Sublethal TKI exposure induces disarrangement of sarcomere and alteration abnormal Ca 2+ -handling in cardiomyocytes. We utilized a well-established protocol for CM differentiation11, which consistently generates high-purity beating CMs derived from human stem cells. To optimize the the dose of TKIs in our study, we first determined the no-observed-adverse-effect-levels (NOAELs) of these drugs for CM differentiation with dose-responsive experiments (Fig. S1A). The NOAEL for TKI drugs (imatinib, sunitinib and vandetanib) was 250 nM, which is lower than their concentrations in blood based on the clinical data 12–17.
TKI exposures were conducted using two experimental designs (Fig. 1A), including: Exposure I, a brief early exposure design, in which the TKIs were added from day 7 (early beating CMs) to day 12 (hereafter referred to short period exposure to TKIs); Exposure II, and a chronic exposure strategy, in which cells were exposed to TKI drugs from day 7 until CMs were collected (after day 20) for analyses. No significant differences in efficiency of CM differentiation were observed between control and TKI-treated groups (Fig. 1B and Fig. S1B). However, in Exposure II, TKIs induced significant disarrangements of myofilaments in differentiated CMs (Fig. 1C-1E), suggesting obvious toxic effects on CMs from chronic exposure to TKI at sublethal level of 250 nM. To evaluate the impacts of drug-exposure on differentiated CMs, spontaneous Ca2+-transients were analyzed in the differentiated CMs after day 30. We observed TKI-induced abnormal Ca2+-handling in both Exposure I and II, including decreases in beating rates and amplitude, increases in time to peak, transient duration 90 and 50 (TD90 and TD50), and impaired calcium recycling (i.e., increased decay tau) (Fig. 1F-1I). Both Exposure I and II of TKIs exhibited similar effects on Ca2+-handling of differentiated CMs.
To determine whether these toxic effects were specific to anti-cancer TKIs, we also evaluated other drugs associated with possible cardiotoxicity at their NOAEL levels, including immunosuppressants (tacrolimus and mycophenolate) and thalidomide18–22 (Fig. S1A). We found that NOEALs of tacrolimus, mycophenolic acid, and thalidomide only caused subtle levels of dysfunction in Ca2+-handling in differentiated CMs (Fig.S2D); while no significant morphological change was observed (Fig. 1C and Fig. S2A-2C). These results demonstrate that, at NOAELs, exposure to TKI drugs caused more severe effects during CM differentiation than non-TKI drugs used in this study.
Transcriptomic analysis revealed correlation between the transcriptional regulatory network and mitochondrial function. To elucidate the changes in gene expression during CM differentiation due to TKI exposure, we performed genome-wide transcriptomic analysis of differentiated CMs using RNA-sequencing (RNA-seq) experiments. Twelve modules (i.e., networks) were identified by weighted correlation network analysis (WGCNA) (Fig. 2A and Fig.S3A), and the full lists of genes and enriched GO terms are shown in Supplemental Tables S1 and S2. Clustering analysis of the transcriptomic profiles showed high concordance between Exposure I and II in TKI-treated CMs (except Exposure I of imatinib) (Fig. 2B). WGCNA analysis identifies a specific gene network (module 1) that is down-regulated in TKI-treated compared to non-TKI groups (i.e., immunosuppressant drugs, thalidomide, and 0.1% DMSO) (Fig. 2B-2C). Many genes known to regulate cardiac development and functions were assigned to this module, such as GATA4, MEF2A, TBX20, and HAND2 (Fig. 2D), and the representative enriched-GO terms of this module are related to “heart development” and “mitochondrial respiratory chain complex I assembly” (Fig. 2E). By transcriptomic profiling of the genes involved in Ca2+-handling, ion channels, and β-adrenergic signaling, we showed that TKIs caused down-regulation of RYR2, CAMK2D, and MYH7, which are closely related to the regulation of CM contraction (Fig. S3B).
In addition, genes involved in oxidative phosphorylation (OXPHOS) (such as NDUFV3 and NDUFA12) and glycolysis (such as HK1 and ALDOA) were also assigned to this module (Fig. 2D). Whereas exposure to TKIs generally led to down-regulated gene expressions in this module, exposure to immunosuppressant and thalidomide drugs had the opposite effect (Fig. S3B), suggesting that TKIs induced adverse effects on both cardiac functions and metabolisms are likely related to the dysregulation of transcription factors (TFs) (such as GATA4) and metabolic genes, both of which exhibited some correlations. Differentiated CMs mainly use OXPHOS to support their large ATP demands 23–25. Therefore, we examined mitochondrial respiratory activity by measuring oxygen consumption rate (OCR). We found that chronic exposure to TKIs decreased the levels of both basal respiration and maximal respiration in differentiated CMs (Fig. 2F). The ATP production was also decreased in TKI-treated cells, and sunitinib-treated cells exhibited the lowest ATP level compared to other groups (Fig. 2G).
These results connect TKI exposure to dysregulation of genes involved in cardiac maturation, contractile function, and metabolism, leading to impaired mitochondrial OXPHOS and decreased ATP productions, and ultimately weakened contraction of CMs. These results exhibited some similarities with a single study by Wang et al. 26, which demonstrated that acute TKI exposure impaired OXPHOS in differentiated CMs.
Genome-wide chromatin-accessibility analysis reveals features of transcription-factor occupancy in TKI-treated cardiomyocytes. Since many of the genes disrupted by TKI exposure encode key transcription factors that are important for CM differentiation (Fig. 2D), we leveraged ATAC-seq to identify chromatin accessibility and altered TF occupancy27,28. Specifically, we measured TF-motif enrichments within differential ATAC-seq peaks between control and TKI-treatment groups, and identified several motifs that are significantly enriched in regions where chromatin accessibility was lost in the TKI-treated groups compared to control (Fig. 3A). These enriched motifs correspond to several TF families that are known to play important roles in cardiac differentiation and functions (such as GATAs, TEADs, and MEF2 family members) (Fig. 3B and Fig. S4A). The control group exhibited higher densities of motifs for GATA4 and TEAD1/3 (Fig. 3C), as well as stronger ATAC-seq signals (accessibilities) around the summit of the motifs for GATA4 and TEAD1/3, compared to that of TKI-treated cells (Fig. 3D-3F and Fig. S4B). The sunitinib-treated cells showed the most significant loss of chromatin accessibility surrounding the TF motifs, suggesting that sunitinib exposure may have the strongest adverse effects on DNA-binding of these TFs in CMs. Altogether, these results suggest that exposure to TKIs leads to significant changes in the binding patterns of these TFs, leading to transcriptional dysregulation that may ultimately underlie the TKI-induced toxicity in CMs.
Integrative genome-wide transcriptomic and chromatin-accessibility analyses revealed GATA4 as an important regulator in response to TKI exposure. Using paired expression and chromatin accessibility (PECA) model, we further integrated both chromatin landscapes and gene expression data to build the transcriptional regulatory networks. It can elucidate the effect of TFs bound to the activated context-specific cis-regulatory elements on the transcription of target genes 29,30. We found that genes within the module 1 from the WGCNA, which are involved in mitochondrial biogenesis and function, are predicted to be regulated by transcription factors known to control heart development, such as GATA4, MEF2A/C and TEAD1/3 (Fig. 4A). These genes include peroxisome proliferator-activated receptor alpha (PPARA) and peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PPARGC1A, encoding PGC-1α, which is a key regulator for mitochondrial biogenesis), complex I (i.e., NDUF gene family) 31–33, and cardiac development (such as TBX5) 34 (Fig. 4A). Importantly, the expression of these TFs was down-regulated in differentiated CMs upon exposure to TKIs (Fig. 4B). In particular, GATA4 exhibits high degree (i.e., more targeted genes likely regulated by GATA4) within the network, indicating important roles of GATA4 in TKI-induced transcriptional dysregulation.
Direct mapping of GATA4 occupancy by ChIP-seq in CM identifies global loss of GATA4-binding in response to TKI exposure (Fig. 4C). Genes regulated by GATA4 are related to cardiac structure (“actin filament binding” and “PDZ domain binding”), homeostasis of ion channels (“ion channel binding”), and signaling pathways (“SMAD binding”). By contrast, sunitinib exposure attenuated the enrichment of relevant GO terms in drug-treated CMs, consistent with loss of GATA4-binding within these regions (Fig. 4D), demonstrating sunitinib induced adverse effects on cardiac dysfunction and sarcomere structure via inhibition of transcriptional activity of GATA4.
It was reported that GATA4 interacts with bromodomain-containing protein 4 (BRD4) to regulate mitochondrial biogenesis and Ppargc1a 35 in CMs of adult mouse. In our study, ChIP-seq experiments carried out in drug-treated CMs identifies a loss of GATA4-binding signals on PPARGC1A (Fig. 4E). Moreover, specific analysis of GATA4 occupancy near PPARGC1A identifies a GATA4-binding site approximately 42-Kb upstream of PPARGC1A (Fig. 4F and Table S3). Loss of GATA4 occupancy at this putative PPARGC1A enhancer in sunitinib-treated CMs was also observed (Fig. 4F). Further integration of the current GATA4 ChIP-seq data with promoter-capture Hi-C data previously reported in CMs 36, we confirmed a direct interaction between the GATA4-binding site with the PPARGC1A promoter anchor in CMs, consistent with long-range enhancer-promoter regulation through DNA looping (Fig. 4G). Thus, we hypothesize that TKIs induce mitochondrial dysfunction in CMs via inhibition of GATA4-mediated networks, ultimately leading to cardiac dysfunctions. This led us to investigate the role of GATA4 in CMs upon exposure to TKI, and to examine whether CMs and their mitochondrial function can be restored through inducing GATA4 expression.
GATA4 upregulation sufficiently restored both sarcomere structure and mitochondrial functions in cardiomyocytes upon TKI exposure. In test our hypothesis above, we created a lentiviral delivery-based dCas9/CRISPR-activation (dCas9/CRISPRa) system based on previous studies 37,38, which allows us to find out whether induction of GATA4 expression will ameliorates TKI-induced toxicity in differentiated CMs. The inducing GATA4 expression via this dCas9/CRISPRa system is doxycycline-inducible (Fig. 5A and Fig. S5A). Because our RNA-seq data showed that GATA4 expression was down-regulated in CMs upon exposure to TKI drugs (Fig. S5B), 2 µg/ml doxycycline was added daily to induce the expression GATA4 from day 7 (Fig. 5B-5C and Fig.S5C).
We observed that the CRSIPR-activation of GATA4 expression has reinstated the disorganization of sarcomeres in TKI treated CMs (Fig. 5D-5E and Fig.S6A-S6C). In addition, in spontaneous calcium transient analysis also indicated that the impaired Ca2+-handling signaling of TKI-treated CMs was restored by GATA4 induction, including increased beating rate and amplitude, faster calcium recycling as the time to peak, TD90, TD50, and decay tau were all decreased (Fig. 5F-5G and Fig.S6D-S6E). These results suggest that TKI-induced toxicity in CMs can be reversibly recovered by enhancing GATA4 expression.
Regarding the mitochondria-related gene, PPARGC1A, as expected its expression was downregulated in differentiate CMs upon TKI exposure and, conversely, upregulated by induction of GATA4 (Fig. 6A). We also observed that chronic TKI exposure significantly reduced mitochondrial DNA (mtDNA) copies compared to control cells, whereas induction of GATA4 increased mtDNA levels (Fig. 6B). In addition, TKI exposure decreased branch size of the mitochondria in differentiated CMs compared to control, which was reverted by the induction of GATA4 (Fig. 6C-6K). Moreover, improved basal and maximal respiration in differentiated CMs were observed after induction of GATA4 in the presence of chronic exposure to the TKIs (Fig. 6L-6N), and this enhancement of mitochondrial respiration by induction of GATA4 was also observed in the control group in the absence of drug exposure (Fig. 6O). In addition, we found that enhancing GATA4 expression promoted ATP production and mitochondrial membrane potential (Fig. 6P-6Q) in the TKI-treated CMs, and the most significant improvement of mitochondrial function was observed in sunitinib-treated group.
Both GATA4 and mitochondria are two important factors controlling cardiac functions and heart development. Besides adult stages, mitochondrial biogenesis plays an important role during heart development, and heart development is highly associated with mitochondrial dynamics, biogenesis, and oxidative metabolism 25,39−41. Thus, in the following section, we investigated the role of GATA4 in mitochondrial biogenesis during differentiation processes (without TKI exposure).
GATA4 regulates mitochondrial biogenesis during early cardiomyocyte differentiation. To understand the relationship between GATA4 expression and mitochondrial biogenesis, we examined the expression of GATA4 during CM differentiation. We observed that GATA4 expression was increased from cardiac progenitor stage (day 6) towards differentiated stage (day 36), which is closely related to the increase of mtDNA contents were also increased during CM differentiation (Fig.S7A-S7B). This trend recapitulated the increases of cardiac mitochondria during both in-vivo and in-vitro cardiac differentiation 40,42. In addition, PPARGC1A, the key regulator for mitochondrial biogenesis, was also evaluated after day 6 (Fig.S7C). These results suggested a coordinated increase between GATA4 and mitochondrial biogenesis.
To examine the role of GATA4 during cardiac differentiation, we conducted inducible knockdown of GATA4 expression from day 7 (i.e., early beating CMs) or day 30 (differentiated CMs), using a lentiviral delivery-based shRNA system (Fig. 7A). This allows us to compare the temporal role of GATA4 at different stages.
We find that knockdown of GATA4 during CM differentiation caused decreases in mtDNA contents, ATP production and mitochondrial potential (Fig. 7B-7D), as well as mitochondrial branch sizes in differentiated CMs, compared to that in control group (Fig. 7E-7G). From Seahorse experiments, we observed that the differentiated CMs exhibited lower mitochondrial respiratory activity after knockdown of GATA4, including attenuated levels of both basal respiration and maximal respiration (Fig. 7H-7I). Knockdown of GATA4 in early beating CMs and later stage CMs exhibited similar results, suggesting that GATA4 effects a long differentiation period of mitochondrial biogenesis and function.
We additionally explored the dynamics of genomic occupancy of GATA4 during CM differentiation using ChIP-seq experiments. GATA4-targeted genes are mainly related to “heart development”, and it showed that GATA4 likely regulated VEGF signaling and the Rho GTPase; genes involved in “muscle contraction” and “mRNA splicing” were also targeted by GATA4 at later differentiation stages (Fig.S7D-S7E). We additionally identified a number of genes related to mitochondrial biogenesis (i.e., PPARGC1A) and respiratory chain complexes (i.e., the NDUF gene family) that were enriched for GATA4 ChIP-seq signal (Table S4). Visual inspection of rising ChIP-seq peak signals confirmed the occupancy of GATA4 on PPARGC1A during cardiac differentiation (from day 8 to day 20) in Fig. 7J. Our qPCR confirmed decreases in PPARGC1A expression is associated with the knockdown of GATA4 (Fig. 7K). Our data also indicates the occupation of GATA4 on other genes that regulate mitochondrial biogenesis and maturation. For example, we observed that estrogen-related receptor gamma (ESRRG, known as an effector of PGC-1α43,44) was targeted by GATA4 (Fig.S8F). ESRRG was known to regulate the expression of cardiac genes via interaction with GATA4 and GATA4-independent metabolic genes 45. Moreover, we also observed that the motif for TEADs was enriched upon occupancy of GATA4 (Fig.S8G). The TEAD family is reported to regulate both cardiac development and mitochondrial biogenesis46–48, and this suggests an interaction or co-occupancy between GATA4 and TEAD members. Interestingly, TKI exposure caused loss of occupancy of both TFs (Fig. 3 and Fig.S4). All of these results suggested that GATA4 was involved in mitochondrial biogenesis and functional maturation via regulation of several key genes (such as PPARGC1A) related to mitochondrial biogenesis during early CM differentiation.