Cyclophosphamide arrhythmogenicitytesting using human-induced pluripotent stem cell-derived cardiomyocytes

Cyclophosphamide (CP) is an anticancer drug, an alkylating agent. Cardiotoxicity of CP is associated with one of its metabolites, acrolein, and clinical cardiotoxicity manifestations are described for cases of taking CP in high doses. Nevertheless, modern arrhythmogenicity prediction assays in vitro include evaluation of beat rhythm and rate as well as suppression of cardiac late markers after acute exposure to CP, but not its metabolites. The mechanism of CP side effects when taken at low doses (i.e., < 100 mg/kg), especially at the cellular level, remains unclear. In this study conduction properties and cytoskeleton structure of human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) obtained from a healthy donor under CP were evaluated. Arrhythmogenicity testing including characterization of 3 values: conduction velocity, maximum capture rate (MCR) measurements and number of occasions of re-entry on a standard linear obstacle was conducted and revealed MCR decrease of 25% ± 7% under CP. Also, conductivity area reduced by 34 ± 15%. No effect of CP on voltage-gated ion channels was found. Conduction changes (MCR and conductivity area decrease) are caused by exposure time-dependent alpha-actinin disruption detected both in hiPSC-CMs and neonatal ventricular cardiomyocytes in vitro. Deviation from the external stimulus frequency and appearance of non-conductive areas in cardiac tissue under CP is potentially arrhythmogenic and could develop arrhythmic effects in vivo.


Fibroblast derivation and reprogramming to the pluripotent state, characterization of iPSC lines and differentiation of iPSCs into cardiomyocytes. Fibroblast derivation and reprogramming
to the pluripotent state, characterization of iPSC line m34Sk3, differentiation of iPSCs into cardiomyocytes were performed as previously described in Materials and methods section in Ref. 14 . In brief, dermal fibroblasts isolated from a skin biopsy of healthy donor were nucleofected with episomal vectors expressed OCT4, SOX2, KLF4, L-MYC, and LIN28 (Addgene IDs #41855-41858, #41813-41814). Surface for cell plating was coated with Geltrex LDEV-Free hESC-Qualified Reduced Growth Factor Basement Membrane Matrix. Reprogramming to the pluripotent state was performed as described in https ://tools .therm ofish er.com/conte nt/sfs/manua ls/epi5_ episo mal_ipsc_repro gramm ing_man.pdf. Spontaneous differentiation of the cell lines was carried out through embryoid bodies formation. iPSCs were cultivated for several passages under feeder-free conditions. Directed differentiation of iPSCs into cardiomyocytes was triggered by adding the RPMI 1640 medium (Lonza) contained B27 supplement minus insulin (Thermo Fisher Scientific) and 8 μM CHIR99021 (Sigma-Aldrich) for 48 h. The first cell contractions were observed from day 9 of differentiation. The flow cytometry data for cardiac markers of patient-specific iPSC-derived CMs from healthy donor that used in this study (m34Sk3 cell line) is provided in Ref. 15 , the efficiency of directed iPSC differentiation into cardiomyocytes was 47%. The differentiated cells were subjected to metabolic selection to isolate cardiomyocytes, almost all cells that passed the metabolic selection expressed cardiac troponin T on day 45. When the culture reached 50 days, the optical mapping occurred.

Preparation of neonatal rat ventricular cardiomyocytes, preparation of samples. Preparation
of NRVMs, preparation of samples were performed as previously described in Materials and methods section in Ref. 16 . In brief, isolation and seeding of NRVMs were performed according to the Worthington protocol (http://www.worth ingto n-bioch em.com/NCIS/defau lt.html). For cell seeding and cultivation, 13-and 21-mm glass coverslips were covered with human fibronectin (Imtek) and placed in Petri dishes and 24-well culture plates. iPSCs were seated in the wells of 24-well sterile plates covered with Geltrex. The differentiation protocol started on day 3-4 after plating.
Protocol of optical mapping. Optical mapping was carried out as previously described 13,14 . The setup included high-speed video camera (Andor IXon3, Andor Technologies), a mercury lamp (Olympus U-RFL-T), an optical microscope (Olympus MVX10), a filter cube (Olympus U-M49002XL), and an impulse generator (Vellemann, PCGU-1000), a platinum point electrode and a reference circular electrode. www.nature.com/scientificreports/ Optical mapping of human iPSC-derived cardiomyocytes occurred 50 days after the start of differentiation protocol. Samples were incubated in a sterile medium at 37 °C with the fluorescent calcium-dependent Fluo-4 AM (Invitrogen, USA) dye in concentration 4 μg/ml for 30 min. Then the dye solution was exchanged with a sterile Tyrode's solution (Sigma T2145) containing 136.9 mM NaCl, 2.6 mM KCl, 1.8 mM CaCl 2 , 1.1 mM MgCl 2 , 0.4 mM NaH 2 PO 4 , 11.9 mM NaHCO 3 , 5.6 mM d-glucose, pH = 7.4 with free Ca 2+ concentration of 1.80 mM.
At first, the samples were checked for the presence of spontaneous beating. Then the passage of square pulses of 1 Hz frequency was initiated. If impulses of 1 Hz were captured by a tissue, the frequency of square pulses was increased in increment of ≤ 0.5 Hz. Each step was checked whether the impulses of set frequency are captured. When the tissue was not able to capture each of the external impulses (for example, 1:2 capturing) of set frequency, the value of frequency on the previous step was the MCR. After capturing the CV and MCR controls, the solution of CP (Baxter) in Tyrode's was added to the cell culture at concentrations: 213 μM, 639 μM, 852 μM, 1065 μM in the increasing order. The CV and MCR were measured ≤ 10 min after the addition of CP solution. The CV was measured at a frequency of 1 Hz, for the MCR measurements the frequency was increased from 1 to 5 Hz in increments of ≤ 0.5 Hz. Three independent differentiations of m34Sk3 cell line were done with n = 2, n = 3 and n = 3 independent runs of experiments respectively. The experiments were carried out at 37 °C, 0.03-0.05% CO 2 .
As the metabolism of individual organism is specific, the correspondence between concentrations in vitro and concentrations in mg/kg could be estimated approximately. Setting the mass of the human body 60 kg and the blood volume Patch-clamp. Whole-cell currents were recorded using the perforated patch-clamp technique in single cardiomyocytes, which were isolated from neighboring cells. As a perforating agent, Amphotericin B in DMSO was used at a final concentration of 0.24 mg/ml 17 . A cover slip with cardiac cells was placed in the recording chamber mounted on the stage of the Olympus IX71 inverted microscope table. The pipette and the extracellular solutions used in these protocols are listed below.
Voltage clamp experiments were performed as previously described in Materials and Methods section in Ref. 19 . Patch pipettes were pulled from borosilicate glass (BF150-86-10 Sutter Instrument, USA) with tip resistances of ~ 3 MΩ when placed into the experimental solution. The pipette offset was corrected to zero just prior to the formation of a gigaohm (GΩ) seal. After formation of the GΩ seal, the pipette capacitance was cancelled using the amplifier fast capacitance cancellation settings. Electrical access to the cell by perforation was indicated by the appearance of slow capacitance currents that increased the amplitude and rate of decay when more amphotericin pores formed in the membrane enclosed by the patch pipette. The access resistance was monitored using the slow whole-cell capacitance cancellation settings on the amplifier. Once the access resistance decreased below 12 MΩ, the experiment was started. Series resistance was compensated if required.
Whole-cell currents evoked by ramping up stimuli from − 120 to + 50 mV was examined over a 200-ms period, with a holding potential (HP) of − 80 mV (using a prestep: − 80 to − 120 mV for 100 ms) 20 . The voltagedependence of the peak Na + currents was determined by measuring peak inward currents for cells depolarized from − 80 to + 15 mV in 5-mV increments, which were applied for 200 ms. To detect L-type Ca 2+ currents without contamination from Na + currents, a 100-ms prepulse to − 40 mV from a HP of − 80 mV was used 21,22 . The peak ICaL was measured at 0 mV. Outward IKr was elicited by a 5-s depolarizing pulse from − 40 mV to + 50 mV in 10-mV increments (HP of − 40 mV). IKr was isolated as an E4031-sensitive current 23 . Typically, the membrane capacitances measured with pCLAMP10.2 software ranged from 20 to 50 pF.
Immunofluorescent staining. The protocol used for fixation and immunocytochemistry of the samples was made due to recommendations from https ://www.abcam .com/proto cols/immun ocyto chemi stry-immun ofluo resce nce-proto col. Immunofluorescent staining was performed as previously described 13 . Cells were fixed for 10 min in 4% paraformaldehyde, permeabilized for 10 min in 0.4% Triton-X100. Cells were further incubated for 30 min in blocking buffer (1% bovine serum albumin in phosphate-buffered saline, PBS), overnight at 4C with primary antibodies and for 1 h at room temperature with secondary antibodies. Cells were washed twice for 15 min in PBS. Nuclei were stained with DAPI. Primary antibodies (working dilution-1:100)-sarcomeric alpha-actinin (Abcam, ab9465). Secondary antibodies (Thermo Fisher Scientific, working dilution-1:400)-

Assessment of the structural integrity of the hiPSC-CMs cytoskeleton. Normally, α-actinin in
cardiomyocytes is associated with F-actin, which is organized in several parallel bundles. These structures are greatly affected by CP. The degree of this effect was estimated using the Fast Fourier Transform and Directionality plugin in ImageJ (NIH, Maryland, USA, http://rsb.info.nih.gov/ij) software. It allowed to analyze a distribution of structures orientation present in the input image. F-actin parallel bundles cross-linked by α-actinin appear in the distribution charts as peak values compared to the baseline intensity. Damage of cellular structures leads to a decrease in peak intensities and an increase in baseline intensity. The contrast of the highest peak was calculated according to Weber's definition 24 as: where C denotes contrast; Imax is the maximal preferred structures orientation; Ibl is the baseline intensity calculated by Directionality plugin.
Data processing and statistics. Data processing was performed as previously described 16 . All videos from optical mapping and images from the confocal microscope were processed in the ImageJ. The activation maps were built using Wolfram Mathematica. The statistical significance was determined using a one-way ANOVA followed by Fisher's least significant difference for comparisons among groups. Values of p < 0.05 were considered statistically significant.

Results
Optical mapping of hiPSC-CMs with CP. In the first set of experiments, efficiency assessment of three differentiations that were performed in this study was done on the basis of optical mapping fluorescence data. Using Fluo-4 dying of obtained cardiac tissue, amplitude maps were constructed and the percent of cardiomyocytes was calculated. The percentage of CMs in the tissue, and, therefore, average efficiency of each of three independent differentiations was 42% (n = 4), 51% (n = 6) and 50% (n = 9) (Fig. 1).
In the second set of experiments, the CP dose-dependences of the conduction velocity (CV) and maximum capture rate (MCR) were measured in the hiPSC cardiac tissue. Three independent differentiations of m34Sk3 cell line were done with n = 2, n = 3 and n = 3 independent runs of experiments. Negative control data of CV (Fig. 2a,c) and MCR (Fig. 2b,d) was obtained during incubation of the cells with Tyrode's solution. In 1 of the 8 runs of experiments, normal propagation stopped after the addition of 724 µM (~ 15 mg/kg) of CP, and in another 2 runs after the addition of 852 µM (~ 18 mg/kg). In the other 5 runs normal propagation stopped after the addition of 1065 µM (~ 22 mg/kg) of CP (Fig. 2e). Figure 2f illustrates MCR values for each run. No considerable difference in CV was found during CP treatment (Fig. 2g) up to 852 µM (~ 18 mg/kg). The MCR was stable within the margin of error after the addition of 213 µM (~ 4 mg/kg) of CP. The MCR fell by 25% ± 7% in comparison to the control values after the addition of 852 µM (~ 18 mg/kg) of CP (Fig. 2h).
In the third set of experiments, re-entry occurrence was checked on a standard linear obstacle in hiPSC cardiac tissue (n = 8) after the addition of 213 µM (~ 4 mg/kg), 639 µM (~ 13 mg/kg), or 852 µM (~ 18 mg/kg) of CP. While increasing the frequency of the external impulses from 1 to 5 Hz in increments of ≤ 0.5 Hz, no reentry was observed (Fig. 3).
However, arrhythmogenicity diagram on Fig. 3d illustrates the increase of non-capturing probability with the increase of CP dose. The impulses of the highest frequencies captured in control (4 Hz and 5 Hz) were not captured in all of the samples under 852 µM CP.
Differentiation efficiency   (Fig. 4). The ion currents of voltage-dependent ion channels were fixed unchanged after 5, 10, 15, 20, 25 min after the start of exposure to CP. After 30 min and more, vacuoles appear in isolated cells, the cells begin to die by visual observation under a microscope in the presence of 630 μM (~ 13 mg/kg) CP.
Optical mapping of neonatal rat ventricular myocytes (NRVMs) with CP. The CP dose-dependences of CV and MCR were measured in neonatal rat ventricular myocytes (NRVM) monolayers. In 1 of the 5 runs of experiments, normal propagation stopped after the addition of 639 µM (~ 13 mg/kg) of CP, and in another 4 runs after the addition of 852 µM (~ 18 mg/kg) (Fig. 5a). Figure 5b illustrates MCR values for each run. No considerable difference in CV was found during CP treatment (Fig. 5c) up to 639 µM (~ 13 mg/kg). The MCR was stable within the margin of error after the addition of 213 µM (~ 4 mg/kg) of CP. The MCR fell by 33% ± 9% in comparison to the control values after the addition of 639 µM (~ 13 mg/kg) of CP (Fig. 5d). The conductivity area of excitation wave at 852 µM (~ 18 mg/kg) of CP was 77 ± 13% (n = 5, p < 0.05 vs control).

Immunocytochemistry of hiPSC-CMs and NRVMs. Samples with ~ 50 hiPSC-CMs (both isolated and
in clusters) were prepared. Control samples and samples after 5, 10, 20, and 30 min incubation with 213 µM (~ 4 mg/kg) of CP were fixated and further stained for F-actin, α-actinin, and DAPI (Fig. 6a-f). Modification of the α-actinin structure under CP was found (Fig. 6b, Table 1). Distribution of structures orientation present in the image is characterized as C-value (calculation of the C-value is described in Materials and Methods section, paragraph 7). C-value for normal, undamaged cardiomyocytes that were presented in control samples, was 1.53 ± 0.67 (n = 9) and for cardiomyocytes with damaged cytoskeleton after 30 min incubation with 213 µM (~ 4 mg/kg) of CP-0.64 ± 0.18 (n = 9, p < 0.001 vs control).
Samples with ~ 100 isolated of NRVMs were prepared. Control samples and samples taken after 5, 10, 20, and 30 min of incubation with 213 µM (~ 4 mg/kg) of CP were fixated and further stained for α-actinin. Modification of the α-actinin structure under CP was also found (Fig. 7).
The proportion of CMs with a normal cytoskeleton (i.e., "undamaged") in relation to the total number of CMs per sample was calculated for each time interval of incubation with CP (Table 1). For hiPSC-CMs samples, the number of "undamaged" isolated CMs and "undamaged" CM clusters were summarized, and the proportion to the total number of isolated CMs and CM clusters for each time interval was calculated. The boundary time at which the number of undamaged cells was close to that of the control (a 9% and 22% decrease for hiPSC-CMs and NRVMs, respectively) was 5 min.

Discussion
CP is an anti-cancer drug that is used in the treatment of rheumatoid arthritis, lupus erythematosus, multiple sclerosis, neuroblastoma, and other types of cancer and it is also used in transplantology 25 . Unfortunately, its action is often associated with cardiovascular side effects such as atrioventricular block, tachyarrhythmias, heart failure, and myocarditis when taken in high doses (i.e., 100-200 mg/kg) 1-6 . While heart failure and myocarditis are caused by metabolites of CP 8 , atrioventricular block and tachyarrhythmias might be associated with the mechanisms of arrhythmia occurrence under CP influence on a cellular level, particularly when small doses of the drug are taken (i.e., less than 100 mg/kg).
To directly test the arrhythmogenic properties of CP and not its metabolites, we used an experimental model based on CMs obtained from the induced pluripotent stem cells of a healthy donor. 3 values were measured: the conduction velocity (CV), maximum capture rate (MCR), and number of occasions of re-entry on a standard linear obstacle (Supplementary Fig. 1) after the application of CP using the optical mapping method. The obtained 3 values compose the arrhythmogenicity test previously published by the authors 13,14 . Electrophysiological parameters (action potential duration APD, stable value of CV, the response to periodic stimulation in the range of physiological values etc.) of these cells were measured by Slotvitsky M.M., Agladze K.I. et al. previously 26 and reached the same values as in mature cardiomyocytes at the 50th day of differentiation. As a spontaneous activity could be either a limit cycle or stochastic parameter depends on stochastic gating of transmembrane currents and of calcium release channels, it was not investigated in this study 27,28 .
The CV was stable within the margin of error under 213-852 µM (~ 4-18 mg/kg) CP. The level of Cx43 in a cardiomyocyte tissue/monolayer is connected with changes of conduction velocity. The significant reduction of intercellular coupling is required to cause minor slowing of conduction velocity 29,30 . Recent study provided the dependence of CV on Cx43 level in strands of ventricular myocytes. The average value of the CV was > 1.5 times slower in cell strands combining 70% wild type cells and 30% Cx43 knock out cells vs 100% wild type cell samples (p < 0.05). 50% wild type cells and 50% Cx43 knock out cells combination showed CV reduce > 5 times vs 100% wild type cell samples (p < 0.05) 31 and the conduction block. Thus, the fact that conduction velocity was stable within the margin of error after applying CP in this study leads to the conclusion that intercellular contacts might not be influenced by 213-852 µM (~ 4-18 mg/kg) of CP.
However, in this study, a decrease in the MCR of the hiPSC cardiac tissue of up to 25% ± 7% under the influence of 852 μM (~ 18 mg/kg) was observed. The conductivity area of excitation wave was dose-dependently www.nature.com/scientificreports/ Our study shows that CP has no effect on the voltage-dependent ICaL, IKs and IKr, and INa ion channels (Fig. 4) in the concentration of 630 μM (~ 13 mg/kg) with an experiment duration ≤ 30 min. Nevertheless, CP caused exposure time-dependent changes in the α-actinin structure in the part of the seeded CMs (Table 1). It was detected both in human iPSC-CMs from a healthy donor and in NRVMs. Moreover, detachment process of the cells' edge occurred after 30 min of incubation with CP (see SEM micrographs of hiPSC-CMs in Supplementary  Fig. 3). Such structural changes may lead to the reduction of the conductivity area and cause the MCR decrease www.nature.com/scientificreports/ under the influence of CP. This is consistent with the fact that clinical QT prolongation may also be a consequence of structural injury to the myocardium 32 . Thus, a possible mechanism of the CP effect on CMs in vitro is shown to be related to the structural injury of the cells rather than to effects on the voltage-gated ion channels. As shown in previous works, CP showed no response according to the hiPSC-CM (iCell Cardiomyocytes) arrhythmic risk model developed by Guo et al. 33 based on real-time cellular impedance measurement. The absence of re-entry on a standard linear obstacle in this study proves these results. Re-entry formation presence under CP was checked on the whole range of frequencies of the external impulses from 1 to 5 Hz in increments of ≤ 0.5 Hz. No re-entry on the standard linear obstacle in hiPSC cardiac tissue was observed during CP treatment. Three drugs with confirmed proarrhythmic risk in clinic and known effects on voltage-dependent ion channels were used as the positive control (experiments were conducted on the current experimental setup) and published before 13,14 . Lidocaine and E-4031 were tested on patient-specific hiPSC cardiac tissue from a healthy donor (one cell line, ISMA6L) 13 . Lidocaine, fast sodium channel blocker, at the concentration 100 µM induced re-entry formation on the sharp end of the obstacle at 2 Hz, higher than 2 Hz frequencies of external stimulation were not captured by the tissue. E-4031, hERG-channel blocker, in a concentration of 1.6 µM induced re-entry formation on the sharp end of the obstacle at 1.7 Hz and lower frequencies of the external impulses, higher than 1.7 Hz frequencies were not captured by the tissue. Erythromycin was tested on the same model used in this study, patient-specific hiPSC cardiac tissue from a healthy donor, cell line m34Sk3 14 . Erythromycin, IKr channel blocker, at the concentrations 15-30 µM induced re-entry formation on the sharp end of the obstacle at 2-2.5 Hz, higher frequencies of external stimulation were not captured by the tissue. As discussed above, the re-entry formation is connected with voltage-gated ion channel blockage, which was not found during CP treatment. Arrhythmogenicity diagram on Fig. 3d illustrates that frequencies higher than 4 Hz were not captured by the tissue under 639 µM (~ 13 mg/kg) CP and frequencies higher than 3,33 Hz were not captured by the tissue under 852 µM (~ 18 mg/kg) CP, whereas 4 Hz and 5 Hz were captured in control. Deviation from the external stimulus frequency could be regarded as arrhythmic beats and should be considered when using CP for disease treatment. Moreover, appearance of non-conductive areas in cardiac tissue under CP could develop arrhythmia in vivo. www.nature.com/scientificreports/ Limitations iPSC-CMs are often regarded as immature electrophysiologically 34,35 . Nevertherless, in this research tissue of 50 days old human iPSC-CMs were used for cardiotoxicity testing. Electrophysiological characteristics of these cells were measured previously 26 and reached the same values as in mature cardiomyocytes after the 50th day of differentiation. Selected patient-specific iPSC line (m34Sk3 line) from healthy donor is well studied by authors in all aspects related to this work: the structure of hiPSC-CMs tissue 26 , maturation 26 , reference tests for arrhythmogenicity with hERG channels blockage 14,26 , studies of electrophysiology and ion channels (E4031) 26 . The selected differentiation protocol was optimized and published earlier 14 . Stable differentiation efficiency was obtained 15,26 (Fig. 1). However, since one line of patient-specific human iPSC (m34Sk3) is used in this work, each particular conclusion cannot be generalized to other cell lines of human iPSC-CMs.
Compliance with ethical standards. The cell line is provided by the E. Meshalkin Novosibirsk Scientific Research Institute of Circulation Pathology and handling approved by the Institute of Circulation Pathology Ethics Committee (#27, March 21, 2013). The generation of iPSC line from cells donated by patient with informed consent described in Refs. 14,15 . All experiments and procedures were performed in accordance with principles for human experimentation as defined in the 1964 Declaration of Helsinki and its later amendments and were approved by the Scientific Council of the MIPT Life Science Center.
All applicable international, national, and/or institutional guidelines for the care and use of cell lines were followed.

Conclusions
This study examined the conduction properties and cytoskeleton structure of hiPSC cardiac cells (m34Sk3 patient-specific cell line) obtained from a healthy donor under low doses (i.e., < 100 mg/kg) of CP. CP has distinct negative effect on cardiac tissue in vitro which is manifested in the reduction of maximum capture rate (the maximum rate at which each stimulus from the electrode was followed by a response) and the conductivity area.
Maximum capture rate decreased up to 25% ± 7% after application of 852 µM (~ 18 mg/kg) cyclophosphamide for ≤ 10 min. The conductivity area of excitation wave was 66 ± 15% under the influence 1065 µM (~ 22 mg/kg) of CP for ≤ 10 min. CP has no direct effect on re-entry formation, measured according earlier proposed method of cardiac tissue optical mapping 13,14 . It could be explained by absence of effect on voltage-gated ion channels in patch-clamp experiments in this study on hiPSC cardiac cells (m34Sk3 patient-specific cell line) obtained from a healthy donor. Immunocytochemical labeling revealed an exposure time-dependent disruption of α-actinin, which indicated the CP influence on the cardiomyocytes structure and, consequently, on conduction properties of the cardiac tissue. Thus, CP arrhythmogenicity testing showed deviation from the external stimulus frequency under 639 µM (~ 13 mg/kg) and 852 µM (~ 18 mg/kg) CP and appearance of non-conductive areas in cardiac tissue, which is potentially arrhythmogenic and could develop arrhythmic effects in vivo.

Data availability
The datasets generated and analysed during the current study, plugins for data processing are available from the corresponding author on reasonable request.