Mesenchymal stem cell-derived extracellular vesicles as probable triggers of radiation-induced heart disease

Background Radiation-induced heart disease has been reported, but the underlying mechanisms remain unclear. Mesenchymal stem cells (MSCs), also residing in the heart, are highly susceptible to radiation. We examined the hypothesis that the altered secretion of extracellular vesicles (EVs) from MSCs is the trigger of radiation-induced heart disease. Methods By exposing human placental tissue-derived MSCs to 5 Gy γ-rays, we then isolated EVs from the culture medium 48 h later and evaluated the changes in quantity and quality of EVs from MSCs after radiation exposure. The biological effects of EVs from irradiated MSCs on HUVECs and H9c2 cells were also examined. Results Although the amount and size distribution of EVs did not differ between the nonirradiated and irradiated MSCs, miRNA sequences indicated many upregulated or downregulated miRNAs in irradiated MSCs EVs. In vitro experiments using HUVEC and H9c2 cells showed that irradiated MSC-EVs decreased cell proliferation (P < 0.01), but increased cell apoptosis and DNA damage. Moreover, irradiated MSC-EVs impaired the HUVEC tube formation and induced calcium overload in H9c2 cells. Conclusions EVs released from irradiated MSCs show altered miRNA profiles and harmful effects on heart cells, which provides new insight into the mechanism of radiation-related heart disease risks.


Background
In the 1960s, clinicians first recognized that cardiovascular complications occurred in patients who underwent radiotherapy (RT) for chest tumors [1]. Subsequent studies involving patients who received relatively high thoracic RT doses demonstrated an excess risk of radiation-induced heart disease (RIHD) [2][3][4]. The incidence of RIHD declined considerably with decreased cardiac radiation exposures using modern RT techniques (intensity-modulated RT, image-guided RT, etc.) [5]. Indeed, Darby and van den Bogaard et al. identified that the mean heart doses linearly correlated with the risk of RIHD [5,6]. However, the minimum threshold dose remained unclear. Epidemiological data revealed increased heart disease risks in atomic bomb survivors with individually estimated doses over 0.5 Gy [7]. In addition, individuals with other traditional risk factors (e.g., hypertension, hyperlipidemia, diabetes mellitus, smoking), or at a young age, seemed to be more vulnerable to developing RIHD [8]. The rising radiation exposure potentiality of medical imaging, galactic cosmic, or terrestrial ionizing resources makes it reasonable to worry about RIHD.
The clinical presentations of RIHD are currently well documented, including coronary artery atherosclerosis, pericarditis, cardiomyopathy, and conduction defects [9]. RIHD usually takes years or decades to manifest, making it challenging to interpret the underlying cellular and molecular mechanisms. Currently, ionizing radiation causes endothelial dysfunction and inflammatory responses, preceding the development of atherosclerosis, cardiac fibrosis, and tissue remodeling [9]. Sustained DNA lesions, oxidative stress, mitochondrial dysfunction, epigenetic regulation, and telomere erosion are also related to the development of RIHD [10][11][12]. Notably, these molecular changes interact with each other and act diversely in different types of heart cells. Therefore, the pathophysiological mechanisms of RIHD are unclear.
Classical radiobiology identifies that cells with high proliferative rates and immature features are more susceptible to ionizing radiation. Cardiomyocytes, the primary cell type in the heart, are postmitotic and incapable of proliferating. Hence, the dogma heart, as a radioresistant organ, has lasted for a long time. However, apart from cardiomyocytes, other cells such as microvascular endothelial cells, fibroblasts, and recently identified mesenchymal stem cells (MSCs) [13] also reside in the heart. Although there is no consensus on the number of MSCs in various tissues/organs, rare resident MSCs are known to play a critical role in maintaining tissue homeostasis. Following myocardial infarction, cardiac CD45 -CD44 + DDR2 + MSCs proliferated and exhibited typical characteristics with multipotent differentiation capacity and clonogenic expansion [14]. Previous studies identified that injuries to hematopoietic stem cells (0.01 to 0.05% in the bone marrow) after radiation exposure are acknowledged to contribute to a future increased leukemia risk [15]. We have recently demonstrated that cardiac MSCs are highly radiosensitive [16,17], and whole-body irradiation with 3 Gy γ-rays impairs the endogenous regeneration of infarcted mouse hearts [18]. Thus, MSCs theoretically reveal higher radiosensitivity than other mature heart tissue cells such as endothelial cells and cardiomyocytes. In response to different stimuli, MSCs release abundant extracellular vesicles (EVs), which are essential mediators of intercellular communication [19]. Hypoxia-primed bone marrow MSCs promoted cardiac function in a mouse model of myocardial infarction via upregulated EV miR-125bmediated cell protection [20]. Thus, we hypothesize that radiation exposure alters the secretion of EVs from MSCs, which subsequently initiates/triggers the damage of other heart tissue cells with less radiosensitivity.
By exposing human placenta-derived mesenchymal stem cells (hp-MSCs) to 5 Gy γ-rays, we investigated the radiation-induced change in the secretion of EVs from hp-MSCs in this study. We further evaluated the potential role of EVs from irradiated hp-MSCs in regulating the survival and function of heart tissue cells in vitro.

Radiation exposure and EV isolation
hp-MSCs (passaged 2-5, 3×10 5 cells) were plated on a 10-cm dish. The next day, the culture medium was aspirated, and the cells were washed with PBS to remove the residual FBS. Fresh culture medium supplemented with 10 ml 10% exosome-depleted FBS (System Biosciences) was added. Then, the hp-MSCs were exposed to 0 or 5 Gy γ-rays at a dose rate of 1 Gy/min using a PS-3100SB γ-ray irradiation system with a Cs source (Pony Industry Co., Ltd. Osaka, Japan) [16].
After 48 h of incubation, culture medium (10 ml) from hp-MSCs that irradiated or not was collected for EV isolation as previously described with minor modifications [22]. Briefly, the culture medium was centrifuged at 300 g for 3 min, at 4°C and 2000 g for 30 min to remove cell debris and apoptotic bodies. The supernatant was ultracentrifuged at 4°C and 100,000 g for 120 min to collect EVs. Then, the pellet was washed with PBS and underwent another step of ultracentrifugation at 4°C and 100,000 g for 120 min to concentrate and purify EVs. Finally, the pellet was resuspended in PBS and passed through a 0.22-μm filter (Millex) for further experiments or stored at −80°C. EVs isolated from a 30-ml culture medium were pooled together as one sample for further experiments.

EVs imaging and size distribution
The image of EVs (n=1) was taken by a transmission electron microscope. Briefly, 5 μl of EVs was dropped on a copper net and incubated at room temperature (RT) for 5 min. Then, excess liquids were removed by filter paper. Five microliters of 1% phosphotungstic acid was added to the copper mesh and incubated for 1 min at RT. Excess liquids were also removed by filter paper. Deionized water was added to the copper mesh to remove excess dye solution. EVs were observed under an electron microscope after drying at RT.
The size distribution of EVs (n=1) was estimated by nanoparticle tracking analysis using a Particle Metrix Zeta view.

EV protein preparation and Western blotting
The protein concentration of EVs (n=4) was tested by Micro BCA protein analysis as described in the instructions (Thermo Scientific Pierce 23235). Expression of the EVs markers CD63 and TSG101 (System Biosciences) was verified by Western blot analysis (n=1).

EV miRNA sequences
EVs from three donors of hp-MSCs were pooled for miRNA sequences. EV miRNAs were analyzed using gene chip miRNA 4.0 by Filgen Company. Briefly, EV miRNAs were extracted using the miRNeasy® Serum/ Plasma Kit. The extracted miRNAs (150 ng) were further concentrated by Micro Vac TM , and their volume was adjusted to 8 μl using nuclease-free water. Then, the hybridization solution was prepared by mixing the hybridization master mix with a biotin-labeled sample according to the manual (FlashTag™ Biotin HSR RNA Labeling Kit for GeneChip™ miRNA Arrays). The array that added hybridization solution was incubated in the GeneChip™ Hybridization Oven 645 for 18 h (48°C, 60 rpm). Later, the array was washed using GeneChip™ Fluidics Station 450. Finally, the array was screened by the GeneChip™ Scanner 3000 7G and analyzed according to the GeneChip™ Command Console (AGCC) 4.0 User Manual.

Uptake of EVs
EVs (n=1) derived from non-irradiated (non-irradiated-EVs) and irradiated hp-MSCs (irradiated-EVs) were labeled with the PKH26 Red Fluorescent Cell Linker Kit (Sigma-Aldrich) according to the manufacturer's protocol with minor modifications. Non-irradiated or irradiated-EVs diluted in PBS were added to 1 ml diluent C (Sigma-Aldrich). In parallel, 4 μl PKH26 dye was added to 1 ml diluent C and incubated with the EV solution (10 μg/ml) for 4 min. To bind excess dye, 2 ml 0.5% BSA/PBS was added. The labeled EVs were washed at 100,000 g for 1 h, and the EV pellet was diluted in 100 μl PBS and used for uptake experiments. PKH26 labeled non-irradiated or irradiated-EV were cultured with the HUVEC and H9c2 cell lines, respectively. Images of EV uptake were taken after coculture at 3 and 24 h using confocal microscopy.

Evaluation of cell proliferation and DNA damage
To evaluate the effects of EVs on cell proliferation and DNA damage, HUVEC (4×10 4 ) and H9c2 cells (4×10 4 ) were seeded on 4-well chamber culture slides. After 72 h of culture with 10 μg/ml non-irradiated-EVs or irradiated-EVs (n=3), the cells were washed with PBS and fixed in 4% paraformaldehyde for 10 min. After the incubation of protein block serum-free (DAKO) with 0.01% Triton X-100, the cells were incubated with Ki67 monoclonal antibody (SolA15, Invitrogen), anti-53BP1 antibody (ab36823, Abcam), or anti-gamma H2A.X (ab2893, Abcam), followed by associated Alexa flour 488-conjugated second antibody. Nuclei were stained with DAPI, the positively stained cells were counted under fluorescence microscopy at 200-fold magnification, and 20 fields per section were randomly selected for quantitative counting. The percentage of positive cells in each field was calculated as (positively stained cells/all cells in the field)×100%.

Annexin-V flow cytometry
To evaluate the effects of EVs on HUVEC and H9c2 cell apoptosis, HUVEC (3×10 5 ) and H9c2 (3×10 5 ) cells were seeded on 10-cm culture dishes. After 48 h of culture with 10 μg/ml non-irradiated-EVs or irradiated-EVs (n= 3), the cells were collected and washed with cold D-PBS by centrifugation for 5 min at 500×g at 4°C. Cells treated with 3% formaldehyde in a buffer for 30 min were included as a positive control. The cell pellets were suspended with 100 μl cold D-PBS and then added 5 μl of Annexin V-FTIC solution, and 2.5 μl dissolved PI were added as described in the manual (Beckman Coulter). The samples were kept on ice and incubated for 10 min in the dark. Finally, 400 μl ice-cold 1× binding buffer was added to the samples for further experiments. Flow cytometry analysis was performed using a FACSCalibur (Becton Dickinson, Franklin Lakes, NJ, USA). The acquired data were analyzed using Cell Quest software (Becton Dickinson).

Tube formation
Corning® Matrigel® Matrix (356230) was thawed overnight on the ice at 4°C according to the guidelines in the manual. All pipets and procedures were previously kept on ice. Then, 289 μl chilled Corning Matrigel® matrix into 24-well culture plates to avoid air bubbles. Plates were incubated at 37°C for 30-60 min. The medium remaining was removed carefully without disturbing the matrix layer, and the plates were ready to use. HUVEC cells were previously cocultured with 10 μg/ml nonirradiated-EVs or irradiated-EVs for 48 h (n=3). Then, 300-μl cell suspensions were collected and added to each well and incubated at 37°C, in a 5% CO 2 atmosphere. Tube formation was observed 3 h later under microscopy at 100-fold magnification and 10 fields per section were randomly selected for quantitative counting. The photo was further analyzed by ImageJ.

Intracellular calcium detection
Intracellular calcium was examined by loading H9c2 cells with Fluo 3 (Dojindo Molecular Technologies, Inc.) according to the instructions. Briefly, 3×10 5 H9c2 cells were previously seeded on 10-cm culture dishes and cocultured with 10 μg/ml non-irradiated-EVs or irradiated-EVs for 48 h (n=3). Cells were harvested and then plated on 96-well plates. The culture medium was carefully removed without injuring the cells. Cells were washed gently with PBS and then incubated with loading buffer at 37°C for 1 h. Loading buffer was removed carefully, and a warm recording medium was added. Finally, the fluorescence was examined by a multifunctional microplate detector.

Statistical analysis
All the results are presented as the mean ± SD. The statistical significance was determined by one-way ANOVA followed by Turkey's multiple comparisons test (Graph-Pad Prism). Differences were considered significant when P < 0.05.

Characterization of hp-MSCs and hp-MSC EVs
Primarily expanded hp-MSCs exhibited a fibroblast-like morphology (Fig. 1A) and were identified as the biological properties of MSCs according to their expression pattern on the cell surface markers CD44, CD105, CD90, CD73, CD45, and CD34 ( Fig. 1 B, C). To investigate the impact of IR on the EVs secretion, hp-MSCs (passaged 2-5) were exposed to 5 Gy γ-rays and the medium was collected 48 h later for EVs isolation by ultracentrifugation. The successful isolation of EVs was confirmed by electron microscopy (Fig. 1 D), nanoparticle track analysis (Fig. 1E), and Western blot analysis of the expression of membrane markers of CD63 and TSG101 (Fig. 1F). The size distribution (Fig. 1E) and protein concentration (Fig. 1G) were not obviously different between the EVs from the non-irradiated and irradiated MSCs. These data indicated very limited changes in the amount and size distribution of EVs from hp-MSCs within 48 h after exposure to 5 Gy γ-rays.

EVs from non-irradiated and irradiated hp-MSCs exhibited different miRNA expression profiles
We further measured the expression of miRNAs in EVs by gene chip miRNA 4.0. In contrast to the small changes in the secretion amount and size distribution, the analysis of miRNA sequences indicated that many miRNAs in irradiated MSC-EVs were upregulated or downregulated more than two-fold relative to the levels of the miRNAs of non-irradiated hp-MSCs (Fig. 1H). The top 20 miRNAs of upregulation (Table 1) or downregulation (Table 2) in EVs from irradiated hp-MSCs were listed in the table. Only miR-4655-5p was upregulated over three-fold, but miR-4443, miR-7110-5p, miR-520g-3p, miR-382-5p, miR-424-3p, miR-3197, and miR-6824-5p were downregulated over three-fold in irradiated hp-MSC EVs.

Uptake of EVs by HUVEC and H9c2 cells
Then, we evaluated the biological effects of EVs from the non-irradiated and irradiated hp-MSCs on endothelial cells and cardiomyocytes. By culturing HUVEC and H9c2 cells with the supplement of PKH26-labeled EVs (10 μg/ml) in the medium, the uptake of EVs by cells was observed using a confocal microscope. Red fluorescence was clearly detectable in the cytoplasm at 3 h and further enhanced after 24 h (Fig. 2 A, B). However, the uptake of EVs from either non-irradiated or irradiated hp-MSCs was quite similar by HUVEC and H9c2 cells, demonstrating that EVs from non-irradiated or irradiated hp-MSCs could be internalized by the HUVEC and H9c2 cells.

EVs from irradiated hp-MSCs significantly impaired the survival of HUVEC and H9c2 cells
The impact of non-irradiated and irradiated hp-MSC EVs on cell proliferation and DNA damage was observed after 72 h of coculture using an immunofluorescence assay (Fig. 3A, D). Ki67 expression in HUVEC and H9c2 cells was increased by non-irradiated hp-MSC EVs but decreased by irradiated hp-MSC EVs (Fig. 3B, C). Due to the different specificities of antibodies between HUVEC and H9c2 cells, DNA damage in cells was detected by the formation of 53BP1 or γ-H2AX foci in nuclei. EVs   Non-irradiated-EVs EVs isolated from conditioned medium of non-irradiated hp-MSCs Irradiated-EVs EVs isolated from conditioned medium of irradiated hp-MSCs from non-irradiated hp-MSCs significantly decreased the percentages of cells with 53BP1 or γ-H2AX foci compared with the control (Fig. 3E, F). However, the percentages of cells with 53BP1 or γ-H2AX foci were significantly increased by EVs from irradiated hp-MSCs compared with the non-irradiated hp-MSCs (Fig. 3E, F). All these results indicated that irradiated hp-MSC EVs impaired the proliferation and induced DNA damage in HUVEC and H9c2 cells. Cell apoptosis was also evaluated after 48 h of coculture using an Annexin-V flow cytometry assay (Fig. 4A,  D). The non-irradiated hp-MSC EVs significantly protected the HUVEC and H9c2 cells from apoptosis ( Fig. 4 B, E). However, the apoptosis of HUVEC and H9c2 cells was less decreased by EVs from irradiated hp-MSCs than by EVs from non-irradiated hp-MSCs (Fig. 4B, E). Additionally, the necrosis of HUVEC cells was less decreased by EVs from irradiated hp-MSCs than by EVs from nonirradiated hp-MSCs (Fig. 4 C). However, neither the non-irradiated nor the irradiated hp-MSC EVs affected the necrosis of H9c2 cells (Fig. 4 F). In contrast to nonirradiated hp-MSC EVs, irradiated hp-MSC EVs showed a very poor ability to protect HUVEC and H9c2 cells from apoptosis.

EVs from irradiated hp-MSCs revealed functional impairments to HUVEC and H9c2 cells
To evaluate the potential roles of EVs from irradiated hp-MSCs in cell function, we observed HUVEC cells tube formation (Fig. 5A) and calcium transient in H9c2 cells (Fig. 5D) 48 h after coculture. The tube formation of HUVEC cells was significantly increased by EVs from non-irradiated hp-MSCs but slightly decreased by EVs from irradiated hp-MSCs (Fig. 5B, C). However, the calcium transients of H9c2 cells were significantly increased by EVs from irradiated hp-MSCs, but were not changed significantly by EVs from non-irradiated hp-MSCs (Fig. 5D). These results indicated the functional impairment of irradiated hp-MSC EVs to HUVEC and H9c2 cells.

Discussions
Emerging findings have identified the contribution of stem cell injury to radiation-induced tissue toxicity [23,24]. EVs seem to be essential mediators of communication between MSCs and heart cells [25]. Resident MSCs in the heart are known to play an essential role in cardiac homeostasis [26], and their dysfunction may contribute to heart disease development [27]. Thus, following radiation exposure, the relatively highly radiosensitive MSCs may secrete specific EVs to induce injury to heart tissue cells, including cardiomyocytes and endothelial cells, which ultimately develop heart disease. To verify our hypothesis, we exposed hp-MSCs to 5 Gy γ-rays and then evaluated how EVs from irradiated hp-MSCs affect the biological properties of HUVEC and H9c2 cells.
We successfully isolated EVs from non-irradiated or irradiated hp-MSCs culture medium using ultracentrifugation, which was confirmed by electron microscopy and the expressions of CD63 and TSG101. As we isolated EVs for experiments before the publication of MISE2018 [22], we missed to examine the ratio of MSC-derived EVs have been demonstrated to possess regenerative potential comparable to the regenerative potential of MSCs [28]. We next investigated the potential effects of EVs on heart tissue cells using HUVEC and H9c2 cells. Correspondingly, EVs from nonirradiated hp-MSCs clearly showed beneficial effects on Values are the mean ± SD. ns P > 0.5, *P < 0.5, **P < 0.1, ***P < 0.01, ****P < 0.001 cell proliferation, DNA damage, and cell apoptosis. In contrast, EVs from irradiated hp-MSCs revealed much less beneficial, and sometimes even harmful effects on HUVEC and H9c2 cells. We also aimed to identify whether EVs from irradiated hp-MSCs impair the functions of HUVEC and H9c2 cells. MSC-derived EVs promote angiogenesis by transferring signals to endothelial cells [29]. Tube formation of HUVEC cells was facilitated by EVs from non-irradiated hp-MSCs, but not irradiated hp-MSCs. The homeostasis of calcium transients is a crucial factor for maintaining normal cardiac rhythm [30]. Following ischemia/reperfusion injury, the internal levels of calcium in H9c2 cells increased in a time-dependent manner [31], and calcium overload in H9c2 cells may further accelerate reperfusion injury [32]. We found that calcium transients in H9c2 cells were significantly enhanced by EVs from only the irradiated hp-MSCs, indicating the induction of calcium overload in cells. All these data suggested the harmful, rather than beneficial effects of EVs from irradiated MSCs in heart cells.
This study has several limitations (but is not limited to) that need further addressing. First, the culture scale of cardiac MSCs makes it difficult for us to obtain enough EVs for further experiments. As a proof-ofconcept of study, we used MSCs from human placental tissues instead of the resident MSCs of the heart because of the availability of cell sources. MSCs are highly variable and heterogonous depending on source tissue and age. Therefore, it is of interest to estimate whether varied MSCs would respond differently to ionizing radiation. Second, hp-MSCs were exposed to only 5 Gy at a high dose rate (1 Gy/min) which is equivalent to the daily dose generally used in clinical radiotherapy for cancers. As the biological effects on cells vary greatly depending on the dose and dose rate of radiation [50,51], it is necessary to evaluate the quantity and quality of EVs from MSCs by exposing the cells to different doses and dose rates of radiation. Third, we analyzed only miRNAs, but radiation exposure may also change other components such as lipids, proteins, and lncRNAs in EVs. Otherwise, we did not try to further confirm the role of each upregulated or downregulated miRNA in EVs from irradiated MSCs to HUVEC and H9c2 cells. Most importantly, we examined our hypothesis using only in vitro cell models, and animal models that mimic the in vivo changes after IR are worth studying. Finally, the pathological development of RIHD is complex, influenced by comprehensive factors released from all cells that resided in the heart (cardiomyocytes, endothelial cells, fibroblasts, stem cells). Herein, the interaction between these cells needed further investigation.

Conclusions
Overall, although the number of EVs secreted from hp-MSCs was not changed by 5 Gy γ-ray exposure, EVs from irradiated hp-MSCs caused damage to HUVEC and H9c2 cells. Our preliminary data from an in vitro study demonstrated that EVs from MSCs may indirectly contribute to radiation-induced heart disease. Further studies, including interventional experiments and in vivo mouse models, are required to confirm our concept.