Identification and characterization of hADSC‐derived exosome proteins from different isolation methods

Abstract Exosomes are secreted into the extracellular space by most cell types and contain various molecular constituents, which play roles in many biological processes. Adipose‐derived mesenchymal stem cells (ADSCs) can differentiate into a variety of cell types and secrete a series of paracrine factors through exosomes. ADSC‐derived exosomes have shown diagnostic and therapeutic potential in many clinical diseases. The molecular components are critical for their mechanisms. Several methods have been developed for exosome purification, including ultracentrifugation, ultrafiltration, density gradient purification, size‐based isolation, polymer precipitation and immuno‐affinity purification. Thus, we employed four methods to isolate exosomes from the hADSC culture medium, including ultracentrifugation, size exclusion chromatography, ExoQuick‐TC precipitation and ExoQuick‐TC ULTRA isolation. Following exosome isolation, we performed quantitative proteomic analysis of the exosome proteins using isobaric tags for relative and absolute quantification (iTRAQ) labelling, combined with 2D‐LC‐MS/MS. There were 599 universal and 138 stably expressed proteins in hADSC‐derived exosomes. We proved that these proteins were potential hADSC‐derived exosomes markers, including CD109, CD166, HSPA4, TRAP1, RAB2A, RAB11B and RAB14. From the quantitative proteomic analysis, we demonstrated that hADSC‐derived exosome protein expression varied, with lipopolysaccharide (LPS) treatment, in the different isolation methods. Pathway analysis and proliferation, migration and endothelial tube formation assays showed varying effects in cells stimulated with hADSC‐derived exosomes from different isolation methods. Our study revealed that different isolation methods might introduce variations in the protein composition in exosomes, which reflects their effects on biological function. The pros and cons of these methods are important points to consider for downstream research applications.


| INTRODUC TI ON
Exosomes are a discrete population of small extracellular vesicles (EVs), 30-150 nm in size, and secreted into the extracellular space from most cell types. 1 Exosomes represent a mode of intercellular communication though they contain various molecular constituents, including DNA, RNA, proteins and lipids. 2,3 Exosomes may play a role in immune response, signal transduction, antigen presentation, metabolism and cancer development. [4][5][6][7] Exosomes from lung spheroid cell could attenuate and resolve bleomycin-and silica-induced fibrosis. 8 Exosomes from bone marrow-derived mesenchymal stem cells combine with atorvastatin pretreatment significantly improved cardiac function and promoted blood vessel formation. 9 Adipose-derived mesenchymal stem cells (ADSCs) are derived stromal cells originating from stromal-vascular fragments of adipose tissue, with promising therapeutic potential. 10 ADSCs can differentiate into a variety of cell types and secrete a series of paracrine factors that function in cell-to-cell communication, immunoregulation, angiogenesis, revascularization and tissue regeneration. [11][12][13] Some paracrine factors are secreted through exosomes. Various studies have demonstrated that ADSC-derived exosomes have diagnostic and therapeutic potentials in many clinical diseases. [14][15][16][17][18] Lipopolysaccharide (LPS) derived from gram-negative bacteria which stimulating immune cell activity and triggering the inflammatory response. Several studies showed that LPS stimulates growth factors production in mesenchymal stromal cells (MSCs). 19 LPS enhanced survival of engrafted MSCs through VEGF expression and protect MSCs against apoptosis. 20,21 In rat model, LPSpreconditioned MSCs transplantation can reduce apoptosis of myocardium and enhance cardiac function. 21 LPS known to induce exosomes release and exosomes from LPS-stimulated macrophages increase neuroprotection and functional improvement after ischaemic stroke. 22,23 ADSC paracrine levels of VEGF, FGF and EGF were also induced after LPS treatment and showed the therapeutic potential in acute lung injury. 24 However, in current study the molecular function is still not understood.
The molecular components of ADSC-derived exosomes are critical for their mechanisms of action. Exosome isolation and characterization are important for their application in biomedical sciences.
However, there is no gold standard procedure for exosome purification. Several methods have been developed, including ultracentrifugation, 25 ultrafiltration, 26 density gradient purification, 2 size-based isolation, 27 polymer precipitation and immuno-affinity purification. 4,28,29 The vesicle purity, yield and components may depend on the methods used. Thus, we analysed the protein components of hADSC-derived exosomes, obtained using different isolation methods.
In this study, we isolated exosomes from the hADSC culture medium using four methods, including ultracentrifugation, size exclusion chromatography, ExoQuick-TC precipitation and ExoQuick-TC ULTRA isolation. Following isolation, we performed quantitative proteomic analysis of the exosome proteins using isobaric tags for relative and absolute quantification (iTRAQ) labelling, combined with 2D-LC-MS/MS. Using these analysis techniques, we investigated the protein components in hADSC-derived exosomes, using different isolation methods. For culture media collection, cell culture was limited to eight passages. 1 × 10 6 hADSC cells were cultured in 10 mL serum-free media, with/without 1 μg/mL lipopolysaccharides (LPS, L3755, SIGMA), for 24 hours. The culture media (500 mL) were harvested and centrifuged at 300 g for 5 minutes to remove cells and cell debris. The supernatants were concentrated using the Amicon ® Ultra-15 (UFC900324, Merck-Millipore) and transferred into new tubes for further use.

| Ultracentrifugation (UC)
Exosomes were separated from cell culture media via multiple centrifugation steps, as per the method described by Lin et al. 30 Briefly, concentrated media were centrifuged at 2000 g for 20 minutes, then at 10 000 g for 30 minutes. The supernatant was harvested and centrifuged at 110 000 g for 60 minutes. The pellet was resuspended in phosphate-buffered saline and stocked for further use.
Briefly, concentrated media were centrifuged at 3000 g for 15 minutes. The supernatant was transferred into a new tube, and equal volumes of the ExoQuick-TC solution were added. After mixing, they were refrigerated at 4°C overnight, at least 12 hours, then centrifuged at 1500 g for 30 minutes. The supernatant was discarded, and pellet was resuspended in PBS and stocked for further use.

| ExoQuick-TC ULTRA isolation (TCU)
Exosomes were purified from cell culture media using, the ExoQuick-TC ® ULTRA EV isolation kit (EQULTRA-20TC-1, system Biosciences), according to the manufacturer's instructions. Briefly, the exosome pellet from the ExoQuick-TC was resuspended in 200 μL buffer B and equal volumes of buffer A were added. After mixing, they were loaded into a prepared purification column and incubated at room temperature on a rotating shaker for 5 minutes.
Exosomes were eluted by centrifuging at 1000 g for 30 seconds.

| qEV10 size exclusion column purification (qEV)
Exosomes were purified from the cell culture media using qEV10 size exclusion columns (iZON science), according to the manufacturer's instructions. Briefly, 10 mL of concentrated media was loaded into qEV size exclusion columns, followed by elution with phosphate-buffered saline (PBS). Five millilitres of each fraction was collected and quantified, with a spectrophotometer (Tecan Sunrise™). Based on the exosomes size distribution and protein level, the fractions containing the vesicles were pooled and concentrated using the Amicon ® Ultra-15 (UFC900324, Merck-Millipore).

| Nanoparticle tracking analysis (NTA)
Size determination and concentration measurements of hADSC exosomes were performed on NanoSight NS300 (Malvern Instruments, UK). All samples were diluted in PBS to a final volume of 1 mL. Following settings were set according to the manufacturer's software manual: camera used sCMOS mode and set camera level to

| Dynamic light scattering (DLS)
The size and zeta potential of exosomes was measured by dynamic light scattering coupled with a Zetasizer Nano ZS system (Malvern Instruments, UK). Briefly, exosomes samples from four methods were diluted to 1 mL of PBS and gently mixed to provide a homogeneous solution. The homogeneous solution was put in a disposable cuvette and transferred to a Malvern Clear Zeta Potential cell for the Zeta potential measurement. The data were analysed through Dispersion Technology Software V7.01 supplied by the Malvern Zetasizer Nano ZS. The mean particle diameter was calculated from the measured particle distributions, and polydispersity index (PdI) was given as a measure of the size ranges of the solution.

| Transmission Electron Microscopy (TEM)
The bi-lipid layer of exosomes was characterized with TEM, which was commissioned to MA-tek at Hsin-chu, Taiwan. Briefly, the iso-

| Exosome protein extraction and iTRAQ labelling
Exosome proteins were purified using the T-PER tissue protein extraction reagent (78510, Thermo Scientific). The protein samples were desalted using the Amicon ® Ultra-15 (Merck-Millipore) and quantified using the BCA protein assay (23225, Thermo Scientific Pierce). with iodoacetamide at 37°C for 30 minutes. After the protein samples were digested using sequencing grade modified trypsin (V511A, Promega), they were dried using SpeedVac. Next, the peptides were reconstituted in 10 µL iTRAQ dissolution buffer and mixed with 30 µL iTRAQ labelling reagents at RT for overnight (Applied Biosystems Inc, Foster City). iTRAQ-labelled samples were dried using SpeedVac for further analysis.

| 2D LC-MS/MS
The iTRAQ-labelled samples were analysed using the Q ExactiveTM HF mass spectrometer (Thermo Fisher, San Jose), coupled with the Thermo Scientific™ UltiMate™ 3000 RSLCnano HPLC System, which was commissioned to the clinical proteomics core laboratory of Chang Gung Memorial Hospital at Linkou, Taiwan. Briefly, the iTRAQ-labelled peptides were pooled and desalted using Sep-Pak C18 cartridges (Waters). The desalted peptides were dried using SpeedVac and re-suspended in 0.5% trifluoroacetic acid. The peptide mixtures were loaded onto a C18 column (EASY-Spray™) and separated using 0.1% formic acid solution, with varying amounts of acetonitrile (5%-80%). The top abundant fifteen precursor ions, within 375-1400 m/z scan range, were dynamically selected for further fragmentation in high collision dissociation (HCD) mode, with normalized collision energy set to 33 ± 1. In full MS scan, the resolution was set to 60 000 at m/z 200, AGC target to 3e6 and maximum inject time to 50 ms In MS/MS scan, the resolution was set to 15 000, AGC target to 5e4 and maximum injection time to 100 ms The release of the dynamic exclusion of the selected precursor ions was set to 20 seconds.

| Database search and protein quantification
The raw MS data were queried using the Mascot search algorithm

| Antibodies
The commercially available primary antibodies used in this study included the following: monoclonal rabbit anti-CD9 (92726, Abcam) and polyclonal rabbit anti-Rab7 (2094, cell signalling). The secondary antibodies used for Western blotting included HRP-conjugated goat anti-rabbit IgG antibodies purchased from GE Healthcare.
Exosomes (30 µg) from different isolation methods were added, then cultured for 24 hours. Finally, 10 µL of the WST-1 Assay Dye Solution was added to each well and the plate incubated in the cell culture incubator overnight. Finally, the plates were measured at 450 nm using a microplate reader.
Images were acquired using a light microscope. After 8 hours, images were acquired again. The migration areas were measured using the ImageJ software.

| Endothelial tube formation assay
The Angiogenesis Assay (K905-50, BioVision) was used to measure the tube forming ability of HUVEC upon hADSC-derived exosome treatment. 96-well plates were pre-coated with extracellular matrix gel. HUVECs (1 × 10 4 ) were seeded in the 96-well and treated with 30 µg of exosomes, from different isolation methods. After 6 hours, the culture media were removed and the cells washed with wash buffer. The cells were stained, with staining dye at 37°C for 30 minutes. The images were acquired using a fluorescence microscope.

| Statistical analyses
All data were processed using the GraphPad Prism 5 for Windows (version 5.01). Variables were presented as the mean ± standard deviation (SD). Pairwise comparisons were performed using the Student's t-test and represented with a P value. All statistical tests were two-tailed, and differences were considered significant at P < .05.

| Exosomes isolation and identification
Exosomes were isolated from 500-mL culture media. After different methods isolating, the exosome proteins yield was measured using absorbance at 280 nm and particles number was measured using NTA. The data showed that exosome proteins yield and particle number in UC group was higher than other groups ( Figure S1).
The identification of exosomes from different isolated methods bases on size and morphology. The size distribution of exosomes was measured using DLS and NTA ( Figure S2A,B), and the bi-lipid layer morphology of exosomes was display by TEM ( Figure S2C)

| Generation of hADSC-derived exosome proteomic data sets
To identify the protein components in hADSC-derived exosomes, we performed an iTRAQ-based quantitative proteomic analysis to analyse differentially expressed proteins from different isolation methods, including ultracentrifugation, qEV-10 size exclusion chromatography, ExoQuick-TC precipitation and ExoQuick-TC ULTRA isolation ( Figure 1). hADSC were treated, with or without LPS, and the culture media harvested for further analyses. Exosomes purified by different isolation methods were labelled with 4-plex iTRAQ reagents of varying masses (114-117). The experimental design is summarized in Table S1. The exosomes from ultracentrifugation and qEV-10 size exclusion chromatography were grouped (Group 1) and those from the ExoQuick-TC precipitation and ExoQuick-TC ULTRA isolation into another group (Group 2).
We identified 1461 proteins and quantified 1136 proteins in Group 1 via 2D-LC-MS/MS analysis. In Group 2, we identified 897 proteins and quantified 801 proteins in Group 2 (Table S2). There were 599 exosome proteins from the four isolation methods, belonging to many protein families (Figure 2A, Table S3 Annexins (ANXA 1-6), ribosomal protein, etc CD9 and Alix are common exosome markers only found in Group 1; conversely, CD81 and TSG101 were only found in Group 2. We used Western blotting to verify the MS analysis results. The data showed that CD9 was detected in Group 1 (UC and qEV) but not in Group 2 (TC and TCU).
Rab7 was detected in Group 1 and Group 2 ( Figure S3). These results consist with MS analysis (Tables S2 and S3).

| Characterization of hADSC-derived exosome proteins
Using Gene Ontology analysis, we showed the cellular distribution of 599 hADSC-derived exosome proteins ( Figure 2B), which included 23.4% cell part, 12.1% membrane, 29.8% organelle and 5.7% extracellular region. A majority of the biological processes were cellular and metabolic processes, biological regulation, response to stimuli, localization and signalling ( Figure 2C).

| Differences in the four kinds of isolation method
To investigate the differences in exosome component distribution down-regulated) and 311 in the TCU isolation method (205 up-and 106 down-regulated) ( Figure 3A and Table S4). There is a difference in the number of exosome proteins regulated, after LPS treatment, in the four isolation methods.
Next, we found 50 differentially expressed proteins, with twofold changes in both UC and qEV, but with different expression profiles ( Figure 3B, Table S5). In the TC and TCU groups, there were 49 proteins differentially expressed, with similar profiles. Surprisingly, there was almost no intersection of these two protein groups.
Through gene ontology analysis, we showed the pathways the differentially expressed proteins are involved in ( Figure 3C). In all groups, these differentially expressed proteins were involved in the integrin signalling pathway and inflammation. In UC, these differentially expressed proteins were also involved in the cholecystokinin receptor (CCKR) signalling pathway, oxidative stress response and angiogenesis. In qEV, these proteins were involved in the CCKR signalling pathway, EGF receptor signalling pathway and cytoskeletal regulation. In TC, they were involved in cytoskeletal regulation, angiogenesis, and EGF and PDGF receptor signalling pathways. In TCU, they were involved in cytoskeletal regulation, angiogenesis and the EGF receptor signalling pathway.
Together, these results indicated that the different isolation methods affected the protein expression profiles after LPS treatment.

| The biological function of LPS-induced hADSC-derived exosomes
Pathway analysis, using WST-1 proliferation, migration and endothelial tube formation assays, the biological function of LPSinduced hADSC-derived exosomes, obtained from the different isolation methods, was evaluated. In the WST-1 proliferation assay, hADSC-derived exosomes were added to HUVEC, with/without LPS treatment. LPS-induced exosomes significantly increased cell proliferation in qEV and UC, but not in TCU ( Figure 4A). However, LPSinduced exosomes decreased cell proliferation in TC. In the wound healing migration assay, LPS-induced exosomes did not affect the cell migration ability, except in TC ( Figure 4B).
Next, we used the tube formation assay to evaluate angiogenesis. HUVECs were seeded in Matrigel-coated 96-well, and hADSCderived exosomes, with/without LPS, added. Through quantitative analysis of tube area and total tube length, LPS-induced exosomes significantly decreased tube formation ability in UC and TC ( Figure 4C). However, the tube formation ability of HUVECs increased after LPS-induced exosome treatment in qEV. In TCU, LPSinduced exosomes increased total tube length, but not tube area.
Together, these results indicated that LPS-induced exosomes affected cell proliferation, migration and tube formation. However, the results were inconsistent with exosomes isolated using different methods.

| D ISCUSS I ON
In this study, we used ultracentrifugation, size exclusion chromatography, ExoQuick-TC and ExoQuick-TC ULTRA precipitation to isolate exosomes from hADSC culture medium. Quantitative proteomic analysis was performed to identify and quantify the protein content in the exosomes. We showed 599 proteins, which belonged to the same protein family, in all four isolation methods. CD antigen and heat shock proteins are commonly used exosome markers. 6,31,32 Other protein families were also identified in the hADSC-derived exosomes such as RAB, proteasome proteins, Annexins and ribosomal proteins (Table S3). Through iTRAQ labelling, we found 138 proteins that were stably expressed in hADSC-derived exosomes, irrespective of the isolation method. These proteins are potential markers for hADSC-derived exosomes, including CD109, CD166, HSPA4, TRAP1, RAB2A, RAB11B and RAB14. Although some proteins are commonly used as exosome markers such as CD9, CD81, Alix and TSG101, we found CD9 and Alix only in Group 1 (UC and qEV); conversely, CD81 and TSG101 were only found in Group 2 (TC and TCU). We do not rule out the effect of experimental limitations on this result. Contamination in mass spectrometry analyses is a major problem, which leads to ion suppression and interferes with protein identification. 33,34 One of the most common contaminants is polyethylene glycol (PEG) in samples. 35 Exosomes were isolated via F I G U R E 3 Hierarchical cluster analysis of differentially expressed proteins in the hADSC-derived exosome. A, Proteins with a 2-fold change in the four isolation methods were compared in each group. B, Hierarchical clustering of exosome proteome was performed via unsupervised hierarchical classification, and distance trees were constructed from differentially expressed proteins in each group. The isolation methods are shown in columns and proteins in rows. The heat map scale of fold-change from −2 (green) to 2 (red). C, Pathway analysis of differentially expressed proteins in each group, based on universal Gene Ontology annotation terms precipitation using polymers in TC and TCU. We could not eliminate these polymers in sample preparation and they interfere with pro- This may explain the varying effects observed in the functional assay.
Exosomes are promising diagnostic, prognostic, therapeutic and drug delivery tools in clinical settings. Uniformity and quality are major challenges associated with exosome application. Many studies revealed the efficiency, yield and purity, size-distribution, RNA/ protein quality and miRNA composition of isolated exosomes. 28,37,38

F I G U R E 4
The effect of LPS-induced exosomes. HUVECs were treated with hADSC-derived exosomes, obtained with or without LPS treatment, followed by WST-1 proliferation (A), migration (B) and endothelial tube formation (C) assays. The results are presented as the means ±SDs; * indicates significance P < .05, ** indicates significance P < .01, as assessed by the Student's t-test These reports revealed that different isolation methods introduced variations in exosomal component distribution. Each method has certain benefits and drawbacks such as purity, isolation scale and desired ease for a particular application. Through our quantitative proteomic analysis, we provided a set of hADSC-derived exosomal marker proteins, which were independent of the isolation method. It may be beneficial to identify exosomes and develop affinity chromatography for their isolation.

| CON CLUS ION
In conclusion, our study revealed that different isolation methods might introduce diversity in the protein composition of exosomes, which reflects their various effects on biological function. We focused on the protein composition in this study. The DNA, RNA and lipid contents and amount varied, with methodological differences.
The pros and cons of these methods are important points to consider for downstream research applications.