研究背景与目的：

心血管疾病是最重要的公共卫生问题之一。心脏毒性是药物常见且严重的不良反应之一，也是心血管疾病的一个重要危险因素。临床前心脏安全性评价体系的完善将对药物开发后期昂贵的经济损耗和患者的生活产生深远的影响。治疗药物引起的线粒体毒性是心脏毒性的主要原因，严重威胁着患者的生命安全，然而线粒体毒性测试还未纳入常规心脏安全筛查程序。另外，评价模型的优化，尤其是人源心肌细胞模型的使用，将进一步优化临床前心脏安全性评价体系。本研究尝试建立预测心脏线粒体毒性的人源心肌细胞模型，优化心脏安全性评价体系，避免药物心脏毒性，减轻心血管疾病负担。

研究方法及结果：

本研究首先以细胞活率为检测指标，使用人原代心肌细胞对部分常见临床治疗药物进行了心脏毒性的筛查，发现该细胞模型检测的药物心脏毒性与药物的临床心脏毒性高度一致。通过对比成人原代心肌细胞和人诱导多能干细胞来源的心肌细胞两种心肌细胞模型，发现二者对于不同药物的心脏毒性预测结果不一致，成人原代心肌细胞对其中54.5%的药物更敏感，IC₅₀比人诱导多能干细胞来源的心肌细胞的低达177倍；人诱导多能干细胞来源的心肌细胞对其中27.3%的药物更敏感，IC₅₀比成人原代心肌细胞低约3到6倍，因此，成人原代心肌细胞对药物的敏感性更高。为进一步将心脏毒性研究从细胞水平深入到细胞器水平，更准确地模拟药物对人心肌细胞的线粒体毒性，本研究系统评估了成人原代心肌细胞和人诱导多能干细胞来源的心肌细胞对心脏毒性代表药物瑞德西韦的毒性反应，包括细胞活性、电生理、线粒体含量、膜电位、通透性转换孔、耗氧量、复合体活性检测、活性氧和乳酸分泌，发现了瑞德西韦对成人原代心肌细胞中线粒体的各项指标均造成了严重的损伤，还导致了动作电位的延长，这与瑞德西韦诱发QT延长的临床反应相一致。与此同时，在人诱导多能干细胞来源心肌细胞模型中，本研究仅检测出少量线粒体功能障碍，并未检测出电生理功能异常。通过进一步的机制研究发现人诱导多能干细胞来源心肌细胞利用了线粒体自噬恢复了主要线粒体功能，而该修复机制在成人原代心肌细胞中并不存在，且人为诱导成人原代心肌细胞中的线粒体自噬并未逆转线粒体功能障碍。因此，本研究在人原代心肌细胞中对21个潜在线粒体保护药物进行了7个线粒体不同参数筛选，发现线粒体活性氧清除药物依达拉奉，苔藓抑素，阿拉泊韦和7-羟基-3-（4'-甲氧基苯基）香豆素，在抑制活性氧产生的同时，也纠正了成人原代心肌细胞的电生理异常。而线粒体膜电位保护药物与线粒体通透性转换孔抑制药物均无此作用，提示了异常的线粒体活性氧蓄积是瑞德西韦导致心肌细胞电活动异常的分子基础。最后，为进一步验证成人原代心肌细胞作为心脏线粒体药物毒性预测模型的可靠性，本研究对18个具有不同类型线粒体毒性的药物进行了上述7个线粒体关键参数的通量化筛选，发现成人原代心肌细胞共检测到了68个线粒体指标的异常，不仅涵盖了各药物在其他模型中已报道的53个，还揭示了15个（22%）尚未被报道的线粒体异常。

研究结论：

本研究证实了：（1）在细胞水平，成人原代心肌细胞模型在筛查药物心脏毒性中具有与临床观察结果高度一致的预测效果；（2）在线粒体水平，成人原代心肌细胞能够真实反映瑞德西韦导致的心脏线粒体毒性，线粒体自噬保护机制的缺乏是成人原代心肌细胞线粒体受到广泛损伤的重要原因，并发现瑞德西韦导致心脏毒性的机制与线粒体超氧化物过量产生有关；（3）使用成人原代心肌细胞线粒体为检测指标可进行药理学和毒理学筛选，并具有较高的可行性、敏感性和准确性，为药物心脏安全性通量化筛选提供了较好的范例。本研究将为临床前心脏安全性评价体系的优化提供理论基础和数据支撑，为制药业的经济利益、患者的健康以及心血管疾病的防控奠定基础。

Background and objective

Cardiovascular disease is one of the most important public health problems. Cardiotoxicity is one of the most common and serious adverse drug reactions and an important risk factor for cardiovascular diseases. Improvements in the preclinical cardiac safety evaluation system are expected have a profound impact on the costs during late drug development and patients’ well-being. Mitochondrial toxicity caused by therapeutic drugs is a main cause of cardiotoxicity, which seriously threatens the patients’ lives. However, mitochondrial toxicity tests have not been incorporated into routine cardiac safety screening procedures. In addition, the optimization of evaluation models, especially the use of human derived cardiomyocyte models, will further improve the preclinical cardiac safety evaluation system. The aim of this study is to establish a human cardiomyocyte model for cardiac mitochondrial toxicity prediction, optimize the cardiac safety evaluation system, avoid drug cardiotoxicity, and reduce the burden of cardiovascular disease.

Methods and results

This study first screened a set of clinical drugs in human primary cardiomyocytes (hPCMs) by cell viability detection and found the results was highly consistent with the clinical cardiotoxicity. Through the comparison of two human cardiomyocyte models, hPCMs and human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs), we observed differing prediction capabilities of drug-induced cardiotoxicity. hPCMs were more sensitive to 54.5% of cardiotoxic drugs, and IC₅₀ was up to 177.7 times lower than that of hiPSC-CMs. While hiPSC-CMs are more sensitive to 27.3% of these drugs, and IC50 is about 3 to 6 times lower than that of hPCMs. These results indicat the high sensitivity of hPCMs for cardiotoxic drugs detection. To accurately model mitochondrial toxicity induced by cardiotoxic drugs on human cardiomyocytes, we systematically evaluated the toxic responses of hPCMs induced by anti-viral agent remdesivir (RDV), through the assessment of cell activity, electrophysiology, mitochondrial content, mitochondrial membrane potential (MMP), mitochondrial permeability transition pore (mPTP), oxygen consumption rate (OCR), complex activity detection, mitochondrial reactive oxygen species (mtROS), and lactate secretion. Our results showed that RDV caused widespread mitochondrial abnormalities and prolonged action potential in hPCMs, which was consistent with the clinical response of RDV-induced QT prolongation. In contrast, hiPSC-CMs exhibited moderate mitochondrial dysfunction and no electrophysiological abnormalities. Further mechanistic studies revealed that a mitochondrial function recovery mechanism, i.e. mitophagy, in hiPSC-CMs that maintains the health of the mitochondrial network. However, this repair mechanism was absent in hPCMs, and artificially inducing mitophagy did not restore mitochondrial function. Therefore, we screened 21 potential cardioprotective compounds with different mitochondrial protective abilities by evaluating the 7 key mitochondrial parameters, including MMP, mPTP, maximal respiration, spare respiratory capacity, mitochondrial complex I activity detection, mtROS, and lactate secretion. Compounds that restored mtROS, including Edaravone (Eda), Bryostatin1 (Bry), Debio025 (Deb), and 7-hydroxy-3 – (4' –methoxyphenyl) coumarin (C12), were identified to prevent RDV-induced AP prolongation. By contrast, neither MMP nor mPTP protection drugs had this effect, suggesting that mtROS accumulation is the molecular basis of abnormal electrical activity in hPCMs. Finally, in order to further verify the reliability of hPCMs as an effective safety screening model for cardiac mitochondrial toxicity, 18 drugs with different mechanisms of mitochondrial toxicities using the same parameters. The results showed that hPCMs detected 68 mitochondrial dysfunctions which not only covered 53 mitochondrial disorders reported by other models, but also revealed 15 (22%) previously unknown mitochondrial abnormalities.

Conclusions

This study firstly confirmed that hPCMs highly predicted drug induced cardiotoxicity which consistents with the clinical observation at the cellular level. Secondly, hPCMs can truly reflect the cardiac mitochondrial toxicity caused by RDV, and the lack of protection mechanism mediated by mitophagy is an important reason for extensive mitochondria damage in hPCMs. In addition, it was found for the first time that the mechanism of RDV induced cardiotoxicity was related to the excessive mtROS production, which confirmed the internal relationship between mitochondrial damage and cardiac arrhythmia. Thirdly, this study systematically elucidated the feasibility, sensitivity, and accuracy of using hPCMs as a cardiomyocyte model for pharmacological and toxicological screening in a high-throughput manner at the mitochondrial level, which provides a good example for drug cardiac safety screening. Our data are expected to facilitate the optimization of preclinical cardiac safety assessment, and ensure the economic benefits of pharmaceutical industries, patient health, as well as the prevention and control of cardiovascular diseases.