In-depth characterization of a mouse model of postoperative atrial fibrillation

INTRODUCTION

Atrial fibrillation (AF) is the most common arrhythmia and affects 4% of adults older than 60^[1]. Common risk factors for AF include hypertension, valvular heart disease, heart failure, obstructive sleep apnea, obesity, and chronic kidney disease^[2]. One important risk factor for AF is surgery, and AF developing during the days following surgery is termed postoperative AF (POAF). POAF typically occurs 1-3 days after surgery^[3] and portends a 5-fold higher risk of chronic AF^[4] and a 2-fold higher risk of stroke and mortality^[5,6]. Rates of POAF are 15%-42% after cardiac surgery^[7] and 10%-20% after non-cardiac surgery^[6,8,9].

POAF is thought to arise due to transient triggers such as autonomic activation and post-surgical inflammation acting on a preexisting substrate to promote triggered activity and reentry^[9]. Evidence for the role of autonomic activation comes from the significant reduction in POAF with beta-blocker therapy^[10] as well as higher circulating norepinephrine^[11] and aldosterone^[12] levels observed in POAF patients. Evidence for the role of inflammation is less consistent among experimental studies, but higher preoperative and postoperative white blood cell counts are seen in POAF patients^[13]. Structural and cellular alterations such as left atrial fibrosis^[14] and decreased atrial autophagy^[15] have also been linked to POAF development in mice.

Experimental studies of POAF are limited. Human atrial tissue samples provide information about predisposing substrate but are typically collected during the surgical procedure days prior to POAF development and thus do not provide information about the substrate at the time of arrhythmia initiation. Current animal models of POAF include sterile pericarditis, pericardiotomy, and atriotomy^[16-18]. However, unlike in the clinical setting with patients undergoing cardiac surgery, the currently available mouse models do not account for the ischemic and hemodynamic insults inherent to cardiac surgery. Moreover, to our knowledge, a comprehensive and powered analysis of the electrophysiologic (EP) and AF inducibility of these models in a large cohort of mice has not been conducted. Here, we describe a model of POAF consisting of a bi-atrial pericardiectomy with cross-clamping of the thoracic aorta to simulate cardiothoracic (CT) surgery. We systematically characterize the AF inducibility and inflammatory EP signature in our POAF model using over 100 wild-type mice.

MATERIALS AND METHODS

Mouse colonies: All experiments were carried out using wild-type male and female C57BL/6J mice purchased from Jackson Labs (strain #000664).

Experimental protocols: Experiments were carried out according to NIH Guidelines on the Use of Laboratory Animals. All procedures were approved by the Baylor College of Medicine Institutional Animal Care and Use Committee, protocol number AN-4044. A total of 131 mice (48% male) were included in this study (45 Sham, 86 THOR). All animals were given one injection of buprenorphine subcutaneously at a dose of 1 mg/kg at least 1 h before surgery. Animals were monitored every day after surgery until terminal EP studies were performed 72 h after the surgery. Meloxicam, which is often administered every 24-48 h postoperatively in mice, was withheld per protocol due to the potential confounding effect of its non-steroidal anti-inflammatory mechanism of action on the pathways involved in POAF (see Discussion).

Animal model: Our mouse model of POAF was designed to recapitulate CT surgery [Figure 1]. Anesthesia was induced by placing mice in an induction chamber filled with 4% isoflurane in 100% oxygen until the mice lost consciousness. Mice were intubated and ventilated at a tidal volume of 150 µL and respiratory rate of 175 breaths/minute, and anesthesia was maintained using 2.5% v/v isoflurane in 100% oxygen. Body temperature was maintained at 37.0 °C ± 0.5 °C using a heated platform guided by a rectal thermometer. After dissection through the skin and subcutaneous tissue, the pectoralis major and minor muscles were separated laterally and the thoracic cavity was exposed through the 2nd or 3rd intercostal space [Figure 1A]. Bi-atrial pericardiectomy [Figure 1B and C] and cross-clamping of the thoracic aorta for 20 s [Figure 1D] were performed. For the sham procedure, the same skin and muscle dissection was performed, but the chest cavity was not entered.

Figure 1. Cardiothoracic surgery workflow. (A) Schematic illustration of exposed chest cavity following dissection through 2nd or 3rd intercostal space. (B) Fibrous pericardium covering the left atrium prior to pericardiectomy. (C) Status after bi-atrial pericardiectomy for both atria, with pericardium removed. (D) Cross-clamping of ascending aorta. Ao: Aorta; LA: left atrium; LAD: left anterior descending coronary artery; RA: right atrium; RCA: right coronary artery.

Programmed electrical stimulation: 72 h after CT surgery, AF inducibility was analyzed by performing a right heart catheterization to conduct intracardiac pacing as previously described^[19]. Briefly, the mouse was anesthetized (1.75% v/v isoflurane in 100% oxygen). The subcutaneous tissue superficial to the right external jugular vein was dissected, and proximal and distal control of the right external jugular vein was secured with 6-0 sutures. A 1.1F octapolar catheter (EPR-800, Millar Instruments, Houston, TX) was then inserted into the right external jugular vein, through the superior vena cava, and into the right atrium and ventricle. Correct placement of the catheter was verified by looking for the appropriate P wave deflection after intra-atrial pacing and the appropriate QRS deflection after intra-ventricular pacing. All EP protocols were performed at 1.75% v/v, isoflurane/oxygen and rectal temperature between 37.0 °C ± 0.5 °C.

Following baseline recordings, the sinus node recovery time (SNRT) was determined by pacing the right atrium at a basic cycle length (BCL) of 100 milliseconds (ms) for 15 s and then measuring the amount of time from the last pacing impulse until the first spontaneous sinoatrial node beat^[20]. The SNRT was repeated at a BCL of 90 ms. The atrial effective refractory period (AERP) was determined by applying a series of atrial pacing trains at a BCL of 100 ms, each with a coupled premature stimulus at the end (referred to as S2). The S1-S2 was decreased by 2 ms from 70 ms to 20 ms. The AERP was defined as the longest S1-S2 interval that failed to propagate into the ventricles. AF inducibility was determined by performing the A-burst protocol^[21], which applies a series of 2-s bursts starting at a BCL of 40 ms. Each 2-s burst decreases the BCL by 2 ms until a BCL of 20 ms is reached. The burst pacing protocol was performed in triplicate per mouse and AF inducibility was defined as a positive AF event lasting at least one second in at least two (out of three) positive events. A positive AF event was defined by irregularly irregular RR intervals without discernable P waves on electrocardiogram (ECG).

RNA isolation and cDNA preparation: Total RNA was isolated from the atrial appendages of six Sham mice without AF and six THOR mice with AF. Tissue was harvested immediately after EP pacing. Both the left and right atrial appendages were used for all samples. After harvesting, the heart underwent retrograde perfusion with cold phosphate-buffered saline. Atrial samples were then frozen in liquid nitrogen for 5 min and stored at -80 °C. To isolate total RNA, atrial samples were ground up and then 200 µL of TRIzol (Ambion) was added to each sample. Samples were spun down the samples for 60 s at 16,000 rpm. The Direct-zol RNA MiniPrep (Zymo Research) was used to isolate total RNA. RNA concentrations were measured using a nanodrop and stored at -80 °C. To generate cDNA, reverse transcription was carried out using the iScript Reverse Transcription Supermix (BioRad) with 500 ng of RNA per sample.

qPCR: Quantitative polymerase chain reaction (qPCR) was conducted for IL-6, IL-1, IL-18, IL-10, interferon gamma (IFN-γ), tumor necrosis factor alpha (TNF-α), and TGF-β1 with the primers listed in Supplementary Table 1. Three technical replicates were performed per sample. Six individual biological replicates were used per group (six Sham and six THOR mice). A total of 10 µL reaction solution was used per well: 5 µL of SYBR Green Master Mix (Thermo Fisher Scientific), 0.5 µL of 40× Yellow Sample Buffer (Thermo Fisher Scientific), 1 µL of 10 µM forward primer, 1 µL of 10 µM reverse primer, 2 µL of cDNA (prepped as described above), and 3 µL of distilled water. Fold changes in gene expression were calculated using the delta-delta Ct method as previously described^[22]. Gene expression fold changes (relative to GAPDH) in THOR mice were then normalized to those in Sham mice by dividing by the average Sham fold change. To compare gene expression between Sham and THOR mice, two-sample t-tests were conducted comparing the delta-delta Ct values due to the non-linearity of the fold-change values.

Statistical analyses: All statistical analyses were performed using RStudio version 1.4.1717 and GraphPad Prism version 9.3.1. Two-sided P values less than 0.05 were considered statistically significant. Chi-square tests were performed for categorical independent and dependent variable data. Fisher’s exact tests were used in lieu of chi-square tests when expected counts were less than five in any group. Two-sample t-tests were performed for binary independent and continuous dependent variable data. Wilcoxon rank-sum tests were used in lieu of 2-sample t-tests when the dependent variable was ordinal (e.g., number of AF events). Outliers were defined as > 2 standard deviations above or below the mean.

RESULTS

Baseline characteristics: A total of 131 mice (48% male) were included in the study - 45 Sham and 86 THOR. Intraoperative mortality was 4% (N = 5 deaths) due to bleeding complications. Mice who died intraoperatively were not included in final mouse counts or subsequent analyses. Mice ages ranged from 7- to 27-week-old (median age was 14 weeks). Mice were analyzed by sex and tertiles of age (≤ 11 weeks, 12-19 weeks, and ≥ 20 weeks). ECG metrics are shown by treatment group [Table 1], tertiles of age [Table 2], sex [Table 3], and AF inducibility [Table 4]. All pairwise comparisons between groups did not reach statistical significance except for the comparison of PR interval by age group (ANOVA P = 5.6E-05). Subsequent pairwise comparisons demonstrated that PR interval differed between the ≤ 11-week vs.≥ 20-week(P = 5.2E-05) and the 12-19-week vs. ≥ 20-week (P = 0.016) age groups but not the ≤ 11-week vs. 12-19-week age group (P = 0.023) at Bonferroni significance (P < 0.017, results not shown).

Overall AF inducibility: In the overall study, 33 out of 86 (38.4%) THOR mice exhibited inducible POAF, while 8 out of 45 (17.8%) Sham mice had POAF. Representative ECG tracings showing sinus rhythm and AF are shown in Figure 2A. The proportion of POAF events in THOR mice was greater than IN Sham mice (P = 0.027, Figure 2B). The number of inducible AF events (out of three) was higher in THOR vs. Sham mice (P = 0.0089, Figure 2C). These results were primarily driven by the greater number of THOR mice with 2 (out of 3) AF events [Figure 2C], since the proportion of THOR and Sham mice with 3 (out of 3) was similar. Of the mice who had POAF, the AF duration trended longer in THOR mice (mean = 34.9 s) compared to Sham (mean = 30.2 s), but the difference was not statistically significant (P = 0.41, Figure 2D).

Figure 2. (A) Representative ECG tracings showing sinus rhythm (left) and AF (right) after the intracardiac burst atrial burst protocol. (B) Percent of mice in sinus rhythm (blue) and AF (red) in Sham (n = 45) and THOR (n = 86) mice. AF was defined as at least two (out of three) positive AF events after the A-burst protocol. (C) Percent of mice with zero, one, two, and three events after the A-burst protocol in Sham (n = 45) and THOR (n = 86) mice. (D) AF duration in Sham (n = 19) and THOR (n = 46) mice. Only mice with inducible AF episodes are included. Values plotted are equal to the sum of AF duration over the three A-burst protocols. Data are reported as mean ± SEM. P values in panels (B and C) were obtained from chi-square tests. P value in panel (D) was obtained from two-sample Student’s t-tests. *P < 0.05. P ≥ 0.05 denoted ns. ECG: Electrocardiogram; AF: atrial fibrillation; SR: sinus rhythm.

AF inducibility by age: Mice were analyzed according to tertiles of age to identify an age range at which the difference in AF inducibility after cardiac surgery differed the most between sham and THOR mice. Odds of inducible POAF in THOR were similar to Sham in the youngest tertile (odds ratio [OR] = 0.97) and greater than Sham in the middle (OR = 11.7) and older (OR = 3.11) tertiles [Figure 3A-C]. Only the middle tertile (12-19 weeks old) showed a statistically significant difference in AF inducibility (P = 8.2E-03, Figure 3B). The number of AF events was nominally significantly different between Sham and THOR mice in the youngest (P = 0.075) and middle (P = 0.053) tertiles [Supplementary Figure 1].

Figure 3. AF inducibility by age. (A) Percent of mice in sinus rhythm (blue) and AF (red) in Sham (n = 14) and THOR (n = 33) mice in the bottom age tertile. (B) Percent of mice in sinus rhythm (blue) and AF (red) in Sham (n = 14) and THOR (n = 34) mice in the middle age tertile. (C) Percent of mice in sinus rhythm (blue) and AF (red) in Sham (n = 17) and THOR (n = 20) mice in the oldest age tertile. AF was defined as at least two (out of three) positive AF events after the A-burst protocol. P values were obtained from chi-square tests. *P < 0.05. P ≥ 0.05 denoted ns. AF: Atrial fibrillation; SR: sinus rhythm.

AF inducibility by sex: The odds of POAF were greater in male and female THOR mice. Only female mice exhibited statistically significantly greater AF inducibility after THOR compared to Sham (P = 6.9E-03, Figure 4A). There was no significant difference in AF inducibility between THOR and Sham male mice (P = 0.71, Figure 4B). Female THOR mice had a statistically significantly greater number of AF events compared to Sham (P = 0.039, Supplementary Figure 2A). There was no difference in the number of AF events for male Sham vs. THOR mice (P = 0.15, Supplementary Figure 2B).

Figure 4. AF inducibility by sex. (A) Percent of female mice in sinus rhythm (blue) and AF (red) in n = 16 Sham and n = 52 THOR mice. (B) Percent of male mice in sinus rhythm (blue) and AF (red) in n = 10 Sham and n = 53 THOR mice. AF was defined as at least two (out of three) positive AF events after the A-burst protocol. P values were obtained from chi-square tests. *P < 0.05. P ≥ 0.05 denoted ns. AF: Atrial fibrillation; SR: sinus rhythm.

PCR amplification of inflammatory genes: Gene expression of IL-6, IL-1β, IL-18, IL-10, IFN-γ, TNF-α, and TGF-β1 was evaluated by qPCR in atrial tissue of six Sham and six THOR mice. GAPDH was used as the housekeeping gene. Normalized fold changes of gene expression (relative to the Sham group) are shown in Figure 5. Compared to Sham, THOR mice had a statistically significantly greater expression of IL-6(P = 1.82E-04, Figure 5A), IL-18 (P = 1.7E-03, Figure 5B), and TGF-β1 (P = 0.046, Figure 5C). There were no statistically significant differences in IL-1β, IL-10, IFN-γ, and TNF-α expression between Sham and THOR mice [Supplementary Figure 3].

Figure 5. qPCR amplification of IL-6, IL-18, and TGF-β1 mRNA levels in atrial tissue. Normalized fold changes of gene expression (relative to Gapdh) in IL-6 (A), IL-18 (B), and TGF-β1 (C) calculated using the delta-delta CT method. *P < 0.05. P ≥ 0.05 denoted ns. n = 6 Sham, n = 6 THOR mice. qPCR: Quantitative polymerase chain reaction.

DISCUSSION

In this study, we characterized a novel mouse model of POAF in a cohort of 131 mice. The surgery-related mortality in our study was around five percent. All mortalities occurred intraoperatively due to bleeding complications at an incidence of ~4%, which is similar to that associated with coronary artery bypass grafting in humans (3%)^[23]. No mice died during the 72 h postoperative period. The absolute risk of developing POAF during the first three days post-surgery was 38.4% (33/86) in THOR and 17.8% (8/45) in Sham mice, corresponding to a 2.87-fold increase in odds of POAF in THOR compared to Sham mice. This effect size is comparable to POAF in humans, which occurs in ~30% of patients after cardiac surgery^[24], although our study assessed the incidence of inducible but not spontaneous AF. The inducibility of AF is an index for the development of an AF-maintaining substrate and is commonly assessed in non-surgical mouse models since only a few (transgenic) mouse lines exhibit spontaneous AF^[25,26]. As AF in mice requires supraphysiologic triggers, baseline POAF risk may be higher than those seen in humans, but this increase in risk may be negated by the healthy baseline status of mice in our study compared to humans undergoing cardiac surgery.