Pro-arrhythmic effects of elevated branched chain amino acid levels Short title: Branched chain amino acids and cardiac arrhythmia

Vincent Portero PhD, Thomas Nicol PhD, Svitlana Podliesna MSc, Gerard A. Marchal MSc, Antonius Baartscheer PhD, Simona Casini PhD, Rafik Tadros MD PhD, Jorien L. Treur PhD, Michael W.T. Tanck PhD, I. Jane Cox PhD, Fay Probert PhD, Tertius A. Hough PhD, Sara Falcone PhD, Leander Beekman MSc, Martina Müller-Nurasyid PhD, Gabi Kastenmüller, Christian Gieger PhD, Annette Peters PhD, Stefan Kääb MD PhD, Moritz F. Sinner MD PhD, Andrew Blease PhD, Arie O. Verkerk PhD, Connie R. Bezzina PhD, Paul K. Potter PhD, Carol Ann Remme MD PhD


Introduction
Cardiac arrhythmias comprise a major health and economic burden and are associated with significant morbidity and mortality, including cardiac failure, stroke and sudden cardiac death (SCD). SCD remains a leading cause of mortality in the Western world, accounting for up to 20% of all natural deaths, and up to 50% of all cardiovascular deaths. 1 Arrhythmias typically occur in the setting of an underlying pathology, including myocardial ischemia, structural derangements, and co-morbidities such as hypertension. 1,2 Furthermore, patients suffering from metabolic disorders (diabetes, obesity) and heart failure are at increased risk for arrhythmias and SCD. [3][4][5][6][7][8] Despite decades of research, few anti-arrhythmic therapeutic options exist due to the complexity of underlying pathologies. Development of efficient preventive and therapeutic strategies is essential but is as yet hampered by incomplete knowledge of disease mechanisms.
Over the last decade, we and others have employed several novel approaches to identify new disease mechanisms and pathways regulating cardiac electrical (dys)function, including genomic studies in rodents. 9,10 Phenotype driven screens have been very successful in identifying novel genes and alleles associated with disease, and resolving gene function. 11,12 In N-ethyl-N-nitrosourea (ENU) mutagenesis screens, pedigrees of mice with randomly induced point mutations are screened for a range of phenotypes and affected mice are subsequently used to map and clone the causative allele. 13 Employing such a phenotype-driven ENU screen, we here identified a mutant mouse line presenting with sudden death and a homozygous nonsense mutation in the Bcat2 gene. Affected mice displayed a drastic increase of plasma branched chain amino acid (BCAAvaline, isoleucine, leucine) levels due to the incomplete catabolism of BCAAs, enhanced arrhythmia susceptibility, cardiac conduction and repolarization disturbances, and excessive pro-arrhythmic intracellular calcium dysregulation in isolated cardiomyocytes. In support of a role for BCAAs in modulation of cardiac electrical function, Downloaded from https://academic.oup.com/cardiovascres/advance-article/doi/10.1093/cvr/cvab207/6303607 by guest on 02 July 2021 CVR-2021-0172-R1 7 plasma BCAA levels were positively correlated to ECG indices of conduction and repolarization in individuals from the general population. 14 Studies in human pluripotent stem cell-derived cardiomyocytes furthermore confirmed a direct pro-arrhythmic effect of elevated BCAAs, and demonstrated the crucial involvement of the mammalian target of rapamycin (mTOR) activation in mediating these effects. Our findings thus demonstrate a causative link between elevated BCAAs and arrhythmia, which has implications for furthering our understanding of mechanisms underlying lethal arrhythmias, particularly those occurring in disease states associated with elevated BCAA levels such as diabetes and metabolic syndrome. [4][5][6][14][15][16][17] Downloaded from https://academic.oup.com/cardiovascres/advance-article/doi/10.1093/cvr/cvab207/6303607 by guest on 02 July 2021 isoflurane and maintained at 0.8-1.2% isoflurane in oxygen. Euthanasia was performed by 100% CO2 administration (20% v/v per minute), followed by cervical dislocation.

Generation of mutagenized pedigrees
The details of the ENU program have been described previously. 11 Briefly, C57BL/6J male mice (G0) were treated with ENU doses of 1 x 120mg kg -1 , and then 2 x100mg kg -1 with a week between each dose. The mice were then bred with wild-type 'sighted C3H' (C3H.Pde6b+) females. 12 The resulting G1 males were bred with wild-type C3H.Pde6b+ females to produce G2 females, which were backcrossed to their G1 fathers to generate G3 offspring. The G3 mating scheme results on average in 1 in 8 G3 mice homozygous for any particular ENU-induced mutation. Binomial probability calculations indicate that a cohort of 50 single sex mice will yield 4 homozygotes with a probability of ~85% and 5 with a probability of ~75%. Initial identification of the sudden death phenotype was in G3 mice (batch A) but subsequent studies employed mice backcrossed ten generations to the C3H.Pde6b+ background (batch B). Measurements and experiments were performed in mice of either sex at the age of 4-5 weeks, unless otherwise specified (as further specified in the Data Supplement).

Mapping and DNA sequencing
DNA from affected mice and littermates was prepared from ear biopsies and used for linkage mapping utilizing the Illumina GoldenGate Mouse Medium Density Linkage Panel (Gen-Probe Life Sciences Ltd, UK). DNA from the founder G1 mouse was prepared for whole genome sequencing (WGS) using the Nucleon BACC2 Genomic DNA Extraction System (GE Healthcare Life Sciences). After library generation, a single lane of paired-end sequencing (100nt) was performed employing the Illumina HiSeq platform (Oxford Genomics Centre, Wellcome Trust Centre for Human Genetics) and analyzed as previously described. 11 Genetic variants identified by WGS were confirmed by pyrosequencing (see Data Supplement).

NMR methodology and biochemical screen
Because of the severe nature of the phenotype and risk of early death, the phenotyping of this line was limited primarily to post mortem analysis. The function of the Bcat2 gene suggested the possibility of abnormal levels of metabolites such as branched chain amino acids. To prevent the unnecessary death of mice, urine and plasma of male and female mice aged between 4 and 5 weeks of age was analysed by NMR. Details on urine and plasma sample preparation, NMR analysis, and estimation of the valine and xleucine (leucine and isoleucine combined) concentrations from the NMR spectra are provided in the Data Supplement. Further analysis of plasma was carried out by a clinical chemistry analysis of terminal plasma samples; further details are provided in the Data Supplement and Suppl. Table S1. Full length cDNA clone of BCAT2 (Dharmacon) was ligated into the pCMV6-AN-Myc vector (Origene), and the C1121T mutation was introduced by Q5 site directed mutagenesis. Plasmids were transfected into HEK293T cells followed by Western blot analysis as detailed in the Data Supplement.

Cardiac structural abnormalities
To assess the presence of cardiac hypertrophy, heart weight/body weight and heart weight/tibia length ratios were calculated. Histological assessment of cardiac tissue was carried out on hematoxylin and eosin stained tissue obtained from 6-week-old G3 mice. Both female and male mice were included in the quantification.

ECG measurements, arrhythmia inducibility testing and optical mapping
Downloaded from https://academic.oup.com/cardiovascres/advance-article/doi/10.1093/cvr/cvab207/6303607 by guest on 02 July 2021 Surface ECG measurements were performed in mice anesthetized by isoflurane inhalation using the Powerlab acquisition system (ADInstruments). Cardiac arrhythmia inducibility was assessed in isolated, Langendorff-perfused hearts. Atria and ventricles were stimulated at a basic cycle length (BCL) of 120 ms and arrhythmia inducibility was evaluated using up to 3 extrastimuli (S1-S2-S3) and burst pacing. Optical mapping was carried out in Langendorffperfused hearts paced at a BCL of 120 ms from the center of the ventricular epicardial surface.
Optical action potentials were analyzed, and local activation defined as the maximum positive slope of the action potential was calculated using custom software. Measured local activation times were used to construct ventricular activation maps and calculate conduction velocity in longitudinal and transversal directions. For all electrophysiological measurements, both female and male mice (aged 4-5 weeks) were used. More details can be found in the Data Supplement.

Action potential measurements
Details on isolation of ventricular mouse cardiomyocytes and action potential (AP) measurements are provided in the Data Supplement. In short, APs were measured at 36 °C and were elicited at 2 Hz in mouse cardiomyocytes and 1 Hz in hPSC-CMs. Typically, hPSC-CMs have a small or even complete lack of the inward rectifying potassium current (IK1).
Consequently, hPSC-CMs have a depolarized resting membrane potential (RMP) and are frequently spontaneously active. 18 To overcome these conditions, we injected an in silico IK1 with kinetics of Kir2.1 channel through dynamic clamp, 19 as previously described and validated. 20,21 Consequently, cells became quiescent with a RMP of around −82 mV. We analysed RMP, AP amplitude (APA), maximal AP upstroke velocity (Vmax) and APD at 20, 50, and 90% repolarization (APD20, APD50, and APD90, respectively). The fast-pacing protocol used to count and quantify early after depolarizations (EADs), delayed after depolarizations (DADs) and triggered action potentials (TAPs) consisted of 20 pulses at a frequency of 5 Hz followed by a 10 second pause. Average results were obtained from 5 consecutive traces.

Intracellular calcium measurements
Calcium transients were measured at 37 0 C in Tyrode solution (see Data Supplement) in isolated myocytes stimulated at 6 Hz using the fluorescent probe Indo-1 as described previously. 22 Calcium after-transients were elicited in myocytes stimulated at 6 Hz in which stimulation was stopped and spontaneous activity was recorded for 10 s. 23 Rapid cooling (RC), which causes complete depletion of calcium from the sarcoplasmic reticulum (SR), was used to estimate SR calcium content as described previously. 23 Calcium transient decay time was determined using a mono-exponential fit.

Association between plasma BCAA levels and ECG measures in the KORA F4 Study
The community-based KORA Study (Cooperative Health Research in the Augsburg Region) was conducted in Augsburg, Southern Germany. 14

Statistical analysis
No statistical methods were used to predetermine sample size. Differences between two unpaired groups were assessed using two-tailed t-tests when following a normal distribution and Mann-Whitney Rank sum test if normality test failed. Differences comparing 3 groups, when normally distributed were assessed using one-way ANOVA with Student-Newman-Keuls post-hoc test and using one-way ANOVA on ranks (Dunn's method post-hoc) when data were not normally distributed. All statistical tests were performed using sigma stat 3.5 software (Systat Software, Inc). Variability in all plots and graphs is presented as the s.e.m. All P < 0.05 were considered to be significant. *P < 0.05; **P ≤ 0.01; #P ≤ 0.001. Summary statistics are depicted in Supplemental Tables S2-9. To determine a relation between ECG measures and BCAAs in the KORA population, linear regression models were fitted, adjusting for sex.

Sudden death phenotype in homozygous Bcat2 C1121T mice
Following ENU mutagenesis of C57BL/6J male mice (G0) and successive rounds of breeding with C3H.Pde6b+ females (Fig. 1A), G3 cohorts of mutagenized animals were produced for longitudinal phenotyping studies as part of the Harwell Ageing Screen. 11 Mice from a single pedigree, consisting of mixed background G3 mice died suddenly with no obvious cause and in the absence of preceding symptoms such as tremoring or gait abnormalities, between 6 and 9 weeks of age (batch A; Fig. 1B). Mapping was performed on the basis that homozygous ENU induced mutations must lay in a region of C57BL/6J homogeneity and a region was identified on chromosome 7 spanning from 28.8Mb to 56.1Mb. Whole genome sequencing of the G1 founder male revealed two high confidence coding mutations within the mapped region, Bcat2 C1121T and Siglech C751T resulting in a Q300* and Q219* nonsense mutation, respectively.
Pyrosequencing confirmed that both mutations were present in individuals affected by the sudden death phenotype but absent from unaffected controls. Further breeding eliminated Siglech as a candidate by narrowing the mapped region and confirmed that all affected mice suffering sudden death were homozygous for the nonsense mutation Bcat2 C1121T in the gene encoding branched chain aminotransferase 2 (Suppl. Fig. S1). No heterozygous Bcat2 C1121T or wild-type mice succumbed to sudden death before the age of eight weeks and no homozygous mice lived beyond this age (batch B; Fig. 2A). The Bcat2 C1121T mutation results in an early stop (Q300*) and consequently a truncated BCAT2 protein lacking the CXXC motif essential for enzyme catalytic activity and substrate orientation (Suppl. Fig. S2). Plasma clinical chemistry revealed increased levels of triglycerides, fatty acids, and glycerol, as well as LDL in homozygous Bcat2 p.Q300*/p.Q300* mice and intermediate levels in heterozygous Bcat2 +/p.Q300* mice (Suppl. Fig. S3 and Suppl. Table S3). Homozygous Bcat2 p.Q300*/p.Q300* mice exhibited Downloaded from https://academic.oup.com/cardiovascres/advance-article/doi/10.1093/cvr/cvab207/6303607 by guest on 02 July 2021 increased potassium but decreased chloride concentration in plasma, and increased iron levels.
Whilst significant, none of these data provide a clear explanation for the sudden death phenotype.

Accumulation of branched chain amino acids in plasma and urine of Bcat2 p.Q300*/p.Q300*
BCAT2 is located in the mitochondria of most cells of the body, and is responsible for catabolizing BCAAs (leucine, isoleucine, and valine), essential amino acids that must be obtained from dietary sources. Following catabolism to their respective α-keto acids (BCKA), BCAAs are further catabolized by the branched-chain keto acid dehydrogenase complex (BCKDC), and the final catabolic products (acetyl-CoA and succinyl-CoA) are ultimately consumed in mitochondria for respiration through the tri-carboxylic acid cycle. 24 In line with this, NMR analysis of plasma revealed an up to 10-fold increase in plasma levels of BCAAs (but no increase in BCKAs, which were under the detection threshold) in Bcat2 p.Q300*/p.Q300* mice aged 5 weeks ( Fig. 2B-D). Similar accumulation of BCAAs was observed in the urine of mutant mice aged 5 weeks (Suppl. Fig. S4). Hence, the identified mutation resulted in a truncated BCAT protein with consequent accumulation of BCAAs but not BCKAs.
Heterozygous Bcat2 +/p.Q300* mice did not display sudden death or any other overt phenotype up to the age of 6-8 months, showed normal BCAA plasma levels ( Fig. 2B-D), and hence were not investigated further.

Bcat2 p.Q300*/p.Q300* mice display cardiac electrical alterations and increased arrhythmia inducibility
Given the absence of overt prior symptoms in Bcat2 p.Q300*/p.Q300* mice, we hypothesized a cardiac basis for their sudden death. No cardiac structural changes such as hypertrophy, dilation or fibrosis were observed in Bcat2 p.Q300*/p.Q300* hearts upon histological examination (Suppl.  Table 2). We therefore performed detailed electrophysiological analysis in mice aged 4-5 weeks, well before the age of 6 weeks at which they started dying suddenly.

Since intracellular calcium dysregulation often underlies pro-arrhythmic DADs and
TAPs, we next explored calcium homeostasis in isolated cardiomyocytes.  Table S5). Sarcoplasmic reticulum (SR) content measured by rapid cooling was also significantly increased in Bcat2 p.Q300*/p.Q300* cardiomyocytes (Suppl. Fig.   S8A-B), while the ratio between diastolic [Ca 2+ ]i and SR content was not different between groups (Suppl . Table S5). Moreover, mRNA expression levels of Slc8a1 (NCX) and Atp2a2 (SERCA) were unchanged (Suppl. Fig. S9, Suppl. Table S2). Fig. 5C shows typical examples of calcium after-transients recorded following a fast-pacing protocol (20 pulses at 6 Hz) using field stimulation. These calcium after-transients were separated into two groups based on their amplitude: 1) after-transients of low amplitude considered not to trigger an AP, and 2) high-amplitude after-transients capable of triggering an AP (i.e. TAP). The total number of calcium after-transients (triggered and non-triggered combined) was significantly higher in Bcat2 p.Q300*/p.Q300* compared to Bcat2 +/+ cardiomyocytes ( Fig. 5C-D Table S5). Nevertheless, noradrenaline significantly increased the number of calcium aftertransients in both Bcat2 +/+ and Bcat2 p.Q300*/p.Q300* cardiomyocytes (Suppl . Table S5).

Direct pro-arrhythmic effects of BCAAs in human PSC-derived cardiomyocytes
To demonstrate a direct causal link between elevated BCAAs and pro-arrhythmia, we evaluated the cellular electrophysiological consequences of increased BCAA levels in cultured human PSC-derived cardiomyocytes (hPSC-CMs). Incubation of hPSC-CMs for 5 days with BCAA levels similar to those observed in plasma of Bcat2 p.Q300*/p.Q300* mice resulted in significant prolongation of APD90 (Fig. 6A-B, Suppl. Table S6), and significantly increased the occurrence of pro-arrhythmic early and delayed afterdepolarizations (EAD/DAD) as compared to hPSC-CMs cultured in control medium ( Fig. 6C-D, Suppl. Table S6), hence recapitulating the findings observed in Bcat2 p.Q300*/p.Q300* cardiomyocytes. Incubation of valine alone induced similar APD90 prolongation and increased EAD/DAD incidence, while alanine (as a general amino acid control) had no effects (Suppl. Fig. S10A-D, Suppl. Table S8). Similar to Bcat2 p.Q300*/p.Q300* cardiomyocytes, hPSC-CMs cultured in medium with increased BCAA concentration also displayed intracellular calcium dysregulation, including increased diastolic [Ca 2+ ]i and after-transients ( Fig. 7A-D, Suppl. Table S6), further confirming a direct proarrhythmic effect.

Pro-arrhythmic effects of elevated BCAA levels are mediated by mTOR pathway activation
In addition to their role in energy balance, BCAAs function as potent nutrient signal molecules to promote protein synthesis, cellular metabolism, and cell growth in a mammalian target of rapamycin (mTOR)-dependent manner, 24,25 and elevated BCAA concentrations have been shown to lead to chronic induction of mTOR activity and increased oxidative stress. 26,27 Indeed, cardiac tissue from Bcat2 p.Q300*/p.Q300* mice showed significantly higher expression levels of both mTOR total protein and activated P-mTOR as compared to Bcat2 +/+ (normalized for GAPDH), in addition to an increased P-mTOR to mTOR ratio, indicating mTOR pathway activation ( Fig. 5E-F, Suppl. Table S2). In line with mTOR pathway activation, the mRNA expression level of the pro-hypertrophic marker Anf was also significantly increased in Bcat2 p.Q300*/p.Q300* hearts (Suppl. Fig. S9, Suppl. Table S2). The functional contribution of mTOR pathway activation to the observed pro-arrhythmic effects of BCAAs was subsequently demonstrated in hPSC-CMs. Incubation with the mTOR inhibitor rapamycin (500 nmol/L) prevented the APD90 prolongation and EAD/DAD incidence (Fig. 6A-D), as well as the calcium dysregulation and after-transient occurrence (Fig. 7A-D) induced by increased BCAA levels (Suppl . Table S6), whereas it had no electrophysiological effects in the absence of high BCAA levels (Suppl . Table S7). These observations demonstrate the functional involvement of mTOR pathway activation in the pro-arrhythmic effects of elevated BCAA levels.

Plasma BCAA levels correlate with ECG parameters in the general population
Finally, to confirm a modulatory role of BCAAs on cardiac electrical function in humans, we investigated the relation between ECG measures and BCAA plasma levels in 2304 participants (mean age 51.5±10.6 years; 52.6% women) from the German community-based KORA F4 Study. 14 Fig.   S11). Regression results between ECG measures and BCAAs are presented in Table 1. We observed significant sex-adjusted associations between fasting valine plasma levels and the ECG parameters PR interval, QRS duration, and QTc interval. This association was consistent across men and women for the PR interval, and maintained effect directionality for QRS duration and QTc interval. Overall, the effect size was strongest for PR interval (beta: 0.0335±0.0085 ms/µmol/l; p<0.001), for which we also identified a significant association with xLeucine (i.e. the combination of Leucine and Isoleucine; beta: 0.0270±0.0125 ms/µmol/l; p<0.05). Together with the observations in Bcat2 p.Q300*/p.Q300* mice, these findings indicate that higher plasma BCAA levels are associated with altered cardiac conduction and repolarization.

Discussion
Through a phenotype-driven random mutagenesis screen 11 we here identified a mutant mouse line presenting with a striking, highly penetrant phenotype of sudden death without prior symptoms. The causative mutation was determined to be a nonsense mutation in the Bcat2 gene resulting in a defect in BCAA catabolism and the accumulation of high levels of BCAAs in plasma and urine. Detailed electrophysiological studies unravelled a robust arrhythmogenic phenotype in affected mice, and additional studies in hPSC-CMs confirmed a direct proarrhythmic effect of elevated BCAAs levels. We therefore identified for the first time a role for BCAA dysregulation in modulating cardiac electrophysiology and risk for cardiac arrhythmias and sudden death.
BCAT2 is located in the mitochondria of most cells of the body, and is responsible for catabolizing branched chain amino acids (BCAAs; leucine, isoleucine, and valine), essential amino acids that must be obtained from dietary sources. Following catabolism by either BCAT2 or BCAT1 (the cytosolic isoform) to their respective α-keto acids (BCKA), BCAAs are further catabolized by the branched-chain keto acid dehydrogenase complex (BCKDC). 24 The final catabolic products of BCAAs are consumed in mitochondria for respiration through the tricarboxylic acid cycle. Defects in BCAA catabolism have been associated with the development of Maple Syrup Urine disease (MSUD), an autosomal recessive metabolic disorder associated with brain damage caused by mutations in genes of the branched chain BCKDC complex. 24 Here, the block in BCAA catabolism prevents the catabolism of BCKAs, the catabolic product resulting from the action of the two BCAT enzymes, cytosolic and mitochondrial. MSUD patients exhibit a failure to thrive and neurological deterioration thought to be associated with the accumulation of BCKAs. 24 This was mirrored in a different, previously published mouse line with an ENU-induced Bcat2, which resulted in the 5' splice site in Bcat2 and the deletion of exon 2 with an absence of mature BCAT2 protein. 28 These mutant mice displayed failure to thrive, reduced body weight and early death; no details were provided on the early death phenotype, and cardiac (electrophysiological) characterisation was not performed. Analysis of the plasma revealed increased levels of both BCAAs (albeit it to a lower extent as compared to Bcat2 p.Q300*/p.Q300* mice) and BCKAs, with the latter prevented from entering the mitochondria and therefore accumulating. In contrast, our Bcat2 p.Q300*/p.Q300* mutation results in the production of a truncated protein, no differences in body weight, an accumulation of BCAAs but not BCKAs, and a sudden death phenotype. In particular, the absence of neuronal damage is likely due to the fact that BCKAs in Bcat2 p.Q300*/p.Q300* mice were not elevated and hence glutamate signalling not disturbed. Mice with targeted deletion of exon 4-6 of Bcat2 demonstrated lower body weight and exercise intolerance, while their BCKA plasma levels remained normal. 29 The lack of accumulation of BCKAs in our mice implies that the BCKA transport mechanism is intact in Bcat2 p.Q300*/p.Q300* mice but not in the previously described Bcat2 deficient mice, as catabolism of BCKAs occurs within the mitochondria. 24 This suggests that the truncated BCAT2 Q300* protein has impaired transamination activity yet retained BCKA transport activity 30,31 either in the truncated protein or by maintaining the interaction of BCAT2 with a separate transport protein. This is supported by the observation that in our mice the concentration of valine is higher than that observed in the null mice. Hence, our novel mouse model presenting specific BCAA accumulation in the absence of BCKAs allows the investigation of the physiological effects of an excess of BCAAs, and the dissection of the transamination and BCKA transport activities of BCAT2. Additionally, all murine Bcat2 models described so far were generated in C57BL/6J mice, a murine strain known to present distinct mitochondrial calcium and reactive oxygen species handling properties demonstrated to be protective in heart failure models. 32 In our present study, after the initial ENU mutagenesis step, mice were backcrossed onto the C3H background strain which may explain the severe phenotype severity observed in Bcat2 p.Q300*/p.Q300* mice.
Our findings identify a profound effect of elevated BCAA levels on intracellular calcium handling. Despite the substantial increase in diastolic and systolic [Ca 2+ ]i in Bcat2 p.Q300*/p.Q300* cardiomyocytes, the Ca 2+ gradient across the SR membrane and the decay time of the Ca 2+ transient remained unaltered, arguing against underlying alterations in phosphorylation state.
Taken together, we consider the altered diastolic [Ca 2+ ]i as the main driver for the observed pro-arrhythmic calcium dysregulation in Bcat2 p.Q300*/p.Q300* cardiomyocytes. Elevated diastolic Ca 2+ , which has been shown to increase both the systolic Ca 2+ as well as the SR Ca 2+ content, 33 may be the consequence of various processes affecting ionic gradients across cell membranes, including altered energetic metabolism. Together with the increase in SR Ca 2+ , elevated diastolic Ca 2+ leads to a higher open probability of the ryanodine channels and the likelihood of spontaneous Ca 2+ release from the SR. Subsequent calcium removal by the sodium/calcium exchanger (NCX) may then lead to membrane (after)depolarization and onset of arrhythmia.
Of note, late sodium current density was not altered and hence unlikely to have contributed to the AP prolongation, EAD incidence and Ca 2+ dysregulation observed in in Bcat2 p.Q300*/p.Q300* cardiomyocytes. In line with the in vivo and ex vivo findings in the Bcat2 p.Q300*/p.Q300* mice, atrio-ventricular conduction (PR interval) and repolarization (QT interval) ECG traits were significantly correlated with BCAA plasma concentrations in the KORA cohort. These results indicate a modulatory effect of BCAAs on electrophysiological characteristics of both the working myocardium and the cardiac conduction system, suggesting that ventricular arrhythmias as well as conduction defects could be triggered by elevated BCAA levels.
Conduction through the central part of the AV-node is mainly calcium-driven, and we have recently demonstrated that intracellular calcium dysregulation may significantly contribute to AV-conduction. 34 Hence, BCAA-induced calcium dysregulation may affect AV-conduction through prolonged repolarization and/or reduced gap junctional conductance within the central AV-node. Additionally, while myocardial fibrosis was not increased in Bcat2 p.Q300*/p.Q300* mice, minor structural remodelling within the AV-nodal area may still have impacted on AVconduction. Finally, the AV-nodal region is highly innervated and a modulatory role of BCAAs on neuronal function has been previously demonstrated in neurons, 35 providing another potential mechanism by which AV-conduction may have been affected. Interestingly, gender differences in electrophysiological effects of BCAAs appear to be present in both humans ( Table 1) and mice (see Suppl. Table S9), which may be due to intrinsic BCAA concentration differences between men and women, 36 and/or gender-related differences in ion channels, exchangers and calcium handling. 37,38 Importantly, the electrophysiological changes and pro-arrhythmic events observed in hPSC-CMs incubated with elevated BCAA levels confirm a direct pro-arrhythmic effect of BCAAs in cardiomyocytes. To date, BCAT2 human mutations have been reported in only six individuals, associated mainly with developmental delay and autistic features. 39 Affected individuals displayed elevated plasma BCAA levels of up to 4 mmol/L (valine), with plasma BCKA levels remaining unchanged as in Bcat2 p.Q300*/p.Q300* mice. Interestingly, valine levels were generally more elevated than leucine and isoleucine in BCAT2 mutation carriers, 39,40 similar to our findings in Bcat2 p.Q300*/p.Q300* mice. Valine appears to play a central role in BCAA-induced electrophysiological abnormalities, as evidenced by our findings in hPSC-CMs that increased levels of valine only were sufficient to induce pro-arrhythmic effects. Moreover, we observed the strongest correlation with ECG parameters for valine in the KORA cohort.
While MSUD patients also have elevated BCAA plasma levels, arrhythmias have not been reported but cardiac electrophysiological assessment is likely not routinely performed.
Similarly, arrhythmias have as yet not been reported in BCAT2 mutation carriers, but our results CVR-2021-0172-R1 26 suggests that an in-depth cardiac electrophysiological investigation and follow-up could be beneficial for these patients as well as individuals suffering MSUD.
Acute BCAA administration has been shown to reduce infarct size and exert cardioprotective effects during myocardial ischemia-reperfusion. 41 In contrast, chronic BCAA administration was found to exacerbate cardiac dysfunction and remodelling following myocardial infarction, and furthermore increased myocardial vulnerability to ischemiareperfusion injury by enhancing glycolysis and fatty acid oxidation. 42,43 Taken together, these findings and our current observations suggest time-dependent effects triggered by BCAAs, with acute administration presenting potential benefits while a chronic exposure would be deleterious for cardiac (electrical) function. Crucially, we here demonstrate pro-arrhythmic actions of chronically elevated BCAA levels through direct effects on the cardiomyocyte level outside of the cardiac ischemia-reperfusion context.
The striking rapamycin-mediated reversal of pro-arrhythmic events in hPSC-CMs as well as the absence of effect of rapamycin in non-BCAA treated cells suggest a central role of mTOR on the electrophysiological dysfunctions triggered by elevated BCAAs. A number of studies have implicated increased reactive oxygen species (ROS) production secondary to mTOR pathway activation in the detrimental effects of BCAAs supporting our current observations. 35,44 ROS-driven oxidative stress is known to induce pro-arrhythmic calcium dysregulation in cardiomyocytes as observed in our Bcat2 p.Q300*/p.Q300* mice. 45 Interestingly, two recent studies have shown that low-BCAA diets can improve life-span through a decreased activation of the mTOR pathway, suggesting that dietary intervention could represent a preventive option in patients with elevated BCAA plasma levels. 46,47 The activation of the mTOR molecular pathway has also been linked to cardiac hypertrophy and structural remodelling. 48,49 While Bcat2 p.Q300*/p.Q300* mice showed absence of fibrosis and normal cardiac dimensions, they did display increased ventricular transcript levels of Anf, indicating that pro-hypertrophic remodeling, likely secondary to mTOR pathway activation, preceded sudden death. We hypothesize that BCAA levels in affected mice rose drastically within a short period of time following weaning, causing severe electrophysiological changes and sudden death before more overt cardiac remodelling such as fibrosis could occur. BCAAs have also been shown to regulate AMP-activated protein kinase (AMPK), 25 which functions as a metabolic sensor in cardiomyocytes. Interestingly, AMPK can directly modulate sarcolemmal ion channels and transporters, providing an additional pathway through which BCAAs can impact on cardiomyocyte electrophysiology. 50,51 Crucially, the pro-arrhythmic potential of the clinically observed increased BCAA levels may be enhanced by conditions that further compromise either BCAA catabolism or cardiac conduction and repolarization, such as for instance metabolic disturbances, certain QT-prolonging drugs and structural alterations (cardiac hypertrophy, fibrosis). BCAA dysregulation is furthermore an increasingly recognized feature of diabetes mellitus, obesity and heart failure, acquired disorders associated with an increased risk for arrhythmias and SCD. [4][5][6][15][16][17]52 Our current observations suggest that enhanced BCAA plasma levels as observed in metabolic syndromes may underlie, at least in part, the increased susceptibility to cardiac arrhythmias and SCD observed in these patients, constituting a potential target for the prevention and treatment of cardiac arrhythmias in this setting.

Limitations
The ENU mutagenesis approach is a powerful tool for identifying potential novel genes and pathways, but is inherently limited by the fact that a potential role for additional ENU-induced mutations besides Bcat2 p.Q300*/p.Q300* cannot be completely ruled out. Nevertheless, the sudden death phenotype remained after backcrossing the mice onto another genetic background, strengthening the involvement of the Bcat2 p.Q300*/p.Q300* mutation. Importantly, the identification of the Bcat2 mutation led to the discovery of elevated BCAA plasma levels in Downloaded from https://academic.oup.com/cardiovascres/advance-article/doi/10.1093/cvr/cvab207/6303607 by guest on 02 July 2021 affected mice and the subsequent establishment of their pro-arrhythmic effects in an independent system (hPSC-CMs). While we were not able to establish whether Bcat2 p.Q300*/p.Q300* mice died from cardiac arrhythmia (given the young age at which sudden death occurred), the robust arrhythmogenic electrophysiological alterations in both Bcat2 p.Q300*/p.Q300* mouse cardiomyocytes and hPSC-CMs incubated with elevated levels of BCAAs confirmed a pro-arrhythmic effect of the latter. Bcat2 p.Q300*/p.Q300* mice presented with severely elevated BCAA levels whereas BCAA plasma levels in the KORA cohort were markedly lower. However, KORA participants provided a fasting blood sample, were not supplementing BCAAs for nutritional purposes, and were considered healthy (i.e. not metabolically compromised). Hence, even under normal circumstances in the general population, BCAAs are associated with cardiac conduction and repolarization. It will be an interesting future investigation to study if strongly increased BCAA levels in humans, as it is the case with nutritional supplementation or in the setting of metabolic disorders such as obesity or diabetes, result in even more pronounced ECG changes and/or arrhythmias. Milder elevations of BCAA levels following a longer exposure may also trigger cardiac structural remodelling through mTOR pathway activation. Both electrophysiological and structural consequences may be of significance in patients presenting an altered BCAA metabolism as well as in athletes supplementing their diet with BCAAs.

Conclusion
Using an unbiased approach, we have identified BCAA metabolism as a novel modulator of cardiac electrical function and a mediator of cardiac arrhythmias and sudden death. This new mechanism linking BCAAs to pro-arrhythmia may not only be relevant for inherited BCAT2 deficiency, but also for acquired metabolic disorders such as diabetes, obesity and heart failure in which BCAA metabolism is impaired. Since cellular BCAA metabolism is reflected by

Data availability statement
The data underlying this article will be shared on reasonable request to the corresponding authors.     proton NMR spectra of free-fed plasma from Bcat2 +/+ , Bcat2 +/p.Q300* , and Bcat2 p.Q300*/p.Q300* mice at 5 weeks of age. The NMR spectra were plotted such that the αC1 proton of glucose at 5.23ppm (not shown) was of similar intensity in each data set. The prominent metabolites shown in the expanded region are assigned to: 1) lactate, 2) glucose, 3) including β-glucose, 4)