Inhibition of activin A receptor signalling attenuates age-related pathological cardiac remodelling

Ageing is characterised by a progressive loss of physiological integrity leading to impaired organ function and increased risk of death. Disruption to tissue homeostasis is a major cause of diseases such as cardiovascular diseases (North and Sinclair, 2012), cancer (Niccoli and Partridge, 2012) and neurological disorders (Licher et al., 2019). Ageing is the consequence of the impairment of multiple interconnected pathways such as protein turnover, disregulation of signalling pathways, DNA repair and telomere attrition. These failing mechanisms lead to an accumulation of damage at the cellular level with an associated increase in oxidative stress and inflammation (López-Otín et al., 2013).

Cardiomyopathies such as heart failure are a major public health issue, with a worldwide prevalence of over 23 million (Ambrosy et al., 2014). Ageing significantly increases risk factors such as coronary artery disease (Jenča et al., 2021), hypertension (Hansson, 1998) and metabolic disorders such as obesity (Kenchaiah et al., 2002; Tune et al., 2017). These disorders are characterised by pathological remodelling leading to cardiac dysfunction, hypertrophy (Lazzeroni et al., 2016), inflammation (Murphy et al., 2020), fibrosis (Querejeta et al., 2004) and gene expression.

Although normal ageing is slow and progressive, there are a number of genetic mutations that significantly increase the rate of ageing, in a condition called progeria. The progeroid syndrome can be induced by mutations in the nuclear structural protein lamin A leading to Hutchinson–Gilford progeria syndrome (HGPS) (Ahmed et al., 2018). Defects in DNA repair and replication systems also result in progeria leading to Werner syndrome (WS) (Oshima et al., 1993) and Bloom syndrome (Arora et al., 2014). A defect in the nucleotide excision repair system due to a mutation in the Ercc1-Xpf endonuclease system leads to several human disorders such as Cockayne syndrome and Xeroderma pigmentosum (Karikkineth et al., 2017), Fanconi anaemia (Kashiyama et al., 2013) and cerebro-oculo-facio-skeletal syndrome (Jaspers et al., 2007). Cardiovascular diseases are the main cause of death in progeroid syndromes (Ahmed et al., 2018; Oshima et al., 1993; Crome and Kanjilal, 1971; Yuen et al., 2001). In HGPS, children suffer from severe atherosclerosis with a 50% loss of vascular smooth muscle cells in the media of the aorta and an increase of the adventitia thickness due to increased fibrosis (Olive et al., 2010). Mice with WS display increased plasma cholesterol, triglycerides, insulin blood levels and interstitial cardiac fibrosis due to elevated levels of oxidative stress (Massip et al., 2006). In Ercc1^Δ/− progeroid mice, the enzymatic activity of the endonuclease is reduced to 5% (Gurkar and Niedernhofer, 2015). An impaired DNA repair system as a result of Ercc1 deficiency was found to lead to a systemic ageing process due to the accumulation of DNA damage, leading to a maximum lifespan of 24 weeks (Niedernhofer et al., 2006; Dollé et al., 2011). Ercc1 deficiency also leads to sarcopenia (Alyodawi et al., 2019) and neurological disorders (Sepe et al., 2016; Sharma et al., 2018). Importantly, potential effects of Ercc1 deficiency on the heart have not been extensively determined. A previous study has shown that the Ercc1^Δ/− progeroid mouse hearts are smaller than controls from week 1 of age. The Ercc1^Δ/− hearts decreased in weight from week 10, whereas control hearts continued to increase (Dolle et al., 2011). This suggests that the onset of progeria is associated with significant cardiac changes that have yet to be fully characterised.

It has been shown that circulating levels of activin A were increased in patients with heart failure (Yndestad et al., 2004). Circulating levels of activin A and myostatin were shown to have negative impacts on cardiac tissue physiology through binding to activin receptor type IIB (encoded by ACVR2B) (Loumaye et al., 2015). Soluble activin receptor type IIB (sActRIIB) has been used experimentally as a ligand trap to block receptor signalling with beneficial effects on cardiac muscle in heart failure (Lee et al., 2005). The sActRIIB treatment also showed beneficial effects during normal cardiac ageing by preserving cardiac function (Roh et al., 2019). We have previously shown that sActRIIB treatment of Ercc1^Δ/− mice significantly improves muscle mass, tone and function (Alyodawi et al., 2019). In this study, we investigated whether sActRIIB treatment of Ercc1^Δ/− mice could improve pathological cardiac remodelling and contractile function associated with progeria.

To determine the phenotype of the Ercc1^Δ/− progeroid mice with and without sActRIIB treatment, body weight, heart weight and tibia length were measured in the four experimental groups. The body weight of Ercc1^Δ/− mice was decreased by 57% in comparison to control mice (P<0.000001) (Fig. 1A). sActRIIB had no impact on the body weight of Ercc1^Δ/− mice. The body weight of control mice increased by 16% following sActRIIB treatment (P<0.000123).

The heart weight of Ercc1^Δ/− mice was reduced by 49% in comparison to control mice at 16 weeks (P<0.000001). sActRIIB treatment had no impact on Ercc1^Δ/− hearts. However, sActRIIB treatment decreased the heart weight in control mice (P<0.035282) (Fig. 1B).

The tibia length of Ercc1^Δ/− mice was reduced by 12% in comparison to control mice at 16 weeks (P<0.000507). sActRIIB had no impact on the tibia length of Ercc1^Δ/− mice. However, sActRIIB-treated control mice showed a 9% increase in tibia length (P<0.012632) (Fig. 1C).

The heart weight to tibia length ratio gives an indication of cardiac size relative to growth (Yin et al., 1982). The heart weight to tibia length ratio was reduced by 42% in Ercc1^Δ/− mice compared with control mice (P<0.000001) (Fig. 1D). A similar result was observed for treatment with the soluble activin receptor. sActRIIB treatment of Ercc1^Δ/− mice had no significant impact on the heart weight to tibia length ratio. However, sActRIIB treatment of control mice showed a 16% decrease of heart weight to tibia length ratio (P<0.038378). These data were also expressed as a ratio of heart weight to body weight (Fig. S1).

To determine cardiac function in Ercc1^Δ/− progeroid mice following sActRIIB treatment, mice were assessed at 12 and 16 weeks and the results were averaged (for control, Ercc1^Δ/− and Ercc1^Δ/− sActRIIB cohorts). The left ventricle mass was assessed by ultrasound for Ercc1^Δ/− and control mice and was decreased by 56% in comparison to the control hearts (P<0.0001) (Fig. 1E). sActRIIB had no impact on left ventricular mass.

The left ventricle internal diameter at the end of diastole for Ercc1^Δ/− mice was reduced by 20% in comparison to the control mice (P=0.0047) (Fig. 1F). The left ventricle internal diameter at the end of diastole for Ercc1^Δ/− sActRIIB mice was reduced by 30% in comparison to the control mice (P=0.0003). The interventricular septum thickness in diastole of Ercc1^Δ/− mice was reduced by 32% in comparison to the control mice (P=0.0317) (Fig. 1G), but was not altered by sActRIIB treatment.

The stroke volume for Ercc1^Δ/− mice was reduced by 58% in comparison to the control (P=0.0001) (Fig. 1H), a feature not changed by sActRIIB treatment. Cardiac output was reduced by 68% Ercc1^Δ/− mice, but was not altered by sActRIIB treatment (Fig. 1I).

Next, we examined systolic function. The ejection fraction of Ercc1^Δ/− mouse hearts was reduced by 24% in comparison to control, whereas sActRIIB treatment significantly increased it (Fig. 1J). There was no significant difference in the ejection fraction between control and Ercc1^Δ/− sActRIIB hearts. The fractional shortening of Ercc1^Δ/− mice was reduced by 31% in comparison to the control mice, and sActRIIB treatment significantly increased it (Fig. 1K). There was no significant difference in the fractional shortening between control and Ercc1^Δ/− sActRIIB hearts.

Picrosirius Red staining was performed to visualise the cardiac fibrosis. The results showed that there was no change in Ercc1^Δ/− hearts relative to control hearts. However, Ercc1^Δ/− mice treated with sActRIIB showed decreased fibrosis (P<0.002349) (Fig. 2A-E). Collagen I staining was also performed by immunostaining since it makes up 80% of the cardiac extracellular matrix (ECM). As with Picrosirius Red staining, there was no change in collagen I in Ercc1^Δ/− mice compared with control mice. However, sActRIIB treatment resulted in a significant decrease in both control and Ercc1^Δ/− mouse hearts, by 28% (P<0.016173) and 22% (P<0.044248), respectively (Fig. 2F-J). These changes were associated with a significant increase in tissue interstitial space in Ercc1^Δ/− mice compared with control mice (Fig. S2).

To determine the extent of accumulated tissue DNA damage, left ventricular tissue sections were stained for γH2AX, which detects DNA double-strand breaks (Mah et al., 2010). γH2AX staining was significantly increased in Ercc1^Δ/− hearts compared with control hearts (Fig. 3A-E) (P<0.000001). In Ercc1^Δ/− hearts treated with sActRIIB, there was a decrease in γH2AX staining compared with untreated Ercc1^Δ/− hearts (P<0.023025).

There is a close link between the production of reactive oxygen species and subsequent cellular damage in the form of lipid peroxidation, protein oxidation and DNA damage (Cooke et al., 2003; Dizdaroglu and Jaruga, 2012). We determined the extent of oxidative stress using dihydroethidium (DHE) staining as previously described (Wang and Zou, 2018) (Fig. 3F-J). DHE staining increased significantly in Ercc1^Δ/− hearts compared with control hearts (P<0.014995) (Fig. 3J). Ercc1^Δ/− mouse hearts had significantly less DHE staining following treatment with sActRIIB (P<0.038102) (Fig. 3J).

To determine whether the changes to heart weight represented changes to cardiomyocyte size, sections were stained for N-cadherin and wheat germ agglutinin (WGA). N-cadherin allowed longitudinal myocyte length to be determined, and WGA allowed the transverse area to be measured (Fig. 4A-F). The Ercc1^Δ/− mouse hearts displayed a reduced cardiomyocyte length in comparison with control hearts (P<0.004042) (Fig. 4E). Moreover, sActRIIB treatment normalised myocyte length in Ercc1^Δ/− hearts (P<0.039810) (Fig. 4E). The average myocyte width was unchanged in Ercc1^Δ/− mouse hearts compared with control hearts. However, sActRIIB treatment of Ercc1^Δ/− mice resulted in a significant increase in width compared with untreated Ercc1^Δ/− mice (P<0.034548) (Fig. 4F).

To determine whether the microvasculature was altered between the four experimental groups, left ventricular tissue sections were stained using an isolectin B4 antibody, which identifies endothelial cells (Fig. 4G-K). The Ercc1^Δ/− mouse hearts showed no difference in staining compared with control mouse hearts. However, the control mice treated with sActRIIB showed increased staining compared with control mice alone suggesting an increased microvasculature (P<0.000626) (Fig. 4K).

To mechanistically understand the impact of the Ercc1 mutation and how the activin signalling blockade influences cardiac remodelling, RNA-sequencing (RNAseq) analysis was performed on the four experimental groups. The results of this analysis indicated a large number of differentially expressed genes. The heat maps showing the top 30 most significantly altered genes from a pairwise comparison are shown in Fig. 5A-D.

The differentially expressed genes were then clustered by their biological gene ontology to determine which cellular functions had been affected. Genes associated with cell adhesion were the most significantly regulated between the control and Ercc1^Δ/− progeroid mice (Fig. 6A). The chemokine-mediated signalling pathway and chemotaxis were the most significantly altered processes between the untreated and treated Ercc1^Δ/− progeroid mice (Fig. 6B). The chemokine-mediated signalling pathway and chemotaxis were also the most significantly altered processes between the control mice and those treated with sActRIIB (Fig. 6C). Finally, cell division and cell cycle genes were the most significantly altered between the treated control and the treated Ercc1^Δ/− progeroid mice (Fig. 6D).

The differentially expressed genes related to oxidative stress or mitochondrial function and myocyte growth that were significantly different for the two experimental conditions, control versus Ercc1^Δ/− mice, and untreated Ercc1^Δ/− mice versus Ercc1^Δ/− mice treated with sActRIIB, were extracted from the RNAseq analysis (Table 1). The analysis showed that for genes related to oxidative stress, there was an upregulation of leukotriene C4 synthase (Ltc4s), and a downregulation of sulfiredoxin 1 (Srxn1), uncoupling protein 3 (Ucp3) and aquaporin 8 (Aqp8) in the Ercc1^Δ/− progeroid mice compared to control mice. For untreated Ercc1^Δ/− mice versus Ercc1^Δ/− mice treated with sActRIIB, there was an upregulation of Aqp8 and a downregulation of P450 family 2 subfamily E member 1 (Cyp2e1) and Ltc4s in Ercc1^Δ/− mice treated with sActRIIB. We also examined genes associated with myocyte growth and hypertrophy under the two experimental conditions shown in Table 1. For control versus Ercc1^Δ/− mice, secreted frizzled related protein 2 (Sfrp2), myostatin (Mstn), SLIT and NTRK like member 4 (Slitrk4) and myosin light chain 7 (Myl7) were all upregulated, whereas TNF receptor superfamily member 12A (Tnfrsf12a) was downregulated in Ercc1^Δ/− mice. For untreated Ercc1^Δ/− mice versus Ercc1^Δ/− mice treated with sActRIIB, fibroblast growth factor 6 (Fgf6) was upregulated, whereas WNT inhibitory factor 1 (Wif1), Slitrk4, Sfrp2 and Myl7 were downregulated in Ercc1^Δ/− mice treated with sActRIIB.

The RNAseq study also highlighted alternatively spliced genes that were differentially expressed from the four experimental groups. These were clustered by their gene ontology using gene ontology resources (Mi et al., 2019). Alternatively spliced genes associated with muscle development and organisation were the most affected between control and the Ercc1^Δ/− mouse hearts (Fig. 7A). Genes related to metabolism were the most significantly affected between the untreated Ercc1^Δ/− mice and Ercc1^Δ/− mice treated with sActRIIB (Fig. 7B). Genes related to muscle development, hypertrophy and sarcomere organisation were the most significantly affected between control mice and those treated with sActRIIB (Fig. 7C). Finally, genes associated with cardiac muscle contraction, sarcomere organisation and hypertrophy were the most affected between the sActRIIB-treated control mice and sActRIIB-treated Ercc1^Δ/− mice (Fig. 7D).

The differentially expressed genes highlighted in the RNAseq study were compared to the differentially expressed alternatively spliced genes to identify those affected by both processes (Fig. 7E-H). The comparison between control and Ercc1^Δ/− mouse hearts indicated 22 overlapping genes (Fig. 7E, Table S1). The pairwise comparison between untreated Ercc1^Δ/− mice and Ercc1^Δ/− mice treated with sActRIIB did not show any overlapping genes (Fig. 7F). The comparison between untreated control mice and control mice treated with sActRIIB showed 11 overlapping genes (Fig. 7G, Table S1), whereas the comparison between control sActRIIB-treated mice and Ercc1^Δ/− sActRIIB-treated mice indicated 28 overlapping genes (Fig. 7H, Table S1).

In this study, we investigated both the impact of Ercc1 depletion and the blockade of activin signalling on cardiac remodelling and gene expression. This study shows for the first time that Ercc1^Δ/− progeroid mouse hearts displayed significant DNA damage associated with increased oxidative stress. These hearts also had smaller cardiomyocytes and reduced systolic function. Activin blockade significantly reduced DNA damage, oxidative stress and cardiac fibrosis, induced myocyte hypertrophy and improved systolic function in Ercc1^Δ/− hearts. Gene expression analysis from RNAseq identified a large number of differentially expressed and alternatively spliced genes in all groups. The analysis showed that activin blockade decreased pro-oxidant and concomitantly increased antioxidant gene expression. Additionally, activin blockade also decreased the expression of anti-hypertrophic genes and increased the expression of hypertrophic genes. Mechanistically, these changes help to explain the improved morphology and function of Ercc1^Δ/− mouse hearts treated with sActRIIB. Our data also suggest that sActRIIB treatment of Ercc1^Δ/− mice also significantly attenuates cardiac inflammation as a result of decreased cellular damage.

Oxidative stress is a major source of protein and DNA damage and lipid peroxidation, and is associated with cardiac dysfunction. Our RNAseq analysis showed that the antioxidant genes Srxn1, Ucp3 and Aqp8 were all downregulated in Ercc1^Δ/− hearts. The upregulation of Srxn1 has been shown to increase the survival of cardiac progenitor cells and astrocytes by protecting them against oxidative stress (Li et al., 2018; Zhou et al., 2015). Aqp8 expression is a marker of mitochondrial function (Ikaga et al., 2015), and has been shown to be protective by transporting H₂O₂ out of pancreatic cells in the context of diabetes mellitus (Krüger et al., 2021). The downregulation of Aqp8 in Ercc1^Δ/− hearts would lead to increased oxidative stress and mitochondrial dysfunction. Another antioxidant Ucp3 that was decreased in Ercc1^Δ/− hearts has been shown to neutralise oxidative stress in myotubes (Barreiro et al., 2009). Taken together, the downregulation of these antioxidants would result in a significant oxidative stress in Ercc1^Δ/− hearts. sActRIIB treatment increased the expression of Aqp8, suggesting improved mitochondrial function and reduced oxidative stress as indicated in Fig. 8.

In addition to a decrease in tissue antioxidants, we show increased expression of the pro-oxidant Ltc4s in Ercc1^Δ/− hearts. Ltc4s has been shown to induce oxidative stress in chemotherapy, resulting in DNA damage (Dvash et al., 2015), so the increase in Ercc1^Δ/− hearts would be associated with inflammation-induced oxidation. Our RNAseq analysis showed that sActRIIB downregulated Ltc4s, suggesting reduced inflammation, and another pro-oxidative enzyme Cyp2e1 in the Ercc1^Δ/− mouse hearts. Cyp2e1 has been shown to induce oxidative stress in cardiomyocytes, resulting in apoptosis and impaired contractile function (Zhang et al., 2011). Moreover, inhibition of Cyp2e1 has been shown to improve cardiomyocyte contractility (Ren et al., 2019). The activin signalling pathway has been associated with increased oxidative stress. For example, myostatin was shown to increase oxidative stress through activation of NF-κB which in turn leads to TNF-α secretion (Sriram et al., 2011). Therefore, inhibiting the activin A/myostatin pathway may present a novel means of reducing oxidative stress in the heart. Despite the accumulation of DNA damage and intracellular oxidative stress, Ercc1^Δ/− progeroid mouse hearts did not develop significant fibrosis. Although Ercc1^Δ/− hearts displayed an increase in tissue interstitial space, presumably as a result of cell necrosis and apoptosis, there was no subsequent increase in cardiac fibrosis. These findings indicate that Ercc1^Δ/− mice may have functionally impaired cardiac fibroblasts (Robinson et al., 2018). In humans, mutations in the ERCC1 gene have been shown to affect the activity of fibroblasts (Jaspers et al., 2007), suggesting that DNA damage may directly impair the activity of these cells. In another progeroid model induced by a nuclear lamin A mutation, ECM protein synthesis by fibroblasts was impaired via the Wnt/β-catenin pathways, suggesting that progeria does not always display the hallmarks of normal ageing (Hernandez et al., 2010). Chemokine activity by cardiac macrophages has been shown to lead to impaired fibronectin and collagen deposition. This was associated with impaired remodelling of the ECM through the upregulation of matrix metalloproteinase-2 (Mmp2) (Deleon-Pennell et al., 2017).

Our data showed that inhibition of activin signalling resulted in a reduction of cardiac tissue fibrosis and collagen in both Ercc1^Δ/− and control mice. Increased activin signalling is associated with tissue fibrosis as a result of cardiac fibroblast proliferation and differentiation through the activation of the p38-MAPK and ERK1/2 pathway (Hu et al., 2016). Myostatin expression also induced interstitial fibrosis by the activation of the TAK1-MKK3/6 pathway leading to the production of collagen I (Biesemann et al., 2015). This suggests that inhibition of the activin signalling pathway could be used to reduce cardiac fibrosis in ageing, diseased hearts and other conditions that increase fibrosis.

To determine some additional mechanistic aspects in the phenotypes associated with the four treatment groups, we used mRNA sequencing to identify genes that were differentially expressed and spliced. A large number of genes were shown to be significantly targeted in the different treatment groups, and these were grouped by gene ontology. Interestingly, altered expression of a large number of genes suggested a disrupted circadian cycle in Ercc1^Δ/− hearts compared with control hearts. Clock genes such as Arntl (Bmal1), Per1, Per2, Per3, Nr1d2, Ciart, Npas2, Nfil3 and Dbp were all significantly altered. Arntl is expressed out of phase with its physiological repressors, the Per genes: Per1, Per2 and Per3 (Cao et al., 2021). A human study published by Chen et al. (2016), showed that ageing was associated with a decrease of Arntl expression, which was associated with a loss of the rhythmicity of the Per genes. Arntl is an important regulator of ageing mechanisms as its activity is vital for cell division and repair. Indeed, the deletion of Arntl in mice resulted in a progeroid syndrome with a shorter lifespan, smaller organs, increase of oxidative stress, loss of subcutaneous fat and other ageing-related diseases (Kondratov et al., 2006).

When Ercc1^Δ/− mice were treated with sActRIIB, there were far fewer changes observed in differentially expressed genes. The gene ontology of biological processes suggested that the main gene targets were those associated with the suppression of inflammation. This suggests that the inhibition of activin signalling significantly attenuates tissue inflammation in progeria. The increased inflammation in Ercc1^Δ/− mice is thought to be via activation of the NF-κB signalling pathway (Tilstra et al., 2012). Zhao and colleagues showed that NF-κB activation was associated with increased DNA damage and the activation of DNA damage response pathways (Zhao et al., 2020). Other upregulated chemokines in Ercc1^Δ/− hearts such as Ccl8, Ccl11 and Ccl12 and secreted factors such as Retnla have been shown to be involved in the late phase of inflammation (Yu et al., 2015). Ccl8 gene expression under the regulation of NF-κB (Halle et al., 2010) was associated with uremic cardiomyopathy (Amador-Martínez et al., 2021). Ccl11, or eotaxin, is also known as the protein of senility (Khavinson et al., 2016) and is mainly produced in the heart by the resident fibroblast and the macrophages. Its secretion is required for the eosinophil chemoattraction in myocarditis (Diny et al., 2016). Ccl11 was also shown to be involved in angiotensin II-induced cardiac hypertrophy through the downstream regulation of pro-hypertrophic genes such as ANP, BNP and BMHC (Zhang et al., 2018).

Our data indicate that the cardiomyocytes from Ercc1^Δ/− hearts were significantly smaller than control hearts and were associated with decreased systolic function. The RNAseq analysis showed that the smaller myocytes were associated with increased expression of the anti-hypertrophic genes Sfrp2, Mstn and Sltrk4. Sfrp2 inhibits hypertrophy through the Wnt/β-catenin pathway (Wei et al., 2020) whereas MSTN attenuates hypertrophy through its impact on autophagy (Qi et al., 2020). In addition to the increased expression of anti-hypertrophic genes, Ercc1^Δ/− hearts also expressed lower levels of the pro-hypertrophic gene Tnfrsf12a. Tnfrsf12a has been shown to be an important inducer of cardiac hypertrophy (Yerra et al., 2021). However, the size of the cardiomyocytes relative to heart weight suggests that Ercc1^Δ/− hearts are also likely to have fewer contractile cells. As cardiomyocyte proliferation ceases early in the neonatal period, reduced myocyte number is likely to have occurred during development. Specifically, Ercc1^Δ/− hearts had shorter myocytes whereas myocyte width remained unchanged. Myocyte size is regulated by both preload and afterload and is associated with the addition of sarcomeres in series or parallel, respectively (Opie et al., 2006). Cardiac size can also be affected in conditions that induce cachexia such as cancer and viral infections through the NF-κB pathway (Harvey and Leinwand, 2014). In cancer, the inhibition of NF-κB increased left ventricular mass and function by inducing cardiomyocyte hypertrophy. NF-κB-mediated inflammation also leads to the upregulation of the ubiquitin ligases MuRF1 and atrogin-1 that are known to be involved in myostatin- or activin A-induced skeletal muscle atrophy by increasing protein degradation and inhibiting protein synthesis (Wysong et al., 2011). The activation of MuRF1, atrogin-1 and Foxo3a increase protein degradation through the ubiquitin-proteasome machinery and autophagy (Wang et al., 2015; Ding et al., 2017). MuRF1 and atrogin-1 can also inhibit the canonical cardiac hypertrophic pathway through calcineurin (Razeghi and Taegtmeyer, 2006) and Akt through the downregulation of mTOR signalling pathway (Razeghi and Taegtmeyer, 2005). Ercc1^Δ/− hearts treated with sActRIIB had improved systolic function as measured by increased ejection fraction and fractional shortening. This was associated with increased myocyte hypertrophy. However, stroke volume and cardiac output were not significantly altered, suggesting that changes to filling properties could also be altered by sActRIIB, as shown in a previous study (Rodgers et al., 2009). The RNAseq study highlighted the downregulation of the anti-hypertrophic genes Wif1, Slitrk4, Sfrp2 and Myl7 in the Ercc1^Δ/− mice treated with sActRIIB. Myl7 has been shown to inhibit cardiac hypertrophy (Sun et al., 2021), whereas Wif1 impacts hypertrophy through inhibition of the Wnt/β-catenin pathway (Lu et al., 2013). Additionally, sActRIIB treatment of Ercc1^Δ/− mouse hearts increased the expression of Fgf6, which is associated with protein synthesis and muscle hypertrophy (Xu et al., 2021). sActRIIB treatment shifts the balance in favour of hypertrophic myocyte growth and, in combination with the reduction in oxidative stress, results in improved systolic function as illustrated in Fig. 8.

Our data showed that although sActRIIB treatment increased the cardiac vasculature in control mouse hearts, it had no impact on Ercc1^Δ/− progeroid hearts. These data indicate that the cellular hypertrophy induced by sActRIIB treatment was not accompanied by significant angiogenesis. The lack of angiogenesis normally associated with myocyte hypertrophy could negate some of the beneficial effects by increasing the diffusion distance for oxygen and nutrients. Durik et al. (2012) showed that Ercc1^Δ/− progeroid mice developed vascular cell senescence, vasodilator dysfunction, vascular stiffness and elevated blood pressure. These vascular changes and the inhibition of angiogenic pathways could explain the impaired response to sActRIIB treatment in Ercc1^Δ/− hearts, despite the increased cardiomyocyte hypertrophy. It is interesting that sActRIIB treatment significantly decreased cardiac size in control hearts, whereas the size of Ercc1^Δ/− hearts remained unchanged. These data are in contrast to skeletal muscle, in which sActRIIB treatment is known to significantly increase muscle mass (Alyodawi et al., 2019). These differences indicate that skeletal and cardiac muscles have differing responses to activin signalling inhibition. Our work here showed that unlike Ercc1^Δ/− mice, treatment of control mice with sActRIIB resulted in a decrease in the weight of the heart. The impact of sActRIIB on cardiac gene expression changes in control mice is clearly evident in heatmap plots (Fig. 5B). Indeed, scrutiny of the gene expression changes through the heatmaps suggests that the impact of sActRIIB was greater in control hearts than in the Ercc1^Δ/− hearts. It is therefore important to understand the cellular origin of these changes in gene expression and their functional consequence for the control mice. The most obvious mediator of these gene changes is the cardiomyocyte population. However, analysis of cardiomyocyte size showed that there were no significant changes induced by sActRIIB treatment of control mice. Therefore, if changes are being induced in cardiomyocytes, they are not affecting their size. It is also worth contemplating the non-contractile component of the heart as a possible reason for the decreased heart weight following treatment with sActRIIB. Conceivable explanations for this point come to light by interrogating the Gene Ontology (GO) term enrichment analysis from the differentially expressed genes, which reveals that a number of activities related to changes in the ECM may have changed, e.g. regulation of protein secretion, changes in cell adhesion molecules and, importantly, regulation of proliferation. Therefore, the changes in these pathways could act on the cardiac fibroblast population resulting in fewer cells, which ultimately produce less ECM as evidenced by our quantification of collagen I (Fig. 2). However, we suggest that the decrease in collagen I levels induced by sActRIIB may not reflect changes in fibrosis (a pathological state) but rather the ECM normally formed to couple structure with function. This seems to be the most plausible source of loss of heart weight. Changes in ECM may also have impacts on myocardial compliance, thereby affecting filling properties during diastole.

Finally, our study shows for the first time differences in gene targets between those that were differentially expressed and alternatively spliced in Ercc1^Δ/− mouse hearts. A small number of genes were both alternatively spliced and expressed in the treatment groups, with the exception of the Ercc1^Δ/− mice that were treated with sActRIIB. This suggests that in Ercc1^Δ/−-induced progeria, the coordination between mRNA synthesis and post-transcriptional regulation is impaired and likely blunts the response to activin-signalling inhibition. Alternative splicing may provide an important means of regulating cellular activities in post-mitotic cells such as cardiomyocytes. In cardiomyopathy, alternative splicing has been shown to target calcium signalling, sarcomere organisation and apoptosis (Hu et al., 2017; Dlamini et al., 2015; Kong et al., 2010). These findings are in keeping with our data indicating that splicing-induced functional changes in proteins are important for regulating cardiac muscle architecture (Kolur et al., 2021; D'Amario et al., 2018), calcium homeostasis (Torac et al., 2014; Wanitchakool et al., 2017), inflammation (Comerford et al., 2006; Wu et al., 2019; Morgan and Harris, 2015) and circadian rhythmicity (Young, 2006; Wang et al., 2021). The numerous changes in response to sActRIIB treatment are likely due to signalling changes downstream of the activin receptor. Studies have shown that inhibition of this pathway has significant impact on both cardiac and skeletal muscle, resulting in both morphological and changes in gene expression (Lodberg, 2021).

In conclusion, we have shown for the first time that a model of attenuated DNA repair is associated with pathological cardiac remodelling and gene expression. However, a significant amount of this phenotype is attenuated by inhibition of the activin signalling pathway using soluble activin receptor treatment. The inhibition of this pathway improved cardiac systolic function, decreased cardiac fibrosis, oxidative stress and DNA damage. Moreover, gene expression analysis showed that inhibition of the activin signalling pathway led to an increase in antioxidants and a decrease in pro-oxidants that explains the decreased oxidative stress. There was also increased expression of pro-hypertrophic and a decrease in anti-hypertrophic genes that was associated with myocyte hypertrophy. We propose that these two mechanisms combine to improve cardiac contractile function following sActRIIB treatment. These data strongly suggest that inhibition of activin signalling may provide a novel means of improving cardiac remodelling that is associated with ageing and pathological stimuli that lead to heart failure.

All work was done in compliance with the Institutional Ethical Committee. The experiments were performed under a project licence from the United Kingdom Home Office in agreement with the Animals (Scientific Procedures) Act 1986. The University of Reading Animal Care and Ethical Review Committee approved all procedures. Animals were humanely sacrificed via Schedule 1 killing using carbon dioxide.

This study was performed mostly on male mice with a hybrid C57BL/6-FVB F1 background (Weeda et al., 1997), except for the RNAseq analyses, for which the untreated Ercc1^Δ/− and Ercc1^Δ/− sActRIIB-treated group contained one female. Transgenic Ercc1^Δ/− and control mice were bred as previously described (Dolle et al., 2011; Weeda et al., 1997). Mice were kept in ventilated cages under specific pathogen-free conditions (20-22°C, 12-12 h light-dark cycle) and fed with food and water ad libitum. The Ercc1^Δ/− mice were smaller than the control mice, so the food was given within the cages and water bottles with long nozzles and were used from around 2 weeks of age. Mice were bred and maintained on AIN93G synthetic pellet. The myostatin/activin pathway was inhibited from week 7 to week 16 by intraperitoneal injection (IP) with 10 mg/kg of sActRIIB twice a week. Mice were sacrificed at week 16 as this is the timepoint when death began to occur. Once their body weights were measured with a precision balance, animals were sacrificed, then both tibia were measured with digital callipers, hearts were harvested for the study and the weights were measured with an analytical balance as previously described (Alyodawi et al., 2019). The tissue was stored at −80°C prior to use.

Harvested hearts were embedded in Optimal Cutting Temperature (OCT) medium before being stored at −80°C for at least 24 h. Then transverse sections were collected by cryo-sectioning the OCT blocks at a thickness of 5 µm at −20°C and harvested on slides, before being stored at −80°C.

The function of the heart was determined with the Vevo 2100 Ultrasound imager on untreated wild-type mice (n=4), untreated Ercc1^Δ/− progeroid mice (n=3) and sActRIIB-treated Ercc1^Δ/− progeroid mice (n=4) aged 12 weeks and 16 weeks. The cardiac function imaging was performed on M-mode short axis on mice under isoflurane anaesthesia. Prior to starting the cardiac function imaging, the Vevo 2100 software (FUJIFILM Visualsonics) was set up with cardiology and cardiac packages. The image was calibrated for the width at 11.36 mm and depth at 11 mm. To perform the anaesthesia, the mouse was placed in the induction chamber and the oxygen flow was calibrated to 1 l/min and the isoflurane vaporiser to 5%. Once the mouse lost its righting reflex and its breathing pattern became deeper, the isoflurane was decreased to 1.5%. The short axis view M-mode images were analysed by using the Vevo 2100 software in order to measure the cardiac physiological parameters, the left ventricle mass, the stroke volume, the cardiac output, the ejection fraction, the fractional shortening, the left ventricle internal diameter at the end of diastole and the interventricular septum thickness in diastole. The results were presented as a mean of the cardiac function assessed at 12 and 16 weeks of age. There were no age discrepancies between the experimental groups.

To visualise cardiac fibrosis, sections were stained with Picrosirius Red staining kit (ab150681, Abcam). Sections were fixed in Bouin solution for 15 min at 56°C at room temperature, then washed for 15 min at room temperature in distilled water. Sections were stained in Picrosirius Red for 1 h at room temperature under the fume hood. Slides were differentiated in acidified water, dipped twice. Sections were then washed and dehydrated three times in 100% ethanol before being cleared in xylene for 5 min. Coverslips were mounted with DPX mounting medium. The images were analysed using a bright field microscope (Nikon TE-200) and 10 pictures per animal were captured of the left ventricle area at 400× magnification. The quantification of the Picrosirius Red staining was carried out with ImageJ software 1.53h. The original picture was changed to RGB stack, and only the green channel was kept, and the fibrotic area was measured by the threshold method.

5 μm-thick cardiac tissue sections were incubated in permeabilization buffer 0.5% Triton-X 100 for 15 min before being washed three times for 5 min in 1× PBS. The slides were incubated for 30 min in blocking buffer containing 5% fetal calf serum before being washed three times for 10 min in 1× PBS. Sections were incubated overnight at 4°C with the rabbit polyclonal anti-collagen I primary antibody (1:200, ab34710, Abcam). Slides were then washed three times for 10 min in 1× PBS before being incubated at room temperature for 1 h with the chicken anti-rabbit Alexa Fluor 488 secondary antibody (1:200). Sections were then washed three times for 5 min. Coverslips were mounted on them with DAPI mounting medium, and slides were kept in the dark at 4°C. The images were analysed using fluorescence microscopy (AxioImager Epifluorescent System) and 10 pictures per animal were taken in the left ventricle area at 400× magnification. The quantification of the collagen I staining was carried out with ImageJ software 1.53h by a threshold method to get the percentage of the area covered by collagen I.

5 μm-thick sections were fixed in methanol–acetone (50:50) solution at −20°C for 10 min, then washed. Sections were incubated in permeabilization buffer containing 0.5% Triton-X 100 for 15 min before being washed in 1× PBS. Then, the slides were incubated for 30 min in blocking buffer containing 5% fetal calf serum before being washed in 1× PBS. Sections were incubated overnight at 4°C with the primary antibody anti-N-cadherin (1:200, ab18203, Abcam). Slides were then washed in 1× PBS before being incubated at room temperature for 1 h with the secondary antibody Alexa Fluor 568 (1:200) and the probes of the WGA (1:1000, W11261, Thermo Fisher). Sections were then washed, and coverslips were mounted with DAPI mounting medium. The images were taken using fluorescence microscopy (AxioImager Epifluorescent System) and 10 pictures per animal were captured in the left ventricle area with the double staining at 200× magnification to measure the length and the width of the cardiomyocytes (Figs 3 and 4). For these features, 20 cells were analysed per picture with ImageJ software 1.53h before statistical analysis. The width was measured from the WGA line to WGA line (green), whereas the length was measured from the N-cadherin line to N-cadherin line (red) (intercalated disk). For these features, 20 cells were analysed per image, then each picture was averaged before doing statistical analysis.

5 μm-thick sections were firstly fixed in 5% paraformaldehyde at room temperature for 15 min, then washed in 1× PBS for 15 min. The sections were then incubated for 15 min in permeabilization buffer containing 0.5% Triton-X 100, washed three times for 5 min in 1× PBS, and incubated for 30 min in blocking buffer containing 5% goat serum. Slides were finally incubated overnight at 4°C with the primary antibody anti- γH2AX (1:200, 9718, Cell Signaling Technology), then washed in 1× PBS before being incubated at room temperature for 1 h with the secondary antibody Alexa Fluor 568 (1:200) (Table 1). Coverslips were mounted with DAPI mounting medium. The images were analysed with fluorescence microscopy (AxioImager Epifluorescent System) and 10 pictures per animal were captured in the left ventricle area with the double staining at 400× magnification to highlight the DNA damage within the nuclei. The analysis was performed with ImageJ software 1.53h. Two counts were performed, one for total number of nuclei and the other for the nuclei positive for γH2AX staining. Finally, for the quantification of DNA damage, we calculated the percentage of positive nuclei.

5 μm-thick sections were first left at room temperature to dry for 15 min, then rehydrated with 1× PBS two times for 2 min. Slides were then incubated with 10 μM DHE (Table 1) for 30 min at 37°C in the dark and then washed with 1× PBS two times for 2 min. Coverslips were mounted with DAPI mounting medium, and slides were kept in the dark at 4°C. The images were analysed using fluorescence microscopy (AxioImager Epifluorescent System) and 10 pictures per animal were captured in the left ventricle area at 200× magnification. The image analysis was done with ImageJ software 1.53h by measuring the intensity of the staining.

RNAseq and alternative splicing was performed with 20 mg of the left ventricle from three mice for each experimental condition: control and Ercc1^Δ/− mice, and control and Ercc1^Δ/− mice treated with sActRIIB. RNA extraction, library preparations, sequencing reactions and bioinformatics analyses were performed as a blind study at GENEWIZ^®, LLC (South Plainfield, NJ, USA).

One-way ANOVA parametric test was used followed by a Tukey post-hoc test to compare the mean of control mice to Ercc1^Δ/− progeroid mice that were untreated and treated with sActRIIB. The results are expressed as mean±s.d. The levels of significance were set at *P≤0.05, **P<0.01 and ***P<0.001. The statistical analysis was performed using GraphPad Prism 9.2.0. The enrichment of gene ontology terms of the differentially expressed genes was tested using Fisher exact test with GeneSCF v1.1-p2 and performed by GENEWIZ^®. The enrichment of gene ontology terms of the differentially alternative spliced gene was performed with the online tool Gene Ontology Resources. The P-value was tested by Fisher exact test and corrected with a false discovery rate test to yield any false negatives.

The authors thank Dr Eva Kevei for her supervision and helpful discussions during this project.