Models and Mechanisms to Evaluate Tissue Engineered Vascular Graft Stenosis

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy

in the Graduate School of The Ohio State University

Elizabeth S. Clark

Graduate Program in Comparative Veterinary Medicine

The Ohio State University

2017

Committee:

Dr. Christopher K. Breuer, Advisor

Dr. Keith Gooch

Dr. Krista La Perle

Dr. Joy Lincoln

Copyrighted by

Elizabeth Suzanne Clark

2017

Abstract

Congenital heart disease represents the most common form of birth defect and is present in up to 1% of all live births. Of particular interest to our group is the treatment of single ventricle disease, specifically hypoplastic left heart syndrome (HLHS) in which left-sided heart structures including the left ventricle, aorta, and mitral valve are malformed. We utilize tissue engineering principles to provide palliative treatment with

Tissue-Engineered Vascular Grafts (TEVG) that allow for growth of the child’s own tissues. The TEVG overcomes limitations present in other composite grafts including thrombogenicity and lack of growth capacity. One limitation of our TEVG is that a small percentage of children develop critical stenosis requiring the utilization of interventional cardiovascular techniques. Developing both large and small animal models of TEVG stenosis will allow us to better understand clinical data sets as well as provide models to better characterize the development of neotissue formation and stenosis.

The long-term goal of this project is to develop a second-generation composite vascular graft for use in children. The objective of this study is to elucidate the mechanism of TEVG neotissue formation in immunocompetent mice (Mus musculus) and characterize neotissue formation in a sheep (Ovis aries) model. Echocardiography, angiography, and magnetic resonance imaging are routinely used to evaluate children that have received the TEVG to treat HLHS. While angiography remains the clinical gold

ii standard for evaluation of the lumen it only indirectly evaluates the vessel wall.

Therefore, we sought to evaluate intravascular ultrasound (IVUS) to characterize neotissue formation in an in vivo ovine surgical model. IVUS demonstrated close correlation to angiographic and histologic changes. Additionally, IVUS measurement of graft lumen morphometry correlates with angiography measurements. This provides further data to support the use of IVUS in a clinical setting. Future preclinical experiments will utilize this modality as an additive approach to characterize the development of TEVG stenosis and neotissue remodeling over time, with specific regard to response to treatment with both interventional and therapeutic approaches.

Previous research by our lab C57BL/6 mice has suggested that the transforming growth factor-β (TGF-β)/Smad2-3 pathway is involved in the development of TEVG stenosis. While the specific role of the TGF-β pathway in the development of stenosis remains unknown, this pathway is upregulated in stenotic TEVG. To this end, this work seeks to evaluate the TGF-β pathway in Cdh5-lineage cells in vivo in mice implanted with TEVG. A conditional genetic murine model was developed with endothelial-derived

(Cdh5) loss of the TGF-β receptors, TβRI and TβRII. Although TEVG have a low incidence of stenosis in both treated and control animals, Cdh5 TGF-β pathway modulation results in altered neotissue formation including collagen production and polymeric degradation at early timepoints. We rationalize that by elucidating the cell(s) of origin for the neotissue and the role of TGF-β receptor pathway will enable the development of targeted therapies in combination with TEVGs to improve clinical outcomes for children.

iii

Acknowledgments

This work is dedicated to my mother for inspiring my love of science: you have shared your love of science with me from a young age - from car rides discussing the origins of the universe as a young child, to adventures in streams collecting macroinvertebrates and perusing mathematical textbooks in the library of Southern

Illinois University as a middle schooler. You have taught me that the pursuit of science can be as enjoyable as discovering the answer to your hypothesis.

To my father who instilled the value of commitment and loyalty to a chosen passion. You always provided whatever materials were required for my various pursuits, whether that was music, sport, or school. I believe your insistence on ‘making life easy’ by pursuing certain career opportunities were intentional misdirections - that you knew I would oppose – to send me on this trajectory to pursue my DVM, PhD, and ACVP board certification.

Ido and Lexie, my loves, you have provided an emotional bedrock of support during the course of this training program. Ido, thank you for your commitment to our relationship and my success. While supporting my endeavors you also served to remind me of my values in life, and that occasionally items needed re-prioritizing. As someone that delves wholeheartedly, and sometimes overwhelmingly into a pursuit, this has served

iv to re-balance my life. I’m very much looking forward to seeing where life’s adventures take us now. To Lexie, my dog: you have been on this journey since the start of veterinary school in 2007. You have tolerated long days, takeout food, and late nights.

You’ve earned all the chewy bones and walks to Central Park that you desire.

Thank you to Dr. Krista La Perle for accepting my request to visit the Ohio State

University for a rotation my fourth year of veterinary school. It was this experience the ultimately led to my application for the Combined PhD/Anatomic Pathology Residency

Program. You have been a mentor to me both personally and professionally these past few years.

To my PI, Dr. Christopher Breuer, thank you for the opportunity to become a member of your research team. You created a truly unique environment in which I could participate in cutting edge translational research. While the lab has changed greatly since you first established your group in Columbus, OH in 2012, you always inspire our team and project a vision to direct our work. You have offered unwavering support, flexibility, understanding, and compassion during this wild ride called a PhD.

This work is also dedicated in memoriam to Richard (Pappy) Tennant and

Leonardo and Carmen Kashy. Leo and Carmela – thank you for raising the man I came to love. I’m joyous to send a note up to heaven to tell you that, yes, I finally finished my

PhD.

Vita

December 2, 1984 ...... Born, Mount Vernon, Illinois

2007 ...... Bachelor of Science, Molecular and Cellular

Biology, The University of Illinois at

Urbana-Champaign

2012 ...... Doctor of Veterinary Medicine, The

University of Illinois at Urbana-Champaign

2012 to present ...... Anatomic Pathology Resident and Post-

Doctoral Fellow, The Ohio State University

Publications

1. Pepper VK, Clark ES, Best CA, Onwuka EA, Sugiura T, Heuer ED, Moko LE, Miyamoto S, Miyachi H, Berman DP, Cheatham SL, Chisolm JL, Shinoka T, Breuer CK, Cheatham JP. Intravascular Ultrasound Characterization of a Tissue-Engineered Vascular Graft in an Ovine Model. (2017) Journal of Cardiovascular Translational Research, January 17. [Epub ahead of print]

2. Clark ES, Best C, Onwuka E, Sugiura T, Mahler N, Bolon B, Niehaus A, James I, Hibino N, Shinoka T, Johnson J, Breuer CK. Effect of cell seeding on neotissue formation in a tissue engineered trachea. (2016) Journal of Pediatric Surgery 51(1):49-55.

3. Clark ES, Pepper VK, Best CA, Onwuka EA, Yi T, Tara S, Cianciolo R, Baker P, Shinoka T, Breuer CK. A mouse model of endocardial fibroelastosis. (2015) Cardiovascular Pathology 24(6):388-94.

4. Henderson SE, Clark ES, Stromberg PC, Radin MJ, Wellman ML. Pathology in Practice. Canine Morbillivirus infection. (2015) Journal of the American Veterinary Medical Association 247(12):1375-7.

5. Rudinsky AJ, Clark ES, Russell DS, Gilor C. Adrenal insufficiency secondary to lymphocytic panhypophysitis in a cat. (2015) Australian Veterinary Journal 93(9):327-31.

Fields of Study

Major Field: Comparative Veterinary Medicine

vii

Table of Contents

Abstract ...... ii

Acknowledgments ...... iv

Vita ...... vi

Publications ...... vi

Table of Contents ...... viii

List of Figures ...... xiii

List of Tables...... xiv

Abbreviations ...... xv

Chapter 1: Introduction ...... 1

1.1 Congenital Heart Disease & Hypoplastic Left Heart Syndrome ...... 1

1.2 Polymeric TEVG Construct: A Scaffold for Neovascularization ...... 4

1.3 Vasculogenesis and Blood Vessel Structure...... 5

1.3.1 Formation of the Tunica Media ...... 8

1.3.2 Tunica Media, Tunica Adventitia, and the Extracellular Matrix ...... 10

1.4 Clinical Trial Results for the Treatment of Hypoplastic Left Heart Syndrome...... 12 viii

1.5 Animal models ...... 15

1.5.1 Ovine Models ...... 17

1.5.2 Murine Models ...... 22

1.5.3 Mechanisms of TEVG Stenosis ...... 27

1.5.4 TGF- Pathway ...... 28

1.5.5 Endothelial-to-Mesenchymal Transition ...... 30

1.6 Study Objectives ...... 35

Chapter 2: Intravascular Ultrasound Characterization of a Tissue-Engineered Vascular

Graft in an Ovine Model ...... 38

2.1 Abstract ...... 38

2.2 Introduction ...... 39

2.3 Materials and Methods ...... 41

2.3.1 Scaffold Fabrication ...... 41

2.3.2 Bone Marrow Harvest ...... 42

2.3.3 Scaffold Seeding ...... 44

2.3.4 Implantation ...... 44

2.3.5 Interventional Monitoring ...... 45

2.3.5.1 Angiography...... 45

2.3.5.2 Intravascular Ultrasound (IVUS) ...... 46

2.3.5.3 Balloon Angioplasty ...... 46

2.3.5.4 Animal Euthanasia ...... 46

2.3.6 Tissue Processing, Histology, and Immunofluorescent Staining ...... 47

2.3.7 Animal Subjects Statement ...... 48

2.3.8 Statistical Analysis ...... 48

2.4 Results ...... 49

2.4.1 Animal Outcomes ...... 49

2.4.2 IVUS Characterization of Neovessel Formation ...... 51

2.4.2.1 IVUS Characterization Over Time ...... 51

2.4.2.2 IVUS Findings over the Length of the Graft ...... 53

2.4.3 Comparison of Angiography and IVUS by Quantification of Lumen Geometry . 55

2.4.4 Characterization of Anomalies...... 57

2.4.4.1 Strengths of IVUS ...... 57

2.4.4.2 Weaknesses of IVUS ...... 61

2.5 Discussion ...... 61

2.6 Conclusions ...... 66

Chapter 3: Cdh5 Deletion of TβRI and TβRII Modulates Neotissue Formation in Tissue-

Engineered Vascular Grafts ...... 69

3.1 Abstract ...... 69

3.2 Introduction ...... 70

3.3 Materials and Methods ...... 72

3.3.1 Transgenic mouse model ...... 72

3.3.2 Graft production and Implantation ...... 75

3.3.3 ECHO ...... 75

3.3.4 Explant and Tissue Collection ...... 76

3.3.5 Histochemical Staining, Immunohistochemistry, and Immunofluorescence ...... 77

3.3.6 Image Analysis ...... 78

3.3.7 Quantitative RT-PCR ...... 79

3.3.8 Immunoblot analysis ...... 80

3.3.9 Statistical Analysis ...... 80

3.4 Results ...... 81

3.4.1 Graft Phenotyping and Lineage Tracing ...... 81

3.4.2 Graft Remodeling and Stenosis...... 88

3.4.3 PCR ...... 97

3.4.4. Immunoblotting analysis ...... 99

3.5 Discussion ...... 103

3.6 Conclusions ...... 110

Chapter 4: Conclusions and Future Directions ...... 112

4.1 Conclusions ...... 112

4.2 Future small animal experiments ...... 117

4.3 Ovine Models ...... 120

4.3.1 Assessment of Efficacy and Safety of Balloon Angioplasty for the Treatment of

TEVG stenosis ...... 121

4.3.2 Modalities to Evaluate Early Characterization of Microcalcification in Ovine

TEVG ...... 128

4.3.3 Future Approaches for the Treatment of TEVG Stenosis ...... 132

4.3.4 TEVG Utilization of Drug Eluting Compounds ...... 133

4.4 Final Thoughts ...... 135

Bibliography ...... 137

xii

List of Figures

Figure 1: Schematic of neovessel formation in a mouse model...... 33

Figure 2: TGF- signaling pathway...... 34

Figure 3: TEVG Seeding and Imaging Methods...... 43

Figure 4: IVUS and Histology Over Time...... 52

Figure 5: IVUS and Histology Along the Graft Length...... 54

Figure 6: Luminal Geometry: IVUS vs. Angiographic Quantification...... 56

Figure 7: Abnormalities Encountered in Model: IVUS versus Angiography...... 59

Figure 8: Correlation of IVUS Findings to Ex Vivo Pathology...... 60

Figure 9: Schematic for cre-mediated recombination and expression of mT/mG in Cdh5- cells...... 74

Figure 10: Genotyping of offspring for Cdh5cre.ERT2+/+, Tgfbr1f/f, and Tgfbr2 f/f...... 75

Figure 11: TEVG Implant, ECHO, and Explant. Continued...... 76

Figure 12: Direct fluorescent phenotyping of TEVG for mG (membrane green) and mT

(membrane red)...... 83

Figure 13: Expression of TβRI, TβRII, and SMAD2 in TEVG ...... 84

xiii

List of Tables

Table 1: Animal Outcomes…………………………………………………………….50

xiv

Abbreviations

α-SMA: alpha smooth muscle actin

ALK: activin receptor-like kinase

ALK5: TGF-β receptor 1

ANOVA: analysis of variance

BA: balloon angioplasty

BM-MNC: bone marrow-derived mononuclear cells

BP: base pairs

CD31: see PECAM-1

CHD: congenital heart disease

CI: confidence interval

CT: computed tomography

DAPI: 4',6-diamidino-2-phenylindole

DNA: deoxyribonucleic acid

DTR: diphtheria toxin receptor

ECHO: echocardiogram

ECL: enhanced chemiluminescence

ECM: extracellular matrix

EMT/Endo-MT: endothelial to mesenchymal transition

FDA: Food and Drug Administration

FGF: fibroblast growth factor

GAG: glycosaminoglycans

GFP: green fluorescent protein

HLHS: hypoplastic left heart syndrome

IF: immunofluorescence

IHC: immunohistochemistry

IL-1: interleukin 1

IL-8: interleukin 8

IDE: Investigational Device Exemption

IND: Investigational New Drug

IVUS: intravascular ultrasound

LAP: latency associated peptide

Mac-3: Lysosomal-associated membrane protein 2

MAPK: mitogen-activated protein kinases

MMP: matrix metalloproteinase

MNGC: multinucleated giant cells

MRI: magnetic resonance imaging

OCT: optical coherence tomography

OCT: optimal cutting temperature

PBS: phosphate buffered solution

PCLA: polycaprolactone and polylactic acid

PCNA: proliferating cell nuclear antigen

xvi

PDGF: platelet derived growth factor

PECAM-1: platelet endothelial cell adhesion molecule-1

PGA: polyglycolic acid

PMSF: phenylmethane sulfonyl fluoride

PO: per os

PVDF: polyvinylidene fluoride

RFP: red fluorescent protein

RNA: ribonucleic acid

RT-PCR: reverse transcriptase polymerase chain reaction

SCID: Severe Combined Immune Deficient

SDS-PAGE: sodium dodecyl sulfate polyacrylamide gel electrophoresis

SPECT: single photon emission computed tomography

TCPC: total cavo-pulmonary connection

TGF-β: transforming growth factor beta

TβRI: transforming growth factor beta receptor I (Alk5)

TβRII: transforming growth factor beta receptor II

TEVG: tissue engineered vascular graft

TNF-α: tumor necrosis factor alpha

VC: vena cava

VE-cadherin: vascular endothelial cadherin

VEGF: vascular endothelial growth factor

VSMC: vascular smooth muscle cell

xvii vWF: von Willebrand Factor

WT: wildtype

xviii

Chapter 1: Introduction

1.1 Congenital Heart Disease & Hypoplastic Left Heart Syndrome

Congenital Heart Disease (CHD) and Hypoplastic Left Heart Syndrome (HLHS) form the impetus for our research. These conditions can be treated with surgical approaches and provide increased longevity and quality of life for children born with these diseases. CHD encompasses a broad spectrum of pathophysiologic defects that include structural defects in septation, heart valves, atrial and ventricular chambers, ventricular outflow systems, great vessels, and heterotaxy (left-right patterning). CHD represents the most common form of birth defect and is present in up to 1% of all live births. (1, 2) Many of the underlying causes for these abnormalities are multifactorial and include genetic, environmental, and viral causes. (1-3) Single ventricle disease, which is recognized as a class of CHD, also represents a diverse spectrum of disease that includes hypoplasia of the left and right ventricles, and tricuspid and pulmonary valve atresia. (4,

5) Hypoplastic left heart syndrome occurs in 0.1-0.25/1000 live births and contributes to the 1% incidence of worldwide congenital heart disease. (6-8) HLHS is primarily characterized by anomalous left heart structures, including the left ventricle, but may also present with a diverse combination of anomalies that involve the mitral and aortic valves

1 and the aortic outflow tract. This condition is significant in that it is life-threatening when left untreated. (9)

Treatment approaches for congenital heart disease include surgical correction, partial correction, or palliation. Current treatments for HLHS, with the exception of cardiac transplantation, are palliative in that this approach does not re-establish normal blood flow or anatomy, but is of benefit in that it reduces the associated pathophysiologic implications of the disease. Major advances were made in the treatment of HLHS in the

20th century with the introduction of new surgical techniques, including Norward, Glenn, and Modified Fontan surgeries. These three surgeries are utilized in a staged approach often within the first year of life. (9) Following the Norwood, which shunts blood between the pulmonary arteries and aorta, the Glenn shunt closes the pulmonary artery shunt and redirects systemic blood by connecting the superior vena cava to the pulmonary artery. In the final sugery, the Modified Fontan, the inferior vena cava is connected to the pulmonary circuit. (10) Due to the spectrum of HLHS and the novelty of the Fontan in the latter half of the 20th century there has been variation and incremental progress in the surgical approaches used to complete the Fontan circuit. Today the total cavo-pulmonary connection (TCPC) is the most commonly utilized approach. This approach results in a parallel system that contributes to both the pulmonary and systemic circulation while bypassing the atretic left ventricle; this is in contrast to the normal physiology in which the pulmonary and systemic circulations are in series. (11)

Utilization of the Fontan surgery to achieve the TCPC earlier in life is precluded by the

2 elevated pulmonary pressures in the neonatal population. Other surgical approaches include a hybrid approach, hybrid-bridge-to-Norwood, and cardiac transplantation.

Cardiac transplantation has been limited by the availability of donor tissue and the requirement for life-long immunotherapy. (9) Further implementation of the hybrid approach has been limited by highly variable survival rates in the pediatric population at independent study sites. (9, 12)

The modified Fontan results in a TCPC by utilization of either an intracardiac lateral tunnel or extracardiac baffle TCPC conduit approach. Limitations of the intracardiac lateral tunnel include surgical complexity requiring the atrium to be opened and has inherent risks such as the development of arrhythmias. However, this approach does allow for the growth of the native atrial tissue. The extracardiac TCPC requires the use of a vascular conduit graft. (11) Growth potential of the extracardiac TCPC is dependent on the graft material selected. While the Modified Fontan surgery is considered the standard of care, progress in the treatment of this disease has been limited by the availability of healthy vascular tissue from autologous donor sites. (13) To address this tissue, surgeons have turned to the use of polymeric graft constructs such as Gore- tex® (polytetrafluoroethylene), Dacron® (polyethelene terephthalate) and the Tissue

Engineered Vascular Graft (TEVG) which is composed of polyglycolic acid (PGA) fibers and polycaprolactone and polylactic acid (PCLA) to craft de novo vascular conduits for the Fontan circuit. (14) Gore-tex® and Dacron® grafts are limited by their ability to grow and remodel over time. Biomechanical limitations of the Dacron® grafts have led

3 surgeons to primarily use Gore-tex®. (15, 16) Clinically, Dacron® grafts have been noted for graft stenosis (defined as 50% narrowing in the diameter of the graft). (17)

Gore-tex® grafts demonstrate narrowing (defined as narrowing of 10-30%), but do not typically stenose. To address these limitations surgeons often implant oversized vascular grafts which can affect the flow dynamics and increased thrombogenicity. (18) Other research has focused on the use of decellularized tissue scaffolds and the use of bioreactors. (19, 20) However, these approaches have yet to see clinical translation and lack approval for the treatment of orphan diseases as an Investigational New Drug (IND) submission to the Food and Drug Administration (FDA) (TEVG) or use in the surgical suite.

1.2 Polymeric TEVG Construct: A Scaffold for Neovascularization

A synthetic graft, the Tissue Engineered Vascular Graft, was developed with tissue engineering principles to provide an enhanced treatment for congenital heart disease that would allow for the growth of vascular tissues. The clinical and pre-clinical work on the TEVG utilizes a biodegradable polymeric graft composed of polyglycolic acid (PGA) fibers that are knitted into a tube and coated with a 50:50 copolymer of polycaprolactone and polylactic acid (PCLA). Biologically relevant aspects of the copolymer include minimal immunogenicity and biodegradability; it has been deemed safe for utilization for surgical implantation. With regard to efficacy, the tensile strength of the construct must support the hemodynamic load of the patient, while encouraging

4 cellular migration, proliferation, differentiation, and maturation of endothelial cells and vascular smooth muscle cells, discussed below, together with the appropriate elaboration of extracellular matrix constituents. (Figure 1.1 a) This coordinated tissue patterning has been termed neotissue formation. Additionally, an important feature of the graft is that it must undergo polymeric degradation over time and eventually be completely bioresorbed.

1.3 Vasculogenesis and Blood Vessel Structure

One of the key events in neotissue formation is the development of de novo blood vessels that constitute the Tissue Engineered Vascular Graft. This vascular patterning is required to provide an endothelial-lined non-thrombogenic surface to promote the laminar flow of blood. This blood vessel formation can occur in not only within the

TEVG but also by the formation of collateral blood vessels in the event of TEVG stenosis. The blood vessels deliver the currency of life: oxygen-rich blood is provided to organs while removing metabolic waste. Veins and arteries within the vascular tree are defined by three regions: the tunica intima, tunica media, and tunica adventitia. This simple scheme belies the complex nature of their functional zones. A single layer of endothelial cells resting on a basement membrane comprises the tunica intima. These endothelial cells maintain cell-cell junctions between one another and the underlying vascular smooth muscle cells within the tunica media. Separating the tunica intima and media is the internal elastic lamina in larger caliber vessels. The position within the

5 vascular tree directly correlates with the thickness of the vascular smooth muscle cell layer that is arranged in concentric layers. Extending beyond the tunica media is the tunica adventitia. Both the tunica media and adventitia are comprised of extracellular matrix components described in detail below. (21) The TEVG is implanted in a large- caliber venous structure characterized by low pressure, high flow system that contains a thinner tunica media and elastic laminae. Ultimately TEVG implantation can be deemed successful if the body is able to recapitulate these three functional and interdependent regions of the vessel by the formation of neotissue, full polymeric degradation, and eventual reconstitution by a fully functional neovessel.

Endothelialization is one of the earliest events observed from explanted TEVG implants and is a prerequisite for successful TEVG reconstitution. Vascular endothelial growth factor (VEGF) family members are the primary drivers for postnatal vasculogenesis. (22) They direct capillary sprouting of tip and stalk cells in a highly coordinated fashion by regulation of the expression of Notch receptors and their ligand

Delta-like-4 (DLL4). (23, 24) These sprouts form the migratory highway to direct the migration of mature endothelial cells. This migration requires the downregulation of cell- cell junction marker vascular endothelial (VE)-cadherin and platelet endothelial cell adhesion molecule-1 (PECAM-1/CD31) and upregulation of integrins. (25, 26) The early

TEVG lacks the typical substrates of collagen (α1β1, α2β1 receptors), laminins (a3b1, a6b1, α6β4), and fibronectin (a4b1, α5β1) and these or other receptors may be co-opted for migration on the polymeric substrates and acute fibrin deposition from the

6 bloodstream. (26) These markers are expressed in all levels of the blood vessel: ranging from the basement membrane of endothelial cells, extending through the tunica media to the tunica adventitia. Upon arriving at their destination endothelial cells must re-establish cell-cell junctions and expression of both anti- and pro-thrombotic cell surface markers.

Ultimately to function these cells must establish a vascular lumen, downregulate their proliferation cycles, and associate with sub-intimal vascular smooth muscle cells, among other cell types. (22) Various markers can be utilized to characterize maturation of vascular endothelium including: CD34, CD31 (PECAM-1), and von Willebrand Factor

(vWF). Endothelial cells produce vWF, which can then be either stored within Weibel-

Palade bodies or secreted. In contrast, CD31 is a transmembrane protein expressed on the surface of endothelial cells that is involved with VE-cadherin in leukocyte transmigration into the underlying vessel. (27)

A successful TEVG will ultimately be lined by a fully functional endothelial cell- lined intima that maintains a vascular lumen and appropriately balances pro- and anti- coagulant properties. Our studies, summarized below, have demonstrated Factor VIII+ and vWF+ intima lining the TEVG in ovine and murine models at acute and late time points. (28, 29) A unique challenge posed to endothelia constituting a TEVG utilized in a

Modified Fontan surgery in a child is the mixing of both arterial and venous blood streams, potentially requiring both intracellular and cell surface modifications for responsiveness to both arterial and venous systems. There is differential regulation of

7 markers, such as the Eph receptor tyrosine kinases that bind ephrin ligands that drive arterial and venous angiogenesis, which could have important implications for tissue patterning of the TEVGs. (22, 30, 31) Work by our group has demonstrated that by 6 months post-implant, neovessels derived from TEVG express Eph-B4, further supporting differentiation along a venous lineage that colocalizes with vWF+ neointima. (28)

Ultimately, though, continued maintenance of vascular integrity and response to metabolic demands requires appropriate communication between endothelium and vascular smooth muscle cells. (21)

1.3.1 Formation of the Tunica Media

In addition to endothelial cells, vascular smooth muscle cells (VSMC) are significant contributors to neotissue formation and vascular homeostasis. These cells produce extracellular matrix components, and respond to vasodilatory and vasoconstrictive signals. Vascular smooth muscle cells respond to circulating cues from both the blood stream and local endothelial cells, such as nitric oxide and endothelin-1, to regulate vascular tone, proliferation, and cellular hypertrophy. (32) These cells exist along a spectrum of two phenotypes that are intricately linked with function: a contractile, non-proliferative phenotype that is responsible for vessel tone and a synthetic phenotype that contributes to extracellular matrix production. The majority of markers utilized to characterize post-natal VSMC are more prevalent in contractile than synthetic cells and include -smooth muscle actin, muscle myosin heavy chain, SM22, calponin,

8 and smoothelin. These markers localize within the cellular cytoplasm to contractile elements including filaments and actin-associated structures. Smemb is unique in that this contractile filament marker predominately labels synthetic VSMC. Other markers used to characterize synthetic VSMC include caldesmon and vimentin. These cells often have upregulated production of extracellular matrix components, such as collagens, and markers of extracellular matrix remodeling, such as the matrix metalloproteinases

(MMPs). (33, 34) In the adult these cells predominately exist in their contractile phenotype except in response to vascular pathology in which they contribute to the remodeling of the extracellular matrix through phenotypic modulatory switching. During this process VSMCs are observed migrating to the subintimal regions of the vascular wall. (35) Interestingly, the extracellular matrix components also serve an important role in promoting dichotomous phenotypes for VSMC. Heparin, fibrillar collagens I and IV, and laminin all promote a contractile, non-proliferative phenotype. In contrast, monomeric collagen I, fibronectin, and hyaluronin promote VSMC proliferation and synthetic phenotype. (33) Soluble factors, such as platelet derived growth factor (PDGF), basic fibroblast growth factor (bFGF), and TGF-1 also promote VSMC contractile phenotype. (34) Other roles in TEVG stenosis for TGF- will be discussed in following sections.

Within a completely polymeric TEVG scaffold infiltrating cells must create ECM de novo. The large porous structure initially allows the diffusion of molecules due to a porosity of 25-50 nm. Research in mouse models has shown early TEVG to be highly

9 dynamic with acceleration of collagen I and III production in the first two weeks, with increases in collagen IV, GAGs, and MMP9 up to 6 months post-implant. VSMC must respond and contribute to this highly dynamic microenvironment that potentially requires numerous modulatory switches between their contractile and synthetic phenotypes. This is consistent with previous data demonstrating functional heterogeneity of VSMC within blood vessels and potential for switching between phenotypic switching. (34, 36)

Our studies, which are described below within the animal model context, have demonstrated the importance of macrophage phenotype on the development of TEVG stenosis with M1 macrophages predominating the lesions. Inflammatory cells and mediators, including macrophages and TNF-a, IL-1a, IL-8 can promote a pro-synthetic and proliferative VSMC phenotype. (34) In addition to highlighting the origin of VSMC in neotissue future studies should characterize their phenotypic patterns at various time points and in stenotic versus patent grafts. This data will enable the development of targeted therapeutic strategies to modulate the development of stenosis by either proliferation of VSMC of their production of extracellular matrix components.

1.3.2 Tunica Media, Tunica Adventitia, and the Extracellular Matrix

The tunica adventitia and extracellular matrix (ECM) are significant in that they are the source of resident stem cells and provide biomechanical support to the vessel, respectively. The tunica adventitia is the outermost layer of the vascular wall. It is

10 comprised of extracellular matrix components and is thought to be the source of resident stem cells. (37, 38) Recent reviews have underscored the complexity of this layer including a role in endothelial and VSMC interactions, a potential source of VSMC within the vessel wall, and as a mediator of inflammation. (39) Extracellular matrix components comprising native vessels include the collagen family members, elastins, and proteoglycans. These components provide structural support to the vessel and modulate the access to growth factors. Collagen I, which predominately comprises the tunica media, also plays a role as mediator of angiogenesis. (37) While the extracellular matrix

(ECM) may contribute relatively little percentage-wise to the overall vessel composition it has been proposed to play a critical role in immunoprivilege, and provide an environment for stem cell niches. (21, 40) The role of extracellular matrix on VSMC contractile and synthetic phenotype is discussed above.

Data from TEVG mouse models demonstrate the importance of collagen I and III production at early timepoints within the graft. Elaboration of these ECM components has been suggested to contribute to vascular graft stability that increases with loss of structural polymeric graft support. Additionally, in studies outlined below, MMP2 and

MMP9 activities are differentially regulated at early and late time points, suggesting that there is ongoing remodeling of the newly deposited graft ECM.

Also of particular importance to our research is the localization of TGF-1 by its latency associated peptide (LAP) and fibrillin which sequesters the ligand from receptors

11 on endothelial, VSMC, and other cells. Further highlighting the significant role of this microenvironment, aberrant structural organization of ECM components can manifest as vasculopathies seen in children such as Marfan Syndrome and Loeys-Dietz Syndrome that data suggest are mediated by aberrant fibrillan organization and availability of TGF-

. In these diseases there is increased bioavailability of TGF- due to the inability of latency associated peptide (LAP) to sequester the ligand. (37, 41) In order to optimally modulate growth and remodeling of the TEVG will require the appropriate induction of

ECM components that can store and release growth factors in a manner comparable to that of a native vessel.

1.4 Clinical Trial Results for the Treatment of Hypoplastic Left Heart Syndrome

Early clinical results for the treatment of Hypoplastic Left Heart Syndrome with the TEVG served to provide a vascular conduit that could grow and remodel to create a new blood vessel. These surgeries were successful but stenosis was identified in some clinical patients; these results motivate our preclinical research in ovine and murine models. The Tissue Engineered Vascular Graft was developed to address the key tenants of the tissue engineering paradigm: a biologically benign scaffold to provide structural support for the growth and differentiation of cellular constituents that ultimately replace the defective or absent tissue structure. (42) To this end, a polymeric scaffold comprised of PGA and PCLA was developed that could be seeded with autologous bone-marrow derived mononuclear cells. This construct has advanced growth potential in comparison

12 with Gore-tex® and Dacron® grafts in that TEVG demonstrate decreased thrombogeniticy and the ability to grow and remodel over time. (18)

After initial characterization with both small and large animal models the TEVG was utilized in a Japanese clinical trial in 2000-2004 to treat children born with HLHS. Early results of the trial deemed the device safe and efficacious for the treatment for staged palliative surgery for HLHS. Mid-term results of the first TEVG implantation were promising with limited acute events such as thrombosis and stenosis. Mid-term results showed that the graft was capable of growth and remodeling, but that there was a low- incidence of critical stenosis that was either successfully treated with angioplasty (16% of patients), stenting, or was safely monitored over time. (38, 43, 44) This result does not appear to be unique to the TEVG: vascular graft narrowing of variable clinical severity has also been characterized in the treatment for HLHS with a Fontan circuit with

Dacron® and Gortex® conduits. (17, 45) The development of critical stenosis requiring angioplasty (75% loss of graft diameter) of the TEVG has been the primary driver for experimental designs and surgical approaches for preclinical models interrogated during this dissertation work.

Tissue formation that occurs with the TEVG is characterized as neotissue formation, neo-intimal hyperplasia with Gore-tex® graft, and intimal hyperplasia in vein graft models. Clinically, TEVG stenosis shares similarities with intimal hyperplasia characterized in adult cardiovascular disease in that the tissue alters the luminal

13 geometry, flow dynamics, and predisposes to thrombosis. Aberrant neotissue formation can result in stenosis that is defined as loss of more than 50% of the nominal diameter with a 75% or greater decrease considered critical stenosis. These changes are defined both echocardiographicly and with catheter-based assessments Additionally, pressure measurements have defined increased transgradient pressures across stenotic TEVG.

Finally, angiography has characterized luminal abnormalities and luminal filling defects after the administration of iodinated contrast. In adults, stenosis is commonly identified after angiography, stent placement (in-stent restenosis) or with atherosclerotic disease.

(46, 47) Non-atherosclerotic stenosis is characterized as neointimal (myointimal) hyperplasia, in which vascular smooth muscle cells proliferate and migrate under the endothelium and produce excess extracellular matrix (ECM). (48) Altered cellular organization and extracellular matrix components could potentially have significant implications for neotissue formation and vascular functionality due to alter biomechanical and physiologic functionality. We hypothesize that VSMCs contribute to TEVG stenosis.

However, due to the success of the Japanese clinical trial, we lack tissue specimens to evaluate this hypothesis. Therefore, the development of pre-clinical models to recapitulate this process is paramount to our ability to develop treatment approaches in the clinical cohort.

Treatments for adult cardiovascular disease have been extended to the pediatric patient population, including children with TEVGs. Treatments include interventional angioplasty with balloon dilation and cutting balloons. When these therapies fail the next

14 treatment of choice is stent placement. Due to the prevalence and significance of neointimal hyperplasia and stenosis the development of drug-coated/eluting stents and catheter balloons is an active area of research. (49, 50) One of the significant differences between interventional angioplasty techniques between the pediatric and adult populations is that the latter is treated with first and second-generation drug coated balloons and stents coated that employ rapamycins (everolimus, sirolimus, zotarolimus) and taxanes (paclitaxel) to modulate vascular smooth muscle cell proliferation. (51, 52)

These compounds have been utilized as anti-neoplastic therapies and inhibit cell proliferation and migration, which would not support their use in our clinical population.

1.5 Animal models

Our lab uses a variety of preclinical models to better characterize the development of

TEVG stenosis to characterize novel imaging modalities and the molecular mechanisms that underpin TEVG stenosis. Our tissue engineered vascular grafts have been characterized in vivo across species including the mouse, sheep, and human; ex vivo characterization has also been performed on murine and ovine tissues. (28, 38, 53-56)

Our studies have demonstrated that our TEVGs function as vascular conduits and have the ability to repair and remodel over time. In both ovine and murine models by six months post-implant the majority of the scaffold is gone; in its place is a functional neovessel. These neovessels exhibit many of the required characteristics of native vessels: endothelial-lined neo-intima promotes a non-thrombogenic surface, and vascular

15 smooth muscle cells show dynamic changes in response to pressure and volume suggesting the neo-media may be functioning to regulate blood vessel tone. Despite these successes, one of the primary clinical limitations of the graft is the incidence of stenosis in clinical cohorts.

This proposal seeks to characterize the contribution of both adjacent endothelial cells and VSMCs, via targeted modulation of TGF-β pathway in endothelial cells, to the development of TEVG stenosis in an immunocompetent mouse model. Identifying the cell(s) of origin and significant regulatory pathways would enable us to better direct neotissue formation and predict graft/neotissue responses to interventional therapies.

Hypothetically, vascular smooth muscle cells within TEVG could derive from one of several cell sources: including seeded bone marrow-derived mononuclear cells (BM-

MNC), circulating stem cells, adjacent endothelial cells, vascular smooth muscle cells, or peri-adventitial cells. Several research laboratories, including ours, have also investigated the role of endothelial-to-mesenchymal transition (Endo-MT) in the cardiovascular system by utilizing a Tie-2 and Cdh5 reporting system to demonstrate lineage-traced endothelium constituted vascular neotissue and neovessels. However, we seek to specifically trace Cdh5 cells with loss of TβRI and TβRII to dissect the role of endothelial cells on TEVG stenosis and neotissue formation. Determining the cell(s) of origin and predominant molecular pathways will better characterize the TEVG and ultimately lay the foundation for the development of a second generation TEVG for use in children.

1.5.1 Ovine Models

Sheep (Ovis aries) provide a valuable pre-clinical tool to evaluate TEVG in vivo with clinical imaging of grafts that are identical to those used in the FDA clinical trial.

These tools allow us to provide relevant imaging data to aid in treatment approaches and data interpretation. Sheep (Ovis aries) have been historically used for cardiovascular studies, having a similar anatomy and physiology to the human heart. Physiologic similarities include comparable systemic and intracardiac blood pressures, cardiac output, heart rate, and size and structure of the heart. (57) Calcification of the vascular structures in surgical transplantation studies is noted for its increased incidence when compared to other large animal models. Additionally, calcification often exceeds the spectrum and severity of calcific disease noted in humans. (58) Benefits of working with sheep include their slower growth rate, ease of surgical manipulation, and ease of husbandry. (57) We have utilized this model for our surgical implantation studies to mimic similar sizes and growth rates to those seen in our pediatric population.

Our research group has utilized juvenile lamb models to evaluate the efficacy and safety of the TEVG. Initially, TEVG models utilized vascular-derived cells which were expanded in culture and seeded on to TEVG. This was the first proof-of-concept study for the ovine TEVG surgical implantation for use as a vascular interposition graft. (59)

Due to potential limitations in acquiring autologous, healthy tissue from human donors the next phase of the study evaluated the use of bone marrow derived mononuclear cells

17 for seeding of TEVG with both culture-free and culture systems. These initial studies served to demonstrate the ease of aseptically isolating BM-MNC to seed TEVG, as well as validate surgical approaches and model. (28, 60) Clinically, vascular derived cells and

BM-MNC seeded TEVG were evaluated in vivo with echocardiogram, magnetic resonance imaging (MRI), and computed tomography (CT). These modalities were used to assess vascular graft patency and growth. In several studies 1-3 animals were used as controls with unseeded TEVG that demonstrated increased incidence of acute thrombosis and stenosis. (28, 59) Furthermore, MRI and CT demonstrated somatic growth of BM-

MNC seeded TEVG. (28, 55) In vivo, this data imaging could not differentiate between graft dilation versus true somatic growth. However, interpretation of the data for the effect of bone marrow seeding is limited in these studies by the limited number of control

(unseeded TEVG) animals utilized.

Cellular constitution and differentiation has been assessed by evaluation of immunohistochemical markers for Factor VIII, von Willebrand Factor (vWF), Ephrin-B4, and deoxyribonucleic acid (DNA) content. Early TEVG experiments evaluated DNA content at 11 and 24 weeks post-implant, with older TEVG demonstrating decreased

DNA, which was interpreted as tissue remodeling. (59) Both vWF labeled neo-intima and

-SMA labeled neo-media were characterized as early as 4 weeks post-implant. (60) At late timepoints TEVG lumina demonstrated Factor VIII (FVIII), von Willebrand Factor

(vWF), and Ephrin-B4 which are markers of endothelial and venous differentiation, respectively. Concentric layers of alpha-smooth muscle actin (-SMA), one marker of

18 vascular smooth muscle cell differentiation, has also been characterized underlying the endothelial-lined lumen. (28) These results provide data to support that there is appropriate cellular differentiation or migration of mature cells along several important cell lineages including endothelial cell and a mesenchymal cell population suggestive of

VSMC differentiation at various time points of graft maturation.

Graft remodeling has been characterized by assessing collagen and elastin content, matrix metalloproteinase (MMP) activity, and polymeric degradation. Seeded

TEVGs with vascular-derived cells were initially evaluated for collagen content with assays for hydroxyproline as well as the histochemical stain Masson’s Trichrome. By hydroxyproline assay, grafts demonstrated collagen content that began to approach that of the native adjacent pulmonary artery. The presence of elastin was also confirmed. (59)

Results of a second study of BM-MNC seeded grafts demonstrated that by 6 months post- implant TEVG demonstrate similar amounts of extracellular matrix constituents, including collagen, elastin, and glycosaminoglycans as their native vessel counterparts.

The one unseeded control for the BM-MNC TEVG demonstrated increased collagen content, but decreased elastin content when compared to seeded TEVG; GAG content was comparable between the groups. (28) Another study comparing BM-MNC seeded to unseeded TEVG described a “collagen dominated” TEVG by 6 months post-implant; collagen assay did not detect a significant difference between the two groups. However, the assay evaluated acid- and pepsin-soluble type I-V collagens, but not covalently cross- linked collagen, which is a marker of maturation. (55) Other research groups have

19 employed in vitro TEVG culture systems that employ both dynamic and static seeding systems. Interestingly, dynamically seeded TEVG implanted in vivo in the pulmonary artery position in an ovine model demonstrated increased levels of collagen. (61)

MMPs play an important role in extracellular matrix (ECM) modeling and have been evaluated in numerous studies of TEVG. (55, 61) MMP localization was assessed with the radiolabeled isotope 99mTc-RP805 by single photon emission computed tomography (SPECT). 99mTc-RP805 non-specifically targets MMPs, including MMP-2, -

3, 7-, -9, -12, and -13 and has been used to assess the severity of myocardial infarction.

(62) MMP activity was increased in BM-MNC seeded TEVG when compared to the adjacent native vena cava and was significantly increased when compared to unseeded

TEVG suggesting that even at late time-points graft seeding is associated with graft remodeling. (55) Results from these experiments suggest some degree of graft remodeling with increased levels of collagen, elastins, and MMPs. These studies are unified in undertaking several approaches, both biochemical and histochemical, to evaluate the collagen content of the TEVG from 1 month to 6 months post-implant.

While collagen was not assessed with regard to stenotic/patent graft phenotype (due to the limited number of unseeded animals) patent grafts that remodeled over time demonstrated increased collagen content when compared to their native vein counterparts. This data suggests that elaboration of collagen within the TEVG is a significant process in TEVG remodeling regardless of seeding methodology or seeded cell phenotype.

In addition to evaluation of device function, assessments for pathology, including the development of calcium deposits were undertaken. Calcification of medical devices can result in significant impediments to device functionality and tissue restoration. (63)

This pathologic process appears to be accelerated in the pediatric population when compared to the adult population. The mechanism has yet to be fully elucidated and is likely partially dependent on the specific biomaterials utilized. (58, 63) In vascular cell- derived seeded TEVG increased calcium content was present when compared to native vessels, however no mineralization was identified histologically. (59) In contrast, von

Kossa (a histochemical stain for mineral) identified mineral aggregates in BM-MNC seeded TEVG. (28) Utilization of interventional angioplasty techniques to treat TEVG stenosis may result in devitalization of, or hemorrhage into, regions of neotissue within the TEVG that could predispose to calcific lesions. Therefore, it will be important to develop in vivo techniques for monitoring the development of mineral within TEVG.

Significant steps have been made towards the development of an ovine model of

TEVG stenosis. Studies have demonstrated the surgical feasibility of the model, optimization of cellular seeding, and somatic growth and tissue remodeling up to 6 months post-surgery. Both “open” and “closed” seeding techniques were compared with an ovine model, demonstrating similar efficacies, but significant improvements towards decreased anesthetic time were characterized with the “closed” seeding method. (64)

While TEVG have been historically evaluated as models of patency (seeded grafts)

21 versus stenosis there is limited data regarding the effect of unseeded grafts. Future studies will need to employ additional animals for control groups for statistical robustness and to characterize a reproducible model of stenosis. Additionally, review of the historical data has suggested that collagen forms at early time points and is present in the mature TEVG.

It has yet to be determined if this neotissue represents an intermediary step to the path of a mature blood vessel or if this change represents an abnormally fibrotic neovessel.

Further studies to assess TEVG functionality and biomechanical characterization is needed. Once a reproducible model of TEVG stenosis is developed, further assessment with in vivo modalities such as intravascular ultrasound and angiography are indicated.

These modalities will allow for real time assessment of TEVG stenosis, dilation, growth, and remodeling.

1.5.2 Murine Models

Evaluation of TEVG stenosis in mouse (Mus musculus) models allow us to undertake mechanistic studies to dissect the signaling pathways and potential pharmacologic treatment approaches. Small animal models, such as the mouse present the ideal background in which to study the role of genetics, molecular pathways, and tissue environment. Elegant studies can be constructed to evaluate the role of genes and effects of treatments. One benefit of these small animals is that they are genetically and phenotypically well characterized, with known lesion incidence for most background strains, including our most commonly utilized mouse strain: C57BL/6. Therefore, we

22 have sought to generate a conditional genetic model of endothelial-specific loss of the

TGF-β receptors, TβRI and TβRII on a mouse model that had been backcrossed to

C57BL/6. Backcrossing to C57BL/6 was performed to evaluate the effect of TGF-β pathway modulation due to the high incidence of TEVG stenosis observed in this strain.

Other specific challenges with working with mouse models include their small size and inherent technical challenges associated with surgical manipulation. However, these technical challenges have been overcome to generate a reproducible model of TEVG implantation and stenosis in a variety of murine models.

Our lab has pioneered the surgical approaches for the implantation of vascular interposition grafts of polymeric vascular grafts in both the arterial and venous positions in Severe Combined Immune Deficient (SCID) and C57BL/6 mice. (53, 65, 66) Initial

TEVG studies with SCID/bg mice identified graft infiltration at 3 weeks post-implant by multinucleated giant cells (MNGC) that labeled with Mac-3, a macrophage marker. Scant collagen formation by Gomori Trichrome histochemical staining was noted within the

TEVG, however a fibrous capsule was noted in the neo-adventitia. (66) After successfully demonstrating the feasibility of implanting 20-21 gauge TEVG grafts the next goal was to evaluate the effect of seeding human-derived BMCs on TEVG. This study demonstrated that the cells initially seeded on TEVG disappear within the first week of implant while host-derived mononuclear cells populate the TEVG. Interestingly, when compared to unseeded TEVG at chronic timepoints seeded grafts demonstrated markedly increased mononuclear cell infiltration. Similar to results from ovine studies, at

23 early timepoints cells express markers of smooth muscle cell differentiation (-SMA) and of endothelialization (vWF). (54) The results of the studies demonstrated the important temporal role cell seeding plays in graft remodeling and neotissue formation.

After initial work with the SCID/bg model the next phases of the study utilized immunocompetent C57BL/6 mice. These studies have demonstrated that TEVG seeded with BM-MNC disappear within the first week of implantation and do not contribute to graft neotissue formation. Mice were irradiated and received transplanted bone marrow labeled with green fluorescent proteins from animals of the opposite sex. CD45+ BM-

MNCs were initially seeded on the graft but were no longer present within the graft by 7 days post-implantation. Instead, the CD45+ cells were been replaced by monocytes and macrophages. (54, 67) At later time points a majority of cells repopulating both the neointimal and neo-medial compartments originated not from the bone marrow, but from the adjacent vasculature by absence of colocalization of the Y chromosome with vWF or calponin. In contrast, the majority of F4/80 cells within the graft, a macrophage marker, colocalized with a marker for the Y chromosome indicating their bone marrow origin.

(67) By 6 months post-implant cells within the graft did not label for stem cell markers stem cells antigen-1 (Sca-1), CD117/Mast/stem cell growth factor receptor (c-kit), and

Sox-2 or sex determining region Y-box 2. (67)

Time course studies of extracellular matrix constitution and remodeling were undertaken to assess grafts at 1-, 2- and 4-weeks post implant. Grafts seeded with male

C57BL/6-derived BM-MNC were implanted into C57BL/6 female mice. TEVG demonstrated collagens I and III increasing from weeks one to two by Masson’s

Trichrome staining and rt-PCR, while collagen IV steadily increased with the highest levels at week 4. Type III collagen preceded the formation of collagen I. By week 4 levels of collagens I and III had decreased. Hydroxyproline assay demonstrated that total collagen levels were highest at weeks 1 and 2. (68) In contrast to the previous studies with SCID/bg mice, C57BL/6 mice demonstrated increased amounts of collagen when comparing histochemical staining (Gomori Trichrome for SCID/bg and Masson’s

Trichrome for C57BL/6) at 3 versus 4 weeks post-implant, respectively. In the C57BL/6 mice collagen formation was noted around graft polymers and immediately underlying the neointimal layer. This may suggest differences in the animal model as well as the influence of modified inflammatory pathways for cells derived from the bone marrow.

There may also be differences in macrophages derived from the bone marrow from male and female mice. Interpretation of this study is also limited by the lack of a control cohort to compare the natural timecourse of remodeling.

Within this cohort, graft remodeling and maturation were assessed by the development of elastins, GAGs, and MMPs. Fastin™ colorimetric assay and qPCR showed the highest elastin levels by 4 weeks, and GAGs were also at their highest levels at this time by Blysan™ colorimetric assay. MMP-2 and MMP-9 activity were assessed with histochemical stains, qPCR, and in situ zymography. (68) While MMP-2 levels increased over the timecourse of the study, MMP-9 levels were the highest at early

25 timepoints. This is in accordance with the role of MMP-9 in innate foreign body reactions in response to biomaterials. (69) This work proposed that the increases in MMP-2 is driven by their production by mesenchymal cells within the graft as a response to a pro- inflammatory environment. (68)

Construction of TEVG with polyglycolic acid (PGA) fibers and polycaprolactone and polylactic acid (PCLA) confers biocompatability and biodegradability to the TEVG. The primary mechanism of TEVG degradation is by hydrolysis of the ester bonds. (70)

Polymeric degradation is typically assessed by visual confirmation of photomicrographic sections of the graft and/or assessment of polymeric birefringence under polarized light.

(68) Degradation of the polymeric fibers is complete by about 6 months post-implant.

(71) Additionally, macrophages that infiltrate the TEVG may also contribute to polymeric degradation. Previous studies in our lab have demonstrated shifts in the

M1/M2 ratios, or classically and alternatively activated macrophages, in stenotic and patent grafts. A shift away from the M1 phenotype, characterized as in intermediary

M2/M1 phenotype was observed in stenotic grafts, while treatment with an inhibitor of

TβRI, that precluded TGF- pathway signal transduction resulted in TEVG with increased M1 macrophages and stenosis rates. (72, 73) Further research on the mechanisms of polymeric degradation, especially in the context of the TGF- pathway and macrophage activation have been undertaken.

To further evaluate the role of macrophage infiltration on TEVG formation a

C57BL/6 macrophage-depletion model was employed and assessed at 3-days, 1-, 2-, and

8.5-weeks. TEVGs from macrophage-depleted mice had decreased incidence of TEVG stenosis when compared to undepleted mice, however the former mice also demonstrated a paucity of neotissue formation and cellular differentiation along endothelial and VSMC lineages and the production of collagen. There was also differential activation of macrophage phenotypes with depleted mice shifting towards a classical M1 phenotype and undepleted mice shifted to a M2, or alternative, phenotype. M2 macrophages play an important role in tissue remodeling and angiogenesis. (74, 75) M1 macrophages, initially seen in both experimental groups, may be needed to incite the initial inflammatory cascade to response to the graft as a foreign body. (76) In this study their absence in macrophage-depleted animals was associated with decreased tissue remodeling including cellular differentiation and production of extracellular matrix (collagen III), however the study does not define the age of the grafts that this assessment occurred. The results of this study are consistent with previously published data highlighting the role of the inflammatory system in the modulation of vascular repair and remodeling. (77, 78)

1.5.3 Mechanisms of TEVG Stenosis

The mouse model has allowed us to identify several significant pathways that contribute to TEVG stenosis, including endothelial to mesenchymal transition, and the TGF- pathway. While these studies have identified that seeded cells do not contribute to

27 neovessel reconstitution they do play a beneficial role in decreasing TEVG stenosis. Data from murine studies has demonstrated the important role of the phenotypic modulatory switching of macrophages between M1 and M2 activation. While M1 activation is initially needed within the graft, ultimately a M2 macrophage phenotype would optimally promote tissue remodeling because prolonged presence of M1 macrophages are seen with increased stenosis. Theoretically, macrophages are one several important sources for

TGF-1 within the graft, in addition to endothelial cells and VSMC. This ligand can promote endothelial cells to undergo endothelial to mesenchymal transition and contribute to proliferation of a mesenchymal cell phenotype within the graft. In contrast,

TGF-1 promotes a contractile phenotype and decreased synthesis and proliferation of

VSMC. We hypothesize that modulation of these pathways could ultimately decrease

TEVG stenosis.

1.5.4 TGF- Pathway

Directed targeting of the TGF-β pathway has successfully reduced the incidence and severity of TEVG stenosis in our mouse models and provides the foundation for mechanistic studies to evaluate its role in macrophages, endothelial cells, vascular smooth muscle cells. TGF-β receptors, including TβR-I and TβR-II are expressed on cells throughout the body, including endothelial cells, macrophages, and smooth muscle cells, which are of interest to our research group. (79) The TGF-β signaling pathway plays a significant role during embryogenesis, homeostatic maintenance of normal post-natal

28 cells, and phenotypic switching of VSMC. Significantly, complete loss of TβRII during embryogenesis prevents vasculogenesis, and partial loss results in cardiovascular malformations. (80) During homeostasis these receptors form heterodimers that interact with their ligand, TGF-β, to induce several different signaling cascades, including the canonical nuclear agonists Smad2/3. (Figure 2) (79, 81) In our Tie2 Endo-MT study, we identified a statistically significant increase in TGF-β levels in stenotic versus non- stenotic TEVG. This increase could be abrogated by the administration of SB431542 a small molecule that specifically targets TβRI. Vascular stenosis has also been attenuated with TGF-β antibody antagonism, and knockdown of the canonical TGF-β pathway

SMAD2/3 with shRNA. (82)

TGF-β1 has been implicated in VSMC phenotypic modulatory switching and is thought to promote a contractile phenotype and downregulates synthetic programming that results in extracellular matrix remodeling. (33) In experiments evaluating endothelial-to-mesenchymal transition (Endo-MT), decreased TGF-β1 resulted in expression of early markers of VSMC (-SMA, and SM22a) but not late markers (MHC, calponin, smoothelin). VSMC not derived from Endo-MT demonstrated both early and late markers. (82) One mechanism in which TGF-β1 works to further promote VSMC differentiation is via Notch signaling. (80) Other in vivo studies have also characterized a correlation between TGF-β1 levels and -SMA levels. (83) Our previous studies with a

Tie2 reporter to characterize EMT demonstrated that antagonizing TβRI resulted in decreased stenosis and also an overall decrease in -SMA and calponin positive cells;

29 these regions are also associated with increased TGF-β1 expression. (29, 84) Further dissecting the origins of VSMC and the relevant molecular pathways involved in TEVG neomedia reconstitution and stenosis development will provide the foundation for new therapeutic approaches for evaluation in the pre-clinical ovine model.

1.5.5 Endothelial-to-Mesenchymal Transition

Endothelial-to-mesenchymal transition (Endo-MT) occurs when endothelial cells downregulate markers of mature endothelium and upregulate expression of mesenchymal markers to express and intermediate phenotype that may contribute to TEVG stenosis.

TGF-β1 is essential to the process of endo-MT. Endothelial culture systems have demonstrated Endo-MT occurs via a Smad-independent pathway via upregulation of

Snail1. (85) Our research indicates that Endo-MT may contribute to TEVG stenosis.

Endothelial-to-mesenchymal transition occurs when endothelial cells downregulate markers of mature endothelium (CD31, VE-cadherin) and upregulate expression of mesenchymal markers such as -SMA, fibroblast specific protein 1, and transgelin, among others. This process is observed both in utero during normal cardiogenesis and postnatally in response to vascular disease. (86) In vitro cell culture studies have demonstrated that this process is required by the transcription factors Snail, Slug, and

Twist. These transcription factors regulate transmembrane receptor complexes involved in cell-cell junction stability. (87) Other in vivo arterial graft models have demonstrated the role of the TGF-β/Smad2/3-Slug signaling pathway in graft stenosis. (82) Several

30 studies undertaken by our lab and others have provided data to support the role of endothelial-to-mesenchymal transition in the development of TEVG stenosis by modulation of the TGF- pathway.

Lineage tracing of endothelial cells within TEVG were undertaken with Tie-2cre and Cdh5-creERT2 mouse models to evaluate for Endo-MT. For the first time, experimental groups both received unseeded TEVG in order to create a stenosis model.

Tie2 and Cdh5-lineage cells were confirmed to reconstitute the TEVG and that greater percentages colocalize with SMA in occluded TEVG. Stenosed grafts also demonstrated increased RNA levels of TGF-β1 and TβR1 at 1- and 2-weeks post-implant. Treatment of mice with SB431542, a non-specific inhibitor of the TGF-B pathway resulted in decreased colocalization of these markers. However, RNA transcripts were not assessed by PCR in the treated cohort. This data suggests that the TGF-β pathway plays a role in

Endo-MT within murine TEVG. (84) However, this model non-specifically targeted the

TGF-β pathway in all cells within the graft. To address the question of TGF-β origin and primary cell target C57BL/6 mice with seeded and unseeded TEVG. Stenotic TEVG at acute (2 week) timepoints demonstrated increased TGF-β1 and latent TGF-β binding protein 1 (LTBP1), so 6-month implant studies were designed to assess how short-term administration SB431542 could modulate the development of unseeded TEVG when compared to traditional seeded and unseeded TEVG. Mice that received SB431542 within the first two weeks of TEVG implant achieved similar patency rates and as seeded

TEVG. These results demonstrate that early temporal targeting of the TGF-β pathway can

31 positively influence TEVG remodeling and neovessel formation at late timepoints.

Interestingly, when returning to evaluate the 2-week implant model monocyte/macrophage numbers were not significantly different between SB431542 treated and control animals. However phenotypic differences in these cell populations were characterized with control animals exhibiting a shift away from M1 to the M2

(classical) phenotype, but SB431542 animals favoring an M1 phenotype. (73) The results of these studies suggest that the fate of stenotic and patent TEVGs are at least partially dependent on the TGF-β pathway potentially via modulation of Endo-MT, and macrophage and VSMC phenotypic modulatory switching. Future studies will need to dissect their intricate and temporal roles in neotissue formation with respect to modulation of the TGF-β pathway.

- 3

Stenosis Patency

w -

Figure 1: Schematic of neovessel formation in a mouse model. a) A PGA/PLCL copolymer graft is implanted as a caudal vena cava interposition graft. By as early as two weeks in our mouse models endothelial cells (CD31+) line the neointima and macrophages (F4/80+) have migrated to the wall of the TEVG. Smooth muscle cells (α-SMA+) are also present within the neomedia of the graft. By 6 months post-implant the majority of the polymer has degraded and the neovessel demonstrates all three functional layers: neointima, neomedia, and neoadventitia. b) Dichotomous in vivo response to TEVG implant. TEVG implanted as vascular interposition grafts in C57BL/6 mice will demonstrate either stenosis or patency by 2 weeks post-implant. Scale bar = 50 µm

Figure 2: TGF- signaling pathway.

A) TGF-1 is sequestered in the extracellular matrix by latency associated peptide (LAP). When released the ligand associates with the TRII receptor. TRII tphosphorylates TRI and heterodimerizes. TRI autophosphorylates to induce the intracellular canonical (pSmad2/3) and non-canonical (p-ERK1/2) signaling pathways, among others. B) Schematic of receptor association.

1.6 Study Objectives

The objectives of this dissertation serve our long-term goal to develop a second- generation composite vascular graft for use in children by elucidating the mechanism of

TEVG stenosis in immunocompetent mice (Mus musculus) and sheep (Ovis aries) models. Ultimately the findings from these studies will improve the clinical outcomes for children on the FDA clinical trial with the TEVG by providing a better understanding of the interventional strategies employed to treat TEVG stenosis. Data from both the large and small animal models will result in targeted therapies for this patient population.

The first aim seeks to modulate the development of TEVG neotissue stenosis in a large animal model by targeting endothelial cells via the TGF-β signaling pathway and/or the use of interventional angioplasty. The working hypotheses are: therapeutic tears impact neotissue formation and the TGF-β pathway by modulating vascular smooth muscle cells. It is hypothesized that neotissue formation and pathologic stenosis can be modulated with interventional angioplasty techniques and/or pharmacomodulation.

Chapter 2 validates the use of intravascular ultrasound (IVUS) for the evaluation of

TEVG stenosis and graft remodeling over time. IVUS is compared to the current gold standard modality of assessment: angiography. We demonstrate that IVUS is superior to angiography for evaluation of 3D geometry and stenosis. IVUS is also validated against histomorphometric assessments of neotissue formation. Finally, this study demonstrates a

35 natural time course for endothelialization and vascular smooth muscle cell proliferation within the TEVG. Data generated from this study provides data to evaluate neotissue formation and polymeric degradation in vivo with IVUS in our patient cohort.

Chapter 3 provides data to support the second and third aims of the dissertation.

Aim 2 seeks to identify the source(s) of vascular smooth muscle cells in TEVG neotissue.

The working hypotheses are: 1) adjacent vascular smooth muscle cells form neotissue via migration, proliferation, and phenotypic modulatory switching. 2) endothelial-to- mesenchymal transition is responsible for a subset of vascular smooth muscle cells present in neotissue. The third aim seeks to characterize the role of the TGF-β receptor pathway in endothelial cells in the development of TEVG neotissue. The working hypotheses are: neotissue formation is modulated through the TGF-β pathway; and 2) developing a double TβRI and TβRII knockout in endothelial cells (Cdh5 or cadherin 5 lineage traced cells) will result in decreased incidence of TEVG stenosis by modulating neotissue formation. The study described in Chapter 3 utilizes a Tgfbr1f/f; Tgfbr2f/f;Cdh5- creERT+/+;mTom-EGFP constitutive inducible model to evaluate endothelial-cell specific loss of TβRI and TβRII on the development of TEVG stenosis. Results from this study demonstrate a population of GFP+/α-SMA+ and GFP+/vWF- cells within the TEVG suggesting that endothelial-to-mesenchymal transition has occurred. Additionally, tamoxifen-treated mice demonstrated decreased TEVG neotissue formation when compared to vehicle-treated animals providing further data to support the role of TGF- in TEVG neotissue formation in the absence of stenosis. Finally, we show that there is

36 modulation of both canonical and non-canonical TGF- signaling between tamoxifen and vehicle-treated TEVG.

Chapter 2: Intravascular Ultrasound Characterization of a Tissue-Engineered Vascular

Graft in an Ovine Model

2.1 Abstract

Patients who undergo implantation of a tissue-engineered vascular graft (TEVG) for congenital cardiac anomalies are monitored with echocardiography, followed by magnetic resonance imaging (MRI) or angiography when indicated. While these methods provide data regarding the lumen, minimal information regarding neotissue formation is obtained. Intravascular ultrasound (IVUS) has previously been used in a variety of conditions to evaluate the vessel wall. The purpose of this study was to evaluate the utility of IVUS for evaluation of TEVGs as an in vivo modality in our ovine model with the purpose of correlating tissue characterization with the FDA clinical patient cohort. Eight sheep underwent implantation of TEVGs either unseeded or seeded with bone marrow-derived mononuclear cells. Angiography, IVUS, and histology were directly compared. Intima, media, and graft were identifiable on IVUS and histology at multiple time points. There was strong agreement between IVUS and angiography for evaluation of luminal diameter. IVUS offers a valuable tool to evaluate the changes within TEVGs, and clinical translation of this application is warranted.

2.2 Introduction

Approximately 40,000 babies born in the US are affected by congenital cardiac anomalies every year (7). Nearly 25% of these children will require surgical interventions that can require synthetic patches or conduits (88, 89). Our lab has pioneered a biodegradable tissue-engineered vascular graft (TEVG) for use in these conditions that is composed of polyglycolic acid (PGA) fibers that are knitted into a tube and coated with a 50:50 copolymer of polycaprolactone and polylactic acid (PCLA). In contrast to typical synthetic conduits, tissue engineered scaffolds offer the potential for growth in this pediatric population, while also mitigating some of the concerning complications related to traditional vascular prostheses such as infection and chronic inflammation. The TEVG was applied in a first-in-man clinical trial in 25 children undergoing an extracardiac Fontan procedure (38, 90). Results from this trial identified that the grafts demonstrated growth, but there was a 30% incidence of stenosis and half of these patients required transcatheter interventions. A better understanding of in vivo tissue analysis with imaging techniques will allow for the characterization of neotissue formation and direct treatment modalities in the clinical cohort.

Currently, the TEVGs are followed in a similar fashion to other vascular conduits used in the surgical reconstruction of congenital cardiac anomalies, which typically includes routine echocardiographic imaging. If required, additional imaging modalities, such as computed tomography (CT), MRI, angiography, or intravascular ultrasound

(IVUS) may be used to evaluate complications associated with the TEVG, such as the development of stenosis or aneurysmal dilation. While these imaging modalities have been used to characterize traditional synthetic grafts in both animals and humans, only echocardiogram, MRI, pressure measurements, and angiography have routinely been used within the patients receiving TEVGs. One limitation of echocardiogram and angiography is the inability to provide adequate analysis of the vessel wall. Transthoracic echocardiography does not have adequate resolution to interrogate the morphometry and structure of the TEVG wall. Similarly while angiography remains the clinical gold standard for evaluation of the lumen, it only indirectly evaluates the vessel wall.

IVUS has more recently been utilized in a wide variety of intravascular conditions. It is most frequently described in the adult population with placement of stents for the treatment of aneurysms or atherosclerosis (91-95). The advantages of this technique are a limitation of contrast use, a 2-dimensional view of the vessel lumen, an ability to determine true from false lumens, and accurate assessment of stent apposition to the luminal wall (96-98). In addition, IVUS can be used to evaluate pathology within the walls of vessels, as with atherosclerotic plaques or peripheral vascular disease (98-100).

Studies show that histology can correlate directly with IVUS, identifying necrotic, lipidic, fibrotic, and calcific areas within vessel walls (99, 101).

The use of IVUS in pediatric conditions is less thoroughly described. However, it has been used to evaluate children with pulmonary hypertension, valvular disease, pulmonary

40 artery stenosis, coarctations, and Kawasaki disease. Specifically, findings of medial hypertrophy and intimal hyperplasia have been correlated with increased pulmonary hypertension (102-105). IVUS has also been used in a fashion similar to that in adults in examining the placement of intravascular stents (91, 106). These conditions share a commonality in that they all lead to vessel wall changes, suggesting that the use of IVUS in patients receiving TEVGs would allow surveillance of growth, remodeling, and the development of pathologic lesions within the walls of the evolving neovessel, which would surpass the capabilities of traditional angiographic imaging. These changes in vessel morphology have been previously characterized histologically in both murine and ovine models, but association with IVUS could provide a potentially useful tool for monitoring TEVGs in the clinical setting. Therefore, the purpose of this study is to examine the correlation between histology and IVUS imaging and to evaluate the agreement in measures of vessel morphology between IVUS and angiography using an ovine model of TEVG implantation. IVUS has not previously been utilized to characterize neotissue formation or synthetic graft materials within the pediatric population.

2.3 Materials and Methods

2.3.1 Scaffold Fabrication

TEVG scaffolds (12-20 mm inner diameter, 1.0 mm wall thickness, 13 cm length) were provided by Gunze Ltd. (Tokyo, Japan). Scaffolds were fabricated from seamless tubes of knitted poly-glycolic acid (PGA) fibers coated with a 50:50 copolymer solution of polycaprolactone and polylactic acid (PCLA) following GMP protocols as previously described (90). The scaffolds used in this study are identical to those used in our current clinical trial (FDA IDE 14127). (Figure 3 a, b)

2.3.2 Bone Marrow Harvest

4 of 8 juvenile sheep weighing between 18-35 kilograms (Dorset and Finn-Dorset crosses, PH Dorsets, Ohio) underwent bone marrow harvest for graft seeding for the seeded group. The unseeded group did not undergo bone marrow aspiration. On the day before (n=2) or the morning of surgery (n=2), marrow was harvested from juvenile sheep weighing between 22-28 kg. Sheep were anesthetized using ketamine (10 mg/kg) and diazepam (0.5 mg/kg) for induction and isoflurane (1-4%) for maintenance. Animals were placed in the lateral recumbent position, and the area overlying the iliac crest was shaved and prepped in the standard sterile fashion. A 2mm incision was made over the bone and an aspiration needle was inserted. Heparinized syringes (100 units/ml) were used to aspirate 4.60.9 ml/kg of bone marrow.

Figure 3: TEVG Seeding and Imaging Methods. (a, b) Tissue-engineered vascular graft prior to seeding. (c) Seeding of graft with BM- MNC. (d) A representative angiographic image with sites of measurement; abdominal CaVC (line), low CaVC (dotted line), proximal anastomosis (arrow head), midgraft (dashed line), distal anastomosis (arrow), and high CaVC (double line). (e) Comparison of histology and IVUS showing concentric circles of probe, lumen, endothelium, neotissue, graft, and surrounding connective tissue.

2.3.3 Scaffold Seeding

Following harvest, the bone marrow was processed through Ficoll density gradient separation as previously described to isolate the bone marrow mononuclear cells

(BM-MNCs) (29). In brief, the marrow is filtered through 100 μm cell strainers to remove clot and bone spicules. A 1:1 dilution with PBS is performed and the bone marrow is layered onto Ficoll 1077 (Sigma-Aldrich, St. Louis, MO, USA). After centrifugation, the plasma and mononuclear cell layers are isolated. The mononuclear cell layer undergoes 2 washes with PBS to yield a cell pellet that is diluted in 20 mL of

PBS for the purposes of seeding the graft. The cells are vacuum-seeded onto the graft and

TEVG is incubated in plasma at 37°C, 5% CO2 until implantation. (Figure 3 c)

2.3.4 Implantation

Animals (n=8) were placed in a lateral recumbent position on the operating table.

Anesthesia was maintained with isoflurane. The right thorax was prepped and draped in the normal sterile fashion. An incision was made in the seventh intercostal space and the thoracic caudal vena cava (CaVC) was isolated. After proximal and distal control was obtained, the vessel was divided. An unseeded (n=4) or seeded (n=4) tissue-engineered vascular graft was sewn in place using a running absorbable monofilament suture and covered with thrombin glue. The chest wall, overlying muscle, and skin layers were re- approximated with absorbable braided suture.

2.3.5 Interventional Monitoring

Animals were scheduled for monitoring immediately post-TEVG implant and 1-,

2-, 3- and 6- months with additional imaging if emergent disease was noted.

2.3.5.1 Angiography

After sedation and intubation, the animals were placed in a left lateral recumbent position and the right cervical region was prepped in a sterile fashion. The right internal jugular vein was cannulated and a 9-French sheath (Terumo, Somerset, NJ) inserted. A wire and catheter were passed through the right atrium, into the caudal vena cava

(CaVC), and across the graft. A 6-French Multi-track catheter (B. Braun, Bethlehem,

PA) was then used to obtain an angiogram through instillation of ioversol 68%

(Mallinckrodt Pharmaceuticals, Raleigh, NC). The images were measured at 7 points: abdominal CaVC, low CaVC, proximal anastomosis, midgraft, distal anastomosis, high

CaVC, and the area of most severe stenosis. (Figure 3 d) Stenosis was defined as 50% narrowing of the mean luminal diameter with critical stenosis defined as 75% or greater narrowing.

2.3.5.2 Intravascular Ultrasound (IVUS)

A 0.035 digital IVUS catheter (Volcano, San Diego, CA) was advanced past the graft. This was used to obtain images at the same 7 points measured during angiography.

Longitudinal images were obtained without the use of a sled. These were then analyzed using proprietary imaging software to obtain a cross-sectional luminal area. The software generates a maximum and minimum mean diameters based on the annotated points identifying the inner lumen of the TEVG; data utilized for diameter represent the mean of the two numbers for a given image slice at a defined section within the graft. The software also calculates area based upon input parameters defining, in this case, the

TEVG lumen.

2.3.5.3 Balloon Angioplasty

Balloon angioplasty was performed based on clinical discretion based on clinical signs, and evidence of TEVG stenosis.

2.3.5.4 Animal Euthanasia

At the study end point (180 days) or as clinically indicated (evidence of TEVG stenosis, or other clinical conditions such as renal failure, loss of TEVG integrity), animals were deeply sedated with ketamine (20 mg/kg) and diazepam (0.02-0.08 mg/kg), followed by

46 induction of secondary pneumothorax. A complete autopsy was performed at time of

TEVG explantation. (Figure 3 e)

2.3.6 Tissue Processing, Histology, and Immunofluorescent Staining

TEVG explants were fixed in 10% neutral buffered formalin at 4°C, trimmed, dehydrated, and embedded in paraffin. Tissue samples, which were four-five μm serial sections, were prepared and stained with hematoxylin and eosin (H&E) and Masson’s

Trichrome. Immunofluorescent (IF) stains were performed to identify α-smooth muscle actin (-SMA) and von Willebrand’s Factor (vWF) positive cells in the TEVG neotissue.

Briefly, sections were deparaffinized, rehydrated, and antigens retrieved via the citrate buffer method (pH 6.0, 90°C). Sections were blocked for non-specific antibody binding

(3% normal goat serum) and incubated overnight at 4°C with an antibody cocktail of mouse anti-human α-SMA (1:100 Dako) and rabbit anti-human vWF (1:200, Dako) which both cross-react with the ovine antigens. Antibody binding was detected by subsequent incubation with goat-anti-mouse Alexa-Fluor® 647 (1:300 Life

Technologies) and goat-anti-rabbit Alexa-Fluor® 488 (1:300, Life Technologies) followed by nuclear counterstaining with DAPI (Life Technologies). Photomicrographs were acquired with a Zeiss Axio Observer Z.1 microscope with Zeiss Axiocam 503 (dark field) and 105 (bright field) digital cameras. Exposure time was informed by the appropriate negative controls (secondary antibody and 3% goat serum for the primary incubation step).

2.3.7 Animal Subjects Statement

No human studies were carried out by the authors for this article. All institutional and national guidelines for the care and use of laboratory animals were followed and approved by the appropriate institutional committees.

2.3.8 Statistical Analysis

Lumen diameter measurements from angiography were compared to lumen diameters calculated from IVUS luminal area measurements

( ).

Similarly, lumen area measurements from IVUS were compared to lumen areas calculated from angiographic luminal diameter measurements

( ( ) ).

The degree of TEVG stenosis (%) at each time point was calculated by dividing the change in angiographic lumen diameter or IVUS lumen area by the corresponding implantation measurement. IVUS and angiographic data were considered paired if they were derived from the same level of the TEVG or native CaVC at the same time point.

Statistical analyses were performed using GraphPad Prism 7 (GraphPad Software, Inc.,

CA). Type I error was controlled at α = 0.05 and two-sided p values < 0.05 were considered statistically significant. Associations among variables were evaluated using

Pearson correlation coefficients, and significantly correlated variables were further subjected to Bland-Altman agreement analysis.

2.4 Results

2.4.1 Animal Outcomes

A total of 8 animals underwent implantation (4 unseeded, 4 seeded). Of these animals 75% (n=6) did not reach the end point of 6 months. In vivo complications included stenosis (n=4), graft dilation (n=1), thrombosis (n=4), wound infection (n=1), pericardial tamponade (n=1), pneumonia (n=1), and renal insufficiency (n=3). (Table 1)

TEVG Days on Study Post Cause of Death Stenosis (>75%) Graft dilation Thrombosis Wound Pericardial Renal Mineralization Implant (Acute) Infection Tamponade Insufficiency (Histology Confirmed) X275 Unseeded 74 Anesthetic associated Undetermined cause of death 1 0 1 0 0 0 0 Prolonged cross clamp time; X267 Unseeded 0 Surgical associated loss of anastomotic integrity 0 0 1 0 0 0 0 X277 Unseeded 70 Elective euthanasia No underlying pathology 1 0 1 0 0 0 0 X274 Unseeded 36 Anesthetic associated Pleural effusion 1 0 1 0 0 0 0 5901 Seeded 180 Elective euthanasia 1 0 0 0 0 0 1 4937 Seeded 12 Surgical associated Renal failure 0 0 0 0 0 0 0 5029 Seeded 196 Elective euthasia 0 0 0 0 0 1 1 Cardiac arrest post-imaging 5018 Seeded 53 Anesthetic associated procedure 0 1 0 1 1 1 1

Table 1. Animal Outcomes.

Animals were on study from 0-196 days post-TEVG implant with associated causes of death and other study outcomes. Causes of

50 death were classified as anethestic associated (non-surgical associated), surgical associated, or elective euthanasia. Outcomes are

binary (0 = not present or observed; 1 = present or observed).

2.4.2 IVUS Characterization of Neovessel Formation

2.4.2.1 IVUS Characterization Over Time

IVUS identified the presence (Figure 4 a) or absence of luminal thrombus deposition at the time of implantation. There was a paucity of tissue formation within the intima, media, or adventitia of the graft. At approximately 1-month post-implantation, the neointima had developed and was lined by an endothelium and neomedia inside the lumen of the graft (represented by a hyperechoic laminar layer) (Figure 4 b). The endothelial lining became more clearly defined as the tissue organized and remodeled between 3 and 6 months (Figure 4 c). At 6 months, as the graft degraded, the hyperechoic polymeric layers became less clearly defined on IVUS (Figure 4 d).

0 Days 30 Days 60-90 Days 180 Days

a.1 b.1 c.1 d.1

V I

5 mm

a.2 b.2 c.2 d.2

& H

500 µm

a.3 b.3 c.3 d.3

T M

500 µm

a.4 VWF VWF c.4 VWF d.4 VWF SMA b.4 SMA SMA SMA DAPI DAPI DAPI DAPI

F I

100 µm

Figure 4: IVUS and Histology Over Time. A comparison of representative ovine IVUS and histochemical stains at 0 (a), 30 (b), 60- 90 (c), and 180 (d) days showing the evolution of the neotissue development and graft degradation. (a, b, c, d.1) IVUS at the level of the midgraft. (a, b, c, d.2) H&E photomicrographs (2.5x) of the TEVG. (a, b, c, d.3) Photomicrographs of Masson’s Trichrome stains (2.5x) of the TEVG. (a, b, c, d.4) Immunofluorescent images of TEVG immunolabeled for smooth muscle (alpha-SMA: red) and endothelium (von Willebrand’s Factor: green) (10x). Neovessel layers are outlined as followed: dotted line – lumen; dashed line – graft polymer; solid line – adventitia. IVUS and histologic images are from Day 0 are different, but correlative animals. (n=5 animals; Day 0 n=2 unseeded; Day 30 n=1 unseeded; Day 60-90 n=1 unseeded; Day 180 n=1 seeded)

2.4.2.2 IVUS Findings over the Length of the Graft

IVUS findings of an unseeded animal on post-operative day 74. IVUS of the low

CaVC demonstrated thin-walled vessel consistent with histologic findings. At the proximal and distal anastomosis (Figure 2.3 b and d), thickening of the neointimal and neomedial layers was present in both the IVUS and histologic images. The hyperechoic inner layer (white granular “tissue”) seen on IVUS correlated with the neointima on both

Masson’s Trichrome and IF analyses. The neomedia was hypoechoic (gray to black granular “tissue”) on IVUS and characterized by a higher density extracellular matrix

(including collagen) as seen on trichrome and vascular smooth muscle cells as seen with immunofluorescence. The midgraft section (Figure 5c) was noted to have a thinner layer in this area, which correlated to a thinner neointima and neomedia (Figure 5).

U a b c d e

V I

a.1 b.1 c.1 d.1 e.1

U V

I IVUS

5 mm

a.2 b.2 c.2 d.2 e.2 *

M *

500 µm a.3 VWF b.3 VWF c.3 VWF d.3 VWF e.3 VWF SMA SMA SMA SMA SMA

DAPI DAPI DAPI DAPI DAPI

F I

100 µm

Figure 5: IVUS and Histology Along the Graft Length.

TEVGs at measured points along the length of the graft 74 days post-implantation. The top panel demonstrates the longitudinal IVUS images with points of interest annotated during manual manipulation of the catheter; (a.1-3): proximal CaVC; (b1-3): proximal anastomosis; (c.1-3): midgraft; (d.1-3): distal anastomosis; (e.1-3): distal CaVC. (a, b, c, d.2) As sled was not used, there is no correlation to the length of graft. Photomicrographs of TEVG over the length of the graft (Masson’s Trichrome). (a, b, c, d.3): Immunofluorescent images of TEVG immunolabeled for vWF (green) and SMA (red). Neovessel layers are outlined as followed: dotted line – lumen; dashed line – graft polymer; solid line – adventitia; asterisk - suture. While an endothelial lining is present along the length of the graft, differential vascular smooth muscle cell labeling along the length of the graft can be appreciated. (n=1, unseeded)

2.4.3 Comparison of Angiography and IVUS by Quantification of Lumen

Geometry

Lumen diameter approximations from IVUS imaging were significantly correlated with angiographic measurements (Fig. 6 a, Pearson r = 0.8669, p < 0.0001).

Conversely, calculated luminal area derived from angiographic imaging were not significantly correlated with IVUS area measurement (Fig. 6 b, Pearson r: 0.2886, p =

0.0832). When the degree of stenosis at each time point was calculated based on the corresponding measurements from the implant, a significant correlation between IVUS and angiography was discovered (Fig. 6 c, Pearson r = 0.6952, p < 0.0001). Significant correlations prompted quantification of the agreement between IVUS and angiography by

Bland-Altman analysis to investigate whether these modalities could be used interchangeably for the determination of lumen diameter and degree of stenosis. With regards to approximation of lumen diameter, the mean difference between IVUS and angiography was 0.739±2.055mm (Fig. 6 d, 95% Confidence Interval: 0.077 to 1.401 mm, 95% Limits of Agreement: -3.288 to 4.767 mm). While quantification of stenosis was significantly correlated between the two modalities, the agreement between IVUS and angiography was not robust (Fig. 6 e, Bias: 1.863±36.84%, 95% Confidence Interval

-5.749 to 9.475%, 95% Limits of Agreement: -70.35 to 74.08%).

Figure 6: Luminal Geometry: IVUS vs. Angiographic Quantification. A significant correlation between direct measurement of lumen diameter from angiography and calculated lumen diameter from IVUS was discovered (a, Pearson r = 0.8669, 95% CI: 0.7549 to 0.928mm, R2 = 0.7515, p < 0.0001). The correlation between direct measurement of lumen area from IVUS and calculated lumen area from angiography was not significant (b, Pearson r = 0.2886, 95% CI: -0.03903 to 0.5603mm2, R2 = 0.08331, p = 0.0832). In addition, a significant correlation between IVUS and angiography was found for determination of TEVG stenosis (c, Pearson r = 0.6952, 95% CI: 0.5703 to 0.7888%, R2 = 0.4834, p < 0.0001). Negative values indicate cases of dilation, which may confirm growth of the TEVG over time. Bland-Altman agreement analysis (difference versus mean) between IVUS and angiography revealed a strong agreement between the two modalities in measurement of lumen diameter (d, Bias: 0.739±2.055mm, 95% CI: 0.077 to 1.401mm, 95% Limits of Agreement: -3.288 to 4.767mm), however, the agreement between IVUS and angiography for determination of stenosis was not as robust (e, Bias: 1.863±36.84%, 95% CI: -5.749 to 9.475%, 95% Limits of Agreement: -70.35 to 74.08%). Dashed horizontal lines identify the limits of agreement between the two modalities. (n=47 angiographic two plane image points; n=37 IVUS image slices with image points; correlative images from the same TEVGs sections of animals at time points were sampled at 1-, 2-, 3-, and 6-months post implant)

2.4.4 Characterization of Anomalies

2.4.4.1 Strengths of IVUS

Characteristics of the graft that were clearly identifiable on angiography included anastomotic narrowing, stenosis, dilation, and compression (Figure 7). These problems are easily identified with IVUS, and correlated with angiographic imaging. While IVUS demonstrated these changes, it may be less likely to identify some of the sequelae of the pathology, such as the collateralization visualized with severe stenosis (Figure 7 b).

However, angiography may not be able to readily identify narrowing or other pathology that occurs due to its single axis view. An example of this pathology is seen in Figure

5d.1, which shows the narrowing of the vessel in the anterior/posterior direction as it traverses diaphragm. This anomaly may be less readily identified with angiography when only one view is acquired. In addition, IVUS can specifically identify and characterize intimal depositions, such as thrombus (Figure 8 a). In this comparison, the relationship between the graft and thrombus was more specifically characterized by

IVUS than angiography where only the luminal narrowing could be identified.

Following balloon dilation, IVUS was also able to identify therapeutic tears (Figure 8 b).

While this change was apparent on angiography, its spatial relationship to the graft, neotissue, and surrounding tissues could not be evaluated. A tear through the neotissue

57 and graft, as well as containment within the surrounding tissue was clearly identifiable both on IVUS and histology (Figure 8 b).

Figure 7: Abnormalities Encountered in Model: IVUS versus Angiography. Several pathologic conditions are better evaluated with imaging techniques. Narrowing is visible at anastomotic site on angiography, but can also be seen using the transverse and longitudinal images on IVUS (a.1-3). Similar results are seen with evaluation of stenosis (b.1-3), however the IVUS images demonstrate neotissue development (arrow head) at this site. Additionally, collateralization can be seen (arrow). While rare, dilation can occur, and is equally visible with angiography and IVUS (c.1-3). External compression may or may not be seen with angiography (d.1), while the combined views of IVUS (d.2-3) allow for identification of compression as the VC traverses the diaphragm. (n=4; A n=1 seeded; B n=1 unseeded; C n=1 seeded; D n=1 unseeded, but before TEVG surgical implant) Due to the dynamic nature of the neotissue and graft material, in addition to sampling timepoints histologic assessment could not convey some pathologic changes described above.

Thrombus Therapeu c Tear

a.1 b.1

n A 1 cm 1 cm

a.2 b.2

V I

2.5 mm 2.5 mm

a.3 F b.3

E T &

H T G NT 300 µm 500 µm

Figure 8: Correlation of IVUS Findings to Ex Vivo Pathology. Angiography, IVUS, and H&E comparison of thrombus (a.1-3) and therapeutic tear (b.1- 3) following balloon dilation. Thrombus can be visualized with both imaging formats and by histology (Hematoxylin and eosin; 5x). The shadow of the contained leak can be seen on angiography, but the relationship to the neotissue and graft material is better appreciated with IVUS and histology. The line on angiography denotes correlative position of IVUS and histologic images. F: fibrin; T: tear; G: TEVG; NT: neotissue. (n=3; A1 unseeded; A2-3 unseeded; B1-3 unseeded)

2.4.4.2 Weaknesses of IVUS

Histopathologic assessment offers a greater subgross resolution of neotissue formation, extracellular matrix composition, and polymeric graft degradation in comparison to IVUS. Important findings specific to histology, not identified by IVUS include mineralization or microcalcification (histologically confirmed in 3 of 8 sheep).

Sheep have been characterized as pro-calcific, with mineralization often exceeding the spectrum and severity of calcific disease noted in humans. (58) Additionally, with IVUS, as the graft matured and the scaffold degraded, the tissue layers were less readily discerned. This finding is highlighted at the 180 day time point shown in Figure 4 d.

2.5 Discussion

Currently, patients who receive TEVGs are closely monitored clinically and with serial echocardiogram. If indicated, these children may also undergo MRI or angiography. While these studies can identify dilation, stenosis, or changes in intravascular flow, they are less informative regarding the changes that are occurring within the wall of the neovessel. Data interpretation generated by these studies is limited and correlation with histologic findings in the clinical setting is also limited. Our previous work with murine models has characterized the natural time course of neotissue formation that occurs following TEVG implantation, as it progresses to a neovessel that has similar characteristics to that of native vein (72, 107). Time-course studies over 6

61 months have characterized both in vivo changes by ultrasound, as well as ex vivo with histology and biomechanical testing. As with echocardiography in the clinic, this technique is useful to determine patency versus stenosis, and provides some limited information regarding flow. However, histology continues to provide the majority of data for analysis of changes within the vessel wall in preclinical settings.

While histologic analysis has been the gold standard in both our murine and ovine models, IVUS offers the potential to evaluate the progression of changes, including neotissue formation and polymeric graft degradation, within our ovine model over time in vivo within the same animals, obviating the need for serial explantation. The results described in this study highlight the capacity of IVUS to follow changes that occur within the wall of the graft over time, including acute and chronic responses to angioplasty. In comparison to histology, the progressive changes in the endothelium, underlying neotissue, and graft degradation clearly correlate with IVUS findings. In the current study, IVUS is able to demonstrate all relevant graft components at early time points, but by 180 days these findings are no longer present. These results suggest that IVUS will be able to monitor not only neotissue development, but the degradation of the graft over time.

The IVUS analysis over the length of the graft also provides support for the use of the modality for graft analysis. Our previous work has shown that the cells seeded onto the graft modulate the innate humoral and cell-mediated inflammatory response,

62 stimulating ingrowth of cells from the adjacent native vessel (72). IVUS successfully captures the differences in thickness of the neovessel wall over the length of the graft, and these changes can be correlated with histologic images. These results suggest that

IVUS can be used to follow the formation of an endothelial lining, the development of a neomedia comprised of vascular smooth muscle cells, and the eventual maturation of the neotissue. As IVUS correlates with angiographic and histologic findings, further evaluation of this technique within the ovine model may allow characterization of changes within these tissue planes that could be used to identify grafts that will progress to stenosis or, alternatively, stenotic segments that will resolve without need for angiographic intervention. One limitation may be the availability of this modality and clinicians that are experienced with IVUS for evaluation of this specific indication.

With regards to the capacity of IVUS to quantify vessel geometry, there was a significant correlation between lumen diameters approximated from IVUS measurements and angiographic measures of lumen diameters at the same level of the graft and there existed a clear agreement between these two modalities. This indicates that IVUS is able to determine the measurements obtained by both the one-dimensional angiography while also providing a two-dimensional area, which one-plane angiography was unable to approximate. The data sets utilized to generate these data included: 1) Diameter measurements: 2-plane angiographic measurements and 3-D IVUS measurements that acquired minimum and maximum diameters; the data utilized for IVUS was an average of these two numbers. 2) Area measurements: area was calculated based on 2-plane

63 angiographic measurements and compared to 3-D IVUS measurements of area. Hence, different data sets for IVUS were utilized for this analysis and could influence their associated statistical analysis and interpretation. However, in the pre-clinical study the limitations of 2-plane angiography were identified and confirmed by limited comparison to 3D reconstructions of 180 degree angiographic imaging. This comparison demonstrated that the 2-plane lateral angiogram often did not capture the most significant differences in TEVG stenosis and dilation (data not shown). Therefore, we propose that

IVUS is the best modality to assess the asymmetrical lumens of the ovine vena cava at physiological pressure, and is therefore better suited to evaluate TEVG growth and remodeling in this model. Future studies should employ additional diagnostic imaging approaches, such as CT and MRI to characterize the in vivo formation of mineral within

TEVG. Further, traditional angiography cannot be used with confidence to approximate a luminal area when compared to IVUS. Graft lumen area may better approximate IVUS measurements when 3D angiography is employed, but this requires a significant increase in both contrast and radiation. Taken along with the advantages of IVUS evaluation of vascular neotissue described herein, the lack of strong agreement with regards to percent stenosis determinations between IVUS and angiography further supports the use of IVUS for longitudinal surveillance of TEVGs in vivo. Clinically, the ability of IVUS to obtain an accurate luminal area and diameter at physiologic conditions should allow a significant reduction of contrast and radiation exposure, an especially important concern in the pediatric population undergoing multiple cardiac catheterizations. Further research in our animal model and in the clinic will be needed to further assess this benefit.

The ability of IVUS to identify abnormalities has previously been documented including its ability to demonstrate vessel interfaces with stent use (91, 106) and to clearly identify false lumens (96). These strengths are similar to those we have characterized above when evaluating TEVGs. Examples include the ability to monitor the degradation of the graft and the ability to identify therapeutic tears and the resultant contained fluid collections in the surrounding tissue. While IVUS has many strengths, there are several weaknesses when compared to histologic techniques. While previous studies have been able to view calcific plaques within the vessel wall (100), three of the specimens demonstrated microcalcifications that were not identified on either angiography or IVUS (data not shown). Mineral aggregates were present within the midgraft, underlying neotissue comprised of endothelial cells and smooth muscle cells.

Additionally, once the graft degrades and tissue maturation has occurred at 6 months, tissue planes are not as easily identified on IVUS. However, the findings at this point are similar to those in the native vena cava, and are confirmed by histologic similarity between native and tissue-engineered vena cava at late time points. While commonly utilized imaging techniques such as angiography, CT, MRI, or echocardiogram are unable to visualize these tissue planes, optical coherence tomography (OCT) has been used experimentally to evaluate coronary plaques. This technique uses infrared light and has a high resolution of 10-15µm, but has a limited penetration into tissue (1.5 mm)

(108). While this technique might provide a more detailed picture of tissue closest to the

65 luminal surface, the depth of penetration would limit its application for imaging of the

TEVG.

Data from the current study demonstrates that serial IVUS can be used safely in vivo in animals that have undergone implantation of a TEVG in the thoracic vena cava.

The use of this technology complements angiography and provides a more detailed examination of the neovessel wall. Additionally, IVUS is able to identify common abnormalities from graft implantation and their spatial relationship to the graft material.

Our data suggests that the measurements obtained by IVUS agree with angiographic measurements and provide superior quantification of neovessel geometry, which could benefit patients in the clinical setting by decreasing contrast and radiation exposure.

Finally, IVUS offers the potential to follow both animal models and patients with TEVGs longitudinally, allowing direct correlative data from one subject over time.

2.6 Conclusions

IVUS characterization of neotissue correlates closely with angiographic and histologic changes. IVUS measurement graft lumen morphometry correlates with angiography measurements. Utilization of these modalities used in an additive fashion, with consideration for patient safety, can provide further analysis of the maturation of tissue-engineered vascular grafts.

Author Roles:

Elizabeth S. Clark: Study design and management, acquisition, analysis, and interpretation of data, manuscript preparation, surgical assistant

Victoria Pepper: Data acquisition, analysis, and interpretation of data, manuscript preparation, surgical assistant

Cameron Best: Data acquisition, statistical analysis, interpretation of data, surgical assistant, manuscript preparation

Ekene Onwuka: Data acquisition, surgical assistant, manuscript preparation

Tadahisa Sugiura: Primary surgeon, surgical reports

Eric Heuer: Surgical assistant, ex vivo assistance with data acquisition

Lilamarie Moko: Ex vivo data acquisition and interpretation (her contributions formed the basis for her poster submission to the MDSR Medical Student Research

Scholarship Symposium), compilation of medical record data

Shinka Miyamoto: Surgical observation

Hideki Miyachi: Surgical observation

Daren Berman: Primary interventional cardiologist, data acquisition, critical review of manuscript

Sharon Cheatham: Primary interventionalist team, data acquisition, critical review of manuscript

Joanne Chisolm: Primary interventionalist team, data acquisition, critical review of manuscript

Toshiharu Shinoka: Primary surgeon, critical review of manuscript

Christopher K. Breuer: Study design, data interpretation, and final approval of manuscript

John Cheatham: Primary interventional cardiologist, data acquisition and interpretation, study design and final approval of manuscript

Chapter 3: Cdh5 Deletion of TβRI and TβRII Modulates Neotissue Formation in Tissue-

Engineered Vascular Grafts

3.1 Abstract

TGF-β1 has been implicated in the development of tissue engineered vascular graft (TEVG) stenosis. This ligand has been shown to modify macrophage phenotype and may induce endothelial-to-mesenchymal transition in endothelial cells derived from the adjacent native vein. Here we demonstrate that TEVG neotissue formation, specifically collagen formation, is altered in vivo in a conditional genetic murine model of endothelial-derived (Cdh5) loss of the TGF-β receptors, TβRI and TβRII with recombination events marked by green fluorescent protein (GFP). When compared to mice with intact TGF-β signaling pathway, tamoxifen treated mice demonstrate altered neotissue formation with reduced collagen content and polymeric degradation. Despite loss of TGF-β signaling pathway in endothelial cells GPF+ Cdh5 lineage cells that are

GFP+/vWF+, GPF+/ α-SMA+, and GFP-/ α-SMA+ constitute the TEVG to promote neotissue formation. This study demonstrates that endothelial cells contribute to a sub- population of vascular smooth muscle/mesenchymal cells within TEVG and that they indirectly regulate TEVG neotissue formation. These results further support the important and early temporal role that the TGF-β pathway plays in vascular graft

69 neotissue formation within our TEVG models and the role of endothelial cells on whole- graft remodeling.

3.2 Introduction

Tissue engineered vascular grafts (TEVG) have been employed in phase I clinical trials in children born with hypoplastic left heart syndrome to create de novo vascular structures. (38, 44) Acutely, these structures must support the cardiovascular load and demonstrate anti-thrombogenic properties. Over time, the ideal vascular graft allows for growth of appropriately differentiated vascular layers that respond to physiologic changes and stimuli, and support somatic growth, while meeting the body’s metabolic demands.

(109) A more thorough understanding of the molecular mechanisms and cellular origins that underpin early graft modeling and remodeling is required to inform clinical expectations for subacute responses to angioplasty. (71)

Currently, the development of stenosis remains one of the major barriers to expanded clinical utilization of TEVG. We have previously demonstrated in mouse (mus musculus) and ovine (ovis aries) models that the neovessels are comprised of an endothelial cell lined lumen (CD31-positive), smooth muscle and mesenchymal cells (α-

SMA and calponin-positive cells), and macrophages (F4/80-positive). (67, 71) These cell lineages are present in varying ratios in both stenotic and patent neovessels over the natural time course of TEVG remodeling. (73) We previously hypothesized that

70 excessive TGF-β signaling may contribute to TEVG stenosis. Currently, the cellular sources of TGF-β1 within the TEVG are unknown; as a de novo graft the scaffold does not initially contain ligands or growth factors embedded in its matrix. It is possible that grafts seeded with bone-marrow derived mononuclear cells provide an initial source for

TGF-β1. At later timepoints a variety of cells such as macrophages, endothelial cells, and

VSMC may contribute to production of this ligand in both unseeded and seeded TEVG.

Previously we have targeted TβRI, which non-specifically targeted all cells within the

TEVG, to result in a shift in macrophage phenotype from M1 to M1-M2 and a reduction in stenosis. (72) Directly targeting macrophages within the TEVG with clodronate mouse model and a CD11b-diphtheria toxin-receptor (DTR) transgenic mouse model resulted in severe deficits in neotissue formation and failure of the graft to constitute with endothelial or VSMC. (72) Therefore, we sought to more specifically characterize the roles of other cell populations within the TEVG with regard to neotissue formation. Other studies have employed lineage-tracing of Tie-2 and Cdh5-cre cells (endothelial-origin) within TEVG to demonstrated that some of these cells gain VSMC/mesenchymal markers. (84) TGF-β is required for this process, termed endothelial-to-mesenchymal transition, to occur. (110) Therefore, we sought to evaluate whether the TGF-β pathway is required in endothelial cells for neotissue formation, graft stenosis, and VSMC phenotype switching.

For this study, we sought to specifically characterize the loss of TGF-β signaling through disruption of its receptors, TβRI and TβRII in lineage-traced endothelial cells

71 using a conditional genetic model in vivo. An unseeded TEVG model was used in mice with a C57BL/6 background, similar to our previous studies, due to the consistent incidence of stenosis in this model. Utilization of a model of stenosis would enable the characterization of the role for the TGF-β signaling pathway with specific regard to endothelial cells and their dynamic relationships with other cells that populate the TEVG, such as vascular smooth muscle cells and macrophage. Therefore, this study seeks to collectively induce the conditional knockout of both TβRI and TβRII in order to unambiguously downregulate the TGF-β signaling cascade. This strategy will allow characterization of endothelial cellular origins, and their recruitment, maturation, and differentiation into the TEVG under conditions of TGF-β inhibition.

3.3 Materials and Methods

3.3.1 Transgenic mouse model

A conditional compound transgenic mouse, Tgfbr1f/f; Tgfbr2f/f;Cdh5- creERT+/+;mTom-EGFP that was backcrossed nine generations to the C57BL/6 strain was generated to temporally evaluate the loss of Tgfbr1 and Tgfbr2 in endothelial-lineage cells in vivo (designated here R1/R2TM+ or TM-). (Figure 9) Tgfbr1f/f mice are floxed for the Alk5/Tgfbr1 gene which is excised from Cdh5 cells when treated with tamoxifen.

Tgfbr2f/f mice are floxed for the Tgfbr2 gene which is excised from Cdh5 cells when treated with tamoxifen. TβRII initiates the intracellular TGF-β signaling cascade by

72 transphosphorylation of TβRI upon association with the TGF-β1 ligand. Cdh5-creERT+/+ mice utilize a tamoxifen-sensitive estrogen reporter to enable lineage tracing of Cdh5

(cadherin-5 or VE-cadherin, a junctional adhesion marker of endothelial cells) with mTom-EGFP mice with membrane-bound reporters with tomato (red fluorescent protein) and green fluorescence protein (GFP) within the cell membrane. Cdh5 cells expressing cre excise mTom and express GFP. Briefly, Tgfbr1tm1.1Karl/KulJ [JAX 028701] mice were bred to B6.129S6-Tgfbr2tm1Hlm/Nci obtained from A.L. Moses (Vanderbilt

University, Nashville, Tennessee, USA). Offspring were crossed with B6.129(Cg)-

Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP)Luo/J [JAX 007676]. The next generation of offspring were bred with Cdh5-creERT2 (gift from R.H. Adams, Max Planck Institute,

Münster, Germany). Homozygotes were selected for breeding and experimental procedures. At 8-10 weeks of age, both male and female mice received 20 mg/kg tamoxifen (5 mg/mL working solution solubilized in corn oil) (Fisher Scientific) with one intraperitoneal injection (RI/RIITM+) or vehicle treatment (corn oil) (RI/RIITM-). (Figure

10) Genotyping for the loxP allele, WT allele, as well as loss or presence of endogenous mRNA for Tgfbr1 or Tgfbr2 were performed with PCR primers as previously characterized. (8, 9) Primer sets included: Cdh5CRE-ERT2 5’

GCCTGCATTACCGGTCGATGCAACGA 3’ (forward) and 5’

GTGGCAGATGGCGCGGCAACACCATT 3’ (reverse); Alk5flox 5’

ACTCACATBTTGGCTCTCACTGTC 3’ (forward) and 5’

AGTCATAGAGCATGTGTTAGAGTC 3’ (reverse). For the Tgfbr2 exon 2 two primer combinations were used to assess the presence of the LoxP allele and loss of the floxed

73 allele after recombination: lox1F 5’ TAAACAAGGTCCGGAGCCCA 3’ and lox1R 5’

ACTTCTGCAAGAGGTCCCCT 3’; lox1F allele as above in combination with lox2R 5’

AGAGTGAAGCCGTGGTAGGT 3’. Mice were surgically implanted with TEVG one week after the completion of tamoxifen treatment.

Tamoxifen

Endothelial cells

Vascular smooth muscle cells

TEVG Implant TGF-β1

Figure 9: Schematic for cre-mediated recombination and expression of mT/mG in Cdh5- cells.

Tamoxifen was administered by one intraperitoneal injection at 8-10 weeks of age. Cre- mediated excision of the mT cassette encoding membrane tdTomato, a type of RFP, occurs 1-7 days with ~80-90% efficiency. Ultimately endothelial cells express membrane EGFP on their cell membranes. One week later TEVG were implanted. In this model VSMC derived from Cdh5 cells via Endo-MT could potentially express membrane- associated GFP.

Cdh5 Tg r1( (Alk5) Tg r2

BP - + BP +/- +/+ BP +/+ +/-

600 350 loxP 540 loxP 200 WT 420 WT

Figure 10: Genotyping of offspring for Cdh5cre.ERT2+/+, Tgfbr1f/f, and Tgfbr2 f/f. BP: base pairs. lox P: allele with loxP insertion. WT: wild type allele. +/-: heterozygous. +/+: homozygous.

3.3.2 Graft production and Implantation

Custom made 19G grafts were generated from a 50:50 copolymer sealant solution of ε-caprolactone and L-lactide (Absorbable Polymers International) and a nonwoven polyglycolic acid mesh (Biomedical Structures) as previously described. (10) TEVG were implanted as caudal vena cava renal interposition grafts as previously described.

(11) (Figure 11 A)

3.3.3 ECHO

Mice (n=39), aged 10-12 weeks of age, that had received either intraperitoneal injections of tamoxifen or vehicle (corn oil) underwent serial echosonographic evaluation at 7 and 14 days post-implant (VisualSonics Vevo 2100) following an anesthestic protocol of 1.5% isoflurane vaporized with oxygen at a flow rate of 1 L/min. These time

75 points represented 1- and 2- weeks post-TEVG implant and the second ultrasound was performed on the day of euthanasia. (Figure 11 B)

3.3.4 Explant and Tissue Collection

TEVG were explanted two weeks post-implantation from mice aged 12-14 weeks old (n=39, 11 treated males, 11 treated females, 11 control males, and 6 control females).

Animals were euthanized with intraperitoneal injections of ketamine (300 mg/kg) and xylazine (30 mg/kg). in vivo the vasculature was perfused first with phosphate buffered solution (Gibco). Sections of TEVG underwent direct immunofluorescent analysis

(Olympus SZX7, Olympus DP71, Olympus CellSens Dimensions version 1.15). TEVG were either frozen in liquid nitrogen and stored at -80C or placed in neutral buffered 10% formalin. (Figure 11 C)

A B C SV

Figure 11: TEVG Implant, ECHO, and Explant. Continued.

Continued from page 76. A. In situ photograph of the interpositional caudal vena cava TEVG implant on the day of surgery. B. In situ image of patent (non-stenotic) TEVG at 2 weeks post-implant. C. In situ photograph of TEVG at two weeks post implant with right kidney (RK), colon (C), and seminal vesicle (SV). Dashed scale bars : 3 mm.

3.3.5 Histochemical Staining, Immunohistochemistry, and Immunofluorescence

For histology, grafts were fixed in 10% neutral buffered formalin for 24 hours, followed by dehydration with a glucose concentration gradient (10%, 20%, 30%) (Fisher

Scientific). Grafts were embedded frozen in optimum cutting tissue medium (O.C.T.), frozen, and stored at -80°C until immunofluorescence (IF) analysis. TEVG analysis included characterization of histomorphometry, polymer degradation, cellularity, and extracellular matrix deposition. Routine histochemical stains were undertaken including hematoxylin and eosin and Masson’s Trichrome. Immunofluorescent and immunohistochemical analysis was undertaken with primary antibodies against GFP

(1:300; Abcam ab13970), CD31 (1:50; Abcam ab7388), vWF (1:300; Abcam ab218333),

α-SMA (1:500; Dako M0851 Clone 1A4), Collagen I (1:300; ab6577, Abcam), Ki67

(1:300; Abcam ab15580), PCNA (1:300; Abcam 18197), TGF- (1:200, Abcam ab

66043), TRI (1:500, Abcam ab31013), and TRII (1:200, Abcam ab186838). Secondary fluorophore-conjugated antibodies included goat anti-rabbit IgG (1:300; Vector

Laboratories); goat anti-mouse (1:300; Vector Laboratories); goat anti-rat (1:300; ab150159, Abcam), and donkey anti-chicken (1:300; Jackson Immuno). Sections were mounted and stained with 20 L DAPI (63698, Life Technologies).

3.3.6 Image Analysis

Slide analysis was undertaken with the Zeiss Axio Imager.A2 microscope (Carl

Zeiss GmbH). Image analysis was undertaken with FIJI (NIH). No antibody, negative control slides were employed (1’ incubation with 3% goat serum followed by 2’ incubation with flurophore labeled antibody; initial experiments utilized slides that were blocked and imaged without antibody utilization to characterize autofluorescence).

Negative control images were thresholded in FIJI for background fluorescence and overexposure for DAPI (4',6-diamidino-2-phenylindole), 488 nanometers (nm), and 647

(nm), followed by secondary thresholding of fluorescent slides, and a third threshold that applied the secondary thresholding levels to the control slides (settings acquired from control slides often resulted in overexposure to treated slides, which were gated with lower exposures and reapplied to control slides to assess underexposure). Exposures were modified to achieve standard fluorescence exposure curves; brightness and contrast were modified to highlight the presence or absence of colocalization. Exposure, brightness, and contrast settings were then applied to all treated and control .pzi files, and converted to

.jpg files, which underwent color de-convolution for two-color colocalization anlysis with

FIJI Coloc 2 software (NIH). Colocalization protocols were employed to generate

Mander’s correlation coefficients of colocalized voxels to characterize fluorescent marker colocalization independent of signal intensity and density with the Costes thresholding correction. (111) Regions of interest included the neo-intima, neo-media, and graft polymer regions. (Figure 15)

3.3.7 Quantitative RT-PCR

PCR analysis was performed on selected markers for the TGF-β pathway (Tgbr1,

Tgbr2, Tgfb1), extracellular matrix and mesenchymal markers (Col1a1, MMP2, Acta2,

Eln), and proliferation (PCNA). TEVG were placed in Trizol (Fisher Scientific) and homogenized (TissueLyser II, Qiagen) and underwent chloroform treatment, RNA extraction, and DNAse treatment (Total RNA Purification Kit, Norgen Biotek). RNA yield was characterized by a Nanodrop 2000 (Thermo Fisher Scientific) and for RNA

Integrity Analysis (Agilent 2100 Bioanalyzer, Biomedical Genomics Laboratory,

Nationwide Children’s Hospital). Equivalent amounts of cDNA were reverse transcribed using the Maxima First Strand Synthesis Kit (Thermo Fisher Scientific). PCR was performed in duplicate using Thermo Maxima Probe/ROX qPCR Master Mix and Roche

Universal Probe/Primer sets (Roche Diagnostics Corp). Relative expression was determined using 2-Ct analysis and normalized to the ribosomal transcript Rp113a

(NM_012423.2). Primer sequences: Tgbr1 (NM_009370.2). 5’ cagctcctcatcgtgttgg

(forward) and 5’ cagaggtggcagaaacactg (reverse). Tgfbr2 (NM_009371.2) 5’ cgttcaagcagacggatgt (forward) and 5’ gctcgtaatccttcacttctcc(reverse). Tgfb1

(NM_011577) 5’ tcagacattcgggaagcagt (forward) and 5’ tcagacattcgggaagcagt (reverse).

Col1a1 (NM_007742) 5’ acctaagggtaccgctgga (forward) and 5’ gagctccagcttctccatctt

(reverse). MMP2 (NM_008610) 5’ gtgggacaagaaccagatcac (forward) and 5’ gcatcatccacggtttcag (reverse). Acta2 (NM_007392) 5’ cccacccagagtggagaa (forward) and

5’ acatagctggagcagcgtct (reverse). PCNA (NM_011045) 5’ ctagccatgggcgtgaac (forward) and 5’ gaatactagtgctaaggtgtctgcat (reverse). Eln (NM_007925) 5’ gctgatcctcttgctcaacc

(forward) and 5’ gggaactccaccaggaagtc (reverse).

3.3.8 Immunoblot analysis

TEVG tissue lysates were prepared with tissue lysis buffer (0.2% Triton X-100,

10 ug/ml aprotinin, 10 ug/mL leupeptin, 0.5 mM PMSF, 500 um Na3VO4), followed by centrifugation (14,000 rpm, 10:00 minutes, at 4°C). Protein (35 ug) was separated by

SDS-PAGE and transferred to PVDF membranes. Primary antibodies used for immunoblotting were raised against TβRI (1:1000; Abcam ab31013), TβRII (1:1000;

Abcam ab186838), p-Smad2 (1:1000; Abcam ab53100), p-ERK 1/2 (1:1000; Cell

Signaling 9911), and GAPDH (1:2000; Novus Bio NB300-221), followed by horseradish peroxidase-conjugated secondary antibodies (1:5000, Fisher 31460; 1:5000, Cell

Signaling 7074). Two ECL substrates were used separately for chemiluminescence detection of bound antibodies (SuperSignal™ West Dura Extended Duration Substrate and Pierce™ ECL Western Blotting Substrate, Thermo Fisher). Immunoblot band densitometry was evaluated with ImageJ gel analysis software; data were normalized to

GAPDH. For this data set the values for the two p-Smad2 values were averaged.

3.3.9 Statistical Analysis

Data are presented as mean standard error of the mean. Prism 6.0 (GraphPad

Prism Software) was used to evaluate paired, two-tailed t-tests for comparisons between groups, and one-way ANOVA for multiple groups with p < 0.05 considered significant.

3.4 Results

3.4.1 Graft Phenotyping and Lineage Tracing

Cre-mediated recombination was induced in endothelial-lineage cells traced with the expression of membrane-associated EGFP and tdTomato, TβRI, TβRII, and SMAD2.

Direct and indirect immunofluorescent analysis demonstrated GFP expression in

RI/RIITM+ but not RI/RIITM- TEVG. Strong immunolabeling of TβRII and SMAD2 were demonstrated in RI/RIITM- and was decreased in neo-intima, media, and TEVG of

RI/RIITM+ TEVG. (Figures 12, 13, and 14) von Willebrand Factor (vWF)-labeled endothelial cells, a cell surface marker of endothelium, were present in the neointima, neomedia, and neoadventitia, but vWF only colocalized with GFP in RI/RIITM+ grafts.

(Figure 14) By immunofluorescence, GFP-labeled endothelium was present lining the neointima as well as small vascular channels that infiltrate the neomedia and neoadventitia, and a greater percentage of the graft labeled with GFP than vWF. In areas of the graft where both GFP and vWF label endothelial-origin cells, colocalization of

GFP+ in vWF+ areas and vWF+ in GFP+ areas within the whole grafts was equivalent.

There was significantly more vWF-GFP fluorescent co-localization seen in the intima than the media. (Figure 13)

Direct Immunofluorescence

I R

Figure 12: Direct fluorescent phenotyping of TEVG for mG (membrane green) and mT (membrane red). A-C) TEVG from Tgfbr1f/f; Tgfbr2f/f;Cdh5-creERT+/+;mTom-EGFP mice treated with vehicle lack GFP fluorescence. D-F) Tgfbr1f/f; Tgfbr2f/f;Cdh5-creERT+/+;mTom-EGFP mice treated with tamoxifen demonstrate cre-mediated GFP expression within TEVG. Dashed circle: Aortae demonstrate elastin auto-fluorescence. Dashed square: thin anastomosing branches within TEVG wall in E, F. Arrow: Punctate foci within vehicle treated mice represent autofluorescence of graft polymer.

Figure 13: Expression of TβRI, TβRII, and SMAD2 in TEVG A, 1-4) Immunohistochemical (IHC) evaluation of TβRI in the intima, media, and TEVG demonstrated equivocal immunolabeling of RI/RIITM+ and RI/RIITM- TEVG. B, 1-4) IHC evaluation of TβRII in the intima, media, and TEVG demonstrated strong immunolabeling in RI/RIITM- (control) intima and neovascular channels in media. Macrophages within the RI/RIITM- also strongly label for this marker. While macrophages within the RI/RIITM+ TEVG (treated) demonstrate less intense immunolabeling. Continued.

Continued from page 84. C, 1-4) IHC evaluation of SMAD2 in the intima, media, and TEVG demonstrated strong immunolabeling in RI/RIITM- (control) intima and neovascular channels in media. Macrophages within the RI/RIITM- TEVG (control) also strongly label for this marker. While macrophages within the RI/RIITM+ TEVG (treated) demonstrate less intense immunolabeling.

Figure 14: TEVG Lineage Tracing by of colocalization of GFP and vWF in TEVG. Continued. 86

Continued from page 86. ImageJ Coloc2 processed indirect immunofluorescent images to generate Mander’s coefficients for colocalization of voxels. Data from 11-13 week old mice that had received either tamoxifen or vehicle treatment 21 days prior. A) RI/RIITM+ grafts demonstrate significantly more GFP than vWF within their TEVG (***P<0.0002) suggesting there are GPF+/vWF- cells within the TEVG. B) Equivalent proportions of GFP-vWF and vWF-GFP colocalize in RI/RIITM+ TEVG, demonstrating that there is moderate colocalization of these markers, but that there are vWF+/GFP- and GFP+/vWF- cells present within the TEVG. C) In RI/RIITM+ grafts more vWF-GFP colocalization is identified in the intima than the media. (**P<0.0068). For both RI/RIITM+ and RI/RIITM- TEVG more GFP-vWF and vWF-GFP colocalizes in the intima than the media, but the results are not statistically significant. D) Coloc2 generated data reproducing thresholded voxel colocalization of 488 (vWF) and 647 (GFP) channels of a TEVG. E-K: IF TEVG labeled for GFP (488) and vWF (647) demonstrating colocalization patterns within the intima and medi consistent with the FIJI image analysis that generated Mander’s coefficient data. Bars are standard error of the mean. G: TEVG; L: Lumen. n=6 per group.

3.4.2 Graft Remodeling and Stenosis

Unexpectedly, both RI/RIITM+ and RI/RIITM- grafts demonstrated a low incidence of stenosis with 2% all controls animals and 4.5% all treated animals demonstrating stenosis (9% control males [n=1], 16% control females [n=1], 0% treated males, 9% treated females [n=1]). Historically seeded grafts demonstrate higher patency rates than unseeded grafts (represented here in this study). In comparing patent TEVG between the groups, grafts were widely patent with no significant difference between lumen area

(p=0.45; 5.6 ± 0.99 x105 versus 6.6 ± 0.7 x105 mm2) and graft area (p=0.10; 2.0 ± 0.22 x106 versus 2.5 ± 0.14 x106 mm2, respectively). TEVG graft-luminal area was significantly increased in vehicle versus tamoxifen treated animals (p<0.05; 2.5 ± 0.17 x

106 versus 3.2 ± 0.18 x 106 μm2, respectively), suggesting expansion of the graft with tissue elements. (Figure 15) By polarization, there was a significant difference in the amount of graft polymer in vehicle versus tamoxifen treated animals (p<0.0409; control:

10.0 ± 0.87% versus 7.4 ± 0.50% in treated). (Figure 16)

Masson’s Trichrome analysis of control and treated grafts demonstrated no statistically significant differences in extracellular matrix composition (p<0.49; 17.7 ±

2.4% versus 20.7 ± 3.5%). (Figure 16) In both experimental groups α-SMA highly colocalized with Col I (p<0.0092; treated: 0 0.783 ± 0.0486; control: 0.942 ± 0.02154), with significantly higher colocalization in RI/RIITM-. Between treatment groups, there was significantly more α-SMA-Col I than Col I-α-SMA colocalization (p<0.0001 for 88 both groups). Low colocalization of Col I with α-SMA was characterized (treated: 0.1392

± 0.03053; control: 0.1435 ± 0.02702). (Figure 16) Within RI/RIITM+ TEVG had significantly more α-SMA than GFP-labeled tissue (P<0.0198), with more α-SMA-GFP colocalization than GFP- α-SMA. Immunofluorescent analysis of GFP and -SMA demonstrate that there is very low colocalization of GFP--SMA, but high colocalization of -SMA-GFP which suggests that a small percentage of -SMA cells are of endothelial origin (p<0.0001). (Figure 18)

Figure 15: TEVG stenosis rates do not differ between RI/RIITM- and RI/RIITM+ mice, with both demonstrating high levels of luminal patency. A) Incidence of stenosis. There were no statisically significant differences between treated and control animals B-D) Histomorphometric measurements and representative TEVG regions. B) The entire TEVG graft area was significantly decreased in treated versus untreated animals (p<0.05; 2.5 ± 0.17 x 106 versus 3.2 ± 0.18 x 106 μm2, respectively). C) TEVG graft-luminal area was decreased in treated versus untreated animals (p=0.44; 5.6 ± 0.99 x105 versus 6.6 ± 0.7 x105 μm2, respectively). Continued.

Continued from page 90. D) TEVG luminal area was decreased in treated versus untreated animals (p=0.10; 2.0 ± 0.22 x106 versus 2.5 ± 0.14 x105 μm2, respectively). E-G) Ex vivo TEVG explant photographs demonstrating the regions of the graft: adventitia (E), graft area (F), and intima/lumen (G). E-G represent animals with patent grafts. Bars are standard error of the mean. n=5 samples per group for B-D; n=39 for A.

Figure 16: Extracellular matrix deposition and polymeric graft degradation kinetics at two weeks post-implant. A-C) TEVG were stained with Masson’s Trichrome and demonstrated no significant difference in extracellular matrix deposition at this timepoint increased in treated versus control animals (p<0.49; 17.7 ± 2.4% versus 20.7 ± 3.5%), respectively. Continued.

Continued from page 92. D-F) Indirect immunofluorescent images of collagen I in the TEVG. Significantly more tissue immunolabels with collagen I in RI/RIITM- than RI/RIITM+ TEVG. (P<0.0155) G-I) Indirect immunofluorescent images of -SMA in the TEVG. Immunolabeling of -SMA in RI/RIITM- is increased, but did not achieve statistical significance. J-L) Photomicrographs of hematoxylin and eosin stained slides were polarized in one plane. TEVG graft polymer is significantly increased in treated versus untreated animals (p<0.0409) respectively. Bars are standard error of the mean. G: TEVG; L: lumen. n=5-6 per group.

RI/RIITM+ RI/RIITM-

A B

l l

o o

6 6

488: aSMA 488: aSMA

C Col I D Col I L α-SMA L α-SMA C Coloc Coloc o

V G G c

z e

Figure 17: Colocalization of -SMA and collagen I in TEVG

Continued.

Continued from page 94. RI/RIITM- TEVG demonstrate increased colocalization of synthetic vascular smooth muscle cell markers than RI/RIITM+. ImageJ Coloc2 processed indirect immunofluorescent images to generate Mander’s coefficients for colocalization of voxels. A-B) Colocalization scatter plot for -SMA (488) and collagen I (647) for RI/RIITM- and RI/RIITM+ TEVG. RI/RIITM- demonstrates decreased colocalization and RI/RIITM+ demonstrates increased colocalization. C-D) Coloc2 generated data reproducing thresholded voxel localization of 488 (collagen I) and 647 (-SMA) in the TEVG with colocalization of voxels. E) Collagen I has low colocalization with -SMA (M1 = Mander’s 1) in both RI/RIITM- than RI/RIITM+ TEVG. -SMA has high colocalization with collagen I (M2 = Mander’s 2) in both RI/RIITM- than RI/RIITM+ TEVG. There are statistically significant differences between RI/RIITM+ M2 and RI/RIITM- M2 (P<0.0092), RI/RIITM+ M2 and RI/RIITM+ M1 (P<0.0001), and RI/RIITM- M2 and RI/RIITM- M1 (P<0.0001). Bars are standard error of the mean. G: TEVG; L: lumen. n=6-7 per group.

Figure 18: Vascular smooth muscle cell origins in TEVG. A-B) Indirect immunofluorescent images utilized for Coloc2 analysis. C) There is significantly more -SMA than GFP in RI/RIITM+ TEVG. (P<0.0198) D) GFP rarely colocalizes with -SMA but there is a significant increase in colocalization with -SMA- GFP. (P<0.0001) Bars are standard error of the mean. n=4 per group.

3.4.3 PCR

PCR was undertaken on whole graft lysates representing a heterogenous population of cellular constituents to enable the characterization of differential signaling pathways in TEVG with intact and lost TGF-β signaling pathways within Cdh5 cells.

Data were analyzed with the 2-ΔΔCT method. TGFβR1(P<0.02) transcripts were significantly increased in the RI/RIITM- group when compared to the RI/RIITM+ treatment group. TGFβR2 or TGFβ1 transcripts were not significantly different between groups.

RT-qPCR identified no significant differences in fold change for extracellular matrix markers matrix metalloproteinase 1 (MMP1), alpha-smooth muscle actin (Act2a) or proliferation marker PCNA. However, a statistically significant difference was identified between transcript levels of collagen I (Col1a1) with RI/RIITM- demonstrating a higher fold change compared to RI/RIITM+. (P<0.02). All data were normalized to Rpl13a.

(Figure 19)

Figure 19: Quantitative PCR Analysis with 2-ΔΔCT method.

PCR was performed on selected markers with regard to VSMC phenotype (contractile versus synthetic), extracellular matrix modeling, proliferation, and the TGF-β pathway. A) Grafts from RI/RIITM- (CTL) animals express increased levels of collagen I (*p<0.02). There were no significant differences between RI/RIITM- (CTL) and RI/RIITM+ (Tx) animals with regard to matrix metalloproteinase 2 (MMP2), -smooth muscle actin (Acta2), or proliferating cell nuclear antigen transcripts (PCNA). B) There was a significant difference in transcript levels for TGFb1R (*p<0.02) with a greater fold change for RI/RIITM- (CTL) when compared to RI/RIITM+ (Tx). There were no significant differences in other TGF-β pathway markers, including TGFb2R and TGFb1, between these two groups. Error bars are standard error of the mean. n=5-6 per group.

3.4.4. Immunoblotting analysis

TβRI, TβRII, p-Smad2, and pERK1/2 protein levels were differentially modified between treated TEVG from males and females when compared to control animals.

When separated by sex, treated females had significantly less TβRI (P<0.0116) than control females. Interestingly, RI/RIITM+ males demonstrated higher levels of TβRI than their untreated counterparts. While TβRII protein levels were increased in RI/RIITM+ animals, when separated by sex, RI/RIITM+ females had significantly greater TβRII protein levels than RI/RIITM- females (P<0.0281) and RI/RIITM+ males (P<0.0355).

RI/RIITM+ animals had significantly increased protein levels of p-Smad2 (P<0.0461), with both males and females contributing to this difference. There was also a significant difference in ERK 1/2 (p42/44 MAPK) signaling in treated and untreated animals, independent of sex (P<0.0482). (Figure 20)

A RI/RIITM- RI/RIITM+

TGFBRI

TGFBRII

pSMAD2

pERK 1/2

GAPDH

Figure 20: Western blotting analysis demonstrates modulation of both canonical and non- canonical TGF-β signaling pathway. Representative immunoblots for TGFBRI, TGFBRII, pERK1/2, pSMAD2, and GAPDH. TGFBRI, TGFBRII, and pSMAD2. GAPDH served as a loading control. Continued.

100

Figure 20 Continued: Western blotting analysis demonstrates modulation of both canonical and non-canonical TGF-β signaling pathway. Continued.

101

Continued from pages 100-101. B-C) TGFBRI protein levels are increased in treated animals. When separated by sex, treated females have significantly less TGFBRI than control females. (*p<0.0116) Treated males have higher levels of TGFBRI than their untreated counterparts. D-E) TGFBRII protein levels are increased in treated animals.

When separated by sex, treated females have significantly more TGFBRII than control females and treated males. (*p<0.0281 and #p<0.0355, respectively) Treated females have higher levels of TGFBRII in this model. F-G) Treated animals have significantly increased protein levels of pSMAD2, with both males and females contributing to the significant increase. (^p<0.0461) The two pSMAD2 bands were averaged. This data is consistent with upregulation of the TGF-β signaling pathway, as indicated by increased levels of TGFBRII in treated females. While not statistically significant, TGFBRI levels are increased in treated males. H-I) There is a significant difference in pERK 1/2 signaling in treated and untreated animals, independent of sex. (^p<0.0481). pERK2 (top band) was utilized for analysis. Bar is standard error of the mean. n=12 animals per group; 6 males and 6 females.

102

3.5 Discussion

TEVG stenosis is the most significant clinical limitation observed in our patient cohort. Characterization of the mechanisms of stenosis will be critical for clinical progress in the treatment of congenital heart disease. This study sought to develop a model of murine TEVG stenosis in order to evaluate the role of the TGF-β pathway in endothelial cells. Several research groups have employed lineage-tracing models in murine arterial and venous models to characterize the cellular origin and transformation of vascular endothelium. Duncan et al. and Cooley et al. demonstrated that Tie2-cre and

Cdh5-cre lineage cells contribute to neovessel remodeling in part by upregulating the expression of myogenic protein markers, including α-SMA and calponin. These authors posit that this shift represents endothelial-to-mesenchymal transition (Endo-MT). TGF-β1 is required for this process to occur. Interestingly, our lab’s work has demonstrated increased co-localization of both Tie2-cre and Cdh5-cre derived cells with α-SMA in stenotic grafts. This suggests that endothelial-to-mesenchymal transition may contribute to this pathologic process. In these studies, increased RNA transcript levels of TGF-β and

TGFBRI were demonstrated in occluded and stenosed grafts. Previous studies have demonstrated that vascular stenosis can be attenuated by inhibition with SB431542 (a specific inhibitor of TβRI), TGF-β antibody antagonism, and knockdown of the canonical

TGF-β pathway SMAD2/3 with shRNA. (82, 84) Further support for non-canonical

TGF-β signaling has been supported by studies demonstrating inhibition of endothelial-

103 to-mesenchymal transition of mitral valve endothelial cells with losartan treatment. (112)

Losartan represents an attractive approach for the treatment of TEVG stenosis because it is already approved for use in the pediatric population. Use of Losartan for the modulation of aberrant TGF- signaling has been supported by recent work on a murine model of Loeys-Dietz syndrome in which there is aberrant structural organization of

ECM components, including fibrillan, which modifies the microenvironment availability of TGF-β. (113) However, recent studies evaluating the clinical utility of losartan have not demonstrated that these results translate to the successful treatment of diseases such as Marfan Syndrome, which overlaps pathophysiologically with Loeys-Dietz Syndrome.

(114) In order to assess potential mechanisms of TGF- signaling on TEVG stenosis we sought to evaluate its role in post-natal endothelial cells in vivo in TEVG grafts.

While the studies characterized above suggest the early and crucial role that the

TGF-β pathway plays in endothelial-to-mesenchymal transition and the development of graft stenosis, interpretation is limited by the lack of specificity of the Tie2-cre reporter.

Tie2 has also been identified in myeloid-derived progenitor cells. (35) Additionally, these models nonspecifically and globally target the TGF-β pathway. Here, we demonstrate that loss of TβR1 and TβR2 can be induced post-natally in a time-sensitive, lineage traceable method in endothelial cells. These cells can be lineage traced within TEVG where they constitute both the neointima as well as neovascularize the TEVG neomedia and neoadventitia. GFP labeled cells strongly co-localize with vWF labeled cells, and vice versa. However, higher proportions of GFP-vWF and vWF-GFP are colocalized in

104 the intima than the media. These results suggest that at two weeks post-implant, GFP- positive cells derived from native endothelial cells have migrated into the graft but do not fully express endothelial markers (e.g. vWF). Additionally, there is significant α-

SMA+/GFP+ colocalization within the TEVG, further suggesting that a population of

Cdh5 origin cells have modified their cellular programming towards a mesenchymal phenotype.

It has been proposed that TEVG polymeric degradation decreases biomechanical support and that cellular constituents respond by production of extracellular matrix components to support neovessel integrity. (68) Consistent with our other studies, TEVG from animals with intact endothelial TGF-β signaling produce a collagenous matrix that coincides with graft polymer degradation. This study demonstrates more neotissue formation, comprised of collagen I, and increased levels of -smooth muscle actin by immunofluorescent labeling in control animals when compared to the knock-out model.

Interestingly, a smaller proportion of collagen I colocalizes with -SMA while a higher proportion of α-SMA strongly colocalizes with collagen I. This latter colocalization is significant between treatment groups as well as the same parameters within the treatment group. This suggests that VSMCs have upregulated cellular programming for pro- synthetic phenotype in the TEVG environment. Endothelial and VSMC co-culture systems have demonstrated the important regulatory effect endothelial cells exert on

VSMC. In isolation VSMC cultures are pro-synthetic and proliferative, but when grown adjacent to endothelial cells they demonstrate a phenotypic switch to a contractile cell

105 that inhibits the synthesis of collagen. (115, 116) Additionally, TGF-β1 has been implicated in VSMC phenotypic modulatory switching and is thought to promote a contractile phenotype by downregulating synthetic programming that results in extracellular matrix remodeling. (33) Studies have utilized TGF-β1 neutralizing antibody in attempts to promote pro-synthetic VSMC phenotypes, but increasing titrations were not observed to have an effect, which led the authors to interpret that TGF-β1 neutralization has no regulatory effect on VSMC phenotype. (115) This is in contrast to this study in which intact TGF-β signaling in endothelial cells resulted in increased collagen deposition. Other studies evaluating the role of biomaterials on VSMC extracellular matrix production demonstrate that TGF-β1 present within the scaffold can promote VSMC synthetic phenotype. (117) Ultimately the TEVG represents a significantly more complicated system than in vitro approaches with co-culture systems and VSMC culture on biomaterials. More studies are required to evaluate the critical interplay of cells that constitute the TEVG.

Another explanation is the effect of other cells on VSMC phenotype, such as macrophages or other inflammatory cells within the TEVG influence VSMC by production of TGF-β1 or other inflammatory mediators such as TNF-a, IL-1a, and IL-8.

(34) Previous studies have demonstrated that M1 macrophages predominate stenotic

TEVG. In vivo this phenotype can be shifted from M1 to the M2 (classical) phenotype with treatment of SB431542, a specific TβRI inhibitor. (73) Studies designed to evaluate mechanisms underlying atherosclerotic plaque formation have employed macrophage and

106

VSMC co-culture systems. In vitro macrophages can inhibit VSMC proliferation, promote dedifferentiation, and promote apoptosis. (118-123) Studies with this system also demonstrate that VSMC will increase their production of matrix metalloproteinases

(MMPs) in response to macrophages. (124, 125) These results appear heterogeneous with regard to promotion of VSMC contractile, proliferative, and synthetic phenotypes. While this study did not specifically evaluate the role of macrophage phenotype within the

TEVG increased levels of TβRII and SMAD2, a downstream TGF-β pathway marker, were observed in control TEVG when compared to treated grafts. Indirect evidence of macrophage function was characterized by the significant differences in graft polymer between treated and control TEVG; animals with endothelial TGF-β pathway modulation demonstrated significantly more graft polymer at 2 weeks post-implant. While one mechanism of TEVG degradation is hydrolysis, other research groups have characterized the important role macrophages and multinucleated giant cells play in polymer degradation. (70, 126)

In contrast to the study by Duncan et al, in which transcripts of TGF-β and TβRI were increased in stenotic grafts at days 7 and 14, our untreated animals (which represent a pro-fibrotic graft group) saw no significant differences in the levels of TGF-β and

TβRII. However, in alignment with Duncan et. al’s finings, TβRI transcripts were significantly increased in RI/RIITM- when compared to RI/RIITM+. (84) An important difference between these studies and data sets is reflected in the low incidence of stenotic grafts within our experimental model. Our research group has utilized TEVG in the

107

C57BL/7 mouse model and have robust data sets on the incidence of stenosis for this strain. Therefore, to achieve similar stenosis rates, mice utilized in this study were backcrossed nine generations to C57BL/6 mice. However, stenosis rates were significantly lower for both control and treated animals in this study. Differences in the animal model, tamoxifen dosing, and/or polymeric graft construct may all have contributed to this change. There were no statistically significant differences observed between sexes. For Duncan’s study the incidence of stenosis for Tie2 mice was 66%

(n=6 of 9) and for Cdh5 mice 57% (n= 4 of 7). In this study, stenosis ranged from 0-16% with 0 or 1 stenotic animals per group. One other important factor may be the utilization of 19 G TEVG. Historically, the laboratory has utilized TEVG ranging from 19-21 G in size.

Decreased TGFβRI, TGFβRII, pSMAD2, and pERK2 in RI/RIITM- mice at 2 weeks post-implant may reflect a late remodeling time point in a pro-fibrotic microenvironment. Additional studies should employ earlier timepoints to better characterize the kinetics of the TGF- pathway mediators. Elevated TGFβRI and

TGFβRII in treated animals may suggest aberrant TGF-β signaling in non-endothelial origin cells that comprise the graft such as smooth muscle cells, fibrocytes, myofibroblasts, and macrophages. Upregulation of p-Smad2 is consistent with upregulation of the canonical TGF-β signaling pathway, as indicated by increased levels of TGFβRII in treated females and males. In vitro co-culture studies of valve endothelial cells and valve interstitial cells have demonstrated the important role endothelial cells in

108 modulating EMT and transformation interstitial cells within their regional microenvironment. (127) Increased p-Smad2 protein levels could portend a late pro- fibrotic event for these grafts. Alternatively, they may represent aberrant signaling cascades in which the graft will remain patent. This provides the impetus for further chronic studies to evaluate the role of dysregulated TGF- in endothelial cells on whole- graft remodeling. Increased p-ERK 1/2 in the treated group suggests that there is also upregaulation of non-canonical TGF-β pathway signaling via p-ERK1/2/MAPK pathway.

This latter result is consistent with other studies in which MAPK signaling was increased after postnatal inactivation of TGFβR2 in Myh11-lineage cells, which are of smooth muscle cell origin. (128) Ultimately one of the challenges in interpreting the Western blotting data is contextualizing and reconciling the transgenic construct and differential

TGF-β pathway regulation observed in males and females. This does not appear to be unique to this model. Other studies have characterized differential sex-dependent modulation of the TGF-β pathway in a diversity of experimental models. (129, 130) The mechanisms underpinning this sexual dimorphism have yet to be fully elucidated.

Future studies with this model will further refine the role that endothelial cells have on adjacent cells within their tissue microenvironment. While this study has shown that dysregulating the TGF-β pathway in endothelial cells alters both canonical and non- canonical signaling pathways in whole graft explants, future studies will isolate these individual cell populations to differentially characterize their individual pathway modifications.

109

3.6 Conclusions

The results of this study suggest that specifically targeting the TGF-β pathway in endothelial cells effects whole-graft remodeling at acute time points, including collagen deposition, polymer degradation, and VSMC phenotype. IHC demonstrated loss of TβRII and SMAD2 in endothelial cells and decreased levels in TEVG macrophages. TEVG with intact endothelial TGF-β signaling demonstrate increased TβR1 by PCR and do not negatively regulate the production of collagen by VSMC. Downregulation of TGF-β signaling in endothelial cells results in decreased VSMC collagen production and increased residual polymer. In both treated and control TEVG VSMC demonstrate increased colocalization of α-SMA and collagen I suggesting the predominance of a pro- synthetic phenotype at 2 weeks; loss of endothelial TGF-β signaling resulted in a statistically significant decrease in this synthetic phenotype. Finally, it was demonstrated that a portion of VSMC in TEVG with disrupted TGF-β signaling pathways co-label with

α-SMA and GFP, supporting their origins from Cdh5 (endothelial) cells. Future studies will be required to further characterize late term changes in grafts with TGF-β signaling modulation as well as differential responses to graft modeling that are time course and sex-specific. These results provide the foundation for a better understanding the role of vascular patterning and neotissue formation with differentiated cell lineages and potential implications for therapeutic approaches at all stages of life.

110

Author Roles:

Elizabeth S. Clark: Study design and management, acquisition, analysis, and interpretation of data, manuscript preparation

Brittany Shutes: Data acquisition, manuscript preparation

Kevin Blum: Data acquisition and analysis, interpretation of data, manuscript preparation

Nathan Mahler: Data acquisition

Tai Yi: Primary surgeon, data acquisition

Y. Jiao: Mouse model development

George Tellides: Mouse model development, intellectual input, critical review and final approval of manuscript

Christopher K. Breuer: Data interpretation, study design, critical review and final approval of manuscript

111

Chapter 4: Conclusions and Future Directions

4.1 Conclusions

This work has utilized both large (ovis aries) and small (mus musculus) in vivo animal models to characterize neotissue formation. The ovine model characterized the development of a TEVG stenosis model and novel imaging modalities while the murine study evaluated the role of the TGF-β pathway in endothelial cells with regard to neotissue formation, graft remodeling, and cell constitution. Ovine studies have demonstrated that we are able to experimentally produce models of both TEVG stenosis and patency. The large animal model was evaluated by both angiography and intravascular ultrasound (IVUS) to characterize neotissue formation. In Chapter 2 we demonstrated that not only does IVUS generate comparable measurements to angiography, the current standard of evaluation, but that IVUS is a superior modality to evaluate the TEVG with regard to characterization of 3D geometry and stenosis.

Correlation of IVUS generated images with histomorphometric analysis characterized the presence of clot deposition, neotissue remodeling, polymeric graft degradation, and therapeutic tears. (Figures 4, 5, 8) We demonstrated that histologic assessment is a more sensitive modality to assess certain processes such as neotissue formation, extracellular matrix composition, and polymeric graft degradation. This study also followed the

112 natural time course of TEVG modeling and remodeling with IVUS and histology. These results will provide the foundation to proceed with thesis objective 1: modulation of the development of TEVG neotissue stenosis in a large animal model by targeting vascular smooth muscle cells via the TGF-β signaling pathway and/or the use of interventional angioplasty. Early data shown below in section 4.3.1 demonstrate early TEVG responses to interventional angioplasty techniques. (Figures 23, 24, 25, 26) Additionally, preliminary data on the utilization of microcomputed tomography for the characterization of TEVG is also described in section 4.3.2. (Figures 27, 28) Future studies will employ the use of pharmacomodulation for the treatment of TEVG stenosis, including losartan,

SB431542, and other anti-proliferative drugs that both non-specifically and specifically target the vascular smooth muscle cell layer.

Chapter 3 focused on the second and third aims of this work: identification of the source(s) of vascular smooth muscle cells in TEVG neotissue and the characterization of the role for the TGF-β receptor pathway in the development of TEVG neotissue. To this end, the murine study in Chapter 3 utilized a Tgfbr1f/f; Tgfbr2f/f;Cdh5-creERT+/+;mTom-

EGFP constitutive inducible model to evaluate endothelial-cell specific loss of TβRI and

TβRII on the development of TEVG stenosis. This study demonstrated that endothelial- cell specific recombination could be induced in a lineage-traceable Cdh5 model. Cdh5- lineage endothelial cells expressing mG (EGFP) were characterized in this model with both ex vivo direct and indirect immunofluorescent techniques. GFP-labeled cells contributed to both the neointima and small vascular channels within the neomedia.

113

(Figures 12, 14) GFP-labeled cells demonstrated several immunolabeling patterns, including: GFP+/vWF+, GFP+/vWF-, and a small proportion of GFP+/-SMA+. These results demonstrate that 1) GFP+ lineage Cdh5 cells contribute to the neovascularization and demonstrate markers of mature endothelium (vWF), 2) a portion of GFP+ cells do not express vWF, with a significantly higher proportion in the neomedia than neointima, and 3) while the majority of GFP+ cells do not express -SMA, a small proportion of -

SMA positive cells strongly colocalize with GFP. These results suggest that there are

GFP+ lineage cells that demonstrate variable phenotypic differentiation. The presence of

-SMA+/GFP+ cells suggests that endothelial-to-mesenchymal transition has occurred.

Additionally, the colocalization of -SMA and collagen I within the TEVG were characterized. There is strong, significant colocalization of -SMA-Col I in both control and treated grafts, while Col I positive areas rarely colocalize with -SMA. This suggests that -SMA are exhibiting of myofibroblastic phenotype within the TEVG.

The third aim of this dissertation sought to characterize the role of the TGF-β pathway in the development of TEVG stenosis, however one limitation of this study is that luminal stenosis was rarely observed in the Tgfbr1f/f; Tgfbr2f/f;Cdh5- creERT+/+;mTom-EGFP model. However, when treated with tamoxifen, these mice demonstrated significant differences in neotissue formation with regard to decreased collagen deposition and decreased polymeric degradation when compared to control animals, while supporting TEVG constitution with endothelial, VSMC, and macrophage lineage cells. We hypothesize that differences in luminal stenosis rates may be attributed 114 to use of a transgenic mouse model or optimization of polymeric graft production. The

Tgfbr1f/f; Tgfbr2f/f;Cdh5-creERT+/+;mTom-EGFP had been backcrossed nine generations to the C57BL/6 strain, which is the model our research group has previously utilized to characterize stenosis. Current studies in our research lab are focusing on isolating the variables of graft production between a commercial collaborator and our research group; the results of these studies are beyond the current scope of this project.

Importantly, this study demonstrates that mice with intact endothelial-specific

TGF-β signaling (RI/RIITM-) have significantly more collagen I deposition present at two weeks. This data shows that RI/RIITM- TEVG are more pro-fibrotic than RI/RIITM+

TEVG. This study focused on collagen I because, along with collagen III, it is one of the predominant collagen forms within the tunica media. (37) Collagen I mutations are significant in that they are implicated in both embryonic lethality as well as post-natal vasculopathies and functional deficiencies in mouse models. (131, 132)

Our results also demonstrated that significantly less graft polymer in the TEVG

RI/RIITM+ mice. In another study a C57BL/6 model implanted with sex-mismatched BM-

MNC seeded TEVG demonstrated progressive polymeric degradation at 1-, 2- and 4- weeks post-implant with scant polymer remaining at 4 weeks. These mice had intact

TGF- β signaling pathways. This study associated the loss of polymers with increased levels of collagen IV, GAGs, and elastin, but decreased collagens I and III at late timepoints. They proposed that with the degradation of polymers there was a concomitant

115 decrease in biomechanical support that necessitated the production of extracellular matrix components to support neovessel integrity. (68) Masson’s trichrome staining of TEVG with intact and lost TGF-β signaling demonstrated no significant differences in extracellular matrix components by image analysis of Masson’s Trichrome stained slides.

One mechanism of polymeric degradation of polyesters is by hydrolysis of ester bonds; our graft is composed of polyglycolic acid (PGA) and a copolymer of polycaprolactone and polylactic acid (PCLA). (70) Hypothetically, differences in extracellular matrix composition, cellularity, and phenotype between the experimental groups could also contribute to the availability of polymeric fibers to this mechanism of degradation.

Alternatively, modulation of the TGF-β pathway in endothelial cells may alter recruitment and activation of macrophages in the TEVG that also contribute to polymer degradation. Previous studies by our group have highlighted the role of macrophage activation, differential M1 and M2 phenotype expression, and the formation of stenosis.

(72) Utilization of SB431542, a specific inhibitor of TβRI, in a C57BL/6 model, demonstrated decreased F4/80+ M1 (classically activated) macrophages within TEVG.

(73, 133-135) Other studies in our lab have identified increased levels of M2 macrophages, alternatively activated cells involved in tissue remodeling, are associated with increased TEVG stenosis. (72)

Ultimately, results from Chapter 3 provided data to support that Cdh5-lineage cells contribute to both mature endothelium, a less differentiated endothelial-phenotype cell, and some vascular smooth muscle cells. The presence of GFP+/-SMA+ cells

116 suggests a role for endothelial-to-mesenchymal transition within the TEVG. Investigation of other sources of vascular smooth muscle cells will provide future avenues of research for our laboratory. Results from this work provide further data to support the pro-fibrotic role for the TGF-β receptor pathway in the TEVG model and drive therapeutic approaches for future clinical utilization.

4.2 Future small animal experiments

Future studies with the Tgfbr1f/f; Tgfbr2f/f; Cdh5-creERT+/+;mTom-EGFP model will more finely dissect the individual role of the endothelial cells, vascular smooth muscle cells, and their intermediary phenotypes with regard to their phenotypic programming, modulation of signaling pathways, and interdependent interactions. One approach that can be employed is flow cytometry or laser capture microdissection to isolate the cells of interest. Protocols for the characterization of markers for CD31, α-

SMA, EGFP, and tdTomato (RFP) have been optimized for use with the TEVG model.

Preliminary studies with this model in our lab have demonstrated that while cells can be isolated from murine TEVG and native vessels, the process is time-consuming, and yields relatively low cell numbers (C57BL/6 TEVG: 1.1x106 cells, n=3; Cdh5.mTmG 7.5x105 cells). (Figure 21) These results demonstrate the relative paucity of cells in Cdh5.mTmG

TEVG when compared to C57BL/6 TEVG; alternatively, these differences may reflect differences in the liberation from cells embedded within the TEVG at early timepoints.

Future approaches may require batching of experimental groups or for the experiment to

117 be repeated in triplicate. Early results suggest that further thresholding optimization is required to isolate the GFP signal (a small percentage of cells within C57BL/6 mice were gated as GFP positive and were aberrantly co-localized with CD31). This study demonstrated that CD31 does not approximate GFP and vice versa, and that PE (tomato) and α-SMA should also be evaluated with separate channels. Overall, by flow cytometric analysis, Cdh5.mTmG grafts contain low percentages of cells that label for CD31 when compared to C57BL/6 TEVGs. Both groups demonstrate low numbers of events for α-

SMA. This is discordant with numerous studies undertaken by our lab, and therefore may require further optimization of the gating strategy, or limitations of isolation of this cell population from the TEVG. The majority of the cells of interest for Aim 2 are located within the neo-medial regions of the graft when compared to endothelial cells lining the neointimal surface; this localization likely contributes to differential success in their isolation from the graft. Alternative methodologies for their isolation could also be further evaluated.

118

BL/6 Cdh5.mTmG

1 GFP 1 GFP

2 CD31 2 CD31

3 SMA 3 SMA

4 PE 4 PE

5 Unlabeled cells 5 Unlabeled cells

Figure 21: Total percent events for C57BL/6 and Cdh5.mTmG. C57BL/6 (n=3) yielded 1.1x106 cells, with 43.7% of cells labeled for GFP, CD31, α- SMA, and PE (tomato). Cdh5.mTmG (n=3) yielded 7.5x105 cells, with 34.1% of cells labeled for GFP, CD31, SMA, and PE (tomato).

40.00% 35.00% 30.00% 25.00% 20.00% 15.00% Cdh5 10.00% BL/6 5.00% 0.00% BL/6 %GFP from Cdh5 %CD31 CD31 from %SMA from from GFP %tomato total tomato from total from SMA from total from total

Figure 22: Colocalization of markers for C57BL/6 and Cdh5.mTmG. Channels were gated for immunolabeled-GFP, PE (tomato), CD31, and α-SMA.

119

4.3 Ovine Models

Our recent work with the large animal model has demonstrated a reproducible model of TEVG stenosis. We demonstrated that in addition to angiography that TEVG can be assessed safely in vivo with IVUS. IVUS was demonstrated to correlate well with one-plane angiographic measurements of luminal diameter, and was superior to the current gold-standard modality of angiography with regard to assessment of luminal area.

While IVUS will allow for chronic evaluation of tissue remodeling and polymeric graft degradation there are some limitations to its utilization. Furthermore, we demonstrated that IVUS is superior to angiography for the characterization of TEVG stenosis. Our study also demonstrated that while IVUS may be suggestive of changes such as thrombus deposition, and therapeutic tears, histology can better characterize the diverse spectrum of tissue response, including mineralization. While we have gained more information regarding the in vivo assessment of TEVG stenosis in our human cohort and large animal study, we have yet to robustly characterize treatment modalities that are employed clinically (such as angioplasty) in our preclinical model. Furthermore, this model will enable us to evaluate the tissue responses to angioplasty as well as potential pharmacologic approaches for the treatment of this vascular pathology.

120

4.3.1 Assessment of Efficacy and Safety of Balloon Angioplasty for the Treatment

of TEVG stenosis

Stenosis can be treated clinically with balloon angioplasty (BA), cutting balloons, or stent placement. The advantage of these techniques is that they are minimally invasive and do not require long anesthetic induction times. However, the use of stents in growing children necessitates surgery later in life. Fortunately, the majority of our patients with stenosis have been successfully treated with balloon angioplasty for stenosis. (14)

However, it is currently unknown how the TEVG will respond to balloon angioplasty, the type of balloon that should be utilized (low versus high compliance), and long-term effects of endovascular interventions on neo-vessel development. We have previously utilized 25-35 kg juvenile lambs (Ovis aries) which received unseeded TEVGs (Gunze

Medical, Ltd.) implanted into thoracic caudal vena cava interposition grafts (n=4) to promote TEVG stenosis. Sheep recovered and survived for up to 73 days post-surgery.

Animals underwent serial angiography and BA at 0, 30-40, and 60-70 days to evaluate graft patency and remodeling. BA was performed using a low-pressure compliant balloon catheter (Tyshak II) in cases of stenosis. Graft tissue was harvested at each time point from one animal for histologic analysis.

Stenotic TEVGs were treated with BA on post-operative days 0, 30, and 60.

Acute stenosis was identified in unseeded TEVG, which was highly compliant to BA, but did not significantly alter lumen patency or mean pressure gradients (Figures 22-25).

121

Acute stenosis is attributed to deposition of fibrin (identified by histology in Animal 1) on the luminal surface of the TEVG. For 30- and 60-day TEVG median luminal patency was 52.9% [range: 22.1-55.7%]. BA increased lumen diameters to a median of 61.4%

[27.9-64.3%]. Pre-intervention patency rates approach previously defined graft stenosis

(50% narrowing in the diameter of the graft, but not critical stenosis (75% loss of graft diameter) Pressure gradients pre-intervention were a mean of 14 mmHg [8-17 mmHg] and BA decreased mean pressure gradients by a median of 4.0 mmHg [3.5-9.0 mmHg].

This represents a significant physiologic decrease in trans-gradient graft pressure that approaches the upper limits of trans-gradient graft pressures in the human clinical population. Histology of unseeded TEVG demonstrated neotissue formation, including neo-intima and neo-media as early as 36 days post-implantation.

Important findings from this model highlight some important differences between this large animal model and our patient cohort. First, pre-implantation these animals demonstrate higher venous pressures than children. Secondly, post-implant these animals not only demonstrate higher trans-gradient pressures than children, but these higher pressures appear to be clinically insignificant for these animals. These results contrast from the clinical picture in which intervention is considered when trans-gradient pressures between 2-5 mmHg are recorded and associated clinical abnormalities are identified. These model differences could be used to inform modifications of the surgical approaches used in pre-clinical studies.

122

Ultimately, our initial work has shown that an ovine model using cell-free TEVGs recapitulates clinical TEVG stenosis and that that TEVG geometry and remodeling can be captured by serial angiographic, pressure measurements, and IVUS. However, clinically, a cell-free TEVG is not approved so we continue to utilize seeded TEVG to evaluate neotissue remodeling. TEVG stenosis can be successfully treated with BA during acute time points, but should be further evaluated to gain more information on graft and neotissue pressure tolerance, and long-term responses to this treatment. These results and future studies will improve current clinical practice and the safety of TEVG angioplasty. Future directions will evaluate the response of TEVG seeded with bone marrow derived mononuclear cells to balloon angioplasty and characterize neotissue formation.

123

Pre-Dilation Balloon Dilation Post-Dilation

A B C *

0 *

a D E F

e * *

t *

a *

o P G H I *

0 6

Figure 23: Angiography: TEVG stenosis responds to interventional angioplasty. TEVG can be imaged with angiography to characterize angioplasty. Unseeded TEVG from Animal 3 demonstrates stenosis on post-surgical days 0 (A, B, C), 30 (D, E, F), and 60 (G, H, I) that responds to BA. Red(*): stenosis. White(*): waist. Yellow(*): luminal filling defect. n=1.

124

Pre-Dilation Post-Dilation

A D

s o

P B C E F

Figure 24: IVUS: TEVG stenosis responds to interventional angioplasty.

Intravascular ultrasound images of unseeded TEVG from Animal 4 on post-surgical day 60. TEVG pre- (A), and post- (D) balloon dilation. Representative intravascular ultrasound images of the pre- and post-dilated TEVG at the proximal (B, E), middle (C, F) portions of the graft. n=1.

125

h 80

i g

n 60

y A2 30 b

A3 30 )

A4 30 %

( 40 A3 60

y A4 60

t a

P 20

m u

L 0 Pre-dilation Post-dilation

Figure 25: Changes in luminal Patency in response to balloon angioplasty. Angiographic measurements at 30- and 60- days post-implant for animals 2, 3, and 4 with unseeded TEVG. Median luminal patency was 52.9% [range: 22.1-55.7%] before BA. Therapeutic intervention increased lumen diameters to a median of 61.4% [27.9-64.3%].

126

)

g 20

(

n 15

i d

a A2 30 r A3 30

G A4 30 10

e A3 60

r A4 60

s e

r 5

a e

M 0 Pre-dilation Post-dilation

Figure 26: Mean pressure gradient response to balloon angioplasty. Pressure measurements at 30- and 60- days post-implant for animals 2, 3, and 4. Pressure gradients pre-intervention were a mean of 14 mmHg [8-17 mmHg]. BA decreased mean pressure gradients by a median of 4.0 mmHg [3.5-9.0 mmHg].

127

4.3.2 Modalities to Evaluate Early Characterization of Microcalcification in

Ovine TEVG

Current draft guidance from the FDA, “General Considerations for Animal

Studies for Medical Devices” of ex vivo imaging modality assessment includes radiography, micro-computed tomography (microCT), and histomorphometric assessment of calcification. We have performed initial studies to evaluate these modalities for a chronic ovine (ovis aries) TEVG model to characterize the presence or absence of microcalcification.

In vivo, 2D and 3D rotational angiography were utilized while ex vivo methodology included microCT and histology. After formalin fixation for 24-48 hours, grafts underwent 3D scanning with a GE eXplore Locus Micro CT. Each specimen was positioned horizontally on a vertical gantry bed. Scanning was undertaken with the

Microview protocol of 45 m resolution, 360-degree scan at 80 kVp and 450 A producing a 720 slice view. TEVG were scanned with and without contrast (oxilan).

Post-processing was completed with Vitrea® Software with scoring for calcium gated voxels. Utilization of contrast did not improve scanning resolution and in some runs served to obscure tissue planes. Artifacting was observed occasionally that was attributed to the containment device for the TEVG, the presence of metal clips and circumferential markers composed of omnipaque material that had been applied during the operative procedures. However, the presence of the metal clips did not preclude the evaluation of

128 the TEVG. MicroCT data sets were successfully utilized for the early assessment of microcalcification by gross examination and histomorphometric assessment. (Figures 26,

27)

Results from these preliminary studies demonstrate that microcalcification can be identified with microCT but not in vivo imaging modalities such as angiography. Future studies could employ traditional ex vivo radiography of TEVG explants. Data generated from imaging modalities can optimally direct tissue-sampling approaches for histomorphometric assessment. While microCT may provide better 3D spatial resolution, the higher cost and limited availability of this specialty equipment, software requirements, and user expertise may limit the pre-clinical utilization of microCT.

129

a.1 a.2

a.3 a.4

b.1 b.2

b.3 b.4

Figure 27: Vitrea reconstructions of microCT data with calcium gating. Representative Vitrea reconstructions of microCT data of longitudinal and cross sections of TEVG with gating for calcium a) An animal with multifocal aggregates within the TEVG wall highlighted by the gating protocol seen on longitudinal (1, 2, 3) and cross section (4). b) An animal with a focal aggregate within the wall of the TEVG seen on longitudinal (1, 2, 3) and cross section (4).

130

Figure 28: MicroCT used for histologic assessment of microcalcification. Correlatvive images of TEVG microcalcification as characterized by a) microCT, b) gross analysis, and c) histology with hematoxylin and eosin staining. Histology did not confirm that the material highlighted at the proximal and distal aspects of the TEVG by the microCT program was mineral.

131

4.3.3 Future Approaches for the Treatment of TEVG Stenosis

Future approaches to evaluate preclinical models of TEVG stenosis may employ pharmacologic agents that either mechanistically target the TGF-β pathway, such as

SB431542, or utilization of a drug that is commonly used in the pediatric population, such as losartan. SB431542 is a specific antagonist of the TGF-β type I receptor. (73,

133-135) Losartan, an angiotensin II receptor antagonist, is also of interest to researchers for the treatment of vascular stenosis. (136) Its proposed mechanism of action is via non- canonical targeting of the TGF-β pathway. (113)

Previous studies by our group with mice treated with SB431542 have demonstrated that administration of this drug is associated with a significant decrease in

TEVG stenosis, decreased EMT, and fewer VSMC present in neotissue. (84) The first experiment will need to assess the delivery mechanisms, tissue distribution, and pharmacokinetics of SB42152 within an ovine model. Next, the efficacy of this drug to decrease the incidence of TEVG stenosis should be assessed; the two study groups employed could include seeded TEVG and seeded TEVG + SB431542 administration; alternatively, unseeded TEVG could be employed as the control group.

Our murine studies (unpublished data) have demonstrated that administration of losartan decreases the incidence and severity of TEVG stenosis. There has been limited research undertaken on this drug in the ovine model. Losartan’s function as an

132 angiotensin-II receptor antagonist has been evaluated in adult sheep, including its effects on renal perfusion. (137-139) In addition to its potential therapeutic benefit for the treatment of TEVG stenosis losartan could also provide potential renal protective effects in the ovine model. Due to prolonged clamp times necessitated by the surgical procedure, acute renal failure with elevations in blood urea nitrogen and creatinine is occasionally observed due to ischemia-reperfusion syndrome. This drug is an ideal candidate for clinical translation because there is FDA-approved pediatric dosing for the drug. Another large animal experiment could utilize losartan (0.7 mg/kg/day PO up to 50-100 mg – representing a clinically relevant dose used in the pediatric population) treatment to decrease the incidence of TEVG stenosis in seeded TEVG; study groups will include seeded TEVG and seeded TEVG + losartan administration. Losartan would be compounded and provided orally to the animals, and as above, unseeded TEVG could be employed as the control group. Ultimately we anticipate that our Phase II graft for clinical use will result in a decreased incidence of clinical stenosis due to improved knowledge regarding tissue formation, tissue response to angioplasty, as well as approaches for modulation of tissue formation such as pharmacomodulation.

4.3.4 TEVG Utilization of Drug Eluting Compounds

Mechanisms of and treatment approaches for vascular restenosis have been a focus of cardiovascular medical device assessment. Three components of this process defined by Costa and Simon (2005) include neointimal hyperplasia, elaboration of

133 extracellular matrix components, and remodeling of the vessel wall. (140) Infiltration of host inflammatory cells and proliferation of resident vascular smooth muscle cells are some of the earliest temporal events in vascular repair. Modulation of these potentially pathologic processes with pharmacologic agents to treat vascular restenosis have both been evaluated clinically. Rapamycins (everolimus, sirolimus, zotarolimus) and taxanes

(paclitaxel) have been evaluated for their clinical utility in modulating the development of in-stent restenosis. (51, 52, 141, 142) Paclitaxel, a taxane, inhibits smooth muscle cell proliferation by arresting the cell between the G2 and M phase and prevents migration by inhibition of microtubule depolarization. (143, 144) Rapamycins function as inhibitors of cell proliferation and promote a contractile phenotype. (145) However, clinical utility of these drugs for the treatment of pediatric TEVG stenosis may be limited by their safety and efficacy. Widespread adoption of some of these drugs has been limited by the limited healing and cellular coverage seen in adults.

Additionally, there are several important differences between drug-coated cardiovascular devices and our pre-clinical and clinical TEVG. First, FDA-approved drug-coated devices include stents and balloons that are utilized to treat arterial disease.

In contrast, the pre-clinical TEVG is implanted in venous structures: the caudal abdominal vena cava of the mouse, and the caudal thoracic vena cava in the sheep. In children with hypoplastic left heart syndrome the inferior graft is exposed to the venous system, while the superior portion of the graft is exposed to an arterial environment.

There are significant differences in vascular structures such as the presence of interior

134 and external elastic lamina, variation in thickness of the tunica media, and amount and organization of extracellular matrix components including elastin and collagens.

Additionally, cellular phenotype and signaling pathways are adapted to high and low oxygen saturation. These vascular structures must also respond to significantly different biomechanical forces including vascular pulsatility and distensibility. An optimal drug- embedded TEVG will allow for optimal smooth muscle cell proliferation, production of extracellular matrix components such as elastin and collagen, and contain the appropriate balance of contratile and synthetic apparati.

4.4 Final Thoughts

The primary barrier to further clinical utilization of the TEVG is understanding the mechanisms and pathogenesis of TEVG stenosis. We have demonstrated that both ovine and murine models can recapitulate components of TEVG neotissue formation in order to study the underlying mechanisms, treatment approaches, and modalities to assess this pathology. Our pre-clinical large animal model is already yielding information that can be used clinically to interpret imaging data from children in the Phase I clinical trial.

We have a better understanding of graft remodeling and interpretation of these changes within the graft with IVUS and angiography. Data generated from the murine transgenic models will allow us to further interrogate the role of the TGF- pathway in not only endothelial cells, but also its effects on endothelial-to-mesenchymal transition and the

135 vascular smooth muscle compartment of the tunica media. Ultimately these findings will provide the foundation for the development of the next generation of TEVG.

136

Bibliography

1. V. Garg, Insights into the genetic basis of congenital heart disease. Cell Mol Life Sci 63, 1141-1148 (2006). 2. M. E. Mitchell, T. L. Sander, D. B. Klinkner, A. Tomita-Mitchell, The molecular basis of congenital heart disease. Semin Thorac Cardiovasc Surg 19, 228-237 (2007). 3. M. J. Sutherland, S. M. Ware, Disorders of left-right asymmetry: heterotaxy and situs inversus. Am J Med Genet C Semin Med Genet 151C, 307-317 (2009). 4. A. F. Corno, "Functionally" univentricular hearts: impact of pre-natal diagnosis. Front Pediatr 3, 15 (2015). 5. C. Frescura, G. Thiene, The new concept of univentricular heart. Front Pediatr 2, 62 (2014). 6. D. van der Linde et al., Birth prevalence of congenital heart disease worldwide: a systematic review and meta-analysis. J Am Coll Cardiol 58, 2241-2247 (2011). 7. J. I. Hoffman, S. Kaplan, The incidence of congenital heart disease. J Am Coll Cardiol 39, 1890-1900 (2002). 8. M. Abu-Harb, E. Hey, C. Wren, Death in infancy from unrecognised congenital heart disease. Arch Dis Child 71, 3-7 (1994). 9. C. E. Greenleaf, J. M. Urencio, J. D. Salazar, A. Dodge-Khatami, Hypoplastic left heart syndrome: current perspectives. Transl Pediatr 5, 142-147 (2016). 10. G. Stefanelli et al., Early and intermediate-term (10-year) results of surgery for univentricular atrioventricular connection ("single ventricle"). Am J Cardiol 54, 811-821 (1984). 11. M. Gewillig, The Fontan circulation. Heart 91, 839-846 (2005). 12. M. Galantowicz et al., Hybrid approach for hypoplastic left heart syndrome: intermediate results after the learning curve. Ann Thorac Surg 85, 2063-2070; discussion 2070-2061 (2008). 13. T. Shinoka, C. Breuer, Tissue-engineered blood vessels in pediatric cardiac surgery. Yale J Biol Med 81, 161-166 (2008). 14. E. W. Dean, B. Udelsman, C. K. Breuer, Current advances in the translation of vascular tissue engineering to the treatment of pediatric congenital heart disease. Yale J Biol Med 85, 229-238 (2012). 15. A. Amodeo et al., Extracardiac Fontan operation for complex cardiac anomalies: seven years' experience. J Thorac Cardiovasc Surg 114, 1020-1030; discussion 1030-1021 (1997). 16. S. Pashneh-Tala, S. MacNeil, F. Claeyssens, The Tissue-Engineered Vascular Graft-Past, Present, and Future. Tissue Eng Part B Rev, (2015). 17. T. J. van Brakel et al., High incidence of Dacron conduit stenosis for extracardiac Fontan procedure. J Thorac Cardiovasc Surg 147, 1568-1572 (2014). 137

18. J. T. Patterson et al., Tissue-engineered vascular grafts for use in the treatment of congenital heart disease: from the bench to the clinic and back again. Regen Med 7, 409-419 (2012). 19. A. H. Huang et al., Design and Use of a Novel Bioreactor for Regeneration of Biaxially Stretched Tissue-Engineered Vessels. Tissue Eng Part C Methods 21, 841-851 (2015). 20. S. Dimitrievska et al., Click-coated, heparinized, decellularized vascular grafts. Acta Biomater 13, 177-187 (2015). 21. B. Lilly, We have contact: endothelial cell-smooth muscle cell interactions. Physiology (Bethesda) 29, 234-241 (2014). 22. R. H. Adams, K. Alitalo, Molecular regulation of angiogenesis and lymphangiogenesis. Nat Rev Mol Cell Biol 8, 464-478 (2007). 23. R. C. Sainson et al., Cell-autonomous notch signaling regulates endothelial cell branching and proliferation during vascular tubulogenesis. FASEB J 19, 1027- 1029 (2005). 24. M. Hellstrom et al., Dll4 signalling through Notch1 regulates formation of tip cells during angiogenesis. Nature 445, 776-780 (2007). 25. D. Vestweber, VE-cadherin: the major endothelial adhesion molecule controlling cellular junctions and blood vessel formation. Arterioscler Thromb Vasc Biol 28, 223-232 (2008). 26. U. R. Michaelis, Mechanisms of endothelial cell migration. Cell Mol Life Sci 71, 4131-4148 (2014). 27. J. R. Privratsky, P. J. Newman, PECAM-1: regulator of endothelial junctional integrity. Cell Tissue Res 355, 607-619 (2014). 28. M. P. Brennan et al., Tissue-engineered vascular grafts demonstrate evidence of growth and development when implanted in a juvenile animal model. Ann Surg 248, 370-377 (2008). 29. N. Hibino et al., Comparison of human bone marrow mononuclear cell isolation methods for creating tissue-engineered vascular grafts: novel filter system versus traditional density centrifugation method. Tissue Eng Part C Methods 17, 993-998 (2011). 30. J. Torres-Vazquez, M. Kamei, B. M. Weinstein, Molecular distinction between arteries and veins. Cell Tissue Res 314, 43-59 (2003). 31. R. H. Adams et al., Roles of ephrinB ligands and EphB receptors in cardiovascular development: demarcation of arterial/venous domains, vascular morphogenesis, and sprouting angiogenesis. Genes Dev 13, 295-306 (1999). 32. G. Kapakos, A. Bouallegue, G. B. Daou, A. K. Srivastava, Modulatory Role of Nitric Oxide/cGMP System in Endothelin-1-Induced Signaling Responses in Vascular Smooth Muscle Cells. Curr Cardiol Rev 6, 247-254 (2010). 33. S. S. Rensen, P. A. Doevendans, G. J. van Eys, Regulation and characteristics of vascular smooth muscle cell phenotypic diversity. Neth Heart J 15, 100-108 (2007).

138

34. J. A. Beamish, P. He, K. Kottke-Marchant, R. E. Marchant, Molecular regulation of contractile smooth muscle cell phenotype: implications for vascular tissue engineering. Tissue Eng Part B Rev 16, 467-491 (2010). 35. G. Wang, L. Jacquet, E. Karamariti, Q. Xu, Origin and differentiation of vascular smooth muscle cells. J Physiol 593, 3013-3030 (2015). 36. M. G. Frid, E. P. Moiseeva, K. R. Stenmark, Multiple phenotypically distinct smooth muscle cell populations exist in the adult and developing bovine pulmonary arterial media in vivo. Circ Res 75, 669-681 (1994). 37. J. A. Eble, S. Niland, The extracellular matrix of blood vessels. Curr Pharm Des 15, 1385-1400 (2009). 38. N. Hibino et al., Late-term results of tissue-engineered vascular grafts in humans. J Thorac Cardiovasc Surg 139, 431-436, 436 e431-432 (2010). 39. M. W. Majesky, X. R. Dong, V. Hoglund, W. M. Mahoney, Jr., G. Daum, The adventitia: a dynamic interface containing resident progenitor cells. Arterioscler Thromb Vasc Biol 31, 1530-1539 (2011). 40. G. Tellides, J. S. Pober, Inflammatory and immune responses in the arterial media. Circ Res 116, 312-322 (2015). 41. N. Takeda et al., Pathophysiology and Management of Cardiovascular Manifestations in Marfan and Loeys-Dietz Syndromes. Int Heart J 57, 271-277 (2016). 42. R. P. Lanza, R. S. Langer, J. Vacanti, Principles of tissue engineering. (Academic Press, San Diego, CA, ed. 2nd, 2000), pp. xli, 995 p. 43. T. Shin'oka, Y. Imai, Y. Ikada, Transplantation of a tissue-engineered pulmonary artery. N Engl J Med 344, 532-533 (2001). 44. T. Shin'oka et al., Midterm clinical result of tissue-engineered vascular autografts seeded with autologous bone marrow cells. J Thorac Cardiovasc Surg 129, 1330- 1338 (2005). 45. C. Lee et al., Midterm follow-up of the status of Gore-Tex graft after extracardiac conduit Fontan procedure. Eur J Cardiothorac Surg 31, 1008-1012 (2007). 46. E. P. Moran, D. K. Agrawal, Increased expression of inhibitor of apoptosis proteins in atherosclerotic plaques of symptomatic patients with carotid stenosis. Exp Mol Pathol 83, 11-16 (2007). 47. J. P. Carrozza, Jr. et al., Angiographic and clinical outcome of intracoronary stenting: immediate and long-term results from a large single-center experience. J Am Coll Cardiol 20, 328-337 (1992). 48. R. F. Neville, A. N. Sidawy, Myointimal hyperplasia: basic science and clinical considerations. Semin Vasc Surg 11, 142-148 (1998). 49. J. V. Betala, E. M. Langan Iii, M. LaBerge, Drug-coated percutaneous balloon catheters. Crit Rev Biomed Eng 42, 193-212 (2014). 50. G. Katz, B. Harchandani, B. Shah, Drug-eluting stents: the past, present, and future. Curr Atheroscler Rep 17, 485 (2015). 51. Y. Cheng, M. B. Leon, J. F. Granada, An update on the clinical use of drug- coated balloons in percutaneous coronary interventions. Expert Opin Drug Deliv 13, 859-872 (2016). 139

52. A. Kalra et al., New-Generation Coronary Stents: Current Data and Future Directions. Curr Atheroscler Rep 19, 14 (2017). 53. T. L. Mirensky et al., Tissue-engineered arterial grafts: long-term results after implantation in a small animal model. J Pediatr Surg 44, 1127-1132; discussion 1132-1123 (2009). 54. J. D. Roh et al., Tissue-engineered vascular grafts transform into mature blood vessels via an inflammation-mediated process of vascular remodeling. Proc Natl Acad Sci U S A 107, 4669-4674 (2010). 55. M. R. Stacy et al., Targeted imaging of matrix metalloproteinase activity in the evaluation of remodeling tissue-engineered vascular grafts implanted in a growing lamb model. J Thorac Cardiovasc Surg 148, 2227-2233 (2014). 56. M. Watanabe et al., Tissue-engineered vascular autograft: inferior vena cava replacement in a dog model. Tissue Eng 7, 429-439 (2001). 57. P. A. Iaizzo, Heart valves : from design to clinical implantation. (Springer, New York, 2013). 58. F. J. Schoen, R. J. Levy, Calcification of tissue heart valve substitutes: progress toward understanding and prevention. Ann Thorac Surg 79, 1072-1080 (2005). 59. T. Shinoka et al., Creation of viable pulmonary artery autografts through tissue engineering. J Thorac Cardiovasc Surg 115, 536-545; discussion 545-536 (1998). 60. J. D. Roh et al., Construction of an autologous tissue-engineered venous conduit from bone marrow-derived vascular cells: optimization of cell harvest and seeding techniques. J Pediatr Surg 42, 198-202 (2007). 61. I. Cummings et al., Tissue-engineered vascular graft remodeling in a growing lamb model: expression of matrix metalloproteinases. Eur J Cardiothorac Surg 41, 167-172 (2012). 62. H. Su et al., Noninvasive targeted imaging of matrix metalloproteinase activation in a murine model of postinfarction remodeling. Circulation 112, 3157-3167 (2005). 63. R. J. Levy et al., Cardiovascular implant calcification: a survey and update. Biomaterials 12, 707-714 (1991). 64. H. Kurobe et al., Comparison of a closed system to a standard open technique for preparing tissue-engineered vascular grafts. Tissue Eng Part C Methods 21, 88-93 (2015). 65. A. Goyal et al., Development of a model system for preliminary evaluation of tissue-engineered vascular conduits. J Pediatr Surg 41, 787-791 (2006). 66. J. D. Roh et al., Small-diameter biodegradable scaffolds for functional vascular tissue engineering in the mouse model. Biomaterials 29, 1454-1463 (2008). 67. N. Hibino et al., Tissue-engineered vascular grafts form neovessels that arise from regeneration of the adjacent blood vessel. FASEB J 25, 2731-2739 (2011). 68. Y. Naito et al., Characterization of the natural history of extracellular matrix production in tissue-engineered vascular grafts during neovessel formation. Cells Tissues Organs 195, 60-72 (2012). 69. J. M. Anderson, A. Rodriguez, D. T. Chang, Foreign body reaction to biomaterials. Semin Immunol 20, 86-100 (2008). 140

70. K. A. Rocco, M. W. Maxfield, C. A. Best, E. W. Dean, C. K. Breuer, In vivo applications of electrospun tissue-engineered vascular grafts: a review. Tissue Eng Part B Rev 20, 628-640 (2014). 71. V. K. Pepper et al., Intravascular Ultrasound Characterization of a Tissue- Engineered Vascular Graft in an Ovine Model. J Cardiovasc Transl Res, (2017). 72. N. Hibino et al., A critical role for macrophages in neovessel formation and the development of stenosis in tissue-engineered vascular grafts. FASEB J 25, 4253- 4263 (2011). 73. Y. U. Lee et al., TGF-beta receptor 1 inhibition prevents stenosis of tissue- engineered vascular grafts by reducing host mononuclear phagocyte activation. FASEB J 30, 2627-2636 (2016). 74. F. O. Martinez, S. Gordon, The M1 and M2 paradigm of macrophage activation: time for reassessment. F1000Prime Rep 6, 13 (2014). 75. K. L. Spiller et al., The role of macrophage phenotype in vascularization of tissue engineering scaffolds. Biomaterials 35, 4477-4488 (2014). 76. M. E. Ogle, C. E. Segar, S. Sridhar, E. A. Botchwey, Monocytes and macrophages in tissue repair: Implications for immunoregenerative biomaterial design. Exp Biol Med (Maywood) 241, 1084-1097 (2016). 77. C. E. Bergmann et al., Arteriogenesis depends on circulating monocytes and macrophage accumulation and is severely depressed in op/op mice. J Leukoc Biol 80, 59-65 (2006). 78. H. D. Danenberg et al., Macrophage depletion by clodronate-containing liposomes reduces neointimal formation after balloon injury in rats and rabbits. Circulation 106, 599-605 (2002). 79. J. Massague, TGF-beta signal transduction. Annu Rev Biochem 67, 753-791 (1998). 80. X. Guo, S. Y. Chen, Transforming growth factor-beta and smooth muscle differentiation. World J Biol Chem 3, 41-52 (2012). 81. T. M. Holm et al., Noncanonical TGFbeta signaling contributes to aortic aneurysm progression in Marfan syndrome mice. Science 332, 358-361 (2011). 82. B. C. Cooley et al., TGF-beta signaling mediates endothelial-to-mesenchymal transition (EndMT) during vein graft remodeling. Sci Transl Med 6, 227ra234 (2014). 83. D. J. Grainger, J. C. Metcalfe, A. A. Grace, D. E. Mosedale, Transforming growth factor-beta dynamically regulates vascular smooth muscle differentiation in vivo. J Cell Sci 111 ( Pt 19), 2977-2988 (1998). 84. D. R. Duncan et al., TGFbetaR1 inhibition blocks the formation of stenosis in tissue-engineered vascular grafts. J Am Coll Cardiol 65, 512-514 (2015). 85. S. Piera-Velazquez, Z. Li, S. A. Jimenez, Role of endothelial-mesenchymal transition (EndoMT) in the pathogenesis of fibrotic disorders. Am J Pathol 179, 1074-1080 (2011). 86. L. Xiao, A. C. Dudley, Fine-tuning vascular fate during endothelial-mesenchymal transition. J Pathol 241, 25-35 (2017).

141

87. T. Kokudo et al., Snail is required for TGFbeta-induced endothelial-mesenchymal transition of embryonic stem cell-derived endothelial cells. J Cell Sci 121, 3317- 3324 (2008). 88. R. M. Simeone et al., Pediatric inpatient hospital resource use for congenital heart defects. Birth defects research. Part A, Clinical and molecular teratology 100, 934-943 (2014). 89. C. A. Russo, A. Elixhauser, in Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. (Rockville (MD), 2006). 90. T. Shin'oka et al., Midterm clinical result of tissue-engineered vascular autografts seeded with autologous bone marrow cells. J Thorac Cardiov Sur 129, 1330-1338 (2005). 91. C. A. C. Pedra, C. Fleishman, S. F. Pedra, J. P. Cheatham, New imaging modalities in the catheterization laboratory. Current Opinion in Cardiology 26, 86-93 (2011). 92. E. B. Diethrich, K. Irshad, D. B. Reid, Virtual Histology and Color Flow Intravascular Ultrasound in Peripheral Interventions. Seminars in vascular surgery 19, 155-162 (2006). 93. J. T. Lee, T. D. Fang, R. A. White, Applications of intravascular ultrasound in the treatment of peripheral occlusive disease. Seminars in vascular surgery 19, 139- 144 (2006). 94. B. J. Pearce, W. D. Jordan, Jr., Using IVUS during EVAR and TEVAR: improving patient outcomes. Seminars in vascular surgery 22, 172-180 (2009). 95. G. A. Wallace et al., Intravascular ultrasound is a critical tool for accurate endograft sizing in the management of blunt thoracic aortic injury. Journal of vascular surgery 61, 630-635 (2015). 96. C. J. Marrocco et al., Intravascular ultrasound. Seminars in vascular surgery 25, 144-152 (2012). 97. S. J. Park, S. J. Kang, R. Virmani, M. Nakano, Y. Ueda, In-stent neoatherosclerosis: a final common pathway of late stent failure. Journal of the American College of Cardiology 59, 2051-2057 (2012). 98. B. D. Gogas, V. Farooq, P. W. Serruys, H. M. Garcia-Garcia, Assessment of coronary atherosclerosis by IVUS and IVUS-based imaging modalities: progression and regression studies, tissue composition and beyond. The international journal of cardiovascular imaging 27, 225-237 (2011). 99. H. M. Garcia-Garcia, B. D. Gogas, P. W. Serruys, N. Bruining, IVUS-based imaging modalities for tissue characterization: similarities and differences. The international journal of cardiovascular imaging 27, 215-224 (2011). 100. E. Madssen et al., Coronary atheroma regression and plaque characteristics assessed by grayscale and radiofrequency intravascular ultrasound after aerobic exercise. The American journal of cardiology 114, 1504-1511 (2014). 101. H. Fatakdawala et al., Fluorescence Lifetime Imaging Combined with Conventional Intravascular Ultrasound for Enhanced Assessment of Atherosclerotic Plaques: an Ex Vivo Study in Human Coronary Arteries. Journal of cardiovascular translational research 8, 253-263 (2015). 142

102. M. Ishii et al., Evaluation of pulmonary artery histopathologic findings in congenital heart disease: an in vitro study using intravascular ultrasound imaging. Journal of the American College of Cardiology 26, 272-276 (1995). 103. J. Xu et al., Intravascular ultrasound assessment of regional aortic wall stiffness, distensibility, and compliance in patients with coarctation of the aorta. American heart journal 134, 93-98 (1997). 104. R. W. Day, L. Y. Tani, Pulmonary intravascular ultrasound in infants and children with congenital heart disease. Catheterization and cardiovascular diagnosis 41, 395-398 (1997). 105. R. M. Berger, A. H. Cromme-Dijkhuis, W. C. Hop, M. N. Kruit, J. Hess, Pulmonary arterial wall distensibility assessed by intravascular ultrasound in children with congenital heart disease: an indicator for pulmonary vascular disease? Chest 122, 549-557 (2002). 106. S. R. Veeram Reddy et al., A novel design biodegradable stent for use in congenital heart disease: mid-term results in rabbit descending aorta. Catheter Cardiovasc Interv 85, 629-639 (2015). 107. R. Khosravi et al., Biomechanical diversity despite mechanobiological stability in tissue engineered vascular grafts two years post-implantation. Tissue engineering. Part A 21, 1529-1538 (2015). 108. N. S. van Ditzhuijzen et al., Invasive coronary imaging in animal models of atherosclerosis. Netherlands heart journal : monthly journal of the Netherlands Society of Cardiology and the Netherlands Heart Foundation 19, 442-446 (2011). 109. S. Tara et al., Vessel bioengineering. Circ J 78, 12-19 (2014). 110. P. Ten Dijke, A. D. Egorova, M. J. Goumans, R. E. Poelmann, B. P. Hierck, TGF- beta signaling in endothelial-to-mesenchymal transition: the role of shear stress and primary cilia. Sci Signal 5, pt2 (2012). 111. K. W. Dunn, M. M. Kamocka, J. H. McDonald, A practical guide to evaluating colocalization in biological microscopy. Am J Physiol Cell Physiol 300, C723-742 (2011). 112. J. Wylie-Sears, R. A. Levine, J. Bischoff, Losartan inhibits endothelial-to- mesenchymal transformation in mitral valve endothelial cells by blocking transforming growth factor-beta-induced phosphorylation of ERK. Biochem Biophys Res Commun 446, 870-875 (2014). 113. E. M. Gallo et al., Angiotensin II-dependent TGF-beta signaling contributes to Loeys-Dietz syndrome vascular pathogenesis. J Clin Invest 124, 448-460 (2014). 114. R. V. Lacro et al., Atenolol versus losartan in children and young adults with Marfan's syndrome. N Engl J Med 371, 2061-2071 (2014). 115. R. J. Powell, J. Hydowski, O. Frank, J. Bhargava, B. E. Sumpio, Endothelial cell effect on smooth muscle cell collagen synthesis. J Surg Res 69, 113-118 (1997). 116. M. F. Fillinger, L. N. Sampson, J. L. Cronenwett, R. J. Powell, R. J. Wagner, Coculture of endothelial cells and smooth muscle cells in bilayer and conditioned media models. J Surg Res 67, 169-178 (1997).

143

117. B. K. Mann, R. H. Schmedlen, J. L. West, Tethered-TGF-beta increases extracellular matrix production of vascular smooth muscle cells. Biomaterials 22, 439-444 (2001). 118. D. Proudfoot, C. Fitzsimmons, J. Torzewski, D. E. Bowyer, Inhibition of human arterial smooth muscle cell growth by human monocyte/macrophages: a co- culture study. Atherosclerosis 145, 157-165 (1999). 119. B. G. Halloran, J. J. Grange, B. J. So, B. T. Baxter, Macrophage products inhibit human aortic smooth muscle cell proliferation and alter 1 alpha (I) procollagen expression. Ann Vasc Surg 11, 80-84 (1997). 120. A. Shioi et al., Induction of bone-type alkaline phosphatase in human vascular smooth muscle cells: roles of tumor necrosis factor-alpha and oncostatin M derived from macrophages. Circ Res 91, 9-16 (2002). 121. X. Li, M. Y. Speer, H. Yang, J. Bergen, C. M. Giachelli, Vitamin D receptor activators induce an anticalcific paracrine program in macrophages: requirement of osteopontin. Arterioscler Thromb Vasc Biol 30, 321-326 (2010). 122. J. J. Boyle, P. L. Weissberg, M. R. Bennett, Human macrophage-induced vascular smooth muscle cell apoptosis requires NO enhancement of Fas/Fas-L interactions. Arterioscler Thromb Vasc Biol 22, 1624-1630 (2002). 123. J. J. Boyle, P. L. Weissberg, M. R. Bennett, Tumor necrosis factor-alpha promotes macrophage-induced vascular smooth muscle cell apoptosis by direct and autocrine mechanisms. Arterioscler Thromb Vasc Biol 23, 1553-1558 (2003). 124. E. Lee et al., Human vascular smooth muscle cell-monocyte interactions and metalloproteinase secretion in culture. Arterioscler Thromb Vasc Biol 15, 2284- 2289 (1995). 125. Y. Zhu, Y. Hojo, U. Ikeda, M. Takahashi, K. Shimada, Interaction between monocytes and vascular smooth muscle cells enhances matrix metalloproteinase-1 production. J Cardiovasc Pharmacol 36, 152-161 (2000). 126. Z. Xia et al., Macrophage-mediated biodegradation of poly(DL-lactide-co- glycolide) in vitro. J Biomed Mater Res A 79, 582-590 (2006). 127. K. Shapero, J. Wylie-Sears, R. A. Levine, J. E. Mayer, Jr., J. Bischoff, Reciprocal interactions between mitral valve endothelial and interstitial cells reduce endothelial-to-mesenchymal transition and myofibroblastic activation. J Mol Cell Cardiol 80, 175-185 (2015). 128. W. Li et al., Tgfbr2 disruption in postnatal smooth muscle impairs aortic wall homeostasis. J Clin Invest 124, 755-767 (2014). 129. D. Liu et al., Skeletal muscle gene expression in response to resistance exercise: sex specific regulation. BMC Genomics 11, 659 (2010). 130. G. Kararigas et al., Sex-dependent regulation of fibrosis and inflammation in human left ventricular remodelling under pressure overload. Eur J Heart Fail 16, 1160-1167 (2014). 131. J. Lohler, R. Timpl, R. Jaenisch, Embryonic lethal mutation in mouse collagen I gene causes rupture of blood vessels and is associated with erythropoietic and mesenchymal cell death. Cell 38, 597-607 (1984).

144

132. O. Rahkonen et al., Mice with a deletion in the first intron of the Col1a1 gene develop age-dependent aortic dissection and rupture. Circ Res 94, 83-90 (2004). 133. G. J. Inman et al., SB-431542 is a potent and specific inhibitor of transforming growth factor-beta superfamily type I activin receptor-like kinase (ALK) receptors ALK4, ALK5, and ALK7. Mol Pharmacol 62, 65-74 (2002). 134. S. K. Halder, R. D. Beauchamp, P. K. Datta, A specific inhibitor of TGF-beta receptor kinase, SB-431542, as a potent antitumor agent for human cancers. Neoplasia 7, 509-521 (2005). 135. N. J. Laping et al., Inhibition of transforming growth factor (TGF)-beta1-induced extracellular matrix with a novel inhibitor of the TGF-beta type I receptor kinase activity: SB-431542. Mol Pharmacol 62, 58-64 (2002). 136. J. Zhu et al., Losartan ameliorates "upstream" pulmonary vein vasculopathy in a piglet model of pulmonary vein stenosis. J Thorac Cardiovasc Surg 148, 2550- 2557 (2014). 137. J. M. Bonnet, R. Boivin, Comparative study of the cardiovascular effects of losartan in normal and in water- and salt-depleted sheep. J Vet Med Sci 62, 925- 931 (2000). 138. R. R. Singh, Y. R. Lankadeva, K. M. Denton, K. M. Moritz, Improvement in renal hemodynamics following combined angiotensin II infusion and AT1R blockade in aged female sheep following fetal unilateral nephrectomy. PLoS One 8, e68036 (2013). 139. J. Ullman, S. Eriksson, M. Rundgren, Losartan increases renal blood flow during isoflurane anesthesia in sheep. Acta Anaesthesiol Scand 45, 1168-1175 (2001). 140. M. A. Costa, D. I. Simon, Molecular basis of restenosis and drug-eluting stents. Circulation 111, 2257-2273 (2005). 141. J. E. Sousa et al., Sustained suppression of neointimal proliferation by sirolimus- eluting stents: one-year angiographic and intravascular ultrasound follow-up. Circulation 104, 2007-2011 (2001). 142. M. C. Morice et al., A randomized comparison of a sirolimus-eluting stent with a standard stent for coronary revascularization. N Engl J Med 346, 1773-1780 (2002). 143. X. Zhou, J. Li, D. F. Kucik, The microtubule cytoskeleton participates in control of beta2 integrin avidity. J Biol Chem 276, 44762-44769 (2001). 144. E. K. Rowinsky, R. C. Donehower, Paclitaxel (taxol). N Engl J Med 332, 1004- 1014 (1995). 145. D. Rosner, N. McCarthy, M. Bennett, Rapamycin inhibits human in stent restenosis vascular smooth muscle cells independently of pRB phosphorylation and p53. Cardiovasc Res 66, 601-610 (2005).

145