Role of Collection Media on the Biological Activity of Extracellular Vesicles From hTERT‐Immortalised Mesenchymal Stromal Cells

ABSTRACT

Extracellular Vesicles (EVs) from Mesenchymal Stromal Cells (MSCs) are promising cell‐free therapeutics due to their ability to modulate immune responses, promote tissue repair and replicate many benefits of their parental cells. However, developing standardised, clinically translatable MSC‐EV preparations remains difficult because of biological and technical variability. Factors such as cell source, donor differences, passage number and culture conditions affect EV yield and function, while enrichment methods, purity and media composition further complicate interpretation of EV‐specific effects. Although prior studies have explored isolation methods and serum contaminants, the impact of different collection media on EV bioactivity is poorly understood. Here, we used hTERT‐immortalised MSCs, which preserve the therapeutic properties of primary MSCs, to minimise cellular variability. EVs were collected in four different xeno‐free media, enriched via Tangential Flow Filtration (TFF) and assessed in terms of their anti‐inflammatory, anti‐fibrotic, gap closure and proliferative potential compared with unconditioned, TFF processed collection media. Our results demonstrate that collection media significantly affect MSC‐EV biological and functional properties, highlighting the need to carefully select media for standardised EV production. This study advances standardisation in EV research and supports the development of consistent, clinically relevant MSC‐EV products.

1. Introduction

Extracellular vesicles (EVs) are nanosized, membrane‐enclosed particles secreted by virtually all cell types, mediating intercellular communication through the transfer of proteins, lipids and nucleic acids (Théry et al. 2018; Welsh et al. 2024). EVs derived from mesenchymal stromal cells (MSCs) have attracted significant attention for their potential to reproduce many of the therapeutic effects of their parental cells, including immunomodulation, tissue regeneration and anti‐inflammatory activity (Jafarinia et al. 2020; Liang et al. 2014; Romanelli et al. 2019; Spees et al. 2016). EVs represent a promising, cell‐free, off‐the‐shelf, scalable alternative for all regenerative medicine approaches using living MSCs (Jafarinia et al. 2020; Kou et al. 2022). However, the translation of MSC‐EVs into standardised therapeutic products is limited by biological and technical variability (Chen and Li 2025).

The biological activity of EVs is influenced by multiple factors, starting with the heterogeneity of primary MSC populations due to differences in cell source, donor variability or population doubling levels of the producer cells (Giebel and Lim 2025; Ragni et al. 2019; Tertel et al. 2023). Using immortalised MSC lines offers a solution to this, as it provides stable, scalable and reproducible cell factories for EV production (Brancolini et al. 2026; Hindle et al. 2024; Lyu et al. 2023; Tong et al. 2025; Wolbank et al. 2009).

Furthermore, upstream and downstream bioprocessing, as well as methodological factors, critically determine purity and yield of the EV preparations (Dilsiz 2024; Kawai‐Harada et al. 2024; Yamashita et al. 2016). Among these, cultivation system, for example, 2D versus 3D culture conditions, bioreactors (Kusuma et al. 2022; Mas‐Bargues and Borrás 2021), as well as the harvest and enrichment strategies (Brennan et al. 2020; Dilsiz 2024; Yamashita et al. 2016) play a major role in defining the quantity and quality of the resulting EV preparations. In these technological questions, one critical source of variability lies in the composition of the expansion and collection media used during EV production. Media influence both the quantity and molecular composition of secreted EVs (Karttunen et al. 2022) as well as their measured bioactivity (Jimenez et al. 2023). Fetal bovine serum (FBS), even when used in its exosome‐depleted form (Chen and Li 2025; Pham et al. 2021; Shelke et al. 2014), can introduce bovine‐derived vesicles and soluble proteins that may interfere with EV isolation and their functionality (Angelini et al. 2016; Lehrich et al. 2018; Pham et al. 2021; Shelke et al. 2014; Urzì et al. 2022), and it also poses potential biosafety concerns (Lehrich et al. 2021; Urzì et al. 2022). Alternative supplements such as human platelet lysate (hPL) and chemically defined serum‐free formulations, have demonstrated substantial effects on EV yield and function (Cañas‐Arboleda et al. 2020; Gimona et al. 2017; Hemeda et al. 2014; Tancharoen et al. 2019; Zhang et al. 2025). Despite these advances, systematic comparisons of multiple expansion or collection media under identical conditions remain scarce (Karttunen et al. 2022; Zhao et al. 2022), and distinguishing EV‐specific bioactivity arising from soluble media components remains an unresolved challenge (He et al. 2022; Karttunen et al. 2022; Kronstadt et al. 2023). Although several studies have shown that medium‐derived vesicles, proteins or lipids can mimic or mask EV‐mediated effects in bioassays (Giebel and Lim 2025; Kronstadt et al. 2023), the inclusion of unconditioned medium controls processed identically to conditioned medium has rarely been implemented (Gimona et al. 2021; Nguyen et al. 2024). Therefore, we here compare how expansion and collection medium composition affect the biological activity of MSC‐derived EVs. Using hTERT‐immortalised Bone Marrow‐derived MSCs (BM‐MSC/TERT292), we compared EVs collected in four different combinations of commercially available xeno‐free media for MSCs expansion and/or EV enrichment. As controls, unconditioned media (UCM) were processed through the same TFF workflow and tested alongside EV samples. By integrating standardised functional assays, as well as process‐matched controls, and a stable MSC source, this work provides an integrated framework to evaluate how upstream conditions shape EV functionality and establishes methodological standards for reliable EV‐based therapeutic development.

2. Materials and Methods

2.1. Cells and Culture Conditions

2.1.1. Mesenchymal Stem Cell Maintenance and Adaptation to Different Expansion Media

BM‐MSC/TERT273 cells (Evercyte GmbH) were passaged twice a week following the protocols provided by the manufacturer upon reaching about 75%–90% confluence. Cell culture roux flasks were pre‐coated for 2 h at room temperature with fresh Animal component‐free cell attachment substrate (ACF, StemCell Technologies, 1:300 dilution). Mesencult ACF Plus medium (StemCell Technologies) was supplemented with Mesencult ACF Plus 500X supplement (StemCell Technologies), GlutaMAX‐I 100X (Gibco) and 200 µg/mL G418 (InvivoGen) and used for cell culturing. Cells were then washed twice with PBS (Gibco) and incubated with CTS TrypLE Select Enzyme solution (20 µL/cm², Gibco) at 37°C until complete detachment from the culture flasks. The cell suspension was then collected in Mesencult ACF Plus medium and centrifuged at 300 × g for 5 min. The cells were cultured in the pre‐coated flasks in a humidified environment with 5% CO₂ with a 1:4 to 1:6 split ratio. For cells adaptation to RoosterNourish, RoosterBasal 2.0‐CC (RoosterBio) medium was supplemented with either 1% or 2% (the manufacturer's protocol specifies 2%) RoosterBooster‐MSC‐XF (RoosterBio) and 200 µg/mL G418 (InvivoGen). After cell detachment, the cell suspension was then collected in Rooster Nourish MSC‐XF medium and centrifuged at 300 × g for 5 min. The cells were cultured in CellBIND tissue culture flasks (Corning) in a humidified environment with 5% CO₂ with a 1:1.5 to 1:4 split ratio in RoosterBasal medium supplemented with 1% RoosterBooster or with a 1:4 to 1:6 split ratio with 2% RoosterBooster.

2.1.2. Cell Line Maintenance

For conducting the anti‐inflammatory assay, BV‐2 murine microglial cells (IRCCS Azienda Ospedaliera Universitaria San Martino—IST) were propagated in a mixture of Dulbecco's Minimum Essential Medium Eagle (DMEM 1 g/L glucose, Sigma–Aldrich) and Dulbecco's Minimum Essential Medium Eagle (DMEM 4, 5 g Glucose/L, Gibco) at 1:1.9 ratio (final Glucose concentration of 2 g/L) supplemented with 8.8% FBS (PAN Biotech) and 100X Pen/Strep (Thermo Fisher Scientific, 10,000 U/mL). Cells were washed once with PBS (Gibco), detached with 0.05% Trypsin/EDTA (Thermo Fisher Scientific), and cell suspension was centrifuged at 170 g × 5 min. Resuspended cells were propagated at a seeding density of 6667 cells/cm² every second day or of 1333 cells/cm² every third day. Human foreskin fibroblasts fHDF/TERT166 (Evercyte GmbH) were cultured in DMEM/F12 (PAN Biotech), supplemented with 10% FBS (PAN Biotech) and 100 µg/mL G418 (InvivoGen). Cells were passed twice a week at a 1:4 split ratio using 0.05% Trypsin‐EDTA (Gibco). EBM (Lonza) supplemented with bovine brain extract (BBE), human epidermal growth factor (hEGF), hydrocortisone, ascorbic acid (components of the EGM SingleQuots supplement kit, Lonza), 10% FBS (PAN Biotech) and 20 µg/mL G418 (InvivoGen) was used to propagate telomerised human umbilical vein endothelial cells HUVEC/TERT2 (Evercyte GmbH). Cells were expanded twice a week on Gelatine (Sigma–Aldrich) pre‐coated flasks with a split ratio of 1:8 using 0.05% Trypsin‐EDTA (Gibco).

2.2. EV Preparation and Molecular Characterisation

2.2.1. Extracellular Vesicle Enrichment

For EV production, BM‐MSC/TERT292 cells were cultured for 48 h in G418‐free Mesencult ACF Plus medium (Stem Cell Technologies) and in G418‐free RoosterNourish‐MSC‐XF (RoosterBio) supplemented with 2% v/v of RoosterBooster‐MSC‐XF (RoosterBio). Cell density and viability were assessed using the Trypan blue exclusion method using a Cell‐Chip Hemocytometer (Bioswisstec) after the cells had reached a confluency of 70%–85% and the supernatant was collected. The conditioned medium was first centrifuged at 700 × g for 5 min at 4°C, followed by centrifugation at 2000 × g for 10 min at 4°C and supernatant and filtration through a 0.22‐µm PVDF‐filter to remove bigger particles. The supernatant was kept for up to 4 weeks prior to EV enrichment in aliquots at –80°C. Cells growing in Mesencult ACF Plus medium were further cultured for 48 h in OptiMEM (Gibco), while cells growing in RoosterNourish‐MSC‐XF medium were cultured for an additional 48 h in RoosterCollect‐EV (RoosterBio) before harvesting of conditioned medium and processing as described above. Hollow fibre columns with 300 kDa cut‐off (MIDIKROS 65CM 300K MPES 0.5 MM FLL X FLL, Repligen) were used for the first step of TFF enrichment of EVs from the four different collection media. The conditioned medium was concentrated down to 10% of the initial volume and washed abundantly with ice‐cold HEPES (Sigma–Aldrich, 20 mM in cell grade H₂O and 0.22 µm filtered) using twice the diafiltration volume of the conditioned media. Following TFF enrichment of the conditioned medium using Midi hollow fibre columns, the residual 30 mL volume was further processed using 300 kDa cut‐off Micro hollow fibre columns (MicroKros 20CM 300K MPES 0.5 MM MLL X FLL, Repligen). TFF‐processed samples were alternated with ice‐cold HEPES washes throughout the process and the samples (from 30 mL starting volume) were concentrated down to 1–2 mL and stored in aliquots at –80°C until further analysis. For the medium control used in the biological assays, 50 mL of unconditioned Mesencult ACF Plus, OptiMEM, RoosterNourish‐MSC‐XF and RoosterCollect‐EV media were processed following the protocol described above and concentrated down to 500–600 µL and stored in aliquots at –80°C until further analysis. To evaluate inter‐batch variability in particle release, purity and particle size, three independent enrichments were performed for each of the four media used for EV collection and for unconditioned media (UCM) controls.

2.2.2. Analysis of Particle Number and Size Distribution

The ZetaView BASIC PMX‐120 (Particle Metrix GmbH) was utilised to evaluate the concentration, size distribution, median and mode sizes of EVs by nanoparticle tracking analysis (NTA). ZetaView (version 8.05) was used to calculate particle count and size distribution. The ZetaView device was calibrated with the supplied standard beads; for the analysis of the EV preparations, the sensitivity was set to 80, the shutter to 100, the temperature to 23.0°C and the frame rate to 30. The EV samples were diluted with filtered PBS before particle analysis. Three replicates of each particle measurement were recorded and evaluated, each with a different batch of EVs obtained from independent EV enrichment processes.

2.2.3. Transmission Electron Microscopy (TEM)

BM‐MSC/TERT292‐derived EVs enriched in different collection media were visualised via electron microscopy, and samples were prepared as described previously (Hausjell et al. 2023). Formvar‐coated copper grids were placed in a 10 µL sample drop (2E+08‐2E+09 particles, diluted in ddH₂O) for 5 min. After carefully removing excess sample with filter paper (Whatman), grids were promptly cleaned with distilled water before floating in a drop of 2% glutaraldehyde for 10 min to fix. After a quick wash with distilled water, negative staining was done with uranyl acetate replacement stain (UAR_EMS Stain, EMS‐Electron Microscopy Sciences) by floating the grids on a drop of UAR twice for 10 s and again for 60 s, gently blotting off the excess stain after each time. Grids were allowed to air dry before being imaged with a 160 kV FEI Tecnai G2 transmission electron microscope.

2.2.4. Protein Content Quantification

The protein concentration in the BM‐MSC/TERT292 EV preparations enriched in Mesencult ACF Plus medium, RoosterNourish‐MSC‐XF medium and the respective TFF enrichment controls was quantified using the Bicinchoninic Acid (BCA) protein assay (Thermo Fisher Scientific), according to instructions supplied by the manufacturer. EVs were diluted 1:20 to 1:100 in filtered PBS, whereas UCM controls were diluted 1:5 and 1:20 before protein detection. The protein concentration in the BM‐MSC/TERT292 EV preparations enriched in OptiMEM, RoosterCollect‐EV and the respective UCM controls was quantified using the microBicinchoninic Acid (microBCA) protein assay (Thermo Fisher Scientific), according to instructions supplied by the manufacturer. EVs were diluted 1:20 to 1:100 in filtered PBS, whereas UCM controls were diluted 1:4 and 1:10 before protein detection. A Spark multimode microplate reader (Tecan) was used to detect absorbance at 562 nm. The purity of EV preparations was determined by calculating the ratio of particle number, as indicated by NTA, to protein content, as measured by the BCA assay. The quantification was repeated three times, with each batch of EVs obtained from a different EV enrichment process.

2.2.5. Analysis of EV Marker Expression by Western Blot Analysis

BM‐MSC/TERT292 cells were collected, and the cell pellets (4 × 10⁶ cells) were lysed with 300 µL of 1× RIPA buffer (Merck Millipore), stored on ice and vortexed five times every 10–15 min for 1 h. The cell lysate was centrifuged at 15,000 × g for 10 min at 4°C and the supernatant was then collected. The protein concentrations in the supernatants were measured using the BCA assay (Thermo Fisher Scientific) as per the manufacturer's instructions. BM‐MSC/TERT292 EVs enriched in different collection media were diluted in 0.22 µm filtered PBS to a final concentration of 5.7E+10 p/mL in 58 µL. BM‐MSC/TERT292 EV preparations enriched in Mesencult ACF Plus and RoosterNourish‐MSC‐XF medium were diluted to a final concentration of 1.7E+11 p/mL and volume of 58 µL. Each EV sample was treated with 6.5 µL of 10X RIPA buffer (Merck Millipore) and 0.65 µL of Halt Protease Inhibitor Cocktail 100X (Thermo Fisher Scientific) prior to further processing. Cell lysates (24 µg), 1.5E+09 particles from each collection medium, and 5E+09 particles from EV preparations enriched in Mesencult ACF Plus medium and RoosterNourish‐MSC‐XF medium were mixed with NuPAGE LDS ample Buffer (Thermo Fisher Scientific) and NuPAGE Sample Reducing Agent (Thermo Fisher Scientific) and heated at 90°C for 10 min. Sample proteins and the Chameleon Duo Pre‐Stained Protein Ladder (Licor) were separated using a NuPAGE Novex 4%–12% Bis‐Tris Protein Gel (Thermo Fisher Scientific) and NuPAGE MOPS SDS Running Buffer (Thermo Fisher Scientific). The proteins from the gel were transferred to a Trans‐Blot Turbo Midi 0.2 µm PVDF Transfer Packs (Bio‐Rad) nitrocellulose membrane using the Mix MW function of the Trans‐Blot Turbo Transfer System. The membrane was stained with Ponceau S solution (Sigma–Aldrich) and then blocked with 50% Intercept (TBS) Blocking Buffer (Licor) in TBS (Thermo Fisher Scientific). Following blocking, the membrane was incubated overnight at 4°C with primary antibodies in v/v 50%/50% Intercept (TBS) Blocking Buffer (Licor)/TBS (Thermo Fisher Scientific) supplemented with 0.1% Tween‐20 (Sigma–Aldrich). The following primary antibodies were used: TSG101, abcam, ab125011, 1:1,000; Syntenin‐1, Origene, TA504796, 1:1,000; Calnexin, GeneTex, GTX101676, 1:1000, CD81 (B‐11), Santa Cruz, sc‐16602. The membrane was washed three times with TBS (Thermo Fisher Scientific) mixed with 0.1% Tween‐20 (TBS‐T) before being incubated with the relevant secondary antibody (LI‐COR; anti‐mouse IgG, 926–68072, 1:7500; anti‐rabbit IgG, 925–32213, 1:7500) for 90 min at room temperature. Finally, the membrane was rinsed three times with TBS‐T, once with TBS and imaged using the BioRad ChemiDoc imaging system at 680 and 800 nm.

2.2.6. Analysis of EV Lipid Composition by Fluorescence‐Triggered Flow Cytometry Analysis

The fluorescence‐triggered flow‐cytometry approach used in this study follows the previously established methodology by Oesterreicher et al. (2020), in which EVs are labelled with the lipid‐membrane—selective dye CellMask Green (CMG) and event detection is initiated by the fluorescence signal of CMG‐positive particles, ensuring detection of only vesicles enclosed by a dye‐incorporating lipid bilayer (Arraud et al. 2016). The configuration of the FT‐FC workflow, including the gating strategy was performed as described by Oesterreicher et al. (2020). For CMG staining and subsequent antibody labelling, 80 µL of pre‐diluted BM‐MSC/TERT292 EVs isolated in different collection media (final particle concentration 1E+10 p/mL), as well as 80 µL of the respective UCM controls (final particle concentration 1E+10 p/mL for Mesencult ACF Plus medium and RoosterNourish‐UCM, while OptiMEM and RoosterCollect‐EV UCM were used undiluted, and results adjusted for the initial particle concentration) were transferred into 1.5 mL tubes. A total of 20 µL of a 1:2000 CMG working solution was added, followed by incubation for 20 min at 37°C in the dark. Samples were mixed thoroughly and incubated for 30 min on ice in the dark. Control conditions included PBS with and without CMG, as well as EV samples processed without CMG, to verify the absence of CMG aggregates. For the lysis control, 1 µL of RIPA buffer (50 mM Tris‐Cl pH 7.4; 150 mM NaCl; 1% Triton X‐100; 0.5% sodium deoxycholate; 0.1% SDS; 0.1% PMSF) was added to 40 µL of CMG‐stained preparations and incubated for 30 min light‐protected prior to acquisition.

2.2.7. EV Surface Protein Profiling by Multiplex Bead‐Based Flow Cytometry Assay

BM‐MSC/TERT292 EVs enriched in different collection media were characterised in terms of surface marker expression by Multiplex bead‐based EV analysis followed by a flow cytometric analysis (MACSPlex EV Kit MSC, human130‐136‐864, Miltenyi Biotec). First, 1.0 E+09 particles from each EV preparation were diluted with MACSPlex buffer (MPB) to a final volume of 130 µL and loaded onto wells of a pre‐wet and drained MACSPlex 96‐well 0.22 µm filter plate. 5 µL of MACSPlex Exosome Capture Beads (containing 39 different antibody‐coated bead subsets) were added to each well and incubated overnight; filter plates were kept light‐protected during overnight incubation (14–16 h) on an orbital shaker at 450 rpm at RT. Beads were then washed with 200 µL/well of MPB and centrifuged at 300 × g at RT for 3 min. Next, 145 µL of MPB and 5 µL of APC‐conjugated detection antibody cocktail (anti‐CD9, anti‐CD63 and anti‐CD81) were added to each well to counterstain EVs bound by capture beads with detection antibodies. Plates were then incubated on an orbital shaker at 450 rpm protected from light for 1 h at RT. Upon incubation, plates were first washed twice by adding 200 µL/well of MPB and centrifuged at 300 × g at RT for 3 min, followed by addition of 200 µL/well and incubation for 15 min at RT on an orbital shaker at 450 rpm protected from light. Wells were once again drained and 200 µL of MPB was added/well prior to resuspension of the beads by pipetting and transferring of the beads to V‐bottom 96‐well microtiter plate (Thermo Fisher Scientific). CytoFLEX S (Beckman Coulter) was used for flow cytometric analysis. All samples were automatically mixed and 150–170 µL of samples were immediately loaded to and acquired by the instrument. Flow cytometric data were analysed by CytExpert software (Beckman Coulter). Prior to further data processing, the respective Median fluorescence intensity (MFI) values of matched non‐EV buffer (PBS) that were treated as EV‐containing samples (buffer/medium + capture beads + antibodies) were subtracted from all 39 capture beads subsets for background correction.

2.3. Analysis of the Biological Activity of EV Preparations

2.3.1. Anti‐Inflammatory Assay

To investigate the anti‐inflammatory properties of BM‐MSC/TERT292‐derived EVs enriched in different collection media, 4 × 10⁴ BV‐2 cells/well were seeded in serum‐containing cell expansion media (see Section 2.1.2) in a 96‐well plate and cultured at 37°C with 5% CO₂ and ambient oxygen for 24 h. All treatments were carried out in starvation medium (combination of DMEM 1 g/L glucose, Sigma–Aldrich and DMEM 4.5 g Glucose/L, Gibco at 1:1.9 ratio), with a final volume of 100 µL/well. For each treatment, BV‐2 cells were tested in parallel with (in quadruplicate) and without (in duplicates) stimulation with 1 µg/mL of lipopolysaccharide (LPS) from E. coli O127:B8 (Sigma–Aldrich). Cells were rinsed with 80 µL of starvation media, followed by treatment with a EVs at a concentration ranging from 1E+08 to 2E+09 particles/mL or UCM, tested at the same volume used to test EV enriched in the respective medium at 1E+09 p/mL (adjusted for enrichment factor) or, when volume was a limiting factor with the highest volume usable for the assay. As negative control, a constant volume of 5 µL of 20 mM HEPES was used per well; as a positive control, 2.5 µM Dexamethasone (Sigma–Aldrich, 1 mM dissolved in DMEM‐Ham's F12 1:1) was added. NO secretion was assessed 24 h after treatment using Griess Reagent (Promega) according to the manufacturer's instructions. and absorbance was measured at 535 nm using a Spark multimode microplate reader (Tecan).

2.3.2. Anti‐Fibrosis Assay

To evaluate the anti‐fibrosis properties of BM‐MSC/TERT292‐derived EVs enriched in different collection media, 2.5 × 10³ fHDF/TERT166 cells/well were seeded in starvation medium (DMEM/Ham's F‐12 1:1 supplemented with 0.5% FBS, 100 µg/mL Primocin (InvivoGen) and 100 µg/mL G418) into a 96‐well plate. Prior to treatment, cells were cultured at 37°C with 5% CO₂ and ambient oxygen for 24 h. All treatments were carried out in starvation medium with a final volume of 100 µL/well and performed in sextuplicate. To induce alpha smooth muscle actin (α‐SMA) 100 pg/mL of rhTGFbeta‐1 (Abcam) was added to each well, followed by treatment with a fixed concentration of 1E+09 EVs/mL or UCM, tested at the same volume used to test EV‐enriched in the respective medium (adjusted for enrichment factor) or the highest volume usable for the assay. As negative control, a constant volume of 5 µL/well of 20 mM HEPES was used; as positive control, 2 µM PP2 (Avantor/VWR, stock diluted 1:25) was used. The treatment was repeated on Days 2 and 5 following the initial treatment. Twenty‐four hours after the last treatment, cells were fixed with Roti‐Histofix (4% paraformaldehyde, Carl Roth) and permeabilised by adding a blocking solution (5% BSA/0.2% Triton‐X‐100 [Sigma–Aldrich] in PBS Ca²⁺/Mg²⁺). The following primary antibodies were used for overnight cell staining at 4°C: anti‐alpha smooth muscle actin [1A4] antibody (Abcam) and isotype control mouse IgG2a (Sigma–Aldrich), diluted 1:200 and 1:40 in blocking solution (5% BSA in PBS, Sigma–Aldrich). Cells were stained for 1 h at RT with a donkey anti‐mouse AF594 secondary antibody (Jackson, 1:500) in blocking solution and α‐SMA induction was measured using a Spark multimode microplate reader (Tecan) with excitation at 580 nm and emission at 615 nm.

2.3.3. Wound Healing Assay

BM‐MSC/TERT292‐derived EVs enriched in different collection media were assessed for wound healing ability by inserting Culture‐Insert 2 Well in μ‐Dish 35 mm (ibidi) into the wells of a 24‐well plate. HUVEC/TERT2 cells were seeded with at a fix cell density of 7.5 × 10³ cells/cm² in full expansion medium into the inserts and cultured at 37°C with 5% CO₂ and ambient oxygen for 24 h prior to insert removal and treatment. Except for the positive control, for which complete expansion medium was used, treatment was performed in starvation medium (EBM basal medium supplemented with hydrocortisone, hEGF, ascorbic acid, BBE and G418 at the same concentration as in complete medium), and 500 µL of medium was added/well. HUVEC/TERT2 cells were treated with a fixed concentration of 1E+09 EVs/mL or UCM, tested at the same volume used to test EV enriched in the respective medium (adjusted for enrichment factor) or the highest volume usable for the assay. As negative control, a constant volume of 5 µL/well of 20 mM HEPES was used. An EVOS M5000 microscope was used to image the gaps after the inserts were removed (time 0), as well as 16, and 40 h after treatment. The percentage of gap closure was measured by comparing the free area after 16‐ and 24‐h post‐treatment to the free area at time 0.

2.3.4. Alamar Blue and Proliferation Assay

To evaluate the effect of BM‐MSC/TERT292‐derived EVs on cell proliferation and cell metabolism upon enrichment from different collection media, 2.5 × 10³ fHDF/TERT166 cells/well were seeded in Starvation Mediums (DMEM/Ham's F‐12 1:1 supplemented with 0.5% FBS [Sigma–Aldrich], Primocin [InvivoGen] and 100 µg/mL G418) into a 96 well plate. Prior to treatment, cells were cultured at 37°C with 5% CO₂ and ambient oxygen for 24 h. Except for the negative control, for which Medium supplemented with 10% FBS was used, treatments were performed in Starvation Medium, and 100 µL of medium was added/well. Fixed concentrations of 2.5E+08 and 1E+09 EVs/mL were used for each treatment, that was performed in quadruplicates. Similarly, HUVEC/TERT2 cells were treated with the same UCM‐medium volume of the BM‐MSC/TERT292‐derived EV enriched in the respective medium (adjusted for the enrichment factor) or, when volume was a limiting factor with the highest volume usable for the assay. As negative control, a constant volume of 5 µL/well of 20 mM HEPES was used. The metabolic activity of fHDF/TERT166 was assessed 72 h after treatment by replacement of cell expansion medium with 100 µL/well of complete cell expansion medium supplemented with 10% v/v of Alamar Blue – Cell Viability Reagent (Invitrogen). Absorbance measurement was done at 555 nm using a Spark multimode microplate reader (Tecan) at time 0 and 4 h after Alamar Blue addition. Next, cells were fixed with 4% Histofix (CarlRoth) for 10 min at RT and stained for 15 min at RT with Vybrant dye cycle violet (Thermo Fisher Scientific, 1:2500 dilution in PBS Ca²⁺/Mg²⁺, 100 µL/well). The staining solution was removed; cells were rinsed with PBS Ca²⁺/Mg²⁺ and kept light‐protected at 4°C until imaging and automatic‐cell counting with an EVOS M5000 microscope.

2.3.5. ROS Production Assay

To assess the anti‐oxidant properties of BM‐MSC/TERT292‐derived EVs enriched in different collection media, 6 × 10³ fHDF/TERT166 cells/well were seeded in starvation medium (DMEM/Ham's F‐12 1:1 supplemented with 0.5% FBS, 100 µg/mL Primocin (InvivoGen) and 100 µg/mL G418) into a 96 well plate and cultured for 24 h at 37°C with 5% CO₂ and ambient oxygen. All treatments were carried out in starvation medium with a final volume of 100 µL/well and performed in quadruplicate. Cells were treated with fixed concentrations of 1E+09 and 2.5E+8 EVs/mL or UCM, tested at the same volume used to test EV enriched in the respective medium (adjusted for enrichment factor) or the highest volume usable for the assay prior to H₂O₂ treatment 24 h after treatment, cells were washed twice with HBSS (Thermo Fisher Scientific) and stained with 5 µM 2′,7′‐dichlorodihydrofluorescein diacetate (H2DCFDA, Invitrogen, diluted in HBSS) for 30 min at 37°C with 5% CO₂ and ambient oxygen. Cells were then washed once with DMEM/Ham's F‐12 basal medium prior to 400 µM H₂O₂ treatment (diluted DMEM/Ham's F‐12 basal medium, stock 30% hydrogen peroxide solution, Sigma–Aldrich) for 30 min at 37°C with 5% CO₂ and ambient oxygen. As a negative control, untreated cells with a constant volume of 5 µL/well of 20 mM HEPES was used. H₂O₂ alone treated cells served as a positive control. H2DCFDA unstained cells were used for background correction. Finally, H₂O₂‐mediated ROS production was measured using a Spark multimode microplate reader (Tecan) with excitation at 492 nm and emission at 530 nm.

2.3.6. Tube Formation Assay

The neoangiogenic potential of BM‐MSC/TERT292 EVs enriched in different collection media was assessed using cooled μ‐Slide 15 Well 3D slides (ibidi) pre‐coated with 10 µL/well of ice‐cold Matrigel (Reduced growth factor matrigel: Geltrex LDEV‐Free hESC qualified, Life Technologies), which was placed in a humidified chamber for 60 min at 37°C with 5% CO₂ and ambient oxygen to allow Matrigel polymerization. 1.5 × 10⁴ HUVEC/TERT2 cells were seeded/well in EBM (Lonza) basal media and immediately treated with a fixed concentration of 1E+09 EVs/mL or UCM, tested at the same volume used to test EV enriched in the respective medium. Cells were then cultured at 37°C with 5% CO₂ and ambient oxygen for 6 h prior to imaging. All treatments were carried out in triplicate; EV‐untreated cells seeded in EGM were used as negative control, and EV‐untreated cells seeded in EBM (Lonza) supplemented with bovine brain extract (BBE), human epidermal growth factor (hEGF), hydrocortisone, ascorbic acid (components of the EGM SingleQuots supplement kit, Lonza) were used as positive control. Images were analysed using the Angiogenesis Analyzer plug in of ImageJ software. For each well, three images covering the whole well area were taken with an EVOS M5000 microscope and analysed, and the average of the three images was used for further statistical analysis.

2.4. Statistical Analysis

GraphPad Prism 10.0 (GraphPad Software, Inc. San Diego, CA, USA) was used for all the statistical analysis. At least three different TFF enrichments have been analysed for the characterization of BM‐MSC/TERT292 EVs enriched in different collection media (as well as for the respective UCM controls) in terms of particle secretion, purity and particle size. At least four technical replicates have been used to assess the biological potential of BM‐MSC/TERT292‐derived EVs (as well as of the respective UCM controls). Bars represent mean ± SD. The population doubling times of BM‐MSC/TERT292 and ASC/TERT300 cells in the different expansion media and the particle size of BM‐MSC/TERT292‐derived EVs and the respective UCM controls were analysed with one‐way ANOVA. The biological activity of BM‐MSC/TERT292‐derived EVs enriched in the different collection media were analysed with one‐way ANOVA and Dunnett's test. For BM‐MSC/TERT292 particle release and purity in the different collection media and for comparing the protein content and particle concentration of BM‐MSC/TERT292‐derived EVs in respect to the correspondent UCM control, One‐way Anova test was performed on Log10 transformed data to account for scale difference. The wound healing properties of BM‐MSC/TERT292‐derived EVs enriched in different collection media (as well as of the respective UCM controls) were assessed using Two‐way Anova and Dunnett´s test. Outliers were identified using Grubbs method (alpha = 0.05). Significance levels are reported as follows: *0.01 ≤ p < 0.05; **0.001 < 0.01; ***0.001 < 0.0001 and ****p < 0.0001.

3. Results

3.1. Effect of Different Expansion Media on BM‐MSC/TERT292 Cells

To assess how different cell expansion media affect the growth properties of telomerised MSCs cell lines, BM‐MSC/TERT292 cells were cultivated in two different commercial media, in which they displayed continuous growth, and compared in terms of cell morphology, growth potential and population doubling time (PDT). Growth curves of BM‐MSC/TERT292 cells in Mesencult‐ACF, the expansion medium in which these cells have been initially adapted after isolation, and in RoosterNourish‐MSC‐XF medium supplemented with 1% or 2% Booster (Figure 1A) were assessed. Therefore, cells were initially thawed in Mesencult‐ACF medium; during the first split after thawing, the medium was switched to RoosterNourish‐MSC‐XF medium supplemented with 2% Booster during the centrifugation step. After two passages in RoosterNourish‐MSC‐XF + 2% Booster, a parallel condition with Roster + 1% Booster was tested. Despite an initial slight increase in cell growth, BM‐MSC/TERT292 growing in RoosterNourish‐MSC‐XF supplemented with 2% Booster showed a typical 1:6 split ratio performed twice a week (Figure 1A) with an average population doubling time (PDT) of 0.75 per day (Figure 1B) comparable with cells growing in Mesencult‐ACF. However, cells in RoosterNourish‐MSC‐XF supplemented with 1% Booster showed slower cell growth (Figure 1A) and lower PDT after switching them at Day 10 from 2% Booster to 1% (Figure 1B).

FIGURE 1.

Growth optimisation of BM‐MSC/TERT292 in different media. (A) Growth curves of BM‐MSC/TERT292 cells in Mesencult (orange dots), RoosterNourish‐MSC‐XF supplemented with 2% Rooster Booster (green dots) or 1% Rooster Booster (light blue dots). The black arrow indicates the change from 2% to 1% Booster. (B) Population Doubling Time of BM‐MSC/TERT292 cells in Mesencult and RoosterNourish‐MSC‐XF media supplemented with 2% or 1% Rooster Booster. (C) Representative microscopy images of BM‐MSC/TERT292 cells cultivated in the different media (scale bar: 50 µm). Data are represented as mean ± SD; One‐way Anova was used, and significance is shown as *p < 0.05, **p < 0.01.

Morphologically, cells in RoosterNourish‐MSC‐XF supplemented with 2% Booster showed similar morphological characteristics compared to cells growing in Mesencult‐ACF, with a slightly more elongated and heterogeneous phenotype (Figure 1C). BM‐MSC/TERT292 cells in RoosterNourish‐MSC‐XF supplemented with 1% Booster showed an even more heterogeneous and elongated phenotype, accompanied by reduced cell density relative to cells growing in Mesencult‐ACF (Figure 1C). Overall, these data show adaptability of BM‐MSC/TERT292 cells to different expansion media, as well as the need for increased Booster percentages in RoosterNourish‐MSC‐XF medium to support cell growth. Therefore, we decided to use RoosterNourish‐MSC‐XF medium supplemented with 2% Booster and Mesencult‐ACF media in this study.

3.2. Effect of Different Collection Media on BM‐MSC/TERT292‐Derived EVs

Next, we assessed the combination of expansion media and EV collection media for the enrichment of EVs from BM‐MSC/TERT292 cells (see Figure S1A). On the one hand, we tested Mesencult‐ACF for cell cultivation and for EV harvesting as compared to switching to Opti‐MEM reduced serum medium for EV collection. Opti‐MEM is a standard collection medium extensively used by many research laboratories, as well as in GMP‐grade facilities for production of biologics such as EVs. On the other hand, we used RoosterNourish‐MSC‐XF as standardised, serum‐free and xeno‐free cultivation medium for expansion and harvesting as compared to switching to RoosterCollect‐EV for harvesting. This process was intentionally choosen to maintain process consistency, although the manufacturer recommends the use of dedicated collection medium. RoosterCollect‐EV is a chemically defined, xeno‐free and protein‐free medium designed for clinical‐grade EV production upon cell culturing in RoosterNourish‐MSC‐XF.

BM‐MSC/TERT292 showed typical MSC morphology not only when cultured for 48 h for EV collection in Mesencult‐ACF and RoosterNourish‐MSC‐XF, but also in OptiMEM and RoosterCollect‐EV (Figure 2A). Therefore, we decided to further proceed with EV collection and enrichment from BM‐MSC/TERT292, to elucidate the effect of four different viable combinations of growth and collection media on EV properties. For simplicity, BM‐MSC/TERT292‐derived EVs enriched in RoosterNourish‐MSC‐XF, RoosterCollect‐EV, Mesencult‐ACF and OptiMEM media will be named respectively: Nourish‐EVs, EV Collect‐EVs, Mesencult‐EVs and OptiMEM‐EVs.

FIGURE 2.

Characterisation of BM‐MSC/TERT292‐derived extracellular vesicles enriched from different collection media. (A) Representative microscopy images of BM‐MSC/TERT292 cells cultivated in the different media at 48 h after the last medium exchange as the time point of EV harvest (Scale bar: 50 µm). (B) Particle release, (C) particle size and (D) purity of BM‐MSC/TERT292‐derived EVs enriched from different collection media. Data are represented as mean ± SD; for particle release and purity, One‐way Anova was performed on Log₁₀ transformed data to account for scale difference. For particle size, One‐way Anova was performed on untransformed data. Significance is shown as *p < 0.05, **p < 0.01, ****p < 0.0001 (n = 3 independent EV samples/group, each sample from an independent EV batch). (E) Summary table of mean particle release, particle size and purity for each medium used for EV collection, and relative SD values. (F) Western blot analysis of common EV surface (CD81) and intraluminal (Syntenin, TSG101) markers, as well as negative EV markers (Calnexin) in BM‐MSC/TERT292 cell lysate and secreted EVs in the different collection media. For each EV sample, protein lysate from 1.6E+09 particles was added/lane. (G) Transmission electron microscopy analysis of BM‐MSC/TERT292‐derived EVs. Black arrow: EVs, Yellow arrow: protein aggregate; Orange arrow: coated‐like particle surface. Scale bars: 250 nm.

BM‐MSC/TERT292 cells showed higher particle numbers when cultured in RoosterNourish‐MSC‐XF (1.77E+04 particles/cell/day), followed by EV collection performed in Mesencult‐ACF (3.87E+03 particles/cell/day), RoosterCollect‐EV (7.28E+02 particles/cell/day) and OptiMEM (3.92E+02 particles/cell/day) (Figure 2B). BM‐MSC/TERT292‐derived EVs showed similar particle sizes upon enrichment in four different collection media (Figure 2C). OptiMEM‐EVs and EV Collect‐EVs showed higher particle purity as defined by particles per mg of protein (Webber and Clayton 2013), followed by EVs enriched in RoosterNourish‐MSC‐XF and Mesencult‐ACF (Figure 2D). These three parameters are summarised in Figure 2E.

We further assessed the influence of collection media on surface and intraluminal marker expression in BM‐MSC/TERT292‐derived EVs by Western blot analysis (Figure 2F) loading 1.5E+09 particles per lane. Both EVs enriched in RoosterCollect‐EV and OptiMEM showed CD81 and Syntenin enrichment, and positivity for TSG101; Calnexin signal was barely detectable in the EV preparations compared to cell lysates, indicating minimal ER contamination. However, although the same particle number was loaded, only low‐to‐barely detectable signals of CD81 and Syntenin were visible for Mesencult‐EVs and Nourish‐EVs. Only when loading a 3‐fold higher particle count (Figure S1B), EVs enriched in Mesencult‐ACF showed clear signals for CD81, Syntenin and TSG101, and with absence for Calnexin. Nourish‐EVs showed low signals of common EV markers even upon increased particle loading for protein detection.

Finally, we visualised EVs in our preparations by TEM (Figure 2G). Interestingly, the imaging of Nourish‐EVs showed clear presence of protein aggregates and impurities (bottom left image, yellow arrow) co‐isolated with EVs (bottom, corner‐left image, black arrow), in line with particle purity and Western blot data. On the other hand, Mesencult‐EVs showed the presence of EVs carrying a rougher surface (bottom, middle‐right image, orange arrow), but no presence of protein aggregates and impurities was visible. Overall, these data demonstrate successful enrichment of BM‐MSC/TERT292 EVs across the different collection media used for EV enrichment. Considering the high particle number, low particle‐to‐protein ratio, in combination with the Western blot results, we suggest that a elevated particle count might derive from the cultivation media rather than the harvest medium. To note, these findings highlight the importance of adhering to manufacturer recommendations when using RoosterBio growth and collection media, as BM‐MSC/TERT292 exhibited optimal growth only when the expansion media was supplemented with 2% RoosterBooster‐, as per supplier's optimised formulation (Figure 1), and EVs showed the highest purity when cells were expanded in RoosterNourish‐MS and subsequently cultured in RoosterCollect‐EV for enrichment.

3.3. Characterisation of the EV‐Like Properties of the Unconditioned Media Controls

To assess how the conditioned media compared with their unconditioned counterparts in terms of particle abundance and functional activity, unconditioned media were incubated under identical conditions for 48 h at 37°C in the same incubators as the conditioned counterparts with cells. Subsequently, UCM underwent the same enrichment workflow‐including centrifugation, filtration and TFF processing‐ as applied to conditioned medium.

First, we compared particle concentration in EV preparations and in the respective UCM controls. Of note, particle concentrations of Nourish‐derived EV preparations were comparable to that of their corresponding UCM control, suggesting Nourish unconditioned media might contain high particle components which are indistinguishable from the secreted EVs. On the other hand, EV collection from BM‐MSC/TERT292 cells in RoosterCollect‐EV showed around 10‐fold higher particle concentration compared to UCM, followed by OptiMEM by about 3‐fold and Mesencult‐ACF by about 2‐fold (Figure 3A), while particle counts per volume varied over a range of three orders of magnitude. Particle sizes upon TFF enrichment were in the range of 120–150 nm, with only particles enriched from unconditioned RoosterCollect‐EV medium showed significantly higher average size (Figure 3B). Next, we compared the protein concentration of UCM to EV enrichments in the respective medium (Figure 3C). EVs enriched in RoosterNourish‐MSC‐XF showed a trend of minimal increase in the protein content when compared to RoosterNourish‐MSC‐XF UCM, showing similar tread as particle numbers (Figure 3A) suggesting Nourish medium is the richest medium in terms of protein content and particles followed by Mesencult‐ACF (Figure 3C,D). Interestingly, when compared to UCM, Mesencult‐EVs showed the highest significant increase in protein content/mL of almost 6‐fold, followed by EV Collect‐EVs by around 3‐fold. A trend of an increase of OptiMEM‐EVs was observed. These data are summarised in Figure 3D. Finally, we quantified lipid stain positive BM‐MSC/TERT292‐derived EVs and the respective UCM controls by CMG staining and flow cytometry (Oesterreicher et al. 2020). As control, detergent‐treated samples were used to confirm lipid specificity of the measured signals. Surprisingly, EVs enriched in RoosterNourish‐MSC‐XF showed lower lipid stain‐positive particles when compared to the respective UCM control. In contrast to this and similar to the protein content data, EV‐Collect EVs and OptiMEM‐EVs showed a marked increase for lipid stain positive particles when compared to their respective UCM controls (Figure 3E). No differences between Mesencult‐EVs and UCM control were detected. Taken together, these data show that, depending on the media used for EV enrichment, the contribution of cellular versus media‐derived particles is not easily differentiated.

FIGURE 3.

Comparison of the physicochemical properties of EVs and the respective UCM controls. (A) Comparison of particle concentration from UCM controls and BM‐MSC/TERT292‐derived EVs in the different collection media. Particle concentrations from each UCM control and EV sample were normalized on an enrichment factor of 100. (B) Comparison of particle sizes of UCM controls and BM‐MSC/TERT292‐derived EVs in the different collection media. (C) Comparison of protein content/mL of unprocessed CM between UCM controls and BM‐MSC/TERT292‐derived EVs in the different collection media. One‐way Anova was performed on Log₁₀ transformed data to account for scale differences. For particle size, One‐way Anova was performed on untransformed data. Significance is shown as *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 (n = 3 independent EV samples/group, each sample from an independent EV batch). (D) Summary table of mean particle concentration and protein content for UCM controls and BM‐MSC7TERT292‐derived EVs from each medium used for EV collection, and relative SD values. Fold change between EVs and UCM control values was calculated for each medium used for EV collection and reported in the table. (E) Comparison of CMG+ events in UCM controls and BM‐MSC7TERT292‐derived EVs in the different collection media prior and after triton X‐100 treatment. Data are shown as relative to untreated EVs (intact) and fold change between EVs and UCM values was calculated for each medium used for EV collection and reported in the graph (n = 2 independent EV samples/group).

3.4. Surface‐Marker Profiling of BM‐MSC/TERT292‐Derived EVs Upon Enrichment in Different Collection Media

We next evaluated how the combinations of growth and collection media influence the surface marker profiles of BM‐MSC/TERT292‐derived EVs using equal particle counts in the MSC‐MASCPlex assay (Figure 4). EVs enriched from different media displayed distinct surface marker profiles, whereas EVs obtained by three independent enrichments of the same medium exhibited highly comparable profiles (Figure 4A). This observation was confirmed by Principal Component Analysis (Figure 4B), indicating strong batch‐to‐batch reproducibility and a clear collection medium‐dependent effect on EV surface marker composition, whereby the two defined collection media, Opti‐MEM and RoosterCollect‐EV, seem more similar to each other. Of note, all EV preparations were negative for CD11b, CD14, CD19, CD45 and CD79, well‐known negative markers for MSCs and MSC‐EVs (Dominici et al. 2006; Nguyen et al. 2024).

FIGURE 4.

Surface marker profiling of BM‐MSC/TERT292‐derived extracellular vesicles enriched from different collection media by MACSPlex analysis. (A) Heatmap visualization of individual MACSPlex marker expression in BM‐MSC/TERT292 EVs enriched in different collection media. The color gradient reflects the background‐subtracted Median APC Fluorescent Intensities of each EV preparation included in the analysis. (B) Score plot of the Principal Component (PC) analysis for the MACSPlex detection of surface marker‐expression on BM‐MSC/TERT292 EVs enriched in different media. Oval shapes represent the 95% Confidence Interval. (C) Median APC Fluorescence Intensity of the EV‐markers CD9, CD63 and CD81 and (D) of the MSC‐specific markers CD73, CD90 and CD105 in BM‐MSC/TERT292 EVs enriched in different collection media by MACSPlex analysis. (E) Median APC Fluorescence Intensity of the Platelet and Endothelial cell‐derived markers CD31, CD41b and CD42P and CD42a in BM‐MSC/TERT292 EVs enriched in different collection media by MACSPlex analysis. For each medium, three different EV preparations were analysed (n = 3 independent EV samples/group, each sample from an independent EV batch; the numbers next to the heatmap represent each independent EV batch used for the analysis) and data are represented as mean ± SD were applicable. One‐way Anova was used, and significance is shown as *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Both OptiMEM‐ and EV Collect‐ EVs showed high signals of classical EV markers (CD9, CD81, CD63) (Figure 4C) and MSC‐associated markers (CD73, CD90, CD105) (Figure 4D), with Opti‐MEM‐EVs overall exhibiting the highest marker signals. In contrast, Mesencult‐ and Nourish‐EVs exhibited reduced expression of canonical EV markers (Figure 3B) and showed minimal to no detectable levels of MSC‐specific markers (Figure 4D), except for CD73 in both preparations and CD90 in Mesencult‐derived EVs. This data aligns with the higher purity of OptiMEM‐ and EV Collect‐ EVs compared to Mesencult‐ and Nourish‐EVs, as shown by Western blot and TEM analysis (Figure 2F,G). Notably, Nourish‐EVs showed clear enrichment for platelet/endothelial cell‐markers, such as CD31, CD41b, CD62P and CD42a (Figure 4E), suggesting a co‐enrichment of hPL‐derived EVs contained in the medium. Collectively, these results demonstrate low batch‐to‐batch variability within each medium and highlight the strong influence of collection medium on EV surface marker profiling. These findings further align with the elevated particle concentrations and protein levels detected in unconditioned nourish medium (Figure 3A,C), emphasising the need to integrate multiple downstream analytical approaches when evaluating EV preparations.

3.5. Effect of Different Collection Media on Biological Activity of BM‐MSC/TERT292‐Derived EV

In the following chapters, we assessed how different combinations of growth and collection media influence the biological properties of BM‐MSC/TERT292‐derived EVs. We compared the activity of the various EV preparations across six cell‐based assays evaluating distinct bioactivities, including immunomodulatory and anti‐fibrosis activity, ‘wound healing’ (scratch assay), pro‐proliferative activity, angiogenesis and anti‐oxidative capacity. All these assays were performed using unconditioned media as controls.

First, we assessed the effect of different collection media on the anti‐inflammatory potential of BM‐MSC/TERT292‐derived EVs (Figure 5A). After treating LPS‐stimulated BV‐2 cells with either BM‐MSC/TERT292 EVs enriched in different collection media at concentrations ranging from 1E+08 to 2E+09 particles/mL or with Dexamethasone as a positive control, the secretion of nitric oxide (NO) was measured 24 h later. In detail, all EV preparations were tested at fixed concentrations of 5E+08 and 1E+09 particles/mL: due to the lower particle concentration of the starting preparations, EVs OptiMEM and EV Collect‐EVs were tested at a particle concentration as low as 1E+08 particles/mL, and could not be tested at 2E+09 particles/mL—unlike Mesencult‐ and Nourish‐EVs (Figure 5A). When treating LPS‐stimulated BV‐2 cells with a particle concentration of 1E+08 p/mL, OptiMEM‐EVs showed strong anti‐inflammatory potential (70.3% NO reduction), while EVs enriched in RoosterCollect‐EV showed a lack of activity (Figure S2A). All EV preparations showed anti‐inflammatory activity upon treatment of LPS‐stimulated BV‐2 cells with a concentration of 5E+08 p/mL (Figure S2B), with EVs in OptiMEM showing a higher reduction of NO secretion (63.5%), followed by EVs in RoosterNourish‐MSC‐XF (47.5%), EVs in RoosterCollect‐EV (34.0%) and in Mesencult‐ACF (30.0%). Similarly, all enriched EVs tested at 1E+09 particles/mL were anti‐inflammatory, with slightly higher activity of OptiMEM‐EVs, with an NO reduction of 73.5% compared to LPS‐stimulated BV‐2 cells. EVs enriched in Mesencult‐ACF reduced the signal by 68.4% reduction, EV Collect‐EVs by 59.0% reduction and Nourish‐EVs (53.7% reduction) (Figure S2C). Interestingly, when treating LPS‐stimulated BV‐2 cells with a higher particle concentration of 2E+09 particles/mL, Mesencult‐EVs showed higher anti‐inflammatory potential (89.3% reduction of NO secretion), while Nourish e ‐EVs showed an increase of NO secretion (26.4% increase compared to LPS‐stimulated BV‐2 cells) upon treatment with 2E+09 p/mL (Figure S2C). To note, EV Collect‐EVs and Mesencult‐EVs showed a clear‐dose dependency, while EVs enriched in Nourish lost activity at the higher concentration; EVs enriched in OptiMEM showed a high, similar to Dexamethasone anti‐inflammatory potential at all the tested concentrations (Figure 5A). The lack of activity of Nourish‐EVs at 2E+09 p/mL compared to the strong activity of OptiMEM‐EVs at a concentration as low as 1E+08 p/mL, suggests that the composition of the collection medium may influence the effective therapeutic dose of theEV preparations. These findings also indicate that high particle concentrations may overload the recipient cells, potentially causing membrane stress, endosomal saturation or metabolic burden, which in turns may mask the therapeutic effects of the EVs (Hagey et al. 2023).

FIGURE 5.

Evaluation of the biological potential of BM‐MSC/TERT292‐derived EVs enriched from different collection media. (A) Evaluation of the dose‐dependency of the anti‐inflammatory activity of BM‐MSC/TERT292 EVs enriched from different collection media on NO secretion by LPS‐stimulated BV‐2 cells. LPS‐stimulated BV‐2 cells were treated with increasing EV concentrations ranging from 1E+08 to 2E+09 p/mL. 2.5 µM Dexamethasone (Dexa, green dotted line) was included as positive control. Data are represented as mean ± SD and data were normalised on the LPS‐stimulated BV‐2 cells (set at 100%, red dotted line). (B) Assessment of the anti‐fibrosis effect of BM‐MSC/TERT292 EVs enriched from different collection media on α‐SMA induction by TGF‐β1‐stimulated fHDF/TERT166 cells. 20 mM HEPES was included as negative control and 2 µM PP2 was included as positive control. TGF‐β1‐stimulated fHDF/TERT166 cells were treated with a fixed EV concentration of 1E+09 p/mL. (C) Quantification of the wound closure of HUVEC/TERT2 cells upon BM‐MSC/TERT292‐derived EVs treatment. Wound closure was measured at 0, 16 and 24 h after the removal of culture insert and treatment. 20 mM HEPES was included as negative control and complete expansion medium (Medium + FBS) was included as positive control. HUVEC/TERT2 cells were treated with a fixed EV concentration of 1E+09 p/mL. Data are represented as mean ± SD. One‐way Anova was used for the anti‐inflammatory and anti‐fibrosis assay; data were normalised on the LPS‐stimulated BV‐2 cells and TGF‐β1‐stimulated fHDF/TERT166 cells respectively (set at 100%). Two‐way Anova was used for the Wound Healing assay. Significance is shown as *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 (n = 4 technical replicates for the anti‐fibrosis and anti‐inflammatory assays, n = 3 biological replicates for the wound‐healing assay). (D) Representative microscope images of the HUVEC/TERT2 cells' wound closure 24 h following culture insert removal and treatment (photos were taken at 100× lens magnification).

To assess the anti‐fibrotic properties of the different EV preparations, telomerised human dermal skin fibroblasts (fHDF/TERT166) were exposed to TGF‐β as a fibrosis inducer. α‐SMA immunofluorescence staining and microscopical quantitation was used as a read‐out using the kinase inhibitor PP2 as a positive control for anti‐fibrosis activity (Figure 5B). Nourish‐EVs and EV Collect‐EVs showed the highest and comparable reduction of α‐SMA induction (39.5% and 36.3%, respectively), followed by EVs in Mesencult‐ACF (23.3% reduction); EVs enriched in OptiMEM did not show significant anti‐fibrosis activity.

Furthermore, we used a scratch assay to test if EVs activate migration/proliferation in telomerised human umbilical vein endothelial cells (HUVEC/TERT2) as a surrogate for wound healing (Figure 5C,D). EV Collect‐EVs and Mesencult‐EVs showed higher gap closure potential both at 16 and 24 h after treatment (32.1% and 28.8% of gap closure at 24 h, respectively), followed by EVs enriched in OptiMEM (21.7% of gap closure at 24 h, respectively) (Figure 5C,D, Figure S3A). Interestingly, Nourish‐EVs showed the weakest gap closure ability (non‐significant 12.0% of gap closure at 24 h). Overall, these data show that the usage of different combinations of growth and collection media have an impact on the anti‐inflammatory, anti‐fibrosis and wound healing potential of BM‐MSC/TERT292‐derived EVs.

3.6. Bioactivity of Unconditioned Media

As previously observed, BM‐MSC/TERT292‐derived EV preparations, as well as the respective UCM controls enriched in RoosterNourish‐MSC‐XF and Mesencult‐ACF medium exhibited the highest particle concentrations, protein and lipid content (Figure 3A,C,E). Moreover, Nourish‐EVs showed enrichment for platelet/endothelial cell‐markers, suggesting a co‐enrichment of hPL‐derived components (Figure 4E). Thto determine the extent to which UCM might contribute to the observed biological effects, we assessed the activity of UCM alone across the selected in vitro bioassays (Figure 6). For all bioactivity assays, EV preparations were tested based on particle numbers. Due to the limited particle concentraions of UCM controls enriched using RoosterCollect‐EV and OptiMEM, these UCM controls were instead tested based on volume and enrichment factors. While this approach enables comparison with EV samples, it represents a distinct normalization strategy.

FIGURE 6.

Evaluation of the biological activity of UCM controls. (A) Evaluation of the anti‐inflammatory activity of the different UCM controls on NO secretion by LPS‐stimulated BV‐2 cells. 20 mM HEPES was included as negative control and 2.5 µM Dexamethasone (Dexa) was included as positive control. LPS‐stimulated BV‐2 cells were treated with the same UCM volume of the BM‐MSC/TERT292‐derived EV enriched in the respective medium (adjusted for the enrichment factor). (B) Assessment of the anti‐fibrosis effect of the different UCM controls on α‐SMA induction by TGF‐β1‐stimulated fHDF/TERT166 cells. 20 mM HEPES was included as negative control and 2 µM PP2 was included as positive control. TGF‐β1‐stimulated fHDF/TERT166 cells were treated with the same UCM volume of the BM‐MSC/TERT292‐derived EV enriched in the respective medium (adjusted for the enrichment factor). (C) Quantification of the wound closure of HUVEC/TERT2 cells upon TUCM control treatment. Wound closure was measured at 0, 16 and 24 h after the removal of culture insert and treatment. 20 mM HEPES was included as negative control and complete expansion medium (Medium + FBS) was included as positive control. HUVEC/TERT2 cells were treated with the same UCM volume of the BM‐MSC/TERT292‐derived EV enriched in the respective medium (adjusted for the enrichment factor). Data are represented as mean ± SD. One‐way Anova was used for the anti‐inflammatory and anti‐fibrosis assay; data were normalised on the LPS‐stimulated BV‐2 cells and TGF‐β1‐stimulated fHDF/TERT166 cells respectively (set at 100%). Two‐way Anova was used for the Wound Healing assay. Significance is shown as *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 (n = 4 technical replicates for the anti‐fibrosis and anti‐inflammatory assays, n = 3 biological replicates for the wound‐healing assay). (D) Representative microscope images of the HUVEC/TERT2 cells' wound closure 24 h following culture insert removal and treatment (photos were taken at 100× lens magnification).

In terms of anti‐inflammatory effects of the different UCM controls on LPS‐stimulated BV‐2 cells, only Mesencult‐ACF demonstrated significant anti‐inflammatory potential, reducing NO secretion to 35.8% (Figure 6A). Next, we assessed the anti‐fibrotic properties of the different UCM controls on TGF‐β‐ stimulated fHDF/TERT166 (Figure 6B). While OptiMEM‐UCM and EV‐Collect‐UCM showed no detectable anti‐fibrotic activity, both Nourish‐UCM and Mesencult‐UCM enriched in Mesencult‐ACF exhibited measurable activity. Taken together, these results suggest that in addition to the co‐enrichment of hPL‐derived components within the EV preparations (Seidelmann et al. 2021), factors originating from serum‐free and hPL‐free media may also contribute to the observed anti‐inflammatory and anti‐fibrotic properties of the MSC‐derived EVs.

Next, we assessed the gap‐closures potential of UCM controls on the HUVEC/TERT2‐based scratch assay, as a surrogate in vitro assay of wound healing (Figure 6C,D and Figure S3B). None of the UCMs had marked effects. However, only a slight but significant reduction by about 9.5% of the gap area both at 16‐ and 24‐h timepoints by RoosterNourish‐MSC‐XF UCM was observed, similar to the activity shown by Nourish‐EVs (Figure 5C,D, Figure S3A). Overall, these data show that care needs to be taken in interpreting the source of bioactivity when using EV enriched secretome, as co‐enriched non‐EV components from protein and particle rich media can contaminate the EV preparation and mediate an EV‐independent activity.

3.7. Effect of BM‐MSC/TERT292‐Derived EVs and Unconditioned Media Controls on Cell Proliferation and Metabolism

Next, taking into account that prior literature has reported that EVs can enhance cellular proliferation (da Fonseca Ferreira et al. 2017) as well as improve mitochondrial function (Zhang et al. 2023), and that metabolic activity does not always correlate with DNA content (Quent et al. 2010) we decided to evaluate both parameters—by cell count and Alamar Blue assay—in our EV preparations and in the corresponding UCM controls. Therefore, fHDF/TERT166 were cultured in Starvation Medium (Medium + 0.5% FBS) and either treated with BM‐MSC/TERT292‐derived EVs enriched in the different collection media at a particle concentration of 1E+09 p/mL or with cell expansion medium supplemented with 10% FBS as positive control. Seventy‐two hours after treatment, fHDF/TERT166 metabolism was evaluated by Alamar Blue assay, followed by Vybrant dye cycle violet staining and automatic cell count by fluorescence microscopy.

Interestingly, all the EV preparations showed a significant increase in Alamar Blue signals, with Mesencult‐EVs showing the highest increase (30.8% compared to untreated cells), followed by EVs enriched in RoosterCollect‐EV (21.9% increase), in OptiMEM (12.9% increase) and in RoosterNourish‐MSC‐XF (6.0% increase) (Figure 7A). The proliferative potential of the different EV preparations was further assessed by counting cell numbers (Figure 7B,C). Once again, Mesencult‐EVs showed the highest effect (49.9% increase in cell count) on cell proliferation; EVs enriched in RoosterCollect‐EV and RoosterNourish‐MSC‐XFshowed similar activity (24.0% and 17.4% increase respectively), while EVs enriched in OptiMEM did not show a significant increase in cell counts. The same results were confirmed by microscopy images of serum‐starved fHDF/TERT166 cells 72 h after treatment (Figure 7C). Once again, both the metabolic and proliferative activity of BM‐MSC/TERT292 EVs enriched in different collection media resulted to be dose‐dependent, as showed by the treatment of serum‐starved fHDF/TERT166 with EVs at a concentration of 2.5E+08 p/mL (Figure S4A,B). Interestingly, while Mesencult‐EVs, OptiMEM‐EVs, EV Collect‐EVs and Nourish‐EVs all showed effect on cell metabolism (15.8%, 8.7%, 7.8% and 4.7% increase in Alamar Blue signal compared to untreated cells, respectively) (Figure S4A), no EV treatment showed increase in cell counts compared to serum‐starved fHDF/TERT166 cells (Figure S4B).

FIGURE 7.

Effect of BM‐MSC/TERT292‐derived EVs on fHDF/TERT166 proliferation and metabolism upon enrichment in different collection media. (A) Alamar Blue measurement in fHDF/TERT166 cells upon 72 h treatment in Starvation Medium with BM‐MSC/TERT292‐derived EVs enriched from different collection media. Alamar Blue values were expressed as a percentage of the mean signal from untreated cells (Starvation Medium, set at 100%). (B) Automatic cell count measurement of fHDF/TERT166 cells upon 72 h treatment in Starvation Medium with BM‐MSC/TERT292‐derived EVs enriched from different collection media. fHDF/TERT166 cells were treated with a fixed EV concentration of 1E+09 p/mL. Data are represented as mean ± SD. One‐way Anova was used, and significance is shown as **p < 0.01, ***p < 0.001, ****p < 0.0001 (n = 4 technical replicates). (C) Representative microscopy images of Vybrant dye cycle violet stained fHDF/TERT166 cells upon 72 h treatment in Starvation Medium with BM‐MSC/TERT292‐derived EVs enriched from different collection media. fHDF/TERT166 cells were treated with a fixed EV concentration of 1E+09 p/mL. Scale bars: 250 µm.

When evaluating the impact of the different UCM controls on proliferating fHDF/TERT166 cells, OptiMEM and RoosterCollect‐EV media exhibited no detectable metabolic (Figure 8A) or proliferative (Figure 8B) activity, in line with all the previous observations in bioassays and unlike their respective EV preparations. In contrast, both RoosterNourish‐MSC‐XF and Mesencult‐ACF unconditioned media controls showed higher metabolic and proliferative effects compared to the respective BM‐MSC/TERT292‐derived EVs, with an increase of 14.9% and 45.9% in Alamar Blue signal and of 49.9% and 123.5% in cell count respectively (Figure 8A,B). Consistent with these findings, microscopy of serum‐starved fHDF/TERT166 cell 72 h post‐ treatment confirmed the metabolic and proliferative effects observed for the UCM conditions (Figure 8C). Taken together, these data demonstrate that the choice of collection medium substantially influences the apparent metabolic and cell proliferative effects attributed to EV preparations.

FIGURE 8.

Effect of UCM controls on fHDF/TERT166 proliferation and metabolism. (A) Alamar Blue measurement in fHDF/TERT166 cells upon 72 h treatment in Starvation Medium with different UCM controls. Alamar Blue values were expressed as a percentage of the mean signal from untreated cells (Starvation Medium, set at 100%). (B) Automatic cell count measurement of fHDF/TERT166 cells upon 72 h treatment in Starvation Medium with different UCM controls. fHDF/TERT166 cells were treated with the same UCM volume of the BM‐MSC/TERT292‐derived EV enriched in the respective medium (adjusted for the enrichment factor). Data are represented as mean ± SD; One‐way Anova was used, and significance is shown as **p < 0.01, ****p < 0.0001 (n = 4 technical replicates). (C) Representative microscopy images of Vybrant dye cycle violet stained fHDF/TERT166 cells upon 72 h treatment in Starvation Medium with the same UCM volume of the BM‐MSC/TERT292‐derived EV enriched in the respective medium (adjusted for the enrichment factor). Scale bars: 250 µm.

3.8. Anti‐Oxidant and Neoangiogenic Potential of BM‐MSC/TERT292‐Derived EVs and Unconditioned Media Controls

Finally, we assessed the anti‐oxidant and the neoangiogenic properties of BM‐MSC/TERT292‐derived EVs enriched in different collection media and of the corresponding UCM controls. First, fHDF/TERT166 were treated with BM‐MSC/TERT292‐derived EVs enriched in the different collection media at a particle concentration of 1E+09 p/mL and 2.5E+08 p/mL or with 5% 20 mM HEPES as a negative control. Twenty‐four hours after treatment, oxidative stress was induced in EV‐treated fHDF/TERT166 cells by H₂O₂ treatment, and reactive oxygen species (ROS) production was evaluated by measurement of H2DCFDA staining. Surprisingly, upon treatment with BM‐MSC/TERT292‐derived EVs at a concentration of 1E+09 p/mL, EVs enriched in RoosterNourish‐MSC‐XF and in Mesencult‐ACF showed a slight but significant pro‐oxidative effect (Figure 9A, Figure S5A), while OptiMEM‐EVs were the only preparation showing an anti‐oxidant effect in H₂O₂‐tretaed fHDF/TERT166 (Figure 9A, Figure S5B). Consistent with the previously observed lack of anti‐inflammatory activity of Nourish‐EVs at 2E+09 p/mL (Figure 5A), these data further suggest that an excessively high particle concentration may pose stress on the recipient cells, thereby affecting the therapeutic efficacy of EV preparations (Hagey et al. 2023).

FIGURE 9.

Effect of BM‐MSC/TERT292‐derived EVs on ROS production and tube formation upon enrichment in different collection media. (A) H2DCFDA measurement in BM‐MSC/TERT292 EVs pre‐treated fHDF/TERT166 cells upon H₂O₂—mediated oxidative stress induction. H2DCFDA values were expressed as a percentage of the mean signal from H₂O₂ only—treated cells (Hepes + H₂O₂, set at 100%). A fixed concentration of 5% 20 mM Hepes was kept constant in all wells during the BM‐MSC/TERT292‐derived EVs pre‐treatment. fHDF/TERT166 cells were treated with fixed EV concentrations of 2.5 E+08 and 1E+09 p/mL. Data are represented as mean ± SD. One‐way Anova was used, and significance is shown as **p < 0.01, ***p < 0.001, ****p < 0.0001 (n = 4 technical replicates). (B) Evaluation of the tube‐formation ability by HUVEC/TERT2 cells terms of number of segments, (C) number of nodes and (D) total segment length upon BM‐MSC/TERT292‐derived EVs treatment. HUVEC/TERT2 cells were treated with a fixed concentration of 1E+09 p/mL. Data are represented as mean ± SD. One‐way Anova was used, and significance is shown as **p < 0.01, ***p < 0.001, ****p < 0.0001 (n = 3 technical replicates). (E) Representative microscopy images of tube formation and image analysis by the Angiogenesis Analyzer plug in of ImageJ in HUVEC/TERT2 cells upon 6 h treatment with BM‐MSC/TERT292‐derived EVs enriched from different collection media. Nodes are shown as red dots surrounded by a blue circle, branches are coloured in blue, branches in green, master segments in yellow and mashes in light blue.

Next, the angiogenic potential of BM‐MSC/TERT292‐derived EVs enriched in different collection media, along with their corresponding UCM controls was assessed by evaluating the tube‐formation potential of HUVEC/TERT2 cells.

EVs enriched in RoosterCollect‐EV, OptiMEM and RoosterNourish‐MSC‐XF induced a comparable and significant increase in multiple angiogenic parameters, including the number of segments (Figure 9B), nodes (Figure 9C), junctions (Figure S5C), total length of the segments (Figure 9D) and total length of the branches (Figure S5D). In contrast, Mesencult‐EVs showed the lowest, non‐significant increase across all parameters (Figure 9B–D and Figure S5C,D), highlighting the influence of the collection medium on the EV therapeutic potential in an assay‐dependent manner. Representative images further illustrate these differences (Figure 9E).

When evaluating the antioxidant effect of UCM controls in the H₂O₂‐based fHDF/TERT166 model, none of the collection media exhibited significant changes in the H2DCFDA measurement compared to the positive H₂O₂‐treated control (Figure 10A, Figure S6A). Similarly, UCM controls did not significantly affect tube‐formation in HUVEC/TERT2 cells (Figure 10B–D, Figure S6B,C), which was confirmed by microscopy images (Figure 10E). Taken together, these findings demonstrate that the choice of collection medium significantly influences the biological activity of MSC‐derived EVs, while the effects of the UCM controls remain assay‐dependent.

FIGURE 10.

Effect of UCM controls on ROS production and tube formation. (A) H2DCFDA measurement in UCM pre‐treated fHDF/TERT166 cells upon H₂O₂—mediated oxidative stress induction. H2DCFDA values were expressed as a percentage of the mean signal from H₂O₂ only—treated cells (Hepes + H₂O₂, set at 100%). A fixed concentration of 5% 20 mM Hepes was kept constant in all wells during the UCM pre‐treatment. fHDF/TERT166 cells were treated with the same UCM volume of the BM‐MSC/TERT292‐derived EV at 1.0E+09 p/mL concentration enriched in the respective medium (adjusted for the enrichment factor). Data are represented as mean ± SD. One‐way Anova was used, and significance is shown as **p < 0.01, ***p < 0.001, ****p < 0.0001 (n = 4 technical replicates). (B) Evaluation of the tube‐formation ability by HUVEC/TERT2 cells terms of number of segments, (C) number of nodes and (D) total segment length upon UCM controls treatment. HUVEC/TERT2 cells were treated with the same UCM volume of the BM‐MSC/TERT292‐derived EV enriched in the respective medium (adjusted for the enrichment factor). Data are represented as mean ± SD. One‐way Anova was used, and significance is shown as **p < 0.01, ***p < 0.001, ****p < 0.0001 (n = 3 technical replicates). (E) Representative microscopy images of tube formation and image analysis by the Angiogenesis Analyzer plug in of ImageJ in HUVEC/TERT2cells upon 6 h treatment with different UCM controls. Nodes are shown as red dots surrounded by a blue circle, branches are coloured in blue, branches in green, master segments in yellow and mashes in light blue.

4. Discussion

In this study, we sought to address a critical and often underexplored variable in the EV field—the choice of collection medium—and to determine how it influences the yield, biochemical properties and biological activity of hTERT‐immortalised MSC‐derived EVs. While great effort has been put in the understanding of how different isolation strategies affect the biological and biochemical properties of the isolated EVs (Brennan et al. 2020; Dong et al. 2020; Welsh et al. 2024; Williams et al. 2023; Yamashita et al. 2016; Zhao et al. 2022), to our knowledge, only a few studies have compared how different collection media influence the final properties of isolated EVs (Figueroa‐Valdés et al. 2021; Suthumchai et al. 2025). Moreover, no comprehensive parallel evaluation has been yet conducted that compares processed unconditioned media with EV‐preparations enriched in the same media, particularly with respect to their bioactivity.

Using hTERT‐immortalised MSCs reduced donor‐ and passage‐related variability (Giebel 2017; Jérémy et al. 2024; Maumus et al. 2020; Scheiber et al. 2022) and ensured consistent EV production, as these cells retain key therapeutic characteristics of primary MSCs (Brancolini et al. 2026; Hindle et al. 2024). This strategy enabled us to focus specifically on the impact of expansion media and collection media combinations, isolating its effects from intrinsic cellular variability. Indeed, a major source of misinterpretation in EV studies is the lack of appropriate medium‐only controls (Nguyen et al. 2020; Welsh et al. 2024), since residual proteins, lipids and nanoparticles can mimic EV‐associated activity (Auber et al. 2019; Kronstadt et al. 2023; Shekari et al. 2023).

By applying the full enrichment workflow to unconditioned medium, we generated appropriate controls and compared their physicochemical and biological profiles with EV‐containing preparations, providing a rigorous assessment of medium‐derived versus EV‐specific effects. This approach revealed that, despite the challenges associated with the adaptation of MSCs to different media, BM‐MSC/TERT292 showed adaptability to different expansion media and collection media. However, differences in MSC adaptability to different media may arise due to their origin from different donor or tissues (Winkel et al. 2020). Moreover, the choice of collection medium severely affected particle release by BM‐MSC/TERT292 cells, as well as particle purity. Overall, it is of interest to see how EV samples enriched in Mesencult‐ACF and RoosterNourish‐MSC‐XF with similar purity values showed different biochemical properties, denoting the issue of calculating EV‐sample purity only based on protein content and particle concentrations, thus overestimating the purity of samples enriched in non‐EV particles.

The surface marker profiling by MACSPlex EV Kit MSC (Fernández‐Pérez et al. 2025) successfully confirmed the MSC‐origin of the different EV preparations. The analysis further confirmed the co‐isolation of hPL‐derived components in Nourish‐EVs, as well as the lower purity of Mesencult‐EVs compared to OptiMEM‐ and EV Collect‐EVs, as demonstrated by the lower MSC and EV‐specific marker expression. Overall, these data highlight the importance of such analysis not only to validate the MSC‐origin of the enriched EVs, but as a valid tool to better define the composition and purity of the different EV preparations (Nguyen et al. 2024).

Next, when comparing the biological activity of BM‐MSC/TERT292‐derived EVs to similarly processed UCM, no major differences in terms of bioactivity were measurable between EV preparations and UCM controls enriched in RoosterNourish‐MSC‐XF, as well as for those enriched in Mesencult‐ACF (Table 1). To note, when calculating the activity‐to‐protein ratio, Mesencult‐ACF UCM as lower protein amount was needed to reach comparable effects to these of the respective EV preparation. On the other hand, clear differences were quantified when comparing EV preparations enriched in OptiMEM and RoosterCollect‐EV, as well as upon comparison with the respective UCM controls, which did not show activity in any of the tested bioactivity assays. In detail, BM‐MSC/TERT292 EVs enriched in Mesencult‐ACF and the respective unconditioned media control showed comparable anti‐inflammatory activity, suggesting that not only co‐enriched non‐EV components from fetal bovine serum mediates anti‐inflammatory activity (Kronstadt et al. 2023), but also components from serum‐free and xeno‐free media can contaminate the EV preparation and mediate an EV‐independent activity. On the other hand, RoosterNourish‐MSC‐XF UCM showed a significant gap closure potential not measurable when treating with other UCM controls, in accordance with the known in vitro activity of human cord blood platelet in wound healing (Losi et al. 2019).

TABLE 1.

Summary of the biological effect of BM‐MSC/TERT292‐derived EVs upon enrichment in different collection media and of the respective UCM controls.

	Collection media	Nourish		EV Collect		Mesencult		OptiMEM
	Sample	EVs	UCM	EVs	UCM	EVs	UCM	EVs	UCM
AINF	Activity (%)	53.7	9.6	59.0	/	68.4	35.8	73.5	/
	Activity/Protein (µg)	3.7	1.0	10.7	/	2.8	9.0	17.2	/
AFIB	Activity (%)	39.5	36.4	36.3	/	23.3	29.0	/	/
	Activity/Protein (µg)	1.7	1.8	4.1	/	0.6	6.0	/	/
WH	Activity (%)	12.0	9.5	32.1	/	28.8	5.5	21.7	/
	Activity/Protein (µg)	0.8	1.0	5.8	/	1.2	1.4	5.1	/
Metabolic	Activity (%)	6.0	14.9	21.9	/	30.8	25.9	12.9	/
	Activity/Protein (µg)	0.4	1.6	4.0	/	1.2	6.5	3.0	/
Proliferative	Activity (%)	17.4	49.9	24.0	/	49.9	123.5	/	/
	Activity/Protein (µg)	1.2	5.6	4.4	/	2.0	31.1	/	/
ROS	Activity (%)	20.6% *	/	/	/	27.4%*	/	/	/
	Activity/Protein (µg)	1.4	/	/	/	1.1	/	/	/
Tube formation	Activity (%)	469.3	/	533.3	/	211.6	/	521.4	/
	Activity/Protein (µg)	32.4	/	97.8	/	8.5	/	121.3	/

Note: The biological activity of EVs (at 1E+9 p/mL) and UCM controls is given as % of the adequately stimulated/treatment controls. The activity to protein ration (Activity/Protein) was estimated by normalising the Activity (%) to the protein amount (µg) used for each EV treatment (1E+09 p/mL) or for the respective UCM control treatment (Volume). For the ROS production assay, * indicates a pro‐oxidant effect. For the tube formation assay, the average of the activity in terms of number of segments, number of junctions, number of nodes, total segments length and total branching length was used to indicate and used for the calculation of the activity to protein ratio.

Both BM‐MSC/TERT292 EVs and the UCM controls enriched in Mesencult‐ACF and RoosterNourish‐MSC‐XF showed significant anti‐fibrosis activity, suggesting that not only the contamination of EV preparation with hPL‐derived components (Seidelmann et al. 2021), but also co‐enrichment of factors from serum‐free and hPL‐free medium could affect α‐SMA expression in TGF‐β stimulated fibroblasts. To note, we observed slightly higher activity in terms of pro‐proliferative and metabolic effect in unconditioned media controls enriched with RoosterNourish‐MSC‐XF and Mesencult‐ACF media compared to the respective BM‐MSC/TERT292 EV preparations. This aligns with our prior observations that co‐isolation of medium derived components by TFF enrichment could affect the redout of the biological assay used for EV testing non only when using hPL containing‐media (RoosterNourish‐MSC‐XF), but also serum‐free and xeno‐free media (Mesencult‐ACF) (Whittaker et al. 2020). Moreover, upon enrichment of BM‐MSC/TERT292‐derived EVs in OptiMEM‐ and RoosterCollect‐EV, we showed the EVs in OptiMEM mediated the highest anti‐inflammatory potential, while showing no anti‐fibrosis effect. EVs enriched in RoosterCollect‐EV, on the contrary, showed lower anti‐inflammatory activity, but higher wound healing, proliferative and anti‐fibrosis capacity (Table 1).

We therefore confirmed that even upon TFF enrichment of EVs in commonly used serum‐reduced, xeno‐free EV collection medium—such as OptiMEM and RoosterCollect‐EV (Bost et al. 2022; Gaesser et al. 2024; Jeske et al. 2022; Karttunen et al. 2022)—the biological activity of the final EV preparation differs depending on the combination of growth and collection media used for EV enrichment. No increased cell counts or increased cellular oxidoreductases (metabolic) activity was measurable upon treatment of fHDF/TERT166 cells with OptiMEM unconditioned medium in none of the tested conditions. These results show the importance of the technique used for EV enrichment and the experimental set‐up for EV testing, as increased proliferation and myogenic differentiation potential were observed by Hanson et al., upon treatment of GW4869‐treated murine myoblasts with non‐EV Liquid Chromatography fractions and polymer precipitated EV preparations from murine C2C12 myotubes cultured in OptiMEM for EV collection (Hanson et al. 2023).

Interestingly, only BM‐MSC/TERT292‐derived EVs, but not the corresponding UCM controls, exhibited significant activity in both angiogenesis and ROS‐related assays. This observation suggests that co‐enriched factors within the EV preparations might confer different biological activities in these in vitro assays (Gimona et al. 2021). Furthermore, our data indicate that the effective dose of EV preparations is highly dependent on the specific biological activity being assessed. The lack of anti‐inflammatory activity of Nourish‐EVs at 2E+09 p/mL, together with the observed increase in ROS production in H₂O₂‐stressed cells following treatment with Nourish‐EV and Mesencult‐EV at 1E+09 p/mL, suggests that excessively high concentrations of EVs‐particularly those derived from lipid‐rich media, may induce lipid peroxidation. Such effects could mask or even counteract the intended therapeutic activity. This is consistent with previous reports and underscores the importance of optimised collection conditions, as also reflected in the availability of dedicated collection media formulations (Chiaradia et al. 2021; Cho et al. 2026). Taken together, these findings highlight the need for careful titration of EV dosing in both in vitro and in vivo, taking into account variability introduced by different collection media, as well as production and purification strategies (Witwer et al. 2019)

Further work is still needed to define the molecular mechanisms by which medium composition modulates EV properties (Li et al. 2015). Proteomic, lipidomic and transcriptomic analyses could clarify how specific formulations influence cargo loading, membrane composition and EV‐related signalling pathways. Extending this framework to 3D cultures or bioreactor systems will also determine whether the observed medium‐dependent effects persist under scalable, dynamic production conditions.

In conclusion, our study provides clear experimental evidence that the collection medium composition plays a decisive role in shaping both the quantitative and qualitative aspects of BM‐MSC/TERT292‐derived EVs. We showed that medium‐derived components and co‐isolated factors may affect the therapeutic properties of the isolated EVs, and this aspect needs to be carefully considered when translating MSC‐EV production into disease‐specific applications. Furthermore, our findings indicate that, for each therapeutic indication, thorough evaluation and comparison are necessary to determine the most potent EV preparations. Overall, these results not only clarify a key source of variability in EV research but also provide practical guidance towards the development of reproducible, scalable and clinically translatable EV‐based therapeutics, in alignment with Good Manufacturing Practice (GMP) requirements (Humbert et al. 2025; Watson et al. 2018). Ultimately, the principles demonstrated here—the inclusion of proper controls, the use of stable cell sources, and the standardisation of assays—can be extended beyond MSC‐EVs to other EV‐producing systems. As the field moves towards regulatory maturity, such methodological rigour will be indispensable for establishing the quality, consistency and functional relevance of extracellular vesicle‐based products in regenerative medicine and beyond.

Author Contributions

Alessia Brancolini: investigation, writing – original draft, methodology, validation, visualization, writing – review and editing, formal analysis, data curation. Madhusudhan Reddy Bobbili: investigation, visualization, writing – review and editing, methodology, formal analysis, data curation, supervision. Matthias Wieser: investigation, supervision, validation, methodology. Johanna Gamauf: investigation, methodology, validation. Marieke Theodora Roefs: investigation, methodology, validation. Elsa Arcalis: investigation, methodology, visualization, formal analysis. Silke Aldrian: resources, writing – review and editing. Sebastien Couillard‐Despres: resources, writing – review and editing, methodology. Lara Bieler: methodology, writing – review and editing, resources. Johannes Grillari: conceptualization, funding acquisition, writing – review and editing, data curation, supervision, resources, project administration. Regina Grillari‐Voglauer: data curation, supervision, resources, project administration, writing – review and editing, funding acquisition, conceptualization.

Funding

This project has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska‐Curie grant agreement No. 101072766.

Conflicts of Interest

J.G. and R.G.V. are cofounders and shareholders of Evercyte GmbH, Vienna, Austria, and of TAmiRNA GmbH, Vienna, Austria. Evercyte GmbH and TAmiRNA GmbH provided support in the form of salaries for authors A.B., R.G.V., M.W., J.G.A. and M.R. All other authors have no commercial, proprietary or financial interest in the products or companies described in this article. RoosterBio kindly provided the RoosterBooster‐MSC‐XF and RoosterCollect‐EV media for this study. RoosterBio had no influence on experimental design, data acquisition, or interpretation.

Supporting information

Supplementary Figures: jev270298‐sup‐0001‐FigureS1‐S6.docx

Data Availability Statement

The paper and its supplementary materials provide the data that support the findings of this study.