https://orcid.org/0000-0001-8910-0902
Abstract
Cellular senescence is a stress-related or aging response believed to contribute to many cardiac conditions; however, its role in atrial fibrillation (AF) is unknown. Age is the single most important determinant of the risk of AF. The present study was designed to (i) evaluate AF susceptibility and senescence marker expression in rat models of aging and myocardial infarction (MI), (ii) study the effect of reducing senescent-cell burden with senolytic therapy on the atrial substrate in MI rats, and (iii) assess senescence markers in human atrial tissue as a function of age and the presence of AF.
AF susceptibility was studied with programmed electrical stimulation. Gene and protein expression was evaluated by immunoblot or immunofluorescence (protein) and digital polymerase chain reaction (PCR) or reverse transcriptase quantitative PCR (messenger RNA). A previously validated senolytic combination, dasatinib and quercetin, (D+Q; or corresponding vehicle) was administered from the time of sham or MI surgery through 28 days later. Experiments were performed blinded to treatment assignment. Burst pacing-induced AF was seen in 100% of aged (18-month old) rats, 87.5% of young MI rats, and 10% of young control (3-month old) rats (P ≤ 0.001 vs. each). Conduction velocity was slower in aged [both left atrium (LA) and right atrium (RA)] and young MI (LA) rats vs. young control rats (P ≤ 0.001 vs. each). Atrial fibrosis was greater in aged (LA and RA) and young MI (LA) vs. young control rats (P < 0.05 for each). Senolytic therapy reduced AF inducibility in MI rats (from 8/9 rats, 89% in MI vehicle, to 0/9 rats, 0% in MI D + Q, P < 0.001) and attenuated LA fibrosis. Double staining suggested that D + Q acts by clearing senescent myofibroblasts and endothelial cells. In human atria, senescence markers were upregulated in older (≥70 years) and long-standing AF patients vs. individuals ≤60 and sinus rhythm controls, respectively.
Our results point to a potentially significant role of cellular senescence in AF pathophysiology. Modulating cell senescence might provide a basis for novel therapeutic approaches to AF.
Senescent cells accumulate in the atria of aged rats and rats with left ventricular dysfunction due to myocardial infarction (MI). These conditions were associated with atrial fibrosis, an important pathological atrial fibrillation (AF) substrate, along with enhanced AF susceptibility (left panel). MI rats treated with senolytic compounds, which selectively eliminate senescent cells by inducing apoptosis, showed less atrial fibrosis and AF sustainability, along with suppression of markers of cellular senescence, particularly in myofibroblasts and endothelial cells (right panel).
This manuscript was handled by Guest Editor, Thomas F. Lüscher.
Time of primary review: 15 days
See the editorial comment for this article ‘Is ageing a modifiable risk factor for atrial fibrillation?’, by J. Heijman and C.T. Madreiter-Sokolowski, https://doi.org/10.1093/cvr/cvae040.
1. Introduction
Atrial fibrillation (AF) is the most common sustained arrhythmia observed in clinical practice.1,2 Age is the single important AF risk factor: AF incidence progressively doubles with each decade after 65 years of age.3 A range of cardiovascular conditions, including left ventricular (LV) dysfunction, valvular heart disease, and myocardial infarction (MI), interact with age to cause left atrial (LA) remodelling and further increase AF risk.4,5 However, the mechanisms underlying the important age dependence of AF occurrence remain poorly understood.
Growing evidence indicates that acquired cellular senescence plays an important role in the pathophysiology of cardiac pathologies including cardiac hypertrophy, cardiotoxicity, MI, and heart failure.6–10 Cellular senescence is a stress-related and aging response that classically causes cell-cycle arrest in dividing cells, regulated through two main pathways: p16–retinoblastoma protein and p53–p21.11,12 More recent work points to closely related pathways and changes in non-dividing cells like cardiomyocytes.6 The main proteins in the senescence pathways, like p16, p21, and p53, are widely used as markers of senescence. Senescent cells are resistant to apoptosis and secrete proinflammatory cytokines, growth factors, and matrix remodelling proteases, manifesting a ‘senescence-associated secretory phenotype’ (SASP).13,14
Two studies have investigated the expression of senescence markers in the heart and their potential association with AF promotion.15,16 Xie et al.15 showed that greater expression levels of senescence markers in atrial tissue are associated with more extensive atrial fibrosis and a greater AF recurrence likelihood in patients undergoing combined valve replacement/MAZE surgery. Jesel et al.16 showed stepwise increases in atrial p53 and p16 expression between sinus rhythm (SR), paroxysmal AF, and permanent AF groups. While these studies suggest an association between senescence markers and AF risk, they do not provide any direct evidence of causality.
Recent work has revealed that senolytic compounds, which selectively eliminate senescent cells by inducing apoptosis, may reduce the adverse cardiac consequences of ischaemia reperfusion and neurohormonal activation.17–20 These studies used navitoclax, a compound that suppresses Bcl2 overexpression in senescent cells,19,20 or a combination of the tyrosine kinase inhibitor dasatinib (D) and the flavonoid quercetin (Q).17
Here, we address the potential roles of atrial cell senescence in the pathophysiology of AF. Specific goals included (i) to evaluate AF susceptibility and senescence marker expression in atrial tissue of rat models of aging and MI; (ii) to study the effect of reducing atrial senescent-cell burden on AF susceptibility and the atrial substrate of post-MI rats with senolytic therapy [a combination of dasatinib and quercetin (D + Q)]; and (iii) to evaluate senescence marker expression in atrial tissue from younger (≤60 years old) vs. older (≥70) human patients and in patients with long-standing persistent (chronic) AF (cAF) vs. SR.
2. Methods
2.1 Animal models
All experimental and animal-handling procedures were approved by the Animal Ethics Committee of the Montreal Heart Institute and were conducted in accordance with the Canadian Council on Animal Care and National Institute of Health Guide for the Care and Use of Laboratory Animals. Male Sprague–Dawley rats were obtained from Envigo (Indiana, USA) at 20 months (aged group), 3 months (young control), or 2 months (young MI). For the senolytic study, the male Sprague–Dawley rats (2 months) were obtained from Charles River Laboratories (Montreal, Quebec). To induce MI, 2-month-old rats were injected subcutaneously (sc) with buprenorphine (0.05 mg/kg) and anaesthetized with 2% isoflurane. The left anterior descending (LAD) coronary artery was ligated via left thoracotomy; sham rats underwent the same procedure without LAD ligation. Buprenorphine (0.05 mg/kg sc) was given at 6 and 24 h and ketoprofen (5 mg/kg sc) at 6 h post-operatively. For all MI and sham rats, all in vivo, ex vivo, and tissue collection procedures were performed on Day 28 post-operatively (at 3 months of age). Rats were euthanized by isoflurane overdose followed by cardiac excision.
2.2 Drugs and chemicals
Dasatinib (D-3307) was obtained from LC Laboratories (Woburn, MA, USA). Quercetin (Q4951) was obtained from Sigma-Aldrich (St. Louis, MI, USA). Blebbistatin (B592490) was obtained from Toronto Research Chemical Inc (North York, ON), and di-4-ANEPPS (90134-00-2) was obtained from Cayman Chemical Company (Ann Arbor, MI, USA). TaqMan probes, including senescence markers, p16, p21, p53, galactosidase beta 1 (Glb1), and SASP markers, colony stimulating factor 2 (Csf2), chemokine (C–C motif) ligand 2 (Ccl2), insulin-like growth factor-binding protein 3 (Igfbp3), insulin-like growth factor-binding protein 5 (Igfbp5), and interleukin 6 (Il-6), were obtained from Thermo Fisher Scientific (Waltham, MA, USA). Primary antibody sources for the rat study were as follows: p16 [for immunofluorescence: ab54210, Abcam (Cambridge, UK); for WB: 1E12E10 (Thermo Fisher Scientific)], p53 (ab26 from Abcam), p21 (ab109199), Glb1 (ab128993), Serpine2 (ab134905), chemokine (C–C motif) ligand 20 (Ccl20, ab9829), matrix metalloproteinase 9 (Mmp9, ab76003), Mmp2 (ab92536), α-smooth muscle actin (αSMA, ab5694), troponin I (ab47003), vimentin (3932S), and p16 (for the human study: 80772) from Cell Signaling (Danvers, MA, USA) and p21 antibody (human study, 05-655) from Sigma-Aldrich (St. Louis, MI, USA), CD31 (AF3628-SP) from R&D System (Minneapolis, MN, USA), and GAPDH (5G4-6C5) from Hytest Ltd (Turku, Finland). The 4′,6-diamidino-2-phenylindole (DAPI), wheat germ agglutinin (WGA, W32466) and secondary antibodies, 555 donkey anti-mouse (A31570), 488 donkey anti-rabbit (A21206), and 488 donkey anti-goat (A11055) were obtained from Invitrogen (Waltham, MA, USA).
2.3 Senolytic study
After surgery, MI and sham rats were allocated (in a blinded fashion) to the following groups: sham (vehicle); sham (D + Q), MI (vehicle), and MI (D + Q). D + Q or vehicle treatment started 4 h post-operatively. Rats received vehicle (20% PEG-400 in normal saline) or the senolytic combination (5 mg/kg dasatinib + 50 mg/kg quercetin) by oral gavage once daily for 3 consecutive days beginning the day of MI, with two courses of therapy separated by 2 weeks, a regimen chosen based on prior work.17
2.4 Echocardiography
Echocardiograms were recorded for aged and young rats (at end-study) and young MI study rats (both sham and MI, at baseline and end-study) under 2% isoflurane anaesthesia using a phased-array 10S probe (4.5–11.5 MHz) in a Vivid 7 Dimension system (GE Healthcare Ultrasound, Horten, Norway). M-mode echocardiogram was obtained in parasternal long-axis views to measure LA diameter at both end cardiac systole and diastole (LADs and LADd). M-mode echocardiogram was also obtained in parasternal short-axis view at the level of papillary muscles to measure LV diameter at both end cardiac systole (LVDs) and diastole (LVDd). LV mass was calculated using the Reffelmann formula, and LV fractional shortening (FS) was calculated by FS = (LVDd − LVDs)/LVDd × 100. In apical four-chamber view, right atrium (RA) minor (horizontal) diameter at end cardiac systole (RADs) was obtained by two-dimensional echocardiogram, tricuspid annulus plane systolic excursion (TAPSE) by M-mode echocardiogram, and the right ventricular lateral systolic velocity (SR) by tissue Doppler imaging. Pulsed-wave Doppler was used to record trans-mitral flow in apical four-chamber view. Peak velocity of early-filling E-wave, atrial-filling A-wave, and E-wave deceleration time was measured.
2.5 Transoesophageal electrophysiological study
Transoesophageal electrophysiological study was performed in vivo with a 4-F quadripolar catheter (2 mm interpolar distances. St. Jude Medical #401993, Saint Paul, MI, USA). To assess atrial arrhythmia inducibility, 25 Hz burst pacing (pulse width 2 ms, 4× threshold voltage) was applied for 3 s, with 12 bursts separated by 2-s intervals; the cycle was repeated three times. AF was defined by a rapid (>800 bpm), irregular atrial rhythm ≥ 1 s. AF duration was the mean duration of all induced AF episodes. Surface electrocardiogram (ECG) and catheter signals were recorded and analysed using iox2 software (v.2.8.0.13, EMKA Technologies, Paris, France) and ECG auto (c3.5.5.25, EMKA Technologies, Paris, France). The experimenter was blinded to group identity throughout the experimental protocol and analysis.
2.6 Ex vivo optical mapping
The heart was excised and perfused via the aorta with Krebs solution at 10 mL/min and 37°C. After 20 min for stabilization, a recirculating solution containing blebbistatin (15 µmol/L) was used to suppress mechanical contraction. Di-4-ANEPPS (10 µmol/L, 0.1 mL) was injected. Fluorescence signals were recorded at 2000 frames-per-second with a charge-coupled device camera (CardioCCD, Red Shirt Imaging, Decatur, GA) focused on a region up to 8 mm square in the LA free wall. Conduction velocity and action potential duration to 50 and 80% repolarization (APD50 and APD80) were measured blinded to group assignment with a MATLAB custom-written algorithm.21
The camera utilized in our optical mapping studies has a spatial resolution of approximately 162.5 µm × 162.5 µm per pixel, derived from a camera field of 13 mm × 13 mm divided over an 80 × 80 pixel grid. To calibrate the focus, a thin transparent ruler is placed closely above the tissue. The camera’s focus is then adjusted, and reference images are captured for subsequent spatial resolution calculations. We selected regions with high-quality staining and signal clarity for camera focus, recording, and analysis.
To quantify conduction velocity, we first subtracted the background fluorescence for each pixel and then normalized it to peak amplitude. Activation for each pixel was identified by the maximum rate of fluorescence intensity rise during the upstroke. Subsequently, the activation times across all pixels within the designated region of interest served to generate isochronal maps of each wavefront. Based on the isochronal maps, we calculated activation vectors that encapsulate the magnitude and direction of local wavefront propagation by determining the spatial gradient of the average activation time map. Mathematically, this vector was represented by two orthogonal components, denoted as [Tx, Ty] [Eq. (1)], which defines the direction of wavefront propagation.
Conduction velocity at each pixel was determined from each activation vector [Eq. (2)]. These initial calculations were based on unit distance pixels and were later scaled by the appropriate spatial resolution parameters to derive actual overall conduction velocities. The final measurements are visually depicted in colour-coded maps featuring uniformly spaced activation vectors. The overall conduction velocity was determined from the mean of all individual vector conduction velocities.
2.7 Histology
Formalin-fixed, paraffin-embedded samples were cut at 6 µm thickness. To measure fibrosis, the samples were stained with Masson’s trichrome solution. Premier 9.3 Software (Media Cybernetics) was used to quantify fibrosis on Masson’s trichrome-stained images. The protein expression of p16 in LA tissue was measured with immunofluorescence as follows: the formalin-fixed, paraffin-embedded LA sections were used. The antigen retrieval process was performed with citrate buffer solution (10 mM citric acid, 0.05% Tween 20) in distilled water and the pH was adjusted to approximately 6.0 with sodium hydroxide (NaOH). Tissue sections were deparaffinized in xylene, followed by rehydration in a graded series of ethanol solutions. The tissue sections were submerged in the citrate buffer solution and heated for 25 min at a sub-boiling temperature (about 95°C) to facilitate antigen retrieval. Subsequently, the slides were rinsed in phosphate-buffered saline (PBS). To minimize non-specific binding, tissue sections were incubated in 4% bovine serum albumin (BSA) solution in PBS, for 2 h at room temperature, and the sections were then incubated with primary antibodies overnight at 4°C. The primary antibodies were diluted to their optimal working concentrations in PBS containing 1% BSA: p16 primary antibody (1/400) co-stained with either vimentin (1/200) or αSMA (1/100) or troponin I (1/100) or CD31 (1/100). After overnight primary antibody incubation, the sections were washed three times with PBS for 5 min each. DAPI, WGA (to stain membrane), and secondary antibodies (1/750) were then applied, and the sections were again washed after 1 h of incubation with secondary antibodies. The entire LA sections double stained for p16 and cell-selective markers were scanned (Aperio VERSA Brightfield Scanner) and automatically analysed with Visiomorph software (Visiopharm, Hoersholm, Denmark). All analyses were performed and analysed blinded to rat group identity. WGA membrane staining was used to delineate cell borders, and cardiac cells were identified with positive troponin I (cardiomyocytes), vimentin (fibroblasts), αSMA (myofibroblasts), and CD31,(endothelial cells) staining (cells with a signal intensity that is 25% greater than the background in the green channel). Cells containing DAPI-positive nuclei were considered to calculate % of p16+ cells.
2.8 Human studies
RA appendages (RAAs) were collected from patients (>18 years) undergoing open-heart surgery. Patients who were included in the study of aging and cell senescence [younger (≤60; n = 18) and older (≥70; n = 19) patients] had no history of paroxysmal, persistent, or permanent AF. Patients who were included in the study of AF and cell senescence had either a history of >6-month continuous AF (the cAF group; n = 11) or were in normal SR at the time of surgery and had no prior AF history (SR group; n = 11) (see Supplementary material online, Tables S1 and S2). We were not able to include RAAs from certain procedures, namely, off-pump and redo procedures, emergency surgery, or cases for which the surgeon judged it not appropriate to excise the RAA. All participants gave their written informed consent. The study was designed and conducted to conform to the principles of the Declaration of Helsinki and was approved by the University Hospital Essen’s ethical review boards (no. 12-5268-BO).
2.9 Immunoblot (human samples)
Proteins were isolated from human atrial tissue homogenates, and protein levels were determined using western blot according to standard protocols.22 Antibodies used in the study are listed below above in the Drugs and Chemicals section, at the following dilutions: p16 (1/1000), p21(1/1000), and GAPDH (1/20000). We used the appropriate near-infrared fluorophore dyes (IRDye, all 1:20,000, LI-COR Biosciences, Lincoln, NE) as secondary antibodies and imaged with an Odyssey Infrared Imaging System (LI-COR Biosciences).
2.10 Immunoblot (rat samples)
The samples in liquid nitrogen were crushed with a small grinding rod, and 200 μL lysis buffer (150 mM NaCl, 50 mM Tris-HCl pH7.5, 0.1% TritonX-100, 10% glycerol, 0.1%SDS, 1 mM phenylmethanesulfonyl fluoride) and 10 μL of Halt™ Protease Inhibitor Cocktail [100× (Thermo Scientific, # 78429) in 1 mL buffer] were added. Samples were homogenized with a Polytron and centrifuged at 12 000 g for 15 min. Supernatant was collected and stored at −20°C until use.
Proteins were separated by electrophoresis on 4–20% sodium dodecyl sulfate-polyacrylamide gels and transferred electrophoretically onto polyvinylidene difluoride membranes. The total protein on membrane was determined by using No-Stain Protein Labeling Reagent (Invitrogen, #A4449) and detected with the Bio-Rad ChemiDoc Imaging System. Tris-buffered saline (TBS) containing 0.2% (volume/volume) Tween-20 and 5% (weight/volume) non-fat milk were used to block the membranes. Membranes were then incubated overnight at 4°C with primary antibodies diluted in TBS-containing 0.2% Tween. Membranes were washed with TBS-Tween three times for 15 min at room temperature. The membrane was further incubated with secondary antibodies conjugated to horseradish peroxidase for 1 h and underwent three extensive additional 15-min washes at room temperature. Membranes were exposed to the SuperSignal™ West Pico PLUS Chemiluminescent Substrate (Thermo Scientific, #34577) and chemiluminescent signal was detected with the Bio-Rad ChemiDoc Imaging System.
Protein levels were quantified with Image Lab Software (Bio-Rad, Hercules, CA, USA). All expression data are expressed relative to total protein from the same samples on the same blots.
2.11 Digital PCR (human samples)
Total RNA was extracted from human RAAs using the RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. Subsequently, 500 ng RNA was transcribed into cDNA using the reverse transcription kit (Applied Biosystems; Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s instructions. TaqMan probes were used for hydroxymethylbilane synthase (HMBS: Hs00609296_g1), beta-2-microglobulin (B2M: Hs99999907_m1), GATA-binding protein 4 (GATA4: Hs00171403_m1), glyceraldehyde-3-phosphate dehydrogenase (GAPDH: Hs02758991_g1) as housekeeping genes, cyclin dependent kinase inhibitor 1A (CDKN1A: Hs00355782_m1), CDKN2A: Hs00923894_m1, TP53: Hs99999147_m1, Galactosidase Beta 1 (GLB1: Hs01035168_m1), growth differentiation factor 15 (GDF15: Hs00171132_m1), transforming growth factor beta-2 (TGFB2: Hs00234244_m1), insulin-like growth factor-binding protein 5 (IGFBP5: Hs00181213_m1), IGFBP7: Hs00266026_m1, and matrix metalloproteinase 2 (MMP2: Hs01548727_m1). The reactions were run on a QIAcuity Digital polymerase chain reaction (PCR) System (Qiagen, Hilden, Germany) using a 96-well Nanoplate with 8500 partitions and the QIAcuity master mix according to the manufacturer’s instructions. The cycling programme used for the reactions: 2 min at 50°C, followed by 10 min at 95°C, a total of 35 cycles (15 s at 95°C and 1 min at 60°C), and 30 s at 60°C. The relative messenger RNA (mRNA) levels to controls were calculated from the copies per microliter values and normalized to the geometric mean of HMBS, B2M, GATA4, and GAPDH.
2.12 qPCR (animal tissues)
Tissue samples were freshly isolated on ice, snap-frozen in liquid nitrogen, and homogenized in QIAzol Lysis Reagent. Extraction of RNAs was performed with RNeasy Mini Kit (Qiagen, Hilden, Germany). Reverse transcriptase PCR (RT-PCR) was performed with Applied Biosystems Thermal Cycler Step One Plus (Thermo Fisher Scientific, Waltham, MA, USA). TaqMan probes were used for p16 (Cdkn2a: Rn00580664_m1), p21(Cdkn1a: Rn00589996_m1), Glb1: Rn01403379_m1, B2m: Rn00560865_m1, Gapdh: Rn01775763_g1, Hprt1: Rn01527840_m1, Gdf15: Rn00570083_m1, Igfbp3: Rn00561416_m1, Igfbp5: Rn00563116_m1, colony stimulating factor 2 (Csf2: Rn01456850_m1), interlukin-6 (Il-6: Rn01410330_m), Serpine2: Rn01400467_m1, p53 (Tp53: Rn00755717_m1), chemokine (C–C motif) ligand 2 (Ccl2: Rn00580555_m1), Collagen1a1 (Col1a1: Rn01463848_m1), Collagen3a1 (Col3a1:Rn01437681_m1), and SYBR green primers for matrix metalloproteinase 9 (Mmp9) (forward: TCCAGTAGACAATCCTTGCAATGTG; reverse: CTCCGTGATTCGAGAACT TCCAATA), C-X-C Motif Chemokine Ligand 1 (Cxcl1) (forward: ACTCAAGAATGGTCGCGAGG; reverse: ACGCCATCGGTGC AATCTAT), tumour necrosis factor-alpha (forward: GTGATCGGTCCCAACAAGGA; reverse: CTTGGTGG TTTGCTACGACG), insulin-like growth factor-binding protein 4 (Igfbp4) (forward: GGAGCTGTCGGAAATCGAAG; reverse: GAAGCTGTTGTTGGGATGCTC), chemokine (C–C motif) ligand 20 (Ccl20) (forward: TTCACAACACAGATGGCCGA; reverse: CAGCGCACACGGATCTTTTC), insulin-like growth factor-2 (Igf2) (forward: CGCTTCAGTTTGTCTGTTCGG; reverse: GGCCTGAGAGGTAGACACG), and transforming growth factor-2 Tgfß2 (forward: CCATGACATGAACCGACCCT; reverse: TGCCG TACACAGCAGTTCTT). All mRNA values were normalized to the geometric mean of B2m, Hprt1, and Gapdh.
2.13 Statistical analysis
Statistical analyses were performed using GraphPad Prism 9 (San Diego, CA, USA) and SAS release 9.4 (SAS Institute Inc., Cary, NC, USA). For normally distributed data (assessed by Shapiro–Wilk test), we used Student’s t-test (for two-group only comparisons), one-way analysis of variance (ANOVA) (followed by Dunnett’s test), or two-way ANOVA (followed by t-tests with Bonferroni’s multiple comparison test). Homogeneity of variances among groups was verified with the Brown–Forsythe test. For non-normal data, Kruskal–Wallis or Mann–Whitney tests were used. Categorical variables (like AF inducibility) were analysed with Fisher’s exact test. Results are expressed as mean ± SEM and two-tailed P < 0.05 was considered statistically significant.
3. Results
3.1 Effects of aging and MI on AF vulnerability, electrophysiology, and fibrosis
Transoesophageal stimulation induced AF (see Supplementary material online, Figure S1) in all aged rats (100%, P < 0.0001 vs. control) and the majority of MI rats (87.5%, P = 0.001) but in only one rat in the young control group (10%, Figure 1A). Because of the rare inducibility of AF in young control rats, AF duration was not meaningfully quantifiable in this group (see Supplementary material online, Figure S1D). Figure 1B shows representative activation maps for each group. LA conduction velocity was significantly slower in young MI and aged rats. APD50 and APD80 were significantly greater in aged rats, while they did not show significant changes in young MI (Figure 1C; Supplementary material online, Figure S2B, original recordings shown in Supplementary material online, Figure S2C). RA conduction velocity and RA repolarization were significantly slower in aged rats (see Supplementary material online, Figure S2A). ECG analysis revealed significant increases in P-wave duration and P–R interval in both MI and aged rats (see Supplementary material online, Figure S3A and B), consistent with atrial conduction slowing. QRS duration was not significantly different among groups (see Supplementary material online, Figure S3C), while QT intervals were significantly increased in aged rats (see Supplementary material online, Figure S3D). LA myocardial collagen content was significantly increased, indicating tissue fibrosis, in MI and aged rats compared with young control rats (Figure 1D). RA myocardial collagen content was significantly greater in aged rats compared with young control rats (see Supplementary material online, Figure S4A).
AF inducibility, optical mapping and fibrosis quantification in LA of young control (3 months), young MI (3 months), and aged (20 months) rats. (A) Percentage of rats with inducible AF (n = 8–10; Fisher’s exact test). (B) LA activation maps at BCL 100 ms from a control, young MI, and elderly rat. (C) Mean ± SEM LA conduction velocity and APD50 (n = 6–7; two-way ANOVA followed by Bonferroni test; the P-values indicate Bonferroni-corrected differences between groups from data pooled from all BCLs). (D) Dot-plot graphs show mean ± SEM percentage fibrosis in LA. Points represent results from an individual animal (n = 5–6; one-way ANOVA with Dunnett’s test). AF, atrial fibrillation; APD50, action potential duration to 50% repolarization; BCL, basic cycle length; LA, left atrium; MI, myocardial infarction; SR, sinus rhythm.
3.2 Cardiac function and structure in aged and MI rats
Echocardiography revealed a significant reduction in LA FS in both aged and MI rats (see Supplementary material online, Figure S5A) and increased LA dimensions in aged but not MI rats (see Supplementary material online, Figure S5A). E-wave deceleration time and E/A wave ratio were not altered significantly (see Supplementary material online, Figure S5B), but LV FS% and LV ejection fraction (LV EF%) were reduced and LV diastolic dimension (LVDd) was increased with both MI and aging (see Supplementary material online, Figure S5C). LV systolic dimension (LVDs) increased in aged but not MI rats (see Supplementary material online, Figure S5C). LV mass normalized to body weight was reduced in aged rats but not MI rats (see Supplementary material online, Figure S5C). TAPSE suggested right ventricular dysfunction in MI, and RA dimension was increased in aged rats (see Supplementary material online, Figure S5D). Overall, the most consistent changes were reduced LA emptying function, impaired LV systolic function, and LV diastolic dilation in both aged and MI animals.
3.3 Expression of senescence markers in MI and aged rats
The mRNA expression of senescence markers was measured by quantitative PCR (qPCR) and was increased for p16 in LA of both MI and aged rats (Figure 2A), while no significant change was noted for p21, p53, or Glb1. Immunohistochemical analysis revealed increased protein expression of p16 in LA of MI and aged rats (Figure 2B). In LA, among typical SASP markers, upregulation was noted for Igfbp3, (in MI rats), Csf2 (in aged rats), Mmp9 (in both MI and aged rats), and Serpine2 (in aged rats; Figure 3B and Supplementary material online, Figure S6). In RA, the mRNA expression of the senescence marker p16 and SASP markers Csf2, Serpine2, and Mmp9 were upregulated in aged rats compared with the young control group, while there were no changes in the MI rats (Figure 3A–C). Other SASP markers did not differ among groups (see Supplementary material online, Figure S7).
Senescence marker expression in LA. (A) mRNA expression (n = 6; one-way ANOVA with Dunnett’s test); (B) immunofluorescence for p16 (red) and DAPI (blue; n = 4; one-way ANOVA with Dunnett’s test). DAPI, 4′,6-diamidino-2-phenylindole; LA, left atrium; MI, myocardial infarction; qPCR, quantitative polymerase chain reaction; RA, right atrium.
Senescence marker expression in RA and components of SASP in LA and RA. (A) RA senescence marker mRNA expression (n = 6; one-way ANOVA with Dunnett’s test); (B) LA SASP component mRNA expression (n = 6; one-way ANOVA with Dunnett’s test); (C) RA SASP component mRNA expression (n = 6; one-way ANOVA with Dunnett’s test). LA, left atrium; MI, myocardial infarction; RA, right atrium.
3.4 Effects of senolytic therapy on AF substrate and cardiac function
The data described above suggest senescent-cell accumulation in the atria of both MI and aged rats. To test the role of cellular senescence in AF occurrence, we administered a senolytic drug combination (D + Q) to MI rats, beginning the day of MI (Figure 4A). D + Q significantly reduced AF inducibility 28 days post-MI (Figure 4B). Transoesophageal stimulation (performed and analysed blinded to group identity) induced AF in 89% of MI rats treated with vehicle, but none of the MI rats were treated with D + Q or sham rats (P < 0.0001; Figure 4B). The mean duration of AF in the MI vehicle group was 25 ± 11 s, but limited or no inducibility could not be quantified in sham vehicle, sham D + Q, or MI D + Q groups. In D + Q-treated MI rats, there were no significant changes observed in P-wave duration. However, we observed significant reductions in the PR interval (see Supplementary material online, Figure S3E–H). On echocardiography, D + Q treatment had no effects on LA or LV function or structure (see Supplementary material online, Figure S8).
Study design, AF inducibility changes, optical mapping, and fibrosis quantification with D + Q senolytic therapy. (A) Study design; (B) AF inducibility; (C) LA activation maps at BCL 100 ms from a sham, an MI vehicle, and a MI D + Q rat; (D) mean ± SEM LA conduction velocity and APD50 (n = 5–6; two-way ANOVA with Bonferroni test; the P-values indicate Bonferroni-corrected differences between groups from data pooled from all BCLs); and (E) fibrosis analysis. Dot-plot graphs show mean ± SEM LA fibrous tissue content (n = 5; two-way ANOVA with Bonferroni test). Points represent results from individual animals. AF, atrial fibrillation; APD50, action potential duration to 50% repolarization; BCL, basic cycle length; D + Q, dasatinib and quercetin; MI, myocardial infarction; SR, sinus rhythm.
Figure 4C shows representative activation maps for each group. LA conduction velocity was significantly slower in MI groups vs. sham. D + Q significantly attenuated the LA conduction velocity slowing caused by MI (Figure 4D) while reducing changes in LA APD80 (see Supplementary material online, Figure S9). There was no significant difference between the vehicle and treatment groups in LA APD50 (Figure 4D). Consistent with the changes in LA conduction, LA fibrosis was significantly reduced in the MI group treated with D + Q compared with the MI vehicle group (D + Q: 2.4 ± 0.2% vs. vehicle: 5.4 ± 0.8%, P = 0.001, Figure 4E), as was RA fibrosis (see Supplementary material online, Figure S4B).
3.5 Effects of senolytic therapy on senescence markers
The expression of selected genes encoding proteins involved in senescence (Figure 5A), fibrosis (Figure 5B), and SASP (Figure 5C) was examined by qPCR.
LA fibrosis, cell senescence, and SASP marker gene expression without and with D + Q therapy. mRNA expression for (A) senescence markers p16, p21, p53, and Glb1 (n = 5–8; two-way ANOVA with Bonferroni test), (B) fibrosis markers Mmp2, Col 1a1, Col 3a1, and Tgfβ2 (n = 5–8; two-way ANOVA with Bonferroni test), and (C) SASP components Igfbp4, Ccl20, Serpine2, and Igf2 (n = 5–8; two-way ANOVA with Bonferroni test). Ccl20, chemokine (C–C motif) ligand 20; Igf2, insulin-like growth factor 2; Igfbp4, insulin-like growth factor binding protein-4; Mmp2, matrix metalloproteinase 2; SASP, senescence-associated secretory phenotype; Tgfβ2, transforming growth factor 2.
The mRNA expression of p16 in LA was highly variable, and while it increased quantitatively in MI vehicle and not in MI D + Q, the differences were not significant after Bonferroni correction (Figure 5A). The expression of p21 and Serpine2, Igfbp4, and Ccl20 mRNA was significantly reduced by D + Q treatment in MI rats (Figure 5A–C). Other SASP markers with no changes in LA and RA are shown in Supplementary material online, Figures S10 and S11. The protein expression of selected senescence and SASP markers was assessed by western blot. The expression of p16 and p21 protein was significantly greater in the MI vehicle vs. sham group. Their expression levels decreased quantitatively in D + Q-treated MI rats and were no longer significantly different from sham; however, the differences in p16 and p21 between D + Q- and vehicle-treated rats did not reach the nominal significance level (see Supplementary material online, Figure S12). The protein expression levels of p53, Ccl20, Serpin2, Mmp2, and Mmp9 were significantly higher in the MI vehicle and were decreased significantly with D + Q (see Supplementary material online, Figure S12). Original full-length immunoblot images are provided in Supplementary material online, Figures S13–S16.
In the LV, p16 mRNA expression was upregulated in the peri-infarct and infarct zones of the MI rats and significantly reduced by D + Q in the infarct zone (see Supplementary material online, Figure S17). The mRNA expression of Mmp2, a profibrotic SASP marker, was upregulated in MI vehicle rats compared with sham vehicle and substantially reduced in the MI D + Q group (Figure 5B). The mRNA expression of Col1a1 was significantly reduced by D + Q treatment vs. vehicle (Figure 5B).
To quantify senescence in specific cardiac cell types, we double-stained cross-sectioned samples with p16 along with cell-type selective markers in LA tissues of MI rats (Figure 6). D + Q treatment did not alter the number of vimentin-positive cells expressing p16 but significantly reduced the percentage of p16-expressing cells in the CD31-expressing (endothelial cell) population. In sections not including blood vessels, αSMA-positive cells expressing p16 were significantly reduced by D + Q, while the number of troponin I-positive cells expressing p16 was unaffected. These results suggest that D + Q effects resulted in significant clearance of senescent myofibroblasts and endothelial cells in post-MI LA but not quiescent fibroblasts or cardiomyocytes.
Double immunofluorescence for p16 and cardiac cell-type markers in LA of sham and MI rats treated with vehicle or D + Q. (A) Fibroblast marker, vimentin (n = 4–5; unpaired t-test). (B) Endothelial cell marker, CD 31 (n = 4–5; unpaired t-test). (C) Myofibroblast marker, αSMA (n = 5–8; unpaired t-test). (D) Cardiomyocyte marker, troponin I (n = 8; unpaired t-test). All panels show cell marker staining (green), staining for p16 (red), and DAPI (DAPI, blue). D + Q, dasatinib and quercetin; DAPI, 4′,6-diamidino-2-phenylindole; MI, myocardial infarction.
3.6 Senescence markers in human atrial tissues
To test the potential translational relevance of our findings, we analysed senescence markers in human RAAs. RAA samples from older patients showed significantly higher protein expression of p16 and p21 (Figure 7A). Human RAAs also showed significantly greater mRNA expression of many senescence and SASP markers including p16, p21, GLB1, and GDF15 genes with aging (Figure 7B and Supplementary material online, Figure S18). The RAAs of cAF patients showed significantly greater gene expression for TGFβ2 and protein expression for p21 than for SR patients (Figure 7C and D). Original full-length immunoblot images are provided in Supplementary material online, Figures S19–S21. While there were no statistically significant differences in clinical variables for SR vs. cAF patients in Supplementary material online, Table S2, body mass index (BMI) had P-values between <0.1 and >0.05. To determine whether BMI differences might have affected the results, we assessed the correlations between BMI and senescence marker expression and found none (see Supplementary material online, Figure S22).
Senescence marker mRNA and protein expression in human atrial tissue. (A) Immunoblot analysis of senescence markers, p16 and p21, in RAAs of younger (≤ 60) and older (≥70) patients (n = 15–19; unpaired t-test for parametric or Mann–Whitney test for non-parametric variables). (B) mRNA expression measured by dPCR in RAA of younger (≤60) and older (≥70) patients (n = 15–19; unpaired t-test for parametric or Mann–Whitney test for non-parametric variables). (C) Immunoblot analysis of senescence markers, p16, and p21 in RAA of SR and cAF patients (n = 11; unpaired t-test for parametric or Mann–Whitney test for non-parametric variables). (D) mRNA expression, as measured by dPCR in RAA of SR and cAF patients (n = 6; unpaired t-test for parametric or Mann–Whitney test for non-parametric variables). cAF, chronic atrial fibrillation; dPCR, digital polymerase chain reaction; GFD15, growth differentiation factor 15; RAA, right atrial appendage; SR, sinus rhythm; TGFβ2, transforming growth factor 2.
4. Discussion
In this study, we examined the role of cellular senescence in the pathophysiology of AF using rat models of atrial remodelling associated with aging and MI, as well as human atrial tissue. We found evidence for the accumulation of senescent cells in the atria of both elderly rats and rats with LV dysfunction due to recent MI, accompanying increased AF susceptibility and atrial fibrosis. As a test of the pathophysiological role of cell senescence, we used senolytic therapy (D + Q) to clear senescent cells and found that it attenuated the atrial conduction and fibrotic changes caused by MI, along with the associated AF vulnerability. We evaluated the potential relevance of cell senescence to human AF by examining senescent-cell markers in human atrial tissue, noting enhanced expression of multiple markers in elderly patients and those with long-standing persistent AF. Our results provide direct evidence supporting the involvement of atrial cell senescence in producing an AF substrate due to LV dysfunction post-MI and provide indirect evidence relevant to aging-related AF by showing that cellular senescence is detectable both in rats showing an aging-related AF substrate and in elderly humans.
4.1 The association of AF with aging and MI
Age is well known to be the single most important determinant of the risk of AF3 and recent work points to a primary role of biological age.23 The available literature indicates that aging-related AF results from multiple structural, electrophysiologic, and molecular changes caused by aging.3,4 Among these changes, the potential contribution of atrial fibrosis to aging-related AF has been confirmed in several clinical and experimental studies.24–26 Progressive atrial conduction abnormalities occur with aging in humans27,28 and are believed to be important contributors to increasing susceptibility to AF with age.
While the accumulation of senescent cells is a hallmark of aging, accelerated cell senescence is also believed to play a pathophysiological role in a range of cardiac disease conditions, including heart failure, MI, drug-induced cardiotoxicity, and ischaemia reperfusion injury.6 Almost all the work to date has been performed in ventricular tissues and cell lines. The present work provides the novel insight that MI-induced LV dysfunction causes an accumulation of senescence markers in the atria and that senolytic-cell clearance prevents adverse atrial remodelling and the development of an AF-promoting substrate.
4.2 Potential role of cellular senescence in increased AF susceptibility
While two clinical studies have shown that increased atrial expression of senescence markers accompanies AF progression and atrial fibrosis,15,16 we are unaware of any studies directly evaluating the role of cellular senescence in experimental AF. Further work is needed to establish more clearly how cellular senescence affects AF. One potential avenue is clearly via the induction of tissue fibrosis, which plays a significant role in AF26 and is well documented to be associated with aging-related AF susceptibility.4 This notion is consistent with the prevention of atrial fibrosis that we noted with senolytic therapy in MI rats and with the associated clearance of senescent αSMA-positive cells. However, the role of senescent cells in cardiac fibrosis is complex and both beneficial and detrimental effects have been reported.6 Whether cellular senescence can contribute to other recognized AF-promoting mechanisms like atrial connexin dysregulation,29 Ca2+-handling abnormalities,30 or ion current abnormalities4 remains to be established. Furthermore, it will be important to clarify to what extent the effects of cellular senescence are due to the cells directly affected, as opposed to paracrine actions on neighbouring cells due to SASP mediators released by senescent cells.
4.3 Senolytic therapy
A variety of senolytic compounds have been documented experimentally to clear senescent cells; some of these are presently under clinical investigation.6 Senolytic targets anti-apoptotic pathways such as BCL-2 and phosphoinositide 3-kinase (PI3K)–AKT in senescent cells, thus inducing targeted senescent-cell death.31 The global elimination of senescent cells with navitoclax, a senolytic compound, in aged mice reduces ventricular fibrosis and hypertrophy.18 The administration of navitoclax prior to and following MI improved adverse LV remodelling and post-MI survival in elderly mice.32 We chose to use D + Q because of promising results with this combination in preclinical studies and ongoing clinical trials.6 For example, the administration of D + Q in aged mice reduced LV fibrosis.17 D + Q prevented mitochondrial DNA-induced inflammation and prolonged the survival of cardiac allografts.33
4.4 Potential role of senescent myofibroblasts and endothelial cells in AF
Double staining with p16 and cell-type selective markers (Figure 6) points to myofibroblasts and endothelial cells as senescent cell types cleared by D + Q in post-MI LA. Atrial fibroblasts from patients with cAF show significant myofibroblast differentiation34; however, the role of senescent myofibroblasts in AF has not been studied. Meyer et al.35 showed that myofibroblasts are the predominant cardiac cell population undergoing senescence in the LV of mice with transverse aortic constriction (TAC). These researchers showed that genetic ablation of p53 or p16 results in reduced senescence, associated with increased LV fibrosis. In contrast, we found that clearance of senescent myofibroblasts with D + Q was associated with reduced LA fibrosis. The differences between studies may be due to differences between atrial and ventricular responses or to technical differences like senolytic or genetic approach, models used (TAC vs. MI), or species studied (mouse vs. rat). The role of endothelial cells is unknown but might be related to the generation of fibrosis via endothelial–mesenchymal transition or via secreted profibrotic SASP products.36,37
4.5 Limitations
We studied the effects of senolytic therapy on atrial pathology underlying the AF substrate induced by LV dysfunction resulting from LV dysfunction as caused by acute MI. Our findings support the concept that pathologies can cause cardiac abnormalities by inducing accelerated cellular senescence.6 While we showed that aging atria demonstrate accumulation of cellular senescence, we did not study directly the effects of senolytic therapy on aging-related atrial remodelling. We had access to a limited number of aged rats, which allowed us to carry out the initial phase of the study supporting the concept that cellular senescence contributes to AF susceptibility post-MI and with aging. However, when we performed the follow-up senolytic study, we were unable to obtaining enough aged rats to conduct the desired experiments. It would be of interest to determine whether senolytic interventions can also prevent senescent-cell accumulation and AF with aging.
While D + Q is a well-established senolytic therapy,38 the risk of off-target drug effects can never be fully discounted. It would be interesting to evaluate the effects of senescent-cell clearance in genetically engineered animals on atrial remodelling.39
The cell type mediating the effects of cellular senescence on cardiac pathology is an important, but often ignored, issue.6 We used immunohistochemistry to address this question and found evidence for a possible role of myofibroblasts and endothelial cells (Figure 6). We recognize that our identification of the cell types mediating the response to senolytics is not definitive. In vitro experiments could be used to investigate further the role of fibroblasts vs. myofibroblasts, but we found that senescent fibroblasts failed to proliferate in culture (consistent with cell-cycle arrest due to senescence). In vitro senescence induction (e.g. with irradiation or doxorubicin) could be used in cultured non-senescent fibroblasts and myofibroblasts, but the relationship of these artificially induced forms of senescence to naturally occurring in vivo senescence is unclear.
Our young MI and control rats were studied at 3 months; our aged rats at 20 months. It is difficult to extrapolate directly from rat age to human age. In a recent review,39 the results of a survey completed by 611 labs were reported. Ages of rats studied as ‘adult’ ranged from 8 to 16 weeks, with 8–12 weeks being the most commonly used.40 A Medline search with the term ‘aged rat AF’ identified six studies since 2011, with rat ages in the aged groups ranging 11–24 months (mean 20.4),25,41–45 so our aged rats (20 months) were very close to the mean value in the literature. Typically, 20-month-old rats are considered equivalent to 60-year-old humans.40 This must be considered in relating our findings in elderly rats to our human studies (Figure 7).
Our human studies were conducted in RAA tissues. Human LAA tissue is rarely available from cardiac surgical procedures, whereas a piece of the RAA is routinely removed at cannulation for extracorporeal circulation. The very limited number of human LAA samples available did not allow meaningful analysis on LAA. We acknowledge that LA and RA might have different senescence-related characteristics, so extrapolations should be cautious. For the rat work, we focused on the LA because coronary artery ligation causes primarily LV infarction and dysfunction and the LA is more strongly affected than the RA.46 Nevertheless, we obtained and presented rat RA data for fibrosis, optical mapping, and qPCR (Figure 3; Supplementary material online, Figures S2, S4, S7, and S11). The changes in RA p16 mRNA levels in aged rats are similar to those in older humans. The human studies have the inevitable limitations of clinical variability and sample availability. There are of course many important species-related differences that need to be considered.47
In MI rats, RA fibrosis was reduced with senolytic therapy but senescence markers were not significantly changed, which is not surprising given that the RA changes are well upstream of LV dysfunction, which much more directly affects the LA.46 Senolytic therapy had a moderate effect on conduction velocity in MI rats, whereas its impact on AF inducibility was more striking. This may reflect a non-linear relationship between conduction changes and AF or the role of other arrhythmia determinants.
No alterations occurred in indices of ventricular function with senolytic treatment (see Supplementary material online, Figure S17). In contrast, atrial fibrosis, conduction, and susceptibility to AF were strongly affected. However, p16 levels in the MI zone were reduced by senolytics. Therefore, while ventricular function was unchanged, we cannot exclude the possibility that some remote or neurohumoral factors related to ventricular senescence but unrelated to ventricular function might have contributed to atrial tissue protection.
The specific molecular pathways involved in cell senescence might differ in different contexts like LV dysfunction vs. aging and/or in humans vs. rats. Further work is needed to clarify the specific molecular pathways, signals, and mediators involved and to assess how senescent-cell clearance by D + Q affects them.
The tissue samples utilized in the initial phases of the study (comparison of young, MI, and aged groups) were over 3 years old when we undertook western blot studies, leading to significant protein degradation that made the western blots unfeasible. Consequently, we could not perform western blot analyses for these studies.
5. Conclusions
In the present study, we obtained evidence for the accumulation of senescent cells in the atria of aged rats and in the atria of rats with LV dysfunction due to MI, in association with a pathological AF substrate. Furthermore, we found that senolytic therapy was able to suppress markers of cellular senescence, prevent atrial fibrosis, and obviate AF substrate development in post-MI rats. The potential translational relevance of these findings is supported by the observation of overexpression of senescent-cell markers in RA tissue from older humans and those with long-standing persistent AF. These findings have important potential implications for understanding the mechanisms linking cardiac pathology to AF occurrence and for the development of innovative therapeutic approaches.
Supplementary material
Supplementary material is available at Cardiovascular Research online.
Authors’ contributions
M.M. designed work with S.N., performed all the animal experiments, analysed the results, prepared the figures, and wrote the paper. P.N. performed the qPCR experiments and was involved in immunofluorescence troubleshooting and assisted with data interpretation. I.H.A.-T. and A.S. performed the experiments and analysed the data related to western blots and dPCR in human samples. M.K. performed the human surgeries, provided the human atrial tissue, and interpreted the clinical data. R.H. performed the AF inducibility experiments. F.X. analysed optical mapping experiments. J.X performed and analysed the western blot experiments. N.V. analysed the immunofluorescence data. E.T. and G.F. assisted with design and data interpretation. J.C.T. supervised the experiments and blind analyses of echocardiography. M.G.S. and J.F.T. co-supervised the experiments and blind analyses for histological experiments. D.D. conceptualized, supervised, and funded the human sample experiments. S.N. conceived and funded the project, supervised all Montreal Heart Institute trainees and technical staff, including particularly Dr Mehdizadeh, throughout the project, and was intimately involved in the preparation, writing, and finalization of the paper.
Acknowledgements
The authors thank Anna Nozza for biostatistical advice and data analysis, Nathalie L’Heureux, Chantal St-Cyr, Yanfen Shi, Marie-Elaine Clavet-Lanthier, Colombe Roy, Joanne Vincent, Simone Olesch, and Ramona Löcker for their technical support, and Lucie Lefebvre and Jennifer Bacchi for secretarial assistance.
Funding
This study was supported by the Canadian Institutes of Health Research (148401 to S.N.), Heart and Stroke Foundation of Canada (22-0031958 to S.N.), Fonds de Recherche du Quebec- Santé (doctoral-training scholarship to M.M.), and the National Institutes of Health (R01HL164838, R01HL136389, R01HL163277, R01HL131517, R01HL08959, R01HL160992, and R01HL165704 to D.D.), the European Union (large-scale network project MAESTRIA No. 965286 to D.D.).
Data availability
All raw data supporting the findings from this study are available from the corresponding authors upon reasonable request.
References
Here, we show that cellular senescence, a stress response, is an important contributor to atrial fibrillation (AF) pathophysiology. More specifically, we noted enhanced expression of senescence markers in the atria of aged rats and rats with left ventricular dysfunction due to myocardial infarction (MI), along with elderly patients and those with persistent AF. Senolytic therapy reduced AF susceptibility in MI rats. Since senolytic drugs are already being studied in clinical trials, and since there are senescence-related molecular targets potentially amenable to biologic therapies, this work may open up new treatment opportunities for AF.
Author notes
Conflict of interest: none declared
