Cell culture model using h-iPSCs mimicking cardiac lipotoxicity phenotype in vitro
H-iPSCs were cultured under cardiomyocyte (CM)-induction condition for 7 days and followed by further culturing in standard medium (SM) for another 20 days [30]. Flow cytometry (FCM) analysis showed that 80.53 ± 6.56% of cells were α-actinin positive cells. Furthermore, ventricular marker myosin regulatory light chain 2 (MLC2v) positive cells represent 69.3 ± 4.1% of the total number of cells (Fig. S1).
To mimic diabetic conditions in vitro, we exposed 28 days old CMs to diabetic-like medium (DM) for 3 days [31]. We observed the accumulation of intracellular lipid droplets (497.2 ± 168.87% in DM, p < 0.003) (Fig. 1a and 1b), as well as a loss of the regular sarcomeric pattern of CMs (Fig. 1a and 1c). Also, the result revealed that the cells had reduced MLC2v positive area (48.41 ± 19.36% in DM, p < 0.001) (Fig. 1c and 1d) while the α-actinin positive area was not affected (Fig. 1c and 1d). Furthermore, we found that the gene expression of cardiac brain natriuretic peptide (BNP) was significantly up-regulated (+ 155.06%, p < 03.) in CMs exposed to DM for 3 days (Fig. 1e). Moreover, we identified alteration in genes levels of mitochondrial biogenesis and genes related to glucose and fatty acid (FA) oxidative metabolism in the CMs exposed to diabetic-like conditions, such as glucose transporter 4 (GLUT4) (-65.43%, p < 0.001), peroxisome proliferator-activated receptor alpha (PPARα) (-83.61%, p < 0.001), PPAR gamma coactivator-1 alpha (PGC-1α) (-74.78%, p < 0.001), carnitine palmitoyltransferase-IB (CPT-1B) (-63.57%, p < 0.001), cluster of differentiation 36 (CD36) (+ 165.01%, p < 0.02), pyruvate dehydrogenase kinase 4 (PDK4) (+ 272.69%, p < 0.002), and PPARγ (+ 189.09%, p < 0.03) compare with untreated CMs (Fig. 1e). Diabetic-like conditions induced a decrease in the beating rate by 67.44 ± 3.62% (p < 0.03) (Fig. 1f, online movies 1 and 2). This was accompanied by reduced expression of sarcoplasmic/endoplasmic reticulum Ca2+ ATPase 2a (SERCA2a) (-73.79%, p < 0.001) and altered relaxation-contraction dynamics with a significant increase in abnormal Ca2+ transient frequency (16.7 ± 3.2%) in CMs cultured in diabetic-like condition compared with the untreated CMs (3.5 ± 1.6%) (Fig. 1g and 1 h, Online movies 3 and 4).
To determine whether abnormal lipid accumulation occurs in h-iPSCs-derived-CMs cultured in DM can lead to cardiac lipotoxicity, we analyzed the ceramide content in CMs with mass spectrometry-based analysis. We found significantly increased in total ceramide levels (295.6 ± 78.3 in SM versus 498.3 ± 108.85 nM/10 µl in DM, p < 0.02) in CMs exposed to the DM (Fig. 1i). To investigate the role of de novo ceramide synthesis pathway in cardiotoxicity, the protein level of SPTLC1 was detected by western blot. Assessment of the protein level of SPTLC1 (p < 0.002) in CMs showed a significant increase in h-iPSCs-derived-CMs cultured in DM (Fig. 1j).
Accumulation of ceramides is known to increase mitochondrial ROS production [34]. Therefore, we evaluated mitochondrial oxidative stress in h-iPSCs-derived-CMs using MitoSOX staining. The result demonstrated an increased ROS production (p < 0.001) in CMs exposed to DM (Fig. 1k). Further, increased cleaved-caspase-3 level (p < 0.02) observed in the CMs exposed DM confirmed oxidative damage (Fig. 1l). In conclusion, the presented data suggest that lipotoxicity and oxidative stress were induced in the CMs when exposed to a diabetic-like condition.
Overexpression of SPTLC1 for endogenous ceramide induction
To further investigate the role of ceramides in diabetic cardiotoxicity, CMs were transfected with SPTLC1-overexpression plasmids, a subunit of the SPT complex critical for ceramide de novo synthesis. Protein levels of SPTLC1 increased (p < 0.002) after 24 h (Fig. 2a) compared to the control plasmid (CTRL). Moreover, a significant decrease in protein levels of SPTLC1 (p < 0.002) was noted in myriocin-treated SPTLC1-overexpression-CMs compared to non-treated CMs (Fig. 2a).
Overexpression of SPTLC1 in CMs resulted in increased total ceramide levels (p < 0.02). Among all ceramide species analyzed, it was found that the levels of long-chain (C16:00 and C18:00) and very long-chain (C24:00) ceramide species were much in SPTLC1-overexpressed-CMs compared with CTRL–CMs (Table 1). Furthermore, higher levels of apoptosis and cellular oxidative stress were confirmed by increased levels of cleaved caspase-3 (1.72 ± 1.13, p < 0.008) (Fig. 2b) and MitoSOX staining (1.9 ± 0.2, p < 0.001) (Fig. 2c and 2d), respectively. However, the effect of SPTLC1-overexpression was abolished following myriocin treatment (Fig. 2a-d).
Table 1
Ceramide species determined in SPTLC1-overexpressed CMs (nM/10 µl sample)
N-acyl chain | Empty (CTRL) ± SD | SPTLC1 ± SD |
C14:0 | 2,74 ± 0,86 | 1,82 ± 0,7 |
C16:0 | 53,57 ± 5,59 | 102,31 ± 10,44* |
C18:0 | 76,93 ± 6,07 | 166,17 ± 8,1* |
C20:0 | 16,31 ± 7,54 | 16,83 ± 6,32 |
C22:0 | 26,50 ± 12,06 | 23,4 ± 9,92 |
C24:0 | 20,6 ± 12,25 | 97,88 ± 26,42* |
Total | 196.66 ± 44.36 | 408.5 ± 62.1* |
Data are means ± SD; n = 4 for all groups. * p < 0.05 considered significant compared to control plasmid (CTRL) using an unpaired two-tailed t-test compared to SM-CMs. |
Effects of SPTLC1-overexpression on mitochondrial structure and auto/mitophagy
Next, we investigated the association between SPTLC1-overexpression and mitochondrial morphology and auto/mitophagy in CMs. For examination of the mitochondria and lysosomes, MitoTracker and LysoTracker were used, respectively. Our results revealed significant decreased (p < 0.001) MitoTracker signal (Fig. 3a and 3b) and increased (p < 0.001) LysoTracker area in the SPTLC1-overexpressed-CMs (Fig. 3a and 3c). Furthermore, these cells showed co-localization between LysoTracker and MitoTracker positive area (p < 0.001) (Fig. 3a and 3d). These findings suggested that SPTLC1-overexpression resulted in increased mitochondria fragmentation, which was associated with reduced mitochondria positive area and elevated mitophagy in CMs.
To further confirm this hypothesis, we analyzed mRNA expression levels of mitochondrial dynamic genes (Table 2). We investigated mitochondrial fission as indicated by dynamin-related protein 1 (DRP1) and mitochondrial fission factor (MFF) or mitochondrial fusion, as indicated by mitofusin 2 (MFN2) and optic atrophy 1 (OPA1), as well as auto/mitophagy markers such as microtubule-associated protein 1 light chain 3 beta (LC3B) and PETEN-induced kinase 1 (PINK-1) (Fig. 3e-g).
Table 2
Mitochondrial dynamics mRNA expression levels in SPTLC1-overexpressed-CMs
| Empty (CTRL) Untreated | SPTLC1Untreated | Empty (CTRL) + Myriocin | SPTLC1 + Myriocin |
Mitochondrial fission gene |
DRP1 | 1.0 ± 0.13 | 2.54 ± 0.17* | 1.04 ± 0.11 | 1.17 ± 0.43# |
MFF | 1.0 ± 0.12 | 1.98 ± 0.19* | 1.17 ± 0.43 | 1.19 ± 0.22# |
Mitochondrial fusion gene |
MFN2 | 1.0 ± 0.20 | 0.56 ± 0.14* | 0.83 ± 0.27 | 1.08 ± 0.19 |
OPA1 | 1.0 ± 0.18 | 0.49 ± 0.16* | 0.86 ± 0.36 | 1.09 ± 0.36 |
Data are fold changes ± SD; n = 3 for all groups. * p < 0.05 compared to CTRL-CMs and # p < 0.05 compared to SPTLC1-overexpression-CMs using two-way ANOVA Tukey’s multiple comparisons test. |
Overexpression of SPTLC1 in h-iPSCs-derived-CMs significantly increased levels of the DRP1 and MFF (Table 2). Furthermore, the auto/mitophagic protein levels of LC3B (2.6 ± 0.4, p < 0.001), as determined by the ratio of lipidated LC3-II (autophagic) to non-lipidated LC3-I (non-autophagic), (Fig. 3e and 3f) and PINK-1 (2.5 ± 0.7, p < 0.002) (Fig. 3e and 3 g) were elevated. Moreover, these cells showed decreased mRNA levels of the mitochondrial fusion MFN2 and OPA1 compared with CTRL-CMs (Table 2).
Incubation of SPTLC1-overexpression-CMs with myriocin had no effect on MFN2 and OPA1 expression compared to the myriocin-treated CTRL-CMs (Table 2). Parallel to this, our results showed a significant reduction in the DRP1 and MFF level in SPTLC1-overexpression-CMs treated with myriocin compared to the non-treated CMs (Table 2). Furthermore, decreased expression of LC3B (p < 0.001) (Fig. 3e and 3f) and PINK-1 (p < 0.002) (Fig. 3e and 3 g) were noticed in myriocin-treated SPTLC1-overexpressed-CMs compared to the non-treated SPTLC1-overexpressed-CMs, indicating decreased auto/mitophagy.
Mitochondrial function in SPTLC1-overexpression-CM
We hypothesized that ceramide is a crucial signaling component controlling mitochondrial function. OCR and ECAR were measured continuously throughout the experimental period at baseline and stressed condition using a Seahorse analyzer (Fig. 4a and 4b). We observed a reduction in FA ß-oxidation (Fig. 4a) and glucose oxidation (Fig. 4b). Furthermore, decreased mitochondrial basal respiration (18.8 ± 2.18 versus 8.0 ± 1.26, p < 0.01), respiratory capacity (31.87 ± 2.2 versus 10.32 ± 1.3, p < 0.01), and ATP production (15.41 ± 2.37 versus 6.84 ± 0.77, p < 0.01) (Fig. 4c and 4d) were observed, indicating an impaired mitochondrial function in the SPTLC1-overexpressed-CMs. Incubation of SPTLC1-overexpressed-CMs with myriocin improved FA ß-oxidation (Fig. 4a) and glucose oxidation (Fig. 4b) as well as ATP production (6.84 ± 0.77 versus 10.58 ± 1.51, p < 0.02) compared to non-treated SPTLC1-overexpressed-CMs (Fig. 4d).
In addition, protein levels of PGC-1α (0.47 ± 0.1, p < 0.009) (Fig. 4e and 4f) and AMP-activated protein kinase (p-AMPKα) (0.42 ± 0.12, p < 0.006) (Fig. 4g and 4 h) were decreased in SPTLC1-overexpressed-CMs. Interestingly, myriocin abolished the adverse effect of SPTLC1 overexpression in the CMs (Fig. 4e-h). This alteration in PGC-1α, a regulator of mitochondrial biogenesis, and p-AMPKα, a regulator of cellular energy and auto/mitophagy, confirmed our assumption that ceramide accumulation is involved in mitochondrial dysfunction.
Effects Of Ceramide Accumulation On Cardiac Insulin Signaling
Accumulation of ceramides in tissues leads to impairment of the Akt signaling by activating protein phosphatase 2A (PP2A) or protein kinase Cζ (PKCζ) [35, 36]. To further understand the signaling mediated by ceramide, the insulin-dependent phosphorylation of Akt and p-GLUT4 was measured after stimulating the CMs with insulin (100 nmol/L) [37]. The p-GLUT4 (1.92 ± 0.52, p < 0.001) and p-Akt/Akt (3.47 ± 0.17, p < 0.001) (Fig. 5a and 5b) were significantly blocked in stimulated SPTLC1-overexpressed-CMs with 100 nmol/L insulin for 10 min compared with CTRL–CMs. Interestingly, the levels of p-Akt/Akt (2.47 ± 0.56, p < 0.001) and p-GLUT4 (1. 83 ± 0.680, p < 0.03) were significantly increased by blockade of the serine SPT-specific inhibitor myriocin under the same condition (Fig. 5a and 5b). Together, the decreased phosphorylation of Akt (Ser 473) and p-GLUT4 in SPTLC1-overexpressed-CMs suggested a decreased insulin action.
Next, we analyzed whether inhibition of PP2A or PKC would interfere with the insulin-dependent phosphorylation of Akt. To this end, we used the PP2A and PKC inhibitors okadaic acid and BIM-1, respectively, in the presence or absence of insulin (Fig. 5c-f). Importantly, only the PP2A inhibitor significantly prevented the effect of SPTLC1 overexpression and the p-Akt increased significantly (2.08 ± 0.49, p < 0.006) compared to non-treated SPTLC1-CMs (Fig. 5c and 5d). Together, the western blot analysis revealed that activation PP2A in the SPTLC1-overexpressed-CMs resulted in the inability of insulin to activate the Akt.
Myriocin and okadaic acid improve cardiac structure and function in diabetic-CMs
We hypothesize that inhibition of ceramide accumulation or inactivation of PP2A results in improved diabetic-CMs functions. To prove this concept, CMs were cultured in DM either with myriocin or okadaic acid. Interestingly, we found that inhibition SPT with myriocin or PP2A with okadaic acid reduced diabetic-induced CMs intracellular lipid droplet accumulation (Fig. S2a and S2b) and improved cardiac contractility (9.3 ± 2.7 beats/min in DM versus 27.6 ± 7.4 beats/min in DM + myriocin, p < 0.001) or PP2A (9.3 ± 2.7 beats/min in DM versus 51.3 ± 4.7 beats/min in DM + okadaic, p < 0.001) (Fig. S2c).
Also, inhibition of SPT or PP2A as well as mitochondrial complex I significantly reduced the effect of DM-induced intracellular ceramide accumulation in the CMs versus control, as shown in Table 3. C16:00 and C18:00 ceramides were significantly reduced by SPT inhibitor myriocin or PP2A inhibitor okadaic acid. Only rotenone, the inhibitor of mitochondrial complex I, had no significant effect on reducing the C18:00 ceramide level in CMs co-incubated in the diabetic-like condition (Table 3). No significant change was noticed in C24:00 ceramide in h-iPSCs-derived-CMs treated with myriocin or okadaic acid under diabetic-like condition compared to the CMs in SM (Table 3). Furthermore, our data showed that inhibition of ceramide accumulation could prevent oxidative injury (Fig. S2d-f). Parallel to this result, we showed that inactivated PP2A, during exposing the CMs to diabetic-like condition, was accompanied by a significant increase in the Akt insulin-sensitivity and glucose oxidation related proteins, p-Akt (61.99 ± 6.93%, p < 0.001), p-AMPK (54.78 ± 4.62%, p < 0.001), p-GLUT4 (82.78 ± 23.17%, p < 0.01) and PPARα (71.16 ± 13.19%, p < 0.001) compared with DM (Fig. S2g-k). It was obvious that inhibition of the mitochondrial complex I did not significantly affect the levels of p-GLUT4, p-AMPK, and p-Akt following exposure to a diabetic-like condition for the same period (Fig. S2g-k).
Table 3
Alteration of the levels of different ceramide species in the h-iPSC-derived-CMs (nM/10 µl sample)
N-acyl chain | Standard medium (SM) | Diabetic medium (DM) | Myriocin (SM) | Myriocin (DM) | Okadaic acid (SM) | Okadaic acid (DM) | Rotenone (SM) | Rotenone (DM) |
C14:0 | 4.75 ± 2.35 | 9.25 ± 6.25 | 2.38 ± 0.521 | 3.09 ± 2.47 | 3.93 ± 2.92 | 3.51 ± 3.56 | 1.31 ± 0.24 | 1.34 ± 0.33 |
C16:0 | 179.92 ± 34.24 | 335.73 ± 36.40* | 126.13 ± 25.85 | 179.12 ± 27.88# | 157.9 ± 39.93 | 177.86 ± 26.96# | 142.23 ± 39.45 | 155.57 ± 41.74# |
C18:0 | 62.17 ± 6.17 | 89.27 ± 9.40* | 49.56 ± 5.81 | 55.03 ± 8.63# | 52.62 ± 7.60 | 66.02 ± 5.99# | 60.27 ± 5.94 | 76.5 ± 7.11 |
C20:0 | 15.92 ± 5.67 | 19.76 ± 4.22 | 14.81 ± 7.76 | 15.03 ± 8.92 | 16.26 ± 8.47 | 20.66 ± 10.48 | 15.09 ± 6.85 | 14.05 ± 8.10 |
C22:0 | 22.03 ± 11.01 | 26.88 ± 10.36 | 22.23 ± 10.91 | 23.62 ± 13.34 | 22.12 ± 13.32 | 25.48 ± 14.46 | 22.68 ± 10.96 | 21.42 ± 12.42 |
C24:0 | 20.97 ± 7.62 | 41.03 ± 9.10* | 19.28 ± 9.99 | 20.72 ± 10.16 | 19.94 ± 11.34 | 22.22 ± 11.86 | 20.02 ± 12.56 | 29.29 ± 15.58 |
Total | 305.76 ± 67.08 | 498.59 ± 83.11* | 234.41 ± 60.81 | 296.61 ± 71.43# | 272.76 ± 83.61 | 315.76 ± 73.34# | 261.61 ± 76.04 | 298.17 ± 85.31# |
Data are means ± SD; n = 3 for all groups. * p < 0.05 compared to SM-CMs and # p < 0.05 compared to DM-CMs using two-way ANOVA Tukey’s multiple comparisons test. |
Inhibition of ceramide accumulation rescues morphological and function in diabetic-CMs
In consistence with the results presented above, inhibition of the SPT and inactivate the PP2A improves mitochondrial function in diabetic-CMs.
It was found that both myriocin and okadaic acid decreased the level of auto/mitophagy markers such as LC3-II (-1.41 fold in myriocin, p < 0.001 and − 0.72 fold in okadaic acid, p < 0.02) and PINK-1 (-1.36 fold in myriocin, p < 0.001 and − 1.76 fold in okadaic acid, p < 0.001) in compared to untreated DM-CMs, as well as increased PGC-1α (+ 0.54 fold in myriocin, p < 0.04 and + 0.57 fold in okadaic acid, p < 0.004), the mitochondrial biogenesis regulator (Fig. 6a-d). To further confirm these findings, MitoTracker positivity and the ratio of LysoTracker and MitoTracker co-localized area were analyzed (Fig. 6e-h). Following myriocin or okadaic acid treatment, the DM-cultured cells showed a significant increase in mitochondria marker expression. Furthermore, a decrease of co-localization between the LysoTracker and MitoTracker area was observed, indicating a reduction in auto/mitophagy (Fig. 6h).
Interestingly, the toxic effect of ceramide on mitochondrial function was abolished in the presence of myriocin or okadaic acid under DM condition (Fig. 6i-l). Also, analysis of CM oxidative metabolism using a Seahorse analyzer showed increased mitochondrial activity and enhanced OCR and ECAR (Fig. 6i and 6j). Furthermore, it was obvious that PP2A inhibition abolished the effect of ceramide and increased ATP production under the basal and stressed conditions (Fig. 6k and 6 l). In conclusion, the data presented here indicate a critical role of PP2A in the mitochondrial dysfunction of lipotoxic cardiomyopathy.