The Regenerative Potential of Amniotic Fluid Stem Cell Extracellular Vesicles: Lessons Learned by Comparing Different Isolation Techniques

Extracellular vesicles (EVs) derived from amniotic fluid stem cells (AFSCs) mediate anti-apoptotic, pro-angiogenic, and immune-modulatory effects in multiple disease models, such as skeletal muscle atrophy and Alport syndrome. A source of potential variability in EV biological functions is how EV are isolated from parent cells. Currently, a comparative study of different EV isolation strategies using conditioned medium from AFSCs is lacking. Herein, we examined different isolation strategies for AFSC-EVs, using common techniques based on differential sedimentation (ultracentrifugation), solubility (ExoQuick, Total Exosome Isolation Reagent, Exo-PREP), or size-exclusion chromatography (qEV). All techniques isolated AFSC-EVs with typical EV morphology and protein markers. In contrast, AFSC-EV size, protein content, and yield varied depending on the method of isolation. When equal volumes of the different AFSC-EV preparations were used as treatment in a model of lung epithelial injury, we observed a significant variation in how AFSC-EVs were able to protect against cell death. AFSC-EV enhancement of cell survival appeared to be dose dependent, and largely uninfluenced by variation in EV-size distributions, relative EV-purity, or their total protein content. The variation in EV-mediated cell survival obtained with different isolation strategies emphasizes the importance of testing alternative isolation techniques in order to maximize EV regenerative capacity.


Different isolation techniques generated EV preparations with variable protein content, relative purities, and expression of EV markers. Total protein. The protein content was variable across
the different AFSC-EV preparations (Fig. 2a). We found a similar protein content among the EV preparations obtained with UC, ExoQuick, and TEIR (UC vs. ExoQuick, p = 0.1; UC vs. TEIR, p = 0.5; ExoQuick vs. TEIR, p = 0.2), whereas, the EV protein content was significantly different between Exo-PREP and qEV isolations (p < 0.05).
Preparation purity. To define the purity of each preparation, we measured the ratio between the number of EVs obtained with each technique and the corresponding protein content. We did not observe a correlation between the total number of EVs isolated with each technique and the corresponding total protein content [p = 0.25, r = 0.6 (95% CI −0.56 to 0.97); Fig. 2b].
Expression of EV-related markers. AFSC-EV populations isolated with UC, Exo-PREP, ExoQuick, and TEIR had detectable levels of typical EV protein markers Hsp70, CD63, Flotillin-1, and TSG101 analyzed by Western Blot (Fig. 2c). qEV preparations only had detectable levels of CD63 protein expression. All AFSC-EV isolation techniques showed no evidence of residual cellular debris, as evidenced by a lack of H3K27me3 protein expression.
Scientific RepoRts | (2019) 9:1837 | https://doi.org/10.1038/s41598-018-38320-w AFSC-EV capability of regenerating injured lung epithelium is preserved by increasing the dose of administered EVs. To validate that the number of EVs is important for AFSC-EV biological effect, we investigated if increasing the dosage of the AFSC-EV preparation with the lowest concentration (qEV) could reduce the rate of cell death. We found that treatment with AFSC-EVs isolated by qEV reversed the rate of cell death back to control levels with >40% by volume (1.3 ± 1%, p = n.s. vs. control; Fig. 3e]. Moreover, when multiple fractions of qEV preparations were pooled (7)(8)(9)(10)(11), the cell death rate was significantly reduced to control levels (p = n.s. to control, p < 0.05 to nitrofen). We also investigated whether the expression of an EV related marker would correlate to the number of EV particles and to the cell death rate, and we did not find a significant correlation ( Supplementary  Fig. S1A,B). Lastly, we investigated if decreasing volumes of AFSC-EVs had the opposite effect observed at 10% by volume treatment by administering UC AFSC-EVs in decreasing doses from 10% to 1.25%. We observed that AFSC-EVs from 5%, 2.5%, and 1.25% by volume were able to rescue cell death back to control levels, though the rate of cell death showed an increasing trend that was not significantly different ( Supplementary Fig. S2A).
To confirm the impact of the isolation strategies on EV biological properties, we also compared the effect of the different preparations on cell migration with a scratch assay. Nitrofen injury significantly impaired the ability of A549 cells to migrate by approximately 55% (p = 0.002 compared to control; Supplementary Fig. S2B). Administration of AFSC-EVs from all preparations improved cell migration rate back to control levels (p = n.s. relative to control; p = 0.008 nitrofen vs. UC AFSC-EVs; p = 0.07 nitrofen vs. Exo-PREP AFSC-EVs; p = 0.04 nitrofen vs. ExoQuick AFSC-EVs; p = 0.02 nitrofen vs. TEIR AFSC-EVs; p = 0.04 nitrofen vs. qEV AFSC-EVs) with a beneficial impact that had a similar trend as observed in cell death rate.

Discussion
The present study shows that commonly used EV isolation techniques can isolate AFSC-EVs that have typical EV morphology and protein markers. However, AFSC-EV size, protein content, preparation purity, and number of EVs varied between different isolation techniques. We also noticed that AFSC-EVs isolated with different techniques had different effects on a model of damaged lung epithelium. Interestingly, among all the EV characteristics, the number of EVs administered was the most important parameter responsible for AFSC-EV regenerative potential.
The EV isolation techniques employed in this study are commonly used methods in the field. The variability observed with these techniques have also been reported by other authors evaluating isolation methods for EVs from other sources beyond AFSCs 30,31 . Although our experimental methods did not include an analytical technique to separate AFSC-EVs from conditioned medium based on EV size distributions, we found that the variable EV size distribution did not have a significant impact on AFSC-EV regenerative potential in an in vitro lung injury model. EVs represent a heterogeneous mixture of vesicles of different sizes released from prokaryotic and eukaryotic cells 32 . It is not fully understood how size contributes to biological function of EVs. Among all EVs, exosomes (~30 to 100 nm in size) 20,33 have been classically considered the subpopulation with potent, protective, and pathological functions, and therefore with higher potential as diagnostic or therapeutic tools 34 . For instance, Keerthikumar and colleagues demonstrated a stronger proliferative and pro-migratory effect of exosomes than larger EVs secreted by neuroblastoma cell lines 35 . However, some studies have shown that EVs of other sizes can be more effective than exosomes in a given biological context. For example, Minciacchi et al. reported that large EVs (1-10μm) were more efficient than small EVs secreted by prostate cancer cell lines in fibroblast reprogramming and endothelial cell tube formation 36 . In order to better address the significance of EV-sizes being able to impact their biological effect, focus has now shifted to developing technologies that isolate EV-preparations with smaller size distributions or that determine the relative proportion of EVs of different sizes more accurately in a given sample 34 .
In our study, size did not appear to correlate with AFSC-EV regenerative capacity as preparations with a similar mean EV size (ExoQuick, UC, and TEIR) had different biological effects in the lung injury model. Further suggestion that size is not essential for AFSC-EV potential comes from the dose curve experiment (Fig. 3e). When we treated injured lung epithelium with the smallest and least abundant EV preparation (qEV), we found little to no protection. However, when we increased the number of AFSC-EVs from the same preparation and matched the concentration of larger EV preparations (UC, ExoQuick, Exo-PREP), we observed a similar protective effect.
Additional evidence that specifically AFSC-EV-size does not dictate biological function can be found in the literature, where different isolation techniques led to collection of AFSC-EVs of variable sizes, but all equally effective in various experimental models ( Table 2)  As such it appears that EVs of different sizes may deliver beneficial regenerative effects.
Having established that EV size distribution was not crucial to AFSC-EV mediated biological responses, we next assessed the role of EV protein content. In our study, we found variability in the total protein content of AFSC-EVs, with a significant difference between the preparation with the highest protein content (Exo-PREP) and the one with the lowest (qEV) (Fig. 2b). The ratio between protein content and the number of EVs has been previously used as an indicator of EV purity 37 . In our study, ExoQuick had the lowest protein concentration with highest number of particles isolated in keeping with a high degree of EV-purity. Conversely, qEV was the preparation with the lowest protein content and the least number of particles, which might be explained by obligatory dilution of EV containing samples by SEC.
As also observed for EV size, protein content does not appear to influence the regenerative capability of AFSC-EVs. In fact, the dose curve experiment (Fig. 3e) showed that higher doses of qEV isolated AFSC-EVs had similar biological effects on a lung injury model as preparations with higher protein content. Lastly, we found variability in the total number of AFSC-EVs isolated using different techniques. This variability has also been reported in other studies, where different techniques were assessed to isolate other EV populations 31,38 . In particular in our study, the highest EV-yields were obtained with UC, ExoQuick and Exo-PREP. Interestingly, when we tested the effects of equal volumetric additions of AFSC-EVs isolated with different techniques on the in vitro model of lung injury, we noticed that only AFSC-EV preparations with a high particle yield (UC, ExoQuick and Exo-PREP) had a beneficial effect in reducing cell death. This finding suggested a possible dose effect related to the number of EVs isolated, as also supported by the negative correlation of EVs isolated with the cell death rate. To test dose responsiveness, we elected to titrate increasing amounts of the smallest-least protein containing-EVs (qEV) into the nitrofen lung injury model. We aimed to deliver doses of the qEV isolated EVs to obtain numbers of EVs which would resemble the other isolation techniques (Supplementary Table ST2). By adding 20% by volume of the qEV group, we approximated the TEIR group, while adding 60% by volume more closely resembled the UC, ExoQuick and Exo-PREP groups (Supplementary Table ST2). Increasing the number of qEV derived EVs increased the cell survival advantage in the lung injury model in a dose dependent fashion. We confirmed this response by pooling additional fractions eluted by qEV columns included more EVs, and obtaining a reduction on cell death rates similar to >40% treatment by volume. This confirmed that EV count is crucial for the biological effect of an EV preparation, regardless of the isolation technique used. Moreover, we elected to titrate decreasing amounts of UC AFSC-EV preparations into the nitrofen lung injury model, as this was one of the preparations that brought the rate of cell death back to control levels. Interestingly, when AFSC-EVs isolated using UC were administered in decreasing doses, we observed a trend towards an increase in cell death rate, albeit not significantly different.
Lastly, we investigated the effects of different AFSC-EV preparations on another outcome measure, such as cell migration. We specifically studied cell migration as an outcome measure of tissue homeostasis, as nitrofen exposure impairs lung epithelial cell migration 39 , and EV administration has been reported to be beneficial in cell culture 40 . Interestingly, we observed a positive effect on the ability of nitrofen-injured A549 cells to migrate with the addition of all AFSC-EV preparations to varying degrees ( Supplementary Fig. 2B).
The concept that EVs have variable dose-dependent effects has also been observed in other disease models [41][42][43] 43 . The concept of EV dose-dependent effect has lately become more topical, as we are approaching an era of EV based therapeutics,  that involves the consideration of how EV storage and stability affect EV biological functions 44 . In an effort to reduce variability in EV preparations, some authors have proposed to develop tools to estimate the efficacious EV dose, including fingerprinting assays and potency assay 45 . EV dose could be quantified by fingerprinting assays using surrogate indicators, such as EV markers or microRNAs, and/or by potency assays assessing the ability of a preparation to elicit the desired biologic effect in vitro or in vivo 45 . Although standardization of EV isolation is unlikely to be achieved due to variables in study design and starting material, there is great value in reporting detailed descriptions during comparative studies as described by the broader EV community [46][47][48] . The present study confirms that AFSC-EVs offer potential as beneficial effectors in regenerative medicine. In recent years, EVs have gained significant interest in the field of regenerative medicine, as they exert an effect that is similar and sometimes greater than that of their parent cells 49,50 . To date, the effects of AFSC-EVs have been investigated mainly in models with translational potential towards clinical application, such as chemotherapy-induced ovarian failure 23 , skeletal muscle atrophy 24 , and Alport syndrome 6 . In these studies, AFSC-EVs have shown immunomodulatory 22,26 , proangiogenic 6,25 , antiapoptotic 23,24 and anti-inflammatory effects 24,25 . In the present study, we have used an in vitro model of lung injury as a platform to test different EV isolation techniques. Interestingly, we have also observed for the first time that AFSC-EVs have a beneficial anti-apoptotic effect on injured respiratory epithelium. It has been reported that AFSCs hold regenerative potential in the lung, as they can integrate and differentiate into epithelial lung lineages 9 , reduce lung fibrosis 51 , and repair damaged alveolar epithelial type 2 cells 52 , which are important for surfactant production 53 . Moreover, AFSCs have a reparative effect in nitrofen-mediated models of pulmonary hypoplasia via an undetermined paracrine mechanism 10,54 . In the present study, using an established nitrofen-mediated model of lung injury, we have shown for the first time that AFSC-EVs are the paracrine effector of AFSCs in lung disease. This was confirmed by the lack of beneficial effect observed when EV-depleted AFSC-CM was used.
In conclusion, techniques that are based on differential sedimentation, solubility or size-exclusion chromatography are able to isolate AFSC-EVs with typical EV morphology and protein markers. The variability observed in EV size and protein content did not significantly affect AFSC-EV biological function. Conversely, EV count influenced AFSC-EV beneficial effect on a model of lung injury, in a dose dependent fashion. We advise other investigators working with EVs to consider EV concentration as a variable that could have an impact on AFSC regenerative potential.
Moreover, in the present study, we have shown for the first time that AFSC-EVs exert a beneficial effect in an in vitro model of lung injury. This adds to the current literature, where AFSC-EVs have been reported to hold great potential in a variety of disease models. Further studies are needed to understand the beneficial effect that AFSC-EVs may exert in the context of lung disease.

Materials and Methods
Isolation of AFSC-EVs from cell culture. AFSC cell culture. Amniotic fluid stem cells (AFSCs) were isolated from a pregnant rat at embryonic day E12 by c-Kit+ selection as previously described 2 . In brief, AFSCs were grown to 90% confluence in a humidified 37 °C, 5% CO 2 chamber in alpha-minimal Essential Media (αMEM, Gibco, ThermoFisher, Waltham, MA) supplemented with 20% Chang supplements (Irvine Scientific, Santa Ana, CA), 15% fetal bovine serum (FBS, ThermoFisher Scientific, Waltham, MA), and 0.5% Penicillin/Streptomycin (ThermoFisher Scientific, Waltham, MA). AFSCs were cultured for 18 hours in 7.5% exosome-depleted FBS (ThermoFisher, Waltham, MA) in αMEM. For each of the EV isolation methods described below, 2 mL of AFSC-CM from 4 × 10 6 AFSCs was used. To eliminate the possibility that bacterial contaminants contributed to EVs in the conditioned medium, we confirmed that AFSCs were mycoplasma free (PCR Mycoplasma Detection Kit, Richmond, BC).
Ultracentrifugation isolation of AFSC-EVs. Residual cells and debris were cleared from AFSC-CM by differential centrifugation at 300 × g followed by 1200 × g both for 10 minutes at room temperature. Next, the supernatant was filtered (0.20 µm cellulose filter, Corning, NY) then ultracentrifuged at 100,000 × g for 14 h at 4 °C (swing bucket rotor on minimum acceleration/break setting, SW 32Ti Beckman Coulter, Brea, CA). Post-ultracentrifugation the pellet was resuspended in 500 μL of phosphate buffered saline and either used immediately or stored at −20 °C for up to six months. Supernatants obtained from the ultracentrifugation was used as EV-depleted CM.
Reagent-based isolation of AFSC-EVs. TEIR (ThermoFisher Scientific, Waltham, MA), ExoQuick-TC (System Biosciences, Palo Alto, CA) and Exo-PREP (HansaBioMed, Basel, Switzerland) kits were used as per the manufacturers' recommended protocols. The AFSC-EVs preparations were re-suspended in 500 μL of phosphate buffered saline and either used immediately or stored at −20 °C for up to six months.
Size-exclusion chromatography isolation of AFSC-EVs. Size-exclusion chromatography (SEC) was used to isolate EVs from pre-cleared AFSC-CM using qEV (IZON, Cambridge, MA) as per manufacturer's protocol. In all experiments, Fraction 7 was collected and used for subsequent EV characterizations and experiments as per manufacturer's recommendations. The AFSC-EVs preparations were either used immediately or stored at −20 °C for up to six months.
Characterization of AFSC-EVs isolated by different methods. AFSC-EVs obtained using the different isolation strategies ( TEM. AFSC-EV morphology was assessed using TEM. For TEM EV-analyses, AFSC-EV preparations from each isolation strategy were mixed with equal volumes of 4% paraformaldehyde in PBS (Electron Microscopy Sciences, Hatfield, PA), and adhered to formvar-carbon coated copper grids described below. These grids were formed with 0.5% formvar solution (Electron Microscopy Sciences, Hatfield, PA) from powder in ethylene dichloride. A glass slide was dipped in the formvar, allowed to dry, its edges were scored then pushed into a water bath to float the film off the slide. The grids were then placed on the film, and flipped out of the water with parafilm. The sheet of formvar coated grids were then placed in a Cressington carbon evaporator where carbon was applied. EV preparations were fixed again with 2% glutaraldehyde in PBS, contrasted with uranyl-oxalate (Electron Microscopy Sciences, Hatfield, PA) and embedded in methyl cellulose-UA. AFSC-EVs were visualized on a Tecnai 20 (FEI, Hillsboro, OR) from 25 kx to 100 kx magnification.
Nanoparticle Tracking Analysis. To determine EV sizes, 300 µL of AFSC-EVs (undiluted EVs from the different isolation strategies) were analyzed by the NanoSight LM10 system (NanoSight Ltd, Salisbury, UK). 30-second videos of EVs (number of experiments n = 3, number of videos n = 8) were collected, averaged, and analyzed using LM10 NTA equipped with a 65 mWatt 405 nm violet laser (NanoSight Ltd, Salisbury, UK). For size calibration, 100 nm polystyrene beads (Malvern Instruments, Saint-Laurent, Canada) were used as previously described 55 . At capture with sCMOS camera on NTA 3.1 (machine) Build 3.1.46 (software), the temperature was 22 °C.
Protein analysis and relative purity of EV preparations. To investigate the total protein yield of the different AFSC-EV preparations, we used the Pierce Bradford assay (ThermoFisher Scientific, Waltham, MA). To estimate the relative purity of AFSC-EVs obtained from the different isolation strategies, the ratio of particle counts measured by NTA (number of EV particles/mL) divided by protein concentration (µg/µL) measured using the Pierce Bradford assay was compared, as previously described 37 .
Regenerative potential of AFSC-EVs in a model of lung injury. Immortalized adenocarcinomic alveolar basal epithelial cells (A549; Sigma Aldrich, St. Louis, MO) were grown until confluence in Dulbecco Modified Eagle Medium F-12 (DMEM, ThermoFisher Scientific, Waltham, MA) supplemented with 10% FBS and 0.5% Penicillin/Streptomycin (ThermoFisher Scientific, Waltham, MA). A549 cells were injured with nitrofen (40 μM in DMSO; Sigma Aldrich, St. Louis, MO) for 18 h. After this time, select groups of injured A549 cells were then treated with 10% by volume of AFSC-EVs in DMEM from the different isolation methods. Control groups included: uninjured and untreated (no nitrofen, no AFSC-EVs, DMEM only) A549 cells and injured but untreated (nitrofen only, no AFSC-EVs) A549 cells. After 24 hours, cell death rates were assessed in all groups. Live A549 cells were identified by calcein staining (1 μM) while dead cells were identified with ethidium staining (3 μM) (Live/Dead TM cytotoxicity kit, ThermoFisher Scientific, Waltham, MA). Five 20X magnification fields of cells were imaged using an inverted fluorescence microscope (Leica DMI6000B, Wetzlar, Germany) per replicate were assessed by two independent researchers, and averaged to identify the percentage of dead cells.
To assess whether different isolation techniques had an effect on the AFSC-EV regenerative potential based on EV characteristics such as size and yield, we correlated these two with the rate of cell death.
Finally, we performed a dose analysis of the AFSC-EV isolated using qEV by titrating 10%, 20%, 40%, and 60% of the preparation by volume in DMEM as treatment to nitrofen-injured A549 cells. We also administered decreasing doses of AFSC-EVs isolated using UC at 10%, 5%, 2.5%, and 1.25% by volume treatment. Live/Dead TM cytotoxicity assay was performed as described above after 24 h treatment with increasing doses of AFSC-EV isolated with qEV and decreasing doses of AFSC-EV isolated with UC.
To test the ability of AFSC-EV in promoting cell migration, an artificial would was created in cell culture wells on nitrofen-injured cells with a sterile pipette tip. Nitrofen-injured cells were then treated with AFSC-EVs isolated from the different methods at 10% by volume. Control cells did not receive the nitrofen injury prior to the formation of an artificial wound. Cells were incubated with Hoechst 33342 (1:2000 in PBS; ThermoFisher, Waltham, MA) for 10 minutes at room temperature. Immunofluorescence imaging was conducted after 6 hours and 12 hours to assess the rate of cell migration in each experimental group in two biological replicates. At least n = 25 fields at the cell front were taken per condition.
Scientific RepoRts | (2019) 9:1837 | https://doi.org/10.1038/s41598-018-38320-w Statistics. Comparisons between groups were conducted using Kruskal-Wallis one-way analysis of variance with Dunn's multiple comparison test. Data are reported as mean ± standard deviation. For correlation analysis, we calculated the Pearson correlation coefficient and reported the p-value, Pearson r, and 95% confidence interval. All experiments were performed at least three times. All statistical tests were performed using GraphPad PRISM Version 6.0e.