemental Table S4).

Lower Expression of Extracellular Matrix Genes in Exercised HCM Mice

The effect of exercise training on the development of fibrosis was assessed by gene expression and protein abundance. We found a lower expression of collagen 1α1 (Col1a1) and -2 (Col1a2) and collagen 3α1 (Col3a1) in exercised compared with sedentary HCM mice, suggesting reduced production of structural collagens in response to exercise initiated before HCM induction (Fig. 3A). In addition, lysyl oxidase (Lox) and fibronectin 1 (Fn1) also had lower expression in exercised HCM mice, suggesting less collagen cross linking and altered organization (Fig. 3A). Exercise did not alter the expression level of these genes in WTs (Supplemental Table S5). The lower expression of Fn1 was accompanied by a trend toward less fibronectin protein measured by immunoblotting (P = 0.10) (Fig. 3B). However, when analyzing LV collagen protein content, we did not observe differences, neither when assessed by Masson’s trichrome staining of midventricular sections (Fig. 3C) nor when measured by HPLC in pulverized LVs (Fig. 3D). Furthermore, global native T1 values were similar for LVs in sedentary and exercised mice (Fig. 3E). Of note, T1 values were higher in the hypertrophic regions of the LV than in the nonhypertrophic region, irrespective of group (not shown), but the ratio of T1 values in the hypertrophic to nonhypertrophic regions increased more in sedentary mice than in exercised mice (Fig. 3F).

Effects of Exercise Training on Expression of Proinflammatory and Damage-Associated Molecular Patterns-Related Genes in HCM Mice

We hypothesized that the lower expression of collagen mRNA after exercise training was due to reduced inflammation. Therefore, we explored the expression levels of a selection of 96 inflammation-related genes. We detected expression of 85 genes (Supplemental Table S2), of which 19 were associated with damage-associated molecular patterns (DAMPs) (Fig. 4). Of these, we found lower expression of genes encoding biglycan (Bgn, 29% lower, P = 0.0416), versican (Vcan, 22% lower, P = 0.0341), and fibronectin 1 (Fn1, 45% lower, P = 0.0150), in the exercised HCM group compared with sedentary mice, and a trend toward lower expression of tenacin C (49% lower, P = 0.0924) and syndecan 1 (26% lower, P = 0.0772) (Fig. 4). Furthermore, there was also a trend toward lower fibromodulin (55% lower, P = 0.0708) expression in exercised HCM mice. Somewhat surprisingly, we found no significant differences in the expression of the key inflammatory mediators transforming growth factor-β1 (Tgfb1), interleukin-1β (Il1b), interleukin-18 (Il18), or tumor necrosis factor-α (Tnfa) (Supplemental Fig. S3).

Exercise Leads to Phosphorylation of Akt and p38 in Cytoplasmic Fraction of HCM LV Tissue

Having found lower expression of profibrotic genes in exercised HCM mice, we performed immunoblotting for key intracellular signaling proteins. We observed a higher degree of phosphorylation of both protein kinase B (Akt) at Ser473, glycogen synthase kinase-3β (GSK-3b), and (p38 mitogen-activated protein kinase) in exercised HCM mice (Fig. 5), but found no differences phosphatidylinositol 3-kinase (PI3K), mammalian target of rapamycin (mTOR), or Akt at Thr308 between exercised and sedentary HCM mice (Fig. 5). In addition, there was no difference in phosphorylation of mothers against decapentaplegic homologs 2 and 3 (SMAD2/3), nuclear factor-κβ (NF-κβ), c-Jun kinase (JNK) or extracellular signal-regulated protein kinases 1 and 2 (ERK1/2) (Supplemental Table S6).

Effects of Exercise Training on Arrhythmia Propensity in Hearts from HCM Mice

Fibrosis is closely associated with risk of arrhythmias in HCM (33). Therefore, we explored the propensity for arrhythmias in exercised and sedentary HCM mice. We did not observe a statistically significant increase in the incidence of arrhythmias after pacing and isoprenaline provocation in isolated hearts from exercised HCM mice, although three of eight hearts from exercised mice presented a single episode of VT each after burst pacing, compared with zero of eight in the sedentary group (Supplemental Fig. S4). All mice completed the 6-wk exercise protocol without adverse events.

Since other mechanisms are probably more important for arrhythmias in mice than fibrosis (34), we also quantified mRNA expression of key ion channels and calcium handling proteins. No differences were observed between exercised and sedentary mice (Supplemental Fig. S5).

Exercise Training Attenuates Profibrotic Signaling in HCM Mice

Our observation of lower expression of Col1a1, Col1a2, Col3a1, Fn1, and Lox in exercised HCM mice suggests an attenuation of fibrosis development. We did not expect altered collagen abundance within the relatively short time frame of the exercise training protocol for our HCM mice. However, we did observe a trend toward less fibronectin, which is an important molecule in the initial phases of collagen assembly (35). Based on this, we speculate that exercise training might affect the structure of the ECM both through production and cross linking. In addition, we found a lower increase in the ratio of native T1 signal between hypertrophic and nonhypertrophic regions in the exercised mice, indicating regional differences that were too subtle to detect with other methods. It is possible that a different HIIT protocol could have resulted in a detectable difference in collagen amount.

Our findings are partly in line with results presented by Wang et al. (19), who saw that exercise of a light-chain HCM mouse model led to reduction of transforming growth factor (TGF)-β receptor and other proinflammatory and profibrotic genes and signaling pathways. Furthermore, Hong et al. (15) showed that exercise could suppress LOX2/TGFβ-mediated fibrosis signaling in an aging rat model of hypertension, whereas Kwak et al. (16) showed that exercise could ameliorate age-induced fibrosis and ECM remodeling in rats. Unlike these studies, we did not observe a difference in the expression of key inflammatory genes, e.g., Tgfb1, Il1b, Il18, Tnfa, or actin-α2 (acta2) and only a trend toward less Il6, which suggests a different pathway for fibrosis development than an inflammatory activation of myofibroblasts.

Exercise is known to have an anti-inflammatory effect (36), but since we did not have evidence for Tgf-β-mediated inflammation, we performed a qPCR array of a broad set of inflammatory genes, including DAMPs. It has been hypothesized that alterations in the ratio between DAMPs and suppressing/inhibiting DAMPs (SAMPs) could be involved in the induction and resolvement of inflammation (37). In our study, almost all DAMPs with attenuated expression in exercised HCM mice originate from the ECM, and no difference in expression was seen in the major DAMP receptors, i.e., Toll-like receptor 2 and 4. Thus, our results indicate effects on ECM expression and structure rather than a clear anti-inflammatory response from exercise in the HCM mice.

Signaling Pathways Affected by Exercise Training in HCM Mice

Our analysis included the activation of signaling pathways known to be involved in fibrosis. Of note, we found increased phosphorylation of Akt and p38 in the exercised HCM mice. Akt can increase cellular metabolism and be involved in cardiovascular pathological processes (38). Akt signaling seems to be beneficial when activated in physiological conditions, whereas long-term overexpression can lead to cardiac dysfunction (39). Interestingly, Shiraishi et al. (40) showed that accumulation of Akt to the nucleus mediated inhibition of apoptosis, without hypertrophic remodeling, suggesting that the nuclear-translocated activated Akt is beneficial. We only observed increased phosphorylation on site Ser473, not Thr308. In addition, GSK3b, which is phosphorylated downstream of Akt (Ser473), had a higher degree of phosphorylation. Since PI3K was similar between groups, we investigated whether Akt could be regulated upstream by mTOR but saw no differences between exercised and sedentary HCM mice.

Exercise is known to activate p38 (41), but the interpretation of increased p38 phosphorylation in our study is not straightforward. It has been shown that attenuation of p38 activity in cardiomyocytes leads to hypertrophy (42, 43), whereas cardiac fibroblast-specific p38 activity promotes hypertrophy (44). Since we used the whole LV tissue, we cannot be certain which cells p38 activity was from, but since cardiac fibroblasts secrete p38 activity-dependent Il6, which we did not observe, we assume that our activated p38 did not arise from cardiac fibroblasts.

The induction of the HCM phenotype by CsA complicates the interpretation of some of the results in this study. CsA is an immunosuppressor and inhibits the calcineurin pathway and mitochondrial permeability transition pore. We have shown that CsA induces the hypertrophic phenotype in R403Q mice and not in WTs (see Supplemental Fig. S1). In addition, both exercised and sedentary HCM mice had similar levels of LV hypertrophy, which suggests that exercise did not solely affect the metabolism of CsA.

Functional Consequences of Exercise in HCM Mice

The exercise-trained HCM mice in our study increased their exercise capacity, as evidenced by higher running distance and running speed at V̇o_2peak compared with the sedentary mice. Attenuated LA dilatation is a strong argument for the beneficial effect of exercise on cardiac function, as LA dilatation is a sensitive and robust marker of cardiac dysfunction in mice (45). However, we cannot rule out peripheral effects, e.g., in skeletal muscle function or capillarization. Furthermore, quantifying exercise effects in mice are not straightforward due to the lack of universal testing protocols and criteria for exhaustion. In addition, energy metabolism is different between mice and humans and depends on external factors such as temperature and food intake and internal factors such as weight, tissue composition, and circadian rhythm (46, 47). We did not find clear indications of improvements in systolic or diastolic cardiac function, but these parameters are less established in mice and especially difficult to interpret in a setting of myocardial hypertrophy with supranormal fractional shortening. In a study by Ravassa et al., the authors investigated the effects of a potential drug against heart failure. They found that the drug caused a reduction in LA volume in patients with less collagen synthesis and cross linking (measured indirectly through serum proteins) (48). The drug did not cause any alterations in LV morphology (EDV or mass), suggesting that there might be mechanisms of LA dilatations that are dependent on collagen cross linking and organization (48). This is consistent with our findings of reduced gene expression of both collagen 1 and 3 but also fibronectin and lysyl oxidase that are involved in collagen cross linking and organization, even though these were measured in the LV tissue. It is also possible that exercise caused alterations to the LV morphology or function that we could not detect in this study, either because of the timing or sensitivity of our measurements, or that LA dilatation is easier to affect through exercise. Of course, we cannot rule out that the mechanisms through which exercise affects LV gene expression, and LA dilatation can be uncoupled, parallel processes that might not influence each other directly.

We did not observe clear effects on the propensity for arrhythmias in our experiments. Three hearts from exercised HCM mice had one episode of provoked VT each. In general, sudden cardiac death due to arrhythmias in mouse models is extremely rare, and our data cannot conclude that exercise had any effects on the propensity for arrhythmias.

Translational Perspectives

Although any extrapolations from our study to other models and humans should only be made with great caution, our findings add mechanistic insights to existing data that indicate beneficial effects of exercise in HCM. Cardiac stiffness is an important aspect of diastolic function, and the effects on ECM organization (rather than just abundance) indicated by our results might explain the positive associations between exercise and diastolic function in people with HCM. This could explain the favorable functional effects of exercise seen by Dejgaard et al. (10) and Sheik et al. (13) despite no differences in replacement fibrosis between athletes and nonathletes. The long-term effects of different exercise protocols in genotype-positive HCM need further study. In this regard, results from prospective studies on exercise in HCM (NCT03335332 and NCT02549664) are greatly anticipated.

Study Limitations

Since the use of CsA for phenotype induction in this model is well established only in male R403Q mice, the effect of exercise on female R403Q mice has not been investigated here.

Conclusion

Genotype-positive HCM mice that performed HIIT before and during the development of the HCM phenotype had attenuated LA dilatation and no difference in fibrosis despite lower expression of key ECM genes. Gene expression analysis indicated lower expression of ECM-located inflammatory mediators. These findings support the hypothesis that exercise training at an early age does not have negative effects on cardiac function, hypertrophy, or fibrosis in individuals with an HCM-causative genotype but show mild favorable adaptations.