A deep learning platform to assess drug proarrhythmia risk.

Drug safety initiatives have endorsed human iPSC-derived cardiomyocytes (hiPSC-CMs) as an in vitro model for predicting drug-induced cardiac arrhythmia. However, the extent to which human-defined features of in vitro arrhythmia predict actual clinical risk has been much debated. Here, we trained a convolutional neural network classifier (CNN) to learn features of in vitro action potential recordings of hiPSC-CMs that are associated with lethal Torsade de Pointes arrhythmia. The CNN classifier accurately predicted the risk of drug-induced arrhythmia in people. The risk profile of the test drugs was similar across hiPSC-CMs derived from different healthy donors. In contrast, pathogenic mutations that cause arrhythmogenic cardiomyopathies in patients significantly increased the proarrhythmic propensity to certain intermediate and high-risk drugs in the hiPSC-CMs. Thus, deep learning can identify in vitro arrhythmic features that correlate with clinical arrhythmia and discern the influence of patient genetics on the risk of drug-induced arrhythmia.

Metabolic Maturation Increases Susceptibility to Hypoxia-induced Damage in Human iPSC-derived Cardiomyocytes.

The development of new cardioprotective approaches using in vivo models of ischemic heart disease remains challenging as differences in cardiac physiology, phenotype, and disease progression between humans and animals influence model validity and prognostic value. Furthermore, economical and ethical considerations have to be taken into account, especially when using large animal models with relevance for conducting preclinical studies. The development of human-induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs) has opened new opportunities for in vitro studies on cardioprotective compounds. However, the immature cellular phenotype of iPSC-CMs remains a roadblock for disease modeling. Here, we show that metabolic maturation renders the susceptibility of iPSC-CMs to hypoxia further toward a clinically representative phenotype. iPSC-CMs cultured in a conventional medium did not show significant cell death after exposure to hypoxia. In contrast, metabolically matured (MM) iPSC-CMs showed inhibited mitochondrial respiration after exposure to hypoxia and increased cell death upon increased durations of hypoxia. Furthermore, we confirmed the applicability of MM iPSC-CMs for in vitro studies of hypoxic damage by validating the known cardioprotective effect of necroptosis inhibitor necrostatin-1. Our results provide important steps to improving and developing valid and predictive human in vitro models of ischemic heart disease.

Designing Novel BCR-ABL Inhibitors for Chronic Myeloid Leukemia with Improved Cardiac Safety.

Development of tyrosine kinase inhibitors (TKIs) targeting the BCR-ABL oncogene constitutes an effective approach for the treatment of chronic myeloid leukemia (CML) and/or acute lymphoblastic leukemia. However, currently available inhibitors are limited by drug resistance and toxicity. Ponatinib, a third-generation inhibitor, has demonstrated excellent efficacy against both wild type and mutant BCR-ABL kinase, including the "gatekeeper" T315I mutation that is resistant to all other currently available TKIs. However, it is one of the most cardiotoxic of the FDA-approved TKIs. Herein, we report the structure-guided design of a novel series of potent BCR-ABL inhibitors, particularly for the T315I mutation. Our drug design paradigm was coupled to iPSC-cardiomyocyte models. Systematic structure-activity relationship studies identified two compounds, 33a and 36a, that significantly inhibit the kinase activity of both native BCR-ABL and the T315I mutant. We have identified the most cardiac-safe TKIs reported to date, and they may be used to effectively treat CML patients with the T315I mutation.

Reengineering Ponatinib to Minimize Cardiovascular Toxicity.

Small molecule tyrosine kinase inhibitors (TKI) have revolutionized cancer treatment and greatly improved patient survival. However, life-threatening cardiotoxicity of many TKIs has become a major concern. Ponatinib (ICLUSIG) was developed as an inhibitor of the BCR-ABL oncogene and is among the most cardiotoxic of TKIs. Consequently, use of ponatinib is restricted to the treatment of tumors carrying T315I-mutated BCR-ABL, which occurs in chronic myeloid leukemia (CML) and confers resistance to first- and second-generation inhibitors such as imatinib and nilotinib. Through parallel screening of cardiovascular toxicity and antitumor efficacy assays, we engineered safer analogs of ponatinib that retained potency against T315I BCR-ABL kinase activity and suppressed T315I mutant CML tumor growth. The new compounds were substantially less toxic in human cardiac vasculogenesis and cardiomyocyte contractility assays in vitro. The compounds showed a larger therapeutic window in vivo, leading to regression of human T315I mutant CML xenografts without cardiotoxicity. Comparison of the kinase inhibition profiles of ponatinib and the new compounds suggested that ponatinib cardiotoxicity is mediated by a few kinases, some of which were previously unassociated with cardiovascular disease. Overall, the study develops an approach using complex phenotypic assays to reduce the high risk of cardiovascular toxicity that is prevalent among small molecule oncology therapeutics. SIGNIFICANCE: Newly developed ponatinib analogs retain antitumor efficacy but elicit significantly decreased cardiotoxicity, representing a therapeutic opportunity for safer CML treatment.

Human iPSC modeling of heart disease for drug development.

Human induced pluripotent stem cells (hiPSCs) have emerged as a promising platform for pharmacogenomics and drug development. In cardiology, they make it possible to produce unlimited numbers of patient-specific human cells that reproduce hallmark features of heart disease in the culture dish. Their potential applications include the discovery of mechanism-specific therapeutics, the evaluation of safety and efficacy in a human context before a drug candidate reaches patients, and the stratification of patients for clinical trials. Although this new technology has the potential to revolutionize drug discovery, translational hurdles have hindered its widespread adoption for pharmaceutical development. Here we discuss recent progress in overcoming these hurdles that should facilitate the use of hiPSCs to develop new medicines and individualize therapies for heart disease.

Stars in the Night Sky: iPSC-Cardiomyocytes Return the Patient Context to Drug Screening.

iPSC-derived cardiomyocytes have the potential to revolutionize the discovery of new medicines for serious heart conditions; however, heart failure remains a major cause of mortality worldwide. In this issue of Cell Stem Cell, Fiedler et al. (2019) describe using iPSC-derived cardiomyocytes to screen new chemical entities, discovering a small molecule for ischemic injury.

High-dose intramyocardial HMGB1 induces long-term cardioprotection in sheep with myocardial infarction.

In rodents with acute myocardial infarction (AMI), high mobility group box 1 (HMGB1) injection has produced controversial results. Given the lack of data in large mammals, we searched the dose that would promote angiogenesis and expression of specific regenerative genes in sheep with AMI (protocol 1) and, subsequently, use this dose to study long-term effects on infarct size and left ventricular (LV) function (protocol 2). Protocol 1: Sheep with AMI received 250 µg (high-dose, n = 7), 25 µg (low-dose, n = 7) HMGB1, or PBS (placebo, n = 7) in 10 intramyocardial injections (0.2 ml each) in the peri-infarct area. Seven days later, only the high-HMGB1-dose group exhibited higher microvascular densities, Ki67-positive cardiomyocytes, and overexpression of VEGF, Ckit, Tbx20, Nkx2.5, and Gata4. Protocol 2: Sheep with AMI received HMGB1 250 µg (n = 6) or PBS (n = 6). At 60 days, HMGB1-treated sheep showed smaller infarcts (8.5 ± 2.11 vs. 12.2 ± 1.97% LV area, P < 0.05, ANOVA-Bonferroni) and higher microvascular density (capillaries, 1798 ± 252 vs. 1266 ± 250/mm2; arterioles, 18.3 ± 3.9 vs. 11.7 ± 2.2/mm2; both P < 0.01). Echocardiographic LV ejection fraction, circumferential shortening, and wall thickening increased from day 3 to 60 with HMGB1 (all P < 0.05). Conclusion: in ovine AMI, high-dose HMGB1 induces angio-arteriogenesis, reduces infarct size, and improves LV function at 2 months post-treatment.

Effect of poly (l-lactic acid) scaffolds seeded with aligned diaphragmatic myoblasts overexpressing connexin-43 on infarct size and ventricular function in sheep with acute coronary occlusion.

Diaphragmatic myoblasts (DM) are stem cells of the diaphragm, a muscle displaying high resistance to stress and exhaustion. We hypothesized that DM modified to overexpress connexin-43 (cx43), seeded on aligned poly (l-lactic acid) (PLLA) sheets would decrease infarct size and improve ventricular function in sheep with acute myocardial infarction (AMI). Sheep with AMI received PLLA sheets without DM (PLLA group), sheets with DM (PLLA-DM group), sheets with DM overexpressing cx43 (PLLA-DMcx43) or no treatment (control group, n = 6 per group). Infarct size (cardiac magnetic resonance) decreased ∼25% in PLLA-DMcx43 [from 8.2 ± 0.6 ml (day 2) to 6.5 ± 0.7 ml (day 45), p < .01, ANOVA-Bonferroni] but not in the other groups. Ejection fraction (EF%) (echocardiography) at 3 days post-AMI fell significantly in all groups. At 45 days, PLLA-DM y PLLA-DMcx43 recovered their EF% to pre-AMI values (PLLA-DM: 61.1 ± 0.5% vs. 58.9 ± 3.3%, p = NS; PLLA-DMcx43: 64.6 ± 2.9% vs. 56.9 ± 2.4%, p = NS), but not in control (56.8 ± 2.0% vs. 43.8 ± 1.1%, p < .01) and PLLA (65.7 ± 2.1% vs. 56.6 ± 4.8%, p < .01). Capillary density was higher (p < .05) in PLLA-DMcx43 group than in the remaining groups. In conclusion, PLLA-DMcx43 reduces infarct size in sheep with AMI. PLLA-DMcx43 and PLLA-DM improve ventricular function similarly. Given its safety and feasibility, this novel approach may prove beneficial in the clinic.

Allogeneic Mesenchymal Stromal Cells Overexpressing Mutant Human Hypoxia-Inducible Factor 1-α (HIF1-α) in an Ovine Model of Acute Myocardial Infarction.

BACKGROUND: Bone marrow mesenchymal stromal cells (BMMSCs) are cardioprotective in acute myocardial infarction (AMI) because of release of paracrine angiogenic and prosurvival factors. Hypoxia-inducible factor 1-α (HIF1-α), rapidly degraded during normoxia, is stabilized during ischemia and upregulates various cardioprotective genes. We hypothesized that BMMSCs engineered to overexpress mutant, oxygen-resistant HIF1-α would confer greater cardioprotection than nontransfected BMMSCs in sheep with AMI. METHODS AND RESULTS: Allogeneic BMMSCs transfected with a minicircle vector encoding mutant HIF1-α (BMMSC-HIF) were injected in the peri-infarct of sheep (n=6) undergoing coronary occlusion. Over 2 months, infarct volume measured by cardiac magnetic resonance (CMR) imaging decreased by 71.7±1.3% (P<0.001), and left ventricular (LV) percent ejection fraction (%EF) increased near 2-fold (P<0.001) in the presence of markedly decreased end-systolic volume. Sheep receiving nontransfected BMMSCs (BMMSC; n=6) displayed less infarct size limitation and percent LVEF improvement, whereas in placebo-treated animals (n=6), neither parameters changed over time. HIF1-α-transfected BMMSCs (BMMSC-HIF) induced angio-/arteriogenesis and decreased apoptosis by HIF1-mediated overexpression of erythropoietin, inducible nitrous oxide synthase, vascular endothelial growth factor, and angiopoietin-1. Cell tracking using paramagnetic iron nanoparticles in 12 additional sheep revealed enhanced long-term retention of BMMSC-HIF. CONCLUSIONS: Intramyocardial delivery of BMMSC-HIF reduced infarct size and improved LV systolic performance compared to BMMSC, attributed to increased neovascularization and cardioprotective effects induced by HIF1-mediated overexpression of paracrine factors and enhanced retention of injected cells. Given the safety of the minicircle vector and the feasibility of BMMSCs for allogeneic application, this treatment may be potentially useful in the clinic.

Vascular endothelial growth factor overexpression does not enhance adipose stromal cell-induced protection on muscle damage in critical limb ischemia.

OBJECTIVES: Critical limb ischemia complicates peripheral artery disease leading to tissue damage and amputation. We hypothesized that modifying adipose stromal cells (ASCs) to overexpress human vascular endothelial growth factor 165 (VEGF) would limit ischemic muscle damage to a larger extent than nonmodified ASCs. APPROACH AND RESULTS: Rabbits with critical hindlimb ischemia were injected with allogeneic abdominal fat-derived ASCs transfected with plasmid-VEGF165 (ASCs-VEGF; n=10). Additional rabbits received nontransfected ASCs (ASCs; n=10) or vehicle (placebo; n=10). One month later, ASCs-VEGF rabbits exhibited significantly higher density of angiographically visible collaterals and capillaries versus placebo (both P<0.05) but not versus ASCs (both P=NS). Arteriolar density, however, was increased in both ASCs and ASCs-VEGF groups (both P<0.05 versus placebo). ASCs-VEGF and ASCs showed comparable post-treatment improvements in Doppler-assessed peak systolic velocity, blood pressure ratio, and resistance index. Ischemic lesions were found in 40% of the muscle samples in the placebo group, 19% in the ASCs-VEGF group, and 17% in the ASCs groups (both P<0.05 versus placebo, Fisher test). CONCLUSIONS: In a rabbit model of critical limb ischemia, intramuscular injection of ASCs genetically modified to overexpress VEGF increase angiographically visible collaterals and capillary density. However, both modified and nonmodified ASCs increase arteriolar density to a similar extent and afford equal protection against ischemia-induced muscle lesions. These results indicate that modifying ASCs to overexpress VEGF does not enhance the protective effect of ASCs, and that arteriolar proliferation plays a pivotal role in limiting the irreversible tissue damage of critical limb ischemia.

Efficient plasmid-mediated gene transfection of ovine bone marrow mesenchymal stromal cells.

BACKGROUND AIMS: Given the close similarity between ovine and human cardiomyocytes, sheep models of myocardial infarction and heart failure are increasingly used in studies of stem cell-mediated heart regeneration. In these studies, mesenchymal stromal cells (MSCs) are frequently employed. To enhance the paracrine effects of these MSCs, ex vivo transfection with genes encoding growth factors has been proposed. Although viral vectors exhibit higher transfection efficiency than plasmids, they entail the risks of uncontrolled transgene expression and immune reactions that preclude repeated administration. Our aim was to optimize the efficiency of plasmid-mediated transfection of ovine MSCs, while preserving cell viability. METHODS: Varying amounts of diverse cationic lipids were used to obtain the reagent-to-DNA mass ratio showing highest luciferase activity. Transfection efficiency (flow cytometry) was tested on plasmid-green fluorescent protein-transfected MSCs at increasing DNA mass. RESULTS: Lipofectamine LTX 5 µL and Plus reagent 4 µL with 2 µg of DNA yielded 42.3 ± 4.7% transfection efficiency, while preserving cell viability. Using these transfection conditions, we transfected MSCs with a plasmid encoding human vascular endothelial growth factor (VEGF) and found high VEGF protein concentrations in the culture supernatant from day 2 (1968 ± 324 pg/mL per µg DNA) through at least day 12 (888 ± 386 pg/mL per µg DNA) after transfection. CONCLUSIONS: Plasmid-mediated transfection of ovine MSCs to over-express paracrine heart-regenerative growth factors is feasible and efficient and overcomes the risks and limitations associated with the use of viral vectors.

Reference values for echocardiographic parameters and indexes of left ventricular function in healthy, young adult sheep used in translational research: comparison with standardized values in humans.

Ovine models of ischemic heart disease and cardiac failure are increasingly used in translational research. However, reliable extrapolation of the results to the clinical setting requires knowing if ovine normal left ventricular (LV) function is comparable to that of humans. We thus assessed for echocardiographic LV dimensions and indexes in a large normal adult sheep population and compared them with standardized values in normal human adults. Bidimensional and tissue Doppler echocardiograms were performed in 69 young adult Corriedale sheep under light sedation. LV dimensions and indexes of systolic and diastolic function were measured. Absolute and body surface areanormalized values were compared to those for normal adult humans and their statistical distribution was assessed. Normalized dimensions (except for end diastolic diameter) as well as ejection fraction and fractional shortening fell within the ranges established by the American Society of Echocardiography and European Association of Echocardiography for normal adult humans. Normalized end diastolic diameter exceeded the upper normal limit but got close to it when correcting for the higher heart mass/body surface area ratio of sheep with respect to humans. Diastolic parameters also fell within normal human ranges except for a slightly lower mitral deceleration time. All values exhibited a Gaussian distribution. We conclude that echocardiographic parameters of systolic and diastolic LV performance in young adult sheep can be reliably extrapolated to the adult human, thus supporting the use of ovine models of human heart disease in translational research.

An ovine model of postinfarction dilated cardiomyopathy in animals with highly variable coronary anatomy.

Studies on cardiac regeneration require large mammalian models of dilated cardiomyopathy (DCM) after acute myocardial infarction (AMI), and pig and sheep models are increasingly used in this field of preclinical research. Given the large interindividual variability in ovine left anterior descending artery (LAD) anatomy, protocols based on the coronary arteries to be ligated often lead to significant variation in infarct sizes and hence to heterogeneous results, ranging from no ventricular remodeling to acute, lethal left ventricular (LV) failure. We designed an ovine model of postinfarction DCM based on estimated infarct size rather than on a predetermined menu of coronary artery ligatures. In seven adult sheep we induced an anterolateral AMI of approximately 25% of the LV mass by ligating the branches of the LAD that, by visual inspection, would lead to such an infarct size. In 10 to 12 weeks, LV end-diastolic volume more than doubled and LV end-systolic volume almost tripled. LV ejection fraction decreased dramatically, as did LV percent fractional shortening and LV percent wall thickening. Infarct size (planimetry) was approximately 25% of the LV endocardial surface. We conclude that in sheep, an anterolateral AMI of approximately 25% of the LV mass--regardless of the coronary branches ligated to attain that infarct size--results in a model of postinfarction DCM that may prove useful in preclinical research on myocardial regeneration.