Uncompensated mitochondrial oxidative stress underlies heart failure in an iPSC-derived model of congenital heart disease

Summary

Hypoplastic left heart syndrome (HLHS) is a severe congenital heart disease with 30% mortality from heart failure (HF) in the first year of life, but the cause of early-HF remains unknown. HLHS patient induced pluripotent stem cell-derived cardiomyocytes (iPSC-CM) showed early-HF is associated with increased apoptosis, mitochondrial respiration defects, and redox stress from abnormal mitochondrial permeability transition pore (mPTP) opening and failed antioxidant response. In contrast, iPSC-CM from patients without early-HF showed normal respiration with elevated antioxidant response. Single cell transcriptomics confirmed early HF is associated with mitochondrial dysfunction accompanied by endoplasmic reticulum (ER) stress. These findings indicate uncompensated oxidative stress underlies early-HF in HLHS. Importantly, mitochondrial respiration defects, oxidative stress and apoptosis were rescued by treatment with sildenafil to inhibit mPTP opening or TUDCA to suppress ER stress. Together these findings point to the potential use of patient iPSC-CM for modeling clinical heart failure and the development of therapeutics.

eTOC blurb

Xu et al. demonstrated feasibility of modeling early heart failure using patient induced pluripotent stem cell derived cardiomyocytes. They observed that heart failure is linked to increased apoptosis, mitochondrial dysfunction, redox and endoplasmic reticulum stress, all of which were rescued by Sildenafil or TUDCA, suggesting potential for therapy.

Graphical Abstract

Introduction

Congenital heart disease (CHD) is one of the most common birth defects affecting 0.5% of live births (Feinstein et al., 2012). Hypoplastic left heart syndrome (HLHS) is a severe CHD in which the left ventricle (LV) and aorta are small and nonfunctional. While survival with HLHS is made possible by surgical palliation that recruits the RV to become the single pumping chamber (Oster et al., 2013), there remains high morbidity and mortality. The 10-year transplant free survival stands at only 39–50%(Driscoll et al., 1992; Garcia et al., 2020; Gentles et al., 1997). However, the greatest risk is in the first year of life with 30% mortality reported (Alsoufi et al., 2016; Cleves et al., 2003; Tweddell et al., 2002). While HLHS patients have complicated clinical course, the early mortality is largely associated with ventricular dysfunction with rapid progression to acute heart failure (HF) (Garcia et al., 2020). Unfortunately, therapies developed for HF in adults have been ineffective for treating HF in HLHS(Hsu et al., 2010; Shaddy et al., 2007). Without insights into the underlying mechanisms driving HLHS early-HF, the clinical management of this patient population is largely empirical.

Investigations into the mechanism of HLHS-HF have been hampered by difficulty in obtaining human heart tissue for analysis. An alternative strategy entails in vitro disease modeling using induced pluripotent stem cells (iPSC) and their differentiated derivatives such as cardiomyocytes (iPSC-CM), endothelial/endocardial cells, and other cell types. While this has been successfully deployed for investigating HLHS disease mechanisms (Gaber et al., 2013; Hrstka et al., 2017; Jiang et al., 2014b; Kobayashi et al., 2014; Miao et al., 2020; Paige et al., 2020), no studies have explored the possibility of using iPSC-CM to model disease outcome. Particularly compelling is the question of why only some HLHS patients develop early-HF, and what is the cause for HF. The feasibility to model HLHS-HF in iPSC-CM is suggested by our previous studies of a mouse model of HLHS (Liu et al., 2017). We found cell autonomous defects were associated with prenatal/neonatal lethality from HF in the Ohia HLHS mutant mice. In the present study, we showed mouse iPSC-CM generated from the HLHS mice replicated defects observed in the HLHS mouse heart tissue, confirming the defects are cell autonomous and thus suitable for in vitro iPSC-CM modeling. Generating iPSC and iPSC-CM from HLHS patients dichotomized by clinical outcome, either with or without acute early-HF (Xinxiu Xu, 2018), we further investigated and studies provided surprising insights not only into possible causes for early-HF in HLHS, but they also uncovered mechanisms that may protect against early-HF in HLHS patients surviving heart transplant free.

RESULTS

Cell Autonomous Mitochondrial defects in the Ohia HLHS Mouse Model

The Ohia HLHS mice exhibit mid to late gestation lethality with acute heart failure characterized by severe pericardial effusion with poor cardiac contractility. This is associated with decreased cardiomyocyte proliferation and increased apoptosis (Liu et al., 2017). Ultrastructural analysis showed the myocardium with poorly organized thin myofilaments and altered mitochondrial morphology (Liu et al., 2017). Dynamic changes in mitochondria morphology play an important role in the developmentally regulated metabolic switch from glycolysis to oxidative phosphorylation, a process that also plays a critical role in cardiomyocyte differentiation (Hom et al., 2011). This entails closure of the mitochondrial permeability transition pore (mPTP) and formation of a mitochondrial transmembrane potential (ΔΨ_m) mediating oxidative phosphorylation. Using primary cardiomyocyte explants from the E14.5 Ohia HLHS mouse heart, we measured the ΔΨ_m, in cardiomyocytes from the right and left ventricle (RV, LV). A reduction was observed in both the RV and LV cardiomyocytes (Figure 1A), but mitochondrial mass was unchanged (Extended Data Figure S1A).

Figure 1. Mouse and human HLHS iPSC and iPSC-CM show differentiation and functional defects.

(A) Mitochondrial transmembrane potential (ΔΨm) measured with TMRE/Mitotracker Green in CM from LV (n=124 CM), RV (n=86 CM) of E13.5 wildtype (n=3) and LV (n=105 CM), RV (n=118 CM) of Ohia HLHS mutant (n=3) embryos.

(B) Mouse iPSC and iPSC-CM were generated from WT and Ohia mouse embryonic fibroblasts.

(C) Ki67 showed reduced proliferation in HLHS (n=1800) vs. WT (n=2800) iPSC-CM.

(D) Myh6/Myh7 transcript ratio decreased in Ohia HLHS iPSC-CM, indicating maturation defect. (WT n=6, HLHS n=5)

(E) Reduced beat frequency in Ohia iPSC-CM clusters (n=17) compared to WT (n=38).

(F) Mitochondrial membrane potential (ΔΨm) was reduced in Ohia HLHS iPSC-CM (n=43) vs. WT (n=56).

(G) Seahorse Analyzer was used to assess mitochondrial function (n=3 experiments).

(H) iPSC and iPSC-CM were generated from control and 10 HLHS patients dichotomized by cardiac outcome. Group I:transplant free survival >5 years old, Group II:death or heart transplanted <1 year old.

(I) Atrial (MYL7,SLN) and ventricular (MYL2,TNNT2) marker gene expression show iPSC-CM are ventricle-type.

(J) Increased MYH6/MYH7 ratio indicated Group II cardiomyocyte maturation defect.

(K) Ki67 immunostaining showed reduced Group II iPSC-CM proliferation.

(L, M) Myofibrillar disarray (L) observed in Group II iPSC-CM (M) stained for cTnT (Green) and α-actinin (Red). Control, n=163; Group I. n=42; Group II, n=192.

(N-P) Group II iPSC-CM defects seen with altered beat frequency (N) and calcium transient (O,P).

(Q-S) Reduced fractional shortening (Q), contraction rate (R) and relaxation rate (S) in Group II iPSC-CM.

Mean±SEM with Student’s t-test or ANOVA. Box plot analyzed with Kruskal-Wallis statistics. (H-S): (I) n=3,3,3 subjects. (J) n=3,5,3 subjects. (K, L,N) n=3,6, 3 subjects. (P) n=3,6,4 subjects. (Q-S) n=3(17),4(23),4(38) subjects (cardiomyocytes) analyzed.

To determine whether the abnormal open state of the mPTP is a cell autonomous defect, we generated iPSC from Ohia fibroblasts and differentiated them into iPSC-CM (Figure 1B). These Ohia iPSC-CM showed reduced cell proliferation with lower Myh6/Myh7 ratio indicating a cardiomyocyte differentiation defect (Figure 1C,D;Figure S1C-E), phenotypes reminiscent of those observed in the Ohia HLHS myocardium. Poor cardiac function was also indicated by reduced beat frequency (Figure 1E). Mitochondrial function was assessed with measurement of ΔΨ_m and oxygen consumption rate (OCR) using the Seahorse Flux Analyzer (Figure 1F,G). This uncovered mPTP and mitochondrial respiration defects in the Ohia iPSC and iPSC-CM. While the iPSC showed lower respiratory reserve and respiratory maxima (Figure S1B), the iPSC-CM exhibited reduction in basal respiration, proton production, respiratory reserve, and respiratory maxima (Figure 1G;Figure S1F,F’). Together these findings indicate the mitochondrial dysfunction, and proliferation and differentiation defects in the Ohia HLHS heart tissue are cell autonomous.

Generating HLHS Patient iPSC-CM

The finding that Ohia iPSC-CM replicated defects in the HLHS heart tissue suggested HLHS patient derived iPSC-CM may have utility for investigating acute early-HF in HLHS patients. For this study, we generated iPSC from 10 HLHS patients, including six >5-year-old with transplant free survival (Group I) (Figure 1H;Figure S2A), and four that died (n=3) or survived (n=1) with heart transplant at <1 year old (Group II). In addition, we also generated iPSC from 3 healthy subjects as controls. Using standard iPSC-CM differentiation protocol (Burridge et al., 2015), iPSC-CM at Day 18–22 of differentiation were used for the subsequent analysis.

Impaired Cardiomyocyte Differentiation and Contractile Dysfunction

The iPSC-CM generated were predominantly ventricle-like as shown by high expression of ventricular markers TNNT2 and MYL2, but low expression of atrial marker MYH7 (Churko et al., 2018) and SLN (sarcolipin)(Josowitz et al., 2014)(Figure 1I). The Group II iPSC-CM had fewer cardiac troponin T (cTnT/TNNT2) positive cells with higher MYH6/MYH7 ratio, indicating poor differentiation (Jiang et al., 2014a) (Figure S2F;Figure 1J). Group II iPSC-CM also showed reduced Ki67, but increased pH3 immunostaining, suggesting cell cycle disturbance with metaphase arrest (Figure 1K;Figure S2G,H), reminiscent of findings in the Ohia HLHS LV (Liu et al., 2017). Poor cardiomyocyte differentiation was indicated by low expression of cTnT (A-band) and α-actinin (Z-disc) containing myofilaments together with increased myofibrillar disarray (Figure 1L, M;Figure S2K). This was associated with increase in roundness (Figure S2I), but no change in cardiomyocyte cell size (Figure S2J).

Further assessment of cardiomyocyte contractile function showed the Group II iPSC-CM have lower beat frequency with reduced calcium transients (Figure 1N-P;Figure S2O-Q; Supplemental Video 1&2). The profile of calcium transients (Cyganek et al., 2018) confirmed the majority (84–89%) of the iPSC-CM are ventricle-like (Figure S2L-N). Analysis of the cardiomyocyte contractile motion by high resolution video microscopy showed reduced fractional shortening accompanied by decreased contraction and relaxation rates in the Group II but not Group I iPSC-CM. This was associated with reduction in the diastolic sarcomere length, but not systolic sarcomere length (Figure 1Q-S;Figure S2R,S;Supplemental video 3). While these findings suggest impaired calcium handling and poor contractile function, they may also reflect the iPSC-CM reduced beating rate.

Mitochondrial Dysfunction in HLHS Patient iPSC-CM

Various parameters of mitochondrial function were assessed in the iPSC-CM. A marked decrease in mitochondrial membrane potential (ΔΨ_m) was observed in the Group II iPSC-CM (Figure 2A). OCR measurements showed mitochondrial respiration defects with reduction in basal respiration, ATP production, respiratory reserve, and maximal respiratory capacity (Figure 2B). For Group I iPSC-CM, only respiratory reserve and maximal respiratory capacity showed significant change (Figure 2B). These same two parameters also were reduced in the undifferentiated iPSC of Group II patients, findings reminiscent of the Ohia mouse iPSC (Figure S2C,D & S1B). In contrast, Seahorse assessment of glycolysis with ECAR (extracellular acidification rate) measurements showed no significant change in either basal or oligomycin-treated ECAR between the Group II vs. control and Group I iPSC-CM (Figure S2T-V). Western blotting showed no change in abundance of the electron transport chain (ETC) complexes (Figure S3A,B).

Figure 2. Mitochondrial dynamics and respiration defects in HLHS iPSC-CM.

(A) ΔΨm measured using TMRE/MTG staining in SIRPa⁺ iPSC-CM.

(B) Seahorse Analyzer measurement showed respiration defects in HLHS iPSC-CM.

(C) MitoSOX staining of SIRPa⁺ iPSC-CM show elevated mitochondrial ROS in Group II.

(D,E) TUNEL and γ-H2AX staining show increased apoptosis (D) and DNA damage (E) in Group II.

(F,G) Mitotracker red showed hyperfused mitochondria with increased linkage constant in Group II CM.

(H) qPCR of genes mediating mitophagy and mitochondrial dynamics.

(I) LysoTracker Deep Red-staining showed reduced lysosome in Group II.

(J) Basal OCR (>3 duplicates) of heart tissue from two 19-year-old HLHS patients and 15-year-old cardiomyopathy patient with doxorubicin cardiotoxicity.

(K). Respiratory control ratio (RCR) for 2 control and 2 HLHS neonatal patient heart extracts. V_max measured with succinate as substrate and two ADP concentrations.

Mean±SEM with ANOVA. Number of subjects analyzed: (A,D) n=3,6,3. (B,C,F) n=3,5,3. (E) n=3,4,3, (H) n=3,3,3 (I) n=3, 4, 3.

Consistent with the uncoupling of oxidative phosphorylation in the Group II iPSC-CM, we observed a marked increase in mitochondrial reactive oxygen species (ROS) indicated by increased MitoSOX staining (Figure 2C) (Hom et al., 2011). Also observed was a reduction in nitric oxide (NO), suggesting perturbation of protein nitrosylation required for normal mitochondrial respiration (Figure S3F). The mitochondrial respiration defects and increase in ROS in the Group II iPSC-CM were accompanied by increased apoptosis and activation of a DNA damage response, findings similar to those observed in Ohia (Liu et al., 2017) and human HLHS fetal heart tissue (Gaber et al., 2013) (Figure 2D,E;Figure S3J). Parallel analysis of the Group I iPSC-CM showed no significant change in these parameters.

Perturbation of Mitochondrial Dynamics

As the regulation of mitochondrial fission and fusion play important roles in metabolic and redox regulation, its disturbance can contribute to cardiomyocyte death in HF (Marin-Garcia and Akhmedov, 2016). Using confocal imaging, we assessed mitochondrial mass and morphology. In the Group II but not Group I iPSC-CM, mitochondrial linkage constant and cluster length were increased, while mitochondrial mass was unchanged, indicating increase in mitochondrial fusion and/or decrease in mitochondrial fission in the Group II iPSC-CM (Figure 2F,G;Figure S3C-E). Group II iPSC-CM also showed decreased expression of DRP1 (DNML), regulating mitochondrial fission, and increase in MFN promoting mitochondrial fusion (Figure 2H). Expression of BNIP3/NIX regulating mitophagy were decreased in Group II but increased in Group I iPSC-CM (Figure 2H). Lysosomes, which are involved in mitophagy, were significantly reduced in only Group II (Figure 2I). These findings indicate the Group II hyperfused mitochondria likely arise from altered mitochondrial dynamics from increased mitochondrial fusion and decreased mitophagy. In contrast, Group I iPSC-CM exhibited more normal mitochondrial dynamics accompanied by increase mitophagy.

Mitochondrial Respiration Defects in the Left Ventricle of HLHS Human Heart Tissue

To explore the clinical relevance of the mitochondrial respiration defects in the HLHS patient derived iPSC-CM, heart tissue was obtained from HLHS patients undergoing heart transplant. Analysis of mitochondrial respiration showed basal respiration was reduced in the HLHS-LV vs. RV tissue, but such LV-RV difference was not observed in heart tissue from age-matched heart transplant patient with doxorubicin cardiotoxity induced HF (Figure 2J). In additional analysis of two HLHS and two control neonatal subjects, we observed the HLHS-LV was more sensitive to lower ADP and did not respond to higher ADP, indicating possible adaptation to bioenergetic stress and reduced respiratory capacity (Ventura-Clapier et al., 2011) (Figure 2K;Figure S3G). Western blotting showed no change in ETC components (Figure S3H,I). These findings suggest LV specific mitochondrial respiration defects in HLHS hearts.

Defects in Yap-Regulated Antioxidant Response

Activation of an antioxidant defense pathway occurs during cardiac development with metabolic transition to mitochondrial respiration (Perrelli et al., 2011; Tsutsui et al., 2011). This pathway is regulated by transcription factor NRF2 (Itoh et al., 1999) together with PITX2 and YAP (Tao et al., 2016). Together they regulate the expression of antioxidant genes that scavenge ROS to prevent oxidative stress. These transcription factors also have critical roles in cardiac regeneration and repair, with YAP also regulating heart organ size (Heallen et al., 2013; von Gise et al., 2012; Zhou et al., 2015). YAP is also known to regulate mitochondrial fission (Huang et al., 2018).

We observed NRF2 and PITX2 transcripts are both reduced in Group II iPSC-CM, but in Group I, PITX2 was elevated but NRF2 was unchanged (Figure 3A;Figure S4F,G). In contrast, YAP1 transcripts showed no change in either Group I or Group II iPSC-CM (data not shown). Analysis of downstream antioxidant pathway genes revealed up regulation of thioredoxin (TXN), peroxiredoxin 1 (PRDX1), glutathione peroxidase 1 (GPX1), and superoxide dismutase 2 (SOD2) in the Group I iPSC-CM, but in Group II, expression was either unchanged or downregulated, such as for PRDX1 (Figure 3A; Figure S4A-C, F). In the Group I iPSC-CM, we also observed increased expression of HIF1α, transcription factor regulating cell stress response to hypoxia. This was associated with increased expression of VEGF, a known downstream transcriptional target of HIF1α (Figure 3A) (Guimaraes-Camboa et al., 2015).

Figure 3. HLHS iPSC-CM with failed antioxidant response show NRF2, YAP1, and PITX2 cytoplasmic localization.

(A) qPCR showed antioxidant and HIF pathway genes are upregulated in Group I, and either unchanged or downregulated in Group II iPSC-CM. Inset: mitochondrial antioxidant pathway.

(B-E) Immunostaining shows NRF2, PITX2, YAP, β-catenin nuclear localization defects in Group II (see FigS4 H-K). N/C = nuclear/cytoplasmic ratio.

(F) qPCR quantification of YAP and YAP/β-actinin downstream target genes.

(G) Summary diagram of mitochondrial defects, redox stress, and failed antioxidant response in Group II iPSC-CM.

Mean±SEM, one-way ANOVA, or box plot with Kruskal-Wallis statistics (A-F). Number of control, Group I, Group II subjects or CM analyzed: (A,F) n=3, 3, 3 subjects. (B, C) n=130, 117, 64 CM (D) n=90, 78, 79 CM, (E) n=44, 15, 17 CM

Antibody staining showed nuclear localization of NRF2/PITX2/YAP were reduced in the Group II iPSC-CM, while in Group I, only PITX2 showed a modest reduction (Figure 3B-D; Figure S4H,J,K). However, total YAP and β-catenin protein expression were unchanged (Figure S4D,E). Downstream YAP target genes, NRG1 and MYC (Artap et al., 2018), were reduced in the Group II iPSC-CM. In contrast, the opposite was observed in Group I with NRG1 upregulated, while MYC was unchanged (Figure 3F). However, nuclear localized β-catenin and transcripts for two downstream YAP/β-catenin targets, BIRC5 and SNAI2, were reduced in both Groups I and II iPSC-CM (Figure 3E,F;Figure S4I). Together these findings show Group II iPSC-CM failed to mount an effective antioxidant response (Figure 3G), while in Group I, the antioxidant capacity is expanded, and may promote redox homeostasis.

Inhibition of mPTP Opening Rescues Mitochondrial Respiration and Apoptosis

The low ΔΨ_m observed in the mouse heart cardiomyocytes and Group II iPS-CM might reflect an mPTP closure defect during differentiation (Hom et al., 2011). Therefore, we use the Seahorse Analyzer to investigate whether mitochondrial defects in the HLHS iPSC-CM can be rescued by compounds promoting mPTP closure (Martel et al., 2012). We focused our analysis on sildenafil, as it is known to promote mPTP closure and is also commonly used among CHD patients for its vasodilatory effects (Galie et al., 2005). Sildenafil showed rescue down to 0.1 μM, which restored not only ΔΨ_m but also maximal mitochondrial respiration. This reduced mitochondrial ROS to levels similar to the Group I and control iPSC-CM (Figure 4A-C;Figure S5A-C). Sildenafil is also known to affect NO production, but normal NO level was restored only at 1.0 μM concentration (Figure 4D)(Prabhu et al., 2013). Cell proliferation, apoptosis (Figure 4E,F), and YAP nuclear trafficking were rescued at ten times lower dose of 0.01 mM (Figure 4G). However, β-catenin nuclear trafficking was not rescued (Figure 4H;Figure S5D).

Figure 4. Inhibition of mitochondrial membrane permeability rescues mitochondrial respiration and apoptosis.

(A-H) Sildenafil rescued Group II iPSC-CM including ΔΨm (A), maximum OCR (B), mitochondrial ROS (C), NO level (D), cell proliferation (E), apoptosis (F), and YAP (G) and β-catenin (H) nuclear localization.

(I-L) Bongkrekic acid (BKA), but not carboxyatractyloside (CAT) rescued ΔΨm (I), maximum OCR (J), mitochondrial ROS (K) and YAP nuclear localization (L).

Bar graphs, mean±SEM, one-way ANOVA. Box plot analyzed by Kruskal-Wallis. n≥3 independent repeats. Subjects analyzed: Control n=3, Group I n=4 or 5, and Group II, n=3 or 4.

To determine if mPTP closure defects underlie the ΔΨ_m reduction in the Group II iPSC-CM, we tested the effects of cyclosporin A (CsA), another compound known to target mPTP closure. CsA treatment rescued ΔΨ_m in the Group II iPSC-CM, supporting ΔΨ_m reduction as likely arising from a mPTP closure defect (Figure S5F). This was supported by additional analyses using BKA (bongkrekic acid), and CAT (carboxyatractyloside). BKA is an adenine nucleotide translocator (ANT) antagonist that promotes closure of the mPTP, while CAT is an ANT antagonist that promotes opening of the mPTP (Hom et al., 2011). As expected, BKA but not CAT restored the ΔΨ_m, together with the restoration of mitochondrial maximal and basal OCR. This was accompanied by the reduction of mitochondrial ROS to levels similar to control and Group I iPSC-CM, and rescue of YAP nuclear localization (Figure 4I-L,Figure S5E). Similar treatment of control iPSC-CM with CAT caused repression of respiration, while BKA had no effect (Figure S5G).

Single Cell Transcriptome Profiling

We performed single cell RNAseq (scRNAseq) on iPSC-CM from two Group II patients, patient 7042 with heart transplant at 11 months and patient 7052 deceased at 2 months, Group I patient 7464 surviving transplant free at 7 years of age, and healthy control subject 1053. Data was obtained from 4403 cardiomyocytes forming 9 clusters (Figures 5A and S6A-C). Marker gene (Biendarra-Tiegs et al., 2019) analysis showed these cardiomyocytes were largely of ventricular identity (Figure S6D). Clusters 0 (CM I), 1 (CM II) and 5 (CM III) comprising the majority of cells are well differentiated cardiomyocytes of increasing maturation (Figure S6E,F; Supplemental Spreadsheet 2). Group II vs. control comparison yielded the greatest number of differentially expressed genes (DEGs) (Figure 5B). Enrichment was observed for mitochondrial related pathways in all three clusters, suggesting Group II mitochondrial defects likely arise early in cardiomyocyte differentiation (Figure 5C). In contrast, Group I vs control yielded the fewest DEGs. These were associated with heart development and muscle organ development terms in Clusters 0 and 1, and mitochondrial related terms in Cluster 5 (Figure 5E). Group II vs. Group I comparisons yielded apoptosis and cell death in Clusters 0 and 1, (Figure 5D), and tRNA modification and noncoding RNA in Cluster 5.

Figure 5. Single cell RNAseq showed mitochondrial pathways enriched with early-HF.

(A). Single cell RNAseq yielded 9 clusters (Figure S6E). Clusters 0,1, 5 are well differentiated CM of increasing maturation. Clusters 4, 6 are proliferating CM at G2/M and S phase. Cluster 2 include CM undergoing apoptosis, while Cluster 7 exhibit ER stress/UPR. Clusters 3, 8 show enrichment for oxidative phosphorylation and hypertrophic cardiomyopathy.

(D) DEGs recovered in each cluster with pairwise comparisons.

(C-E) Pathway recovered from DEGs in pairwise group comparisons in Clusters 0,1,5.

(F) DEGs from all pairwise comparisons for Clusters 0,1, 5 and top Biological Processes recovered.

(G) Percentage mitochondrial related DEGs (n represents number of mitochondrial DEG observed) for each group in F.

(H. Mitochondrial related DEGs in clusters 0, 1, 5 and top Biological Processes recovered.

(F, H) Color scale showed relative maximum and minimum value. Grayscale showed adjusted p value (−log₁₀FDR) of each GO term and circle size showed gene count in each GO term.

(I) Hierarchical clustering based on DEGs from Group II (7052 and 7042) vs. Group I (7464). Asterisk denotes region with 28 DEGs upregulated only in 7052 (Spreadsheet 2).

(J) Violin plot show transcripts for 6 of 28 DEGs from panel I.

Combining DEGs from all pairwise comparisons showed the number of DEGs increased with disease severity (Figure 5F). Only control 1053 yielded terms related to muscle and muscle contraction. Group I 7464 recovered protein translation and cell cycle, and Group II 7042 yielded mitochondria and mitochondrial translation. In Group II 7052, hypoxia pathways were observed, but mitochondrial terms were also recovered in DEGs shared with 7042 (Figure 5F; Supplemental Spreadsheet 2). Overall, high percentage of DEGs were mitochondrial related (Figure 5G). Heatmap comprising only mitochondrial related DEGs was nearly identical to that for all DEGs (Figure 5H vs. F), indicating genes with the highest fold change are mostly mitochondrial related. Interestingly pathways related to mitochondrial translation, elongation, and termination were recovered in both 7464 (Group I) and 7042 (Group II), but with only 60 DEGs in 7464, vs. 182 DEGs in 7042 (Figure 5H;Supplemental Spreadsheet 2). Differing from 7042, mitochondrial DEGs in 7052 were hypoxia related, confirming recovery of these same pathways in all DEG analysis (Figure 5F). Mitochondrial DEGs shared between 7042/7052 included ATP synthesis and oxidative phosphorylation, suggesting bioenergetic deficits in Group II patients.

DEG analysis based on Group II vs. Group I comparison yielded further evidence of the effective dichotomization of HLHS patients into two functional groups (Figure 5I). Profiling upregulated DEGs showed the two Group II patients are similar to each other, while Group I is similar to control (Figure 5I). This analysis also recovered 28 genes highly expressed only in patient 7052 (see region denoted by asterisk in Figure 5I) - 14 are related to mitochondria, hypoxia and/or cell death, including EGLN3 encoding prolyl hydroxylase, an oxygen sensor promoting HIF1α degradation (Figure 5J;Supplemental Spreadsheet 2). Assembly of a protein interactome incorporating 26 of these genes showed enrichment for hypoxia, apoptosis, and oxidative stress, indicating a functional network contributing to early-HF in patient 7052 (Figure S7A,B;Supplemental Spreadsheet 2).

Molecular Chaperone Rescues Mitochondrial Respiration and Apoptosis

Recovery of ER stress and unfolded protein response (UPR) in Cluster 7 was notable (Figure 6A), given ER stress has been associated with HF (Schiattarella et al., 2019) and can be triggered by mitochondrial dysfunction and oxidative stress. This pathway has not been investigated previously in the context of HLHS. Real time PCR analysis confirmed elevated expression of genes (XBP1, ATF4, ATF6) associated with the three conserved ER stress pathways (Figure 6B). All three pathways were elevated in patient 7052, and two (ATF4, ATF6) were elevated in 7042. In contrast, Group I patient 7464 showed no change relative to control (Figure 6B). Similar analysis of three downstream ER stress target genes HSPA5, DDIT3, and DNAJC3 showed all were upregulated in 7052, but only DDIT3 and DNJC3 were elevated in 7042. In contrast, all three genes were down regulated in Group I patient 7464 (Figure 6B). These same cell stress related genes were also up regulated in the Ohia HLHS heart, consistent with their severe HF phenotype (Liu et al., 2017). To assess the potential functional impact of UPR on the HLHS iPSC-CM, we treated the iPSC-CM with Tauroursodeoxycholic acid (TUDCA), a molecular chaperone known to promote protein folding and suppress ER stress. TUDCA promoted mPTP closure in the Group II iPSC-CM, restored ΔΨ_m, reduced mitochondrial ROS, and rescued NO production (Figure 6C-E). TUDCA also rescued YAP nuclear translocation, restored cardiomyocyte proliferation and blocked apoptosis (Figure 6F-H). Furthermore, closing the mPTP with sildenafil and BKA normalized expression of ER stress pathway genes (XBP1,ATF4,ATF6) and ER stress targets (HSPA5,DDIT3,DNAJC3) in Group II iPSC-CM (Figure 6I). Together these findings suggest ER stress and UPR may contribute to early-HF in Group II patients.

Figure 6. ER stress in the iPSC-CM and its suppression rescued mPTP closure and apoptosis.

(A) ER stress is top pathway in Cluster 7.

(B) Real time PCR confirmed ER stress genes elevated in Group II CM.

(C-H) TUDCA rescued ΔΨm (C), ROS (D), NO level (E), YAP nuclear localization (F), restored cell proliferation (G) and suppressed apoptosis (H)

(I) Sildenafil and BKA rescued ER stress.

Bar graphs, mean±SEM with Student’s t-test. Box plots with Mann-Whitney statistics. (B, I) n=3 repeats. (C-H) n≥3 repeats. Subjects analyzed: Control n=3, Group I n=4 or 5, and Group II, n=2 (7042,7052).

Enrichment of variants associated with mitochondrial metabolism

Given HLHS is well described as having a genetic etiology, we further investigated the whole exome sequencing data available for 6 of our 10 HLHS patients (Group II 7042, Group I 7131,7400,7438,7434,7464). High impact variants comprising unique loss-of-function variants were recovered (Figure S7C;Supplemental Spreadsheet 3). Using Webgestalt/KEGG pathway enrichment analysis, four genes were identified as significantly associated with metabolic pathways (OXA1L,NNMT,NEU3,ALDH7A1) (Supplemental Spreadsheet 3). An protein-protein interactome (PPI) was constructed using these four genes to explore connections to Hippo signaling, a pathway that plays a critical role in regulating YAP degradation and nuclear translocation (Meng et al., 2016). The interactome showed enrichment for Hippo and Wnt signaling, heart development and many mitochondrial-related terms, including regulation of mitochondrial membrane permeability (Figure 7A,B;Supplemental Spreadsheet 3).

Figure 7. Damaging variants and pathway enrichment analysis.

(A,B) Protein-protein interactome constructed with four genes (black nodes) recovered Hippo/Wnt signaling and mitochondrial and cardiac related genes

(C) Metascape analysis of pathogenic variants from 41 HLHS patients. Size of the collective node represents percentage of genes originating from the corresponding gene list. Group I*:22 HLHS patients surviving without heart transplant. Group II*:19 HLHS patients who died or had heart transplant.

(D) Proportion of mitochondrial related genes (Mito-genes) among all genes recovered from Group II* vs. Group I* HLHS patients. Chi-square test.

(E) Pathway enrichment of the 19 mitochondrial-related genes recovered from Group II* HLHS patients that intersected with the MitoCarta inventory of mitochondrial genes (see Supplementary Spreadsheet 3).

(F) Fraction of Group II* patients with pathogenic variants in mitochondrial-related genes stratified by age of death or heart transplant (HTx/Death).

(G) Among Group II* patients, those who died or had heart transplant at earlier age (≤1) showed enrichment of pathogenic variants in mitochondrial-related genes. P value obtained with one-tailed Wilcoxon rank-sum test.

(H) Mitochondrial genes from human heart tissue at 5–15 weeks gestation from scRNAseq data (Cui et al., 2019). Circle size is percentage of cells expressing the gene (Exp%) and color shows average expression with Z-transform.

Building on this finding, we interrogated the WES data from another 41 HLHS patients comprising 19 patients who died or had heart transplant (Group II*) and 22 HLHS patients surviving transplant-free beyond 5 years of age (Group I*). Interrogating for pathogenic variants comprising unique loss-of-function variants or predicted damaging missense or splice variants yielded 160 genes from the unfavorable outcome group and 195 genes from the favorable group (Figure 7C;Supplemental Spreadsheet 3). Rendering these genes in a network plot using Metascape recovered terms such as “Ion channel transport”, “Mitochondrial gene expression”, and “Mitochondrial translation” in association with genes from the unfavorable group, while “lipid location, response to IL-17” were associated with the favorable group (Figure 7C;Supplemental Spreadsheet 3). Some pathways were shared by both groups such as calcium signaling, MAPK signaling, and nervous system development.

Examining the genes recovered for overlap with an expanded MitoCarta-related inventory of mitochondrial genes yielded 19 genes from the unfavorable and 11 from favorable group (Calvo et al., 2012; Pagliarini et al., 2008) (Supplemental Spreadsheet 3). We observed significantly higher percentage of mitochondrial genes with pathogenic variants among the HLHS patients who died or received heart transplant (Group II*) as compared to those surviving heart transplant free (Group I*). (p=0.035) (Figure 7D). ToppGene analysis of the overlapping genes recovered from the Group II* patients yielded multiple mitochondrial and metabolic pathways, but no pathway enrichment was observed for the overlapping genes recovered from the Group I* patients (Figure 7E;Supplemental Spreadsheet 3).

To further assess whether pathogenic variants in the mitochondrial genes may impact outcome, we investigated the relative proportion of mitochondria-related pathogenic variants in HLHS patients stratified by age of death or heart transplantation. This analysis showed HLHS patients who died or received heart transplant at ≤1 year of age had more pathogenic mitochondrial-related variants than those who died or had heart transplant at >1 year of age (p=0.03) (Figure 7F,G). We noted nearly all the overlapping mitochondrial related genes recovered are highly expressed in cardiomyocytes of the human fetal heart (Cui et al., 2019), supporting a role in HLHS pathogenesis (Figure 7H). These findings indicate mitochondrial variants as having a significant genetic modifying role on death and heart transplant outcomes in HLHS.

DISCUSSION

Our objective was to investigate the molecular etiology of early-HF seen in some HLHS patients. Analysis of iPSC-CM from our HLHS mouse model and HLHS patients revealed cell autonomous defects involving failure in mPTP closure. Thus, mitochondrial defects seen in the HLHS mouse heart were replicated in the mouse iPSC-CM. This was associated with defects in mitochondrial respiration and poor cardiomyocyte differentiation. These mitochondrial defects in the mouse heart and iPSC-CM were replicated in iPSC-CM of HLHS patients with early HF, suggesting a common cell autonomous mechanism involving mitochondrial defects underlying early-HF in HLHS.

For these studies, we selected HLHS patients with extreme phenotypes comprising death or surviving with heart transplant at less than one year of age (Group II), as the first year of life poses the greatest risk with 30% mortality (Oster et al., 2013). For comparison, HLHS patients surviving transplant free >5 years of age were also recruited (Group I). Using these two HLHS group comparisons and control subjects, we interrogated a myriad of parameters such as cardiomyocyte differentiation, myocyte contractility and calcium handling, mPTP closure, mitochondrial dynamics, respiration, and regulation of the antioxidant pathway. From this comprehensive analysis, we showed the iPSC-CM from Group II patients closely resembled each other, while the Group I patients were more similar to control. This was corroborated with scRNAseq analysis, which showed transcriptome profiles of two Group II patients are similar to each other, but very different from the Group I patient.

In the Group II iPSC-CM, we uncovered severe oxidative stress arising from mitochondrial dysfunction. This was associated with a reduction in ΔΨ_m that was rescued by sildenafil, BKA, or CsA, all compounds that promote mPTP closure. This suggests the ΔΨ_m reduction may reflect a mPTP closure defect. Although it is possible that mitochondrial respiration defects might also contribute to the low ΔΨ_m, such as from changes in the expression of the three uncoupling proteins UCP1,2,3, this was not observed in the scRNAseq data. We also noted that the mitochondrial respiration defects in the Group II iPSC-CM were accompanied by altered mitochondrial dynamics and reduced mitophagy. When combined with a failed antioxidant response, this would exacerbate the redox stress to enhance apoptosis and increase DNA damage. Also observed were severe defects in cardiomyocyte differentiation with poor myocyte function. Cardiomyocyte differentiation and maturation are known to be regulated by mPTP closure (Hom et al., 2011) and a metabolic switch to mitochondrial respiration (Mills et al., 2017; Nakano et al., 2017). Skeletal myoblast differentiation is regulated by a similar metabolic transition (Fortini et al., 2016a), and this transition was also shown to be modulated by mitophagy (Fortini et al., 2016b).

ScRNAseq showed the up regulation of ER stress together with apoptosis in a small cluster of Group II iPSC-CM. In contrast, the three major cardiomyocyte cell clusters (I, II, III) showed only the disturbance of mitochondrial related processes (Figure 5A), but ER stress was not observed. This suggests the ER stress likely arises secondary to the mitochondria associated increase in ROS, exacerbating the oxidative stress to induce apoptosis in the context of a poor antioxidant response (see Graphical Abstract). Recent studies have in fact shown an important role for ER stress and UPR in HF (Schiattarella et al., 2019). Of significant interest from a therapeutic standpoint, apoptosis in the Group II iPSC-CM can be rescued using sildenafil (Ascah et al., 2011) to inhibit mPTP opening or TUDCA to suppress UPR. This was associated with the reduction of mitochondrial ROS, recovery of mitochondrial respiration, and restoration of YAP nuclear translocation. Together these findings support mitochondrial mediated oxidative stress as underlying the acute early-HF in HLHS. The scRNAseq analysis further suggests this may involve defects in the HIF1α pathway, altered mitochondrial translation, and bioenergetic deficits, findings that will need to be further investigated in future studies.

In contrast to Group II iPSC-CM, the Group I iPSC-CM show similarities to that of control with near normal mitochondrial respiration and normal mitochondrial dynamics without oxidative stress nor increase in apoptosis. Nevertheless, the Group I iPSC-CM have reduced mitochondrial respiratory reserve and reduced maximal respiration, indicating an overall reduction in total respiratory capacity. Importantly, nuclear localization of NRF2, YAP1, PITX2 was maintained, albeit with some reduction observed for PITX2. This was associated with striking gene expression changes that included elevated expression of many antioxidant genes, and the elevated expression of HIF1α and its downstream target genes. Genes regulating mitophagy were also elevated, while MFN1, gene regulating mitochondrial fusion was down regulated. Significantly, key mediators of all three ER stress pathways were downregulated. Together these findings suggest the maintenance of mitochondrial dynamics in conjunction with the suppression of oxidative and ER stress by a vigorous NRF2/YAP/PITX2 mediated antioxidant response may provide protection from early-HF in Group I patients. As Group I iPSC-CM also showed better differentiation with improved myocyte contractile function, these factors also may contribute to improved clinical outcome.

The WES analysis showed pathogenic variants in HLHS patients with unfavorable outcome are enriched for genes in mitochondrial related pathways. While the genetic causes for HLHS remains largely unknown, pathogenic variants in mitochondrial related pathways may contribute to the pathogenesis of HLHS or they may act as genetic modifiers affecting clinical outcome. It is worth noting Sap130, one of the two genes causing HLHS in the Ohia mouse model is known to regulate genes involved in mitochondrial metabolism via the Sin3A complex (Pile et al., 2003), suggesting the developmental etiology of HLHS may involve the disturbance of mitochondrial metabolism (Liu et al., 2017). We note mounting evidence of the integral role for metabolism and mitochondrial respiration in the regulation of a wide range of developmental processes (Mills et al., 2017).

In summary, our findings point to the common involvement of mitochondrial dysfunction in HLHS regardless of HF outcomes. This is supported by another study that also reported mitochondrial defects in HLHS iPSC-CM (Paige et al., 2020). With the outcome-based iPSC modeling, we showed the mitochondrial dysfunction and oxidative stress underlie early-HF in HLHS, while a hyper-elevated antioxidant response may provide protection from oxidative and ER stress to prevent early HF. Together these findings suggest early-HF is the result of uncompensated mitochondrial mediated oxidative stress. The observed altered regulation of YAP1 suggests the tantalizing possibility that the mitochondrial defects also may contribute to the LV hypoplasia in HLHS, a question that warrants further studies.

We also showed possible therapeutic intervention with the targeting of mPTP closure with sildenafil or suppression of UPR with TUDCA. We note sildenafil is already being used empirically to threat HF associated with pulmonary hypertension(Guglin et al., 2016). Suppression of UPR, such as with TUDCA, may be another therapeutic path. TUDCA is currently in clinical trial for amyotrophic lateral sclerosis(Elia et al., 2016). Providing antioxidant might be another therapeutic course, although we found ascorbic acid did not rescue mitochondrial defects in the Group II iPSC-CM. Overall, our iPSC modeling has yielded mechanistic insights into the underlying causes for early-HF in HLHS and suggest possible evidence-based therapies that will need to be further investigated. These findings also point to the efficacy of patient iPSC-CM stratified by clinical outcome for modeling heart failure.

Limitations of the Study

One limitation of our study is the inclusion of iPSC-CM from only 10 patients. However, this compares favorably to other studies that typically include iPSC from only one to three patients, and no study had controlled for outcome (Gaber et al., 2013; Hrstka et al., 2017; Jiang et al., 2014b; Kobayashi et al., 2014; Miao et al., 2020; Paige et al., 2020). Nevertheless, the generalizability of our findings will require future confirmation with analysis of iPSC-CM from additional patients. Another potential limitation is specificity of the acute early-HF. This has been addressed by a previous clinical study examining the causes of death in single ventricle patients after their first stage Norwood procedure. The predominant cause of death was determined to be cardiovascular, with only 7% attributable to multisystem organ failure (Ohye et al., 2012). The relevance of our findings to HF in older HLHS patients will require further studies, as our focus was on acute early-HF in patients less than one year old. While additional factors may contribute to HF in older patients, the involvement of mitochondrial dysfunction is likely. This is suggested by the enrichment of mitochondrial-related pathogenic variants in the expanded WES analysis of 41 HLHS patients that included older patients with heart transplant.

STAR*METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Cecilia W. Lo (cel36@pitt.edu).

Materials availability

iPSC lines may be made available by the PI for studies related to CHD after specific approvals are obtained per IRB regulations.

Data and code availability

WES data of a part of the HLHS (n = 20) is publicly available at the dbGAP under phs001256. v1.p1. WES data of the remainder of subjects (n =21) will be available on request under the condition of approval of the ethical committee of University of Pittsburgh and material transfer agreement. Human iPSC-CM scRNAseq data is publicly available at GEO under accession number GSE146341. Publicly available single-cell RNA-seq data of human fetal hearts (Figure 7E) was downloaded from GEO database under accession number GSE106118.

MATLAB scripts used to process datasets of sarcomere contractility and calcium transient had been deposited in Dataverse (https://doi.org/10.7910/DVN/9BMEY7; https://doi.org/10.7910/DVN/9ZWTAL). scRNA-seq data analyses were performed in open-source R programming environment (v3.6.1) (https://www.rproject.org). Code for scRNAseq is freely available upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Mouse Strain

E13.5-E14.5 Ohia HLHS mouse (Sap130m/m;Pcdha9m/m) or CRISPR HLHS mouse (Sap130m/m;Pcdha9m/m) and littermate controls were used for primary cardiomyocytes explants from heart tissue. Mouse embryo fibroblasts used for mouse iPSC generation were generated from E14.5 – 17.5 mouse embryos (See Supplemental Spreadsheet 1). All mice were housed, treated, and handled in accordance with the guidelines set forth by the University of Pittsburgh Institutional Animal Care and Use Committee and the National Institutes of Health’s Guide for the Care and Use of Laboratory Animals.

Human Blood, Cells, and Surgical Tissue

Cells, heart tissue and blood were obtained from HLHS patients recruited from Children’s Hospital of Pittsburgh of UPMC with informed consent under a human study protocol approved by the University of Pittsburgh Institutional Review Board (Supplemental Spreadsheet 1). For infants and minors, informed consent was obtained from the legal guardian. Some human heart tissues were obtained from the Molecular Atlas of Lung Development Program (LungMAP) Consortium distributed by Human Tissue Core (HTC). Donor tissue was supplied through the United Network for Organ Sharing for Western blot and isolated mitochondrial OCR measurements.

METHOD DETAILS

Production of patient iPS cells

Mouse embryonic fibroblasts were reprogrammed using the CytoTune-iPS Sendai Reprogramming kit(Fusaki et al., 2009). Human fibroblasts or lymphoblastoid cells were transfected with four episomal plasmids(Okita et al., 2011) using electroporation. iPSCs clones were identified by immunofluorescent staining of pluripotency marker Oct4 and Nanog and qPCR analysis of stem cell markers (Supplemental Figure S2B)(Xu et al., 2013). All antibody and primer sequence information are provided in Supplemental Spreadsheet 1.

Independent iPSC clones were isolated from which one was selected from each subject for detailed experimental analysis. Overall, 13 iPSC lines, 6 from HLHS patients in Group I, 4 from HLHS patients in Group II, and also 3 control subjects were used for all experiments in this study. Given the 72% basal OCR difference observed between the Group II HLHS vs. control iPSC-CM, the 3 control and 4 Group II iPSC-CM provide 80% power with a two-sided test with significance level of 0.05. We note while independent iPSC clones from one subject are often used to demonstrate reproducibility of findings, a previous study showed using transcriptome profiling showed the importance of using iPSCs from different subjects rather than multiple sister iPSC clones to distinguish disease-associated mechanisms from genetic background effects in disease modeling (Schuster et al., 2015). A summary of the iPSC lines used in the various experiments in this study are shown in Supplemental Spreadsheet 1, 5. Subject analysis details.

Production of iPS derived cardiomyocytes

The iPSC cells were seeded on BD Matrigel pre-coated plates for 2–3 days under mTESR1 media then switched to CDM3 media consisting of RPMI 1640, BSA, B27, 213 μg/ml Vitamin C (Ascorbic acid) and 6 μM CHIR99021(Burridge et al., 2014). After 2 days the media was replaced with CDM3 media containing RPMI 1640, BSA, 10 μM XAV939, 213 μg/ml Vitamin C, and BSA. Medium was changed every two days. CDM3 media without XAV939 and Vitamin C was used after 14 days. Finally, beating cells are observed at ~8–10 days after initiating reprograming and further analyzed in the following days (Day18–22). A summary scheme of the cardiac differentiation protocol is shown in Figure S2E.

Immunofluorescence staining

Cells were fixed with 4% paraformaldehyde with 0.1% Triton X-10, followed by blocking in 5% goat serum, then staining overnight with primary antibody in 0.5% bovine serum albumin/phosphate-buffered saline (BSA/PBS). After washing in PBS, incubation with secondary antibody was performed in 0.5% BSA/PBS for 1 hour at room temperature and nuclei were stained with 2 μg/ml Hoechst 33342 (Life Technologies). Images were acquired using the Leica SP8 confocal or Leica DMI6000-SD microscopes. Antibody information in Supplemental

Myofibrillar and cell roundness analyses

Human iPSC-CMs were stained by DAPI, cTNT and a-actinin. Myofibrillar disarray was quantified with analysis of confocal images captured at 20X magnification, examining for myofilaments with well-defined sarcomeric structures. Cell circularity or roundness was calculated in the iPSC-CM using measurements of cell area and perimeter and the equation 4πXarea/perimeter2, with value of 1.0 indicating a perfect circle, and value<1 indicating eccentricity or elongated shape.

Analysis of mitochondrial calcium transients

iPSC-CMs cultured in chamber slides were loaded with 1 μM Rhod-2 (Molecular Probes, Life Technologies, Carlsbad, CA, USA) in Hank’s balanced salts modified buffer (HBSS, pH 7.4) for 15 minutes at 37°C and washed twice for 15 minutes in HBSS. The slides were placed on a temperature-regulated microscope stage and kept at 37°C. Fluorescence images were acquired using the Leica DMI6000-SD fluorescence microscope. Post-acquisition analyses of changes in intracellular calcium transients (CaTs) were performed by plotting mean signal intensity as a function of time using ImageJ time series analyzer package (NIH, Bethesda, MD; Version: 2.0.0-rc-69/1.52K), polynomial smoothing (2nd order, 6 neighbors) and GraphPad Prism 9. The final analysis of CaTs was conducted using MATLAB (Mathworks). Time to peak of CaTs was defined as the time from start to the maximum of the calcium transient. Peak to peak interval is defined as period length (s). Half decay is defined as the duration from peak to half amplitude. CaTD20 was defined as the time from the maximum of the transient until 20% signal decay. For each cell, the individual data set represents a mean of 6–12 CaTs recordings. Frequency distribution and Gaussian distribution methods (Prism 9) were used to fit the overall distribution of the entire hiPSC-CM population. The crossover value was defined as the threshold that divided CaTDs into atrial and ventricular like groups. The data shown in Figure S2 M represent the Rhod-2 intensity obtained from 724 iPSC-CM from 3 controls and 9 HLHS patients generated in 9 independent experiments.

Analysis of sarcomere contractility in iPSC-CM

Single iPSC-CM cell videos were collected by Leica DMI 3000B microscope and videos of human iPSC-CM (200 Hz) containing striated sarcomere were analyzed using a custom MATLAB code (available upon request) written to apply the fast Fourier transform (FFT) algorithm to each frame (approximately 1400 frames per video) to compute the spatial frequencies of the sarcomeres. The frequency (f₀) of the highest amplitude peak of the FFT within a user defined range was identified, and sarcomere length (L) in each frame was calculated by taking the reciprocal of f₀, $L=\frac{1}{f_0}$. The user defined range was determined and optimized to ensure that the maximum and minimum measured sarcomere lengths always occurred within this range. The sarcomere length in units of pixels was then converted to units of micrometers, 1μm = 4.58 pixels (100X magnification video), and plotted as a function of time (seconds). From the plots of sarcomere length versus time, the point of maximum sarcomere length immediately before contraction (t₁, max), minimum sarcomere length (Systolic length) during contraction (t₂, min), 701 and maximum sarcomere length (Diastolic length) immediately after contraction (t₃, max) were 702 identified for each contraction. Fractional shortening (FS) was calculated as $FS=\frac{max-min}{max}$, contraction rate (CR) was calculated as $CR=\frac{max-min}{t_2-t_1}$, and relaxation rate (RR) was calculated as $RR=\frac{max-min}{t_3-t_2}$.

RNA extraction, real-time PCR, and transcript splicing analysis

Total RNA was isolated using the miRNeasy micro-Kit (QIAGEN) with on-column DNase I digestion (QIAGEN). cDNA was prepared with high-capacity RNA to cDNA kit (Applied Biosystems). Real-time PCR was conducted using 7900HT Fast Real Time PCR System. All primer sequence information is provided in Table S1.

Seahorse Analyzer analysis of oxygen consumption rate

For cell oxygen consumption rate (OCR) quantification, 20,000 iPS-CMs or 20000 iPSCs were seeded into each well of a Seahorse XFe96 cell culture plate and cultured for 2 days for adherence to the culture plate. On day of measurement, the medium was changed to pre-warmed Seahorse assay medium, and OCR determined using the Seahorse XF Cell Mito Stress Kit (Agilent). Basal respiration was measured in unstimulated cells. Afterwards, oligomycin (1 μM) was added to quantify respiration coupled to ATP production and proton leak followed by carbonyl cyanide-4-(trifluoromethoxy)-phenylhydrazone (FCCP) injection (0.5 μM-(Burridge et al., 2016) to assess maximal cellular respiration (respiratory capacity). Finally, antimycin A (1 μM) and 719 rotenone (1 μM) were used to assess non-mitochondrial respiration. This is summarized in 720 Supplemental Figure S3F’. For mouse and human heart tissue OCR quantification, freshly dissected 2mm X 2mm heart tissue blocks were seeded into each well of a Seahorse XF24 islet capture microplate and OCR were measured using the same Seahorse XF Cell Mito Stress Kit to obtain the basal respiration rate in unstimulated cells during two cycles of measurement.

Analysis of inner mitochondrial transmembrane potential

Embryonic left and right ventricles were freshly dissected and dissociated with papain to generate primary cardiomyocytes for live imaging as previously described (Hom et al., 2011). Briefly, this entailed loading live explanted cardiomyocytes, iPSC, and iPSC-CM for 35 minutes with tetramethylrhodamine ethyl ester (TMRE, 20 nM, Invitrogen, Cat# T-669) and Mito Tracker Green (MTG, 200 nM, Invitrogen, Cat# M-7514) in Hepes-Tyrode's buffer, washed, and equilibrated for 20 minutes in the same buffer. The live cells were then imaged using epifluorescence microscopy. Mitochondrial membrane potential (Δψ_m) was quantified as the ratio of TMRE to MTG intensity (Galmiche et al., 2011).

Mitochondrial network analysis

MitoTracker Red (100 μM) was loaded into live cells following manufacturer’s recommendations, and then cells were fixed in 4% paraformaldehyde/PBS at 37°C for 15 minutes. Cells were permeabilized in 0.2% Triton X-100/PBS for 10 minutes and then immunostained for cTnT while DNA was labeled with Hoechst. Mitochondria were imaged using a Leica SP8 confocal with a 40x / 1.3NA objective. Acquisition settings and deconvolution were done with the guidance of SVI Huygens software, and images were post-processed in ImageJ (NIH, Bethesda, MD; Version: 2.0.0-rc-69/1.52K) with unsharp mask (radius 2; mask weight 0.7), background subtraction, and the tubeness filter (sigma = 0.25 microns) to highlight mitochondrial filaments. Mitochondria were segmented with Skeletonize 2D/3D. Mitochondrial networks were then analyzed (“Analyze Skeleton”) using BoneJ. Clusters with 20 branches or more were used for measuring average branch length and linkage statistics. To quantify the degree of mitochondrial consolidation, clusters were ranked from most to least branches (see graphs) and a mono-exponential decay curve is fit to the resulting data. The curve’s decay constant is then inverted so that higher values reflect more-linked mitochondrial networks.

Mitochondrial DNA copy number assays

DNA was extracted from human iPSC-CM. qPCR was performed and mitochondrial DNA copy number was determined by normalizing results from primers targeted to mtDNA-tRNA-Leu (Forward: 5'-CAC CCA AGA ACA GGG TTT GT-3' and Reverse: 5′-TGGCCATGG GTA TGT TGT TA -3′) against results from primers targeted to nuclear B2-microglobulin (Forward: 5'-TGCTGT CTC CAT GTT TGA TGT ATC T-3' and Reverse: TCT CTG CTC CCC ACC TCT AAG T-3'(Rooney et al., 2015).

Reactive oxygen species, nitric oxide and lysosome measurements

To quantify reactive oxygen species (ROS), nitric oxide level and lysosome abundance, iPSC-CMs were incubated at 37 °C for 30 minutes with 5 μM MitoSOX, 5 μM DAF-FM diacetate and 1 μM LysoTracker Red DND-99 (Life Technologies), respectively. For NO measurement, additional 15–30 minutes incubation could complete de-esterification of the intracellular diacetates. CD172a(SIRPα/β) was used as a human iPSC-CM marker (Dubois et al., 2011). Live cell fluorescent imaging was conducted using the Leica DMI6000-SD microscope.

Human heart tissue and iPSC-CM Western blotting

Cryopreserved LV and RV tissue or iPSC-CM were homogenized and processed for Western blotting using a ChemiDoc (Biorad) with Image J image processing (Beutner et al., 2017; Beutner et al., 2014). Antibodies from Abcam and BioRad were used and included: OXPHOS Rodent Cocktail (ab110413), AC (#154856), Starbright 700 (anti-mouse), Starbright 520 (anti-rabbit).

Isolation of mitochondria and oxygen consumption assay

Mitochondria were recovered from freshly dissected heart tissue (~140 mg) after homogenization followed by differential centrifugation and resuspension in EGTA/EDTA-free isolation (Beutner et al., 2017; Beutner et al., 2014). Oxygen consumption was measured at room temperature in respiration medium with a Clark oxygen electrode (Hansatech) using published protocols. Cytochrome c (50μM) and atractyloside (100μM) were used to test mitochondrial membrane integrity. Substrate-mediated respiration (state 2 or V0), maximal respiration (state 3 or Vmax), and RCR (Vmax over V0) were calculated.

Single-cell RNA sequencing

Previous study proved there is no significance difference between iPSC-CM from day 21 and day 30 (funakoshi et al., 2018), hence, the iPSC-CM differentiated at day 22 were choose as scRNAseq samples. The iPSC-CM from three patients and one control was prepared for single cell RNAseq. The iPSC-CMs were disaggregated using cold active protease [10 mg/ml Bacillus Licheniformis protease; Creative Enzymes NATE0633) and 125 U/ml DNase (Applichem, A3778) incubated on ice with trituration 5–7 minutes, then 5% bovine serum albumin (BSA) was added, and cells were filtered by 100 μm cell strainer and the cells pelleted, then re-suspended in 200ul PBS/BSA. Trypan blue exclusion was used to quantify cell viability, and the volume was adjusted to 200,000 cells/ml for 10X chromium single-cell RNA-seq. Pair-end library preparation was carried out using the V3 version (10X Genomics). Single-cell droplet libraries from ~10K cells from each suspension were generated using the 10X Genomics Chromium controller with the Chromium Single Cell 3’ GEM Library and Gel Bead Kit v.3 and the Chromium Chip B Single Cell kit (1 GEMs per sample, expected recovery ~6k cells per GEM). All samples were barcoded with the Chromium i7 Multiplex Kit. All libraries were pooled and sequenced across two lanes of a HiSeq4000, 150bp paired end reads with a target coverage of 20k fragments per cell. All samples were uniquely indexed, mixed, and evenly distributed into the Illumina HiSeq 4000 for sequencing.

Single-cell RNA-Sequencing Data Analysis

Single-cell sequencing data was processed using the Cell Ranger (version 3.1.0) count pipeline using the human reference genome GRCh38 and annotations from Ensembl (version 93). Quality control and filtering were performed using scater (McCarthy et al., 2017) (v1.18.6). For each sample, cells with library size less than 500, number of detected genes less than 300 or greater than 6,000, or mitochondrial percentage greater than 4 times the median absolute deviation (MAD) from the median value were excluded. Additionally, top 3% cells ranked by the doublet score (hybrid) calculated using the scds R package (Bais and Kostka, 2020) (v1.6.9) were excluded. Only non-ribosomal genes with at least 1 count in ≥5 cells were considered. We adapted the approach of Kannan et al. [https://doi.org/10.1101/2020.04.02.022632] for cell type classification using SingleCellNet(Tan and Cahan, 2019) (v0.1.0) and further limited to cells classified as "cardiac muscle cells" yielding a data for 13,954 genes across 8,094 cells. We performed downstream analyses using the Seurat package (Stuart et al., 2019). To focus on high-quality CMs, we further removed cells with total library size less than 1,400 or number of detected genes ≤ 800, or percentage of mitochondrial gene counts greater than 20%. This yielded a final set of 877, 1,718, 1,434, 374 cells for 1053, 7042, 7052 and 7464, respectively, for downstream analysis.

We normalized total count per cell to 10,000 and find top 2000 highly variable genes in each sample. Integration of cells from different samples and batch correction were performed using IntegrateData function in Standard procedure of Seurat 3. Scaled data after integration was used for principal component analysis (PCA) and top 30 dimensions were used for neighbor detection and Louvain clustering (resolution = 0.5). UMAP was drawn for the visualization of single-cell data in reduced dimensions.

Differentially expression analysis was conducted using student t-test in Scanpy(Wolf et al., 2018). We compared differentially expressed genes of clusters and sample groups, as well as samples and sample groups per cluster (Figure 5C-F,G, S5F). Genes with FDR < 0.05 in tests were selected as DEGs. ToppGene(Chen et al., 2009) was used for gene enrichment analysis and Gene Ontology (Biological Process) terms and coexpression of MSigDB were used for annotations of gene lists. The strength of associations was represented by -log10(FDRToppGene) (Figure 5C-F,G). Gene modules of cardiomyocyte clusters were generated using 200 most significantly upregulated genes (Figure S5E) and their top enriched Gene Ontology (Biological Process) terms were used for annotating cluster identities. Similarity between these clusters were evaluated using Pearson correlation of genes in gene modules. Cell cycle scores were calculated in Seurat using CellCycleScoring function and cell cycle phases were inferred accordingly (Figure S5C).

Whole exome sequencing analysis

Whole-exome capture was carried out on 6 Caucasian HLHS subjects with iPSC and 41 HLHS subjects (including the 6 HLHS subjects) at BGI Americas. Genomic DNA from venous blood was captured with Agilent V4 Exome Capture kit. Sequencing was performed on the Illumina HiSeq2000 platform with 100 paired-end reads, or the Illumina HiSeq4000 with 150 paired-end reads at 100× coverage. Sequence reads were mapped to the reference genome (hg19) with BWA-MEM(Arakawa et al., 2010) and further processed using the GATK(McKenna et al., 2010) Best Practices workflows, which include duplication marking, and base quality recalibration. Single nucleotide variants (SNVs) and small indels (InDels) were detected using GATK HaplotypeCaller and annotated by ANNOVAR (Wang et al., 2010). High quality variants were recovered that: 1) passed GATK Variant Score Quality Recalibration (VSQR); 2) have minimum 5 supported reads; 3) have genotype quality ≥ 20 or 60 for SNVs or InDels, respectively; 4) SNVs or InDels not within 10bp or 5bp of an indel, respectively.

Variants with minor allele frequency (MAF) less than 0.01 in GnomAD (Karczewski et al., 2020) or Kaviar database(Glusman et al., 2011) were retained for downstream analyses. Only loss-of-function (LoF) mutations (nonsense, canonical splice-site, frameshift indels, and start loss), likely damaging missense variants (D-Mis) and non-frameshift indels were considered potentially damaging. Missense variants were considered likely pathogenic if it was predicted to be damaging by at least three out of nine prediction scores available via dbNSFP v3.5a (Liu et al., 2016). All filter processes are shown in Figure S7. As a further filter, only pathogenic variants with allele count of at least two in Group I* and absent in Group II* would be retained as pathogenic variants in Group I*, and similarly, only pathogenic variants with allele count of at least two in Group II*, but absent in Group I* would be retained as pathogenic variants in Group II*.

Functional enrichment and interactomes analysis

Webgestalt KEGG pathway analysis (http://webgestalt.org/) was performed for unique LoF variants from 6 HLHS cohort. The interactomes of the four genes harboring unique variants in the Group II HLHS patient was assembled by including their protein-protein interactions (PPIs) collected from BioGRID (Stark et al., 2011) and HPRD (Prasad et al., 2009), and novel-PPIs predicted by High-precision PPI Prediction (HiPPIP) model (Ganapathiraju et al., 2016) focusing on short path connections to the Hippo pathway. Hippo pathway genes were extracted from KEGG (Kanehisa et al., 2008). Enrichment Analysis Tool available on Gene Ontology (GO) website which uses PANTHER (Mi et al., 2021) on the backend, was used to find biological process terms associated with the interactome genes with statistical significance. It computes fold enrichment of the genes in the input list over the expected value. Fold enrichment>1 and fold enrichment<1 showed that the annotation is overrepresented and underrepresented in the list respectively. It presents p-value determined by Fisher's exact test with FDR correction, and a cut-off of 0.05 was used to select significantly enriched annotations.

Enrichment of mitochondrial genes in HTx/Death group

Mitochondrial genes were obtained from MitoCarta database (Calvo et al., 2016) comprising merge of versions v1.0, v2.0, and v3.0. The mitochondrial gene list is shown in Supplemental sheet 3 (Tab 11). To assess whether pathogenic variants in mitochondrial-related genes were enriched in HTx/Death patient group, Chi-square test was conducted to estimate P value (Figure 7D). Next, the Group II* patients were further stratified by age of death or heart transplant - either at less than or equal to one year of age or more than one year of age. Statistical analysis was conducted using one-tailed Wilcoxon rank-sum test (Figure 7G).

QUANTIFICATION AND STATISTICAL ANALYSIS

Standard statistical analyses were performed using GraphPad Prism 9. D’Agostino & Pearson normality test and Shapiro-Wilk normality test were used to test if the data had a Gaussian distribution. For Gaussian distribution, data are presented as bar graphs and expressed as mean±SEM, either Unpaired t-test (Two- tailed) or One-way ANOVA (FDR B&Y correction were used for multiple comparisons) were applied. Data without Gaussian distribution showed by box plot (Line at median and minimum-maximum were represented by the top /bottom of box), either non-parametric Mann-Whitney test (Two-tailed) or Kruskal-Wallis tests (FDR B&Y correction were used for multiple comparisons) were used. The experiments were not randomized. The investigators were not blinded to allocation during experiments and outcome assessment.

Supplementary Material

2. Supplemental Spreadsheet 1: Patient information, iPSC production, antibodies, and primer sequences. Related to STAR Methods.

1. miPSCs generated and used in this study

2. HLHS patient medical history

3. Patient iPSCs reprogramming

4. Subject analysis details (mouse)

5. Subject analysis details (human)

6. Statistical details

3. Supplemental Spreadsheet 2: Single cell RNAseq related information. Related to Figure 5,6.

1. Marker gene list for C0–8 (Figure S5 E)

2.1–2.9. GO enrichment analysis for Cluster 0–8 (Figure S5 E)

2.10. HCM related genes in Clusters C3 and C8 (Figure 5 A)

3.1. DEG No. in Each Cluster (Figure 5 B)

3.2–3.10. GO enrichment analysis of DEG in C0/C1/C5 under different comparisons (Figure 5 C-E)

4.1 All DEGs in Figure 5F

4.2 ToppGene of Control 1053 DEGs (Figure 5F)

4.3 ToppGene of Patient 7464 DEG (Figure 5F)

4.4 ToppGene of Group II 7042/7052 Shared DEG (Figure 5F)

4.5 ToppGene of Patient 7042 DEG (Figure 5F)

4.6 ToppGene of Patient 7052 DEG (Figure 5F)

5.1. Mitochondrial DEGs (Figure 5H)

5.2 ToppGene of Control 1053 Mitochondrial DEGs (Figure 5H)

5.3. ToppGene analysis of Patient 7464 Mitochondrial – DEGs (Figure 5H)

5.4. ToppGene of Group II Mitochondrial-DEG (Figure 5H)

5.5. ToppGene of Patient 7042 Mitochondrial-DEG (Figure 5H)

5.6. ToppGene of Patient 7052 Mitochondrial-DEG (Figure 5H)

6. Patient 7052 – 28 upregulated DEGs (Figure 5G)

7.1. Protein-Protein Interactome network genes (Figure S6A)

7.2. PPI BiNGO Biological Process Pathway Enrichment (Figure 5I)

8.1–8.2. GO enrichment analysis of DEG of Group II VS Control in C7 (Figure 6A)

4. Supplemental Spreadsheet 3: Whole exome sequencing and interactome analysis related information. Related to Figure 7.

1. Description 2. Unique LoF (loss of function) genes from Group II patient (Figure 7A)

3. High impact variants from Group II patient (Figure 7A)

4. Webgestalt/KEGG pathway enrichment of unique LoF genes from Group II patient (OXA1L, NNMT, NEU3, ALDH7A1) (Figure 7A)

5. Protein-protein interactome GO Biological Processes

6. GO Biological Processes in Figure 7B

7. Genes in GO Biological Processes in Figure 7B.

8. Unique gene with variants in 41 HLHS cohort (Figure 7C)

9. Pathogenic variant list in 41 HLHS cohort (Figure 7C)

10. Metascape-GoEnriched (Figure 7C)

11. Mitochondrial gene list (Figure 7D)

12. Overlap of mitochondrial and total genes with pathogenic variants in 41 HLHS cohort (Figure 7D)

13. Pathogenic variants in mitochondrial-related genes (Figure 7D)

14. ToppGene analysis of those 19 overlapping genes (Figure 7E)

6

Sup-video-1_hips-cm_beating, Related to Figure 1: Videomicroscopy showing contraction of human iPSC-CM. The iPSC-CM from control subject and Group I beat faster than iPSC-CM from Group II. Scale bar = 250 μm.

7

Sup-video-2_hips-cm_Ca, Related to Figure 1: Calcium transients in the iPSC-CM are visualized using Rhod-2. Note faster propagation of calcium transients in iPSC-CM from control subject and Group I patients as compared to that of Group II. Scale bar = 250 μm.

8

Sup-video-3_hips-cm_single_cell, Related to Figure 1: Videomicroscopy recording of individual beating iPSC-CM from control subject, Group I and Group II patients. Robust contractions are seen in cardiomyocytes from control and Group I, but only weak contractions are seen in Group II. Scale bar = 10 μm.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-Active-β-Catenin (Anti-ABC) Antibody, clone8E7	Sigma-Aldrich, Millipore	Millipore Cat# 05-665, RRID:AB_309887
Anti-PITX2/RGS antibody	Abcam	Abcam Cat# ab98297, RRID:AB_10887751
Recombinant Anti-Nrf2 antibody [EP1808Y] - C-terminal	Sbcam	Abcam Cat# ab62352, RRID:AB_944418
Anti-Sarcomeric Alpha Actinin antibody	Abcam	Abcam Cat# ab137346, RRID: AB_2909405
APC anti-human CD172a/b (SIRPα/β) Antibody	BioLegend	BioLegend Cat# 323809, RRID:AB_11219399
ki67	Abcam	Abcam Cat# ab16667, RRID:AB_302459
Lab Vision™ TroponinT, Cardiac Isoform Ab-1,Mouse Monoclonal Antibody	ThermoFisher	ab Vision Cat# MS-295-P1, RRID:AB_61808
MF20---Myosin heavy chain, sarcomere (MHC)	DSHB	DSHB Cat# MF 20, RRID:AB_2147781
OXPHOS Rodent Cocktail	Abcam	Abcam Cat# ab110413, RRID:AB_2629281
Total OXPHOS Human WB Antibody Cocktail	Abcam	Abcam Cat# ab110411, RRID:AB_2756818
ph3Anti-HistoneH3 (phosphoS10) antibody [mAbcam14955](AlexaFluor®488)	abcam	Abcam Cat# ab197502, RRID:AB_2909407
rH2ax Antibody	Bethyl	Bethyl Cat# A300-081A, RRID:AB 203288
StarBright Blue 520 Goat Anti-Rabbit IgG	Biorad	Bio-Rad Cat# 12005870, RRID:AB_2884949
StarBright Blue 700 Goat Anti-Mouse IgG	Biorad	Bio-Rad Cat# 12004159, RRID:AB_2884948
VDAC1 antibody (AC)	Abcam	Abcam Cat# ab154856, RRID:AB_2687466
SOD-2 (E-1Q)	Santa Cruz	Santa Cruz Biotechnology Cat# sc-137254, RRID:AB_2191808
Purified Mouse Anti-DLP1 antibody	BD Bioscience	BD Biosciences Cat# 611112,RRID:AB_398423
HIF-1 alpha Antibody (H1alpha 67)	Santa Cruz	Santa Cruz Biotechnology Cat# sc-53546, RRID:AB_629639
YAP(YAP1)	DSHB	DSHB Cat# YAP1 8J19, RRID:AB_2619554
Dyes
DAF-FM diacetate	ThermoFisher	D23842
Lyso Tracker™ Deep Red	ThermoFisher	L12492
Mitosox	ThermoFisher	M36008
Mitotracker Red	ThermoFisher	M7512
Mitotracker Green	ThermoFisher	M7514
Rhod2	ThermoFisher	R1245MP
TMRE	ThermoFisher	T669
DAF-FM diacetate	ThermoFisher	D23842
Lyso Tracker™ Deep Red	ThermoFisher	L12492
Mitosox	ThermoFisher	M36008
Mitotracker Red	ThermoFisher	M7512
Mitotracker Green	ThermoFisher	M7514
Rhod2	ThermoFisher	R1245MP
TMRE	ThermoFisher	T669
Deposited data
Single cell RNA sequencing data of human iPSC-CM, related to Figure 5	This paper	GEO: GSE146341
Single cell RNA sequencing data of human fetal hearts, related to Figure 7H	(Cui et al., 2019)	GEO: GSE106118
Whole exome sequencing data	This paper	dbGAP: phs001256. v1.p1
Matlab codes for Calcium calculation	This paper	Dataverse:https://doi.org/10.7910/DVN/9ZWTAL
Critical commercial assays
e-Mycoplus Mycoplasma PCR DetectionKit	bocascientific	25235
In Situ Cell Death Detection Kit, TMRred	sigma	12156792910Roche
Matrigel® hESC-Qualified Matrix, *LDEV-free	Corning	354277
mTeSR	STEMCELLTechnologies	85850
Nucleofector™ Kits for Human Dermal Fibroblast (NHDF)	Lonza	VPD-1001
RPMI 1640 Medium, noglucose	Thermofisher	11879020
Seahorse XF Cell Mito Stress Test Kit	Agilent	Cat#103015-100
Oligonucleotides
See Table S1	N/A	N/A
Recombinant DNA
pCXLE-hOCT3/4-shp53-F	Addgen	RRID:Addgene_270 77
pCXLE-hSK	Addgen	RRID:Addgene_270 78
pCXLE-hUL	Addgen	RRID:Addgene_270 80
pCXLE-EGFP	Addgen	RRID:Addgene_270 82
Software and algorithms
CellRanger	10XGenomics	Cell Ranger, RRID:SCR_017344
ChemiDoc	Bio-rad	Bio Rad ChemiDoc MP Imaging System, RRID:SCR_019037
Graph Pad Prism	GraphPad Software	GraphPad Prism, RRID:SCR 002798
ImageJ	ImageJ	ImageJ, RRID: SCR_003070
LASX	Leica Application Suite X	Leica Application Suite X,RRID:SCR_013673
MATLAB	MathWorks	MATLAB, RRID:SCR_001622
Rstudio	R Core	R Project for Statistical Computing, RRID: SCR_001905
Seahorse Wave Desktop Software	Agilent	Seahorse Wave, RRID:SCR_014526
Seurat 3	(Butler et al., 2018)	SEURAT, RRID: SCR_007322
BWA-MEM Version 0.7.15	(Li and Durbin, 2010)	BWA, RRID:SCR_010910
GATK Version 3.4-0	(Van der Auwera et al., 2013)	GATK, RRID:SCR_001876
ANNOVAR	(Wang et al., 2010)	ANNOVAR, RRID:SCR_012821
ToppGene	(Chen et al., 2009)	ToppGene Suite, RRID:SCR_005726
dbNSFP	(Liu et al., 2016)	dbNSFP, RRID:SCR_005178
WebGestalt	(Liao et al., 2019)	WebGestalt: WEB- based GEne SeT AnaLysis Toolkit, RRID:SCR_006786)
BioGRID	(Stark et al., 2011)	Biological General Repository for Interaction Datasets (BioGRID), RRID:SCR_007393
HPRD	(Keshava Prasad et al., 2009)	Human Protein Reference Database, RRID:SCR_007027
PANTHER	(Mi et al., 2021)	PANTHER, RRID:SCR_004869
Scanpy 1.8.2	(Wolf et al., 2018)	Scanpy, RRID:SCR_018139
BioRender	Created with BioRender.com	Biorender, RRID:SCR_018361

Highlights.

• Patient iPSC derived cardiomyocyte were used to model clinical heart failure

• Heart failure is linked to mitochondrial defects, redox stress, apoptosis.

• Failed antioxidant response is associated with heart failure outcome.

• Sildenafil or TUDCA rescued uncompensated redox stress, suggesting possible therapy