SYNOVIAL EXTRACELLULAR MATRIX and SYNOVIAL MESENCHYMAL STEM CELLS ARE CHONDROGENIC in VITROAND in VIVO DISSERTATION Presented In

SYNOVIAL EXTRACELLULAR MATRIX AND SYNOVIAL MESENCHYMAL STEM CELLS ARE CHONDROGENIC IN VITRO AND IN VIVO

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the in the Graduate School of The Ohio State University

Nathalie Ann Reisbig

Graduate Program in Comparative and Veterinary Medicine

The Ohio State University

2018

Dissertation Committee:

Alicia Bertone, DVM, PhD, DACVS, DACVSMR (advisor)

Prosper Boyaka, PhD

Margaret Mudge, DACVS, DACVECC

Teresa Burns, DVM, PhD, DACVIM

Copyrighted by

Nathalie Reisbig

2018

ABSTRACT

Osteoarthritis (OA) is a progressive disease associated with cartilage injury and is the most common form of arthritis, affecting millions of people worldwide. Cartilage healing and treatments aimed at this is challenging due to the inherently limited repair capability of cartilage.

Our overall objective was to create a bioactive equine synovium scaffold, sConstruct, by infusing decellularized synovial-derived extracellular matrix (sECM) with synovial- derived mesenchymal stem cells (sMSCs). The goal was to produce a sConstruct that could house normal or engineered sMSC and that would have little immune reaction while improving cartilage healing.

The first part of this thesis investigates the potential of seeding the sECM with sMSCs to create a bioactive sConstruct. Synovium and synoviocytes were harvested from the femoropatellar and medial femorotibial joints from equine cadavers. sMSCs were cultured in monolayer and not treated or cotransduced with green fluorescent protein

(GFP) and human bone morphogenetic protein (BMP)-2. The sECM was decellularized with 0.1% peracetic acid (PAA) and seeded with sMSCs (0.5 X 10 6 cells/0.5 mL) by use of a 30% serum gradient. Cell migration, differentiation, and distribution into the sECMs were determined by CD90, viability, histologic morphology, fluorescence microscopy results and expression of GFP, BMP-2, hyaluronic acid (HA), and proteoglycan (PG). At

ii day 14, sMSCs were viable and had multiplied 2.5-fold in the sECMs. The sECMs seeded with sMSCs had a significant decrease in CD90 expression and significant increases in HA and PG expression. Seeding with sMSCs-BMP-2 enhanced the expression of BMP-2 and increased soluble HA and PG, indicated production of anabolic agents and sMSC differentiation in the scaffold. Because BMP-2 can promote repair of damaged cartilage, such a bioactive scaffold could be useful for treatment of injured cartilage.

The second portion of the thesis has two parts; 1) an in vitro model similar to that explained above but co-cultured with chondrocytes, and 2) placing sConstructs in the synovium in juxtaposition the a lesion in a rodent cartilage damage model. 1) In vitro survival, distribution, and chondrogenic potential of the sConstructs were assessed. sConstructs in co-culture with chondrocytes increased chondrocyte proliferation, viability, and Col II production, greatest in BMP-2-sConstructs. Chondrocyte presence increased the production of HA, PG, and BMP-2 by the sConstructs in a positive feedback loop. 2) sECM alone, GFP- or BMP-2-sConstructs were implanted in synovium adjacent to clinically created full-thickness rat-knee cartilage lesions. At 5 weeks, the lesion area and implants were resected. Gross anatomy, adjacent articular cartilage growth and subchondral bone repair were scored; peripheral, central and cartilage lesion measurements taken. For all scores and measurements, sConstruct implants were significantly greater than controls, greatest with the BMP-2-sConstructs.

Immunohistochemistry demonstrated migration of endogenous cells into the sECM, with greater cellularity in the constructs with intense positive GFP staining confirming iii engraftment of implanted sMSC and continued gene expression. Overall, we found that exposing cartilage to sConstructs was chondrogenic in vitro and in vivo and resulted in substantially increased growth in vivo. This effect was mediated, in part, by soluble ECM and cell factors and upregulation of anabolic growth proteins, such as BMP-2.

Lastly, we proposed that allogenic sConstructs could be a possible treatment of cartilage damage without eliciting a strong immune response. In addition, we wanted to understand in more detail what the sECM consisted of and what happens to it when seeded with sMSCs. Like above an in vitro co-culture was used, but now with peripheral blood mononuclear cells (PBMCs) rather than chondrocytes. Surface markers of sMSCs (CD44,

CD45, CD90, MHCI and MHCII) and of PBMCs (CD11b, MHCI and MHCII) in addition to biomarkers such as, IL-1, IL-1ra, IL-6, IL-10, TGF-beta, TNF-alpha and IFN- gamma, were investigated in vitro. The proteomic structure of sConstructs, sECM and sMSCs was analyzed. When adding allogenic PBMCs, sConstructs caused a low level of inflammatory response (increased IL-1, IL-6 and TNF-alpha in the media and increased

CD11b on PBMCs) as compared with an immunologic activator control,

Lipopolysaccharides (LPS). The inflammatory level of response followed a distinct pattern with sECMs the highest, then sMSCs and then sConstructs. Correspondingly, the

PBMC impact on sConstructs increased the production of anti-inflammatory cytokines

(IL-1ra and TGF-beta). This indicates that the sMSCs, as they are migrating and maturing into the sECMs, are becoming less immunogenic. Proteomics of the sConstructs substantiate the above conclusions. There were few changes when comparing proteomes of sConstructs to sConstructs co-cultured with PBMCs. There were significant changes iv when comparing sConstructs to its original constituents and there was an increase in sMSC metabolic proteins in the sConstructs.

In summary, we showed that it is possible to make a allogenic sConstruct that releases growth factors and anabolic agents over a longer period of time. Also, sConstructs surgically implanted adjacent to cartilage damage can significantly improve cartilage and subchondral bone repair, and potentially prevent the progression of OA. In addition, exposing the sConstructs and constituents to PBMCs in so-culture had little pro- inflammatory effect, and this affect was dampened by the addition of sMSCs. The proteomics supported this finding and showed that sECMs seeded with sMSCs becomes a unique product, the sConstruct. This work is “proof of concept” that sConstructs surgically implanted adjacent to cartilage damage can significantly improve cartilage and subchondral bone repair, with a low risk of a detrimental immune reaction.

ACKNOWLEDGMENTS

I would like to extend my sincere appreciation and gratitude to my primary supervisor

Dr. Alicia Bertone. Also, I would like to acknowledge my Advisory Committee members for their dedication and support throughout my research.

I express my sincere thanks to the many student, technicians, and secretarial staff who let their time, efforts, and expertise to this project: Dr. Becky Lovasz, Dr. Hayam Hussein,

Erin Pinnell, Logan Scheuermann, Michael Palillo, Haley Steiner and Dr. Liwen Zhang.

In addition, I would like to thank the Comparative Pathology and Mouse Phenotyping

Shared Resource / Histology and Immunohistochemistry Laboratory, and the

Proteomics Shared Resource (PSR) Laboratory at the Ohio State University, especially

Dr. Michael Freitas for his invaluable knowledge, support and patience.

Thank you most of all to my family and friends. To my dad, Richard Reisbig, for the hours spent discussing, formulating and formatting this work, and my mom for all her patience and support, no matter what. To Dr. James Carmalt, for always pushing me to continue, and his attention to detail. This work would not have been possible without them and their belief in me.

VITA

2002...... Diploma, Ringstabekk Skole, Oslo, Norway

2004...... International Baccalaureate, Oslo, Norway

2012...... Vet. Med. Leipzig, Germany

2012 -2013 ...... Intern, Large Animal Surgery, Leipzig University

2013 -2016 ...... Resident, Equine Surgery, The Ohio State University

2016 to present ...... Resident, Alternative Track Equine Sport Medicine

and Rehabilitation, The Ohio State University

2016 to present……………………Graduate Research Associate, Department of

Veterinary Bioscience, The Ohio State University

vii

PUBLICATIONS

Peer reviewed manuscripts of original research.

Reisbig N, Hussein H, Pinnell E, Bertone A. (2016) Comparison of four methods for generating decellularized equine synovial extracellular matrix. Am J Vet Res, 77(12):1332-1339.

Kaido M, Kilborne A, Sizemore J, Reisbig N, Aarnes T, Bertone A. (2016) Effects of repetition within trials and frequency of trial sessions on quantitative parameters of vertical force peak in horses with naturally occurring lameness. Am J Vet Res, 77 (7): 756-65.

Bertone A, Reisbig N, Kilborne A, Kaido M, Salmanzadeh N, Lovasz R, Sizemore J, Scheuermann L, Kopp R, Zekas L, Brokken. (2017) Prospective Controlled Clinical Trial for the Injection of Dental Pulp Tissue Particles for the Treatment of Equine Lameness Conditions, Front Vet Sci. 10;4:31.

Muir S, Reisbig N, Baria M, Kaeding C, Bertone A (2017) Plasma protein concentration using a polyacrylamide device increased plasma-origin IGF-1 that was additive to platelet-WBC-origin TGF- ß and IL-1 receptor antagonist concentration. Under review Am J Sport Med, Nov 2017.

Reisbig N, Hussein H, Pinnell E, Bertone A. (2018) Evaluation of equine synovial- derived extracellular matrix scaffolds seeded with equine synovial-derived mesenchymal stem cells. Am J Vet Res. 2018;79(1):124–33.

Reisbig N, Pinnell E, Scheuerman L, Bertone AL. (2018) Synovium Constructs Stimulate Chondrogenesis In Vitro and Cartilage Healing In Vivo. Under review, PLoS ONE, 2018.

Reisbig N, and Bertone AL. (2018) Immune and Signaling Proteins of Allogeneic Stem Cell-Extracellular Matrix Scaffold Interactions. Submitted, PLoS ONE, 2018.

viii

Presenting Author at International and National Published Abstracts and Proceedings

Reisbig, N, Hussein, H.A.G.H, Pinnell, E., Bertone, A.L. (2015) Decellularization to Produce Biological Synovial Extracellular Matrix Scaffolds. (abstr) in Proceedings. European College of Veterinary Surgeons Annual Scientific Meeting.

Reisbig N, H Hussein, E Pinnell, AL Bertone. (2015) Decellularization to Produce Biological Synovial Extracellular Matrix Scaffolds. Adv Vet Med Res Day, OSU CVM Book of Abstracts: MCB-15. Poster Competition.

Reisbig N, Hussein H, Pinnell E, Bertone AL. (2016) Characterization of living synovial extracellular matrix scaffolds for gene delivery (abstr), in Proceedings. American College of Veterinary Surgeons Annual Meeting, Nashville, Tennessee.

Muir S, Reisbig N, Baria M, Kaeding C, Bertone A (2018): Plasma protein concentration using a polyacrylamide device increased plasma-origin IGF-1 that was additive to platelet-WBC-origin TGF- ß and IL-1 receptor antagonist concentration. Presented American College of Sport Medicine and Rehabilitation, 3rd place equine abstract, Phoenix, Arizona.

FIELDS OF STUDY

Major Field: Comparative and Veterinary Medicine Veterinary Clinical Sciences

TABLE OF CONTENTS

ABSTRACT ...... ii

ACKNOWLEDGMENTS ...... vi

VITA ...... vii

PUBLICATIONS ...... viii

TABLE OF CONTENTS ...... x

LIST OF TABLES ...... xiii

LIST OF FIGURES ...... xiv

CHAPTER 1: INTRODUCTION SUMMARY ...... 1

1.1 Figure ...... 6

CHAPTER 2: EVALUATION OF EQUINE SYNOVIAL-DERIVED

EXTRACELLULAR MATRIX SCAFFOLDS SEEDED WITH EQUINE SYNOVIAL-

DERIVED MESENCHYMAL STEM CELLS ...... 7

2.1 Abstract ...... 8

2.2 Introduction ...... 10

2.3 Materials and Methods ...... 14

2.4 Results ...... 20

2.5 Discussion ...... 22

2.6 Figures ...... 30

CHAPTER 3: SYNOVIUM CONSTRUCTS STIMULATE CHONDROGENESIS IN

VITRO AND CARTILAGE HEALING IN VIVO ...... 37

3.1 Abstract ...... 38

3.2 Introduction ...... 40

3.3 Materials and Methods ...... 44

3.4 Results ...... 53

3.5 Discussion ...... 59

3.6 Tables ...... 65

3.7 Figures ...... 68

CHAPTER 4: IMMUNE AND SIGNALING PROTEINS OF ALLOGENEIC STEM

CELL-EXTRACELLULAR MATRIX SCAFFOLD INTERACTIONS ...... 76

4.1 Abstract ...... 77

4.2 Introduction ...... 79

4.3 Materials and Methods ...... 83

4.4 Results ...... 92

4.5 Discussion ...... 99

4.6 Tables ...... 111

4.7 Figures ...... 138

CHAPTER 5: DISSERTATION SUMMARY ...... 143

REFERENCES ...... 152

xii

LIST OF TABLES

Table 1. Preparation prior to in vitro co-culture experiment start Day 0 ...... 65

Table 2. Assays performed on 3,7, and 14 days ...... 66

Table 3. Scoring criteria for chondrocyte morphology, lesion gross anatomy, adjacent articular cartilage growth and subchondral bone repair ...... 67

Table 4. Cell locality categories used to group sConstruct proteins selected by EdgeR q <

0.028...... 111

Table 5. E/M proteins ...... 113

Table 6. C/M and C/E proteins in sConstructs ...... 119

Table 7. Elevated sConstruct proteins not found in E/M+/0/- ...... 130

xiii

LIST OF FIGURES

Figure 1. Three phase approach to creating an sConstruct ...... 6

Figure 2. Diagram depicting the seeding process for sMSCs into a sECM ...... 30

Figure 3. sMSC cell growth during the seeding of sECMs ...... 31

Figure 4. sMSC CD90 expression during the seeding of sECM ...... 32

Figure 5. Histology of sECM during the seeding with sMSCs ...... 33

Figure 6. Transduced sMSC BMP-2 expression during the seeding of sECMs ...... 34

Figure 7. sMSC HA expression during the seeding of sECMs...... 35

Figure 8. sMSC PG expression during the seeding of sECMs ...... 36

Figure 9. Histomorphometric measurements of lesion filling ...... 68

Figure 10. Co-culture chondrocyte cell counts and intracellular Col II production ...... 69

Figure 11. Co-cultured chondrocyte monolayers and morphology scores ...... 70

Figure 12. Co-culture sConstruct sMSC soluble biomarker concentrations ...... 71

Figure 13. Co-culture sConstruct sMSC count, viability, and maturity ...... 72

Figure 14. Lesion with sConstructs implants; smooth fibrocartilage repair tissue, adjacent articular cartilage and subchondral bone repair ...... 73

Figure 15. Lesion with sConstruct implants; growth scores and filling measurements ... 74

Figure 16. sConstructs recovered from implant site ...... 75

Figure 17. sMSCs with surface markers as a % of total cells...... 134

xiv

Figure 18. PBMCs with surface markers as a % of total cells ...... 135

Figure 19. Concentration of soluble inflammatory biomarkers in co-cultures ...... 136

Figure 20. Concentration of soluble anti-inflammatory biomarkers in co-cultures...... 137

Figure 21. Heat map and protein abundance ...... 138

Figure 22. E/M set protein abundance by cell locality category...... 139

Figure 23. sConstruct protein abundance by cell locality category...... 140

Figure 24. The union of C/M+, C/M-, C/E+, C/E-, and E/M+/0/- sets...... 141

Figure 25. sConstructs proteins not found in sECMs and sMSCs by cell locality and functional categories...... 142

CHAPTER 1: INTRODUCTION SUMMARY

This summary is a synopsis of the introductions of: Reisbig N, Hussein H, Pinnell E, Bertone A. Evaluation of equine synovial-derived extracellular matrix scaffolds seeded with equine synovial-derived mesenchymal stem cells. Am J Vet Res. 2018;79(1):124–33.

Reisbig N, Pinnell E, Scheuerman L, Bertone AL. (2018) Synovium Constructs Stimulate Chondrogenesis In Vitro and Cartilage Healing In Vivo. Under review, PLoS ONE, 2018.

Reisbig N, and Bertone AL. (2018) Immune and Signaling Proteins of Allogeneic Stem Cell-Extracellular Matrix Scaffold Interactions. Submitted, PLoS ONE, 2018

1 In the United States Osteoarthritis (OA) has the highest disability rate and health cost of any single disease. It is estimated to cost industrialized countries 1-2.5% of total gross domestic production [1]. In sport horses, diagnosis of OA normally leads to the end of the animal’s career and in many cases euthanasia. It is characterized as an irreversible degenerative joint disease with synovial inflammation and articular cartilage loss [2].

Micro-fracture, the most common surgical treatment, results in frail fibro-cartilage repair tissue even after long recovery period [3–5]. Because the rehabilitation time is long and often requires restricted movement of the joint such treatments are unacceptable in horses. New surgical treatments used to repair cartilage do so by directly grafting the injured cartilage site, such as Autologous Chondrocyte Implantation (ACI) [6] and

Osteochondral Autograft Transfer System (OATS) [7]. Both methods have shown good outcomes, the new tissue is still inferior in quality. Mesenchymal stem cell therapies, both as cells alone injected near the cartilage damage [8–10] or in scaffolds placed on the damaged tissue [11–15], have been explored extensively. They have yet to show significant improvement over surgical methods. The limited success of all the therapies appears to be that, due to the avascular nature of the joint and the density of the cartilage tissue, injured cartilage does not receive the continual-appropriate, bioactive anabolic mediators over the entire healing time period.

In this work, we have taken a novel approach of constructing a biological vehicle that could overcome this limitation. To do this we used synovium components, synovium mesenchymal stem cells (sMSCs) assimilated into a synovium extracellular matrix

(sECM), a synovium Construct (sConstruct). The therapy is envisioned to make the

2 sConstruct from “off the shelf” synovium constituents and then suture it to the animal’s synovium in juxtaposition to the cartilage damage using microsurgery. This sConstruct, made from components that are known to be involved in the healing process would create the appropriate changing growth environment around the damaged cartilage as the healing process progresses [16].

Synovium tissue was selected due to its high metabolic activity and a rich vasculature. It has the potential to create significantly elevated quantities of proteins secreted directly into the joint that could immerse the articular cartilage [17]. Synovium is a rich source of tissue-specific MSCs [18–20]. Synovial MSCs (sMSCs) was selected for our study because it can differentiate to other tissue types, e. g. cartilage, have rapid phenotypic differentiation and produce supportive joint-specific biomediators. Synovium-origin

MSCs have been shown in vitro to have a superior effect on chondrogenesis over all other MSCs [16,20,21].

To increase the efficacy of MSCs, techniques to place cells in the damaged area using a

MSC-seeded scaffold, gel, or aggregate implants has been developed [12–15]. Although cartilage is a natural choice for a scaffold, it has been shown to be a suboptimal scaffold for healing joint injury [14]. Current thought is that only a synthetic scaffold using biological-derived MSC or chondrocyte constituents may grow and mature along with the regenerating cartilage. Thus many investigations have focused on developing synthetic bio-constructs that mimic a natural scaffold [22].

3 In this work, we have taken the unique approach of using synovium tissue to make a extracellular matrix (sECM) natural scaffold [16]. The fundamental tissue structure of the sECM should retain the collagen components, the porosity of the native tissue, cell surface receptors such as integrins, serve as a reservoir for growth factors, and provide a substrate for cell attachment and migration [23]. In vitro, ECMs have been shown to encourage the formation of tissue specific phenotypes [24,25].

The work here follows a 3-phase approach (Figure 1). First, the development of appropriate decellularization procedures to produce the sECM was completed in our previous study [26] (Figure 1A). Second, to develop a reliable method for seeding and integrating sMSCs into the sECM forming an sConstruct (Figure 1B, Chapter 2 in this dissertation[16]). Third, showing that the sConstructs are chondrogenic in vitro and enhance cartilage repair in vivo (Figure 1C, Chapter 3, [27]).

The sConstructs could be deleterious causing an immune response when implanted in joints. We selected the constituents of the sConstructs with this in mind. Due to the

MSCs being recognized as 'immune privileged' it is thought this enables MSC transplantation across major histocompatibility barriers [28–30]. However, there have been some clinical trials reporting an immune response [28]. Suitably prepared ECMs, exhibiting low DNA concentrations and correspondingly short base pair lengths [31], have been used successfully as allogenic or even as xenogenic grafts [32,33]. The sConstructs, the combined sECM and sMSC’s, used allogenically, have a high potential of being a hypoimmunogenic.

4 In this work, we have tested the potential for an immunological response caused by the sConstructs (Chapter 4, [34]). They were exposed to peripheral blood mononuclear cells

(PBMCs) in vitro measuring both the inflammatory and anti-inflammatory effects. By analyzing the proteome of the sConstructs we were able to delineate the protein changes due to any immunological response.

Our goal was to have sConstructs that could be modified to provide the right biofactors for enhanced structured cartilage growth. We wanted sMSCs in the sConstruct to distribute, proliferate, differentiate retain viability and produce soluble biomediators. By including inserted transgenes, such as the chondrogenic bone morphogenetic protein two

(BMP-2, see Figure 1B), the anabolic biomediator concentration could be increased.

BMP-2 was chosen because of its strong anabolic properties and our laboratory’s access and previous work with these reagents [35]. We have confirmed in vitro that sConstructs produce greater concentrations of the growth factor BMP-2 from transduced sMSCs seeded into sECMs (BMP-2-sConstructs) compared to untransduced-sConstructs

(Chapter 2, [16]).

This work is a confirmation of a “proof of concept”. sConstructs have the potential to be used as a therapy for treating OA.

1.1 Figure

Figure 1. Three phase approach to creating an sConstruct A) Phase 1. Developing an sECM. See references [16,26] B) Phase 2. Developing an sConstruct from sMSCs and sECMs. C) Phase 3. Testing the sConstruct in vitro

CHAPTER 2: EVALUATION OF EQUINE SYNOVIAL-DERIVED

EXTRACELLULAR MATRIX SCAFFOLDS SEEDED WITH EQUINE

SYNOVIAL-DERIVED MESENCHYMAL STEM CELLS

Published ahead of dissertation; Reisbig N, Hussein H, Pinnell E, Bertone A. Evaluation of equine synovial-derived extracellular matrix scaffolds seeded with equine synovial- derived mesenchymal stem cells. Am J Vet Res. 2018;79(1):124–33.

2.1 Abstract

Our objective was To create a bioactive synovium scaffold by infusing decellularized synovial-derived extracellular matrix (sECM) with synovial-derived mesenchymal stem cells (sMSCs). Synovium from the femoropatellar and medial femorotibial joints from equine cadavers was used for the sECM. sMSCs were cultured in monolayer and not treated or cotransduced to enhance expression of green fluorescent protein (GFP) and human bone morphogenetic protein (BMP)-2. The sECM was decellularized with 0.1% peracetic acid and then seeded with sMSCs (0.5 X 10 6 cells/0.5 mL) by use of a 30% serum gradient. Samples were evaluated on days 0, 3, 7, and 14. Cell migration, differentiation, and distribution into the sECMs were determined by cell surface marker

CD90, viability, histologic morphology, and fluorescence microscopy results and expression of GFP, BMP-2, hyaluronan (HA), and proteoglycan (PG). At day 14, sMSCs were viable and had multiplied 2.5-fold in the sECMs. The sECMs seeded with MSCs had a significant decrease in CD90 expression and significant increases in HA and PG expression. The sECMs seeded with sMSCs cotransduced with GFP or BMP-2 had a significant increase in BMP-2 expression. sECM seeded with sMSCs or sMSCs cotransduced with GFP or BMP-2 yielded a bioactive synovial scaffold. Expression of

BMP-2 by sMSCs cotransduced to enhance expression of BMP-2 or GFP, and an accompanying increase in both HA and PG, indicated production of anabolic agents and synoviocyte differentiation in the scaffold. Because BMP-2 can promote repair of

8 damaged cartilage, such a bioactive scaffold could be useful for treatment of injured cartilage.

2.2 Introduction

Cartilage degeneration in humans, as seen in osteoarthritis or traumatic injury, is estimated to cost industrialized countries 1% to 2.5% of total gross domestic production

[36]. Cartilage repair is limited because of the low regenerative ability of the tissue and the progressive nature of cartilage lesions caused by trauma or diseases [37,38]. Current treatment strategies, such as mosaicplasty, autologous chondrocyte injection, or microfracture, have differing success rates, but long-term results typically are unsatisfactory [3,4,37]. A general drawback of these therapeutic strategies is that the newly formed tissue lacks the structural organization of cartilage and has inferior mechanical properties, compared with those of native tissue [3].

One area investigated for promoting cartilage structural repair or replacement has been the use of scaffolds, specifically when combined with regenerative cells and anabolic agents, for targeted cartilage delivery [11]. For synthetic scaffolds, a polycaprolactone nanofiber scaffold seeded with chondrocytes transduced with the AdBMP-2 gene resulted in greater and accelerated chondrogenesis in vitro [22]. Although such synthetic scaffolds can potentially provide a structure for chondrocyte growth, methods currently have not provided the ideal structure on which to build healthy hyaline cartilage [39].

Alternatively, ECM can be processed to provide a biological scaffold with the advantages of proteins and structure for optimal cellular ingrowth and differentiation that is tissue specific. Generation of biological scaffolds usually involves methods to decellularize the 10 tissue to create a matrix scaffold. Decellularized scaffolds from several tissues (eg, musculoskeletal, cardiovascular, urogenital, and integumentary structures) [40], have been found to be successful for promoting structural cell growth in vitro [41,42], maintaining tissue-specific cell phenotypes [41,43–45], promoting cell differentiation

[46], inducing tissue-specific differentiation [47], and enhancing chemotaxis of lineage- directed progenitor cells [48–50].

Use of cartilage ECM scaffolds for the treatment of damaged cartilage would seem the obvious choice to repair cartilage defects; however, such scaffolds have not been successful because of the matrix composition [51]. Cartilage ECM has a high matrix density with few spaces for cells; therefore, it does not take up a substantial number of chondrocytes. Cartilage ECM (when seeded with chondrocytes) yields scaffolds with few actively growing cells [52]. Thus far, scaffold regenerative cell–constructs for cartilage treatment have been unsatisfactory.

The purpose of the study reported here was to investigate a novel alternative to the use of biological scaffolds that would not serve as a replacement for cartilage; rather, it would ultimately be used as a transplant placed in direct apposition to the cartilage injury and serve as a biological booster for focal regeneration. This method may be complimentary or a replacement for current strategies that involve injecting MSCs or vectors containing anabolic genes or proteins directly into a damaged joint without targeted application. The central goal for the study was to determine whether decellularized synovial scaffolds could be successfully seeded with a synovial-origin cell and could serve to deliver anabolic genes locally. The scaffold in this application would serve as a vehicle that 11 positions both regenerative cells and anabolic proteins in juxtaposition to the damaged cartilage. The scaffold would not be designed to be part of the structured healed cartilage, although some cells could migrate to the site; instead, it would support the structured growth and development for repair of the injured cartilage on the basis of proximity.

Synovium was selected as the scaffold because synovium has a rich vasculature and is highly metabolic; hence, it can produce relatively high amounts of proteins that would be directly secreted into the joint fluid and bathe the articular cartilage [17]. In addition, over time, the tissue would be anticipated to integrate and be biodegraded. The ECM scaffolds have the ability to promote maintenance of specific cell phenotypes and differentiation states. This appears to be related to differences in tissue-specific architecture and molecular composition that have been observed in a number of experiments [18]. Such a scaffold, if prepared correctly, should have the tissue-specific architecture and molecular composition [42] that promote differentiation of sMSCs [18].

It was decided to use sMSCs because they are simple to obtain, can release anabolic agents that induce ordered cartilage growth [18], can be successfully transduced with

AdBMP-2 [53], and differentiate into chondrocytes under appropriate conditions [18].

With these attributes, such MSC scaffolds could provide a regenerative potential and an anabolic potential. The MSC scaffold may be able to assist with local cartilage repair and also improve the health of existing cartilage in situ or of transplanted chondrocytes

[54,55].

Information on synovial scaffolds with low residual cell and DNA content with preserved synovial villous architecture has been reported by our laboratory group [26].The ECM 12 scaffolds with MSCs or low DNA content both have low allogenic rejection rates and thus could have the potential to be used nonautologously. Our hypothesis was that similar decellularized synovial scaffolds could be successfully seeded with allogeneic synovial- origin cells by use of a serum gradient and would overexpress a protein anabolic to cartilage, which would yield a living biological scaffold for future transplantation. Our objectives were to use histologic examination and fluorescent cell tracking to measure the localization and engraftment of seeded synovial cells as well as to confirm gene expression of BMP-2 by these cells.

2.3 Materials and Methods

Sample. Villous synovium was aseptically harvested from the femoropatellar and medial femorotibial joints of 3 adult (< 7 years old) equine cadavers. Horses were euthanized for reasons unrelated to orthopedic problems and did not have a history or current signs of lameness. Joints were macroscopically inspected for any abnormalities, which included abnormal synovial fluid, thickened or inflamed synovium, hemorrhage in or around the joint, or visible cartilage damage. Horses with joints that had abnormalities were excluded. For horses with joints without abnormalities, all villus synovium that could be obtained was harvested. Synovium was dissected from the underlying fat and fibrous layer of the joint capsule and placed in Dulbecco modified Eagle medium1 for transport.

Experimental design. A portion of the harvested synovium was used to prepare sECMs, and a portion was used for isolating sMSCs. Some of the sMSCs were transduced with

E1-A–deleted AdBMP-2 and AdGFP. The sMSCs were assigned such that they were allogeneic with sECMs.

Purified sECMs were seeded with no cells, untreated sMSCs, or transduced sMSCs (the day of seeding is designated as day 0). Supernatants of cell-scaffold constructs were

1 Gibco, Gaithersburg, Md. 14 collected and pooled for time periods 0 to 3, 4 to 7, and 8 to 14 days. Supernatants were assayed for BMP-2, HA, and PG concentrations.

On days 3, 7, and 14, samples of cell-scaffold constructs were histologically prepared or digested for microscopic analysis or flow cytometry, respectively. Light microscopy was used to evaluate cell migration, and fluorescent microscopy was used to monitor cells transduced with GFP. Flow cytometry was used for determining cell viability or the presence of the stem cell marker CD90. The analysis was conducted over a 14-day period because previous studies [56,57] for the preparation of bioactive cell-scaffold constructs revealed continual growth during that period.

Culture of sMSCs. Synovium was aseptically divided into small pieces and digested in

0.02% collagenase type II solution2 (37°C for 5 hours); cells were filtered (70-µµm cell strainer), and collected cells were passaged at > 90% confluence. Passages 3 to 6 were chosen for transduction and seeding because this ensured that > 95% of the cells would express a multipotent MSC phenotype with minimal presence of lymphocytes, natural killer cells, and macrophages [58,59]. Flow cytometry was conducted at the time of seeding to determine baseline CD90 expression.

AdBMP-2 and ADGFP vector and synovial cell transduction. Replication-deficient E1-

A–deleted AdBMP-2 (adenoviral vector encoding a 1,547 base-pair open reading frame segment of human BMP-2) and AdGFP under the control of the cytomegalovirus promoter were generated [38] and titered by use of a plaque assay and then were stored at

2 Sigma Aldrich, Steinheim, Germany. 15 –80°C. The sMSCs were incubated at multiplicity of infection [53] of 100 with AdBMP-

2 and AdGFP in coculture (day 0 was the first day of culture). On day 3, cells were evaluated for GFP expression by use of fluorescent microscopy. The sMSCs and transduced sMSCs were treated with trypsin for 3 minutes, washed twice with PBS solution, and suspended in supplemented -modified Eagle medium3 at a concentration of

1.0 X 106 cells/mL.

Preparation of decellularized sECM. Synovium for use as scaffolds was pooled within each horse and divided by use of a dissecting microscope into 1 X 1-cm sheets. Sheets then were cut into standard disks with an 8-mm biopsy punch. Thickness of individual disks ranged from 5 to 7 mm.

The decellularization method consisted of incubation at 37oC in 0.1% peracetic acid4 and

4% ethanol for 6 hours with mechanical agitation. Samples were washed twice (15 min/wash) with PBS solution and twice (15 min/wash) with deionized water, and the process then was repeated as described elsewhere.5

Samples of sECM from each horse were randomly assigned (each sample, and the sMSCs from same horse, received a number and was assigned to another sample with a different number in a blinded fashion) to a control group (no cells) or to 1 of 2 sMSC groups

(sECM with sMSCs or sECM with transduced sMSCs) such that at least 27 sECM samples from each horse (81 total for the 3 horses) were allotted to each of the 3

3 Sigma Aldrich, Steinheim, Germany 4 Sigma Aldrich, Steinheim, Germany 5 Thermo Fisher Scientific, Wilmington, Del. 16 treatment groups in triplicate for harvest at each of 3 time points. Samples of sECM were frozen at –80oC in dry ice in ethanol and used within 1 month. Extra samples were maintained at –80oC and, when needed as a replacement, were used within 2 months. sECM seeding with sMSCs. Thawed sECMs were placed in 6-mm inserts6 designed to fit into standard 12-well culture plates and incubated for 3, 7, or 14 days. The sECM alone, sECM with sMSCs, or sECM with transduced sMSCs were cultured at 37oC in supplemented -modified Eagle medium with a 30% FBS7 gradient (10% FBS in the insert medium and 40% FBS in the well; Figure 2) [22]; culture medium was also supplemented with penicillin (100 U/mL), streptomycin (100 g/mL), and amphotericin

(250 ng/mL). Triplicate samples from each of the 3 horses for sECM alone, sECM seeded with 0.5 X 106 sMSCs, or sECM seeded with transduced sMSCs and incubated for 3, 7, or 14 days were prepared. Seeding was allogeneic such that sECMs from a horse were seeded with sMSCs from that same horse.

Flow cytometry. On day 3, 7, or 14, sECMs were digested by use of the previously described method. The sMSCs from digested sECMs containing transduced or untreated sMSCs were transferred into separate tubes, stained by incubation with the fluorescent chemical 7-aminoactinomycin D8 (7AAD) for 2 hours, washed twice with PBS solution, and analyzed with flow cytometry. Cell viability was determined by detecting the fraction of cells stained with 7AAD and reported as the percentage of the total number of cells

6 Corning Costar Transwell, Sigma Aldrich, Steinheim, Germany 7 BD Biosciences, San Diego, USA 8 BD Biosciences, San Diego, USA 17 counted. Transduced and untransduced sMSCs were washed twice with 2 mL of PBS solution and centrifuged at 300 X g for 3 minutes. Cells were counted with a hemacytometer, and 5 X 106 cells were placed in a tube. Then, 0.2 L of anti-human

CD90 primary monoclonal antibody9 were added, and cells were incubated at 37°C for 2 hours. After incubation was completed, cells were washed twice with PBS solution, resuspended in 200 L of PBS solution, and analyzed by use of flow cytometry. These represented cells that had been seeded on the scaffold, had proliferated, and were in the scaffold at the end of the incubation period.

Histologic characterization. Synovium, sECM alone, sECMs seeded with sMSCs, and sECMs seeded with transduced sMSC were fixed, thinly sectioned (thickness, 8 µm), stained with H&E stain, and evaluated by use of polarized light. Unstained and unfixed sECMs seeded with transduced and untreated sMSCs were evaluated with fluorescent microscopy on days 3, 7, and 14. Slides were evaluated by 2 investigators (NAR and

ALB) to assign a consensus score for histologic appearance; the investigators were not aware of the source of the sections. For sections stained with H&E stain, scores were assigned for cell content and migration by use of a scale from 0 to 4 as follows: 0 = few or no cells and no migration, 1 = moderate cell concentration in an area that represented <

10% of the scaffold, 2 = high cell concentration in an area that represented 25% of the scaffold, 3 = high cell concentration in an area that represented > 50% of the scaffold, and 4 = high cell concentration throughout the scaffold.

9 BD Biosciences, San Diego, USA 18 Quantifying BMP-2, HA, and PG concentrations. All medium from the 6-mm inserts was collected each day and stored at –20°C. Inserts were refilled with 10% FBS medium.

Every 3 days, the medium was removed and refreshed with 40% FBS -modified Eagle medium. The collected medium was pooled for days 1 to 3, 4 to 7, and 7 to 14. Samples were analyzed by use of ELISAs to quantify BMP-2,10 HA,11 and PG12 concentrations.

Data analysis. Numeric data (cell loss; percentage of cell viability and CD90 expression; and HA, PG and BMP-2 concentrations) were analyzed by means of the Shapiro-Wilk method to confirm a normal distribution. With horse as a random factor, an ANOVA was performed to determine whether values differed on the basis of horse or treatment group.

Horse did not have a significant effect, and analysis was performed as a 2-way ANOVA

(group and time), which was followed by a Tukey posttest. All analyses were performed with standard software13; values of P < 0.05 were considered significant. Data were expressed as mean ± SEM.

10 BMP2 Quantikine, R & D Systems, Minneapolis, Minn. 11 Hyaluronan Quantikine, R & D Systems, Minneapolis, Minn. 12 Proteoglycan assay kit, Rheumera, Redmond, Wash. 13 SPSS, version 22.1, IBM, Fullerton, Calif. 19

2.4 Results

Mean number of viable cells increased 2.5-fold in the sECMs by day 14; there was no significant difference between sECMs with sMSCs or sECMs with transduced sMSCs

(Figure 3). Cell death peaked on day 3, with a mean decrease in viable cells of 39% in all groups.

At baseline, all sMSCs had CD90 expression on > 70% of cells, which is consistent with cells that have stem cell–like characteristics. The CD90 differentiation surface marker on sMSCs decreased to a mean of 51% of cells on day 3, 10% of cells on day 7, and 30% of cells on day 14 (Figure 4). Percentage of CD90 expression in viable cells was significantly (P < 0.001) lower at day 14 than at day 3, which suggested cell differentiation. An overall decrease in total GFP fluorescence with concurrent fluorescent clusters in the sECMs seeded with transduced sMSCs was observed over the 14-day period, which indicated cell dilution with proliferating nonexpressing daughter cells

(Figure 5).

Cell proliferation was also apparent for microscopic examinations of H&E-stained seeded sECMs (Figure 5). This included a cell pattern that initially started at day 3 as a focal area of high cell content on the surface and became more dispersed on the surface of the scaffold; by day 14, cells reached halfway through the sECMs.

20 The sECM alone had a high baseline concentration of BMP-2, but only sECMs seeded with transduced sMSCs had a significant (P < 0.001) increase in BMP-2 concentration in the media, compared with the BMP-2 to concentration in sECM seeded with sMSCs or sECM seeded with untransduced sMSCs (Figure 6). Release of BMP-2 into the media from sECMs, regardless of whether they were seeded with sMSCs, revealed a pattern of decrease over the 14-day period. Correspondingly, there was a 2.5-fold increase in cell number over that same period.

Concentration of HA, a nonsulfated anionic glycosaminoglycan found in ECMs, was significantly (P < 0.001) greater in sECMs seeded with transduced sMSCs than in sECMs seeded with untreated sMSCs at all time points (Figure 7). The sECMs seeded with sMSCs had significantly (P = 0.004) greater HA concentrations in the media, compared with the HA concentration for sECMs alone for days 3 and 7, but the HA concentrations did not differ significant at day 14.

Concentrations of PG, a major protein component of ECM, were consistently present in sECM (mean, 5.1 g/mL) throughout the study (Figure 8). The addition of sMSCs to the scaffold (cell-scaffold construct) significantly (P = 0.03) increased the PG concentration

1.5-fold (mean, 7.5 g/mL) in the initial sample representing days 0 to 3. In constructs containing sMSCs, the PG concentration did not differ over time. In the cell-scaffold constructs containing transduced sMSCs, PG concentration further increased 1.3-fold by days 7 and 14, which was significantly greater than the PG concentration in cell-scaffold constructs containing untreated sMSCs.

2.5 Discussion

Results of the study reported here indicated that a viable engrafted cell-scaffold construct can be generated from synovium and secrete an anabolic agent into soluble medium. The cell infusion or decellularization method (or both) caused an initial decrease in viable cells that rebounded within a 14-day incubation period, with sMSCs proliferating to a

2.5-fold increase over the number of seeded cells. Histologic analysis confirmed that the cells migrated into the scaffold, starting with clusters that spread across the surface, and then penetrated deeper into the interior of the sECM. Simultaneously, transduced cells increased their production of HA and PG (markers for extracellular component production), and all cells had a concurrent loss of CD90 (loss of stem cell characteristics), which were suggestive of maturation and transformation of the sMSCs. The transduced sMSC-scaffold constructs had a greater concentration of BMP-2 in the medium, which suggested that these cells would be able to increase BMP-2 concentrations in synovial fluid in vivo. The BMP-2 concentrations were associated with greater PG content in these transduced constructs and were within the lower amounts of confirmed biological activity

[17].

The present investigation was designed to provide proof of concept for the generation of a biological viable scaffold in a laboratory setting. This study was not designed to recapitulate or replace synovium or to develop a commercial cellular product. To manufacture these constructs will require further characterization of cells and scaffolds to

22 meet federal standards for manufacturing and commercialization. The study involved the use of cells and matrix of synovial origin to retain characteristics supporting a synovial phenotype by use of routine laboratory cell culture techniques that are highly repeatable.

The sECM constructs in the present study had general characteristics of a potential bioactive scaffold [17]. If the scaffolds show promise for mice in vivo, further efforts would be needed to define the cells and scaffolds to determine collagen composition and structure. In addition, proteomic analysis would be needed for other anabolic and anti- inflammatory mediators that might be present in the scaffold.

It is challenging to induce cell migration into ECMs, and numerous techniques have been explored, which have included an electric charge, perfusion, and pressure [22]. Similar to another study [22] conducted by our laboratory group, the study reported here involved the use of a 30% chemotactic gradient (from 10% to 40% FBS) to induce sMSC migration. The experimental design with the 3-fold gradient also caused an initial osmotic pressure gradient of > 6.9MPa. Analysis of the results for the present study indicated that the combined chemotactic gradient and osmotic pressure were successful in seeding and growing sMSCs in sECMs. The migration was consistent and had a repeatable pattern over time. Cells started on the surface, often in clusters, and then moved into the interior of the sECMs. Morphologically, these cells were elongated and aligned in threads of cells onto the collagen framework of the scaffold. Mesenchymal development in the embryonic limb bud starts in a cellular condensation process, which leads to the development of cell-cell junctions and subsequent chondrocytic

23 differentiation [60–62]. These observations may support a similar propagation of the sMSCs, whereby the clusters differentiate and then migrate into the sECMs.

After cultures were incubated for 3 days, 39% of the initial seeded cells were dead for both transduced and untransduced sMSCs, which is similar to a result observed in another study [53]. Because this was seen for both transduced and untransduced cell-scaffold constructs, it was unlikely to have been attributable to cytotoxic properties of the adenovirus transfection. Possibly, residual peracetic acid could have contributed to cell death, but variability among samples would have been expected if this were the case.

More likely, the procedure caused an initial environmental shock (high chemotactic gradient combined with osmotic pressure), which resulted in cell death. Reducing the multiplicity of infection may decrease initial cell death and may still yield sufficient amounts of BMP-2 [55], however, the study reported here started with viable transduced cells that had fully survived the transduction process, and the reduction in cell number occurred regardless of transduction. Cell death occurred early after seeding; by 14 days, sMSCs had increased 2.5-fold, were extensively found within the sECM, and had a biological signature of mature synovial cells (as determined on the basis of morphology and matrix production). The cells were producing anabolic growth factors known to positively influence cartilage repair. In the present study, MSCs were characterized by use of CD90, which is considered an important marker for stem cell-like characteristics

[63,64]. The definition of MSC is controversial in regenerative medicine and is moving toward the addition of other modifiers to capture the capabilities and differentiation status of a stem cell (eg, totipotent vs multipotent, immortal, or resting progenitors). Examples

24 of these broader definitions include statements such as “Stem cells are a class of undifferentiated cells that are able to differentiate into specialized cell types”[65] or “A cell from which a variety of other cells can develop through the process of cellular differentiation” [66]. The monolayer cultures used in the present study drive mesenchymal cells toward dedifferentiation. There is disagreement as to whether these cells should be used to mimic tissue-specific cell types. Synovial cells isolated and cultured by use of the same techniques as were used in the study reported here have multiple other markers of stem cell-like characteristics, and CD90 expression is a clear reflection of the cell state, as indicated by these other markers [58]. For the present study, we used the term sMSC as the description for the cells at the start of the experiments.

These cells came from synovium, adhered to plastic (an inclusion criterion for mesenchymal cells), and expressed CD90 (> 70% expression at the start of the study).

Expression of CD90 is a widely accepted marker of undifferentiation (fetal cells, thymocytes, and cancer cells) [67]. Investigators of other studies [18,68–70] conducted by use of the same techniques that were used in the present study have termed these cells sMSCs.

The significant decrease of CD90 expression as sMSCs migrated into the sECMs can reflect sMSC differentiation induced by the matrix of the scaffold and a transition from the dedifferentiating conditions of monolayer culture. Initially, CD90 was expressed by most cells (> 70%), but expression decreased to 10% after incubation with the sECMs for

7 days. Subsequently, the proportion of cells expressing CD90 increased to 28% at 14 days; during that same period, there was a > 2-fold increase in viable cells. These results

25 indicated that after culture for 7 days, most cells were mature but lacked the CD90 marker; after culture for 14 days, there was a higher concentration of rapidly dividing young but immature cells that expressed CD90, with older mature cells that lacked CD90 expression and that divided less. These results are consistent with those of another study

[71] conducted by use of a collagen scaffold, whereby it was found that MSCs differentiated and migrated during infusion depending on the scaffold material.

The concentration of BMP-2 in the media was greatest in the cell-scaffold constructs seeded from transduced sMSCs. Although the sECM alone had the lowest concentration of BMP-2, it yielded a high concentration of BMP-2, which supported the potential the biological scaffold may have to induce paracrine and autocrine responses even without cell seeding. The BMP-2 concentration decreased slightly, although not significantly, in all groups (including unseeded sECMs) during culture for 14 days. This may have represented leaching of the endogenous BMP-2 in the sECM because the slope of the decrease was similar among all 3 groups. This observation, combined with the increase in cell numbers, supported sustained BMP-2 production by the original cells. This observation supported the fact that the adenovirus vector (AdBMP-2) was replication deficient and hence did not transfer the BMP-2 gene to the daughter cells during cell division, which could explain the steady production of BMP-2 [72].

Concentrations of BMP-2 protein in the range of 170 to 180 ng/mL can have physiologic importance for bone formation or chondrogenesis [60,61]. If these cell-scaffold constructs were placed in juxtaposition to damaged cartilage, the local BMP-2 concentration would be much higher and may have a positive effect on surrounding cells. 26 Additionally, BMP-2 efficacy does not require equal enhanced production by every cell because paracrine and autocrine effects of BMP-2 will occur on adjacent cells [55]. The complete duration of expression BMP-2 was not evaluated in the study reported here, but it can reportedly last for at least 1 month [53]. The significant increase in soluble BMP-2 for the transduced sMSCs provided a boost to the BMP2 concentrations released from the sECM. The combination of the sECM and BMP-2–expressing cells generated a soluble

BMP-2 concentration in the media within the biologically active range, which supported the contention that the seeded sECM could serve as a bioactive scaffold in vivo.

Increased HA production supported biological activity of sMSCs that is unique to mature synovial-fibroblast cells. The production of HA combined with the decrease of CD90 expression supported the synovial phenotype of the sMSCs. In vivo, HA is produced by synoviocytes and released into a joint; HA plays an important role in the protection of articular cartilage and transport of nutrients to cartilage. In patients with rheumatoid arthritis, it has been reported that HA acts as an anti-inflammatory substance by inhibiting the adherence of immune complexes to neutrophils or by protecting the synovial tissues from the attachment of inflammatory mediators [62].

The production of PG was also indicative that the sMSCs were of synovial lineage, phenotype, and function. The sECM consisted of heparan sulfate, dermatan sulfate, and lower concentrations of chondroitin sulfate, a compound found in high concentrations in cartilage [73]. Transduced sMSCs had significantly greater concentrations of PG than did constructs of sECMs with untreated sMSCs, which reflected upregulation of cell metabolism or targeted matrix production (or both). Enhanced BMP-2 production of 27 transduced sMSC–sECM constructs may stimulate the production of PG in the sECM during cell growth [74]. Although BMP-2 is a potent chondrogenic protein in certain environments, it appears that the synovial condition is not one of those environments.

Cartilage metaplasia was not seen in the synovial constructs of the present study and has not been seen in the synovium of rodents injected with MSCs transduced with AdBMP-2

[75] or in horses with AdBMP-2 injected into joints [76]. Furthermore, cartilage metaplasia has not been detected when adenoviral vector–transduced chondrocytes secreting BMP-7 were transplanted into joints of horses [72]. Further in vitro co-culture experiments with chondrocytes would yield important information on the cell-scaffold construct response to chondrogenic signals.

Other strategies for delivery of anabolic or anti-inflammatory proteins to cartilage are also being developed. They include direct delivery of vectors containing BMP-2 [77], insulin-like growth factor-1 [78], or interleukin-1 receptor antagonist [79] to joints, direct delivery of BMP-2–transduced neocartilage grafts [53], and intra-articular injection of

MSCs with [75] or without [80] engineering to produce chondrogenic mediators. These strategies have the advantage of only requiring injection of a joint; therefore, patients would not require invasive surgery. However, these methods have adverse effects that include toxic effects attributable to vectors as well as systemic effects. When acting as vectors, allogeneic cells or engineered cells move directly into the large and vascular surface area of the synovium, where they can cause perivascular cuffing of lymphocytes in synovium at 60 days after injection of MSCs and elevated vector antibody concentrations in contralateral joints [76,79]. Intra-articular injection of MSCs without

28 engineering may provide some beneficial effects [80], but the benefits of cell engineering would not be achieved. These injection methods could also be used concomitantly with the approach described in the present study.

Increased numbers of viable cells, cell migration, and engraftment into the scaffold and evidence of cell differentiation supported that sECMs seeded with sMSCs or transduced sMSC resulted in a bioactive synovial cell–scaffold construct. Compared with results for sMSCs, transduced sMSCs produced greater BMP-2 concentrations and greater HA and

PG concentrations, which indicated gene production and enhanced synovial function, which accelerated the local phenotype in these cells. It has been reported [55,74] that

BMP-2 can support chondrogenesis, so the bioactive cell–scaffold construct described in the present study has the potential to promote cartilage repair in a paracrine manner.

Analysis of results of the study reported here suggested that sECMs seeded with synovial-origin stem-like cells or genetically engineered sMSCs may promote cartilage repair.

2.6 Figures

Figure 2. Diagram depicting the seeding process for sMSCs into a sECM An 8-mm-diameter sECM was placed in a 6-mm insert. Untransduced or transduced sMSCs (500,000 cells in 0.5 mL) was added to the insert. Notice the 30% FBS gradient.

Figure 3. sMSC cell growth during the seeding of sECMs Mean ± SEM number of cells for sECMs seeded with sMSCs, as measured by use of flow cytometry and 7-aminoactinomycin D, after incubation for up to 14 days (day 0 = seeding and initiation of culture). A—Total number of cells (viable plus dead; triangle and solid line) and the number of viable cells (total number of cells minus the number of cells stained with 7-aminoactinomycin D; squares and dotted line ). B—Total number of sMSCs (triangles and dashed line), number of viable sMSCs (squares and dotted line), total number of transduced sMSCs (triangles and solid line), and number of viable transduced sMSCs (squares and solid line) after incubation for up to 14 days. There was no significant (P ≥ 0.05) difference between the number of untransduced and transduced sMSCs at any time point.

Figure 4. sMSC CD90 expression during the seeding of sECM Mean ± SEM percentage of cells expressing CD90 in cell-scaffold constructs seeded with untransduced sMSCs (triangles and solid line) and transduced sMSCs (squares and dotted line) and the total number of viable cells (diamonds and dashed line) after incubation for 0, 3, 7, and 14 days (day 0 = seeding and initiation of culture). There was no significant (P ≥ 0.05) difference between the number of untransduced and transduced sMSCs at any time point. On day 0, the percentage of sMSCs that expressed CD90 was more than 70%.

Figure 5. Histology of sECM during the seeding with sMSCs Light and fluorescent microscopy of seeded sECM samples on day 3, 7 and 14. The left column of pictures are sECM seeded with sMSCs and the right column are sECM seeded with transduced sMSCs. Arrows show cells migrating into sECM.

Figure 6. Transduced sMSC BMP-2 expression during the seeding of sECMs Mean ± SEM BMP-2 concentration after incubation for up to 14 days in constructs of sECMs seeded with transduced sMSCs (triangles and solid line) or untransduced sMSCs (squares and dotted line) and in unseeded sECMs (diamonds and dashed line). Media were pooled for all days within a time period. There is a low endogenous turnover of BMP-2 in the unseeded sECMs. *Within a time period, value differs significantly (P < 0.05) from the value for the other groups.

Figure 7. sMSC HA expression during the seeding of sECMs Mean ± SEM HA concentration after incubation for up to 14 days in constructs of sECMs seeded with transduced sMSCs (triangles and solid line) or untransduced sMSCs (squares and dotted line) and in unseeded sECMs (diamonds and dashed line). There is a low endogenous turnover of HA in the unseeded sECMs. †Within a time period, value differ significantly (P < 0.05) from the value for the unseeded sECMs. See Figure 6 for remainder of key.

Figure 8. sMSC PG expression during the seeding of sECMs Mean ± SEM PG concentration after incubation for up to 14 days in constructs of sECMs seeded with transduced sMSCs (triangles and solid line) or untransduced sMSCs (squares and dotted line) and in unseeded sECMs (diamonds and dashed line). *Within a time point, value differs significantly (P < 0.05) from the value for the unseeded sECMs. †Within a time point, value differs significantly (P < 0.05) from the value for the untransduced sMSCs.

CHAPTER 3: SYNOVIUM CONSTRUCTS STIMULATE

CHONDROGENESIS IN VITRO AND CARTILAGE HEALING IN VIVO

Submitted and under review ahead of dissertation; Reisbig N, Pinnell E, Scheuerman L, Bertone AL. (2018) Synovium Constructs Stimulate Chondrogenesis In Vitro and Cartilage Healing In Vivo. Under review, PLoS ONE, 2018.

3.1 Abstract

Osteoarthritis (OA) is a progressive disease associated with cartilage injury and its inherently limited repair capability. Synovium-based cellular constructs (sConstructs) are proposed as possible treatments. Equine sConstructs were produced from decellularized synovium-based extracellular matrix scaffolds (sECM) seeded with synovium-derived mesenchymal stem cells (sMSC) and engineered to express green fluorescent protein

(GFP), or bone morphogenetic protein-2 (BMP-2). Survival, distribution, and chondrogenic potential of the sConstructs in vitro and in vivo were assessed. sConstructs in co-culture with chondrocytes increased chondrocyte proliferation, viability, and Col II production, greatest in BMP-2-sConstructs. Chondrocyte presence increased the production of hyaluronic acid (HA), proteoglycan (PG), and BMP-2 by the sConstructs in a positive feedback loop. sECM alone, or GFP- or BMP-2-sConstructs were implanted in synovium adjacent to clinically created full-thickness rat-knee cartilage lesions. At 5 weeks, the lesion area and implants were resected. Gross anatomy, adjacent articular cartilage growth and subchondral bone repair were scored; and peripheral, central and cartilage lesion measurements taken. For all scores and measurements, sConstruct implants were significantly greater than controls, greatest with the BMP-2-sConstructs.

Immunohistochemistry demonstrated migration of endogenous cells into the sECM, with greater cellularity in the constructs with intense positive GFP staining confirming engraftment of implanted sMSC and continued gene expression. In summary, exposing

38 cartilage to sConstructs was chondrogenic in vitro and in vivo, and resulted in substantially increased growth in vivo. This effect was mediated, in part, by soluble ECM and cell factors and upregulation of anabolic growth proteins, such as BMP-2. This work is “proof of concept” that sConstructs surgically implanted adjacent to cartilage damage can significantly improve cartilage and subchondral bone repair, and potentially prevent the progression of OA.

3.2 Introduction

Osteoarthritis (OA) has the highest disability rate and health cost of any single disease in the United States [1]. It is an irreversible degenerative joint disease characterized by articular cartilage loss and synovial inflammation [2]. The most widely used surgical treatment, micro-fracture, results in weak fibro-cartilage repair tissue even after long convalescence [3–5]. New surgical treatments, such as Autologous Chondrocyte

Implantation (ACI) [6] and Osteochondral Autograft Transfer System (OATS), are used to repair cartilage by directly grafting the injured cartilage site [7]. Although both methods have shown good outcomes, cell yield in the grafts is low and the new tissue is still inferior in quality. A long period of rehabilitation is necessary to avoid overload of the avascular grafted cartilage and is a significant limitation. The grafted injured site does not receive a continued, renewable source of bioactive anabolic mediators. We have suggested a different strategy to overcome this weakness [26][26]: Use of a microenvironment paracrine-like feedback loop via a bioactive transplant in the synovium nearby a cartilage injury. Such a transplant, made from synovium components

(sConstructs) has not, to our knowledge, been reported in the literature.

Synovium is a rich source of both pericyte and tissue-specific mesenchymal stem cells

(MSCs) [18–20]. We selected synovial MSCs (sMSCs) for our study to both retain the potential to differentiate to other tissue types (cartilage) for migrating cells, and rapid phenotypic differentiation and production of supportive joint-specific biomediators, such

40 as hyaluronic acid (HA). Other studies have shown that MSCs in joints due produce biomediators [6,13,81]. Treatments using autologous MSC solutions have tried to overcome the lack of growth nutrients around the damaged area [6]. MSCs injected into the joint improved cartilage repair, but MSCs were not found in the lesions [81]. These data support that MSCs produce biomediators that enhance the growth environment.

To enhance the efficacy of MSCs, cells have been placed in the damaged area using a

MSC-seeded scaffold, gel, or aggregate implant [12–15]. Cartilage, a natural choice for a scaffold, has been a suboptimal scaffold for healing joint injury [14]. Current thought is that a synthetic scaffold using biological-derived MSC or chondrocyte constituents may grow and mature along with the regenerating cartilage. The bio-construct could provide the appropriate growth environment over the whole healing period. These approaches still placed the bio-scaffolds in, or on, cartilage sites.

Many investigations have focused on developing a synthetic bio-constructs that mimic a natural scaffold that is still placed in, or on, cartilage [22]. We have taken the unique approach of using synovium tissue to make a natural scaffold [16] and potentially heal nearby cartilage. We selected synovium-origin tissue for our ECM scaffolds. The biological nature of sECMs should retain the porosity of the native tissue, the collagen components, cell surface receptors such as integrins, serve as a reservoir for growth factors, and provide a substrate for cell attachment and migration [23]. Additionally, in vitro, ECMs have been shown to encourage the formation of tissue specific phenotypes

[24,25]. We have reported our decellularization procedures to produce synovium-derived extra-cellular matrix (sECM) [26] and our sMSC seeding procedures to show that sMSC 41 will retain viability, distribute, proliferate, differentiate and produce soluble biomediators, including inserted transgenes, such as the chondrogenic bone morphogenetic protein two (BMP-2) [16]. BMP-2 was chosen because of its strong anabolic properties and our laboratory’s access and more than a decade of work with these reagents [35]. Indeed, we confirmed in vitro that sConstructs can produce greater concentrations of the growth factor BMP-2 from transduced sMSCs seeded into sECMs

(BMP-2-sConstructs) compared to untransduced-sConstructs [16]. Use of sMSCs transduced with green-fluorescent protein (GFP) was included in these studies for cell tracking, confirmation of successful transduction, in vitro. [17,18]

A few reports, mostly in vitro, have demonstrated that MSCs can be chondrogenic [18].

Synovium-origin MSCs have been shown in vitro to have a superior effect on chondrogenesis over all other MSCs [20,21]. Here, in this study, we have created an in vitro experimental design to show “proof of concept” that our sConstructs, made from synovium, could influence chondrocytes when not in direct contact by communicating via soluble factors in fluid (media). This co-culture model mimics the situation in joints with the synovium and cartilage communicating only via synovial fluid since articular cartilage is avascular. This in vitro system also can determine the presence of a feedback loop in which stimulated chondrocytes could further stimulate the sConstruct.

Chondrocyte morphology, cell viability, and intracellular Collagen Type II (Col II) were analyzed. The chondrocyte impact on sMSCs in sConstructs was followed by measuring the sMSC’s growth in cell counts, viability, and BMP-2 production. In addition, sMSC maturation is quantified by determining the concentration of soluble biofactors,

42 hyaluronan (HA) and proteoglycan (PG) excreted into the medium. Once demonstrated, we proceeded with in vivo studies.

Using our experience with the nude rodent model to study knee cartilage healing [55,77] we determined the healing potential of the sConstructs. The time point of healing, 5 weeks, was selected such that we could detect either a delay or acceleration of repair among our comparison groups. These rat knee lesions generally heal around 8 weeks.

Longer duration studies run the risk of missing the window such that the lesions could all be healed. For our in vivo studies, sConstructs were implanted in the sub-patellar synovium directly adjacent to clinically-produced, athymic-nude-rat-knee lesions on the femur. After a 5-week healing period, knees were resected and lesion morphology, fibrocartilage repair tissue, subchondral bone repair, and cartilage growth are examined and quantitated. Host lesion effects on sConstructs was determined by en bloc resection of the synovium implant site, evaluating the morphology, and tracking sMSC cell integration.

Our overall hypothesis was that sConstructs would enhance both chondrogenesis in vitro and cartilage repair in vivo. Additionally, BMP-2-sConstucts would be superior to the bioactivity of other sConstructs. If this hypothesis is correct, then this study offers a

“proof of concept” that sConstructs and their derivative transduced-sConstructs are an alternative therapy for OA.

3.3 Materials and Methods

General experimental methods. The following general methods were used through all the experiments. Cells were cultured under sterile conditions, at 37oC in Dulbecco's

Modified Eagle's Medium (DMEM), supplemented with penicillin,100 U/mL, streptomycin, 100 μg/mL, and amphotericin 250 ng/mL (supplemented DMEM)14 and

10% fetal bovine serum 15, in T75 flasks with shaking. This supplemented DMEM was used in all cell cultures unless specified otherwise. For co-culture studies, Transwell 12- well culture plates with snap-in 6 mm diameter inserts designed for the plates were used16

. Each lower well held approximately 0.6 mL of solution, and the insert held approximately 0.1 ml. Cells were counted on a hemocytometer or by flow cytometry done on an Accuri Cytometer17. Enzyme-linked immunosorbent assay (ELISA) kits followed the instructions of the vendors.

In Vitro study design. Cell and tissue preparation began 30 days before the start of the in vitro study with the fresh harvest of synovium and cartilage tissues for preparation of the sECM, sMSC, and isolated chondrocytes (Table 1) [16,26]. Once prepared, all cells and scaffolds were frozen and thawed prior to the start of the experiments that started on Day

0 on the time line detailed in Table 1. All experiments were performed comparing our

14 Gibco, Gaithersburg, Md. 15 FBS, Sigma Aldrich, Steinheim, Germany 16 Transwell Culture Insert, Corning Costar, Sigma Aldrich, Steinheim, Germany 17 Ann Arbor, MI 44 groups, described later, both with and without the chondrocyte monolayer culture in the bottom well. The in vitro study harvested media and specimens on Day 3,7, and 14 to study early influence among the biologics and mimic the time anticipated for clinical application (within 2 weeks) for our in vivo portion of the study. For each study with and without chondrocytes, nine replicates were performed (three horses, 3 samples each in triplicate) with allogeneic assignment of sMSC and sECM.

Chondrocyte, sMSC and sECM preparation. Equine villous synovium from the medial femoral patellar joint and articular cartilage from the femoral condyles were harvested

[26] from knee joints of healthy adults (2-7 years old), euthanized for reasons unrelated to this study. Briefly, the joints were aseptically prepared with 2 5min scrubs using 4% chlorhexidine and sterile H2O. Sterile gloves and instruments were used to enter and harvest tissue. The joints were macroscopically inspected for any abnormalities and excluded if any were noted. Chondrocytes and sMSCs were purified from the cartilage and synovium aseptically [16,53], divided into small pieces and digested in 0.02% collagenase type II solution18 at 37°C for 5 hours; cells were filtered (70-m cell strainer), and cultured. Cell monolayers were grown to a > 90% confluence and passaged

3 to 6 times prior to use. This ensured a uniform population of sMSCs in which >95% of the cells express a multipotent MSC phenotype with minimal presence of lymphocytes, natural killer cells and macrophages [58,59]. Caplin et.al. [19] have shown that prepared

MSCs can consist of tissue phenotype sMSC and perivascular cells, pericytes, pMSCs.

18 Sigma Aldrich, Steinheim, Germany 45 No attempt here has been made to quantify nor classify the pMSCs from the tissue sMSCs. sECM was decellularized from harvested synovium by immediately dividing the tissue into 1x1 cm sheets using a dissecting microscope, then incubating them at 37°C in 0.1% peracetic acid (PAA) and 4% ethanol for 6 hours with mechanical agitation twice

[26]. Samples of sECM were prepared using a 6 mm diameter biopsy punch, frozen wet at –80°C, surrounded by dry ice in ethanol. All the components were used within 30 days of preparation. sMSC transduction. Replication-deficient, E1-A-deleted adenoviral (Ad) vectors encoding for a 1547 base-pair open reading frame segment of human bone morphogenic protein-2 (Ad-BMP-2) under the control of the cytomegalovirus promoter were generated

[38], titered by plaque assay, and stored at -80°C. sMSCs to be transduced were incubated at a multiplicity of infection (MOI) of 100 [53] with Ad-BMP-2. On day -5 prior to the experiment, the untranduced and transduced cells (BMP-2-sMSC) were evaluated for BMP-2 expression by ELISA, the percent of viable cells and then suspended to a concentration of 1.0 × 10 6 cells/mL. sConstructs preparation. Approximately 50 sECM samples from each of three horses were thawed to have three different biologic samples run in triplicate distributed for each group: controls (sECM alone) and two sConstruct types; sConstructs with 1) untransduced sMSCs (untransduced-sConstructs) and 2) BMP-2 transduced sMSC

(BMP-2-sConstructs), seeded with 1 x106 untransduced or transduced-sMSCs, respectively. A 30% FBS, gradient from insert to well was made by having 10% FBS in the insert and 40% FBS in the well to attract the sMSC to seed deep within the sECM 46 [22]. The gradient was maintained by replacing the medium in the inserts and wells every

24 hrs. The seeding was allogeneic; sECMs from one animal were never seeded with the same animal´s sMSCs.

Chondrocyte preparation and experiment start. Single passaged Chondrocytes (1 x 103 chondrocytes/well) were cultured in 10% FBS for 3 days in 12 well culture plates prior to the experiment. The experiment was initiated (day 0) when Transwell inserts containing sECM or sConstructs were placed into the wells with the cultured chondrocytes. Media, approximately 0.7 ml, from the wells and Transwell inserts was collected daily (stored at

-20°C) and replaced with fresh media, 10% FBS supplemented DMEM.

In vitro sConstruct-chondrocyte co-culture assays. Table 2 is a summary of the assays performed on the chondrocytes-sConstructs co-cultures on days 3, 7, and 14.

Chondrocyte and sMSC cell counts, viability, and maturation – Cells on culture plates were trypsinized. sECMs and sConstructs from inserts were digested with collagenase as described above for synovium and cartilage. Cells were counted with a hemocytometer and flow cytometry. Cell viability was determined by detecting the fraction of cells stained with 7-aminoactinomycin D19, (7AAD). Cells were incubated with 7AAD for 2 hours, washed twice with PBS solution, and analyzed by flow cytometry. sMSCs maturation was determined by % cluster differentiation 90 (CD90). After cells were counted, 5 X 106 cells were combined with 0.2 L of anti-human CD90 primary

19 BD Biosciences, San Diego, CA 47 monoclonal antibody20 (Clone 5E10 directly conjugated to phycoerythrin), incubated at

37°C for 2 hours, washed, resuspended in 200 L of PBS, and analyzed using flow cytometry.

HA, PG, BMP-2 and Coll II determinations - ELISA kits were used as per manufacturer’s recommendations to determine the concentrations of HA21 , PG22 , BMP-223, and Col II24

. For HA, PG and BMP-2 co-culture medium from the insert and well was collected daily, frozen to -20oC, pooled for days 1-3, 4-7, and 8-14 then analyzed. For Col II concentrations, chondrocytes in the wells were lysed and effluents analyzed.

Morphology of sConstructs and chondrocytes from co-cultures. sECM and sConstructs from inserts were formalin fixed, thinly sectioned (8 m), and stained with hematoxylin and eosin (H&E) for subjective microscopic evaluation for architecture, cellularity and density of the eosin staining. For the chondrocyte morphology, the Transwell inserts were removed from the culture plates and the chondrocyte monolayers in the wells were fixed, stained with toluidine blue and microphotographed under standard conditions. The microphotographs were scored from 0 to 4 with 0.5-point increments for phenotype

(Table 3) by a trained investigator blinded to the treatment group [82]. One score was determined from a consensus score from multiple areas in the well.

20 BD Bioscience, San Diego, USA 21 Hyaluronan Quantikine, R&D Systems, Minneapolis, Minn. 22 Proteoglycan assay kit, Rheumera, Redmond, Wash. 23 BMP2 Quantikine, R&D Systems, Minneapolis, Minn. 24 MyBioSource, Inc., San Diego, CA USA 48 In vivo rat lesions with sConstruct implants. Twelve female athymic nude rats25 (10–12 weeks of age; 24 knees) were used with four groups (n= 6 knees each), suture alone

(control), sECM alone, GFP- or BMP-2-sConstructs (created as described above), following an approved Institutional Animal Care and Use protocol. Implants were randomized to different knees and rats. After the rats had 1 week of acclimation, anesthesia was induced with 5% isoflurane26, then maintained with 1% isoflurane delivered via nose cone. The surgical site was aseptically prepared and draped. Each knee had bilateral surgery; a full-thickness articular cartilage lesion was created in the intertrochlear groove of the femur until subchondral bone was penetrated, representing a marrow-stimulating technique. Specifically, a medial parapatellar approach was performed and the trochlear groove exposed by sub-luxating the patella. In the trochlear groove, a cartilage lesion, 2.5 mm diameter by ≈ 1.5 mm depth, was created using a motorized burr 27. In the control group (6 knees), suture was placed in the sub-patellar synovium directly adjacent to the trochlear groove. Constructs, prepared as described above, were removed from the Transwell inserts and cut to a 3 mm diameter with a biopsy punch. The assigned sConstruct implant was sutured (6-0 non-absorbable) to the synovium, similar to the control suture, directly adjacent to the lesion. The arthrotomy was closed with absorbable suture28. Buprenorphine29 0.05 mg/kg was administered intramuscularly for pain. Prophylactic Augmentin30 0.35 mg/mL was given 1 day

25 Charles River Laboratories, Wilmington, MA, USA 26 GlaxoSmithKline, Research Triangle Park, NC, USA 27 Dremel, Mount Prospect, IL, USA 28 Polydioxanone, Ethicon, USA 29 GlaxoSmithKline, Research Triangle Park, NC, USA 30 GlaxoSmithKline, Research Triangle Park, NC, USA 49 preoperatively and 2 weeks postoperatively in the animals’ drinking water. The rats were allowed to roam freely in their cage. Rats were observed daily for incision swelling or drainage, chewing, ambulation, appetite and changes in behavior.

Gross anatomy, articular cartilage growth, subchondral bone repair scores and lesion filling measurements. After 5 weeks, rats were euthanized with CO2. Rats had no sign of infection or lameness at any time. The articular surface of the knees was examined grossly and photographed for gross anatomy scoring. For articular cartilage repair and subchondral bone damage scores and all measurements, the distal femur was resected en bloc, fixed in 10% neutral-buffered formalin31 (NBF) for 48 hours, decalcified 32

(Decalcifier-S) for 12 hours, paraffin-embedded, and serially sectioned, 30 µm apart, in the sagittal plane, and stained with toluidine blue. Scoring was performed by two independent investigators (NR and ALB) blinded to implant group. Scoring criteria was a modification of an OA scheme described by Gerwin et al. [83] (Table 3). Scores ranged from 0 to 4, where 4 was growth like original tissue and 0, maximum damage. Five measurements were taken for lesion filling (Figure 9); a) width of opening halfway between the lesion bottom and the projected original surface, b) width, at the same depth as measurement a, of opening of physiological cartilage, c) depth of the opening at the center, d) depth at 100µm from edge of original lesion, and e) depth of original lesion, from the projected cartilage surface to the estimated original tidemark. All measurements

31 Sigma Aldrich, Steinheim, Germany 32 U.S. Biotex, Webbville, KY, USA 50 were obtained using imaging software (Image J, NIH, USA) and the reported values of all scores and measurements was the mean of three consecutive sections [84].

Lesion-sConstruct morphology - Implants were identified in the synovium by the suture and cut out en bloc, then carefully dissected free, fixed in neutral buffered formalin, paraffin-embedded, sectioned (7m), stained with H&E. GFP Immunohistochemistry was done as previously described [10]. In brief, formalin fixed synovial biopsy specimens were sectioned (4 m) and stained with polyclonal rabbit anti-human GFP antibody at a dilution of 1:100 and counterstained with biotinylated goat anti-rabbit antibody (Vector

Laboratories, Burlingame, CA) at a dilution of 1:200. then analyzed subjectively for cellular content, integration and tissue density.

Statistics. There was no significant difference among horses, among samples or among technical replicates (P>0.05).

Numeric continuous outcomes (CD90 expression (%); cell counts, % viability, HA, PG and BMP-2 concentrations) were analyzed by means of the Shapiro-Wilk method to confirm normality of variance followed by a 2-way ANOVA (group and time) and posttest multiple comparisons using Proc Mixed statistical models33.

The ordinal categorical outcomes in vitro chondrocyte morphology, and in vivo gross anatomy, smooth fibrocartilage repair, and subchondral bone repair were examined using

33 SAS Institute Inc, Cary, NC, USA 51 Genmod statistical models. Repeated variables were considered nested with animal as a random factor.

The in vivo cartilage histomorphometric data (lesion width, depth, and percentages) were examined for normality of variance (Shapiro-Wilks test) and the data were examined using linear regression within multi-level mixed-effects models with treatment groups as a fixed factor, controlling for rat, leg and slide replicate within the model.

High intra-observer (Pearson r = 0.92–0.99) and inter-observer (intra-class correlation,

0.94–0.99) have been reported demonstrating reliability of our elementary and complex histologic scoring systems for articular cartilage repair [85,86]. The intra-observer correlation was high (Pearson r =0.91-0.99). Power calculations were made using means, standard deviations, sample sizes and 5% alpha-error level of all outcomes that differed significantly by treatment groups.

Significance level was set at p < 0.05 for all analyses. Data were expressed as mean ±

SEM.

3.4 Results

For all the studies, sMSCs prior to seeding had a cell count of 1.0 x106 cells per scaffold, a cell viability as measured by 7AAD > 90%, a CD90 expression > 70%, and, for sMSC-

BMP2 a soluble BMP-2 mean of 1.04 x 10-5 ng/cell. For the in vivo studies, the transduced

GFP-sConstructs were used as an additional control for transduced BMP-2-sConstructs.

The GFP- and BMP-2-MSCs had transduction efficiencies > 58% at 100 MOI. In all the samples, there was no significant difference among the different horses.

In Vitro Studies sConstructs impact on Chondrocytes. Co-cultured chondrocyte cell counts were affected by sConstructs (Figure 10A). At day 3, chondrocyte cell counts were inhibited by BMP-

2-sConstructs > untransduced-sConstructs > sECM. This was reversed by day 14 when chondrocytes co-cultured with BMP-2-sConstructs had a 1.9-fold increase in counts over chondrocytes alone. sECMs and untransduced-sConstructs enhanced chondrocyte counts

1.4-fold and 1.5-fold fold, respectively, with no significant difference. Co-cultured chondrocyte cell viability, for all samples remained above 83 % and was not significantly affected by co-culturing with sConstructs. Chondrocyte intracellular Col II concentration mirrored the cell count measurements (Figure 10B). After 14 days, Col II increased 2.9-,

1.8-, and 1.6-fold in chondrocyte co-cultures with BMP-2-sConstructs, untransduced- sConstructs, and sECMs, respectively.

53 Chondrocytes changed from elongated shaped cells, with relatively low cytoplasm-to- nucleus ratio to a chondrocyte phenotype; larger polyhedral-shaped cells with a high cytoplasm-to-nucleus ratio (Figure 11F). The chondrogenesis was most dramatic in the

BMP-2-sConstructs (Figure 11E and F). In the BMP-2-sConstruct co-cultures, already at day 3 had significant numbers of chondrocyte phenotype cells and by day 7 all cells had changed. Chondrocyte change as measured by morphology scores (Figure 11G) followed the same pattern as chondrocyte cell counts, BMP-2-sConstructs > untransduced- sConstructs > sECMs) reflecting the influence of the bioactive biologic preparations in the insert had on the chondrocytes.

Chondrocyte impact on sConstructs. Presence of chondrocytes in the co-cultures significantly increased the sMSC counts and % sMSC CD90 positive cells than sMSC in groups cultured without the chondrocytes. (p<0.001; Figure 12) This reflects an influence of chondrocytes on the sMSC proliferation and maturation. Specifically, untransduced- sConstructs and transduced-sConstructs (Figure 12), rebounded by day 14 to a 2.5-fold increase in untransduced-sConstructs and 2.9-fold increase in BMP-2-sConstructs when cultured with chondrocytes, significantly greater than without chondrocytes. Cell viability was not influenced by presence or absence of chondrocytes (Figure 12B)

Presence of chondrocytes in the co-cultures also significantly increased the levels of soluble biomediators in media (HA, PG, BMP-2) from sECM and sConstructs compared to sECMs and sConstructs cultured without chondrocytes. (p<0.01; Figure 13) This confirmed the feedback influence that the chondrocytes also had on the biologic preparation in the insert, resulting in greater expression of synovial phenotype bioactive 54 mediators. After 14 days, it is clear that BMP-2-sConstructs > untransduced-Constructs > sECM in production of HA (p < 0.01), PG (p < 0.05), and BMP-2 (p < 0.001).

Chondrocytes alone had a residual level of PG concentration (Figure 13B) and the significant increase in PG by the sECM and sConstructs with chondrocytes is of the same magnitude as produced by chondrocytes alone. However, for HA, the increase is much greater than the low level produced by chondrocytes alone, confirming release of HA by sECM and higher production in sConstructs, enhanced by the presence of chondrocytes.

Also, noteworthy, sECMs released a low and sustained level of endogenous HA, PG, and

BMP-2 during the 14 days of incubation, but significantly lower than the sConstruct-

BMP-2.

In all groups, with and without chondrocyte co-culture, the sMSCs followed a growth pattern (Figure12) that is characteristic of sMSCs infiltrating sECMs [16]. By day 3, cell counts decreased 77% for untransduced-sConstructs and 48% for transduced-sConstructs

(Figure 12A), then rebounded by day 14, with both the untransduced- and transduced- sConstructs in co-culture significantly higher (p<0.01). sMSC viability as measured by

7AAD (Figure 12B) showed a less pronounced but similar pattern that returned to day 14 baseline values on Day 0 of > 80%. Cell maturity as measured by CD90 expression decreased from > 70% to < 30% in day 7, then rebounded slightly by day 14 (Figure

12C).

55 BMP-2-sConstruct sMSCs cell counts and cell maturity (CD90 expression) was significantly higher than untranduced-sConstructs after 14 days incubation (Figure 12A and C).

In Vivo Studies sConstruct implant impact on rat knee lesions. Post-operative assessments were performed daily and showed no pain, chewing, inflammation, redness nor leaky wounds in the surgical area. After a 5-week healing period there were dramatic differences between lesions with sConstructs, and those with suture or sECM implants. A gross visual review of the lesion showed that both the GFP- and BMP-2-sConstruct implants produced smoother edges to the lesion, a smooth fibrocartilage that filled the lesion, less adjacent cartilage degradation and few osteophytes. Photomicrographs of lesions confirmed the gross visual review (Figure 14). Lesions with BMP-2-sConstruct implants

(Figure 14A) were > 50% filled with smooth fibrocartilage repair tissue integrated with subchondral bone. Adjacent articular cartilage repair showed anabolic characteristics including chondrocyte/lacunae duplets, strong histochemical staining of PG and maintenance of cartilage thickness. Control knees (Figure 14B) with suture or sECM implants had almost no fibrocartilage filling, and an irregular subchondral bone plate without cartilage integration. The adjacent cartilage was thinner with less intense histochemical staining and there was a loss of cellularity.

In BMP-2-sConstruct implants (Figure 14C), subchondral bone was re-established, calcified chondrocytes were evident across most of the defect, the tidemark was re-

56 establishing, and the surface of the cartilage lesion and adjacent cartilage was smoother.

Control knees (Figure 14D) with only a suture had repair tissue that extended into the marrow with an irregular, less defined, tidemark area and obvious marrow changes.

The sConstructs demonstrated a clear pattern of repair tissue growth enhancement when the gross anatomy, adjacent articular cartilage growth and subchondral bone repair was scored; BMP-2-sConstructs > GFP-sConstructs > sECMs (Figure 15). The gross anatomy and subchondral bone repair scores (compare Figures 15A to 15C) showed a significant progression in all the samples (p < 0.01 for all implants).

To determine the amount of filling from lesion measurements (Figure 9), relative ratios were calculated and graphed (Figure 15D-G); filling from bottom (1-c/e), edge (1-d/e), sides inwards, (1-a/2.5 mm), and peripheral cartilage repair (1-a/b). Lesion implants with

BMP-2-sConstructs showed the most filling (p <0.01) followed by GFP-sConstructs. sECM implants had some enhanced growth at the edges (p <0.01) as well as in the peripheral cartilage repair (p < 0.01). sECMs and GFP-sConstructs implants had no significant difference in the adjacent articular cartilage scores (Figure 15B) or edge filling (Figure 15E) (p >0.05).

Implants in rat synovium. Implants harvested from the rat knee synovium are shown in

Figure 16. Suture only specimens had normal synovium with synovial cells layered over a loose adventia with distinctive suture tracks (Figure 16A). The sECM implantation site

(Figure 16B) had a denser area of tissue, identified as the implant, with cells dispersed throughout a strongly eosinophilic scaffold. The GFP- and BMP-2-sConstruct implants

57 (Figures 16C and D) showed a patch of firm tissue with less eosinophilic tissue area including focal areas of cell congregation with a significantly greater number of cells.

The GFP-sConstruct implant (Figure 16C insert) had many GFP intensely stained cells

(brown stained cells). When other implants were stained with GFP immunochemistry there was a mild background coloring of some cells and matrix but no intensely stained cells. BMP-2-sConstructs (Figure 16D insert) had the greatest cell density that had permeated the implant reducing collagen density.

3.5 Discussion

This study conclusively shows that sConstructs enhance chondrogenesis in vitro and cartilage healing in vivo. It is a “proof of concept” that sMSC synovium scaffold

Construct has the potential to be a vehicle creating the correct microenvironment for the repair of damaged joints to treat cartilage injury and OA.

The results strongly support a signaling loop mechanism between sConstructs and chondrocytes, both in vitro and in vivo. The sConstruct’s anabolic biofactors amplify a paracrine, restoration and maturation response of chondrocytes that, in turn, impacts the sConstruct’s own maturation and biofactor output. In vitro, sConstruct-chondrocyte co- cultures showed increased chondrocyte and sMSC cell counts, increased intracellular chondrocyte Collagen II and increased production of soluble HA and BMP-2, respectively. In vivo, rat knee lesions with sConstruct implants, showed markedly improved articular cartilage growth and subchondral bone repair. Correspondingly, resected sConstruct implants had substantial morphological changes with a marked increased cell integration.

It was striking how much the increased BMP-2 concentrations reinforced the feedback loop. BMP-2-sConstruct co-cultures boosted the chondrocyte maturation and intracellular

Col II production. After 14 days of BMP-2-sConstruct co-culture, chondrocyte intracellular Col II levels were > 2-fold that of any other samples. After just 3 days of incubation, co-cultured chondrocytes were reverting to one typical of a hypertrophied cell 59 or chondroblast. Correspondingly, BMP-2-sConstruct’s MSCs had higher cell counts and produced more HA than untransduced-sConstructs. The results also show that BMP-2 delivered an anabolic boost to sMSCs proliferation and differentiation that overcame the initial loss of viability, presumably due to cytotoxic Ad vector transduction [22] and from the seeding procedure [16]. The production of BMP-2 by BMP-2-sConstructs in co- cultures was marked, with soluble BMP-2 concentrations higher than those considered to be in the bioactive range (> 170-180 ng/mL) [60,61].

The strong effect of BMP-2 was also observed in vivo. BMP-2-sConstruct implants considerably improved gross anatomy, adjacent articular cartilage growth and subchondral bone repair over that of other implants. The morphological changes of the implants were by far the most pronounced in the BMP-2-sConstruct.

The BMP-2 effects observed here are in line with previous studies. BMPs have been shown to be involved in stimulating MSC differentiation and cell recruitment, along with other functions in chondrogenesis and osteogenesis [87,88]. There was no sign of any deleterious effects of BMP-2 on cartilage [53], no bone growth or overgrowth into the synovium with implanted BMP-2-sConstructs. sECMs seem to contain endogenous BMP-2. sECMs contained no MSCs, nor did chondrocytes alone produce BMP-2, yet significant amounts of soluble BMP-2 was found in sECMs cultured with chondrocytes or alone. It is known that ECMs can be a reservoir of biomediators [89] and it appears that the method of sECM preparation [26] does not eliminate reserves of BMP-2. BMP-2 concentrations were lower in sECM-

60 chondrocyte co-cultures than in sECMs alone. A plausible explanation is the ability of chondrocytes to take up BMP-2 [90], thus lowering the co-culture concentration of the growth factor as compared with sECM alone.

The sECM implants, in vivo, caused a small but significant enhancement of lesion fill as well as a migration of endogenous cells into the sECM scaffold. This may have been caused by increased local BMP-2 concentration producing a growth environment for repair of the lesion. However, ECMs could have other endogenous growth biofactors as well as structural properties [91] that could have activated 1) changes in local chondrocytes or 2) the migration of growth inducing endogenous cells that contributed to the enhanced lesion repair.

HA acts as a boundary lubricant and an anti-inflammatory substance that inhibits the adherence of immune complexes to neutrophils and protects the synovial tissues from the attachment of inflammatory mediators [62]. sConstruct’s increased levels of HA production in co-cultures with chondrocytes, suggests they should have a positive effect on joints healing from cartilage damage. PG concentrations significantly increased for sConstructs alone [16] or as shown here in chondrocyte co-cultures. Chondrocytes alone produced PG, that could account for the observed increase in other groups.

The in vitro and in vivo results correlated remarkably well. Both showed that BMP-2- sConstruct gave the highest chondrogenesis and healing potential. Importantly, the in vitro and in vivo results had the same trend; BMP-2-sConstructs > GFP-sConstructs or untransduced-sConstructs > sECM. This study provided both in vitro and in vivo testing

61 methods for future studies. While BMP-2 was used in this study as an example of how transduced sMSCs may impact chondrogenesis, other growth factors alone or in combination enhance chondrogenesis of sMSCs [21,92] and should be tested to improve the efficacy of sConstructs.

In this study, 1x106 sMSCs were used for seeding the sConstructs in the inserts, while

1x103 chondrocytes were cultured in the wells. The sMSC cell amounts were chosen because previous studies [16] had shown a cellular distribution pattern that had infiltrated the sECMs after 7 days using the 30% FBS gradient to augment the seeding process.

Chondrocytes, at 1x103 cells, made a confluent monolayer of morphologically transformed cells at day 7, when co-cultured with sConstructs. The surface area for cell attachment was anticipated to be greater in the complex matrix than the flat surface of the well, corresponding to the difference in seeding numbers selected. Thus, the conditions in this study gave extremely pronounced results, good for an in vitro test model. Other cell concentrations could be tested in the future to optimize this model.

Limitations of this study included the relatively short duration of in vitro incubation times. The in vitro 14 day was chosen because of the chondrocyte cell growth patterns.

Within 7 days of co-culture, all the chondrocytes had matured, and the wells were fully covered with cells. We selected a time anticipated to show both delay or acceleration of chondrogenesis. By 14 days the sECMs were well seeded with viable cells. We also anticipated a less than two-week time period from culture to implantation at surgery, in either our rodents or in human patients in the future. We focused our efforts on this time window. Longer term studies would be needed to study toxicity or shelf-life of the 62 sConstructs. Another limitation was the short in vivo incubation time. Although we selected the time point of healing of 5 weeks to catch the window to detect either a delay or acceleration of repair among our comparison groups, longer term studies would be needed to demonstration the durability of the repair cartilage and the development of OA in these knees. Current studies on OA treatments using large animal models are expected to have 6 months, or more, of observation in this final stage in vivo models. By restricting the treatment time, we were not able to see the overall healing effect of the sConstruct.

The 5-week treatment period was also selected to allow adequate time for measurable cartilage healing, without complete degradation of the sConstruct implanted in the synovium. We desired to study the sMSC as well as the cartilage healing. The MSCs prepared from synovium may have contained several types of MSCs [19]. While this may be considered a limitation, for this proof of concept study, a wide variety of MSC cell types might have led to the positive results. Further work to optimize the conditions for sConstruct implants, such as type of MSC, the seeding period prior to implantation, cell seeding number, and scaffold size all needs needs further study.

63 implanted as xenogeneic grafts [32,33]. The combination of sECM scaffold and sMSC’s in the sConstruct have great potential as a hypoimmunogenic product that could be used allogenically. Preliminary results from our laboratory indicate there is a very minor immune response when sConstructs were co-cultured with peripheral blood mono nuclear cells (PBMCs).

To further study the immune component of syngeneic verses allogenic sConstructs,

MSCs should be thoroughly tested in immunocompetent models. Understanding the mechanisms of scaffold-cell interactions could refine the treatment process by balancing anabolic and anti-inflammatory pathways. Proteomic data on the sECM will direct future work on the mechanism of implantation effects on cartilage.

In conclusion, the synergistic potential of sConstructs on distant articular cartilage has been clearly demonstrated. To evaluate modified sConstructs, this work has developed a fast in vitro method that correlates well with in vivo results. It is a “proof of concept” that sConstructs could be used as a possible treatment for OA in the future

64 3.6 Tables

Table 1. Preparation prior to in vitro co-culture experiment start Day 0

Day < -30 days -8 days -6 days -5 days -3 days Day 0 sMSC sConstruct Chondrocyte Experiment Cells/Scaffolds Biologic Preparation transduction preparation preparation start sECM inserts sECM decellularized sECM discs thawed at followed alone from fresh synovium, room temperature for 3 and placed into sECMs cut into 6 mm  discs hrs and placed in the 6 plate wells and frozen (-80C). mm  insert. containing chondrocytes ELISA for BMP2 in the Media sMSCs prepared from sMSC 106 viable cells in 10% from sMSC-BMP2. confirmed sMSCs fresh synovium and transduced with FBS were placed on transduction and BMP2 greater passaged. Ad-BMP2. sECM in insert. than sMSC. Chondrocytes thawed, 103 cells in 10% FBS Chondrocytes Chondrocytes isolated were placed into each followed alone from digestion of fresh well of the 12 well plate Chondrocytes and in the plate cartilage and frozen at - for the experiments wells of co- 80C. assigned to co-culture. cultures Medium was changed daily. Inserts were placed into sConstruct inserts 12 well plates containing were placed into 40% FBS. Medium sConstruct plate wells replaced daily, 10% FBS containing in insert, 40% FBS in chondrocytes. well.

Table 2. Assays performed on 3,7, and 14 days

Characteristic Assay Type Sample Preparation sConstructs* Flow cytometry sConstructs removed from inserts and digested with collagenase. Labelled sMSC cell count, viability & maturation Microscopy 7AAD and CD90 for flow and trypan blue vital stain for microscopic count. (hemocytometer)

Medium**

Every 24 hrs, medium from inserts and wells was collected Soluble HA, PG, & BMP-2 conc. ELISA and pooled for three time points; 0-3, 4-7, 8-14-day samples.

Chondrocytes*** Flow cytometry Removed medium. Cells in wells were lifted with trypsin. Labelled 7-AAD Cell count & viability Microscopy for flow and trypan blue vital exclusion stain for microscopic count. (hemocytometer)

Coll II concentration ELISA Removed medium. Cells in wells were lysed and effluent assayed.

Removed medium, fixed cells in well, and toluidine blue stained. Cell Cell morphology Microphotometry monolayers were photographed under controlled conditions.

* Number of samples per characteristic = 3 horses x 3 samples/horse x 3 replicates x 3 co-culture groups ( sECMs, untransduced-sConstructs, BMP-2- sConstructs) x 3 time periods= 3 x 3 x 3 x 3 x 3 = 243 samples. ** Number of samples per characteristic (HA, PG, BMP-2) = same as sConstructs except 4 co-culture groups (additionally media from chondrocytes alone) = 3 x 3 x 3 x 4 x 3 = 324 samples *** Number of samples per characteristic = 3 horses x 3 samples/horse x 3 replicates x 4 co-culture groups (chondrocytes alone, sECMs, untransduced- sConstructs, BMP-2-sConstructs) x 3 time periods= 3 x 3 x 3 x 4 x 3= 324 samples.

66 Table 3. Scoring criteria for chondrocyte morphology, lesion gross anatomy, adjacent articular cartilage growth and subchondral bone repair

Gross Adjacent Articular Score Chondrocyte Morphology Subchondral Bone Repair Anatomy Cartilage Growth  > 75% of cells with chondral morphology typical of a  No damage 4 hypertrophied cell or chondroblast  No abnormalities (cellular enlargement and rounded  Reestablished tidemark  Normal shape with greater cytoplasm-to-  Reestablished subchondral bone  Intact morphology nucleus ratio) plate cartilage  Normal cellularity  < 25% of cells with early chondral  Calcified zone of cartilage surface  Intense toluidine morphology (elongated, tapered, reformed blue stain spindle- or stellate- shaped cells,  No fragmentation 3.5  Normal thickness with relatively low cytoplasm-to-  No bone marrow changes nucleus ratio; confluence often leads to mounding of cells)  Normal cellularity  Surface  >75% cartilage 3  No marrow changes or, if  50 to 75% of cells with chondral roughening thickness present, minimal and focal morphology typical of a  No presence  Intense toluidine  Increased thickening of hypertrophied cell or chondroblast of cartilage blue stain subchondral bone subjacent to  25 to 50 % of cells with early lesions or  Lacunae doublets the area of greatest articular 2.5 chondral morphology osteophyte present cartilage lesion severity. formation  No chondrone formation  Moderate  Minimal to mild focal  Deeper 2 cellularity fragmentation of calcified defects  50-75% cartilage cartilage of tidemark, (fibrillation,  25 to 50% of cells with chondral thickness  Mesenchymal change in fissures) not morphology typical of a  Moderate Intensity marrow (fibroblastic cells) involving the hypertrophied cell or chondroblast of toluidine blue involving about 1/4 of subchondral  50 to 75% of cells with early stain subchondral region under lesion 1.5 bone chondral morphology  No lacunae  Increased thickening of  Mild doublets subchondral bone subjacent to osteophyte the area of greatest articular formation  No chondrone formation cartilage lesion severity  Low cellularity  Mild to marked fragmentation  Erosions  25-50% cartilage (multiple larger areas) of 1 down to thickness calcified cartilage bone loss  < 25% of cells with chondral subchondral  Low intensity of  Mesenchymal change in morphology typical of a bone (less toluidine blue marrow up to 3/4 of total area, hypertrophied cell or chondroblast than 5 mm stain  Areas of marrow  25% to 50% of cells with early diameter)  No lacunae chondrogenesis may be evident 0.5 chondral morphology  Presence of doublets  No major collapse of articular multiple cartilage into epiphyseal bone osteophytes  Chondrone formation (definite depression in surface).  No cellularity  Large  Low toluidine blue  Marked to severe fragmentation erosions stain of calcified cartilage  No chondral morphology down to  No lacunae  Mesenchymal change in 0  < 25% of cells with early chondral subchondral doublets marrow > 3/4 of area morphology bone  Chondrone  Articular cartilage has collapsed  Osteophyte formation into the epiphysis to a depth of formation  <25% cartilage 250 mm or less from tidemark thickness 67 3.7 Figures

Figure 9. Histomorphometric measurements of lesion filling Five measurements were taken for lesion filling a) width of opening halfway between the lesion bottom and the projected original surface b) width, at the same depth as measurement a, of opening of physiological cartilage, c) depth of the opening at the center, d) depth at 100µm from edge of original lesion, and e) depth of original lesion, from the projected cartilage surface to the estimated original tidemark. Red dashed line represents the size of the original lesion (2.5 mm wide x ~ 1.5 mm deep).

Figure 10. Co-culture chondrocyte cell counts and intracellular Col II production Cell counts (A) and Col II concentration (B) from chondrocytes cultured alone (white), with sECM (green), untransduced-sConstructs (blue) or BMP-2-sConstructs (red) on day 3, 7, and 14.

Figure 11. Co-cultured chondrocyte monolayers and morphology scores Chondrocytes co-cultured with sECM or BMP-2-sConstructs. Selected photomicrographs, 20X, (A-F) and scores (G) from chondrocytes cultured alone (white), with sECM (green), untransduced-sConstructs (blue) or BMP-2-sConstructs (red) on day 3 and 7.

Figure 12. Co-culture sConstruct sMSC soluble biomarker concentrations HA (A), PG (B) and BMP-2 (C) from chondrocytes alone (white), sECM alone (green hatched), sECM chondrocyte co-culture (green), untransduced-sConstructs alone (blue hatched), untransduced-sConstructs chondrocyte co-cultures (blue) and BMP-2- sConstructs alone (red hatched), and BMP-2-sConstructs chondrocyte co-cultures (red) from media collected 1-3, 4-7, and 8-14 days.

Figure 13. Co-culture sConstruct sMSC count, viability, and maturity Total cell count (A), % 7AAD (B), and % CD90 expression (C) with untransduced- sConstructs alone (blue hatched), untransduced-sConstructs in co-culture (blue), BMP-2- sConstructs alone (red hatched) and BMP-2-sConstructs in co-culture (red) after incubation for 3, 7, and 14 days. 72

Figure 14. Lesion with sConstructs implants; smooth fibrocartilage repair tissue, adjacent articular cartilage and subchondral bone repair Illustrative (median representative) photomicrographs (40x) of histochemical toluidine blue stained sections of rat knee lesions after 5 weeks of healing. Specimens selected for the photograph had the middle histology score for that group and thereby is a mid- representative of the differences. A) Lesions with BMP-2-sConstruct implants where the white arrow ( ) points to smooth fibrocartilage repair tissue, the black arrow () to chondrocyte/lacunae duplets, and the red dot (), the histochemical staining for PG. B) Image as Figure 14A but from a control knee with suture only. C) Lesions with BMP-2- sConstruct implants where black arrow () points to re-established subchondral bone and dotted line (- - - - -) is re-established tidemark, D) Similar image to Figure 14C from a control knee with suture only. Scale bar: 200 m.

Figure 15. Lesion with sConstruct implants; growth scores and filling measurements Gross anatomy (A), articular cartilage (B), and subchondral bone repair scores (C) and bottom (D), edge (E), sides inwards (F) and peripheral cartilage (G) measurement ratios. Lesion with no implant (white), with sECM (green), GFP-sConstructs (blue) or BMP-2- sConstructs (red).

Figure 16. sConstructs recovered from implant site Histophotomicrographs (2X) of implants recovered from rat knee synovium. A) suture only with suture tracks (), B) sECM, C) GFP-sConstruct and D) BMP-2-sConstruct. Inserts are 20X magnification and C) was stained immunohistochemically for the presence of GFP.

CHAPTER 4: IMMUNE AND SIGNALING PROTEINS OF ALLOGENEIC

STEM CELL-EXTRACELLULAR MATRIX SCAFFOLD INTERACTIONS

Submitted ahead of dissertation: Reisbig N, and Bertone AL. (2018) Immune and Signaling Proteins of Allogeneic Stem Cell-Extracellular Matrix Scaffold Interactions Submitted, PLoS ONE, 2018.

4.1 Abstract

Osteoarthritis is a progressive disease associated with cartilage injury and is the most common form of arthritis, affecting millions of people worldwide. We propose that allogeneic synovium-based cellular constructs (sConstructs) would have potential as a tissue-engineered implant biotherapy provided the viable implants had limited immune and inflammatory response from the host and could provide synergy to generate chondrogenic proteins. Equine sConstructs were produced from decellularized synovium- based extracellular matrix scaffolds (sECM), seeded with synovium-derived mesenchymal stem cells (sMSC) and co-cultured with, or without, allogeneic peripheral blood mononuclear cells (PBMCs). Concentrations of sMSC surface markers (CD44,

CD45, CD90, MHCI and MHCII) and PBMCs (CD11b, MHCI and MHCII) in addition to biomarkers, IL-1, IL-1ra, IL-6, IL-10, TGF-beta, TNF-alpha and IFN-gamma, were measured. The proteomic structure of sConstructs, sECM and sMSCs was also analyzed.

Allogeneic PBMCs had minimal inflammatory or immune response, seen primarily to the sECM. The presence of the sMSC in the sECM (sConstruct) abrogated this response.

Correspondingly, PBMC with sConstructs increased the production of anti-inflammatory cytokines (IL-1ra and TGF-beta). This indicated that sConstructs were less inflammatory than either component (sECM or sMSC) alone and was associated with proteome changes as the components integrated. Proteomic data also concurred that few changes occurred to sConstructs co-cultured with PBMCs. SECM or sMSC proteome differed

77 from each other and sConstructs differed from each of these components. Proteins that differed, stratified to functional (Gene Ontology) characteristics of matrix (structural proteins) or cells (metabolic proteins). However, proteins unique to the sConstruct were both anabolic, early matrix, and cell proliferation proteins. The combination of lower inflammatory profiles, enhanced anti-inflammatory profiles, immunotolerance by allogeneic PBMCs, and a protein enriched integrating tissue allo-implant, supported the potential for this implant to support nearby articular cartilage.

4.2 Introduction

The use of decellularized extracellular matrices (ECMs) as a scaffold have become increasingly popular as a regenerative treatment [91,93,94]. The ECMs can be either implanted into a tissue defect alone, or re-seeded with appropriate cells before implantation [95,96]. ECMs have become more popular than synthetic scaffolds due to their structural and biological potentials [91,97,98]. Although decellularized, the ECMs retain the porosity of the native tissue, the collagen components, and many growth factors, all of which promote improved cell seeding and survival which can drive the implanted cells toward differentiation [57,91].

Mesenchymal stem cells (MSC) are non-hematopoietic stromal cells that are capable of differentiating into, and contributing to, the regeneration of tissues such as bone, cartilage, muscle, ligament, tendon, and fat [99]. They maintain their ability to multiply in culture, and retain their growth and multi-lineage potential, differentiating into many cell lines, such as osteocytes, chondrocytes and adipocytes [99,100]. MSCs are identified by the expression of many molecular markers including CD105 (SH2) and CD73 (SH3/4) and are negative for the hematopoietic markers CD34, CD45, and CD14[100]. The properties of MSCs make these cells potentially ideal candidates for tissue engineering.

The most common sources of MSC’s are bone marrow, adipose tissues, and recently fibroblast-like cells, isolated from the peripheral blood of multiple species, including human and rat. They have been demonstrated to possess the capacity to differentiate into 79 several mesenchymal lineages [101,102]. The response when implanted into different tissues is dependent on the MSCs origin [103]. However, characterization of the responses between site-specific and systemic MSCs has yet to be elucidated. In the joint, synovial MSCs (sMSC) are site specific and a natural component of the joint. Although the sMSC are not pluripotent they are defined as a multipotent mesenchymal stem cell

[104] from which differentiation into osteoblastic, chondrogenic, and adipogenic differentiation has been reported [105–107].

MSC have been injected, unbound, into joints [8,9,108], tendons and ligaments [109–

111], intra-peritoneally [112,113] and intra-venously. MSCs have been shown to migrate to sites of injury in animals, when transplanted systemically [114], However the magnitude of cell loss to migration or leakage out of the third compartment space is unknown [115–117]. Here, we propose the strategy to place a seeded scaffold directly adjacent to a defect to retain cells. This ensures proximity to the site of injury, and may provide high, local concentrations of the cells and their products.

We have shown in our prior work that retaining the MSCs adjacent to the damaged area in a synovium construct enhanced cartilage healing in athymic rat knee lesions [27]. The bio-construct was made from decellularized synovium extracellular matrix (sECM) [16] seeded with synovium MSCs (sMSCs) [16] producing a bioactive scaffold, the sConstruct. The results demonstrated that not only was the sConstruct anabolic to healing cartilage, but that there was a paracrine effect resulting in greater growth of the sConstruct. Additionally, genetically modified sMSCs, containing the growth factor bone morphogenic protein (BMP-2) showed a further enhancement in chondrogenesis and 80 stable cartilage regeneration in this athymic rat knee model. The bio-construct provided the appropriate growth environment throughout the healing period.

Our data showing improvement in cartilage repair in immunocompromised rats did not measure the immunologic and inflammatory response due to the allogeneic source of biomaterials. The ultimate goal of this work was to choose sConstructs of frozen and thawed components (sMSCs and sECM low in DNA) for immunotolerance and potential for cell seeding a few days before potential implantation during surgery to address the cartilage injury. MSCs are known to inhibit maturation and antigen-presenting capacity of dendritic cells [118,119] and decrease proliferation and cytotoxicity of natural killer cells [120,121]. MSCs also suppress the adaptive immune response, by dampening both

CD4+ helper and CD8+ cytotoxic T cell proliferation and exertion of their respective functions [122]. Appropriately prepared (DNA concentration and base pair length) ECM scaffolds, can be implanted as xenogenic grafts [32][33] with little immune response. The combination of sECM scaffold and sMSC’s have the potential as hypoimmunogenic products that could be used allogenically. The process of freezing and thawing has also been used to dampen immune response to tissues [123].

In this study, we investigated the sConstruct potential for immunologic response, using a co-culture model containing the sConstruct, the individual components (sECM, sMSC) and allogeneic peripheral blood mononuclear cells (PBMCs). The PBMC response is determined by quantifying the anti- or pro-inflammatory factors, Interleukin 1 (IL-1),

Interleukin 6 (IL-6), tumor necrosis factor- (TNF-) or interferon-γ (IFN-) and alterations of surface markers. MSCs broadly suppress T-cell activation and proliferation 81 in vitro via a plethora of soluble and cell contact-dependent mediators. These mediators may act directly upon T cells or indirectly via modulation of antigen-presenting cells and other accessory cells [124].

The paracrine effects we observed with sConstructs and chondrocytes in co-culture [27] may also occur between sConstructs and immune cells in co-culture and impact the efficacy of cartilage growth enhancement. By comparing the proteomes of sConstructs with samples co-cultured with PBMCs, any paracrine effects by the immunologic cells could be identified and measured. We used proteomics in our current study, to quantify protein changes induced in the sConstructs during an immune challenge. The proteome information, obtained here, will not only provide an understanding of the mechanisms for increased cartilage healing capabilities, but also serve as the basis for designing future sConstructs with potentially lower immune response [125,126].

The specific objectives of this study were to use an in-vitro co-culture model to determine

1) the allogeneic PBMC immunologic response to sConstructs and components, 2) the impact of allogeneic PBMC on sConstruct and components, and 3) proteome changes in the sConstructs by these interactions of cells (sMSC, PBMC) and sECM. Our hypothesis was that sConstructs would induce a low but measurable immune/inflammatory response that would be abated by the presence of sMSCs, and correspondingly, there would be an insignificant paracrine effect on the sConstruct by the presence of allogeneic PBMC.

4.3 Materials and Methods

Co-culture experimental designs. MSC, sECMs and sConstructs were from 3 different horses. Two groups of samples were prepared 1) control cultures and 2) immune response co-cultures. The controls included 4 groups: sMSCs, sECM, sConstructs and PBMCs cultured alone. The immune response had 4 groups of co-cultures: sMSCs-PBMC, sECM-PBMC, sConstructs-PBMC, and PBMCs cultured in lipopolysaccharide (LPS).

Each of the eight groups of 3 horses was performed in triplicate for a total of 72 samples.

After 48 hours incubation, PBMC anti- and pro-inflammatory soluble factors were measured by ELISA and sMSC and PBMC surface markers were measured using flow cytometry. sMSC, sECM and sConstruct preparation. Equine villous synovium was harvested from knee joints of healthy adults (2-7 years old), euthanized for reasons unrelated to this study, and processed to obtain isolated synovial mesenchymal stem cells (sMSC) and synovium-based extra cellular matrix (sECM) as previously published [26].

Synovium was divided into small pieces under sterile conditions and digested in 0.02% collagenase II solution (37°C, 5 hours), filtered (70μm cell strainer), and collected cells were passaged at > 90% confluence. Passage 3 to 6 was chosen for seeding as this ensures that >95% of the cells will express a multipotent mesenchymal stem cell phenotype with minimal presence of lymphocytes, natural killer cells and macrophages

[58]. 83 Synovium to be used as scaffolds was divided under a dissecting microscope into 1 ×

1cm sheets. Decellularization was performed by incubation at 37°C in 0.1% peracetic acid (PAA) (Sigma Aldrich, Steinheim, Germany) and 4% ethanol for 6 hours with mechanical agitation and then repeated to produce a synovium origin extracellular matrix

(sECM) [26]. SECM samples were frozen at –80°C surrounded by dry ice in ethanol and used within 1 month. sConstructs were prepared by thawing 6 mm diameter sECMs and placing them in 6-mm inserts (Corning Costar Transwell, Sigma Aldrich, Steinheim, Germany) designed to fit into standard 12-well culture plates. SECM alone and sECM with sMSCs were all cultured in supplemented DMEM with a 30% fetal bovine serum (FBS, Sigma Aldrich,

Steinheim, Germany) gradient by a 10% FBS in the insert medium and 40% FBS in the well at 37°C in DMEM also supplemented with penicillin (100 U/mL), streptomycin (100

μg/mL), and amphotericin (250 ng/mL). SECM (6 mm diameter) were seeded with 1 x106 sMSCs for 5 days before in vitro co-culture with PBMCs. The seeding was allogeneic such that no sECMs from one animal was seeded with the same animals sMSCs.

PBMCs isolation and co-cultures. PBMCs were isolated by Ficoll density gradient

(Ficoll- PaqueTM Premium, GE Healthcare, Piscataway, NJ) from 60 mL of heparinized venous blood from 2 horses by following established company protocols (Stem Cell

Technologies, Vancouver, BC). PBMC from each horse was randomly assigned to all 3 of the sMSC, sECM and sConstruct samples from one of the three synovium donating horses. The samples were tracked during the experiment and there was no significant 84 difference observed due to the different source of PBMCs. The PBMCs were applied to each sMSC well at the start of the in vitro study in a 8:1 ratio (8 × 106 PBMC added onto

1 x 106) construct.

PBMC surface marker determination. The activation of a cell-mediated immune response in the PBMC was evaluated 48h after co-culture start by flow cytometry for the number of 7AAD% (viability), CD4+, CD8+, MHCI, MHCII and CD11b positive

PBMCs. Monoclonal mouse anti-equine CD4 antibody (Clone CVS4, Thermo Fischer

Scientific Inc., Rockford, IL) and secondary FITC rat anti-mouse antibody (Clone A85-1,

BD Biosciences, San Jose, CA), Mouse anti Horse CD8 antibody, (clone CVS21, BD

Biosciences, San Jose, CA) and secondary Human anti Mouse IgG2a (clone AbD06344,

BD Biosciences, San Jose, CA), Mouse anti Horse MHC Class I Monomorphic antibody

(clone CVS22, BD Biosciences, San Jose, CA) and secondary Human anti Mouse IgG2a

(clone AbD06344, BD Biosciences, San Jose, CA), Mouse anti Horse MHC Class II

Monomorphic antibody (clone CVS20, BD Biosciences, San Jose, CA), Mouse anti

Human, Equine CD11b (clone 238446, R&D systems) and secondary with secondary

Human anti Mouse IgG1 antibody (clone AbD04639, BD Biosciences, San Diego, USA) was used and 7AAD were applied to the PBMC samples. Antibody staining was done in a total volume of 200 μl of FACS buffer. Cells were incubated at 4°C for 20 min in darkness, and then washed with 2 ml of FACS buffer by centrifugation. sMSC surface marker determination. The sMSC cluster of differentiation (CD) 90,

CD44 and CD45 and major histocompatibility complex (MHC) I and II were evaluated before seeding, after seeding and in PBMC co-culture with and without seeding. Anti- 85 human CD90 primary monoclonal antibody (BD Biosciences, San Diego, USA), CD44

Mouse anti Horse CD11a/CD18 antibody (clone CVS18, BD Biosciences, San Diego,

USA) with secondary Human anti Mouse IgG1 antibody (clone AbD04639, BD

Biosciences, San Diego, USA) was used, Mouse anti Horse MHC Class I Monomorphic antibody (clone CVS22, BD Biosciences, San Jose, CA) and secondary Human anti

Mouse IgG2a (clone AbD06344, BD Biosciences, San Jose, CA), and Mouse anti Horse

MHC Class II Monomorphic antibody (clone CVS20, BD Biosciences, San Jose, CA) with secondary Human anti Mouse IgG1 antibody (clone AbD04639, BD Biosciences,

San Diego, USA). Antibody staining was done in a total volume of 200 μl of FACS buffer. Cells were incubated at 4°C for 20 min in darkness, and then washed with 2 ml of

FACS buffer by centrifugation.

Pro-, anti-inflammatory and growth factor determination. ELISAs were performed on the supernatant culture media after 48h of co-culture with PBMCs for IL-1, IL-1ra,

TGF- , TNF-, IFN-, IL-6 and IL-10 concentrations using commercially available kits

(R&D Systems, Minneapolis, MN). Lipopolysaccaride (LPS) was used to activate

PBMCs providing a positive control (2g/ml).

Proteomics

Protein extraction and digestion. Proteins were extracted from the samples using RIPA buffer and the extract was reduced, alkylated and digested with trypsin (protein:enzyme

30:1) at 37C for overnight. The following day, acetic acid was added to the sample to

86 quench the reaction. The samples were dried in a vacufuge and resuspended in 20 L of

50 mM acetic acid. Peptide concentration was determined by nanodrop (A280nm).

Orbitrap Fusion. Capillary-liquid chromatography-nanospray tandem mass spectrometry

(Capillary-LC/MS/MS) of protein identification was performed on a Thermo Scientific orbitrap Fusion mass spectrometer equipped with an EASY-Spray™ Sources operated in positive ion mode. Samples were separated on an easy spray nano-column (PepmapTM

RSLC, C18 3µ 100A, 75µm X150mm Thermo Scientific) using a 2D RSLC HPLC system from Thermo Scientific. Each sample was injected into the µ-Precolumn

Cartridge (Thermo Scientific,) and desalted with 0.1% Formic Acid in water for 5 minutes. Mobile phase A was 0.1% Formic Acid in water and acetonitrile (with 0.1% formic acid) was used as mobile phase B. Flow rate was set at 300nL/min. Typically, mobile phase B was increased from 2% to 35% in 80 min and then increased from 35-

55% in 15min and again from 55%-90% in 5 min and then kept at 90% for another 5 min before being brought back quickly to 2% in 1 min. The column was equilibrated at 2% of mobile phase B (or 98% A) for 20 min before the next sample injection. MS/MS data was acquired with a spray voltage of 1.7 KV and a capillary temperature of 275 °C is used.

The scan sequence of the mass spectrometer was based on the preview mode data dependent TopSpeed™ method [127]: the analysis was programmed for a full scan recorded between m/z 400 – 1600 and a MS/MS scan to generate product ion spectra to determine amino acid sequence in consecutive scans starting from the most abundant peaks in the spectrum in the next 3 seconds. To achieve high mass accuracy MS determination, the full scan was performed at FT mode and the resolution was set at

87 120,000. MSn was performed using ion trap mode to ensure the highest signal intensity of MSn spectra using both CID (for 2+ and 3+ charges) and ETD (for 4+-6+ charges) methods. The CID fragmentation energy was set to 35%. Dynamic exclusion is enabled with a repeat count of 1 within 30s and a low mass width and high mass width of 10ppm.

Sequence information from the MS/MS data was processed by converting the .raw files into a merged file (.mgf) using MS convert (ProteoWizard). The resulting mgf files were searched using Mascot Daemon by Matrix Science version 2.3.2 (Boston, MA) and the database searched against Uniprot Horse database (071618). The mass accuracy of the precursor ions was set to 10ppm, accidental pick of 1 13C peaks was also included into the search. The fragment mass tolerance was set to 0.5 Da. Considered variable modifications were oxidation (Met), deamidation (N and Q) and carbamidomethylation

(Cys). Four missed cleavages for the enzyme were permitted. A decoy database was also searched to determine the false discovery rate (FDR) and peptides were filtered according to the FDR. The significance threshold was set at p<0.05 and bold red peptides required for valid peptide identification. Proteins with less than 1% FDR as well as a minimal of two significant peptides detected were considered as valid proteins. sConstruct protein comparisons

Relative abundance of proteins. Using cluster analysis (Gene Cluster 3.0, http://bonsai.hgc.jp/~mdehoon/software/cluster/) the 703 proteins from the LC/MS/MS was further narrowed down by requiring at least two of the three observations with a value of >= 2. The 441 proteins selected by this method was used for the statistical

88 analysis (Further detail on the bioinformatics and statistical program EdgeR is detailed under Statistics). Results from EdgeR are presented as +1 and -1 (increase in former part of the set or increase in latter part of the set, respectively, with a q value < 0.028) and 0

(increase or decrease in any comparison, with a q value > 0.028 and < 0.050). Edge R also provided further evaluation using a heat map

(https://bitbucket.org/TreeView3Dev/treeview3/) and significantly expressed abundant proteins were identified using PANTHER (http://www.pantherdb.org) [128]. Further analysis of the abundance protein sets sECM to sMSC (E/M+/-), sConstruct to sMSC

(C/M+/-), sConstruct to sECM (C/E+/-) and sConstruct-PBMC to sConstruct (CPC+/-) was performed by 1) reviewing all the PANTHER identified proteins with their corresponding Gene Ontology (GO) categories [129,130] as produced by PANTHER, 2) creating a manageable group of “Cell Locality” categories (Table 4) where each protein could be assigned and 3) assigning the proteins to the appropriate cell locality category.

In most cases each protein had several GO categories associated with it. If a protein had several of the categories in the group, it was assigned to that group. Proteins that did not have GO categories were reviewed in the literature and assigned a group according to its cellular location. The proteins in the sets (E/M, C/M, C/E, and CP/C) were organized by category and then Log2-Fold Change (Log (FC)), with the highest Log (FC) value at the beginning of each category (Table 5, 6, and 7). During this analysis several contaminant blood components were found in the identified proteins (Serum Albumin, Histidine-Rich

Glycoprotein, Inter-Alpha-Trypsin Inhibitor Heavy Chain H4, Hemoglobin Subunit Beta,

Immunoglobulin Heavy Constant Gamma 1-Related, Angiotensinogen, Antithrombin-III,

89 Prothrombin and Fibrinogen Gamma Chain). These were not included in the final identified protein sets or their evaluation.

Elevated and decreased proteins observed in sConstructs, not found in sECMs and sMSCs

To determine how the sConstruct proteome differs from the constituent sMSC and sECM proteomes, we determined the proteins that were in the sConstructs yet not found in the corresponding constituents. This was done by determining all proteins in C/M+, C/M-,

C/E+ and C/E- that were not included in E/M-/0/+ set. The union of the sets C/M+, C/M-,

C/E+, C/E- and E/M+0- was determined in a Venn graph [131])

(http://www.interactivenn.net/index.html) and all proteins outside the E/M=/0/- set were was - identified in PANTHER [128]. These proteins were also grouped in cellular location categories (Table 4). During the categorization process, it became obvious that several of the proteins had related functions. A set of function categories (antioxidant, metabolic process, collagen related, cytoskeletal related, ECM modification, glycolysis, iron metabolism, mRNA, protein processing, cellular transport and other) was made. The proteins were categorized to these function categories by a review of the literature.

Statistical analysis. Numerical data (percentage cell viability, MHC1, MHCII and

CD11b expression; IL-1, TNF-, IFN-, IL-1ra, TGF- , IL-6 and IL-10 concentrations) were analyzed by means of the Shapiro-Wilk method to confirm they were normally distributed. With groups as a random factor, ANOVA was performed to determine whether values varied with treatment group followed by Tukey post-test. All analyses

90 were performed with standard software (SAS 9.4 Software); values of p < 0.05 were considered to differ. Data were expressed as mean ± SEM.

Relative quantitation of the LC-MS/MS data was performed using the label-free approach described by Liu et al. and Colinge et al. [132,133]. Significance analysis of relative protein abundance from spectral count data was determined by using the EdgeR bioconductor package [134]. Peptide spectral count distributions were modeled using a

Poisson/negative binominal distribution and normalized to the respective spectral count library size [134–136]. Differences in protein abundance were evaluated based upon an exact text for the over-dispersed data [136]. Results were presented as +1 and -1 (increase in former part of the set or increase in latter part of the set, respectively, with a q value <

0.028) and 0 (increase or decrease in any comparison, with a q value > 0.028 and <

0.050). False discovery was controlled by applying a Benjamini-Hochberg multi-test correlation (α = 0.05) to final p-values [137]. Counts per million (CPM) were calculated as the base 2 log of the normalized average counts across a row divided by one million.

4.4 Results

Immune and inflammatory profiles of PBMCs and sMSCs

Response of PBMC cellular profile to the presence of sConstructs (Figure 17). Figure

17 shows PBMC viability, as determined by 7AAD, and induced PBMC differentiation as determined by MHC I, MHC II and CD11b surface markers. PBMC death was as anticipated in culture, 45-50%, and did not differ among the groups, nor did the PBMC

MHCII expression. The MHCI expression had a significant decrease from PBMCs alone to PBMCs co-cultured with sConstruct (p = 0.044). CD11b expression significantly increased from PBMCs alone to PBMCs co-cultured with sECM (p = 0.008), but no significant difference from PBMCs alone to PBMCs co-cultured with the sConstruct, demonstrating the presence of sMSC in the sConstruct was associated with lower CD11b expression.

All the culture conditions (controls and co-cultures) displayed a similar composition of lymphocyte subpopulations, i.e. 40–58% CD4+, 9–16% CD8+ and 10–14% CD4/CD8 double-positive T cells indicating no influence of the presence of sConstruct on immune activation of PBMCs.

Characterization of sMSCs in sConstructs and the influence of PBMCs (Figure 18). sMSCs were characterized by the % of cells expressing CD90, CD44, CD45 and MHC II, before seeding the sECMs as well as before and after seeding the sMSCs with, and

92 without, co-culture with PBMCs. Before seeding the sMSCs had a CD90 of > 95%,

CD44 of > 90%, CD45 of < 5%, MHC I of < 2%, and a MHC II of < 5%, demonstrating stem characteristics. After seeding for 5 days (sConstructs), but before PBMC co-culture, sMSCs expressed CD90 > 75%, CD44 > 80%, CD45 < 8%, MHC I < 2%, MHC II < 8%.

In all samples MHC I remained below 2% and was not graphed. In PBMC co-cultures,

CD45 and MHC II expression was always <10%, Figure 2B).

The influence on pro-inflammatory profile in the presence of sConstructs (Figure 19).

There was a 2.1-, 48-, 5.9-, and 6.2-fold increased response of IFN-, TNF-, IL-6, and

IL-1, respectively, in the activated (LPS) PBMCs over PBMCs-sConstruct co-cultures.

While all the inflammatory cytokines tested had increased concentrations when the

PBMCs were co-cultured with sMSCs, sECMs and sConstructs, the response was minimal compared to LPS activated PBMCs. Even though the response was low, the co- cultures showed a distinct response pattern; specifically, sMSCs had the lowest response, then sConstructs and the greatest inflammatory response was in sECM co-cultures. The presence of sMSC in the sECM (i.e. the sConstruct) reduced the inflammatory mediator release associated with the sECM.

The influence on the anti-inflammatory profile in the presence of sConstructs (Fig 20A and B). There were significant increases in anti-inflammatory TGF-β and IL-1ra concentrations in sMSC- and sConstruct-PBMC co-cultures over sECMs co-cultures, sConstructs alone, and PBMCs alone. The presence of sMSC, and further together with the sECM (sConstruct), produced greater release of anti-inflammatory mediators (Figure

93 4); the highest concentrations when co-cultured with PBMCs were in the sConstruct > sMSCs > sECMs.

The IL-1ra to IL-1 ratios (Fig 4 C) show that the PBMC co-cultures that included sMSCs, had significantly greater ratios of IL-1ra to IL-1. IL-1 concentrations were low. sConstruct abundance proteins and the influence of PBMC co-culturing

The cluster analysis and heat map (Figure 21A) revealed that the sMSC and the sECM had the greatest number of differentially expressed proteins, while the sConstruct had additional groups of proteins increased and decreased not seen in the sECM or the sMSC.

Figure 21B shows the relative abundance of the detected proteins found in each set of comparisons, E/M+/-, C/M+/-, C/E+/- and CP/C +/-. All sets containing sMSCs (E/M and

C/M) had the highest levels of detected proteins; E/M+/- had a combined total of 186,

C/M+/- had 137 and C/E+/- 98. CPC set (sConstructs-PBMCs/sConstruct) had only two detected significant proteins, both found only in the sConstruct-PBMC demonstrating little protein composition change between these groups. sECM and sMSC abundance proteins (Table 5 and Figure 22). Table 5 lists the proteins with their Log (FC) and assigned category in the E/M+/- (Table 5). Figure 22 are pie charts of the proteins in the sets. 79 and 71 E/M+ and E/M- proteins, respectively, out of the 103 and 83 detected, were identified and categorized. Figure 22A shows that sECMs abundance proteins (E/M+) has a high amount of extracellular matrix - region proteins (32%, the combined percentage of extracellular and matrix and region proteins), 94 13 of them with a Log (FC) > 7 (Table 5). There were substantial numbers of cytoplasmic

(17%), cytoplasmic membrane (9%), and nuclear proteins (16%), but the overall abundance of these proteins was low relative to sMSCs (E/M-). In the sECM proteins

(E/M+, Table 5) one protein specifically stood out, Transforming Growth Factor- beta induced protein IG-H3 (Log (FC) = 10.17). The sECM contained an abundance of extracellular matrix proteins, describing the primary composition of this scaffold.

The E/M- set (sMSC proteins also found in the sECMs, Figure 22B) is mainly a mixture of cellular component proteins, with one exception, the abundance of cytoskeleton proteins (27%). In Table 5, we noted one striking anomaly in the sMSC proteins (E/M-),

Insulin-like growth factor II (Log (FC) = 6.5). There were no other identifiable growth factors in the E/M+/- comparisons. The sMSC contained an abundance of cellular component proteins, describing the primary composition of cells. sConstructs abundance proteins. Table 6 lists the proteins with their Log (FC) and assigned category in the C/M+/- and C/E+/- (Table 6). Table 6 column A is the rank according to the highest log (FC) of all the proteins in the set and column B are other sConstruct sets (overlapping sets) where the proteins are found. Figure 23 are pie charts of the proteins in the sets. Of the 179 and 98 detected C/M+/- and C/E+/- proteins, 158 and 91, respectively were identified and categorized (Table 6). The relative abundance of the proteins in the categories is shown in the pie charts of Figure 23. In general, as expected, the CE sets showed that sConstructs contained an abundance of matrix component proteins when compared to sECM (and less cellular components) and the CM

95 sets showed an abundance of cellular component proteins when compared to sMSC (and less matrix component proteins).

There was a high turnover of extra cellular region - matrix proteins; all the pie charts

(Figure 23A, B, C, and D) are dominated by these categories. The extracellular proteins were in C/M+ = 31%, C/M- = 10%, C/E+ = 31%, C/E- = 67%. In the C/M+ and C/E+

(Figure 23A and B) there was substantial increases in cytoplasm, endoplasmic reticulum, and cytoplasmic-plasma membrane proteins; the combined amounts were C/M+ = 28%, and C/E+ = 46%. There was a dramatic loss in cytoskeletal proteins as shown in the C/M- and E/M- charts (36% and 14 % respectively, Figure 23C and D). The Transforming

Growth Factor- beta induced protein IG-H3 was found in the C/E- set of proteins indicating an increase of this protein in the sConstruct.

The identified proteins reflecting the sConstructs (Table 6) demonstrated regenerative features within the sConstruct, such as multiple collagen and collagen-related proteins seen early in collagen production and collagenases needed for remodeling. The C/M+ and C/E+ sets, represented elevated protein levels, Collagen Alpha-1(III) Chain, Collagen

Alpha-1(VI) Chain, Collagen Alpha-1(XII) Chain, Collagen Alpha-1(XIV) Chain,

Collagen Alpha-2(VI) Chain, Collagen Alpha-3(VI) Chain, Collagen Alpha-5(VI) Chain,

Collagen Alpha-6(VI) Chain 72 Kda Type Iv Collagenase, Procollagen C-Endopeptidase

Enhancer 1, Procollagen-Lysine, and 2-Oxoglutarate 5-Dioxygenase 1. In the C/E-, proteins with decreased levels, there were Collagen Alpha-1(VI) Chain, Collagen Alpha-

2(VI) Chain and Collagen Alpha-3(VI) Chain.

96 sConstruct abundance proteins, not found in sECMs and sMSCs. Figure 24 is a Venn diagram demonstrating the union of sets by numbers of proteins, specifically unique for the sConstructs and therefore not overlapped by sECM or sMSC. All protein numbers reported were present in sConstructs as either increased or decreased in abundance relative to the latter portion of the set. Specifically, 17 proteins were unique and abundant relative to sMSC, 22 proteins were unique and abundant to sECM, 4 proteins were unique, but less abundant that sMSC, and 1 protein was unique and but less abundant to the sECM. Table 7 lists the proteins determined using the Venn diagram. In Table 7, the cell locality category, function category, and, in Column A, the set location of the proteins is listed. In total there were 69 elevated proteins (C/M+ and C/E+) where 66 were identified (Table 7) and 9 decreased proteins detected and 7 identified (C/M- and

C/E-, Table 7). The high number of elevated proteins compared with the low number of decreased proteins shows that the sConstruct sMSCs are in a state of active protein buildup (growth) compared with sMSCs The cell locality chart (Figure 25A) indicates elevated proteins throughout the cell with higher amounts in the cytoplasm (30%) and extracellular region (23%) that are new to the sConstruct arrangement and supports cellular integration, function, and alterations that occur when in contact with the sECM and not without contact with the sECM. When categorizing the elevated proteins by their cell locality several function protein subcategories became apparent (Table 7 Function

Category). Figure 25B shows that the 3 largest subgroups were glycolysis (20%), cellular transport (18%) and other metabolic processes (14%). Three collagen proteins were contained in both the C/M + and C/E- (Collagen Alpha-1(VI) Chain, Collagen Alpha-

2(VI) Chain and Collagen Alpha-3(VI) Chain). This means that in the C/M+, the 97 sConstructs had higher levels when compared with sMSCs. C/E- means there are higher levels in the sECMs than in the sConstruct. It appears that there is a replacement of the sECM collagens by the sConstructs sMSCs. In addition, 4 proteins stood out due to their specific function. Three were immunomodulatory proteins (Stromelysin-2, Glia-derived

Nexin, and Translocator Protein) and one a participant in the biosynthesis of glycosaminoglycan’s (UDP-glucose 6-dehydrogenase). sConstruct-PBMC proteome vs sConstruct. Only two proteins were found specifically in the CPC+ set. Fibrinogen Gamma Chain, considered a contaminate (see Discussion). The other has not, at this time, been identified and its function is unknown.

4.5 Discussion

The allogeneic PBMCs co-cultured with the sConstructs caused a low, but detectable, inflammatory response as compared with a PBMC immunologic activator (LPS). The level of response followed a distinct pattern, most stimulated by the sECMs, and least by the sMSCs. Correspondingly, the PBMC co-culture with sConstructs increased the production of anti-inflammatory markers and cytokines, more so than sMSCs or sECMs.

Additionally, the low and steady expression of CD4+ and CD8+ PBMCs supports a limited cellular immune response. In concert, these findings indicated that the sMSCs integrating into the sECMs have an immunomodulatoy role that serves to promote a less inflammatory environment with less immune response. Proteomics of the sConstructs substantiated these observations and interpretations. There were few changes in the sConstruct proteome when comparing sConstructs not in co-culture to sConstructs co- cultured with PBMCs, showing that the presence of the PBMCs did not substantially alter the progress of the integration of the sMSC with the sECM. When comparing the sConstructs proteome to those of its original components, sMSCs and sECMs, there was an abundance of change indicating an active environment during integration and development of the bioactive scaffold. Independently, the sMSCs in the sConstructs had substantial increased metabolic activity enzymes with increased in many extracellular matrix proteins. In conclusion, our data demonstrated the development of a viable integrating tissue that is biologically active with immune-, inflammatory-modulating

99 capabilities with little activation of the peripheral blood immune cells. Our data supports that the properties of this tissue has potential as an allo-implant. Combined with our work in vivo with this implant [30], this study supports further work in immunocompetent and larger animal models.

It is known that mesenchymal stem cells should express high levels of cell surface markers like CD90 and CD44 while expressing low levels of others, like CD45 and MHC

II [138]. The markers CD90 and CD44 reflect the cells undifferentiated state and a low or decreased expression would indicate adhesion and differentiation into a specific cell lineage, i.e. loss of multipotency. The decrease of CD90 and CD44 after seeding was similar to that found in earlier publications and is most likely due to the interaction of the sMSCs with the sECM stimulating differentiation [16]. MHC expression should be low in sMSCs as stem cells are generally immunotolerant and express less major histocompatibility antigen [139]. Our sMSC were low in both MHC I and II as anticipate and reported [139]. It has been reported that several pro-inflammatory proteins, such as

IL-6 and IFN-, can increase the expression of both MHC I and MHC II on mesenchymal stem cells and hence remove their immune-privilege status [140,141]. Here the co-culture with PBMCs and increased cytokines did not significantly change the surface marker expression on the sMSCs, indicating they did not decrease the sMSC immune-privilege status.

The construct and its components had no effect on the viability or the MHC II expression of the PBMCs, however it did significantly increase CD11b between groups control/sConstruct and sMSC/sECM. CD11b is a marker of activated monocytes and has 100 been indicated in early stages of activation of inflammation [142]. An increased expression of CD11b, as seen in this study, has been found on recently activated CD8+ cells [142]. These cells react to MHC I, also seen increased on the PBMCs here, and results in the production of TNF- and IFN-, IL-1 and IL-6. However, some studies have reported that CD11b plays significant immunoregulatory roles [143]. The overall effect of the immune reaction and pro-inflammatory factors of the increase in CD11b, as seen in this study, will need further evaluation.

The cytokines were chosen due to their catabolic and inflammatory potential in joint disease. The cytokines IL-1β, TNF-α and IL-6 have been widely studied in equine joint disease [144–146]; IL-1β and TNF-α induce a synergistic effect in cartilage degradation and are characterized as purely catabolic cytokines [147]. IL-1β and TNF-α were both significantly increased compared to the PBMCs alone, but in comparison to the positive control (LPS) the increase seen in this study was very low. IL-1β, TNF-α and IL-6 all showed a similar trend to increase with exposure to sECM, not with sMSC, and this response to ECM is mitigated with by adding the sMSC to the scaffold. The LPS positive control was several fold changes higher than all other groups, so this effect was minimal and unknown clinical significance. The addition of the sECM to the PBMCs resulted in more cytokine production than the sMSCs, and the sMSCs in the scaffold mitigated this effect seen with sECM. Consistently, the addition of sMSCs to the sECM in the sConstructs had an immunomodulatory effect to reduce the inflammatory profile, increase the anti-inflammatory profile and to decrease the MHC expression.

101 IL-6 is considered to be a modulatory cytokine, with both pro- and anti-inflammatory effects, and is rapidly produced after joint insult [144,146]. In osteoarthritis, the net effect of catabolic/anti-catabolic and anabolic processes is determined by a complex interplay of cytokines and growth factors [147,148]. IFN- is an important activator of macrophages and inducer of MHC I on all cells and MHC II on antigen presenting cells expression. It is, similarly to IL-6, immune-stimulatory and immune-modulatory. It is produced by Th and Tc cells, and to lesser extent macrophages, mucosal epithelial cells and NK [149]. In this study no significant increase in IFN- was found and only a moderate amount of IFN- was produced in the LPS positive control.

Overall, the pro-inflammatory cytokines chosen in this study supports the idea sMSCs are immune-privileged. In addition, it became clear that the sMSCs, when in the sConstruct, can dampen the pro-inflammatory response to the sECM.

Anti-inflammatory cytokines are important in modulating and controlling the inflammatory response, either by being a direct antagonist (like IL-1ra is to IL-1) or by influencing the response of immune cells. In contrast to the inflammatory cytokines, the anti-inflammatory cytokines had highest concentration in the sConstruct-PBMC co- culture. This was seen with all the ant-inflammatory cytokines except IL-10. TGF- is a multifunctional set of peptides that controls proliferation, differentiation, and other functions in many cell types. TGF-β1 plays an important role in controlling the immune system, and shows different activities on different types of cell, or cells at different developmental stages. Most immune cells (or leukocytes) secrete TGF-β1[150]. As in many systems, the effects of TGF-β on T cells are, in part, a function of their state of 102 differentiation, and such effects often are altered by the nature of the activating signals.

The significant increase in TGF-β seen in the sConstructs and sMSCs may explain the immunomodulatory effects seen on the pro-inflammatory cytokines in these 2 groups. IL-

1ra is a member of the interleukin 1cytokine family. IL-1ra is secreted by various types of cells including immune cells, epithelial cells, and adipocytes, and is a natural inhibitor of the pro-inflammatory effect of IL-1β. This protein inhibits the activities of IL-1 β, and modulates a variety of interleukin 1 related immune and inflammatory responses

[151].The ratio of IL-1β to IL-1ra seen in this study supports the overall anti- inflammatory effects the sMSCs have, both alone and in the sConstruct.

Interestingly, it has been found that MSCs may only modulate immunosuppression when they are first stimulated by inflammatory cytokines, such as tumor necrosis factor (TNF) and interleukin- (IL-) 1[152]. MSCs not only respond to inflammatory cytokines but also produce immune-regulatory secretors that mediate the process of inflammation in response. For example, a large number of indoleamine 2,3-dioxygenase (IDO) in humans, nitric oxide (NO) in mice, and chemokines produced by MSCs play a key part in MSC- mediated immunomodulation[153] These are factors not studied in this work and could potentially explain the sMSC immune modulatory effect seen here.

When considering the low pro-inflammatory response together with the ant-inflammatory profile of the sConstruct in vitro, it is possible the sConstruct will elicit little immune response in vivo and support an overall anti-inflammatory and anabolic joint environment for a longer period of time. 103 Proteomics is the study of the proteome, the protein composition of biological components and cells. This approach examines thousands of proteins in a high throughput way, which has several advantages over examining the transcriptional level or an individual protein level. For one, mRNA levels do not always correlate with protein levels due to post-translational modifications. Further, mRNA levels do not provide information on protein quality, and this is especially true for the highly post- translationally modified proteins found in an ECM. In addition, a proteomics approach provides direct information about a mixture of proteins, including quantity and quality

(e.g., presence of post-translational modifications) [154]. Post-translational modifications can dramatically alter the signaling transduction networks that link intercellular and extracellular communication [155,156]. There are many post-translational modifications relevant to the ECM include glycosylation, citrullination, and proteolytic processing[155]. The latter is of special importance for cytokine regulation of ECM

[157][158]. For the purposes of our work here, evaluation of the protein milieu that serves as the microenvironment and influences cells and tissue development was our focus.

Our sECM proteome results show that sECM is made of more various extracellular proteins than sMSCs as would be expected, however, the presence of significant abundance of cellular components in the E/M- set, means that the sECMs contained cellular protein remnants. These proteins were mostly cytoskeletal proteins, most likely caught or in some way attached to the sECMs. Cellular proteins associated with nuclear

104 content was only 12% and of those proteins, none were identified as DNA related proteins.

Transforming Growth Factor-beta induced protein IG-H3 was found in the sECM and the sConstructs in addition to the expected extracellular matrix components. IG-H3 promotes cell adhesion [159,160] and binds to collagens [161,162], fibronectin and glycosaminoglycans [163,164] suggesting that IG-H3 might play roles in development processes and tissue modeling. In addition, prominent expression of IG-H3 transcripts has been detected in chondrogenic tissue destined for cartilage or bone formation. This expression occurred during the cell recruitment stage for the formation of a cartilaginous model. IG-H3 expression was observed in all developing areas of axial, craniofacial, and appendicular primordial cartilage, including areas of vertebral cartilage [164]. This protein could be beneficial both for the sMSCs as they migrate through the sECM and through leaching from the sECM and sConstruct into the environment surrounding cartilage injury.

This study was not an endeavor to characterize the sMSC proteome but focused on the sECM and sConstruct. However, IGF II, was found with a high Log (FC) in the proteomic sMSC sets, and has multiple functions similar to those of IGF-I. These include cell proliferation, differentiation, growth, migration, and survival. Despite its high circulating levels in postnatal life, IGF II is probably more important in regulating embryonic/fetal growth in vivo [165]. The abundance of IGF II seen in the sMSCs in monolayer when compared to the sECM was not expressed when the sMSCs were in the sConstruct (C/M- set). This may be suggestive of differentiation of the sMSCs. IGF II 105 may promote cartilage repair, as is a potent function for IGF 1 [166], and would seem to be an alternate cell associated source of IGF.

In the sConstruct, there were 66 elevated, (i.e. found more abundant in the sConstruct) identified proteins that were not significantly present in the sECM or sMSCs alone. There was only 7 decreased proteins (i.e. not present in the sConstructs but in the sECM and/or sMSC). The elevated proteins when categorized on function were associated with glycolysis, transport, collagen and collagen related proteins, ECM modifiers and antioxidants. These elevated proteins support the conclusion that the sConstruct is a new viable tissue product, not merely the sum of sECM and sMSCs. Based on the function categories, the sConstruct had an increased cell metabolism, with corresponding increase in antioxidants to reduce the excess metabolic oxidative stress. The sMSCs appeared to be decreasing in their cytoskeletal structure with a turnover and buildup of extracellular matrix proteins. There was both elevated and decreased levels of Collagens and collagen related enzymes, indicating a replacement of some of the sECM collagens by the sConstructs sMSCs.

In the elevated sConstruct proteins, not observed in sMSCs and sECMs, 4 proteins stood out due to their specific function; 3 immunomodulatory proteins (MMP10, Glia-derived

Nexin, and Translocator Protein) and 1 participant in the biosynthesis of glycosaminoglycan’s (UDP-glucose 6-dehydrogenase). MMP10 is induced in response to injury, infection, or transformation in essentially all tissues and was initially considered a catabolic enzyme. Several studies have contradicted this and the widespread expression of MMP10 among tissues suggests that this proteinase serves critical roles in the host 106 response to environmental insults. It has been shown that MMP10 serves a protective role in acute infection by moderating the pro-inflammatory activity of macrophages. Hence,

MMP10 appears to play a beneficial role in mitigating the deleterious responses to injury and infection by promoting the conversion of pro-inflammatory macrophages toward immunosuppressive cells [167]. In addition, MMP-10 has been implicated in the controlling the tissue remodeling activity of macrophages and moderating scar formation during wound repair [168]. Finally, it has been postulated that, although chondrocyte migration in vivo remains little studied, that MMP10 may promote migration of chondrocytes in an attempted wound healing response through release of matrix-bound factors such as FGF-2 [169]. These studies indicate that the sConstruct increased levels of MMP10 may be highly beneficial for cartilage repair when sConstructs are used as a treatment.

Glia-derived Nexin regulates matrix accumulation and coagulation by inhibiting thrombin, plasmin, tPA, and uPA. Glia-derived Nexin was shown to be induced by proinflammatory cytokines in different cell types. IL-1beta, TGF-beta, and TNF-alpha induced Glia-derived Nexin in neurons and muscle cells [170,171], chronic exposure to

TNF-alpha in rat fibroblast-like synoviocytes and mouse endothelial cells [172,173], and stimulation of human monocytes with lipopolysaccharide were shown to up-regulate the expression of Glia-derived Nexin [174]. Also, Glia-derived Nexin has been detected at high levels in atherosclerotic plaques and was suggested to play a protective role against aggression of proteases under inflammatory conditions [174]. Most interestingly, Glia- derived Nexin indeed prevented IL-1beta/basic fibroblast growth factor–induced articular

107 cartilage loss through the inhibition of plasmin, thus preventing the subsequent activation of matrix metalloproteinases (MMPs) in rabbits [175]. Given the protective role of Glia- derived Nexin against cartilage loss by inhibiting plasmin and by averting the subsequent activation of cartilage catabolic MMPs [176].

Translocator protein 18 kDa (TSPO) was previously known as the peripheral benzodiazepine receptor (PBR) in eukaryotes, where it is mainly localized to the mitochondrial outer membrane. Considerable evidence indicates that it plays regulatory roles in steroidogenesis. Oxidative stress appears to be a common theme, and porphyrins have often been implicated as having a role in regulating it. Porphyrins bind to TSPO with micromolar affinity [177].

UDP-glucose dehydrogenase (UGDH) catalyzes the transformation of UDP-glucose to

UDP-glucuronic acid, a key precursor for the synthesis of the glycosaminoglycan chain in proteoglycans, hyaluronan, chondroitin sulfate, and heparan sulfate. These glycosylated compounds are common components of the extracellular matrix and likely play roles in signal transduction, cell migration. Stimulating UGDH enzyme activity with transforming growth factor β (TGF-β) results in enhanced GAG synthesis in articular chondrocytes [178].

When looking at the sConstructs compared with its constituents, the increases in

Procollagen and decreases in Collagen Alpha I is an indicator that the sMSCs are rebuilding the 3-dimensional structure of the ECM.

108 There were only two unique proteins detected in the sConstruct-PBMC/sConstruct set.

Only one could be identified, Fibrinogen Gamma Chain. Fibrinogen is made and secreted into the blood primarily by liver hepatocyte cells. Blood platelets and their precursors, bone marrow megakaryocytes, while once thought to make fibrinogen, are now known to take up and store but not make the glycoprotein [179]. The most likely source of this protein would be platelet contamination during the isolation of the PBMCs. Similarly, serum albumin and coagulation factors was found in all the groups, most likely due to contamination from the FBS used in the media. Further washing cycles may remove this issue. This leaves only one protein of unknown name and function was found to differ from the sConstruct proteomes in sConstruct-PBMCs. Our immune studies indicated that there were the sConstructs had an increased anti-inflammatory effect on the PBMCs. In the in vitro PBMC co-culture, the sConstruct sMSCs did not change surface characteristics, the PBMCs increased the % of CD11b expression slightly and there was an increase in anti-inflammatory mediators with little to no pro-inflammatory cytokines.

The proteomics supports the concept of little pro-inflammatory effect on the sConstruct when co-cultured with PBMCs, as only 1 protein was found (with unknown function).

Here we show, in vitro, that allogeneic sConstructs cause low immune recognition by dendritic cells and low inflammatory response. The immunomodulatory effects of sMSCs in co-culture with PBMCs was clearly shown both alone and in the sConstruct through suppression of pro-inflammatory mediators and greater abundance of anti-inflammatory cytokines, with little changes in sMSC surface markers. In total, these data support an immune-tolerance of the sConstruct compared to the sECM alone, a characteristic

109 brought on by the sMSC. In addition, the proteomics of sConstructs paints a picture of highly active metabolic sMSCs integrating into the sECMs, including active remodeling of the extracellular matrix structure.

110

4.6 Tables

Table 4. Cell locality categories used to group sConstruct proteins selected by EdgeR q < 0.028.

GO Categories used to assign to a cell Cell Locality Category locality category Cytoplasm (GO:0005737) Cytosol (GO:0005829) Cytoplasm Cell (GO:0005623) Intracellular (GO:0005622) Cytoplasm (GO:0005737) Plasma membrane (GO:0005886) Cytoplasmic microtubule (GO:0005881) Cytosol (GO:0005829) Cytoplasmic Membrane Lysosomal membrane (GO:0005765) Apical plasma membrane (GO:0016324) Intracellular (GO:0005622) Membrane (GO:0016020) Actin cytoskeleton (GO:0015629) Actin filament (GO:0005884) Stress fiber (GO:0001725) Cytoskeleton Unconventional myosin complex (GO:0016461) Intermediate filament (GO:0005882) Cytoskeleton (GO:0005856) Endoplasmic reticulum (GO:0005783) Endoplasmic Reticulum Golgi membrane (GO:0000139) Extracellular matrix (GO:0031012) Extracellular Matrix Cell-cell junction (GO:0005911) Extracellular matrix (GO:0005578) Extracellular space (GO:0005615) Extracellular Region Extracellular region (GO:0005576) Mitochondrion (GO:0005739) Mitochondria Mitochondrial matrix (GO:0005759) Nucleus Nucleosome (GO:0000786)

111 Table 4. Cell locality categories used to group sConstruct proteins selected by EdgeR q < 0.028.

GO Categories used to assign to a cell Cell Locality Category locality category Nucleus (GO:0005634) Nucleolus (GO:0005730) Nucleoplasm (GO:0005654) Nuclear chromatin (GO:0000790) External side of plasma membrane (GO:0009897) Plasma Membrane Plasma membrane (GO:0005886) Integral component of membrane (GO:0016021) Ribosome Ribosome (GO:0005840) Other

112 Table 5. E/M proteins

E/M+* Sorted first by cell locality category, then by Log (FC).

Cell Locality Protein Name (PANTHER family ID name: Subfamily list name) Log (FC) Categories PURINE NUCLEOSIDE PHOSPHORYLASE (PTHR11904:SF12) 7.93 14-3-3 PROTEIN BETA/ALPHA (PTHR18860:SF28) 6.13 INTER-ALPHA-TRYPSIN INHIBITOR HEAVY CHAIN H1 5.87 (PTHR10338:SF106) GLUTATHIONE S-TRANSFERASE MU 3 (PTHR11571:SF133) 5.66 GLUTATHIONE S-TRANSFERASE THETA-3 (PTHR43917:SF9) 5.48 ADENOSYLHOMOCYSTEINASE (PTHR23420:SF14) 5.34 RETINAL DEHYDROGENASE 1 (PTHR11699:SF140) 5.23 Cytoplasm

ALDO-KETO REDUCTASE FAMILY 1, MEMBER C-LIKE 5.08 (PTHR11732:SF404) PEROXIREDOXIN-2 (PTHR10681:SF102) 4.98 14-3-3 PROTEIN GAMMA (PTHR18860:SF22) 4.84 CYTOSOLIC 10-FORMYLTETRAHYDROFOLATE DEHYDROGENASE 4.60 (PTHR11699:SF120) GLUTATHIONE S-TRANSFERASE MU 2 (PTHR11571:SF144) 3.64 RAB GDP DISSOCIATION INHIBITOR ALPHA (PTHR11787:SF3) 2.70 RAS GTPASE-ACTIVATING-LIKE PROTEIN IQGAP1 (PTHR14149:SF15) 5.40 ANNEXIN A3 (PTHR10502:SF25) 4.97 A-KINASE ANCHOR PROTEIN 12 (PTHR23209:SF4) 4.97 Cytoplasm L-XYLULOSE REDUCTASE (PTHR44252:SF2) 4.91 Membrane

CALPAIN SMALL SUBUNIT 1 (PTHR46735:SF1) 4.70 RAS-RELATED PROTEIN RAP-1B-RELATED (PTHR24070:SF393) 4.46 ALCOHOL DEHYDROGENASE [NADP(+)] (PTHR11732:SF385) 4.36 MYOSIN-10 (PTHR45615:SF24) 5.44 Cytoskeleton PLASTIN-2 (PTHR19961:SF35) 4.86 Endoplasmic EPOXIDE HYDROLASE 1 (PTHR21661:SF69) 4.50 Reticulum COLLAGEN ALPHA-3(VI) CHAIN (PTHR24020:SF13) 12.90 TRANSFORMING GROWTH FACTOR-BETA-INDUCED PROTEIN IG-H3 10.17 (PTHR10900:SF82) MIMECAN (PTHR46269:SF1) 9.71 COLLAGEN ALPHA-6(VI) CHAIN (PTHR22588:SF5) 9.17 Extracellular Matrix BIGLYCAN (PTHR45712:SF11) 8.54 DECORIN (PTHR45712:SF14) 8.42 PROLARGIN (PTHR45712:SF8) 8.30 ASPORIN (PTHR45712:SF2) 8.28

113 E/M+* Sorted first by cell locality category, then by Log (FC).

Cell Locality Protein Name (PANTHER family ID name: Subfamily list name) Log (FC) Categories COLLAGEN ALPHA-2(VI) CHAIN (PTHR24020:SF29) 8.07 COLLAGEN ALPHA-1(XIV) CHAIN (PTHR24020:SF15) 7.90 BASEMENT MEMBRANE-SPECIFIC HEPARAN SULFATE 7.75 PROTEOGLYCAN CORE PROTEIN (PTHR10574:SF273) COLLAGEN ALPHA-1(VI) CHAIN (PTHR24020:SF18) 7.74 COLLAGEN ALPHA-5(VI) CHAIN (PTHR22588:SF4) 7.03 LAMININ SUBUNIT BETA-2 (PTHR10574:SF36) 6.81 DERMATOPONTIN (PTHR15040:SF2) 5.89 FIBROMODULIN (PTHR45712:SF4) 5.31 COLLAGEN ALPHA-1(III) CHAIN (PTHR24023:SF604) 5.23 PROTEIN-GLUTAMINE GAMMA-GLUTAMYLTRANSFERASE 2 5.17 (PTHR11590:SF6) ALPHA-1B-GLYCOPROTEIN (PTHR11738:SF3) 5.08 NIDOGEN-1 (PTHR46513:SF6) 4.88 NIDOGEN-2 (PTHR12352:SF23) 4.53 TENASCIN-X (PTHR46708:SF3) 4.16 COFILIN-1 (PTHR11913:SF17) 3.12 ACTIN-RELATED PROTEIN 3 (PTHR11937:SF175) 2.57 SUBFAMILY NOT NAMED (PTHR24020:SF34) 2.53 VON WILLEBRAND FACTOR A DOMAIN-CONTAINING PROTEIN 1 6.51 (PTHR24020:SF21) FIBRINOGEN ALPHA CHAIN (PTHR47221:SF1) 5.98 XANTHINE DEHYDROGENASE/OXIDASE (PTHR11908:SF80) 5.40 FIBRINOGEN GAMMA CHAIN (PTHR19143:SF338) 5.37 Extracellular MICROFIBRIL-ASSOCIATED GLYCOPROTEIN 4 (PTHR19143:SF225) 5.37 Region

COMPLEMENT C4-A-RELATED (PTHR11412:SF86) 3.99 ADENYLYL CYCLASE-ASSOCIATED PROTEIN 1 (PTHR10652:SF1) 3.37 SIGNAL RECOGNITION PARTICLE RECEPTOR SUBUNIT BETA 2.55 (PTHR11485:SF34) ANNEXIN A2-RELATED (PTHR10502:SF18) 1.16 DIHYDROPYRIMIDINASE-RELATED PROTEIN 2 (PTHR11647:SF56) 6.98 GLUTATHIONE S-TRANSFERASE P (PTHR11571:SF214) 5.72 Mitochondria ANNEXIN A6 (PTHR10502:SF19) 2.15 HISTONE H2A TYPE 2-B (PTHR23430:SF72) 6.81

ANNEXIN A4 (PTHR10502:SF28) 6.27 Nucleus SELENIUM-BINDING PROTEIN 1 (PTHR23300:SF0) 5.72 TRANSKETOLASE (PTHR43195:SF3) 5.35

114 E/M+* Sorted first by cell locality category, then by Log (FC).

Cell Locality Protein Name (PANTHER family ID name: Subfamily list name) Log (FC) Categories HISTONE H2B TYPE 1-B (PTHR23428:SF28) 5.31 HYDROXYACYL-COENZYME A DEHYDROGENASE, 5.01 MITOCHONDRIAL (PTHR43561:SF3) ADENOSINE KINASE (PTHR45769:SF3) 4.99 CYTOSOL AMINOPEPTIDASE (PTHR11963:SF23) 4.93 CORE HISTONE MACRO-H2A.1 (PTHR23430:SF20) 4.90 PROTEIN DJ-1 (PTHR43444:SF2) 4.76 PROTEIN S100-A1 (PTHR11639:SF66) 4.67 ANNEXIN A7 (PTHR10502:SF140) 4.59 HEAT SHOCK 70 KDA PROTEIN 1A-RELATED (PTHR19375:SF223) 2.34 SUBFAMILY NOT NAMED (PTHR11571:SF208) 5.84 Other FATTY ACID-BINDING PROTEIN, ADIPOCYTE (PTHR11955:SF83) 6.64 MYELOID-ASSOCIATED DIFFERENTIATION MARKER 5.66 (PTHR17068:SF3) Plasma SYNAPTIC VESICLE MEMBRANE PROTEIN VAT-1 HOMOLOG 4.57 Membrane (PTHR44054:SF1)

6-PHOSPHOGLUCONATE DEHYDROGENASE, DECARBOXYLATING 4.46 (PTHR11811:SF45) SYNAPTOPHYSIN-LIKE PROTEIN 1 (PTHR10306:SF9) 4.35 * E/M+ = sECM to sMSC (+) comparison set.

115 E/M- proteins sorted first by cell locality category, then by Log (FC).

Cell Locality Protein Name (PANTHER family ID name: Subfamily list name) Log (FC) Categories TRANSGELIN-2 (PTHR18959:SF41) 7.55 CALDESMON (PTHR18949:SF0) 6.32 CYTOSKELETON-ASSOCIATED PROTEIN 4 (PTHR45161:SF1) 5.92 KERATIN, TYPE II CYTOSKELETAL 6B-RELATED (PTHR45616:SF39) 4.55 KERATIN, TYPE II CYTOSKELETAL 1 (PTHR45616:SF33) 4.30 TROPOMYOSIN ALPHA-4 CHAIN (PTHR19269:SF40) 4.13 TROPOMYOSIN BETA CHAIN (PTHR19269:SF46) 3.94 MYOSIN REGULATORY LIGHT CHAIN 12A (PTHR23049:SF57) 3.70 MYOSIN LIGHT POLYPEPTIDE 6 (PTHR23048:SF7) 3.64 Cytoskeleton ACTIN, CYTOPLASMIC 1 (PTHR11937:SF192) 3.40

MYOSIN-9 (PTHR45615:SF16) 3.37 KERATIN, TYPE II CYTOSKELETAL 5 (PTHR45616:SF32) 3.21 TRANSGELIN (PTHR18959:SF40) 2.77 ALPHA-ACTININ-4 (PTHR11915:SF425) 2.42 ACTIN, CYTOPLASMIC TYPE 5 (PTHR11937:SF243) 2.30 FRUCTOSE-BISPHOSPHATE ALDOLASE A (PTHR11627:SF1) 1.91 TUBULIN BETA CHAIN (PTHR11588:SF61) 1.86 ACTIN, ALPHA CARDIAC MUSCLE 1 (PTHR11937:SF176) 1.75 VIMENTIN (PTHR45652:SF5) 0.99 CALUMENIN (PTHR10827:SF52) 6.13 ENDOPLASMIC RETICULUM RESIDENT PROTEIN 29 (PTHR12211:SF0) 5.54 TRANSLOCON-ASSOCIATED PROTEIN SUBUNIT ALPHA 5.46 (PTHR12924:SF0) PROCOLLAGEN-LYSINE,2-OXOGLUTARATE 5-DIOXYGENASE 2 4.60 (PTHR10730:SF6) SERPIN H1 (PTHR11461:SF27) 4.56 THIOREDOXIN DOMAIN-CONTAINING PROTEIN 5 (PTHR45672:SF3) 4.55 GLUCOSIDASE 2 SUBUNIT BETA (PTHR12630:SF1) 3.95 Endoplasmic Reticulum PROTEIN DISULFIDE-ISOMERASE (PTHR18929:SF101) 3.59 ENDOPLASMIN (PTHR11528:SF54) 3.06 COATOMER SUBUNIT BETA' (PTHR19876:SF2) 2.99 CALNEXIN (PTHR11073:SF11) 2.75 DOLICHYL-DIPHOSPHOOLIGOSACCHARIDE--PROTEIN 2.59 GLYCOSYLTRANSFERASE 48 KDA SUBUNIT (PTHR10830:SF0) PEPTIDYL-PROLYL CIS-TRANS ISOMERASE B (PTHR11071:SF387) 2.49 RAS-RELATED PROTEIN RAB-1A (PTHR24073:SF212) 2.01

116 E/M- proteins sorted first by cell locality category, then by Log (FC).

Cell Locality Protein Name (PANTHER family ID name: Subfamily list name) Log (FC) Categories INSULIN-LIKE GROWTH FACTOR II (PTHR46886:SF1) 6.50 PROTEIN S100-A11 (PTHR11639:SF60) 6.36 PROTEIN S100-A6 (PTHR11639:SF80) 6.15 PLASMINOGEN ACTIVATOR INHIBITOR 1 (PTHR11461:SF49) 5.00 CALRETICULIN (PTHR11073:SF16) 4.79 60 KDA HEAT SHOCK PROTEIN, MITOCHONDRIAL (PTHR45633:SF3) 4.09 Extracellular VITAMIN K-DEPENDENT PROTEIN S (PTHR24040:SF0) 3.37 Region

THROMBOSPONDIN-1 (PTHR10199:SF78) 2.74 INTEGRIN BETA-1 (PTHR10082:SF28) 2.38 PROTEIN DISULFIDE-ISOMERASE A3 (PTHR18929:SF191) 2.31 ORNITHINE AMINOTRANSFERASE, MITOCHONDRIAL 6.24 (PTHR11986:SF18) PEROXIREDOXIN-4 (PTHR10681:SF146) 5.80

PROHIBITIN-2 (PTHR23222:SF1) 4.60 Mitochondria 10 KDA HEAT SHOCK PROTEIN, MITOCHONDRIAL (PTHR10772:SF0) 4.56 ASPARTATE AMINOTRANSFERASE, MITOCHONDRIAL 2.96 (PTHR11879:SF10) HSC70-INTERACTING PROTEIN-RELATED (PTHR45883:SF2) 2.41 NUCLEASE-SENSITIVE ELEMENT-BINDING PROTEIN 1 5.89 (PTHR11544:SF68) THIOREDOXIN (PTHR10438:SF18) 5.71 HETEROGENEOUS NUCLEAR RIBONUCLEOPROTEIN M 4.83 (PTHR23003:SF6) 78 KDA GLUCOSE-REGULATED PROTEIN (PTHR19375:SF144) 3.85 HETEROGENEOUS NUCLEAR RIBONUCLEOPROTEINS C1/C2 2.97 (PTHR13968:SF3) HETEROGENEOUS NUCLEAR RIBONUCLEOPROTEIN A1-RELATED 2.74 (PTHR24012:SF689) HEAT SHOCK PROTEIN HSP 90-ALPHA-RELATED (PTHR11528:SF34) 2.37 Nucleus SUBFAMILY NOT NAMED (PTHR23428:SF189) 2.33 HETEROGENEOUS NUCLEAR RIBONUCLEOPROTEIN U 2.29 (PTHR12381:SF11) NUCLEOLIN (PTHR15241:SF92) 2.11 HEAT SHOCK PROTEIN HSP 90-BETA (PTHR11528:SF79) 1.63 SUBFAMILY NOT NAMED (PTHR23115:SF225) 1.56 HISTONE H4 (PTHR10484:SF163) 1.38 HETEROGENEOUS NUCLEAR RIBONUCLEOPROTEINS A2/B1 1.33 (PTHR15241:SF113) HISTONE H1.2 (PTHR11467:SF86) 1.25 60S ACIDIC RIBOSOMAL PROTEIN P2 (PTHR21141:SF5) 6.00 Ribosome

117 E/M- proteins sorted first by cell locality category, then by Log (FC).

Cell Locality Protein Name (PANTHER family ID name: Subfamily list name) Log (FC) Categories 60S RIBOSOMAL PROTEIN L12 (PTHR11661:SF8) 3.42 60S RIBOSOMAL PROTEIN L18 (PTHR10934:SF2) 2.56 UBIQUITIN-60S RIBOSOMAL PROTEIN L40 (PTHR10666:SF268) 2.19 60S RIBOSOMAL PROTEIN L22 (PTHR10064:SF2) 1.98 60S RIBOSOMAL PROTEIN L13 (PTHR11722:SF0) 1.88 60S RIBOSOMAL PROTEIN L7 (PTHR11524:SF12) 1.68 * E/M- = sECM to sMSC (-1) comparison set.

118

Table 6. C/M and C/E proteins in sConstructs

C/M-* proteins in sConstructs sorted first by cell locality category, then by Log (FC). Column A is the rank according to the highest log (FC) of all the C/M- proteins and the B column are other sets containing the same protein.

Protein Name Cell Locality Log (FC) B C (PANTHER family ID name: Subfamily list name) Category PROFILIN-1 (PTHR13936:SF14) 1.64 Cytoplasm KERATIN, TYPE II CYTOSKELETAL 1 3.60 17 (PTHR45616:SF33) KERATIN, TYPE II CYTOSKELETAL 6B-RELATED 8.64 1 (PTHR45616:SF39) CALDESMON (PTHR18949:SF0) 5.41 4 MYOSIN REGULATORY LIGHT CHAIN 12A 5.27 5 (PTHR23049:SF57) TROPOMYOSIN ALPHA-1 CHAIN (PTHR19269:SF41) 4.80 9 INTERMEDIATE FILAMENT FAMILY 4.14 10 TROPOMYOSIN BETA CHAIN (PTHR19269:SF46) 4.06 11 MYOSIN-9 (PTHR45615:SF16) 3.80 14 TROPOMYOSIN ALPHA-4 CHAIN (PTHR19269:SF40) 3.70 15 KERATIN, TYPE II CYTOSKELETAL 5 3.62 16 (PTHR45616:SF32) Cytoskeleton CYTOSKELETON-ASSOCIATED PROTEIN 4 2.77 C/E+ (PTHR45161:SF1) ACTIN, ALPHA CARDIAC MUSCLE 1 2.40 (PTHR11937:SF176) ACTIN, CYTOPLASMIC 1 (PTHR11937:SF192) 2.23 MYOSIN LIGHT POLYPEPTIDE 6 (PTHR23048:SF7) 2.17 ACTIN, CYTOPLASMIC TYPE 5 (PTHR11937:SF243) 2.16 TRANSGELIN (PTHR18959:SF40) 2.15 VIMENTIN (PTHR45652:SF5) 1.83 C/E- TRANSGELIN-2 (PTHR18959:SF41) 1.72 C/E+ TUBULIN BETA CHAIN (PTHR11588:SF61) 1.68 TUBULIN ALPHA-1B CHAIN (PTHR11588:SF251) 1.39 C/E- ALPHA-ACTININ-4 (PTHR11915:SF425) 1.31 ENDOPLASMIC RETICULUM RESIDENT PROTEIN 29 5.03 7 (PTHR12211:SF0)

SERPIN H1 (PTHR11461:SF27) 3.01 Endoplasmic C/E+ CALUMENIN (PTHR10827:SF52) 1.94 Reticulum C/E+ PROTEIN DISULFIDE-ISOMERASE 1.86 C/E+ (PTHR18929:SF101) INSULIN-LIKE GROWTH FACTOR II (PTHR46886:SF1) 7.39 Extracellular 2 VITAMIN K-DEPENDENT PROTEIN S Region 7.14 3 (PTHR24040:SF0) 119 C/M-* proteins in sConstructs sorted first by cell locality category, then by Log (FC). Column A is the rank according to the highest log (FC) of all the C/M- proteins and the B column are other sets containing the same protein.

Protein Name Cell Locality Log (FC) B C (PANTHER family ID name: Subfamily list name) Category INTEGRIN BETA-1 (PTHR10082:SF28) 2.46 APOLIPOPROTEIN A-I (PTHR18976:SF11) 2.45 C/E- 60 KDA HEAT SHOCK PROTEIN, MITOCHONDRIAL 1.64 (PTHR45633:SF3) ANNEXIN A1 (PTHR10502:SF17) 1.39 ORNITHINE AMINOTRANSFERASE, 5.04 6 MITOCHONDRIAL (PTHR11986:SF18) STRESS-70 PROTEIN, MITOCHONDRIAL-LIKE 4.00 12 PROTEIN Mitochondria ASPARTATE AMINOTRANSFERASE, 2.40 MITOCHONDRIAL (PTHR11879:SF10) HSC70-INTERACTING PROTEIN-RELATED 1.82 (PTHR45883:SF2) NUCLEOPHOSMIN-LIKE PROTEIN 4.97 8 THIOREDOXIN (PTHR10438:SF18) 3.39 18 HETEROGENEOUS NUCLEAR 3.28 19 RIBONUCLEOPROTEIN M (PTHR23003:SF6) HETEROGENEOUS NUCLEAR 3.11 RIBONUCLEOPROTEINS C1/C2 (PTHR13968:SF3) 78 KDA GLUCOSE-REGULATED PROTEIN 2.70 (PTHR19375:SF144) HISTONE H4 (PTHR10484:SF163) 2.55 SUBFAMILY NOT NAMED (PTHR23428:SF189) 2.21 NUCLEASE-SENSITIVE ELEMENT-BINDING 2.21 C/E+ PROTEIN 1 (PTHR11544:SF68) Nucleus HISTONE H1.2 (PTHR11467:SF86) 2.18 DESMIN (PTHR45652:SF2) 2.14 C/E- HETEROGENEOUS NUCLEAR 1.94 RIBONUCLEOPROTEIN U (PTHR12381:SF11) HISTONE H2A TYPE 1 (PTHR23430:SF29) 1.91 HEAT SHOCK PROTEIN HSP 90-ALPHA-RELATED 1.69 (PTHR11528:SF34) HETEROGENEOUS NUCLEAR 1.65 RIBONUCLEOPROTEINS A2/B1 (PTHR15241:SF113) PEPTIDYL-PROLYL CIS-TRANS ISOMERASE B 1.56 (PTHR11071:SF387) PRELAMIN-A/C (PTHR45721:SF5) 1.20 RIBOSOME BINDING PROTEIN 1 3.99 13 60S RIBOSOMAL PROTEIN L7 (PTHR11524:SF12) 3.27 20 UBIQUITIN-60S RIBOSOMAL PROTEIN L40 2.47 Ribosome (PTHR10666:SF268) 60S RIBOSOMAL PROTEIN L18 (PTHR10934:SF2) 2.26 60S ACIDIC RIBOSOMAL PROTEIN P2 1.80 C/E+ (PTHR21141:SF5) 120 C/M-* proteins in sConstructs sorted first by cell locality category, then by Log (FC). Column A is the rank according to the highest log (FC) of all the C/M- proteins and the B column are other sets containing the same protein.

Protein Name Cell Locality Log (FC) B C (PANTHER family ID name: Subfamily list name) Category 60S RIBOSOMAL PROTEIN L13 (PTHR11722:SF0) 1.72 * C/M- = sConstruct to sMSC (-1) comparison set.

121 C/M+* proteins in sConstruct sorted first by cell locality category, then by Log (FC). Column A is the rank according to the highest log (FC) of all the C/M+ proteins and the B column are other sets containing the same protein.

Cell Protein Name Log Locality A B (PANTHER family ID name: Subfamily list name) (FC) Category C/E GAMMA-ENOLASE (PTHR11902:SF10) 7.84 9 + PURINE NUCLEOSIDE PHOSPHORYLASE (PTHR11904:SF12) 7.73 10 14-3-3 PROTEIN BETA/ALPHA (PTHR18860:SF28) 7.40 19 C/E 14-3-3 PROTEIN GAMMA (PTHR18860:SF22) 6.92 + C/E PURINE NUCLEOSIDE PHOSPHORYLASE (PTHR11904:SF12) 6.82 + C/E HOSPHOGLYCERATE MUTASE 1 (PTHR11931:SF15) 6.81 + FERRITIN LIGHT CHAIN (PTHR11431:SF47) 6.38 GLUTATHIONE S-TRANSFERASE MU 3 (PTHR11571:SF133) 6.30 C/E FERRITIN HEAVY CHAIN (PTHR11431:SF37) 6.20 + GLYCOGEN PHOSPHORYLASE, BRAIN FORM C/E 6.19 (PTHR11468:SF3) Cytoplasm +

RAB GDP DISSOCIATION INHIBITOR BETA (PTHR11787:SF1) 6.08 GLUTATHIONE S-TRANSFERASE THETA-3 (PTHR43917:SF9) 5.83 PEROXIREDOXIN-2 (PTHR10681:SF102) 5.50 INTER-ALPHA-TRYPSIN INHIBITOR HEAVY CHAIN H1 5.48 (PTHR10338:SF106) C/E AP complex subunit beta 5.46 + ADP-RIBOSYLATION FACTOR 4 (PTHR11711:SF110) 5.33 PHOSPHOGLUCOMUTASE-2 (PTHR45745:SF3) 5.22 GLUTATHIONE S-TRANSFERASE MU 2 (PTHR11571:SF144) 3.29 C/E CLATHRIN HEAVY CHAIN 1 (PTHR10292:SF7) 3.07 + RAB GDP DISSOCIATION INHIBITOR ALPHA 2.72 (PTHR11787:SF3) RAS GTPASE-ACTIVATING-LIKE PROTEIN IQGAP1 7.24 20 (PTHR14149:SF15) Cytoplasm C/E ANNEXIN A8 (PTHR10502:SF133) 6.61 membrane + L-XYLULOSE REDUCTASE (PTHR44252:SF2) 5.15 PLASTIN-2 (PTHR19961:SF35) 6.17 C/E MACROPHAGE-CAPPING PROTEIN (PTHR11977:SF13) 6.16 + C/E TROPOMYOSIN ALPHA-4 CHAIN (PTHR19269:SF40) 6.15 Cytoskeleton + MYOSIN-10 (PTHR45615:SF24) 5.98 C/E PHOSPHOGLUCOMUTASE-1 (PTHR22573:SF37) 5.74 + 122 C/M+* proteins in sConstruct sorted first by cell locality category, then by Log (FC). Column A is the rank according to the highest log (FC) of all the C/M+ proteins and the B column are other sets containing the same protein.

Cell Protein Name Log Locality A B (PANTHER family ID name: Subfamily list name) (FC) Category CYTOPLASMIC DYNEIN 1 HEAVY CHAIN 1 5.67 (PTHR10676:SF314) ALPHA-CENTRACTIN (PTHR11937:SF370) 5.08 PROCOLLAGEN-LYSINE,2-OXOGLUTARATE 5- Endoplasmic C/E 5.95 DIOXYGENASE 1 (PTHR10730:SF5) Reticulum + COLLAGEN ALPHA-3(VI) CHAIN (PTHR24020:SF13) 11.73 1 C/E- TRANSFORMING GROWTH FACTOR-BETA-INDUCED 9.46 2 C/E- PROTEIN IG-H3 (PTHR10900:SF82) COLLAGEN ALPHA-6(VI) CHAIN (PTHR22588:SF5) 9.05 4 DECORIN (PTHR45712:SF14) 8.35 6 MIMECAN (PTHR46269:SF1) 8.02 7 C/E- BIGLYCAN (PTHR45712:SF11) 7.87 8 COLLAGEN ALPHA-5(VI) CHAIN (PTHR22588:SF4) 7.68 12 BASEMENT MEMBRANE-SPECIFIC HEPARAN SULFATE 7.67 13 PROTEOGLYCAN CORE PROTEIN (PTHR10574:SF273) COLLAGEN ALPHA-1(XIV) CHAIN (PTHR24020:SF15) 7.46 15 PROLARGIN (PTHR45712:SF8) 7.45 16 C/E GLIA-DERIVED NEXIN (PTHR11461:SF48) 7.41 18 +

ASPORIN (PTHR45712:SF2) 7.14 Extracellular COLLAGEN ALPHA-2(VI) CHAIN (PTHR24020:SF29) 6.68 Matrix C/E-

COLLAGEN ALPHA-1(III) CHAIN (PTHR24023:SF604) 6.55 C/E 72 KDA TYPE IV COLLAGENASE (PTHR10201:SF29) 6.39 + PROCOLLAGEN C-ENDOPEPTIDASE ENHANCER 1 C/E 6.35 (PTHR45645:SF1) + COLLAGEN ALPHA-1(VI) CHAIN (PTHR24020:SF18) 6.13 C/E- C/E EMILIN-1 (PTHR15427:SF1) 5.98 + FIBULIN-2 (PTHR24039:SF32) 5.78 LAMININ SUBUNIT BETA-2 (PTHR10574:SF36) 5.73 C/E- FIBROMODULIN (PTHR45712:SF4) 5.72 TARGET OF NESH-SH3 (PTHR23197:SF10) 5.50 C/E STROMELYSIN-2 (PTHR10201:SF215) 5.49 + C/E SUBFAMILY NOT NAMED (PTHR24020:SF34) 3.54 + LUMICAN 9.07 Extracellular 3 C/E- Region

FIBRILLIN 1 8.78 5

123 C/M+* proteins in sConstruct sorted first by cell locality category, then by Log (FC). Column A is the rank according to the highest log (FC) of all the C/M+ proteins and the B column are other sets containing the same protein.

Cell Protein Name Log Locality A B (PANTHER family ID name: Subfamily list name) (FC) Category SERPIN FAMILY 7.71 11 ALPHA-1-ANTITRYPSIN FAMILY 7.55 14 C/E CATHEPSIN B (PTHR12411:SF16) 7.09 + VON WILLEBRAND FACTOR A DOMAIN-CONTAINING 7.00 PROTEIN 1 (PTHR24020:SF21) C/E PROSAPOSIN (PTHR11480:SF36) 6.85 + C/E GLUCOSE-6-PHOSPHATE ISOMERASE (PTHR11469:SF4) 6.84 + TENASCIN (PTHR46708:SF1) 6.60 C/E HEAT SHOCK PROTEIN BETA-6 (PTHR45640:SF2) 5.94 + XANTHINE DEHYDROGENASE/OXIDASE (PTHR11908:SF80) 5.64 PLASMINOGEN (PTHR24261:SF13) 5.60 PHOSPHATIDYLETHANOLAMINE-BINDING PROTEIN 1 5.44 (PTHR11362:SF28) COAGULATION FACTOR XIII A CHAIN (PTHR11590:SF42) 5.28 C/E BETA-GALACTOSIDASE (PTHR23421:SF61) 5.08 + ADENYLYL CYCLASE-ASSOCIATED PROTEIN 1 4.11 (PTHR10652:SF1) C/E CLUSTERIN (PTHR10970:SF1) 2.13 + DIHYDROPYRIMIDINASE-RELATED PROTEIN 2 6.43 C/E- (PTHR11647:SF56) GLUTATHIONE S-TRANSFERASE P (PTHR11571:SF214) 6.28 SUPEROXIDE DISMUTASE [MN], MITOCHONDRIAL C/E 6.18 (PTHR11404:SF6) + GLYCEROL-3-PHOSPHATE DEHYDROGENASE [NAD(+)], Mitochondria 5.68 CYTOPLASMIC (PTHR11728:SF32) VACUOLAR PROTEIN SORTING-ASSOCIATED PROTEIN 35 5.65 (PTHR11099:SF0) NADH-CYTOCHROME B5 REDUCTASE 3 (PTHR19370:SF121) 5.15 CATALASE (PTHR11465:SF9) 2.04 HISTONE H2A TYPE 2-B (PTHR23430:SF72) 7.43 17 HISTONE H2B TYPE 1-B (PTHR23428:SF28) 6.79 ANNEXIN A4 (PTHR10502:SF28) 6.61 Nucleus

SELENIUM-BINDING PROTEIN 1 (PTHR23300:SF0) 6.37 ADENINE PHOSPHORIBOSYLTRANSFERASE C/E 6.22 (PTHR32315:SF3) + CULLIN-ASSOCIATED NEDD8-DISSOCIATED PROTEIN 1 6.02 (PTHR12696:SF1) 124 C/M+* proteins in sConstruct sorted first by cell locality category, then by Log (FC). Column A is the rank according to the highest log (FC) of all the C/M+ proteins and the B column are other sets containing the same protein.

Cell Protein Name Log Locality A B (PANTHER family ID name: Subfamily list name) (FC) Category ADENOSINE KINASE (PTHR45769:SF3) 5.98 C/E PEROXIREDOXIN-6 (PTHR43503:SF4) 5.64 + TRANSKETOLASE (PTHR43195:SF3) 5.60 TRYPTOPHAN--TRNA LIGASE, CYTOPLASMIC 5.44 (PTHR10055:SF1) UDP-GLUCOSE 6-DEHYDROGENASE (PTHR11374:SF35) 5.39 PROTEIN DJ-1 (PTHR43444:SF2) 5.33 HYDROXYACYL-COENZYME A DEHYDROGENASE, 5.21 MITOCHONDRIAL (PTHR43561:SF3) C/E 5'-NUCLEOTIDASE (PTHR11575:SF25) 5.15 + ANNEXIN A6 (PTHR10502:SF19) 2.05 14-3-3 PROTEIN EPSILON (PTHR18860:SF17) 1.43 SUBFAMILY NOT NAMED (PTHR11571:SF208) 6.48 Other FATTY ACID-BINDING PROTEIN, ADIPOCYTE 6.66 (PTHR11955:SF83) 6-PHOSPHOGLUCONATE DEHYDROGENASE, 6.16 Plasma DECARBOXYLATING (PTHR11811:SF45) Membrane C/E WD REPEAT-CONTAINING PROTEIN 1 (PTHR19856:SF0) 5.27 + COATOMER SUBUNIT GAMMA-1 (PTHR10261:SF3) 5.27 * C/M+ = sConstruct to sMSC (+1) comparison set.

125 C/E+* proteins in sConstruct sorted first by cell locality category, then by Log (FC). Column A is the rank according to the highest log (FC) of all the C/E+ proteins and the B column are other sets containing the same protein.

Cell Protein Name Log (FC) Locality A B (PANTHER family ID name: Subfamily list name) Category GAMMA-ENOLASE (PTHR11902:SF10) 7.95 1 C/M+ PURINE NUCLEOSIDE PHOSPHORYLASE 6.93 6 C/M+ (PTHR11904:SF12) PHOSPHOGLYCERATE MUTASE 1 (PTHR11931:SF15) 6.91 7 C/M+ FERRITIN HEAVY CHAIN (PTHR11431:SF37) 6.30 14 C/M+ SH3 DOMAIN-BINDING GLUTAMIC ACID-RICH-LIKE 4.67 PROTEIN 3 (PTHR12232:SF3) MILK FAT GLOBULE-EGF FACTOR 8 SPLICE VARIANT 4.35 AP COMPLEX SUBUNIT BETA 3.47 C/M+ PHOSPHOGLYCERATE KINASE 1.93 C/M+ GLYCERALDEHYDE-3-PHOSPHATE DEHYDROGENASE Cytoplasm 1.93 (PTHR10836:SF81) TRIOSEPHOSPHATE ISOMERASE (PTHR21139:SF15) 1.92 FRUCTOSE-BISPHOSPHATE ALDOLASE A 1.89 (PTHR11627:SF1) GLYCOGEN PHOSPHORYLASE, BRAIN FORM 1.79 C/M+ (PTHR11468:SF3) ALPHA-ENOLASE (PTHR11902:SF12) 1.52 PHOSPHOGLUCOMUTASE-1 (PTHR22573:SF37) 1.50 C/M+ CLATHRIN HEAVY CHAIN 1 (PTHR10292:SF7) 1.39 C/M+ Pyruvate kinase 1.33 14-3-3 PROTEIN GAMMA (PTHR18860:SF22) 1.21 C/M+ TRANSGELIN-2 (PTHR18959:SF41) 6.82 8 C/M- Cytoskeleton MACROPHAGE-CAPPING PROTEIN (PTHR11977:SF13) 6.26 16 C/M+

TROPOMYOSIN ALPHA-4 CHAIN (PTHR19269:SF40) 6.26 17 C/M+ PROCOLLAGEN-LYSINE,2-OXOGLUTARATE 5- 6.05 19 C/M+ DIOXYGENASE 1 (PTHR10730:SF5) CALUMENIN (PTHR10827:SF52) 5.18 C/M- PROCOLLAGEN-LYSINE,2-OXOGLUTARATE 5- 5.03 DIOXYGENASE 2 (PTHR10730:SF6) THIOREDOXIN DOMAIN-CONTAINING PROTEIN 5 4.87 (PTHR45672:SF3) Endoplasmic DOLICHYL-DIPHOSPHOOLIGOSACCHARIDE--PROTEIN Reticulum 3.60 GLYCOSYLTRANSFERASE SUBUNIT 1 (PTHR21049:SF0) CYTOSKELETON-ASSOCIATED PROTEIN 4 3.33 C/M- (PTHR45161:SF1) PROTEIN DISULFIDE-ISOMERASE (PTHR18929:SF101) 1.75 C/M- ENDOPLASMIN (PTHR11528:SF54) 1.73 SERPIN H1 (PTHR11461:SF27) 1.56 C/M-

126 C/E+* proteins in sConstruct sorted first by cell locality category, then by Log (FC). Column A is the rank according to the highest log (FC) of all the C/E+ proteins and the B column are other sets containing the same protein.

Cell Protein Name Log (FC) Locality A B (PANTHER family ID name: Subfamily list name) Category Glucosidase II alpha subunit 5.37

STROMELYSIN-2 (PTHR10201:SF215) 5.60 Extracellular C/M+ COLLAGEN ALPHA-1(XII) CHAIN (PTHR24020:SF17) 2.95 Matrix GLIA-DERIVED NEXIN (PTHR11461:SF48) 7.52 2 C/M+ PROTEIN S100-A6 (PTHR11639:SF80) 7.20 3 CATHEPSIN B (PTHR12411:SF16) 7.20 4 C/M+ PROSAPOSIN (PTHR11480:SF36) 6.96 5 C/M+ PROTEIN S100-A11 (PTHR11639:SF60) 6.70 10 72 KDA TYPE IV COLLAGENASE (PTHR10201:SF29) 6.50 11 C/M+ PROCOLLAGEN C-ENDOPEPTIDASE ENHANCER 1 6.45 12 C/M+ (PTHR45645:SF1) CALRETICULIN (PTHR11073:SF16) 6.40 13 CATHEPSIN K (PTHR12411:SF55) 5.69 BETA-GALACTOSIDASE (PTHR23421:SF61) 5.18 Extracellular C/M+ Region SERINE PROTEASE HTRA1 (PTHR22939:SF13) 5.04 GLUCOSE-6-PHOSPHATE ISOMERASE (PTHR11469:SF4) 4.87 C/M+ GLUCOSYLCERAMIDASE (PTHR11069:SF23) 4.56 SERPIN B6 (PTHR11461:SF204) 4.47 PLASMINOGEN ACTIVATOR INHIBITOR 1 4.35 (PTHR11461:SF49) CLUSTERIN (PTHR10970:SF1) 3.10 C/M+ 60 KDA HEAT SHOCK PROTEIN, MITOCHONDRIAL 2.54 (PTHR45633:SF3) EMILIN-1 (PTHR15427:SF1) 2.23 C/M+ HEAT SHOCK PROTEIN BETA-6 (PTHR45640:SF2) 1.95 C/M+ SUBFAMILY NOT NAMED (PTHR24020:SF34) 1.00 C/M+ SUPEROXIDE DISMUTASE [MN], MITOCHONDRIAL 6.29 15 C/M+ (PTHR11404:SF6) PEROXIREDOXIN-4 (PTHR10681:SF146) 6.20 18 PROLYL 4-HYDROXYLASE SUBUNIT ALPHA-1 Mitochondria 6.00 20 (PTHR10869:SF101) TRANSLOCATOR PROTEIN (PTHR10057:SF5) 4.77 VOLTAGE-DEPENDENT ANION-SELECTIVE CHANNEL 2.41 PROTEIN 2 (PTHR11743:SF12) NUCLEASE-SENSITIVE ELEMENT-BINDING PROTEIN 1 4.67 C/M- (PTHR11544:SF68) Nucleus EUKARYOTIC INITIATION FACTOR 4A-II 5.44 (PTHR24031:SF84) 127 C/E+* proteins in sConstruct sorted first by cell locality category, then by Log (FC). Column A is the rank according to the highest log (FC) of all the C/E+ proteins and the B column are other sets containing the same protein.

Cell Protein Name Log (FC) Locality A B (PANTHER family ID name: Subfamily list name) Category 5'-NUCLEOTIDASE (PTHR11575:SF25) 5.25 C/M+ POLY(RC)-BINDING PROTEIN 1 (PTHR10288:SF96) 2.61 PEROXIREDOXIN-6 (PTHR43503:SF4) 2.34 C/M+ ADENINE PHOSPHORIBOSYLTRANSFERASE 1.74 C/M+ (PTHR32315:SF3) PROTEIN FAM180A (PTHR34034:SF2) 5.03 Other ANNEXIN A8 (PTHR10502:SF133) 6.71 9 C/M+ AP-2 COMPLEX SUBUNIT BETA (PTHR11134:SF9) 5.57 PLasma Membrane

AP-2 COMPLEX SUBUNIT ALPHA-2 (PTHR22780:SF30) 5.04 WD REPEAT-CONTAINING PROTEIN 1 (PTHR19856:SF0) 3.28 C/M+ 60S ACIDIC RIBOSOMAL PROTEIN P2 (PTHR21141:SF5) 5.18 Ribosome C/M- * C/E+ = sConstruct to sECM (+1) comparison set.

128 C/E-* proteins in sConstructs Sorted first by cell locality category, then by Log (FC). Column A is the rank according to the highest log (FC) of all the C/E- proteins and the B column shows other sets containing the same protein.

Cell Protein Name Log Locality A B (PANTHER family ID name: Subfamily list name) (FC) Category ERYTHROCYTE MEMBRANE PROTEIN BAND 4.1 LIKE 2.74 2 2 TUBULIN ALPHA-1B CHAIN (PTHR11588:SF251) 1.31 Cytoskeleton 20 C/M- VIMENTIN (PTHR45652:SF5) 0.84 C/M- MIMECAN (PTHR46269:SF1) 2.59 3 C/M+ TENASCIN-X (PTHR46708:SF3) 2.54 4 COLLAGEN ALPHA-3(VI) CHAIN (PTHR24020:SF13) 2.06 6 C/M+ LAMININ SUBUNIT BETA-2 (PTHR10574:SF36) 1.97 8 C/M+ Extracellular

DERMATOPONTIN (PTHR15040:SF2) 1.69 Matrix 12 COLLAGEN ALPHA-1(VI) CHAIN (PTHR24020:SF18) 1.69 13 C/M+ TRANSFORMING GROWTH FACTOR-BETA-INDUCED 1.60 14 C/M+ PROTEIN IG-H3 (PTHR10900:SF82) COLLAGEN ALPHA-2(VI) CHAIN (PTHR24020:SF29) 1.53 16 C/M+ LUMICAN 1.48 18 C/M+ MICROFIBRIL-ASSOCIATED GLYCOPROTEIN 4 2.00 7 (PTHR19143:SF225) COMPLEMENT C4-A-RELATED (PTHR11412:SF86) 1.73 10 Extracellular APOLIPOPROTEIN A-I (PTHR18976:SF11) 1.71 Region 11 C/M- ANNEXIN A2-RELATED (PTHR10502:SF18) 1.60 15 SPECTRIN ALPHA, NON-ERYTHROCYTIC 1 1.48 17 C/M+ DIHYDROPYRIMIDINASE-RELATED PROTEIN 2 1.43 Mitochondrial 19 C/M+ (PTHR11647:SF56) DESMIN (PTHR45652:SF2) 2.29 5 C/M- HISTONE H2A TYPE 1 (PTHR23430:SF29) 1.74 Nucleus 9

HEAT SHOCK 70 KDA PROTEIN 1A-RELATED 1.22 (PTHR19375:SF223) MYELOID-ASSOCIATED DIFFERENTIATION MARKER Plasma 3.14 1 (PTHR17068:SF3) Membrane * C/E- = sConstruct to sECM (-1) comparison set.

129 Table 7. Elevated sConstruct proteins not found in E/M+/0/-

Elevated sConstruct proteins not found in E/M+/0/-* sorted first by cell locality category, then by Log (FC). Column A is the protein set where the protein was found.

Protein Name Log Cell Locality Function (PANTHER family ID name: Subfamily A (FC) Category Category list name) GAMMA-ENOLASE (PTHR11902:SF10) 7.95 Metabolic Process C/M+,C/E+ PURINE NUCLEOSIDE PHOSPHORYLASE 6.93 Metabolic Process C/M+,C/E+ (PTHR11904:SF12) PHOSPHOGLYCERATE MUTASE 1 6.91 Glycolysis C/M+,C/E+ (PTHR11931:SF15) GLUCOSE-6-PHOSPHATE ISOMERASE 6.84 Glycolysis C/M+,C/E+ (PTHR11469:SF4) FERRITIN LIGHT CHAIN 6.38 Iron Metabolism C/M+ FERRITIN LIGHT CHAIN (PTHR11431:SF47) 6.38 Iron Metabolism C/E+ FERRITIN HEAVY CHAIN 6.30 Iron Metabolism C/M+,C/E+ (PTHR11431:SF37) GLYCOGEN PHOSPHORYLASE, BRAIN 6.19 Glycolysis C/M+,C/E+ FORM (PTHR11468:SF3) RAB GDP DISSOCIATION INHIBITOR 6.08 Cellular Transport C/M+ BETA (PTHR11787:SF1) AP COMPLEX SUBUNIT BETA 5.46 Other C/M+,C/E+ Cytoplasm PHOSPHATIDYLETHANOLAMINE- 5.44 Metabolic Process C/M+ BINDING PROTEIN 1 (PTHR11362:SF28) ADP-RIBOSYLATION FACTOR 4 5.33 Cellular Transport C/M+ (PTHR11711:SF110) PHOSPHOGLUCOMUTASE-2 5.22 Glycolysis C/M+ (PTHR45745:SF3) SH3 DOMAIN-BINDING GLUTAMIC ACID- 4.67 Antioxidant C/E+ RICH-LIKE PROTEIN 3 (PTHR12232:SF3) MILK FAT GLOBULE-EGF FACTOR 8 4.35 Cellular Transport C/E+ SPLICE VARIANT PHOSPHOGLYCERATE KINASE 2.21 Glycolysis C/M+,C/E+ GLYCERALDEHYDE-3-PHOSPHATE 1.93 Glycolysis C/E+ DEHYDROGENASE (PTHR10836:SF81) TRIOSEPHOSPHATE ISOMERASE 1.92 Glycolysis C/E+ (PTHR21139:SF15) ALPHA-ENOLASE (PTHR11902:SF12) 1.52 Glycolysis C/E+ PYRUVATE KINASE 1.33 Glycolysis C/E+ MACROPHAGE-CAPPING PROTEIN Cytoskeletal 6.26 C/M+,C/E+ (PTHR11977:SF13) Related TROPOMYOSIN ALPHA-4 CHAIN Cytoskeletal 6.26 C/M+,C/E+ (PTHR19269:SF40) Related PHOSPHOGLUCOMUTASE-1 Cytoskeleton 5.74 Glycolysis C/M+,C/E+ (PTHR22573:SF37) CYTOPLASMIC DYNEIN 1 HEAVY CHAIN 5.67 Cellular Transport C/M+ 1 (PTHR10676:SF314) ALPHA-CENTRACTIN (PTHR11937:SF370) 5.08 Cellular Transport C/M+

130 Elevated sConstruct proteins not found in E/M+/0/-* sorted first by cell locality category, then by Log (FC). Column A is the protein set where the protein was found.

Protein Name Log Cell Locality Function (PANTHER family ID name: Subfamily A (FC) Category Category list name) PROCOLLAGEN-LYSINE,2- OXOGLUTARATE 5-DIOXYGENASE 1 6.05 Collagen Related C/M+,C/E+ (PTHR10730:SF5) GLUCOSIDASE II ALPHA SUBUNIT 5.37 Endoplasmic Glycolysis C/E+ DOLICHYL- Reticulum DIPHOSPHOOLIGOSACCHARIDE-- 3.60 Protein Processing C/E+ PROTEIN GLYCOSYLTRANSFERASE SUBUNIT 1 (PTHR21049:SF0) STROMELYSIN-2 (PTHR10201:SF215) 5.49 Other C/M+,C/E+ Extracellular COLLAGEN ALPHA-1(XII) CHAIN 2.95 Matrix Collagen Related C/E+ (PTHR24020:SF17) GLIA-DERIVED NEXIN (PTHR11461:SF48) 7.52 Other C/M+,C/E+ CATHEPSIN B (PTHR12411:SF16) 7.20 ECM Modification C/M+,C/E+ PROSAPOSIN (PTHR11480:SF36) 6.96 Metabolic Process C/M+,C/E+ TENASCIN (PTHR46708:SF1) 6.60 Metabolic Process C/M+ 72 KDA TYPE IV COLLAGENASE 6.50 ECM Modification C/M+,C/E+ (PTHR10201:SF29) PROCOLLAGEN C-ENDOPEPTIDASE 6.45 Collagen Related C/M+,C/E+ ENHANCER 1 (PTHR45645:SF1) HEAT SHOCK PROTEIN BETA-6 5.94 Cellular Transport C/M+,C/E+ (PTHR45640:SF2)

CATHEPSIN K (PTHR12411:SF55) 5.69 Extracellular ECM Modification C/E+ BETA-GALACTOSIDASE Region 5.08 Glycolysis C/M+,C/E+ (PTHR23421:SF61) SERINE PROTEASE HTRA1 5.04 Protein Processing C/E+ (PTHR22939:SF13) PROTEIN FAM180A (PTHR34034:SF2) 5.03 Other C/E+ GLUCOSYLCERAMIDASE 4.56 Metabolic Process C/E+ (PTHR11069:SF23) SERPIN B6 (PTHR11461:SF204) 4.47 Protein Processing C/E+ CLUSTERIN (PTHR10970:SF1) 3.10 Cellular Transport C/M+,C/E+ Cytoskeletal EMILIN-1 (PTHR15427:SF1) 2.23 C/M+,C/E+ Related SUPEROXIDE DISMUTASE [MN], 6.29 Glycolysis C/M+,C/E+ MITOCHONDRIAL (PTHR11404:SF6) PROLYL 4-HYDROXYLASE SUBUNIT 6.00 Collagen Related C/E+ ALPHA-1 (PTHR10869:SF101) GLYCEROL-3-PHOSPHATE DEHYDROGENASE [NAD(+)], 5.68 Mitochondria Antioxidant C/M+ CYTOPLASMIC (PTHR11728:SF32) VACUOLAR PROTEIN SORTING- ASSOCIATED PROTEIN 35 5.65 Cellular Transport C/M+ (PTHR11099:SF0) TRANSLOCATOR PROTEIN 4.77 Protein Processing C/E+ (PTHR10057:SF5)

131 Elevated sConstruct proteins not found in E/M+/0/-* sorted first by cell locality category, then by Log (FC). Column A is the protein set where the protein was found.

Protein Name Log Cell Locality Function (PANTHER family ID name: Subfamily A (FC) Category Category list name) VOLTAGE-DEPENDENT ANION- SELECTIVE CHANNEL PROTEIN 2 2.41 Cellular Transport C/E+ (PTHR11743:SF12) CATALASE (PTHR11465:SF9) 2.04 Antioxidant C/M+ ADENINE PHOSPHORIBOSYLTRANSFERASE 6.22 Metabolic Process C/M+,C/E+ (PTHR32315:SF3) CULLIN-ASSOCIATED NEDD8- DISSOCIATED PROTEIN 1 6.02 Protein Processing C/M+ (PTHR12696:SF1) PEROXIREDOXIN-6 (PTHR43503:SF4) 5.64 Antioxidant C/M+,C/E+ TRYPTOPHAN--TRNA LIGASE, 5.44 Nucleus Metabolic Process C/M+ CYTOPLASMIC (PTHR10055:SF1) EUKARYOTIC INITIATION FACTOR 4A-II 5.44 mRNA C/E+ (PTHR24031:SF84) UDP-GLUCOSE 6-DEHYDROGENASE 5.39 Antioxidant C/M+ (PTHR11374:SF35) POLY(RC)-BINDING PROTEIN 1 2.61 mRNA C/E+ (PTHR10288:SF96) 14-3-3 PROTEIN EPSILON 1.43 Other C/M+ (PTHR18860:SF17) ANNEXIN A8 (PTHR10502:SF133) 6.71 Other C/M+,C/E+ AP-2 COMPLEX SUBUNIT BETA 5.57 Cellular Transport C/E+ (PTHR11134:SF9) WD REPEAT-CONTAINING PROTEIN 1 Cytoskeletal 5.27 Plasma C/M+,C/E+ (PTHR19856:SF0) Related Membrane COATOMER SUBUNIT GAMMA-1 5.27 Cellular Transport C/M+ (PTHR10261:SF3) 5'-NUCLEOTIDASE (PTHR11575:SF25) 5.25 Metabolic Process C/M+,C/E+ AP-2 COMPLEX SUBUNIT ALPHA-2 5.04 Cellular Transport C/E+ (PTHR22780:SF30) * E/M+/0/1 = sECM to sMSC (+1), (0), and (-1) comparison sets.

132 Decreased sConstructs proteins not found in E/M+/0/-* Sorted first by cell locality category, then by Log (FC). Column A is the protein set where the protein was found.

Protein Name Log Cell Locality (PANTHER family ID name: Subcategory A (FC) Category Subfamily list name) TROPOMYOSIN ALPHA-1 CHAIN Cytoskeletal 4.80 C/M- (PTHR19269:SF41) Related Cytoskeleton TUBULIN ALPHA-1B CHAIN Cytoskeletal 1.39 C/E- ,C/M- (PTHR11588:SF251) Related

APOLIPOPROTEIN A-I (PTHR18976:SF11) 2.45 Extracellular Other C/E-,C/M- ANNEXIN A1 (PTHR10502:SF17) 1.39 Region Other C/M-

DESMIN (PTHR45652:SF2) 2.14 Other C/E-,C/M- Nucleus HISTONE H2A TYPE 1 (PTHR23430:SF29) 1.91 Other C/E-,C/M- Cytoskeletal PRELAMIN-A/C (PTHR45721:SF5) 1.20 Cytoskeleton C/M- Related * E/M+/0/1 = sECM to sMSC (+1), (0), and (-1) comparison sets.

133

Figur1 1. sMSCs with surface markers as a % of total cells. A) CD90 and CD44, B) CD45 and MHC II.

Figure 17. sMSCs with surface markers as a % of total cells. A) CD90 and CD44, B) CD45 and MHC II.

134

A 60

) 40

(

D 30 A

A 7 20

0 C C C C B 60 M M M M B B B B P P P P - - t- C M c 50 S C ru M E st s Cultures Type n o

) 40 sC

(

I 30 C H p = 0.044

M 20 10

0 -… C C C ct C 60 M M M u B B B tr P P s P - - n C M o 50 S Culture CType C M E s s s ) 40

(

I I 30

C H

M 20

0 … t- C C C c D 60 M M M u B B B tr P P s P - - n C M o 50 S Culture TCype C M E s s s ) 40

(

b 30

1 p = 0.008 D 20 p = 0.023 C 10 0 C C C C BM BM BM M P -P -P PB C M ct- S C ru sM sE st on sC Culture Type

Figure 18. PBMCs with surface markers as a % of total cells Values are mean ± SEM of A) 7AAD B) MHCI, C) MHCII and D) CD 11b

135

Figure 19. Concentration of soluble inflammatory biomarkers in co-cultures Values are mean ± SEM of A) IFN-, B) TNF-, C) IL-6, and D) IL-1.

136

Figure 20. Concentration of soluble anti-inflammatory biomarkers in co-cultures Values are mean ± SEM of A) TGF-, B) IL-1ra, C) IL-1ra/IL-1

137 4.7 Figures

Figure 21. Heat map and protein abundance A) Heat map of the sMSC, sECM, and sConstruct, sConstructs co-cultured with PBMC. Most comparable samples are ordered from left to right. B) The number of significant overlapping proteins found when the sample groups were compared; sECM to sMSC (E/M+/-), sConstruct to sMSC (C/M+/-), sConstruct to sECM (C/E+/-) and sConstructs co-cultured with PBMC to sConstructs (CP/M+/-). .

138

Figure 22. E/M set protein abundance by cell locality category. Number of proteins as a percentage of number of all proteins found in the set. A) E/M+ set and B) E/M- set.

139

Figure 23. sConstruct protein abundance by cell locality category. Number of proteins as a percentage of number of all proteins found in the set. A) C/M+, B) C/M-, C) C/E+, and D) C/E-.

140

Figure 24. The union of C/M+, C/M-, C/E+, C/E-, and E/M+/0/- sets. The union of the sets C/M+, C/M-, C/E+, C/E- and E/M+0- compared using the Venn diagram. Values in parenthesis are the number of detected proteins in each set used to do the union. Only values of interest, sConstruct proteins not found in the original sECM and sMSC proteomes (compliments and intercepts of C/M+, C/M-, C/E+, C/E-), are shown.

141

Figure 25. sConstructs proteins not found in sECMs and sMSCs by cell locality and functional categories. Number of proteins as a percentage of number of all proteins found in the set. 142

CHAPTER 5: DISSERTATION SUMMARY

This summary is a synopsis of the conclusions of: Reisbig N, Hussein H, Pinnell E, Bertone A. Evaluation of equine synovial-derived extracellular matrix scaffolds seeded with equine synovial-derived mesenchymal stem cells. Am J Vet Res. 2018;79(1):124–33.

Reisbig N, Pinnell E, Scheuerman L, Bertone AL. (2018) Synovium Constructs Stimulate Chondrogenesis In Vitro and Cartilage Healing In Vivo. Under review, PLoS ONE, 2018.

Reisbig N, and Bertone AL. (2018) Immune and Signaling Proteins of Allogeneic Stem Cell-Extracellular Matrix Scaffold Interactions. Submitted, PLoS ONE,, 2018

143 The purpose of the studies reported here was to investigate a novel alternative to the use of biological scaffolds; rather than the traditional use of scaffolds as a filler of a defect here we focus on using it as a transplant placed in direct apposition to the cartilage injury.

In this way the scaffold can serve as a biological booster for focal regeneration.

A viable engrafted cell-scaffold construct was generated from sECM and sMSCs that secrete anabolic agents into soluble medium in vitro, with and without genetic engineered sMSCs (Chapter 2). It was conclusively shown that sConstructs enhance chondrogenesis in vitro and cartilage healing in vivo (Chapter 3). When using allogenic PBMCs in a co- culture in vitro, sConstructs caused a low level of inflammatory response as compared with an immunologic activator. Also, the sMSCs decreased the immune response indicating that the sMSCs, as they are integrating into the sECMs, results in a less immunogenic sConstruct. Proteomics of the sConstructs substantiate the above conclusions (Chapter 4).

This is a “proof of concept”; a sConstruct has the potential to function as a vehicle creating a microenvironment supporting the healing of cartilage in damaged joints with minimal immune response.

By using the decellularized synovial scaffolds, sECMs, we successfully seeded the sECM with allogeneic synovial-origin cells, sMSCs, by use of a serum gradient (Chapter 2). By genetically engineering sMSCs we showed that it was not detrimental to the migration of the sMSCs in the sECM and that the engineered sMSCs produced high levels of the inserted protein, in this case BMP-2.

144 Inducing cell migration into ECMs can be challenging. Numerous techniques have been explored, including an electric charge, perfusion, and pressure [22]. In this study a 30% chemotactic gradient (from 10% to 40% FBS) to induce sMSC migration (similarly to another study conducted by our laboratory group [22]). The use of the combined chemotactic gradient and osmotic pressure successfully seeded sMSCs in sECMs. The migration pattern of the cells was consistent and repeatable pattern over time.

Maturation and transformation of the sMSCs was seen and BMP-2 transduced sMSCs increased their production of HA and PG (markers for extracellular component production), with a concurrent loss of CD90 (loss of stem cell characteristics). As expected, the transduced sConstructs had a greater concentration of BMP-2 in the medium, suggesting that these cells would be able to increase BMP-2 concentrations in synovial fluid in vivo.

Significant cell death was seen early on after seeding, but by day 14, sMSCs had increased 2.5-fold. At this point the sMSCs were found throughout the sECM with a biological signature of mature synovial cells. The cells were producing anabolic growth factors known to positively influence cartilage repair.

For these studies, the term sMSC was used as the description for the synovium-derived cells. These cells came from synovium, adhered to plastic (an inclusion criterion for mesenchymal cells), and expressed CD90 (> 70% expression at the start of the study). In later experiments (Chapter 4) these cells were further characterized and found to express

145 the traditional markers for MSCs; high levels of cell surface markers like CD90 and

CD44 while expressing low levels of others, like CD45 and MHC II [138].

CD90 expression significantly decreased as the sMSCs migrated into the sECMs. This could be due to sMSC differentiation induced by the matrix of the scaffold. These results are similar to those found in another study [71] using a collagen scaffold.

In vivo, HA is produced by synoviocytes and released into the joint and plays important roles both in the transport of nutrients to cartilage and protection of articular cartilage. An increase in HA production supports sMSC biological activity and is unique to mature synovial-fibroblast cells. The combination of HA production and the decrease of CD90 expression supported the synovial phenotype of the sMSCs. In addition, the PG production was indicative of a synovial lineage, phenotype, and function. Transduced sMSCs had significantly greater concentrations of PG and HA than the constructs of sECMs with untreated sMSCs, hence there was a upregulation of cell metabolism or targeted matrix production (or both).

The concentration of BMP-2 in the media was greatest in the sConstructs seeded from transduced sMSCs. The greatest BMP-2 production was found with the transduced sMSCs and, in addition, this group also had greater HA and PG concentrations. This indicates increased gene production and enhanced synovial function. Although the sECM alone had the lowest concentration of BMP-2, it yielded a level of BMP-2. This supports the potential of the biological scaffold, even without cell seeding, of inducing paracrine and autocrine responses. During the 14-day co-culture the BMP-2 concentration

146 decreased slightly, although not significantly, in all groups (including unseeded sECMs) and may be due to leaching of the endogenous BMP-2 from the sECM.

Considering that a BMP-2 concentration in the range of 170 to 180 ng/mL can have physiologic importance for bone formation or chondrogenesis [60,61], if the sConstruct was placed in the synovium close to the damaged cartilage, the local BMP-2 concentration may have a positive effect on surrounding cells and cartilage healing.

Cartilage metaplasia was not seen in the synovial constructs and was also not seen in the in vivo rat model (Chapter 3).

By successfully seeding the sECM with sMSCs we proved that a bioactive sConstruct could be made and that it has the potential to improve the joint environment and healing in injured joints or joints with OA.

The natural next phase was to investigate the effect of the sConstruct on chondrocytes and vice versa, first in vitro and then in vivo (Chapter 3). For the in vitro study a co- culture with chondrocytes was performed. In vivo work was performed on a cartilage damage model in athymic rats (immunosuppressed). After 14 days the sConstruct-BMP-2 co-cultures boosted the chondrocyte maturation and intracellular Col II production. At this point in the sConstruct-BMP-2 co-culture, chondrocyte intracellular Col II levels were > 2-fold that of any other groups. Within 3 days of incubation, co-cultured chondrocytes morphology reverted to that of a typical hypertrophied cell or chondroblast.

The sConstruct-BMP-2’s sMSCs also produced more HA and had higher cell counts than untransduced-sConstructs. In vivo, the effect of BMP-2 was also observed with

147 sConstruct-BMP-2 implants having considerably improved gross anatomy, adjacent articulate cartilage growth and subchondral bone repair over that of other implants. The morphological changes of the damaged cartilage were by far the most pronounced in the

BMP-2-sConstruct group.

The BMP-2 effects observed here are similar to those seen in previous studies. BMPs have been shown to be involved in stimulating MSC differentiation and cell recruitment, along with other functions in chondrogenesis and osteogenesis [87,88]. As previously seen during seeding (Chapter 2) no sign of any negative effects of BMP-2 on cartilage

[53], no bone growth or overgrowth into the synovium with implanted BMP-2- sConstructs was noted.

In vivo, the sECM implants caused a some (significant) increase of lesion fill in addition to migration of endogenous cells into the sECM scaffold. This may be due to increased local BMP-2 from the sECM alone (seen in Chapter 2) producing a growth environment.

The levels of HA and PG production also increased in sConstruct co-cultures with chondrocytes.

Both the in vitro and the in vivo studies showed that BMP-2-sConstruct had the highest healing potential. Also, important to note is that the in vitro and in vivo results showed a similar trend; BMP-2-sConstructs > GFP-sConstructs or untransduced-sConstructs > sECM. The results in both the in vivo and in vitro strongly support a feedback mechanism between sConstructs and chondrocytes. The sConstruct’s anabolic biofactors resulted in a paracrine effect with restoration and maturation responses of chondrocytes that, in turn,

148 impacted the sConstruct’s own maturation and biofactor output. In vitro, sConstruct- chondrocyte co-cultures showed increased chondrocyte and sMSC cell counts, increased intracellular chondrocyte Collagen II and increased production of soluble HA and BMP-

2, respectively. In vivo, rat knee lesions with sConstruct implants, showed markedly improved articular cartilage growth and subchondral bone repair. Correspondingly, resected sConstruct implants had substantial morphological changes with a marked increased cell integration.

The positive results, both in vivo and in vitro, had one major limitation; none of the work had been performed with the influence of the immune system. Using a similar system as used with the chondrocyte co-culture, the chondrocytes were replaced with PBMCs and incubated with the sConstructs for 48h (Chapter 4).

When co-cultured with allogenic PBMCs in vitro, sConstructs caused a low level of inflammatory response when compared with Lipopolysaccharide (LPS) an immunologic activator. A distinct pattern was seen in the level of response: with sECMs the highest, then sMSCs and last sConstructs. Similarly, the impact of the PBMCs on sConstructs only slightly increased the production of anti-inflammatory markers and cytokines. This led to the conclusion that the sMSCs, as they are growing into the sECMs, are becoming less immunogenic. The proteomics of the sConstructs resulted in similar conclusions.

Only few changes were noted when comparing proteomes of sConstructs to sConstructs co-cultured with PBMCs. When comparing sConstructs to its original constituents there were significant changes and an increase in sMSC metabolic proteins in the sConstructs was noted. 149 The overall goal for the study was to determine whether decellularized synovial scaffolds could be seeded with cells derived from the synovium and could serve to deliver anabolic proteins locally. Rather than the traditional use of scaffolds, as a defect filler, the scaffold in this application would serve as a vehicle for regenerative cells and anabolic proteins placed in juxtaposition to the damaged cartilage. The present investigation was designed to provide “proof of concept” for the generation of a biological viable scaffold in a laboratory setting. Significant work, both in vitro and in vivo, before this could be considered for manufacturing and commercial use. The study involved the use of cells and matrix of synovial origin to retain characteristics supporting a synovial phenotype by use of routine laboratory cell culture techniques that are highly repeatable.

Although we choose sMSCs due to the phenotype and ability to produce metabolic agents that supports cartilage the same procedure would most probably be successful with other types of MSCs. Comparison in vitro of MSCs from different sources, such as bone marrow derived MSCs, blood derived MSCs and fetal MSCs, would give valuable information on the specific potentials of each of these groups. Similarly, only transduced

BMP-2 sMSCs were studied. Other anabolic growth factors know to support chondrogenesis, such as IGF-1, could prove more beneficial. Even combinations of different MSCs and MSCs transduced with different anabolic agents might demonstrate the best effect.

A limitation was the use of athymic nude rats to eliminate any contribution that might arise from an immunologic reaction by the equine sConstructs. MSCs have been recognized to be 'immune privileged' which is thought to enable MSC transplantation 150 across major histocompatibility barriers and may result in the creation of “off-the-shelf” therapies consisting of MSCs grown in culture [28–30]. However, there have been some clinical trials reporting an immune response [28]. Similarly, decellularized ECM scaffolds, appropriately prepared (DNA concentration and base pair length [31]), can be implanted as xenogeneic grafts [32,33]. The combination of sECM scaffold and sMSC’s in the sConstruct have great potential as a hypoimmunogenic product that could be used allogenically. For this a in vivo model to further study the immune component of syngeneic verses allogenic sConstructs, MSCs should be thoroughly tested in immunocompetent models. Understanding the mechanisms of scaffold-cell interactions could refine the treatment process by balancing anabolic and anti-inflammatory pathways. The proteomic data on the sConstructs and constituents will need further analyzing to clearly understand the pathways that will affect healing of cartilage and modulation of the joint environment.

151

REFERENCES

152

1. Helmick CG W-CS. United States Bone and Joint Initiative: The Burden of Musculoskeletal Diseases in the United States (BMUS). 2018. 2. Rossdale PD, Hopes R, Winffield-Digby JN, Al E. Epidemiological study of wastage amoung racehorses. Am J Vet Res. 1985;116(3):66–9. 3. Hunziker EB. The elusive path to cartilage regeneration. Adv Mater. 2009;21(32– 33):3419–24. 4. Kreuz PC, Erggelet C, Steinwachs MR, Krause SJ, Lahm A, Niemeyer P, et al. Is Microfracture of Chondral Defects in the Knee Associated With Different Results in Patients Aged 40 Years or Younger? Arthrosc - J Arthrosc Relat Surg. 2006;22(11):1180–6. 5. Walker-Bone, K., Javaid, K., Arden, N., & Cooper C. Medical management of osteoarthritis. Br Med J. 2000;321(7266):936–40. 6. Frisbie DD, Stewart MC. Cell-based therapies for equine joint disease. Vet Clin North Am Equine Pract. 2011 Aug;27(2):335–49. 7. Hohmann, Erik Tetsworth K. Large osteochondral lesions of the femoral condyles: Treatment with fresh frozen and irradiated allograft using the Mega OATS technique. Knee. 2016;23(3):436–41. 8. Vangsness T, Farr J, Boyd J, Dellaero DT, Mills CR, Leroux-Williams M. Adult Human Mesenchymal Stem Cells Delivered via Intra-Articular Injection to the Knee Following Partial Medial Meniscectomy. J Bone Jt Surg. 2014;96:90–8. 9. Emadedin M, Aghdami N, Taghiyar L, Fazeli R, Moghadasali R, Jahangir S, et al. Intra-articular injection of autologous mesenchymal stem cells in six patients with knee Osteoarthritis. Arch Iran Med. 2012;15(7):422–8. 10. Pigott JH, Ishihara A, Wellman ML, Russell DS, Bertone AL. Investigation of the immune response to autologous, allogeneic, and xenogeneic mesenchymal stem cells after intra-articular injection in horses. Vet Immunol Immunopathol. 2013;156(1–2):99–106. 11. Kock L, van Donkelaar C, Ito K. Tissue engineering of functional articular cartilage: the current status. Cell Tissue Res. 2012;347:613–627. 12. Rai V, Dilisio MF, Dietz NE, Agrawal DK. Recent strategies in cartilage repair: A systemic review of the scaffold development and tissue engineering. J Biomed Mater Res - Part A. 2017;105(8):2343–54. 13. Whitney KE, Liebowitz A, Bolia IK, Chahla J, Ravuri S, Evans TA, et al. Current perspectives on biological approaches for osteoarthritis. Ann N Y Acad Sci. 2017;1410(1):26–43. 14. Kiyotake EA, Beck EC, Detamore MS. Cartilage extracellular matrix as a biomaterial for cartilage regeneration. Ann N Y Acad Sci. 2016;1383(1):139–59.

153 15. Sridharan BP, Laflin AD, Holtz MA, Pacicca DM, Wischmeier NK, Detamore MS. In vivo evaluation of stem cell aggregates on osteochondral regeneration. J Orthop Res. 2017;35(8):1606–16. 16. Reisbig NA, Hussein HA, Pinnell E, Bertone AL. Evaluation of equine synovial- derived extracellular matrix scaffolds seeded with equine synovial-derived mesenchymal stem cells. Am J Vet Res. 2018;79(1):124–33. 17. Brown BN, Badylak SF. Extracellular Matrix as an Inductive Scaffold for Functional Tissue Reconstruction. Transl Regen Med to Clin. 2015;11–29. 18. Lee CM, Kisiday JD, McIlwraith CW, Grodzinsky AJ, Frisbie DD. Synoviocytes protect cartilage from the effects of injury in vitro. BMC Musculoskelet Disord. 2013;14(1):54. 19. Caplan AI. New MSC : MSCs as Pericytes Are Sentinels and Gatekeepers. 2017;(December 2016):10–3. 20. Sakaguchi Y, Sekiya I, Yagishita K, Muneta T. Comparison of human stem cells derived from various mesenchymal tissues: Superiority of synovium as a cell source. Arthritis Rheum. 2005;52(8):2521–9. 21. Shirasawa S, Sekiya I, Sakaguchi Y, Yagishita K, Ichinose S, Muneta T. In vitro chondrogenesis of human synovium-derived mesenchymal stem cells: Optimal condition and comparison with bone marrow-derived cells. J Cell Biochem. 2006;97(1):84–97. 22. Gorsline RT, Tangkawattana P, Lannutti JJ, Yamaguchi M, Kaeding CC, Bertone AL. Accelerated chondrogenesis in nanofiber polymeric scaffolds embedded with BMP-2 genetically engineered chondrocytes. J Biomed Sci Eng [Internet]. 2010;03(09):908–16. Available from: http://www.scirp.org/journal/doi.aspx?DOI=10.4236/jbise.2010.39121 23. Chen SS, Fitzgerald W, Zimmerberg J, Kleinman HK, Margolis L. Cell-cell and cell-extracellular matrix interactions regulate embryonic stem cell differentiation. Stem Cells. 2007;25(3):553–61. 24. Ingber DE. Mechanical signaling and the cellular response to extracellular matrix in angiogenesis and cardiovascular physiology. Vol. 91, Circulation Research. 2002. p. 877–87. 25. Bissell MJ, Barcellos-Hoff MH. The influence of extracellular matrix on gene expression: is structure the message? J Cell Sci Suppl. 1987;8:327–43. 26. Reisbig NA, Hussein HA, Pinnell E, Bertone AL. Comparison of four methods for generating decellularized equine synovial extracellular matrix. Am J Vet Res. 2016;77(12):1332–9. 27. Reisbig NA, Pinnell E, Scheuerman L, Bertone AL. Synovium Constructs Stimulate Chondrogenesis In Vitro and Cartilage Healing In Vivo. PLoS One. 2018;In review. 28. Ankrum JA, Ong JF, Karp JM. Mesenchymal stem cells: Immune evasive, not 154 immune privileged. Nat Biotechnol. 2014;32(3):252–60. 29. Chen X, Armstrong MA, Li G. Mesenchymal stem cells in immunoregulation. Immunol Cell Biol [Internet]. 2006;84(5):413–21. Available from: http://www.scopus.com/inward/record.url?eid=2-s2.0- 33747839997&partnerID=40&md5=e5be044b9a27204bff05e9182ae689ce 30. Ryan JM, Barry FP, Murphy JM, Mahon BP. Mesenchymal stem cells avoid allogeneic rejection. J Inflamm. 2005;2. 31. Mecham RP. Overview of extracellular matrix. Curr Protoc Cell Biol. 2012;(SUPPL.57). 32. Bigham AS, Dehghani SN, Shafiei Z, Torabi Nezhad S. Xenogenic demineralized bone matrix and fresh autogenous cortical bone effects on experimental bone healing: Radiological, histopathological and biomechanical evaluation. J Orthop Traumatol. 2008;9(2):73–80. 33. Fan X, Chen P, Fu Y. [Xenogenic corneal acellular matrix as carrier for reconstruction of biological cornea epithelium-scaffold-endothelium compound]. Zhonghua Yan Ke Za Zhi [Internet]. 2007;43(5):437–41. Available from: http://ovidsp.ovid.com/ovidweb.cgi?T=JS&PAGE=reference&D=med5&NEWS= N&AN=17706094 34. Reisbig NA, Bertone AL. Immune and Signaling Proteins of Allogeneic Stem Cell-Extracellular Matrix Scaffold Interactions. PLoS One. :Submitted. 35. Erickson DM, Harris SE, Dean DD, Harris MA, Wozney JM, Boyan BD, et al. Recombinant bone morphogenetic protein (BMP)-2 regulates costochondral growth plate chondrocytes and induces expression of BMP-2 and BMP-4 in a cell maturation-dependent manner. J Orthop Res. 1997;15(3):371–80. 36. Hiligsmann M, Cooper C, Arden N, Boers M, Branco JC, Luisa Brandi M, et al. Health economics in the field of osteoarthritis: An Expert’s consensus paper from the European Society for Clinical and Economic Aspects of Osteoporosis and Osteoarthritis (ESCEO). Semin Arthritis Rheum. 2013;43(3):303–13. 37. Widmer WR, Buckwalter KA, Hill MA, Fessler JF, Ivancevich S. A technique for magnetic resonance imaging of equine cadaver specimens. Vet Radiol Ultrasound. 1999;40(1):10–4. 38. Bertone AL, Pittman DD, Bouxsein ML, Li J, Clancy B, Seeherman HJ. Adenoviral-mediated transfer of human BMP-6 gene accelerates healing in a rabbit ulnar osteotomy model. J Orthop Res. 2004;22(6):1261–70. 39. Li WJ, Tuli R, Okafor C, Derfoul A, Danielson KG, Hall DJ, et al. A three- dimensional nanofibrous scaffold for cartilage tissue engineering using human mesenchymal stem cells. Biomaterials. 2005;26(6):599–609. 40. Badylak SF. Xenogeneic extracellular matrix as a scaffold for tissue reconstruction. Transpl Immunol. 2004;12(3–4):367–77. 41. French KM, Boopathy A V., Dequach JA, Chingozha L, Lu H, Christman KL, et 155 al. A naturally derived cardiac extracellular matrix enhances cardiac progenitor cell behavior in vitro. Acta Biomater. 2012;8(12):4357–64. 42. Crapo, Peter; Gilbert, Thomas; Badylak S. An overview of tissue and whole organ dece. Biomaterials [Internet]. 2011 Apr;29(6):997–1003. Available from: http://www.embase.com/search/results?subaction=viewrecord&from=export&id= L51263258%5Cnhttp://dx.doi.org/10.1016/j.biomaterials.2011.01.057 43. Sellaro TL, Ravindra AK, Stolz DB, Badylak SF. Maintenance of Hepatic Sinusoidal Endothelial Cell Phenotype In Vitro Using Organ-Specific Extracellular Matrix Scaffolds. Tissue Eng [Internet]. 2007;13(9):2301–10. Available from: http://www.liebertonline.com/doi/abs/10.1089/ten.2006.0437 44. Sellaro TL, Ranade A, Faulk DM, McCabe GP, Dorko K, Badylak SF, et al. Maintenance of human hepatocyte function in vitro by liver-derived extracellular matrix gels. Tissue Eng Part A [Internet]. 2010;16(3):1075–82. Available from: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2863084&tool=pmcen trez&rendertype=abstract 45. Allen RA, Seltz LM, Jiang H, Kasick RT, Sellaro TL, Badylak SF, et al. Adrenal extracellular matrix scaffolds support adrenocortical cell proliferation and function in vitro. Tissue Eng Part A [Internet]. 2010;16(11):3363–74. Available from: http://www.ncbi.nlm.nih.gov/pubmed/20528677 46. Zhang Y, He Y, Bharadwaj S, Hammam N, Carnagey K, Myers R, et al. Tissue- specific extracellular matrix coatings for the promotion of cell proliferation and maintenance of cell phenotype. Biomaterials. 2009;30(23–24):4021–8. 47. Cortiella J, Niles J, Cantu A, Brettler A, Pham A, Vargas G, et al. Influence of Acellular Natural Lung Matrix on Murine Embryonic Stem Cell Differentiation and Tissue Formation. Tissue Eng Part A [Internet]. 2010;16(8):2565–80. Available from: http://www.liebertonline.com/doi/abs/10.1089/ten.tea.2009.0730 48. Brennan EP, Tang XH, Stewart-Akers AM, Gudas LJ, Badylak SF. Chemoattractant activity of degradation products of fetal and adult skin extracellular matrix for keratinocyte progenitor cells. J Tissue Eng Regen Med. 2008;2(8):491–8. 49. Crapo PM, Medberry CJ, Reing JE, Tottey S, van der Merwe Y, Jones KE, et al. Biologic scaffolds composed of central nervous system extracellular matrix. Biomaterials. 2012;33(13):3539–47. 50. Beattie AJ, Gilbert TW, Guyot JP, Yates AJ, Badylak SF. Chemoattraction of Progenitor Cells by Remodeling Extracellular Matrix Scaffolds. Tissue Eng Part A [Internet]. 2009;15(5):1119–25. Available from: http://www.liebertonline.com/doi/abs/10.1089/ten.tea.2008.0162 51. Hutmacher D. Sca!olds in tissue engineering bone and cartilage. Biomaterials. 1904;21:1–15. 52. Ioku K, Kamitakahara M, Ikeda T. Functional Biomaterials for Bone Regeneration. J Biomed Mater Res Part A. 2012;783(2010):2728. 156 53. Ishihara A, Zachos TA, Bartlett JS, Bertone AL. Evaluation of permissiveness and cytotoxic effects in equine chondrocytes, synovial cells, and stem cells in response to infection with adenovirus 5 vectors for gene delivery. Am J Vet Res. 2006;67(7):1145–55. 54. Cao L, Yang F, Liu G. The promotion of cartilage defect repair using adenovirus mediated Sox9 gene. Biomaterials. 2011;32:3910–3920. 55. Ng VY, Jump SS, Santangelo KS, Russell DS, Bertone AL. Genetic engineering of juvenile human chondrocytes improves scaffold-free mosaic neocartilage grafts basic research. Clin Orthop Relat Res. 2013;471(1):26–38. 56. Jung DY, Yamada T, Tsuchiya T, Ryu SC, Han DW. Effects of cell-seeding methods of human osteoblast culture on biomechanical properties of porous bioceramic scaffold. Biotechnol Bioprocess Eng. 2010;15(2):341–8. 57. Faulk DM, Wildemann JD, Badylak SF. Decellularization and cell seeding of whole liver biologic scaffolds composed of extracellular matrix. Vol. 5, Journal of Clinical and Experimental Hepatology. 2015. p. 69–80. 58. Rosengren S, Boyle DL, Firestein GS. Acquisition, culture, and phenotyping of synovial fibroblasts. Methods Mol Med. 2007 Jan;135:365–75. 59. Cagnard N, Letourneur F, Essabbani A, Devauchelle V, Mistou S, Rapinat A, et al. Interleukin-32, CCL2, PF4F1 and GFD10 are the only cytokine/chemokine genes differentially expressed by in vitro cultured rheumatoid and osteoarthritis fibroblast-like synoviocytes. Eur Cytokine Netw. 2005;16(4):289–92. 60. Seeherman H, Li R, Bouxsein M, Kim H, Li XJ, Smith-Adaline EA, et al. rhBMP- 2/calcium phosphate matrix accelerates osteotomy-site healing in a nonhuman primate model at multiple treatment times and concentrations. J Bone Jt Surg - Ser A. 2006;88(1):144–60. 61. Vanhatupa S, Ojansivu M, Autio R, Juntunen M, Miettinen S. Bone Morphogenetic Protein-2 Induces Donor-Dependent Osteogenic and Adipogenic Differentiation in Human Adipose Stem Cells. Stem Cells Transl Med [Internet]. 2015;4(12):1391–402. Available from: http://doi.wiley.com/10.5966/sctm.2015- 0042 62. Šoltés L, Mendichi R, Lath D, Mach M, Bakoš D. Molecular characteristics of some commercial high-molecular-weight hyaluronans. Biomed Chromatogr. 2002;16(7):459–62. 63. Ranera B, Lyahyai J, Romero A, José F, Rosa A, Luisa M, et al. Veterinary Immunology and Immunopathology Immunophenotype and gene expression profiles of cell surface markers of mesenchymal stem cells derived from equine bone marrow and adipose tissue. Vet Immunol Immunopathol [Internet]. 2011;144(1–2):147–54. Available from: http://dx.doi.org/10.1016/j.vetimm.2011.06.033 64. Crovace A, Lacitignola L, Rossi G, Francioso E. Histological and immunohistochemical evaluation of autologous cultured bone marrow 157 mesenchymal stem cells and bone marrow mononucleated cells in collagenase- induced tendinitis of equine superficial digital flexor tendon. Vet Med Int. 2010;2010. 65. Crosta P. What Are Stem Cells? Stem Cells [Internet]. 2014;Accessed M:53–68. Available from: https://linkinghub.elsevier.com/retrieve/pii/B9780124115514000039 66. stem cell. (n.d.). The American Heritage® New Dictionary of Cultural Literacy, Third Edition. [Internet]. [cited 2017 Jul 3]. Available from: http://www.dictionary.com/browse/stem-celldictionary.com 67. Fan J, Varshney RR, Ren L, Cai D, Wang D-A. Synovium-Derived Mesenchymal Stem Cells: A New Cell Source for Musculoskeletal Regeneration. Tissue Eng Part B Rev [Internet]. 2009;15(1):75–86. Available from: http://www.liebertonline.com/doi/abs/10.1089/ten.teb.2008.0586 68. De Bari C, Dell’Accio F, Tylzanowski P, Luyten FP. Multipotent mesenchymal stem cells from adult human synovial membrane. Arthritis Rheum. 2001;44(8):1928–42. 69. Fan J, Ren L, Liang R, Gong Y, Cai D, Wang DA. Chondrogenesis of synovium- derived mesenchymal stem cells in photopolymerizing hydrogel scaffolds. J Biomater Sci Polym Ed. 2010;21(12):1653–67. 70. Fellows CR, Matta C, Zakany R, Khan IM, Mobasheri A. Adipose, bone marrow and synovial joint-derived mesenchymal stem cells for cartilage repair. Front Genet. 2016;7(DEC):1–20. 71. Trappmann B, Gautrot J, Connelly J. Extracellular-matrix tethering regulates stem- cell fate.pdf. Nat Mater. 2012;11:742–742. 72. Dormond E, Perrier M, Kamen A. From the first to the third generation adenoviral vector: What parameters are governing the production yield? Biotechnol Adv. 2009;27(2):133–44. 73. Osago H, Shibata T, Hara N, Kuwata S, Kono M, Uchio Y, et al. Quantitative analysis of glycosaminoglycans, chondroitin/dermatan sulfate, hyaluronic acid, heparan sulfate, and keratan sulfate by liquid chromatography-electrospray ionization-tandem mass spectrometry. Anal Biochem. 2014;467:62–74. 74. Li J, Kim KS, Park JS, Elmer WA, Hutton WC, Yoon ST. BMP-2 and CDMP-2: Stimulation of chondrocyte production of proteoglycan. J Orthop Sci. 2003;8(6):829–35. 75. Information S. Harnessing Traction-Mediated Manipulation of the Cell-Matrix Interface to Control Stem Cell Fate Supplemental Materials. Nat Mater. 2010;9:1– 34. 76. Naito H, Dohi Y, Zimmermann W-H, Tojo T, Takasawa S, Eschenhagen T, et al. The Effect of Mesenchymal Stem Cell Osteoblastic Differentiation on the Mechanical Properties of Engineered Bone-Like Tissue. Tissue Eng Part A [Internet]. 2011;17(17–18):2321–9. Available from: 158 http://www.liebertonline.com/doi/abs/10.1089/ten.tea.2011.0099 77. Zachos T, Diggs A, Weisbrode S, Bartlett J, Bertone A. Mesenchymal stem cell- mediated gene delivery of bone morphogenetic protein-2 in an articular fracture model. Mol Ther. 2007;15(8):1543–50. 78. Ortved K, Wagner B, Calcedo R, Wilson J, Schaefer D, Nixon A. Humoral and Cell-Mediated Immune Response, and Growth Factor Synthesis After Direct Intraarticular Injection of rAAV2-IGF-I and rAAV5-IGF-I in the Equine Middle Carpal Joint. Hum Gene Ther [Internet]. 2015;26(3):161–71. Available from: http://online.liebertpub.com/doi/10.1089/hum.2014.050 79. Goodrich LR, Grieger JC, Phillips JN, Khan N, Gray SJ, McIlwraith CW, et al. ScAAVIL-1ra dosing trial in a large animal model and validation of long-term expression with repeat administration for osteoarthritis therapy. Gene Ther. 2015;22(7):536–45. 80. Mardones R, Jofré CM, Tobar L, Minguell JJ. Mesenchymal stem cell therapy in the treatment of hip osteoarthritis. J Hip Preserv Surg [Internet]. 2017;4(2):159– 63. Available from: https://academic.oup.com/jhps/article- lookup/doi/10.1093/jhps/hnx011 81. Frisbie DD, Smith RKW. Clinical update on the use of mesenchymal stem cells in equine orthopaedics. Equine Vet J. 2010;42(1):86–9. 82. Henderson SE, Santangelo KS, Bertone AL. Chondrogenic effects of exogenous retinoic acid or a retinoic acid receptor antagonist (LE135) on equine chondrocytes and bone marrow-derived mesenchymal stem cells in monolayer culture. Am J Vet Res. 2011;72:884–92. 83. Gerwin N, Bendele AM, Glasson S, Carlson CS. The OARSI histopathology initiative - recommendations for histological assessments of osteoarthritis in the rat. Osteoarthr Cartil. 2010;18:24–34. 84. Ishihara A, Shields KM, Litsky AS, Mattoon JS, Weisbrode SE, Bartlett JS, et al. Osteogenic gene regulation and relative acceleration of healing by adenoviral- mediated transfer of human BMP-2 or -6 in equine osteotomy and ostectomy models. J Orthop Res. 2008;26(6):764–71. 85. Moojen DJ, Saris DB, Auw Yang KG, Dhert WJ VA. The correlation and reproducibility of histological scoring systems in cartilage repair. Tissue Eng. 2002;8:627–634. 86. Orth P, Zurakowski D, Wincheringer D, Madry H. Reliability, Reproducibility, and Validation of Five Major Histological Scoring Systems for Experimental Articular Cartilage Repair in the Rabbit Model. Tissue Eng Part C Methods. 2012;18(5):329–39. 87. Kim B-S, Yu D-A, Han J. Stimulation of Chondrogenic Differentiation of Mesenchymal Stem Cells. Int J Stem Cells [Internet]. 2012;5(1). Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3840987/pdf/ijsc-05-16.pdf 88. Luu HH, Song W-X, Luo X, Manning D, Luo J, Deng Z-L, et al. Distinct Roles of 159 Bone Morphogenetic Proteins in Osteogenic Differentiation of Mesenchymal Stem Cells. Wiley Intersci. 2007;11(4):1609–12. 89. Badylak SF. The extracellular matrix as a biologic scaffold material. Biomaterials. 2007;28(25):3587–93. 90. Shu B, Zhang M, Xie R et al. BMP2, but not BMP4, is crucial for chondrocyte proliferation and maturation during endochondral bone development. J Cell Sci. 2011;124(20):3428–40. 91. Badylak SF, Freytes DO, Gilbert TW. Extracellular matrix as a biological scaffold material: Structure and function. Vol. 5, Acta Biomaterialia. 2009. p. 1–13. 92. Pei M, He F, Vunjak-Novakovic G. Synovium-derived stem cell-based chondrogenesis. Differentiation [Internet]. 2008;76(10):1044–56. Available from: http://dx.doi.org/10.1111/j.1432-0436.2008.00299.x 93. Rhodes JM, Simons M. The extracellular matrix and blood vessel formation: not just a scaffold. J Cell Mol Med. 2007;11(2):176–205. 94. Wolf MT, Daly KA, Reing JE, Badylak SF. Biologic scaffold composed of skeletal muscle extracellular matrix. Biomaterials. 2012;33(10):2916–25. 95. Cushing MC, Jaeggli MP, Masters KS, Leinwand LA, Anseth KS. Serum deprivation improves seeding and repopulation of acellular matrices with valvular interstitial cells. J Biomed Mater Res - Part A. 2005;75(1):232–41. 96. Sabetkish S, Kajbafzadeh AM, Sabetkish N, Khorramirouz R, Akbarzadeh A, Seyedian SL, et al. Whole-organ tissue engineering: Decellularization and recellularization of three-dimensional matrix liver scaffolds. J Biomed Mater Res - Part A. 2015;103(4):1498–508. 97. Brown JA, Holcombe SJ, Southwood LL, Byron CR, Embertson RM, Hauptmann JG. End-to-Side Versus Side-to-Side Jejunocecostomy in Horses: A Retrospective Analysis of 150 Cases. Vet Surg. 2015;44(4):527–33. 98. Crapo PM, Gilbert TW, Badylak SF. An overview of tissue and whole organ decellularization processes. Biomaterials. 2011 Apr;32(12):3233–43. 99. Pittenger MF, Mackay AM, Beck SC, Jaiswal RK, Douglas R, Mosca JD, et al. Multilineage potential of adult human mesenchymal stem cells. Science (80- ). 1999;284(5411):143–7. 100. Barry FP, Murphy JM. Mesenchymal stem cells: clinical applications and biological characterization. Int J Biochem Cell Biol. 2004;36(4):568–84. 101. Koerner J, Nesic D, Romero JD, Brehm W, Mainil-Varlet P, Grogan SP. Equine peripheral blood-derived progenitors in comparison to bone marrow-derived mesenchymal stem cells. Stem Cells. 2006;24(6):1613–9. 102. Dhar M, Neilsen N, Beatty K, Eaker S, Adair H, Geiser D. Equine peripheral blood-derived mesenchymal stem cells: Isolation, identification, trilineage differentiation and effect of hyperbaric oxygen treatment. Equine Vet J. 2012;44(5):600–5. 160 103. Nakao N, Nakayama T, Yahata T, Muguruma Y, Saito S, Miyata Y, et al. Adipose tissue-derived mesenchymal stem cells facilitate hematopoiesis in vitro and in vivo: advantages over bone marrow-derived mesenchymal stem cells. Am J Pathol. 2010;177(2):547–54. 104. Dasuri K, Antonovici M, Chen K, Wong K, Standing K, Ens W, et al. The synovial proteome: analysis of fibroblast-like synoviocytes. Arthritis Res Ther. 2004;6(2):R161–8. 105. J. H. Differentiated cells and the maintenance of tissues. In: Molecular Biology of the Cell. 1994. p. 1139–93. 106. Lysy PA, Smets F, Sibille C, Najimi M, Sokal EM. Human skin fibroblasts: From mesodermal to hepatocyte-like differentiation. Hepatology. 2007;46(5):1574–85. 107. Haniffa MA, Collin MP, Buckley CD, Dazzi F. Mesenchymal stem cells: The fibroblasts’ new clothes? Vol. 94, Haematologica. 2009. p. 258–63. 108. Pigott JH, Ishihara A, Wellman ML, Russell DS, Bertone AL. Investigation of the immune response to autologous, allogeneic, and xenogeneic mesenchymal stem cells after intra-articular injection in horses. Vet Immunol Immunopathol [Internet]. 2013;156(1–2):99–106. Available from: http://dx.doi.org/10.1016/j.vetimm.2013.09.003 109. Christmann U, van Loon G. ECG of the Month. J Am Vet Med Assoc. 2013;242(2):165–7. 110. Van Loon VJF, Scheffer CJW, Genn HJ, Hoogendoorn AC, Greve JW. Clinical follow-up of horses treated with allogeneic equine mesenchymal stem cells derived from umbilical cord blood for different tendon and ligament disorders. Vet Q. 2014;2176(September 2014):1–6. 111. Lacitignola L, Staffieri F, Rossi G, Francioso E, Crovace A. Survival of bone marrow mesenchymal stem cells labelled with red fluorescent protein in an ovine model of collagenase-induced tendinitis. Vet Comp Orthop Traumatol. 2014;27(3):204–9. 112. Ueno T, Nakashima A, Doi S, Kawamoto T, Honda K, Yokoyama Y, et al. Mesenchymal stem cells ameliorate experimental peritoneal fibrosis by suppressing inflammation and inhibiting TGF-β1 signaling. Kidney Int. 2013;84(2):297–307. 113. Parys M, Nelson N, Koehl K, Miller R, Kaneene JB, Kruger JM, et al. Safety of Intraperitoneal Injection of Adipose Tissue-Derived Autologous Mesenchymal Stem Cells in Cats. J Vet Intern Med. 2016;30(1):157–63. 114. Roufosse C a, Direkze NC, Otto WR, Wright N a. Circulating mesenchymal stem cells. Int J Biochem Cell Biol. 2004;36(4):585–97. 115. Edwards-Milewski ML, Morello SL, Zhao Q, Mattan-Bell C. The Effect of Intravenous Regional Perfusion of the Distal Limb With Amikacin Sulfate on Wounds Healing by Second Intention in Horses. Vet Surg. 2016;45(1):125–32.

161 116. Omori Y, Honmou O, Harada K, Suzuki J, Houkin K, Kocsis JD. Optimization of a therapeutic protocol for intravenous injection of human mesenchymal stem cells after cerebral ischemia in adult rats. Brain Res. 2008;1236:30–8. 117. Honmou O, Houkin K, Matsunaga T, Niitsu Y, Ishiai S, Onodera R, et al. Intravenous administration of auto serum-expanded autologous mesenchymal stem cells in stroke. Brain. 2011;134(6):1790–807. 118. Le Blanc K, Davies LC. Mesenchymal stromal cells and the innate immune response. Immunol Lett. 2015;168(2):140–6. 119. Aggarwal S, Pittenger MF. Human mesenchymal stem cells modulate allogeneic immune cell responses. Blood. 2005;105(4):1815–22. 120. Grazia Maria Spaggiari, Andrea Capobianco, Stelvio Becchetti MCM, Moretta L. Mesenchymal stem cell–natural killer cell interactions: evidence that activated NK cells are capable of killing MSCs, whereas MSCs can inhibit IL-2–induced NK- cell proliferation. Blood. 2006;107(4):1484–90. 121. Spaggiari GM, Capobianco A, Abdelrazik H, Becchetti F, Mingari MC, Moretta L. 2011 - cytotoxicity , and cytokine production role of indoleamine Mesenchymal stem cells inhibit natural killer – cell proliferation , cytotoxicity , and cytokine production role of indoleamine 2 , 3-dioxygenase and prostaglandin E2.pdf. 2018;111(3):1327–34. 122. Benvenuto F, Ferrari S, Gerdoni E, Gualandi F, Frassoni F, Pistoia V, et al. Human Mesenchymal Stem Cells Promote Survival of T Cells in a Quiescent State. Stem Cells. 2007;25(7):1753–60. 123. Reikerås O, Sigurdsen UW, Shegarfi H. Impact of freezing on immunology and incorporation of bone allograft. J Orthop Res. 2010;28(9):1215–9. 124. Duff MM, Ritter T, Ceredig R, Griffi MD. Mesenchymal stem cell eff ects on T- cell eff ector pathways. 2011;1–9. 125. Gallagher WM, Lynch I, Allen LT, Miller I, Penney SC, O’Connor DP, et al. Molecular basis of cell–biomaterial interaction: Insights gained from transcriptomic and proteomic studies. Biomaterials. 2006;27(35):5871–82. 126. Araújo‐ Gomes N, Romero‐ Gavilán F, Sánchez‐ Pérez AM, Gurruchaga M, Azkargorta M, Elortza F, Martinez‐ Ibañez M, Iloro I, Suay J GI. Characterization of serum proteins attached to distinct sol–gel hybrid surfaces. J Biomed Mater Res Part B. 2018;106B:1477–85. 127. Tu CT, Li JL, Shen S, Sheng Q, Shyr Y, Qu J. Performance Investigation of Proteomic Identification by HCD/CID Fragmentations in Combination with High/Low-Resolution Detectors on a Tribrid, High-Field Orbitrap Instrument Chengjian Tu, Jun Li, Shichen Shen, Quanhu Sheng, Yu Shyr, Jun Qu . 2016; 11(7. PLoS One. 2016;11(7). 128. Mi H, Huang X, Muruganujan A, Tang H, Mills C, Kang D, et al. PANTHER version 11: Expanded annotation data from Gene Ontology and Reactome pathways, and data analysis tool enhancements. Nucleic Acids Res. 162 2017;45(D1):D183–9. 129. Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, et al. Gene ontology: tool for the unification of biology. {T}he {G}ene {O}ntology {C}onsortium. Nat Genet [Internet]. 2000;25(1):25–9. Available from: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3037419/ 130. The Gene Ontology Consortium. Expansion of the {Gene} {Ontology} knowledgebase and resources. Nucleic Acids Res. 2017;45(D1):D331--D338. 131. Heberle H, Meirelles VG, da Silva FR, Telles GP, Minghim R. InteractiVenn: A web-based tool for the analysis of sets through Venn diagrams. BMC Bioinformatics. 2015;16(1). 132. Liu, H., Sadygov, R.G., Yates JR. A model for random sampling and estimation of relative protein abundance in shotgun proteomics. Anal Chem. 2004;76:4193–201. 133. Colinge, J., Chiappe, D., Lagache, S., Moniatte, M., Bougueleret L. Differential proteomics via probabilistic peptide identification scores. Anal Chem. 2005;77:596–606. 134. Robinson, M.D.M., McCarthy, D.J.D., Smyth GKG. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. In: Transactions of the IRE Professional Group on 2010. 2010. p. 139–40. 135. Robinson, M.D.M., Smyth GKG. Moderated statistical tests for assessing differences in tag abundance. Bioinformatics. 2007;23:2881–7. 136. Robinson, M.D.M., Smyth GKG. Small-sample estimation of negative binomial dispersion, with applications to SAGE data. Biostatistics. 2008;9:321–32. 137. Benjamini, Y., Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Ser B. 1995;289–300. 138. Paebst F, Piehler D, Brehm W, Heller S, Schroeck C, Tárnok A, et al. Comparative immunophenotyping of equine multipotent mesenchymal stromal cells: An approach toward a standardized definition. Cytom Part A. 2014;85(8):678–87. 139. Wang M, Yuan Q, Xie L. Review Article Mesenchymal Stem Cell-Based Immunomodulation : Properties and Clinical Application. 2018;2018. 140. Hardy RS, Hülso C, Liu Y, Gasparini SJ, Fong-Yee C, Tu J, et al. Characterisation of fibroblast-like synoviocytes from a murine model of joint inflammation. Arthritis Res Ther. 2013;15(1):R24. 141. Petinati NA, Kapranov NM, Bigil’deev AE, Popova MD, Davydova YO, Gal’tseva I V, et al. Changing the Properties of Multipotent Mesenchymal Stromal Cells by IFN$γ$ Administration. Bull Exp Biol Med. 2017 Jun;163(2):230–4. 142. Christensen JE, Andreasen SO, Christensen JP TA. CD11b expression as a marker to distinguish between recently activated effector CD8(+) T cells and memory cells. Int Immunol 2001 Apr;13(4)593-600. 2001;13(4):593–600. 143. Fagerholm SC, MacPherson M, James MJ, Sevier-Guy C LC. The CD11b-integrin

163 (ITGAM) and systemic lupus erythematosus. Lupus. 2013;22(7):657–63. 144. Bertone AL, Palmer JL JJ. Synovial fluid cytokines and eicosanoids as markers of joint disease in horses. Vet Surg. 2001;30(6):528–38. 145. Van Den Boom, R., P. A. J. Brama, G. H. Kiers, J. De Groot A, Barneveld PRVW. The influence of repeated arthrocentesis and exercise on matrix metalloproteinase and tumour necrosis factor a activities in normal equine joints. Equine Vet J. 2004;36:155–9. 146. Ley C, Ekman S, Elmén A, Nilsson G, Eloranta ML. Interleukin-6 and tumour necrosis factor in synovial fluid from horses with carpal joint pathology. J Vet Med Ser A Physiol Pathol Clin Med. 2007;54(7):346–51. 147. Goldring SR GM. The role of cytokines in cartilage matrix degeneration in osteoarthritis. Clin Orthop Relat Res 2004 Oct;(427 Suppl)S27-36. 2004;427 Suppl:27–9. 148. Grimaud, Eve Heymann, Dominique Rédin F. Recent advances in TGF-β effects on chondrocyte metabolism: Potential therapeutic roles of TGF-β in cartilage disorders. Cytokine Growth Factor Rev. 2002;13(3):241–57. 149. Haller D, Blum S, Bode C, Hammes WP, Schiffrin EJ. Activation of Human Peripheral Blood Mononuclear Cells by Nonpathogenic Bacteria In Vitro: Evidence of NK Cells as Primary Targets. Infect Immun. 2000 Feb;68(2):752 LP- 759. 150. Letterio JJ RA. Regulation of immune responses by TGF-beta. Annu Rev Immunol. 1998;16:137–61. 151. Perrier S, Darakhshan F, Hajduch E. IL-1 receptor antagonist in metabolic diseases: Dr Jekyll or Mr Hyde? FEBS Lett. 2006;580(27):6289–94. 152. Abomaray F, al Jumah M, Alsaad K. Phenotypic and functional characterization of mesenchymal stem/multipotent stromal cells fromdecidua basalisof human term placenta. Stem Cells Int. 2016;2016. 153. Su J, Chen X, Huang Y. Phylogenetic distinction of iNOS and IDO function in mesenchymal stem cell-mediated immunosuppression in mammalian species. Cell Death Differ. 2014;21(3):388–96. 154. KY D, LE de CB, RA L, ML L. Extracellular matrix proteomics in cardiac ischemia/reperfusion: the search is on. Circ Res. 2012;125:746–8. 155. Dj L, AC B-J, E V, MR L. Post-translational modifications of the extracellular matrix are key events in cancer progression: opportunities for biochemical marker development. Biomarkers. 2011;16:193–205. 156. M T, M M. From genomics to proteomics. Nature. 2003;422:193–7. 157. Prange T, Carr EA, Stick JA, Garcia-Pereira FL, Patterson JS, Derksen FJ. Cervical vertebral canal endoscopy in a horse with cervical vertebral stenotic myelopathy. Equine Vet J. 2012;44(1):116–9.

164 158. Patterson NL, Rugmani P, De Castro Bras L, Li Y, Andrews TG, Aune GJ, et al. Using Proteomics to Uncover Extracellular Matrix Interactions During Cardiac Remodeling. Proteomics Clin Appl. 2013;7:516–27. 159. Kim SJ, Im YH, Markowitz SD BY. Molecular mechanisms of inactivation of TGF-beta receptors during carcinogenesis. Cytokine Growth Factor Rev 2. 2000;11(1–2):159–68. 160. LeBaron RG, Bezverkov KI, Zimber MP, Pavelec R, Skonier J, Purchio AF. βIG- H3, a novel secretory protein inducible by transforming growth factor-β, is present in normal skin and promotes the adhesion and spreading of dermal fibroblasts in vitro. J Invest Dermatol. 1995;104(5):844–9. 161. Hashimoto K, Noshiro M, Ohno S, Kawamoto T, Satakeda H, Akagawa Y, et al. Characterization of a cartilage-derived 66-kDa protein (RCD-CAP/βig-h3) that binds to collagen. Biochim Biophys Acta - Mol Cell Res. 1997;1355(3):303–14. 162. Rawe IM, Zhan Q, Burrows R, Bennett K, Cintron C. Beta-Ig: Molecular cloning in situ hybridization in corneal tissues. Investig Ophthalmol Vis Sci. 1997;38(5):893–900. 163. Billings PC, Charles Whitbeck J, Adams CS, Abrams WR, Cohen AJ, Engelsberg BN, et al. The transforming growth factor-β-inducible matrix protein βig-h3 interacts with fibronectin. J Biol Chem. 2002;277(31):28003–9. 164. Ferguson JW, Mikesh MF, Wheeler EF, LeBaron RG. Developmental expression patterns of Beta-ig (βIG-H3) and its function as a cell adhesion protein. Mech Dev. 2003;120(8):851–64. 165. MunetakaShimizu. Subchapter 19C - Insulin-Like Growth Factor-II. In: Handbook of Hormones Comparative Endocrinology for Basic and Clinical Research. 2016. p. 164–6. 166. Zhang Z, Li L, Yang W, Cao Y, Shi Y, Li X, et al. The effects of different doses of IGF-1 on cartilage and subchondral bone during the repair of full-thickness articular cartilage defects in rabbits. Osteoarthr Cartil. 2017;25(2):309–20. 167. McMahan RS, Birkland TP, Smigiel KS, Vandivort TC, Rohani MG, Manicone AM, et al. Stromelysin-2 (MMP10) Moderates Inflammation by Controlling Macrophage Activation. J Immunol. 2016;197(3):899–909. 168. Rohani M, McMahan R, Razumova M, Hertz A, Cieslewicz M, Pun S, et al. MMP-10 Regulates Collagenolytic Activity of Alternatively Activated Resident Macrophages. J Invest Dermatol. 2015;135(10):2377–84. 169. Baker N, Sharpe P, Culley K, Otero M, Bevan D, Newham P, et al. Dual regulation of metalloproteinase expression in chondrocytes by WISP3/CCN6. Arthritis Rheum. 2012;64(7):2289–99. 170. Mbebi C, Hantai D, Jandrot-Perrus M, Doyennette M, Verdiere-Sahuque M. Protease nexin I expression is up-regulated inhuman skeletal muscle by injury- related factors. J Cell Physiol1999;179305–14. 1999;179:305–14.

165 171. Vaughan P, Cunningham D. Regulation of protease nexin-1 synthesis and secretion in cultured brain cells by injury-related factors. J Biol Chem. 1993;268(268):3720–7. 172. Rajashekhar G, Grow M, Willuweit A, Patterson C, Clauss M. Divergent and convergent effects on gene expression and function in acute versus chronic endothelial activation. Physiol Genomics. 2007;31:104–13. 173. Zer C, Sachs G, Shin J. Identification of genomic targets downstream of p38 mitogen-activated protein kinase pathway mediating tumor necrosis factor- signaling. Physiol Genomics. 2007;31:343–51. 174. Mansilla S, Boulaftali Y, Venisse L, Arocas V, Meilhac O, Michel J. Macrophages and platelets are the major source of protease nexin-1 in human atherosclerotic plaque. Arter Thromb Vasc Biol. 2008;28:1844–50. 175. Stevens P, Scott R, Shatzen E. Recombinant human protease nexin-1 prevents articular cartilage-degradation in the rabbit. Agents Actions Suppl. 1993;39:173–7. 176. Maciejewska-Rodrigues H, Al-Shamisi M, Hemmatazad H, Ospelt C, Bouton M, Jager D, et al. Functional Autoantibodies Against Serpin E2 in Rheumatoid Arthritis. ARTHRITIS Rheum. 2010;62:93–104. 177. LI F, Liu J, Liu N, Kuhn LA, Garavito M, Ferguson-Miller S. Translocator Protein 18 kDa (TSPO): An Old Protein with New Functions? Biochemistry. 2016;55(20):2821–31. 178. Wen Y, Li J, Wang L, Tie K, Magdalou J, Wang H. UDP-glucose dehydrogenase modulates proteoglycan synthesis in articular chondrocytes: its possible involvement and regulation in osteoarthritis. Arthritis Res Ther 2014; 16(6) 484. 2014;16(6):484. 179. Casini A, de Moerloose P N-AM. Clinical Features and Management of Congenital Fibrinogen Deficiencies. In: Seminars in Thrombosis and Hemostasis. 2016. p. 366–74.

166