Neonatal rat cardiomyocytes as an in vitro model for circadian rhythms in the heart

Circadian rhythms are biorhythms with a 24-hour period that are regulated by molecular clocks. Several clinical and animal models have been developed to analyze the role of these rhythms in cardiovascular physiology, disease and therapy, but a convenient in vitro model that mimics both molecular and functional circadian e ﬀ ects of the heart is not available. Therefore, we established a neonatal rat cardiomyocyte model that recapitulates in vivo circadian rhythmicity, as measured by anti-phasic oscillatory mRNA expression of two core clock genes, Bmal1 and Per2 and that shows functional dependence on the clock as indicated by an oscillating response in apoptosis induced by doxorubicin, hydroperoxide or hypoxia. In addition, perturbation of the cardiac clock by the use of several compounds including Resveratrol and Ex-527 was found to result in loss of functional rhythmicity. This indicates that neonatal rat cardiomyocytes are a good model to investigate the cardiac cir- cadian clock as well as a system that allows for fast and easy preclinical testing of the in ﬂ uence of compounds on circadian rhythmicity that might have crucial e ﬀ ects on cardiac health.


Introduction
Circadian rhythms allow the body to anticipate diurnal environmental changes [1,2]. In humans, these rhythms are regulated by multiple clocks: one central master clock located in the suprachiasmatic nucleus of the brain, and distinct peripheral clocks that are present in almost every organ and cell. The clock pathway relies on the oscillatory expression of core clock genes such as CLOCK, ARNTL, PER and CRY [3], which results in the rhythmic expression of clock-controlled output genes (CCGs). According to tissue physiology, CCGs vary per organ and have important functional implications. In the cardiovascular system, circadian rhythms influence various physiological features such as metabolism, electrophysiological characteristics, (cardiac) hormone receptor functionality and coagulation [4][5][6][7]. In addition, the clock plays an important role in a number of cardiovascular disorders. The incidence of several diseases, such as myocardial infarction and sudden cardiac death follows a diurnal pattern [8][9][10][11], and disruption of rhythmicity by genetic defects, genetic manipulation or sleep disturbance, is involved in cardiac pathophysiology [12][13][14].
Discovery of the importance of the circadian clock in organ function resulted in several studies that investigated circadian rhythmicity in the heart [13,15,16]. In addition, the interest in the use of circadian rhythms in (pharmacological) therapy is rising [17][18][19]. Preclinically, several animal models have been applied to uncover the contribution of circadian rhythmicity to cardiac physiology, disease and (pharmacological) therapy [13,20,21]. An easy to use in vitro system to analyze cardiac circadian rhythmicity on a transcriptional, protein and functional level, however, is currently lacking.
A handful of experimental studies used adult rodent cardiomyocytes to mimic circadian rhythmicity in the heart [22]. Indeed, in vitro cultured rodent ventricular cardiomyocytes retain their circadian rhythms, even in the absence of any environmental input [22]. While these studies have provided important insights, adult cardiomyocytes are difficult to obtain: it takes time to breed animals to adulthood, isolation and culture are difficult and time-consuming, and since adult cardiomyocytes do not proliferate, the number of cells derived per heart is relatively small. In addition, adult cardiomyocytes cannot be kept in culture long enough to allow the analysis of several circadian cycles.
This limits their suitability for interventional studies, for example to test whether the effect of newly developed drugs is time-dependent and whether they interfere with the intrinsic cardiomyocyte circadian clock.
In the current study, we propose neonatal rat cardiomyocytes (nrCMs) as an easy in vitro system to study molecular and functional circadian rhythmicity in the heart and prove that it can serve as a model to test clock interfering characteristics of multiple compounds.

Isolation of neonatal rat cardiomyocytes
All experiments were carried out in accordance with the Guide for the Care and Use of Laboratory Animals, with prior approval by the Animal Experimentation Ethics Committee, Utrecht University, The Netherlands.
Ventricular cardiomyocytes were isolated from 1-day-old neonatal Wistar rats (Charles River). After sacrifice, hearts were excised and flushed with Solution A (NaCl 8 mg/L, KCl 0,4 mg/L, glucose 1 g/L, Na 2 HPO 4 ·2H 2 O 60 mg/L, KH 2 PO 4 60 mg/L, phenol red 20 mg/L and HEPES 4,77 mg/L in MilliQ pH 7.2-7.4) to get rid of any remaining blood. Atria and large vessels were removed and the ventricles were cut in 1mm 3 pieces. Tissue pieces were transferred to a glass flask containing 14 mL Solution A supplemented with 750 μL trypsin (2,5%; #15090046, GIBCO) and shaken for 15 min at 37°C. The tissue/suspension mix was subsequently pipetted up and down several (± 20) times using a glass pipet to detach cells from the tissue pieces. Supernatant was transferred to a new tube, pelleted (3 min at 1100 RPM without brake) and re-suspended in 5 mL culture medium (Ham's F10 without Ca 2 + and Mg 2 + (#31550-023, Gibco) supplemented with 1% penicillin-streptomycin (#DE17-602E, Lonza), 1% L-glutamine (BE17-605E, Lonza) and 5% fetal bovine serum (#F7524, Sigma)). New Solution A and trypsin were added to the remaining tissue pieces and the same procedures were followed until no tissue pieces were left (approximately 5 cycles). Cell suspensions were combined, filtered using a sterile non-woven compress (#45847, Cutisoft), and plated on uncoated culture dishes (#430167, Corning). After 2 h, non-adhering cells were collected, counted and plated as a confluent monolayer on laminin-coated (10 mg/L in Solution A, #11243217001, Roche) culture dishes (35 mm, #353001, Falcon). After 20 h, medium was replaced to remove dead cells.

Spontaneous beating
nrCMs were cultured in laminin-coated culture dishes and synchronized. Between 9 and 51 h (with 6-hour intervals), spontaneous beating frequency was measured by manual counting of cardiomyocyte contractions on 6 locations during 1 min (Nixon light microscope, objective 10×).

Statistical analysis
Data are presented as mean ± standard error of mean. Circadian rhythmicity was assessed via RAIN, a non-parametric method detecting arbitrary wave-forms in biological data [30]. Student's t-test was used to compare non-circadian differences between groups. P-values < 0.05 were considered statistically significant.

Neonatal rat cardiomyocytes show a functional molecular clock
In contrast to adult cardiomyocytes, neonatal rat cardiomyocytes (nrCMs) are easy to derive in large quantities and can be cultured for prolonged periods. We therefore investigated the use of these cells to study circadian rhythmicity. To analyze the presence of a circadian clock in nrCMs, we synchronized nrCMs via serum shock [27]. 24 hour period significance was assessed by the nonparametric algorithm RAIN [30]. Bmal1 expression, as measured by qRT-PCR, significantly oscillated in a diurnal manner (RAIN, P < 0.05; Fig. 1a). To further validate the presence of a functional clock we investigated anti-phasic oscillatory expression of Bmal1 and Per2, a hallmark of a functional circadian clock, using lentiviral promoter-based Bmal1-and Per2-dLuc (destabilized luciferase) reporter constructs. nrCMs were transduced with both constructs, synchronized and bioluminescence was measured using a LumiCycle32. Anti-phasic bioluminescent oscillations for Bmal1 and Per2 were detected over the course of 4.5 days (Fig. 1b). In addition to mRNA oscillations, western blot analysis for BMAL1 revealed rhythmic protein levels in nrCMs (RAIN, P < 0.0005; Fig. 1c) with peaks that followed gene expression with a delay of approximately 6-12 h (Fig. 1a). From these data, we conclude that nrCMs contain a functional molecular circadian clock at the mRNA as well as at the protein level.

Neonatal rat cardiomyocytes show 24-hour rhythmicity in beating frequency and induced damage response
A functional circadian clock drives rhythmic output of clock-controlled genes, which allows for functional tissue-specific oscillations. In the heart, central and peripheral circadian clocks together, control heart rate [31]. To test intrinsic functional rhythmicity in nrCMs, we therefore measured beating frequency in a temporal manner. After serum shock-based synchronization, circadian rhythmicity was observed when counting the number of beats per minute over the course of two days (RAIN, P < 0.05; Fig. 2a). In contrast, cells that only underwent a medium change 24 h before recording did not show rhythmicity (RAIN, P = 0.99; Fig. 2a). These results show that nrCMs do not only possess a molecular clock, but display functional rhythmicity as well.
UBC, a stress-inducible polyubiquitin gene, has previously been shown to oscillate in a diurnal manner in the murine heart and in human ES cell-derived cardiomyocytes [32,33]. In human ES cell-derived cardiomyocytes, this result in a strongly connected oscillatory network of stress-associated transcripts. Therefore, we tested whether synchronized nrCMs display circadian stress-responsive behaviour.
First, we exposed nrCMs to doxorubicin, an anti-cancer drug with known cardiotoxic side effects, in a time-dependent manner. Serum shock based synchronization itself did not induce apoptosis as measured by a Caspase-Glo 3/7 assay (two-tailed Student t-test, P = 0.95; Supplemental Fig. S1a). When synchronized nrCMs were exposed to doxorubicin however, significant circadian rhythms were found in their damage response, when normalized to background values (Methods) (RAIN, P < 1E-07; Fig. 2b; Supplemental Fig. S1b), which were absent in non-synchronized cultures (RAIN, P = 0.58; Supplemental Fig. S1c). Additional TUNEL staining on doxorubicin treated nrCMs showed a similar oscillating output (RAIN P < 0.005 and P < 0.0005 for nontreated and doxorubicin-treated samples; Supplemental Fig. S1d), confirming the time-dependent vulnerability of nrCMs to doxorubicin.
Next, we hypothesized that this time-dependent vulnerability of nrCMs is not limited to doxorubicin and therefore analyzed whether the use of a different stressor, the pro-oxidant tert-butyl hydroperoxide (tBHP) would lead to equal results. Indeed, a circadian apoptotic response pattern was observed for tBHP (RAIN, P < 0.0005; Fig. 2c). In vivo, the most common cardiac stressors are ischemia and ischemia followed by reperfusion, which are known to follow a diurnal pattern [34]. To mimic ischemia in vitro, we cultured nrCMs in hypoxic conditions (1% O 2 ) during 3 h at a 6-hour interval. To mimic ischemia/ reperfusion, 3-hour hypoxia conditions were followed by 2 h of normoxia (5% CO 2 and 20% O 2 ). The apoptotic response of nrCMs after hypoxia treatment as well as hypoxia/normoxia followed a significant 24-hour pattern (RAIN P < 0.005 and P < 0.05, respectively; Fig. 2d). These results show that, similar to in vivo hearts, nrCMs respond to ischemia in a circadian manner, which further underscores the relevance of using nrCMs when aiming to assess the role of rhythmicity in response to stressors.

Neonatal rat cardiomyocytes function as a good model for molecular cardiac clock interference in vitro
Many pharmacological drugs and compounds affect the circadian clock. Because of the important role of circadian rhythmicity in cardiovascular physiology and disease, an in vitro tool that is able to predict the effect of compounds on the molecular and functional clock could be useful for drug development and testing. To determine whether nrCMs can be used for this purpose, we treated nrCMs with resveratrol, a Fig. 1. Neonatal rat cardiomyocytes possess a functional circadian clock. (a) Bmal1 mRNA expression levels in synchronized neonatal rat cardiomyocytes (nrCMs) as determined by qRT-PCR. Lowest value was set to 1. Expression levels were normalized to the non-oscillatory housekeeping gene PPIA. Significance of rhythmicity across 48 h was analyzed using the RAIN algorithm and is indicated (*P < 0.05). (b) Bmal1-Luc and Per2-Luc bioluminescence in synchronized nrCMs across 108 h. Data was detrended and normalized to the highest value. Measurements were done using a LumiCyle32. Representative tracks are shown. (c) BMAL1 protein expression as determined through western blot across 48 h. (d) Relative levels protein levels that were deducted from western blots (n = 3). Lowest value was set to 1. Ponceau staining was used as a loading control. Significance of rhythmicity across 48 was analyzed using the RAIN algorithm and is indicated (****P < 0.0005). compound of interest for its potentially beneficial effects on atherosclerosis, hypertension and ischemia/reperfusion [35], which has also been linked to the circadian clock [36,37]. Using our Per2-dLuc bioluminescence reporter system, we observed a dose-dependent decrease of Per2 amplitude (Fig. 3a), which indicates dampening of the molecular clock upon administration of resveratrol. Congruently, we observed a dose-dependent dampening of circadian oscillations upon administration of Ex-527 (Sirtinol) (Fig. 3b), a compound previously linked to both cardiomyocyte stress [38], and the core clock system via the sirtuin SIRT1 [39]. Low concentrations of each compound (resveratrol or Ex-527) had moderate effects, but importantly, high dosage significantly decreased the amplitude of Per2-dLuc levels (Student's t-test: resveratrol, P < 0.05; Ex-527, P < 0.005; Fig. 3c). Additionally, both compounds lengthened the circadian period at high dose (Student's t-test P < 0.05 for resveratrol, and P < 0.05 for Ex-527 respectively). Dampened circadian amplitudes have been correlated with premature aging [40] and clock disturbance has been observed in numerous pathologies (hypertension, diabetes, sleep disturbance and cancer [1,14]). Gaining insight in the consequences of compounds on the cardiac clock is thus essential considering the pros and cons of using for hypoxia or hypoxia followed by normoxia treatment across 48 h. Significance of rhythmicity in (a-d) across 48 was analyzed using the RAIN algorithm and is indicated (ns: not significant, *P < 0.05, **P < 0.005, ***P < 0.0005). a specific compound. As illustrated here, nrCMs would be an excellent system to model how pharmacological therapeutics affect clock dynamics.

Neonatal rat cardiomyocytes lose functional rhythmicity upon clock disturbance
As both resveratrol and Ex-527 were found to dampen the molecular clock in nrCMs (Fig. 3a-d), we questioned whether this would also affect functional circadian output. Resveratrol showed a cardioprotective effect on nrCMs under basic culture conditions (Student's t-test P < 0.05; Fig. 4a). Nevertheless, the strong pro-apoptotic effect of doxorubicin (Student's t-test P < 0.05) could not be reversed by resveratrol (Fig. 4a). In contrast, when our second compound Ex-527 was added, basal apoptosis levels increased significantly (Student's t-test P < 0.005; Fig. 4B). Administration of Ex-527 in combination with doxorubicin had an additive apoptotic effect when compared to doxorubicin alone (Student's t-test, P < 0.0005; Fig. 4b).
Analysis of the effect of both compounds on the oscillating response to doxorubicin (Fig. 2c) revealed that both resveratrol and Ex-527 abolished the circadian variation of the apoptotic response (RAIN, resveratrol, P = 0.29, Ex-527, P < 0.37; Fig. 4c,d). This shows that disturbing the molecular clock of nrCMs through the use of compounds can lead to impaired circadian functionality.

Discussion
Circadian rhythmicity plays an important role in physiological, biochemical, and behavioral processes and is increasingly considered in pharmacological studies [19]. Typically, pharmacological studies are done in animal models, where time-of-day dependent drug efficacy or toxicity has to be analyzed at various time-points, requiring an undesired, but necessary large number of animals. In addition, there has been a shift in attention from central, neurohumoral control of 24-h rhythms to peripheral, cellular regulation [13]. In animals experiments, these different mechanisms cannot be uncoupled completely, possibly blurring circadian effects. Therefore, alternative, more specific assays at earlier preclinical test stages are needed. In cardiovascular disease this is challenging because cardiomyocytes are difficult to obtain or to maintain in culture for a sufficient long time without deterioration of intrinsic characteristics.
In the current study, we present neonatal rat cardiomyocytes (nrCMs) as a convenient system to model the circadian clock of the heart. We show that nrCMs have a molecular and functional clock with 24-hour rhythmicity in spontaneous contraction as well as resistance to induced apoptosis by several stressors. Importantly, we demonstrate that different pharmacological compounds influence and disrupt the cardiac clock both at the molecular and functional level.
Our findings correspond to in and ex vivo studies that investigated circadian rhythmicity in the heart. Clock components (BMAL1 and PER2), stress tolerance, and contraction rate are regulated by the cardiomyocyte circadian clock and all follow a 24-hour pattern in vivo [4,31]. More specifically, in a rodent model Durgan et al. demonstrated that cardiomyocyte stress tolerance is lowest when Bmal1 mRNA expression levels start to rise [34]. Our in vitro assay shows similar results: maximum apoptosis (doxorubicin exposure 20-26 h after synchronization) coincided with low, increasing Bmal1 mRNA levels. Second, in vitro circadian spontaneous contraction rhythmicity is comparable to physiological rhythmicity: heart rate in mice (both in and ex vivo) peaks in the middle of the active period, when Bmal1 mRNA expression is high [20]. In correspondence, our nrCM experiments show maximal spontaneous beating frequencies when Bmal1 mRNA is high.
Prior to using expensive and long-lasting animal or clinical studies, nrCMs would thus be a good and easy-to-use first stage model to study the cardiomyocyte molecular clock in vitro, as well as the effect of compounds on circadian rhythmicity that might have crucial effects on cardiac health.

Financial support
This work was supported by the Netherlands Organization for Health Research and Development (ZonMW Veni 91612147), Fig. 4. Resveratrol and Ex-527 impair time-dependent apoptotic response to doxorubicin in neonatal rat cardiomyocytes. (a) Apoptotic levels in neonatal rat cardiomyocytes (nrCMs) treated with doxorubicin (dox), resveratrol or a combination compared to non-treated cells (control). (b) Similar analysis as in a) for Ex-527 treatment. (c) Relative doxorubicin-induced apoptosis levels in nrCMs treated with resveratrol across 48 h. Apoptotic levels were normalized to baseline apoptosis levels of non-treated cells.