Introduction

Light chain amyloidosis (AL) is a systemic proteotoxic disease defined by the formation and accumulation of immunoglobulin light chain (LC) amyloid fibrils within many target tissues causing their dysfunction. Cardiac involvement is frequent and constitutes the main prognosis factor1,2. Patients suffer mainly from restrictive cardiomyopathy leading to severe congestive heart failure, arrhythmias, and sudden death3,4.

Despite important progress in disease handling, therapeutical approaches for this type of cardiac amyloidosis are still insufficient and necessitate a better understanding of the molecular basis of pathogenetic events5,6,7. At the cellular level, amyloid fibrils and their soluble oligomeric precursors are deposited around cardiomyocytes inducing cytotoxicity and apoptosis [reviewed in8. Most experimental evidence originates from in vitro and ex vivo studies using isolated adult rodent cardiomyocytes and hearts, and cardiomyocyte-like cell lines. In vivo studies were performed with the simple animal models of C. elegans and zebrafish. A general observation was that LC treatment causes contractile dysfunction, including reduced contractility with concomitant reduced calcium transient amplitude and prolonged relaxation in isolated cardiomyocytes9,10,11,12, impaired systolic function and reduced cardiac output in zebrafish11,12,13,14,15, and reduced pumping rate in C. elegans16. Transient perfusion of ex vivo isolated mouse heart with patient-derived LCs resulted in a progressive increase in end-diastolic pressure, closely resembling severe diastolic dysfunction in hearts of AL cardiac amyloidosis (AL-CA) patients17. The latter experiment highlights the necessity to pursue studies with more complex experimental models. Unfortunately, the obtention of an appropriate higher-order animal model able to reproduce the main characteristics of the disease in vivo has proved challenging18. Besides this difficulty, it is known that human cardiac physiology differs from rodents, rendering necessary the use of a human cardiomyocyte-based experimental system to further advance our knowledge on this disease19,20.

Currently, protocols allowing obtention of cardiomyocytes from human induced pluripotent stem cells (hiPSC-CMs) are being successfully established21. These human cardiomyocytes have been widely proposed as highly promising in vitro disease models. A major concern, however, is the highly variable degree of hiPSC-CM maturity. To address this problem, more recently, microtissue-like cardiac spheroids were generated by admixing hiPSC-CMs with other cardiac cells, especially fibroblasts22. In this three-dimensional model, multicellular crosstalk promotes maturation of hiPSC-CMs by enhancing cardiomyocyte structure, contractility and mitochondrial respiration. In fact, spheroids can better reproduce the tissue architecture and cellular interactions in heart, enabling cardiomyocytes to develop morphological and functional characteristics closer to those observed in vivo. These improvements include the development of communicating junctions, the alignment of myofibrils, and a better integration of mechanical and electrical stimuli, more closely recapitulating the conditions of the beating heart.

In this study, we have set up a human cellular disease model mimicking AL-CA, to study the cytotoxic impact of free amyloidogenic LCs. Here we report changes in cell structure, contractility and calcium handling induced by patient-derived LCs in a 3D human cardiac spheroid model based on hiPSC-CM and human primary cardiac fibroblasts.

Results

Purification, cytotoxicity, and structural impact of amyloidogenic patient-derived light chains on human cardiomyocytes

Our aim was to observe the impact of amyloidogenic light chains (LCs) on human cardiomyocytes derived from induced pluripotent stem cells (hiPSC-CMs). We first purified LCs from urine of three patients (Pa1, Pa2, Pa3) suffering from AL-CA, using preparative size-exclusion chromatography (Fig. 1; see Materials and Methods). Basic characteristics of patients are described in Supplementary Table S1. All three patients’ hearts were highly affected (European stage III for Pa1 and Pa3; stage II for Pa2), although the severity of the disease differed among patients, as indicated by NT-proBNP and TnT (TnT-HS) serum levels and functional evaluations. Isolated LCs were recovered from eluted samples corresponding to the absorbance peak eluted at an apparent molecular mass near 44 kDa and confirmed by western blot on the collected fractions (Fig. 1A, Supplementary Fig. S1). Only a subset of fractions was retained for experimental use in order to minimize frequent contamination with serum albumin often present in patients’ urine because of concomitant kidney damage (Supplementary Fig. S2). SDS-PAGE under nonreducing conditions and western blot analysis of recovered samples indicated the presence of both LC monomers and dimers (Fig. 1B,C). Purified LCs were named Pa1-λ, Pa2-κ and Pa3-κ after patient number and their lambda (λ) or kappa (κ) subtypes, and were used to treat cardiomyocytes at a monomer concentration of 100 µg/ml. This concentration lies within the midrange of pathological serum concentrations observed in patients with AL cardiomyopathy23, and was used in this study throughout experiments.

Fig. 1
figure 1

Purification and characterization of amyloidogenic free light chains (LCs) from patients’ urine. (A) Size exclusion chromatograms showing LC elution profiles. Peaks corresponding to eluted fractions selected for LC purification are indicated by arrows. (B, C) PAGE analysis of purified LCs under reduced (R) and non-reduced (NR) conditions after concentration by Amicon ultrafiltration: (B) Ponceau S staining; (C) western blot analysis with antibodies against human lambda and kappa LC. Purified LCs appear as monomers (25–30 kDa) and dimers (50–55 kDa). Full length images of nitrocellulose membranes are shown in Supplementary Fig. S1.

Full size image

Two human induced pluripotent stem cell (hiPSC) lines were derived from nucleated blood cells of healthy donors and used to generate cardiomyocytes (hiPSC-CMs) according to the biphasic activation/inhibition of the Wnt pathway differentiation protocol24. Cardiomyocytes were then treated with the purified amyloidogenic LCs for 3 days before analysis (Fig. 2A). We first evaluated the cytotoxic capacity of purified LCs using the MTT cytotoxicity assay in 2D cultures of cardiomyocytes treated with LCs (Fig. 2B). In this test, cell viability is evaluated by changes in metabolic activity, and is inversely proportional to cytotoxicity. Indeed, we observed a trend of increase in cardiomyocyte cytotoxicity after a 7 day-exposure to all tested LCs, when compared with non-treated cells. Increase in cytotoxicity varied among the 3 LCs (22.0%±2.6%, 29.0%±0.8% and 44.7%±13.0% for Pa1-λ, Pa2-κ and Pa3-κ, respectively) and was statistically significant for Pa3-κ.

Fig. 2
figure 2

Patient-derived LCs exert toxicity and perturb cellular structure in human iPSC-derived cardiomyocytes (hiSPC-CMs). (A) Schematic protocol of hiPSC-CM treatment with purified LCs. (B) Increased cytotoxicity measured by the MTT assay in LC-treated, as compared to non-treated (control), hiPSC-CM cultures. Control, n = 4; Pa1-λ, n = 3; Pa2-κ, n = 2; Pa3-κ, n = 3 independent experiments. (C) Quantification of cardiac alpha-actinin levels based on immunofluorescence labeling of LC-treated, as compared to non-treated (control), hiPSC-CM cultures. Field views: control, n = 11; Pa1-λ, n = 5; Pa2-κ, n = 7; Pa3-κ, n = 4. (D) Representative images of cardiac alpha-actinin immunolocalization in LC-treated and non-treated (control) hiPSC-CM cultures. (Lower panel), magnified areas indicated by dashed rectangles in upper panel. Bars, 50 μm. Quantitative results are expressed as mean values ± SEM. *p < 0.05; **p < 0.01.

Full size image

We next investigated the impact of amyloidogenic LCs on the structure and function of cardiomyocytes. Previous in vitro studies showed that the contractile capacity of adult rat cardiomyocytes was compromised by the presence of amyloidogenic LCs9,10,11,12. We therefore turned our attention to contractile proteins and the contractile function of LC-treated cardiomyocytes. We first analyzed by immunostaining the subcellular distribution of the cardiac isoform of alpha-actinin, a key protein for the organization of the cardiomyocyte contractile apparatus (Fig. 2C,D). The observed disruption of alpha-actinin striated pattern with Pa2-κ and Pa3-κ LCs (Fig. 2D, higher magnification insets) strongly suggested the disorganization of the entire sarcomeric network. Moreover, quantification of immunofluorescence staining (Fig. 2C) showed an important and statistically significant decrease in protein levels of alpha-actinin for Pa2-κ (34.0%±9.2%, p ≤ 0.05) and Pa3-κ (55.2%±6.1%, p ≤ 0.05). Cardiomyocytes treated with Pa1-λ LCs were clearly less affected, showing a slight alpha-actinin decrease (10.9%±3.4%, p > 0.05). It is noteworthy that, the greater the toxicity of one LC (Fig. 2B), the greater its impact on contractile proteins (Fig. 2C), suggesting a correlation between LC cytotoxicity and its impact on cardiomyocyte structural organization.

Establishment of a human cardiac spheroid AL-CA disease model to study contractility and calcium handling

Subsequently, in order to analyze contractile function, we decided to work with a more robust cellular model displaying tissue-like properties. We adopted the organoid model, a 3D cell culture system resulting from self-assembly of different tissue-specific cell types, also known as spheroid. We obtained 10,000-cell spheroids by admixing hiPSC-CMs with human primary cardiac fibroblasts in a ratio 8.5:1.5, according to the protocol shown in Fig. 3A. LCs were added 24 h before and during spheroid formation, as well as in all medium changes, to ensure that amyloidogenic LCs integrated the entire structure. Before functional analysis, we first checked whether LCs had an impact on overall spheroid formation, because differences in shape and size could affect later measured parameters. Morphological analysis of spheroids in terms of area (Fig. 3B), perimeter (Fig. 3C) and circularity (Fig. 3D) showed no differences across all conditions, meaning that spheroid compaction was not affected by the presence of amyloidogenic LCs. We also confirmed by immunostaining the uniform distribution of LCs inside spheroids (Fig. 3E). Immunostaining was carried out on sections of frozen spheroids, to make sure that labeled LCs are indeed present inside spheroids and in direct contact with cells. Therefore, any functional impact of amyloidogenic LCs are due to their effects on cells and not to other reasons, like poor spheroid formation.

Fig. 3
figure 3

Establishment of a 3D human cardiac spheroid model of AL amyloidosis. (A) Schematic protocol of human cardiac spheroid formation and purified LC treatment. (B–D) Quantification of area, perimeter and circularity of spheroids. Analyzed spheroids: control, n = 45; Pa1-λ, n = 44; Pa2-κ, n = 30; Pa3-κ, n = 30. Results are expressed as mean values ± SEM. (E) Immunofluorescence localization of purified LCs on spheroid cryosections. LC staining accumulates between cardiomyocytes when spheroids form in the presence of LCs, but not in control (no LCs during spheroid formation). Bars, 50 μm.

Full size image

To analyze LC impact on contractility, we first determined the percentage of beating spheroids in each condition (Fig. 4A). In the absence of LCs (non-LC treated controls), about 80% (79.2%±12.5%) of electrically paced at 1 Hz spheroids were able to contract, versus only about 50% (53.1%±16.4%), 40% (39.6%±20.2%) and 35% (35.7%±18.0%) of spheroids when treated with Pa1-λ, Pa2-κ and Pa3-κ, respectively. Therefore, contractility can be blocked by the presence of these different amyloidogenic LCs. As expected, the extent of this blocking correlates with the degree of disruption of alpha-actinin network (Fig. 2C,D). Next, because contractility is directly linked to the cytosolic calcium oscillations, we analyzed calcium transients in spheroids paced at 1 Hz using a non-ratiometric calcium probe. Muscle contraction and relaxation are mediated by the alternating inflow and outflow of calcium into the cell, mainly due to Ryanodine Receptors (RyR) and Sarcoplasmic-Endoplasmic Reticulum Calcium ATPase (SERCA) proteins, respectively25. Measurement parameters included the total duration of calcium transients (CD), the time to reach the peak intracellular calcium concentration (time to peak; Ton) and the time for calcium concentration to fall back to the baseline (time to decay; Toff), as defined in Fig. 4B. CD remained unchanged in all conditions (Fig. 4C), whereas treatment with LCs resulted in statistically significant decrease of Ton (control, 170ms; Pa1-λ, 129ms; Pa2-κ, 119ms; Pa3-κ, 127ms) (Fig. 4D) and a trend of increasing Toff (control, 538ms; Pa1-λ, 586ms; Pa2-κ, 560; Pa3-κ, 552ms) (Fig. 4E). These results suggest that cytosolic calcium increase occurs more rapidly in the presence of amyloidogenic LCs, with concomitant slower calcium reuptake by SERCA into the sarcoplasmic reticulum. The activity of the SERCA pump in the sarcoplasmic reticulum could be one important factor influencing this modulation of calcium entry and exit from the cytoplasm. We therefore examined the subcellular pattern of this protein by immunostaining of the cardiac isoform SERCA2a on spheroid frozen sections (Fig. 4F). Interestingly, we observed a change in the distribution of SERCA2a inside several cardiomyocytes of LC-treated spheroids, as compared to controls. SERCA2a displayed a non-uniform accumulation and appeared to be highly compacted (Fig. 4F, higher magnification insets), resembling aggregated protein as seen in other experimental conditions26. These observations applied to all tested LCs.

Fig. 4
figure 4

Patient-derived LCs perturb contractile function and calcium transients of human cardiac spheroids. (A) Percentage of LC-containing beating spheroids during 5 min pace at 1 Hz. Control, n = 4; Pa1-λ, n = 4; Pa2-κ, n = 4; Pa3-κ, n = 3 independent experiments. (B) Graphic definition and (C–E) measurements of calcium transient parameters of beating LC-containing and control spheroids; CD, calcium duration; Ton, time to peak; Toff, time to decay. Analyzed spheroids: control, n = 23; Pa1-λ, n = 17; Pa2-κ, n = 15; Pa3-κ, n = 15, from at least 3 independent experiments. Results are expressed as mean values ± SEM. *, p < 0.05; **, p < 0.01; ****, p < 0.0001. (F) Representative images of SERCA2a immunolocalization on LC-containing and control spheroid cryosections. (Lower panel), magnified areas indicated by dashed rectangles in upper panel. Bars, 50 μm.

Full size image

Discussion

In this work, we have created a human cellular model for AL-CA in the form of cardiac spheroids based on human iPSC-derived cardiomyocytes, and used it to investigate the impact of patient-issued amyloidogenic LCs on cardiac contractility and calcium handling. Tissue spheroids evolved from a continuous effort to obtain in vitro models able to better reproduce and predict biological responses of living organisms. Human cardiac spheroids have been successfully used before to measure contractility22. Here, to better mimic cardiac amyloidosis, we purified soluble LCs from patients’ urine, introduced them during the formation of spheroids, and subsequently confirmed by immunostaining their uniform accumulation among cells. We show that spheroids treated with amyloidogenic LCs display impaired contractility and abnormal calcium cycling. Impaired contractility was observed in at least half of LC-treated spheroids (53–65% versus 20% in untreated controls), which were totally unable to contract. In support of this observation, the disrupted subcellular pattern and reduced protein levels of alpha-actinin suggested that myofibrils get disorganized in the presence of LCs. Calcium transients were also affected, showing a faster time to peak values as a main effect. This result may be interpreted in different ways, including a reduced resting cytosolic free calcium concentration, shown by a lower peak of calcium transients in LC-treated cells. Unfortunately, the use of a non-ratiometric calcium probe did not allow us to measure the amplitude of calcium transients. However, the unchanged total duration of transients indicates that an alternative explanation is not excluded. A higher resting calcium concentration, possibly related to a defective calcium uptake by sarcoplasmic reticulum protein SERCA2a, would also result in a faster time to peak during contraction. SERCA2a is particularly interesting in the context of cardiomyopathies27,28. Decreased SERCA2a expression and/or activity has been reported in the hearts of patients with dilated and ischemic cardiomyopathies; enhancing or restoring SERCA2a activity improved cardiac function and prevented the progression of heart failure in various animal models; cardiac-specific conditional SERCA2a knockout mice exhibited severe diastolic dysfunction and developed end-stage heart failure within 7 weeks [reviewed in28]. Here we report for the first time that amyloidogenic LCs could have an effect on SERCA2a protein in human cardiomyocytes. We noticed that in the presence of LCs, intracellular SERCA2a staining often appeared particularly dense, reminiscent of aggregated SERCA2 proteins as observed in other diseases26. Protein aggregation could be responsible for disrupting calcium flux by reducing SERCA2a pumping activity, leading to abnormal accumulation of intracellular resting calcium. Other factors can also affect intracellular calcium cycling in the presence of amyloidogenic LCs. Post-translational modifications are known to impact RyR2 and SERCA2a activity28,29. In our disease model, of particular interest are oxidative protein modifications due to elevated intracellular levels of ROS, one of the hallmarks of cytotoxicity caused by LCs9,11,12,30. Direct oxidation of SERCA2a at the cytosolic cysteine 674 decreases calcium pump activity in cardiomyocytes [reviewed in31]; conversely, reversible redox modifications increase RyR2 activity by increasing the channel’s sensitivity to luminal SR calcium. Enhanced activity of oxidized RyR2 would also be consistent with a faster time to calcium peak values in paced LC-treated spheroids and could not be excluded. We may hypothesize that the combination of enhanced RyR2 function and reduced SERCA2a activity can cause in the long-term depletion of calcium SR stores, leading to diminished calcium release during systole and finally to impaired contractility. The results of the present study suggest that these conditions may apply in the presence of amyloidogenic LCs. Our disease model would be adequate to address these hypotheses in future studies.

Finally, it is important to point out that the definite attribution of observed effects to isolated amyloidogenic LCs requires comparison with patient-derived non amyloidogenic LCs (e.g., multiple myeloma), which were not accessible during this work. Future studies will address this issue. In favor of a direct role for LCs, impairment of contractility and deregulation of calcium cycling in cardiomyocytes were previously attributed to isolated amyloidogenic LCs in various experimental models9,10,11,12,13,14,15,16. We also noticed that cardiomyocytes treated with LCs from three different patients displayed similar defects, albeit with varying degrees of severity when compared to untreated control, at least for certain parameters. Such variations would be consistent with the patient-specific features of amyloidogenic LCs and the varying clinical manifestations of this disease32. In particular, cells treated with Pa1 LCs suffered milder damage. This was surprising because of patient 1’s severe stage IIIa cardiomyopathy. Interestingly, this patient had a delay time between first symptoms and diagnosis 4 times longer than the other two patients. One explanation may be that Pa1 LCs had a milder cytotoxic activity, in agreement with our results, and therefore a longer period of time would be required for manifestation of cardiac pathogenic effects related to amyloidosis.

In conclusion, we report that the presence of amyloidogenic patient-derived LCs affected human cardiomyocytes with disorganization of myofibrils, impaired contractility and perturbation of calcium handling during calcium transients. Human cardiac spheroid models for AL-CA can be a valuable tool for further dissecting the pathogenetic mechanisms of the disease. Furthermore, they could serve to develop in vitro systems suitable to predict cardiotoxicity of different amyloidogenic LCs, and to test therapeutic treatments.

Materials and methods

Light chain purification

Light chains (LC) were obtained from 24 h patients’ urine (French Referral Centre for Cardiac Amyloidosis, Henri-Mondor Hospital, AP-HP), conserved at -20 °C until purification. All anonymized samples were obtained with patients’ informed consent. The study complied with the Declaration of Helsinki and was approved by the Committee for the protection of persons (CPP) Tours Ouest 1 (Protocol CNRIPH n°21.03.30.40253). LCs were purified by size exclusion chromatography according to previously reported procedures17,33 with modifications. Briefly, 50 ml of urine were precipitated in 65% ammonium sulfate overnight at 4 °C, and centrifuged for 15 min at 3,900 g. Pellets were resuspended in 10 mM HEPES-100 mM NaCl buffer, pH 7.4, and dialyzed against HEPES-NaCl buffer for 24 h at 4 °C. LC-containing samples were recovered from Superdex200 26/600 size exclusion chromatography on Äkta Go system (Cytiva) and concentrated in 50 mM phosphate buffer, pH 7.2, by 6 rounds of dilution/centrifugation at 3,900 g using Amicon Ultra-15 devices (Sigma, Ultracel-10 K). Purity of isolated LCs was analyzed by western blot.

Human induced pluripotent stem cell (hiPSC) generation and culture

The two hiPSC lines (CW30318C and CX30356C) used in this study were reprogrammed from blood cells of healthy donors with no cardiac pathology and were purchased from Cellular Dynamics International. Cells were grown and amplified on Matrigel-coated culture dishes with daily replacement of mTeSR + medium (StemCell Technologies) and passaged (ratio 1:30) when confluent at around 80% using ReLeSR (StemCell Technologies).

Generation of human cardiomyocytes

Three cardiomyocyte batches were generated from the two hiPSC lines CW30318C (batch LS74) and CX30356C (batches OX58 and OX60) according to the biphasic activation/inhibition of the Wnt pathway differentiation protocol24. Briefly, at 80% confluence, hiPSCs were detached with accutase, counted, then seeded at optimal density. After two days in culture, cardiac differentiation was initiated by replacing mTeSR + medium with RPMI medium (Life Technologies) enriched with 2% insulin-free B27 supplement (ThermoFisher Scientific) containing 9µM CHIR99021 (StemCell Technologies) to activate the Wnt pathway. This medium was replaced by insulin-free RPMI + B27 after 24 h. Two days later, cells were treated with insulin-free RPMI + B27 containing 5µM IWP2 (Tocris Bioscience) for 48 h to inhibit the Wnt pathway, then the medium was replaced by insulin-free RPMI + B27. After two days, the culture medium was replaced by RPMI enriched with a 2% supplement of B27 with insulin and changed every two days. After 8 days of differentiation, the cells contracted spontaneously. On Day 12, cardiomyocytes were detached with TrypLE Select 10X (Life Technologies) (incubation at 37 °C for 10 min), suspended in RPMI containing 20% fetal bovine serum (Thermo Fisher Scientific), filtered through a 70 μm cell strainer and frozen in Cryomedium PSC (Life Technologies) to create a homogeneous cardiomyocyte bank. The three batches of differentiated cells had a similar cardiomyocyte content exceeding 80% and did not display statistically significant differences among them for time to peak of calcium transients, a parameter which changes after LC treatment (Supplementary Fig. S3). For experiments, thawed cardiomyocytes were cultured for one week before treatment. Independent experiments were defined by the use of a newly thawed batch of cells originating each time from one differentiation. All three batches were used through experiments. Untreated cells were used as controls and the results from independent experiments were pooled.

Spheroids

Spheroids were formed from the combination of hiPSC-derived cardiomyocytes (hiPSC-CM) and human cardiac fibroblasts (NHCF-V, Lonza, Cat#: CC-2904) as previously described34, according to the protocol shown in Fig. 3A. Briefly, on Day 1, hiPSC-CMs were thawed and plated on a Matrigel-coated plate (Corning) in “complete” RPMI medium containing RPMI medium (Gibco) supplemented with B-27 supplement, KO serum, ROCK inhibitor (5µM) and 1% P/S. On Day 2, KO serum and ROCK inhibitor were omitted from medium. On Day 7, hiPSC-CMs were treated or not with 100 µg/ml purified LCs, and fibroblasts were thawed in DMEM + Glutamax medium (4.5 g/L glucose, with pyruvate), 10% SVF, 1% P/S. On Day 8, all cells were detached with TrypLE 10X, counted, and seeded in a 96-well round-bottom ultra-low attachment plate in a ratio of 85% hiPSC-CM (8500 cells) and 15% fibroblasts (1500 cells) in 100 µl of complete RPMI medium per well, with 100 µg/ml LCs added. Next day, 50 µl of complete medium was added, then changed after 4 days with addition of 100 µg/ml LCs. The cells self-aggregated to form a spheroid within 7 days. On Day 15, spheroids were either frozen for immunofluorescence experiments or analyzed for contractility and calcium transients with an inverted Leica DMI8 microscope (Leica) equipped with environmental control chamber adjusted to 37 °C and 5% CO2, as follows. Spheroids were incubated for 1 h with 1 mM Fluo-4AM probe (Life Technologies), then placed in a drop of Matrigel under a flow of RPMI + B27, 1% P/S medium, and paced at 1 Hz. Video-based measurements of contractility and calcium transients were analyzed after 5 min of stimulation using ImageJ and Matlab software respectively.

MTT test

Light chain (LC) toxicity was tested on hIPSC-derived cardiomyocytes using an MTT cell viability assay. Cells were seeded in 96-well plates at a density of 30,000 cells/cm2. After 24 h in culture, the cells were treated with the various LCs at a concentration of 100 µg/ml, for a period of 7 days. The medium was changed at Day 2 and Day 5. The MTT viability assay (ab211091, Abcam) was performed as follows. Culture medium was replaced by medium RPMI + B27, 1% P/S, without LCs and with the addition of MTT reagent. After incubation for 3 h, the solvent was added for 15 min with agitation, to release the reaction product. The reduction of MTT tetrazolium salt (3-4(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide) to formazan by mitochondrial succinate dehydrogenase can only take place in living cells with viable mitochondria.

Immunofluorescence

hiPSC-CMs in culture were fixed in 4% paraformaldehyde solution for 15 min at room temperature. After washing with PBS, cells were permeabilized with 0.1% Triton-PBS buffer for 10 min saturated with 5% BSA-PBS buffer for 1 h at room temperature and incubated with primary antibodies against alpha-actinin (A7811, Sigma) and human immunoglobulin light chain subtypes kappa (ab134929, Abcam) and lambda (ab124719, Abcam). Spheroids were fixed in 4% paraformaldehyde solution for 1 h at room temperature. After washing with PBS, spheroids were subjected to successive sucrose baths, put in OCT and frozen in liquid nitrogen-cooled isopentane. Spheroid sections were cut to a thickness of 5 μm using a cryostat. Sections were permeabilized with 0.1% Triton-PBS buffer for 10 min, saturated with 5% BSA-PBS buffer for 1 h at room temperature, and incubated with primary antibodies against alpha-actinin (A7811, Sigma), SERCA2a (sc-376235, Santa Cruz) and human immunoglobulin light chain subtypes kappa (ab134929) and lambda (ab124719). Immunofluorescence images were taken on a Leica DMi8 fluorescence microscope with a 40X objective, and treated with Leica Thunder software for computational clearing. Fluorescence quantification was carried out using ImageJ software. A threshold was applied to quantify grey levels. Alpha-actinin levels (Fig. 2C) were calculated according to the formula:

$$ (mean~ gray~ value) \times (\% labeled ~area) / number ~of ~nuclei.$$

Results were presented as fold changes as compared to control.

Statistical analysis

All experimental data are presented as mean ± standard error of the mean (SEM). Normality was checked using the Shapiro-Wilk test and, if necessary, non-parametric tests were used. When multiple comparisons were required, a one-way ANOVA test combined with Dunnett’s multiple comparison tests was used if normality was respected, otherwise a Kruskal-Wallis test combined with Dunn’s multiple comparison tests was used. p values of less than 0.05 (*), 0.01 (**), 0.001 (***) and 0.0001 (****) were considered statistically significant. Data were analyzed and presented using GraphPad Prism 10.