AMPK activators increased ATP content, mitochondrial membrane potential, mitochondrial content and mtDNA:nDNA ratio in hiPSC-CMs
To examine the role of AMPK activators on the maturation of hiPSC-CMs, we performed a dose screening of AMPK activators on several features of mitochondrial maturation, including ATP content, the ratio of mtDNA:nDNA and mitochondrial membrane potential. The 2D hiPSC-CM cultures (> 90% pure cardiomyocytes) were treated with AMPK activators AMP-mimetric 5-aminoimidazole-4-carboxamide ribonucleoside (AICAR), a thienopyridone derivative (A-769662), a benzimidazole derivative (EX229 or Compound 991) and metformin at various concentrations (see Additional file 1: Results on dose screening of AMPK activators). Compared with DMSO control, all four activators increased the ATP content and mitochondrial membrane potential after 14 days of treatment, and A-769662 and EX229 also increased mtDNA:nDNA ratio in hiPSC-CMs derived from both IMR-90 hiPSCs (Additional file 1: Fig. S1) and SCVI-273 hiPSCs (Additional file 1: Fig. S2). Based on all the three measurements, A-769662 at 100 µM (A100) and 200 µM (A200) and EX229 at 10 µM (E10) and 50 µM (E50) were selected in subsequent studies.
We further examined the effect of AMPK activators on mitochondria maturation in 3D culture of hiPSC-CMs. Cardiac spheres were generated by microscale tissue engineering from hiPSC differentiation cultures [12, 13]. The maturation outcomes were then measured after cardiac spheres were treated with A-769662 or EX229 for 7 or 14 days starting on differentiation day 14 (Fig. 1A). The 3D cultures contained highly enriched hiPSC-CMs (> 90% pure cardiomyocytes; Additional file 2: Video S1) as analyzed by flow cytometry (Fig. 1B) and high-content imaging (Fig. 1C). The hiPSC-CM purity of the cultures remained similar after these cells were treated with A-769662 or EX229 (Fig. 1B–D, detailed in Additional file 1: Results). At treatment day 7, both EX229 and A-769662 significantly increased the ATP content (Fig. 1E). Cells treated with E10 had the highest ATP content levels among all tested conditions (> 1.7-fold higher in E10-treated cells than that in DMSO-treated cells) (Fig. 1E). At treatment day 14, the effect of elevated ATP content with treatment at optimal doses showed the same trend as at treatment day 7 (Additional file 1: Fig. S3B).
We also investigated the effect of these two AMPK activators on the mitochondrial membrane potential using TMRM, a cell-permeant and cationic dye that accumulates in active mitochondria with intact membrane potentials. At treatment day 7, cells in all four tested conditions showed increased mean fluorescence intensity of TMRM compared with DMSO-treated cells (Fig. 1F). In agreement with the effect on ATP content, EX229 was more potent in promoting mitochondrial membrane potential than A-769662. Similar increased TMRM levels were also observed in 3D hiPSC-CMs at treatment day 14 (Additional file 1: Fig. S3C). E10, E50 and A200 also increased mean fluorescence intensity of MitoTracker Red, another indicator of mitochondrial membrane potential at treatment day 7 (Fig. 1G) and day 14 (Additional file 1: Fig. S3D).
In addition, we examined the expression of TOM20, an indicator of mitochondrial content, in 3D hiPSC-CMs treated with A-769662 and EX229. Both quantitative methods of ArrayScan and flow cytometry detected elevated expression of TOM20 in E10-, E50- and A200-treated cultures (Fig. 1H and I; Additional file 1: Fig. S3E and F).
We further evaluated mitochondrial DNA development by measuring mtDNA:nDNA ratio in 3D hiPSC-CMs treated with A-769662 and EX229 (Fig. 1J; Additional file 1: Fig. S3G). Compared with DMSO treatment, all other treatments resulted in the increased mtDNA:nDNA ratio of ND1/SDHA and ND1/LPL. Higher levels of increase were observed in cells treated with E10, E50 and A200 than with A100. E10, E50 and A200 also increased the mtDNA:nDNA ratio of mt-CO2/SDHA and mt-CO2/LPL, and E10-treated cells had the highest mtDNA:nDNA ratio among conditions, reaching a more than two-fold increase in mtDNA:nDNA ratio compared with that of DMSO-treated cells.
These results showed that AMPK activators increased ATP content, mitochondrial membrane potential and mtDNA:nDNA ratio in hiPSC-CMs, promoting mitochondrial development.
AMPK activators promoted mitochondrial function and FAO
To characterize mitochondrial function, we measured major aspects of mitochondrial coupling and respiratory control—basal respiration, ATP production, maximal respiration, spare respiratory capacity and non-mitochondrial respiration—by the sequential additions of oligomycin (ATP synthase inhibitor), FCCP (an uncoupler of oxidative phosphorylation) and rotenone and antimycin A (electron inhibitors) using the Seahorse Extracellular Flux Analyzer. At treatment day 7, AMPK activator-treated cells had higher basal, maximal respiration, spare respiratory capacity, non-mitochondrial respiration and ATP production than DMSO-treated cells (Fig. 1K). Among the treatment conditions, E10 resulted in the highest basal, maximal respiration, ATP production, spare respiratory capacity and non-mitochondrial respiration (Fig. 1K). At treatment day 14, the effect of E10 on basal, maximal respiration, ATP production and non-mitochondrial respiration remained similar to treatment day 7 (Additional file 1: Fig. S3H).
We next examined if enhanced mitochondrial function in AMPK activator-treated cells coincided with the metabolic changes in the usage of energetic substrates. Compared with DMSO-treated cells, EX229-treated hiPSC-CMs had increased fatty acid uptake as detected by a fluorometric assay kit, and EX229 was more potent in enhancing the fatty acid uptake than A-769662 (Fig. 1L; Additional file 1: Fig. S3I). The consumption of glucose did not significantly change among conditions during monitoring for 4 days after the treatment (Fig. 1M). These observations indicate that treatment with EX229 boosted the usage of fatty acid.
Since EX229 was more potent than A-769662 in promoting mitochondrial function and enhancing the fatty acid uptake, we also examined the level of FAO and glycolysis in EX229-treated hiPSC-CMs. By monitoring OCR using the Seahorse XF24 Extracellular Flux Analyzer, which is an established assay for FAO measurement , cells were treated with ETO, a specific inhibitor of CPT1 which is an enzyme that mediates internalization of fatty acids into the mitochondrial matrix for oxidation, and the level of FAO was then calculated by the ETO-induced reduction in OCR normalized to the number of cells. Compared with DMSO-treated cells, EX229-treated hiPSC-CMs had higher levels of basal respiration and FAO (Fig. 1N). The level of glycolysis in EX229-treated hiPSC-CMs was not significantly different from that in the DMSO-treated cells (Fig. 1O). Therefore, AMPK activator EX229 promoted mitochondrial function and FAO.
AMPK inhibitor reduced mitochondrial maturation and abolished the effect of AMPK activators on mitochondrial maturation
To further evaluate the role of AMPK in mitochondrial maturation of hiPSC-CMs, we examined the effect of a specific chemical inhibitor of AMPK, Compound C, on mitochondrial membrane potential, mitochondrial content and ATP content in 3D hiPSC-CMs. Following 7 days of treatment with Compound C, levels of mitochondrial membrane potential (Fig. 2A), mitochondrial content (Fig. 2B) and ATP content (Fig. 2C) in 3D hiPSC-CMs were reduced compared with cultures treated with DMSO control. In addition, the levels of mitochondrial membrane potential (Fig. 2A), mitochondrial content (Fig. 2B) and ATP content (Fig. 2C) in 3D hiPSC-CMs co-treated with an AMPK activator and Compound C were lower than in cultures treated with AMPK activator alone (comparing E10 + CC vs. E10, E50 + CC vs. E50, A100 + CC vs. A100 and A200 + CC vs. A200).
We further examined if Compound C affected mitochondrial function by the Seahorse Mito Stress assay. Following 7 days of Compound C treatment of 3D hiPSC-CMs (derived from IMR90 hiPSCs), the levels of basal respiration, maximal respiration, ATP production, spare respiratory capacity and non-mitochondrial oxygen consumption were lower than in cultures treated with DMSO control (Fig. 2D and E). In addition, Compound C treatment abolished increased mitochondrial function in AMPK activator-treated cultures (Fig. 3C, D). After 14 days of the treatment, Compound C also reduced ATP content (Additional file 1: Fig. S4A), mitochondrial membrane potential (Additional file 1: Fig. S4B), mitochondrial content (Additional file 1: Fig. S4C) and mitochondrial function (Additional file 1: Fig. S4D and E).
In addition, we also examined the effect of Compound C in hiPSC-CMs derived from another cell line. The negative effect of Compound C on ATP content, mitochondrial membrane potential, mitochondrial content and mitochondrial function was also observed in 3D hiPSC-CMs derived from SCVI273 hiPSCs at treatment day 7 (Additional file 1: Fig. S5) and day 14 (Additional file 1: Fig. S6).
Taken together, these results indicate that AMPK inhibitor Compound C reduced ATP content, mitochondrial content, mitochondrial membrane potential and mitochondrial function in 3D hiPSC-CMs derived from both IMR90 hiPSCs and SCVI273 hiPSCs, supporting a role of AMPK in mitochondrial maturation of hiPSC-CMs.
AMPK knockdown using small interfering RNA inhibited mitochondrial maturation
We next investigated the role of AMPK in mitochondrial maturation by genetic knockdown of AMPK, which was achieved by the expression of two sets of small interfering RNAs (siRNAs) that targeted PRKAA1 (encoding AMPKα1) and PRKAA2 (encoding AMPKα2). hiPSC-CMs derived from IMR90 hiPSCs were transfected with AMPK siRNAs or a scrambled control siRNA followed by E10 treatment for 7 days. Compared with the control siRNA, AMPK siRNAs caused efficient knockdown of PRKAA1 and PRKAA2 expression at 50–200 nM (Fig. 3A). In hiPSC-CMs treated with AMPK activator E10 or DMSO, levels of PRKAA1 and PRKAA2 expression remained lower in AMPK siRNAs cultures than those in the control siRNA cultures after siRNA treatment for 7 days (Fig. 3B).
We then examined the effect of AMPK knockdown on mitochondrial maturation in cultures treated with E10 or DMSO. Levels of ATP content, TMRM and TOM20 in cultures treated with AMPK siRNAs were significantly lower than those in the control siRNA cultures (Fig. 3C–E), indicating knockdown of AMPK reduced ATP content, mitochondrial membrane potential and mitochondrial content. Compared with the control siRNA cultures, AMPK siRNAs cultures also had decreased fatty acid uptake (Fig. 3F) and decreased mtDNA/nDNA ratio (Fig. 3G). In addition, AMPK knockdown abolished the positive effect of E10 on ATP content, mitochondrial membrane potential, mitochondrial content, mtDNA/nDNA ratio and fatty acid uptake (Fig. 3C–G).
We further examined the outcomes of AMPK knockdown on mitochondrial function using Seahorse Mito Stress assay. AMPK siRNAs cultures had significantly lower levels of basal respiration, maximal respiration, ATP production, spare respiratory capacity and non-mitochondrial oxygen consumption than did the control siRNA cultures (Fig. 3H). AMPK knockdown also abolished the positive effect of E10 on these parameters associated with mitochondrial function (Fig. 3H).
Therefore, the results from both pharmacological and genetic inhibition of AMPK consistently support a functional role of AMPK in mitochondrial maturation of hiPSC-CMs.
AMPK activators promoted structural maturation, calcium handling, electrophysiology and contractility index
To evaluate structural maturation, we performed immunostaining of α-actinin, a major component of the Z-line, on hiPSC-CMs treated with EX229 and A-769662 for 7 days and quantified sarcomere length, cell area, cell perimeter and length/width ratio (Additional file 1: Fig. S7 and Fig. 4A). Compared with sarcomere length of DMSO-treated cells (1.658 ± 0.011 µm), sarcomere length significantly increased in E10- (1.861 ± 0.013 µm), E50- (1.949 ± 0.017 µm) and A200- (1.868 ± 0.016 µm) treated hiPSC-CMs (Fig. 4A). In addition, DMSO-treated cells were smaller (3.037 ± 103 µm2) than those treated with E10 (5.096 ± 164 µm2), E50 (4.892 ± 173 µm2) and A200 (4.385 ± 153 µm2) based on cell area (Fig. 4A). Similarly, E10, E50 and A200 treatment increased cell perimeters (Fig. 4A). E10 and A200 also increased length/width ratio of the cells (Fig. 4A).
We next investigated the calcium handling properties of hiPSC-CMs using line-scan confocal imaging after cells were loaded with the intracellular calcium dye, Fluo-4AM. E10- and A100-treated cells displayed significantly higher calcium transient amplitude (Fig. 4B). All treatments with AMPK activators led to significantly higher maximal rise slope and maximal decay slope than DMSO control (Fig. 4B). Time to rise half amplitude was significantly shorter with E10 and E50 treatments. E10- and A200-treated cells had significantly shortened the time to peak amplitude and the time to decay half amplitude. Notably, only E10-treated cells had decreased rise tau and decay tau, making E10 the most potent treatment with enhanced calcium transient kinetics based on all parameters examined (Fig. 4B). In addition, enhanced calcium transient kinetics were also observed in E10-treated hiPSC-CMs after 14 days of the treatment (Additional file 1: Fig. S3K). Together, these results indicate that AMPK activators significantly increased the kinetics of calcium transients, a functional characteristic of more mature hiPSC-CMs.
We further investigated the electrophysiological properties of hiPSC-CMs using FluoVolt probe and confocal imaging. E10-treated hiPSC-CMs had significantly higher peak amplitude and shorter peak rise time. The upstroke velocity was faster in both E10- and A200-treated cells while the action potentials APD50 and APD80 did not change (Fig. 4C). In addition, compared with DMSO-treated cells, E10- and E50-treated 3D hiPSC-CMs had higher average maximum contraction (DMSO, 150.37 ± 8.18 µm/s; E10, 182.78 ± 9.33 µm/s and E50, 177.21 ± 7.20 µm/s) and relaxation (DMSO, 69.03 ± 4.92 µm/s; E10, 100.90 ± 6.11 µm/s and E50, 94.83 ± 5.63 µm/s) velocities (Fig. 4D).
Together, these results indicate that treatment of AMPK activators promoted the functional maturation of hiPSC-CMs.
AMPK activator-treated 3D hiPSC-CMs had increased expression of genes involved in controlling mitochondrial properties and cardiac structural and functional features
We next investigated the effect of AMPK activator treatment on gene expression profile (we chose E10 for the experiments given it showed most potent effect on maturation phenotypes). RNA-sequencing (RNA-seq) analysis was performed on 3D hiPSC-CMs treated with E10 or DMSO for 7 days as well as three heart tissue samples from pediatric left ventricle (LV). At the threshold of adjusted P value < 0.05, we identified 1969 differentially expressed genes (DEGs) in E10-treated cells compared with DMSO-treated cells, including 888 upregulated genes and 1081 downregulated genes (Fig. 5A). DEGs (9817) were identified in LV samples compared with DMSO-treated cells, including 4992 upregulated genes and 4825 downregulated genes (Fig. 5B). Common DEGs were visualized in a heatmap showing a differential gene expression pattern across the three groups (Fig. 5C). Specifically, there were 1055 commonly expressed DEGs between E10-treated cells and LV compared with DMSO-treated cells, including 456 upregulated and 599 downregulated DEGs (Fig. 5D). Top commonly expressed DEGs between E10-treated hiPSC-CMs and LV were listed in Additional file 1: Table S3.
We performed GO analysis on the DEGs in E10- vs. DMSO-treated hiPSC-CMs and LV vs. DMSO-treated hiPSC-CMs. Strikingly, a large number of biological processes were commonly enhanced in E10-treated cells and LV. Most top upregulated GO terms represented mitochondrial metabolism including cellular respiration, aerobic respiration, oxidative phosphorylation, ATP metabolic process, respiratory electron transport chain, nucleoside triphosphate metabolic process, proton transmembrane transport, acyl-CoA metabolic process, tricarboxylic acid cycle and mitochondrial transport (Fig. 5E). These upregulated GO terms also included fatty acid metabolic process, lipid localization, FAO, lipid transport and heart contraction, representing the upregulation of fatty acid metabolism, cardiac structural and functional development in E10-treated cells (Fig. 5E). In addition, E10 downregulated GO terms associated with extracellular matrix organization, extracellular structure organization, skeletal system development, mesenchyme development, embryonic organ development and neural tube development (Fig. 5E).
The investigation of individual gene changes revealed a number of DEGs that are critical for mitochondrial function in E10-treated hiPSC-CMs (Fig. 5F). These included MT-ND1 (mitochondrially encoded NADH dehydrogenase 1), MT-CO2 (mitochondrially encoded cytochrome c oxidase II), COQ10A (coenzyme Q10A), CKMT2 (creatine kinase, mitochondrial 2), OPA1 (OPA1 mitochondrial dynamin-like GTPase), MFN2 (mitofusin 2), UCP2 (uncoupling protein 2), UCP3 (uncoupling protein 3), NDUFB5 (NADH:ubiquinone oxidoreductase subunit B5), SLC2A4 (solute carrier family 2 member 4), PDHA1 (pyruvate dehydrogenase E1 subunit alpha 1), PDHX (pyruvate dehydrogenase complex component X) and COX (cytochrome c oxidase) subunits COX5A, COX6A2 and COX7A2. The upregulation of MT-ND1 and MT-CO2 was consistent with the result in mtDNA/nDNA ratio. The increased expression of SLC2A4 coincided with the theory that the switch of glucose transporter from SLC2A1 to SLC2A4 occurs along with the CM maturation. PDH complex consists of key enzymes that link the glycolysis metabolic pathway to the tricarboxylic acid (TCA) cycle and catalyze the process of pyruvate decarboxylation by converting pyruvate to acetyl-coenzyme A and carbon dioxide, where PDHA1 plays a key role in maintaining the function of PDH complex. DLAT (dihydrolipoamide S-acetyltransferase) and DLD (dihydrolipoamide dehydrogenase) which encode the other two subunits of PDH were also upregulated, suggesting the enhanced capacity of pyruvate decarboxylation. We also found multiple DEGs that encode the key enzymes in TCA cycle, including CS (citrate synthase), ACO2 (aconitase 2), IDH3A (isocitrate dehydrogenase (NAD( +)) 3 catalytic subunit alpha), OGDH (oxoglutarate dehydrogenase), DLST (dihydrolipoamide S-succinyltransferase), DLD, SDHB (succinate dehydrogenase complex iron sulfur subunit B), SDHD (succinate dehydrogenase complex subunit D), FH (fumarate hydratase) and the cytosolic isozyme MDH1 (malate dehydrogenase 1). In complex I of the respiratory chain, in addition to NDUFB5, some other subunits of NADH:ubiquinone oxidoreductase were also upregulated, including NDUFA4, NDUFA5, NDUFA10, NDUFAB1, NDUFAF4, NDUFB3, NDUFB9, NDUFS1, NDUFS2 and NDUFS3. SDHB and SDHD in complex II together with ubiquinol-cytochrome c oxidoreductase complex (UQCR) subunits UQCR10, UQCRB, UQCRC1, UQCRC2, UQCRFS1 and UQCRH in complex III of the respiratory chain were upregulated. Additionally, we also observed the upregulation of some other COX subunits in complex IV including COX4I1, COX6B1, COX6C, COX7B, COX7C and the COX assembly subunit COX14. Genes that encode ATP synthase, for example, ATP5F1A (ATP5A1, ATP synthase F1 subunit alpha), ATP5F1B (ATP5B, ATP synthase F1 subunit beta) and ATP5F1C (ATP5C, ATP synthase F1 subunit gamma), were also upregulated. Our mitochondrial function test on E10-treated cells using Seahorse showed significant increased proton leak, which is known to be activated by enhanced level of oxidative phosphorylation . UCP2 and UCP3 together with another upregulated gene, SOD2 (superoxide dismutase 2), enhanced their expression to mediate the potential oxidative stress induced by the increased proton leak [14,15,16,17].
Several key genes involved in fatty acid metabolism were upregulated in E10-treated cells (Fig. 5G). FABP3 (fatty acid-binding protein 3) is a critical gene in transporting long-chain fatty acids from cytoplasm to mitochondria. SLC27A4, also known as FATP4, is a fatty acid transport protein that functions in translocation of long-chain fatty acids cross the plasma membrane. PPARGC1A (peroxisome proliferator-activated receptor gamma coactivator 1-alpha), also known as PGC-1a (PPARG coactivator 1 alpha), is a master regulator of mitochondrial biogenesis. ACSL1 (acyl-CoA synthetase long-chain family member 1) plays a key role in lipid biosynthesis and fatty acid degradation. Other key upregulated genes that participated in FAO were upregulated in E10-treated cells, included HADHB (hydroxyacyl-CoA dehydrogenase trifunctional multienzyme complex subunit beta), FASN (fatty acid synthase), the enzymes of acyl-CoA dehydrogenase family (ACADs) ACAD10, ACADM, ACADSB, ACAT1 (acetyl-CoA acetyltransferase 1) and ATP-binding cassette (ABC) transporters ABCA5, ABCB4, ABCC5, ABCC9, ABCD1, ABCD3, ABCG1 and TAP1.
Additionally, E10-treated cells had upregulated PPARD (peroxisome proliferator-activated receptor delta), which was recently identified for its role in inducing metabolic and contractile maturation of hiPSC-CMs . E10-treated cells also had upregulated MB (myoglobin) which transfers oxygen from the cell membrane to the mitochondria, indicating increased cellular respiration in mitochondria. These results were consistent with improved mitochondrial function in E10-treated hiPSC-CMs.
E10 treatment also upregulated genes that play an important role in the maturation of electrophysiology and Ca2+ handling of cardiomyocytes (Fig. 5H). These genes included KCNJ2 (potassium inwardly rectifying channel subfamily J member 2), PLN (phospholamban), ATP2A2 (ATPase sarcoplasmic/endoplasmic reticulum Ca2+ transporting 2), TNNI3 (troponin I3, cardiac type), NPPA (natriuretic peptide A), ADRB1 (adrenoceptor beta 1), KCNN2 (potassium calcium-activated channel subfamily N member 2), KCNQ1 (potassium voltage-gated channel subfamily Q member 1), SCN1B (sodium voltage-gated channel beta subunit 1), SCN3B (sodium voltage-gated channel beta subunit 3), CACNG4 (calcium voltage-gated channel auxiliary subunit gamma 4), MYL3 (myosin light chain 3), MYBPC3 (myosin-binding protein C3), TMEM38A (transmembrane protein 38A) and EHD3 (EH domain containing 3). In addition, we observed the decreased expression of the automaticity ion channel HCN4 (hyperpolarization-activated cyclic nucleotide-gated potassium channel 4). Notably, many of these upregulated genes were reported as cardiomyocyte maturation markers. For example, TNNI3 represents one of the outstanding cardiomyocyte maturation markers . In addition, majority of these upregulated genes in E10-treated cells were also upregulated in LV samples compared with DMSO-treated cells, although the differences in expression levels of some genes were also noted (Additional file 1: Fig. S8).
Together, these observations—increased expression of genes involved in controlling mitochondrial properties, cardiac structure and cardiac function—show that E10-treated hiPSC-CMs had an improved molecular signature indicative of more mature cardiomyocytes.
Proteomics analysis revealed upregulation of key regulators on mitochondrial function and fatty acid metabolism in E10-treated 3D hiPSC-CMs
To further examine the effect of AMPK activator on hiPSC-CM maturation, we performed quantitative proteomic analysis on 3D hiPSC-CMs treated with E10 vs. DMSO for 7 days. Of 5113 proteins analyzed, E10 treatment resulted in 272 upregulated and 41 downregulated proteins (absolute fold change > 1.3, P < 0.05) (Fig. 6A).
GO term analysis was performed using the differentially expressed proteins. Notably, in E10-treated hiPSC-CMs, top upregulated GO terms of biological process (BP) were associated with mitochondrial function and fatty acid metabolism (Fig. 6B). Based on cellular component (CC) and molecular function (MF), top GO terms were also related to mitochondrial metabolism (Fig. 6C). KEGG pathway analysis of the upregulated proteins in the E10-treated cells indicated upregulation of pathways associated with oxidative phosphorylation, metabolic pathways and citrate cycle (Fig. 6C). In addition, fatty acid degradation, cardiac contraction and PPAR signaling were also present in the upregulated KEGG pathways.
The String protein–protein interaction analysis identified a large number of differentially expressed proteins that are involved in the oxidative phosphorylation pathway. The high connectivity among these proteins further indicated their important roles in oxidative phosphorylation (Fig. 6D).
Detailed examination of the upregulated differentially expressed proteins identified the key proteins associated with mitochondrial biogenesis, TCA cycle and oxidative phosphorylation (Fig. 6E). These upregulated proteins included eight enzymes that catalyze the TCA cycle—CS, ACO2, IDH3A, IDH3B (isocitrate dehydrogenase (NAD(+)) 3 non-catalytic subunit beta), IDH3G (isocitrate dehydrogenase (NAD(+)) 3 non-catalytic subunit gamma), OGDHL (oxoglutarate dehydrogenase L), OGDH, DLST, DLD, SUCLG1 (succinate-CoA ligase GDP/ADP-forming subunit alpha), SUCLA2 (succinate-CoA ligase ADP-forming subunit beta), SDHA, SDHB, FH and MDH2 (malate dehydrogenase 2). In agreement with RNA-seq, E10 treatment upregulated SLC2A4, MT-CO2, CKMT2, PDHA1, PDHX, DLAT, DLD and another PDH E1 subunit PDHB.
In agreement with the observation in RNA-seq data, many important proteins in mitochondrial electron transport chains were upregulated in E10-treated cells. Specifically, we observed upregulated proteins in complex I—NADH dehydrogenase MT-ND2 and MT-ND5—and several NADH:ubiquinone oxidoreductase subunits—NDUFAF3, NDUFA12, NDUFA4, NDUFS2, NDUFA9, NDUFA5, NDUFAB1, NDUFA1, NDUFA3, NDUFS5, NDUFA8, NDUFC2, NDUFAF4, NDUFS8, NDUFB9, NDUFB1, NDUFB8, NDUFB11, NDUFS4, NDUFB3, NDUFB5, NDUFA2, NDUFA7, NDUFA11, NDUFA10 and NDUFS7. Upregulation of SDHA and SDHB in complex II was detected with upregulated SDHB being detected by RNA-seq as well. Additionally, UQCR subunits—UQCR10, UQCRB, UQCRC1 and UQCRH—in complex III and terminal components of the respiratory chain—COA6, COX6B1, COX6C, COX7A2 and COX7B—were upregulated. In line with our RNA-seq data, several subunits of ATP synthase F1 and F0 complex were upregulated, including ATP5A1, ATP5B, ATP5C1, ATP5O (ATP5PO, ATP synthase F1 subunit OSCP), ATP5H (ATP5PD, mitochondrial F0 complex, subunit D), ATP5I (ATP5ME, mitochondrial F0 complex, subunit E), ATP5L (ATP5MG, mitochondrial F0 complex, subunit G) and ATP5J (ATP5PF, mitochondrial F0 complex, subunit F). Also, E10-treated cells had increased expression of several solute carriers (SLC) family 25 members, including SLC25A24, SLC25A11, SLC25A5, SLC25A13 and SLC25A4. These results indicated an increased capacity of electron translocation for respiration and ATP synthesis in E10-treated cells.
Upregulation of many important proteins involved in fatty acid metabolism was also indicated. These included fatty acid-binding proteins FABP3 and FABP7, which are the intracellular fatty acid-binding proteins that regulate the uptake and transport of long-chain fatty acids into cells. The expression of SLC27A4 also increased, suggesting enhanced capacity of fatty acid uptake in E10-treated cells upon entry in cells and within cells. E10-treated cells had increased expression of acyl-CoA synthases (ACSs, which catalyze the process of fatty acid activation into fatty acyl-CoAs), including long-chain family members ACSL1, ACSL3 and short-chain family members ACSS2, ACSS3. Intriguingly, we found the upregulation of CPT1B (carnitine palmitoyltransferase 1B), carnitine acyl-carnitine translocase SLC25A20 and CPT2 (carnitine palmitoyltransferase 2). All of which are essential proteins in transporting long-chain fatty acyl-CoAs from the cytoplasm into mitochondria. Notably, E10-treated cells had upregulated MYLCD (malonyl-CoA decarboxylase, which catalyzes the breakdown of malonyl-CoA that inhibits CPT1, thus increasing the rate of CPT1-mediated fatty acid transport). E10 treatment also resulted in the upregulation of several enzymes which are essential to FAO process, including ACOX1 (acyl-CoA oxidase 1), ACADM, HADHA (hydroxyacyl-CoA dehydrogenase trifunctional multienzyme complex subunit alpha), HADHB and HADH (hydroxyacyl-CoA dehydrogenase). In addition, CROT (carnitine O-octanoyltransferase) and CRAT (carnitine O-acetyltransferase), BDH1 (3-hydroxybutyrate dehydrogenase 1) and ACAA1 (acetyl-CoA acyltransferase 1) involved in peroxisomal fatty acid oxidation as well as the key enzymes EC1 and EC2 in β-oxidation of unsaturated fatty acids were also upregulated. Collectively, the upregulation of the above essential proteins in E10-treated cells demonstrated increased consumption of fatty acid for the cell energy demand. Furthermore, comparison of the RNA-seq with proteomic data revealed a subset of overlapping genes and proteins that were differentially expressed in E10-treated cells. Specifically, 90 of them were upregulated and 25 were downregulated, respectively (Fig. 6G).
Together, these results revealed that E10 upregulated key regulators of mitochondrial function and FAO, which was consistent with improved metabolic maturation in E10-treated 3D hiPSC-CMs.
Metabolomics identified enhanced energy-related metabolic pathways in E10-treated hiPSC-CMs
To gain additional understanding of the metabolic differences associated with AMPK activation, we performed untargeted metabolomic analysis using HILIC (ultra-high-resolution mass spectrometry with hydrophilic interaction liquid chromatography) to compare the metabolic profile of E10-treated hiPSC-CMs with their DMSO-treated counterparts. Of the 9903 features analyzed, 917 features significantly changed in E10-treated hiPSC-CMs from DMSO controls (P < 0.05). Principal component analysis (PCA) of the metabolic profiles between E10-treated hiPSC-CMs and DMSO controls showed a clear separation of metabolome between the two cultures (Fig. 7A). Unsupervised two-way hierarchical analysis of the significant features confirmed the distinction of metabolic profiles between the E10-treated hiPSC-CMs and DMSO controls (Fig. 7C, Metabolic feature table in Additional file 3).
To identify the significantly affected pathways, we performed pathway enrichment analysis on the 917 metabolic features that were different between the groups and identified 12 pathways that were enriched in E10-treated hiPSC-CMs. Top pathways included multiple mitochondria and energy metabolism-related activities, including lysine metabolism, carnitine shuttle, vitamin B12 (cyanocobalamin) metabolism, glycerophospholipid metabolism, ubiquinone biosynthesis, arginine and proline metabolism, pentose phosphate pathway, TCA cycle, phosphatidylinositol phosphate metabolism, urea cycle/amino group metabolism, androgen and estrogen biosynthesis and metabolism, pentose and glucuronate interconversions (Fig. 7D).
In E10-treated cells, two intermediates of the glycolysis pathway, D-glucose-6-phosphate and fructose-1,6-bisphosphate, were significantly decreased, while the level of D-glucose and pyruvate was not significantly different compared with DMSO-treated cells (Fig. 7E). In addition, the level of lipoamide was significantly increased in E10-treated cells (Fig. 7F). Lipoamide acts as an essential cofactor for mitochondrial 2-ketoacid dehydrogenases that are key enzymes in TCA cycle, which include PDH, OGDH and branched-chain ketoacid dehydrogenase (BCKDH) [20,21,22]. Combined results in RNA-seq and Proteomics on upregulated PDH, OGDH and BCKDH, the increased level of lipoamide further suggested an enhanced TCA cycle in E10-treated hiPSC-CMs based on our findings in GO and KEGG pathway analysis.
Among the upregulated pathways, carnitine shuttle system is responsible for the transport of fatty acids into mitochondria for subsequent β-oxidation. Upon cellular entry, fatty acids are activated into fatty acyl-CoAs by ACSs followed by conjugation to carnitine forming fatty acyl-carnitines catalyzed by CPT1B and located in the outer mitochondrial membrane. Fatty acyl-carnitines then diffuse across the mitochondrial membrane by action of carnitine acyl-carnitine translocase (SLC25A20) located on inner mitochondrial membrane. Once within mitochondria, carnitine is released from fatty acyl-carnitines by CPT2 and transferred back to cytoplasm, while fatty acyl-CoAs enter β-oxidation. Interestingly, E10 treatment significantly reduced the level of fatty acyl-carnitines in E10-treated cells compared to DMSO control (Fig. 7G and H). Specifically, we found marked, lower levels of short-chain fatty acyl-carnitine propionyl-carnitine, two medium-chain fatty acyl-carnitines (octadecenoyl-carnitine and hexadecenoyl-carnitine) and five long-chain fatty acyl-carnitines (L-palmitoyl-carnitine, linoelaidyl-carnitine, stearoyl-carnitine, arachidyl-carnitine and myristoyl-carnitine). Together, they imply elevated fatty acid β-oxidation in E10-treated cells.
Overall, these findings suggested AMPK activation-induced metabolic changes and promoted metabolic flux in energy-related metabolisms in hiPSC-CMs. These results demonstrate consistency with our findings in RNA-seq and proteomics analyses and reveal the upregulated expression profile of key regulators in carnitine shuttle system which remarkably facilitated fatty acids transportation and β-oxidation.
AMPK activator-treated hiPSC-CMs enabled pathological modeling
Based on aforementioned assessments of multiple features associated with metabolic maturation and extensive omic analyses, we concluded that E10 treatment could drive significantly increased FAO and metabolic maturation in 3D hiPSC-CMs. In addition, the expression of ADRB1 (adrenergic receptor B1) was significantly upregulated in AMPK-activated hiPSC-CMs (Fig. 8A). Given that the metabolic switch from FAO to glycolysis is associated with heart failure , we examined if E10-treated 3D hiPSC-CMs could model pathological response and recapitulate some of the changes related to heart failure including glycolysis enhancement, lipid accumulation and apoptosis. Three-dimensional hiPSC-CMs were first treated with E10 for 7 days to promote cardiomyocyte maturation. The cells were then treated with isoproterenol (100 µM), a small molecule adrenergic agonist to simulate pathological conditions for 6 days in hypoxic environment and low-glucose medium. The treatment with pathological stimuli increased glycolysis, glycolytic capacity and glycolytic reserve compared with no pathological stimuli in E10-treated cells, as detected by the extracellular acidification rate (ECAR; index of glycolytic activity) using the Seahorse assay (Fig. 8B). In contrast, the treatment of pathological stimuli did not alter glycolysis, glycolytic capacity and glycolytic reserve in immature cells (without E10 treatment) (Fig. 8B). Consistently, E10-treated cells but not immature cells had significant upregulation of the glycolysis-related genes including SLC2A1, GAPDH, LDHA and PKM2 under the treatment with pathological stimuli compared with cells without the treatment with pathological stimuli (Fig. 8C). These results indicate that E10-treated hiPSC-CMs had increased glycolysis when challenged with pathological stimuli, a pathological response associated with heart failure .
The pathological stimuli also induced the upregulation of triacylglycerol synthesis-associated gene CD36 (cluster of differentiation 36) and PLIN2 (Perilipin 2; a marker of lipid accumulation) in E10-treated 3D hiPSC-CMs but not in immature cells (Fig. 8D). In addition, the expression of HSL (LIPE; Lipase E, hormone-sensitive type) that hydrolyzes stored triglycerides to free fatty acids was downregulated in E10-treated hiPSC-CMs when challenged with pathological stimuli (Fig. 8D). Consistently, an increased accumulation of lipid droplets was observed in E10-treated hiPSC-CMs when challenged with pathological stimuli as detected by ArrayScan analysis of Nile red staining (Fig. 8E). Increased accumulation of lipid droplets is also a pathological response associated with heart failure .
In addition to the change in glycolysis and lipid accumulation, the pathological stimuli resulted in significantly increased cell death (Fig. 8F–I). Following the treatment of pathological stimuli, E10-treated cells had twofold decreased ATP content while immature hiPSC-CMs had 29% decreased ATP content compared with parallel cultures without the treatment of pathological stimuli. Similarly, the treatment of pathological stimuli resulted in higher magnitude of decrease in cell viability and cell counts in E10-treated cells compared with immature cells. In addition, increased caspase 3/7 activity was observed in E10-treated cells when challenged with the pathological stimuli.
These results suggest that E10-treated hiPSC-CMs responded to pathological stimuli, which was not achievable in immature hiPSC-CMs. Thus, E10-treated hiPSC-CMs have advantages for pathological modeling.