Regulation of the Human Vascular Smooth Muscle Cell Transcriptome by Extracellular Matrix Stiffness

University of Pennsylvania ScholarlyCommons

Publicly Accessible Penn Dissertations

2017

Regulation Of The Human Vascular Smooth Muscle Cell Transcriptome By Extracellular Matrix Stiffness

Christopher Yu University of Pennsylvania, [email protected]

Follow this and additional works at: https://repository.upenn.edu/edissertations

Part of the Cell Biology Commons, and the Genetics Commons

Recommended Citation Yu, Christopher, "Regulation Of The Human Vascular Smooth Muscle Cell Transcriptome By Extracellular Matrix Stiffness" (2017). Publicly Accessible Penn Dissertations. 2710. https://repository.upenn.edu/edissertations/2710

This paper is posted at ScholarlyCommons. https://repository.upenn.edu/edissertations/2710 For more information, please contact [email protected]. Regulation Of The Human Vascular Smooth Muscle Cell Transcriptome By Extracellular Matrix Stiffness

Abstract Arterial stiffness is a risk factor for several cardiometabolic diseases and is caused by pathological remodeling of the vascular extracellular matrix (ECM). Vascular smooth muscle cells (VSMCs) respond to ECM stiffness by proliferating, migrating, and further remodeling the vascular ECM, thus contributing to vascular disease like atherosclerosis and hypertension. VSMCs along the vasculature are highly diverse as they arise from different embryologic origins, reside in ECMs of diverse compositions, and are exposed to various mechanical forces. This dissertation aims to understand how ECM stiffness regulates the transcriptional response of VSMCs from different origins, namely aortic (Ao) and coronary (Co) VSMCs. We conducted deep sequencing of RNA from Ao and Co VSMCs grown on engineered polyacrylamide hydrogel surfaces tuned to physiologic and pathologic stiffness. Using several bioinformatic approaches, we compared the transcriptional landscapes in Ao and Co VSMCs by looking at whole-gene level expression, splicing, and conservation properties, with a focus on long non-coding RNAs (lncRNAs), as they compose a significant portion of the unexplored VSMC transcriptome. We found evidence suggesting that the overall transcriptional response to stiffness may be conserved across species and cell types, and that stiffness significantly dictates VSMC transcriptional identity over contributions from embryologic origins. However, we also discovered instances of origin-specific stiffness esponsesr in stiffness-mediated lncRNA expression and stiffness-mediated splicing. We identified a highly correlated network of adjacent stiffness-sensitive lncRNAs-protein coding gene pairs, which led us to experimentally interrogate the lncRNA PACER as a regulator of stiffness-mediate expression. We surprisingly found that PACER’s previously established regulatory pathway is absent in VSMCs. Using enrichment methods, we identified TBX5 and show that it is a stiffness-sensitive transcription factor specific ot Co VSMCs. We also demonstrated a new role for MALAT1 as a lncRNA regulator of stiffness-dependent VSMC proliferation and migration. Thus, this dissertation reveals many novel characteristics of the VSMC stiffness-regulated transcriptome that may have clinical utility in understanding and managing the pathogenesis of arterial stiffening.

Degree Type Dissertation

Degree Name Doctor of Philosophy (PhD)

Graduate Group Bioengineering

First Advisor Daniel J. Rader

Second Advisor Richard K. Assoian

Keywords Extracellular matrix stiffness, Long non-coding RNAs, Transcriptome, Vascular smooth muscle cells, Vascular stiffness Subject Categories Cell Biology | Genetics

This dissertation is available at ScholarlyCommons: https://repository.upenn.edu/edissertations/2710 REGULATION OF THE HUMAN VASCULAR SMOOTH MUSCLE CELL TRANSCRIPTOME BY

EXTRACELLULAR MATRIX STIFFNESS

Christopher Yu

A DISSERTATION

Bioengineering

Presented to the Faculties of the University of Pennsylvania

Partial Fulfillment of the Requirements for the

Degree of Doctor of Philosophy

2017

Supervisor of Dissertation Co-Supervisor of Dissertation

______

Daniel J. Rader, M.D. Richard K. Assoian, Ph.D.

Seymour Gray Professor of Molecular Medicine Professor of Pharmacology

Graduate Group Chairperson

______

Ravi Radhakrishnan, Ph.D.

Professor of Bioengineering; Professor of Chemical and Biomolecular Engineering

Dissertation Committee:

(Chair): Peter F. Davies, Ph.D., Sc.D., Professor of Pathology & Laboratory Medicine; Professor of Bioengineering

Rebecca G. Wells, M.D., Professor of Medicine

Christopher S. Chen, M.D., Ph.D., Professor of Biomedical Engineering (Boston University)

REGULATION OF THE HUMAN VASCULAR SMOOTH MUSCLE CELL TRANSCRIPTOME BY EXTRACELLULAR MATRIX STIFFNESS

2017

Christopher Ku Yu

This work is licensed under the Creative Commons Attribution- NonCommercial-ShareAlike 3.0 License

To view a copy of this license, visit https://creativecommons.org/licenses/by-nc-sa/3.0/us/ Dedication

I dedicate this thesis to my parents (Robin H. Yu, M.D., and Estrelita Ku, M.D.), who prioritized and spared no expense on the education and life opportunities for me and my brothers. This thesis and my pursuit of a combined MD/PhD degree would never have been possible without their life-long support.

I also dedicate this thesis to my incredible wife, Sue Lee, who put her own Ph.D studies on hold to take care of her sick mom. She has the biggest heart for family. Without her constant support, I could not have completed my dissertation.

En Taro Adun

iii

ACKNOWLEDGMENTS

Early in my Ph.D, I assumed that my prior research experiences in high school and college would have made my Ph.D experience smooth and straightforward. But it became quickly apparent that I had a lot to learn. There’s a lot of people, programs, and organizations to thank along the way for teaching me and providing me scientific, financial, and social support.

First, I have to thank all the mentors and teachers I had before starting my Ph.D. who gave me a foundation for critical thinking and the scientific method.

Thanks to everyone in the Rader Lab, Assoian Lab, and Chen Lab. I was lucky to be immersed in three different scientific environments utilizing microfabrication/tissue engineering, cell and molecular biology, computation/next-generation sequencing and human genetics.

From the Rader Lab: Deepti Abbey, Guha Arunkumar, Rob Bauer, Xin Bi, Linda Carmichael, Stephanie DerOhannessian, Susannah Elwyn, Nicholas Hand, Sumeet Khetarpal, Hyein Kim, Wen Lin, Nicholas Lyssenko, Dawn Marchadier, Minal Mehta, Sylvia Nuremberg, Amirth Rodrigues, Kevin Trindade, Cathy Warford. From the Assoian Lab: Nate Bade, Yong Ho Bae, Paola Castagnino, Beth Hawthorne, Shulin Liu, Keeley Mui, Ziba Razinia, Ryan Von Kleeck, and Tina Xu. From the Chen Lab: Colin Choi, Peter Galie, and Michael Yang.

I’d also like to thank others from other labs and cores. The Reily Lab (when they were at Penn), especially, Rachel Ballantyne and Chenyi Xue. The CDB microscopy core and its director Andrea Stout. The ITMAT biomechanics core, especially Jeff Byfield.

Thanks my committee members Peter Davies and Rebecca Wells for their never-ending support through both the ups and downs of a PhD Study. Keeping me on track during thesis committee meetings and being very accommodating to me.

The NIH funding sources: HL119346 and T32HL007954-17

I would also like to acknowledge all my friends for all their support and for the shared fun times through all the years. All the people I started med school, the group collectively known as zergling rush, especially Mike Sheng, Ray Zhang, James Hui, and Nabil Thalji. My long-time friends from college, Yang Li and Sophie Su. To the friends I met along the Ph.D path, especially Rosa Maria Alvarez, Keeley Mui, Kristy Ou, and Sean Yeldell. All the members of Penn Lions, for keeping me physically active, culturally active, and feel young.

Thanks to my family. My mom, my dad, and my brothers, Robinson and Michael. My dad got sick and passed away toward the end of my Ph.D, but our family stuck together to support each other through that difficult time. As this thesis is dedicated to my mom and dad, I literally could not have completed this dissertation without them.

Lastly, thank you so so much to my loving wife, Sue Lee. She has given me unlimited support in every aspect of my life, not only during my Ph.D studies, but almost every day in the last 10 years of knowing each other. She will discuss with me ad nauseam any scientific and non-scientific issue that I cannot believe or understand. She has helped iv me with all my grammar issues, even in writing of this dissertation, and practiced presentations with me. She has always been there to keep me sane through all the strings of failed experiments and turbulent times in life. I love her very much and owe her my deepest gratitude.

ABSTRACT

REGULATION OF THE HUMAN VASCULAR SMOOTH MUSCLE CELL TRANSCRIPTOME BY

EXTRACELLULAR MATRIX STIFFNESS

Christopher Yu

Daniel J. Rader, M.D.

Richard K. Assoian, Ph.D.

Arterial stiffness is a risk factor for several cardiometabolic diseases and is caused by pathological remodeling of the vascular extracellular matrix (ECM). Vascular smooth muscle cells (VSMCs) respond to ECM stiffness by proliferating, migrating, and further remodeling the vascular ECM, thus contributing to vascular disease like atherosclerosis and hypertension. VSMCs along the vasculature are highly diverse as they arise from different embryologic origins, reside in ECMs of diverse compositions, and are exposed to various mechanical forces. This dissertation aims to understand how ECM stiffness regulates the transcriptional response of VSMCs from different origins, namely aortic

(Ao) and coronary (Co) VSMCs. We conducted deep sequencing of RNA from Ao and

Co VSMCs grown on engineered polyacrylamide hydrogel surfaces tuned to physiologic and pathologic stiffness. Using several bioinformatic approaches, we compared the transcriptional landscapes in Ao and Co VSMCs by looking at whole-gene level expression, splicing, and conservation properties, with a focus on long non-coding RNAs

(lncRNAs), as they compose a significant portion of the unexplored VSMC transcriptome. We found evidence suggesting that the overall transcriptional response to stiffness may be conserved across species and cell types, and that stiffness significantly dictates VSMC transcriptional identity over contributions from embryologic origins.

However, we also discovered instances of origin-specific stiffness responses in stiffness- mediated lncRNA expression and stiffness-mediated splicing. We identified a highly

vi correlated network of adjacent stiffness-sensitive lncRNAs-protein coding gene pairs, which led us to experimentally interrogate the lncRNA PACER as a regulator of stiffness- mediate expression. We surprisingly found that PACER’s previously established regulatory pathway is absent in VSMCs. Using enrichment methods, we identified TBX5 and show that it is a stiffness-sensitive transcription factor specific to Co VSMCs. We also demonstrated a new role for MALAT1 as a lncRNA regulator of stiffness-dependent

VSMC proliferation and migration. Thus, this dissertation reveals many novel characteristics of the VSMC stiffness-regulated transcriptome that may have clinical utility in understanding and managing the pathogenesis of arterial stiffening.

vii

TABLE OF CONTENTS

ACKNOWLEDGMENTS ...... IV

ABSTRACT ...... VI

TABLE OF CONTENTS ...... VIII

LIST OF TABLES ...... XI

LIST OF ILLUSTRATIONS ...... XII

LIST OF APPENDIX CONTENTS...... XIII

CHAPTER 1: INTRODUCTION ...... 1

Clinical Importance of Vascular Stiffening ...... 1

Clinical Risk Factors for Arterial Stiffening...... 1

Mechanism of Vascular Stiffening ...... 2 Extracellular Matrix Contributions to Stiffness ...... 2 Cellular Contributions to Stiffness ...... 3

Mechanotransduction: How Cells Sense Stiffness ...... 4

Mechanotransduction: Transcriptional Response ...... 7

Specific Mechanotransduction Mechanisms in Vascular Smooth Muscle Cells ...... 8 ECM-stiffness-dependent VSMC functions...... 9 Cell-ECM interaction during VSMC stiffness-sensing ...... 9 Intracellular signaling during VSMC stiffness-sensing ...... 10

Vascular Heterogeneity ...... 11 Vascular bed mechanical property diversity ...... 11 Vascular smooth muscle heterogeneity ...... 11

Dissertation Goals ...... 13

CHAPTER 2: CHARACTERIZATION OF THE VASCULAR SMOOTH MUSCLE CELL STIFFNESS-SENSITIVE TRANSCRIPTOME ...... 14

Introduction ...... 14

Transcriptome landscapes and determinants of diversity ...... 14

Long non-coding RNAs ...... 16 viii

Alternative Isoforms and Splicing ...... 16

Bioinformatic pipelines for characterizing transcriptome landscapes ...... 18 Machine-learning: Principal component analysis and hierarchical clustering ...... 18 Selecting Candidates and Predicting Function ...... 19

Summary ...... 21

METHODS ...... 23

RESULTS ...... 29

The Landscape of the Stiffness-sensitive Transcriptome of Co and Ao VSMCs (Gene Expression Level) ...... 29

The Landscape of the Stiffness-sensitive Splicing in Co and Ao VSMCs (Isoform Level) .. 41

DISCUSSION ...... 51

Implications of Splicing Observations ...... 53

CHAPTER 3: REGULATORS OF THE STIFFNESS SENSITIVE TRANSCRIPTOME .. 57

Introduction ...... 57

Mechanosensitive Transcription Factors ...... 57 Cell/Context Specific Transcription Factors ...... 59

The Role of lncRNAs as Transcriptional Regulators ...... 59 Long non-coding RNAs: Mechanisms of action ...... 60 The long non-coding RNA, PACER ...... 62

METHODS ...... 63

RESULTS ...... 67

Unbiased Motif Analysis Identifies TBX5 as a novel stiffness-sensitive transcription factor ...... 67

Transcriptome-wide correlation analysis reveals coordinated expression of stiffness- sensitive lncRNA and stiffness-sensitive protein coding gene pairs ...... 71

Interrogation of the lncRNA, PACER as a potential regulator of nearby stiffness-sensitive protein coding genes ...... 78

DISCUSSION ...... 84

TBX5 ...... 84

Transcriptome-wide Correlation ...... 86

PACER ...... 86 ix

CHAPTER 4: MALAT1, A LNCRNA REGULATOR OF STIFFNESS-SENSITIVE VSMC FUNCTIONS ...... 89

Introduction ...... 89

Discovery of MALAT1 ...... 89

Function of MALAT1: Proliferation, Migration, and Splicing...... 90

MALAT1 as a competitive endogenous RNA ...... 91

MALAT1 knockout Mice...... 91

Role for MALAT1 during pathology ...... 92

METHODS ...... 93

RESULTS ...... 96

Identification of MALAT1 as a regulator of stiffness-dependent phenotypes ...... 96

DISCUSSION ...... 103

CHAPTER 5: DISCUSSION ...... 105

Basic Mechanotransduction ...... 105

Clinical/Translational Implications ...... 108

APPENDIX ...... 112

BIBLIOGRAPHY ...... 120

LIST OF TABLES

Table 1. Enriched annotations of differentially spliced genes ...... 45

Table 2. Differentially spliced genes with SH3 domains ...... 45

Table 3. Expression quantitative trait loci (eQTL) for cardiovascular disease-related GWAS variants associated with stiffness-sensitive genes...... 49

Table 4. VSMC-relevant transcription factor motifs enriched in stiffness-sensitive genes using MEME/TOMTOM Suite ...... 69

Table 5. Top correlated stiffness-sensitive lncRNA-protein coding gene pairs...... 77

Table 6. Identification of lncRNA regulators of stiffness-dependent cell responses (Predicted Categorical Pathways)...... 98

Table 7. Identification of lncRNA regulators of stiffness-dependent cell responses (Predicted Functions)...... 99

LIST OF ILLUSTRATIONS

Figure 1. Experimental and computational flow chart for interrogating the VSMC stiffness-mediated transcriptome...... 31

Figure 2. Expression Landscape of Aortic and Coronary VSMCs ...... 32

Figure 3. Expression Landscape of Whole Aortic and Coronary Artery Tissue from GTEx Database ...... 33

Figure 4. Heatmap of the VSMC Stiffness-sensitive Gene Expression ...... 36

Figure 5. Cell-specific expression of stiffness-sensitive genes ...... 37

Figure 6. Sequence Conservation and Tissue Expression Analysis of the VSMC Stiffness-Sensitive Genes...... 38

Figure 7. Predicted Stiffness-dependent cellular functions ...... 40

Figure 8. Stiffness-sensitive splicing in Co and Ao VSMCs ...... 42

Figure 9. Schematic diagrams of differentially spliced genes containing SH3 domains ..46

Figure 10. Schematic diagrams of stiffness-sensitive spliced genes and relationship to cardiovascular trait GWAS-identified variants...... 50

Figure 11. Stiffness-dependent localization of TBX5 and YAP...... 70

Figure 12. Increased lncRNA-gene pair correlation with pair proximity...... 73

Figure 13. Increased lncRNA-gene pair correlation with pair proximity and stiffness- sensitivity...... 75

Figure 14. Coordinate stiffness-sensitive expression of PACER, PTGS2, and PLA2G4A ...... 81

Figure 15. Regulation of PTGS2 and PLA2G4A by PACER ...... 82

Figure 16. The role of NFkB in stiffness-sensitive PACER-PTGS2 expression...... 83

Figure 17. MALAT1, a lncRNA regulator of stiffness-dependent cell functions in primary VSMCs...... 100

Figure 18. MALAT1, a lncRNA regulator of stiffness-dependent cell functions in hTERT- immortalized Co VSMCs...... 101

Figure 19. MALAT1 knockdown inhibits EdU incorporation, cyclin expression and wound closure in Co VSMCs on tissue-culture plastic (TCPS)...... 102

xii

LIST OF APPENDIX CONTENTS

Appendix 1. Table of Cardiovascular Disease Traits used in GWAS analyses in Chapter 2 ...... 112

Appendix 2. Ingenuity Pathway Analysis Enriched Upstream “Transcriptional Regulators” ...... 113

Appendix 3. KEGG enriched pathways...... 115

Appendix 4. REACTOME Enriched Pathways (STIFF) ...... 116

Appendix 5. REACTOME Enriched Pathways (SOFT) ...... 117

Appendix 6. Correlated Stiffness-sensitive lncRNA-Protein Coding Gene Pairs ...... 119

xiii

CHAPTER 1: INTRODUCTION

Clinical Importance of Vascular Stiffening

Arterial stiffening is a long-recognized mechanical property important in cardiovascular health. Arterial stiffness has been under investigation since the early 19th century, when

Thomas Young, who was a physician and physicist, quantified heart and arterial stiffness and derived Young’s (elastic) modulus, a now generalizable material property1. Arterial stiffening is a natural result of aging2–4 and is a risk factor for and consequence of multiple cardiometabolic diseases including hypertension, atherosclerosis, coronary artery disease, stroke, diabetes, renal disease, and all-cause cardiovascular mortality2,5–

15. Clinically, arterial stiffness is measured indirectly using surrogate parameters such as pulse pressure and pulse wave velocity (the gold standard), or through pressure/diameter measurements using magnetic resonance imaging14.

Clinical Risk Factors for Arterial Stiffening

Although arterial stiffness is measurable and highly predictive of disease, there are no therapies targeting stiffness directly and management focuses on addressing modifiable risk factors such as lifestyle modifications16, smoking17, dyslipidemia18, diabetes and hypertension19,20. Age and genetic factors contribute to arterial stiffness. Arterial stiffening progresses with aging, with inflections during puberty and menopause, and further accelerates into older age. At advanced ages, males also have a higher rate of progression of arterial stiffening than females2–4,21. Analysis of several hundred families in the Framingham Heart Study cohort found that arterial stiffness is a heritable trait14 and subsequent genome-wide association studies have linked several arterial stiffness- associated traits such as pulse pressure and pulse wave velocity to single nucleotide polymorphisms (SNP)22–24. Several monogenic disorders also result in altered vascular 1 biomechanics and premature stiffening such as mutations in FBN1 (fibrillin-1, leading to

Marfan’s syndrome)25,26, mutations in ELN (elastin, leading to Willam’s syndrome and supravalvular aortic stenosis)27, and LMNA (lamin A, leading to progeria)28.

Mechanism of Vascular Stiffening

Tissue stiffening is a hallmark of aging and disease in many organ systems not only arterial vasculature. In general, tissue stiffening results from cellular remodeling of the extracellular matrix through increased deposition and crosslinking of matrix proteins, but also through changes in matrix components. In the vasculature, both changes in matrix stiffness, intrinsic cellular stiffness and cell contractility/tone contribute to arterial stiffening.

Extracellular Matrix Contributions to Stiffness

During development, the vasculature is formed from a primitive endothelial plexus that rearranges to form lumens and branches and subsequently recruits vascular smooth muscle cells (VSMCs). These VSMCs interact with the developing vasculature by depositing concentric layers of extracellular matrix (ECM). VSMCs secrete a myriad collection of ECM proteins (e.g. collagens, elastin, and proteoglycans), crosslinkers (e.g. lysyl oxidases), and matrix-metalloproteinase. The basis for vascular wall stiffness is formed from the balance and organization of these ECM proteins (e.g. fibrillar collagens provide load bearing tensile properties, while elastin provides elastic properties). VSMCs only deposit and crosslink elastin during the postnatal period, thus the elastic properties of the vascular wall are, in a sense, afforded only early in development. Through aging and disease, elastin progressively degrades without replenishment and fibrillar collagens are extensively deposited, leading to imbalances in the ECM and progressive stiffening29–31. 2

Biomechanical studies suggest that the collagen/elastin ratio of the vessel wall significantly determines arterial stiffness. At physiologic pressures/distension, the vessel wall resides in an elastic regime dominated by elastin. As degradation occurs, the same pressures and distension shift into a stiffer regime dominated by collagens30,31. The loss of elastin occurs naturally during aging and is accelerated in several pathologic states.

This occurs through the upregulation of matrix-metalloproteinases (e.g. MMP-2, -9, -12) which directly degrade elastin resulting in fragmentation, disorganization, and increased arterial stiffness32,33. Human mutations in the elastin and fibrillin-1, a scaffold for elastin, result in severely reduced and disorganized elastic lamina resulting in premature arterial stiffening. In osteogenesis imperfecta, mutations in collagen I are associated with loss in tensile properties leading to aortic root dilatation and aortic valve insufficiency, coronary artery dissection, and aneurysms34–37.

Cellular Contributions to Stiffness

In addition to the material properties of the ECM, properties of vascular cells also contribute significantly to overall arterial stiffening. During adulthood, VSMCs can exist in a continuum of phenotypes between a “differentiated/contractile” state and a “dedifferentiated/synthetic” state. Under normal physiologic conditions, VSMCs are thought to be in a predominantly contractile state where they are non-proliferative, non-secretory, and primarily function to regulate vascular tone and blood flow. During disease, VSMCs can de-differentiate into the synthetic state, gaining proliferative and migratory capacity as well as secreting and crosslinking new ECM, such as fibrillar collagens and fibronectin, but not elastin, promoting vascular wall hypertrophy, luminal narrowing, and stiffening38. 3

Endothelial cells lining the arterial lumen serve many roles as a protective barrier and regulator of vascular homeostasis. Endothelial cells release potent vasoactive factors such as nitric oxide and endothelin-1 that control VSMC tone and also prevent VSMC dedifferentiation into the synthetic phenotype39,40. During disease, endothelial dysfunction occurs resulting in loss of the protective barrier function, leading to inflammation and dysregulated VSMCs. Pathologic increases in vascular tone, VSMC contractility, and thus intrinsic VSMC stiffness, contribute to total vascular wall stiffness41,42. Endothelial and VSMC dysfunction can also result from chronic exposure to elevated blood glucose, inflammation and renin-angiotensin signaling underlying the association of arterial stiffness with diabetes, hypertension, and renal dysfunction19.

Lastly, arterial stiffening is both a cause and consequence of arterial stiffness. Synthetic substrates mimicking pathologic stiffness increased endothelial cell inflammatory markers and increased VSMC expression of matrix proteins and crosslinkers43–45.

Several in vivo models are consistent with this positive feedback mechanism of arterial stiffening begetting more stiffening. In an obese mouse model, diet-induced arterial stiffening measured by pulse wave velocity preceded elevated systolic blood pressures and matrix expression9. This phenomenon was similarly seen in rat models of pulmonary hypertension and liver fibrosis, where matrix stiffness, measured by atomic force microscopy, preceded further matrix remodeling and fibrosis45,46.

Mechanotransduction: How Cells Sense Stiffness

The vasculature is constantly exposed to mechanical stimuli that are critical to vascular morphogenesis and function: cyclic stretch, shear stresses, hydrostatic pressure, and

4 microenvironmental stiffness. Vascular cells sense these external stimuli through a series of mechanotransduction events including membrane bound receptors for cell- matrix adhesions (integrins), cell-cell adhesions (cadherins), cell-fluid flow interfaces

(e.g. VEGFR2/PECAM-1/VCAM47), and stretch-sensitive receptors (e.g. ion-channels

48,49). Cells then integrate these external stimuli through a series of downstream signaling events leading to changes in cellular phenotype. I will focus primarily on mechanotransduction events specific to matrix stiffness-sensing, as it is the focus of this thesis, but the contributions of the other aforementioned mechanical forces are of equal significance in vascular physiology and disease.

Much like how we pull on an object with more or less effort to perceive its resistance/stiffness, early studies in the 1980-1990’s visualized and quantified cellular traction forces50–52 and developed platforms to directly show that cells can sense an external substrate’s resistance to intracellularly-generated traction forces53,54. Notably,

Pelham and Wang developed a collagen-coated polyacrylamide hydrogel substrate, still extensively used in stiffness-sensing studies today, that is ideally elastic, can span physiologic to pathologic elastic moduli, and most importantly precisely control matrix stiffness while keeping cell-adhesion density constant. Their original study showed that matrix stiffness controlled focal adhesion formation, and since then, several other tunable stiffness platforms have been engineered using other natural and synthetic polymers to understand cell stiffness-sensing and response.

Critical to stiffness-sensing is the ability of the cell to engage the ECM through integrins.

Integrins are a family of heterodimeric, transmembrane proteins that directly bind to

ECM proteins and can become activated upon engagement and with applied force. The 5 activation of integrins results in a complex cascade of intracellular signaling events, in constant feedback, that trigger the formation and maturation of focal adhesions which physically link the ECM to the internal actin cytoskeleton. Small RhoGTPases (e.g.

RhoA, Rac1) are activated during this process and promote further organization of the cytoskeleton through increased actin-myosin generated forces. The forces propagate inwards to the nucleus through the LINC (linker of nucleoskeleton and cytoskeleton) complex but also back out toward the integrin/focal adhesion complex, resulting in a feedback control loop of further integrin activation until cell generated forces and resistive forces from the matrix are balanced and tensional homeostasis is achieved. At pathologic stiffness, this process is greatly altered and tensional homeostasis can be disrupted55–58.

The initial events in force propagation occur rapidly and do not require changes in gene expression or new protein synthesis as the acute response utilizes already present cellular components in a series of second messenger, phosphorylation, polymerization, and nucleotide-exchange events. However, the initial events ultimately result in longer time-scaled transcriptional changes that ultimately govern critical cellular processes like proliferation and differentiation, and also induce further expression of focal adhesion and cytoskeletal genes. These transcriptional changes occur at several levels involving transcription factors in the immediate early response, and regulation at the epigenetic

(e.g. histone modification) and post-transcriptional level (e.g. microRNA). Several of these processes will be discussed below.

Mechanotransduction: Transcriptional Response

Forces generated from stiffnesss-sensing and stretching were found to repress gene expression through PRC2 (polycomb-repressor complex 2)-mediated histone modifications (e.g. H3K27me3 and H3K9me2,3). These led to heterochromatin and transcriptional silencing of several mechanosensitive genes, and this effect was dependent on transduction through LINC complex proteins (e.g. nesprins, lamins, and emerin)59–61. These forces not only influence chemical modifications of chromatin, but can also physically stretch, position, and de-condense chromatin to enable transcription

62,63. Furthermore, the nucleus itself contains mechanosensitive elements and like the cytoskeleton, the nucleoskeleton responds by stiffening in response to applied force64.

Several mechanosensitive transcription factors have been described. MAL (MRTF-A or

MLK1) is a transcriptional co-activator that potentiates the activity of serum response factor (SRF) to regulate expression of genes in myogenesis65. YAP/TAZ are also transcriptional co-activators that potentiate the activity of TEAD transcription factors to regulate proliferation and differentiation66. Both YAP/TAZ and MAL nuclear localization are regulated by stiffness-driven actin polymerization and nucleoskeletal/cytoskeletal tension64,67–69. NF-kB also responds to matrix stiffness70 and has been extensively observed in response to shear stresses, requiring Rho-GTPase activity (Rac1) and cytoskeletal organization71. Gene targets of early transcription factor activation may also feedback to further strengthen upstream mechanotransduction events such as the transcription of focal adhesion complex proteins72.

Matrix stiffness also regulates gene expression co- and post-transcriptionally. In breast tumors and endothelial cells, the expression of fibronectin-EDB, VEGF-165b, and PKC- 7

βII isoforms are upregulated by matrix stiffness via activation of serine/arginine rich (SR) splicing factors73. MicroRNAs (miRNAs), a class of short non-coding RNAs that target mRNAs for degradation, also control the post-transcriptional response to stiffness. For example, matrix stiffness increases miR-18a to reduce levels of its targets (e.g. PTEN) contributing to malignant phenotypes in breast cancer74. miR-203, which is inhibited by stiffness in breast epithelium, targets ROBO-1/Rac1 to regulate cell contractility and focal adhesion complexes75.

Indeed, transcriptional regulation is a complex process. New layers of transcriptome properties and classes of regulators are being exponentially discovered due to advances in next-generation sequencing technologies and transcriptome-wide bioinformatic algorithms. For example, the thousands of recently discovered long non-coding RNAs,

RNA-modifications/editing, and novel splicing patterns are only beginning to be mechanistically implicated in critical cellular functions. As the group of studies discussed above have yielded significant insight into mechanotranduction mechanisms, whether and how mechanotransduction interfaces with these new transcriptome layers is intriguing.

Specific Mechanotransduction Mechanisms in Vascular Smooth Muscle Cells

The prior section discussed mechanotransduction events elucidated from multiple cellular systems and engineering platforms. Much of these mechanisms are intact in

VSMCs and will be discussed in detail in this section.

As briefly mentioned above, changes in ECM material properties and vascular cells phenotypes during development and disease can dictate how various mechanical stimuli are transmitted to and sensed by cells. For example, changes in vascular stiffness often 8 coincide with changes in vascular wall thickness, luminal diameter, and intimal barrier permeability, resulting in altered cell exposure to cyclic wall stresses and fluid shear stresses. Again, this section will focus mainly on the VSMC response to matrix stiffness, but the role of these other important mechanical forces on VSMCs has been reviewed by others76,77.

ECM-stiffness-dependent VSMC functions

Early studies using synthetic PEG-hydrogels suggested that matrix stiffness may regulate VSMC phenotypic modulation, where increased stiffness favored the synthetic phenotype78–80. On polyacrylamide-hydrogels, our lab and others showed that VSMCs increased expression of ECM proteins, were more proliferative and motile on stiffer matrices, and that VSMC migration was durotactic (i.e. migrating toward increasing stiffness)44,45,78,81,82. Matrix stiffness also enhanced the VSMC response to platelet- derived growth factor and its receptor (PDGF/PDGFR), a potent regulator of VSMC phenotypic switching in both development and disease83. VSMC basal contractility (non- stimulated) and active contractility (vasoagonist-stimulated) were both increased at higher stiffness substrates, and this effect was independent to changes in VSMC differentiation state84. In vivo stiffness models, such as femoral-wire injury and pharmacologic inhibition of lysyl oxidase, were consistent with these in vitro findings44,45,81. These studies demonstrate that matrix stiffness regulates key VSMC- specific processes critical in arterial disease progression.

Cell-ECM interaction during VSMC stiffness-sensing

VSMC stiffness-sensing begins with integrin engagement with the underlying ECM protein. Different integrins are activated by different ECM proteins and can affect how

VSMCs may sense the underlying matrix stiffness, resulting in altered VSMC phenotypes85. For example, fibronectin, which is present at low levels in normal arteries is elevated during injury and inflammation and engages mechanosensitive integrins (e.g.

86 α5β1) . Studies which independently changed fibronectin concentrations and substrate stiffness showed that the amount of fibronectin dictated stiffness-driven VSMC migration78. Furthermore, stiffness-mediated increases in VSMC cytoskeletal tension and durotactic migration was greater on fibronectin-conjugated hydrogels than on those conjugated with laminin, a “healthy” basement membrane protein typically disrupted during disease87–90. Mechanotransduction events through VSMC cell-ECM interactions are also influenced and balanced by cell-cell adhesions91,92.

Intracellular signaling during VSMC stiffness-sensing

Our lab recently identified a VSMC mechanosensing pathway where stiffness is transduced through phosphorylation of FAK, p130Cas and Rac1 to regulate stiffness- mediated increases in proliferation, cytoskeletal tension, and cellular stiffening93. A series of in vitro and in vivo studies also identified a potential positive-feedback signaling pathway for matrix stiffening involving cyclooxygenase-2 (COX2) and lysyl oxidase

(LOX). These studies showed that matrix stiffness increases the expression of LOX, collagens, and fibronectin. Stiffness also inhibited COX2 expression, which is necessary to suppress stiffness-mediated expression of collagens and fibronectin44,45,94,95. At the time of writing this dissertation, a study elucidated that the stiffness-sensitive transcriptional co-activator, YAP/TAZ and its DNA-binding partner, TEAD, are necessary for stiffness-mediated expression of LOX and COX-2 in VSMCs96. YAP activation also interferes with SRF/myocardin transcriptional activity leading to repression of contractile/differentiated phenotype genes97.

Vascular Heterogeneity

The arterial vasculature is an extremely diverse active conduit of multiple cell types from multiple embryonic origins experiencing multiple environmental forces.

Vascular bed mechanical property diversity

Vascular bed mechanical stimuli are extremely diverse along the length of the arterial tree: the large diameter aorta contains multiple layers of elastic lamina experiencing pulsatile, high pressure, high velocity flow; medium and smaller muscular arteries (e.g. coronary and mesenteric arteries) have fewer elastic layers; arterioles experiencing less pressures and velocities; and capillaries with laminar non-pulsatile flow are often only lined by a single layer of mural cells. These regional differences can also be appreciated in the observation that atherosclerotic plaques do not develop uniformly but rather site- specifically due to differences in local blood flow properties. The compositional balance between the different ECM proteins (e.g. collagens, elastins, proteoglycans) is also diverse along the vascular bed, contributing to differences in overall mechanical properties and cellular responses through differential integrin engagement30,31,98.

Vascular smooth muscle heterogeneity

VSMCs arise from various distinct embryologic origins: proepicardium (coronary arteries), neural crest (aortic outflow tract-ascending aorta), and somatic and splanchnic mesoderm (descending thoracic and abdominal aorta). VSMCs along the vasculature are also phenotypically plastic and exist in a continuum of phenotypes ranging from

“contractile/differentiated” to “synthetic/proliferative” depending on stimuli. As a result of differing embryologic origins and exposure to mechanical stimuli, regional variations in vascular cell transcriptomes and proteomes exist98–101. For example, VSMCs isolated

11 from different segments of the arterial tree vary in their expression of vasoagonist receptors102,103, cytoskeletal proteins, and transcription factors104, and cell stiffness98.

Furthermore, VSMCs have origin-specific responses to biochemical and mechanical stimuli: response to vasoagonists102,103, propensity to form calcifications105, proliferation, matrix production, integrin expression, and response to TGF-beta106–108. Several studies also show that origin-specific VSMC phenotypes and responses may be relatively stable and maintain their properties when transplanted to different microenvironments, suggesting that VSMC phenotype heterogeneity in vivo may be more genetically controlled than controlled by the local microenvironment109–114. Thus, VSMC diversity and its implication in disease is complex and covered in several comprehensive reviews109,110,115,116.

Most of these lineage-specific VSMC differences and matrix-dependent VSMC responses have been elucidated using predominantly non-human cell and animal models. Recently, Cheung et. al derived VSMCs representing three distinct origins

(neural crest, somitic mesoderm and proepicardium/lateral plate mesoderm) using human-induced pluripotent stem cells, and demonstrated VSMC-origin-specific responses to cytokines and expression patterns of matrix remodeling proteins117.

Understanding the stiffness-dependent differences in human aortic and coronary

VSMCs, which arise from the neural crest and proepicardium, respectively, will contribute to our understanding of differences in aortic and coronary artery disease and will be a goal of this dissertation.

Dissertation Goals

Given the importance of arterial stiffness as a normal consequence of aging and its close association to several diseases, the critical role VSMCs play in responding to and contributing to pathological matrix stiffening, and regional differences in vascular matrix properties and in cellular embryologic origins, this dissertation aims to understand the relative contribution of VSMC origin and matrix stiffness in dictating the VSMC transcriptome landscape. As the in vivo milieu of matrix stiffening is complex, resulting from many mechanical, cellular, and biochemical stimuli, this dissertation will utilize polyacrylamide hydrogel substrates to interrogate only the consequences of stiffness.

Several unbiased approaches will be employed to characterize the VSMC stiffness- sensitive transcriptome, to identify potential transcriptional regulatory networks in the stiffness-response, and to generate several hypotheses that will be subsequently tested.

A focus will be placed on long non-coding RNAs (discussed in detail in the following chapters), as they compose a large portion of the unexplored VSMC transcriptome and have been implicated in key cellular processes.

CHAPTER 2: CHARACTERIZATION OF THE VASCULAR SMOOTH MUSCLE CELL

STIFFNESS-SENSITIVE TRANSCRIPTOME

Introduction

Recent advances in next-generation sequencing and high-throughput technology has enabled the rapid discovery of nucleic acid sequences and modifications found in cells: genomes, transcriptomes, epigenomes, methylomes, etc. Unlike microarrays, which only allowed for detection of pre-defined targets, next generation sequencing allows for unbiased discovery and quantification of both known and novel nucleic acid sequences.

This has allowed for the rapid expansion of knowledge about the human transcriptome, the set of all RNA molecules transcribed from genomic DNA.

In the last decade, the known human transcriptome has expanded from being predominantly composed of protein-coding transcripts to predominantly non-coding transcripts. In early 2009, the GENCODE project reported that 47,553 human genes were known (15,739 of these genes being non-coding RNAs). By the end of 2016,

58,219 human genes were known with about 24,355 of those being non-coding RNAs118.

While this represented an increase of 10,666 known genes, the number of known transcripts increased even further from 132,067 to 199,325, representing newly discovered isoforms of previously well-studied genes as well as previously unannotated genes. While our ability to detect and define novel genes and transcripts have rapidly evolved, the vast majority of the transcriptome is still functionally unknown.

Transcriptome landscapes and determinants of diversity

Transcriptome landscapes are extremely complex and diverse between cell-types, tissue-types, individual organisms, and physiologic-pathologic state. Several efforts have

14 utilized RNA-sequencing to define the transcriptomes of various cell types and tissues from thousands of different humans and other organisms (GTEx, ENCODE, NIH

Roadmap Epigenomics)119–121. These multi-cellular and multi-tissue studies demonstrate that there are multiple layers of transcriptome diversity forming the basis of cell, tissue, and organism diversity.

In particular, transcriptome complexity arises partly from cell and tissue-type specific gene expression (including transcription factors), alternatively spliced isoforms, alternative transcription initiation/termination isoforms, and post-transcriptional RNA regulators (e.g. microRNAs and editing)122. Human genomic variations, such as single nucleotide polymorphisms (SNPs) and copy number variations (CNVs), also lead to transcript diversity in the form of expression quantitative trait loci (eQTL), splicing quantitative trait loci (splicing-QTL) and allelic-specific expression (ASE), where a variant is statistically correlated to changes in gene expression pattern123.

As discussed in Chapter 1, the vasculature is heterogeneous across the body, and different vascular beds and their VSMCs arise from different developmental origins.

Multiple studies (including GTEx and NIH Roadmaps Epigenomics)119,121 have profiled the transcriptomes of different arteries (e.g. aorta, coronary, tibial), but these data represent the expression not only from VSMCs but from all other vascular cells such as endothelial cells, macrophages, and fibroblasts. Thus, while the transcriptomes of whole vascular tissue from different regions have been interrogated, the transcriptome landscapes of origin-specific VSMCs have not yet been explored or directly compared.

Long non-coding RNAs

Long non-coding RNAs (LncRNAs) were among the thousands of newly identified transcripts as a result of RNA-sequencing. LncRNAs are defined as transcripts greater than 200 nucleotides in length which can be spliced and polyadenylated, but with no coding potential. They contribute significantly to transcriptome diversity as lncRNA expression patterns exhibit higher cell- and tissue-specificity than protein-coding genes124–126. Furthermore, even if their expression is not specific, lncRNA functionality can be cell-type specific. A large lncRNA-inhibition screen in multiple cell types identified lncRNAs that controlled cell proliferation in only a few cell types, despite being ubiquitously expressed in all cell types127.

LncRNAs have been implicated in biological, developmental, and pathological processes and act through mechanisms such as chromatin modifications, cis-regulation of target genes, and post-transcriptional regulation of mRNA processing125,128,129. Several studies have identified VSMC-specific and VSMC-functional lncRNAs such as SENCR, lincRNA- p21, MYOSLID, and SMILR130–134. These lncRNAs were found to be involved in processes such as TGF-β signaling, actin-stress fiber formation, and regulation of transcription factors like MAL, suggesting that lncRNAs could play a role in mechanotransduction. To my knowledge, no lncRNA has been implicated in the cellular response to mechanical stimuli and this chapter serves to identify lncRNAs that respond to matrix stiffeness.

Alternative Isoforms and Splicing

The expression of alternative gene isoforms is cell and tissue specific135 and significantly contributes to proteome diversity136. While the number of known isoforms has also grown

16 exponentially, the functional consequences of most isoforms are unknown. However, in contrast to lncRNAs, the mechanism and consequence of alternative splicing is relatively better understood. Alternative splicing is regulated by several factors including splice-site sequence, cis-regulatory sequences in the pre-mRNA, and expression of RNA-binding proteins and splicing factors. Splicing can result in changes in mRNA stability, localization, or reading-frame/translation affecting protein function137. Fibronectin (FN1) is one of the earliest genes identified to have alternatively spliced isoforms (e.g. notably cellular and plasma FN1) with functional consequences. With the advent of deep sequencing, more FN1 isoforms were discovered in different developmental context.

In VSMCs, alternative splicing plays a large role in phenotype modulation. In synthetic/de-differentiated VSMCs, the RNA-binding protein, polypyrimidine tract binding protein (PTBP1), is upregulated, which results in exon exclusion from several differentiated VSMCs genes138. Another RNA-binding protein, Quaking (QKI) is upregulated in human neointimal injury and in synthetic VSMCs. QKI promotes the expression of a Myocardin isoform that is pro-proliferative, while in differentiated

VSMCs, QKI is low and a pro-contractile Myocardin isoform is expressed139. A global analysis VSMC splicing also showed that genes in contractile/differentiated VSMCs tend to have more intron-retention and “poison exons” resulting in pre-mature stop codons138.

As alternative splicing appears to regulate VSMCs during phenotypic modulation and in neointimal lesions, which are characteristically stiffer81,93, we hypothesize that matrix stiffness alters alternative splicing events through differential regulation of splicing factors and RNA-binding proteins. Indeed, in a gene-specific manner, a recent study identified that matrix stiffness regulated alternative splicing of fibronectin in endothelial

17 cells73. In contrast, I will approach this question using an unbiased survey of the VSMC transcriptome in response to stiffness.

Bioinformatic pipelines for characterizing transcriptome landscapes

Although next-generation sequencing technology has been able to produce data even more rapidly, the number of predictive tools and knowledge-databases lag behind. For example, unlike protein coding genes, lncRNAs do not possess conserved protein coding domains with which we can predict their biological and cellular function.

Nevertheless, several strategies have been employed to analyze large datasets and characterize global transcriptome properties or identify specific candidate targets.

Machine-learning: Principal component analysis and hierarchical clustering

In this context, machine learning broadly refers to unbiased statistical methods to find characteristics and insights in data. Although “machine learning” has become a buzz- phrase in the last few years due to the explosion of big data, it is conceptually old and encompasses familiar methods such as linear regression as well as new, more complex techniques. The two most common machine-learning techniques used with transcriptome level data are principal component analysis (PCA) and hierarchical clustering.

PCA aims to condense high-dimensionality (>50,000 genes per sample) of transcriptome-wide data into several “principal components”. Instead of plotting a

50,000-dimensional plot (where each axis represents one gene) which cannot be visualized, PCA aims to represent the variation of those 50,000 dimensions in only 2 or 3 dimensions which can be easily visualized in a traditional plot (where each axis is a principal component). These principal components (which mathematically are 18 eigenvectors) are derived to represent the maximal variations in gene expression. Each principal component is uncorrelated to all other principal components, and typically, 2 - 3 principal components encompass most of the variability in the data. Plotting samples on a PCA plot allows visual assessment of overall similarities and differences between samples, and whether the samples segregate into interesting groups. How samples segregate in a PCA plot is useful for visualizing what sample characteristics is associated with maximal gene expression variation140.

Hierarchical clustering also allows us to take high-dimensionality gene expression data and determine “distances” between genes and/or samples, and mathematically define groups of genes/samples. Genes that behave similarly across all samples are “close in distance” and tend to group together, and samples whose transcriptomes behave similarly across all genes are clustered together141.

Selecting Candidates and Predicting Function

Methods to identify novel stiffness-mediated cellular processes and potential functional candidate genes include sequence and gene set enrichment, conservation analyses, and correlation analyses.

In enrichment analyses, an entire transcriptome is searched for over-represented gene signatures, or, in other words, expression patterns that occur more likely than random.

These signatures are sets of genes grouped by any characteristic. Commonly used grouping characteristics include genes that function in a common cellular process, share similar genomic positions, or respond to a common stimulus. This general method is

19 employed in this chapter and in subsequent chapters. Specifics about this method relevant to other chapters will be discussed in those sections.

Sequence information obtained through RNA-sequencing also enables several analysis opportunities to identify novel transcriptome characteristics and responses, such as the identification of novel and unannotated genes, the identification of novel splicing patterns and transcripts, and conservation analyses. Conservation at the sequence and structural level is often a characteristic used as an indicator that an RNA or protein contains evolutionarily important functionality. Conservation as a predictive tool is particularly evident in that homologous protein domains often have similar functions. In part, microRNA conservation analyses also enabled the discovery of previously unknown microRNA genes, the prediction of their targets, and the understanding of their biogenesis. In comparison to protein coding genes and microRNAs, lncRNAs are significantly less conserved. However, several lncRNAs with conserved properties (e.g. sequence level, syntenic position, exon structure, tissue-specific expression) have been identified with cross-species functionality142. The lncRNAs discussed in the introduction of Chapter 3 and interrogated in Chapter 4 represent functionally conserved lncRNAs with sequence and positional conservation.

Correlation analysis is also a particularly powerful tool in identifying networks of functional gene relationships. Early analyses of microarray data utilized expression pattern correlations to identify sets of tightly co-expressed genes. Some sets were later experimentally determined to be regulated by common upstream mechanisms.

This method can also be used to infer the function of an unknown gene from the known function of a highly-correlated gene143–146, and will be used for predicting lncRNA 20 function, described more in following chapter. Correlation analyses can prioritize potentially human disease-relevant genes from a longer list. For example, one approach is to use observations from genome-wide association studies (GWAS). GWAS discovers common human variants associated with human diseases and traits. Although GWAS does not prove a causal link between a variant and a disease, it is often inferred that genes near a GWAS variant (within the same haplotype-block) are functionally relevant in those diseases. Indeed, several efforts have demonstrated GWAS-identified variants to be functionally important in disease pathogenesis147. Thus, a list of candidate genes in

VSMCs, for example, could be prioritized by selecting genes that are in proximity to

GWAS-identified variants in vascular disease, as they may be significant in vascular disease processes.

Lastly, all the outputs from these aforementioned analyses can feedback into larger machine-learning algorithms to improve predictive power. For example, multiple datasets have been incorporated into machine learning models to predict whether a lncRNA is a functional growth modifier. In this method, characteristics about experimentally-validated growth-modifying lncRNAs (e.g. specificity of expression, correlated expression to nearby coding genes, number of exons, proximity to disease SNPs, proximity to epigenetic marks) can be used to build a model that predicts whether an unknown lncRNA is likely to be a growth-modifying lncRNA or not127.

Summary

In this chapter, I define and characterize the stiffness-sensitive transcriptome in human aortic and coronary VSMCs. To achieve this, I performed deep RNA-sequencing of

VSMCs grown on physiologically soft and pathologically stiff hydrogels matrices. I used

21 donor-matched aortic and coronary VSMCs such that observed differences in aortic and coronary VSMC expression are due to the VSMC-origin and not from a confounding donor-specific effect. This chapter establishes the similarities and differences in the transcriptome response of aortic and coronary VSMCs to matrix stiffness in gene expression, alternative splicing, and predicted functional consequences of transcriptome differences.

METHODS

Cell Culture on Stiffness-tunable Hydrogels

Donor-matched human primary aortic (Ao) and coronary (Co) artery smooth muscle cells were cultured with smooth muscle growth medium-2 (SmGM2, Lonza) with all supplements. Donor-matched primary VSMCs were used for the RNA-seq experiments.

Primary and hTERT-immortalized Co VSMCs (XS12C1, Lonza) were also used for investigating MALAT1 and PACER biology in Chapter 3 and 4. VSMCs were serum- starved for 48 hours in smooth muscle basal medium (SmBM, Lonza) supplemented with heat-inactivated fatty acid-free bovine serum albumin (1 mg/mL) before plating on hydrogels with SmGM2.

Preparation of Polyacrylamide Hydrogels

Physiologically soft (2-4 KPa) and pathologically stiff (20-25 KPa) hydrogel matrices coated with human fibronectin were fabricated as previously described148,149. These elastic moduli are comparable to those reported for human atherosclerotic lesions150 and have been shown previously to affect VSMC proliferation44,81. Furthermore, we used fibronectin as a vascular disease relevant cell adhesion matrix, as fibronectin expression is limited in normal healthy arteries but increased in disease such as atherosclerosis151,152. We prepared polyacrylamide hydrogels as previously described148,149. Briefly, acrylamide/bis-acrylamide solutions were polymerized on glutaraldehyde-functionalized coverslips at ratios 0.3%/7.5% for pathologically stiff (20 to

25 KPa) and 0.03%/7.5% for physiologically soft (2 to 4 KPa) with acrylic acid N- hydroxysuccinimide ester. Hydrogels were then functionalized with 5 µg/mL human fibronectin overnight and blocked with 1 mg/mL bovine serum albumin prior to cell 23 seeding. Hydrogels were made on coverslips of varying sizes (50-mm for the RNAseq study, 18-mm for validation, knockdown, and overexpression studies, and 12-mm for

EdU incorporation and migration studies).

RNA-sequencing and Bioinformatic Analysis

Four sets (biological replicates) of serum-starved donor-matched Ao and Co VSMCs were plated at 10,000 cells/cm2 for 24 hours on 50-mm soft and stiff hydrogels (4 samples per set, total 16 samples). Cells were seeded at 10,000 cells/cm2, a relatively low density, to minimize cell-cell interactions such that transcriptional changes resulted primarily from cell-ECM stiffness interactions. Total RNA was isolated from cells using

RNeasy Mini Kit (QIAGEN) and TRIzol reagent following the manufacturer’s instructions.

Ribosomal RNA was depleted using Illumina Ribo-Zero Gold, and libraries were prepared using a strand-specific TruSeq RNA library Prep kit (Illumina). Samples were then sequenced on a HiSeq2500, 100bp paired-end (Illumina). Reads were mapped against hg19 using STAR153. Alignment files were then filtered for high quality reads

(score >= 30) and sorted using SAMtools154. Counts per gene were determined using

HTSeq155 and human GENCODE v19 annotations118. Normalization for expression and differential expression were done using DESeq2156. Stiffness-sensitive genes were defined based on a false-discovery rate cut-off < 1% for each Ao and Co VSMCs. For conservation, tissue-specificity, pathway, and correlation analyses, stiffness-sensitive genes were determined by combining Ao and Co VSMC data with a false-discovery rate cut-off < 1%.

These data have been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO Series accession number GSE100081

(https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE100081).

Expression Analysis and Transcriptome-level Clustering

We defined a gene to be expressed in cultured Ao and Co VSMCs if the average number of normalized counts determined by DESeq2 was greater than 8. We also assessed expression cut offs of up to 100-200 normalized counts and found similar qualitative findings between Ao VSMCs and Co VSMCs. To assess gene expression in whole human coronary and aortic tissues, we used expression data from GTEx

(downloaded from the GTEx portal, gtexportal.org). An RPKM > 0.7 cut off was used to classify a gene as being expressed in each tissue type.

We defined lncRNAs as genes annotated in GENCODE with the following transcript_type terms: lincRNA, antisense, processed_transcript, sense_intronic, and sense_overlapping.

Principal component analysis and hierarchical clustering were performed in R using the top 500 variant genes among all RNA-seq samples. Euclidean distances were used for complete-linkage hierarchical clustering. Top variant genes were determined by calculating the expression variance of each gene across all samples (Ao and Co VSMCs cultured on both soft and stiff). Heatmaps were generated in R using the pheatmap package. Venn diagrams were generated in R using the VennDiagrams package.

Conservation and Tissue Specificity Analysis 25

To assess the sequence conservation of stiffness-sensitive v. insensitive genes, we used the pre-computer PhastCons 46-way (vertebrate) scores that were downloaded from the UCSC browser (http://genome.ucsc.edu/). Scores were calculated for each exon per gene accounting for the length of each exon. Cumulative distributions were determined and plotted for stiffness-sensitive and stiffness-insensitive genes. A

Kolmogorov–Smirnov test was performed to determine statistical significance between distributions.

For each stiffness-sensitive and insensitive gene identified in Ao and Co VSMCs, we counted the number of tissues where that gene had a median RPKM > 0.7 in the GTEx database. Cumulative distribution functions of the number of tissues expressing VSMC- identified (in this study) stiffness-sensitive v. insensitive genes were generated. A

Kolmogorov–Smirnov test was performed to determine statistical significance between these distributions.

Pathway Analysis and Correlation Analysis

Differential gene expression lists generated by DESeq2 using combined Ao and Co

VSMC expression values were uploaded into Ingenuity Pathway Analysis (QIAGEN). A

FDR < 1e-7 and an absolute LogFC > 1 were used to select analysis-ready molecules.

Top enriched “categories” and “functions” were assessed. We identified lncRNAs attributed to each significant category or function. These enriched lncRNAs were then ranked based on the number of occurrences within the significant category or functions.

KEGG and REACTOME terms were enriched using the Gene Set Enrichment Analysis tool (GSEA, Broad Institute)157. For each VSMC cell type, genes were ordered based on 26 their FDR values resulting from differential expression analysis (DESeq2). This ordered list was used as an input for GSEA. Significant term enrichment was determined using an FDR <1% for KEGG terms or PWER < 1% for REACTOME terms.

To characterize the transcriptome landscape locally to each lncRNA, Spearman’s correlation coefficients and p-values were calculated for all pairs of genes within the same chromosomes using the combined Ao and Co VSMC expression data. We focused on all lncRNAs and plotted correlation scores vs. distance from the lncRNA in base pairs. To determine statistically significant correlated lncRNA-gene pairs, we corrected for multiple hypothesis testing (Bonferroni, e.g. p-value threshold for significance was p < 1e-5 for all lncRNA-gene pairs within 50kb from the lncRNA, plotted as red dots). Histogram plots of significant correlations were compared between stiffness-sensitive lncRNAs and an equivalent number of the most variant (calculated by variance across all Ao and Co VSMCs) stiffness-insensitive lncRNAs.

To identify putative functional stiffness-sensitive lncRNA-protein coding gene pairs, we ranked the top stiffness-sensitive lncRNAs (fold change > 2) that were significantly correlated with a stiffness-sensitive protein coding gene (with fold change > 2) with an association p-value < 1e-3 and within 100kb away from each other. The top stiffness- sensitive lncRNA-protein coding gene pairs are presented in Table 5.

Differential Splicing Analysis using MAJIQ

Fasta files and STAR-aligned files from the RNA-seq experiment were used as inputs into MAJIQ to quantify the PSI (percent spliced in), which is the relative abundance of each local splice variations (LSV, or splice junction) in each gene158. For each LSV, the 27 delta PSI (change in PSI between soft and stiff matrices) threshold cutoff was set at 20%

(E(dPSI) > 0.2). MAJIQ and VOILA were run for Co VSMC and Ao VSMC separately.

Association of Stiffness-sensitive Splicing to Cardiovascular GWAS traits.

Differentially spliced genes as defined above were associated with SNP variants in the

NHGRI-EBI GWAS Catalog (www.ebi.ac.uk/gwas) Version 1.0. The genomic coordinate for each SNP was translated into hg19 coordinates using the track: “GWAS catalog”, table: “gwasCatalog” from the UCSC genome table browser. Only SNPs relevant to vascular disease and genome-wide significant (p < 5 x10-8) were used for analysis. The list of relevant vascular diseases used in this analysis is in Appendix 1. Only genes within 50kB of the disease-associated SNP were considered as “associated” to stiffness- sensitive genes.

RESULTS

The Landscape of the Stiffness-sensitive Transcriptome of Co and Ao VSMCs

(Gene Expression Level)

We implemented deep RNA-sequencing to comprehensively define gene expression in donor-matched aortic (Ao) and coronary (Co) vascular smooth muscle cells (VSMCs) in response to physiologically soft and pathologically stiff hydrogel matrices (Figure 1).

Using GENCODE annotations, we identified a total of 17,716 genes expressed in either

Ao or Co VSMCs and in either soft or stiff conditions (Figure 2A); 93.2% were expressed in both cell lineages (DESeq2 mean normalized counts > 8). Of all expressed genes,

13,574 were protein-coding genes, and 2,379 were long non-coding RNAs. Further inspection revealed protein-coding genes are more commonly expressed in both lineages of VSMCs than lncRNAs (95.7% v 82.9%). These proportions are consistent with expression data obtained from whole human coronary and aortic tissues in the

GTEx database (Figure 3). The degree of uniquely expressed lncRNAs (17.1%) and protein coding genes (4.3%) may reflect the unique developmental origin and transcriptome program of Ao and Co VSMCs.

Principal component analysis of the top 500 most variant genes revealed that matrix stiffness is a major contributor to expression differences (as is it most segregated by the first principal component) and accounts for approximately 54% of variation in gene expression. VSMC origin is also a major contributor accounting for about 36% of the expression variance (Figure 2B). Consistent with this analysis, unsupervised hierarchical clustering with the 500 most variant genes revealed that Ao and Co VSMCs cultured on the same matrix stiffness clustered together (Figure 2C). This clustering is similar when

29 clustering based only on protein coding genes or even lncRNAs despite their reduced overlap in expression between Ao and Co VSMCs (Figure 2D).

Figure 1. Experimental and computational flow chart for interrogating the VSMC stiffness- mediated transcriptome. 31

Figure 2. Expression Landscape of Aortic and Coronary VSMCs

(A) Venn Diagrams depicting the overlap of expressed genes (defined as normalized counts > 8) in Ao VSMCs and Co VSMCs by type. (B) Principal component analysis plot of the first two principal components using the top 500 variant genes (all genes). (C) Hierarchical clustering of all Ao and Co VSMC samples using the top 500 variant genes (all genes). (D) Principal component analysis and hierarchical clustering of all Ao and Co VSMC samples by either the top 500 variant protein coding genes or lncRNA genes

Figure 3. Expression Landscape of Whole Aortic and Coronary Artery Tissue from GTEx Database

(A) Comparison of expression overlap of human aortic tissue and coronary artery tissue from GTEx. Venn diagrams depicting the overlap of expressed genes (defined as RPKM > 0.7) in aortic and coronary tissue expression data.

Global Characteristics of Stiffness-sensitive Gene Expression of lncRNAs and Protein-coding Genes

We conducted pair-wise differential expression analysis and found that a total of 5818 genes are differentially expressed by matrix stiffness (stiffness-sensitive, SS) in either

Ao or Co VSMCs (FDR<1%) (Figure 4A,Figure 4B). From here on, we will define SS genes as the following: Origin-Specific (OS-SS) are genes identified as SS in only Ao

OR Co VSMCs; Origin-Non-Specific (ONS-SS) are genes identified as SS in both Ao

AND Co VSMCs; ALL-SS are genes identified as SS in either Ao OR Co VSMC (see

Figure 4B for a graphical representation).

Of the genes identified as stiffness-sensitive, 3098 (53.2%) are ONS-SS, with 2842 being protein coding genes and 157 being lncRNAs. Interestingly, lncRNAs are more likely to be OS-SS than protein coding genes. That is, 57% (206/363) of ALL-SS lncRNAs are OS-SS, compared to only 45% (2377/5219) of ALL-SS protein coding genes are OS-SS (Figure 4C).

We also found that if a gene was SS in either VSMC (ALL-SS), that gene was more likely to be expressed (DESeq2 normalized count > 8) in both VSMC lineages than a gene that was stiffness-insensitive (Figure 5). For example, 93% of ALL-SS lncRNAS are expressed in both VSMCs and 88% of OS-SS lncRNAs are expressed in both

VSMCs. This is in comparison to stiffness-insensitive lncRNAs, where only 83% of stiffness-insensitive lncRNAs are expressed in both VSMCs. This is similar, but to a lesser degree, for protein coding genes. 98% of ALL-SS protein-coding genes are expressed in both VSMCs and 97% of OS-SS protein coding genes are expressed in both VSMCs, while 96% of stiffness insensitive protein coding genes are expressed in

34 both VSMCs (Figure 5A, Figure 5B). We also analyzed these percentages over different

DESeq2 count thresholds (used to define if a gene was expressed) and found that these qualitative findings hold true across a large range of cut-off values (Figure 5C). Taken together, these observations suggest that genes classified as SS tend to be expressed in an origin non-specific manner (found to be expressed in both VSMCs), and that the

SS transcriptome may be conserved. Conversely stated, stiffness-insensitive genes were more likely to be expressed in an VSMC-origin specific manner.

Indeed, we find that SS genes were significantly more conserved at sequence level using phastCons scores (Figure 6, Left). SS protein coding genes and lncRNAs both exhibited increased conservation over stiffness-insensitive genes. Furthermore, using data from the GTEx project, we find that SS genes are expressed (cutoff of RPKM>0.7) in more tissue types than stiffness-insensitive genes (Figure 6, Right). The SS transcriptome appears to be a conserved cellular response, which suggests that SS lncRNAs, by way of their increased conservation across species and tissue-types, may control stiffness-dependent functions.

Figure 4. Heatmap of the VSMC Stiffness-sensitive Gene Expression

(A) Heatmap of stiffness-sensitive genes (FDR < 0.01) in both Co VSMC and Ao VSMCs. Heatmap colors correspond to log2 (normalized counts). (B) Legend for diagram of origin-specific stiffness-sensitive genes (OS-SS), origin-non-specific stiffness-sensitive genes (ONS-SS), or stiffness-sensitive genes in either Ao or Co VSMCs (ALL-SS). (C) Venn diagrams depicting the overlap in stiffness-sensitive genes in Ao and Co VSMCs.

Figure 5. Cell-specific expression of stiffness-sensitive genes

(A) Venn diagrams depicting the number of cell-specific stiffness-sensitive genes (by FDR<0.01) that are cell-specific-expressed (by normalized count > 8) or expressed in both Ao and Co VSMCs. (B) Overlap of expressed stiffness-sensitive genes (identified in either Ao or Co VSMCs) in Ao or Co VSMCs. 93% of ALL-SS lncRNAs (identified as stiffness-sensitive in either Ao or Co VSMCs) are expressed (normalized count > 8) in both Ao and Co VSMCs compared to 83% of stiffness-insensitive lncRNAs. More stiffness-sensitive genes are expressed in both cell types than stiffness-insensitive genes. (C) Percent overlap as in panel B but comparing stiffness- sensitive vs. all genes per subtype over various expression cut-offs (DESeq2 normalized count from 0 – 200).

Figure 6. Sequence Conservation and Tissue Expression Analysis of the VSMC Stiffness- Sensitive Genes

Cumulative distributions of conservation (phastCons) scores and tissue expression (GTEx) of stiffness-sensitive v. stiffness-insensitive genes. In-set figure of lncRNA distribution shows zoomed-in distribution at conservation scores 0 to 0.15. A right-shifted cumulative distribution demonstrates higher conservation or expression in more tissues. P-values determined using a Kolmogorov–Smirnov test.

To identify putative stiffness-dependent functions common and unique to Aortic and

Coronary VSMCs, we performed unbiased enrichment analyses using Gene Set

Enrichment Analysis157 and Ingenuity Pathway Analysis (QIAGEN) (Figure 7). Both methods identified several highly enriched cellular processes in both Ao and Co VSMCs: cell cycle, growth, death, motility, morphology, RNA post-transcriptional modifications

(such as splicing).

Figure 7. Predicted Stiffness-dependent cellular functions

(A) Venn Diagrams representing the common and unique gene sets enriched from the KEGG and REACTOME databases. (B) Selected common KEGG and REACTOME pathways enriched in both Ao and Co VSMCs. The full list appears in the Appendix. (C) Top categorical pathways enriched in shared stiffness-sensitive genes in Ao and Co VSMCs using Ingenuity Pathway Analysis

The Landscape of the Stiffness-sensitive Splicing in Co and Ao VSMCs (Isoform

Level)

Another level of transcriptome diversity and regulation is alternative splicing events.

Splicing was a highly predicted process regulated by stiffness in both Ao and Co VSMCs

(Figure 5, Terms: KEGG-Spliceosome, Reactome-mRNA_Splicing, Ingenuity-RNA post- transcriptional modifications). Within these gene sets were stiffness-sensitive RNA- binding proteins and splicing factors such as PTBP1 (upregulated by stiffness (Ao, Co) log2FC (-0.55, -0.6), FDR (3e-3, 8e-4)), previously identified to regulate splicing in VSMC phenotypic modulation. To assess splicing events in response to stiffness, we used the

MAJIQ (Modeling Alternative Junction Inclusion Quantification) algorithm suite158 to detect and quantify local splice variations (or splicing events). In contrast to stiffness- sensitive gene expression, stiffness-sensitive splicing was less similar between Ao and

Co VSMCs. Only 61 genes (12.8%) were differentially spliced in both Ao and Co VSMCs

(Figure 8A). About one-third (23) of these 61 genes were also differentially expressed at the gene level in both VSMCs, while another one-third (23) of these genes were only differentially spliced without change in their gene-level expression (Figure 8B). Thus, alternative splicing appears to be a more origin-specific layer of the stiffness-sensitive transcriptome than gene level expression.

Figure 8. Stiffness-sensitive splicing in Co and Ao VSMCs

(A) Venn Diagram depicting the number of genes differentially spliced by stiffness in Ao and Co VSMCs. A gene was considered differentially spliced by stiffness if that gene had a local splice variation (LSV or splice junction) with > 20% differential exon/intron inclusion (delta PSI). (B) Four-way Venn diagram comparing differentially expressed vs. differentially spliced genes in Ao and Co VSMCs.

Global properties of differentially spliced genes

Using the Database for Annotation, Visualization and Integrated Discovery (DAVID) tool159,160, we looked for properties annotated to differentially spliced genes in both Ao and Co VSMCs. Table 1 depicts the list of enriched annotations and terms associated with differentially spliced genes. Surprisingly, a significant number of differentially spliced genes are annotated as “phosphoproteins” in the uniprot database (35 out of 61).

Several terms are also associated to known mechanotransduction components: cytoskeleton and actin-binding proteins. Four genes containing SH3 domains were also enriched in this analysis (Table 2). SH3 domains (Src homology 3 domains) are protein domains often found in signal transduction proteins allowing for protein-protein interactions especially in cytoskeletal and small GTPase signaling pathways161,162.

Whether these four genes and their SH3 domains are involved in stiffness-dependent downstream effects is unknown.

Thus, we further analyzed how the differential splicing events in these four genes may alter their properties. Surprisingly, the stiffness-mediated splicing events in these four genes directly involved their SH3 domains (Figure 9). We found that matrix stiffness in 3 out of 4 of these genes (SASH1, MACF1, and MIA3) increased the relative abundance of exon junctions consistent with transcripts lacking SH3 domains (Figure 9 A, B, D).

And at the gene expression level, both MACF1 [(Ao,Co), log2FC (0.77, 0.9), FDR (9e-6,

4e-6)] and MIA3 [(Ao,Co), log2FC (1.1, 1.3), FDR (9e-10, 2e-10)] were down-regulated by stiffness. In contrast, SH3PXD2A (Figure 9 B) had a more complex local splicing pattern with multiple acceptor exons and 1 common splice donor exon. For SH3PXD2A, stiffness increased the relative abundance of transcripts containing 3-4 SH3 domains

(Figure 9 C, green junction), and decreased the relative abundance of a novel junction 43 not annotated in the GENCODE annotations (Figure 9 C, red junction). This novel junction is predicted to cause a frameshift and a premature stop codon resulting in a transcript with only 1 partial SH3 domain. Taken together, stiffness alters the relative abundance of SH3 domains in transcripts of these 4 genes and may affect contribute to stiffness-mediate signal transduction.

Benjamini Category Term Gene Count P-Value Corrected p-Value UP_KEYWORDS Alternative splicing 43 0.000027 0.0037 UP_KEYWORDS Phosphoprotein 35 0.00033 0.023 UP_KEYWORDS Cytoplasm 23 0.0026 0.085 UP_KEYWORDS Cytoskeleton 10 0.0023 0.1 UP_KEYWORDS Acetylation 18 0.0042 0.11 UP_KEYWORDS Actin-binding 5 0.0054 0.12 UP_KEYWORDS Coiled coil 16 0.0081 0.15 UP_KEYWORDS SH3 domain 4 0.018 0.27 UP_KEYWORDS Glycolysis 2 0.077 0.57 UP_KEYWORDS Isopeptide bond 7 0.07 0.6 UP_KEYWORDS Tumor suppressor 3 0.076 0.6 UP_KEYWORDS Nucleus 20 0.061 0.62 UP_KEYWORDS Ubl conjugation 9 0.069 0.63 UP_SEQ_FEATURE splice variant 36 0.000027 0.0083 UP_SEQ_FEATURE compositionally biased region:Poly-Ser 6 0.01 0.65 UP_SEQ_FEATURE compositionally biased region:Poly-Gln 4 0.0077 0.7 UP_SEQ_FEATURE compositionally biased region:Pro-rich 7 0.04 0.96 UP_SEQ_FEATURE short sequence motif:Nuclear localization signal 4 0.06 0.98 UP_SEQ_FEATURE domain:SH3 3 0.076 0.98

Table 1. Enriched annotations of differentially spliced genes

Enriched uniprot terms associated with the 61 commonly differentially spliced genes in both Ao and Co VSMCs. Categories are derived from the uniprot database (uniprot.org). The number of genes associated to each term is denoted under “gene count”.

Gene Symbol Gene Name MACF1 microtubule-actin crosslinking factor 1 MIA3 MIA family member 3, ER export factor SASH1 SAM and SH3 domain containing 1 SH3PXD2A SH3 and PX domains 2A

Table 2. Differentially spliced genes with SH3 domains

Genes associated with the uniprot term SH3 domain and are differentially spliced by stiffness in either Ao or Co VSMCs

Figure 9. Schematic diagrams of differentially spliced genes containing SH3 domains

(Continued below) 46

Figure 9 (Continued). MAJIQ Local Splice Variation (LSV) Diagrams depict the splicing junctions found to be stiffness-sensitive. The yellow square represents an exon that acts as the common splice acceptor/donor. Red, blue and green squares represent exons that act as the differentially included or excluded exon. The percent spliced in (PSI) is the relative abundance of each junction in the LSV. The horizontal bars represent the delta PSI of each junction on stiff or soft matrices.

Below each LSV diagram is a track view of each gene’s structure from the GENCODE V19 Comprehensive Set Track from UCSC Genome Browser. Each isoform and the exon-intron structure is represented in black. The dotted red line represents where an SH3 domain is encoded. Red, blue, green and yellow highlighted regions correspond to the same exon squares in the LSV diagram. Junctions are also depicted above the gene track. (A) Schematic for SASH1. The red junction, increased relative abundance by stiffness, represents an isoform lacking the SH3 domain. (B) Schematic for MACF1. The blue junction, increased relative abundance by stiffness, represents an isoform lacking the SH3 domain. (C) Schematic for SH3PXD2A. The green junction, increased relative abundance by stiffness, represents isoforms containing 4 entire predicted SH3 domains. The red junction, decreased in relative abundance by stiffness is not represented in any of the annotated isoforms and may be a novel junction. This junction would result in a premature stop codon resulting in a transcript containing 1 partial SH3 domain. (D) Schematic for MIA3. The blue junction, increased in relative abundance by stiffness represents an isoform lacking the SH3 domain.

Association of stiffness-sensitive spliced genes with cardiovascular disease-

GWAS variants

Several GWAS variants have been associated with alternative splice sites and events, termed splicing quantitative trait loci (splicing-QTL)163. We asked whether any of the 61 commonly spliced genes were associated with GWAS variants of cardiovascular traits relevant to aortic disease, coronary disease, or vascular stiffening. Only 2 of the 61 stiffness-spliced genes were associated with GWAS-identified loci: FURIN-FES and

MIA3. Variants near FES and MIA3 were associated with coronary artery disease, myocardial infarction, and blood pressure.

As described previously, the relative abundance of MIA3 containing SH3 domains is reduced by stiffness. Three disease-associated variants are found within the MIA3 gene; one variant (rs35700460) is located in the intron directly upstream of the isoform lacking the SH3 domain, and two (rs17465637 and rs67180937) are located further downstream

47 near constitutive exons (Figure 10 A). The risk-alleles for all three SNPs are eQTLs and associated with reduced MIA3 gene expression (in aortic but not coronary tissue), but none were splicing-QTLs in the GTEx database (Table 3).

Two variants were also identified in the FURIN-FES locus (Figure 10 B). One variant

(rs2521501) located in an intronic region of FES, was associated with blood pressure (a consequence and risk factor of vascular matrix stiffening). This variant was also identified in a myocardial infarction GWAS, but its p-value (2e-7) was just under the genome-wide significance threshold (p < 5e-8). The second variant (rs17514846) is located in an intronic region of FURIN and is associated with coronary artery disease but also with a p-value (3e-7) just under the genome-wide significance threshold. Whether both of these variants are linked to changes in FURIN and/or FES activity is unknown.

The FES intronic variant (rs2521501) is an eQTL for FES, with the risk-allele associated with reduced expression (in aorta and tibial artery but not coronary tissue), but not a splicing-QTL in the GTEx database (Table 3).

As FES was differentially spliced by stiffness and intronic variants in FES were associated with several cardiovascular traits, we further investigated the splicing pattern in FES. Surprisingly, matrix stiffness increased the relative abundance of exon junctions consistent with FES transcripts lacking an SH2 domain (Figure 10 B, blue junction). SH2

(Src homology 2) domains, like SH3 domains, are critical in protein-protein interactions and signal transduction, but specifically allow binding to phosphorylated tyrosine residues161,162. Taken together, these observations suggest that stiffness-mediated splicing of FES and MIA3 could be a mechanism contributing to vascular-related diseases.

Gene Symbol SNP P-Value Effect Size Tissue MIA3 rs67180937 1.3E-06 -0.30 Artery - Aorta MIA3 rs35700460 3.9E-06 -0.28 Artery - Aorta MIA3 rs17465637 1.3E-06 -0.29 Artery - Aorta FES rs2521501 2.6E-10 -0.34 Artery - Aorta FES rs2521501 4.3E-08 -0.21 Artery - Tibial

Table 3. Expression quantitative trait loci (eQTL) for cardiovascular disease-related GWAS variants associated with stiffness-sensitive genes.

These data are from the GTEx (gtexportal.org) database. eQTLs were searched in aorta, coronary artery, and tibial artery tissues. Effect sizes < 0 correspond to reduced expression associated with the risk-allele.

Figure 10. Schematic diagrams of stiffness-sensitive spliced genes and relationship to cardiovascular trait GWAS-identified variants.

(continued below)

Figure 10 (Continued). For each locus (A) and (B), the table of GWAS-identified variants (SNP) and the associated disease trait is listed. The reference for the discovery of each variant is listed: Wain LV23, Ehret GB164, Schunkert H164, Nikpay M165. The location and type of each variant is listed and depicted each schematic (green arrow). (A) Schematic for the MIA3 Locus. Table of the GWAS-identified traits in the MIA3 locus. The blue junction, increased in relative abundance by stiffness represents an isoform lacking the SH3 domain. (B) Schematic for the FURIN-FES Locus. Table of the GWAS-identified traits in the FURIN-FES locus. The blue junction, increased in relative abundance by stiffness represents an isoform lacking the SH2 domain.

DISCUSSION

Although vascular stiffness has been identified as an important risk factor for cardiovascular disease, our understanding of how vascular stiffness alters vascular cellular behavior is incomplete. This study provides the first comprehensive analysis, at the gene-level and isoform-level, comparing the stiffness-sensitive transcriptome in human Ao and Co VSMCs with a focus on lncRNAs. We found that Ao and Co VSMC transcriptome profiles are significantly dictated by matrix stiffness, and that stiffness- sensitive genes (whether protein coding or lncRNA) are more commonly co-expressed in

Ao and Co VSMCs compared to stiffness-insensitive genes. We also found that VSMC stiffness-sensitive genes are overall more species conserved and expressed in more tissue types than stiffness-insensitive genes. Furthermore, we observed that stiffness- sensitive lncRNAs may be more origin-specific response (between Ao and Co VSMCs) than stiffness-sensitive protein coding genes.

We find that ECM stiffness significantly dictates the transcriptomic identity of Ao and Co

VSMCs. Prior studies have suggested that VSMC from different embryologic origins are rather phenotypically stable even when transplanted into disparate vascular microenvironments98,110,111; however, these transplantation studies are uncontrolled for any particular microenvironmental parameter. By using donor-matched Ao and Co

VSMCs, any differences we observe in the transcriptome of Ao and Co VSMCs arise from contributions from embryologic lineage programs or the effect of ECM stiffness. Our clustering analysis suggests that ECM stiffness may override the embryologic-origin contributions to the transcriptome; the transcriptome of Co VSMCs/stiff matrix is more similar to Ao VSMCs/stiff matrix than Co VSMCs/soft matrix. However, we also did identify the presence of OS-SS genes and observed that SS lncRNAs tended to be more origin-specific than SS protein coding genes.

Several studies support this notion that ECM stiffness may significantly dictate cell transcriptomic identity. ECM stiffness and other mechanical parameters are critical in tissue and organ development166. Early studies demonstrated that ECM stiffness can dictate stem cell fate and that different stiffness can promote different lineages167–170.

ECM stiffness has also been shown to alter nuclear mechanics, chromatin organization and the activity of a variety of ubiquitous transcription factors in multiple cell types, leading to global transcription changes56,60,171–173.

In further support of this idea, we found that ALL-SS (identified as stiffness-sensitive in either Ao or Co VSMCs) tended to be expressed in both VSMC lineages and across more tissue types than stiffness-insensitive genes. These stiffness-sensitive genes were also more conserved across species at the sequence level. This suggests that the stiffness-sensitive gene set may represent a critical response module across many cell types and regulate cell-ubiquitous processes, as predicted by enrichment analyses.

Implications of Splicing Observations

At the transcript/isoform-specific level, we found that many splicing regulatory genes were stiffness-sensitive and an unbiased analysis of junctional reads altered by stiffness yielded many potentially stiffness-sensitive splicing events. We found that the stiffness- sensitive splicing was highly origin-specific with only a 13% overlap compared to stiffness-sensitive gene level expression (53% overlap).

The splicing of cytoskeletal proteins is a significant process in development and disease of cardiac, skeletal, and smooth muscle tissues137. Rather unsurprisingly, we found that cytoskeletal and cytoskeletal associated proteins were enriched in stiffness-spliced genes identified in both Ao and Co VSMC. However, rather unexpectedly, we found that

SH3 domain containing genes were also enriched, and that the stiffness-sensitive splicing events are predicted to result in inclusion or exclusion of these domains. Though not shown above, several SH3 domain containing genes were also significantly spliced by stiffness in a VSMC origin-specific manner. RT-PCR analyses will be required to validate these RNA-seq findings.

There is evidence that SH3 domains are important in VSMC-specific stiffness-sensing.

For example, the SH3 domain in p130Cas, a key mechanotransduction protein, is necessary for transducing matrix stiffness into VSMC proliferation and intracellular stiffening93. As stiffness-dependent activation of p130Cas is phosphorylation-mediated, we did not necessarily expect nor find stiffness-mediated splicing of p130Cas. However, we inspected each of the SH3 domain containing genes and found that stiffness is predicted to repress inclusion of the SH3 domain in 3 genes (SASH1, MACF1, and

MIA3) and include multiple SH3 domains in 1 gene (SH3PXD2A).

MIA3 (melanoma inhibitor activity member 3, a.k.a. TANGO1) is an endoplasmic reticulum resident transmembrane protein that functions in the packaging and export of several collagens (I, II, III, IV, VII, and IX), but not fibronectin, within large vesicles. The

SH3 domain in MIA3 binds directly to collagen VII alpha-1 but not collagen I, despite global defects in collagen I deposition in MIA3-knockout mice174,175. Three intronic SNPs within MIA3 are associated with increased risk for coronary artery disease and myocardial infarction. How these SNPs contribute to this increased risk is unknown.

However, each SNP is a significant eQTL for MIA3 gene expression in aortic but not coronary tissues (GTEx database). The risk-alleles for all three SNPs are associated with reduced MIA3 expression. We also found that stiffness reduced MIA3 gene level expression, with a disproportionate reduction in SH3-containing MIA3 transcripts. One

SNP (rs35700460) is located upstream to the alternative start exon of the SH3-lacking

MIA3 isoform, and perhaps contributes to usage of an alternative promoter. Thus, we could speculate that arterial stiffening and MIA3 locus SNPs may additively contribute to

CAD and MI by reducing MIA3 expression, resulting in an abnormal vascular ECM.

Except for MIA3, the specific role and binding partners of the SH3 domain in the remaining 3 genes are relatively unknown, but their overall function may suggest their role in stiffness-sensing. MACF1 (Microtubule actin crosslinking factor 1 or a.k.a. ACF7) coordinates microtubule alignment along F-actin at focal adhesions. MACF1-knockout keratinocytes exhibited larger focal adhesions and impaired migration without changes in phosphorylated FAK, or levels of active RhoA, Rac1, and Cdc42176. Perhaps the SH3 domain in MACF1 is important for its association with the actin cytoskeleton and may partially control stiffness-mediated migration.

SH3PXD2A (SH3 and PX domains 2A, a.k.a. TKS5, Tyr kinase substrate with five SH3 domains), is a critical and specific podosome/invadopodia adaptor protein required for cell invasion177. Podosome/invadopodia formation is upregulated by ECM stiffness in breast epithelium178 and reports have found SH3PXD2A containing podosome in VSMCs in vitro and in vivo179. Perhaps the stiffness-mediated inclusion (soft substrate- associated truncation) of SH3 domains in SH3PXD2A serves to increase invasiveness of

VSMCs during vascular disease. Lastly, almost nothing is known about SASH1 (SAM and SH3 domain containing protein 1), other than its expression has been linked to smoking and atherosclerosis180.

Lastly, we found matrix stiffness was predicted to splice out the SH2 domain of FES.

FES is a non-receptor protein tyrosine kinase implicated in cytoskeletal dynamics associated with cell-cell and cell-matrix adhesions181. It interacts with many signaling proteins implicated in mechanotransduction including p130Cas182. Whether FES plays a direct role in mechanotransduction is unknown, but the stiffness-mediated loss of an

SH2 domain could significantly the activity of FES and its role in cytoskeletal signaling.

Similar to MIA3, variants in FES were associated with vascular disease and are eQTLs with the risk-allele associated with reduced FES expression.

Thus, in comparison to rapid phosphorylation/enzymatic activities attributed to SH2/SH3- domain proteins, stiffness-mediated splicing of these domains may represent another layer of signaling regulation that occurs at a slower, transcriptional rate. Ultimately, whether these SH3 and SH2 domains are important in transducing the stiffness signal into VSMC functions will require further investigation using splicing blocking oligonucleotides and other molecular manipulations.

There are still many questions that can be asked of this data set. Specifically, in the context of splicing, motif analysis of the splice sites could yield insight into the specific splicing factors mediating stiffness-sensitive splicing. Indeed, motif analysis of differentially included exons in contractile vs. synthetic mouse VSMCs led to the identification of PTBP1138. In our dataset, PTBP1 was differentially expressed by stiffness, but surprisingly, none of the PTBP1 target VSMC genes were differentially spliced (not shown). Of note, QKI, a regulator of myocardin splicing that is upregulated in neointimal VSMCs, was not differentially expressed by stiffness. As briefly mentioned in

Chapter 1, matrix stiffness induced phosphorylation of an SR-splicing factor, leading to alternative splicing of fibronectin. Whether SR proteins are activated by phosphorylation in VSMCs could not be predicted using the RNA-seq data, but may be a potential mechanism for the observed stiffness-mediated splicing events.

Our lab recently has performed RNA-sequencing of mouse VSMCs on varying stiffness substrates. Although there will be differences in the mouse v. human transcriptome regulation at this complexity level, a comparison of stiffness-dependent splicing events in this mouse VSMC dataset with our human VSMC dataset may identify conserved events and splice-site sequences to help curate which isoforms are likely to be functional183, especially in the case of the predicted SH3 and SH2 splicing events.

CHAPTER 3: REGULATORS OF THE STIFFNESS SENSITIVE TRANSCRIPTOME

Introduction

Chapter 1 gave an overview of various types of transcriptional regulators in mechanosensing. This section will now specifically address two classes, transcription factors and lncRNAs, and how several were historically identified and predicted to function.

Mechanosensitive Transcription Factors

Several methods have been employed to find regulators of mechanotransduction. In general, the methods rely on finding patterns (known and unknown) that occur more likely than chance.

In one approach, significance of overlap is tested between the set of all observed differentially expressed genes and the set of genes known to be regulated by a transcription factor (gene expression signature). This is done over all known transcription factors. Those signatures with significant overlaps are likely to explain a part of the stimulus-mediated differential expression. This is the method employed by ingenuity pathway analysis(QIAGEN) and DAVID159 bioinformatics tools. This approach was used in part by the authors who identified YAP/TAZ as mechanosensitive transcription factors.

In combination with historical empirical evidence that suggested YAP/TAZ would be stiffness-sensitive, they tested several in-house curated gene signatures for significant overlap and found only YAP/TAZ to be significant68.

Another approach termed gene set enrichment analysis (GSEA) is a method developed by the Broad Institute and utilizes an array of gene signatures, not limited to those modulated by transcription factors, but also containing common transcription factor

57 motifs, genes on common chromosomes, genes in a common curated cellular process

(e.g. KEGG, REACTOME), genes associated with specific cancers, mutant cells, and genes in response to pharmaceutical compounds. Instead of an overlap, GSEA compares a rank-ordered list of differentially expressed genes (often by some statistic, either fold change or p-value). It then determines which gene signatures in its entire database are over-represented at each end of the differential expression list by calculating an enrichment score and empirically determining the p-value for significance.

This method was used to help identify a PRC2 signature in mechanosensing60.

Lastly, motif searching in promoters of differentially expressed genes is an unbiased approach that searches for common over-represented DNA sequences. These sequences are then matched with known DNA-binding motifs. This method is especially useful for sets of genes that have been poorly studied and therefore has no known signature, such as the thousands of newly identified lncRNAs, and can be done in a targeted way with one gene or a set of genes. This general method of promoter searching was used to identify RARG (retinoic acid receptor-G) as a stiffness-dependent regulator of Lamin A expression171.

However, there are limitations to these methods. For example, in motif searching, while a transcription factor binding motif may be enriched upstream of a set of genes, it is not guaranteed that that transcription factor is active and occupying those regions. Similarly, in gene expression signature enrichment, an enriched gene signature for factor A could be explained by another factor(s) that also redundantly activate factor A-target genes

(i.e. multiple factors can activate the same gene). Thus, this could lead to false positives in the enrichment process, especially if factor A is not expressed in the cell of interest and other factors generate a factor A-like signature. 58

Cell/Context Specific Transcription Factors

By extension, there may be cell-specific mechanosensitive transcription factors. As described in Chapter 1, mechanotransduction events often require an intact contractile, cytoskeletal network for force propagation, generation and signaling (e.g. RhoGTPases).

It would then be reasonable to suspect that transcription factors that interact with these machineries would be stiffness-sensitive, but this is not necessarily the case. For example, in the enrichment analyses that identified YAP/TAZ, several other transcriptional regulators (e.g. SRF/MAL, NF-kB, and TGF-β/SMAD) were not enriched despite their known dependence on cytoskeletal changes and mechanical stimuli68.

Indeed, there are cell/tissue-specific mechanosensitive transcription factors, such as

NKX2.5184, whose expression is mostly restricted to cardiac tissues (GTEx). And there are relatively ubiquitous mechanosensitive transcription factors, like the aforementioned

YAP/TAZ173 and also RARG, which controls Lamin-A expression and scales with stiffness over many human and mouse tissue171. But even these ubiquitous factors can have differential effects based on their interacting partners. For example, YAP may differentially interact with its DNA-binding partners (e.g. TEADs, TBX5, SMADs, RUNX2) in different situations to activate gene programs185–187. Thus, while many common pathways exist during mechanotransduction, different cells and different context of stimuli may utilize different transcriptional regulators.

The Role of lncRNAs as Transcriptional Regulators

As mentioned in the introduction to Chapter 2, lncRNAs do not possess coding domains as in protein coding genes, limiting our abilities to predict their function. By combining multiple layers of data from high-throughput sequencing (e.g. RNA-seq, ChIP-seq,

Chromatin-conformation-capture, CHART and ChIRP-seq, RIP-seq, etc) and

59 bioinformatics analyses, several roles of lncRNAs and mechanism of function have been identified. By and large, the primarily elucidated function of lncRNAs is to regulate gene expression, examples which will be discussed below.

Correlation analyses between lncRNAs and nearby coding genes is a commonly used approach to predict lncRNA function at a whole transcriptome level because of its early relative “success” (in quotations because how much failure is attributed to this strategy is rarely reported) in identifying lncRNA function. This “guilt by association” strategy was derived from noticing that lncRNAs and nearby protein coding genes were highly correlated, and by using known functions of nearby coding genes, the function of an unknown lncRNA could be roughly predicted126,188–190.

For example, the identification and elucidation of function for lincRNA-p21 resulted from the observation that many lncRNAs were associated to p53 regulated genes. One of those lncRNAs, lincRNA-p21, was found to be a direct target gene of p53 and subsequent experiments determined that lincRNA-p21 mediates p53-induced apoptosis by binding with hnRNP-K, a ribonucleoprotein, to repress transcription of survival genes191.

Long non-coding RNAs: Mechanisms of action

Over time, more studies identified more lncRNAs that regulated gene expression, not only of nearby genes, but also of distant genes within the same and even on other chromosomes. This section discusses some of these mechanisms.

In one mode, lncRNAs can associate with chromatin modifying complexes to repress expression of target genes. The lncRNA, HOTAIR was found to regulate gene expression at hundreds of sites by binding with PRC2, a polycomb repressor complex 60 protein189,192. XIST is another lncRNA that binds PRC2, and it is transcribed from one X- chromosome to transcriptionally silence the entire second X-chromosome in females to control gene dosage. One study also found that ~20% of expressed lncRNAs were bound to PRC2, and that silencing of select lncRNAs led to upregulation of genes known to be PRC2 targets193. Other complexes known to associate with lncRNAs include

DNMT3b and CTCF: pRNA, a lncRNA that associates with DNMT3b to mediate DNA methylation, and SRA, a lncRNA that binds CTCF to enhance its function to insulate genomic regions194,195. As discussed in Chapter 1, one study showed that cyclic stretch- mediated gene repression acts through PRC2 (in a cytoskeletal tension and emerin/LINC complex dependent manner)60. Thus, it is within reason to suspect that specific “mechanosensitive” lncRNAs may guide PRC2 to mechanosensitive coding gene loci.

As there are many lncRNAs transcribed in relative proximity to protein coding genes, often in enhancer regions and at low abundance, it has been proposed that some of these lncRNAs may represent non-functional by-products of local transcriptional processes (transcriptional noise). However, a critical early study demonstrated that lncRNA transcripts could have enhancer-like functions (enhancer RNAs). During keratinocyte differentiation, several lncRNAs were differentially expressed near coding genes that were known to be involved in differentiation. SiRNA silencing of several candidate lncRNAs were found to reduce expression of their neighboring coding genes, demonstrating that the lncRNA transcript was functional. A series of reporter constructs then demonstrated that it was specifically the lncRNA transcript that potentiated the promoter activity of its target protein coding gene196.

LncRNAs can act locally and globally. As discussed above, HOTAIR and XIST regulate expression in hundreds of sites by associating with PRC2. Some lncRNAs also remain locally anchored at their site of synthesis and enact their influence at distal genomic loci.

For example, HOXA locus genes, through chromosomal loops, are brought into the proximity of the lncRNA, HOTTIP197. Even multiple gene loci on separate chromosomes can be recruited to the lncRNA loci, as is in the case of Firre198.

Another mode of transcriptional regulation by lncRNAs is through regulation of splicing.

One mechanism was observed in NAT/Zeb2. NAT is a lncRNA that is transcribed from the anti-sense strand of the protein coding gene Zeb2. Successful splicing of Zeb2 results in a transcript that is not normally translated, however, expression of NAT, which overlaps the 5’ splice-site, prevents splicing of Zeb2 leading to a translatable product199.

Another was observed in the lncRNA, MALAT1, which regulates splicing by controlling the expression, localization and activation of SR family splicing factors200

Lastly, lncRNAs have been described to act in the cytoplasm and regulate mRNA stability and translation. For example, they have been reported to act as miRNA sponges, or competing endogenous RNAs128,201. In this mode, the lncRNA transcript may contain sequences that can compete and effectively sequester microRNAs that would otherwise degrade a protein coding transcript.

The long non-coding RNA, PACER

Part of this chapter experimentally addresses the role of the lncRNA, PACER (p50- associated COX-2 extragenic RNA), as a potential regulator of stiffness-mediated gene expression. We identified PACER in a transcriptome-wide correlation analysis as a stiffness-sensitive lncRNA whose expression is highly correlated to the nearby stiffness-

62 sensitive coding gene, PTGS2 (shown later in the Results section). Our lab and others have previously found that PTGS2 is an important player in VSMC stiffness-sensing and vascular stiffening44,217, and thus we chose to investigate PACER as a potential upstream regulator of stiffness-mediated PTGS2 expression in VSMCs.

PACER is an 800 bp, single exon, polyadenylated lncRNA located ~200 bp upstream to

PTGS2 and is transcribed in the antisense direction relative to PTGS2. As PACER is located in the known PTGS2 promoter region, its sequence overlaps several known

DNA-binding sites for transcription factors (e.g. NF-κB, AP1, NF-IL6, CREB,

TCF4/LEF1) known to functionally regulate PTGS2202,203. Nevertheless, PACER was shown not to be a transcriptional by-product at the PTGS2 promoter. In human breast epithelial cells and macrophages, PACER positively regulates PTGS2 expression by sequestering p50 NFkB from the PTGS2 promoter. P50 NFkB when homodimerized forms a repressive complex, but when heterodimerized with RelA/p65 NFkB (which contains a transactivating domain) can promote transcription203. Other lncRNAs have also been implicated in NFkB signaling, and also binding to NFkB affecting its ability to bind DNA204. Whether PACER is functional in VSMCs and regulates stiffness-mediated

PTGS2 expression is unknown and will be addressed in this chapter.

METHODS

Unbiased motif searching

We searched predicted promoter and enhancer regions (determined by chromHMM/NIH roadmaps)205,206 in human heart/aorta within the regions 5kb upstream of the transcriptional start site and 1kb downstream of genes that were upregulated by stiffness 63

(combined Ao and Co VSMCs). Enriched motifs were identified using the MEME suite.

Significant motifs were then matched to known transcription factor databases (JASPAR core vertebrates v.2014207, Uniprobe207, and Jolma207) using TOMTOM208. Motifs with known matches were then filtered, keeping matches with e-values < 1, which represents the number of false positives expected at that point in the rank of matches per motif. We further curated the remaining matched transcription factors for those with known documented roles in VSMCs.

Immunofluorescence and quantification of TBX5/YAP localization hTERT-immortalized Co VSMCs were serum starved for 48 hours then cultured on soft and stiff hydrogels for 24 hours in SmGm-2 (Lonza). Cells were fixed in 4% paraformaldehyde for 20 minutes at room temperature, permeabilized in PBS containing

0.5% Triton X-100, 0.11g/mL sucrose, and 0.61 mg/mL MgCl2. Cells were blocked with

1%BSA in PBS for 30 minutes. YAP antibody (mouse monoclonal IgG2a, sc-101199,

Santa Cruz) was used at 0.5 ug/mL (1:200 dilution) and incubated overnight at at 4ºC, in

1% BSA in PBS. TBX5 antibody (rabbit polyclonal) was obtained as a gift from Cathy

Hatcher (Philadelphia College of Osteopathic Medicine, PCOM, Philadelphia PA) and previously described207. Primary incubation for TBX5 was at 0.26 ug/mL overnight at

4ºC, in 1% BSA in PBS. Alexa-488 or Alexa-568 conjugated secondary antibodies (Life

Technologies) to mouse or rabbit were used at 1:200 at room temperature for 1 hour.

Hydrogels were mounted with ProLong® Gold antifade reagent with DAPI (nuclear stain) overnight before imaging.

An epifluorescence microscope with a 20X objective was used to image YAP, TBX5 and

DAPI signals. The area normalized intensity of TBX5 and YAP in both the nucleus and

64 cytoplasm was determined using FIJI imaging software207: nuclear area delineated by the DAPI signal, total cellular area by the boundary of each cell in the phase channel, cytoplasmic area determined as the total cell area minus the nuclear area. The nuclear/cytoplasmic ratio reported is the area-normalized intensity of the nucleus/cytoplasm per cell. Approximately 20 cells were counted per experiment per stiffness over four experiments (n = 4). Per experiment, the average nuclear/cytoplasmic ratio of cells grown on stiff substrates are normalized to ratios from soft substrates which was set to 1.

The correlation between YAP and TAZ localization determined using the nuclear/cytoplasmic ratio per cell over all experiments (n = 148). Reported is the spearman’s correlation coefficient calculated using the Hmisc library in R.

Gene Expression Validation by Quantitative RT-PCR

RNA was extracted from VSMCs using TRIzol reagent according to the manufacture’s protocol and reverse transcribed using the High-Capacity cDNA Reverse Transcription

Kit (Life Technologies). cDNA was subjected to quantitative PCR with the following primer-probe sets from Life Technologies: PTGS2 (Hs00153133_m1). To assay for

PACER expression, we obtained a Custom Plus Taqman Assay (Assay ID: AJWR2Q9,

Life Technologies). The primer-probe set used for 18S rRNA has been previously described148 . Real-time qPCR results were calculated using the ddCT method with 18S rRNA as the reference.

Assessing LncRNA Knockdown

LncRNA knockdown was performed using locked nucleic acid GapmeRs (Exiqon).

Primary and hTERT-immortalized Co VSMCs were transfected with Lipofectamine

RNAiMAX according to the manufacturer’s protocol (Life Technologies) and with

GapmeRs targeting either a negative scramble control (50nM) or PACER (50nM) for 48 hours in Opti-MEM (Life Technologies) in 6-well plates before cells were trypsinized, centrifuged, and replated on 18-mm hydrogels with SmGM2. Sequences for the

GapmeRs are as follows: Negative Control (5'-AACACGCTATAACGC-3'), PACER-ASO-

1 (5'-TTAGCGTCCCTGCAAA-3'), PACER-ASO-2 (5'-GAACTTTAAAACTCGA-3'), and

PACER-ASO-3 (5’-TGCTTAGGACCAGTAT-3’). Degree of knockdown was assessed by quantitative RT-PCR.

Western Blot for NFkB

Serum-starved (48 hours) hTERT-immortalized Co VSMC were grown on soft and stiff hydrogels for 24 hours in SmGm-2 (Lonza) before being washed in PBS and lysed using the NE-PER nuclear and cytoplasmic extraction kit (Thermo Fisher). Cells on tissue- culture plastic were treated with 25ng/µL TNF-alpha for 30 minutes as a control (Abcam, ab9642). Nuclear fractions were separated via the NuPage SDS Page system

(Invitrogen, NP0322BOX). Primary antibodies used were anti-p65 NFkB (1:500, sc-372,

Santa Cruz), anti-Lamin B (1:1000, sc-365214, Santa Cruz). HRP-conjugated anti-rabbit

(1:2500, GE Healthcare Life Sciences, NA934V and NA931V) and Luminata Crescendo

Western HRP substrate (Millipore, WBLUR0100) were used to visualize the signal.

Inhibition of NFkB with BAY-11-7085

Serum-starved (48 hours) hTERT-immortalized Co VSMC were pre-incubated with either

0.3 µM BAY-11-7085 compound (Cayman chemical, 14795) or equivalent volume of

DMSO for 30 minutes prior to trypsin treatment of cells. Cells were subsequently seeded on soft and stiff hydrogels in the presence of either 0.3 µM BAY-11-7085 or DMSO for

24 hours in SmGm-2 before RNA was collected.

Statistical Analyses

Statistical tests for each experiment are included in the figure legends. Briefly, student’s paired t-tests and two-way ANOVA were used in general for comparison between soft and stiff conditions with additional perturbations. For transcriptome-wide correlation analyses, p-values were calculated using the Hmisc library in R and the threshold for significance was Bonferroni corrected per condition.

RESULTS

Unbiased Motif Analysis Identifies TBX5 as a novel stiffness-sensitive transcription factor

Several approaches exist to identify transcriptional regulators that may contribute to the observed transcriptome response to stimuli. We first chose to use unbiased motif analysis of genes differentially expressed by stiffness. We searched for recurring motif sequences in the promoter regions of these genes that had histone marks consistent with active promoters or enhancers. These statistically recurring sequences were then matched to known transcription factor motif databases and curated for transcription factors with documented activity in VSMCs. The resulting transcription factors are listed in Table 4. Among these factors, TBX5, a t-box transcription factor, was also identified in 67

Ingenuity pathway analysis as a potential upstream regulator of the stiffness-sensitive transcriptome (fisher’s exact test, p = 2.1e-4, Appendix 2).

We further focused on TBX5 in Co VSMCs as its expression is restricted to the lungs and heart including the coronary arteries, but not the aorta, and is critical in coronary artery smooth muscle cell development207,209–213. Indeed, TBX5 expression in our RNA- seq was ~10-fold lower in Ao VSMC than in Co VSMCs (Figure 11A). Data from another

RNA-seq study in our lab also showed that TBX5 was further restricted to Co VSMCs and is not expressed in Co endothelial cells (not shown). TBX5 has also been shown to interact directly with the mechanosensitive YAP/TAZ186,214. Thus, we hypothesized that

TBX5 may be a stiffness-sensitive transcription factor in Co VSMCs that may interact with YAP to regulate stiffness-mediated gene expression.

To determine if the nuclear/cytoplasmic localization of TBX5 was stiffness-dependent, we cultured Co VSMCs on soft and stiff hydrogels and quantified TBX5 and YAP subcellular localization using immunofluorescence methods as shown by others68,170,171.

Indeed, we find that nuclear localization of TBX5 was significantly increased on stiff substrates. YAP trended toward higher nuclear localization on stiff substrates (p = 0.053,

Figure 11 B, D). YAP and TBX5 localization signals were also significantly correlated

(spearman coefficient = 0.69, Figure 11C,D), consistent with studies showing direct interaction of YAP and TBX5 to activate their gene targets186,214. YAP/TBX5 gene targets

(BIRC5 and BCL2L1) and YAP/TEAD gene targets (ANKRD1 and CTGF) were also increased on stiff substrates (Figure 11E). Taken together, these data suggest that

TBX5 is a stiffness-dependent transcriptional regulator in Co VSMCs and may interact with YAP to drive stiffness-dependent gene expression.

Table 4. VSMC-relevant transcription factor motifs enriched in stiffness-sensitive genes using MEME/TOMTOM Suite

This table includes the top motifs that were (1) enriched in predicted promoters and enhancers of stiffness-sensitive genes, (2) matched to known transcription factor motif sequences, and (3) were documented to have VSMC specific functions. Motif e-value. Statistical significance of the motif (e.g. the expected number of motifs one would expect to find in random sequences of similar size to the input search set). Match p-value. The probability that the match occurred by random chance according to the null model. Match e-value. The expected number of false positives in matches up this this point. Match q-value. The minimum false discovery rate required to include the match.

Figure 11. Stiffness-dependent localization of TBX5 and YAP.

(continued below)

Figure 11. Stiffness-dependent localization of TBX5 and YAP (continued) (A) Heatmap of the expression of TBX5 in Co and Ao VSMCs from RNA-seq (B) Ratio of nuclear intensity (area normalized) vs. cytoplasmic intensity (area normalized) of TBX5 and YAP in coronary VSMCs grown on stiff and soft polyacrylamide hydrogels. Nuclear/Cytoplasmic intensity ratios were normalized to those on soft hydrogels. Significance was determined using student’s paired t-test (* p < 0.05, $ p =0.053). Error bars are standard errors of the mean, data was generated from measuring 148 cells (87 Stiff, 61 Soft) over n = 4 different sets of soft and stiff hydrogels). (C) The nuclear/cytoplasmic ratio (area normalized) per cell (n = 148) for TBX5 and YAP are plotted on a scatterplot. TBX5 and YAP localization correlation was tested using a spearman correlation test. (D) Representative images of TBX5 and YAP immunofluorescence images of Co VSMCs used for panels (A) and (B). Scale bar = 50um (E) Expression of YAP/TBX5 and YAP/TEAD gene targets in Co VSMCs grown on soft and stiff from the RNA-seq dataset. Y-axis per subplot are DESeq2 normalized expression values. Log2 Fold change and false discovery rate is reported for each gene target.

Transcriptome-wide correlation analysis reveals coordinated expression of stiffness-sensitive lncRNA and stiffness-sensitive protein coding gene pairs

LncRNAs and their neighboring genes are often correlated in their expression, and several studies have demonstrated that some lncRNAs regulate the expression of neighboring coding genes through various mechanisms127,129,215. We asked whether the

VSMCs transcriptome-wide stiffness response may result in part by lncRNA regulation of nearby genes. Thus, we conducted transcriptome-wide correlation analyses between all lncRNA-gene pairs within the same chromosome. We found that the frequency of significantly correlated lncRNA-gene pairs increased based on the proximity of the lncRNA-gene pair (Figure 12). These significant correlations were also predominantly positive. When comparing stiffness-sensitive to insensitive lncRNAs, stiffness-sensitive lncRNAs exhibited a greater number of correlated gene-pairs than stiffness-insensitive genes (Figure 13).

The top 20 most highly correlated stiffness-sensitive lncRNA-protein coding gene pairs are presented in Table 5. The full list of significant correlations is presented in Appendix 71

6. Notably, we find that the lncRNA, RP5-973M2.2 (PACER) is highly associated with the stiffness-sensitive coding gene, PTGS2. PACER has been shown to positively regulate PTGS2 expression in several cell types203,216, and we and others have shown that PTGS2 is important in stiffness-sensing and contributes to vascular stiffening44,217.

Furthermore, several lncRNAs are significantly correlated to matrix and cytoskeletal proteins that may potentially be involved in VSMC stiffness-sensing. EDIL3 (Del-1) is an extracellular matrix protein that regulates VSMC adhesion, migration, and proliferation218. Two nuclear membrane associated proteins are also highly correlated to stiffness-sensitive lncRNAs. SYNE3 (aka Nesprin-3), a component of the LINC (linker of the nucleoskeleton and cytoskeleton) complex which transmits mechanical stress from cytoskeleton to the nucleus, may be important in the stiffness-mediated transcriptional response61,219–221. TMPO (a.k.a lamin-associated polypeptide 2-alpha, LAP2α) is a LEM- domain containing nuclear membrane protein which binds Lamin A, much like Emerin, a critical regulator of nuclear mechanotransduction64,222,223.

Figure 12. Increased lncRNA-gene pair correlation with pair proximity.

(Continued below)

Figure 12 (Continued). Correlation distributions of the top variant 1000 lncRNAs expressed Ao and Co VSMCs over different distances (A) Scatter plot of all lncRNA-gene pair correlation scores. Each dot position represents the distance between a lncRNA and nearby gene (x-axis) and the Spearman’s correlation coefficient (y-axis). All possible lncRNA-gene pairs within the same chromosome were correlation tested and all lncRNAs are centered at the 0-position on the x-axis. Color of each dot represents significance after correction for multiple-hypothesis testing (black dots are insignificant, red dots are correlations with p < 1e-5). (B) Histogram depicting the distance distribution of significantly correlated lncRNA-gene pairs. (C) Histogram depicting the distance distribution of all correlation tested lncRNA-gene pairs. (D) Normalized distributions of significantly correlated lncRNA-gene pairs. The number of significant lncRNA-gene pairs per bin (distance from lncRNA) is normalized by the number of total lncRNA-gene pairs within that bin.

Figure 13. Increased lncRNA-gene pair correlation with pair proximity and stiffness-sensitivity.

(Continued Below)

Figure 13 (Continued). Correlation distributions of all expressed Ao and Co VSMCs lncRNAs (first column), combined Ao and Co VSMCs stiffness-sensitive lncRNAs (second column), stiffness-insensitive lncRNAs (third column). (A) Scatter plot of all lncRNA-gene pair correlation scores. Each dot position represents the distance between a lncRNA and nearby gene (x-axis, within a +/- 50kb range) and the Spearman’s correlation coefficient (y-axis). All possible lncRNA-gene pairs within the same chromosome were correlation tested and all lncRNAs are centered at the 0-position on the x-axis. Color of each dot represents significance after correction for multiple-hypothesis testing (black dots are insignificant, red dots are correlations with p < 1e-5 for all lncRNAs, p < 4e-5 for stiffness-sensitive lncRNAs, p < 2e-5 for stiffness-insensitive lncRNAs). (B) Histogram depicting the distance distribution of significantly correlated lncRNA-gene pairs. (C) Histogram depicting the distance distribution of all correlation tested lncRNA-gene pairs. (D) Normalized distributions of significantly correlated lncRNA-gene pairs. The number of significant lncRNA-gene pairs per bin (distance from lncRNA) is normalized by the number of total lncRNA-gene pairs within that bin.

Top correlated stiffness-sensitive lncRNA-protein coding gene pairs

Distance Spearman Correlation LncRNA Log FC PCG Log FC 2 (kB) 2 R p-value

LINC00341 1.41 -65.7 SYNE3 1.00 0.98 4.00E-11 RP11-54O7.1 1.05 -13.4 SAMD11 1.06 0.95 1.20E-08 AC002116.7 -1.66 -40.4 WDR62 -2.17 0.95 2.70E-08 RP11-800A3.4 2.85 20.7 P2RY2 2.26 0.94 3.90E-08 TMPO-AS1 -1.97 0.9 TMPO -1.72 0.94 3.90E-08 RP11-401P9.4 2.32 97.5 NKD1 2.06 0.94 7.80E-08 RP11-386G11.10 -1.97 -61 TUBA1C -2.21 0.94 1.10E-07 CTD-2269F5.1 1.17 -0.4 EDIL3 1.64 0.93 1.90E-07 CTB-92J24.2 -1.61 12.9 ZNF726 -1.73 0.92 4.30E-07 CTD-2298J14.2 1.37 -2.7 LRFN5 1.98 0.92 4.30E-07 JHDM1D-AS1 1.66 0.2 JHDM1D 1.62 0.92 4.30E-07 RP11-303E16.2 -1.5 24.3 CENPN -1.56 0.91 8.80E-07 RP11-110I1.12 -1.03 -97.5 H2AFX -2.11 0.90 1.70E-06 RP5-973M2.2 (PACER) 1.61 0.2 PTGS2 2.23 0.90 1.70E-06 RP11-386G11.10 -1.97 -3.6 TUBA1B -2.01 0.90 2.10E-06 RP3-462E2.3 1.78 45.9 ALDH2 1.08 0.90 2.50E-06 AC009133.12 1.18 30.5 KIF22 -1.94 -0.89 3.60E-06 CASC15 1.63 71 SOX4 1.74 0.89 5.10E-06 RP1-151F17.2 1.4 2.9 ATXN1 1.08 0.89 5.10E-06

Table 5. Top correlated stiffness-sensitive lncRNA-protein coding gene pairs.

The top significantly correlated stiffness-sensitive lncRNAs-stiffness-sensitive protein coding gene pairs are listed. Protein coding genes and lncRNA genes are listed along with their log2 fold change (log2FC > 0 represents decreased expression by stiffness). The distance between gene pairs, Spearman’s correlation coefficient, and p-value are listed. The PACER-PTGS2 correlation is bolded.

Interrogation of the lncRNA, PACER as a potential regulator of nearby stiffness- sensitive protein coding genes

We further focused on the PACER-PTGS2 pair because of the importance of PTGS2 in

ECM stiffening. Studies by our lab and others have previously suggested that PTGS2 is involved in a positive feedback stiffening mechanism whereby stiffness-mediated suppression of PTGS2 results in enhanced collagen and fibronectin production in

VSMCs and fibroblasts leading to further ECM stiffening44,95. However, how matrix stiffness modules PTGS2 expression is unknown. PTGS2 is a known target gene of several transcription factors in response to several stimuli224. Recent studies in macrophages, mammary epithelial cells, and osteosarcoma cells show that PACER positively regulates PTGS2 expression via NFkB203,216, suggesting that the PACER-

PTGS2 correlation in VSMCs may underlie a functional link. NF-kB family members and pathways were also enriched as potential transcriptional regulators of the stiffness- sensitive transcriptome (GSEA Figure 16A, pathway analysis fisher’s exact test p = 2.4e-

6). Specifically, RelA/p65 activity was also predicted to be inhibited by stiffness (z-score

= 2.9, fisher’s exact test p = 2.4e-3, Appendix 2). Thus, we hypothesized that stiffness may act through PACER and NFkB to regulate stiffness-dependent PTGS2 expression and perhaps other stiffness-sensitive protein coding genes.

We first expanded our correlation analysis of PACER with stiffness-sensitive protein coding genes within 1Mb. We also found that PLA2G4A, which codes for cytosolic phospholipase A2, is also correlated to PACER expression and is stiffness-sensitive in both Ao and Co VSMCs (Figure 14A, B). Interestingly, both PLA2G4A and PTGS2, which are located directly in cis to PACER, are key enzymes in arachidonic acid metabolism and the synthesis of prostacyclin, which regulates stiffness-dependent ECM 78 production44,45. RT-PCR confirmed that increased ECM stiffness reduced both PACER and PTGS2 expression in primary Ao and Co VSMCs (Figure 14C).

To test whether PACER might coordinately regulate the expression of these nearby protein coding genes, we knocked-down and over-expressed PACER in hTERT- immortalized Co VSMCs. We used three different LNA-GapmeR ASO (anti-sense oligonucleotides) to reduce PACER expression levels grown on soft matrices to that as in stiff matrices, but this reduction in PACER did not significantly alter PTGS2 or

PLA2G4A expression (Figure 15A). Although knockdown with PACER-ASO-2 appeared to significantly increase expression of PLA2G4A, this may be an off-target effect as the other two ASOs targeting PACER did not alter PLA2G4A levels. We next used a lentiviral-based expression system (pSLIK225), to overexpress PACER. A diagram of the expression system is in Figure 15C. Consistent with the PACER-knockdown findings, overexpression of PACER in Co VSMCs both on soft and stiff matrices did not significantly change PTGS2 expression (Figure 15B). However, PLA2G4A expression was significantly higher only in VSMCs on soft substrates with PACER overexpression.

We next interrogated the role of NFkB in stiffness-mediate PACER and PTGS2 expression, as NF-kB was previously implicated in PACER-mediated regulation of

PTGS2203. Pathway analyses predict that p65/NFkB pathways are activated by soft hydrogels, consistent with the increase in PTGS2 on soft hydrogels (Figure 14C).

Indeed, the localization of p65 (RelA), the trans-activating subunit of NFkB is stiffness- dependent. However, nuclear localization of p65 is increased on stiff hydrogels compared to soft hydrogels (Figure 16B). This suggests that increased PTGS2 expression on soft hydrogels is independent of p65 NFkB activation. Furthermore, 79 inhibition of NFkB signaling using BAY-11-7085 also did not affect PTGS2 expression on soft hydrogels (Figure 16C). Taken together, these findings demonstrate that the

PACER-NFkB-PTGS2 axis, while discovered in epithelial cells and macrophages, may not be intact in VSMCs.

Figure 14. Coordinate stiffness-sensitive expression of PACER, PTGS2, and PLA2G4A

(A) UCSC Browser track view of PTGS2-PACER-PLA2G4A and their corresponding correlation in expression (from RNAseq) with Spearman’s correlation coefficient and p-value from donor- matched Ao and Co VSMCs grown on soft and stiff hydrogels. (B) PACER (RP5-973M2.2), PTGS2 and PLA2G4A RNA-seq expression from donor-matched Ao and Co VSMCs. Log2FC and FDR are reported above each gene (C) PACER and PTGS2 expression by quantitative-RT- PCR in (Left) primary Ao VSMCs (n = 5, **** p < 0.0001) and (Right) in primary Co VSMCs (n = 7, **** p < 0.0001). Two-way ANOVA, Holm-Sidak’s multiple comparison test.

Figure 15. Regulation of PTGS2 and PLA2G4A by PACER

(A) PACER, PTGS2 and PLA2G4A expression by quantitative RT-PCR after PACER knockdown using 3 different LNA-GapmeRs in hTERT-immortalized Co VSMCs cultured on soft and stiff hydrogels (n = 4, * p < 0.05, ** p < 0.005, *** p < 0.0005, **** p < 0.0001, ns not-significant, Two- way ANOVA, Holm-Sidak’s multiple comparison test). (B) PACER, PTGS2 and PLA2G4A expression by quantitative RT-PCR after lentiviral-based overexpression of PACER in hTERT-immortalized Co VSMCs grown on soft and stiff hydrogels (n = 4, **** p< 0.0001, Two-way ANOVA; Holm-Sidak's multiple comparisons test). (C) Vector map of the lentivirus-based system to overexpress PACER in VSMCs. Shown is the region between the two LTRs.

Figure 16. The role of NFkB in stiffness-sensitive PACER-PTGS2 expression.

(A) Enrichment Score Plot for NFKB pathways using GSEA using RNA-seq of Ao and Co VSMCs cultured on soft and stiff hydrogels. (Left) Enrichment for NFKB using the transcription factor motif database, C3 (Right) Enrichment for TNFalpha signaling via NFKB using the hallmark database, H. (B) Western blot analysis for p65 NFkB in the nuclear fraction of hTERT-immortalized Co VSMCs grown on soft hydrogels, stiff hydrogels or tissue-culture plastic. TNF-alpha was added at 83

25ng/mL for 30 minutes prior to protein collection. Lamin B was used as an internal load control. Boxed in lanes represent different nitrocellulose membranes. (C) RT-qPCR expression of PACER and PTGS2 from hTERT-immortalized Co VSMCs cultured on soft and stiff hydrogels incubated with 0.3uM BAY-11-7085 for 24 hours. Student’s t-test was performed (n =5, *** p < 0.005, ns not-significant)

DISCUSSION

In this chapter, we used several different bioinformatics approaches to identify TBX5 and

PACER as putative transcriptional regulators of matrix stiffness in VSMCs.

TBX5

Our results suggest that TBX5, as it is not expressed in Ao VSMCs, may be a Co

VSMC-specific stiffness-sensitive transcription factor. We demonstrated that TBX5 nuclear localization in Co VSMCs was increased by stiffness and highly correlated to

YAP localization, a previously identified TBX5-binding partner186,214. We also showed that known YAP-TBX5 gene targets were up-regulated by stiffness in Co VSMCs, consistent with the localization data. These findings only suggest a potential role for

TBX5. Further experiments in manipulating of TBX5 expression and TBX5-ChIP-qPCR to verify promoter occupancy will be needed.

Several studies suggest potential mechanisms for how stiffness may localize TBX5. In a series of chick model studies, TBX5 was found to bind directly to LMP4, a PDZ-LIM domain protein that binds directly to F-actin. LMP4 has been shown to repress TBX5 activity, and in its absence, TBX5 becomes predominantly nuclear226–228. This suggests stiffness-mediated TBX5 localization could be due to changes in the actin cytoskeleton.

Stiffness-mediated TBX5 localization in our human Co VSMCs may also be related to changes in VSMC differentiation state. In the developing chick heart, TBX5 localization is spatially and temporally dynamic. In undifferentiated epicardial cells (the precursor to

84 coronary VSMCs that proliferate and migrate), TBX5 is predominantly nuclear. In differentiated epicardial cells (which express VSMC differentiation markers like calponin), TBX5 was both nuclear and cytoplasmic226. And in developed chick coronary artery sections, TBX5 was reportedly exclusively cytoplasmic228. Perhaps, human Co

VSMCs in part become more synthetic/dedifferentiated consistent with their increased proliferative and migratory capacity on stiff matrices (shown in Chapter 4, and prior literature81,89).

We further suspect that TBX5 may regulate stiffness-mediated cell responses such as proliferation/survival and migration in part through partnering with YAP, a key regulator of growth control. In humans, mutations in TBX5 result in Holt-Oram Syndrome, which is characterized by aplastic upper-limb malformations, congenital heart abnormalities, and coronary artery defects213,229,230. TBX5 mutations known to cause Holt-Oram Syndrome lead to reduced YAP/TBX5 association and reduced TBX5 transcriptional activity in cardiomyocytes214. Furthermore, a TBX5-YAP-β-catenin complex, but not TEAD-YAP, drives expression of BIRC5 and BCL2L1, though this study was done in cancer cells186.

We find that stiffness increases expression of BIRC5 and BCL2L1 consistent with increased TBX5-YAP localization, but we also see increased TEAD-YAP targets,

ANKRD1 and CTGF. Whether stiffness requires TBX5 and YAP in VSMCs to induce

BIRC5/BCL2L1 expression will require further studies. Evidence in mice suggests that

TBX5 may control VSMC proliferation and migration. TBX5-knockout mice exhibit poor coronary vasculogenesis, resulting from reduced epicardical cell proliferation and migration into the myocardium212. As TBX5 has been predominantly interrogated in either developing chicks and mice, or cancer cell models, whether TBX5 in mature human Co VSMCs still control these processes will require further work.

Transcriptome-wide Correlation

In this study, we found that SS lncRNAs were significantly correlated with more of their nearby genes when compared to stiffness-insensitive lncRNAs, suggesting that SS lncRNAs may coordinate the stiffness-mediate protein coding response. We identified several highly correlated pairs that were within 100kB of each other listed in Table 5.

Many of the protein coding genes in this list code for nucleoskeletal structures, and they, along with their correlated lncRNA, could play a role in mechanotransduction. Only two of the lncRNAs in this list have been previously studied, CASC15 and RP5-973M2.2

(PACER), and both have been shown to regulate coding gene expression231. However, some of these lncRNA-coding gene pairs may be associated due to stiffness-driven local chromosome changes, whereby clusters of genes are simultaneously co-expressed.

Thus, some of these lncRNAs may be non-functional and transcriptional noise, while others may represent true cis-regulatory functional lncRNAs.

PACER

We hypothesized that PACER may regulate two coding genes, PTGS2 and PLA2G4A.

This was based on the highly correlated lncRNA-coding gene expression patterns, common coding gene function in arachidonic acid/prostacyclin metabolism, and prior reports of a cis-regulatory role of PACER on PTGS2 in three different human cell types

(mammary epithelial, macrophages, osteosarcoma lines)203,216. However, it appears that this PACER-PTGS2 axis may not be present in human VSMCs and/or during stiffness- sensing.

This may be due to our findings that stiffness-mediated regulation of PTGS2 is NFkB- independent, while PACER regulation is NFkB-dependent. We observed that PTGS2

86 expression was upregulated on soft matrices, and this effect was not reversed by inhibition of NFkB with BAY-11-7085 (BAY-11-7085 is an inhibitor of IκBα- phosphorylation, which when phosphorylated releases NFkB to become active).

Furthermore, the transactivating p65 NFkB subunit was predominantly nuclear on stiff substrates, as seen in other studies70,232,233. Given these findings, it makes sense that manipulation of PACER, and thus NFkB, would not have an effect on stiffness-regulated

PTGS2 expression.

Nevertheless, PACER may still have a stiffness-dependent and NFkB-dependent role by its previously reported role in sequestering p50 NFkB. As nuclear p65 was increased by stiffness, perhaps the simultaneous reduction in PACER may lead to less sequestered p50, and thus influence the p50/p65 NFkB balance. It is possible that this leads to global expression changes in stiffness-dependent NFkB gene targets (identified in our enrichment analyses) because PACER, unlike several locally acting lncRNAs, is not restricted to its site of transcription (~25% of its transcripts are cytoplasmic in macrophages)203.

However, we also acknowledge that PACER’s function may be cell-type specific and non-functional in VSMCs, despite it being expressed and stiffness-dependent. This concept was demonstrated in a library-based lncRNA knockdown study over seven different cell lines which identified lncRNAs that were ubiquitously expressed, but functional in only 1 of 7 cell types127. There is also the possibility that the PACER-PTGS2 axis may be present in primary VSMCs (where the correlation analyses identified

PACER-PTGS2) and absent in hTERT-immortalized coronary VSMCs (where all the pathway interrogation experiments were conducted). We think this is unlikely as the 87 expression of PACER and PTGS2 are stiffness-dependent in both primary and hTERT- immortalized VSMCs. Thus, an RNA-seq following manipulation of PACER on soft and stiff conditions will determine whether PACER has no function or have global transcriptional consequences.

Then how does stiffness-regulate PTGS2? Many other transcription factor binding sites

(e.g. AP1, NF-IL6, CREB, TCF4/LEF1) are upstream (-2kb) of PTGS2 and overlap

PACER203,234, and several of these factors have been shown to regulate PTGS2 expression in human VSMCs in response to various stimuli. Our enrichment analyses for potential upstream regulators of the stiffness-sensitive transcriptome also found CREB1

(Ingenuity, fisher’s exact test p = 2.4e-11). However, more likely, at the time of this writing, a study showed that YAP/TAZ and TEAD mediated stiffness-repression of

PTGS2 in pulmonary artery VSMCs96. They showed that both YAP and TAZ were sufficient and necessary for stiffness-dependent PTGS2 repression, and that this effect was TEAD-dependent (using TEAD-binding mutants of YAP and TAZ). Furthermore, stiffness-regulated YAP/TAZ localization is Rho-dependent, consistent with work in our lab showing that the stiffness-mediated PTGS2 response is also dependent on Rho activity94.

CHAPTER 4: MALAT1, a lncRNA regulator of stiffness-sensitive VSMC functions

Introduction

This chapter focuses on MALAT1, a lncRNA that was chosen as a candidate through an unbiased analysis of the stiffness-sensitive transcriptome in VSMCs. How MALAT1 was determined to be the focus of this chapter will be discussed in the results section, but this introduction serves to get the reader acquainted with the current knowledge about

MALAT1, setting the context for the hypothesis generated.

Discovery of MALAT1

MALAT1 (metastasis-associated lung adenocarcinoma transcript-1, a.k.a. NEAT2, nuclear-enriched abundant transcript-2) was discovered in 2003, from a subtractive hybridization screen between primary human non-small cell lung cancer tumors that had metastasized vs. non-metastasized. MALAT1 was the most over-represented clone in this screen, and high MALAT1 expression was highly predictive for overall survival in lung adenocarcinomas, thus leading to its name235. Several years later, the mouse ortholog to MALAT1 was identified in mouse hepatocellular carcinoma236, and many reports have linked increased MALAT1 to a range of different human cancers: breast, renal cell, cervical, colorectal and bladder cancer 237.

MALAT1 is a highly abundant, nuclear-retained transcript (hence its other name,

NEAT2) that is ubiquitously expressed in many human tissues and highly conserved in mammals235,238. MALAT1 is a large transcript ~ 8000 bp long and can be processed into a long (~7000 bp), nuclear-retained segment (which is often referred to as the MALAT1 transcript) and a short (61 bp), cytoplasmic segment (which is referred to as mascRNA,

MALAT1-associated small cytoplasmic RNA)239. MALAT1 has a relatively long half-life (>

12 hrs) due to a triple helical structure on its 3’ end, compared to the ascRNA with half- life < 6 hrs. Little is known about the role of mascRNA, and most studies to date have focused predominantly on MALAT1.

Function of MALAT1: Proliferation, Migration, and Splicing

The role of MALAT1 appears to be complex and perhaps cell and context-dependent.

Key studies regarding the role of MALAT1 will be discussed below.

Within the nucleus, MALAT1, through its two localization motifs, associate specifically with nuclear speckles, which are sites enriched with mRNA processing and splicing factors240. In HeLa cells, MALAT1 was found to regulate splicing programs and cell cycle, and in MALAT1 knockdown HeLa cells, SR splicing factors (SRSF1 and SRSF3) localization and activation were reduced leading to altered splicing in hundreds of genes200. Furthermore, in response to serum-stimulation, MALAT1 relocates and associates demethylated-pc2 (polycomb 2 protein, distinct from PRC2 mentioned in prior chapters) with growth control genes within the nuclear speckle. This pc2-MALAT1 association promotes cell cycle progression by activating E2F1 and transcription of its targets241, and knockdown of MALAT1 results in aberrant cell cycle200,241.

In contrast, one study used siRNA to knockdown MALAT1 in 10 different cell lines

(mouse and human) and found proliferation defects in some cells lines but not others242.

In zinc finger nuclease MALAT1 knockout A549s (human lung cancer cell), no difference in splicing compared to wild-type was found. No differential splicing was seen in the same candidate genes as those identified as differentially spliced in HeLa cells, nor in other selected candidates243. Instead, they found that MALAT1-knockout A549 cells had altered gene expression in metastasis-associated genes, resulting in functional defects

90 in migration and metastasis. A role for MALAT1 in migration has also been described in

HUVECs, but the knockdown of MALAT1 resulted in increased migration along with reduced proliferation244. Lastly, these MALAT1-regulated proliferation, apoptosis, and migration effects have been observed in many different cancers as well245. Thus,

MALAT1 appears to be critical for several cell functions in a cell specific manner.

MALAT1 as a competitive endogenous RNA

In addition to RNA processing within the nuclear speckle, MALAT1 has been proposed to act as a competitive endogenous RNA (ceRNA), as several microRNA seed sequences have been found in the long MALAT1 transcript. In human renal cell carcinoma, MALAT1 and miR205 interact to control E-cadherin expression246. In mouse myoblasts, MALAT1 coordinates with miR133 (to control SRF expression) and miR181a

(to control MYOD activity) for myogenic differentiation247,248. Furthermore, MALAT1 has been shown to complex with Ago2249–251. These studies also suggest that MALAT1 may not just act as a “dummy” diversion/sponge. That is, the repression in MALAT1 by a diverted microRNA may contribute to cell phenotype as significantly as the concomitant increase in that microRNA’s mRNA targets.

MALAT1 knockout Mice

Given that significant phenotypes have been observed in proliferation, migration, splicing and differentiation, the in vivo role of MALAT1 was highly anticipated. At almost the same time, three separate labs generated their own MALAT1 knockout mice and surprisingly found no physiologic defects252–254. MALAT1 knockout mice breeding normal, and their offspring follow expected Mendelian ratios. There were no histopathological abnormalities assessed over 12 different organs. Staining for Ki67

(proliferation) and caspases (apoptosis) were normal. Mouse embryonic fibroblasts 91 isolated from these knockouts had no altered nuclear speckle SR protein distribution or phosphorylation. No differential splicing was observed either in brain and liver tissues.

HeLa cells were analyzed in parallel, and the aforementioned MALAT1-dependent HeLa phenotypes were reproduced252–254. Thus, it appeared that MALAT1 was dispensible for normal development and physiology in mice.

Role for MALAT1 during pathology

The finding that MALAT1 had no in vivo developmental phenotype was striking in comparison to the numerous in vitro and cancer observations. It was hypothesized that perhaps MALAT1 was important during pathological conditions, partially reconciling these seemingly discrepant observations. Indeed, MALAT1 had a phenotypic contribution in insult mouse models.

In a cardiotoxin-induced skeletal muscle injury model, MALAT1-knockout mice had accelerated muscle regeneration after injury by increasing MYOD activity and myogenic differentiation255. When MALAT1 was knocked-out in a MMTV-PyMT (mouse mammary tumor virus) mouse model, tumors were more cystic and differentiated compared to the solid and poorly defined tumors when MALAT1 was expressed. However, the number of tumors was not affected. The increased tumor differentiation was also associated with significantly fewer macro- and micro-metastasic lesions256. However, the loss of

MALAT1 did not appear to be important in pressure-overload hypertrophic cardiomyopathy257

In the vasculature, and as relevant to this chapter, MALAT1-knockdown using antisense oligonucleotides (ASO) in mice had reduced reperfusion after hind limb ischemia injury compared to control ASO. MALAT1 knockout mice also had statistically reduced

92 neonatal retinal vascularization, but the effect size was small. These finding were attributed to reduced proliferation and increased migration from MALAT1-knockdown in human umbilical vein endothelial cells244.

These studies in the MALAT1 knockout mice support the notion that the presence of

MALAT1 is functionally significant on pathologic backgrounds in vivo. How these insults permit a role for MALAT1 is unknown, but we hypothesize that pathologic ECM stiffening plays a role in the function of MALAT1. Indeed, many tumors, especially mammary tumors, and hypoxic/ischemic-insults are characterized by increased ECM stiffening and post-injury fibrosis258–262. Furthermore, as MALAT1 and stiffness have been independently found to regulate proliferation and migration, they may be coordinating with one another to do so. Lastly, MALAT1 appears to have cell type-dependent effects, and while many cell types have been interrogated, the role of MALAT1 in VSMCs is unknown. Thus, this chapter focuses on how MALAT1 was chosen as a candidate gene representing the VSMC stiffness-sensitive response, and the role of MALAT1 in VSMCs and matrix stiffness.

METHODS

Identification of MALAT1 as a putative regulator of stiffness-dependent functions using Ingenuity Pathway Analysis

We used Ingenuity pathway analysis (QIAGEN) to identify lncRNAs that may underlie the predicted functions from gene set enrichment analyses (Figure 7) because it relies on knowledge/literature-based annotations of lncRNA function. As described in Chapter

2, the differential gene expression lists generated by DESeq2 using combined Ao and

Co VSMC expression values were uploaded into Ingenuity Pathway Analysis (QIAGEN).

An FDR < 1e-7 and an absolute LogFC > 1 were used to select analysis-ready molecules. Top enriched “categories” and “functions” were assessed. We identified lncRNAs attributed to each significant category or function. These enriched lncRNAs were then ranked based on the number of occurrences within the significant category or functions. Only 3 lncRNAs were annotated in the ingenuity knowledge base for the enriched categories, and only 2 lncRNAs were annotated to enriched functions.

Gene Expression Validation by Quantitative RT-PCR

RNA was extracted from VSMCs using TRIzol reagent according to the manufacture’s protocol and reverse transcribed using the High-Capacity cDNA Reverse Transcription

Kit (Life Technologies). cDNA was subjected to quantitative PCR with the following primer-probe sets from Life Technologies: MALAT1 (Hs00273907_s1), CCND1

(Hs00765553_m1) and CCNA1 (Hs00171105_m1). The primer-probe set used for 18S rRNA has been previously described148 . Real-time qPCR results were calculated using the ddCT method with 18S rRNA as the reference.

Assessing LncRNA Knockdown

LncRNA knockdown was performed using locked nucleic acid GapmeRs (Exiqon).

Primary and hTERT-immortalized Co VSMCs were transfected with Lipofectamine

RNAiMAX according to the manufacturer’s protocol (Life Technologies) and with

GapmeRs targeting either a negative scramble control (50nM) or MALAT1 (50nM) for 48 hours in Opti-MEM (Life Technologies) in 6-well plates before cells were trypsinized, centrifuged, and replated on 18-mm hydrogels with SmGM2. Sequences for the

GapmeRs are as follows: Negative Control (5'-AACACGCTATAACGC-3'), MALAT1-

ASO-1 (5'-CGTTAACTAGGCTTTA-3'), MALAT1-ASO-2 (5'-ACTAGCGTGTGGAAAG-

3'). Degree of knockdown was assessed by quantitative RT-PCR.

EdU Incorporation Assay

Following knockdown with GapmeRs, primary or hTERT-immortalized Co VSMCs were plated at 10,000 cells/cm2 on 12-mm soft and stiff hydrogels in SmGM2 for 48 hours with

10µM EdU (Life Technologies). Cells were fixed in 4% Paraformaldehyde overnight. EdU was visualized using the Click-iT EdU Imaging Kit (Life Technologies). Nuclei were stained with DAPI. Three fields of view were counted per hydrogel to determine the percent of EdU-positive cells relative to DAPI-stained cells. Relative EdU incorporation was determined by normalizing all conditions to the percent EdU of the control-GapmeR treated cells grown on soft hydrogels.

Migration Assay

For single cell migration studies, following GapmeR knockdown, Co VSMCs were plated at 5,000 cell/cm2 on 12-mm soft and stiff hydrogels overnight. NucBlue (Life

Technologies) was then added (40µL per mL) with fresh SmGM2 30 minutes prior to the start of the migration study. Nuclei were then imaged using fluorescence every 5-10 minutes for 5 hours. Three fields of view were captured per timepoint per condition.

Nuclei were tracked using TrackMate (ImageJ). Nuclei that were not in the field of view for at least 4 hours were not counted. Median cell velocities, total distance (path- dependent), and total displacement (path-independent distance) were calculated.

Relative values for each parameter were determined by normalizing to the respective values on soft hydrogels with negative control GapmeR knockdown. 95

For the scratch wound assay, following GapmeR knockdown, hTERT-immortalized Co

VSMCs were plated on 12-well plates in SmGM2 at 50,000 cells/cm2 overnight to achieve confluence. A uniform scratch was created using a 200µL pipet tip, cell debris was washed, and fresh SmGM2 was replaced. The wound closure was assessed by microscopy.

RESULTS

Identification of MALAT1 as a regulator of stiffness-dependent phenotypes

Most lncRNAs have no known function, and none have been implicated in the response to ECM stiffening. To identify lncRNAs that may play a role in stiffness-mediated cellular responses, we used unbiased enrichment analyses (Ingenuity pathway analysis) and identified several highly enriched cellular processes in both Ao and Co VSMCs (Figure

7C): cell cycle, growth, death, motility, morphology, and RNA post-transcriptional modifications. Among the ingenuity pathway categories and functions, MALAT1 was the most overrepresented lncRNA annotated to these functions (Table 6, Table 7). MALAT1, a highly conserved lncRNA, has been primarily implicated in cancer biology, and, more recently, in endothelial cell function. Whether MALAT1 plays a role in VSMCs or in stiffness-dependent functions is unknown.

RT-PCR validated the RNA-seq finding that matrix stiffening reduces MALAT1 expression in both Ao and Co VSMCs (Figure 17A, Figure 18A). We then focused on using primary and hTERT-immortalized Co VSMCs. Knockdown of MALAT1 with two different LNA-GapmeRs resulted in significant reduction in MALAT1 expression (Figure

17B, Figure 18B). Upon MALAT1 knockdown, Co VSMCs lost their elongated cell

96 morphology (Figure 17C, Figure 18C). MALAT1 knockdown also reduced stiffness- dependent S-phase entry in Co VSMCs measured by EdU incorporation (Figure 17D,

Figure 18D, Figure 19A) consistent with reduced cyclin D1 and cyclin A2 expression levels (Figure 19B). Reduced MALAT1 also decreased stiffness-induced Co VSMCs migration speed, total distance, and displacement (Figure 17E, Figure 18E), and slowed collective migration on a wound closure assay on stiff tissue culture plastic (Figure 19C).

Taken together, these data show that MALAT1 is a positive regulator of VSMC proliferation and migration in response to ECM stiffness. ECM stiffness negatively regulates MALAT1, and this may represent a homeostatic negative feedback mechanism in which the reduction in MALAT1 limits stiffness-induced proliferation and migration.

Ingenuity Pathway Predicted p-value # Molecules MALAT1 NEAT1 DLEU2 Categories Cell Cycle 3.63E-27 538 Cellular Growth and Proliferation 4.22E-26 910 X Cancer 4.01E-24 2057 X X X Organismal Injury and Abnormalities 4.01E-24 2099 X X X Cellular Assembly and Organization 4.48E-23 592 X X DNA Replication, Recombin., Repair 4.48E-23 454 Cell Death and Survival 3.86E-22 873 X Organismal Survival 2.88E-18 608 X X Cellular Development 6.54E-18 806 X Post-Translational Modification 1.54E-16 80 Protein Synthesis 1.54E-16 91 Cellular Movement 2.71E-12 575 X Cellular Function and Maintenance 2.78E-12 431 Cell Morphology 2.34E-11 528 Tissue Development 1.40E-10 483 Cardiovascular Sys Dev. and Func. 3.20E-10 331 Organismal Development 6.58E-10 698 Tumor Morphology 5.50E-09 203 Nucleic Acid Metabolism 2.04E-08 135 Small Molecule Biochemistry 1.01E-07 113 Tissue Morphology 1.78E-07 479 Embryonic Development 2.17E-07 412 Organ Development 2.17E-07 196 Organ Morphology 2.17E-07 153 Skeletal and Muscular Disorders 1.03E-06 115 X RNA Post-Transcriptional Modification 1.47E-06 78 X Cardiovascular Disease 1.54E-06 416

Table 6. Identification of lncRNA regulators of stiffness-dependent cell responses (Predicted Categorical Pathways).

Top categorical pathways identified in Ingenuity Pathway Analysis. Enrichment p-value and number of molecules for each category are listed. An (X) denotes that the corresponding lncRNA is represented within that canonical pathway. Only three lncRNAs were represented for the top enriched pathways listed (i.e. no other lncRNAs were annotated to these pathways in the ingenuity database).

Ingenuity Pathway Predicted Functions p-value MALAT1 NEAT1 Proliferation of cells 4.22E-26 X Necrosis 3.86E-22 X Cell death 1.72E-21 X Apoptosis 2.34E-20 X Organismal death 2.88E-18 X X Morbidity or mortality 3.94E-18 X X Metastasis 1.63E-13 X Cell movement 2.71E-12 X Migration of cells 2.25E-11 X Invasion of cells 3.13E-10 X Smooth muscle tumor 1.06E-06 X Processing of RNA 1.47E-06 X Cell transformation 2.7E-06 X Organization of organelle 9.24E-06 X Formation of nucleus 1.21E-05 X Metastasis of cells 3.29E-05 X Splicing of RNA 3.55E-05 X Leiomyomatosis 0.00012 X

Table 7. Identification of lncRNA regulators of stiffness-dependent cell responses (Predicted Functions).

Top predicted functions identified in Ingenuity Pathway Analysis that contained lncRNAs annotated to those functions. Enrichment p-value and for each category are listed. An (X) denotes that the corresponding lncRNA is represented within that predicted function. MALAT1 and NEAT1 are the only lncRNAs linked to these cellular functions in the ingenuity database.

Figure 17. MALAT1, a lncRNA regulator of stiffness-dependent cell functions in primary VSMCs.

(A) Left, MALAT1 RNA-seq expression from donor-matched Ao and Co VSMCs. Right, MALAT1 expression validation by quantitative-PCR in primary Co and Ao VSMCs (n = 5 Co VSMCs, n =4 Ao VSMCs, **, p < 0.005). (B) MALAT1 knockdown using LNA-GapmeRs in primary Co VSMCs. (C) Phase image of primary Co VSMCs cultured on tissue culture plastic for 48 hrs in SmGm2 after MALAT1 knockdown. Scale bar = 100um. (D) Primary Co VSMCs cultured on soft and stiff hydrogels for 48 hours with serum containing EdU. Values are normalized to non-targeting control LNA-GapmeR on soft hydrogels (n = 4, ** p<0.005; ****p<0.0001). (E) Normalized median velocity, total distance (path-dependent) migrated, and total displacement (path-independent) migrated in primary Co VSMCs over 5 hours. Values are normalized to velocity and distance in non-targeting control LNA-GapmeR on soft hydrogels. (n = 4, * p<0.05, **p<0.005, Two-way ANOVA; Holm-Sidak's multiple comparisons test).

100

Figure 18. MALAT1, a lncRNA regulator of stiffness-dependent cell functions in hTERT- immortalized Co VSMCs.

(A) MALAT1 expression validation by quantitative-PCR in primary Co and Ao VSMCs (n = 7, **, p < 0.005). (B) MALAT1 knockdown using LNA-GapmeRs in hTERT-Co VSMCs. (C) Phase image of hTERT-Co VSMCs cultured on tissue culture plastic for 48 hours in SmGm-2 after MALAT1 knockdown. Scale bar = 100um. (D) hTERT-Co VSMCs cultured on soft and stiff hydrogels for 48 hours with serum containing EdU. Values are normalized to non-targeting control LNA-GapmeR on soft hydrogels (n = 5, * p<0.05; ** p<0.005). (E) Normalized median velocity, total distance (path-dependent), and total displacement (path-independent) migrated in hTERT-VSMCs over 5 hours. Values are normalized to velocity and distance in non-targeting control LNA-GapmeR on soft hydrogels. (n=2-4, * p<0.05, Two-way ANOVA; Holm-Sidak's multiple comparisons test).

101

Figure 19. MALAT1 knockdown inhibits EdU incorporation, cyclin expression and wound closure in Co VSMCs on tissue-culture plastic (TCPS).

(A) Primary and hTERT-immortalized Co VSMCs grown on tissue culture plastic with serum containing EdU for 48 hours following LNA-GapmeR knockdown of MALAT1 (n = 4, ** p<0.005, *** p<0.0005). (B) RT-PCR expression of CCND1 and CCNA1 in hTERT-immortalized Co VSMCs after knockdown of MALAT1. Student’s t-test for LNA-control v. LNA-MALAT1-ASO1. (n = 4, *** p<0.0005). No statistical test performed for LNA-MALAT1-ASO-2 (n=1). (C) Phase images of the closure of the scratch wound in hTERT-immortalized Co VSMCs at 20 hours post-scratch.

102

DISCUSSION

Analysis of stiffness-enriched pathways and gene sets for stiffness-sensitive lncRNAs allowed us to identify MALAT1 as a lncRNA regulator of VSMC stiffness-dependent proliferation and migration. To our knowledge, this is the first study implicating a role for

MALAT1 in VSMCs and its interplay with matrix stiffness. MALAT1 had previously been linked to tumor aggressiveness in in vitro and clinical studies235,246,256,263–267. However, its importance in vascular disease is relatively unknown, with only one study implicating

MALAT1 in endothelial cell function244. While we found that MALAT1-dependent proliferation is similar in both VSMCs and endothelial cells, MALAT1 knockdown enhances migration in endothelial cells while reducing migration in VSMCs.

Despite many in vitro and in vivo tumor model studies linking MALAT1 to critical cell functions, whole-body MALAT1 knockout mice develop normally252–254. We hypothesized that MALAT1 may regulate proliferation and migration during pathologic/abnormal stiffening but is dispensable during physiologic/normal conditions. In support of this stiffness-mediated dependence on MALAT1 hypothesis, our data show that MALAT1 is necessary in VSMC proliferation and migration on stiff matrices, but is less important under physiologically soft matrices. Most in vitro tumor cell studies interrogating MALAT1 have been conducted on culture-plastic dishes (orders of magnitude stiffer than tissue).

Whether this MALAT1-dependence would still be observed in these tumor cell types on softer/physiologic stiffness is unknown. As the tumor microenvironment is stiffer than normal tissue, and the increased ECM stiffness directly contributes to tumor aggressiveness261,262,268–270, perhaps this pathologic stiffening may underlie tumor growth and invasion dependence on MALAT1256,263,264,271. Similarly, the dependence of MALAT1

103 in revascularization after ischemic injury may result from post-injury fibrosis244, while normal vasculature development in MALAT1-knockout mice is unaffected. Thus, vascular stiffening may lead to an aberrant dependence on MALAT1 in VSMCs.

We also find that MALAT1 expression is reduced by ECM stiffness in VSMCs and perhaps acts in a negative feedback mechanism to limit stiffness-mediated proliferation and migration. However, this effect may not necessarily be universal. RT-PCR validated the stiffness-mediated reduction in MALAT1 in the donor-matched Ao and Co VSMCs used for the RNAseq. We also validated this finding in several different donors of Ao and

Co VSMCs, but not in all donors. We believe there may be a level of variability in

VSMCs from specific donors that exhibit stiffness-mediated MALAT1 reduction while others do not. This potential variability may reflect differences in the risk for cardiovascular disease from vascular stiffness in various populations272,273.

104

CHAPTER 5: DISCUSSION

This dissertation was motivated by arterial matrix stiffness, an important clinical parameter for disease, but also an important mechanical property dictating cellular behavior. We broadly addressed the question of how matrix stiffness regulated the transcriptional programs of VSMCs, an important population in responding to and promoting matrix stiffness. Many of the findings from this dissertation generated more hypotheses than it answered, as is often the case in data-mining, bioinformatic approaches. However, we had several a priori goals, including, understanding the relative contributions of VSMC embryonic origin and matrix stiffness in regulating the transcriptome and identifying novel transcriptional regulators of stiffness-sensing by looking at lncRNAs and transcription factors. Working through these initial goals led us to analyze transcriptome-wide splicing patterns and experimentally interrogate several candidates such as MALAT1, PACER and TBX5. This chapter will put into context the findings of the prior chapters in terms of their contributions to basic mechanotransduction biology, clinical implications, and in each case, address important future directions.

Basic Mechanotransduction

From a basic cellular mechanotransduction viewpoint, this dissertation provides insight into stiffness-sensitive transcriptome characteristics likely applicable to many cellular context, not just VSMCs. We found that stiffness-sensitive genes identified in VSMCs are sequence conserved and tend to be expressed across multiple tissues. Thus, the stiffness-sensitive transcriptome response, as a whole, may represent a ubiquitous response, but whether this is necessarily true will require surveying the stiffness- sensitive response in multiple, different cell types. It would be interesting to do this 105 analysis in part by integrating the several publicly available datasets of stiffness- modulated gene expression. Of course, this has its limitations such as batch and platform-bias (mostly microarray, few RNA-seq) effects which will need to be accounted for, but this combined analysis may help identify ubiquitous vs. cell-specific stiffness- sensing transcriptional programs.

This is also the first study to broadly characterize, at a whole transcriptome level, the role of stiffness in two classes of transcriptional complexity: alternative splicing and lncRNAs. Only one study has previously described stiffness-mediated splicing in one gene73, and no study to our knowledge has implicated lncRNAs in the stiffness- response. We found that stiffness-dependent lncRNA expression and alternative splicing are cell-specific, more so than protein coding gene responses. Whether these unique lncRNAs and splicing events confer different phenotypes in Ao and Co VSMCs will require further work. We also identified differentially spliced genes (in both Ao and

Co VSMCs) that may be important in mechanotransduction as they are annotated in cytoskeletal processes, but more interestingly, we found that many directly splice out their SH3/SH2 domains due to stiffness. The implications of this were discussed in

Chapter 2.

Through bioinformatic approaches in Chapter 3, we also identified TBX5 which may be a

Co VSMC-specific, stiffness-sensitive transcription factor. Indeed, much more will need to be studied to validate this hypothesis. A ChIP-qPCR or ChIP-seq for TBX5 will be necessary to demonstrate differential promoter binding due to stiffness. Knockdown and overexpression of TBX5 will also be necessary to elucidate the functional consequences of TBX5 in Co VSMCs and whether documented-TBX5 target genes are indeed

106 activated by stiffness through TBX5. Additionally, direct binding studies between TBX5 and YAP in Co VSMCs will further strengthen the role of TBX5 in stiffness-sensing.

Lastly, we focused on lncRNAs that were stiffness-sensitive in both in Ao and Co

VSMCs: MALAT1 and PACER.

We found that stiffness represses MALAT1, and that MALAT1 was necessary for stiffness-mediated proliferation and migration. Thus, MALAT1 may exist in a feedback loop to limit stiffness-responses. However, we suspect that MALAT1 may also contribute to the observed stiffness-mediated alternative splicing events. In some cells (e.g. HeLa),

MALAT1 alters SR splicing factor expression and activity200. Stiffness has been shown to phosphorylate and activate SR splicing factors 73. Furthermore, in our VSMC data, we found that the expression of many splicing factors including some SR factors were stiffness-sensitive, which led to the splicing term enrichment described in Chapter 2.

Thus, to determine whether MALAT1 coordinates stiffness-mediated splicing in VSMCs, a survey of splicing events and splicing factor machinery while modulating MALAT1 in

VSMC will be required.

MALAT1 may also regulate epigenetic processes relevant to stiffness-sensing.

Epigenetic modifications in the form of histone modifications were not interrogated in this dissertation, but are mechanosensitive, especially in response to cyclic stretch60.

MALAT1 has been shown to affect histone modifications and activity of transcriptional machinery at growth gene loci241,255. Perhaps MALAT1 may regulate VSMC proliferation and migration through such mechanisms. Lastly, MALAT1 has been implicated as a competitive endogenous RNA, titrating the effects of microRNAs, some of which target known mechanosensitive transcription factors like SRF. Recent evidence also suggest

107 that microRNAs dictate stiffness-mediated responses and may even encode cellular memory of exposure to substrate stiffness274,275. It would be interesting if MALAT1 acts as a central regulator of mechanotransduction.

This dissertation also highlights a previously described PACER-PTGS2 axis that is intact in several cell types, but is absent from VSMCs during pathologic stiffening. Although

PACER did not regulate PTGS2 or PLA2G4A, PACER may still regulate other genes through its interaction with NFkB, and act as a stimulus-specific lncRNA. An RNA-seq of

VSMCs with overexpression and knockdown of PACER will be required to determine if

PACER has any transcriptome-wide consequence in VSMCs or during stiffness-sensing.

If not, PACER’s function may be cell-type-specific. Indeed, lncRNAs may be expressed in a cell type, while having no transcriptional regulatory role in that cell127. PACER-

PTGS2 was identified in a transcriptome-wide correlation analysis, in which we detected many highly correlated stiffness-sensitive lncRNA-protein coding gene pairs. How many of these are functionally linked versus pure associations will also require further investigation. However, the overall observation that stiffness-sensitive lncRNAs are more correlated to surrounding genes than stiffness-insensitive genes suggest that some may be functionally linked. Our correlation analyses also show that stiffness-sensitive genes are often near to other stiffness-sensitive genes, suggesting that regions of chromosome, rather than individual genes, may be stiffness-sensitive.

Clinical/Translational Implications

Altered cellular functions due to pathologic vascular matrix stiffness manifest as a clinical predictor and risk factor for cardiovascular disease. Unlike other risk factors such as lipids, blood glucose, and blood pressure, monitoring arterial stiffness is not yet a routine clinical parameter. This is likely due to the relative absence and difficulty of developing 108 therapies directly targeting arterial stiffness. However, treatment of dyslipidemia, hyperglycemia, and hypertension are often associated with reductions in arterial stiffness. For example, angiotensin-2 signaling inhibitors (for hypertension) partially act through anti-fibrotic effects, and AGE (advanced glycation end products)-crosslink breakers remove non-enzymatic crosslinks due to hyperglycemia276,277. Directly treating matrix stiffness is also difficult because it is a critical mechanical property for normal vascular function, and in some cases act as a necessary adaptive process to withstand changes in hemodynamic load. Unfortunately, systemic inhibition of matrix crosslinkers like lysyl oxidase, which is critical in many tissues, can lead to adverse effects like vascular dissection and rupture278,279. One option is targeting cellular responses to stiffness, such as increased Rho-kinase activity. This is the basis for therapies like fasudil, a Rho-kinase inhibitor, which has shown promise in pulmonary artery hypertension and atherosclerosis280,281. While the side effect profile of fasudil is fairly low, inhibition of Rho-kinase activity has been associated with systemic effects such as hypotension282,283. Thus, an ideal therapeutic target for arterial stiffness would be one that is dispensable during physiologic elasticity but mediates the effects of pathologic stiffening, or one that is cell/tissue-type specific to minimize unintended consequences in other tissues.

Our interrogation of lncRNAs, especially MALAT1, in the stiffness response may contribute toward this goal. It appears that MALAT1 is only a critical player during pathologic stiffness, and pharmaceutical inhibition of MALAT1 may have limited off- target effects in non-diseased vasculature. This is supported by the finding that MALAT1 is dispensable for normal development in mice and becomes functionally relevant during several injury models. Our findings in this dissertation support a scenario where

109 pharmaceutical inhibition of MALAT1 could be used to specifically reduce pathologic stiffness-driven VSMC proliferation and migration. To test this scenario, inhibition of

MALAT1, either by MALAT1-knockout or infusion of MALAT1 antisense oligonucleotides during in vivo models of stiffening, such as neointimal formation after wire-injury, or an atherosclerosis model, would be needed. It must be noted that targeting MALAT1 may also lead to inhibition of beneficial regeneration processes such as reperfusion after ischemia244. Furthermore, while dispensable during development and early age, it is not known whether MALAT1-knockout mice exhibit abnormalities during later adulthood.

Mouse arteries also progressively stiffen with age and whether MALAT1-knockout mice exhibit age-related stiffening will be interesting. Translationally, this is important as the incidence of arterial stiffening sharply rises at older age.

We also identified several variants that were significant in GWAS for vascular diseases.

These variants were in proximity to FES and MIA3, which were both spliced by stiffness in Ao and Co VSMCs. Understanding how the splicing of these two genes and their function in the stiffness-response will be a new paradigm for basic mechanotransduction biology, but it also may provide insight into why these two genes may be highly associated with human vascular diseases. Although the variants discussed in Chapter 2 were not splicing-quantitative trait loci for FES or MIA3, a deeper analysis of the loci may identify other variants within the same haplotype-block that may alter splice-site strength or use of an alternative promoter and influence inclusion of a SH2/SH3 domain. If causal variants exist for FES and MIA3, this could be clinically important for risk stratification and developing therapeutics such as exon-skipping oligonucleotides.

Lastly, we identified several origin-specific VSMC responses that could be tissue-specific therapeutic targets. Stiffness-driven differential splicing was highly VSMC-origin specific. 110

As mentioned above, these specific splicing patterns may be amenable to exon-skipping therapies targeting either Ao or Co VSMCs, but their role in pathogenesis will need to be studied first. TBX5 is also a Co VSMC-specific transcription factor (although it is also expressed in the heart, it is not expressed in other VSMCs). It may be a potential therapeutic target with fewer off-target effects than say targeting YAP or SRF, which are more ubiquitous. Several TBX5 animal models and human mutations in TBX5 have provided detailed understanding of its requirement in development and cardiac differentiation, but its role in adulthood is unknown. Inhibiting TBX5 during adulthood could lead to cardiac dysfunction if it is required for maintenance. Thus, much about

TBX5 and whether it has a pathologic role in stiffening and coronary artery disease needs to be established.

In summary, this dissertation integrated an engineered-stiffness culture platform with next-generation sequencing and bioinformatics to reveal novel transcriptional properties in the Ao and Co VSMC stiffness response, especially in lncRNA and splicing. We explored several candidate stiffness-sensitive components and identified potential novel mechanotranduction mechanisms. Looking forward, new classes of transcriptomic/epigenomic properties and characteristics are being identified, and biological data are being generated at exponential rates due to high-throughput technology advances. Analyzing and processing these data into digestible information at both the systems and gene-specific level will be necessary for meaningful insight and better predictions about gene/biological function. Subsequent reductionist approaches are still necessary to validate predictions and establish causality, and also feedback to improve system level prediction methods.

111

APPENDIX

Appendix 1. Table of Cardiovascular Disease Traits used in GWAS analyses in Chapter 2

Trait Name 1 Abdominal aortic aneurysm 2 Aortic root size 3 Aortic stiffness 4 Aortic-valve calcification 5 Arterial stiffness 6 Blood pressure 7 Cardiovascular disease risk factors 8 Cardiovascular heart disease in diabetics 9 Carotid intima media thickness 10 Coronary arterial lesions in patients with Kawasaki disease 11 Coronary artery calcification 12 Coronary artery disease 13 Coronary artery disease or ischemic stroke 14 Coronary artery disease or large artery stroke 15 Coronary heart disease 16 Coronary restenosis 17 Coronary spasm 18 Diastolic blood pressure 19 Heart failure 20 Hypertension 21 Hypertension (young onset) 22 Intracranial aneurysm 23 Large artery stroke 24 Major CVD 25 Myocardial infarction 26 Myocardial infarction (early onset) 27 Peripheral artery disease 28 Pulmonary arterial hypertension (without BMPR2 mutations) 29 Stroke 30 Stroke (ischemic) 31 Stroke (pediatric) 32 Subclinical atherosclerosis traits (other) 33 Systolic blood pressure 34 Systolic blood pressure (alcohol consumption interaction) 35 Thoracic aortic aneurysms and dissections

112

Appendix 2. Ingenuity Pathway Analysis Enriched Upstream “Transcriptional Regulators”

P-values are determined by Fisher’s Exact Test. Regulators mentioned in the text are bolded. Only FDR<1% are shown.

Upstream p-value of BRCA1 2.47E-08 1.95E-07 Transcriptional FDR overlap JUN 3.39E-08 2.59E-07 Regulator TP53 1.43E-47 3.72E-45 SMAD7 5.20E-08 3.86E-07 E2F4 4.53E-47 5.89E-45 GATA1 7.83E-08 5.66E-07 E2F1 6.88E-39 5.96E-37 NKX2-3 9.79E-08 6.88E-07 CCND1 1.47E-33 9.56E-32 FOXO3 1.60E-07 1.09E-06 MYC 2.03E-33 1.06E-31 EP300 2.57E-07 1.71E-06 NUPR1 2.73E-27 1.18E-25 NFYB 2.69E-07 1.72E-06 CDKN2A 2.10E-26 7.80E-25 FOXO1 2.72E-07 1.72E-06 E2F3 2.00E-23 6.50E-22 NEUROG1 4.19E-07 2.59E-06 TBX2 5.07E-22 1.46E-20 NPAT 6.03E-07 3.65E-06 RB1 2.40E-21 6.24E-20 TP63 6.42E-07 3.79E-06 MITF 5.05E-18 1.12E-16 SP1 6.75E-07 3.90E-06 YY1 5.18E-18 1.12E-16 HIF1A 9.15E-07 5.17E-06 FOXM1 1.96E-16 3.92E-15 TAL1 9.36E-07 5.18E-06 E2F2 1.16E-14 2.15E-13 CCNE1 9.87E-07 5.24E-06 MAX 8.09E-13 1.40E-11 SIN3B 9.87E-07 5.24E-06 E2F6 2.51E-12 4.08E-11 HDAC2 1.34E-06 6.97E-06 SMARCB1 2.86E-12 4.37E-11 FOS 2.02E-06 1.03E-05 CREB1 2.40E-11 3.47E-10 HOXA10 3.23E-06 1.58E-05 E2F7 3.42E-11 4.68E-10 PAX3 3.23E-06 1.58E-05 MYCN 5.35E-11 6.96E-10 MYBL2 3.49E-06 1.68E-05 RBL1 9.31E-11 1.15E-09 CBX4 4.92E-06 2.33E-05 KDM5B 1.19E-10 1.41E-09 TCF3 8.43E-06 3.91E-05 YAP1 2.57E-10 2.91E-09 CUX1 9.69E-06 4.42E-05 NFKBIA 3.74E-10 4.05E-09 FOXP3 1.31E-05 5.87E-05 TP73 4.29E-10 4.46E-09 WT1 1.54E-05 6.79E-05 SMARCA4 5.86E-10 5.86E-09 HTT 2.55E-05 1.10E-04 HSF1 7.23E-10 6.96E-09 MEOX2 2.59E-05 1.10E-04 HNF4A 8.76E-10 8.13E-09 APBB1 5.46E-05 2.29E-04 CTNNB1 9.21E-10 8.26E-09 ATF6 5.89E-05 2.43E-04 CEBPB 2.86E-09 2.48E-08 SMARCA2 6.00E-05 2.44E-04 HDAC1 1.06E-08 8.89E-08 ATN1 7.22E-05 2.89E-04 E2F8 1.26E-08 1.02E-07 MEF2D 7.60E-05 2.99E-04 113

NFYA 9.13E-05 3.54E-04 KLF4 2.02E-03 4.91E-03 H2AFX 1.17E-04 4.41E-04 IRX1 2.16E-03 5.13E-03 SMARCE1 1.17E-04 4.41E-04 PURB 2.16E-03 5.13E-03 GLI1 1.19E-04 4.42E-04 ONECUT1 2.17E-03 5.13E-03 NOTCH1 1.29E-04 4.72E-04 PDX1 2.23E-03 5.22E-03 CREBBP 1.38E-04 4.98E-04 SNAI2 2.25E-03 5.22E-03 ARNT 1.43E-04 5.09E-04 SIN3A 2.30E-03 5.29E-03 STAT6 1.56E-04 5.48E-04 RELA 2.43E-03 5.54E-03 TFEB 1.64E-04 5.69E-04 TFDP1 2.49E-03 5.63E-03 MED1 1.78E-04 6.01E-04 SMAD4 2.56E-03 5.74E-03 TWIST1 1.78E-04 6.01E-04 MYB 2.67E-03 5.93E-03 TBX5 2.09E-04 6.97E-04 HOXA9 2.75E-03 5.95E-03 REL 2.57E-04 8.46E-04 KLF2 2.76E-03 5.95E-03 AIP 2.76E-04 8.96E-04 STAT3 2.76E-03 5.95E-03 UXT 2.79E-04 8.96E-04 RUNX1 2.77E-03 5.95E-03 E2F5 2.96E-04 9.39E-04 FLI1 2.89E-03 6.16E-03 SATB1 3.90E-04 1.22E-03 PAX7 3.06E-03 6.47E-03 SQSTM1 4.09E-04 1.27E-03 NFE2L2 3.31E-03 6.94E-03 VHL 4.84E-04 1.48E-03 ID2 3.40E-03 7.04E-03 ATXN1 5.06E-04 1.53E-03 FOSL1 3.41E-03 7.04E-03 ZNF217 6.07E-04 1.81E-03 ID4 3.55E-03 7.27E-03 SRSF2 6.63E-04 1.96E-03 NOTCH3 4.46E-03 9.06E-03 RAD21 6.79E-04 1.98E-03 MYOCD 4.66E-03 9.39E-03 ERG 7.63E-04 2.20E-03 SMAD3 4.90E-03 9.80E-03 CBX3 8.99E-04 2.57E-03 CEBPA 1.05E-03 2.94E-03 EPAS1 1.05E-03 2.94E-03 HIC1 1.19E-03 3.25E-03 KLF6 1.19E-03 3.25E-03 TEAD2 1.20E-03 3.25E-03 PML 1.24E-03 3.32E-03 TFDP2 1.25E-03 3.32E-03 ID3 1.30E-03 3.41E-03 EZH2 1.37E-03 3.56E-03 CDKN2C 1.44E-03 3.67E-03 MEF2C 1.44E-03 3.67E-03 FHL2 1.48E-03 3.74E-03 MXI1 1.50E-03 3.75E-03 MNT 1.56E-03 3.86E-03 HLX 1.58E-03 3.88E-03 114

STIFF Enriched SOFT Enriched

Coronary Aortic Coronary Aortic RIBOSOME RIBOSOME ECM_RECEPTOR_INTERACTION LYSOSOME SPLICEOSOME SYSTEMIC_LUPUS_ERYTHEMATOSUS LYSOSOME ECM_RECEPTOR_INTERACTION SYSTEMIC_LUPUS_ERYTHEMATOSUS SPLICEOSOME FOCAL_ADHESION FOCAL_ADHESION HUNTINGTONS_DISEASE DNA_REPLICATION CYTOKINE_CYTOKINE_RECEPTOR_INTERACTION COMPLEMENT_AND_COAGULATION_CASCADES PARKINSONS_DISEASE CELL_CYCLE TGF_BETA_SIGNALING_PATHWAY PATHWAYS_IN_CANCER PYRIMIDINE_METABOLISM PYRIMIDINE_METABOLISM PATHWAYS_IN_CANCER GLYCOSAMINOGLYCAN_DEGRADATION CELL_CYCLE PURINE_METABOLISM GLYCOSAMINOGLYCAN_DEGRADATION ABC_TRANSPORTERS OXIDATIVE_PHOSPHORYLATION HOMOLOGOUS_RECOMBINATION COMPLEMENT_AND_COAGULATION_CASCADES TGF_BETA_SIGNALING_PATHWAY DNA_REPLICATION HUNTINGTONS_DISEASE NEUROACTIVE_LIGAND_RECEPTOR_INTERACTION PROTEASOME PURINE_METABOLISM MISMATCH_REPAIR CELL_ADHESION_MOLECULES_CAMS TOLL_LIKE_RECEPTOR_SIGNALING_PATHWAY EPITHELIAL_CELL_SIGNALING_IN_HELICOBACTER ALZHEIMERS_DISEASE PARKINSONS_DISEASE BASAL_CELL_CARCINOMA _PYLORI_INFECTION AMINOACYL_TRNA_BIOSYNTHESIS RNA_POLYMERASE DILATED_CARDIOMYOPATHY HYPERTROPHIC_CARDIOMYOPATHY_HCM RNA_POLYMERASE BASE_EXCISION_REPAIR ABC_TRANSPORTERS INOSITOL_PHOSPHATE_METABOLISM HOMOLOGOUS_RECOMBINATION NUCLEOTIDE_EXCISION_REPAIR MAPK_SIGNALING_PATHWAY LEISHMANIA_INFECTION ARRHYTHMOGENIC_RIGHT_VENTRICULAR_CARDI OOCYTE_MEIOSIS OXIDATIVE_PHOSPHORYLATION DILATED_CARDIOMYOPATHY OMYOPATHY_ARVC NUCLEOTIDE_EXCISION_REPAIR OOCYTE_MEIOSIS HYPERTROPHIC_CARDIOMYOPATHY_HCM PHOSPHATIDYLINOSITOL_SIGNALING_SYSTEM MISMATCH_REPAIR AMINOACYL_TRNA_BIOSYNTHESIS WNT_SIGNALING_PATHWAY PPAR_SIGNALING_PATHWAY GLYCOSAMINOGLYCAN_BIOSYNTHESIS_CHONDR GLYCOLYSIS_GLUCONEOGENESIS RNA_DEGRADATION MAPK_SIGNALING_PATHWAY OITIN_SULFATE ARRHYTHMOGENIC_RIGHT_VENTRICULAR_CARDI CITRATE_CYCLE_TCA_CYCLE P53_SIGNALING_PATHWAY HEDGEHOG_SIGNALING_PATHWAY OMYOPATHY_ARVC PROGESTERONE_MEDIATED_OOCYT ALZHEIMERS_DISEASE JAK_STAT_SIGNALING_PATHWAY WNT_SIGNALING_PATHWAY E_MATURATION PROGESTERONE_MEDIATED_OOCYT EPITHELIAL_CELL_SIGNALING_IN_HELICOBACTER BASE_EXCISION_REPAIR CELL_ADHESION_MOLECULES_CAMS E_MATURATION _PYLORI_INFECTION RNA_DEGRADATION GLYCOLYSIS_GLUCONEOGENESIS PPAR_SIGNALING_PATHWAY NEUROACTIVE_LIGAND_RECEPTOR_INTERACTION PATHOGENIC_ESCHERICHIA_COLI_IN PATHOGENIC_ESCHERICHIA_COLI_IN CYTOKINE_CYTOKINE_RECEPTOR_INTERACTION FECTION FECTION CYSTEINE_AND_METHIONINE_METAB UBIQUITIN_MEDIATED_PROTEOLYSIS JAK_STAT_SIGNALING_PATHWAY OLISM PROTEASOME ONE_CARBON_POOL_BY_FOLATE GRAFT_VERSUS_HOST_DISEASE

Appendix 3. KEGG enriched pathways.

Enrichment performed in Gene Set Enrichment Analysis (GSEA, Broad Institute). Listed are the top 25 KEGG Terms per VSMC cell type. Enrichment of terms are segregated by Stiff v. Soft based on the rank-order gene list for GSEA. FDR < 1% for all terms.

115

STIFF Enriched STIFF Enriched (Continued) Coronary Aortic Coronary Aortic 3_UTR_MEDIATED_TRANSLATIONAL_REGULATION 3_UTR_MEDIATED_TRANSLATIONAL_REGULATION MRNA_SPLICING MRNA_SPLICING ACTIVATION_OF_THE_PRE_REPLICATIVE_COMPLEX ACTIVATION_OF_ATR_IN_RESPONSE_TO_REPLICATION_STRESS MRNA_SPLICING_MINOR_PATHWAY MRNA_SPLICING_MINOR_PATHWAY AMYLOIDS ACTIVATION_OF_THE_PRE_REPLICATIVE_COMPLEX PACKAGING_OF_TELOMERE_ENDS PACKAGING_OF_TELOMERE_ENDS CELL_CYCLE AMYLOIDS PEPTIDE_CHAIN_ELONGATION PEPTIDE_CHAIN_ELONGATION CELL_CYCLE_CHECKPOINTS CELL_CYCLE PROCESSING_OF_CAPPED_INTRON_CONTAINING_PRE_MRNA PROCESSING_OF_CAPPED_INTRON_CONTAINING_PRE_MRNA CELL_CYCLE_MITOTIC CELL_CYCLE_MITOTIC RESPIRATORY_ELECTRON_TRANSPORT RNA_POL_I_PROMOTER_OPENING CHROMOSOME_MAINTENANCE CHROMOSOME_MAINTENANCE RNA_POL_I_PROMOTER_OPENING RNA_POL_I_RNA_POL_III_AND_MITOCHONDRIAL_TRANSCRIPTION DNA_REPAIR DNA_REPAIR RNA_POL_I_RNA_POL_III_AND_MITOCHONDRIAL_TRANSCRIPTION RNA_POL_I_TRANSCRIPTION DNA_REPLICATION DNA_REPLICATION RNA_POL_I_TRANSCRIPTION RNA_POL_II_TRANSCRIPTION DNA_STRAND_ELONGATION DNA_STRAND_ELONGATION S_PHASE S_PHASE EXTENSION_OF_TELOMERES EXTENSION_OF_TELOMERES TCA_CYCLE_AND_RESPIRATORY_ELECTRON_TRANSPORT SYNTHESIS_OF_DNA G1_S_TRANSITION G1_S_SPECIFIC_TRANSCRIPTION TELOMERE_MAINTENANCE TELOMERE_MAINTENANCE G2_M_CHECKPOINTS G1_S_TRANSITION TRANSCRIPTION TRANSCRIPTION HIV_INFECTION G2_M_CHECKPOINTS TRANSLATION TRANSLATION HIV_LIFE_CYCLE HIV_LIFE_CYCLE RESPIRATORY_ELECTRON_TRANSPORT_ATP_SYNTHESIS_BY_CHEMI INFLUENZA_LIFE_CYCLE INFLUENZA_LIFE_CYCLE OSMOTIC_COUPLING_AND_HEAT_PRODUCTION_BY_UNCOUPLING_ TRANSPORT_OF_MATURE_TRANSCRIPT_TO_CYTOPLASM LATE_PHASE_OF_HIV_LIFE_CYCLE LAGGING_STRAND_SYNTHESIS NONSENSE_MEDIATED_DECAY_ENHANCED_BY_THE_EXON_JUNCTI NONSENSE_MEDIATED_DECAY_ENHANCED_BY_THE_EXON_JUNCTI MEIOSIS LATE_PHASE_OF_HIV_LIFE_CYCLE ON_COMPLEX ON_COMPLEX MEIOTIC_RECOMBINATION MEIOSIS FORMATION_OF_THE_TERNARY_COMPLEX_AND_SUBSEQUENTLY_T FORMATION_OF_THE_TERNARY_COMPLEX_AND_SUBSEQUENTLY_T MEIOTIC_SYNAPSIS MEIOTIC_RECOMBINATION HE_43S_COMPLEX HE_43S_COMPLEX METABOLISM_OF_MRNA MEIOTIC_SYNAPSIS ACTIVATION_OF_THE_MRNA_UPON_BINDING_OF_THE_CAP_BINDI ACTIVATION_OF_THE_MRNA_UPON_BINDING_OF_THE_CAP_BINDI METABOLISM_OF_NON_CODING_RNA METABOLISM_OF_MRNA NG_COMPLEX_AND_EIFS_AND_SUBSEQUENT_BINDING_TO_43S NG_COMPLEX_AND_EIFS_AND_SUBSEQUENT_BINDING_TO_43S METABOLISM_OF_PROTEINS METABOLISM_OF_NON_CODING_RNA DEPOSITION_OF_NEW_CENPA_CONTAINING_NUCLEOSOMES_AT_T DEPOSITION_OF_NEW_CENPA_CONTAINING_NUCLEOSOMES_AT_ METABOLISM_OF_RNA METABOLISM_OF_PROTEINS HE_CENTROMERE THE_CENTROMERE MITOCHONDRIAL_PROTEIN_IMPORT METABOLISM_OF_RNA SRP_DEPENDENT_COTRANSLATIONAL_PROTEIN_TARGETING_TO_M SRP_DEPENDENT_COTRANSLATIONAL_PROTEIN_TARGETING_TO_ MITOTIC_G1_G1_S_PHASES MITOTIC_G1_G1_S_PHASES EMBRANE MEMBRANE MITOTIC_M_M_G1_PHASES MITOTIC_M_M_G1_PHASES MITOTIC_PROMETAPHASE MITOTIC_PROMETAPHASE INFLUENZA_VIRAL_RNA_TRANSCRIPTION_AND_REPLICATION INFLUENZA_VIRAL_RNA_TRANSCRIPTION_AND_REPLICATION MRNA_PROCESSING MRNA_PROCESSING Appendix 4. REACTOME Enriched Pathways (STIFF)

Enrichment performed in Gene Set Enrichment Analysis (GSEA, Broad Institute). Listed are the top Reactome Terms per VSMC cell type. Enrichment of terms for STIFF based on the rank-order gene list for GSEA. FDR < 1% for all terms.

116

SOFT Enriched Coronary Aortic CHONDROITIN_SULFATE_DERMATAN_SULFATE_METAB OLISM CHONDROITIN_SULFATE_DERMATAN_SULFATE_METABOLISM COLLAGEN_FORMATION COLLAGEN_FORMATION DEVELOPMENTAL_BIOLOGY DIABETES_PATHWAYS EXTRACELLULAR_MATRIX_ORGANIZATION EXTRACELLULAR_MATRIX_ORGANIZATION GENERIC_TRANSCRIPTION_PATHWAY GENERIC_TRANSCRIPTION_PATHWAY GLYCOSAMINOGLYCAN_METABOLISM GLYCOSAMINOGLYCAN_METABOLISM INTEGRIN_CELL_SURFACE_INTERACTIONS HEPARAN_SULFATE_HEPARIN_HS_GAG_METABOLISM METABOLISM_OF_LIPIDS_AND_LIPOPROTEINS INNATE_IMMUNE_SYSTEM NCAM1_INTERACTIONS INTEGRIN_CELL_SURFACE_INTERACTIONS TRANSMEMBRANE_TRANSPORT_OF_SMALL_MOLECUL METABOLISM_OF_LIPIDS_AND_LIPOPROTEINS TRANSPORT_OF_GLUCOSE_AND_OTHER_SUGARS_BILE NCAM1_INTERACTIONS _SALTS_AND_ORGANIC_ACIDS_METAL_IONS_AND_AM TRANSMEMBRANE_TRANSPORT_OF_SMALL_MOLECULES Appendix 5. REACTOME Enriched Pathways (SOFT)

Enrichment performed in Gene Set Enrichment Analysis (GSEA, Broad Institute). Listed are the top Reactome Terms per VSMC cell type. Enrichment of terms for SOFT based on the rank-order gene list for GSEA. FDR < 1% for all terms.

117

Dist Corr p- log2FC FDR log2FC FDR LncRNA PCG Chr R (kB) value PCG PCG lncRNA lncRNA LINC00341 SYNE3 14 65.7 0.979 3.98E-11 1.00 3.00E-08 1.41 3.97E-11 RP11-54O7.1 SAMD11 1 13.4 0.953 1.21E-08 1.06 2.81E-07 1.05 4.44E-04 AC002116.7 WDR62 19 40.4 0.947 2.72E-08 -2.17 6.48E-41 -1.66 5.07E-04 RP11-800A3.4 P2RY2 11 20.7 0.944 3.93E-08 2.26 2.63E-18 2.85 1.47E-16 TMPO-AS1 TMPO 12 0.9 0.944 3.93E-08 -1.72 7.01E-29 -1.97 1.08E-23 RP11-401P9.4 NKD1 16 97.5 0.938 7.80E-08 2.06 3.23E-08 2.32 3.47E-12 RP11-386G11.10 TUBA1C 12 61.0 0.935 1.07E-07 -2.21 1.40E-36 -1.97 4.06E-21 CTD-2269F5.1 EDIL3 5 0.4 0.929 1.94E-07 1.64 4.33E-50 1.17 4.82E-14 CTB-92J24.2 ZNF726 19 12.9 0.921 4.32E-07 -1.73 7.33E-14 -1.61 3.62E-06 CTD-2298J14.2 LRFN5 14 2.7 0.921 4.32E-07 1.98 8.04E-09 1.37 1.14E-05 JHDM1D-AS1 JHDM1D 7 0.2 0.921 4.32E-07 1.62 8.98E-23 1.66 4.01E-18 RP11-303E16.2 CENPN 16 24.3 0.912 8.82E-07 -1.56 6.61E-37 -1.50 2.28E-22 RP11-110I1.12 H2AFX 11 97.5 0.903 1.68E-06 -2.11 3.29E-48 -1.03 4.93E-07 RP5-973M2.2 PTGS2 1 0.2 0.903 1.68E-06 2.23 3.17E-13 1.61 1.28E-11 RP11-386G11.10 TUBA1B 12 3.6 0.900 2.05E-06 -2.01 4.47E-08 -1.97 4.06E-21 RP3-462E2.3 ALDH2 12 45.9 0.897 2.49E-06 1.08 2.21E-20 1.78 6.89E-07 AC009133.12 KIF22 16 30.5 -0.891 3.62E-06 -1.94 2.70E-54 1.18 2.85E-05 CASC15 SOX4 6 71.0 0.885 5.15E-06 1.74 8.08E-08 1.63 2.35E-14 RP1-151F17.2 ATXN1 6 2.9 0.885 5.15E-06 1.08 6.09E-23 1.40 3.78E-07 PRKG1-AS1 DKK1 10 0.2 0.879 7.19E-06 -1.21 5.46E-14 -1.03 5.91E-04 CTD-2267D19.6 CDC6 17 15.6 0.876 8.44E-06 -1.84 1.61E-23 -1.51 1.09E-08 RP11-800A3.4 P2RY6 11 25.5 0.876 8.44E-06 1.17 1.68E-06 2.85 1.47E-16 MCM3AP-AS1 C21orf58 21 94.7 0.874 9.87E-06 -1.27 1.84E-18 -1.12 2.80E-04 RP4-575N6.4 S1PR1 1 0.4 0.874 9.87E-06 -1.20 1.75E-16 -1.41 1.27E-04 RP11-1114A5.4 ARHGEF6 X 65.7 0.871 1.15E-05 1.01 4.76E-11 1.88 3.39E-07 AC005618.6 PCDHGA1 5 4.5 0.865 1.54E-05 1.75 2.83E-13 1.60 1.55E-16 VPS9D1-AS1 SPATA33 16 54.1 0.862 1.78E-05 -1.01 8.16E-18 -1.93 1.56E-07 RP11-443B20.1 CENPO 2 32.5 0.859 2.05E-05 -1.71 1.07E-68 -1.34 2.40E-12 AC005618.6 PCDHGA6 5 47.9 0.856 2.34E-05 1.22 2.34E-10 1.60 1.55E-16 RP11-134L10.1 TMEM140 7 20.3 0.853 2.68E-05 2.11 2.61E-26 2.16 6.02E-08 RP11-161H23.5 TUBA1C 12 84.6 0.853 2.68E-05 -2.21 1.40E-36 -2.12 1.06E-08 RP11-1081L13.4 TMEM86A 11 13.4 0.847 3.42E-05 2.05 8.67E-20 1.87 4.69E-03 RP11-366L20.3 RP11-366L20.2 12 66.3 0.847 3.47E-05 1.96 5.48E-08 1.76 2.36E-04 AC005618.6 PCDHGB3 5 44.1 0.841 4.44E-05 1.26 4.72E-10 1.60 1.55E-16 RP11-161H23.5 C1QL4 12 63.9 0.841 4.44E-05 -1.72 3.18E-04 -2.12 1.06E-08 RP11-253M7.1 KIF23 15 0.8 0.841 4.44E-05 -2.64 7.25E-40 -1.76 1.92E-10 RP1-86C11.7 HIST1H2AH 6 23.9 -0.841 4.44E-05 -2.50 6.19E-36 1.84 4.11E-06 RP11-90B22.1 NRP1 10 5.4 0.834 5.91E-05 1.16 2.83E-12 1.91 2.42E-03 AC005618.6 PCDHGB4 5 61.7 0.832 6.33E-05 1.14 1.27E-14 1.60 1.55E-16 AC005618.6 PCDHGB7 5 91.7 0.829 7.09E-05 1.43 2.83E-13 1.60 1.55E-16 RP1-86C11.7 HIST1H2BJ 6 9.5 -0.826 7.92E-05 -2.03 3.09E-48 1.84 4.11E-06 AC005618.6 PCDHGB1 5 24.1 0.824 8.84E-05 1.61 4.46E-11 1.60 1.55E-16 DNAJC3-AS1 DNAJC3 13 0.2 0.824 8.84E-05 1.09 1.98E-25 1.03 2.49E-05 RP1-86C11.7 HIST1H2AG 6 9.8 -0.824 8.84E-05 -2.48 3.22E-34 1.84 4.11E-06 CTD-2534I21.8 KIF18B 17 0.2 0.821 9.84E-05 -2.49 2.47E-40 -1.89 9.91E-06 RP1-43E13.2 MRTO4 1 41.0 -0.821 9.84E-05 -1.29 7.84E-42 1.29 2.97E-04 RP1-151F17.1 ATXN1 6 0.4 0.818 1.09E-04 1.08 6.09E-23 1.09 5.37E-05 AC009133.12 SEZ6L2 16 78.3 0.815 1.21E-04 1.26 8.59E-13 1.18 2.85E-05 RP11-34D15.2 CWF19L1 10 78.7 -0.813 1.27E-04 -1.02 5.80E-27 2.11 1.20E-03 AC005618.6 PCDHGA10 5 87.0 0.809 1.48E-04 1.18 4.15E-10 1.60 1.55E-16 RP11-243M5.1 GPR133 12 23.7 0.806 1.63E-04 -1.66 8.86E-17 -1.76 2.05E-03 RP11-855A2.5 KPNA2 17 39.1 -0.804 1.77E-04 -2.19 8.03E-55 2.47 1.28E-06 RP11-123O22.1 RP11-180C1.1 4 32.6 0.793 2.46E-04 -2.21 5.15E-18 -1.88 7.55E-06 RP1-86C11.7 HIST1H4I 6 16.1 -0.791 2.62E-04 -2.64 3.58E-35 1.84 4.11E-06 RP11-632K5.3 P4HA3 11 0.3 0.785 3.14E-04 1.34 1.81E-08 3.31 5.62E-10 ACVR2B-AS1 ACVR2B 3 1.0 0.782 3.41E-04 1.35 2.95E-12 1.63 4.68E-05 AC007743.1 CCDC85A 2 1.6 0.779 3.72E-04 -1.35 9.24E-13 -1.14 4.10E-05 AC005618.6 PCDHGB2 5 33.9 0.776 4.04E-04 1.25 4.98E-09 1.60 1.55E-16 RP4-694A7.4 DEPDC1 1 0.5 0.776 4.15E-04 -2.56 3.15E-32 -1.85 2.37E-05 AC009133.12 PRRT2 16 9.4 0.774 4.39E-04 1.07 1.81E-04 1.18 2.85E-05 GAS5-AS1 SERPINC1 1 54.1 0.771 4.77E-04 1.01 3.42E-03 1.41 1.77E-04 LINC01059 GRAMD1B 11 71.2 0.771 4.77E-04 1.27 6.68E-10 1.18 1.57E-05 RP1-86C11.7 HIST1H2BK 6 23.6 -0.771 4.77E-04 -1.33 3.43E-25 1.84 4.11E-06 RP11-92C4.6 COL15A1 9 0.6 0.767 5.30E-04 2.47 5.58E-35 1.37 3.88E-03 RP11-135L13.4 RNF112 17 75.0 0.765 5.60E-04 2.13 4.57E-10 1.06 4.30E-03 118

RP11-490M8.1 CRIM1 2 0.8 0.762 6.05E-04 -1.01 3.82E-19 -1.58 1.20E-13 OSER1-AS1 JPH2 20 23.4 -0.756 7.06E-04 -1.62 2.68E-05 1.08 2.38E-10 GAS5-AS1 CENPL 1 38.5 -0.753 7.61E-04 -1.64 1.36E-26 1.41 1.77E-04 RP11-161H23.5 TROAP 12 49.9 0.753 7.61E-04 -2.91 1.49E-61 -2.12 1.06E-08 RP11-690G19.3 PLXDC1 17 99.9 0.753 7.61E-04 1.16 8.25E-03 1.56 3.34E-20 ANKRD10-IT1 ANKRD10 13 15.9 0.744 9.48E-04 1.53 1.42E-15 1.38 9.71E-05

Appendix 6. Correlated Stiffness-sensitive lncRNA-Protein Coding Gene Pairs

Spearman’s correlation (R) for stiffness-sensitive lncRNA and protein coding genes (PCG) on the same chromosome (Chr). The distance (Dist) between their annotated gene start coordinates is shown in kilobases. Log2FC for LncRNA and PCG represent the combined Ao and Co VSMC data. Log2FC and FDR calculated based on grouping soft and stiff groups together, independent to cell-type. Only correlation p-values < 1e-5 are included.

119

BIBLIOGRAPHY

1. Young T. The Croonian Lecture: On the Functions of the Heart and Arteries. Philos Trans R Soc London. 1809;99:1-31.

2. Gepner AD, Korcarz CE, Colangelo LA, Hom EK, Tattersall MC, Astor BC, Kaufman JD, Liu K, Stein JH. Longitudinal effects of a decade of aging on carotid artery stiffness: the multiethnic study of atherosclerosis. Stroke. 2014;45(1):48-53.

3. AlGhatrif M, Strait JB, Morrell CH, Canepa M, Wright J, Elango P, Scuteri A, Najjar SS, Ferrucci L, Lakatta EG. Longitudinal trajectories of arterial stiffness and the role of blood pressure: the Baltimore Longitudinal Study of Aging. Hypertens (Dallas, Tex 1979). 2013;62(5):934-941.

4. Mitchell GF, Guo C-Y, Benjamin EJ, Larson MG, Keyes MJ, Vita JA, Vasan RS, Levy D. Cross-sectional correlates of increased aortic stiffness in the community: the Framingham Heart Study. Circulation. 2007;115(20):2628-2636.

5. van Popele NM, Grobbee DE, Bots ML, Asmar R, Topouchian J, Reneman RS, Hoeks AP, van der Kuip DA, Hofman A, Witteman JC. Association between arterial stiffness and atherosclerosis: the Rotterdam Study. Stroke. 2001;32(2):454-460.

6. Boutouyrie P, Tropeano AI, Asmar R, Gautier I, Benetos A, Lacolley P, Laurent S. Aortic stiffness is an independent predictor of primary coronary events in hypertensive patients: a longitudinal study. Hypertension. 2002;39(1):10-15.

7. Mitchell GF, Hwang S-J, Vasan RS, Larson MG, Pencina MJ, Hamburg NM, Vita JA, Levy D, Benjamin EJ. Arterial stiffness and cardiovascular events: the Framingham Heart Study. Circulation. 2010;121(4):505-511.

8. Shen L, Zhang YG, Liu M, Qiang DC, Sun XL, Liu L, Jiang YY. Increased arterial stiffness in subjects with pre-diabetes among middle aged population in Beijing, China. Biomed Environ Sci. 2013;26(9):717-725.

9. Weisbrod RM, Shiang T, Al Sayah L, Fry JL, Bajpai S, Reinhart-King CA, Lob HE, Santhanam L, Mitchell G, Cohen RA, Seta F. Arterial stiffening precedes systolic hypertension in diet-induced obesity. Hypertension. 2013;62(6):1105-1110.

10. Kaess BM, Rong J, Larson MG, Hamburg NM, Vita JA, Levy D, Benjamin EJ, Vasan RS, Mitchell GF. Aortic stiffness, blood pressure progression, and incident hypertension. JAMA. 2012;308(9):875-881.

11. Karras A, Haymann J-P, Bozec E, Metzger M, Jacquot C, Maruani G, Houillier P, Froissart M, Stengel B, Guardiola P, Laurent S, Boutouyrie P, Briet M, Nephro Test Study Group. Large artery stiffening and remodeling are independently associated with all-cause mortality and cardiovascular events in chronic kidney disease. Hypertens (Dallas, Tex 1979). 2012;60(6):1451-1457.

12. Laurent S, Boutouyrie P, Asmar R, Gautier I, Laloux B, Guize L, Ducimetiere P, Benetos A. Aortic stiffness is an independent predictor of all-cause and 120

cardiovascular mortality in hypertensive patients. Hypertension. 2001;37(5):1236- 1241.

13. Laurent S, Katsahian S, Fassot C, Tropeano A-I, Gautier I, Laloux B, Boutouyrie P. Aortic stiffness is an independent predictor of fatal stroke in essential hypertension. Stroke. 2003;34(5):1203-1206.

14. Mitchell GF, DeStefano AL, Larson MG, Benjamin EJ, Chen M-H, Vasan RS, Vita JA, Levy D. Heritability and a genome-wide linkage scan for arterial stiffness, wave reflection, and mean arterial pressure: the Framingham Heart Study. Circulation. 2005;112(2):194-199.

15. Tounian P, Aggoun Y, Dubern B, Varille V, Guy-Grand B, Sidi D, Girardet JP, Bonnet D. Presence of increased stiffness of the common carotid artery and endothelial dysfunction in severely obese children: a prospective study. Lancet (London, England). 2001;358(9291):1400-1404.

16. Ashor AW, Lara J, Siervo M, Celis-Morales C, Mathers JC. Effects of exercise modalities on arterial stiffness and wave reflection: a systematic review and meta- analysis of randomized controlled trials. PLoS One. 2014;9(10):e110034.

17. Takami T, Saito Y. Effects of smoking cessation on central blood pressure and arterial stiffness. Vasc Health Risk Manag. 2011;7:633-638.

18. Van Doornum S, McColl G, Wicks IP. Atorvastatin reduces arterial stiffness in patients with rheumatoid arthritis. Ann Rheum Dis. 2004;63(12):1571-1575.

19. Wu C-F. Therapeutic modification of arterial stiffness: An update and comprehensive review. World J Cardiol. 2015;7(11):742.

20. Boutouyrie P, Lacolley P, Briet M, Regnault V, Stanton A, Laurent S, Mahmud A. Pharmacological modulation of arterial stiffness. Drugs. 2011;71(13):1689-1701.

21. Rossi P, Franc?s Y, Kingwell BA, Ahimastos AA. Gender differences in artery wall biomechanical properties throughout life. J Hypertens. 2011;29(6):1023-1033.

22. Tarasov K V., Sanna S, Scuteri A, et al. COL4A1 Is Associated With Arterial Stiffness by Genome-Wide Association Scan. Circ Cardiovasc Genet. 2009;2(2):151-158.

23. Wain L V, Verwoert GC, O’Reilly PF, et al. Genome-wide association study identifies six new loci influencing pulse pressure and mean arterial pressure. Nat Genet. 2011;43(10):1005-1011.

24. Levy D, Larson MG, Benjamin EJ, Newton-Cheh C, Wang TJ, Hwang S-J, Vasan RS, Mitchell GF. Framingham Heart Study 100K Project: genome-wide associations for blood pressure and arterial stiffness. BMC Med Genet. 2007;8 Suppl 1:S3.

25. Vitarelli A, Conde Y, Cimino E, D’Angeli I, D’Orazio S, Stellato S, Padella V, Caranci F. Aortic wall mechanics in the Marfan syndrome assessed by

121

transesophageal tissue Doppler echocardiography. Am J Cardiol. 2006;97(4):571- 577.

26. Crosas-Molist E, Meirelles T, López-Luque J, et al. Vascular Smooth Muscle Cell Phenotypic Changes in Patients With Marfan SyndromeSignificance. Arterioscler Thromb Vasc Biol. 2015;35(4):960-972.

27. Li DY, Faury G, Taylor DG, Davis EC, Boyle WA, Mecham RP, Stenzel P, Boak B, Keating MT. Novel arterial pathology in mice and humans hemizygous for elastin. J Clin Invest. 1998;102(10):1783-1787.

28. Gerhard-Herman M, Smoot LB, Wake N, Kieran MW, Kleinman ME, Miller DT, Schwartzman A, Giobbie-Hurder A, Neuberg D, Gordon LB. Mechanisms of premature vascular aging in children with Hutchinson-Gilford progeria syndrome. Hypertens (Dallas, Tex 1979). 2012;59(1):92-97.

29. Raines EW. The extracellular matrix can regulate vascular cell migration, proliferation, and survival: relationships to vascular disease. Int J Exp Pathol. 2001;81(3):173-182.

30. Wagenseil JE, Mecham RP. Vascular Extracellular Matrix and Arterial Mechanics. 2009:957-989.

31. Kelleher CM, McLean SE, Mecham RP. Vascular Extracellular Matrix and Aortic Development. In: ; 2004:153-188.

32. Liu S-L, Bae YH, Yu C, Monslow J, Hawthorne EA, Castagnino P, Branchetti E, Ferrari G, Damrauer SM, Puré E, Assoian RK. Matrix metalloproteinase-12 is an essential mediator of acute and chronic arterial stiffening. Sci Rep. 2015;5:17189.

33. Chung AWY, Yang HHC, Sigrist MK, Brin G, Chum E, Gourlay WA, Levin A. Matrix metalloproteinase-2 and -9 exacerbate arterial stiffening and angiogenesis in diabetes and chronic kidney disease. Cardiovasc Res. 2009;84(3):494-504.

34. Kalath S, Tsipouras P, Silver FH. Increased aortic root stiffness associated with osteogenesis imperfecta. Ann Biomed Eng. 1987;15(1):91-99.

35. Radunovic Z, Wekre LL, Diep LM, Steine K. Cardiovascular abnormalities in adults with osteogenesis imperfecta. Am Heart J. 2011;161(3):523-529.

36. Matouk CC, Hanbidge A, Mandell DM, Terbrugge KG, Agid R. Osteogenesis Imperfecta, Multiple Intra-Abdominal Arterial Dissections and a Ruptured Dissecting-Type Intracranial Aneurysm. Interv Neuroradiol. 2011;17(3):371-375.

37. Eskola MJ, Niemelä KO, Kuusinen PR, Tarkka MR. Coronary artery dissection, combined aortic valve replacement and coronary bypass grafting in osteogenesis imperfecta. Interact Cardiovasc Thorac Surg. 2002;1(2):83-85.

38. Owens GK, Kumar MS, Wamhoff BR. Molecular Regulation of Vascular Smooth Muscle Cell Differentiation in Development and Disease. Physiol Rev. 2004;84(3):767-801.

122

39. Davignon J. Role of Endothelial Dysfunction in Atherosclerosis. Circulation. 2004;109(23_suppl_1):III-27-III-32.

40. Hergenreider E, Heydt S, Tréguer K, Boettger T, Horrevoets AJG, Zeiher AM, Scheffer MP, Frangakis AS, Yin X, Mayr M, Braun T, Urbich C, Boon R a, Dimmeler S. Atheroprotective communication between endothelial cells and smooth muscle cells through miRNAs. Nat Cell Biol. 2012;14(3):249-256.

41. Sehgel NL, Zhu Y, Sun Z, Trzeciakowski JP, Hong Z, Hunter WC, Vatner DE, Meininger GA, Vatner SF. Increased vascular smooth muscle cell stiffness: a novel mechanism for aortic stiffness in hypertension. AJP Hear Circ Physiol. 2013;305(9):H1281-H1287.

42. Qiu H, Zhu Y, Sun Z, Trzeciakowski JP, Gansner M, Depre C, Resuello RRG, Natividad FF, Hunter WC, Genin GM, Elson EL, Vatner DE, Meininger G a, Vatner SF. Short communication: vascular smooth muscle cell stiffness as a mechanism for increased aortic stiffness with aging. Circ Res. 2010;107(5):615-619.

43. Huynh J, Nishimura N, Rana K, Peloquin JM, Califano JP, Montague CR, King MR, Schaffer CB, Reinhart-King C a. Age-related intimal stiffening enhances endothelial permeability and leukocyte transmigration. Sci Transl Med. 2011;3(112):112ra122.

44. Kothapalli D, Liu S-L, Bae YH, et al. Cardiovascular protection by ApoE and ApoE-HDL linked to suppression of ECM gene expression and arterial stiffening. Cell Rep. 2012;2(5):1259-1271.

45. Liu F, Haeger CM, Dieffenbach PB, et al. Distal vessel stiffening is an early and pivotal mechanobiological regulator of vascular remodeling and pulmonary hypertension. JCI Insight. 2016;1(8).

46. Georges PC, Hui J-J, Gombos Z, McCormick ME, Wang AY, Uemura M, Mick R, Janmey PA, Furth EE, Wells RG. Increased stiffness of the rat liver precedes matrix deposition: implications for fibrosis. Am J Physiol Gastrointest Liver Physiol. 2007;293(6):G1147-54.

47. Tzima E, Irani-Tehrani M, Kiosses WB, Dejana E, Schultz D a, Engelhardt B, Cao G, DeLisser H, Schwartz MA. A mechanosensory complex that mediates the endothelial cell response to fluid shear stress. Nature. 2005;437(7057):426-431.

48. Retailleau K, Duprat F, Arhatte M, Ranade SS, Peyronnet R, Martins JR, Jodar M, Moro C, Offermanns S, Feng Y, Demolombe S, Patel A, Honoré E. Piezo1 in Smooth Muscle Cells Is Involved in Hypertension-Dependent Arterial Remodeling. Cell Rep. 2015;13(6):1161-1171.

49. Schubert R, Brayden JE. Stretch-Activated Cation Channels and the Myogenic Response of Small Arteries.; 2005.

50. Dembo M, Wang Y-L. Stresses at the Cell-to-Substrate Interface during Locomotion of Fibroblasts. Biophys J. 1999;76(4):2307-2316.

123

51. Harris AK, Stopak D, Wild P. Fibroblast traction as a mechanism for collagen morphogenesis. Nature. 1981;290(5803):249-251.

52. Harris A, Wild P, Stopak D. Silicone rubber substrata: a new wrinkle in the study of cell locomotion. Science (80- ). 1980;208(4440):177-179.

53. Pelham RJ, Wang Y l. Cell locomotion and focal adhesions are regulated by substrate flexibility. Proc Natl Acad Sci U S A. 1997;94(25):13661-13665.

54. Choquet D, Felsenfeld DP, Sheetz MP. Extracellular Matrix Rigidity Causes Strengthening of Integrin–Cytoskeleton Linkages. Cell. 1997;88(1):39-48.

55. DuFort CC, Paszek MJ, Weaver VM. Balancing forces: architectural control of mechanotransduction. Nat Rev Mol Cell Biol. 2011;12(5):308-319.

56. Discher DE, Janmey P, Wang Y-L. Tissue cells feel and respond to the stiffness of their substrate. Science. 2005;310(5751):1139-1143.

57. Eyckmans J, Boudou T, Yu X, Chen CS. A hitchhiker’s guide to mechanobiology. Dev Cell. 2011;21(1):35-47.

58. Solon J, Levental I, Sengupta K, Georges PC, Janmey PA. Fibroblast adaptation and stiffness matching to soft elastic substrates. Biophys J. 2007;93(12):4453- 4461.

59. Tan Y, Tajik A, Chen J, et al. Matrix softness regulates plasticity of tumour- repopulating cells via H3K9 demethylation and Sox2 expression. Nat Commun. 2014;5.

60. Le HQ, Ghatak S, Yeung C-YCC, et al. Mechanical regulation of transcription controls Polycomb-mediated gene silencing during lineage commitment. Nat Cell Biol. 2016;18(8):864-875.

61. Alam SG, Zhang Q, Prasad N, Li Y, Chamala S, Kuchibhotla R, KC B, Aggarwal V, Shrestha S, Jones AL, Levy SE, Roux KJ, Nickerson JA, Lele TP. The mammalian LINC complex regulates genome transcriptional responses to substrate rigidity. Sci Rep. 2016;6(1):38063.

62. Nagayama K, Yahiro Y, Matsumoto T. Stress fibers stabilize the position of intranuclear DNA through mechanical connection with the nucleus in vascular smooth muscle cells. FEBS Lett. 2011;585(24):3992-3997.

63. Tajik A, Zhang Y, Wei F, Sun J, Jia Q, Zhou W, Singh R, Khanna N, Belmont AS, Wang N. Transcription upregulation via force-induced direct stretching of chromatin. Nat Mater. 2016;15(12):1287-1296.

64. Guilluy C, Osborne LD, Van Landeghem L, Sharek L, Superfine R, Garcia-Mata R, Burridge K. Isolated nuclei adapt to force and reveal a mechanotransduction pathway in the nucleus. Nat Cell Biol. 2014;16(4):376-381.

65. Wang D-Z, Li S, Hockemeyer D, Sutherland L, Wang Z, Schratt G, Richardson JA, Nordheim A, Olson EN. Potentiation of serum response factor activity by a 124

family of myocardin-related transcription factors. Proc Natl Acad Sci U S A. 2002;99(23):14855-14860.

66. Zhao B, Ye X, Yu J, Li L, Li W, Li S, Yu J, Lin JD, Wang C-Y, Chinnaiyan AM, Lai Z-C, Guan K-L. TEAD mediates YAP-dependent gene induction and growth control. Genes Dev. 2008;22(14):1962-1971.

67. Miralles F, Posern G, Zaromytidou A-I, Treisman R. Actin dynamics control SRF activity by regulation of its coactivator MAL. Cell. 2003;113(3):329-342.

68. Dupont S, Morsut L, Aragona M, Enzo E, Giulitti S, Cordenonsi M, Zanconato F, Le Digabel J, Forcato M, Bicciato S, Elvassore N, Piccolo S. Role of YAP/TAZ in mechanotransduction. Nature. 2011;474(7350):179-183.

69. Huang X, Yang N, Fiore VF, Barker TH, Sun Y, Morris SW, Ding Q, Thannickal VJ, Zhou Y. Matrix stiffness-induced myofibroblast differentiation is mediated by intrinsic mechanotransduction. Am J Respir Cell Mol Biol. 2012;47(3):340-348.

70. Ishihara S, Yasuda M, Harada I, Mizutani T, Kawabata K, Haga H. Substrate stiffness regulates temporary NF-κB activation via actomyosin contractions. Exp Cell Res. 2013;319(19):2916-2927.

71. Tzima E, Del Pozo MA, Kiosses WB, Mohamed SA, Li S, Chien S, Schwartz MA. Activation of Rac1 by shear stress in endothelial cells mediates both cytoskeletal reorganization and effects on gene expression. EMBO J. 2002;21(24):6791-6800.

72. Nardone G, Oliver-De La Cruz J, Vrbsky J, et al. YAP regulates cell mechanics by controlling focal adhesion assembly. Nat Commun. 2017;8:15321.

73. Bordeleau F, Califano JP, Negr?n Abril YL, Mason BN, LaValley DJ, Shin SJ, Weiss RS, Reinhart-King CA. Tissue stiffness regulates serine/arginine-rich protein-mediated splicing of the extra domain B-fibronectin isoform in tumors. Proc Natl Acad Sci. 2015;112(27):8314-8319.

74. Mouw JK, Yui Y, Damiano L, Bainer RO, Lakins JN, Acerbi I, Ou G, Wijekoon AC, Levental KR, Gilbert PM, Hwang ES, Chen Y-Y, Weaver VM. Tissue mechanics modulate microRNA-dependent PTEN expression to regulate malignant progression. Nat Med. 2014;20(4):360-367.

75. Le LT-N, Cazares O, Mouw JK, Chatterjee S, Macias H, Moran A, Ramos J, Keely PJ, Weaver VM, Hinck L. Loss of miR-203 regulates cell shape and matrix adhesion through ROBO1/Rac/FAK in response to stiffness. J Cell Biol. 2016;212(6):707-719.

76. Stegemann JP, Hong H, Nerem RM. Mechanical, biochemical, and extracellular matrix effects on vascular smooth muscle cell phenotype. J Appl Physiol. 2005;98(6):2321-2327.

77. Qiu J, Zheng Y, Hu J, Liao D, Gregersen H, Deng X, Fan Y, Wang G. Biomechanical regulation of vascular smooth muscle cell functions: from in vitro to in vivo understanding. J R Soc Interface. 2013;11(90):20130852-20130852. 125

78. Peyton SR, Putnam AJ. Extracellular matrix rigidity governs smooth muscle cell motility in a biphasic fashion. J Cell Physiol. 2005;204(1):198-209.

79. Peyton SR, Kim PD, Ghajar CM, Seliktar D, Putnam AJ. The effects of matrix stiffness and RhoA on the phenotypic plasticity of smooth muscle cells in a 3-D biosynthetic hydrogel system. Biomaterials. 2008;29(17):2597-2607.

80. Peyton SR, Raub CB, Keschrumrus VP, Putnam AJ. The use of poly(ethylene glycol) hydrogels to investigate the impact of ECM chemistry and mechanics on smooth muscle cells. Biomaterials. 2006;27(28):4881-4893.

81. Klein EA, Yin L, Kothapalli D, Castagnino P, Byfield FJ, Xu T, Levental I, Hawthorne E, Janmey PA, Assoian RK. Cell-cycle control by physiological matrix elasticity and in vivo tissue stiffening. Curr Biol. 2009;19(18):1511-1518.

82. Isenberg BC, DiMilla PA, Walker M, Kim S, Wong JY. Vascular Smooth Muscle Cell Durotaxis Depends on Substrate Stiffness Gradient Strength. Biophys J. 2009;97(5):1313-1322.

83. Brown XQ, Bartolak-Suki E, Williams C, Walker ML, Weaver VM, Wong JY. Effect of substrate stiffness and PDGF on the behavior of vascular smooth muscle cells: Implications for atherosclerosis. J Cell Physiol. 2010;225(1):115-122.

84. Steucke KE, Tracy P V., Hald ES, Hall JL, Alford PW. Vascular smooth muscle cell functional contractility depends on extracellular mechanical properties. J Biomech. 2015;48(12):3044-3051.

85. Beamish JA, He P, Kottke-Marchant K, Marchant RE, Ph D. Molecular regulation of contractile smooth muscle cell phenotype: implications for vascular tissue engineering. Tissue Eng Part B Rev. 2010;16(5):467-491.

86. Friedland JC, Lee MH, Boettiger D. Mechanically Activated Integrin Switch Controls 5 1 Function. Science (80- ). 2009;323(5914):642-644.

87. Thyberg J, Blomgren K, Roy J, Tran PK, Hedin U. Phenotypic modulation of smooth muscle cells after arterial injury is associated with changes in the distribution of laminin and fibronectin. J Histochem Cytochem. 1997;45(6):837- 846.

88. Thyberg J, Hultgårdh-Nilsson A. Fibronectin and the basement membrane components laminin and collagen type IV influence the phenotypic properties of subcultured rat aortic smooth muscle cells differently. Cell Tissue Res. 1994;276(2):263-271.

89. Hartman CD, Isenberg BC, Chua SG, Wong JY. Vascular smooth muscle cell durotaxis depends on extracellular matrix composition. Proc Natl Acad Sci. 2016;113(40):11190-11195.

90. Sazonova O V, Isenberg BC, Herrmann J, Lee KL, Purwada A, Valentine AD, Buczek-Thomas JA, Wong JY, Nugent MA. Extracellular matrix presentation modulates vascular smooth muscle cell mechanotransduction. Matrix Biol. 126

2015;41:36-43.

91. Mui KL, Bae YH, Gao L, Liu S-L, Xu T, Radice GL, Chen CS, Assoian RK. N- Cadherin Induction by ECM Stiffness and FAK Overrides the Spreading Requirement for Proliferation of Vascular Smooth Muscle Cells. Cell Rep. March 2015.

92. Sazonova OV V., Lee KLL, Isenberg BCC, Rich CBB, Nugent MAA, Wong JYY. Cell-Cell Interactions Mediate the Response of Vascular Smooth Muscle Cells to Substrate Stiffness. Biophys J. 2011;101(3):622-630.

93. Bae YH, Mui KL, Hsu BY, Liu S-L, Cretu A, Razinia Z, Xu T, Puré E, Assoian RK. A FAK-Cas-Rac-lamellipodin signaling module transduces extracellular matrix stiffness into mechanosensitive cell cycling. Sci Signal. 2014;7(330):ra57.

94. Hsu BY, Bae YH, Mui KL, Liu S-L, Assoian RK. Apolipoprotein E3 Inhibits Rho to Regulate the Mechanosensitive Expression of Cox2. Kumar S, ed. PLoS One. 2015;10(6):e0128974.

95. Liu F, Mih JD, Shea BS, Kho AT, Sharif AS, Tager AM, Tschumperlin DJ. Feedback amplification of fibrosis through matrix stiffening and COX-2 suppression. J Cell Biol. 2010;190(4):693-706.

96. Dieffenbach PB, Mallarino Haeger C, Coronata AMF, Choi KM, Varelas X, Tschumperlin DJ, Fredenburgh LE. Arterial Stiffness Induces Remodeling Phenotypes in Pulmonary Artery Smooth Muscle Cells via YAP/TAZ-Mediated Repression of Cyclooxygenase-2. Am J Physiol - Lung Cell Mol Physiol. June 2017:ajplung.00173.2017.

97. Xie C, Guo Y, Zhu T, Zhang J, Ma PX, Chen YE. Yap1 protein regulates vascular smooth muscle cell phenotypic switch by interaction with myocardin. J Biol Chem. 2012;287(18):14598-14605.

98. Dinardo CL, Venturini G, Zhou EH, Watanabe IS, Campos LCG, Dariolli R, da Motta-Leal-Filho JM, Carvalho VM, Cardozo KHM, Krieger JE, Alencar AM, Pereira AC. Variation of mechanical properties and quantitative proteomics of VSMC along the arterial tree. Am J Physiol Heart Circ Physiol. 2014;306(4):H505- 16.

99. Civelek M, Manduchi E, Riley RJ, Stoeckert CJ, Davies PF. Coronary artery endothelial transcriptome in vivo identification of endoplasmic reticulum stress and enhanced reactive oxygen species by gene connectivity network analysis. Circ Cardiovasc Genet. 2011;4(3):243-252.

100. Burridge KA, Friedman MH. Environment and vascular bed origin influence differences in endothelial transcriptional profiles of coronary and iliac arteries. AJP Hear Circ Physiol. 2010;299(3):H837-H846.

101. Zhang H, Gu S, Al-Sabeq B, Wang S, He J, Tam A, Cifelli C, Mathalone N, Tirgari S, Boyd S, Heximer SP. Origin-specific epigenetic program correlates with vascular bed-specific differences in Rgs5 expression. FASEB J. 2012;26(1):181- 127

191.

102. Reslan OM, Yin Z, do Nascimento GRA, Khalil RA. Subtype-specific Estrogen Receptor-mediated Vasodilator Activity in the Cephalic, Thoracic, and Abdominal Vasculature of Female Rat. J Cardiovasc Pharmacol. 2013;62(1):26-40.

103. Sufka KJ, Stratton DB, Giordano J. Regional differences in serotonergic contractile sensitivity mediated by 5-HT2 receptors in rat aorta. Artery. 1990;18(1):47-53.

104. Gadson PF, Rossignol C, McCoy J, Rosenquist TH. Expression of elastin, smooth muscle alpha-actin, and c-jun as a function of the embryonic lineage of vascular smooth muscle cells. In Vitro Cell Dev Biol Anim. 1993;29A(10):773-781.

105. Leroux-Berger M, Queguiner I, MacIel TT, Ho A, Relaix F, Kempf H. Pathologic calcification of adult vascular smooth muscle cells differs on their crest or mesodermal embryonic origin. J Bone Miner Res. 2011;26(7):1543-1553.

106. Topouzis S, Majesky MW. Smooth muscle lineage diversity in the chick embryo. Two types of aortic smooth muscle cell differ in growth and receptor-mediated transcriptional responses to transforming growth factor-beta. Dev Biol. 1996;178(2):430-445.

107. Gadson PF, Dalton ML, Patterson E, Svoboda DD, Hutchinson L, Schram D, Rosenquist TH. Differential response of mesoderm- and neural crest-derived smooth muscle to TGF-beta1: regulation of c-myb and alpha1 (I) procollagen genes. Exp Cell Res. 1997;230(2):169-180.

108. Thieszen SL, Dalton M, Gadson PF, Patterson E, Rosenquist TH. Embryonic lineage of vascular smooth muscle cells determines responses to collagen matrices and integrin receptor expression. Exp Cell Res. 1996;227(1):135-145.

109. Majesky MW. Developmental basis of vascular smooth muscle diversity. Arterioscler Thromb Vasc Biol. 2007;27(6):1248-1258.

110. Yoshida T, Owens GK. Molecular determinants of vascular smooth muscle cell diversity. Circ Res. 2005;96(3):280-291.

111. Bochaton-Piallat M-L, Clowes AW, Clowes MM, Fischer JW, Redard M, Gabbiani F, Gabbiani G. Cultured Arterial Smooth Muscle Cells Maintain Distinct Phenotypes When Implanted Into Carotid Artery. Arterioscler Thromb. 2001;21(21):949-954.

112. Haimovici H, Maier N. Fate of Aortic Homografts in Canine Atherosclerosis III. Study of Fresh Abdominal and Thoracic Aortic Implants Into Thoracic Aorta: Role of Tissue Susceptibility in Atherogenesis. Arch Surg. 1964;89:961-969.

113. Haimovici H, Maier N. Experimental canine atherosclerosis in autogenous abdominal aortic grafts implanted into the jugular vein. Atherosclerosis. 13(3):375- 384.

128

114. Haimovici H. The role of arterial tissue susceptibility in atherogenesis. Texas Hear Inst J. 1991;18(1):81-83.

115. Gittenberger-de Groot AC, DeRuiter MC, Bergwerff M, Poelmann RE. Smooth muscle cell origin and its relation to heterogeneity in development and disease. Arterioscler Thromb Vasc Biol. 1999;19(7):1589-1594.

116. Pfaltzgraff ER, Bader DM. Heterogeneity in vascular smooth muscle cell embryonic origin in relation to adult structure, physiology, and disease. Dev Dyn. 2015;244(3):410-416.

117. Cheung C, Bernardo AS, Trotter MWB, Pedersen RA, Sinha S. Generation of human vascular smooth muscle subtypes provides insight into embryological origin-dependent disease susceptibility. Nat Biotechnol. 2012;30(2):165-173.

118. Harrow J, Frankish A, Gonzalez JM, et al. GENCODE: the reference human genome annotation for The ENCODE Project. Genome Res. 2012;22(9):1760- 1774.

119. Lonsdale J, Thomas J, Salvatore M, et al. The Genotype-Tissue Expression (GTEx) project. Nat Genet. 2013;45(6):580-585.

120. Bernstein BE, Birney E, Dunham I, Green ED, Gunter C, Snyder M. An integrated encyclopedia of DNA elements in the human genome. Nature. 2012;489(7414):57-74.

121. Bernstein BE, Stamatoyannopoulos JA, Costello JF, et al. The NIH Roadmap Epigenomics Mapping Consortium. Nat Biotechnol. 2010;28(10):1045-1048.

122. Djebali S, Davis CA, Merkel A, et al. Landscape of transcription in human cells. Nature. 2012;489(7414):101-108.

123. Battle A, Mostafavi S, Zhu X, Potash JB, Weissman MM, McCormick C, Haudenschild CD, Beckman KB, Shi J, Mei R, Urban AE, Montgomery SB, Levinson DF, Koller D. Characterizing the genetic basis of transcriptome diversity through RNA-sequencing of 922 individuals. Genome Res. 2014;24(1):14-24.

124. Hon C-C, Ramilowski JA, Harshbarger J, et al. An atlas of human long non-coding RNAs with accurate 5’ ends. Nature. 2017;543(7644):199-204.

125. Iyer MK, Niknafs YS, Malik R, et al. The landscape of long noncoding RNAs in the human transcriptome. Nat Genet. 2015;47(3):199-208.

126. Cabili MN, Trapnell C, Goff L, Koziol M, Tazon-Vega B, Regev A, Rinn JL. Integrative annotation of human large intergenic noncoding RNAs reveals global properties and specific subclasses. Genes Dev. 2011;25(18):1915-1927.

127. Liu SJ, Horlbeck MA, Cho SW, et al. CRISPRi-based genome-scale identification of functional long noncoding RNA loci in human cells. Science (80- ). 2017;355(6320).

128. Fatica A, Bozzoni I. Long non-coding RNAs: new players in cell differentiation and 129

development. Nat Rev Genet. 2014;15(1):7-21.

129. Engreitz JM, Haines JE, Perez EM, Munson G, Chen J, Kane M, McDonel PE, Guttman M, Lander ES. Local regulation of gene expression by lncRNA promoters, transcription and splicing. Nature. 2016;539(7629):452-455.

130. Congrains A, Kamide K, Katsuya T, Yasuda O, Oguro R, Yamamoto K, Ohishi M, Rakugi H. CVD-associated non-coding RNA, ANRIL, modulates expression of atherogenic pathways in VSMC. Biochem Biophys Res Commun. 2012;419(4):612-616.

131. Wu G, Cai J, Han Y, et al. LincRNA-p21 regulates neointima formation, vascular smooth muscle cell proliferation, apoptosis, and atherosclerosis by enhancing p53 activity. Circulation. 2014;130(17):1452-1465.

132. Bell RD, Long X, Lin M, Bergmann JH, Nanda V, Cowan SL, Zhou Q, Han Y, Spector DL, Zheng D, Miano JM. Identification and Initial Functional Characterization of a Human Vascular Cell-Enriched Long Noncoding RNA. Arterioscler Thromb Vasc Biol. 2014;34(6):1249-1259.

133. Ballantyne MD, Pinel K, Dakin R, et al. Smooth Muscle Enriched Long Noncoding RNA ( SMILR ) Regulates Cell Proliferation. Circulation. 2016;133(21):2050-2065.

134. Zhao J, Zhang W, Lin M, Wu W, Jiang P, Tou E, Xue M, Richards A, Jourd’heuil D, Asif A, Zheng D, Singer HA, Miano JM, Long X. MYOSLID Is a Novel Serum Response Factor-Dependent Long Noncoding RNA That Amplifies the Vascular Smooth Muscle Differentiation Program. Arterioscler Thromb Vasc Biol. 2016;36(10):2088-2099.

135. Wang ET, Sandberg R, Luo S, Khrebtukova I, Zhang L, Mayr C, Kingsmore SF, Schroth GP, Burge CB. Alternative isoform regulation in human tissue transcriptomes. Nature. 2008;456(7221):470-476.

136. Kim M-S, Pinto SM, Getnet D, et al. A draft map of the human proteome. Nature. 2014;509(7502):575-581.

137. Baralle FE, Giudice J. Alternative splicing as a regulator of development and tissue identity. Nat Rev Mol Cell Biol. May 2017.

138. Llorian M, Gooding C, Bellora N, Hallegger M, Buckroyd A, Wang X, Rajgor D, Kayikci M, Feltham J, Ule J, Eyras E, Smith CWJ. The alternative splicing program of differentiated smooth muscle cells involves concerted non-productive splicing of post-transcriptional regulators. Nucleic Acids Res. 2016;44(18):8933- 8950.

139. Van Der Veer EP, De Bruin RG, Kraaijeveld AO, et al. Quaking, an RNA-binding protein, is a critical regulator of vascular smooth muscle cell phenotype. Circ Res. 2013;113(9):1065-1075.

140. Ringnér M. What is principal component analysis? Nat Biotechnol. 2008;26(3):303-304. 130

141. D’haeseleer P. How does gene expression clustering work? Nat Biotechnol. 2005;23(12):1499-1501.

142. Ulitsky I. Evolution to the rescue: using comparative genomics to understand long non-coding RNAs. Nat Rev Genet. 2016;17(10):601-614.

143. Fuller T, Langfelder P, Presson A, Horvath S. Review of Weighted Gene Coexpression Network Analysis. In: Handbook of Statistical Bioinformatics. Berlin, Heidelberg: Springer Berlin Heidelberg; 2011:369-388.

144. Tavazoie S, Hughes JD, Campbell MJ, Cho RJ, Church GM. Systematic determination of genetic network architecture. Nat Genet. 1999;22(3):281-285.

145. Wang J, Wu G, Chen L, Zhang W. Cross-species transcriptional network analysis reveals conservation and variation in response to metal stress in cyanobacteria. BMC Genomics. 2013;14(1):112.

146. Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics. 2008;9:559.

147. Strong A, Patel K, Rader DJ. Sortilin and lipoprotein metabolism: making sense out of complexity. Curr Opin Lipidol. 2014;25(5):350-357.

148. Klein EA, Yung Y, Castagnino P, Kothapalli D, Assoian RK. Cell adhesion, cellular tension, and cell cycle control. Methods Enzymol. 2007;426:155-175.

149. Cretu A, Castagnino P, Assoian R. Studying the effects of matrix stiffness on cellular function using acrylamide-based hydrogels. J Vis Exp. 2010;(42).

150. Lee RT, Richardson SG, Loree HM, Grodzinsky AJ, Gharib SA, Schoen FJ, Pandian N. Prediction of mechanical properties of human atherosclerotic tissue by high-frequency intravascular ultrasound imaging. An in vitro study. Arterioscler Thromb Vasc Biol. 1992;12(1):1-5.

151. Glukhova MA, Frid MG, Shekhonin B V, Vasilevskaya TD, Grunwald J, Saginati M, Koteliansky VE. Expression of extra domain A fibronectin sequence in vascular smooth muscle cells is phenotype dependent. J Cell Biol. 1989;109(1):357-366.

152. Rohwedder I, Montanez E, Beckmann K, Bengtsson E, Dunér P, Nilsson J, Soehnlein O, Fässler R. Plasma fibronectin deficiency impedes atherosclerosis progression and fibrous cap formation. EMBO Mol Med. 2012;4(7):564-576.

153. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P, Chaisson M, Gingeras TR. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29(1):15-21.

154. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, Marth G, Abecasis G, Durbin R, 1000 Genome Project Data Processing Subgroup. The Sequence Alignment/Map format and SAMtools. Bioinformatics. 2009;25(16):2078-2079.

155. Anders S, Pyl PT, Huber W. HTSeq--a Python framework to work with high- throughput sequencing data. Bioinformatics. 2015;31(2):166-169. 131

156. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550.

157. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL. Gene set enrichment analysis : A knowledge-based approach for interpreting genome-wide. 2005.

158. Vaquero-Garcia J, Barrera A, Gazzara MR, González-Vallinas J, Lahens NF, Hogenesch JB, Lynch KW, Barash Y, González-Vallinas J, Lahens NF, Hogenesch JB, Lynch KW, Barash Y. A new view of transcriptome complexity and regulation through the lens of local splicing variations. Elife. 2016;5:1-30.

159. Huang DW, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc. 2009;4(1):44- 57.

160. Huang DW, Sherman BT, Lempicki RA. Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res. 2009;37(1):1-13.

161. Mayer BJ, Baltimore D. Signalling through SH2 and SH3 domains. Trends Cell Biol. 1993;3(1):8-13.

162. Schlessinger J. SH2/SH3 signaling proteins. Curr Opin Genet Dev. 1994;4(1):25- 30.

163. Zhang X, Joehanes R, Chen BH, Huan T, Ying S, Munson PJ, Johnson AD, Levy D, O’Donnell CJ. Identification of common genetic variants controlling transcript isoform variation in human whole blood. Nat Genet. 2015;47(4):345-352.

164. International Consortium for Blood Pressure Genome-Wide Association Studies, Ehret GB, Munroe PB, et al. Genetic variants in novel pathways influence blood pressure and cardiovascular disease risk. Nature. 2011;478(7367):103-109.

165. Nikpay M, Goel A, Won H-H, et al. A comprehensive 1,000 Genomes-based genome-wide association meta-analysis of coronary artery disease. Nat Genet. 2015;47(10):1121-1130.

166. Mammoto T, Ingber DE. Mechanical control of tissue and organ development. Development. 2010;137(9):1407-1420.

167. Engler AJ, Sen S, Sweeney HL, Discher DE. Matrix elasticity directs stem cell lineage specification. Cell. 2006;126(4):677-689.

168. Przybyla L, Lakins JN, Weaver VM. Tissue Mechanics Orchestrate Wnt- Dependent Human Embryonic Stem Cell Differentiation. Cell Stem Cell. 2016;19(4):462-475.

169. Saha K, Keung AJ, Irwin EF, Li Y, Little L, Schaffer D V, Healy KE. Substrate modulus directs neural stem cell behavior. Biophys J. 2008;95(9):4426-4438.

170. Cosgrove BD, Mui KL, Driscoll TP, Caliari SR, Mehta KD, Assoian RK, Burdick 132

JA, Mauck RL. N-cadherin adhesive interactions modulate matrix mechanosensing and fate commitment of mesenchymal stem cells. Nat Mater. 2016;15(12):1297-1306.

171. Swift J, Ivanovska IL, Buxboim A, Harada T, Dingal PCDP, Pinter J, Pajerowski JD, Spinler KR, Shin J-WJ-W, Tewari M, Rehfeldt F, Speicher DW, Discher DE. Nuclear lamin-A scales with tissue stiffness and enhances matrix-directed differentiation. Science. 2013;341(6149):1240104.

172. Janmey PA, Wells RG, Assoian RK, McCulloch CA. From tissue mechanics to transcription factors. Differentiation. 2013;86(3):112-120.

173. Dupont S. Role of YAP/TAZ in cell-matrix adhesion-mediated signalling and mechanotransduction. Exp Cell Res. 2016;343(1):42-53.

174. Saito K, Chen M, Bard F, Chen S, Zhou H, Woodley D, Polischuk R, Schekman R, Malhotra V. TANGO1 facilitates cargo loading at endoplasmic reticulum exit sites. Cell. 2009;136(5):891-902.

175. Wilson DG, Phamluong K, Li L, Sun M, Cao TC, Liu PS, Modrusan Z, Sandoval WN, Rangell L, Carano RAD, Peterson AS, Solloway MJ. Global defects in collagen secretion in a Mia3/TANGO1 knockout mouse. J Cell Biol. 2011;193(5):935-951.

176. Wu X, Kodama A, Fuchs E. ACF7 regulates cytoskeletal-focal adhesion dynamics and migration and has ATPase activity. Cell. 2008;135(1):137-148.

177. Murphy DA, Courtneidge SA. The “ins” and “outs” of podosomes and invadopodia: characteristics, formation and function. Nat Rev Mol Cell Biol. 2011;12(7):413- 426.

178. Alexander NR, Branch KM, Parekh A, Clark ES, Iwueke IC, Guelcher SA, Weaver AM. Extracellular matrix rigidity promotes invadopodia activity. Curr Biol. 2008;18(17):1295-1299.

179. Quintavalle M, Elia L, Condorelli G, Courtneidge SA. MicroRNA control of podosome formation in vascular smooth muscle cells in vivo and in vitro. J Cell Biol. 2010;189(1):13-22.

180. Weidmann H, Touat-Hamici Z, Durand H, Mueller C, Chardonnet S, Pionneau C, Charlotte F, Janssen K-P, Verdugo R, Cambien F, Blankenberg S, Tiret L, Zeller T, Ninio E. SASH1, a new potential link between smoking and atherosclerosis. Atherosclerosis. 2015;242(2):571-579.

181. Greer P. Closing in on the biological functions of Fps/Fes and Fer. Nat Rev Mol Cell Biol. 2002;3(4):278-289.

182. Jücker M, McKenna K, da Silva AJ, Rudd CE, Feldman RA. The Fes protein- tyrosine kinase phosphorylates a subset of macrophage proteins that are involved in cell adhesion and cell-cell signaling. J Biol Chem. 1997;272(4):2104-2109.

133

183. Keren H, Lev-Maor G, Ast G. Alternative splicing and evolution: diversification, exon definition and function. Nat Rev Genet. 2010;11(5):345-355.

184. Dingal PCDP, Bradshaw AM, Cho S, Raab M, Buxboim A, Swift J, Discher DE. Fractal heterogeneity in minimal matrix models of scars modulates stiff-niche stem-cell responses via nuclear exit of a mechanorepressor. Nat Mater. 2015;14(9):951-960.

185. Zhou Q, Li L, Zhao B, Guan K-L. The Hippo Pathway in Heart Development, Regeneration, and Diseases. Circ Res. 2015;116(8):1431-1447.

186. Rosenbluh J, Nijhawan D, Cox AG, et al. β-Catenin-driven cancers require a YAP1 transcriptional complex for survival and tumorigenesis. Cell. 2012;151(7):1457-1473.

187. Tsai BP, Hoverter NP, Waterman ML. Blending hippo and WNT: sharing messengers and regulation. Cell. 2012;151(7):1401-1403.

188. Ebisuya M, Yamamoto T, Nakajima M, Nishida E. Ripples from neighbouring transcription. Nat Cell Biol. 2008;10(9):1106-1113.

189. Guttman M, Amit I, Garber M, et al. Chromatin signature reveals over a thousand highly conserved large non-coding RNAs in mammals. Nature. 2009;458(7235):223-227.

190. Rinn JL, Chang HY. Genome regulation by long noncoding RNAs. Annu Rev Biochem. 2012;81:145-166.

191. Huarte M, Guttman M, Feldser D, et al. A large intergenic noncoding RNA induced by p53 mediates global gene repression in the p53 response. Cell. 2010;142(3):409-419.

192. Gupta RA, Shah N, Wang KC, et al. Long non-coding RNA HOTAIR reprograms chromatin state to promote cancer metastasis. Nature. 2010;464(7291):1071- 1076.

193. Khalil AM, Guttman M, Huarte M, Garber M, Raj A, Rivea Morales D, Thomas K, Presser A, Bernstein BE, van Oudenaarden A, Regev A, Lander ES, Rinn JL. Many human large intergenic noncoding RNAs associate with chromatin- modifying complexes and affect gene expression. Proc Natl Acad Sci. 2009;106(28):11667-11672.

194. Yao H, Brick K, Evrard Y, Xiao T, Camerini-Otero RD, Felsenfeld G. Mediation of CTCF transcriptional insulation by DEAD-box RNA-binding protein p68 and steroid receptor RNA activator SRA. Genes Dev. 2010;24(22):2543-2555.

195. Schmitz K-M, Mayer C, Postepska A, Grummt I. Interaction of noncoding RNA with the rDNA promoter mediates recruitment of DNMT3b and silencing of rRNA genes. Genes Dev. 2010;24(20):2264-2269.

196. Ørom UA, Derrien T, Beringer M, Gumireddy K, Gardini A, Bussotti G, Lai F,

134

Zytnicki M, Notredame C, Huang Q, Guigo R, Shiekhattar R. Long noncoding RNAs with enhancer-like function in human cells. Cell. 2010;143(1):46-58.

197. Wang KC, Yang YW, Liu B, et al. A long noncoding RNA maintains active chromatin to coordinate homeotic gene expression. Nature. 2011;472(7341):120- 124.

198. Hacisuleyman E, Goff L a, Trapnell C, et al. Topological organization of multichromosomal regions by the long intergenic noncoding RNA Firre. Nat Struct Mol Biol. 2014;21(2):198-206.

199. Beltran M, Puig I, Peña C, García JM, Alvarez AB, Peña R, Bonilla F, de Herreros AG. A natural antisense transcript regulates Zeb2/Sip1 gene expression during Snail1-induced epithelial-mesenchymal transition. Genes Dev. 2008;22(6):756- 769.

200. Tripathi V, Ellis JD, Shen Z, Song DY, Pan Q, Watt AT, Freier SM, Bennett CF, Sharma A, Bubulya P a, Blencowe BJ, Prasanth SG, Prasanth K V. The nuclear- retained noncoding RNA MALAT1 regulates alternative splicing by modulating SR splicing factor phosphorylation. Mol Cell. 2010;39(6):925-938.

201. Salmena L, Poliseno L, Tay Y, Kats L, Pandolfi P. A ceRNA Hypothesis: The Rosetta Stone of a Hidden RNA Language? Cell. 2011;146(3):353-358.

202. Chun K-S, Surh Y-J. Signal transduction pathways regulating cyclooxygenase-2 expression: potential molecular targets for chemoprevention. Biochem Pharmacol. 2004;68(6):1089-1100.

203. Krawczyk M, Emerson BM. p50-associated COX-2 extragenic RNA (PACER) activates COX-2 gene expression by occluding repressive NF-κB complexes. Elife. 2014;3(3):e01776.

204. Rapicavoli NA, Qu K, Zhang J, Mikhail M, Laberge R-M, Chang HY. A mammalian pseudogene lncRNA at the interface of inflammation and anti-inflammatory therapeutics. Elife. 2013;2.

205. Ernst J, Kellis M. ChromHMM: automating chromatin-state discovery and characterization. Nat Methods. 2012;9(3):215-216.

206. Kundaje A, Meuleman W, Ernst J, et al. Integrative analysis of 111 reference human epigenomes. Nature. 2015;518(7539):317-330.

207. Hatcher CJ, Goldstein MM, Mah CS, Delia CS, Basson CT. Identification and Localization of TBX5 Transcription Factor During Human Cardiac Morphogenesis. 2000;95(March):90-95.

208. Bailey TL, Boden M, Buske F a, Frith M, Grant CE, Clementi L, Ren J, Li WW, Noble WS. MEME SUITE: tools for motif discovery and searching. Nucleic Acids Res. 2009;37(Web Server issue):W202-8.

209. Basson CT, Bachinsky DR, Lin RC, Levi T, Elkins JA, Soults J, Grayzel D,

135

Kroumpouzou E, Traill TA, Leblanc-Straceski J, Renault B, Kucherlapati R, Seidman JG, Seidman CE. Mutations in human TBX5 [corrected] cause limb and cardiac malformation in Holt-Oram syndrome. Nat Genet. 1997;15(1):30-35.

210. Li QY, Newbury-Ecob RA, Terrett JA, et al. Holt-Oram syndrome is caused by mutations in TBX5, a member of the Brachyury (T) gene family. Nat Genet. 1997;15(1):21-29.

211. Baban A, Postma AV, Marini M, et al. Identification of TBX5 mutations in a series of 94 patients with Tetralogy of Fallot. Am J Med Genet A. 2014;164(12):3100- 3107.

212. Diman NYS-G, Brooks G, Kruithof BPT, Elemento O, Seidman JG, Seidman CE, Basson CT, Hatcher CJ. Tbx5 is required for avian and Mammalian epicardial formation and coronary vasculogenesis. Circ Res. 2014;115(10):834-844.

213. Steimle JD, Moskowitz IP. TBX5: A Key Regulator of Heart Development. Curr Top Dev Biol. 2017;122:195-221.

214. Murakami M, Nakagawa M, Olson EN, Nakagawa O. A WW domain protein TAZ is a critical coactivator for TBX5, a transcription factor implicated in Holt-Oram syndrome. Proc Natl Acad Sci U S A. 2005;102(50):18034-18039.

215. Touma M, Kang X, Zhao Y, Cass AA, Gao F, Biniwale R, Coppola G, Xiao X, Reemtsen B, Wang Y. Decoding the Long Noncoding RNA during Cardiac Maturation: A Roadmap for Functional Discovery. Circ Cardiovasc Genet. 2016;9(5).

216. Qian M, Yang X, Li Z, Jiang C, Song D, Yan W, Liu T, Wu Z, Kong J, Wei H, Xiao J. P50-associated COX-2 extragenic RNA (PACER) overexpression promotes proliferation and metastasis of osteosarcoma cells by activating COX-2 gene. Tumour Biol. 2016;37(3):3879-3886.

217. Avendaño MS, Martínez-Revelles S, Aguado A, et al. Role of COX-2-derived PGE2 on vascular stiffness and function in hypertension. Br J Pharmacol. 2016;173(9):1541-1555.

218. Rezaee M, Penta K, Quertermous T. Del1 mediates VSMC adhesion, migration, and proliferation through interaction with integrin alpha(v)beta(3). Am J Physiol Heart Circ Physiol. 2002;282(5):H1924-32.

219. Starr DA. KASH and SUN proteins. Curr Biol. 2011;21(11):R414-R415.

220. Lovett DB, Shekhar N, Nickerson JA, Roux KJ, Lele TP. Modulation of Nuclear Shape by Substrate Rigidity. Cell Mol Bioeng. 2013;6(2):230-238.

221. Stewart-Hutchinson PJ, Hale CM, Wirtz D, Hodzic D. Structural requirements for the assembly of LINC complexes and their function in cellular mechanical stiffness. Exp Cell Res. 2008;314(8):1892-1905.

222. Dechat T, Korbei B, Vaughan OA, Vlcek S, Hutchison CJ, Foisner R. Lamina-

136

associated polypeptide 2alpha binds intranuclear A-type lamins. J Cell Sci. 2000;113 Pt 19:3473-3484.

223. Wagner N, Krohne G. LEM-Domain proteins: new insights into lamin-interacting proteins. Int Rev Cytol. 2007;261:1-46.

224. Chen G, Wood EG, Wang SH, Warner TD. Expression of cyclooxygenase-2 in rat vascular smooth muscle cells is unrelated to nuclear factor-kappaB activation. Life Sci. 1999;64(14):1231-1242.

225. Shin K-J, Wall EA, Zavzavadjian JR, Santat LA, Liu J, Hwang J-I, Rebres R, Roach T, Seaman W, Simon MI, Fraser IDC. A single lentiviral vector platform for microRNA-based conditional RNA interference and coordinated transgene expression. Proc Natl Acad Sci U S A. 2006;103(37):13759-13764.

226. Camarata T, Bimber B, Kulisz A, Chew T-L, Yeung J, Simon H-G. LMP4 regulates Tbx5 protein subcellular localization and activity. J Cell Biol. 2006;174(3):339-348.

227. Krause A, Zacharias W, Camarata T, Linkhart B, Law E, Lischke A, Miljan E, Simon H-G. Tbx5 and Tbx4 transcription factors interact with a new chicken PDZ- LIM protein in limb and heart development. Dev Biol. 2004;273(1):106-120.

228. Bimber B, Dettman RW, Simon H-G. Differential regulation of Tbx5 protein expression and sub-cellular localization during heart development. Dev Biol. 2007;302(1):230-242.

229. Wu JM, Young ML, Wang TR, Lin SJ, Chang JK, Wei J. Unusual cardiac malformations in Holt-Oram syndrome: report of two cases. Zhonghua Min Guo Xiao Er Ke Yi Xue Hui Za Zhi. 32(2):100-104.

230. HOLT M, ORAM S. Familial heart disease with skeletal malformations. Br Heart J. 1960;22:236-242.

231. Russell MR, Penikis A, Oldridge DA, et al. CASC15-S Is a Tumor Suppressor lncRNA at the 6p22 Neuroblastoma Susceptibility Locus. Cancer Res. 2015;75(15):3155-3166.

232. Previtera ML, Sengupta A. Substrate Stiffness Regulates Proinflammatory Mediator Production through TLR4 Activity in Macrophages. PLoS One. 2015;10(12):e0145813.

233. Sero JE, Sailem HZ, Ardy RC, Almuttaqi H, Zhang T, Bakal C. Cell shape and the microenvironment regulate nuclear translocation of NF- B in breast epithelial and tumor cells. Mol Syst Biol. 2015;11(3):790-790.

234. Chun K-S, Surh Y-J. Signal transduction pathways regulating cyclooxygenase-2 expression: potential molecular targets for chemoprevention. Biochem Pharmacol. 2004;68(6):1089-1100.

235. Ji P, Diederichs S, Wang W, Böing S, Metzger R, Schneider PM, Tidow N, Brandt B, Buerger H, Bulk E, Thomas M, Berdel WE, Serve H, Müller-Tidow C. MALAT-1,

137

a novel noncoding RNA, and thymosin beta4 predict metastasis and survival in early-stage non-small cell lung cancer. Oncogene. 2003;22(39):8031-8041.

236. Lin R, Maeda S, Liu C, Karin M, Edgington TS. A large noncoding RNA is a marker for murine hepatocellular carcinomas and a spectrum of human carcinomas. Oncogene. 2007;26(6):851-858.

237. Gutschner T, Hämmerle M, Diederichs S. MALAT1 — a paradigm for long noncoding RNA function in cancer. J Mol Med. 2013;91(7):791-801.

238. Hutchinson JN, Ensminger AW, Clemson CM, Lynch CR, Lawrence JB, Chess A. A screen for nuclear transcripts identifies two linked noncoding RNAs associated with SC35 splicing domains. BMC Genomics. 2007;8:39.

239. Wilusz JE, Freier SM, Spector DL. 3’ end processing of a long nuclear-retained noncoding RNA yields a tRNA-like cytoplasmic RNA. Cell. 2008;135(5):919-932.

240. Hutchinson JN, Ensminger AW, Clemson CM, Lynch CR, Lawrence JB, Chess A. A screen for nuclear transcripts identifies two linked noncoding RNAs associated with SC35 splicing domains. BMC Genomics. 2007;8:39.

241. Yang L, Lin C, Liu W, Zhang J, Ohgi K, Grinstein J, Dorrestein P, Rosenfeld M. ncRNA- and Pc2 Methylation-Dependent Gene Relocation between Nuclear Structures Mediates Gene Activation Programs. Cell. 2011;147(4):773-788.

242. Tripathi V, Shen Z, Chakraborty A, Giri S, Freier SM, Wu X, Zhang Y, Gorospe M, Prasanth SG, Lal A, Prasanth K V. Long noncoding RNA MALAT1 controls cell cycle progression by regulating the expression of oncogenic transcription factor B- MYB. PLoS Genet. 2013;9(3):e1003368.

243. Gutschner T, Hämmerle M, Eissmann M, Hsu J, Kim Y, Hung G, Revenko A, Arun G, Stentrup M, Gross M, Zörnig M, MacLeod AR, Spector DL, Diederichs S. The noncoding RNA MALAT1 is a critical regulator of the metastasis phenotype of lung cancer cells. Cancer Res. 2013;73(3):1180-1189.

244. Michalik KM, You X, Manavski Y, Doddaballapur A, Zörnig M, Braun T, John D, Ponomareva Y, Chen W, Uchida S, Boon R a, Dimmeler S. Long noncoding RNA MALAT1 regulates endothelial cell function and vessel growth. Circ Res. 2014;114(9):1389-1397.

245. Gutschner T, Hämmerle M, Diederichs S. MALAT1 — a paradigm for long noncoding RNA function in cancer. J Mol Med. 2013;91(7):791-801.

246. Hirata H, Hinoda Y, Shahryari V, Deng G, Nakajima K, Tabatabai ZL, Ishii N, Dahiya R. Long Noncoding RNA MALAT1 Promotes Aggressive Renal Cell Carcinoma through Ezh2 and Interacts with miR-205. Cancer Res. 2015;75(7):1322-1331.

247. Chen X, He L, Zhao Y, et al. Malat1 regulates myogenic differentiation and muscle regeneration through modulating MyoD transcriptional activity. Cell Discov. 2017;3:17002. 138

248. Han X, Yang F, Cao H, Liang Z. Malat1 regulates serum response factor through miR-133 as a competing endogenous RNA in myogenesis. FASEB J. 2015;29(7):3054-3064.

249. Wang X, Li M, Wang Z, Han S, Tang X, Ge Y, Zhou L, Zhou C, Yuan Q, Yang M. Silencing of Long Noncoding RNA MALAT1 by miR-101 and miR-217 Inhibits Proliferation, Migration, and Invasion of Esophageal Squamous Cell Carcinoma Cells. J Biol Chem. 2015;290(7):3925-3935.

250. Leung AKL, Young AG, Bhutkar A, Zheng GX, Bosson AD, Nielsen CB, Sharp PA. Genome-wide identification of Ago2 binding sites from mouse embryonic stem cells with and without mature microRNAs. Nat Struct Mol Biol. 2011;18(2):237-244.

251. Han X, Yang F, Cao H, Liang Z. Malat1 regulates serum response factor through miR-133 as a competing endogenous RNA in myogenesis. FASEB J. 2015;29(7):3054-3064.

252. Zhang B, Arun G, Mao YS, Lazar Z, Hung G, Bhattacharjee G, Xiao X, Booth CJ, Wu J, Zhang C, Spector DL. The lncRNA Malat1 is dispensable for mouse development but its transcription plays a cis-regulatory role in the adult. Cell Rep. 2012;2(1):111-123.

253. Eißmann M, Gutschner T, Hämmerle M, Günther S, Caudron-Herger M, Groß M, Schirmacher P, Rippe K, Braun T, Zörnig M, Diederichs S. Loss of the abundant nuclear non-coding RNA MALAT1 is compatible with life and development. RNA Biol. 2012;9(8):1076-1087.

254. Nakagawa S, Ip JY, Shioi G, Tripathi V, Zong X, Hirose T, Prasanth K V. Malat1 is not an essential component of nuclear speckles in mice. RNA. 2012;18(8):1487- 1499.

255. Chen X, He L, Zhao Y, et al. Malat1 regulates myogenic differentiation and muscle regeneration through modulating MyoD transcriptional activity. Cell Discov. 2017;3:17002.

256. Arun G, Diermeier S, Akerman M, Chang KC, Wilkinson JE, Hearn S, Kim Y, MacLeod AR, Krainer AR, Norton L, Brogi E, Egeblad M, Spector DL. Differentiation of mammary tumors and reduction in metastasis upon Malat1 lncRNA loss. Genes Dev. 2016;30(1):34-51.

257. Peters T, Hermans-Beijnsberger S, Beqqali A, Bitsch N, Nakagawa S, Prasanth K V, de Windt LJ, van Oort RJ, Heymans S, Schroen B. Long Non-Coding RNA Malat-1 Is Dispensable during Pressure Overload-Induced Cardiac Remodeling and Failure in Mice. PLoS One. 2016;11(2):e0150236.

258. Levental KR, Yu H, Kass L, Lakins JN, Egeblad M, Erler JT, Fong SFT, Csiszar K, Giaccia A, Weninger W, Yamauchi M, Gasser DL, Weaver VM. Matrix crosslinking forces tumor progression by enhancing integrin signaling. Cell. 2009;139(5):891- 906.

139

259. Butcher DT, Alliston T, Weaver VM. A tense situation: forcing tumour progression. Nat Rev Cancer. 2009;9(2):108-122.

260. Lokmic Z, Musyoka J, Hewitson TD, Darby IA. Hypoxia and hypoxia signaling in tissue repair and fibrosis. Int Rev Cell Mol Biol. 2012;296:139-185.

261. Chaudhuri O, Koshy ST, Branco da Cunha C, Shin J-W, Verbeke CS, Allison KH, Mooney DJ. Extracellular matrix stiffness and composition jointly regulate the induction of malignant phenotypes in mammary epithelium. Nat Mater. 2014;13(10):970-978.

262. Paszek MJ, Zahir N, Johnson KR, Lakins JN, Rozenberg GI, Gefen A, Reinhart- King C a, Margulies SS, Dembo M, Boettiger D, Hammer D a, Weaver VM. Tensional homeostasis and the malignant phenotype. Cancer Cell. 2005;8(3):241- 254.

263. Hu L, Wu Y, Tan D, Meng H, Wang K, Bai Y, Yang K. Up-regulation of long noncoding RNA MALAT1 contributes to proliferation and metastasis in esophageal squamous cell carcinoma. J Exp Clin Cancer Res. 2015;34:7.

264. Yang M-H, Hu Z-Y, Xu C, Xie L-Y, Wang X-Y, Chen S-Y, Li Z-G. MALAT1 promotes colorectal cancer cell proliferation/migration/invasion via PRKA kinase anchor protein 9. Biochim Biophys Acta. 2015;1852(1):166-174.

265. Yamada K, Kano J, Tsunoda H, Yoshikawa H, Okubo C, Ishiyama T, Noguchi M. Phenotypic characterization of endometrial stromal sarcoma of the uterus. Cancer Sci. 2006;97(2):106-112.

266. Guffanti A, Iacono M, Pelucchi P, et al. A transcriptional sketch of a primary human breast cancer by 454 deep sequencing. BMC Genomics. 2009;10:163.

267. Lai M, Yang Z, Zhou L, Zhu Q, Xie H, Zhang F, Wu L, Chen L, Zheng S. Long non-coding RNA MALAT-1 overexpression predicts tumor recurrence of hepatocellular carcinoma after liver transplantation. Med Oncol. 2012;29(3):1810- 1816.

268. Levental KR, Yu H, Kass L, Lakins JN, Egeblad M, Erler JT, Fong SFT, Csiszar K, Giaccia A, Weninger W, Yamauchi M, Gasser DL, Weaver VM. Matrix crosslinking forces tumor progression by enhancing integrin signaling. Cell. 2009;139(5):891- 906.

269. Butcher DT, Alliston T, Weaver VM. A tense situation: forcing tumour progression. Nat Rev Cancer. 2009;9(2):108-122.

270. Lokmic Z, Musyoka J, Hewitson TD, Darby IA. Hypoxia and hypoxia signaling in tissue repair and fibrosis. Int Rev Cell Mol Biol. 2012;296:139-185.

271. Zhou X, Liu S, Cai G, Kong L, Zhang T, Ren Y, Wu Y, Mei M, Zhang L, Wang X. Long Non Coding RNA MALAT1 Promotes Tumor Growth and Metastasis by inducing Epithelial-Mesenchymal Transition in Oral Squamous Cell Carcinoma. Sci Rep. 2015;5:15972. 140

272. Winston GJ, Palmas W, Lima J, Polak JF, Bertoni AG, Burke G, Eng J, Gottesman R, Shea S. Pulse Pressure and Subclinical Cardiovascular Disease in the Multi-Ethnic Study of Atherosclerosis. Am J Hypertens. 2013;26(5):636-642.

273. Vlachopoulos C, Aznaouridis K, Stefanadis C. Prediction of Cardiovascular Events and All-Cause Mortality With Arterial Stiffness. J Am Coll Cardiol. 2010;55(13):1318-1327.

274. Yang C, Tibbitt MW, Basta L, Anseth KS. Mechanical memory and dosing influence stem cell fate. Nat Mater. 2014;13(6):645-652.

275. Li CX, Talele NP, Boo S, Koehler A, Knee-Walden E, Balestrini JL, Speight P, Kapus A, Hinz B. MicroRNA-21 preserves the fibrotic mechanical memory of mesenchymal stem cells. Nat Mater. 2016;16(3):379-389.

276. Murphy AM, Wong AL, Bezuhly M. Modulation of angiotensin II signaling in the prevention of fibrosis. Fibrogenesis Tissue Repair. 2015;8(1):7.

277. Janić M, Lunder M, Šabovič M. Arterial Stiffness and Cardiovascular Therapy. Biomed Res Int. 2014;2014:1-11.

278. Peacock EE, Madden JW. Administration of beta-aminopropionitrile to human beings with urethral strictures: A prelimary report. Am J Surg. 1978;136(5):600- 605.

279. Ren W, Liu Y, Wang X, Jia L, Piao C, Lan F, Du J. β-Aminopropionitrile monofumarate induces thoracic aortic dissection in C57BL/6 mice. Sci Rep. 2016;6(1):28149.

280. Nohria A, Grunert ME, Rikitake Y, Noma K, Prsic A, Ganz P, Liao JK, Creager MA. Rho kinase inhibition improves endothelial function in human subjects with coronary artery disease. Circ Res. 2006;99(12):1426-1432.

281. Dong M, Yan BP, Liao JK, Lam Y-Y, Yip GWK, Yu C-M. Rho-kinase inhibition: a novel therapeutic target for the treatment of cardiovascular diseases. Drug Discov Today. 2010;15(15-16):622-629.

282. Chan SY, Loscalzo J. Pulmonary Arterial Hypertension. In: Vascular Medicine: A Companion to Braunwald’s Heart Disease. Elsevier; 2013:667-686.

283. Vicari RM, Chaitman B, Keefe D, Smith WB, Chrysant SG, Tonkon MJ, Bittar N, Weiss RJ, Morales-Ballejo H, Thadani U. Efficacy and Safety of Fasudil in Patients With Stable Angina. J Am Coll Cardiol. 2005;46(10):1803-1811.

141