Hypoxia and TNF‐alpha modulate extracellular vesicle release from human induced pluripotent stem cell‐derived cardiomyocytes Authors: Margarida Viola, Maarten P. Bebelman, Renee G. C. Maas, Willemijn S. de Voogt, Frederik J. Verweij, Cor S. Seinen, Saskia C. A. de Jager, Pieter Vader, Dirk Michiel Pegtel, Joost Petrus Gerardus Sluijter Published: 2024 ABSTRACT AbstractExtracellular vesicles (EVs) have emerged as important mediators of intercellular communication in the heart under homeostatic and pathological conditions, such as myocardial infarction (MI). However, the basic mechanisms driving cardiomyocyte‐derived EV (CM‐EV) production following stress are poorly understood. In this study, we generated human induced pluripotent stem cell‐derived cardiomyocytes (hiPSC‐CMs) that express NanoLuc‐tetraspanin reporters. These modified hiPSC‐CMs allow for quantification of tetraspanin‐positive CM‐EV secretion from small numbers of cells without the need for time‐consuming EV isolation techniques. We subjected these cells to a panel of small molecules to study their effect on CM‐EV biogenesis and secretion under basal and stress‐associated conditions. We observed that EV biogenesis is context‐dependent in hiPSC‐CMs. Nutrient starvation decreases CM‐EV secretion while hypoxia increases the production of CM‐EVs in a nSmase2‐dependent manner. Moreover, the inflammatory cytokine TNF‐α increased CM‐EV secretion through a process involving NLRP3 inflammasome activation and mTOR signalling. Here, we detailed for the first time the regulatory mechanisms of EV biogenesis in hiPSC‐CMs upon MI‐associated stressors. INTRODUCTION Myocardial infarction (MI) is the leading cause of death worldwide (Ahmad et al., 2023). Despite current treatment options (Elgendy et al., 2019), heart failure incidence post‐MI is expected to increase in the coming years (Foreman et al., 2018). It is evident that there is an unmet need for new approaches to promote cardiac repair. Extracellular vesicles (EVs) released by the ischemic heart tissue have recently been reported to drive cardiac repair following acute MI and chronic ischemic remodelling (Almeida Paiva et al., 2019; Cheng et al., 2019; Gan et al., 2022; Li et al., 2021; Loyer et al., 2018; Ribeiro‐Rodrigues et al., 2017). EVs are membrane particles that mediate intercellular communication by transporting biomolecules, such as proteins and small non‐coding RNA, between cells (Kalluri & LeBleu, 2020). Cardiomyocyte‐derived EVs (CM‐EVs) influence cardiac repair by controlling fibrosis (Li et al., 2021), promoting angiogenesis (Deddens et al., 2017; Ribeiro‐Rodrigues et al., 2017), stimulating macrophages (Almeida Paiva et al., 2019), and regulating endocrine dysfunction in adipocytes (Gan et al., 2022). After MI, circulating EVs mobilize bone marrow progenitor cells (Cheng et al., 2019) and modulate the inflammatory response (Loyer et al., 2018). Finally, CM‐EVs released by human induced pluripotent stem cell‐derived cardiomyocytes (hiPSC‐CMs) have therapeutic potential for treating MI. Preclinical studies in rodents showed that these EVs improved heart function when injected directly into infarcted hearts (Maring et al., 2019; Santoso et al., 2020; Tominaga et al., 2023) or delivered in a hydrogel (Liu et al., 2018; Wang et al., 2023). EVs are a heterogeneous group of vesicles of different sizes, ranging from 50 nm to several micrometer in size, which are produced through various biogenesis mechanisms. They can originate intracellularly from the fusion of multivesicular bodies (MVBs) with the plasma membrane, called exosomes, or be generated by outward budding of the plasma membrane as ectosomes (Van Niel et al., 2022). The regulation of EV secretion is highly cell‐type and context‐specific, with microenvironmental cues having strong influences on EV type, quantity and cargo (Dixson et al., 2023). MI has been reported to increase cardiac EV release (Anselmo et al., 2021; Loyer et al., 2018) and change EV cargo, such as microRNA content (Gan et al., 2022; Li et al., 2021; Ribeiro‐Rodrigues et al., 2017). However, the regulatory mechanisms and EV biogenesis pathways that underlie these changes in CM‐EV production following MI are poorly understood. This lack of understanding hampers the potential development of therapeutic strategies aimed at modulating CM‐EV production to improve cardiac repair following MI and the design of more effective CM‐EV therapies (de Abreu et al., 2020). The current gap in our understanding of CM‐EV secretion is a consequence of the technical difficulties associated with mechanistic EV studies. (Van Niel et al., 2022) The initial step of most EV research typically involves the isolation of EVs using techniques such as differential ultracentrifugation (dUC) and size exclusion chromatography (SEC), but these require a large starting volume of conditioned media and number of cells, are time‐consuming, and can lead to co‐isolation of protein aggregates (Mol et al., 2017; Sluijter et al., 2017). Several strategies have been developed in the last years to study EV production from small cell numbers without the need for EV isolation. Reporters based on tetraspanins and other EV cargo proteins fused to the bright luciferases NanoLuc (Hall et al., 2012) and ThermoLuc (Koksharov & Ugarova, 2011) have been used to study the mechanisms underlying EV biogenesis directly (Bebelman et al., 2023; Hikita et al., 2018), to identify novel EV modulator drugs (Grisard et al., 2022), and to optimize culture conditions for the production of therapeutic EVs (Bost et al., 2022). Importantly, most CM‐EV secretion studies are currently based on immortalized cell lines (Almeida Paiva et al., 2019; Borosch et al., 2017; Garcia et al., 2015; Gennebäck et al., 2013; Kenneweg et al., 2019; Martins‐Marques et al., 2020; Ribeiro‐Rodrigues et al., 2017) and primary culture (Borosch et al., 2017; Datta et al., 2017; Gan et al., 2022; Garcia et al., 2015; Li et al., 2021; Ottaviani et al., 2022; Wang et al., 2016) of CMs of rodent origin, which are more available, but do not reflect human cardiomyocyte biology. hiPSC‐CMs offer this and several other advantages (Karakikes et al., 2015), including their compatibility with genetic interventions for more mechanistic studies. In this study, we aimed to understand the regulatory mechanisms involved in CM‐EV biogenesis and secretion under basal culture conditions and in combination with known post‐MI stressors, such as nutrient and oxygen deprivation, and exposure to pro‐inflammatory stimuli. To achieve this, we generated hiPSC‐CMs expressing NanoLuc‐tetraspanin reporters, which enabled the study of CM‐EV biogenesis directly from the conditioned media. We identified several pathways involved in tetraspanin‐positive CM‐EV biogenesis during basal culture conditions using small molecule inhibitors. Furthermore, we found that conditions associated with MI modulate CM‐EV release and discovered a role for neutral sphingomyelinase 2 (nSMase2) in hypoxia‐driven EV secretion and a potential role for the NLRP3 inflammasome and mTOR signalling in TNF‐α‐induced EV release. RESULTS hiPSC‐derived cardiomyocytes (hiPSC‐CMs) secrete small extracellular vesicles (sEVs) hiPSC‐CMs are increasingly recognized as a relevant in vitro model to study human cardiomyocyte biology. In addition, these cells release EVs (Anselmo et al., 2021; Gao et al., 2020; Ikeda et al., 2021; Jung et al., 2021; Liu et al., 2018; Louro et al., 2022; Santoso et al., 2020; Schoger et al., 2023; Tominaga et al., 2023; Vaskova et al., 2020; Zwi‐Dantsis et al., 2020), although the variations in the heterogeneity of these EVs remain unclear. The protocol applied here (Maas et al., 2021) enabled the generation of a pure population of beating cardiomyocytes (Figure S1, Video S1). To subsequently investigate if the hiPSC‐CMs generated secrete small EVs (sEVs), we isolated EVs from the conditioned media using dUC and analyzed the EV pellet following 70 min centrifugation at 100K ×g (100K pellet), according to MISEV2018 guidelines (Théry et al., 2018). Western blot confirmed the enrichment relative to cell lysates of classical sEV markers, such as the tetraspanins CD9, CD63 and CD81 (Kowal et al., 2016), syntenin (Baietti et al., 2012), as well as the presence of ectosome marker Annexin A1 (Jeppesen et al., 2019). Depletion of GAPDH, the endoplasmic reticulum marker Calnexin, and the Golgi marker GM130 demonstrated the purity of the isolated sEVs (Figure 1a). Nanoparticle tracking analysis (NTA) (Figure 1b) and transmission electron microscopy (Figure 1c) of the 100K pellet revealed the presence of membrane vesicles with a diameter ranging from 50 to 200 nm and a mean/mode of approximately 90 nm, as expected for sEVs (Van Niel et al., 2018). The contribution of the culture media to the EV measurements is negligible (Figure S2). Overall, these data confirm that the hiPSC‐CMs being used in this research release sEVs. hiPSC‐derived cardiomyocytes (hiPSC‐CMs) secrete small extracellular vesicles (sEVs). (a) Western blot analysis of sEVs isolated by differential ultracentrifugation from the conditioned media of hiPSC‐CMs for common EV markers (CD9, CD63, CD81, Syntenin and Annexin A1) and cell‐enriched markers (GAPDH, GM130 and Calnexin). The arrows highlight the bands that matched the predicted molecular weight. (b) Size distribution profile of sEVs analyzed by nanoparticle tracking analysis (NTA). Results are plotted as mean ± SEM of three independent sEV isolations. (c) Representative transmission electron microscopy (TEM) image of sEVs (scale bar = 200 nm). sEVs were isolated from conditioned media harvested from 12 to 17 million hiPSC‐CMs in (a–c). NanoLuc‐tetraspanin reporters enable direct quantification of sEV secretion from hiPSC‐CMs To study CM‐EV secretion directly without the need for EV isolation, we generated hiPSC‐CMs stably expressing the tetraspanins CD9, CD63 and CD81 tagged at their N‐termini with NanoLuc and a hemagglutinin (HA)‐tag to facilitate antibody‐based detection (Figure 2a). Transduction efficiency and location of the HA‐NanoLuc‐(NL)‐tetraspanin reporters were verified by immunofluorescence microscopy (Figure 2b). HA‐NL‐CD9 and HA‐NL‐CD81 were localized to the plasma membrane and endosomal compartments, whereas HA‐NL‐CD63 was mostly present in endosomal compartments. This is consistent with their endogenous localization (Figure S3). Importantly, western blot analysis of the 100K pellet confirmed that HA‐NL‐tetraspanin reporters were secreted in sEVs of transduced hiPSC‐CMs (Figure S4). NanoLuc‐tetraspanin reporters enable direct quantification of sEV secretion from hiPSC‐CMs. (a) Schematic representation of sEV quantification using HA‐NanoLuc(NL)‐tetraspanin reporters. HA‐NL‐tetraspanin reporters are sorted into exosomes and ectosomes and tetraspanin‐positive EV secretion can be quantified by measuring NanoLuc bioluminescent activity in the culture supernatant. (b) Representative immunofluorescence images of hiPSC‐CMs transduced with HA‐tagged NL‐tetraspanin reporters and stained for HA‐tag and cardiac troponin (scale bar = 20 µm). (c) Representative measurement of NanoLuc activity in cells and conditioned culture supernatant of hiPSC‐CMs expressing HA‐NL‐tetraspanin reporters cultured in a 96‐well format at a density of 70,000 cells per well. Data is represented as mean ± SD. (d) Representative luminescence measurement of size exclusion chromatography (SEC) fractions obtained from the supernatant of hiPSC‐CMs expressing HA‐NL‐tetraspanin reporters. (e) Representative graph of particle number (>50 nm) of pooled SEC fractions analyzed by nanoparticle tracking analysis (NTA) with mean ± SEM. (f) Representative graph of protein concentration of SEC fractions with mean ± SD. Experiments from (d) to (f) were performed in conditioned media collected from 0.5 to 2 million hiPSC‐CMs. Upon addition of the NanoLuc substrate Furimazine, we could reliably measure luminescence in both cells and conditioned culture supernatant of hiPSC‐CMs (Figure 2c). To confirm that the NanoLuc activity in the culture supernatant originated from EV‐associated HA‐NL‐tetraspanin reporters, we separated EVs and soluble proteins using SEC and measured NanoLuc activity in each fraction (Figure 2d). NanoLuc activity peaked in EV‐enriched fractions 8–14, which contain the most particles as confirmed by NTA (Figure 2e), and was nearly absent in fractions 20–26, which are enriched in soluble proteins and extracellular particles that are smaller than the 70 nm pore size of the SEC column (Figure 2f). In summary, we generated hiPSC‐CMs that stably express HA‐NL‐tetraspanin reporters. These cells can be utilized to measure tetraspanin‐positive EV release directly from the conditioned medium of a relatively small number of cells. Screening of chemical modulators identifies pathways involved in tetraspanin‐positive EV biogenesis of hiPSC‐CMs To improve our understanding of CM‐EV biogenesis mechanisms, we subsequently used hiPSC‐CMs expressing HA‐NL‐tetraspanin reporters to screen a range of small molecules known to modulate EV secretion in other cell types (Table S1) (Catalano & O'Driscoll, 2020). The screen was performed in a 96‐well format and the NanoLuc activity in the collected culture supernatant was normalized to the corresponding cellular NanoLuc activity, to account for small differences in the plating of the cells (Figure 3a). Small molecule concentrations were selected based on the literature, and alamarBlue assay was used to ensure that the cell viability was above 80% (Figure S5). As a positive control, we included the vacuolar proton pump H+ ATPase (vATPase) inhibitor Bafilomycin, which has been reported to strongly increase sEV secretion in multiple cell types (Bebelman et al., 2023). Indeed, we observed that Bafilomycin significantly increased HA‐NL‐CD9 (+211%) and ‐CD63 (+343%) secretion, with a similar trend for HA‐NL‐CD81 (+159%) (Figure 3b). Similarly, the actin polymerization inhibitor cytochalasin D and the calpain inhibitor calpeptin significantly increased HA‐NL‐CD9 secretion (cytochalasin D: +73%, calpeptin: +43%) and HA‐NL‐CD63 secretion (cytochalasin D: +80%, calpeptin: +30%) (Figure 3c–e). The PIKfyve inhibitor apilimod has previously been demonstrated to cause an accumulation of late endosomes and stimulate EV secretion in prostate cancer cells. We observed a similar effect in hiPSC‐CMs, where apilimod caused the accumulation of CD63‐positive late endosomes and significantly increased HA‐NL‐CD9 (+45%) and ‐CD63 (+38%) secretion (Figures 3c–e and S6). In contrast, the ROCK inhibitor Y‐27632 significantly reduced HA‐NL‐CD81 secretion (–29%) and showed a similar trend for HA‐NL‐CD9 (–20%) and ‐CD63 (–16%). The neutral sphingomyelinase 2 (nSMase2) inhibitor GW4869, the Niemann–Pick disease type C1 (NPC1) inhibitor U18666A, the Rab27 inhibitor nexinhib20, the VPS34 inhibitor SAR405, the PI3K inhibitor wortmannin and the calcium ionophore A23187 did not significantly affect HA‐NL‐tetraspanin secretion by cardiomyocytes. These findings confirm that cardiomyocytes expressing HA‐NL‐tetraspanins can be used to screen for modulators of CM‐EV secretion and suggest that multiple players participate in the biogenesis of tetraspanin‐positive CM‐EVs. Effect of extracellular vesicle biogenesis modulators on NanoLuc‐tetraspanin secretion by hiPSC‐CMs. (a) Experimental method schematic of a small molecule screen using NanoLuc‐tetraspanin reporters in hiPSC‐CMs. (b) NanoLuc‐tetraspanin secretion upon bafilomycin treatment (0.1 μM) normalized to the DMSO control. (c–e) NanoLuc‐tetraspanin secretion of CD9 (c), CD63 (d) and CD81 (e) analysis using various small molecules involved in sEV secretion: GW4869 (5 µM), wortmannin (0.1 µM), U18666A (5 µM), apilimod (0.5 µM), Y‐27632 (5 µM), cytochalasin D (0.1 µM), calcium ionophore A23187 (1 µM), calpeptin (5 µM), SAR405 (1 µM), and nexinhib20 (1 µM). Results from (b) to (e) are plotted as mean ± SEM of three independent experiments. Statistical analysis was done using a one‐sample, two‐tailed T‐test against a theoretical mean of 1. *p < 0.05. Experiments were performed in a 96‐well format at a density of 70,000 cells per well. Stress conditions modulate tetraspanin‐ positive EV secretion by hiPSC‐CMs MI is characterized by physiological stress due to a lack of blood supply, limiting nutrient and oxygen levels (hypoxia) in cardiac tissue. When ischemic cardiac cells start to die, they release inflammatory cytokines such as tumour necrosis factor‐alpha (TNF‐α), initiating inflammation (Swirski & Nahrendorf, 2018). Hypoxia, nutrient starvation, and TNF‐α have all been linked to changes in EV biogenesis and cargo selection in cardiomyocytes (Anselmo et al., 2021; Borosch et al., 2017; Yang et al., 2018) or other cell types (Dias et al., 2016; Fader et al., 2008; Wei et al., 2017). To gain insight into the underlying mechanisms, we first exposed hiPSC‐CMs expressing HA‐NL‐tetraspanins to cardiac stressors and assessed the effect of these on HA‐NL‐tetraspanin release after 24 and 48 h (Figure 4a–c). Nutrient starvation significantly reduced the secretion of HA‐NL‐CD9 (–32%) and ‐CD81 (–56%) already after 24 h. After 48 h, nutrient starvation significantly reduced HA‐NL‐CD63 (–36%) and ‐CD81 (–34%) secretion, with a similar trend for HA‐NL‐CD9 (–34%). Hypoxic culture conditions increased the secretion of HA‐NL‐CD63 (+67%) and ‐CD81 (+35%) after 48 h (Figure 4a). Although the effect of hypoxia was not significant in this experiment, we observed a significant increase in the secretion of all HA‐NL‐tetraspanin reporters under hypoxic conditions in further experiments (Figure 5a‐c). Interestingly, TNF‐α treatment resulted in a time‐dependent significant increase in HA‐NL‐CD9 (+50%) and ‐CD81 (+39%) secretion after 48 h, and a similar trend for HA‐NL‐CD63 (+42%). Importantly, at the concentration used in this study, TNF‐α does not induce CM cell death (Saraf et al., 2021), as was confirmed using a cell viability assay (Figure S5). Cardiomyocytes are known to express TNF‐α receptors I and II (Schulz & Heusch, 2009), and neutralizing antibodies against both receptors significantly reduced TNF‐α‐induced HA‐NL‐CD81 secretion (–34%) (Figure 4d), suggesting that both TNF receptors contribute to TNF‐α‐induced CD81‐positive EV secretion. Stress conditions modulate extracellular vesicle secretion by hiPSC‐CMs. (a–c) Effect of various stress conditions (nutrient starvation, 10 ng/mL TNF‐α, and hypoxia) on the secretion of NanoLuc‐CD9 (a), NanoLuc‐CD63 (b) and NanoLuc‐CD81 (c) after 24 and 48 h. Results were normalized to control (Ctrl), which consists of maturation media only, and are plotted as mean ± SEM of three independent experiments. Statistical analysis was done using a one‐sample, two‐tailed T‐test against a theoretical mean of 1. *p < 0.05. (d) NanoLuc‐CD81 secretion under control (Ctrl) and TNF‐α conditions in combination with neutralizing antibodies against TNF‐α receptor I (TNF‐α RI) or II (TNF‐α RII). Results are plotted as mean ± SEM of three independent experiments. Comparisons in the control group were analyzed using a one‐sample, two‐tailed T‐test against a theoretical mean of 1. Comparisons within the TNF‐α group were analyzed using a one‐way ANOVA with Dunnett correction against the DMSO condition. *p < 0.05, **p < 0.01. (e–g) Fold change of particle concentration compared to control (e, f) and representative size distribution profiles (g) of sEVs released after 48 h under control, TNF‐α, and hypoxia analyzed by nanoparticle tracking analysis (NTA). Results are plotted as mean ± SEM of four and seven independent experiments for TNF‐α and hypoxia, respectively. Comparisons to control were analyzed using a one‐sample, two‐tailed T‐test against a theoretical mean of 1. (h) Representative western blot analysis of two independent experiments of sEVs released after 48 h under control, hypoxia and TNF‐α for EV markers (HSP70, annexin A1 and CD81). The arrows highlight the bands that matched the predicted molecular weight. sEV samples from equal cell numbers and conditioned medium volumes were loaded. To validate the effects of TNF‐α and hypoxia on total CM‐EV secretion, we isolated CM‐EVs by dUC from the conditioned medium of hiPSC‐CMs cultured under hypoxic conditions or treated with TNF‐α for 48 h and measured the particle number and size distribution using NTA (Figure 4e–g). We observed that hypoxia increased the particle number in all replicates, although with a variation in effect. On average, TNF‐α stimulation caused a minor increase (+22%) in particle concentration, but this effect was not significant. Western blot analysis of EVs confirmed increased levels of the EV markers HSP70, annexin A1, and CD81 in the sEV pellet of hypoxic cells (Figure 4h). The discrepancy of results between NTA and the HA‐NL‐tetraspanin secretion assay suggests that, in addition to tetraspanin‐positive EVs, hiPSC‐CMs also secrete tetraspanin‐negative EVs and/or non‐vesicular extracellular particles, which respond differently to hypoxia and TNF‐α. Alternatively, hypoxia and TNF‐α may modulate the loading density of tetraspanins into EVs. To investigate this possibility, we analyzed the effect of hypoxia and TNF‐α on the percentage of EVs displaying CD63 and/or CD81 by dSTORM imaging (Figure S7). Hypoxia reduced the percentage of CD63‐positive EVs, but not CD81‐positive EVs, and TNF‐α‐treatment did not change the percentage of CD63‐ and CD81‐positive EVs. Thus, modulation of tetraspanin‐loading into EVs alone is not sufficient to explain the discrepancies between NTA and HA‐NL‐tetraspanin measurements. In conclusion, our results suggest that hiPSC‐CMs release multiple EV and particle subpopulations and highlight that MI‐associated stress conditions affect the secretion of tetraspanin‐positive EVs by hiPSC‐CMs. Neutral sphingomyelinase 2 mediates tetraspanin‐positive EV release under hypoxic conditions To further explore the potential mechanisms underlying TNF‐α‐ and hypoxia‐stimulated tetraspanin‐positive EV production, we investigated the effect of EV biogenesis‐modulating small molecules on HA‐NL‐tetraspanin release (Figure 5a–c). Under basal culture conditions, the effects of these modulators were similar between 16 (Figure 3c–e) and 48 h (Figure 5a–c), except for the NPC1‐inhibitor U18666A and calpeptin. U18666A treatment for 48 h significantly increased HA‐NL‐CD9 secretion (+67%), with a similar trend for HA‐NL‐CD63 (+38%) and ‐CD81 (+40%) secretion, whereas calpeptin treatment for 48 h did not affect HA‐NL‐tetraspanin secretion. nSmase2 mediates extracellular vesicle release under hypoxic conditions. (a–c) Effect of EV biogenesis modulators GW4869 (5 µM), Y‐27632 (5 µM), nexinhib20 (1 µM), calpeptin (5 µM), U18666A (5 µM) and SAR405 (1 µM) on NanoLuc‐tetraspanin secretion under control (Ctrl), 10 ng/mL TNF‐α and hypoxic conditions for CD9 (a), CD63 (b), CD81 (c) positive EVs after incubation for 48 h. Results were normalized to control (Ctrl), which consists of maturation media only, and are plotted as mean ± SEM of three independent experiments. Comparisons to control conditions were analyzed using a one‐sample, two‐tailed T‐test against a theoretical mean of 1. Comparisons within the TNF‐α and hypoxia groups were analyzed using a one‐way ANOVA with Dunnett correction against the DMSO condition within each group. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. (d) Schematic representation of CD63‐pHluorin for imaging of MVB‐plasma membrane fusion. (e) Localization of CD63‐pHluorin in hiPSC‐CMs (scale bar = 20 µm). (f, g) Time‐lapse TIRF imaging of a CD63‐pHluorin fusion event in control (f) and TNF‐α‐treated (g) hiPSC‐CMs. (h) MVB‐plasma membrane fusion rate in CD63‐pHluorin hiPSC‐CMs under control, TNF‐α and hypoxic conditions of two independent experiments for each condition. The ROCK inhibitor Y‐27632 significantly decreased HA‐NL‐CD63 (–21%) and ‐CD81 (–32%) secretion in control conditions and showed a similar trend for all HA‐NL‐tetraspanins upon TNF‐α treatment or hypoxic conditions. Interestingly, whereas nSmase2 inhibition by GW4869 did not decrease EV secretion in control conditions or upon TNF‐α treatment, it significantly inhibited HA‐NL‐CD9 (–52%) and ‐CD81 (–42%) secretion under hypoxic conditions. A similar trend was observed for HA‐NL‐CD63 secretion (–36%). Thus, it appears that the mechanisms underlying EV biogenesis in hiPSC‐CMs are context‐dependent. In line with the previously observed discrepancies between HA‐NL‐tetraspanin measurements and NTA (Figure 4), GW4869 did not significantly decrease the total extracellular particle concentration following hypoxia treatment (Figure S8), suggesting that the secretion of tetraspanin‐negative EVs and/or non‐vesicular extracellular particles is nSmase2 independent. The contribution of nSmase2 to HA‐NL‐tetraspanin secretion under hypoxic conditions suggests that hypoxia drives the release of MVB‐derived exosomes (Verweij et al., 2018). This speculation agrees with previous studies revealing that hypoxia induces peripheral trafficking of late endosomal compartments, a limiting step in exosome secretion (Dorayappan et al., 2018; Wang et al., 2023). To investigate whether the TNF‐α‐ and hypoxia increase exosome secretion specifically, we first assessed the effect of these treatments on MVB localization using IF microscopy against the MVB marker CD63. We did not observe differences in MVB localization when compared to control hiPSC‐CMs (Figure S9). Due to their overlapping content and biophysical properties, most EV isolation and characterization strategies, including HA‐NL‐tetraspanin‐based approaches, do not formally distinguish between exosomes and ectosomes. To gain a more conclusive answer, we transduced cardiomyocytes with a CD63‐pHluorin reporter construct, which enables quantification of MVB exocytosis rate as a proxy for exosome release (Figure 5d) (Bebelman et al., 2020; Verweij et al., 2018). Fluorescence microscopy confirmed the presence of CD63‐pHluorin in intracellular compartments, in line with its incorporation into MVBs (Figure 5e). Live cell total internal reflection fluorescence (live‐TIRF) microscopy revealed several sudden bursts in fluorescence intensity corresponding to the fusion of CD63‐pHluorin‐containing MVBs with the plasma membrane (Figure 5f,g, Video S2–S3) (Bebelman et al., 2020). Possibly because the fusion rate under control conditions and upon TNF‐α treatment is very low in hiPSC‐CMs, we failed to measure a significant increase in MVB exocytosis upon TNF‐α exposure (Figure 5h). Importantly, we observed no fusion events under hypoxic conditions, suggesting that hypoxia does not lead to a large increase in MVB exocytosis rate and exosome release. Inflammasome activation and mTOR signalling contribute to TNF‐alpha‐induced HA‐NL‐CD81 secretion TNF‐α signalling upregulates the expression of inflammasome components, and hypoxia‐induced mitochondrial dysfunction can activate the inflammasome, leading to the production of cytokines, such as interleukin‐1 beta (IL‐1β) and interleukin‐18 (IL‐18) (Franchi et al., 2009; Toldo & Abbate, 2018). Several studies have linked inflammasome activation to elevated EV secretion (Budden et al., 2021; Liao et al., 2023; Öhman et al., 2014). To investigate the role of the inflammasome in TNF‐α‐ and hypoxia‐induced EV secretion, we inhibited the inflammasome sensor NACHT, leucine‐rich repeat and pyrin domain‐containing protein 3 (NLRP3) using the small molecule MCC950 (Coll et al., 2015). MCC950 significantly inhibited TNF‐α‐induced HA‐NL‐CD81 secretion (–24%) but did not inhibit basal or hypoxia‐induced HA‐NL‐CD81 secretion, and a similar trend was observed for HA‐NL‐CD63 (–19%) (Figure 6a,b). Considering that TNF‐α signalling can trigger the mechanistic target of rapamycin (mTOR) activation (Lee et al., 2007), and given the role of mTOR signalling in inflammasome formation (Ko et al., 2017), we also tested the effect of the mTOR inhibitor rapamycin and found that it significantly inhibited TNF‐α‐induced HA‐NL‐CD81 secretion (–30%) with a similar trend for HA‐NL‐CD63 (–25%) (Figure 6a,b). In conclusion, our results suggest a role for inflammasome activation and mTOR signalling in TNF‐α‐induced CM‐EV secretion. Inflammasome activation and mTOR signalling contribute to TNF‐alpha‐induced HA‐NL‐CD81 secretion. (a, b) NanoLuc‐tetraspanin secretion of CD63 (a) and CD81 (b) after incubation for 48 h with the inflammasome inhibitor MCC950 or mTOR inhibitor rapamycin under control, TNF‐α and hypoxic conditions. Results are plotted as mean ± SEM of three independent experiments. Comparisons to control conditions were analyzed using a one‐sample, two‐tailed T‐test against a theoretical mean of 1. Comparisons within the TNF‐α and hypoxia groups were analyzed using a one‐way ANOVA with Dunnett correction against the DMSO condition within each group. **p < 0.01, ***p < 0.001. (c) Scheme summarizing the findings of this study. EV biogenesis is context‐dependent in hiPSC‐CMs. Under basal culture conditions, hiPSC‐CMs can secrete exosomes, and ectosomes, which release is mediated in part by ROCK. Nutrient starvation decreases CM‐EV secretion while hypoxia increases the production of CM‐EVs in a nSmase2‐dependent manner. TNF‐α increased CM‐EV secretion through a process involving NLRP3 inflammasome activation and mTOR signalling. MI, myocardial infarction; ROCK, Rho‐associated protein kinase; MVB, multivesicular body; nSMase2, neutral sphingomyelinase 2; TNF‐α, tumour necrosis factor alpha; TNFRI, TNF‐α receptor 1; TNFRII, TNF‐α receptor 2; NLRP3, NLR family pyrin domain containing 3; mTOR, mammalian target of rapamycin. DISCUSSION EVs have emerged as important mediators of intercellular communication in the heart under homeostatic and pathological conditions, such as MI. However, the biogenesis mechanisms that contribute to CM‐EV secretion under these conditions are poorly understood. This hampers the discovery of new therapeutic targets to modulate CM‐EV production and the design of more effective CM‐EV therapies to promote cardiac repair after MI. Here, we generated HA‐NL‐tetraspanin‐expressing hiPSC‐CMs that enabled us to quantify tetraspanin‐positive EV secretion from small numbers of cells. Using this approach, we identified modulators of CM‐EV biogenesis under basal and stress‐associated conditions (Figure 6c). To study EV biogenesis in hiPSC‐CMs, we used a panel of small molecules that target known EV biogenesis pathways (Table S1) and observed that the ROCK inhibitor Y‐27632 decreased tetraspanin‐positive CM‐EV secretion. ROCK has been reported to participate in ectosome release by inducing outward budding of the plasma membrane via actomyosin remodeling. There was no significant effect on HA‐NL‐tetraspanin secretion upon the inhibition of nSmase2, which contributes to ceramide‐dependent intraluminal vesicle budding and MVB formation. Similarly, we did not observe any effect of inhibitors of the endosomal cholesterol‐transporter NPC1 or the small GTPase Rab27, which contribute to exosome secretion by controlling MVB positioning and docking to the plasma membrane, respectively. These findings suggest that, under basal culture conditions, hiPSC‐CMs release more ectosomes than exosomes. In line with this, live TIRF imaging with CD63‐pHluorin revealed that hiPSC‐CMs have a very low MVB‐plasma membrane fusion rate compared to other cell types (Bebelman et al., 2020). hiPSC‐CMs are not unique in their preference for ectosome secretion. Recent studies have reported that the majority of sEVs secreted by HEK293 and HeLa cells originate from the plasma membrane (Bebelman et al., 2023; Fordjour et al., 2022; Mathieu et al., 2021). Similar to our findings in hiPSC‐CMs, NanoLuc‐CD63 secretion from HEK293 cells was not inhibited by GW4869 or U1866A treatment (Bebelman et al., 2023), except when bafilomycin was used to induce exosome secretion (Bebelman et al., 2023). Additionally, we tested other small molecules that have been reported to modulate EV production. mTOR signalling and autophagy can both inhibit and stimulate the release of sEVs, depending on cell type and cellular context (Fan et al., 2020; Gao et al., 2020; Zou et al., 2019). Our results suggest that mTOR signalling and canonical autophagy do not play a role in tetraspanin‐positive EV release from hiPSC‐CMs under basal conditions, as shown using rapamycin and SAR405. However, we found that the PIKfyve inhibitor apilimod, a stimulator of secretory autophagy and exosome secretion in a prostate cancer cell line, increased tetraspanin‐positive CM‐EV release. We observed that apilimod caused an accumulation of CD63‐positive late endosomes in hiPSC‐CMs (Figure S6), which may indicate impaired fusion with lysosomes and could explain the observed increase in EV release. The actin polymerization inhibitor cytochalasin D also increased HA‐NL‐tetraspanin secretion from hiPSC‐CMs, consistent with previous observations in HEK293 cells. Finally, we found that 16 h treatment with the calpain inhibitor calpeptin induced HA‐NL‐tetraspanin release, whereas it is described to decrease EV secretion by other cell types. Interestingly, this stimulatory effect was lost after 48 h, suggesting that short‐term calpeptin treatment leads to a temporary surge in CM‐EV secretion. The rate of EV secretion and the EV biogenesis mechanisms highly depend on the cellular context (Dixson et al., 2023). We sought to explore the effect of MI‐associated stress factors on human CM‐EV secretion. Nutrient starvation decreased HA‐NL‐tetraspanin secretion by hiPSC‐CMs. This is consistent with observations in other cell types (Bebelman et al., 2023; Dias et al., 2016), but different from a previous finding in immortalized H9C2 embryonic cardiomyocytes, where glucose starvation increased EV secretion (Garcia et al., 2015). The decrease in HA‐NL‐tetraspanin secretion upon nutrient starvation may be due to the induction of autophagy and redirection of MVBs into the autophagic pathway, resulting in decreased EV secretion (Fader et al., 2008). Consistent with previous work, we observed that hypoxia increased the production of CD81‐ and CD63‐positive CM‐EVs (Anselmo et al., 2021; Jung et al., 2021; Koncz et al., 2023; Yang et al., 2018). Notably, several other studies did not find a significant effect of hypoxia on CM‐EV secretion. These discrepancies may be attributed to variations in the origin of the cardiomyocytes, EV isolation and quantification methodology, and the duration and method used to induce hypoxia (Borosch et al., 2017; Koncz et al., 2023; Vaskova et al., 2020). The cytokine TNF‐α, which contributes to the inflammatory response following MI, has been reported to stimulate EV release from tumour cells and astrocytes (Wang et al., 2017; Wei et al., 2017), but its role in CM‐EV secretion remained unknown. Our results suggest that stimulation with a non‐toxic concentration of TNF‐α increases tetraspanin‐positive sEV secretion by cardiomyocytes. To shed light on the mechanisms underlying hypoxia‐ and TNF‐α‐induced tetraspanin‐positive EV secretion, we tested the panel of EV‐modulating small molecules on HA‐NL‐tetraspanin secretion under these stress conditions. We found that the nSmase2 inhibitor GW4869 blocked hypoxia‐induced HA‐NL‐tetraspanin secretion, indicating that hypoxia might increase exosome secretion. However, we did not observe any MVB‐plasma membrane fusion events using CD63‐pHuorin following hypoxia treatment. The CD63‐pHluorin imaging data needs to be interpreted with care since the low fusion activity in hiPSC‐CMs precluded an accurate assessment of differences in MVB exocytosis rate. Nevertheless, the complete absence of fusion events, and the lack of Rab27 and NPC1‐involvement in hypoxia‐induced HA‐NL‐tetraspanin secretion, argue against a strong induction of exosome secretion under hypoxic conditions. Further research is needed to understand the membrane origin of hypoxia‐induced nSmase2‐dependent EV secretion by hiPSC‐CMs, especially given that nSmase2 has also been demonstrated to regulate ectosome release. In addition, it would be of interest to investigate whether hypoxia affects nSmase2 expression, localization or activity. None of the EV‐modulating drugs significantly inhibited TNF‐α‐stimulated secretion, although the ROCK inhibitor Y‐27632 appeared to reduce TNF‐α‐stimulated and basal HA‐NL‐tetraspanin secretion to a similar extent. TNF‐α primes the NLRP3 inflammasome and activates mTOR signalling (Franchi et al., 2009; Lee et al., 2007), and both signalling pathways have been linked to enhanced EV secretion (Budden et al., 2021; Gao et al., 2020). Indeed, our results suggest that NLRP3 inflammasome and mTOR signalling contribute to the TNF‐α‐induced secretion of CD81‐positive EVs. Overall, our findings suggest that hypoxia and TNF‐α induce tetraspanin‐positive CM‐EV secretion through different mechanisms. One limitation of HA‐NL‐tetraspanin reporters to study EV biogenesis is that these reporters only enable the quantification of tetraspanin‐positive EVs and not tetraspanin‐negative EVs or non‐vesicular extracellular particles. Furthermore, certain conditions may affect the loading density of tetraspanins into EVs rather than the number of secreted tetraspanin‐positive EVs. However, dSTORM imaging suggests this is not the case for TNF‐α and is only partially true for hypoxia, which affected the percentage of CD63‐, but not CD81‐positive, CM‐EVs. This can result in discrepancies when comparing HA‐NL‐tetraspanin measurements with total particle measurements. In addition, even though the HA‐NL‐tetraspanin‐expressing hiPSC‐CMs allowed the quantification of tetraspanin‐positive EV secretion from small numbers of hiPSC‐CMs (70K) in a 96‐well format, we observed some variation in between experimental replicates. This variability may originate from occasional cell detachment as a result of frequent medium replacements (every 3–4 days) between the seeding of the cells in 96‐well format and the experiments. Nevertheless, by combining HA‐NL‐CD9, ‐CD63, and ‐CD81 measurements, we were able to detect changes in tetraspanin‐positive CM‐EV secretion in response to stress conditions and known EV modulators, highlighting the power of this approach for investigating CM‐EV biogenesis under control and MI‐associated conditions. A major advantage of HA‐NL‐tetraspanin reporters is that they can be used to screen for modulators of EV secretion in a high‐throughput manner. For example, HA‐NL‐tetraspanin reporters have been used to screen a kinase inhibitor library in HEK293 cells (Bebelman et al., 2023) and an FDA‐approved drug library in a breast cancer cell line (Grisard et al., 2022). Alternatively, small interfering RNA (siRNA) and parallel CRISPR screens may be used to study genes involved in CM‐EV generation. HA‐NL‐tetraspanin reporters can also be used to optimize culture conditions for the production of large numbers of therapeutic hiPSC‐CM EVs (Bost et al., 2022). Another exciting possibility is to express HA‐NL‐tetraspanin reporters in genetically modified or patient‐derived hiPSC‐CMs that model cardiomyopathies associated with changes in the secretion or function of CM‐EVs (Gartz et al., 2018; James et al., 2021; Turner et al., 2021). In this case, HA‐NL‐tetraspanin reporters could reveal the impact of genetic mutations or specific cardiomyopathy phenotypes on EV generation mechanisms. Finally, EVs isolated from HA‐NL‐tetraspanin‐expressing cardiomyocytes could be used to study CM‐EV uptake by other cell types in vitro, or even injected into the infarcted mouse heart to study their cardiac retention and biodistribution following MI (Bonsergent et al., 2021; Gupta et al., 2020; Rufino‐Ramos et al., 2022). In conclusion, we generated HA‐NL‐tetraspanin‐expressing hiPSC‐CMs that allowed for quantification of CM‐EV secretion from small numbers of cells (70K) without the need for cumbersome and time‐consuming EV isolation techniques. We used these cells to identify modulators of tetraspanin‐positive CM‐EV biogenesis under basal and MI‐associated stress conditions, demonstrating the power of this multiple‐reporter approach. We observed that under basal culture conditions, hiPSC‐CMs can secrete exosomes, and ectosomes as part of CM‐EV release is mediated by ROCK. Additionally, we demonstrated that EV biogenesis is context‐dependent in hiPSC‐CMs. Nutrient starvation decreased HA‐NL‐tetraspanin secretion, while hypoxia increased the production of tetraspanin‐positive EVs in an nSmase2‐dependent manner. Moreover, TNF‐α increased tetraspanin‐positive EV secretion through a process involving NLRP3 inflammasome activation and mTOR signalling. This work paves the way for larger screens that will improve our understanding of the molecular mechanisms underlying CM‐EV biogenesis in homeostasis and disease, which will aid the development of novel pharmacological strategies for MI. MATERIALS AND METHODS Chemicals GW4869 (D1692), wortmannin (W3144), U18666A (U3633), apilimod (SML2974), Y‐27632 (688001), cytochalasin D (C8273), calcium ionophore A23187 (C7522), nexinhib20 (SML1919), bafilomycin A1 (B1793), MCC950 (538120), and rapamycin (R0395) were purchased from Sigma–Aldrich. Calpeptin (HY‐100223) and SAR405 (HY‐12481) were obtained from MedChem Express. Recombinant human TNF‐α (300‐01A) was acquired from PeproTech. hiPSC‐derived cardiomyocyte differentiation and culture The hiPSC line was obtained from the Stanford Cardiovascular Institute Biobank (SCVI‐273) and non‐enzymatically passaged using 0.5 mM EDTA (Thermo Fisher) every 4 days as previously described (Feyen et al., 2021). hiPSC cells were maintained in 0.1 mg/mL matrigel‐coated 6‐well plates (Corning, 356230) in E8 complete medium (Gibco, A1517001) for at least two passages, and grown to ∼90% confluence before starting cardiac lineage differentiations. hiPSCs were efficiently differentiated as previously described (Maas et al., 2021), using the RPMI/B27 minus insulin medium (Gibco, A1895601) supplemented with 7–8 µM CHIR99021 (cell line specific) (SelleckChem, S2924) for the first 2 days and 2 µM Wnt‐C59 (SelleckChem, S7037) for another 48 h. The proliferation of hiPSC‐CMs can be conducted by removing cell‐to‐cell contacts and small molecule inhibition of GSK3 with CHIR99021 (Buikema et al., 2020). Subsequential multiple low‐density passaging and reintroduction of 2 µM CHIR99021 to the RPMI/B27 media (Gibco, 17504044) results in an expansion. During the expansion phase, hiPSC‐CMs were grown in matrigel‐coated T175 flasks and the media was changed every 2–3 days. To mature the CMs, we used a previously described metabolic maturation medium (Feyen et al., 2020), which consists of DMEM without glucose (Gibco, 11966025) supplemented with 3 mM glucose (Sigma–Aldrich, G7021), 10 mM L‐lactate (Sigma–Aldrich, 71718), 5 µg/mL Vitamin B12 (Sigma–Aldrich, V6629), 0.82 µM Biotin (Sigma–Aldrich, B4639), 5 mM Creatine monohydrate (Sigma–Aldrich, C3630), 2 mM Taurine (Sigma–Aldrich, T0625), 2 mM L‐carnitine (Sigma–Aldrich, C0283), 0.5 mM Ascorbic acid (Sigma–Aldrich, A8960), 1× non‐essential amino acid solution (Gibco, 11140050), 0.5% (w/v) Albumax (Gibco, 11020021), 1× B27 and 1% KnockOut Serum Replacement (KOSR, Gibco, 10828028). Maturation was carried out for at least 3 weeks with media changes every 3–4 days before proceeding to experiments. Mature hiPSC‐CMs were detached by incubation with TrypLE select 10× (Gibco, A1217702) for 15–30 min at 37°C. Cells were then re‐plated into matrigel‐coated 96‐well plates to a density of 70,000 cells/well and cultured in RPMI/B27 supplemented with 10% KOSR and 10 µM Y‐27632 (Selleck Chemicals, S1049). Media was switched to RPMI/B27 the following day and changed to maturation media 2–3 days later. hiPSC‐CMs were kept in maturation media for another 7 days to allow cell recovery before drug testing. All HA‐NL‐tetraspanin assays were performed in maturation media which was refreshed every 3–4 days. EV isolation using differential ultracentrifugation (dUC) For Figure 1, conditioned media was harvested from hiPSC‐CMs cultured in T175 flasks and 6‐well plates. For Figure 4, 2–3 million hiPSC‐CMs were grown per well in 6‐well plates and three full plates per condition were stimulated for 48 h with 2 mL per well of maturation media only (control), 10 ng/mL TNF‐α or exposed to hypoxia by placing the plate in a BD GasPak EZ Anaerobe Pouch System (BD Biosciences, 260683). After 48 h, the conditioned medium was collected and sEV isolation was performed using dUC. To remove cell debris, the conditioned medium was centrifuged for 5 min at 300 × g and for 10 min at 2000 × g. Next, the supernatant was added to ultracentrifuge tubes (Beckman Coulter, 344058) and centrifuged for 30 min at 10 000 × g followed by centrifugation for 70 min at 100,000 × g. The pellet of the last step was resuspended in 50 µL of PBS. All centrifugation steps were done at 4°C and the last two steps were performed using Optima XE‐90 ultracentrifuge (Beckman Coulter, A94471) with rotor SW32Ti (Beckman Coulter, 369694). Size exclusion chromatography (SEC) Conditioned medium of hiPSC‐CMs expressing HA‐NL‐tetraspanin reporters was harvested and centrifuged for 5 min at 500 × g to remove cell debris. Separation of sEVs from soluble proteins was performed using qEVoriginal 70 nm SEC column (IZON, ICO‐70) according to the manufacturer's instructions. A total of 1.5 mL of supernatant was added on top of the SEC column and allowed to enter by gravity. PBS was used as running buffer and fractions of 0.5 mL were collected manually up to 26 fractions. A total of 50 µL of each fraction was added in duplicate into a white 96‐well plate (Greiner Bio‐One, 655075) and 50 µL of Nano‐Glo reagent (Promega, N1110, 1:1000) was added in each well using a multichannel. Luminescence activity was measured after 3‐min incubation using SpectraMax iD3 microplate reader (Molecular Devices) and analyzed using SoftMax Pro 7 Software (Molecular Devices). Protein quantification and western blot EV samples were prepared by lysing EVs directly in concentrated RIPA buffer (Millipore, 20–188) and protease/phosphatase inhibitors (Cell Signalling Technology, 5872). Cells were collected in 1× RIPA buffer containing protease/phosphatase inhibitors and incubated on ice for 30 min. Next, cell lysate was centrifugated at 12,000 × g for 15 min at 4°C and the resultant supernatant was collected and stored at –80°C. Protein concentration was determined using the micro‐BCA protein assay kit (Thermo Scientific, 23235) according to the manufacturer's instructions. The protein concentration of cell lysates was measured by diluting samples 1:50 in PBS. The protein content of SEC fractions was quantified by adding 50 µL of each SEC fraction in duplicate into a 96‐well plate except for protein‐enriched fractions (usually starting on fraction 16), which were diluted 1:25. Measurements were done using Multiskan FC microplate photometer (Thermo Scientific, 51119000) and analyzed with SkanIt software version 3.1 (Thermo Scientific, 5187139). To analyze proteins with western blot, EV samples from equal numbers of cells and equal volumes of conditioned medium were loaded without further dilution. Five micrograms of cell lysate was used. Samples were prepared in NuPAGE LDS loading buffer (Invitrogen, NP0007) and reduced using NuPAGE reducing agent (Invitrogen, NP0009), followed by boiling at 95°C for 10 min. For CD63 detection, non‐reduced samples were used. Subsequently, samples were loaded into Bolt 4%–12% Bis‐Tris Plus 1.0 mm × 15 well gels (Invitrogen, NW04125BOX) and PageRuler Plus (Invitrogen, 26619) was used as a protein ladder. Electrophoresis was done at 100 V for 75 min in NuPAGE MOPS SDS running buffer (Invitrogen, NP0001). Gels were transferred to a polyvinylidene difluoride (PVDF) membrane (Invitrogen, IB24001) using the iBlot2 system for 7 min at 20 V. Membranes were blocked in TBS containing 5% BSA (Roche, 10735086001) for 1 h at room temperature or overnight at 4°C. Primary antibody incubation was done overnight at 4°C in TBS‐0.1%Tween20 (Sigma–Aldrich, 822184) containing 0.5% BSA. Primary antibodies were used against CD9 (Merck Millipore, CBL162, 1:1000), CD63 (BD Biosciences, 556019, 1:1000), CD81 (Santa Cruz, sc‐166029, 1:500‐1000), syntenin (OriGene, TA504796, 1:1000), annexin A1 (Abcam, ab214486, 1:1000), GAPDH (Cell Signalling Technology, 2118, 1:1000), GM130 (BD Biosciences, 610823, 1:500), calnexin (GeneTex, GTX101676, 1:1000), HSP70 (Biolegend, 648001, 1:500), anti‐HA‐tag (Roche, 11867423001, 1:500) and TOMM20 (Abcam, ab186735, 1:1000). Secondary antibodies were diluted 1:7500 in a mixture of 50% v/v of Intercept blocking buffer (LI‐COR, 927–60001) in TBS‐0.1%Tween20 and incubated for 2 h at room temperature in the dark. Secondary antibodies included Alexa 680‐conjugated goat anti‐mouse (Invitrogen, A21057) and IRDye 800CW‐conjugated goat anti‐rabbit (LI‐COR Biosciences, 926–32211). Imaging was achieved using channels 700 and 800 nm of the Odyssey M (LI‐COR Biosciences) and processed with LI‐COR acquisition software (LI‐COR Biosciences). Washing steps were performed between antibody incubations three times for 10 min in TBS‐0.1%Tween20 buffer. Nanoparticle tracking analysis (NTA) NTA was conducted using NanoSight NS500 (Malvern Panalytical). To study particle concentration and size distribution of isolated sEVs, EV samples were diluted in PBS to obtain a particle concentration of 30–100 particles per frame. Particle concentration of pooled SEC fractions was measured without dilution, except for protein‐enriched fractions (21–26) that were diluted 1:5 in PBS. Five videos of 60 or 30 s in the case of pooled SEC fractions, were captured using a camera level of 16 and a detection threshold of 6. Videos were analyzed using NanoSight NTA 3.3 software (Malvern Panalytical). Transmission electron microscopy (TEM) To image sEVs isolated by dUC, the EV sample was added to a carbon‐coated grid (75–200 mesh) for 15 min. Unbound vesicles were removed with PBS washing followed by fixation with 1% glutaraldehyde in PBS buffer for 30 min. Next, samples were counterstained with 2% uranyl‐oxalate for 10 min at room temperature and embedded in a mixture of 1.8% methylcellulose and 0.4% uranyl acetate for at least 10 min at 4°C. Grids were dried for 1 h at room temperature. sEVs were imaged on JEM 1011 TEM microscope (Jeol), equipped with a tungsten filament and a DCC CMOS camera Phurona (Emsis) with a resolution of 12 megapixels. The EM beam was set to a voltage of 80 KV. Plasmids The pLenti6.3‐CD63‐pHluorin and pLenti6.3‐HA‐NL‐CD63 lentiviral plasmids were described previously (Bebelman et al., 2020, 2023). To generate HA‐NL‐CD9 and HA‐NL‐CD81 lentiviral plasmids, CD9 and CD81 cDNA sequences were amplified using the primer pairs 5′‐ataggatccatgccggtcaaaggaggcacc‐3′ + 5′‐atatctagagctagaccatctcgcggttcc‐3′ and 5′‐ataggatccatgggagtggagggctgcacc‐3′ + 5′‐atatctagagctagtacacggagctgttcc‐3′, respectively, to add BamHI and XbaI restriction sites. The amplified CD9 and CD81 fragments were cloned into BamHI/XbaI‐digested HA‐NL‐CD63‐pCMV‐Sport6 vector, to generate HA‐NL‐CD9 and HA‐NL‐CD81, which were subsequently subcloned into the pLenti6.3/TO/V5‐DEST vector (Thermo Fisher Scientific). Generation of HA‐NL‐tetraspanin and CD63‐pHluorin expressing hiPSC‐CMs HA‐NL‐tetraspanin and CD63‐pHluorin expressing hiPSC‐CMs were produced using lentiviral transduction. Lentivirus production was done in HEK293FT cells, which were seeded in 10 cm dishes coated with 0.1% gelatin (Sigma–Aldrich, G1890) and cultured in DMEM (Gibco, 41965039) supplemented with 10% FBS (Corning, 35‐079‐CV) and 1% penicillin‐streptomycin (Gibco, 15140122). HEK293FTs were transfected with plasmid DNA using lipofectamine 3000 (Invitrogen, L3000015) at 70%–80% confluency. A total of 3.7 µg pLenti6.3‐HA‐NL‐tetraspanin plasmids or pLenti6.3‐CD63‐pHluorin plasmid were co‐transfected with 0.9 µg pREV (Addgene, 12253) and 1.8 µg pRRE (Addgene, 12251) packaging plasmids, and 1.1 µg pMD2.G (Addgene, 12259) lentiviral envelope plasmid. P3000 reagent was added to the DNA mix in a 2–1 µg ratio. Next, plasmids were complexed with lipofectamine 3000 using a 1– 1.5 µL ratio in Opti‐MEM medium (Gibco, 31985070) for 15 min at room temperature. Subsequently, the complexes were added on top of HEK293FT cells after removing half of the culture medium and incubated for 18 h before changing the media. Lentivirus‐containing media was collected 48–72 h after transfection. Cell debris was removed by centrifugation at 500 × g for 5 min and filtration using a 0.45 µm SCFA syringe filter (Corning, 516–1954), followed by aliquoting and storage at –80°C until further use. To produce stable HA‐NL‐tetraspanin expressing hiPSC‐CMs, cells were cultured in T175 flasks and 7 mL of lentivirus‐containing media was added on top of 20 mL of expansion media RPMI supplemented with B27 (Gibco, 17504044) and 2 µM CHIR99021 (SelleckChem, S2924), when cells reached 60%–70% confluency. After 48 h, the antibiotic selection was performed for at least 5 days using expansion media containing 5 µg/mL blasticidin S (Gibco, A1113903). hiPSC‐CMs were kept in expansion media for another 2–3 days to allow cell recovery and then cultured in maturation media for at least 21 days before split into matrigel‐coated white 96‐well plates (Greiner Bio‐One, 655083) to perform NanoLuc assay. To generate stable CD63‐pHluorin expressing hiPSC‐CMs, cells were grown in a 6‐well plate and cultured for 5 days in an expansion medium containing 2 mL of lentivirus‐containing medium per well. Antibiotic selection was done in 5 µg/mL blasticidin for 4 days and cells were kept in expansion media for 3 days before starting the maturation process. After 21 days, hiPSC‐CMs were passed into matrigel‐coated CELLview slides (Greiner Bio‐One, 543078) to allow live TIRF imaging. Immunofluorescence microscopy hiPSC‐CMs were grown in CELLview slides (Greiner Bio‐One, 543078) and fixed using 4% PFA for 15 min, followed by washing with PBS. Cells were permeabilized with 0.1% Triton X‐100 (Sigma–Aldrich, T9284) for 5 min, washed with PBS, and blocked in 5% BSA (Roche, 10735086001) in PBS solution. Subsequently, primary antibody incubation was performed in blocking buffer with anti‐HA‐tag (rat, Roche, 11867423001, 1:250) or anti‐CD9 (mouse, Merck Millipore, CBL162, 1:100) or anti‐CD63 (mouse, BD Biosciences, 556019, 1:100) and anti‐cardiac troponin T (rabbit, Abcam, ab45932, 1:300). For Figure S1d, hiPSC‐CMs were grown on coverslips and were stained with anti‐α‐actinin (mouse, Sigma–Aldrich, A7811, 1:300) and anti‐Tomm20 (rabbit, Santa Cruz, sc‐11415, 1:300). Both blocking and primary antibody incubations were done for 1 h at room temperature or overnight at 4°C. Secondary antibody incubation was done together with the nuclei staining using Hoechst 33342 (Invitrogen, H1399, 1:10 000) for 1 h at room temperature in the dark. Alexa Fluor 647‐linked anti‐rat (goat, Invitrogen, A21247) and Alexa Fluor 488‐linked anti‐rabbit (donkey, Invitrogen, A21206), or Alexa Fluor 488‐linked anti‐mouse (donkey, Invitrogen, A21202) and Alexa Fluor 647‐linked anti‐rabbit (donkey, Invitrogen, A31573), or Alexa Fluor 488‐linked anti‐mouse (goat, Invitrogen, A11029) and Alexa Fluor 568‐linked anti‐rabbit (goat, Invitrogen, A11011), were used as secondary antibodies and were diluted 1:400 in blocking buffer. Washing steps after each antibody incubation were done three times for 10 min with PBS. Cells were kept in PBS until imaging with the LSM700 confocal microscope (Zeiss) using ZEN 2011 software (Zeiss). Image acquisition was performed using an oil 63×/1.40 objective and the 405, 488 and 633 nm laser lines. Images were analyzed using Fiji software version 2.14 (Schindelin et al., 2012). For Figure S1d, samples were imaged with 20× and 40× objectives of EVOS M7000 Imaging System (Thermo Fisher Scientific). HA‐NL‐tetraspanin secretion assay HA‐NL‐tetraspanin hiPSC‐CMs were seeded in a white 96‐well plate (Greiner Bio‐One, 655083) at a density of 70,000 cells per well. Drug incubations were performed in 200 µL of maturation media per well and in the presence of 10 ng/mL TNF‐α when mentioned. Nutrient starvation was carried out by culturing the cells in DMEM without glucose (Gibco, 11966025) and without supplements. To expose the cells to hypoxia, plates were incubated in a BD GasPak EZ Anaerobe Pouch System (BD Biosciences, 260683), placing one plate per bag. After the specified time point, 100 µL of conditioned medium was collected into a transparent U‐bottom 96‐well plate using a multichannel and centrifuged at 500 × g to remove any floating cells from the supernatant. Next, 50 µL of supernatant was transferred into a white 96‐well plate (Greiner Bio‐One, 655075) and 50 µL of Nano‐Glo reagent was added using a multichannel and incubated for 3 min. Nano‐Glo assay reagent consisted of a 1:1000 dilution of Nano‐Glo substrate (Promega; N1110) in buffer (100 mM 2‐(N‐Morpholino)ethanesulfonic acid hydrate (M8250; Sigma–Aldrich), 1 mM trans‐1,2‐Diaminocyclohexane‐N,N,N′,N′‐tetraacetic acid monohydrate, 150 mM KCl, 1 mM DL‐Dithiothreitol (D0632; Sigma–Aldrich), 35 mM Thiourea (T7875; Sigma–Aldrich) and 0.5% Tergitol NP‐40 (NP40S; Sigma–Aldrich), pH 6). Luminescence activity was measured using CLARIOstar Plus (BMG Labtech) and analyzed with MARS Data Analysis Software (BMG Labtech). The same was performed on the original plate containing the donor cells by adding 100 µL of Nano‐Glo reagent per well. HA‐NL‐tetraspanin secretion was calculated by first dividing the luminescence of the supernatant with that of the cells, followed by normalization to the DMSO control of each plate. Live TIRF microscopy TIRF imaging was performed as previously described (Bebelman et al., 2020). In short, hiPSC‐CMs transduced with CD63‐pHluorin were seeded in CELLview slides (Greiner Bio‐One, 543078) and exposed to maturation media only (control), 10 ng/mL TNF‐α or incubated in a BD GasPak EZ Anaerobe Pouch System (BD Biosciences, 260683) to mimic hypoxic condition. After 48 h, hiPSC‐CMs were imaged by live TIRF microscopy. A Nikon Eclipse Ti with Perfect Focus System setup was used with a laser bench coupled into a Nikon Apo TIRF 100× N.A. 1.49 oil objective. Images were acquired at 2 Hz, using a Photometrics CoolSNAP HQ2 CCD. Live‐imaging experiments were performed in maturation medium at 37°C and 5% CO2 using a Tokai Hit INUBG2E‐ZILCS Stage Top Incubator. Images were acquired using MetaMorph software (version 7.10.2.240) and fusion activity was quantified as described before (Bebelman et al., 2020). Cell viability assay (alamarBlue) Cell viability was performed using alamarBlue reagent (Invitrogen, DAL1025) diluted 1:10 in maturation media followed by filtration with a 0.22 µm CME syringe filter (Carl ROTH, KH54.1). Incubation with small molecule inhibitors and TNF‐α was done in a transparent 96‐well plate for 48 h using the same concentrations as used for the HA‐NL‐tetraspanin secretion assays. After 48 h, the medium was replaced by 100 µL alamarBlue solution per well and incubated overnight. alamarBlue solution was also added in six empty wells as a blank to control for fluorescence background. The next day, 50 µL of conditioned medium was transferred to a black 96‐well plate. Fluorescence was measured at an excitation wavelength of 560 nm and an emission of 590 nm in CLARIOstar Plus (BMG Labtech) and analyzed using MARS Data Analysis Software (BMG Labtech). Cell viability was calculated by subtracting the fluorescence background of the blank (no cell incubation) from all conditions and normalizing to the DMSO control. Direct stochastic optical reconstruction microscopy (dSTORM) imaging hiPSC‐CMs were grown in 6‐well plates at a density of 2–3 million cells per well. One full plate per condition was stimulated for 48 h with 2 mL per well of maturation media only (control), 10 ng/mL TNF‐α or exposed to hypoxia by placing the plate in a BD GasPak EZ Anaerobe Pouch System (BD Biosciences, 260683). After 48 h, the conditioned medium was collected, and sEV isolation was performed using dUC, as described above. EVs were labelled overnight at 4°C with primary conjugated CD81 (TAPA‐1, Biolegend)—AF647 and CD63 (MEM‐259, ITK diagnostics)—ATTO488. Conjugation of the primary antibodies was performed as detailed previously (De Voogt et al., 2023). The next day, 200 nM MemGlow590 was added and incubated for 30 min at room temperature. EVs were immobilized on chips provided in the EV profiler kit v1 (ONI) according to the manufacturer's protocol. dSTORM imaging was performed in OxEA (Nahidiazar et al., 2016) buffer on a Nanoimager‐S (ONI) equipped with a 100× oil immersion objective (Olympus, NA 1.4). A total of 488, 561 and 640 nm lasers were used with 30%, 50% and 25% laser power, respectively. For each channel, 3000 frames were acquired with an exposure time of 30 ms. An illumination angle of 53° was applied. Three fields of view per sample were acquired (representing the individual data points). Filtering and DBSCAN cluster analysis were conducted using a collaborative discovery (CODI) analysis platform (https://alto.codi.bio/, ONI). Statistical analysis Data was analyzed and visualized using Prism 10.0 software (GraphPad). The sample number and statistical tests used are indicated in the figure legends. Statistical significance is denoted by asterisks with *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. AUTHOR CONTRIBUTIONS Margarida Viola: Conceptualization (lead); data curation (lead); formal analysis (lead); investigation (lead); methodology (lead); visualization (lead); writing—original draft (lead); writing—review and editing (lead). Maarten P. Bebelman: Conceptualization (supporting); data curation (supporting); formal analysis (supporting); investigation (supporting); resources (supporting); validation (supporting); visualization (supporting); writing—original draft (supporting). Renee G. C. Maas: Investigation (supporting); methodology (supporting); resources (supporting); writing—review and editing (supporting). Willemijn S. de Voogt: Formal analysis (supporting); investigation (supporting); methodology (supporting). Frederik J. Verweij: Investigation (supporting); methodology (supporting); resources (supporting); writing—review and editing (supporting). Cor S. Seinen: Investigation (supporting); methodology (supporting). Saskia C. A. de Jager: Supervision (supporting); writing—review and editing (supporting). Pieter Vader: Supervision (supporting); writing—review and editing (supporting). Dirk Michiel Pegtel: Conceptualization (supporting); resources (supporting); supervision (supporting); writing—review and editing (supporting). Joost Petrus Gerardus Sluijter: Conceptualization (supporting); funding acquisition (lead); project administration (lead); supervision (lead); writing—review and editing (supporting). CONFLICT OF INTEREST STATEMENT The authors declare no competing interests. Supporting information Figure S1. Generation and purity control of hiPSC‐CMs. A. Schematic workflow of the hiPSC‐CM generation and culturing. B. Representative gating strategy to select alpha‐actinin positive (α‐actinin+) hiPSC‐CMs. Abbreviations: SSC, side scatter; FSC, forward scatter. C. Flow cytometry analysis of α‐actinin+ hiPSC‐CMs of three individual differentiations as previously described. (Maas et al., 2021) D. Representative immunofluorescence images of hiPSC‐CMs after expansion and stained for Tomm20 (mitochondria) and α‐actinin (cardiac sarcomeres). Scale bar = 100 µm. Figure S2. Particle concentration of 100K pellet obtained from non‐conditioned media consisting of maturation media that has not been in contact with the cells (NC, 2.05 × 1010) and the supernatant of hiPSC‐CMs (sEV, 5.22 × 1011) measured by nanoparticle tracking analysis (NTA). Results are plotted as mean ± SEM of three independent experiments. Figure S3. Representative immunofluorescence images of hiPSC‐CMs after maturation and stained for CD9 or CD63 and cardiac troponin T. Scale bar = 20 µm. Figure S4. Transduced hiPSC‐CMs expressing HA‐NanoLuc‐tetraspanins CD63, CD81, and CD9 and their respective sEVs isolated by differential ultracentrifugation analyzed by western blot for HA‐tag and the cell‐enriched marker TOMM20. Figure S5. Cell viability (alamarBlue assay) of hiPSC‐CMs after exposure to small molecules or proteins used throughout the study. Results are plotted as mean ± SEM (n = 3). Figure S6. Representative immunofluorescence images of hiPSC‐CMs treated for 16 hours with DMSO or 0.5 µM apilimod and stained for CD63 (cyan) and cardiac troponin T (magenta). Scale bar = 20 µm. Figure S7. Representative cluster analysis of dSTORM images of single sEVs released by hiPSC‐CMs after 48 hours under control (Ctrl, maturation media only), TNF‐α, and hypoxia conditions. The percentage of clusters displaying double CD63‐CD81‐, single CD63+, single CD81+, or double CD63+CD81+ phenotype is presented relative to the total number of MemGlow 590+ clusters. Data is plotted as mean ± SD of three technical replicates. Figure S8. Particle concentration of sEVs released after 48 hours under control, hypoxia untreated, and hypoxia treated with 5 µM GW4869 conditions. Data was analyzed by nanoparticle tracking analysis (NTA). Results are plotted as mean ± SEM of two independent experiments. Comparisons to control were analyzed using a one‐way ANOVA with Tukey correction. *p < 0.05. Figure S9. Representative immunofluorescence images of hiPSC‐CMs upon incubation for 48 hours of control, TNF‐α, and hypoxic conditions and stained for CD63 and cardiac troponin. Scale bar = 20 µm. Table S1 Video S1. Morphology of beating hiPSC‐CMs 26 days after maturation imaged using a 10x objective. Video S2. TIRF microscopy of hiPSC‐CMs expressing CD63‐pHluorin. The video is played at 10x speed. Scale bar 10 µm. Video S3. TIRF microscopy of hiPSC‐CMs expressing CD63‐pHluorin after 48 hours incubation with TNF‐α. The video is played at 10x speed. Scale bar 10 µm.