The Integrated Stress Response Potentiates Ponatinib-Induced Cardiotoxicity

Abstract

Background:

Mitochondrial dysfunction is a primary driver of cardiac contractile failure, yet the crosstalk between mitochondrial energetics and signaling regulation remains obscure. Ponatinib, a tyrosine kinase inhibitor (TKI) used to treat chronic myeloid leukemia, is among the most cardiotoxic TKIs and causes mitochondrial dysfunction. Whether ponatinib-induced mitochondrial dysfunction triggers the integrated stress response (ISR) to induce ponatinib-induced cardiotoxicity remains to be determined.

Methods:

Using human induced pluripotent stem cells-derived cardiomyocytes (hiPSC-CMs) and a recently developed mouse model of ponatinib-induced cardiotoxicity, we performed proteomic analysis, molecular and biochemical assays to investigate the relationship between ponatinib-induced mitochondrial stress and ISR and their role in promoting ponatinib-induced cardiotoxicity.

Results:

Proteomic analysis revealed that ponatinib activated the integrated stress response (ISR) in cardiac cells. We identified general control nonderepressible 2 (GCN2) as the eIF2α kinase responsible for relaying mitochondrial stress signals to trigger the primary ISR effector – Activating Transcription Factor 4 (ATF4), upon ponatinib exposure. Mechanistically, ponatinib treatment exerted inhibitory effects on ATP synthase activity and reduced its expression levels resulting in ATP deficits. Perturbed mitochondrial function resulting in ATP deficits then acts as a trigger of GCN2-mediated ISR activation, effects that were negated by nicotinamide mononucleotide, an NAD⁺ precursor, supplementation. Genetic inhibition of ATP synthase also activated GCN2. Interestingly, we showed that the decreased abundance of ATP also facilitated direct binding of ponatinib to GCN2, unexpectedly causing its activation most likely because of a conformational change in its structure. Importantly, administering an ISR inhibition, ISRIB, protected hiPSC-CMs against ponatinib. Ponatinib-treated mice also exhibited reduced cardiac function, effects that were attenuated upon systemic ISRIB administration. Importantly, ISRIB does not affect the antitumor effects of ponatinib in vitro.

Conclusions:

Neutralizing ISR hyperactivation could prevent or reverse ponatinib-induced cardiotoxicity. The findings that compromised ATP production potentiates GCN2-mediated ISR activation have broad implications across various cardiac diseases. Our results also highlight an unanticipated role of ponatinib in causing direct activation of a kinase target despite its role as an ATP-competitive kinase inhibitor.

Graphical Abstract

INTRODUCTION

Cancer is the second leading cause of death after cardiovascular diseases¹. Improvements in cancer therapy have resulted in overall improved survivorship. However, cancer drugs are often associated with cardiotoxicities^2–4. Hence, it is critical to understand the pathophysiology of cancer therapy-associated cardiotoxicity. Chronic myeloid leukemia (CML) is caused by a reciprocal translocation between chromosomes 9 and 22 t(9;22q34;q11) or the Philadelphia (Ph) chromosome resulting in the breakpoint cluster region-Abelson (BCR-ABL) fusion gene^5–7. Ponatinib is a third-generation tyrosine kinase inhibitor (TKI) currently approved for treating CML patients with the gatekeeper mutation BCR-ABL T315I, for which no alternative therapy exists since earlier TKIs are ineffective against this mutation^8–10. Unfortunately, serious cardiac adverse events were reported in patients given ponatinib, leading to its temporary suspension from the market¹¹. Ponatinib has been linked to cardiomyopathy, heart failure, and vascular occlusion in up to 9% of patients^12,13. Since there is no alternative treatment option for CML patients with the T315I mutation, it is crucial to better understand the molecular underpinnings of ponatinib-induced cardiotoxicity and develop strategies to mitigate its adverse cardiotoxic effects.

The heart must adapt to stress conditions that occur as a result of numerous genetic and/or environmental factors. The integrated stress response (ISR) is one of the circuits that responds to stress and serves to restore proteostasis by regulating protein synthesis^14–16. ISR is activated by the phosphorylation of the α subunit of the eukaryotic translation initiation factor (eIF2α) at Ser51. In mammalian cells, this is catalyzed by a family of four eIF2α kinases that are activated by distinct stress stimuli, namely, general controlled non-repressed (GCN2), protein kinase R (PKR), heme-regulated inhibitor (HRI) and PKR-like endoplasmic reticulum kinase (PERK), which sense amino acid depletion, viral double-stranded RNA, heme deficiency and endoplasmic reticulum stress, respectively^15,17–19. Phosphorylation of eIF2α (p-eIF2α) attenuates global protein synthesis but increases the translation of specific mRNAs, including activating transcription factor 4 (ATF4), the master transcriptional regulator of the ISR. This allows the cell to adapt to stress conditions or, alternatively, undergo apoptosis depending on the duration and intensity of cellular stress^15,20–24. Whether the ISR is activated and plays a protective or detrimental role in ponatinib-induced cardiotoxicity is currently unknown.

Mitochondria are essential organelles in cardiomyocytes (CMs) that play a central role in energy homeostasis, metabolism, and signaling^25–27. Various studies have identified a conserved transcriptional program associated with ISR activation that is induced by mitochondrial dysfunction^28–33. Crucially, differences in the various cellular responses to different mitochondrial insults appear to, at least partially, dictate how distinct cells and tissues cope with mitochondrial stress. Ponatinib has been reported to cause mitochondrial toxicity³⁴. We have previously validated the use of human induced pluripotent stem cells (hiPSCs) in defining cancer drug-induced toxicities^35,36. However, whether and how mitochondrial dysfunction triggers the ISR in ponatinib-treated hiPSC-CMs, to what extent ISR activation represents protective mechanisms or contributes to cardiac pathology, and whether the interdependence between these two pathways can be therapeutically targeted to reduce ponatinib-induced cardiotoxicity remain unclear.

Here, we employed both in vitro and in vivo methods to explore the mechanisms underlying ponatinib-induced cardiotoxicity. We show that ponatinib-induced cardiotoxicity is associated with a signature of ISR activation that requires the eIF2α kinase GCN2. Mechanistically, we discover that mitochondrial dysfunction leading to inhibition of ATP synthesis acts as the trigger to activate GCN2. Genetic inhibition of ATP synthase also resulted in GCN2 activation confirming reduced ATP synthesis as a trigger of GCN2-mediated ISR activation. Interestingly, in the case of ponatinib, we provide evidence that reduced concentrations of ATP facilitate direct binding of ponatinib to GCN2, leading to its paradoxical activation of the latter. We further reveal that activation of the ISR promotes ponatinib-induced cardiotoxicity, as both inhibition of GCN2 and the ISR using a small molecular inhibitor, ISRIB, lead to improved outcomes in vitro without affecting the antitumor properties of ponatinib. These results were further validated in vivo using a mouse model of ponatinib-induced cardiotoxicity.

METHODS

Data Availability:

Detailed methods are presented in the Supplementary Material online, Methods. The Major Resources Table is provided in the Supplementary Materials.

RESULTS

Ponatinib induces cellular damage in human iPSC-CMs

We first sought to establish whether human iPSCs can be used to recapitulate ponatinib-induced cardiotoxicity seen clinically and its relevance for in-depth molecular investigations. Three healthy human iPSC lines were differentiated into cardiomyocytes (hiPSC-CMs) according to our previous protocol^35,37, and all hiPSC-CMs exhibited spontaneous beating after day 15 of differentiation (Figure 1A). We initially assessed the effects of ponatinib on cell viability. hiPSC-CMs were treated with ponatinib for 24 hours over a range of concentrations (0.5 to 10 μM). We observed that there was a dose-dependent decrease in cell viability in response to ponatinib compared to the DMSO control group. In agreement with previous studies indicating an LD50 of 4–6 μM ponatinib in hiPSC-CMs^34,38, our results showed that 3 and 10 μM ponatinib treatment, results in cell viability as low as 42.7±1.29% and 13.4±1.15%, respectively (Figure 1B). To avoid excessive cell death confounding our results, subsequent experiments were performed using 1 μM ponatinib, a concentration that elicited ~10% cell death at 24 hours. In support of ponatinib inducing cellular apoptosis, we found that ponatinib treatment elevated the level of cleaved poly (ADP)-ribose polymerase family (c-PARP), which acts as a marker of apoptosis (Figure 1C). Moreover, a significant elevation in cleaved caspase 3 levels was evident in hiPSC-CMs subjected to ponatinib treatment (Figure 1C). Furthermore, immunofluorescence staining of hiPSC-CMs revealed that ponatinib induced significant DNA double-strand breaks, as indicated by an increased average number of γ-H2AX foci in ponatinib-treated hiPSC-CMs compared to vehicle-treated cells (Figure 1D). We also assessed the effects of ponatinib treatment on the functional properties of hiPSC-CMs by measuring contractility using high-speed video microscopy with motion vector analysis³⁹. As expected, we found a marked decrease in the velocity and frequency of contraction of ponatinib-treated hiPSC-CMs (Figure 1E). In summary, these data confirm the role of ponatinib treatment in cardiotoxicity and hiPSC-CMs’ impaired function and serves as a valid platform to investigate the pathophysiology underlying ponatinib-induced cardiotoxicity.

Figure 1. Exposure of hiPSC-CMs to ponatinib leads to cardiotoxicity and impaired cardiomyocyte function.

A, Schematic of the differentiation protocol from hiPSCs to cardiomyocytes. B, Quantification of cell viability by PrestoBlue after 24 hours of treatment with DMSO vehicle (gray) or ponatinib (pink) (ranging from 0.5 to 10 μM). The viability was normalized to the vehicle control. n=9, data represent 3 individual hiPSC lines from 3 independent experiments per group. C, Western blot analysis of c-PARP and c-caspase 3 in hiPSC-CMs after DMSO or ponatinib (1 μM) treatment for 24 hours. GAPDH was used as the loading control. n=9, data represent 3 individual iPSC lines from 3 independent experiments per group. D, Representative images (left) and quantification (right) of immunostaining of the DNA damage marker γ-H2AX (green) in hiPSC-CMs after ponatinib treatment (1 μM) for 24 hours. Cardiomyocytes were counterstained with cardiac troponin T (cTnT) (red), a cardiac marker and DAPI (blue). Scale bar=10 μm. n=6, data represent 3 individual hiPSC lines from 2 independent experiments per group. At least 300 cells were counted in each group. E, Representative contractility traces (left) of hiPSC-CMs after ponatinib treatment (1 μM) for 24 hours. Quantification of hiPSC-CMs beating rates using high-speed video microscopy with motion vector analysis. n=6, data represent 3 individual hiPSC lines from 2 independent experiments per group. Data are presented as mean ± SEM. Data were analyzed using the Kruskal-Wallis test with the Dunn multiple comparisons test in B, and Mann-Whitney U test in C through E.

Inhibition of transcription and translation in ponatinib-treated hiPSC-CMs

We next sought to identify signaling pathways that may explain ponatinib-induced cardiotoxicity. To this end, we decided to take advantage of the reverse-phase protein array (RPPA) platform. This high-throughput proteomics method utilizes antibody binding to quantify protein expression and post-translational modifications, including phosphorylation, acetylation, and protein cleavage (Figure 2A). Considering that the RPPA platform is primarily directed at molecular targets linked to cancer initiation and progression and has been extensively used in cancer research, best evidenced by its inclusion in The Cancer Genome Atlas (TCGA) analysis, we reasoned that this platform may also help define how ponatinib causes cardiotoxicity⁴⁰. Interestingly, we found that there was a significant decrease in many of the measured proteins following ponatinib treatment, suggesting a decline in global protein synthesis, accompanied by upregulation of selected proteins such as asparagine synthetase (ASNS) and alanine, serine, cysteine-preferring transporter 2 (ASCT2) (Figure 2B). To further verify the expression changes of proteins measured by RPPA, we next assessed the levels of ASCT2 and ASNS after treatment with ponatinib by immunoblotting. Consistent with the RPPA results, we observed a significant upregulation of ASCT2 and ASNS at both the mRNA (1.81±0.46-fold and 1.96±0.25-fold, respectively, versus DMSO) and protein levels (Figure 2C–E). Conversely, RPPA revealed inhibition of mTOR signaling, which is necessary for protein synthesis, as indicated by the concomitant decrease in phosphorylated mTOR and ribosomal protein S6 (Figure 2B), results that we validated by immunoblotting (Figure S1A). Ponatinib also increases the levels of phosphorylated AMPK, which is known to inhibit mTOR signaling (Figure S1A)⁴¹. To directly investigate the effects of ponatinib on the global rate of protein biosynthesis, we measured the translation rate in ponatinib-treated cells by puromycin pulse-chase labeling. As expected, the abundance of puromycylated proteins in ponatinib-treated cells was significantly lower than that in DMSO-treated cells (Figure 2F and 2G). We also made use of a click chemistry method based on fluorescent tracking of homopropargylglycine (HPG), a methionine analog that is incorporated into proteins during active protein synthesis and found a significant decline in HPG intensity upon ponatinib exposure compared to DMSO (Figure 2H). Reduced protein synthesis can be caused by limited nucleotide production for RNA synthesis and gene transcription. As judged by the fluorescence intensity of 5-ethynyl uridine (EU), which labels newly synthesized RNA, there was a significant reduction in the number of EU-positive cells following ponatinib exposure, thereby indicating that ponatinib also negatively affects transcription (Figure 2I). Altogether, these findings suggest that ponatinib significantly represses transcription and translation in hiPSC-CMs.

Figure 2. Ponatinib significantly inhibits transcription and translation in hiPSC-CMs.

A, Schematic illustration of the RPPA workflow. B, Heatmap based on RPPA data from 2 hiPSC cell lines (n=4). C, mRNA expression analysis of ASCT2 and ASNS upon 24 hours of treatment with ponatinib (1 μM). n=6, data represent 3 individual hiPSC lines from 2 independent experiments per group. D, Western blot analysis showing the increased expression of ASCT2 and ASNS after 24 hours of ponatinib (1 μM) treatment. GAPDH was used as the loading control. E, Quantification of the western blot results in D. n=9, data represent 3 individual hiPSC lines from 3 independent experiments per group. F, hiPSC-CMs were treated with 10 μM puromycin after ponatinib or DMSO treatment. Puromycin is incorporated into new protein synthesis. Western blot analysis with an antibody against puromycin indicated that ponatinib treatment inhibited protein synthesis. GAPDH was used as the loading control. G, Quantification of puromycin incorporation in F. n=6, data represent 3 individual hiPSC lines from 2 independent experiments per group. H, Newly synthesized proteins were fluorescently labeled in hiPSC-CMs incubated with HPG (green) and DAPI (blue) after ponatinib or DMSO treatment. Representative images (left) and quantification (right) of fluorescent intensity showing a significant decrease in translation levels in ponatinib-treated hiPSC-CMs compared to DMSO-treated hiPSC-CMs. Scale bar=60 μm. n=9, data represent 3 individual hiPSC lines. At least 300 cells were counted in each group. I, Confocal images (left) and scatterplot (right) illustrating EU labeling (green) intensity in hiPSC-CMs 24 hours after ponatinib (1 μM) treatment. Nuclei of hiPSC-CMs were marked by DAPI (blue). Scale bars: 50 μm in the left panel, 10 μm in the right panel. n=21, data represent 3 individual hiPSC lines, from 7 independent experiments per group. At least 300 cells were counted in each group. Data are presented as mean ± SEM. Data were analyzed using the Mann-Whitney U test in C, E, G and H and two-tailed t test in I.

Activation of the integrated stress response (ISR) in ponatinib-treated hiPSC-CMs

In response to diverse environmental and pathological conditions, activation of the ISR leads to global reduction of protein synthesis and the induction of selected mRNAs, including the transcription factor ATF4, to restore cellular homeostasis¹⁵. Translation initiation requires the formation of a ternary complex of methionyl transfer RNA, guanosine 5’-triphosphate (GTP), and the eukaryotic initiation factor 2 (eIF2) complex. The activity of eIF2B, a guanine exchange factor (GEF) that allows continuous ribosome assembly and mRNA translation, is tightly regulated by phosphorylation of the α subunit of eIF2 (p-eIF2α). p-eIF2α inhibits the GEF activity of eIF2B and attenuates global translation during ISR^42–44. Since we observed that most of the proteins differentially affected by ponatinib as measured by RPPA were decreased, along with upregulated levels of ASNS and ASCT2, both of which are known targets of ATF4⁴⁵, a key transcription factor that is upregulated by ISR, we hypothesized that ponatinib elicits stress in cardiomyocytes leading to the activation of ISR. Comparing the protein levels of ATF4 and p-eIF2α in hiPSC-CMs exposed to ponatinib, we observed that the expression of these markers markedly increased (Figure 3A). In further support of ISR activation, ponatinib-treated hiPSC-CMs exhibited elevated mRNA levels of ATF4 (Figure 3B) and signature ISR effector genes, including CHOP and ATF3 (Figure S2A). There are four kinases that mediate the phosphorylation of eIF2α, namely, GCN2, PERK, PKR, and HRI, each of which is activated by different cellular stresses (Figure 3C)¹⁵. To elucidate the driving forces behind ISR activation in ponatinib-treated hiPSC-CMs, we measured the ponatinib-induced ATF4 increase in hiPSC-CMs along with shRNA-mediated knockdown of these four kinases individually (Figure 3D). The knockdown efficacy of each shRNA was confirmed by both RT-qPCR and immunoblotting (Figure 3D and Figure S2B–S2C). Knockdown of GCN2 profoundly inhibited the upregulation of ATF4 by ponatinib, while the knockdown of PERK, PKR, and HRI had no effects (Figure 3D and Figure S2C). This finding suggests that GCN2 activity is responsible for the activation of ISR in ponatinib-treated hiPSC-CMs. Indeed, by following the kinetics of phosphorylated GCN2 and eIF2α as well as ATF4 in hiPSC-CMs treated for 4, 8, 16, and 24 hours, we observed an increase in all these proteins as early as 4 hours after treatment (Figure 3E and Figure S2D). Conversely, we did not observe any changes in the phosphorylation of PERK or PKR (commercially available phospho-HRI antibody is unavailable, Figure S2D). Activation of the unfolded protein response (UPR), which is induced by endoplasmic reticulum (ER) stress, is known to elevate ATF4. PERK is one of the three main branches in regulating the UPR along with activation transcription factor (ATF6) and serine/threonine-protein kinase/endoribonuclease inositol-requiring protein 1α (IRE1α)^46–49. To further rule out the involvement of PERK, we analyzed the expression of the ER chaperone BiP/GRP78, ATF6, and XBP1 s (spliced X-box-binding protein 1, activated by IRE1α) and found no difference in any of these targets between control and ponatinib-treated hiPSC-CMs (Figure S3A–S3C). Consistent with the role of GCN2 in activating ISR, the levels of CHOP, an established target of ATF4, were reduced only in ponatinib-treated hiPSC-CMs coupled with GCN2 knockdown but not the other three kinases or control shRNA (Figure S2E). To further ascertain the role of GCN2, we also employed a pharmacological kinase inhibitor of GCN2 (GCN2iB)²⁹. In hiPSC-CMs, treatment with GCN2iB suppressed the increased p-eIF2α and ATF4 induced by ponatinib treatment (Figure 3F). Moreover, GCN2 knockdown significantly prevented the ponatinib-induced increased expression of ATF4 and its target genes, including CHOP, ASNS, and ASCT2 (Figure 3G). Collectively, these results suggest that ponatinib treatment engages GCN2-eIF2α, leading to ISR activation in hiPSC-CMs.

Figure 3. GCN2 is the eIF2α kinase responsible for ponatinib-induced ISR activation in hiPSC-CMs.

A, Western blot analysis of the levels of p-eIF2α, eIF2α and ATF4 upon treatment with ponatinib (1 μM) or DMSO for 24 hours. GAPDH was used as the loading control. n=9, data represent 3 individual hiPSC lines from 3 independent experiments per group. B, mRNA expression analysis of ATF4 after 24 hours of treatment with ponatinib (1 μM). n=6, data represent 3 individual hiPSC lines from 2 independent experiments per group. C, Schematic summarizing multiple stresses that activate the ISR via four eIF2α kinases (GCN2, PERK, PKR, and HRI) to phosphorylate eIF2α, leading to global inhibition of Cap-dependent mRNA translation while distinctly enhancing ATF4 levels. D, hiPSC-CMs were transduced with shGCN2, shPERK, shPKR and shHRI virus and a negative control (shNC) for 24 hours and then incubated with ponatinib and DMSO. Western blot analysis of GCN2, PERK, PKR, HRI and ATF4 levels. GAPDH was used as the loading control. E, Western blot analysis of p-GCN2 and GCN2 in hiPSC-CMs after DMSO or ponatinib (1 μM) treatment for 24 hours. n=6, data represent 3 individual hiPSC lines from 2 independent experiments per group. F, Western blot analysis of the levels of p-eIF2α, eIF2α and ATF4 upon treatment with DMSO or ponatinib (1 μM) or coadministration with GCN2iB (0.5 μM) for 24 hours. GAPDH was used as the loading control. n=4, data represent 2 individual hiPSC lines from 2 independent experiments. G, hiPSC-CMs were transduced with shGCN2 and the negative control (shNC) for 24 hours and then incubated with ponatinib (1 μM) and DMSO. mRNA expression analysis of ATF4 and its target genes (CHOP, ASNS, and ASCT2). n=3, data represent 3 individual hiPSC lines, each line is from an average of 4 independent experiments. Data are presented as mean ± SEM. Data were analyzed using the Mann-Whitney U test in A, B and E, and Kruskal-Wallis test with the Dunn multiple comparisons test in F and 2-way ANOVA followed by the Sidak’s post hoc test in G.

Mitochondrial defects signal the activation of ISR mainly via GCN2 in hiPSC-CMs

We next investigated the upstream mechanisms driving ponatinib-induced activation of GCN2. As amino acid (AA) deprivation is the canonical trigger of GCN2^20,50, we speculated that decreased AA abundance could be responsible for GCN2 activation in ponatinib-treated hiPSC-CMs. Comparison of aminoacyl-tRNA synthetases (ARSs) between DMSO- and ponatinib-treated hiPSC-CMs revealed an upregulation of multiple cytosolic but not mitochondrial ARSs (Figure S4A). However, supplementing hiPSC-CMs with various AAs did not abrogate the deleterious effects of ponatinib on hiPSC-CMs (Figure S4B). Likewise, LC-MS/MS-based intracellular metabolite profiling of DMSO- and ponatinib-treated hiPSC-CMs did not reveal a significant downregulation of AAs (Figure S4C). As ASNS and ASCT2 are involved in the synthesis and transport of AA and they were upregulated by ponatinib (Figure 2B), we next examined these two proteins more carefully. Aspartate supplementation at a high concentration (since most mammalian cells do not efficiently take up aspartate) did not significantly reverse ponatinib-induced ISR activation (Figure S4D)^51,52. Moreover, we did not detect a significant difference in GLS1 and GLUD1, two key enzymes involved in glutaminolysis in which ASCT2 is a key mediator (Figure S4E)^53,54. These multiple lines of evidence demonstrating the absence of AA deficiency suggest that other factors may be involved in GCN2-mediated ISR activation (Figure S4F).

The ISR is also known to be activated by mitochondrial dysfunction^28,29. To explore the possibility that mitochondrial damage inflicted by ponatinib may underlie ISR activation, we first confirmed that ponatinib induces mitochondrial dysfunction in hiPSC-CMs. As expected, ponatinib treatment resulted in a loss of mitochondrial membrane potential, increased cellular and mitochondrial ROS accumulation, a reduced NAD⁺/NADH ratio, and impaired mitochondrial metabolism (both basal and maximal respiration) (Figure 4A–E, Figure S5A). Immunoblotting of purified mitochondrial fractions also revealed reduced ATP5A levels upon ponatinib treatment, supported by a corresponding decrease in complex V activity (Figure 4F and 4G) and total ATP levels (Figure 4H). We also tested whether ponatinib activates the mitochondrial unfolded protein response (UPR^mt)^55–57. However, the expression of UPR^mt-related genes, including CLPP, YME1L1, HSP60, and CPN10, was indistinguishable between vehicle- and ponatinib-treated hiPSC-CMs (Figure S5B). These results confirmed that ponatinib treatment induces mitochondrial dysfunction in hiPSC-CMs without involving the UPR^mt. Mitochondrial dysfunction-dependent ISR activation is not a universal response and instead depends on the nature of the mitochondrial defects²⁹. Hence, we first tested whether increased mitochondrial ROS acts as a trigger of ISR activation in ponatinib-treated hiPSC-CMs. We treated hiPSC-CMs with ponatinib together with a mitochondria-targeted antioxidant (Mito-TEMPO)⁵⁸. Importantly, Mito-TEMPO when used at a concentration that reduced ponatinib-induced mitochondrial ROS accumulation (Figure S5C) failed to prevent ponatinib-induced increased expression of ATF4 and its downstream target genes, including CHOP, ASNS, ASCT2, PUMA, and TRIB3 (Figure S5D and S5E), suggesting that mitochondrial ROS are not sufficient to trigger ISR activation. Since we observed a reduction in both ATP5A and complex V activity in ponatinib-treated hiPSC-CMs, we then tested whether direct inhibition of the ETC, especially complex V, impacts ISR activation. We treated hiPSC-CMs with an inhibitor of complex I (rotenone), complex III (antimycin), or complex V (oligomycin) at concentrations known to impair mitochondrial respiration^29,59. Rotenone and oligomycin both increased p-eIF2α and ATF4 levels, suggesting that inhibiting the ETC leading to mitochondrial dysfunction in hiPSC-CMs can trigger the ISR (Figure 4I and Figure S6A). Between rotenone and oligomycin, the latter had a much stronger effect in inducing ATF4; likewise, CHOP was upregulated by oligomycin but not rotenone (Figure 4I). As such, we carried out subsequent experiments using oligomycin to specifically test whether inhibition of complex V requires GCN2 to activate ISR similar to that of ponatinib. Interestingly, we found that hiPSC-CMs treated with oligomycin also displayed increased phospho-GCN2 (Figure 4J). Accordingly, similar to ponatinib-treated hiPSC-CMs, knockdown of GCN2 in oligomycin-treated hiPSC-CMs significantly dampened the activation of ISR, as reflected by reduced p-eIF2α and ATF4 (Figure 4K and 4L). This was further supported by the reduced mRNA levels of downstream ISR genes, including ASNS and ASCT2 (Figure S6C). Conversely, oligomycin-induced ISR activation was insensitive to knockdown of PERK or PKR but was significantly reduced by HRI depletion, albeit only to a small extent (Figure S6B and S6C). ISRIB treatment also significantly reduced the oligomycin-induced increase in ATF4 (Figure 4M). Thus, ponatinib treatment most likely impairs ATP production because of reduced complex V activity, which is sensed by GCN2, leading to ISR activation. As ATP production is an essential event for energy-demanding CMs that is dysregulated across multiple cardiac diseases, we sought to determine whether our finding that ponatinib-induced impaired ATP production acts as a trigger of GCN2 phosphorylation is broadly applicable. Knockdown of two different subunits of the ATP synthase were achieved by transduction of hiPSC-CMs with lentiviruses encoding shRNAs against ATP5A and ATP5MG, respectively, for a week, verified by immunoblotting (Figure 4N and 4O). As expected, knockdown of ATP5MG^60,61 resulted in a reduction in both ATP synthase monomers and multimers as measured by blue native polyacrylamide gel electrophoresis (BN-PAGE) using an anti-ATP synthase subunit β antibody, suggesting disruption of the mitochondrial complex assembly (Figures S6D). We confirmed that knockdown of the two subunits of the ATP synthase resulted in a significant decrease in ATP levels (Figure 4P). Similar to the effects of ponatinib and oligomycin, we confirmed that genetic knockdown of ATP synthase subunits is also associated with increased phosphorylation of GCN2 (Figure 4N and 4O), further supporting our hypothesis that reduced ATP levels promote GCN2 phosphorylation resulting in ISR activation.

Figure 4. Inhibition of complex V is responsible for GCN2-mediated ISR activation.

A, hiPSC-CMs were treated with DMSO or ponatinib (1 μM) for 24 hours. Mitochondrial membrane potential was measured using a tetramethylrhodamine methyl ester (TMRE) probe. n=9, data represent 3 individual hiPSC lines from 3 independent experiments per group. B, Intracellular hydrogen peroxide (H₂O₂) was measured in hiPSC-CMs. n=9, data represent 3 individual hiPSC lines from 3 independent experiments per group. C, NAD+/NADH was measured in hiPSC-CMs. n=9, data represent 3 individual hiPSC lines from 3 independent experiments per group. D, E, Representative oxygen consumption rate (OCR) and basal respiration, proton leak, maximal respiration and spare respiration capacity in DMSO- or ponatinib (0.5 and 1 μM)-treated hiPSC-CMs for 24 hours. n=12, data represent 3 individual hiPSC lines from 4 technical replicates per group. F, Mitochondria isolated from hiPSC-CMs treated with DMSO or ponatinib (1 μM) for 24 hours. Western blot analysis of mitochondrial respiratory complex subunits (NDUFA9, SDHA, UQCRC, MTCO1, and ATP5A). GAPDH was used as the loading control. n=3, data represent 3 individual hiPSC lines. G, Mitochondria isolated from hiPSC-CMs treated with DMSO or ponatinib (1 μM) for 24 hours. The complex V activity of mitochondria was measured. n=6, data represent 3 individual hiPSC lines from 2 independent experiments per group. H, The relative amount of ATP was measured in hiPSC-CMs following ponatinib or DMSO treatment. n=6, data represent 3 individual hiPSC lines from 2 independent experiments per group. I, hiPSC-CMs were treated with a mitochondrial electron transport chain (ETC) inhibitor for 24 hours, the complex I inhibitor rotenone (1 μM and 2.5 μM), the complex III inhibitor antimycin (1 μM and 2.5 μM) or the ATP synthase inhibitor oligomycin (1 μM and 2.5 μM). Western blot analysis of ATF4 and CHOP expression. GAPDH was used as the loading control. J, hiPSC-CMs were incubated with oligomycin (2.5 μM) and DMSO for 24 hours. Western blot analysis of the levels of p-GCN2 and GCN2. n=6, data represent 3 individual hiPSC lines from 2 independent experiments per group. K, hiPSC-CMs were transduced with shGCN2 and the negative control (shNC) for 24 hours and then incubated with oligomycin (2.5 μM) and DMSO for 24 hours. ATF4 mRNA expression analysis. n=3, data represent 3 individual hiPSC lines, each line is from an average of 4 independent experiments. L, hiPSC-CMs were transduced with shGCN2 and the negative control (shNC) for 24 hours and then incubated with oligomycin (2.5 μM) and DMSO for 24 hours. Western blot analysis of the levels of GCN2, ATF4, p-eIF2α and eIF2α. GAPDH was used as the loading control. M, Western blot analysis of p-GCN2, GCN2, ATF4, p-eIF2α and eIF2α after 24 hours of drug treatment in DMSO, oligomycin (2.5 μM), ISRIB (200 nM) or co-administration with ISRIB and oligomycin. GAPDH was used as the loading control. n=6, data represent 3 individual hiPSC lines from 2 independent experiments per group. N, O, hiPSC-CMs were transduced with shATP5A, shATP5MG or the negative control (shNC) for 7 days. Western blot analysis of the levels of pGCN2, GCN2, ATP5A and ATP5MG. GAPDH was used as the loading control. n=3, data represent 3 individual hiPSC lines. P, The relative amount of ATP was measured in hiPSC-CMs after transduction with shATP5A, shATP5MG, or the negative control (shNC). n=6, data represent 3 individual iPSC lines from 2 independent experiments. Data are presented as mean ± SEM. Data were analyzed using the Mann-Whitney U test in A through C, F through H, J, N and O, 1-way ANOVA followed by Tukey post hoc in E, 2-way ANOVA followed by the Sidak’s post hoc test in K, and Kruskal-Wallis test with the Dunn multiple comparisons test in M and P.

To further explore the notion that ATP levels determine ISR activation, we asked whether supplementation of ponatinib-treated hiPSC-CMs with nicotinamide nucleotide (NMN), an NAD⁺ precursor, can alleviate the dysfunction caused by ponatinib since complex V is essential for sustaining ATP levels^62–64. Encouragingly, we found that restoring ATP levels with NMN significantly rescued the deterioration of cell viability in ponatinib-treated hiPSC-CMs (Figure 5A). In addition, NMN treatment inhibited p-eIF2α and ATF4 levels, indicating that increasing ATP levels protected hiPSC-CMs against ponatinib-induced mitochondrial damage and GCN2-mediated ISR activation (Figure 5B and 5C). We next examined why decreased ATP levels promote GCN2-mediated ISR activation in the presence of ponatinib. GCN2 has been reported to form constitutive dimers in an inactive antiparallel conformation. Upon activation, it undergoes a conformational change and reorganizes into a parallel active dimer^65,66. A recent study reported a paradoxical observation that neratinib, an ATP-competitive kinase inhibitor, binds to the GCN2 kinase domain promoting its conformational change from an inhibited antiparallel conformation to a parallel active form, resulting in GCN2 activation in the absence of amino acid starvation⁶⁷. As ponatinib is also an ATP-competitive kinase inhibitor, we asked whether ponatinib might directly bind to GCN2 kinase, leading to its direct activation. To assess whether ponatinib competes with ATP to bind GCN2 kinase domain leading to its activation in a similar manner, we performed an ATP displacement assay by incubating recombinant GCN2 with uncharged tRNAs, ponatinib, and a desthiobiotin-ATP probe, which biotinylates GCN2 when the kinase domain is accessible to ATP, followed by avidin capture of biotinylated GCN2. Interestingly, we observed that ponatinib reduced the biotinylation of GCN2 and increased the elution of unbiotinylated GCN2, suggesting that ponatinib competes with ATP to bind to GCN2 (Figure 5D), which may explain why a reduction in ATP facilitates ponatinib-mediated GCN2 activation through increased direct binding. To validate the induction of GCN2 phosphorylation at the biologically relevant site (T899) by ponatinib, we conducted the GCN2 activity using unlabeled ATP and a specific T899 P-GCN2 antibody. Furthermore, we confirmed that ponatinib induced phosphorylation of recombinant GCN2 at the T899 phosphorylation site in a dose-dependent manner (Figure 5E). To determine how ponatinib could potentially elevate GCN2 kinase activity despite apparently occupying the ATP-binding pocket of GCN2, we employed a broader range of ponatinib concentrations. Intriguingly, we observed a biphasic dose response curve, where lower concentrations of ponatinib augmented GCN2 kinase activity, while higher concentration led to its inhibition (Figure 5F). Next, we conducted kinetic analyses by titrating ATP and measuring GCN2 activation (Figure 5G). The addition of ponatinib increased the reaction rate and reduced the ATP concentration required to reach half of the maximal velocity (Figure 5H and 5I). These results suggest that ponatinib binds to one of the GCN2 monomers, leading to conformation alteration that associates with ATP, resulting in an increase in GCN2 enzyme activity. Upon further escalation of ponatinib concentration, both kinase domains become occupied, ultimately leading to reduction in GCN2 kinase activity (Figure 5J).

Figure 5. Ponatinib directly binds to and activates GCN2.

A, Quantification of cell viability by PrestoBlue after 24 hours of DMSO, ponatinib (1 μM) or co-administration with an NAD+ precursor β nicotinamide nucleotide (β-NMN, 200 μM) treatment. The viability was normalized to the vehicle control. n=6, data represent 3 individual hiPSC lines from 2 independent experiments per group. B, C, Western blot analysis of p-eIF2α, eIF2α, ATF4 and c-PARP after 24 hours of drug treatment with DMSO, ponatinib (1 μM) or co-administration with β-NMN (200 μM). GAPDH was used as the loading control. n=3, data represent 3 individual hiPSC lines. D, Schematic of the ponatinib-binding ATP competition assay. Western blot analysis of GCN2 levels in avidin capture (ATP-bound) and eluate (ponatinib-bound) proteins. E, GCN2 activity assay showing autophosphorylation of GCN2 at T899 in the presence of ponatinib. n=3 biological samples. F, The activity assay demonstrates the autophosphorylation of GCN2 at T899 and the phosphorylation of eIF2α at S51, both induced by ponatinib. n=3, data represent 3 individual hiPSC lines. G, Immunoblots of p-GCN2 and GCN2 from in vitro kinase reactions with 7.5 nM GCN2 and indicated concentrations of ATP in the presence of DMSO or 1 μM ponatinib. H, Ratio of active (p-GCN2) to total (GCN2) GCN2 in the reactions fit to Michaelis-Menten nonlinear model. I, Velocity (Vmax) for GCN2 and ATP in presence of DMSO or 1 μM ponatinib derived from panel. n=3 biological samples. J, A model depicts the activation of GCN2 kinase at intermediate concentrations of ponatinib, while higher concentrations result in kinase inhibition. Data are presented as the mean ± SEM. Data were analyzed using Kruskal-Wallis test with the Dunn multiple comparisons test in A, C, E and F, and Mann-Whitney U test in I.

Inhibition of the ISR rescues ponatinib-induced cardiotoxicity in vitro

Because of its central role in regulating cellular homeostasis and survival, the ISR has been implicated in both adaptive and pathophysiological outcomes¹⁵. To evaluate the consequences of ISR activation in ponatinib-treated hiPSC-CMs, we utilized ISRIB, a novel small molecule that blocks the downstream effects of eIF2α phosphorylation and hence dampens ISR activation^43,68–70. We first confirmed that ISRIB treatment successfully blunted ISR activation in ponatinib-treated hiPSC-CMs, as revealed by reduced ATF4 (both at the mRNA and protein levels, Figure 6A and 6B) and several of its target genes, including CHOP, TRIB3, and GADD34A (Figure 6B and Figure S7A). Interestingly, ISRIB treatment suppressed the expression of apoptotic markers (c-PARP and PUMA) in ponatinib-treated hiPSC-CMs, suggesting that ISR activation may be detrimental to the cells in the context of ponatinib exposure (Figure 6A and 6B). Consistent with this finding, ponatinib-induced cell death was reduced by ISRIB treatment (Figure 6C, Figure S7B and S7C). To further scrutinize the effects of ISRIB against ponatinib, we assessed the sarcomeric structure of ponatinib-treated cells by staining against cardiac troponin T and sarcomeric α-actinin. We found that the percentage of cells exhibiting sarcomeric disarray following ponatinib exposure was significantly lower when ISRIB was present, further confirming the protective effects of inhibiting the ISR against ponatinib-induced toxicity (Figure 6D and 6E). Mirroring the protective effects of ISRIB on the viability and structural integrity of ponatinib-treated hiPSC-CMs, we verified using automated video-based contractility analysis that ISRIB partially but significantly reversed the detrimental effects of ponatinib on the contractile properties of hiPSC-CMs, as revealed by improved velocity (Figure 6F) and beating rate (Figure 6G). These results indicate that ISRIB treatment protected against ponatinib-induced hiPSC-CM damage and dysfunction. Given that ISRIB-mediated ISR inhibition protects against ponatinib-induced toxicity, we asked whether activation of the ISR with salubrinal, a selective inhibitor of eIF2α phosphatase complexes, contributes to ponatinib-induced cardiac damage⁷¹. Immunoblotting confirmed that 10 μM salubrinal treatment induced both p-eIF2α and ATF4 levels in hiPSC-CMs for 24 hours (Figure S7D). In contrast to the protective effects of ISRIB in response to ponatinib, the cell viability assay revealed that salubrinal treatment aggravated the ponatinib-induced damage in hiPSC-CMs (Figure S7E). Taken together, these findings confirmed that inhibition of ISR helps to alleviate ponatinib-induced cardiotoxicity in hiPSC-CMs.

Figure 6. ISRIB is protective against ponatinib-induced cardiotoxicity in vitro.

A, hiPSC-CMs were cultured with or without ponatinib (1 μM) and ISRIB (200 nM) for 24 hours. Western blot analysis of p-eIF2α, eIF2α, ATF4 and c-PARP expression. GAPDH was used as the loading control. n=4, data represent 2 individual hiPSC lines from 2 independent experiments per group. B, mRNA expression analysis of the apoptosis markers PUMA and ATF4 and its targets (CHOP, TRIB3 and GADD34A). n=3, data represent 3 individual hiPSC lines, each line is from an average of 4 independent experiments. C, Quantification of cell viability by PrestoBlue after 24 hours of drug treatment. The viability was normalized to the vehicle control. n=9, data represent 3 individual hiPSC lines from 3 independent experiments per group. D, Representative immunostaining of α-actinin (green), cTnT (red) and DAPI (blue) to show sarcomere disorganization in ponatinib- or ISRIB-treated hiPSC-CMs for 24 hours. Scale bar=10 μm, inset scale bar=5 μm. E, Ratio of sarcomere disorganization in ponatinib- or ISRIB-treated hiPSC-CMs. n=12, data represent 3 individual hiPSC lines. At least 600 cells were counted in each group. F, Representative contractility traces of hiPSC-CMs after ponatinib (1 μM) or ISRIB (200 nM) treatment for 24 hours. G, Statistical results of beating rate using high-speed video microscopy with motion vector analysis in F. n=6, data represent 3 individual hiPSC lines from 2 independent experiments per group. H, hiPSC-CMs were transduced with shGCN2, shPERK, shPKR and shHRI virus and the negative control (shNC) for 24 hours and then incubated with ponatinib (1 μM) and DMSO. Quantification of cell viability by PrestoBlue, normalized to the vehicle control, 24 hours after of drug treatment. n=6, data represent 3 individual hiPSC lines from 2 independent experiments per group. I, Western blot analysis of the levels of c-PARP and c-caspase 3 upon treatment with DMSO or ponatinib (1 μM) or co-administration with GCN2iB (0.5 μM) for 24 hours. GAPDH was used as the loading control. J, Representative bright-field images illustrating the cellular morphology of hiPSC-CMs after DMSO, ponatinib (1 μM) or co-administration with GCN2iB (0.5 μM) for 24 hours. Scale bar=100 μm. K, Representative immunostaining of hiPSC-CMs stained with ImageiT dead green stain (green) and Hoechst stain (blue) after treatment with ponatinib (1 μM) or co-administration with GCN2iB (0.5 μM) for 24 hours. Scale bar=60 μm. n=6, data represent 3 individual hiPSC lines from 2 independent experiments per group. At least 300 cells were counted in each group. Data are presented as the mean ± SEM. Data were analyzed using the Kruskal-Wallis test with the Dunn multiple comparisons test in A, C, G, H and K, and 1-way ANOVA followed by Tukey post hoc in B and E.

Since our results point to GCN2 as the kinase responsible for ponatinib-induced ISR activation, we reasoned that depletion of GCN2 should also protect against ponatinib-induced cardiotoxicity. Indeed, cell viability was significantly restored in ponatinib-treated hiPSC-CMs coupled with GCN2 knockdown but not the other three kinases (Figure 6H), and knockdown of GCN2 significantly prevented ponatinib-induced increased expression of the ISR-mediated apoptosis gene CHOP (Figure 3G) and its downstream genes TRIB3, GADD34A, GADD45A, GADD45B and GADD45G and apoptosis-related genes PUMA, DR5, and BAX (Figure S8A). Similarly, GCN2iB-mediated pharmacological inhibition of GCN2 was found to protect against ponatinib-treated hiPSC-CMs, as indicated by reduced expression of c-PARP and c-CASP3 along with improved viability (Figure 6I–6K, and Figure S8B–S8C), further confirming the detrimental role of ISR activation in promoting ponatinib-induced cardiotoxicity.

Inhibition of the ISR ameliorates ponatinib-induced cardiotoxicity in vivo

To gain insights into whether ISRIB ameliorates ponatinib-induced cardiotoxicity in vivo, we utilized a recently published mouse model of ponatinib-induced cardiotoxicity⁷². 8-week-old ApoE^−/− mice were fed a high-fat diet (HFD) for 8 weeks. These mice were subsequently separated into four treatment groups as follows: (i) vehicle (control); (ii) ISRIB (2.5 mg/kg every other day for 14 days)^73–76; (iii) ponatinib (15 mg/kg/day for 14 days); and (iv) ponatinib and ISRIB while continuing the HFD for all groups (Fig. 7A). This concentration of ponatinib was selected based on studies demonstrating therapeutic response in murine models of malignancy, and mice receiving this dosing regimen of ponatinib exhibited cardiac dysfunction^8,72. As expected, we observed significant cardiac dysfunction and adverse cardiac remodeling induced by ponatinib treatment (Figure 7B–7F). Ponatinib-treated mice exhibited a significant decrease in left ventricular fraction shortening (LVFS) and ejection fraction (LVEF) compared to control mice (Figure 7C and 7D), as well as adverse remodeling (Figure 7E and 7F). Importantly, co-administration of both ISRIB and ponatinib prevented the decline in cardiac function, confirming that ISRIB is cardioprotective against ponatinib in vivo (Figure 7B–7F). We then isolated cardiomyocytes from these mice to measure ISR activation. In line with our in vitro data obtained using hiPSC-CMs, ponatinib-treated myocytes also had increased ISR activation, as indicated by higher levels of p-eIF2α and ATF4, and that ATF4 was reduced when ISRIB was present (Figure 7G). To assess the impact of ponatinib on cardiomyocyte apoptosis, we conducted TUNEL (terminal deoxynucleotidyl transferase dUTP nick end labeling) assay on cardiac tissue sections derived from all four groups. We observed a significant increase in the number of apoptotic cardiac myocytes in ponatinib-treated mice, while co-administration of both ISRIB and ponatinib blunted the extent of apoptosis (Figure 7H). These data confirm that ponatinib-induced cardiotoxicity is mediated via the activation of the ISR in mice and that targeting the ISR helps to mitigate the damage caused by ponatinib.

Figure 7. ISRIB is protective against ponatinib-induced cardiotoxicity in vivo.

A, Experimental outline of ponatinib and ISRIB administration in 8-week-old ApoE−/− mice. Mice were subjected to HFD for 8 weeks, after that mice were received an intraperitoneal injection of ISRIB or vehicle at 2.5 mg/kg once every two days and oral administration of ponatinib or vehicle at 15 mg/kg daily for 2 weeks, followed by in vivo studies. B-F, Cardiac function was examined by echocardiography, and the percentage of left ventricle fractional shortening (LVFS), left ventricle ejection fraction (LVEF), left ventricular internal diameter at the end of diastole (LVID;d) and left ventricular internal diameter at the end of systole (LVID;s) n=10 mice. G, After treatment with ponatinib or ISRIB for 2 weeks, cardiomyocytes were isolated from the mice. Western blot analysis of the levels of p-GCN2, GCN2, p-eIF2α, eIF2α and ATF4. GAPDH was used as the loading control. n=6 biological samples. H, Confocal images (left) and scatterplot (right) illustrating TUNEL (red) intensity on cardiac tissue sections. Nuclei were marked by DAPI (blue). Cardiac myocytes were marked by α-actinin (green), Scale bars: 10 μm in the left panel. n=15, 5 biological samples. At least 600 cells were counted in each group. I, Proposed model of ponatinib-induced cardiotoxicity. Data are presented as the mean ± SEM. Data were analyzed using the 1-way ANOVA followed by Tukey post hoc in C through F and H, and Kruskal-Wallis test with the Dunn multiple comparisons test in G.

Although our current results suggest that combining ponatinib with ISRIB could potentially alleviate the prevalence and severity of cardiac injury, it is essential to rule out the possibility that such a strategy could have adverse effects on the antitumor efficacy of ponatinib, which would otherwise negate the use of ISRIB. The application of ponatinib is restricted to treating tumors carrying T315I-mutated BCR-ABL, which occurs in CML, whereby first- and second-generation inhibitors such as imatinib and nilotinib are ineffective^77,78. Accordingly, we tested whether ISRIB negatively affects the potency of ponatinib against the imatinib-resistant K562 (K562-r) human CML cell line⁷⁹. As expected, we found that ponatinib treatment significantly decreased the viability of K562-r cells in a dose-dependent manner Figure S9A). Strikingly, the simultaneous administration of ISRIB and ponatinib not only did not diminish the effects of ponatinib but also further heightened the loss of K562-r cell viability caused by ponatinib alone Figure S9B), indicating that ISRIB treatment protected against ponatinib-induced cardiotoxicity and enhanced the effectiveness of ponatinib against K562-r cells, confirming ISRIB as an amenable therapeutic target to reduce ponatinib-induced cardiotoxicity without compromising its antitumor properties against CML.

DISCUSSION

Ponatinib is among the most cardiotoxic TKIs that have been approved by the FDA¹¹. Although recent efforts have been made in engineering new ponatinib analogs that retain antitumor efficacy with a reduced cardiotoxic profile^80–82, these studies remain experimental, highlighting an unmet gap in better understanding the mechanisms underlying ponatinib-induced cardiotoxicity. Using human iPSC-CMs as a platform to study ponatinib-induced cardiotoxicity, we show that ponatinib-induced cardiac injury is associated with activation of the ISR. Our results further suggest that this activation occurs upon sensing mitochondrial dysfunction caused by ponatinib-mediated inhibition of ATP synthase activity and that GCN2 is the eIF2α kinase involved in relaying these mitochondrial stress signals (Figure 7I). Unexpectedly, we found that the lack of ATP facilitates ponatinib to directly bind to and activate GCN2, an observation that is in contrast with its role in inhibiting kinases. Our results also indicate that ISR activation is detrimental in the context of ponatinib-induced cardiotoxicity, as we found that treatment with ISRIB, a small molecule inhibitor of the ISR, protects cardiomyocytes against ponatinib both in vitro and in vivo, demonstrating that targeting the ISR may be a viable strategy for ameliorating ponatinib-induced cardiotoxicity.

Recognizing that animal models of cardiotoxicity may fail to faithfully reflect the fundamental biology or cardiotoxic responses of the human myocardium, an increasing number of studies have instead relied on iPSC-CMs as an in vitro research tool for understanding the drug-induced cardiotoxicity of cancer drugs^34,35,83. Using iPSC-CMs, Talbert and colleagues reported that ponatinib when used at 10–50x of its C_max (~0.2 μM) led to excessive cardiotoxicity³⁴ and was associated with the inhibition of pro-survival signaling pathways, including the AKT and ERK pathways, results that were corroborated by an independent study using zebrafish and neonatal rat cardiomyocytes³⁸. In this study, we opted for a lower concentration of ponatinib at 1 μM to better understand the underlying mechanisms of ponatinib-induced injury without being confounded by excessive cell death. By combining hiPSC-CMs and RPPA, our proteomics analysis revealed that ponatinib causes ISR activation in hiPSC-CMs, effects that we confirmed independently in isolated cardiomyocytes from ponatinib-treated mice. It remains to be determined whether our results demonstrating that ponatinib activates ISR extend to other major cell types within the heart, including cardiac endothelial cells and fibroblasts.

Although multiple studies have shown that mitochondrial dysfunction is linked to activation of the ISR ^28,29,84, it is also appreciated that the link between mitochondrial dysfunction and the ISR is not universal. Instead, recent evidence suggests that stress responses elicited by different mitochondrial insults differ in a cell-type and stress signal-specific manner²⁹. For example, the depletion of DARS2 (mitochondrial aspartyl-tRNA synthetase) in heart and skeletal muscle leads to the deregulation of mitochondrial protein synthesis in both tissues, yet the activation of adaptive responses is observed predominantly in cardiomyocytes but not skeletal muscle cells⁸⁵. Likewise, a separate study showed that multiple pathways connect mitochondrial dysfunction to ISR activation, dependent on the nature of the mitochondrial defect and the metabolic state of the cell²⁹. Adding another layer of complexity is to define which of the four eIF2α kinases is responsible for relaying mitochondrial defects to trigger ISR, which has only recently started to be unraveled. In this context, we note that ponatinib-induced ISR activation in hiPSC-CMs is dependent on GCN2, as only the inhibition of GCN2 but not other eIF2α kinases blunted the levels of ATF4.

Interestingly, our results further suggest that ponatinib-induced mitochondrial injury in the form of ATP synthase inhibition is the stimulus responsible for activating the ISR via GCN2, which is supported by several observations. First, we observed that ponatinib treatment led to a decrease in both ATP5A levels and complex V activity. Second, we found that treatment with oligomycin mirrored the effects of ponatinib with marked ISR activation. Interestingly, we found that GCN2, but not the other eIF2α kinases, is necessary for signaling oligomycin-induced mitochondrial dysfunction to induce ISR activation in hiPSC-CMs. This is in contrast with several recent publications reporting that HRI but not GCN2 is responsible for oligomycin-induced ISR activation, indicating that there might be a cell-type specific context at play that warrants further investigation^28,86,87. Of note, while we have shown that ponatinib treatment compromises mitochondrial functions of hiPSC-CMs, further work is needed to unravel the exact mechanisms behind the mode of action underlying ponatinib-induced mitochondrial dysfunction which may not be exactly identical to the direct use of complex I and V inhibitors in this study. Moreover, our results indicate that supplementation with NMN but not amino acids or MitoTEMPO negated ponatinib-induced ISR activation, further suggesting that inhibition of ATP synthase is responsible for GCN2-mediated ISR activation instead of the canonical amino acid deficiency often associated with GCN2 activation. We also present evidence suggesting that ponatinib competes with ATP to bind with GCN2, and our results suggest that the lack of ATP increases the direct binding of ponatinib to one of the two GCN2 monomers, allowing for its paradoxical activation via a conformational change that results in increased enzymatic activity through the unbound monomer, which is in agreement with recent studies showing that certain kinase inhibitors unexpectedly activate instead of inhibiting their bound targets, including GCN2^66,67. Future molecular docking studies between ponatinib and GCN2 are required to confirm this model. There also remains the possibility that ponatinib-induced mitochondrial dysfunction can also activate GCN2/ISR without the need for direct ponatinib binding. This is based on our observation that in the presence of oligomycin alone or when subunits of ATP synthases were genetically inhibited, phosphorylation of GCN2 was also observed.

ISRIB is a small molecule that reverses the effects of phosphorylated eIF2α and has been shown to be protective in mouse models of pulmonary fibrosis and neurological disorders but never in the heart^{74–76,88,89}. Here, we show that ISRIB is a promising agent for conferring cardioprotection against ponatinib in a co-morbidity mouse model (ApoE^−/− mice on HFD) of ponatinib-induced cardiotoxicity. The use of this co-morbidity mouse model is required for ponatinib-induced cardiotoxicity to manifest in mice as we observed only impaired stroke volume, but not ejection fraction and fractional shortening in wild-type mice treated with ponatinib (data not shown), in agreement with a previously published study⁸ highlighting the challenges in using rodent models to study cardiotoxicity in humans. There is a need for further in-depth interrogation of ISRIB-mediated cardioprotective effects against ponatinib in this ApoE^−/− mouse model because of the potential presence of other confounders. Notably, our results indicate that aside from its cardioprotective effects, ISRIB appears to enhance the cancer-killing effects of ponatinib against K562-r cells, which is in agreement with studies reporting that inhibition of ISR/GCN2 appears to increase the susceptibility of cancer cells to cell death and that the ISR is tumorigenic^67,90–92. Future studies employing the use of CML lines harboring the T315I mutation will strengthen the potential of ISRIB as a therapeutic option. Moreover, in vivo validation of ISRIB as a cardioprotective agent against ponatinib while retaining its efficacy against tumor growth is warranted. Additionally, it is important to highlight that prolonged systemic targeting of the ISR may not be ideal. Hence, the GCN2-ATF4 pathway that we describe here could potentially be targeted therapeutically to block ISR activation only in cardiac cells that experience mitochondrial dysfunction without globally blocking the ISR in all cells.

Taken together, our study reveals for the first time that ponatinib-induced ISR activation is a major contributor to cardiac dysfunction both in vitro and in vivo. Future work includes a more targeted dosage and time-response of ponatinib in both animal and hiPSC-CM models. These studies will be important to recapitulate the dosage and duration of ponatinib administration in cancer patients, considering the half-life of ponatinib and mirroring the cardiotoxicity timeframe in humans. Future studies will require studying the association between the cardioprotective effects of ISRIB and the activity of GCN2 and ATP synthase. It will also be important to compare the observations from our study, harnessing the use of a specific inhibitor of GCN2 or activator of ATP synthase concurrent with the administration of ponatinib to study the effects on cardiac cell versus cancer cell fate.

Novelty and Significance.

What is Known?

• Ponatinib is a tyrosine kinase inhibitor (TKI) used to treat chronic myeloid leukemia (CML) and is among the most cardiotoxic TKIs. However, the mechanisms underlying its ability to induce cardiotoxicity remain elusive.

• Ponatinib-induced cardiotoxicity is associated with mitochondrial dysfunction.

• Mitochondrial dysfunction is a key factor in cardiac contractile failure. However, whether the interplay between mitochondrial energetics and signaling regulation underlies ponatinib-induced cardiotoxicity has yet to be determined.

What New Information Does This Article Contribute?

• Ponatinib-induced cardiotoxicity is associated with the Integrated Stress Response (ISR) activation signature.

• General control nonderepressible 2 (GCN2) activates the ISR upon sensing reduced ATP as a consequence of mitochondrial dysfunction in hiPSC-CMs.

• Reduced ATP levels also facilitate ponatinib binding to and activation of GCN2, contrary to its role in inhibiting kinases.

• Targeting ISR activation with the small-molecule inhibitor ISRIB can prevent ponatinib-induced cardiotoxicity both in vitro and in vivo.

Novel molecular targeted therapies have led to a dramatic improvement in the prognosis of cancer patients, however, toxicities affecting the cardiovascular system resulting from these treatments have also become evident. Although a targeted approach often exploits the differences between cancer cells and noncancer cells, overlap in fundamental signaling pathways necessary for the maintenance of function and survival in multiple cell types has resulted in systemic toxicities. Ponatinib, a third-generation TKI, is the only choice for treating CML patients with the challenging BCR-ABL T315I gatekeeper mutation resistant to other TKIs. Unfortunately, ponatinib is one of the most common cardiotoxic drugs among TKIs. Therefore, a better understanding of the molecular basis of ponatinib-induced cardiotoxicity and the development of strategies to mitigate its adverse cardiac effects are crucial. This study, using induced pluripotent stem cell-derived cardiomyocytes and mouse animal models, provides evidence that ponatinib-induced mitochondrial dysfunction activates the integrated stress response (ISR) pathway. Neutralizing excessive ISR activation can prevent or reverse ponatinib-induced cardiotoxicity. Mechanistically, mitochondrial dysfunction leads to ATP deficiency, triggering GCN2-mediated ISR activation. The reduced abundance of ATP also facilitates the direct binding of ponatinib to GCN2, unexpectedly leading to its activation, likely due to conformational changes in its structure. Our findings reveal the profound significance of the association between compromised ATP generation and enhanced GCN2-mediated ISR activation in ponatinib-induced cardiotoxicity and may also apply to other cardiovascular diseases in which ATP deficits are present.

Nonstandard Abbreviations and Acronyms:

CML

chronic myeloid leukemia

TKI

tyrosine kinase inhibitor

ISR

integrated stress response

eIF2α

eukaryotic translation initiation factor

GCN2

general controlled non-repressed

PKR

protein kinase R

HRI

heme-regulated inhibitor

PERK

PKR-like endoplasmic reticulum kinase

ATF4

transcription factor 4

hiPSC-CMs

treated human induced pluripotent stem cell-derived cardiomyocytes

RPPA

reverse-phase protein array

ASNS

asparagine synthetase

ASCT2

alanine, serine, cysteine-preferring transporter 2

NMN

nicotinamide nucleotide