key: cord-0332076-vz9akivw
authors: Li, Wener; Luo, Xiaojing; Poetsch, Mareike S.; Oertel, Reinhard; Nichani, Kapil; Schneider, Martin; Strano, Anna; Hasse, Marcel; Steiner, Robert-Patrick; Cyganek, Lukas; Hettwer, Karina; Uhlig, Steffen; Simon, Kirsten; Guan, Kaomei; Schubert, Mario
title: Effects of Hydroxychloroquine and Azithromycin on iPSC-derived Cardiomyocytes: Considerations for the Treatment of COVID-19 Patients
date: 2021-08-19
journal: bioRxiv
DOI: 10.1101/2021.08.19.456950
sha: d9c3a207f6ebe582148bf936ab07cc3d3bf309ea
doc_id: 332076
cord_uid: vz9akivw

Despite known adverse effects of hydroxychloroquine (HCQ) and azithromycin (AZM) on cardiac function, HCQ and AZM have been used as combination therapy in the treatment of COVID-19 patients. Recent clinical data indicate higher complication rates with HCQ/AZM combination treatment in comparison to monotherapy. Here, we used human induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs) to systematically investigate the effects of HCQ and AZM individually and in combination. The clinically observed QT prolongation caused by treatment with HCQ could be recapitulated in iPSC-CMs based on prolonged field potential duration (FPDc). Interestingly, HCQ-induced FPDc prolongation was strongly enhanced by combined treatment with AZM, although AZM alone slightly shortened FPDc in iPSC-CMs. Furthermore, combined treatment with AZM and HCQ leads to higher cardiotoxicity, more severe structural disarrangement, and more pronounced contractile and electrophysiological dysfunctions, compared to respective mono-treatments. First mechanistic insights underlying the synergistic effects of AZM and HCQ on iPSC-CM functionality are provided based on increased Cx43- and Nav1.5-protein levels. Taken together, our results highlight that combined treatment with HCQ and AZM strongly enhances the adverse effects on cardiomyocytes, providing mechanistic evidence for the high mortality in patients receiving HCQ/AZM combination treatment.

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has caused a worldwide 51 pandemic. Several anti-viral drugs have been considered to improve clinical outcomes, including 52 hydroxychloroquine (HCQ), remdesivir, and lopinavir 1 . Attempts using HCQ in combination with 53 azithromycin (AZM) reported first positive results for the treatment of SARS-CoV-2 infected 54 patients, demonstrating reinforced viral load reduction/disappearance in a small 55 number of COVID-19 patients 2, 3 . However, this study has been frequently criticized, because 56 following clinical trials and the meta-analyses could not confirm the efficacy of treatment with HCQ 57 or HCQ in combination with AZM 1, 4, 5 . Moreover, side effects on cardiovascular function have been 58 widely observed during long-term HCQ/AZM combination therapy 6, 7 . 59 Chloroquine (CQ) and HCQ are widely used antimalarial medications and known to inhibit the 60 replication of viruses in vitro 8 . Conduction disorders were reported to occur in 85% of patients after 61 chronic treatment with HCQ (or CQ) and represent one of the main side effects of HCQ 9, 10 . 62

Mechanistic insights from animal models revealed that acute application of HCQ reduces the heart 63 rate by modulating the funny current I f 11, 12 . 64 AZM, a broad-spectrum macrolide antibiotic, was considered a good safety profile until the report 65 of a small absolute increase in cardiovascular deaths during 5 days of AZM therapy 13 . In addition, 66 several cases of AZM-induced QT-interval prolongation were reported in the clinic 12, 14 . 67 A retrospective multicenter study by Rosenberg et al. confirmed that the combination therapy of 68 HCQ and AZM not only potentiated the risk for cardiac arrest, but is further associated with an 69 increased mortality rate 15 . In line with these findings, Wang et al. showed that the treatment with 70 combined HCQ and AZM, but not HCQ or AZM alone, enhanced the susceptibility for ventricular 71

A, Representative brightfield images depicting morphology of iPSC-CMs after 7-day treatments 121 with HCQ and AZM in different concentrations. Scale bar: 100 µm. B, Cell viability after 7-day drug 122 treatment as determined by measurement of formazan formation in the MTT assay. C, LDH activity 123 detected in cell supernatants after 7-day drug treatment. D, Representative brightfield images 124 depicting morphology of iPSC-CMs after 7-day drug treatment and 7-day washout period. Even 125 after washout, iPSC-CMs treated with a combination of high concentrations of AZM and HCQ show 126 severe morphological changes and increased cell death. Scale bar: 100 µm. E, Cell viability after 127 7-day drug washout as determined by using the MTT assay. F, LDH activity detected in 128 supernatants after 7-day drug washout. Data represent technical replicates (points) and means 129 (squares) of each experiment, N = 3-7 independent experiments using iPSC-CMs from 3 healthy 130 donors (iBM76.1, iBM76.3 in green; iWTD2.1, iWTD2.3 in grey, isWT7.22 in pink). Lines and errors 131

show overall mean and SEM. Statistical analysis was performed using one-way ANOVA and 132

Tukey's multiple comparison test. ** p < 0.01, **** p < 0.0001. N.D.not determined. 133

To investigate the effects of HCQ and AZM on iPSC-CM area, sarcomere organization and 136 sarcomere length, we performed immunofluorescence staining to detect α-actinin. To evaluate the 137 effect of AZM and HCQ on cell area, iPSC-CMs were seeded at low density to monitor single cells. 138

Single cells were less resistant to drug treatment compared to cells in monolayer by showing 139 severe morphological changes and cell death, in particular, under treatment with 10 µM HCQ and 140 10 µM AZM either alone or in combination ( Figure 2A ). Therefore, structural analyses of iPSC-CMs 141 were only performed for treatments with lower drug concentrations, for which cell detachment was 142 less evident. 143

The 7-day treatment with 1 µM AZM alone resulted in an increase in cell area ( Figure 2B ). The 144 observed increase in cell area after the 7-day treatment with 1 µM AZM did not persist after 145 washout, but with a slight decrease ( Figure 2C ). After 7-day treatment with 3 µM HCQ alone, 146 iPSC-CMs showed a reduction in cell area, which was not obvious in the groups treated with 1 µM 147 HCQ alone or with HCQ (1 and 3 µM) in combination with 1 µM AZM (Figure 2A and B) . However, 148 after the drug washout, iPSC-CMs treated with 1 and 3 µM HCQ alone or in combination with 1 µM 149 AZM showed smaller cell areas compared to control cells, indicating persistent cellular shrinking 150 ( Figure 2C ). 151

To quantify the effect of HCQ and AZM on sarcomeric organization in iPSC-CMs, the proportion of 152 cells with structurally organized and disorganized sarcomeres were manually determined based on 153 the images of iPSC-CMs stained for α-actinin. Cells with evenly distributed intact sarcomeres 154 across the cell body (occupying > 80% of the cell area) were classified as structurally organized 155 ( Figure 2D , left), while cells with intact sarcomeres distributed exclusively in the center or cell 156 periphery and cells lacking clearly organized ladder-like sarcomeres were classified as structurally 157 disorganized ( Figure 2D , right). Under basal conditions, 61 ± 6% of iPSC-CMs were classified as 158 structurally organized ( Figure 2E ). The relatively high portion of cells with disorganized sarcomeres 159 at basal condition might result from the immaturity of iPSC-CMs undergoing sarcomere assembly. 160

Treatment with 1 µM AZM and 1 µM HCQ alone revealed no effect on the sarcomere organization 161 of iPSC-CMs. An increase in the percentage of structurally disorganized cells was found in cells 162 treated with 3 µM HCQ alone (p = 0.055) or in combination with 1 µM AZM (p < 0.0001, Figure 2E ). 163

As another important aspect of iPSC-CM structure, the sarcomere length was measured in the 164 population of structurally organized cells ( Figure 2D , left). The sarcomere length of iPSC-CMs at 165 basal condition was determined as 2.04 ± 0.05 µm, which is comparable to a sarcomere length of 166 ~2.2 µm observed in mature cardiomyocytes 21 . After 7-day treatment with 1 µM AZM, 3 µM HCQ, 167 or the combination of HCQ (1 and 3 µM) and 1 µM AZM, iPSC-CMs showed a significant reduction 168 in sarcomere length, which was not obvious in the group treated with 1 µM HCQ alone ( Figure 2F ). 169

The strongest reduction in sarcomere length was observed in the group treated with 1 µM AZM 170 combined with 3 µM HCQ, which demonstrates the negative effects of both compounds on the 171 organization of the contractile structures. After the subsequent washout period for 7 days, 172 sarcomere length remained strongly reduced in groups treated with 3 µM HCQ alone and HCQ in 173 combination with AZM and slightly reduced in iPSC-CMs treated with 1 µM AZM or 1 µM HCQ 174 alone ( Figure 2G ). 175

Taken together, these results highlight the negative effect of HCQ and AZM treatments on the 176 structural characteristics of iPSC-CMs and the persistence of their adverse effects even after drug 177 washout for 7 days. images of α-actinin immunostained iPSC-CMs treated with different concentrations of HCQ and 181 AZM for 7 days. B-C, Analysis of cell areas after 7-day drug treatment (B) and after subsequent 7-182 day washout (C). A total of n = 160-240 cells (40 per experiment) from 6 (B) µM HCQ led to even higher beating rates than 10 µM HCQ alone ( Figure 3B ). Moreover, an 215 increased beating rate was also observed in the group treated with 3 µM HCQ in combination with 216 10 µM AZM, which was absent in the cells treated with 3 µM HCQ alone ( Figure 3B ). In terms of 217 contraction time and relaxation time, 10 µM AZM alone showed a progressive reduction, similar to 218 the group treated with 10 µM HCQ alone during the 7-day treatment ( Figure 3C and D, 219

Supplementary Figure 4D and F). The combination of HCQ and AZM enhanced the decrease of 220 contraction and relaxation time in a concentration-and time-dependent manner ( Figure 3C and D) . 221

Of note, the combination of 10 µM AZM with only 1 µM and 3 µM HCQ led to a further reduction in 222 contraction and relaxation time. 223

Overall, these data demonstrate that AZM and HCQ directly affect beating rate, as well as 224 contraction and relaxation behavior of iPSC-CMs in a concentration-and time-dependent manner, 225 while the combination of AZM with HCQ enhances the effects of HCQ on raising the beating rate of 226 iPSC-CMs as well as on decreasing contraction time and relaxation time. and Tukey's multiple comparison test. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. 240

To assess the effect of HCQ and AZM on the heart rhythm, the field potential (FP) analysis in 243 iPSC-CMs were performed using the multi-electrode array (MEA) technique. As shown in Figure  244 4A and B, the corrected FP duration (FPDc) in the control group remained stable while 1 µM AZM 245 showed no effect on the FPDc during the 14-day recording (7-day drug treatment and subsequent 246 7-day washout). However, 10 µM AZM slightly shortened the FPDc of iPSC-CMs, and drug 247 washout could not restore it to the basal level ( Figure 4A and B). Treatment with HCQ at low 248 concentrations (1 µM and 3 µM) had no effect on FPDc, however, iPSC-CMs treated with 10 µM 249 HCQ showed a prolonged FPDc from day 2, which kept rising until day 7 ( Figure 4A and C). The 250 prolongation of FPDc induced by 10 µM HCQ was reversible, as drug washout gradually 251 eliminated this effect. 252

When 1 µM AZM was combined with HCQ (1, 3 or 10 µM), similar effects as HCQ alone were 253 observed, showing the prolongation of FPDc only with 10 µM HCQ, but to a lesser extent ( Figure  254 4D). The combination of 10 µM AZM with 3 µM HCQ significantly and reversibly prolonged the 255 FPDc of iPSC-CMs, which was not observed in cells treated with the combination of 10 µM AZM 256 with 1 µM HCQ ( Figure 4 ). When we combined 10 µM AZM with 10 µM HCQ, the prolonged FPDc 257 in iPSC-CMs was observed from day 3 till day 8 ( Figure 4E ). However, we observed that 55% of 258 iPSC-CMs failed to reveal FP and showed cell death on day 8 (the first day of washout), and 82% 259 of cultures stopped beating at the end of the experiment ( Figure 4A Representative recordings of extracellular FP in spontaneous beating iPSC-CMs under different 269 treatment conditions. iPSC-CMs treated with 10 µM AZM and 10 µM HCQ in combination stopped 270 beating at day 8 (one day after initiation of washout). B, C, Effect of AZM (B) or HCQ (C) on the 271 corrected FPD (FPDc, normalized to day 0) during 7-day treatment and subsequent 7-day 272 washout. D, E, Effects of HCQ (1, 3 and 10 µM) combined with 1 µM AZM (D) or 10 µM AZM (E) 273 on FPDc during 7-day treatment and following 7-day washout. iPSC-CMs derived from four donors 274 were used for MEA recording. For the initial recording (day 0), 10 ≤ n ≤ 13 for all conditions. 275

Spontaneous beating status of iPSC-CMs is listed in Supplementary Table 1. Two-way ANOVA 276

with Bonferroni post-hoc test was used for statistical evaluation (* p < 0.05, ** p < 0.01, *** p < 277 0.001 and **** p < 0.0001). 278

Since conduction disorders were the most frequent side effect that appeared in COVID-19 patients 281 who were administrated with HCQ and AZM 6 , we examined the impact of the two drugs on cardiac 282 conduction velocity (CV) in iPSC-CM model. As shown in Figure 5 , CV of iPSC-CMs in the control 283 group remained stable during the two-week experiment. While cells treated with 1 µM AZM 284 showed a similar conduction trajectory and CV as in the control group, 10 µM AZM led to changes 285 in trajectory and significantly augmented CV in iPSC-CMs, starting on day 3 after drug treatment, 286 but reversing on day 3 after drug washout ( Figure 5A (B) and HCQ (C) for 7 days and following washout for 7 days (normalized to day 0). D, E, CV 300 of iPSC-CMs treated with 1 µM AZM (D) and 10 µM (E) combined with HCQ (1, 3, and 10 µM) 301 during 7-day treatment and following 7-day washout. iPSC-CMs derived from four donors were 302 used for MEA recording. For the initial recording (day 0), 10 ≤ n ≤ 13 wells for all conditions. 303

Spontaneous beating status of iPSC-CMs is listed in Supplementary Table 1. Two-way ANOVA  304 with Bonferroni post-hoc test was used (* p < 0.05, ** p < 0.01, *** p < 0.001, and **** p<0.0001). 305 306 HCQ and AZM synergistically enhance the expression of Cx43 and alter the steady-state kinetics 307 of I Na in iPSC-CMs 308

To gain insights into the molecular mechanism of HCQ/AZM-induced CV augmentation, we 309 analyzed expression of Nav1.5 and Cx43, which are crucial to maintain electrical signal 310 propagation between CMs 24 . Compared to the control group, the expression of Nav1.5 was slightly, 311 but not significantly, higher in iPSC-CMs treated with 10 µM HCQ for 7 days (p > 0.05, Figure 6A , 312 B). Treatment with 10 µM AZM did not change Nav1.5 protein levels (p > 0.05, Figure 6A , B) . 313

Importantly, when we applied 10 µM HCQ combined with 10 µM AZM to iPSC-CMs, we observed a 314 2-fold increase in Nav1.5 protein expression (p > 0.05, Figure 6A , B). In terms of Cx43, 7-day 315 treatment with 10 µM HCQ significantly increased the protein expression by 3-fold (p < 0.01, Figure  316 6A, C). While treatment with 10 µM AZM alone only slightly increased the Cx43 expression (p > 317 0.05), the combination of 10 µM HCQ and 10 µM AZM synergistically quadrupled the expression of 318

Cx43 compared to the control group ( Figure 6A , C). Similar results were observed using 319 immunofluorescence staining, revealing a higher expression as well as a strong intracellular 320 accumulation of Cx43 in iPSC-CMs treated with 10 µM HCQ, 10 µM AZM, and their combination 321 ( Figure 6D ). 322

To further investigate the impact of HCQ and AZM on the function of cardiac sodium channel, we 323 recorded I Na in cells treated with 10 µM HCQ and/or 10 µM AZM for 7 days using an automated Concentrations of HCQ (A) and AZM (B) in cell lysates from iPSC-CMs after the 7-day treatment 361 with HCQ and AZM at different conditions, determined using mass spectrometry. Data represent 362 mean and SEM of N = 3 independent experiments, performed with iPSC-CMs from 2 healthy 363 donors (iBM76.1, iBM76.3 in green; iWTD2.1 in grey). N. D., below detection limit. Statistical 364 evaluation was performed using one-way ANOVA with Tukey's multiple comparison test (* p < 365 0.05). 366

The combination therapy with HCQ and AZM was initially reported to reduce viral load and to 368 improve disease progression of COVID-19 patients 2 , which could not be confirmed in follow-up 369 studies 4, 25 . In contrast, HCQ/AZM combination therapy was associated with increased cardiac 370 complication rates in comparison to monotherapy with HCQ or AZM 17 . In this study, we examined 371 the effects of HCQ and AZM on iPSC-CM structure and function for a period of 7 days, similar to 372 clinical treatment durations of 5 -10 days 2, 4, 18, 25, 26 In this study, treatments with AZM and HCQ alone revealed that both drugs at higher 394 concentrations negatively impact cell viability, morphology, sarcomeric structure, the contractility, 395 and electrophysiological function of iPSC-CMs. At an equimolar concentration of 10 µM, however, 396 a significantly higher cardiotoxic activity of HCQ than that of AZM was observed, as shown by 397 lower MTT reduction to formazan, lower cell density and higher LDH activity after the 7-day 398 treatment (Figure 1 ). Even after 7 days of drug washout, a progressive cardiotoxic effect of HCQ 399 was detected not only by the MTT assay, morphological analysis, but also by the increasing 400 number of iPSC-CM cultures which stopped spontaneous beating (Supplementary Table 1) . 401

Besides the reduced cell viability, treatment with 10 µM HCQ resulted in a progressive increase in 402

FPDc, CV and beating frequency in iPSC-CMs during the 7-day treatment. The increased FPDc 403 was also reported in the guinea pig heart upon acute treatment with 10 µM HCQ alone ex vivo 16 . In 404 our study, we observed a slight reduction of FPDc in iPSC-CMs after the 7-day treatment with 405 10 µM AZM alone, while CV was increased to a similar extend as in cells treated with 10 µM HCQ. 406

Interestingly, AZM led to an initial increase in the beating frequency on day 1, but a decrease to 407 control levels on day 3 and a further decrease until day 7. The AZM-induced increase in the 408 beating rate at day 1 is in line with the previous study showing that treatment of HL-1 CMs with 100 409 µM AZM for 24 hours dramatically increased the spontaneous beating frequency 31 . Although 410 several studies reported the electrophysiological effects of HCQ or AZM in cardiomyocytes in vitro, 411 our study is the first to evaluate HCQ and AZM in terms of the effect of clinically relevant long-term 412 treatment 31 . 413

In agreement with the reduced cell viability and impaired electrophysiological function, iPSC-CMs 414 also showed altered contractile performance. Treatment with 10 µM AZM or HCQ led to decreased 415 contraction and relaxation time as well as highly varying contraction and relaxation velocities, 416

indicating that treatments with AZM or HCQ at a high concentration over a long time period 417 interfere with the ability of iPSC-CMs to contract in a coordinated manner. In a recent publication, 418 the effects of two cardiotoxic drugs, doxorubicin (DOX) and trastuzumab (TRZ), on the viability and 419 function of iPSC-CMs were reported 37 . Unlike in our study, spontaneous beating frequency and 420 electrical propagation of iPSC-CMs were not affected by DOX and TRZ, but the contraction 421 velocity and displacement (or deformation distance) were reduced. These findings point towards 422 different mechanisms of drug-induced cardiac complications induced by AZM and HCQ compared 423 to DOX and TRZ. The adverse effects induced by DOX and TRZ were proposed to be linked to 424 drug-induced mitochondrial dysfunction and altered cardiac energy metabolism 37 . Based on our 425 results, we assume that the HCQ-induced increase in CV and alteration in contraction may be 426 caused by enhanced expression (or accumulation) of Cx43 and altered gating properties of the 427 sodium channel 37 . Acute treatment with HCQ was reported to have an effect on I Na with an IC 50 of 428 113.9 ± 78.3 µM, which may explain the reduction in the electrical signal transmission observed in 429 the guinea pig heart treated with 10 µM HCQ ex vivo 16 . In our study, we observed no effect of 10 430 µM HCQ on I Na after the 7-day treatment, but altered gating properties. This may account for the 431 different effect of HCQ on CV in iPSC-CMs compared to that in the whole heart after acute 432 treatment with 10 µM HCQ. In addition, we cannot exclude the possibilities that these different 433 effects are due to species differences between humans and guinea pigs. 434

It is worth mentioning that the effects of AZM and HCQ at low concentrations (1 or 3 µM) on cell 435 area, and sarcomere structure of iPSC-CMs were relatively mild but failed to recover to the control 436 level after 7 days of drug washout, suggesting that AZM and HCQ may induce persistent, long-437 term damage of iPSC-CM structure. In addition, treatment with AZM caused cellular hypertrophy, 438 as shown by increased cell area and higher membrane capacitance ( Figure to be inhibited by HCQ 45 . However, involvement of ABCB1 in the synergistic effect of AZM and 496 HCQ in iPSC-CMs is unlikely, as RNA-sequencing data from our group as well others reveal that 497 ABCB1 is not expressed in iPSC-CMs 46, 47 . So far, the mechanism for the increased cellular 498 accumulation of AZM and HCQ with combined treatment is unclear. 499

Activation of integrated stress response (ISR) pathway and inhibition of autophagosome formation 500 by AZM and HCQ likely explain the strong intracellular accumulation of Nav1.5 and Cx43. Previous 501 studies showed that application of CQ increased the abundance of Cx43 in neonatal rat ventricular 502 myocytes through its lysosomal inhibiting ability and prolongation of Cx43 turnover 48, 49 . 503 Remarkably, our study shows that the synergistic effect of AZM and HCQ increased Cx43 504 expression by 4-fold, which was significantly higher than the increase in Cx43 protein expression 505 observed by treatment with AZM alone. Additionally, 7-day treatment with AZM and HCQ 506 increased protein expression of Nav1.5 but did not increase sodium current density, suggesting 507 that the availability of functional sodium channels on the membrane was not altered despite the 508 intracellular accumulation 31 . As cardiac conduction is determined not only by sodium channel 509 availability but also by gap junction expression and function, our data suggest that the significantly 510 increased expression of the gap junctional protein Cx43 may contribute to the increased CV in 511 iPSC-CMs after 7-day treatment with HCQ or HCQ and AZM in combination. 512

Taken together, our results reveal that the more severe effects of the combined treatment with 513 AZM and HCQ on viability, structure and functionality of iPSC-CMs may be caused by an 514 increased intracellular accumulation of the drugs. The synergistic upregulation of Cx43 protein 515 levels by AZM and HCQ provide first mechanistic evidence for the increased cardiac complications 516 observed with the combination treatment. 517 518

Aiming to gain mechanistic insights for the increased rates of cardiac complications observed for 520 the combined treatment with AZM and HCQ, we characterized the consequences of the two drugs 521 as well as their combination on the viability, structure and functionality of iPSC-CMs. Despite iPSC-522

CMs represent an important model system to study drug effects on the human heart, different 523 aspects, including the immaturity of the cells and the lack of the multicellular environment limit, the 524 predictive value of our findings. Furthermore, modeling the situation in patients with severe COVID-525 19 may require infection of iPSC-CMs with SARS-CoV-2 before drug treatment to model structural 526 and functional abnormalities, which will make the execution of the study technically challenging. 527

Through the systematic investigation of the effects of AZM and HCQ individually as well as in 529 combination, we show that these two drugs had adverse effects on the viability, structure and 530 functionality of human cardiomyocytes. These adverse effects get more severe when AZM and 531 HCQ are applied in combination, thus recapitulating the higher rates of cardiac complications 532 observed with the AZM/HCQ combination treatment in clinical use. This synergistic activity of AZM 533 and HCQ in iPSC-CMs is likely driven by the increased intracellular accumulation of the drugs 534 when applied in combination. Furthermore, we provide evidence that the HCQ-induced increase in 535 conduction velocity is caused by elevated levels of Cx43, which further increase in combination 536 with AZM. 537 538 539

Human iPSC lines used in this study were reprogrammed from somatic cells of four healthy 542 individuals. The cell lines iWTD2.1/2.3 (UMGi001-A clone 1 and clone 3) and iBM76.1/76.3 543 (UMGi005-A clone 1 and clone 3) were generated from dermal fibroblasts and mesenchymal stem 544 cells, respectively, using STEMCCA lentivirus, and characterized as previously described 46, 50 . The 545 cell lines isWT1.13 (UMGi014-C clone 3) and isWT7.22 (UMGi020-B clone 22) were generated 546 from dermal fibroblasts using the integration-free CytoTune-iPS 2.0 Sendai Reprogramming Kit 547 (Thermo Fisher Scientific), and characterized previously 51 . The iPSC generation was approved by 548 the Ethics Committee of the University Medical Center Göttingen (approval number: 21/1/11 and 549 10/9/15) and used following the approval guidelines. To maintain the growth of iPSCs, a chemically 550 defined E8 medium (Thermo Fisher Scientific) was used, and cells were cultivated on Geltrex 551 (Thermo Fisher Scientific) coated plates at 37°C with 5% CO 2 . The E8 medium was changed on a 552 daily basis and cells at ~85% confluency were passaged using Versene (Thermo Fisher Scientific). 553

Directed differentiation of iPSCs into cardiomyocytes was induced by modulating the WNT 555 signaling cascade as described 52, 53 . In brief, when iPSCs grown on 12-well plates reached 556 80~90% confluency, the medium was changed from the E8 medium to cardio differentiation 557 medium, which composed of RPMI 1640 with Glutamax and HEPES (Thermo Fisher Scientific), 558 0.5 mg/ml human recombinant albumin (Sigma-Aldrich) and 0.2 mg/ml L-ascorbic acid 2-559 phosphate (Sigma-Aldrich). To initiate differentiation, cells were incubated with 4 µM CHIR99021 560 (a GSK3β inhibitor, Millipore) for 48 hours followed by incubation with 5 µM IWP2 (a WNT signaling 561 inhibitor, Millipore) for additional 48 hours. Thereafter, cells were kept in cardio differentiation 562 medium for four days with medium change every second day. The first beating cells were detected 563 on day 8 post differentiation. From day 8, cells were cultivated in RPMI/B27 medium containing 564 RPMI 1640 with Glutamax and HEPES, supplemented with 2% B27 (Thermo Fisher Scientific). 565

To maintain a long-term culture, iPSC-CMs were replated from 12-well plates into 6-well plates at 566 day 20 post differentiation. Briefly, cells were incubated with 1 mg/ml collagenase B (Worthington 567 Biochemical) for 1 hour at 37°C. Detached iPSC-CM clusters were gently collected into a 15 ml 568

Falcon tube and dissociated with 0.25% trypsin/EDTA (Thermo Fisher Scientific) for 8 min at 37°C. 569

Dissociated iPSC-CMs were resuspended in cardio digestion medium (80% RPMI/B27 medium, 570 20% fetal calf serum, and 2 µM thiazovivin) and cultured in Geltrex-coated 6-well plates at a 571 density of 800,000 cells per well for 24 hours. Afterward, iPSC-CMs were cultivated in RPMI/B27 572 medium. 573

To perform functional analyses, 70-day-old iPSC-CMs were dissociated again with collagenase B 574 and trypsin stepwise, and replated for different assays. One week after replating, the cells were 575 treated with HCQ and AZM alone or in combination at different concentrations for 7 days, with daily 576 medium change, followed by a 7-day washout period with RPMI/B27 medium (Supplementary 577 Figure 1 ). HCQ (EMD Millipore) was dissolved in ddH 2 O and AZM (Sigma-Aldrich) was dissolved 578 in DMSO to prepare 10 mM stock solutions, which were aliquoted and stored at -20°C. 579

Video-based analyses were used to examine drug effects on the contractile parameters of iPSC-581

CMs. To this end, iPSC-CMs were replated into Geltrex-coated 48-well plates at a density of 582 60,000 cells per well one week before drug treatment. Videos were obtained using an ORCA Flash 583 4.0 V3 CMOS camera (Hamamatsu, 60 FPS, 1024x1024 pixels resolution) on days 0 (before 584 treatment), 1, 3, 5, and 7 of the treatment-period. Video data were analyzed using the cellular 585 motion analysis software "Maia" (QuoData -Quality & Statistics GmbH) to evaluate the beating 586 properties 54 . Analysis settings were: block size 20.3 µm (16 pixels), frameshift 100 ms, and 587 maximum distance shift 8.9 µm (7 pixels). For every condition, videos were obtained from 3 588 different wells with two videos on different areas of each well. For analysis, data were normalized 589 to control without drugs of the respective day. 590

For immunostainings, iPSC-CMs were seeded into Geltrex-coated 12-well or 6-well plates 592 prepared with coverslips at a density of 15,000 or 200,000 cells per well, respectively. After 593 seeding, cells were cultured for 7 days in RPMI/B27 medium before drug treatment. On day 7 594 (after drug treatment for 7 days) or day 14 (after drug washout for 7 days), cells were washed 2 595 times for 5 minutes in relaxation buffer (PBS supplemented with 5 mM EGTA and 5 mM MgCl 2 ), 596 followed by 2 times wash with PBS and fixation in ice-cold methanol-acetone (7:3, v/v) solution for 597 20 minutes at -20°C. Fixed cells were washed 3 times for 5 minutes with PBS, followed by blocking 598 in 1% BSA for at least 2 hours at 4°C. For staining, cells were incubated with the following primary 599 antibodies: anti-α-actinin, clone EA-53 (1:500; mouse monoclonal, IgG1, Sigma-Aldrich, 7811), Coverslips were mounted on glass slides using Fluoromount-G mounting medium (Thermo Fisher 607 Scientific). Stained iPSC-CMs were imaged using a fluorescence microscope (Keyence BZ-608 X700E). The exposure time was calibrated based on staining controls performed using only 609 secondary antibody. Quantification of cell area was performed based on α-actinin stained iPSC-610 CMs using Cell Profiler 55 and manual analysis with FIJI 56 . Sarcomere-length was determined 611 manually using FIJI as described previously 22 . The amount of structurally organized iPSC-CMs 612 with evenly distributed intact sarcomeres across the cell body (occupying > 80% of the cell area) 613 and disorganized cells was determined using manual counting. 614

For FP measurement, iPSC-CMs were seeded in the cavity containing electrodes of the Geltrex-616 coated CytoView 6-well MEA plates (Axion BioSystems). Around 300,000 iPSC-CMs were first 617 resuspended in 20 µl cardio digestion medium and seeded in the electrode-containing cavity of the 618 MEA plates. One hour later, an additional 1 ml of medium was added into each well, and iPSC-619

CMs were kept in RPMI/B27 medium for one week before drug treatment. For every batch of 620 experiment, at least two wells of iPSC-CMs from different plates were treated with the same 621 condition to avoid plate variability. Spontaneous FP recordings were carried out using the Maestro 622

Edge equipped with AxIS Navigator software (Axion BioSystems) with a sample rate of 12,500 Hz 623 at 37°C with 5% CO 2 . From day 0 (the day before treatment) to day 14 (last day for washout), FPs 624 were recorded daily for all conditions used (Supplementary Figure 1) . Several key parameters 625 including conduction velocity (CV), corrected FPD C (corrected by Fridericia's formula) and inter-626 beat interval were determined using AxIS Navigator, and further analyzed with AxIS Metric Plotting 627 Tool (Axion BioSystems). Spontaneous beating frequency was defined as the reciprocal of 628 averaged inter-beat interval. The mainstream CV values were averaged for one culture. 629

To investigate the effect of high concentration of HCQ and AZM on the function of sodium channel, 631 the properties of I Na were examined in iPSC-CMs treated with 10 µM HCQ alone, 10 µM AZM 632 alone or their combination, respectively. The drug treatment lasted for 7 days with daily medium 633 change, and iPSC-CMs kept in RPMI/B27 medium served as control. Recording of I Na was 634 performed using the Patchliner Quattro (Nanion Technologies GmbH) with low resistance NPC-16 635 chips at room temperature as described previously 52, 57 . In brief, iPSC-CMs were dissociated gently 636 into single cells. Capture of single cells and formation of whole-cell configuration were processed 637 automatically by Patchliner. From a holding potential of -100 mV, I Na was recorded under pulses 638 ranging from -90 to +70 mV for 20 ms in 5 mV increment with an interval of 2 s. Currents were 639 sampled at 25 kHz and low-pass-filtered at 2.9 kHz. 640

Three-month-old iPSC-CMs were treated with 10 µM HCQ, or 10 µM AZM, or the combination of 642 HCQ and AZM for seven days, snap-frozen in liquid nitrogen and stored at -80°C. To detect the 643 expression of specific proteins, cells were lysed by homogenization in RIPA buffer (150 mM NaCl, 644 50 mM Tris, 1.0% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 1 mM EDTA, 10 mM NaF, and 1 645 mM PMSF), supplemented with protease (cOmplete mini, EDTA-free) and phosphatase 646 (PhosSTOP) inhibitors and incubated for 30 min at 4°C with gentle rotation. Cell homogenates 647 were clarified by centrifugation at 14,000 rpm for 20 min at 4°C and protein concentration was 648 measured using a BCA assay following the manufacturer's instruction. 30 µg of proteins were 649 subjected to SDS-PAGE using a 4-15% gradient gel (BioRad) and transferred onto nitrocellulose 650 membranes. Membranes were blocked in 5% milk in TBS-T for 30-45 min at room temperature and 651 probed with anti-Cx43, clone 4E6.2 (1:1000; mouse monoclonal, Merck, MAB3067), anti-Nav1.5 652

(1:200; rabbit polyclonal, Alomone Labs, ASC-005), or anti-EEF2 (1:5000; rabbit polyclonal, 653

Abcam, ab40812) at 4°C overnight, followed by incubation with horseradish peroxidase-conjugated 654 secondary antibodies goat anti-mouse (1:10,000; Sigma Aldrich, A2304) or goat anti-rabbit 655

( 

Cell viability was determined using MTT assay kit (Millipore, CT02) according to the manufacturer's 670 instructions. After drug treatment as well as after drug washout, cells were washed twice with pre-671 warmed PBS and incubated in 200 µl RPMI/B27 medium per well with 0.5 mg/ml MTT for 2 hours 672 at 37°C. Subsequently, 300 µl of isopropanol with 0.04 N HCl was added and samples were mixed 673 thoroughly by pipetting to facilitate cell lysis and the dissolving of formazan. Absorbance was 674 measured at 570 nm (formazan) and 630 nm (reference) using plate reader (Biotek Synergy HTX). 675

Viability was calculated as A 570 -A 630 . 676

Intracellular drug accumulation was determined from cell lysates of the MTT assay using mass 678 spectrometry. After MTT measurement, cell lysates were stored at -20°C for 1-4 days prior to 679 detection. The stability of HCQ and AZM under these conditions was confirmed for up to 7 days at 680 -20°C. 25 µl fresh or thawed cell lysates were diluted with 225 µl 2 mM ammonium acetate buffer, 681 vortexed and centrifuged for 10 minutes (14,000 rpm). 10 µl of the clear supernatants were 682 injected into the LC-MS/MS, which consists of an UltiMate3000 pump, an autosampler (Dionex, 683 ThermoScientific) and an API 4000 Tandem mass spectrometer (ABSciex) using positive 684 Electrospray Ionization (ESI+ ; 4500 V). HCQ and AZM were determined by a Synergi 4µ HydroRP 685 80A column 150 mm x 3.0 mm (Phenomenex, Aschaffenburg, Germany) using a binary gradient 686 with 2 mM ammonium acetate buffer and acetonitrile and a flow rate of 0.5 ml/min. The resulting 687 retention times were 3.0 min for HCQ and 3.2 min for AZM. HCQ and AZM were measured using 688 the multiple reaction monitoring mode (MRM) with nitrogen as collision gas. The method was 689 suitable for the quantification of HCQ and AZM in cell lysates over the range from 20 to 1000 690 ng/ml. Samples with higher concentration were diluted. 691

Statistics 692

Results about cell area are presented as median ± 95% CI and results for the other parameters are 693

presented as mean ± standard error of the mean (SEM). Statistical analysis was performed with 694 GraphPad Prism 9. One-way ANOVA with Tukey's multiple comparison was used for cell viability, 695 cell area, sarcomere length, contractility property, protein expression level, and drug accumulation 696 data. Two-way ANOVA with Bonferroni post-hoc test was used for MEA assay-based FPDc, CV 697 and beating rate data, as well as Patchliner assay-based I Na data. p-value < 0.05 was considered 698 statistically significant. 699 and AZM and HCQ combination groups, respectively, were analyzed; shown are mean and SEM 985 from 6 independent differentiations (B-E) . One-way ANOVA with Tukey's multiple comparison test 986 (B) and two-way ANOVA with Bonferroni post-hoc test (C-E) were used for statistical analysis (* p 987 < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.001). Basal 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 1 µM AZM 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10

711 Funding 712 The work was supported by the Free State of Saxony and the European Union EFRE (SAB 713 projects "PhänoKard" and "PhenoCor" to K. Guan as well as

Guan), by the German Federal Ministry of 715

Schubert was supported by the MeDDrive START grant from the Medical Faculty at TU Dresden

Hasse were financially supported by the Deutsche Forschungsgemeinschaft 719 (DFG, German Research Foundation) -Project Number 288034826 -IRTG 2251: "Immunological 720 and Cellular Strategies in Metabolic Disease

Current Approaches to COVID-19: Therapy and Prevention

Hydroxychloroquine and azithromycin as a treatment of COVID-19: results of an 731 open-label non-randomized clinical trial

In vitro testing of combined hydroxychloroquine and azithromycin on SARS-CoV-2 734 shows synergistic effect

Azithromycin in addition to standard of care versus standard of care alone in 737 the treatment of patients admitted to the hospital with severe COVID-19 in Brazil (COALITION II): a 738 randomised clinical trial

Observational Study of Hydroxychloroquine in Hospitalized Patients with Covid-19. 741

Safety of hydroxychloroquine, alone and in combination with azithromycin, in light 744 of rapid wide-spread use for COVID-19: a multinational, network cohort and self-controlled case 745 series study. medRxiv

Risk of hydroxychloroquine alone and in combination with azithromycin in the 748 treatment of rheumatoid arthritis: a multinational, retrospective study

Will Hydroxychloroquine Still Be a Game-Changer for COVID-19 by Combining 752 Azithromycin?

Cardiac Complications Attributed to 755 Chloroquine and Hydroxychloroquine: A Systematic Review of the Literature

Considerations for Drug Interactions on QTc in 759 Exploratory COVID-19 Treatment

Hydroxychloroquine reduces heart rate by modulating the hyperpolarization-762 activated current If: Novel electrophysiological insights and therapeutic potential

Azithromycin-induced torsade de pointes. Pacing Clin 766 Electrophysiol

Azithromycin and the Risk of Cardiovascular 769 Death

Azithromycin as a cause of QT-interval 772 prolongation and torsade de pointes in the absence of other known precipitating factors

Association of Treatment With Hydroxychloroquine or Azithromycin With In-776

Hospital Mortality in Patients With COVID-19 in

Mechanistic insights into ventricular arrhythmogenesis of hydroxychloroquine and 779 azithromycin for the treatment of COVID-19. bioRxiv

The QT interval in patients with COVID-19 treated with hydroxychloroquine and 782 azithromycin

Risk of QT Interval Prolongation Associated With Use of Hydroxychloroquine 785 With or Without Concomitant Azithromycin Among Hospitalized Patients Testing Positive for 786

COVID-19)

Effect of Chloroquine, Hydroxychloroquine, and Azithromycin on the Corrected QT 789 Interval in Patients With SARS-CoV-2 Infection

Experience With Hydroxychloroquine and Azithromycin in the Coronavirus 792 Disease 2019 Pandemic: Implications for QT Interval Monitoring

Cardiomyocyte maturation: advances in knowledge and implications for 796 regenerative medicine

Enhanced structural maturation of human induced pluripotent stem cell-799 derived cardiomyocytes under a controlled microenvironment in a microfluidic system

Automated Video-Based Analysis of Contractility and Calcium Flux in Human-803

Induced Pluripotent Stem Cell-Derived Cardiomyocytes Cultured over Different Spatial Scales

Determinants of myocardial conduction velocity: implications for 807 arrhythmogenesis

Hydroxychloroquine with or without Azithromycin in Mild-to-Moderate Covid-810 19

Chloroquine and hydroxychloroquine in coronavirus disease 2019 (COVID-19). Facts, 813 fiction and the hype: a critical appraisal

In vitro screening of a FDA approved chemical library reveals potential inhibitors of 816 SARS-CoV-2 replication

Hydroxychloroquine Blood Levels in Systemic Lupus 819 Erythematosus: Clarifying Dosing Controversies and Improving Adherence

In Vitro Antiviral Activity and Projection of Optimized Dosing Design of 823 Hydroxychloroquine for the Treatment of Severe Acute Respiratory Syndrome Coronavirus

CoV-2)

Peak Plasma Concentration of Azithromycin and Treatment Responses in 827

Mycobacterium avium Complex Lung Disease

Azithromycin Causes a Novel Proarrhythmic Syndrome

The newer macrolides: azithromycin and clarithromycin

A dose-ranging study of the pharmacokinetics of hydroxy-836 chloroquine following intravenous administration to healthy volunteers

Development of a population pharmacokinetic 840 model characterizing the tissue distribution of azithromycin in healthy subjects

Hydroxychloroquine, a less toxic derivative of chloroquine, is effective in inhibiting SARS-844

CoV-2 infection in vitro

In vivo activity of the macrolide antibiotics azithromycin, 847 roxithromycin and spiramycin against Toxoplasma gondii

Human-Induced Pluripotent Stem Cell Model of Trastuzumab-Induced Cardiac 851 Dysfunction in Patients With Breast Cancer

Macrolide antibiotics enhance the antitumor effect of lansoprazole resulting in 854 lysosomal membrane permeabilizationassociated cell death

Macrolides sensitize EGFR-TKI-induced non-apoptotic cell death via blocking 857 autophagy flux in pancreatic cancer cell lines

Association of Cardiac Infection With SARS-CoV-2 in Confirmed COVID-19 Autopsy 860 Cases

SARS-CoV-2 Infects Human Engineered Heart Tissues and Models COVID-19

SARS-CoV-2 infection of human iPSC-derived cardiac cells reflects 866 cytopathic features in hearts of patients with COVID-19

BET inhibition blocks inflammation-induced cardiac dysfunction and SARS-CoV-2 869 infection

Existing highly accumulating lysosomotropic drugs 872 with potential for repurposing to target COVID-19

Intracellular ABCB1 as a Possible Mechanism to Explain the Synergistic Effect of 875

Combination in COVID-19 Therapy

Deep phenotyping of human induced pluripotent stem cell-derived atrial and 878 ventricular cardiomyocytes

Metabolic Maturation Media Improve Physiological Function of Human iPSC-881 Derived Cardiomyocytes

Proteolysis of connexin43-containing gap 884 junctions in normal and heat-stressed cardiac myocytes

Chloroquine inhibits autophagic flux by decreasing autophagosome-lysosome 887 fusion

Comparative study of human-induced pluripotent stem cells derived 890 from bone marrow cells, hair keratinocytes, and skin fibroblasts

Efficient generation of osteoclasts from human induced pluripotent stem cells and 894 functional investigations of lethal CLCN7-related osteopetrosis

Establishment of an automated patch-clamp platform for electrophysiological and 898 pharmacological evaluation of hiPSC-CMs

IP3R-Mediated Compensatory Mechanism for Calcium Handling in Human Induced 901

Pluripotent Stem Cell-Derived Cardiomyocytes With Cardiac Ryanodine Receptor Deficiency. Front 902

CellProfiler 3.0: Next-generation image processing for biology

Fiji: an open-source platform for biological-image analysis

Disease Phenotypes and Mechanisms of iPSC-Derived Cardiomyocytes From Brugada 913

Syndrome Patients With a Loss-of-Function SCN5A Mutation

Supplementary Figure 3: Cell viability and contractility in vehicle (0.1% DMSO) treated iPSC-931

Brightfield images of iPSC-CMs after 7-day culture with RPMI/B27 medium (control) or in 932 the presence of 0.1% DMSO vehicle showing that vehicle treatment did not affect CM morphology

Cell density was investigated based on the quantification of cell nucleus counts using 934

Hoechst33342 staining and fluorescence imaging. C, Cell viability examined by MTT assay

Quantification of contractile function using video-based motion vector analysis. Data represent 936 mean and SEM of n = 4 different iPSC-CM differentiations from 3 healthy donors indicated by 937 different colours

Pooled individual replicates from n = 3 different iPSC-CM differentiations

Statistical analysis was performed based on the mean values of the individual experiments using 951 two-way ANOVA and Tukey's multiple comparison test

Supplementary Figure 5: Representative images of iPSC-CMs on MEA plate

Morphology of iPSC-CMs in the vehicle treated (A) and 10 µM AZM and 10 µM HCQ treated (B) 958 groups at day 8. Combination treatment with 10 µM AZM and 10 µM HCQ increased the number of 959 dead cells

during 7-day drug treatment (left) and following 7-day washout (right). iPSC-CMs derived from four 969 donors were used for MEA recording. For the initial recording

Numbers of beating iPSC-CM cultures used for the analysis are listed in Supplementary Table 1

ANOVA with Bonferroni post-hoc test (* p < 0.05, ** p < 0.01, *** p < 0.001, and **** p <

Scheme of beating trace and calculated parameters using Maia motion analysis software. B-G, 944 Effects of 10 µM AZM and 10 µM HCQ during the treatment period of 7 days. Shown are data of 945 the beating rate (B)