Identification of genomic biomarkers for anthracycline-induced cardiotoxicity in human iPSC-derived cardiomyocytes: an in vitro repeated exposure toxicity approach for safety assessment
- 2.2k Downloads
The currently available techniques for the safety evaluation of candidate drugs are usually cost-intensive and time-consuming and are often insufficient to predict human relevant cardiotoxicity. The purpose of this study was to develop an in vitro repeated exposure toxicity methodology allowing the identification of predictive genomics biomarkers of functional relevance for drug-induced cardiotoxicity in human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs). The hiPSC-CMs were incubated with 156 nM doxorubicin, which is a well-characterized cardiotoxicant, for 2 or 6 days followed by washout of the test compound and further incubation in compound-free culture medium until day 14 after the onset of exposure. An xCELLigence Real-Time Cell Analyser was used to monitor doxorubicin-induced cytotoxicity while also monitoring functional alterations of cardiomyocytes by counting of the beating frequency of cardiomyocytes. Unlike single exposure, repeated doxorubicin exposure resulted in long-term arrhythmic beating in hiPSC-CMs accompanied by significant cytotoxicity. Global gene expression changes were studied using microarrays and bioinformatics tools. Analysis of the transcriptomic data revealed early expression signatures of genes involved in formation of sarcomeric structures, regulation of ion homeostasis and induction of apoptosis. Eighty-four significantly deregulated genes related to cardiac functions, stress and apoptosis were validated using real-time PCR. The expression of the 84 genes was further studied by real-time PCR in hiPSC-CMs incubated with daunorubicin and mitoxantrone, further anthracycline family members that are also known to induce cardiotoxicity. A panel of 35 genes was deregulated by all three anthracycline family members and can therefore be expected to predict the cardiotoxicity of compounds acting by similar mechanisms as doxorubicin, daunorubicin or mitoxantrone. The identified gene panel can be applied in the safety assessment of novel drug candidates as well as available therapeutics to identify compounds that may cause cardiotoxicity.
KeywordsCardiotoxicity Human stem cells derived cardiomyocytes Heart failure Transcriptomics Genomic biomarkers In vitro test system Safety assessment
Drug-induced cardiotoxicity is a major safety issue and has to be considered during drug development. Various in vivo and in vitro assays have been developed to assess the adverse effect of lead compounds on cardiac functions. Due to interspecies physiological differences, these assays often do not correctly predict the actual adverse effects of drug candidates on the human heart. Moreover, traditional approaches to toxicological testing involve extensive animal studies, thus making testing costly and time-consuming. Although primate and human primary cardiomyocytes represent highly relevant cell systems, their use is limited by ethical reasons and difficult availability (Anson et al. 2011). Above all, the pharmaceutical industry is struggling with the costly withdrawal of drugs from the market due to toxic effects, often related to cardiotoxicity (Tafuri et al. 2013). Therefore, there is an urgent need for the development of a sensitive, robust and clinically relevant in vitro system with cardiomyocytes for efficacy and safety assessment.
Human embryonic stem cells (hESCs) and human induced pluripotent stem cell (hiPSC)-derived cardiomyocytes have a high physiological relevance and show typical drug-induced changes in electrophysiological properties (Caspi et al. 2009; Reppel et al. 2005; He et al. 2003). Reproducible and large-scale production of highly purified hESCs/hiPSC-derived cardiomyocytes (hESC-CMs/hiPSC-CMs) makes them an attractive source for human cardiotoxicity tests. It is expected that human cardiomyocytes will increase the predictive ability of the adverse effects of potential drugs in humans and may replace or reduce cardiac safety assessment assays based on animal-derived primary cardiomyocytes or cardiac ion channel overexpressing cell lines (Steel et al. 2009).
Among anti-cancer drugs, anthracycline family members such as doxorubicin, daunorubicin and mitoxantrone are known to induce cardiotoxicity (Menna et al. 2012; Paul et al. 2007). Multiple mechanisms such as free radical formation, lipid peroxidation and DNA damage have been proposed to explain the cardiotoxicity of anthracyclines. Additionally, interactions of anthracyclines with the DNA-topoisomerase complex or directly with DNA by intercalation result in disturbances in DNA replication and transcription and have been extensively studied (Minotti et al. 2004). Dose-dependent cardiotoxicity of anthracyclines limits their therapeutic application. Drugs that do not compromise the electrophysiology of the heart can also be cardiotoxic by directly damaging cardiomyocytes at both the subcellular and molecular levels via the formation of reactive oxygen species, DNA damage, mitochondrial damage, apoptosis or disturbed molecular signalling events. Elevated levels of cardiac troponin I (cTnI) and cardiac troponin T (cTnT) in blood correlate well with myocardial injury and act as critical plasma biomarkers for the diagnosis of cardiac damage in clinical and preclinical studies (Babuin and Jaffe 2005; O’Brien 2008; Tonomura et al. 2009). However, high levels of these biomarkers occur only after cardiac damage and can be detected for only a few hours after myocardial infarction and cardiotoxic drug treatment. To avoid drug-induced cardiotoxicity in the future, there is an urgent need to develop sensitive and reliable methods to detect or predict early cardiotoxic events.
In the present study, well-characterized hiPSC-CMs were used as an in vitro system of cardiotoxicity in combination with transcriptomics. Among the anthracyclines, doxorubicin is one of the most successful agents for solid and haematological malignancies in both children and adults. Doxorubicin has been extensively studied in a variety of preclinical models and clinical phases. Here, we developed a methodology allowing single and repeated chronic exposures of human cardiomyocytes to doxorubicin (156 nM). The cardiotoxic effects of doxorubicin were monitored by real-time counting of the beating activity and cytotoxicity. Moreover, global gene expression changes were studied using a transcriptomic approach. Doxorubicin-deregulated expression signature of genes was further analysed in a follow-up study using daunorubicin and mitoxantrone, which also belong to the anthracycline family. Our study demonstrates that the integrative use of the xCELLigence Real-Time Cell Analyser (RTCA) cardio system and toxicogenomics offers a methodology to identify cardiotoxic compounds.
Materials and methods
Cardiomyocyte cell culture
All experiments were performed with purified human iCell Cardiomyocytes® (Cellular Dynamics International, Madison, WI, USA), which were derived from hiPSCs. The cardiomyocytes were supplied as a cryopreserved single cell suspension of a 98 % pure population. The cardiomyocytes were a mixture of spontaneously electrically active atrial-, nodal- and ventricular-like myocytes. These cells exhibit typical biochemical, electrophysiological and mechanical characteristics of normal human heart cells with expected responses upon exposure to exogenous agents. Cryopreserved hiPSC-CMs were thawed in iCell cardiomyocytes plating medium (iCell-PM, Cellular Dynamics International, Madison, WI, USA) per the manufacturer’s instructions. For functional studies, thawed cells were directly plated on a fibronectin-coated (5 µg/cm2, 2 h at 37 °C) E-plate Cardio 96 (ACEA Biosciences, San Diego, CA, USA) at approximately a 25 × 103 cells per well density using iCell-PM. For transcriptomic studies, thawed cells were plated on fibronectin-coated (5 µg/cm2, 2 h at 37 °C) 6-well plates at a 0.4 × 106 per well cell density. Two days later, cells were maintained in iCell cardiomyocyte Maintenance Medium (iCell-MM, Cellular Dynamics International, Madison, WI, USA), with a fresh medium change after every 2 days. The cardiomyocytes were cultured in a standard cell culture incubator at 5 % CO2, 37 °C.
The 10 mM stock solutions (in DMSO) of doxorubicin, daunorubicin and mitoxantrone were purchased from Selleck Chemicals. Stock solutions were stored as small volume aliquots in tightly sealed sterile tubes at −80 °C. Drug dilutions were performed in pre-warmed (37 °C) iCell-MM prior to each drug exposure. Doxorubicin was used as the gold standard reference compound to develop experimental methodology.
xCELLigence RTCA Cardio system
The xCELLigence RTCA Cardio system (ACEA Biosciences, San Diego, CA, USA) is an impedance-based platform for monitoring the real-time beating function of cardiomyocytes. It was used to sensitively and quantitatively detect pro-arrhythmic drug effects on cardiac function and to measure cell viability in real time. Impedance measurements were monitored at regular time intervals. The amount of growth area covered in an E-plate Cardio 96 due to cell adhesion was represented as the Cell Index (CI). A high CI indicates more cell adhesion. Before cell plating, the background impedance of E-plate Cardio 96 (ACEA Biosciences, San Diego, CA, USA) was measured using iCell-PM (50 µl per well). The raw data of cell viability, beating activity and the beating amplitude were acquired using the xCELLigence RTCA Cardio system and analysed using RTCA Cardio software version 1.0 (ACEA Biosciences, Inc, San Diego, CA, USA).
Cell samples were homogenised with QIAzol lysis reagent (Qiagen, Hilden, Germany), and the total RNA was extracted and purified using the RNeasy mini kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. A Nanodrop (ND-1000, Thermo Fisher, Langenselbold, Germany) was used for RNA quantification and purity assessment. RNA integrity was confirmed using the Experion™ automated electrophoresis system (Bio-Rad, Munich, Germany). Extracted RNA was subjected to human gene array processing using Affymetrix’s kits, reagents and instrument setup.
Microarray labelling and hybridization
For microarray gene expression studies, 100 ng of total RNA was used as a starting material. The total RNA samples were amplified and labelled using GeneChip 3′ IVT Express Kit per the manufacturer’s instructions (Affymetrix, High Wycombe, UK). The amplified biotin-labelled RNA (aRNA) samples were purified using magnetic beads, and 15 µg of aRNA was fragmented with fragmentation buffer per the manufacturer’s instructions. Then, 12.5 µg of fragmented aRNA was used to hybridize with Affymetrix Human Genome U133 plus 2.0 array along with the hybridization cocktail solution. For microarray hybridization, gene chips were placed in a GeneChip Hybridization Oven-645 (Affymetrix, High Wycombe, UK) for 16 h at 60 rpm and 45 °C. After incubation, the arrays were washed and stained using the Affymetrix HSW kit on GeneChip Fluidics Station-450. The stained arrays were scanned with Affymetrix GeneChip Scanner-3000-7G, and image and quality control assessments were performed with Affymetrix GCOS software. The generated CEL files were used for further statistical analysis.
Microarray statistical data analysis and functional annotation analysis
Array raw data were quantile normalized using the RMA implementation of the R Affy package (Gautier et al. 2004). Differential expression was determined by the linear model implementation of the R Limma package (Minotti et al. 2004) followed by a Benjamini Hochberg multiple testing correction (1 % FDR). To specifically determine the perturbed transcripts, the expression level of transcripts in the doxorubicin (156 nM)-exposed cell samples was pairwise compared with that of day 2 and day 6 control cell samples, while doxorubicin washout cell sample transcripts were compared with day 14 control cell sample transcripts. The size of change was stated with a threshold value of fold change 2 in absolute scale. Choosing only significantly expressed probe sets, k-means cluster analysis was performed after transcript-wise normalization of signal values to a mean of 0 and an SD of 1 using Euclidean distance measurement and k = 6, using the Cluster 3.0 tool from the Eisen laboratory (Eisen et al. 1999). To further investigate biological functions and the pathway involvement of genes, Database for Annotation, Visualization and Integrated Discovery (DAVID) was used for functional annotation and gene ontology (GO) clustering (Dennis et al. 2003). The GeneCards database was also used to investigate annotative information about genes and its relation to human cardiac disorders (Safran et al. 2010).
mRNA expression analysis using RT2 profiler PCR arrays and real-time PCR
Using 300–500 ng of total RNA, a genomic DNA elimination step and cDNA synthesis were performed with the RT2 First Strand kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. For quantitative comparison of mRNA levels, real-time PCR was performed using custom made RT2 Profiler PCR array (96-well plate) (Qiagen, Hilden, Germany). This array contained 84 target genes, 5 housekeeping genes, 1 genomic DNA control, 3 reverse transcription controls and 3 positive PCR controls. Real-time PCR was performed using RT2 SYBR® Green ROX™ qPCR master mix in an Applied Biosystems 7500 FAST Real-Time PCR System in accordance with the manufacturer’s recommended thermal cycling conditions. The relative gene expression analysis was performed using the 2−ΔΔCt method with the RT2 PCR array data analysis online tool. Expression data were normalized using the geometric mean of 5 housekeeping genes—ACTB, B2M, GAPDH, HPRT1 and RPLP0. A cut-off fold change value of 1.9 was set for significantly deregulated genes and later used to generate the gene list used for Venn diagram analysis.
For immunocytochemistry analysis, control, doxorubicin-exposed and washout iPSC-CMs were fixed with ice-cold 99 % methanol (Roth, Karlsruhe, Germany) for 10 min at −20 °C. Then, cells were permeabilized with 0.3 % Triton X-100 (Sigma-Aldrich, Steinheim, Germany) for 20 min at room temperature. Cells were blocked with 5 % bovine serum albumin (Sigma, Steinheim, Germany) for 1 h at room temperature and incubated with anti-sarcomeric alpha actinin (Abcam, 1:200) and anti-cardiac troponin T (Abcam, 1:200) for 1 h at 37 °C. The cells were washed 3 times with phosphate-buffered saline (PBS) with Ca2+ and Mg2+ for 5 min. Primary antibodies were detected using species matched respective Alexa Fluor-488/568-conjugated secondary antibodies (Invitrogen, Darmstadt, Germany) with 1 h incubation at 37 °C. The cells were washed 3 times with PBS for 5 min and then mounted with Prolong® Gold anti-fade mount with DAPI (Invitrogen, Darmstadt, Germany). Cell images were taken with an Axiovert 200 fluorescence microscope and Axiovision 4.3 software (Carl Zeiss).
Experimental setup for single and repeated exposure
Repeated exposure to doxorubicin induced arrhythmic beats in cardiomyocytes
Genome-wide analysis of cells with and without doxorubicin washout identifies clusters of reversibly and irreversibly altered genes
Overrepresented GO categories and pathways of k-means cluster genes
GO:0065004—protein-DNA complex assembly
GO:0006984—ER-nuclear signalling pathway
GO:0005509—calcium ion binding
GO:0044449—contractile fibre part
GO:0008016—regulation of heart contraction
hsa04115:p53 signalling pathway
GO:0043067—regulation of programmed cell death
GO:0042981—regulation of apoptosis
GO:0019752—carboxylic acid metabolic process
GO:0006520—cellular amino acid metabolic process
GO:0000087—M phase of mitotic cell cycle
Doxorubicin exposure down-regulates genes of cardiac function and up-regulates stress-associated genes
Significantly enriched GO categories and pathways by commonly down-regulated genes between DOX-Day2 and DOX-Day6
SLC8A1, TCAP, ACTA1, MYL3, PGAM2, MYH7, MYH6, TNNI3, GJA5, EDNRA, DES, TNNT1, MYOM2, ARG2, RYR2, ASPH, KCNH2, SCN5A, KCNQ1, CASQ2, HRC, SGCA, MB
KIF23, PRC1, TTK, AURKA, AURKB, GTSE1, KIF2C, FRMD5, DES, ANK2, LMOD2, MYC, TOP2A, LMOD3, TUBA1B, ASPM, KIF14, CDC6, CDK1, KIF11, ACTA1, KIF15, TPX2, LDB3, NUSAP1, MYH7, MYH6, MCM3, TNNT1
MYL2, ACTA1, TCAP, MYL3, LDB3, FHL2, MYH7, MYH6, TNNI3, DES, TNNT1, ANK2, DMD, RYR2
GO:0003013—circulatory system process
MYL2, CORIN, TCAP, MYL3, MYH6, NPR3, ATP1A2, TNNI3, EDNRA, RYR2, KCNH2, SCN5A, KCNQ1
PYGM, SRL, RYR2, CASQ2, HRC
hsa04260:Cardiac muscle contraction
SLC8A1, MYL2, MYL3, ATP1B4, COX6A2, RYR2, MYH7, ATP1A2, MYH6, TNNI3, CACNA2D2
hsa05410:Hypertrophic cardiomyopathy (HCM)
SLC8A1, DES, MYL2, MYL3, DMD, RYR2, MYH7, MYH6, TNNI3, CACNA2D2, SGCA
SLC8A1, DES, MYL2, MYL3, DMD, RYR2, MYH7, MYH6, TNNI3, CACNA2D2, SGCA
PRIM1, NME4, NME2, NME3, NME1-NME2, POLE2, RRM2, POLA1, TK1
hsa05412:Arrhythmogenic right ventricular cardiomyopathy (ARVC)
SLC8A1, DES, DMD, RYR2, CACNA2D2, SGCA, CTNNA3
PRIM1, NME4, ADSSL1, NME2, NME3, NME1-NME2, POLE2, PDE1C, RRM2, POLA1, PAICS
Significantly enriched GO categories and pathways by commonly up-regulated genes between DOX-Day2 and DOX-Day6
GO:0006333—chromatin assembly or disassembly
HIST1H2AB, HIST2H2AA3, HIST2H2AA4, HIST4H4, HIST1H4L, HIST1H4 K, HIST1H2AG, HIST1H2AD, HIST1H2AE, HIST2H4A, HIST2H4B, H2BFS, HIST1H4A, HIST1H2BK, HIST1H4B, HIST1H2BI, HIST1H4E, HIST1H4F
GO:0006974—response to DNA damage stimulus
XRCC4, RAD51C, POLH, ZMAT3, RPS27L, RRM2B, SESN1, RNF8, TRIAP1, CDKN1A, CASP3, XPC, BTG2, BAX, UBR5, AEN, DDB2, PCNA, GADD45A
GO:0033554—cellular response to stress
GADD45A, XRCC4, RAD51C, POLH, ZMAT3, RPS27L, RRM2B, SESN1, RNF8, TRIAP1, DHRS2, GPX1, CASP3, CDKN1A, XPC, BTG2, AEN, UBR5, BAX
ZMAT3, GADD45A, BAX, KIT, STK17A, TP53INP1, PMAIP1, APLP1, GPX1, TRIAP1, CASP3, TNFRSF11B, TMEM173, AEN, FAS, TRAF4, PHLDA1, RNF144B, GAS1, NTN1, TNFRSF10A, TNFRSF10C, TNFRSF10B, TNFRSF10D,
GO:0006979—response to oxidative stress
EGFR, GPX1, DHRS2, SDC1, RRM2B, NQO1, ETV5, ADA, OXR1
GO:0050727—regulation of inflammatory response
GPX1, A2 M, MASP1, ACE2, ITGA2, ADA
GDF15, ACE2,FGFR2, ADAMTS17, A2 M, SORD, IGFBPL1, MASP1, NELL2, JAG1, KIT, LSR, ADA, APLP1, BDNF, TNFRSF11B, SERPINE2, COL27A1, FAS, NRG1, TFPI2, PCSK5, GFOD1, THBS4, EGFR, TMEFF2, FLRT2, ICAM4
TRIAP1, GPX1, BDNF, TNFRSF10D, BAX, FAS, NRG1, ANXA4, GSTP1
hsa04115:p53 signalling pathway
ZMAT3, RRM2B, PMAIP1, SESN2, SESN1, EI24, TP53I3, PPM1D, CDKN1A, CASP3, TNFRSF10B, BAX, DDB2, MDM2, FAS, GADD45A
TNFRSF10A, CASP3, TNFRSF10C, TNFRSF10B, TNFRSF10D, BAX, ENDOD1, FAS, PRKX
Long-term up-regulated genes enriched GO categories and pathways
GO:0006917—induction of apoptosis
TNFRSF10A, CDKN1A, ZMAT3, RRM2B, FAS
GO:0033554—cellular response to stress
XRCC4, CDKN1A, ZMAT3, RRM2B, NEFL, ETV5
GO:0006974—response to DNA damage stimulus
XRCC4, CDKN1A, ZMAT3, RRM2B
hsa04115:p53 signalling pathway
CDKN1A, ZMAT3, MDM2, RRM2B, FAS
Validation of deregulated genes by real-time PCR and follow-up by further cardiotoxic compounds
Gene expression studies (fold regulation) by real-time PCR in hiPSC-CMs after 48 h exposure of doxorubicin (156 nM), daunorubicin (10 nM) and mitoxantrone (3 nM)
Doxorubicin-induced sarcomeric deterioration
For reliable evaluation of cardiotoxicity, human relevant models are urgently required. Current preclinical cardiac safety studies have mainly relied on cardiac ion channels, especially the human ether-a-go–go (hERG) channel, as well as in vivo tests. In the present study, we have established an in vitro cardiotoxicity methodology for the monitoring of early as well as chronic cardiotoxicity events at the cellular, functional and genomic level.
Doxorubicin is an established anti-cancer drug with well-known cardiac injury side effects. Its cumulative dose-dependent cardiotoxicity (Appel et al. 2007; Menna et al. 2012) leads to congestive heart failure (Haq et al. 1985; Ibrahim et al. 1999). Doxorubicin has a tendency to disturb cardiac rhythm and in some cases can cause life-threatening arrhythmia or even sudden death (Lacasse and Bolduc 1992). The doxorubicin-induced disturbances in cardiac function can be detected within a few hours or within 24–48 h following doxorubicin administration by electrocardiography in human and animal hearts (Dindogru et al. 1978; Friess et al. 1985; Kehoe et al. 1978).
Our findings demonstrate that, unlike single exposure, repeated doxorubicin exposure caused arrhythmic beating on day 4 and induced substantial cytotoxicity accompanied with decreased beating amplitude on day 6. Doxorubicin-induced arrhythmic beating indicates disturbed cardiac function, while the decreased amplitude reflects reduced contractile force. Our results are in agreement with findings demonstrating that chronic doxorubicin administration reduces contractile function in rabbit hearts (Boucek et al. 1997).
The GO and KEGG pathway analysis of our transcriptomic studies showed that doxorubicin exposure preferentially suppressed the expression of genes involved in cardiac contraction and pathways related to cardiomyopathies. In addition, doxorubicin exposure also deregulated genes with enriched biological processes such as apoptosis, DNA damage and the oxidative stress response. These observations show that in addition to a general stress response, genes involved in sarcomeric and cardiac muscle contraction are more responsive to doxorubicin exposure. Cardiomyocyte sarcomeres are highly organized structures of myofilaments and are involved in mechanical cardiac contraction. Our transcriptomic data showed significant down-regulation of sarcomere genes in DOX-Day2 and DOX-Day6 cells. Similar findings have also been reported in rat cardiomyocytes, in which chronic doxorubicin exposure induces significant degeneration of sarcomeres (Sussman et al. 1997). Our immunocytochemical analysis showed that compared to control and DOX-Day2 cells, repeated doxorubicin exposure resulted in decreased expression of cardiac troponin T and sarcomeric α-actinin proteins in DOX-Day6 cells, and this expression remained at lower levels even after doxorubicin washout in DOX-Day6WO. In addition, disorganization of myofibrillar structures in the DOX-Day6 cardiomyocytes has been observed. Doxorubicin-induced myofibrillar disarray has also been reported in rat ventricular cardiomyocytes (Sussman et al. 1997; Sawyer et al. 2002).
Regulation of ion homeostasis is one of the essential functional elements during cardiac contraction. Intracellular calcium (Ca2+) is the central regulator of cardiac contraction, and its homeostasis is tightly regulated by Ca2+ ion channels, Ca+2 receptors and Ca+2 binding proteins. Our data showed down-regulation of Ca2+-transporting genes such as RYR2, SLC8A1, CACNA1G, CACNA2D2 and ITPR1 in doxorubicin-exposed hiPSC-CMs. Differential regulation of two calcium release channels (RYR2 and ITPR1) has been reported during end stage heart failure (Go et al. 1995). Chronic rabbit heart studies have shown alterations in Ca2+ release causing abnormalities of contractions and relaxation in doxorubicin-induced cardiomyopathy (Dodd et al. 1993). In addition, we found down-regulation of sodium (Na+) and potassium (K+) ion channel encoding genes, such as KCNQ1, KCNK3, KCNN2, KCNH2, SCN2B and SCN5A. Deregulated Na+ and K+ ion channels play an important role in cardiac arrhythmias and heart failure (Remme and Bezzina 2010; Nabauer and Kaab 1998). In summary, our results are in accordance with animal and clinical studies demonstrating that doxorubicin induces disturbances in cardiac calcium homeostasis as well as altered sodium and potassium ion channel activity. In the present study, doxorubicin exposure significantly deregulated ion homeostasis maintaining genes at the mRNA level accompanied with functional changes in the beating behaviour of hiPSC-CMs.
Mitochondrial dysfunction has been suggested to be involved in doxorubicin-induced cardiotoxicity. However, the exact mechanisms of the suppressive effects by doxorubicin on the mitochondrial electron transport chain, oxidative metabolism and ATP synthesis are not fully understood. As a crucial component of the mitochondrial electron transport chain, cytochrome C oxidase (CCo) and uncoupling protein (UCP) activity influences mitochondrial function at the ATP level. In rat hearts, doxorubicin treatment reduced CCo subunit expression (Chandran et al. 2009) and also down-regulated Ucp2 and Ucp3 expression (Bugger et al. 2011). Down-regulation of UCPs and CCo genes showed an inverse relationship with increased oxidative stress (Akhmedov et al. 2015; Bugger et al. 2011; Srinivasan and Avadhani 2012); CCo dysfunction also has a direct effect on cellular ATP levels. In agreement with these observations, our data also indicated a down-regulation of UCP2 and CCo (Cytochrome C Oxidase Subunit VIa Polypeptide 2) expression, whereas up-regulation was observed for oxidative stress-responsive genes such as NQO1, OXR, GCH1 and GPX1 in doxorubicin-exposed hiPSC-CMs (supplementary Fig. S5). Similarly to our findings, the reduced activity of creatine kinase muscle (CKM) and myoglobin (MB) has been reported in human failing hearts (Braunlin et al. 1986; Nascimben et al. 1996; Obrien et al. 1992). Decreased levels of CKM impair the ATP delivery process to energy-consuming systems, and decreased levels of MB disturb oxygen diffusion and the mechanical functions of the cardiac muscle. In conclusion, the present results suggest that doxorubicin induces increased oxidative stress and impairs mitochondrial ATP synthesis and delivery in cardiomyocytes. Overall, these intracellular mechanisms contribute to the impaired cardiac function observed after doxorubicin treatment in cancer patients.
In accordance with our findings, the expression level of apoptosis genes including BAX and FAS has also been found to be up-regulated in human failing hearts (Latif et al. 2000; Sheppard et al. 2005). Increased expression of BAX and FAS may induce apoptosis and reduce the chances of myocardial recovery. Similarly to our results, increased expression levels of ACE2, NRG1, DUSP4 and LIF have also been found in human heart failure (Goulter et al. 2004; Yan and Morgan 2011; Communal et al. 2002; Eiken et al. 2001). KCNJ2 is also up-regulated in human dilated cardiomyopathy (Szuts et al. 2013). Notably, GDF15 and GPNMB (patent publication number—WO2012072752 A1) have been proposed as diagnostic biomarkers of heart failure (Wang et al. 2010; Kempf and Wollert 2009; Khan et al. 2009) and were also up-regulated in our model system. Therefore, the gene expression responses observed in our established in vitro system showed a high degree of similarity to those of the human heart in vivo.
The Venn diagram analysis of differentially expressed genes in cells exposed to doxorubicin, daunorubicin and mitoxantrone exhibited an overlap of 27 down- and 8 up-regulated genes. The 27 down-regulated genes are mainly involved in sarcomere structure and the regulation of ion homeostasis. Up-regulated genes mainly indicate a general stress response, and they included stress markers such as BAX, FAS, GPX1 and ZMAT3. This observation may help to better understand cellular mechanisms underlying late apoptosis inducing cardiac cell loss many years after anthracycline treatment. Although GDF15 has been reported to represent a biomarker for heart failure, elevated levels have also been found in the cell systems of liver, lung and kidney injury (Hsiao et al. 2000; Zimmers et al. 2005). Thus, it can be interpreted as a marker which indicates cell injury in multiple tissues. The identified 35 genes in the overlap of all three anthracyclines represent an anthracycline-responsive gene consensus expression signature and could be applied as a predictive toxicity signature for potential cardiotoxicants that act by similar mechanisms as doxorubicin, daunorubicin or mitoxantrone.
In the present study, a concentration of 156 nM doxorubicin was chosen because in the hiPSC-CMs this concentration compromises the contractility without causing major cytotoxic effects. At doses of bolus administration of doxorubicin varying between 15 and 90 mg/m2, the initial maximal plasma concentrations in patients are approximately 5 µM (Gewirtz 1999). After this initial peak, the plasma concentration of doxorubicin decreases rapidly, to the range of 25–250 nM, within 1 h. Similar plasma concentrations have also been reported in patients receiving continuous infusions of doxorubicin (Gewirtz 1999). In addition, after a 50 mg/m2 intravenous injection in adult acute myeloid leukaemia patients, a daunorubicin peak plasma concentration range from 120 to 560 nM on day 1 and 8 to 610 nM on day 3 has been observed (Lofgren et al. 2007). The mitoxantrone peak plasma concentrations have been reported to vary from 0.46 to 2.49 µM after 1 h infusion of mitoxantrone (12 mg/m2) and then decrease rapidly to around 10 nM within 5 h (Sundman-Engberg et al. 1993). Therefore, the concentration chosen in the present study is within the therapeutic range of anthracyclines therapy.
In summary, the results obtained by anthracyclines in hiPSC-CMs recapitulated the disturbed cardiac function observed in vivo and in clinical studies. Doxorubicin-induced adverse effects on cardiac function can be detected much earlier at the genomic level before cytotoxicity and arrhythmia can be observed. The combined application of hiPSC-CMs, the xCELLigence RTCA Cardio system and transcriptomics resulted in the identification of an anthracycline consensus signature representing early biological processes that significantly contribute to better understanding of the cardiotoxic effects of compounds both at a cellular and molecular level. The present methodology can allow for first-line in vitro preclinical tests and reduce animal usage in drug safety studies and the costs of safety evaluation. Although it is very likely that the methodology possesses a low false negative rate for severely cardiotoxic compounds, however, whether this approach can avoid false negatives for mildly cardiotoxic compounds and false positives for non-cardiotoxic compounds should be demonstrated by screening of several non- and mild cardiotoxicants.
This work was supported by the ‘Detection of endpoints and biomarkers for repeated dose toxicity using in vitro systems’ (DETECTIVE) Project (FP7 Health Programme, European Commission).
- Chandran K, Aggarwal D, Migrino RQ, Joseph J, McAllister D, Konorev EA, Antholine WE, Zielonka J, Srinivasan S, Avadhani NG, Kalyanaraman B (2009) Doxorubicin inactivates myocardial cytochrome c oxidase in rats: cardioprotection by Mito-Q. Biophys J 96:1388–1398CrossRefPubMedPubMedCentralGoogle Scholar
- Eiken HG, Oie E, Damas JK, Yndestad A, Bjerkeli V, Aass H, Simonsen S, Geiran OR, Tonnessen T, Christensen G, Froland SS, Gullestad L, Attramadal H, Aukrust P (2001) Myocardial gene expression of leukaemia inhibitory factor, interleukin-6 and glycoprotein 130 in end-stage human heart failure. Eur J Clin Invest 31:389–397CrossRefPubMedGoogle Scholar
- Eisen MB, Spellman PT, Brown PO, Botstein D (1999) Cluster analysis and display of genome-wide expression patterns (vol 95, pg 14863, 1998). Proc Natl Acad Sci USA 96:10943Google Scholar
- Ibrahim NK, Hortobagyi GN, Ewer M, Ali MK, Asmar L, Theriault RL, Fraschini G, Frye DK, Buzdar AU (1999) Doxorubicin-induced congestive heart failure in elderly patients with metastatic breast cancer, with long-term follow-up: the M.D. Anderson experience. Cancer Chemother Pharmacol 43:471–478CrossRefPubMedGoogle Scholar
- Lofgren C, Lehmann S, Jonsson-Videsater K, Mollgard L, Linder O, Tidefelt U, Hassan M, Paul C (2007) Higher plasma but not intracellular concentrations after infusion with liposomal daunorubicin compared with conventional daunorubicin in adult acute myeloid leukemia. Ther Drug Monit 29:626–631CrossRefPubMedGoogle Scholar
- Sawyer DB, Zuppinger C, Miller TA, Eppenberger HM, Suter TM (2002) Modulation of anthracycline-induced myofibrillar disarray in rat ventricular myocytes by neuregulin-1 beta and anti-erbB2: potential mechanism for trastuzumab-induced cardiotoxicity. Circulation 105:1551–1554CrossRefPubMedGoogle Scholar
- Szuts V, Menesi D, Varga-Orvos Z, Zvara A, Houshmand N, Bitay M, Bogats G, Virag L, Baczko I, Szalontai B, Geramipoor A, Cotella D, Wettwer E, Ravens U, Deak F, Puskas LG, Papp JG, Kiss I, Varro A, Jost N (2013) Altered expression of genes for Kir ion channels in dilated cardiomyopathy. Can J Physiol Pharmacol 91:648–656CrossRefPubMedGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.