ABSTRACT 
 
 
 

 
Title of Dissertation: MECHANICALLY REGULATED 

PRODUCTION FOR IMPROVED YIELD 
AND THERAPEUTIC POTENCY OF 
MESENCHYMAL STEM CELL 
EXTRACELLULAR VESICLES   

  
 Emily H. Powsner 

Doctor of Philosophy, 2025 
  
Dissertation directed by: Associate Professor, Dr. Steven M. Jay, 

Fischell Department of Bioengineering 
 
 

Mesenchymal stem cells (MSCs) are among the most extensively studied cell types for 

regenerative therapies due to their intrinsic pro-angiogenic and anti-inflammatory 

properties. However, MSC-based therapies face persistent challenges including risks 

of tumorigenicity and spontaneous differentiation. Extracellular vesicles (EVs)- cell-

secreted nanovesicles that shuttle bioactive cargo reflective of their parent cells- offer 

a promising alternative by recapitulating the therapeutic benefits of MSCs while 

avoiding cell-associated risks. Accordingly, the therapeutic relevance of MSC EVs 

extends across numerous indications including cardiovascular disease, inflammatory 

diseases, cancer, and wound healing. Despite their potential, the clinical development 

of MSC EVs remains limited by donor variability, low innate potency, and lack of 

rationally designed scalable manufacturing approaches. Moreover, conventional 



  

strategies to address low potency (e.g. genetic engineering, exogenous cargo loading) 

are generally incompatible with large-scale biomanufacturing and substantially 

increase complexity and cost. Previous studies have shown that bone marrow-derived 

MSCs (BMMSCs) respond favorably to mechanical cues, yet both the cells and their 

EVs are hindered by donor variability and early senescence. To overcome these 

barriers, we leveraged induced pluripotent stem cell-derived MSCs (iMSCs) as a 

renewable and consistent EV source, and harnessed mechanical cues- specifically 

substrate stiffness and cellular confinement- and bioreactor-based cell culture to 

improve EV production and potency. We demonstrated that culturing iMSCs on soft 

substrates enhances both EV yield and angiogenic activity in a stiffness-dependent 

manner, paralleling BMMSC responses. Additionally, we showed that high degrees of 

cell confinement promote superior angiogenic function of the secreted iMSC EVs. 

Building on these findings, we developed and optimized a 3D-printed perfusion 

bioreactor incorporating cell confinement, achieving a 67-fold increase in iMSC EV 

production compared to traditional flask culture while preserving the enhanced 

bioactivity conferred by confinement alone. Importantly, treatment with the bioreactor-

generated EVs accelerated wound healing in a diabetic mouse model compared to 

flask-generated EVs and the vehicle control. Overall, this work establishes 

mechanobiology-based EV manufacturing platforms that address key translational 

bottlenecks in MSC EV therapeutic development. By integrating renewable cell 

sources with scalable, potency-enhancing biomanufacturing strategies, these findings 

advance the path toward consistent, large-scale production of MSC EVs suitable for 

clinical translation.  



  

 
 
 

MECHANICALLY REGULATED PRODUCTION FOR IMPROVED YIELD 
AND THERAPEUTIC POTENCY OF MESENCHYMAL STEM CELL 

EXTRACELLULAR VESICLES 
 
 
 

by 
 
 

Emily H. Powsner 
 
 
 
 
 

Dissertation submitted to the Faculty of the Graduate School of the  
University of Maryland, College Park, in partial fulfillment 

of the requirements for the degree of 
Doctor of Philosophy 

2025 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Advisory Committee: 
Associate Professor Dr. Steven M. Jay, Chair 
Professor Dr. John P. Fisher 
Associate Professor Dr. Kimberly M. Stroka 
Professor Dr. Ian White 
Associate Professor Dr. Amy Karlsson, Dean’s Representative 



  

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

© Copyright by 
Emily Powsner 

2025 
 
 
 
 
 
 
 
 
 
 
 
 
 



 

 

ii 
 

Dedication 

I would like to dedicate this dissertation to my parents. Your work ethics and 

ambition are inspiring. Thank you for always providing me with everything I have 

needed to succeed. Your love and support do not go unnoticed, and I wouldn’t be 

here today without you both. 

 



 

 

iii 
 

Acknowledgements 

First and foremost, I would like to acknowledge my advisor, Dr. Steven Jay, for his 

incredible mentorship. I entered the program directly after undergrad and matched 

into your lab as the youngest member at the time by about 4 years, with no 

experience beyond undergraduate research. You met me where I was and remained 

patient, offering sound advice on everything from research to classes to careers- a 

statement that remains true four years later. While I tended towards hesitancy in 

many moments, you pushed me outside of my comfort zone which has allowed me 

to grow. Your genuine interest and commitment to my career growth throughout 

this entire process has never wavered and because of that, I can confidently say that 

I feel prepared to take this next step. Thank you. 

 

I would also like to acknowledge the rest of my committee members, Dr. John 

Fisher, Dr. Kimberly Stroka, Dr. Ian White, and Dr. Amy Karlsson, as well as Dr. 

Hannah Zierden who served on my proposal committee, for your time, advice, and 

you and your labs’ expertise that led me through my research and helped me get to 

where I am today. 

 

To my past lab mates, Dan, Eli, and Steph, thank you for welcoming me into the 

lab with open arms and for taking the time to teach me countless skills. I would like 

to especially acknowledge Steph, as your work and the many techniques you taught 

me laid the foundation for this thesis. To my current lab mates, Nick, Raith, and 

Talia, thank you for always being down to chat and for providing advice on all 



 

 

iv 
 

things science when I needed it. My animal study would also not have been possible 

without all your help, so thank you. Also, a special thanks to Ian Smith from Dr. 

Stroka’s lab for your expertise and guidance with photolithography and 

mechanobiology, and Pieper Holeman from Dr. Fisher’s lab for your expertise and 

for 3D printing the many bioreactors I went through. I would also like to thank my 

undergrad. Kristin. Your contributions to the work here were invaluable and your 

constant excitement for science was contagious. 

 

My time here would not have been the same, and not nearly as enjoyable, if it 

weren’t for my friends in Bioe, the almost daily lunches with our lunchtime group 

chat, and the frequent third floor coffee runs. To Jackie and Shannon, thank you for 

always being there for anything and everything and for the many cocktails, 

brunches, and girls’ nights that fill my weeks with so much joy and much needed 

stress relief. To my soul sisters- Rachel, Grace, Kerry, and Sam- thank you for 

being constants in my life and for giving me reasons to take breaks from work with 

all the fun trips we do throughout the year. To my family- especially my parents- 

thank you for your unwavering support and for being a place that I know I can 

always go back to. I love and appreciate you all more than you’ll ever know. 

 

This work was supported by the National Institutes of Health (HL141611, 

HL159590 to SMJ; HL007698 to EHP; HD112031 to JPF; R35GM142838 to 

KMS), the National Science Foundation (1750524 to SMJ), and the Maryland Stem 

Cell Research Fund (2023-MSCRF-6126 to SMJ). 



 

 

v 
 

Table of Contents 

 

Dedication ............................................................................................................... ii 

Acknowledgements ................................................................................................ iii 

Table of Contents .................................................................................................... v 

List of Figures ........................................................................................................ ix 

List of Commonly Used Abbreviations .................................................................. 1 

Chapter 1: Introduction ........................................................................................... 3 

Chapter 2: Therapeutic Potential of Extracellular Vesicles and Progress Towards 

Clinical Translation ................................................................................................. 7 

2.1 Extracellular Vesicle Biology ....................................................................... 7 

2.1.1 EV Origin and Biogenesis ..................................................................... 7 

2.1.2 EV Composition .................................................................................... 8 

2.2 Extracellular Vesicle Isolation and Characterization .................................. 10 

2.3 MSC Extracellular Vesicles as Therapeutics .............................................. 13 

2.3.1 Impaired Wound Healing: Diabetic Wounds ....................................... 14 

2.3.2 EVs in Clinical Trials ........................................................................... 18 

2.4 Challenges to the Clinical Translation of Extracellular Vesicles ............... 18 

2.4.1 Cell Source and Donor Variability ...................................................... 19 

2.4.2 Scalable Manufacturing ....................................................................... 20 

2.4.3 Therapeutic Potency ............................................................................. 22 

2.5 Potential Solutions to Translational Challenges ......................................... 23 

2.5.1 Induced Pluripotent Stem Cell-Derived MSCs (iMSCs) ..................... 23 



 

 

vi 
 

2.5.2 Mechanical Cues: Substrate Stiffness .................................................. 24 

2.5.3 Mechanical Cues: Cell Confinement ................................................... 27 

2.5.4 Bioreactor-Based Dynamic Culture ..................................................... 29 

2.6 Conclusion .................................................................................................. 30 

Chapter 3: Mesenchymal Stem Cell Extracellular Vesicle Vascularization 

Bioactivity and Production Yield are Responsive to Cell Culture Substrate 

Stiffness................................................................................................................. 32 

3.1 Introduction ................................................................................................. 32 

3.2 Methods....................................................................................................... 34 

3.2.1 Cell Culture .......................................................................................... 34 

3.2.2 Stiffness Device Fabrication, Preparation, and Seeding ...................... 35 

3.2.3 EV Isolation ......................................................................................... 37 

3.2.4 EV Characterization ............................................................................. 38 

3.2.5 Cell Viability Assay ............................................................................. 40 

3.2.6 Quantifying Cell Growth and Proliferation on PDMS ........................ 40 

3.2.7 Tube Formation Assay ......................................................................... 41 

3.2.8 Gap Closure Assay ............................................................................... 41 

3.2.9 EV miRNA Isolation and Screening .................................................... 42 

3.2.10 EV-Treated HUVEC mRNA Isolation and Screening ....................... 43 

3.2.11 Anti-Inflammatory Assay .................................................................. 44 

3.2.12 Statistical Analysis ............................................................................. 44 

3.3 Results ......................................................................................................... 45 

3.3.1 Substrate Fabrication and EV Characterization ................................... 45 



 

 

vii 
 

3.3.2 Substrate Stiffness Influences BMMSC EV Production and Bioactivity

....................................................................................................................... 49 

3.3.3 Mixing Sylgard 184 and Sylgard 527 PDMS Provides a Wider Range of 

Elastic Moduli ............................................................................................... 51 

3.3.4 iMSC EVs are Mechanoresponsive and Comparable Pro-Angiogenic 

Treatments to BMMSC EVs ......................................................................... 53 

3.3.5 Investigating Potential Mechanistic Changes as a Result of Decreased 

Substrate Stiffness ......................................................................................... 58 

3.4 Discussion ................................................................................................... 61 

Chapter 4: Perfusion Bioreactor Culture Incorporating Mechanical Confinement 

Enhances Mesenchymal Stem Cell Extracellular Vesicle Production and Wound 

Healing Potential ................................................................................................... 69 

4.1 Introduction ................................................................................................. 69 

4.2 Methods....................................................................................................... 71 

4.2.1 Cell Culture .......................................................................................... 71 

4.2.2 Micropillar Device Fabrication, Preparation, and Seeding .................. 73 

4.2.3 3D-Printed Bioreactor Fabrication and Setup ...................................... 75 

4.2.4 EV Separation ...................................................................................... 76 

4.2.5 EV Characterization ............................................................................. 77 

4.2.6 EV Production Quantification .............................................................. 78 

4.2.7 Tube Formation Assay ......................................................................... 79 

4.2.8 Gap Closure Assay ............................................................................... 79 

4.2.9 Anti-Inflammatory Assay .................................................................... 80 



 

 

viii 
 

4.2.10 Molecular Content Analysis .............................................................. 80 

4.2.11 Diabetic Mouse Wound Healing Model ............................................ 82 

4.2.12 Statistical Analysis ............................................................................. 85 

4.3 Results ......................................................................................................... 85 

4.3.1 Mechanical Confinement of MSCs Augments EV Bioactivity ........... 85 

4.3.2 Incorporation of MSC Confinement Within a Perfusion Bioreactor 

Enhances EV Production with Retained Bioactivity .................................... 90 

4.3.3 EVs From Confined iMSCs in Perfusion Bioreactor Culture Improve 

Wound Repair in a Diabetic Mouse Model .................................................. 95 

4.3.4 Mechanical Confinement Affects the Molecular Cargo of iMSC EVs 98 

4.4 Discussion ................................................................................................. 101 

Chapter 5: Conclusions and Future Directions ................................................... 109 

5.1 Conclusions ............................................................................................... 109 

5.2 Future Directions ...................................................................................... 111 

5.3 Contributions ............................................................................................ 112 

5.3.1 Peer-Reviewed Publications .............................................................. 112 

5.3.2 Conference Presentations ................................................................... 115 

5.3.3 Contributions to the Field and to the Jay Lab .................................... 116 

Bibliography ....................................................................................................... 118 



 

 

ix 
 

List of Figures 

Figure 3.1: Substrate and BMMSC EV characterization. ..................................... 47 

Figure 3.2: Incubation in media does not significantly affect the elastic modulus of 

the PDMS substrates. ............................................................................................ 48 

Figure 3.3: Substrate stiffness influences BMMSC EV production and bioactivity.

............................................................................................................................... 50 

Figure 3.4: Mixed Sylgard 184 and Sylgard 527 PDMS substrate characterization.

............................................................................................................................... 52 

Figure 3.5: Softer 184:527 PDMS substrates improve the angiogenic bioactivity of 

BMMSC EVs. ....................................................................................................... 53 

Figure 3.6: iMSC EV characterization. ................................................................. 55 

Figure 3.7: Substrate stiffness affects the pro-angiogenic effect of iMSC EVs 

comparably to BMMSC EVs. ............................................................................... 57 

Figure 3.8: Substrate stiffness does not have a significant effect on the anti-

inflammatory effects of iMSC EVs. ..................................................................... 58 

Figure 3.9: HUVECs treated with 5E9 EVs/mL of soft substrate-generated iMSC 

EVs result in the upregulation of certain angiogenesis-related mRNAs. ............. 59 

Figure 3.10: miRNA qPCR array data of differentially expressed MSC EV-

associated miRNAs in flask vs. soft substrate-generated EVs). ........................... 61 

Figure 4.1: Confinement device fabrication ......................................................... 87 

Figure 4.2: iMSC EV characterization. ................................................................. 87 

Figure 4.3: iMSC EV bioactivity is responsive to producer cell confinement  .... 88 

Figure 4.4: Anti-inflammatory effects of EVs from confined producer cells ....... 89 



 

 

x 
 

Figure 4.5: BMMSC EV bioactivity is responsive to producer cell confinement. 89 

Figure 4.6: Perfusion bioreactor culture incorporating cell confinement enhances 

iMSC EV yield ...................................................................................................... 91 

Figure 4.7: Improved iMSC EV wound healing activity is maintained in the 

confinement-bioreactor system ............................................................................. 93 

Figure 4.8: Anti-inflammatory effect of EVs from confined cells in static and 

optimized bioreactor culture ................................................................................. 94 

Figure 4.9: Figure 5: Bioreactor EVs improve tissue repair in a diabetic mouse 

wound model. ........................................................................................................ 97 

Figure 4.10: Comparison of miRNA profiles between iMSC EVs from different 

culture conditions. ................................................................................................. 99 

Figure 4.11: Proteomics analysis for iMSC EVs from different culture conditions

............................................................................................................................. 101 



 

 

1 
 

List of Commonly Used Abbreviations 

BCA   Bicinchoninic Acid 

BMMSC  Bone Marrow-Derived Mesenchymal Stem cells 

cGMP   Current Good Manufacturing Practices 

DFU   Diabetic Foot Ulcer 

DMEM  Dulbecco’s Modified Eagle Medium 

DNA   Deoxyribonucleic Acid 

ECM   Extracellular Matrix 

ELISA   Enzyme-Linked Immunosorbent Assay 

EV   Extracellular Vesicle 

FBS   Fetal Bovine Serum 

HDMEC  Human Dermal Microvascular Endothelial Cell 

HUVEC  Human Umbilical Vein Endothelial Cell 

IL-6   Interleukin 6 

iMSC Induced Pluripotent Stem Cell-Derived Mesenchymal Stem 

Cell 

iPSC   Induced Pluripotent Stem Cell 

miRNA  Micro RNA 

MSC   Mesenchymal Stem Cell 

NTA   Nanoparticle Tracking Analysis 

PDMS   Polydimethylsiloxane 

RNA   Ribonucleic Acid 



 

 

2 
 

RT-qPCR Reverse Transcription Quantitative Polymerase Chain 

Reaction 

TCPS   Tissue Culture Polystyrene 

TFF   Tangential Flow Filtration 

TNF-𝛼		 	 Tumor Necrosis Factor Alpha 

VEGF   Vascular Endothelial Growth Factor 

  



 

 

3 
 

Chapter 1: Introduction 

Cell-based therapies, particularly mesenchymal stem cell (MSC)-based therapies, 

have gained significant traction within the field of regenerative medicine for both 

their ability to differentiate into multiple lineages and their innate therapeutic 

properties. However, concerns with low and transient engraftment at the site of 

injury in addition to safety, tumorigenicity, and spontaneous differentiation has led 

to slow progress [1, 2]. With that said, substantial research has pointed to a 

paracrine effect, specifically via secreted extracellular vesicles (EVs), as a major 

mechanism of action for MSC therapies that do not require differentiation [3-5]. 

EVs are nanosized, lipid membrane-bound vesicles that play a crucial role in 

intercellular communication through the transfer of bioactive cargo (i.e. RNA, 

DNA, proteins, lipids) [6]. EVs are also able to circumvent tumorigenicity 

concerns, possess low immunogenicity profiles, and can cross biological barriers, 

making them effective therapeutic candidates for a wide range of applications [7-

9]. MSC EVs especially hold promise for a multitude of diseases and indications 

such as cardiovascular disease, wound healing, and neurodegenerative disorders 

due to their ability to confer pro-angiogenic, immunomodulatory, anti-fibrotic, and 

anti-apoptotic effects in recipient cells [10-12]. With the added benefit of 

minimizing safety concerns associated with cells while granting similar, if not more 

potent therapeutic effects, MSC EVs have the potential to transform the field of 

stem cell-based regenerative medicine. 

 



 

 

4 
 

Despite the promise of MSC EVs and their widespread applications as described 

pre-clinically and in clinical trials, successful clinical development is hindered by 

critical challenges including low therapeutic potency, donor cell limitations, and 

lack of scalable manufacturing approaches that can overcome low productivity [13, 

14]. The inherent heterogeneity and complexity of EVs as a biological entity 

already make standardization difficult. Donor-derived MSCs (e.g. bone marrow, 

adipose tissue, etc.) as a source for EVs introduce additional challenges including 

early senescence and inter- and intra-donor variability [15-17]. Moreover, low 

innate potency of naïve MSC EVs necessitates higher and/or more frequent doses, 

increasing requirements for both the batch yield and treatment time which 

subsequently raises costs. Genetic engineering and exogenous loading of bioactive 

cargo to improve potency, while effective, only further increases the complexity 

and regulatory burden associated with the therapeutic. Finally, much of the basic 

and pre-clinical research reported on therapeutic EVs is done within static flask 

culture in small batches [18]. This approach severely limits scalability, as it is 

virtually impossible to achieve clinically relevant EV yields with flask culture and 

necessitates additional research and development of an effective manufacturing 

platform that is compatible with the previously engineered EV of interest. Thus, in 

the interest of improving translational potential, we addressed all three of these 

limitations- low potency, donor cell variability, lack of scalable manufacturing 

processes- in this dissertation. 

 



 

 

5 
 

As an alternative MSC source to primary donor-derived cells, we used MSCs 

differentiated from induced pluripotent stem cells (iPSCs) (iMSCs), as iPSCs are 

self-renewing and can be expanded indefinitely without senescing. Hence, iMSCs 

offer the potential to eliminate donor variability, and the following work establishes 

their compatibility with the described approaches and supports their use as a potent 

pro-angiogenic MSC EV source. 

 

In Chapter 2, we provide an overview of EV biology, the potential of MSC EVs 

as a therapeutic for wound healing, the state of MSC EVs within clinical trials, and 

limitations to their clinical development. We also discuss iMSCs as an alternative 

to donor MSCs, the mechanical cues of substrate stiffness and cell confinement on 

cell behavior, as well as bioreactor-based production as strategies to address these 

limitations. In Chapter 3, we demonstrate the mechanosensitivity of iMSCs to 

substrate stiffness, determining that seeding cells on substrates softer than 0.054 

MPa produce significantly more EVs with augmented angiogenic bioactivity. We 

also report a high-level analysis of the RNA profiles in both the EVs and recipient 

cells. Lastly, in Chapter 4 we show that we can achieve an increase in angiogenic 

bioactivity of iMSC EVs that corresponds with high levels of cell confinement (5 

µm) using polydimethylsiloxane (PDMS) micropillar arrays. We also describe the 

development and optimization of a 3D-printed perfusion bioreactor system that 

incorporates cell confinement, allowing for significantly upregulated production of 

EVs with enhanced therapeutic potency. Cargo differences between flask-

generated and bioreactor-generated EVs are also identified, and we show that these 



 

 

6 
 

EVs also retain their efficacy both in vitro and in vivo in a diabetic mouse wound 

model. Overall, the strategies detailed here provide solutions to some of the most 

critical barriers to clinical translation that could be implemented to bolster the 

eventuality of iMSC EVs in the clinic. 

 



 

 

7 
 

Chapter 2: Therapeutic Potential of Extracellular Vesicles and 

Progress Towards Clinical Translation 

 

2.1 Extracellular Vesicle Biology 

2.1.1 EV Origin and Biogenesis 

EVs are a heterogenous group of lipid membrane-bound nanovesicles that 

are released by all cell types. The overarching term describes various 

subpopulations including exosomes and microvesicles that are largely 

distinguished by their mode of biogenesis. Briefly, exosomes are generated 

via the endocytic pathway where intraluminal vesicles are secreted after 

fusion of the multivesicular body with the plasma membrane [19]. 

Endosomal reorganization also occurs here, where the membrane becomes 

highly concentrated with certain membrane-bound proteins (i.e. 

tetraspanins) such as CD9 and CD63 [19]. Microvesicles are formed by the 

direct outward budding of the plasma membrane upon redistribution of 

phospholipids, cytoskeleton components, and tetraspanins [19, 20]. 

However, the overlap in size and composition makes it difficult to separate 

populations effectively. Thus, most reported studies, including ours, refer 

to a heterogeneous population of both exosomes and microvesicles. 

 



 

 

8 
 

2.1.2 EV Composition 

The cargo encapsulated within EVs consist of nucleic acids, proteins, and 

lipids that reflect that of their producer cells. As such, EVs are mediators of 

intercellular communication by transferring cargo to elicit a response in 

recipient cells. From a therapeutic standpoint, proteins and nucleic acids are 

the key effector molecules. A wide range of proteins exist within EVs that 

play roles in cell adhesion (e.g. integrins, intracellular adhesion molecules), 

intracellular trafficking (e.g. RAB), and EV biogenesis (e.g. tetraspanins, 

ALIX, TSG101) [6, 21, 22]. Tetraspanins are a family of transmembrane 

proteins, and certain ones (i.e. CD9, CD63) are highly enriched in EVs [22]. 

These membrane proteins can also be engineered to preferentially traffic 

exogeneous therapeutic cargo overexpressed within the cells into EVs [21]. 

Other proteins within EVs are more difficult to define given their 

abundance, the EVs’ origin, and the physiological context of the producer 

cells [23].  EV origin especially provides a source of challenge when it 

comes to defining EVs’ proteomics and RNA profiles, as there can be 

inherent batch-to-batch variation, particularly with donor cells. For 

example, two separate studies performing a proteomics analysis on human 

bone marrow-derived MSC (BMMSC) EVs identified 730 and 797 proteins, 

respectively, with only a 60% overlap between the two [24, 25]. In addition, 

universal strategies for loading exogenous therapeutic proteins, especially 

cytosolic proteins, into EVs are lacking due to large variations in size and 

structure [26]. While it is evident that proteins are a bioactive component of 



 

 

9 
 

EVs that are essential to their biogenesis and functionality, further research 

will help to better understand EV biogenesis and mechanisms of action. 

 

Aside from proteins, the other major bioactive constituent of EVs is RNA. 

microRNAs (miRNAs) in particular are the most widely studied type of EV-

associated RNA. miRNAs are non-coding RNAs that are 18-22 nucleotides 

long that have multiple gene targets, thus offering the potential to 

manipulate a range of signaling pathways and subsequent therapeutic 

targets and effects [27-29]. Many reports investigating the therapeutic 

properties of MSC EVs, for instance, identify miRNAs as the functional 

mechanisms behind such bioactivity [30-34]. Like proteins, miRNA 

expression within EVs can also change depending on the producer cell type, 

microenvironment, and physiological context [27]. For this reason, despite 

recognizing that miRNAs are responsible for many of the therapeutic and 

disease-related effects of EVs, characterizing a complete miRNA profile of 

any one type of EV is difficult [27, 35]. Recent literature has also begun to 

realize the therapeutic roles of EV-associated long non-coding RNAs 

(lncRNAs). lncRNAs are greater than 200 nucleotides in length, many with 

active roles in regulating gene expression via chromatin remodeling, 

transcription of both near and distant genes, RNA splicing, post-

transcriptional modifications, and more [36]. Given the diversity in their 

mechanisms of action when it comes to gene regulation, lncRNAs can affect 

numerous aspects of physiology and pathophysiology, making them 



 

 

10 
 

promising therapeutic targets [36]. Moreover, considerable research has 

gone into identifying effective ways to load all types of RNAs into EVs for 

enhanced therapeutic efficacy. For example, it is well established that RNAs 

can be overexpressed in cells, which will then package the RNA into EVs 

during biogenesis and secretion [37]. As a mode of active loading, different 

motif sequences, when tagged on the end of a target RNA, have been 

identified to cause preferential enrichment in EVs [37].  Ultimately, 

continuing to understand EV-associated RNA content and function will 

only enhance the potential of therapeutic EVs. 

 

2.2 Extracellular Vesicle Isolation and Characterization 

There are numerous EV isolation strategies employed within the field. Perhaps the 

most reported method for EV isolation is differential ultracentrifugation (UC) due 

to its simplicity and affordability [38]. Differential UC relies on the separation of 

EVs based on size and consists of multiple centrifugation spins at increasing speeds 

and times, often up to 100,000 xg [19, 38]. Similarly, density gradient 

centrifugation is another method which utilizes a gradient made of sucrose or 

iodixanol in combination with high centrifugal force. With this method, EVs are 

isolated based on their specific density (1.13 – 1.19 g/mL) [38]. While relatively 

simple, UC alone is time-consuming and not easily scalable for large volumes, and 

generates issues with sample purity due to the co-isolation of protein aggregates 

that are of similar size and/or density to EVs [19, 38, 39].  

 



 

 

11 
 

Another EV isolation technique reliant on size is size exclusion chromatography 

(SEC). SEC utilizes a porous stationary phase, which allows for the elution of larger 

particles and aggregates first, followed by smaller vesicles, and finally smaller 

protein contaminants, all in separate fractions [40]. Given the highly specific 

separation that SEC can achieve, especially in the separation of EVs and protein 

contaminants, the purity of the final sample is high compared to UC methods [40]. 

Unfortunately, EV yield is sacrificed, and SEC is most suitable with smaller starting 

volumes, thus requiring an additional preparatory concentration step via UC or 

tangential flow filtration (TFF). Taken together, despite SEC being compatible with 

current good manufacturing practices (cGMP), the potential of SEC as a stand-

alone, large-scale EV isolation method is limited [40, 41]. 

 

TFF is another method that is efficient, scalable, and robust, with the ability to 

isolate EVs from large volumes with relatively high purity and yield [42]. TFF 

utilizes fluid flow parallel to a permeable membrane filter with specific molecular 

weight cutoffs, which allows for separation based on size without filter clogging 

experienced with conventional ultrafiltration [42]. Furthermore, TFF is well-suited 

for large volumes and produces significantly higher EV yields than 

ultracentrifugation or SEC alone, although it does require the conditioned media to 

be depleted of cell debris and large particle contaminants first [42, 43]. TFF is also 

cGMP-compatible, contributing to its potential for larger-scale biomanufacturing 

processes [44]. Although the purity of EVs isolated by TFF is decent, especially 



 

 

12 
 

compared to methods like UC, combining multiple methods for EV isolation and 

concentration can be used to further improve purity. 

 

An alternative method to size-based isolation is to use polymer precipitation with 

polymers such as polyethylene glycol (PEG) to aggregate and co-isolate EVs [38]. 

This isolation strategy is relatively inexpensive and simple, but there is a high 

probability of co-precipitating protein aggregates and other contaminants. 

Moreover, the retention of PEG necessitates additional purification steps [38].  

 

Since there are so many ways to isolate EVs, standardized characterization using 

multiple techniques is important. According to the Minimal Information for Studies 

of EVs (MISEV) guidelines, total EV yield should be reported using technologies 

such as nanoparticle tracking analysis (NTA) which utilizes Brownian motion to 

measure particle size distribution and concentration [45]. Total protein 

concentration as measured by BCA assays is another recommended quantification 

[45]. However, BCA results can often be an overestimation due to protein 

contaminants that are unable to be isolated from the EV sample [45]. Additionally, 

the level of overestimation is partially dependent on isolation method, so total 

protein content measurements should be used in addition to particle concentration, 

and not as a standalone measure. Single vesicles can be detected by methods such 

as transmission electron microscopy to confirm morphology and size. 

 



 

 

13 
 

Once EV yield is determined, it is necessary to confirm their identity via detection 

of certain membrane proteins that have been established as EV markers, as well as 

the lack of cellular markers via western blotting [45]. The major categories for EV 

markers as suggested by the MISEV guidelines are membrane-bound proteins CD63 

or CD81 and cytosolic proteins secreted within EVs such as ALIX and TSG101 

[45]. Calnexin is used as an EV-negative marker, as it is an endoplasmic reticulum 

protein so it should not be present within EVs [45].  

 

2.3 MSC Extracellular Vesicles as Therapeutics 

MSCs have long been the focus of regenerative medicine and stem cell therapies 

due to their ability to differentiate into multiple cell lineages and confer therapeutic 

effects such as immunomodulation and vascularization [46]. However, issues with 

senescence, tumorigenicity, and potential immunogenicity, especially after 

repeated dosing, persist [2]. Studies have also reported low cell viability and 

engraftment post-injection despite positive therapeutic effects, attributing the 

results to paracrine factors (e.g. cytokines, growth factors, EVs) [47-49]. Given the 

bioactive cargo that is present in EVs and their roles as mediators of cell-to-cell 

communication throughout the body, their potential as cell-free therapeutics is 

well-supported. While MSC EVs are being studied for a multitude of applications 

including graft vs. host disease, osteoarthritis, acute kidney injury, respiratory 

disease, and cardiac disease, the work described here focuses on their capacity to 

remedy chronic wounds [50-54]. 

 



 

 

14 
 

2.3.1 Impaired Wound Healing: Diabetic Wounds 

Healthy wound healing consists of 4 phases: hemostasis, inflammation, 

proliferation, and remodeling. Briefly, hemostasis occurs within minutes to 

hours after injury to stop the bleeding via platelet aggregation and formation 

of a blood clot [55]. Next, the inflammatory phase, which includes the 

infiltration of neutrophils, monocytes, and endothelial cells to clear the 

wound of pathogens and initiate the beginning phases of tissue repair, starts 

immediately after injury and reaches a peak around days 4-6 [55]. The 

proliferative phase follows and last for 2-3 weeks [56]. Here, 

vascularization and new tissue formation occurs via angiogenesis, 

granulation tissue formation, and epithelialization [56]. By the end of this 

phase, the wound should be closed, and visible formation of healthy new 

tissue is expected. Finally, matrix remodeling occurs over the course of 

months to years when collagen degrades, the wound contracts, and scar 

tissue forms [56]. 

 

In the case of chronic wounds, aspects of this process are impaired. Among 

the most common types of chronic wounds are diabetic foot ulcers (DFUs), 

which affect an estimated 40-60 million people globally, are associated with 

a 5-year mortality rate of 30% without, or 70% with amputation, and are the 

cause of 80% of all lower extremity amputations [57, 58]. DFUs can be 

separated into different categories based on underlying pathology, but 

patients suffer from either peripheral neuropathy, peripheral artery disease, 



 

 

15 
 

or both, which leads to decreased circulation and therefore a dysfunctional 

inflammatory and proliferative phase of healing [57, 59]. Specifically, a 

prolonged pro-inflammatory macrophage phenotype and impaired anti-

inflammatory M2 polarization leads to excessive inflammatory cytokines, 

decreased growth factors, and extracellular matrix (ECM) disruption which 

significantly delays tissue repair [60]. Additionally, hyperglycemia in 

patients with DFUs negatively affects endothelial cell, keratinocyte, and 

fibroblast function by downregulating growth factors that are critical for 

angiogenesis, thus compromising the proliferative and vascularization 

phase of healing [60]. Hence, an ideal therapeutic for chronic wounds would 

address both the inflammatory and proliferative phase, which MSC EVs 

have the potential to do. 

 

2.3.1.1 Inflammation 

While inflammation is a necessary process in the body’s defense 

against pathogens and injury, prolonged and chronic inflammation 

can be detrimental, and even fatal [61]. Thus, reducing inflammation 

in chronic wounds is critical for a complete and timely healing 

process. A number of cell types are involved in the inflammatory 

response, but macrophages in particular are essential for inducing, 

maintaining, and suppressing inflammation due to their ability to 

polarize from a pro-inflammatory (M1) state to an anti-

inflammatory (M2) state [62, 63]. Upon injury, monocyte 



 

 

16 
 

recruitment and differentiation into M1 macrophages occurs in 

response to endothelial cell activation and secreted inflammatory 

chemokines (e.g. CXCL8, CXCL1, CXCL2, etc.) [64, 65]. 

Following the initial inflammatory response, environmental cues 

such as the clearance of apoptotic cells and the presence of cytokines 

and growth factors (e.g. VEGF, IL-4, IL-13, etc.), in addition to the 

homing of endogenous MSCs, initiate the transition to the M2 state 

to allow for the proper progression of healing and tissue repair [66]. 

 

Importantly, MSC EVs can be taken up by macrophages and help 

modulate the immune response by triggering polarization, making 

them effective anti-inflammatory therapeutics [67, 68]. This 

phenomenon has been reported by many groups, where treatment of 

macrophages with MSC EVs induced a shift towards an anti-

inflammatory phenotype evidenced by the presence of M2 markers 

(e.g. CD206, CD51, CD36, Arg1, CD73) [51, 68-71]. Furthermore, 

the therapeutic anti-inflammatory effects of MSC EVs have 

translated in vivo in mouse models for applications such as 

Niemann-Pick disease, rheumatoid arthritis, ischemia-reperfusion 

injury, wound healing, and Alzheimer’s disease [72-76]. Notably, 

their application for chronic wound healing is promising given the 

prolonged inflammatory state that is characteristic of such wounds. 

 



 

 

17 
 

2.3.1.2 Angiogenesis 

The ability to promote angiogenesis, or the formation of new blood 

vessels, is critical for a wound healing therapy, as the vasculature is 

responsible for delivering oxygen and nutrients necessary for 

regenerating healthy cells and tissue [77]. While angiogenesis is a 

complex, dynamic process involving multiple cell types and 

environmental cues, it is heavily driven by growth factors such as 

VEGF and FGF [78]. Subsequent activation of the PI3K/Akt 

pathway, Src/FAK/p38-MAPK pathway, and others then facilitate 

changes in cell migration, proliferation, and survival to overall 

enhance angiogenesis [79]. While endothelial cells serve as the 

foundation for this process, many cell types work in concert to 

enable effective angiogenesis and tissue repair. 

 

The uptake of MSC EVs by endothelial cells and others associated 

with wound healing (e.g. keratinocytes, fibroblasts, etc.) has been 

thoroughly documented [80-84]. Moreover, their capacity to 

enhance angiogenesis and vascularization is well established in pre-

clinical in vitro models; they have been reported to activate 

angiogenesis-related signaling pathways and upregulate growth 

factors, as well as promote the growth, migration, and capillary-like 

tube formation of recipient cells across numerous separate studies 

[16, 30, 81, 82, 85]. These pro-angiogenic tissue repair effects have 



 

 

18 
 

also been reported in vivo with various rodent wound healing 

models, specifically citing an increase in vascular formation and re-

epithelialization along with a decrease in inflammation and 

oxidative stress [86-90]. 

 

2.3.2 EVs in Clinical Trials 

Beyond pre-clinical evaluation, EVs of all kinds are also being assessed in 

clinical trials with exponential growth over the last 10-15 years [91]. 

According to clinicaltrials.gov, there are currently 166 recruiting, active, or 

completed trials for EVs as an interventional treatment for over 70 different 

indications including COVID-19 (NCT05787288), acute respiratory 

distress syndrome (NCT05354141), cancer (NCT05955521), osteoarthritis 

(NCT06937528), stroke (NCT06612710), cardiomyopathy 

(NCT05774509), and chronic wounds (NCT06825884). The majority of 

trials are evaluating MSC EVs given their multifaceted, intrinsic therapeutic 

effects [91]. However, despite the increasing number of promising clinical 

trials and EV-based companies that exist, successful FDA approval and 

commercialization remain unrealized. 

 

2.4 Challenges to the Clinical Translation of Extracellular Vesicles 

There are a number of factors slowing the development and ultimate clinical 

translation of EV therapies. Limitations surrounding isolation methods, 



 

 

19 
 

heterogeneity and quality control, and storage conditions exist without consensus 

regarding the best ways to address them. With many of these issues, there is 

conflicting information regarding results, study timeline, and characterization 

methods. As an example, one study attempting to determine the best storage 

conditions showed that EVs stored at -20ºC and -80ºC for 4-6 weeks maintained 

their inherent bioactivity, as well as the bioactivity of loaded RNA cargo [92]. 

However, another group using a different EV type reported a decrease in bioactivity 

after storage at -20ºC but not -80ºC after 4 weeks [93]. The recent systematic review 

by Ahmadian et al. outlining an array of studies that use a wide range of different 

EVs, isolation methods, and measures of bioactivity only underscores the 

difficulties in comparing across studies and the lack of standardization within the 

field [94]. More extensive research on storage, isolation, and quality control will 

further the development of EV therapies. However, the enclosed work specifically 

addresses the following translational limitations of cell source and donor 

variability, lack of scalable manufacturing platforms, and low innate potency. 

 

2.4.1 Cell Source and Donor Variability 

MSCs and other stem cells face the issue of donor variability, as MSCs are 

often sourced from individual donors and various different sources (e.g. 

bone marrow, adipose tissue, umbilical cord) [95]. Variability in EV 

bioactivity, potency, and cargo pose issues when comparing efficacy 

between different donors, and subsequently when trying to maintain 

consistency between batches [96, 97]. Work from our lab has also 



 

 

20 
 

demonstrated these differences in both angiogenic and anti-inflammatory 

bioactivity between BMMSC EV donors [16]. To reduce this variability, 

one alternative to donor MSCs is immortalized MSCs- MSCs that have been 

genetically modified to alter properties, namely senescence, of the primary 

cells while maintaining the other phenotypic behaviors [98]. While a 

potentially viable option, immortalized MSCs carry concerns of genetic 

drift and selection pressure, along with the introduction and permanent 

integration of oncogenes whose effects and risks still need to be better 

understood [98]. Another limitation of BMMSCs is the early senescence 

and significant loss in bioactivity of their EVs at early passages [15], which 

substantially limits their scalability. A promising alternative is MSCs 

differentiated from induced pluripotent stem cells (iPSCs) (iMSCs), as 

iPSCs are infinitely renewable and maintain their viability at high passages, 

offering the potential to reduce variability [99].  

 

2.4.2 Scalable Manufacturing 

Most EV production at a pre-clinical scale is reliant on 2D static cell culture. 

While more practical at a small-scale, 2D culture poses significant 

challenges when trying to scale up, as flasks and static culture cannot 

reasonably meet the productivity requirements for clinical, or even some 

pre-clinical trials. To this point, Gupta et al. performed a meta-analysis of 

pre-clinical animal studies, where the median EV dose was 2.75 mg/kg, or 

3.37 x 108 EVs/kg body weight, although the range extended upwards of 



 

 

21 
 

~1000 mg/kg and 1 x 1012 EVs/kg [100]. While total EV sample yield is not 

well-reported, static flask culture is well-documented by our group and 

others to produce a relatively low number of EVs per cell or per volume of 

media [16, 101, 102], making it difficult to meet the requirements of large 

study cohorts. 3D cell culture provides a potential solution, as cell culture 

systems such as bioreactors allow for higher cell densities, exposure to fluid 

flow, and, as a result, higher EV yields [103]. One study demonstrated that 

both MSC EV production and bioactivity were improved within a perfusion 

bioreactor compared to 2D flask culture [16]. Similar findings have been 

reported by numerous groups. For example, MSCs cultured in a hollow-

fiber bioreactor showed superior EV production and efficacy in improving 

acute kidney injury compared to 2D culture [104]. The reported increases 

in EV yield with the use of bioreactors is also not specific to MSCs [16, 

105]. Other approaches to produce EVs as at a larger scale include changes 

in culture conditions (e.g. pH, hypoxia, media additives), and physical 

stimulation (e.g. shear stress, matrix stiffness, acoustic treatment, magnetic 

nanoparticles), although the scalability of some of these methods remains 

uncertain [106]. Since there is not currently a standardized method for large-

scale production of EVs, further research and innovation of the best and 

most feasible approaches is critical for the translation of EV therapeutics. 

 



 

 

22 
 

2.4.3 Therapeutic Potency 

Common approaches to improve EV potency mainly include biochemical 

priming (e.g. growth factors, lipopolysaccharide, cytokines, hypoxia), and 

genetic engineering (e.g. endogenous RNA and protein loading) [107]. 

Conveniently, several of these methods correspond with methods 

mentioned previously that also increase EV secretion. Another approach is 

to manipulate the EVs post-isolation by electroporating or sonicating 

exogenous therapeutic cargo into the vesicles [108]. Since RNAs and 

proteins are particularly effective in targeting signaling pathways, 

overexpressing and delivering these target molecules often offers enhanced 

therapeutic efficacy [109-111]. However, many of these approaches also 

require the use of extraneous reagents and genetic engineering, which 

increases regulatory and downstream purification burdens, production time, 

cost, and sacrifices yield. Moreover, EVs incorporating a multitude of 

diverse cargo from the producer cells points towards a likely synergistic 

mechanism of action that may not be able to be targeted most efficiently by 

overexpressing a single molecule alone. 

 

Alternatively, biophysical/mechanical cues, specifically substrate stiffness 

and cell confinement, have shown to be important regulators of bioactivity 

and phenotype in many cell types [112-116]. Thus, exploiting the cell 

response to these environmental cues is another approach that holds promise 

in improving EV bioactivity [112, 113, 115]. 



 

 

23 
 

 

2.5 Potential Solutions to Translational Challenges 

 2.5.1 Induced Pluripotent Stem Cell-Derived MSCs (iMSCs) 

Given the challenges with donor-sourced MSCs (e.g. variability, 

senescence, loss of bioactivity), finding another cell source is essential to 

move MSC EV therapeutics from bench to bedside. iMSCs are a promising 

solution. iPSCs are self-renewing, can be cultured essentially indefinitely 

without facing senescence, and maintain their bioactivity at high passages 

[117, 118]. Thus, MSCs derived from iPSCs offer the potential to eliminate 

donor variability, as virtually infinite batches of EVs can be produced from 

the same original iPSC lot. Studies have also reported that iMSCs have 

reduced tumorigenic potential compared to donor bone marrow-derived 

MSCs [119, 120]. Supporting the use of iMSCs over donor MSCs, Lee et 

al. compared the immunomodulatory capacity of BMMSCs and iMSCs, 

determining that iMSCs exhibited a more potent immunosuppressive effect 

than donor-matched BMMSCs [118]. Another group generated similar 

results when comparing iMSCs to donor-matched umbilical cord-derived 

MSCs, with iMSCs expressing greater levels of anti-inflammatory factors 

(e.g. NOS1, FOXP3, TGFB1), lower levels of pro-inflammatory cytokines 

(e.g. IL6, CXCL8, IL1B) and more effectively inhibiting T cell proliferation 

[121]. T cell activation is a critical aspect of the immune response, and thus 

the ability to inhibit such proliferation and activation speaks to the robust 

immunomodulatory potential of iMSCs [122].  Still, iMSCs remain 



 

 

24 
 

understudied compared to other MSC sources, especially with regards to 

their EVs. Research also suggests that iMSCs should be considered a 

separate entity from donor-derived MSCs with potentially greater 

therapeutic potential and possibly higher mechanosensitivity [99, 118]. As 

such, this work supports the establishment of iMSCs and evaluates their 

compatibility with the described EV production systems. 

 

2.5.2 Mechanical Cues: Substrate Stiffness 

It has been established that the cell microenvironment can influence many 

aspects of cell behavior both physiologically in vivo and to manipulate 

different properties and contents of cells in vitro and ex vivo. As such, the 

impacts of substrate stiffness on gene expression and bioactivity have been 

substantiated in a variety of studies. For example, Zhuang et al. describe a 

study where MSCs on a softer scaffold were able to shift recipient 

macrophage phenotypes to an anti-inflammatory state more so than MSCs 

on the stiffer scaffold [123]. Specifically, MSCs on the soft substrate 

featured upregulated TSG-6 caused by enriched mitogen-activated protein 

kinase (MAPK) and Hippo signaling pathways and mediated the 

downregulation of inflammatory cytokines in recipient macrophages via 

NF-kB signaling inhibition [123]. In another study, MSCs on stiffer 

hydrogels exhibited more spreading and upregulation of smooth muscle 

markers, while MSCs on softer hydrogels showed upregulation of 

endothelial markers, suggesting the effects of matrix stiffness on 



 

 

25 
 

differentiation fate and phenotype [115]. In the interest of using iMSCs over 

BMMSCs, a 2022 study by Gultian et al. compared the sensitivity of the 

two cell types to substrate stiffness, concluding that iMSCs produced a more 

consistent response across donors with regards to cell area, nuclear yes-

associated protein (YAP), and phosphorylated focal adhesion kinase 

(pFAK) length [99]. Indeed, pFAK can activate various 

mechanotransduction signaling pathways that control proliferation, 

inflammation, migration, and angiogenesis [124, 125]. The effects of 

substrate stiffness are also not exclusive to just MSCs and have been studied 

with other cell types. One group investigated how PC12 neural cells 

responded to anti-cancer topoisomerase inhibitor drugs when grown on 

PDMS substrates ranging from 0.1 kPa to 46.7 kPa, which were meant to 

mimic different tissue types [126]. The authors reported that as substrate 

stiffness decreased, the elastic modulus of the cells themselves also 

decreased, and that softer substrates enhanced the neuroprotective effects of 

the drugs via changes in the EGFR/PI3K/AKT pathway [126].  

 

Given the evident sensitivity of cells to matrix stiffness, it makes sense that 

the cells’ secreted factors, namely EVs, might also be affected. One study 

explored the effects of conditioned media, which contains EVs, from 

BMMSCs grown on soft (0.2 kPa) and stiff (100 kPa) polyacrylamide gels 

[127]. The immunomodulatory potential of the conditioned media was 

significantly greater from cells on the soft substrate compared to the stiff or 



 

 

26 
 

tissue culture polystyrene (TCPS) flask, as measured by macrophage 

phagocytic activity [127]. Additionally, angiogenic bioactivity in vitro was 

enhanced in the conditioned media from the soft substrate [127]. Nasser et 

al. also produced similar results, where seeding BMMSCs on a 5kPa 

substrate produced conditioned media with the highest concentration of 

VEGF and overall pro-angiogenic bioactivity [128]. Substrate stiffness may 

also provide a solution to low EV yield. For instance, a study that seeded 

BMMSCs on 3 kPa alginate hydrogels reported 10-fold more EVs secreted 

per cell than when seeded on TCPS flasks [129]. The cells on the 20 kPa 

hydrogel secreted significantly more EVs per cell than those on flasks as 

well, but to a lesser extent [129]. Interestingly, in the context of cancer, the 

results are more variable. For example, EV secretion from primary breast 

cancer tumor tissue (> 10 kPa) was greater than secretion from healthy, 

tumor-adjacent tissue (< 10 kPa) in one study [130]. A separate study 

compared a much greater range of stiffnesses, but reported significantly 

more secreted EVs by MDA-MB-231 breast cancer cells seeded on soft 

(200 kPa) substrates compared to stiff (3.6 MPa) substrates [131]. Yet 

another group documented in a 2025 pre-print that EV secretion by oral 

squamous cell carcinoma cell line spheroids was increased when seeded on 

stiff (0.9 kPa) compared to soft (0.3 kPa) hydrogels [132]. The diversity in 

study design, results, and stiffness ranges, especially regarding tumor EV 

release only highlights the importance of confirming the effects of matrix 

stiffness with each cell type and application. 



 

 

27 
 

 

2.5.3 Mechanical Cues: Cell Confinement 

Cell confinement is another relevant biophysical cue which occurs in vivo 

with numerous cell types such as cancer cells as they enter and exit the 

bloodstream, and MSCs as they cross endothelial barriers while homing to 

sites of injury [133]. Thus, changes to morphology and bioactivity are not 

entirely surprising as the confinement occurs as cells migrate to perform a 

specific function. In 2017, a correlation between an elongated MSC 

morphology and greater immunosuppressive capacity after IFN-𝛾 

stimulation was reported [134], which suggests the potential of forcing cell 

elongation to manipulate cell behavior. Thus, while cell confinement is not 

congruent with elongation, the mode of confinement that we are interested 

in here is that which causes morphological elongation, where the cell takes 

on a spindle-like shape. Accordingly, one study used an aligned nanofiber 

scaffold that resulted in MSCs, whose source was not identified, organizing 

themselves in one direction and taking on an elongated morphology 

compared to the unaligned controls [135]. The aligned MSCs exhibited a 

greater migratory capacity, as well as an improved ability to promote 

neuronal regeneration and motor function in a rat spinal cord injury model 

[135]. A study by Rao et al. demonstrated universally increased spreading 

and migration of multiple cell types including MSCs, HeLa cells, C2C12 

cells, MDCK cells, and NIH 3T3 cells with micropillar-induced mechanical 

confinement [136]. Notably, they suggested that the confinement-triggered 



 

 

28 
 

increase in migration was dependent on nuclear deformation-induced YAP 

nuclear translocation and independent of actomyosin contractility, which 

has historically been a commonly proposed mechanism [136]. 

 

Regarding EVs specifically, less work has been done to understand the 

effects of cell confinement on EV production and bioactivity, particularly 

with MSCs. One group subjected MDA-MB-231 cells to confinement via 

microchannels, which achieved a more elongated cell morphology as well 

as a significant increase in the number of EVs per cell as the degree of 

confinement was increased [137]. In a different study examining the 

conditioned media, confinement and elongation of BMMSCs achieved by 

encapsulation within a hydrogel resulted in significant upregulation of 

secreted pro-angiogenic growth factors such as VEGF, placental growth 

factor, and transforming growth factor beta 1 (TGF𝛽1) [138]. Endothelial 

cells treated with the conditioned media also organized into more capillary-

like networks than those treated with conditioned media from less confined 

and elongated MSCs [138]. However, since conditioned media consists of 

more than just EVs, further investigation into the effects of confinement on 

MSC EVs specifically, and the scalability of this mechanical cue is 

warranted. 

 



 

 

29 
 

2.5.4 Bioreactor-Based Dynamic Culture 

Bioreactors have long been used in large-scale manufacturing processes for 

cell and cell product-based therapies (e.g. CAR-T cells, antibodies) due to 

their more automated nature and ability to support higher cell densities and 

increase cell productivity [139]. Consequently, dynamic culture and the 

accompanying mechanical stimuli from fluid flow within bioreactors (e.g. 

perfusion, hollow fiber, vertical-wheel, stirred-tank, etc.) is a promising 

scalable strategy to address limited EV yield. Our group has demonstrated 

the efficacy of this approach previously using two different 3D-printed 

perfusion bioreactor designs. In the first study, endothelial cell EV 

production was  enhanced 100-fold with perfusion bioreactor culture 

compared to flask and static scaffold culture, with non-statistically 

significant improvements in vascularization bioactivity [140]. In the second 

study using BMMSCs and a different bioreactor scaffold design, an 83-fold 

increase in EV production was achieved with increased in vivo 

vascularization bioactivity as observed with an increase in CD31+ staining 

of healed diabetic mouse wound tissue [16]. Others have also justified the 

use of bioreactors to augment EV production. Cao et al. cultured umbilical 

cord-derived MSCs in a commercially-available hollow fiber bioreactor to 

produce 20 times the amount of EVs generated with 2D flask culture [104]. 

These bioreactor-generated EVs also exhibited an enhanced ability to 

reduce inflammation and exert renoprotective effects in a cisplatin-induced 

acute kidney injury mouse model [104]. Another hollow fiber bioreactor 



 

 

30 
 

setup was able to produce ~4-fold more immortalized MSC EVs when 

compared to flask culture [141]. With the changes in functional activity that 

have been observed with bioreactor culture, it is important to also note the 

often reported changes in EV cargo, particularly with respect to RNA and 

protein content [16, 140, 142].  Hence, establishing an EV therapeutic using 

a rationally designed, scalable bioreactor system at the early stages of 

development would greatly improve the scale-up and translational potential. 

 

2.6 Conclusion 

MSC EVs hold significant potential to advance the field of regenerative medicine 

and regenerative cell-based therapies, offering the ability to circumvent cell-

specific risks that have contributed to their slow clinical development. Reflecting 

this potential, there has been rapid growth in both clinical trials and industry 

investment, with applications spanning cardiovascular disease, cancer, wound 

healing, and respiratory diseases. Despite this momentum, no FDA approved EV 

products exist to date, as regulatory challenges remain and key limitations- 

including low therapeutic potency, donor variability, and lack of scalable 

manufacturing platforms- continue to hamper their progress. Emerging strategies 

such as iPSC-derived MSCs, mechanical stimulation, and bioreactor-based 

production offer scalable solutions to these translational bottlenecks. However, a 

frequent lack of consideration for scalability during early EV development has been 

detrimental to their success. For example, because EVs are inherently 

heterogenous, the addition of exogenous cargo and reagents associated with genetic 



 

 

31 
 

engineering strategies to improve potency only amplify downstream processing and 

regulatory burden. Moreover, EVs produced from genetically engineered cells in 

small-scale flask culture often fail to retain the same level of therapeutic effects 

when transitioned to bioreactor systems. Therefore, prioritizing the development of 

potent MSC EVs utilizing scalable manufacturing technologies (i.e. mechanical 

cues, bioreactors, iMSCs) early in the development process is critical to realizing 

their clinical potential.  

 



 

 

32 
 

Chapter 3: Mesenchymal Stem Cell Extracellular Vesicle 

Vascularization Bioactivity and Production Yield are 

Responsive to Cell Culture Substrate Stiffness 

 

*This chapter is published in the journal Bioengineering & Translatoinal 

Medicine: 

Powsner EH, Kronstadt SM, Nikolov K, Aranda A, Jay SM. Mesenchymal stem 

cell extracellular vesicle vascularization bioactivity and production yield are 

responsive to cell culture substrate stiffness. Bioengineering & Translational 

Medicine. 2025;10(3):e10743. 

 

3.1 Introduction 

Mesenchymal stem cell-derived extracellular vesicles (MSC EVs) have gained 

significant interest within the field of regenerative medicine based on their pro-

angiogenic, anti-inflammatory, and anti-apoptotic properties. They also hold 

promise for therapeutic applications such as chronic wounds, cardiovascular 

diseases, and inflammatory diseases [143]. However, despite their widespread 

potential and growing traction pre-clinically and in clinical trials, successful FDA 

approval and ultimate clinical translation have not yet been achieved [144]. This 

can be attributed to numerous challenges that MSC EVs and EVs in general face 

that include, but are not limited to, low potency and donor variability [14, 145]. 



 

 

33 
 

 

Most common approaches to overcoming low potency issues include exogenous 

and endogenous loading of therapeutic cargo [107]. Numerous groups have 

demonstrated that overexpressing therapeutic microRNAs (miRNAs) via 

endogenous loading (e.g genetic engineering) or exogenous loading post-EV 

isolation (e.g. sonication, electroporation, etc.) is able to improve their efficacy. For 

example, by overexpressing miR-181a, a miRNA implicated in immune regulation, 

Wei et al. showed that the MSC EVs were able to exert a significantly stronger 

effect in reducing inflammation and improving cardiac function in mice after 

ischemia-reperfusion injury compared to native MSC EVs [74]. Another common 

approach is to precondition the cells with cytokines, growth factors, and/or hypoxia 

to promote certain therapeutic effects [30, 146]. While these methods are effective, 

they require expensive extraneous reagents that, at best, necessitate more intensive, 

costly downstream purification, and, at worst, may be left behind at levels high 

enough to affect EV bioactivity and experimental outcomes, posing additional 

challenges to clinical translation [147, 148]. Alternatively, researchers have shown 

that mechanical stimuli such as substrate stiffness, stretch and compression, and 

fluid shear stress are capable of regulating cellular activity and determining cellular 

fate, and work has begun to investigate the downstream effects of these phenomena 

in secreted EVs [126, 149-151]. Our group has reported on this previously, where 

MSCs subjected to flow-derived shear stress in a perfusion bioreactor produced 

EVs that exerted a significantly more therapeutic angiogenic effect compared to 

both PBS and flask-generated EVs in a mouse wound healing model [16].  



 

 

34 
 

 

Here, using fabricated polydimethylsiloxane (PDMS) devices, we show the 

vascularization bioactivity of EVs from human bone marrow-derived MSCs 

(BMMSCs) is affected by producer cell substrate stiffness, where seeding cells on 

softer substrates promotes the secretion of EVs with greater potency. We also show 

that the softer substrates improve BMMSC EV secretion and overall yield 

compared to conventional tissue culture flasks. Towards the goal of improving the 

scalability of MSC EV therapeutics, we further demonstrate that this 

mechanoresponsiveness is essentially conserved for EVs obtained from MSCs 

derived from induced pluripotent stem cells (iMSCs), which offer an infinite source 

for consistent MSCs without donor variability [152]. Overall, this work suggests 

that adjusting matrix stiffness is a simple, scalable, and cost-effective way to 

significantly improve MSC EV potency for tissue repair applications. These results 

suggest a solution to some of the imminent bottlenecks (i.e. potency, donor 

variability) to clinical translation of MSC EV therapeutics, and the simplicity of 

this approach makes it easy to test and adapt for other mechanoresponsive cell/EV 

types.  

 

3.2 Methods 

3.2.1 Cell Culture  

Human bone marrow-derived mesenchymal stem cells (BMMSCs) were 

purchased from ATCC (PCS-500-012). Human induced pluripotent stem 

cell-derived MSCs (iMSCs) were also purchased from ATCC (ACS-7010). 



 

 

35 
 

Both BMMSCs and iMSCs were cultured in Dulbecco’s Modified Eagle’s 

Medium (DMEM) (Corning; 10-013-CV) supplemented with 10% fetal 

bovine serum (FBS) (Cytiva; SH30910.03), 1% penicillin-streptomycin 

(P/S) (VWR; 45000-652), and 1% MEM non-essential amino acids 

(ThermoFisher Scientific; 11140050) in T-175 tissue culture flasks. All 

BMMSCs were seeded at passage 4 in experiments. 

 

Human umbilical vein endothelial cells (HUVECs) pooled from multiple 

donors were purchased from PromoCell (C-12203) and cultured in 

endothelial growth medium (PromoCell; C-22121) with 1% penicillin-

streptomycin in T-75 tissue culture flasks coated with 0.1% gelatin at 37ºC 

for 1 hour prior to seeding. During the experiments, HUVECs were 

maintained in endothelial basal medium (PromoCell; C-22221) with 0.1% 

FBS and 1% P/S. HUVECs were used at passage 3 or 4 in all experiments. 

 

RAW 264.7 macrophages were purchased from ATCC (TIB71) and 

cultured in DMEM (Corning; 10-013-CV) supplemented with 5% FBS 

(Cytiva; SH30910.03) and 1% penicillin-streptomycin (VWR; 45000-652) 

in T-175 tissue culture flasks. 

 

3.2.2 Stiffness Device Fabrication, Preparation, and Seeding 

Sylgard 184 polydimethylsiloxane (PDMS) (Krayden; DC2065622) was 

made by thoroughly mixing the silicone elastomer base and crosslinker 



 

 

36 
 

reagent, both included within the kit, at different weight/weight ratios as 

specified (5:1, 10:1, 20:1, 33:1, 50:1). PDMS was poured into each 

compartment of 100mm x 15mm compartmentalized petri dishes (Celltreat; 

229684), placed in a desiccator for 20-30 minutes or until all air bubbles 

had been released, and cured for 1 hour at 80ºC. For all devices, the 

thickness of the PDMS layer was kept relatively constant by pouring the 

same amount of PDMS (2 g) was poured into each compartment. 

 

For the mixing experiments, Sylgard 184 was prepared using a 10:1 base to 

curing agent ratio. Sylgard 527 PDMS (Krayden; DC1696742) was made 

by thoroughly mixing Part A and Part B at a 1:1 weight/weight ratio. The 

prepared Sylgard 184 and Sylgard 527 solutions were mixed at the specified 

weight/weight ratios (1:0, 10:1, 5:1, 1:1, 1:5, 1:10, 0:1), and the final 

mixture was poured into each compartment of a 100mm x 15mm 

compartmentalized petri dishes (Celltreat; 229684). Dishes were placed in 

a desiccator for 20-30 minutes or until all air bubbles had been released and 

were cured for 20 hours at 65ºC. Once again, the thickness of the PDMS 

layer was kept constant. 

 

Stiffness devices were cleaned by rinsing with 100% ethanol, deionized 

water, and 100% ethanol again, and drying with filtered air. Devices were 

then plasma treated for 3 minutes, and UV sterilized for 10 minutes. This 

was immediately followed by coating each PDMS surface with a 20 µg/mL 



 

 

37 
 

type I rat tail collagen (Sigma-Aldrich; C3867-1VL) in 1X PBS solution 

and incubated at 37ºC for at least 1 hour, a surface coating procedure to 

make the innately hydrophobic PDMS surface hydrophilic, that has been 

well characterized prior [153, 154]. T-175 flasks were also coated with the 

collagen solution. 

 

Upon seeding, collagen was removed, and the surface was washed twice 

with 1X PBS. Either BMMSCs or iMSCs were seeded at a concentration of 

5000 cells/mL in 2 mL of media per compartment and left to adhere 

overnight at 37ºC before replacing with 4 mL of EV-depleted media per 

compartment for conditioned media collection and subsequent EV isolation. 

250,000 cells were also seeded in a collagen-coated T-175 flask as a control 

group. 

 

3.2.3 EV Isolation 

Conditioned media was collected from the devices and flask and replaced 

with fresh EV-depleted media every day for 3 days and subjected to three 

differential centrifugation spins at 1000 x g for 10 minutes, 2000 x g for 20 

minutes, and 10,000 x g for 30 minutes, taking the supernatant from each 

step to be used in the next. The BMMSC EVs were isolated via 

ultracentrifugation, where the supernatant from the final spin was subjected 

to centrifugation at 118,000 x g for 2 hours. The supernatant was then 

poured off and the pelleted EVs were resuspended in 1X PBS before 



 

 

38 
 

transferring to a Nanosep 300 kDa MWCO spin column (VWR; 29300-636) 

and centrifuging at 8000 x g until PBS was removed. The EVs were washed 

twice in the same spin column and EVs on the filter were resuspended in 

1X PBS and sterile filtered using a 0.2 µm syringe filter. 

 

All iMSC EVs were isolated via tangential flow filtration (TFF), where the 

supernatant from the 30-minute centrifugation spin was passed through a 

0.2 µm filter before isolating the EVs via TFF (Repligen; KrosFlo KR2i 

TFF system). Using a protocol adapted from Heinemann et al., the samples 

were concentrated down to 10-15 mL using a 100 kDa MWCO MidiKros 

mPES membrane (Repligen; D04-E100-05-N) with 6 diafiltration steps and 

a transmembrane pressure of 5 psi [155]. Samples were further concentrated 

using a 100 kDa centrifugal concentrator (Corning; 431486) and 

resuspended in 1X PBS. 

 

3.2.4 EV Characterization 

EV concentration and size was quantified using nanoparticle tracking 

analysis (NTA) using a NanoSight LM10 (Malvern Panalytical Limited). 

Each sample was diluted to achieve 20-100 particles per frame for accurate 

measurement. Three 30 second videos were acquired, and camera levels and 

detection threshold were maintained between EV samples. Total protein 

concentration was measured using a bicinchoninic acid (BCA) assay per the 

manufacturer’s protocol (G-Biosciences; 785-571). 



 

 

39 
 

 

Transmission electron microscopy (TEM) images of the EVs were obtained 

using a negative stain. EV samples were fixed using 4% electron 

microscopy-grade paraformaldehyde (PFA) for 30 minutes at room 

temperature. A carbon film grid (Electron Microscopy Sciences; CF-200-

Cu-25) was placed on a droplet of the EV/PFA solution and then washed by 

placing on a droplet of 1X PBS followed by 1% glutaraldehyde in PBS for 

5 minutes. The grid was then washed using a droplet of MilliQ water and 

then placed on a droplet of uranyl acetate replacement stain for 10 minutes 

(Electron Microscopy Sciences; 22405). Grids were allowed to completely 

dry before imaging using a JEM 2100 LaB6 TEM (JEOL USA 

Incorporated). 

 

EV presence was confirmed using western blotting. Equal amounts of 

protein or cell lysate (as specified within figures) were loaded into gels and 

transferred to a nitrocellulose membrane. Analysis for ALIX (Abcam; 

ab186429), CD63 (ThermoFisher Scientific; 25682-1-AP), TSG101 

(Abcam; ab125011), calnexin (Cell Signaling Technology; 2679), and 

GAPDH (Cell Signaling Technology; 2118). All primary antibodies were 

incubated with the membrane overnight at 4ºC and diluted at a 1:1000 

dilution except for GAPDH, which was diluted at a 1:2000 dilution. The 

next day, goat anti-rabbit IRDye 800CW (LI-COR; 925-32210) was 



 

 

40 
 

incubated with the membrane at a 1:10,000 dilution for 1 hour before 

imaging on an Odyssey CLx imager (LI-COR). 

 

3.2.5 Cell Viability Assay 

The Cell Counting Kit 8 (WST/-8/CCK8) from Abcam was used to measure 

cell viability (Abcam; ab228554). The specified PDMS variations were cast 

into 96-well plates and incubated with collagen solution for at least 1 hour 

at 37ºC. Wells were washed twice with 1X PBS and seeded at a density of 

3000 cells/well. A standard curve was also made by seeding cells in wells 

with no PDMS. The following day, the WST-8 solution was prepared in the 

dark by diluting it 1:10 in EV-depleted media. Media was aspirated out of 

the wells and replaced with 115 µL/well of the WST-8 dilution. The plate 

was incubated for 4 hours at 37ºC. 100 µL from each well was then 

transferred into a new 96-well plate and absorbance was measured via plate 

reader. Fresh EV-depleted media was added back to the cells, and the plate 

was put back into the incubator. This was repeated for the following 3-4 

days. 

 

3.2.6 Quantifying Cell Growth and Proliferation on PDMS 

Specified PDMS mixtures were cast into 24-well plates (3 wells/day for 4 

days) and incubated with collagen solution for at least 1 hour at 37ºC. Wells 

were washed twice with 1X PBS and seeded with 40000 cells/well. The 

following day, media was replaced with EV-depleted media. The next day, 



 

 

41 
 

media was aspirated, cells were washed once with 1X PBS, and cells were 

trypsinized. Cells were then spun down at 270xg, resuspended, and counted 

using a hemocytometer. This was repeated for the following 3 days. 

 

3.2.7 Tube Formation Assay 

To measure in vitro angiogenesis, 48-well plates were coated with 60µl of 

growth factor reduced Matrigel (Corning; 356230) and incubated at 37ºC 

for 30 minutes. P4 HUVECs were then seeded at 35,000 cells/well with 

either endothelial growth media (PromoCell; C-22121) with 1% penicillin-

streptomycin (positive control), endothelial basal media (negative control), 

or endothelial basal media (PromoCell; C-22221) with 0.1% FBS and 1% 

penicillin-streptomycin with 5E9 EVs/mL. 3-6 hours later, tubes were 

imaged using a Nikon Eclipse Ti2 Microscope at 2x magnification and the 

number of fully formed loops were counted. 

 

3.2.8 Gap Closure Assay 

To assess endothelial cell proliferation and migration, 96-well plates were 

coated with 0.1% gelatin and incubated at 37ºC for 1 hour. 15,000 P4 

HUVECs/well were then seeded in endothelial growth media (PromoCell; 

C-22121) with 1% penicillin-streptomycin. Once cells have formed a 

confluent monolayer (~24 hours), a scratch was induced using a p200 

pipette tip before washing with 1X PBS and serum-starving for 1-3 hours 

with endothelial basal media (PromoCell; C-22221) with 0.1% FBS and 1% 



 

 

42 
 

penicillin-streptomycin. Media was then replaced with either endothelial 

growth media (positive control), endothelial basal media (negative control), 

or 5E9 EVs/mL in endothelial basal media, and imaged. 16-20 hours later, 

the cells were imaged in the same locations, and the percent gap closure was 

quantified using ImageJ.  

 

3.2.9 EV miRNA Isolation and Screening 

EVs from iMSCs grown on Sylgard 527 PDMS devices and collagen-coated 

flasks were isolated as described above. 1E11 EVs were lysed using QIAzol 

Lysis Reagent (Qiagen; 79306), RNA was isolated using RNeasy kits 

(Qiagen; 74104), and cDNA was synthesized using the miScript II RT kit 

(Qiagen; 218161) supplementing the kit-provided reverse transcriptase mix 

with E. coli poly(A) polymerase (New England Biolabs; M0276L) and M-

MuLV reverse transcriptase (New England Biolabs; M0253L). Human 

MSC EV-associated miRNAs were screened using the miProfile™ Human 

MSC exosome miRNA qPCR arrays from GeneCopoeia (GeneCopoeia; 

QM046-B6) per the manufacturer’s protocol. Quantitative polymerase 

chain reaction (qPCR) was performed using a QuantStudio 7 Flex qPCR 

system (ThermoFisher Scientific; 4485701) and data were analyzed using 

the delta-delta Ct method normalized to one of the provided housekeeping 

genes (SNORD47). 

 



 

 

43 
 

3.2.10 EV-Treated HUVEC mRNA Isolation and Screening 

P4 HUVECs were seeded at 60,000 cells/well in a 24-well plate. The 

following day, cells were treated with PBS (control), 5E9 particles/mL of 

EVs from iMSCs grown in collagen-coated flasks, or 5E9 particles/mL of 

EVs from iMSCs grown on Sylgard 527 PDMS devices. 24 hours later, cells 

were washed with 1X PBS and lysed using QIAzol Lysis Reagent (Qiagen; 

79306). RNA was isolated using RNeasy kits (Qiagen; 74104), and cDNA 

was synthesized using the miScript II RT kit (Qiagen; 218161) 

supplementing the kit-provided reverse transcriptase mix with E. coli 

poly(A) polymerase (New England Biolabs; M0276L) and M-MuLV 

reverse transcriptase (New England Biolabs; M0253L). Angiogenesis-

related mRNAs were screened using the GeneQuery Human Angiogenesis 

qPCR Array Kit from ScienCell (ScienCell; GK017), per the 

manufacturer’s protocol. qPCR was performed using a QuantStudio 7 Flex 

qPCR system (ThermoFisher Scientific; 4485701). mRNA array data were 

analyzed using the delta-delta Ct method using the average of the provided 

housekeeping genes provided. Results were validated by performing qPCR 

using PowerTrack SYBR Green Master Mix (ThermoFisher; A46109). 

Primer sequences used for validation are listed in Table S1. Data were 

analyzed using the delta-delta Ct method using GAPDH as a housekeeping 

gene. All qPCR results were normalized to the PBS-treated HUVEC group 

and expressed as either the log2 of the fold change as seen in the heat map, 

or the fold change of mRNA as seen in the validation data. 



 

 

44 
 

 

3.2.11 Anti-Inflammatory Assay 

75,000 RAW264.7 mouse macrophages were seeded into each well of a 48-

well plate in DMEM with 5% FBS and 1% P/S. 24 hours later, cells were 

pre-treated with either no treatment (positive control), 1 µg/mL 

dexamethasone (Sigma-Aldrich; D4902-25MG) (negative control), or 5E9 

EVs/mL. The following day, media was replaced with 10 ng/mL 

lipopolysaccharide (LPS) (Sigma-Aldrich; L4391-1MG) in DMEM with 

5% FBS and 1% P/S and left to incubate at 37ºC for 4 hours. The 

conditioned media was then collected and stored at -80ºC until analysis. To 

determine the anti-inflammatory effects of the EVs, pro-inflammatory 

cytokines in the conditioned media were measured using the DuoSet ELISA 

kits for mouse IL-6 (R&D Systems; DY406) and mouse TNF-alpha (R&D 

Systems; DY410). 

 

3.2.12 Statistical Analysis 

All data are presented as the mean +/- standard deviation (SD). An ordinary 

one-way ANOVA with Sidak’s or Tukey’s multiple comparisons tests was 

used to determine statistical significance (p<0.05) between groups for the 

tube formation assays, gap closure assays, elastic moduli data, and EV 

production data. A two-way ANOVA with Tukey’s multiple comparison 

test was used to analyze cell viability and t-tests were used to analyze the 

qPCR validation data. All statistical analyses were performed with Prism 9 



 

 

45 
 

(GraphPad Software). Statistical significance within figures is noted as ns p 

> 0.05, *p < 0.05, **p < 0.01, ***p<0.0005, ****p < 0.0001. 

 

3.3 Results 

3.3.1 Substrate Fabrication and EV Characterization 

In order to understand how matrix stiffness affects MSC EV production and 

bioactivity, PDMS was used due to its high compatibility and tunable 

mechanical properties to create substrates of varying stiffnesses. A previous 

study demonstrated that altering the ratio of base to crosslinker reagent of 

Sylgard 184 PDMS allows for the creation of substrates with differing 

elastic moduli [156]. Therefore, we initially fabricated a range of substrates 

using varying ratios of 184 PDMS base to crosslinker reagent. Mechanical 

testing using a Q800 Dynamic Mechanical Analyzer confirmed that 

substrates with elastic moduli of 250 ± 0.073 kPa, 1.16 ± 0.117 MPa, 2.43 

± 0.299 MPa, and 3.07 ± 0.054 MPa, were created using w/w ratios of 33:1, 

20:1, 10:1, and 5:1, respectively (Figure 3.1A). A CCK8 assay was then 

used to confirm BMMSC viability over 5 days with 10% DMSO as a 

control, as the cells would not be grown on the substrates longer than 4 days 

(Figure 3.1B). We observed that although the cells on the PDMS substrates 

exhibited slightly lower absorbance values than those on the stiffer tissue 

culture polystyrene (TCPS) flasks, all groups except for the DMSO control 

trended upwards, indicating sustained viability. Additionally, despite 

slightly different starting absorbance values on day 1, there was no 



 

 

46 
 

statistically significant difference between groups. We also confirmed that 

the elastic modulus was not affected by incubation with media (Figure 

3.2A). 

 

BMMSCs were seeded on each of these substrates with EV-depleted media, 

and conditioned media was collected over a period of three days. Control 

EVs were from cells on both normal and collagen-coated TCPS flasks. EVs 

were isolated via ultracentrifugation, and the size distribution and 

concentrations of each sample were determined using nanoparticle tracking 

analysis (NTA). The EVs from each group were all within the expected EV 

size distribution ranges, without any statistically significant differences in 

means or modes between the different groups (Figure 3.1C). EVs and cell 

lysates were analyzed via western blotting to confirm EV identity. Here, all 

EV samples were positive for EV-associated markers ALIX, TSG101, and 

CD63, while they were negative for cellular protein marker Calnexin 

(Figure 3.1D). TEM images confirmed the size and spherical morphology 

of EVs from BMMSCs on both the stiffest (flask + collagen) and softest 

(33:1 184 PDMS) substrates, exhibiting no visual differences between EVs 

produced in either cell culture environment (Figure 3.1E). 

 



 

 

47 
 

 
 

 

 

 

 

 

Figure 3.1: Substrate and BMMSC EV characterization. A) Elastic 
moduli of substrates made with varying ratios of Sylgard 184 base to 
crosslinker reagents. All values expressed as mean ± SD (n=3). B) 
Absorbance values indicating cell viability as determined by CCK8 assay 
over 5 days. All values expressed as mean ± SD (n=3). C) Size distribution 
from nanoparticle tracking analysis of EVs isolated from BMMSCs seeded 
on Sylgard 184 PDMS substrates with differing base to crosslinker reagent 
ratios (n=3). D) Representative western blot of BMMSC EVs from each of 
the Sylgard 184 PDMS substrates and the corresponding cell lysates for EV-
positive markers ALIX, TSG101, and CD63 and cellular markers Calnexin 
and GAPDH (15 µg/lane). E) Representative TEM images of BMMSC EVs 
from the softest Sylgard 184 PDMS substrates and collagen-coated flasks. 
Statistical significance was determined by ANOVA; **p<0.01, 
***p<0.001, ****p<0.0001. 



 

 

48 
 

 

 

Figure 3.2: Incubation in media does not significantly affect the elastic 
modulus of the PDMS substrates. Elastic modulus of the A) Sylgard 184 
PDMS devices and B) mixed Sylgard 184 and Sylgard 527 PDMS devices 
before and after being incubated in DMEM a 37ºC for 4 days. All values 
expressed as mean ± SD. All data are representative of at least 2 
independent experiments (n=2). Statistical significance was determined by 
t-test; ns = not significant.  

 



 

 

49 
 

3.3.2 Substrate Stiffness Influences BMMSC EV Production and 

Bioactivity 

Using EV concentration and cell count data, EV production was quantified 

as EV/cell to assess if substrate stiffness affects EV production. The 5:1, 

10:1, 20:1, and 33:1 184 PDMS substrates resulted in EV production 5-fold, 

8-fold, 16-fold, and 15-fold greater than the flask + collagen control, 

respectively (flask + collagen: 1.73E4, flask – collagen: 2.93E4, 5:1: 

9.78E4, 10:1: 1.51E5, 20:1: 2.87E5, 33:1: 2.69 E5) (Figure 3.3A).  

 

We then sought to determine whether substrate stiffness had an effect on 

the angiogenic bioactivity of the generated EVs, since a common goal of 

regenerative medicine therapeutics is to promote vascularization and 

angiogenesis. To do so, we used two in vitro assays incorporating 

endothelial cells (human umbilical vein endothelial cells (HUVECs)) to 

model different stages of vascularization: a gap closure assay to measure 

cell proliferation and migration and a tube formation assay to model 

differentiation and re-organization. 5E9 EVs/mL were used for these 

assays, as this dosing scheme has previously been shown to be effective in 

probing for differences in angiogenic activity of EVs from different culture 

conditions [16]. Importantly, we were able to see significant differences 

between the negative control (endothelial basal media) and the treatment 

groups. 

 



 

 

50 
 

Notably, EVs generated by the BMMSCs cultured on the softest (250 kPa, 

33:1 184 PDMS) substrate caused significantly more cell migration and gap 

closure compared to the flask EVs and EVs from the other PDMS substrate 

groups (Figure 3.3B). This trend of increased angiogenic potential as a 

result of softer substrates was maintained in a tube formation assay, as EVs 

from the 33:1 184 PDMS substrate also significantly enhanced the number 

of loops compared to the rest of the EV groups (Figure 3.3C). 

 

 

Figure 3.3: Substrate stiffness influences BMMSC EV production and 
bioactivity. A) EV production as quantified by EVs per cell from BMMSCs 
seeded on Sylgard 184 PDMS substrates with different base-to-crosslinker 
ratios. EVs used for this data were from 1 day of collection and isolated and 
counted separately from the conditioned media from the other 2 days. After 
media collection, cells were trypsinized and counted (n=3). B) After a 
scratch was induced, HUVECs were treated with BMMSC EVs from the 
different substrates or growth or basal media, and percent gap closure after 
20 hours was evaluated via microscopy (n=3). C) HUVECs were 



 

 

51 
 

resuspended in EV treatments or growth or basal endothelial media, seeded 
in Matrigel-coated wells, and tube formation after 3-6 hours was quantified 
by the number of loops that had formed (n=3). All values expressed as mean 
± SD. All data are representative of at least 3 independent experiments 
(n=3). Statistical significance was determined by ANOVA; *p<0.05, 
**p<0.01, ***p<0.001, ****p<0.0001. 

 

3.3.3 Mixing Sylgard 184 and Sylgard 527 PDMS Provides a Wider Range 

of Elastic Moduli 

To examine a wider range of elastic moduli, we mixed Sylgard 184 and 

Sylgard 527, which is reported to yield PDMS substrates with a range of 

elastic moduli between ~3 kPa and ~2 MPa [157]. We confirmed that we 

could also achieve this range by mixing Sylgard 184, which was prepared 

at a 10:1 base to crosslinker reagent ratio per the manufacturer’s 

instructions, and Sylgard 527 at w/w ratios of 1:0 (pure 184), 10:1, 5:1, 1:1, 

1:5, 1:10, and 0:1 (pure 527) (Figure 3.2B, Figure 3.4A). We also showed 

that BMMSCs grown on these substrates remained viable and generated 

EVs similar to those from the prior characterized PDMS devices (Figure 

3.4B, C). Although the difference in EV production per cell was not 

significant between the 5:1, 1:1, 1:5, 1:10, and 527 groups, they all showed 

increased production when compared to the 184 and 10:1 groups, as well as 

the flask controls (Figure 3.5A). When looking at angiogenic bioactivity, 

treatment with 5E9 EVs/mL from the softer substrates, especially the 527 

PDMS, induced significantly higher levels of gap closure compared to the 

softer PDMS substrates (184, 10:1, 5:1) (Figure 3.5B), continuing the trend 



 

 

52 
 

seen with the 184 PDMS substrate variations. Furthermore, treatment with 

5E9 EVs/mL from the softer substrates, particularly the 139 kPa 1:5, 54 kPa 

1:10, and 3 kPa 527 devices, resulted in significantly greater tube formation 

than EVs from the stiffer PDMS devices and flasks (Figure 3.5C), 

suggesting that the substate stiffness-dependent change in angiogenic 

activity continues past the stiffnesses that were achieved when using only 

Sylgard 184. 

 

 

Figure 3.4: Mixed Sylgard 184 and Sylgard 527 PDMS substrate 
characterization. A) Elastic moduli of stiffness devices made with varying 
ratios of Sylgard 184 and Sylgard 527 PDMS. Values expressed as mean ±  
SD (n=4). B) Absorbance values from a CCK8 assay indicating cell 
viability over 4 days. Values expressed as mean ±  SEM (n=2). C) Size 
distribution from NTA of EVs from BMMSCs seeded on flasks or each 
substrate made with different ratios of Sylgard 184 and Sylgard 527. 
Statistical significance was determined by ANOVA; *p<0.05, ***p<0.001, 
****p<0.0001. 



 

 

53 
 

 

Figure 3.5: Softer 184:527 PDMS substrates improve the angiogenic 
bioactivity of BMMSC EVs. A) EV production quantified as EV per cell 
from BMMSCs seeded on each substrate made with different ratios of 
Sylgard 184 and Sylgard 527 (n=2). EVs used for this data were from 1 day 
of collection and isolated and counted separately from the conditioned 
media from the other 2 days. After media collection, cells were trypsinized 
and counted. B) After a scratch was induced, HUVECs were treated with 
BMMSC EVs from the different substrates or growth or basal media, and 
percent gap closure after 20 hours was evaluated via microscopy (n=3). C) 
HUVECs were resuspended in the different EV treatments or growth or 
basal endothelial basal media, and tube formation after 3-6 hours was 
quantified by the number of loops that had formed (n=3). All values 
expressed as mean ± SD. Statistical significance was determined by 
ANOVA; *p<0.05, **p<0.01, ****p<0.0001. 

 

3.3.4 iMSC EVs are Mechanoresponsive and Comparable Pro-Angiogenic 

Treatments to BMMSC EVs 

Donor-sourced MSCs have inherently variable properties based on the 

characteristics of the donor. Our group has previously demonstrated the 

significant variability in both the angiogenic and anti-inflammatory 



 

 

54 
 

bioactivity of BMMSC EVs from different donors [16]. Additionally, prior 

work from our group has shown a decrease in functional activity of 

BMMSC EVs, specifically in their angiogenic activity, after passage 4, 

which substantially limits their expansion capacity and makes donor MSCs 

unideal for clinically translatable therapeutics [15]. Therefore, we aimed to 

assess whether induced pluripotent stem cell-derived MSCs (iMSCs) are a 

suitable alternative to BMMSCs in the context of enhancing their pro-

angiogenic effect via mechanical regulation. 

 

iMSCs were seeded on the mixed PDMS substrates as described before, 

using only the 184, 10:1 1:1, 1:10, and 527 groups, as the differences in the 

elastic moduli were pronounced enough to still observe a clear trend in 

angiogenic activity. The treatment control substrate was reduced to just 

collagen-coated flasks, as the difference in the flask + collagen and flask – 

collagen groups were consistently not significant. EVs were isolated and 

characterized with NTA to determine size distribution and concentrations 

of the samples, which showed the expected EV diameters (100-200 nm) 

(Figure 3.6A). We then quantified iMSC EV production per cell and found 

that the three softest substrate groups (1:1, 1:10, 527) resulted in a 

statistically significant increase in EV production when compared to flask 

culture (Figure 3.6B). These results were also similar to those produced with 

BMMSC EVs. We also ensured that the iMSCs were viable and could be 

maintained on PDMS as the BMMSCs were (Figure 3.6C). An upward 



 

 

55 
 

trend in cell proliferation was confirmed over the course of 4 days since the 

iMSCs are not in culture longer than 4 days during the EV isolation process. 

TEM images confirmed the spherical EV morphology as expected from 

iMSCs on both TCPS and the 527 PDMS (Figure 3.6D), and western blot 

confirmed EV identity as EVs from all substrates were positive for EV 

markers CD63, ALIX, TSG101, and negative for cellular marker calnexin 

(Figure 3.6E). Finally, we determined that MSC identity was maintained 

after the EV collection process by western blots for the presence of CD90, 

CD73, and CD105, and the absence of CD45 (Figure 3.6F). 

 

 
 

Figure 3.6: iMSC EV characterization. A) EV production as quantified 
by EV per cell by EVs from iMSCs on different PDMS substrates. B) EV 
size and concentration distribution from iMSCs cultured on different PDMS 
substrates as determined by nanoparticle tracking analysis. C) iMSC 
proliferation/viability on PDMS substrates as measured by cell counting 
over 4 days. D) Representative TEM images of F+C EVs and 527 EVs 
confirming morphology. E) Western blot of EV markers CD63, ALIX, and 
TSG101, and EV-negative marker, calnexin, on EVs from each PDMS 
substrate (12µg/lane). F) Western blot of MSC markers CD73, CD105, and 
CD90 and negative marker CD45 on iMSC lysate from each PDMS 
substrate. THP1 cell lysate was used as a positive control for CD45. 



 

 

56 
 

(5µg/lane). All values expressed as mean ± SD. Statistical significance was 
determined by ANOVA; **p<0.01. 

 
Upon investigating iMSCs’ mechanoresponsiveness and their EV 

bioactivity, we observed that like BMMSC EVs, treatment with 5E9 

EVs/mL iMSC EVs caused significantly more gap closure compared to the 

endothelial basal media control (Figure 3.7A). The iMSCs also exhibited 

mechanoresponsiveness, as we show a pronounced substrate stiffness-

dependent effect on bioactivity where the softest substrate induced the most 

gap closure. This trend was reflected in a tube formation assay as well, 

where 5E9 EVs/mL iMSC EVs enhanced the number of loops formed by 

the HUVECs, with a significantly more potent angiogenic effect as 

substrate stiffness decreased (Figure 3.7B). It should be noted that the 

mechanoresponsiveness and trend in substrate stiffness-dependent change 

in bioactivity also continued past the elastic moduli of the softest 184 PDMS 

substrate (33:1, 250 kPa) since the elastic moduli of the 1:1, 1:10, and pure 

527 substrates were 139 kPa, 54 kPa, and 3 kPa, respectively. Moreover, 

since wound healing and tissue repair processes include a shift from a pro-

inflammatory state to an anti-inflammatory state, we also confirmed that 

substrate stiffness does not significantly affect the anti-inflammatory 

activity of the iMSC EVs upon looking at levels of pro-inflammatory TNF-

alpha and IL-6 (Figure 3.8). 

 



 

 

57 
 

 
 

Figure 3.7: Substrate stiffness affects the pro-angiogenic effect of iMSC 
EVs comparably to BMMSC EVs. A) EVs isolated from iMSCs on 
different 184:527 PDMS substrates were used to treat HUVECs after a 
scratch had been induced, and percent gap closure after 20 hours was 
evaluated via microscopy. B) HUVECs were resuspended with the same 
EV groups and seeded, and tube formation after 3-6 hours was quantified 
by the number of loops that had formed. All values expressed as mean ± 
SD. All data are representative of at least 3 independent experiments (n=3). 
Statistical significance was determined by ANOVA; *p<0.05, **p<0.01, 
***p<0.001, ****p<0.0001. 

 



 

 

58 
 

 
 

Figure 3.8: Substrate stiffness does not have a significant effect on the 
anti-inflammatory effects of iMSC EVs. Levels of pro-inflammatory 
cytokines A) IL-6 and B) TNF-alpha in conditioned media of RAW264.7 
cells treated with 5E9 EVs/mL in an LPS-stimulated mouse macrophage 
inflammatory assay, quantified by an E