Semaphorin3f As a Spatial Regulator of Embryogenesis

University of Calgary PRISM: University of Calgary's Digital Repository

Graduate Studies The Vault: Electronic Theses and Dissertations

2019-01-22 Semaphorin3f as a Spatial Regulator of Embryogenesis

Halabi, Rami

Halabi, R. (2019). Semaphorin3f as a Spatial Regulator of Embryogenesis (Unpublished doctoral thesis). University of Calgary, Calgary, AB. http://hdl.handle.net/1880/109507 doctoral thesis

University of Calgary graduate students retain copyright ownership and moral rights for their thesis. You may use this material in any way that is permitted by the Copyright Act or through licensing that has been assigned to the document. For uses that are not allowable under copyright legislation or licensing, you are required to seek permission. Downloaded from PRISM: https://prism.ucalgary.ca UNIVERSITY OF CALGARY

Semaphorin3f as a Spatial Regulator of Embryogenesis

Rami Halabi

A THESIS

SUBMITTED TO THE FACULTY OF GRADUATE STUDIES

IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE

DEGREE OF DOCTOR OF PHILOSOPHY

GRADUATE PROGRAM IN NEUROSCIENCE

CALGARY, ALBERTA

JANUARY, 2019

Abstract

During embryogenesis, cells integrate both spatial and temporal information from their surroundings to influence proliferation, migration, differentiation and physiological functions.

Understanding the molecular mechanisms which confer spatial identity is essential to our understanding of tissue development and human disease. In this thesis I explore multiple roles for the secreted chemotactic ligand Semaphorin3f (Sema3f) in different biological contexts. Using zebrafish (Danio rerio) as a model I take advantage of the duplicated genome to study loss of function of both orthologs, Sema3fa and Sema3fb, in discrete contexts due to their differential expression. First, I show that in the eye Sema3fa produced by progenitors is necessary for the generation of amacrine cells within the temporal retina and the spatially organized transcriptome of stem cells in the ciliary marginal zone (CMZ). Second, I define an endogenous role of Sema3fa to maintain the avascularity of the neural retina and refine the branch pattern of intraocular vessels.

Loss of Sema3fa results in the pathologic angiogenesis of leaky blood vessels into the neural retina.

Last, I unveil a role for Sema3fb produced by cardiomyocyte progenitors in the differentiation of the ventricle of the developing heart. Overall, my work provides the first evidence of a Sema3 involved in retinal progenitor cell and cardiomyocyte differentiation, and elucidates the endogenous role of Sema3fa as a negative regulator of retinal blood vessels in the embryo and adult. My data exemplifies the necessity of spatial information conferred by a single chemotactic molecule, Sema3f, to impact differentiation and cellular biology.

Acknowledgements

I thank Sarah McFarlane for all her help and mentorship over the last five years. You are a great supervisor and I appreciate your support in developing my own skills, both inside and outside of the lab.

Thank you to my committee members, Sarah Childs and Peng Huang, for your input during my PhD. Thank you to both my external examiners Jeff Biernaskie and Vince Tropepe for your readiness to be part of the examination committee.

I thank all the members of the McFarlane lab for their support, guidance and contributions over the years. Thank you to Carrie Hehr for all your assistance in running experiments and maintaining the lab so we can all work at full capacity. Thank you Karen Atkinson-Leadbeater for your career guidance, and Gabriel Bertolesi for all your help in consolidating experimental ideas.

I thank Paula Cechmanek for all the great conversations, both scientific and life related, and for the excellent collaboration we had by the end of our respective PhDs. Thank you Jonathan Yang and Zachary Nurcombe - our chats in the “cove” will be missed! Amira Kalifa and Risa Mori-

Kreiner, I appreciate all the help you have given me this last little while and the passion you have shown for your work is brightening.

I also extend my thanks to all those that have made this PhD memorable in many ways:

Debbie Kurrasch, Cairine Logan, Lian Willets, Dierdre Lobb, Cam Teskey, Andy Bulloch, Kelly

Cook and Carol Schuurmans. Thank you for the conversations, the advice, and the opportunities you have afforded me over the years. To my entire family – Samar, Riad, Rabih, Marwan, Amal,

Stella, Alexander, Anabella, Firas, Rosa and Sarah – thank you for supporting me and asking what experiments I am in the middle of despite not understanding a thing of what I said. Thank you to

the administrative and education personnel – Tianni Song, Lesley Towill and Jason Ng - whom I have worked closely with on side projects.

Thank you to all my friends, both old and new, you have made my time in Calgary wonderful. Thank you to Charlene Watterston, the honorary McFarlane Lab Member - you are an amazing friend and collaborator. Thank you to Rami Abu Zeinab – you are an excellent friend and have been great help in preparing me for my next steps post PhD. Thank you to Raquel Cruz for everything you have done and continue to do for me. And to Matt Lemieux, Becky Klein, Cody

Sahlin, Yamile Jasaui, Tooka Collette, Amlish Munir, Julie Dang, Alexandra Sull, Judith Sull and

Graham Ring for always being there when I needed a break from writing. And thank you to all others I have not directly mentioned here - the last 5 years would not have been the same without you.

Lastly, I thank the Department of Neuroscience, ACHRI, Hotchkiss Brain Institute,

REALISE, AIHS, BFF and NSERC. Without your support and without the opportunities you have afforded me I would not be where I am now. Thank you.

All writing, experiments, quantifications, images and figures were completed by myself, with the following exceptions:

Figures 3.20 and 5.20 – explants of tissues were done by Carrie Hehr. Plastic sectioning and in situ hybridization of several genes were also carried out by Carrie Hehr.

Figures 5.23 and 5.24 – figures made by Dr. Paula Cechmanek

Figures 5.11, 5.16, and 5.19 – confocal images taken by Dr. Paula Cechmanek.

Figures 4.3, 4.4, 4.5, 4.6, 4.7 – confocal images taken by Charlene Watterston.

Dedication

To my Mother, for her continued encouragement.

Table of Contents

Abstract ...... 1 Acknowledgements ...... 2 Dedication ...... 4 Table of Contents ...... 5 List of Tables ...... 9 List of Figures and Illustrations ...... 10 List of Symbols, Abbreviations and Nomenclature ...... 14

CHAPTER ONE: INTRODUCTION ...... 17 1.1 Spatial Cues in Development ...... 17 1.1.1 Neurovascular Guidance ...... 18 1.1.2 Chemotactic Molecules involved in Cellular Guidance ...... 19 1.1.3 Sema3 Signalling During Development ...... 23 1.1.4 Spatially Regulated Tissues in Development ...... 25 1.2 Development of the Zebrafish Neural Retina ...... 28 1.2.1 Organization of the neural retina ...... 28 1.2.2 Development of the zebrafish retina and lens ...... 30 1.2.3 Neurogenesis of the zebrafish retina ...... 31 1.2.4 Neuronal differentiation in the zebrafish retina ...... 32 1.2.5 Development of the RGC axon projection ...... 34 1.2.6 Class 3 Semas in Retinal Development and Injury ...... 35 1.3 Development of the Vasculature of the Zebrafish Retina ...... 35 1.3.1 Hyaloid and Retinal Vasculature Development ...... 36 1.3.2 Choroidal Vasculature Development ...... 38 1.3.3 Blood-Retina-Barrier ...... 40 1.3.4 Age Related Macular Degeneration ...... 40 1.3.5 Class 3 Semas in Orbital Vessel Dynamics ...... 42 1.4 Development of the Zebrafish Cardiovascular System ...... 43 1.4.1 Specification and Differentiation of the Zebrafish Heart ...... 43 1.4.2 Cardiac Chamber Morphogenesis ...... 46 1.4.3 Cardiac Neural Crest and Second Heart Field ...... 47 1.4.4 Class 3 Semas in Heart Development ...... 48 1.5 Hypothesis and Aims ...... 49 1.5.1 General Hypothesis ...... 49 1.5.2 Specific Aim 1. To elucidate the role of Sema3fa in progenitor cells of the retina during neurogenesis...... 50 1.5.3 Specific Aim 2. To investigate an endogenous role for Sema3fa on orbital vasculature in the embryo and adult...... 50 1.5.4 Specific Aim 3. To determine the role of Sema3fb in cardiomyocyte differentiation during cardiogenesis...... 50

CHAPTER TWO: MATERIALS AND METHODS ...... 52 2.1 Animals ...... 52

2.1.1 Zebrafish strains and maintenance ...... 52 2.2 Genetic Manipulation ...... 53 2.2.1 Establishment of sema3fa and sema3fb mutant lines ...... 53 2.3 Gene Expression Detection ...... 55 2.3.1 RNA isolation ...... 55 2.3.2 cDNA Synthesis ...... 55 2.3.3 Real Time Quantitative PCR (RT-qPCR) ...... 58 2.3.4 in situ hybridization ...... 58 2.3.5 Immunolabelling ...... 59 2.3.6 RNA sequencing (RNA seq) ...... 59 2.4 Functional Analyses ...... 60 2.4.1 Intracardial Injections ...... 60 2.4.2 Heart Function Analysis ...... 61 2.4.3 Light exposure to detect c-fos ...... 61 2.4.4 Zebrabox Visual Function Assay ...... 62 2.5 Histology ...... 62 2.5.1 Cryosections ...... 62 2.5.2 Plastic sections ...... 62 2.5.3 Hematoxylin and Eosin Staining ...... 63 2.5.4 Heart Measurements ...... 63 2.5.5 Retinal Cell Counts ...... 63 2.5.6 Retinal Flat Mounts ...... 64 2.6 Imaging ...... 64 2.6.1 Stereoscope ...... 64 2.6.2 Confocal ...... 64 2.7 Statistics ...... 65

CHAPTER THREE: SEMA3FA SIGNALLING DURING RETINOGENESIS ...... 66 3.1 Introduction ...... 66 3.2 Results ...... 68 3.2.1 sema3fa is expressed by RPCs throughout retinogenesis ...... 68 3.2.2 Sema3fa is necessary for initial retinal cell specification ...... 71 3.2.2.1 Generation of the sema3fa genetic mutant ...... 71 3.2.2.2 Sema3fa deficient retina exhibit an anormal retinal cell specification program ...... 71 3.2.3 Sema3fa is involved in RGC and inner nuclear layer specification and differentiation ...... 79 3.2.3.1 Sema3fa deficient embryos have smaller temporal retinas during mid- retinogenesis...... 79 3.2.3.2 Retinal cell type differentiation is largely spared in mid-retinogenesis in sema3fa mutants...... 81 3.2.4 Sema3fa deficient embryos have reduced numbers of amacrine cells in the temporal retina ...... 89 3.2.4.1 Temporal amacrinogenesis is reduced with sema3fa deficiency ...... 89

3.2.4.2 Retinal cell type differentiation is largely spared in late-retinogenesis in sema3fa mutants...... 89 3.2.4.3 Light perception is not robustly impaired with loss of Sema3fa ...... 99 3.2.5 Sema3fa deficient embryos have changed neurogenic potentials in the ciliary marginal zone...... 99 3.2.5.1 The ciliary marginal zone undergoes transcriptional changes with loss of Sema3fa ...... 99 3.2.6 sema3fa signalling is eye tissue autonomous ...... 106 3.2.6.1 Retinal cell type specification is independent of circulating factors ...106 3.2.6.2 Sema3fa may act in a cell autonomous manner to control progenitor biology within the CMZ ...... 106 3.3 Specific Discussion ...... 108 3.4 Significance ...... 121

CHAPTER FOUR: SEMA3FA SIGNALLING IN OCULAR VASCULARIZATION .122 4.1 Introduction ...... 122 4.2 Results ...... 125 4.2.1 semaphorin3fa is expressed by the retinal pigment epithelium in development and adulthood ...... 125 4.2.2 sema3fa mutants present with increased intraocular vascularization ...... 125 4.2.2.1 Establishing a sema3fa mutant transgenic reporter line ...... 125 4.2.2.2 Early intraocular vascularization of the retina is normal ...... 127 4.2.2.3 Hyaloid vessel is compact, and adult retinal vasculature more dense .129 4.2.3 sema3fa mutants have pathologic, leaky vessels that enter the eye ...... 133 4.2.3.1 Choroid plexus undergoes neovascularization in the sema3fa mutant retina ...... 133 4.2.3.2 sema3fa -/- orbital vasculature is leaky ...... 135 4.2.4 sema3fa mutants have apparently normal retinal pigment epitheliums ...... 138 4.2.4.1 Vessel infiltration in sema3fa mutants is not due to an increase in Vegfa ...... 138 4.2.4.2 Retinal pigment epithelium is unperturbed in sema3fa mutants ...... 140 4.2.5 Sema3fa signalling on endothelial cells is likely mediated by Nrp1b ...... 143 4.3 Specific Discussion ...... 143 4.4 Significance ...... 154

CHAPTER FIVE: SEMA3FB SIGNALLING IN HEART DEVELOPMENT ...... 156 5.1 Introduction ...... 156 5.2 Results ...... 158 5.2.1 sema3fb is expressed by the cardiomyocyte progenitor cells of the developing heart ...... 158 5.2.2 Sema3fb is necessary for spatially restricted chamber differentiation ...... 160 5.2.2.1 Generation of the sema3fb genetic mutant ...... 160 5.2.2.2 Embryos that lack Sema3fb exhibit pericardial edema ...... 160 5.2.2.3 sema3fb mutants have smaller hearts that are only delayed slightly in their development...... 162

5.2.2.4 Chamber specific expression patterns are disrupted in the sema3fb mutant...... 167 5.2.2.5 sema3fb mutant hearts do not present sharp boundaries between chambers ...... 173 5.2.3 Small heart phenotype is due to a reduction in cardiomyocyte size ...... 176 5.2.3.1 Cardiomyocytes are specified and develop normally in sema3fb mutants ...... 176 5.2.3.2 Cardiomyocyte contributions from the second heart field and neural crest cell populations do not explain the sema3fb mutant heart phenotype ...... 179 5.2.3.3 Cardiomyocytes are significantly smaller in sema3fb mutants ...... 181 5.2.4 sema3fb mutants have impaired cardiac function ...... 184 5.2.4.1 Cardiac output is significantly reduced in mutants ...... 184 5.2.4.2 Valve development is apparently normal in mutants...... 188 5.2.5 sema3fb signalling is tissue autonomous ...... 188 5.2.5.1 Atrial chamber size deficit is independent of flow ...... 188 5.2.5.2 Nrp2b and Plxna3 are potential receptors mediating Sema3fb signalling ...... 190 5.1 Specific Discussion ...... 190 5.2 Significance ...... 200

CHAPTER SIX: GENERAL DISCUSSION...... 202 6.1 Sema3fa signalling primes temporal RPCs for future amacrinogenesis ...... 203 6.2 Sema3fa signalling as a novel regulator of the stem cell niche ...... 208 6.3 Mining the Sema3fa mutant ...... 212 6.4 Sema3fa signalling in CNV ...... 213 6.5 Sema3fb in the heart ...... 217 6.6 Importance of spatial signalling in development ...... 220 6.7 Concluding Remarks ...... 222

REFERENCES ...... 224

List of Tables

Table 2.1. CRISPR knock-out reagents and detection ...... 54

Table 2.2. Probe list for in situ hybridization ...... 56

Table 6.1. Differentially expressed genes in mutant Sema3fa versus wildtype embryos...... 206

List of Figures and Illustrations

Figure 1.1 The Semaphorin family and Plexin and Neuropilin receptors ...... 21

Figure 1.2 Spatial patterning in the developing nervous system ...... 27

Figure 1.3. Organization of the vertebrate eye ...... 29

Figure 1.4. Development of the zebrafish ocular vasculature ...... 37

Figure 1.5. Development of the zebrafish heart ...... 44

Figure 3.1. sema3fa, and not sema3fb, is expressed by retinal progenitor cells throughout retinogenesis ...... 70

Figure 3.2. Generation of the sema3fa genetic mutant ...... 72

Figure 3.3. sema3fa mutants are physically indistinguishable from wildtype embryos ...... 74

Figure 3.4. Retinal patterning is unaffected in sema3fa mutants ...... 75

Figure 3.5. Early embryonic patterning is normal is sema3fa mutants ...... 77

Figure 3.6. Retinal specification transcription factors are downregulated in sema3fa mutants ... 78

Figure 3.7. Temporal retinal tissue is significantly smaller in sema3fa mutants ...... 80

Figure 3.8. Terminal retinal ganglion cell differentiation occurs in sema3fa mutants in mid retinogenesis ...... 82

Figure 3.9. Amacrine and horizontal cell specification occurs in sema3fa mutants during mid retinogenesis ...... 84

Figure 3.10. Specification of retinal pigment epithelium, and inner and outer nuclear cells occurs normally in sema3fa mutants during mid retinogenesis ...... 87

Figure 3.11. Retinal progenitors undergo cell cycle progression ...... 88

Figure 3.12. sema3fa mutant eyes are the same size as wildtype eyes ...... 90

Figure 3.13. The temporal amacrine cell layer has fewer cells in sema3fa mutants ...... 91

Figure 3.14. Temporal amacrinogenesis and inner plexiform layer development is reduced in sema3fa mutants ...... 94

Figure 3.15. RGC layer forms normally in sema3fa mutants ...... 96

Figure 3.16. Presence of differentiation markers of the retinal pigment epithelium, and inner and outer nuclear cells is normal in sema3fa mutants ...... 97

Figure 3.17. sema3fa mutants retain visual function ...... 100

Figure 3.18. Retinal progenitor cells of the ciliary marginal zone maintain cell cycle markers in sema3fa mutants ...... 102

Figure 3.19. The CMZ of the sema3fa mutants expands spatially restricted transcriptional profiles ...... 104

Figure 3.20. Change in retinogenesis are not due to loss of Sema3fa in tissues outside of the eye ...... 107

Figure 3.21. Nrp2b and Plxna3 are likely receptors for Sema3fa signalling ...... 109

Figure 3.22. Summary of sema3fa roles in the neural retina ...... 111

Figure 3.23. Summary of RNAseq differential gene expression analysis on sema3fa mutants . 116

Figure 4.1. sema3fa is expressed by the retinal pigment epithelium ...... 126

Figure 4.2. Initial entry of the hyaloid vessel into the eye in normal in sema3fa mutants ...... 128

Figure 4.3. Initial intraocular vascular development is normal during embryonic development 130

Figure 4.4. sema3fa deficient embryos present with abnormal intraocular blood vessels ...... 131

Figure 4.5. Increased vascularization of the retina is observed into adulthood ...... 132

Figure 4.6. sema3fa deficient embryos present with choroid plexus overgrowth ...... 134

Figure 4.7. sema3fa deficient embryos present with abnormal extraocular blood vessels ...... 136

Figure 4.8. Sema3fa deficiency permits entry of leaky vessels into the neural retina ...... 137

Figure 4.9. Vessel infiltration persists in the adult sema3fa-/- zebrafish ...... 139

Figure 4.10. vegfa is not induced in the retina of sema3fa mutants ...... 141

Figure 4.11. The RPE appears physically intact in the sema3fa mutant ...... 142

Figure 4.12. The RPE of wildtype and sema3fa mutant eyes express RPE specific genes ...... 144

Figure 4.13. Sema3fa likely signals through Nrp1b...... 145

Figure 4.14. Summary and working model of roles for Sema3fa in regulating vascularization of the neural retina ...... 147

Figure 5.1. sema3fb is expressed by the developing heart ...... 158

Figure 5.2. Generation of the sema3fb genetic mutant ...... 161

Figure 5.3. sema3fb mutants have cardiac edema ...... 163

Figure 5.4. Heart tube extension is disrupted in sema3fb mutant embryos ...... 165

Figure 5.5. Ventricle chamber extension is disrupted in sema3fb mutant embryos ...... 166

Figure 5.6. Ventricles are smaller in sema3fb mutants ...... 168

Figure 5.7. Atrial chamber development is disrupted in sema3fb mutant embryos ...... 169

Figure 5.8. Cardiac chamber specific gene expression is disrupted in sema3fb mutants ...... 170

Figure 5.9. Ventricle identity is disrupted in sema3fb mutants ...... 172

Figure 5.10. Disruption of ventricle identity is sustained in sema3fb mutants...... 174

Figure 5.11. Disruption of chamber specific protein expression with Sema3fb loss ...... 175

Figure 5.12. Gross patterning of the anterior lateral plate mesoderm is not disrupted in sema3fb mutants ...... 177

Figure 5.13. Heart tube extension defect is not due to global developmental delay...... 178

Figure 5.14. sema3fb mutant hearts show no change in proliferation or apoptosis ...... 180

Figure 5.15. Cardiac neural crest cell and second heart field cardiomyocyte contributions are unaffected in sema3fb mutants ...... 182

Figure 5.16. Ventricular cardiomyocytes are smaller in sema3fb mutants ...... 183

Figure 5.17. Ventricular wall thickness is reduced in sema3fb mutants ...... 185

Figure 5.18. sema3fb mutant embryos have reduced cardiac function ...... 186

Figure 5.19. Atrial-ventricular valves develop in the sema3fb mutant ...... 189

Figure 5.20. Atrial size deficit in sema3fb mutants occurs independently of decreased cardiac output ...... 191

Figure 5.21. Sema3fb likely signals through Nrp2b and Plxna3 ...... 192

Figure 5.22. Summary of sema3fb mutant expression analyses ...... 194

Figure 5.23. Working model of Sema3fb signalling in the heart ...... 195

Figure 6.1. Ectopic expression of DM-GRASP in sema3fa mutant retinas...... 209

List of Symbols, Abbreviations and Nomenclature

° C degrees Celsius % percent µM micromolar aa amino acid atoh7 atonal homolog 7 AMD angiogenin Ang Age-related macular degeneration AV atrioventricular AVC atrioventricular canal BCIP 5-bromo-4-chloro-3-indolyl-phosphate BBB blood brain barrier bHLH basic helix-loop-helix BMP bone morphogenetic protein bp base pair BRB blood retina barrier Cas9 CRISPR associated protein 9 ccnd1 Cyclin d1 cdkn1c P57kip2 CF choroid fissure CMZ ciliary marginal zone CNCC cranial neural crest cell CNC cardiac neural crest CNS central nervous system CNV choroidal neovascularization Crip2 cysteine-rich intestinal protein 2 CRISPR clustered regularly interspaced short palindromic repeats Cxcl12 C-X-C motif chemokine 12 Cxcr4 C-X-C chemokine receptor 4 DIG digoxygenin DCC deleted in colorectal cancer Dll4 delta like 4 dpf days post fertilization ECM extracellular matrix Epo erythropoietin ErbB epidermal growth factor receptor FITC fluorescein isothiocyanate Fgf fibroblast growth factor GAP GTPase activating protein h hour hpf hours post fertilization HuC ELAV like RNA binding proteins IGF1 insulin-like growth factor 1 Irx1a iroquois 1a

ISH in situ hybridization L liter Ltbp3a latent tgf beta binding protein 3a mg Milligram MHC1 Myosin heavy chain I ml millilitre Myh6 myosin heavy chain 6 Myh7 myosin heavy chain 7 NBT nitro blue tetrazolium NCC neural crest cell NFAT nuclear factor of activated T cells ng nanogram Nrp neuropilin n number PCNA proliferating cell nuclear antigen PCR polymerase chain reaction PDGF platelet derived growth factor pdgfra platelet derive growth factor receptor a PIGF placental growth factor PFA paraformaldehyde pHH3 phosphorylated histone H3 Plxn plexin Ptf1a pancreas transcription factor 1a PTU 1-phenyl-2-thiourea RA retinoic acid RGC retinal ganglion cell rlbp1b retinaldehyde binding protein 1b RPC retinal progenitor cell RNA ribonucleic acid RPE retinal pigment epithelium rpe65a retinal pigment epithelium-specific protein 65a RT-qPCR real time quantitative PCR Sema semaphorin Shh sonic hedgehog SHF second heart field sgRNA single guide RNA ss somite stage Tbx5a T-box transcription factor Tg transgenic TgfB transforming growth factor beta tnnt2a cardiac troponin t2a TrB2 thyroid hormone receptor TL Tupfel Long Fin TUNEL terminal deoxynucleotidyl transferase dUTP nick end labeling Unc5 uncoordinated 5

Vegf vascular endothelial growth factor Vegfr vascular endothelial growth factor receptor Vldlr very low-density lipid-receptor qPCR quantitative real-time polymerase chain reaction ZO-1 zona-occludens 1

Chapter One: Introduction

Proprioception is the “sense of self”. On an organismal level it refers to the relative position of body parts with respect to each other in space in order to perform a movement. Cells also need to know where they are in space for purposes ranging from proliferation, migration, differentiation and physiological function. But how do cells receive spatial cues? One such mechanism is through the utilization of secreted chemotactic molecules. I have identified multiple roles of the secreted ligand Semaphorin3f (Sema3f) as a spatial information cue across different tissue systems (retina, vasculature, heart). Understanding underlying molecular mechanisms that control spatial signalling is quintessential for understanding human disease (Kholodenko, 2006) and efforts in cellular replacement therapies (Pera and Tam, 2010).

This thesis will look at the role of Sema3f on retinal progenitor cells undergoing neurogenesis and differentiation in Chapter 3. In Chapter 4, I look at the role of Sema3f in providing cues to tissue associated vasculature, and in Chapter 5 I demonstrate that Sema3f can specifically drive differentiation of the ventricle cardiac chamber. Zebrafish express duplicated orthologs of human SEMA3F - sema3fa and sema3fb (Yu and Moens, 2005).

1.1 Spatial Cues in Development

In vertebrates, directed cell migration of a cell or cellular process (e.g. axon) along a predetermined pathway is a classic aspect of spatial signalling. Such pathways are established through chemical gradients of extracellular factors (chemotaxis) (Van Haastert and Devreotes,

2004), adhesion molecules expressed by cells or in the extracellular matrix (haptotaxis), electric field establishment (electrotaxis) and mechanic forces (durotaxis) (Thiery, 1984). Axonal growth cones and tip cells of the nervous and vascular systems, respectively, are responsible for interpreting such spatial signals as attractive or repulsive via changes in their actin cytoskeleton.

1.1.1 Neurovascular Guidance

The tip of a growing axon of a neuron has a specialized, hand-like structure called the growth cone (Tessier-Lavigne and Goodman, 1996). Developing neurons’ axons and dendrites use this structure to navigate through the extracellular matrix (ECM) in order to find their appropriate final resting place and synaptic targets. The growth cone has a dynamic cytoskeletal structure that extends and retracts finger-like filopodia, through F-actin and microtubule polymerization and depolymerization, to assess the environment for molecular guidance cues.

Vascularization of the central nervous system (CNS) occurs primarily through angiogenesis (Fruttiger, 2007; Gariano and Gardner, 2005; Provis, 2001). Angiogenesis is the sprouting of new vessels from pre-existing ones. Endothelial cells (EC) at the distal end of sprouts are called tip cells, while cells that make up the remainder of the vessel are referred to as stalk cells. Tip cells can be identified morphologically and by gene expression profiles as distinct from the stalk cells and allow for the guidance competency of the forming vessel (Siekmann and

Lawson, 2007). Like the axonal growth cone, tip cells use filopodial extensions to make sense of the repulsive and attractive growth cues that exist in the environment. To ensure proper vascular patterning, the number of tip cells is highly regulated by Delta Like 4 (Dll4)-Notch signalling

(Siekmann and Lawson, 2007; Phng and Gerhardt, 2009). Following exposure to the potent proangiogenic protein, vascular endothelial growth factor (Vegf), all cells upregulate Dll4. ECs with the highest Dll4 expression adopt a tip cell fate by increasing their expression of the Vegf receptor

(Vegfr), which makes them more responsive to environmental Vegf. Additionally, Dll4-Notch signalling between stalk cells drives the down-regulation of Vegfr to maintain their quiescence

(Phng and Gerhardt, 2009; Jakobsson et al., 2010).

1.1.2 Chemotactic Molecules involved in Cellular Guidance

Four main families of guidance cues are known: Ephrins, Netrins, Slits and Semaphorins

(Sema). Although identified originally as playing roles in axon guidance, several members within these families also participate in guided angiogenesis (Larrivee et al., 2009). In addition to these classical molecules, growth factors, chemokines and morphogens can also serve as guidance molecules (Zacchigna et al., 2008).

The Ephrin family of ligands are membrane bound proteins that engage in bidirectional, generally repulsive, signalling through Eph receptors (Pitulescu and Adams, 2010). Ephrins are divided into two subclasses; subclass A ligands (EphrinA1-A5) are tethered to the cell membrane while subclass B ligands (EphrinB1-B3) contain a transmembrane domain. Eph receptors constitute the largest family of receptor tyrosine kinases with thirteen members. Similar to the ligand, they are divided into two subclasses; subclass A (EphA1-8) and subclass B (EphB1-B4,

B6) (Kullander and Klein, 2002) . Their function has been well-studied in the retinotopic mapping of the axons of retinal ganglion cells (RGCs) of the eye to the optic tectum and superior colliculus.

Those RGCs located on the temporal-nasal axis of the retina use EphrinA-EphA to map to the tectal anterior-posterior axis, whereas those along the dorsal-ventral axis use EphrinB/EphB to map onto the medial-lateral axis (Feldheim and O’Leary, 2010; Suetterlin et al., 2012) . In vessel sprouting, signalling through EphrinB2 controls tip cell behaviour by positively regulating Vegfr2 trafficking to the membrane, thus influencing vessel guidance (Sawamiphak et al., 2010; Wang et al., 2010b).

The Netrin family of ligands is composed of secreted, matrix bound proteins that signal through two receptor families – Deleted in Colorectal Cancer (DCC) and Uncoordinated 5 (Unc5).

There are three Netrins in mammals (Netrin1/2/4) (Park et al., 2005). The Unc5b receptor is

enriched in the endothelial tip cell (Lu et al., 2004), and the cells form aberrant filopodia in the embryonic lethal Unc5 knockout mouse. Netrin1-Unc5 signalling is likely repulsive in that Netrin1 overexpression inhibits vessel invasion (Adams and Eichmann, 2010). In the nervous system,

Netrin often signals through DCC and is attractive. For instance, Netrin1 expressed at the optic nerve head of the mouse retina is used to attract DCC positive RGC axons for exit through the choroid fissure (Deiner et al., 1997). Similarly, Netrin1 has been identified as a chemoattractant in angiogenesis (Park et al., 2004; Wilson et al., 2006).

Slits are large, secreted glycoproteins (Brose et al., 1999; Kidd et al., 1999) that play roles in axon repulsion and angiogenesis (Legg et al., 2008). There are three Slits in mammals, Slit1-3

(Brose and Tessier-Lavigne, 2000). The Roundabout (Robo) family of transmembrane protein receptors mediate the actions of Slit. Robo1-3 are predominantly found in the nervous system, while Robo4 is specifically found in vascular ECs (Bedell et al., 2005). Robo4 is expressed predominantly in the stalk and not tip cells of mouse retinal vessels, suggesting that the receptor is not required for vascular guidance. Indeed, Robo4 may inhibit Slit2-Robo1 proangiogenic signalling in order to maintain tip to stalk cell ratios (Jones et al., 2008; Morlot et al., 2007; Sheldon et al., 2009; Wang et al., 2003). In the nervous system, generally Slits bind Robos to mediate a repulsive signal for growing axons (Nguyen-Ba-Charvet and Chedotal, 2002).

Semas are a family of secreted, membrane bound and transmembrane proteins involved in axon guidance and angiogenesis. The more than twenty proteins identified are categorized into eight classes of Semas (Sema1-8), of which Sema3, 4, 5,s 6 and 7 are found in vertebrates (Figure

1.1). Class 3 Semas are secreted and capable of long range diffusion, while the remainder are membrane associated or bound (Unified nomenclature for the semaphorins/collapsins. Semaphorin nomenclature committee. 1999; Raper, 2000). Two receptor families are capable of mediating

Figure 1.1 The Semaphorin family and Plexin and Neuropilin receptors. Figure depicting the

8 classes of Semaphorins (Semas) and the Plexin (Plxn) and Neuropilin (Nrp) receptors through which they signal. Of note, the SEMA domain on Semas physically interact with the SEMA domain on Plxn and, in the case of Sema3s, the CUB domain on Nrp. Adapted from (Kruger et al.,

2005; Takamatsu and Kumanogoh, 2012).

Sema signalling; the transmembrane proteins Plexin (Plxn) and Neuropilin (Nrp) (Figure 1.1). The

Plxn family has nine members divided into four classes, PlxnA-D, while Nrps have only two members, Nrp1/2 (Tamagnone et al., 1999; Yazdani and Terman, 2006). While Semas normally signal through Plxns directly, Sema3s require binding to the co-receptor Nrp, which then interacts with Plxn to elicit an intracellular signalling cascade (Carmeliet and Tessier-Lavigne, 2005;

Dickson, 2002; Kolodkin et al., 1997). An exception is Sema3e that can directly bind and activate a Plxn receptor.

In addition to the typical guidance molecules utilized by the nervous and vascular systems, a number of other factors have also been implicated. These include growth factors such as

Fibroblast growth factor (Fgf) and Vegf, as well as the C-X-C chemokine receptor 4 (Cxcr4)

(Zacchigna et al., 2008). For example, transgenic mice expressing dominant negative (DN) FGFR1 in the retinal pigment epithelium (RPE) fail to form normal retinal vasculature (Rousseau et al.,

2000, 2003). Fgf signalling also affects RGC axon guidance by promoting repellent guidance cue expression in the Xenopus forebrain (Atkinson-Leadbeater et al., 2010). Vegf is capable of signalling through both Vegfr and Nrp, and exists as a number of functionally independent isoforms that differ in their affinity to interact with ECM proteoglycan heparan sulfate, which can be necessary for Vegf function (Leung et al., 1989; Tischer et al., 1991). In mice, VegfA has been implicated in retinal vascularization, with mRNA found in the developing lens (Gerhardt et al.,

2003; Mitchell et al., 2006), as well as in the guidance of RGC axons (Erskine et al., 2011). Retinal vessel tip cells express Cxcr4 and respond to the ligand C-X-C motif chemokine 12 (Cxcl12) secreted from the arteries proximal to the optic nerve and retinal astrocytes which line the inner limiting membrane (Strasser et al., 2010). Loss of Cxcr4 signalling results in reduced tip cell

activation and therefore a reduced vascular density in the retina (Unoki et al., 2010). Knockdown of Cxcr4 in zebrafish results in misguided RGC axons (Li et al., 2005).

1.1.3 Sema3 Signalling During Development

Signal transduction of Sema binding to its receptors is generally through manipulation of the actin cytoskeleton, although other downstream targets have been described (Cagnoni and

Tamagnone, 2014; Hota and Buck, 2012; Siebold and Jones, 2013; Tran et al., 2007; Yoshida,

2012; Zhou et al., 2011). Binding of a Sema to Plxn can lead to the activation of the Plxn GTPase- activating protein (GAP) domain, initiating a cascade of signalling through protein kinases,

GTPases and cytoskeletal associated proteins (Cagnoni and Tamagnone, 2014; Hota and Buck,

2012; Tran et al., 2007; Zhou et al., 2011). All Plxn classes exhibit GAP domain activation, which reduces levels of active Ras and Rap GTPases, which in turn reduce the integrity of the cytoskeleton (Mizumoto and Shen, 2013; Oinuma et al., 2004; Wang et al., 2012b; Worzfeld et al., 2014; Yang and Terman, 2012). Outside of cytoskeletal manipulation, Sema/Plxn also control cell-cell and cell-substrate adhesion through downstream Ras inactivation, which negatively regulates integrin-mediated cell adhesion (Kinbara et al., 2003; Oinuma et al., 2006; Toyofuku et al., 2005).

Sema binding to co-receptors, that include Nrps as well as other molecules, modulates downstream Plxn signalling (Castellani et al., 2000; Falk et al., 2005). The molecular mechanisms behind co-receptor modulation of signalling are largely unexplored. The complexity of Sema/Plxn signalling is not only attributed to co-receptor modulation, but also interactions between different

Semas. Semas of different classes can bind the same Plxn class, and Semas of a single class can bind to the same receptor (competitive ligand interactions) and elicit distinct downstream signalling pathways (Fukuda et al., 2013; Hayashi et al., 2012; Takahashi et al., 1998; Wu et al.,

2011). For example, mechanosensory afferent innervation in the Drosophila embryonic CNS is largely regulated by Sema2a and Sema2b, which signal through the same PlxnB receptor (Fukuda et al., 2013; Hayashi et al., 2012; Takahashi et al., 1998; Wu et al., 2011). Restricted expression of sema2b along the longitudinal tracts of the afferents suggests Sema2b may serve as an attractive cue to promote neurite fasciculation, whereas, broad expression of sema2a supports its role as a chemorepellent that prevents improper innervation by the afferents.

Sema3f is one of the best studied members of the Sema3 subclass, especially in the context of human lung cancers, in which it is down regulated (Potiron et al., 2007). During cranial neural crest cell (CNCC) migration in zebrafish, sema3fa (and sema3ga/sema3gb) is expressed in the neural crest cell (NCC) free zones of hindbrain rhombomeres (r) 3 and r5 restrict nrp-expressing

NCCs migration into distinct peripheral streams (Yu and Moens, 2005). Similarly, Sema3f deficiency in mouse results in the crossing of CNCC migratory streams and the improper formation of NCC derived trigeminal ganglia (Gammill et al., 2006; Gammill et al., 2007). Expressed at the boundaries of the rhombomeres, sema3fb in conjunction with sema3gb are required to position fgf20a-expressing neuronal clusters in the center of rhombomere segments, with loss of this boundary formation resulting in an increase in neurogenesis (Terriente et al., 2012).

Sema3f guides axons in a variety of systems. In mouse, Sema3f is expressed in the caudal midbrain and rostral hindbrain, leaving a non-repellent corridor at the midbrain-hindbrain junction.

The axons of Nrp2-expressing trochlear motor neurons project contralaterally to the eye across this corridor (Giger et al., 2000). During glomerular map formation, early arriving olfactory sensory neurons secrete Sema3f into the antero-dorsal olfactory bulb to repel late arriving Nrp- expressing olfactory sensory neuron axons (Takeuchi et al., 2010). In addition to its defined roles

in the nervous system, Sema3f has anti-angiogenic effects on tumour growth in vitro (Potiron et al., 2007) and on pathological retinal vessels in vivo (Sun et al., 2017a).

In addition to their canonical role as guidance cues for migrating cells, axons and angiogenic sprouts, Sema3s can influence other aspects of development including proliferation and polarity, differentiation, and survival (Arbeille et al., 2015; Giacobini and Prevot, 2013; Jongbloets and

Pasterkamp, 2014; Kang and Kumanogoh, 2013; Kumanogoh and Kikutani, 2013; Neufeld et al.,

2012; Roth et al., 2009; Tamagnone, 2012; Xia et al., 2015). In zebrafish, Sema3d is necessary and sufficient to drive the proliferation of neuroepithelial cells of the hindbrain prior to NCC migration, supporting a role in cell cycle progression (Berndt and Halloran, 2006). Circulating

SEMA3B in the cerebral spinal fluid in mouse acts on neural progenitor cells to orient the mitotic spindle, through the inhibition of the microtubule destabilizer Collapsin Repsonse Mediator

Protein 2 (CRMP2), thereby dictating the plane of division in neurogenic zones of the brain

(Arbeille et al., 2015). Recent reports identified SEMA3A as a novel regulator of bone remodeling in mouse (Li et al., 2017). Sensory neuron derived SEMA3A acts on osteoblasts to promote their differentiation via β-catenin, while simultaneously inhibiting osteoclast differentiation by blocking

Phospholipase C Gamma (PLCγ) activation. Together, these functions promote an increase in bone formation and decrease in bone resorption, respectively. In mouse, astrocyte deletion of SEMA3A results in spinal motor neuron apoptosis, suggestive of a role as a survival factor (Molofsky et al.,

2014). Overall, roles outside of guidance mechanisms are coming to light using in vivo models.

1.1.4 Spatially Regulated Tissues in Development

Spatial patterning allows for the specification of diverse identities from a common pool of progenitor cells (Durston et al., 2010). During development, the neural tube is comprised of a pseudostratified sheet of neuroepithelial cells which must regionalize in order to give rise to

neurons and structures of the mature CNS (Lupo et al., 2006). An excellent example of this process is morphogen patterning of the spinal cord. Morphogens are long-range factors that can specify distinct and diverse fates of cell in a concentration dependent manner. In the ventral neural tube of the presumptive spinal cord, Sonic hedgehog (Shh) is secreted by the floor plate and acts as a long- range morphogen which patterns cells by conferring positional information (Figure 1.2A). Cells closest to the ventral floor plate are exposed to the highest environmental concentration of Shh and adopt a fate that is unique to this environment. Shh signalling induces the expression of multiple transcription factors by progenitors, which in turn triggers downstream transcriptional changes and specification of the cells, delineating spatially distinct domains along the dorsoventral axis of the neural tube. Bone morphogenetic protein (BMP) is a morphogen secreted from the roof plate of the dorsal spinal cord, that works alongside Shh to coordinate cell fate specification along the dorso-ventral axis of the spinal cord.

Spatial patterning of progenitors in the eye early in development is also important in establishing the connectivity of RGCs to their brain targets. In the retina, the axons of RGCs make connections to the visual centers of the midbrain, whereby the spatial origin of their cell bodies in the retina is conserved with regards the location of their termination in the midbrain (Stuermer,

1988) (Figure 1.2B). This mapping system is referred to as retinotopic mapping, and is mediated by gradients of Ephrin/Eph signalling; the temporal-nasal axis of the retina uses EphrinA-EphA to map along the tectal anterior-posterior axis, whereas the dorsal-ventral axis of the retina uses

EphrinB/EphB signalling to map along the medial-lateral axis of the tectum (Feldheim and

O’Leary, 2010; Suetterlin et al., 2012). As is observed in the neural tube, morphogen gradients pattern the progenitors of the early developing eye vesicle to confer regional identities within the

Figure 1.2 Spatial patterning in the developing nervous system. A) Schematic representation of a transverse section through the neural tube whereby Sonic hedgehog (Shh, red) is secreted by the notochord (n) and floorplate (fp) to pattern the ventral neural tube. Depending on the concentration of Shh to which progenitors are exposed, they will take on a specific neural fate

(coloured circles). B) Retinotectal topographic map schematic. In the retina, expression of the

EphA receptor by retinal ganglion cells (RGCs) increase nasal (N) to temporal (T), while its ligand,

EphrinA, increases anterior (A) to posterior (P) in the superior colliculus (SC)/optic tectum (OT) axon target. Due to the repulsive signalling between the two, temporal RGCs axons project to the anterior tectum, while nasal RGCs project to the posterior (matched colored circles). Similarly,

EphB expression by RGCs increases dorsal (D) to ventral (V) and EphrinB medial (M) to lateral

(L). Dorsal RGCs project to the lateral aspect of the SC/OT while ventral RGCs connect to the medial aspect.

neuroepithelium, producing nasal-temporal and dorsal-ventral domains. For example, opposing gradients of Fgf and Shh are necessary to specify the nasal and temporal retina, respectively

(Hernández-Bejarano et al., 2015). Once specified, transcriptional domains first establish and then regionalize the eye vesicle such that nasal expression of Foxg1 and temporal expression of Foxd1, amongst a much larger regional transcriptome, trigger downstream differentiation events to induce the expression by RGCs of Eph/Ephrin in a spatially organized, patterned manner (Takahashi et al., 2009).

1.2 Development of the Zebrafish Neural Retina

The vertebrate retina is relatively well conserved across species (Malicki, 1999). The power of using zebrafish as a developmental model comes from their large progeny numbers, rapid ex utero development that allows for discrete genetic and pharmacological interventions, and the availability of transgenic lines with cells of a lineage labelled so that individual cell behaviours can be visualized in vivo by time-lapse microscopy (Dooley and Zon, 2000).

1.2.1 Organization of the neural retina

The neural retina of vertebrates is laminated and consists of three nuclear layers containing cell bodies of the five neural and one glial type of cell, and two plexiform layers where synaptic connections are formed (Malicki, 1999). The basic structure of the vertebrate eye is illustrated in

Figure 1.3. The RGCs form a layer that lines the vitreal surface of the eye, and, as the only output neuron of the retina, send axons to the brain as an optic nerve. The inner nuclear layer consists of the majority of cell types within the retina – amacrine, horizontal, and bipolar interneurons, and

Müller glia. The outer nuclear layer contains rods and cones. Zebrafish have four types of cones

(long-, medium-, short- and UV- wavelength sensitive) and one rod photoreceptor. As cells of the

Figure 1.3. Organization of the vertebrate eye. The retinal pigment epithelium (rpe) lines the back of the retina. Adjacent is the outer nuclear layer (onl) comprised of the rod (r) and cone (c) photoreceptors, which make connections with cells of the inner nuclear layer (inl) at the outer plexiform layer (OPL). The inl is comprised of horizontal cells (hc), amacrine cells (ac), bipolar cells (bc) and the cell bodies of Müller glia (mg). The retinal ganglion cell layer (rgcl) is comprised of retinal ganglion cells (RGCs) which synapse with the inl at the inner plexiform layer (IPL). The rgcl axonal projections make up the optic nerve (ON) which exits the eye. In the periphery of the retina, the ciliary marginal zone (cmz) contains retinal stem cells. Adapted from Cechmanek, 2018.

retina differentiate, they begin to form connections with one another to establish the retinal circuitry (Amini et al., 2018). The inner plexiform layer is comprised of the RGC dendrites that synapse with the processes of amacrine and bipolar cells. Within the outer plexiform layer, the photoreceptor axons synapse with horizontal and bipolar cells. Visual perception relies on the neural circuitry within the retina and the synaptic connections made in the brain. Light enters the eye through the lens, and passes through the RGC layer and the inner nuclear layer to where photons are detected by the photoreceptors of the outer nuclear layer. Light information is then transmitted from the photoreceptors to the RGCs via the bipolar cells, and is modulated by the interneurons of the inner nuclear layer. In teleosts, like zebrafish, a proliferative zone of cells named the ciliary marginal zone (CMZ) is found in the peripheral retina, and permits persistent neurogenesis and eye growth for the life of the organism (Johns and Fernald, 1981).

1.2.2 Development of the zebrafish retina and lens

As in other vertebrates, zebrafish eyes arise from a bilateral thickening and evagination of the anterior neural tube to form the morphologically identifiable optic lobes by the 6 somite stage

(ss) (Malicki, 1999; Schmitt and Dowling, 1994). At approximately the 8-9 ss the posterior portions of these lobes begin to separate from the brain while the anterior – attached - portions persist as the optic stalks. By the 14 ss, the optic lobes invaginate on themselves and the overlying epithelium thickens to form the rudimentary lens. Invagination of this neuroectoderm gives rise to the optic cup, of which the distal/dorsal surface becomes the neural retina, and dorso-medial the retinal pigmented epithelium (Martinez-Morales et al., 2004). Morphogenesis of the optic cup is mostly complete by 24 hours post-fertilization (hpf) and proceeds through a migration of the ventral poles towards each other, meeting at a transient structure known as the choroid fissure

(CF).

The CF serves as the point of entry for the retinal hyaloid vessel, periocular mesenchyme

(POM)/ neural crest cells (NCCs) that provide the support cells for the hyaloid vessel, and the output route for RGC axons that make up the optic nerve. The CF, initially pointing above the yolk sac (more anterior) at 24 hpf, is rotated towards the heart (ventral pole of the optic cup) by 36 hpf, signifying the last major morphogenetic movement in zebrafish eye development. CF closure is complete by 48 hpf (Malicki, 1999; Schmitt and Dowling, 1994), and failure of this closure results in a condition called coloboma (Chang et al., 2006).

1.2.3 Neurogenesis of the zebrafish retina

As for other species, the zebrafish retina contains the six major cell types (Neumann, 2001); ganglion, horizontal, amacrine, bipolar and photoreceptor cells, and the Müller glia.

Differentiation amongst all retinal layers begins in the ventral retina, nasal to the optic nerve exit site, and moves in a temporal direction around the eye. RGCs are first to differentiate and become post mitotic beginning at 28 hpf (Schmitt and Dowling, 1994; Schmitt and Dowling, 1996;

Nawrocki, 1985). Neurogenesis, for all other neural types, is completed by 2.5 days post fertilization (dpf) (Nawrocki, 1985).

Classical birth dating experiments demonstrate that neurons in the retina are born in a stereotyped order conserved across vertebrates (Nawrocki, 1985; Wong and Rapaport, 2009;

Young, 1985). First born are RGCs, then cone photoreceptors, horizontal cells, amacrine, rod photoreceptors, bipolar cells, and lastly the Müller glia (Amini et al., 2018). All cell types of the retina are derived from retinal progenitor cells (RPCs), and it was widely accepted that RPCs change competence over time to give rise to the various cell types of the neural retina in a deterministic fashion (Cepko et al., 1996; Chen et al., 2012). However, lineage studies in both rat

(Gomes et al., 2011) and zebrafish (Boije et al., 2015) suggest that retinogenesis is not entirely

deterministic, and that stochasticity has a major role in the generation of the different neural types.

Large bodies of evidence now suggest that the bias towards particular fates is not due to the terminal division of an RPC, but, instead, a committed precursor undergoing symmetric divisions

(Engerer et al., 2017; Suzuki et al., 2013; Weber et al., 2014). In fact, zebrafish lineage studies have demonstrated that apical divisions (RPE adjacent) of RPCs exclusively generate RGCs and amacrine cells, while all other cell types are born from the symmetric neurogenic divisions of committed precursors: cones divide within the photoreceptor layer (Suzuki et al., 2013), horizontal cells near the outer plexiform layer (Godinho et al., 2007) and bipolar cells in either apical or subapical positions (Weber et al., 2014). In support, a recent review has described the transitional nature of progenitors during vertebrate (murine) retinogenesis, such that progenitor cells are proposed to go through sequential states of potency (multipotent to unipotent) before terminally differentiating (Jin, 2017). It has become increasingly clear that while much of the work done already towards understanding retinogenesis is invaluable, the complex behaviours and heterogeneity of the progenitors and precursors are just only being understood.

In contrast to mammals, neurogenesis continues at the CMZ throughout the life-time of the fish (Johns and Fernald, 1981). The persistent neurogenesis that occurs via stem cells in the CMZ, allows for the fish to grow its retina in concert with its body size. Mammals have CMZ-like regions in the periphery, however, their potential as a source for new neurons are only just being investigated (Kubota et al., 2002; Stenkamp, 2015).

1.2.4 Neuronal differentiation in the zebrafish retina

The retina is differentiated by the interplay between intrinsic and extrinsic mechanisms

(Stenkamp, 2015). RGCs are the first-born neurons in all vertebrates characterized to date (Engerer et al., 2017; Jin, 2017; Malicki, 1999), and Atonal homolog 7 (Atoh7) is key for their genesis.

Atoh7 is expressed during the final division of an RPC (Masai et al., 2005), and its loss, as seen in the zebrafish lakritz mutant, produces a retina with no RGCs, but with all other cell types present

(Kay et al., 2001). Additionally, atoh7 overexpression leads to a preferential increase in the number of RGCs (Sinn et al., 2014). Shh secretion from the optic stalk was postulated initially to initiate RGC differentiation by inducing expression of atoh7 (Masai et al., 2000), with mature

RGCs also expressing Shh to push neighbouring RPCs to an RGC fate (Neumann and Nuesslein-

Volhard, 2000). Characterizing the spatiotemporal expressions of shh and atoh7 by using transgenic (Tg) zebrafish, Tg(shh:GFP) and Tg(atoh7:GFP) identified that the ventro-nasal to temporal retinal differentiation of RGCs actually requires Shh to work in two steps (Masai et al.,

2005). First, Shh promotes cell cycle exit and the expression of atoh7, and then Shh drives atoh7 positive cells to the RGC fate. While atoh7 is necessary for the development of RGCs, it is also expressed by progenitors destined for other fates such as horizontal, cone, rod and amacrine cells

(Feng et al., 2010). Furthermore, in the lakritz (atoh7-/-) mutant an increase in inner nuclear layer cells is found with the absence of RGCs (Kay et al., 2001), and the loss of the murine atoh7 homologue Math5 produces an increase in amacrine, cone, horizontal and rod cells with an accompanied upregulation of factors necessary for their differentiation (Feng et al., 2010). These data suggest that retinogenesis by atoh7/Math5 is mediated through multiple mechanisms to generate the desired retinal neural types.

The INL consists of the majority of cell types within the retina – amacrine, horizontal, bipolar and Müller glia. The molecular control of retinal differentiation for the cells of this layer is still not entirely clear. Knockdown in zebrafish of the transcription factor iroquois 7 (irx7), which is normally expressed by the INL at 52 hpf, results in poorly differentiated cell types of this layer (Zhang et al., 2012). Interestingly, reduced irx7 results in increased pancrease transcription

factor 1 (ptf1a) expression (Zhang et al., 2012), a transcription factor expressed by differentiating amacrine and horizontal cells in zebrafish (Jusuf and Harris, 2009), and necessary and sufficient for the differentiation of both cell types in mouse (Nakhai et al., 2007). The fact that a loss of irx7 induces ptf1a expression in zebrafish, but few terminally differentiated amacrine and horizontal cells are found (Zhang et al., 2012), suggests differences in this regulatory pathway between mice and zebrafish.

The transcriptional regulation of photoreceptor fate in vertebrates has been of intense interest for potential cell replacement therapies (Swaroop et al., 2010). In fish and mouse, thyroid hormone receptor Trβ2 is necessary for the differentiation of long wavelength sensitive cones

(Roberts et al., 2006; Swaroop et al., 2010), and, in mouse, is regulated by the presence of thyroid hormone across the neural retina (Roberts et al., 2006). Whether this is true in zebrafish is still unclear.

1.2.5 Development of the RGC axon projection

In zebrafish, RGC axons are guided out of the eye through the CF between 34-36 hpf, and through the brain to their target in the dorsal midbrain - the optic tectum (Stuermer, 1988). By 72 hpf all axons have innervated the optic tectum. While several molecules are known to guide RGC axons through the brain (reviewed in (Erskine and Herrera, 2007; Erskine and Herrera, 2014)), intraocular guidance mechanisms are largely unexplored. In mice deficient for EphB2 and EphB3 receptors a small number of dorsal RGC axons erroneously bypass the optic disc and misproject into the neural retina (Birgbauer et al., 2000). Similarly, Slit1 and Slit2, and their common receptor

Robo2, also prevent ectopic routing of RGC axons into the neural retina (Thompson et al., 2006;

Thompson et al., 2009). However, only a small subset of ventral axons are misguided in mice mutant for the Slit repulsive guidance pathway, suggesting that other inhibitory signals are present.

The fact that a handful of misguided Slit1/2 and Robo2 deficient RGC axons still manage to reach the optic disc suggests a long-range attractant is present (Thompson et al., 2006; Thompson et al.,

2009). There are only a few defined attractive pathways that promote axon migration towards the optic disc and out through the CF. In zebrafish, the Cxcl12a/Cxcr4b chemokine ligand-receptor is required for such disk-directed growth (Li et al., 2005), while it is responsiveness of DCC- expressing RGC axons to the chemoattractant Netrin1 that guides axons at the disk and through the CF in mouse (Deiner et al., 1997; de la Torre et al., 1997). Additional molecular mechanisms likely guide RGC axons in the retina.

1.2.6 Class 3 Semas in Retinal Development and Injury

Sema3s have not been implicated in normal retinal development, outside of a role for chemotactic guidance of zebrafish RGC axons (Sakai and Halloran, 2006) and Xenopus RGC polarity (Kita et al., 2013), despite their known expression within the developing eye of zebrafish and rat (Callander et al., 2007; de Winter et al., 2004). Instead, SEMA3s have been looked at in pathological situations. Crush of the rat optic nerve cause downregulation of SEMA3F and suggests a possible neuroprotective role of SEMA3F on RGC function and survival (Ko et al.,

2016). In porcine retinal explants, SEMA3A is increased following retinal detachment and inhibits the process formation and cellular orientation of rod photoreceptors, suggesting a negative role in pathological progression following injury (Kung et al., 2017).

1.3 Development of the Vasculature of the Zebrafish Retina

The vertebrate retina is nourished by two vascular systems. The intraocular, or hyaloid/retinal vasculature, and the extraocular, or choroidal vasculature (Campochiaro, 2015;

Saint-Geniez and D’Amore, 2004). Conservation of both independent circulatory systems are

conserved from zebrafish to mammals, and the formation of all orbital vessels in zebrafish is strictly by angiogenesis (Hartsock et al., 2014; Kaufman et al., 2015).

1.3.1 Hyaloid and Retinal Vasculature Development

The hyaloid artery branches from the cranial division of the internal carotid artery and enters the eye through the CF, passing the entirety of the vitreous and contacting the posterior pole of the lens (Isogai et al., 2001; Saint-Geniez and D’Amore, 2004) (Figure 1.4). Little is known of the morphogenesis of the hyaloid vessel in any vertebrate system, however the spatiotemporal dynamics of the vessel have recently been described in zebrafish (Hartsock et al., 2014).

Specifically, the artery enters the fissure at 18 hpf, and contacts the lens by 24 hpf to form a hyaloid loop which encompasses the posterior aspect of the lens. The hyaloid network then become more elaborate and branched between 35 and 84 hpf. Blood circulation within the retina begins around

72 hpf, once the surface vessels - nasal, dorsal and ventral radial vessels - are connected by a duct anastomosed with the hyaloid vasculature (Kitambi et al., 2009). As anterior growth around the lens continues, a statistically significant decrease in the number of hyaloid branches on the lens is observed from 4-5 dpf, referred to as vessel refinement. Additionally, by 5 dpf the blood-retinal- barrier (BRB) of the hyaloid is established, as determined by no observable leakage of angiogram molecules. In mammals, VEGFA plays an essential role in hyaloid vasculature development

(Gerhardt et al., 2003). VEGFA is secreted from the lens and the underlying astrocyte scaffold that lines the inner limiting membrane of the retina (Shui et al., 2003). Interestingly, angiogenesis is misdirected towards neurons with a genetic deletion of VEGFR2, due to a failure of the neurons to endocytose excess VEGF that leads to abnormal sprouting angiogenesis within the retina (Okabe et al., 2014). It is clear that lens derived signals are essential for hyaloid morphogenesis, in that a zebrafish mutant for mab21l2 (Lee et al., 2012), which forms a dysmorphic or completely absent

Figure 1.4. Development of the zebrafish ocular vasculature. A) The hyaloid artery (H) first enters the eye at 18 hours post fertilization (hpf). At 24 hpf (B) the hyaloid loops around the lens

(L) and forms a complex plexus by 72 hpf (C). D) Refinement of the hyaloid plexus simplifies the structure, and the choroid plexus (CP) forming on the outside and back of the neural retina (NR) is visible by 6 dpf. E) The CP continues to envelop the entirety of the back of the eye (red line) while the hyaloid vessel delaminates from the lens and attaches to the inner limiting membrane

(ILM) of the NR from 15 days post fertilization (dpf) onwards. F) In flat mount preparations, the hyaloid (now retinal vasculature) emanates from the optic nerve head (ONH).

lens, has defects in hyaloid branching and refinement, but not in the initial infiltration of the hyaloid into the CF (Hartsock et al., 2014).

At 15 dpf, the hyaloid vasculature begins to detach from the lens in a posterior to anterior fashion (Alvarez et al., 2007). Detachment from the lens and attachment to the inner limiting membrane of the retina is progressive and variable depending on the rate of larval growth. By 60 dpf, however, the vasculature is completely adhered to the retina. At this point, the hyaloid vasculature is now termed the retinal vasculature, and comprises the bulk of the intraocular circulatory system. In contrast, mammals do not undergo the hyaloid-to-retinal transition by way of vessel detachment from the lens (Fruttiger, 2007; Saint-Geniez and D’Amore, 2004). Instead, the hyaloid vessels regress to the site of the optic nerve head, and angiogenesis is reinstated to develop a retinal vessel network. Additionally, the retinal vasculature of many mammals branches and forms intraretinal capillaries that form plexi within both the inner and outer plexiform layers

(Fruttiger, 2007; Provis, 2001). No vessels are present within the nuclear layers of the retina of any vertebrate. Such a phenomenon (intraretinal capillaries) does not occur in the zebrafish retina

(Alvarez et al., 2007), or in mammals such as rabbit, in which the entire retina is nourished by the choroidal vasculature (Saint-Geniez and D’Amore, 2004), likely owing to the thinness of the retinas that require oxygenation.

1.3.2 Choroidal Vasculature Development

The choroid is the vascular layer at the back of the eye lying between the RPE and sclera, and serves as the largest blood supply source for the outer retina, or whole retina in species that have no intraocular vascularization (Nickla and Wallman, 2010) (Figure 1.4). The term choroid is used to encompass a number of different cell types – namely blood vessels, melanocytes, fibroblasts, resident immune cells and ECM-rich connective tissue. In humans, the choroid plexus

forms a dense network at the back of the eye during the second and third month of gestation, and requires the presence of a differentiated RPE (Saint-Geniez and D’Amore, 2004). In zebrafish, the choroid can be identified in zebrafish at 5.75 dpf (Alvarez et al., 2007; Rooijen et al., 2010).

Despite its importance, little is known of its molecular mechanisms underlying its development.

The structure of the choroid has been well described histologically in primates (non-human and human) and divided into 5 layers. From most distal-proximal to the retina these include the suprachoroid, a vascular region containing large blood vessels (Haller’s layer) and medium and small arterioles (Sattler’s layer), the choriocapillaris, and Bruch’s membrane.

The suprachoroid consists of collagen, fibroblasts and melanocytes, and behaves as a transition zone between the choroid and overlying sclera (Krebs and Krebs, 1988). The suprachoroid consists of large endothelial lined spaces referred to as lacunae, which presumably act as lymphatic drainage networks. Additionally, myelinated axons innervating the layer are present from sympathetic, parasympathetic and sensory sources (Nickla and Wallman, 2010).

Below the suprachoroid sits a dense network of small/medium to large arterioles, known as

Sattler’s layer and Haller’s layer, respectively. The vessels decrease in size and give rise to the choriocapillaris.

The choriocapillaris is a dense network of highly anastomosed capillaries that overly

Bruch’s membrane (Bill et al., 1983). From Sattler’s layer the capillaries form hexagonal like structures with fenestrations that almost exclusively face the retina (McLeod et al., 2009). The fenestrations are permeable to proteins and so permit the movement of nutrients, waste and fluid from the RPE/retina to the choroid. The capillaries of the choriocapillaris lay above Bruch’s membrane, the latter serving as the basal lamina for the vessels, and the underlying RPE. Bruch’s membrane is a layered structure of ECM proteins such as collagen (Bill et al., 1983; Nickla and 39

Wallman, 2010), and plays a fundamental role in regulating nutrient and metabolic waste exchange by functioning as part of the BRB.

1.3.3 Blood-Retina-Barrier

The BRB is broken down into two components, an outer- and inner- BRB (Cunha-Vaz et al., 2011). The inner-BRB is established by tight junctions between ECs of the vasculature (Shakib and Cunha-Vaz, 1966). Retinal vessels are covered by processes of astrocytes and Müller glia cells, which, together with EC tight junctions, regulate the diffusion capabilities of the vasculature

(Cunha-Vaz et al., 2011). This type of BRB is necessary for intraocular vessels to prevent leakage and uncontrolled diffusion of metabolites. The outer-BRB regulates the ability of the choriocapillaris to interact with the underlying RPE/retina. While the ECs of the choroid also contain tight junctions, the choriocapillaris is highly fenestrated (Nickla and Wallman, 2010) permitting the passage of larger proteins out of the vessel. The outer-BRB is therefore established by the tight-junctions between RPE cells that prevent intake of nutrients and fluid in an uncontrolled manner.

1.3.4 Age Related Macular Degeneration

Many histological studies have investigated the relationship between photoreceptors, RPE,

Bruch’s membrane and the choriocapillaris (Choi et al., 2013), as changes or atrophy in any of these structures are associated with pathology and disease progression. For example, conditional deletion of RPE-derived VEGF in adult mouse results in loss of the choriocapillaris followed by death of the RPE and photoreceptors (Kurihara et al., 2012). VEGF overexpression in the RPE, however, is insufficient to drive choroidal neovascularization (CNV), a highly pathological condition seen in many choroidal retinopathies (Campochiaro, 2015; Oshima et al., 2004). Instead

VEGF overexpression causes ruptures to Bruch’s membrane that allow ectopic entry of vessels

into the outer retina (Campochiaro, 2015). Thus, the relationship between the different cell types of the outer retina and choroid is essential to maintain normal physiology, however, it is unclear the hierarchical importance of the dysfunction when pathologies are present.

Age-related macular degeneration (AMD) is a disease of the macular retina causing progressive central vision loss (Mitchell et al., 2018). Multifactorial in origin, physiological dysfunctions and genetic susceptibility loci have been identified. Non-genetic risk factors, however, such as smoking, also play fundamental roles in disease onset. Global prevalence is expected to increase to 200 million by 2020 (Wong et al., 2014). Reviews of AMD pathogenesis

(Campochiaro, 2015; Kawa et al., 2014; McLeod et al., 2009; Mitchell et al., 2018) support a slow, two-staged progression of the disease. Early stage-AMD is characterized clinically by focal deposits of drusen (lipids) located subretinally, alongside abnormalities in the RPE integrity and/or pigmentation. Often asymptomatic, early AMD patients can experience mild visual distortion. Late

AMD is subdivided into two types – neovascular, or wet/exudative, and non-neovascular, or atrophic/dry/non-exudative. It is in late AMD that significant loss of vision occurs leading to permanent impairments.

Dry AMD is most common in patients with AMD, and is characterized by slow atrophy or progressive degeneration of the outer retina. Conversely, wet AMD is characterized by neovascularization from the overlying choroid. Clinically this is observed as fluid accumulation in the subretinal space, RPE detachment and fibrous scar tissue, and remains the greatest contributor to vision loss. No treatments currently exist for dry AMD, however, intraocular injections of anti-

VEGF have been used routinely to treat wet AMD with varying efficacy (Cui and Lu, 2017).

Additionally, controversy exists as to whether anti-VEGF therapies pose risks to the retina, RPE

and choroid which use VEGF physiologically, supporting the need for new therapeutic interventions that block CNV in wet AMD.

1.3.5 Class 3 Semas in Orbital Vessel Dynamics

The CNS is vascularized through angiogenic sprouting of blood vessels (Tata et al., 2015).

The molecular and cellular interactions that are critical for CNS vascularization during development, are only just being appreciated for their roles in pathology. Not only do blood vessels need to know where to grow, but also where not to grow. In light of this, multiple Sema3s have been identified as regulators of physiologic and pathologic angiogenesis within the CNS, and more specifically, the retina. The cross-talk between neurons and retinal vasculature is well established, as loss of RGCs results in no retinal vessel development (Sapieha et al., 2008). In ischemic retinopathies, compensatory angiogenesis is initiated to ameliorate ischemia of the tissue, however, this attempted rescue is usually confounded by leaking of pathologic vessels into the vitreous of the eye (Sapieha, 2012). In mouse models of hypoxic stress, RGCs secrete SEMA3A, which likely functions as an anti-angiogenic to push blood vessels into the vitreous (Cerani et al.,

2013; Joyal et al., 2011). Similar findings were also found with RGC-derived SEMA3E (Sun et al., 2017b). SEMA3F is known to promote the avascularity of the outer retina and cornea in mouse

(Buehler et al., 2013; Reuer et al., 2018; Sun et al., 2017a). Sema3f is expressed by the ONL and

RPE in mouse and humans (Buehler et al., 2013). Post-mortem analysis of humans with wet-AMD had a significantly reduced level of SEMA3F in 10 out of 15 patients, potentially suggesting a role for SEMA3F in maintaining avascularity of the outer retina. Viral overexpression of SEMA3F was found to have protective/anti-angiogenic effect and prevent vessel entry into the outer retina with a mouse knockout for Very low-density lipid-receptor (Vldlr), which spontaneously undergoes

CNV, as well with laser-induced damage to Bruch’s membrane to induce CNV (Sun et al., 2017a).

While these data indicate that SEMA3F is sufficient to inhibit pathological angiogenesis of the choroid, and the human expression data suggests a physiological role, a direct endogenous test of

SEMA3F in patterning the vessels providing blood to the eye has not been assessed.

Of note, recombinant SEMA3A (Bai et al., 2014) and SEMA3C (Toledano et al., 2016) has been used as an anti-angiogenics to treat CNV in mouse models. Thus, the possibility of new therapeutic interventions outside of anti-VEGF therapy are becoming a possibility.

1.4 Development of the Zebrafish Cardiovascular System

The heart is the first functional organ established in the developing embryo (DeRuiter et al., 1992). Cardiovascular disorders, both congenital and those acquired later in life, remain the largest contributor to morbidity and mortality in Canada (Manuel et al., 2003). The heart is specified as bilateral fields which fuse at the midline and pass through a series of morphologic and differentiated states in order to establish the vertebrate two-, three- or four-cardiac chambers

(Stainier et al., 1993). In zebrafish, the mature heart is comprised of two cardiac chambers – ventricle and atrium – and two specialized chambers; one which receives blood, the sinus venosus, and, one which functions as a capacitor to perfuse the gill arches, the bulbus arteriosus. Zebrafish have emerged as a powerful model to study cardiovascular development, as they do not initially rely on it for oxygen needs, and therefore severe cardiac mutants can be analyzed for their phenotype much longer period than in any other model system (Liu and Stainier, 2012).

1.4.1 Specification and Differentiation of the Zebrafish Heart

The embryonic heart is composed of two tissue layers – myocardium and endocardium

(Stainier et al., 1993). Myocardial progenitor cells are specified bilaterally in the anterior lateral plate mesoderm (ALPM) by 5 hpf in zebrafish embryos and prior to gastrulation (Figure 1.5). At

Figure 1.5. Development of the zebrafish heart. A) The heart field is specified by 5 hours post fertilization (hpf) with the atrial cardiomyocytes (red) precursors located more medial than the ventricular cardiomyocyte (green) precursors. B) By the 12 somite stage (ss) the cardiomyocytes begin to differentiate and migrate to the midline where the bilateral heart fields fuse into a cone like structure (C) with an endocardial lining (yellow). D) Heart tube elongation occurs from 24 hpf and is followed by twisting and the emergence of the cardiac chambers, ventricle (green) and atrium (red) by 48 hpf (E). F) Chambers continue to grow in size and the presence of the bulbus arteriosus (blue) is now visible. Adapted from Brown et al., (2016).

this early time point, atrial and ventricular progenitor cells are already specified and organized

(Keegan et al., 2004). Through fate map analysis, the chamber specific progenitors remain separated, whereby atrial progenitors are located more ventral in the lateral margin zone as compared to the ventricular progenitors. Meanwhile endocardial progenitors, which go on to line the inside of the heart, show no apparent organization and coalesce at the midline. It remains unclear what maintains the separation of the progenitor pools early and throughout cardiogenesis.

Specification of the heart field by extrinsic regulators is conserved across the vertebrates.

Several signalling molecules have been implicated as positive regulators of cardiac differentiation, including BMPs (Schultheiss et al., 1997), Nodal (Reiter et al., 2001), and FGF8 (Alsan and

Schultheiss, 2002). Also involved are negative regulators, such as Wnts (Marvin et al., 2001) and retinoic acid (RA) (Keegan et al., 2005). In concert, these extrinsic signals initiate the induction of a transcriptional cascade necessary for cardiac development. Of note, GATA-type zinc finger

GATA5 (Reiter et al., 1999) and Basic helix loop helix (Bhlh) factor HAND2 (Yelon et al., 2000), likely work as parallel transcription factor pathways, as loss of either molecule results in a smaller bilateral myocardium. Interestingly, expression of sarcomeric myosin genes, necessary for heart contractility, begins before fusion of the bilateral heart field at the midline of the embryo.

Specifically, atrial myosin (myh6) and ventricular myosin (myh7) are identifiable by the 14 ss

(Yelon et al., 1999), although the onset of expression of myh6 proceeds that of myh7 (Berdougo et al., 2003), suggesting a temporal timing difference in the differentiation of atrial and ventricular myocardial cells. The ability in zebrafish to follow the maturation of different fluorescent proteins in real time has identified that waves of cardiogenic differentiation occur, with differentiation initiated at the 12-15 ss in the future ventricle progenitors, followed by the laterally localized atrial progenitors that continues until 22 hpf (de Pater et al., 2009).

1.4.2 Cardiac Chamber Morphogenesis

Concurrent with cardiogenic differentiation, the bilateral field of progenitors migrate towards the midline, and fuse into a cone shape with the endocardial cells in the middle, lined by the ventricular, and then atrial progenitor cells by 22 hpf (Holtzman et al., 2007; Yelon et al.,

1999). Myocardial epithelium migration requires interactions with the surrounding environment.

In hand2 mutant zebrafish (natter), the bilateral heart field fails to migrate towards the midline as a result of the failure of fibronectin deposition between the myocardium and yolk syncytial layer, and presents as a cardia bifida (Garavito-Aguilar et al., 2010; Trinh and Stainier, 2004).

Additionally, in loss of function mutants for platelet derived growth factor receptor a (pdgfra) the cardiac heart fields fail to migrate to the midline and fuse. Here, the ligand Pdgfa is produced by the underlying endoderm and directs the pdgfra-expressing cardiomyocytes (Bloomekatz et al.,

2017). It is clear that cardiac fusion requires concerted efforts of multiple extrinsic regulators.

Once fused, the cardiac cone undergoes asymmetric morphogenesis (Bakkers, 2011). The cells originally derived from the right cardiac field involute ventrally and migrate to the anterior/left, a process known as cone tilting, or condensing, resulting in a cone with a dorsal- ventral organization (Rohr et al., 2006). The translocation of one pole to the left side of the embryo is concomitant with heart tube extension. The linear heart tube, also known as the first heart field, extends from 24-28 hpf. This process requires normal epithelial polarity (Rohr et al., 2006), and minor amounts of proliferation and integration of cardiomyocytes from other sources such as the cardiac neural crest (Cavanaugh et al., 2015) and second heart field (de Pater et al., 2009).

Once the linear heart tube has elongated, it undergoes a bending towards the right side displacing the future ventricle to the right of the midline by 28 hpf, and future atrium to the left

(Bakkers, 2011; Brown et al., 2016). The underlying cellular mechanisms behind cardiac looping

are not well understood, however, biomechanical forces can contribute to the process (Voronov et al., 2004). During looping, the cardiac chambers – ventricle and atrium – become morphologically distinguishable with an outer and inner curvature to their shape (Christoffels et al., 2004). Blood flow through the heart that begins at 24 hpf promotes cardiomyocyte enlargement and elongation, which intrinsic contractility by the sarcomeres counters in order to prevent dilation (Auman et al.,

2007). The hemodynamic forces the heart is subject to ultimately influence the curvature of the looping tube.

The zebrafish heart is two chambered and therefore does not undergo the septation that occurs in mammals. Cardiac valves are critical, however, to prevent blood flowing back from the ventricle to the atrium when the ventricle contracts. In zebrafish, the atrioventricular (AV) valve first shows signs of differentiation at 36 hpf, when endocardial cells located at the atrioventricular canal (AVC) become cuboidal and express DM-GRASP (Beis et al., 2005). Unlike in amniotes, in zebrafish the cells of the valve do not undergo an epithelial-to-mesenchymal transition, and, instead by 48 hpf the endocardial cushions have formed by invagination of the endocardial cells

(Scherz et al., 2008). Despite this difference, the signalling pathways involved in AV valve development are conserved, and include Notch, nuclear factor of activated T cells (NFAT), epidermal growth factor receptor (ErbB) and transforming growth factor beta (Tgf-β) (Beis et al.,

2005). By 104 hpf, the endocardial cushions are enlarged and differentiated into the valve leaflets.

1.4.3 Cardiac Neural Crest and Second Heart Field

The extension of the linear heart tube cannot be explained entirely by proliferation or cellular shape changes. Cells from the cardiac neural crest (CNC) and second heart field (SHF) contribute to the elongation of the tube and chamber development (Cavanaugh et al., 2015; de

Pater et al., 2009). First, a stream of CNC cells integrates into the linear heart tube between 24-30

hpf, without a bias towards atrium or ventricle, and becomes cardiomyocytes (Cavanaugh et al.,

2015). The integration into the heart tube is concurrent with SHF cardiomyocyte integration, which occurs primarily at the ventricle (de Pater et al., 2009). Interestingly, the contribution of cardiomyocytes from both populations are interconnected. Ablation of CNC not only impairs cardiac development, but also inhibits recruitment of SHF cardiomyocytes (Cavanaugh et al.,

2015). Additionally, the two systems are both dependent on Fgf signalling for proper integration into the arterial pole (Cavanaugh et al., 2015; Marques et al., 2008; de Pater et al., 2009). Later in development (50 hpf+), a second wave of CNC cells invade and populate the ventral aorta and bulbus arteriosus (Cavanaugh et al., 2015). Fgf signalling is necessary for both systems to contribute to the heart, and while it is dispensable for CNC cell integration during the linear heart tube phase, this is not the case for the second wave (Cavanaugh et al., 2015; Marques et al., 2008; de Pater et al., 2009). Pharmacological inhibition of Fgf signalling results in fewer cardiomyocytes integrating into the atrium (de Pater et al., 2009). Interestingly, the SHF cardiomyocyte contribution is more prominent at the arterial pole, however, venous differentiation of cardiomyocytes from the SHF does occur (de Pater et al., 2009). The latter may be partially regulated by BMP signalling, as loss of signalling produces hearts with a reduced atrial size, while ventricles are unaffected (Marques and Yelon, 2009). Ultimately, loss of entry and integration of either population of cardiomyocytes into the first heart field results in smaller hearts with multiple cardiac defects.

1.4.4 Class 3 Semas in Heart Development

Semas and their receptors – Plxn and Nrp –play crucial roles in cardiovascular development

(Epstein et al., 2015; Valdembri et al., 2016). Of the class 3 Semas, Sema3a, Sema3c, and Sema3d are identified as key players in normal cardiogenesis. Sema3a null mice present with altered

sympathetic cardiac innervation patterns that result in sinus bradychardia (slowed heart beat), while cardiac overexpression of Sema3a in mouse reduces sympathetic innervation and produces ventricular tachycardia (fast heart beat) (Ieda et al., 2007). Neural crest derived SEMA3C is necessary for endocardial cushion formation, as Sema3c null mice fail to septate the cardiac outflow tract, resulting in postnatal lethality (Plein et al., 2015). Lastly, Sema3d is expressed by mesocardial reflections located between the splanchnic mesoderm and the venous pole of the heart.

Sema3d null mice present with anomalous pulmonary venous connections, suggesting a role in normally repulsing pulmonary venous ECs towards the left atrium (Degenhardt et al., 2013).

1.5 Hypothesis and Aims

1.5.1 General Hypothesis

Gene expression is tightly controlled in multicellular systems both spatially and temporally. A fundamental question in developmental biology is how cells interpret the environment around them in order to execute downstream functions such as signalling, proliferation, differentiation and morphogenesis. By understanding in greater detail how a single extrinsic cue can provide spatial coordinates to the cells that respond to them, we better understand cellular and tissue development. It is increasingly clear that the originally identified chemotactic

Sema molecules play additional roles outside of coordinating vessel and axonal migration (Alto and Terman, 2017; Jongbloets and Pasterkamp, 2014). I hypothesized that secreted Semas present outside of known axonal or vascular guidance areas in the embryo could act as spatiotemporal cues for developing cells. My preliminary data identified sema3f expression in

RPCs, the RPE and cardiomyocytes. As such, I hypothesize that Sema3f signalling functions in a discrete and tissue specific manner to provide functional spatial information that allow

cellular networks in the embryo to develop. This hypothesis will be addressed by the following

Aims:

1.5.2 Specific Aim 1. To elucidate the role of Sema3fa in progenitor cells of the retina during neurogenesis.

Preliminary data indicates that sema3fa is expressed within a restricted population of RPCs.

Considering the expression occurs prior to when axons and vessels in the eye are being guided, I hypothesize a new progenitor-associated role for Sema3fa function in the embryonic eye. To assess a role for Sema3fa on RPCs, I will: a) characterize the expression of sema3fa during retinogenesis, and b) use CRISPR/Cas9 to generate a loss of function mutant to characterize retinogenesis by cell marker analysis and functional assays.

1.5.3 Specific Aim 2. To investigate an endogenous role for Sema3fa on orbital vasculature in the embryo and adult.

A growing body of literature has implicated SEMA3F as a potential antiangiogenic within the neural retina in mouse and human, with loss of SEMA3F correlated with CNV and AMD in humans. However, an endogenous role for Sema3f in the formation and maintenance of the vessel supply to the eye has not been explored. To assess the role of Sema3f in maintaining avascularity of the retina, I will: a) characterize sema3fa expression in the larval and adult neural retina and

RPE, and b) generate an EC specific fluorescent transgenic on the sema3fa mutant background to assess the intraocular and extraocular vessels in wildtype and mutant embryos and adults.

1.5.4 Specific Aim 3. To determine the role of Sema3fb in cardiomyocyte differentiation during cardiogenesis.

Prior reports have identified the expression of multiple Sema3s within the myocardium of mouse and chick (Epstein et al., 2015; Jin et al., 2006), however, functions were not explored. To assess a role for Sema3fb on cardiomyocytes, I will: a) characterize the expression of sema3fb

during cardiogenesis, and b) use CRISPR/Cas9 to generate a loss of function mutant to characterize cardiogenesis by marker analysis and heart function assays.

Chapter Two: Materials and Methods

2.1 Animals

2.1.1 Zebrafish strains and maintenance

Zebrafish (Danio rerio) were maintained according to standard procedure on a 14-h light/10-

h dark cycle at 28°C. Embryos were obtained by natural spawning, raised in E3 medium

supplemented with 0.25 mg/L methylene blue and staged by convention (Fishman et al., 1997);

(Kimmel et al., 1995). Specifically, adult zebrafish were permitted to spawn for ten minutes before

embryo collection. Embryos were screened at 6 hpf to remove any embryo that was delayed or

unfertilized. This permitted exact staging to be carried out for future time points as all embryos

were properly synced. Pigmentation was inhibited by adding 0.003% (w/v) 1-phenyl-2-thiourea

(PTU) to E3 medium supplemented with 0.25 mg/L methylene blue at 24 hpf.

Wild-type Tupfel Long Fin (TL) was used to identify expression profiles of genes by in situ

hybridization, unless otherwise specified. Three genetic mutants were used (sema3fa, sema3fb Δ19

and sema3fb +10, described in detail below) and maintained as heterozygotes for all initial

experiments, and thereafter maintained as homozygotes and wildtype siblings for future embryonic

collection. Ongoing familial generations were derived from heterozygous incrosses. The Tg(-

6.5kdrl:mCherry)ci5 (Proulx et al., 2010) was used as an outcross for both the sema3fa mutant and

sema3fb Δ19 mutant to label endothelial cells. These lines will be referred to as Tg(kdrl:mCherry),

sema3fa or sema3fb mutants, hereafter. The University of Calgary Animal Care Committee

approved all procedures.

2.2 Genetic Manipulation

2.2.1 Establishment of sema3fa and sema3fb mutant lines

To generate genetic knock-outs using the CRISPR/Cas9 system, sgRNA targeting sema3fa

exon 1 and sema3fb exon 1 were selected following CHOPCHOP query (Montague et al., 2014)

and analysis of secondary structure using Vienna RNAfold Prediction (rna.tbi.univie.ac.at).

sgRNA templates were generated using the 20 bp gene specific oligonucleotide containing the SP6

(5’-GCATTTAGGTGACACTATAGA-3’) promoter sequence (Table 2.1). Template generation

and transcription was carried out in accordance to the protocol described in (Gagnon et al., 2014).

Briefly, the gene specific oligonucleotide was annealed with the Cas9 “constant” oligonucleotide

and single stranded overhangs were filled in by T4 DNA polymerase (New England Biolabs) to

form a double stranded oligonucleotide. All steps were performed in the Thermal Cycler. sgRNA

templates were then gel purified (120bp) and transcribed from using the Sp6 Maxi Kit (Ambion).

sgRNA cleanup was performed using 5 M sodium acetate precipitation and run on a 1% TAE gel

to determine integrity and size. Cas9 mRNA was transcribed from plasmid (Addgene plasmid

#47322) using mMessage Machine T7 (Thermo Fischer).

To generate knock-outs, 1-cell stage embryos were injected with a 1 nl mix of

approximately 56-60 pg sgRNA and 190 pg Cas9 mRNA. Mosaic embryos were raised to

adulthood and crossed with TL fish to identify founders. Heterozygous F1 fish (sema3fa Δ2bp;

sema3fb Δ10bp, sema3fb +10) were outcrossed to TL fish. F2 heterozygous progeny were

intercrossed to generate genotypes (wildtype, heterozygous, homozygous mutant).

To genotype, tissue was collected from single 24 hpf embryos or tricaine methanesulfonate

(MS222; 160mg/L) anesthetized adult caudal fin clippings. Genomic DNA extraction was

Table 2.1. CRISPR knock-out reagents and detection Cas9 Vector/Linearizing Restriction Endonuclease RNA Polymerase pCS2-Cas9 pCS2 / NotI SP6 sgRNA Sequence (5’-3’) RNA [RNA polymerase promoter underlined, gene specific sequence in Polymerase bold, followed by annealing sequence to common oligonucleotide] sema3fa sgRNA GCATTTAGGTGACACTATAGAGAAGACTCGTGGA SP6 ACAGAGGgttttagagctagaaatagcaag sema3fb sgRNA GCATTTAGGTGACACTATAGAGAAGGACAAGAA SP6 GACCCGCGgttttagagctagaaatagcaag crRNA (common AAAAGCACCGACTCGGTGCCACTTTTTCAAGTTGA oligonucleotide) TAACGGACTAGCCTTATTTTAActtgctatttctagctctaaaac Sequencing Sequence (5’-3’) Primer sema3fa forward 5’-CCCATGCAGGACTGATAAATCT sema3fa reverse 5’- AGGAAAGCAGTGGTTGTCATCT sema3fb forward 5’-ATTGCCCCACAAAATAACATTC sema3fb reverse 5’-GTCTACTCTGTGAATTTCCCGC qPCR Primer Sequence (5’-3’) sema3fa forward 5’- GGCACAGGGTTTTCTGCAAG sema3fa reverse 5’- CCAGGCTCCAGTCGGAAAAT sema3fb forward 5’- AGTGACGCATATGGCTCTGC sema3fb reverse 5’- AGGAAGCCTCTTCTGCGAGG

performed in 50µl of 50mM NaOH, boiled for 10 minutes and buffered with 1/10 volume of

100mM Tris-HCl pH 7.4, as described in Meeker et al., (2007) and amplified by PCR using the following primers around the expected mutational site (Table 2.2). Amplicons were sequenced at

University of Calgary DNA Core facility for definitive genotype confirmation.

2.3 Gene Expression Detection

2.3.1 RNA isolation

Total RNA from embryos was prepared using the TRIzol reagent (Invitrogen) extraction method (Peterson and Freeman, 2009). Briefly, 3-5 embryos, at specific developmental ages, were collected in Trizol and homogenized using a 26 ½ gauge syringe. Chloroform was added, vortexed, and centrifuged for 15 minutes at maximum speed to separate phases. The top layer was transferred to a fresh microcentrifuge tube and isopropanol added, vortexed and centrifuged for an additional

5 minutes. The tube was decanted and pellet allowed to air dry. Once dry, 17µl of RNA-se free dH2O, 2µl of Turbo DNAse I 10x transcript buffer and 1µl of Turbo DNAse I (Thermo Fischer) was added and incubated at 37ºC for 15 minutes. Following incubation, 2µl of DNAse Inactivation

Agent (Thermo Fischer) was added, incubated and microcentrifuged. Supernatant was transferred to a fresh microcentrifuge tube and concentration determined by a Nanodrop ND-1000

Spectrophotometer.

2.3.2 cDNA Synthesis

First strand cDNA was made using the Superscript II RT-PCR (Invitrogen, 11904-018) protocol. Specifically, only 100ng of total RNA was converted into cDNA and oligo(dT)12 primers were used to enrich for mRNA transcripts. Final cDNA solution, assuming 100% conversion of

RNA to cDNA was assumed to be 1 ng/µl.

Table 2.2. Probe list for in situ hybridization

Gene Primers (Forward; Reverse) Vector/ RNA [RNA Polymerase Promoter underlined] RE polymerase for antisense atoh7 5’ – TTGAAGAGCCATGAAGCCCC T7 5’ - GAAATTAATACGACTCACTATAGGGTCATTCACAACCCGCCCAAT bmp4a 5’- CTGCCAGGACCACGTAACAT T7 5’- GAAATTAATACGACTCACTATAGGGTGGCGCCTTTAACACCTCAT ccnd1 5’ – GCATTTAGGTGACACTATAGACACTTCCTTGCCAAACTGCC T7 5’ - GAAATTAATACGACTCACTATAGGGCATAGAAGGGGTGGCCAAG G cdkn1c 5’ – GCATTTAGGTGACACTATAGAGAGATTACGAGTGGGAGGCG T7 5’ - GAAATTAATACGACTCACTATAGGGCTCTGTTGGCGGAGCCATTA c-fos pCRII T7 /HindII I crx 5’-GAACCGTCGTGCTAAATGCC T7 5’- GAAATTAATACGACTCACTATAGGGCACCTGGAATTTGTGGTCG crip2 5’-GCCCCAGATGCAGCAAGAAG SP6 5’TCATTTAGGTGACACTATAGACCAGCACCAGTCACAAACACC foxd1 5’-AGGCAACTACTGGACGCTAGACCCTG; T7 5’- GAAATTAATACGACTCACTATAGGGAAACAGACCGTGTAAAAAT ATCACACACTCGGAG foxg1a 5’-GCTAATACGACTCACTATAGGGCAGGAAGAAAAACGGGACGC SP6 5’-GCATTTAGGTGACACTATAGAGATGGGTGAGGGACATGGGG

irx1a 5’- GAGAACAAGGTGACCTGGGG T7 5’- GAAATTAATACGACTCACTATAGGGTGAAGAGGACGAAACGACG A ltbp3a 5’ ACTCACCTTTAGTGCAGCCC T7 5’- GAAATTAATACGACTCACTATAGGGCTTCAAATGGGCGCAAACC C myh6 5’- GGAGTACGTGAAGGGGCAAA T7 5’- GAAATTAATACGACTCACTATAGGGGCTCGTCCCGAAATGAATG C myh7 5’- GCAACTTGGTGAGGGAGGAA T7 5’- GAAATTAATACGACTCACTATAGGGAGCAAGCTTACGGCCTCTTT neurod1 5’- TCAACACACCCTAGAGTTCCG SP6 5’- GCATTTAGGTGACACTATAGA AGGGTGGTGTCAAAGAACGG

neurod4 5’- CAGGAGCAATCGTGGACCTT T7 5’- GAAATTAATACGACTCACTATAGGGGTAGTGAGTCGGATGAGGC G nrp1b 5’-AGGTGTTTGATGGTGCGGAT; SP6 5’-GCATTTAGGTGACACTATAGA CGACGATGCAGCTATCCTGT nrp2b 5’-TCCGGGATGGGAACTCAGAT; SP6 5’-GCATTTAGGTGACACTATAGA ATCGGCCGTCATAGAAGCTG plxna3 pCRII/ SP6 EcoRV ptf1a 5’-AGACCACTTGGACACAGACG T7 5’- GAAATTAATACGACTCACTATAGGGGGATCTTGTTCCTCGGTGG rlbp1b 5’- GCAATGGCGGTTGTTAGTGG T7 5’- GAAATTAATACGACTCACTATAGGTAGCTCTGTGGGCTTGATGC rpe65a 5’- TCAGCCGTTTTGAACACCCT T7 5’- GAAATTAATACGACTCACTATAGGGACTTGGGAACTGCACAACC A sema3fa pCR4 / T7 PmeI sema3fb pCR4 / T3 NotI tbx5 From D. Kurrasch tnnt2a pCRII/ SP6 EcoRV vax2 5’-TGATCAAGCCAGTATGGGCGATGG; T7 5’-TAATACGACTCACTATAGGGATCCTTGCGCTCTATCCTCA vsx2 5’- GAGCACACTGGACTCCTTCC T7 5’- GAAATTAATACGACTCACTATAGGGTGCATCCCTAGAAGCCAGG

2.3.3 Real Time Quantitative PCR (RT-qPCR)

Each RT-qPCR had a 10 µl final volume containing 0.5 µl of cDNA, 500 nM of each primer and 5 µl of SYBR Green QuantiTect RT-qPCR master mix (Qiagen). All primers used are listed in Table 2.1. An Applied Biosystems QuantiStudio 6 Real-Time PCR system was used for all experimental procedures. Gene expressions were normalized relative to reference gene beta- actin. Reactions were performed in three technical replicates, and all results made from three independent biological replicates. Fluorescence was measured after each cycle and a melt curve analysis conducted after to verify purity of sample. Equipment software automatically calculated baseline, threshold Cq, and fold changes.

2.3.4 in situ hybridization

Digoxigenin or FITC-labelled RNA probes were synthesized as previously described

(Thisse and Thisse, 2008). Probes were generated from plasmid templates or through PCR products containing an SP6/T3/T7 RNA polymerase binding sequence on the reverse primer

(Table 2.2). Whole-mount in situ hybridization was performed as previously described with some minor modifications (Thisse and Thisse, 2008). All steps were carried out in microcentrifuge tubes on a nutator, except for staining. Steps with gradient changes in hybridization buffer: 2 x SSC:- and 0.2 x SSC:-PBST solutions were omitted. The 2 x SSC step was carried out at 70 °C and 0.2 x SSC at 37°C. Blocking embryos was omitted and instead directly incubated in anti-DIG-AP, Fab fragments antibody (Roche) solution in blocking buffer at room temperature for 2 hours. Embryos were washed in alkaline buffer, transferred to a 24 well culture dish (Thermo Scientific) and stained by NBT/BCIP (Roche). NBT/BCIP was used at a 2.5/3.5µl/ml ratio, respectively. Once desired staining intensity was reached, embryos were transferred to microcentrifuge tubes, fixed in 4% paraformaldehyde (PFA), and washed in 1 x PBST. Proteinase K digestion was omitted for

embryos 72 hpf and younger. Otherwise, embryos were incubated in 5ug/ml Proteinase K for 30 minutes prior to addition of the riboprobe, to increase permeability. When directly comparing in situ levels between genotypes, embryos were processed in the same vial, with mutant tails clipped to distinguish them from wildtype controls. Slide in situs were processed similarly and coverslipped using Aquapolymount (Polysciences Inc).

To detect apoptotic cells in situ, ApopTag Peroxidase In Situ Apoptosis Detection kit

(Millipore) was used as per the user guide, however diaminobenzidine (DAB) staining was carried out over night to allow for development of the chromogenic reaction on the whole embryo.

2.3.5 Immunolabelling

Embryos were fixed overnight in 4% paraformaldehyde in PBS, and either processed for cryosection or whole-mount immunostaining. Briefly, samples were blocked in 10% bovine serum albumin, 2% normal sheep serum in PBS with 0.1% Triton X-100 (PBST) for 30 minutes and transferred to primary antibody solutions in 1/10 blocking buffer overnight at 4°C. Samples were then washed with PBST, and incubated in secondary antibody (Alexa Flour 488/555 rabbit/mouse) supplemented with 5mg/ml Hoechst in PBST for 45 minutes. Embryos were fixed and imaged or processed further for sectioning. The following primary antibodies were used: calbindin (1:200,

Sigma), GABA (1:3000, Sigma), GS (1:500, Millipore), HuC (1:250, Molecular Probes), MF20

(1:500, DSHB), Pax6 (1:500, Covance and Biolegend), PCNA (1:500; Sigma), pHH3 (1:500,

Millipore), zn5 (1:500, DSHB), zn8 (1:500, DSHB), zn12 (1:500, DSHB), zpr1 (1:1000, ZIRC), zpr2 (1:1000, ZIRC), zpr3 (1:1000, ZIRC).

2.3.6 RNA sequencing (RNA seq)

For RNAseq analysis, six 36 hpf embryos were collected from each genotype and processed for whole RNA extraction, as described above. A total of 9 RNA samples were

generated, representing three technical replicates for each genotype. Total RNA was submitted to

the Center for Health Genomics and Informatics at the University of Calgary, and the RNA

Integrity Number (RIN) quantified by Agilent 220 Tapestation equipment. All samples submitted

had a RIN >8.0. Next, an mRNA library was prepped for each sample using the Illumina TruSeq

Stranded mRNA Library, LT kit. Libraries were then quantified, by Illumina Kapa qPCR Library

Quantitation Assay, to determine whether enough amplifiable molecules were present to ensure

sequencing results were not skewed by library concentration differences. An Illumina NextSeq

500 sequencer was used to provide high-throughput analysis with 400 million read sequencing of

1x75 bp, generating 30 Gb of sequence. Transcripts were bioinformatically analyzed by Dr. Paul

Gordon at the Center for Health Genomics and Informatics. Only transcripts with a differential

expression of a False Discovery Rate of < 0.05 for the Wald Test, and passed the Likelihood Ratio

Test, were reported. RNAseq data visualization was generated in conjunction with Dr. Paul

Gordon (University of Calgary).

2.4 Functional Analyses

2.4.1 Intracardial Injections

Embryos were anesthetized and oriented in a 3% agarose bottom dish for injection.

Embryos were delivered 10mg/ml of 2,000,000 MW tetramethylrhodamine dextran

(ThermoFisher, D7139) via cardiac injection through a glass pulled pipette into the sinus venosus

at 7 or 9 dpf. Samples with successful delivery of dextran (checked for fluorescence immediately

after injection) were fixed overnight at 4ºC in 4% PFA approximately 24 hours post injection.

Vascular infiltration into the neural retina was determined by confocal microscopy on 40 µm

transverse cryosections. For adult experiments, fish were anesthetized in tricaine, injected in the

gill with 5-10 µl of 10mg/ml dextran through a 26 ½ gauge needle, and fixed 4 hours post injection.

Eyes were processed for 20- 40 µm transverse cryosections.

2.4.2 Heart Function Analysis

Embryos (72 hpf) were minimally anaesthetized with minimal tricaine, and positioned on their right hand side laterally on a glass slide in a droplet of water to be imaged by 30 frames per second video microscopy using a Leica DM5500 B microscope equipped with a Leica DFC365

FX. This orientation allows for the best resolution of the embryonic ventricle. Embryos were individually filmed for 15 seconds. Heart rate per minute was determined by counting the number of beats in 15 seconds and multiplying by 4. All other functions were determined by taking frame stills during systole and diastole across three heart beats, and conducting all measurements as averages over three heart beats. Measurements were made in accordance with (Hoage et al., 2012) using ImageJ. Briefly the following equations were used following measurements of ventricle length and width during diastole and systole. Ventricle volume = (0.523) (Ventricle Width)2

(Ventricle Length). Fractional Shortening= (100) (Ventricle Width Diastole-Ventricle Width

Systole)/ Ventricle Width Diastole). Stroke Volume = Ventricle Volume Diastole – Ventricle

Volume Systole. Ejection Fraction = (100) (Stroke Volume/Ventricle Volume Diastole). Cardiac

Output = (Heart Rate) (Stroke Volume).

2.4.3 Light exposure to detect c-fos

Detection of c-fos expression induced by light was carried out as previously defined

(Bertolesi et al., 2014). Briefly, embryos were incubated in the dark from 48 hpf to 72 hpf.

Embryos were then exposed to light (2500 lux) for 30 minutes before fixing and processing for whole-mount in situ hybridization.

2.4.4 Zebrabox Visual Function Assay

Zebrafish visual startle response was conducted as previously described (Scott et al., 2016).

Briefly, embryos (72 hpf or 5 dpf) were habituated in 96-well plates for 30 minutes prior to

experimentation. Plates were placed inside Zebrabox (Viewpoint Live Sciences) and activity

recorded in 1-second time bins for the duration of the 32.5 minute experiment. Zebrabox is

illuminated by infrared light allowing for motion detection in darkness. Embryos are exposed to 5

consecutive flashes of light (960 lux) and their movement recorded. Data was exported and

analyzed using free software Visual Interrogation of Zebrafish Manipulations (VIZN). All

behavioural assays were repeated thrice.

2.5 Histology

2.5.1 Cryosections

Embryos were fixed in 4% PFA overnight at 4°C, at the appropriate stage or following in

situ hybridization or immunolabelling, washed in PBST and placed in 35% sucrose (EM Science)

in PBS overnight at 4°C. Embryos were embedded in molds containing Optimal Cutting

Temperature (OCT; TissueTek), oriented and frozen at -80°C for ten minutes. Sections were made

on a Leica CM 3050S cryostat at 12 µm or 40 µm, and either stored at -80°C for future processing

or coverslipped using Aquapolymount (Polysciences Inc) and stored at -20°C until imaged.

2.5.2 Plastic sections

Embryos were embedded in JB4 medium (Polyscience, Warington, PA) as described in

Sullivan-Brown et al. (2011) and sectioned at 7 µm using a Leica microtome. Briefly, embryos

were fixed in 4% PFS overnight at 4°C, washed in PBS and dehydrated in 100% EtOH. Following

dehydration, embryos were soaked in infiltration solution (2-hydroxyethyl methacrylate/benzoyl

peroxide) until they sank to the bottom of the tube and then transferred to molds filled with

embedding solution (infiltration with N,N-Dimethylaniline), positioned and allowed to harden overnight.

2.5.3 Hematoxylin and Eosin Staining

Haematoxylin and Eosin (H&E) staining of plastic sections was carried out in accordance to the protocol described by Sullivan-Brown et al. (2011) with the following modifications. Acid and base washes were omitted, Harris Modified Hematoxylin Solution (Sigma) and Eosin Y (EMD) were used and regular tap water was used instead of prescribed substitute. Slides were air dried for

15 minutes and coverslipped using Permount (Fischer Scientific).

2.5.4 Heart Measurements

Cardiomyocyte morphometries (area and perimeter) were measured by ImageJ software at 48 hpf. Cells with clearly visible outlines following confocal projection of the zn8 immunolabelling.

Of note, only cells central to the outer curvature of the ventral ventricle were assessed following these restrictions. Circularity was used to distinguish cellular morphologies in which 1 represents a perfect circle. This was calculated by the following formula: Circularity = 4 πArea/Perimeter2.

Measurements were averaged from 5-15 cells per embryo.

2.5.5 Retinal Cell Counts

Cell counts were made from single 40 µm central retinal sections at 72 hpf. All sections were counterstained by Hoechst. Specifically, RGC counts were made from cells within the RGC layer that are HuC+, amacrine cell counts were made from cells within the INL that are Pax6+ and bipolar cells were made from the Hoechst label of cells that were HuC-/Pax6- and did not have horizontal cell morphology or present in the horizontal cell layer.

2.5.6 Retinal Flat Mounts

Adult retinal flat mounts were prepared as previous described (Raymond et al., 2014).

Briefly, adult kdrl transgenic wildtype and homozygous mutants were anesthetized in Tricaine

(Sigma), cervically dislocated and fixed in 4 % paraformaldehyde. Eyes were submerged in PBS and enucleated, anterior segment removed and short cuts made to relax cup into flat preparation.

Retinas were placed on slides and coverslipped using AquaPolymount.

Primary vessel lengths and diameters were measured as previously described (Bozic et al.,

2018). Retinal flatmounts were imaged by confocal analysis in 1 µm slices. Maximal projections were made and vessel of the optic nerve head determined. From optic nerve head base, primary vessel length measurements were made to when the vessel bifurcates. This same vessel was measured for width. Intercapillary distance measurements were made near the periphery of the retinal flatmounts and averaged across all quadrants, an area where retinal arterial capillaries anastomose with circumferential vein capillaries (Cao et al., 2008).

2.6 Imaging

2.6.1 Stereoscope

Embryos were imaged over a 1% agarose dish using a Stemi SV 11 microscope and an

AxioCam HRc camera with AxioVision software. Cryosections and plastic sections were imaged on a Zeiss Axioplan 2 and an AxioCam MRc camera with AxioVision software. Images were cropped, brightness/contrast adjusted and compiled using Inkscape 0.92.3.

2.6.2 Confocal

For confocal microscopy, live embryos were imaged on the Zeiss LSM 700 microscope after mounting in 0.8% - 1% low melting point agarose (Invitrogen) on a glass bottom dish. Slices were taken at intervals from 1-5 µm on a 10 or 20x objective and subjected to 2 times averaging. Similar

parameters were used for slides, however, mounting was omitted. Z-stacks were processed in Zen

Blue as maximal projections and compiled using Inkscape 0.92.3.

2.7 Statistics

All statistical analysis was performed using Prism 7 software (Graph Pad). Unpaired, non-

parametric tests were exclusively used in all statistical tests as I did not test each individual data

set for normality in distribution, which could result in artifactual statistical significance doing

ordinary ANOVA’s and t-tests. For situations containing two sets of data, a Mann Whitney U test

was used. When groups of three or more data sets were tested, a non-parametric one way ANOVA

with a Kruskal-Wallis test was used.

All data sets for quantification (qualitative scorings or absolute measurements) were

analyzed in a blinded fashion.

Chapter Three: Sema3fa signalling during retinogenesis

3.1 Introduction

The mature neural retina of vertebrates contains seven differentiated cells types: retinal ganglion cells (RGCs), amacrine cells, bipolar cells, horizontal cells, cone and rod photoreceptors, and Müller glia (Cepko et al., 1996; Schmitt and Dowling, 1994). All these cell types develop from a pool of retinal progenitor cells (RPCs) found within the developing optic cup. Amazingly, the process to generate each of these cell types is achieved through a mix of stochastic and deterministic pathways in which cells are born in a temporally regulated fashion – first RGCs, followed by the cells of the inner nuclear layer (INL) and cones, and lastly the rod photoreceptors of the outer nuclear layer (ONL) and the Müller glial cells (Chen et al., 2012; Gomes et al., 2011;

Jin, 2017). Specification, commitment, proliferation, and differentiation towards a single cell fate is through the integration of the activity of intrinsic transcription factors and extrinsic signalling pathways (Amini et al., 2018; Cepko, 1999; Dyer and Cepko, 2001; Jin, 2017). Little is known, however, about how at the molecular level specific cell fates are chosen by a progenitor cell during retinogenesis.

Intrinsic factors, such as transcription factors, are well known for their roles in progenitor specification, and cell cycle progression and cessation (Cepko, 1999; Dyer and Cepko, 2001; Farhy et al., 2013; Luo et al., 2012; Wong et al., 2015). Generally, these same intrinsic factors are reemployed in the commitment and differentiation of cell specific precursors towards distinct cell types within the neural retina (Inoue et al., 2002; Luo et al., 2012; Sinn et al., 2014). Extrinsic factors, such as Fibroblast growth factor 8 (Fgf8) and Fgf3, are necessary for retinal differentiation through induction of the transcription factor atonal 7 (atoh7) in a spatially restricted region of the

central and ventronasal retina of chick and fish, respectively (Martinez-Morales et al., 2005).

However, morphogens like Fgf, diffuse and act in a concentration dependent manner (Hillenbrand et al., 2016), and considering the small size of the zebrafish retina are likely to be present in the environment of all RPCs. This raises the question as to whether additional levels of spatially restricted environmental signals modulate the cellular dynamics of RPCs.

Class 3 Semaphorins (Semas) are secreted Semas used for spatial chemotactic guidance of

vessels, axons and neural crest cells (Alto and Terman, 2017; Fujisawa, 2004; Jongbloets and

Pasterkamp, 2014; Mecollari et al., 2014). Interestingly, Sema3s have non-guidance roles in that

Sema3d promotes proliferation of zebrafish hindbrain neuroepithelial cells (Berndt and Halloran,

2006), and SEMA3B at the ventricular, apical side of the murine neural tube controls cell division

of neuroepithelial cells by dictating the orientation of the mitotic spindle (Arbeille et al., 2015).

SEMA3 signalling is mediated through the canonical Plexin (Plxn) receptors and coreceptor,

Neuropilin (Nrp) (Epstein et al., 2015). No neural retina phenotype is described in Sema3f

knockout mice, despite evidence of its expression within RPCs and differentiated retinal neurons

(Blackshaw et al., 2004; Buehler et al., 2013).

In this chapter I take advantage of the zebrafish model as a tool to study the development

of retina, which is complete within the first three days of life (Avanesov and Malicki, 2010). I

demonstrate that sema3fa is a novel extrinsic regulator of retinogenesis by using a CRISPR/Cas9

generated sema3fa mutant. I show that sema3fa, and not its orthologue sema3fb, is expressed

within RPCs of the temporal retina throughout retinogenesis, and subsequently transiently in

differentiated RGCs. Later in the larvae, sema3fa expression becomes restricted to the progenitors

of the ciliary marginal zone (CMZ) and INL cells. Loss of sema3fa signalling disrupts retinal cell

type specification, but not the initial specification and patterning of RPCs. Within the embryonic

retina, Sema3fa loss disrupts the differentiation transcriptome of progenitors that produce RGCs

and amacrine cells. Ultimately, however, progenitors in the absence of Sema3fa do exit the cell

cycle to produce these two cell types, though amacrine cells numbers are reduced significantly and

specifically in the temporal retina. Additionally, the organization of the progenitors within the

CMZ of the larval retina of mutants is changed. This is the first time a Sema is reported to control

within a developing tissue the production of neurons from a progenitor cell in a spatial manner.

3.2 Results

3.2.1 sema3fa is expressed by RPCs throughout retinogenesis

The expression patterns of the eight members of class 3 Semas in zebrafish over visual

system development was reported previously (Callander et al., 2007), with sema3fa as the only

sema3 expressed by presumptive RPCs, prior to when retinal neurons are generated and grow

axons. In order to confirm the expression pattern in progenitor cells, I analyzed from 15-72 hours

post fertilization (hpf) the spatiotemporal expression of both sema3fa and sema3fb, which are

duplicated orthologs of human SEMA3F (Figure 3.1K-L) (Yu and Moens, 2005). During bilateral

eye vesicle elongation (13 ss, 15hpf) (Figure 3.1A,F) and invagination around the lens to form

the eye vesicle (18 hpf) (Figure 3.1A’,F’) (Schmitt and Dowling, 1994), both orthologues are

expressed by the temporal RPCs. sema3fa expression persists in the temporal RPCs through the

completion of optic cup morphogenesis (24 hpf) (Figure 3.1B,B’), and on through the onset of

neurogenic differentiation (36 hpf) (Figure 3.1C,C’). The central retina is post-mitotic by 48 hpf,

with progenitors residing in the peripheral retina and the stem-cell niche of the CMZ (Hu and

Easter, 1999). sema3fa expression is restricted to post-mitotic RGCs of the RGC layer by 48 hpf

(Figure 3.1D,D’), and likely bipolar cells of the outer INL by 72 hpf (Figure 3.1E,E’). CMZ

expression of sema3fa persists from 48 hpf (Figure 3.1D-E,D’-E’) and into adulthood (data not

Figure 3.1. sema3fa, and not sema3fb, is expressed by retinal progenitor cells throughout retinogenesis. Whole-mount in situ hybridization (A-E) and respective sections (A’-E’) of staged embryos show persistent, but dynamic, expression of sema3fa transcript throughout retinal development. Specifically, expression is restricted to progenitors of the temporal retina (TR), retinal ganglion cell layer (arrow), differentiated outer portion of the inner nuclear layer (black arrowhead) and the ciliary marginal zone (CMZ, red asterisk). sema3fb expression is also detected by whole-mount in situ hybridization (F-J) and respective sections (F’-J’) of staged embryos.

While sema3fb is expressed in the temporal retina between the 13 ss–18 hpf (F,F’), neural expression is lost and mRNA becomes restricted to the lens (l) and hyaloid vasculature (blue arrowhead). Cladogram (K) and clustal similarity matrix (L) show high conservation of sequence between the orthologues sema3fa and sema3fb (81.53%), and between zebrafish sema3fa and human SEMA3F (71.09-72.7%). M) Model of retinal neurogenesis throughout developmental stages. Bars represent different cell types and when they differentiate/exit the cell cycle, based off amalgamated literature. A: anterior; P: posterior; hpf: hours post-fertilization; N: nasal; T: temporal. RGC: retinal ganglion cell; AC: amacrine cell; HC: horizontal cell; C: cone; BC: bipolar cell; R: rod; M: Muller glia. All scale bars: 50 µm.

shown). The sema3fb orthologue is expressed by the lens from 18–36 hpf (Figure 3.1F’,G-H,G’-

H’), and in the vasculature of the hyaloid artery from 48 -72 hpf (Figure 3.1I’J, I’-J’). In the mature retina (72 hpf), sema3fb is expressed in a spotty manner in horizontal and amacrine cells (Figure

3.1J, J’). Overall, I conclude that sema3fa is the only sema3 expressed during and throughout retinogenesis by RPCs as new neurons are born in the retina (Figure 3.1M).

To determine protein expression of Sema3fa, I used two antibodies. One was generated against zebrafish Sema3fa (GeneTex) and the other against mouse Sema3F. Both antibodies failed to recognize any protein by Western blot, and the mouse Sema3F antibody appeared to label multiple Sema3s within the developing zebrafish embryo by immunohistochemistry.

3.2.2 Sema3fa is necessary for initial retinal cell type specification

3.2.2.1 Generation of the sema3fa genetic mutant

Through CRISPR/Cas9 gene editing technology (Gagnon et al., 2014), I generated a sema3fa loss of function mutant, by using three sgRNAs (Table 2.1) to target exon 1 (Figure 3.2A), which produced a single founder with a 2 bp deletion. The 2 bp deletion produced a predicted protein size of 76 aa due to a premature truncation within the 500 aa SEMA domain (Figure 3.2B) that is necessary to elicit intracellular signalling (Tamagnone et al., 1999). In the absence of a useful antibody to specifically detect Sema3fa, I used RT-qPCR to detect relative transcript levels of mRNA isolated from 48 hpf wildtype and homozygous mutant embryos. There was a clear reduction in sema3fa mRNA levels in the mutant as compared to wildtype (Figure 3.2C), suggestive of nonsense mediated mRNA decay (Hentze and Kulozik, 1999).

3.2.2.2 Sema3fa deficient retina exhibit an abnormal retinal cell specification program

Sema3f null and conditional mouse knockouts have been analyzed for differences in neural connectivity across many brain regions, CNC cell ganglia formation and behavioural phenotyping

Figure 3.2. Generation of the sema3fa genetic mutant. (A) Chromosomal overview of the sema3fa locus targeted by CRISPR/Cas9 mutagenesis to exon 1 (guide RNA’s: A1-3) and primers used to sequence the mutation. UTR: untranslated region; F: forward primer; R: reverse primer.

(B) Schematic representation of wildtype and premature stop codon mutant proteins. sema3fa -/- have a 2 bp deletion (dashes) which produces a predicted product of 76 amino acids (aa). Striped patterns represent missense amino acids (miss.aa). (C) RT-qPCR of sema3fa mRNA in wildtype and sema3fa-/- embryos at 48 hpf suggests nonsense mediated decay of mRNA transcript (N=2).

Error bars represent standard error of the mean (SEM).

(Gammill et al., 2006, 2007; Ito et al., 2008; Matsuda et al., 2010, 2016; Sahay et al., 2003;

Takeuchi et al., 2010; Walz et al., 2007). These analyses support the importance of Sema3F as a chemotactic molecule during normal brain development. Genotype analysis of pre- and peri-natal mice found a normal Mendelian distribution. In the zebrafish sema3fa mutants I assessed both head-to-tail body size and lateral eye size. No differences were found between gross body lengths in the different genotypes (Figure 3.3A-B), and so I measured the ratios of eye to body size in embryos from 36–72 hpf, in order to assess whether eye size differs between the different genotypes (wildtype, heterozygote, homozygote). The eye to body ratio was comparable across genotypes (Figure 3.3C). Overall, these data suggest no gross abnormalities in the size or development of sema3fa mutant embryos.

The axons of RGCs, the only output neuron of the retina, cross the midline at the optic chiasm and terminate in the optic tectum in a spatially organized manner (Stuermer, 1988). Spatial organization of axon terminals is dictated by the location of the cell bodies within the retina. This mapping of location of the cell body onto the site of termination of their axons within the tectal target is known as retinotopic mapping (Lemke and Reber, 2005). The initial establishment of polarized axes within the neural retina is fundamental to this process. Considering the numerous reports of Sema3f in axonal guidance (Ito et al., 2008; Matsuda et al., 2010; Sahay et al., 2003;

Takeuchi et al., 2010; Walz et al., 2007), and its expression in zebrafish RPCs at the onset of retinal neurogenesis (Figure 3.1B-C), I analyzed the expression of naso-temporal and dorso-ventral patterning markers within the 24 hpf eye by in situ hybridization. No discernible differences were observed in the expression label across genotypes for the temporal retina marker foxd1 (Figure

3.4A-D) (Takahashi et al., 2009), nasal retina marker foxg1a (Figure 3.4E-H) (Xuan et al., 1995), ventral axis marker vax2 (Figure 3.4I-L) (Barbieri et al., 1999), or dorsal axis marker tbx5a

Figure 3.3. sema3fa mutants are physically indistinguishable from wildtype embryos.

Representative bright field images of wildtype (A) and homozygous sema3fa mutants (B) at 72 hpf show no obvious morphological defects. C) To determine potential changes on eye size, anterior-posterior measurements were made of the eye and body axis across developmental stages

(n=5-10 per genotype and stage). Plot represents means (±SEM) of eye/body length ratios indicating no significant differences across genotypes. Scale bar: 1 mm.

Figure 3.4. Retinal patterning is unaffected in sema3fa mutants. Whole-mount eyes (dashed outline) imaged laterally at 24 hpf and 36 hpf for different patterning genes. Temporal tissue

(foxd1, A-D) and nasal tissue (foxg1a, E-H) are patterned normally at 24 hpf (N=1, n’s in panels) and 36 hpf (N=3, n’s in panels). Ventral tissue (vax2, I-L) and dorsal tissue (tbx5a, M-P) are also patterned normally at 24 hpf (N=1, n’s in panels) and 36 hpf (N=2, n’s in panels). N: nasal; L: lens; T: temporal. Scale bar: 50 µm.

(Figure 3.4M-P) (Chapman et al., 1996). I conclude that establishment of retinal patterning is normal in sema3fa mutant embryos.

sema3fa is expressed throughout the temporal eye vesicle during optic cup morphogenesis

(13 ss, Figure 3.1A), retinal cell type specification (28 hpf +), and RPC cell cycle exit and terminal differentiation (36 hpf +, Figure 3.1C-E). To determine if Sema3fa plays a role at any of these steps I sequentially analyzed embryos at each phase of retinogenesis. By the 13 ss, the optic vesicle has elongated, and RPCs have acquired identities that are maintained (Schmitt and Dowling,

1994). I assessed the expression of two key factors in retinal development; fgf8a is a morphogen that patterns the forebrain and midbrain (Suzuki-Hirano and Shimogori, 2009) and is important for

RPC development, while pax6a is a transcription factor that functions in eye and forebrain development (Georgala et al., 2011). I found no significant differences in the expression of fgf8a

(Figure 3.5A-B) across the body of the embryo, and no differences in expression of pax6a within the eye vesicle (Figure 3.5C-D). Together, these data suggest that Sema3fa loss does not disrupt the specification of RPCs.

Next, I asked whether the production of early born neurons was impacted by the loss of

Sema3fa. I analysed by ISH the expression of two key basic helix loop helix (bHLH) transcription factors, atoh7 (Kay et al., 2001) and neurod4 (Inoue et al., 2002), involved in the specification of

RGCs and amacrine cells, respectively. Expression of both atoh7 (Figure 3.6A-D) and neurod4

Figure 3.6E-F) was reduced dramatically in the retinas of over 70% (n=14/18; n=14/20, respectively) of homozygous mutants as compared to wildtype embryos (n=15; n=20, respectively). Interestingly, reduced expression was not obviously confined to any specific retinal axis. Overall, these data suggest that early retinal progenitor specification is normal, however, once

Figure 3.5. Early embryonic patterning is normal is sema3fa mutants. Embryos were processed for RNA in situ hybridization for the morphogen fgf8a (A-B) and the eye specification transcription factor pax6a (C-D). At the 13 ss, sema3fa mutants display normal patterning of fgf8a as viewed in lateral wholemounts (N=1), and pax6a (N=1) as viewed dorsally. A: anterior; e: eye; mhb: midbrain-hindbrain boundary; nk: neural keel; P: posterior; s: somite. Scale bar in A-B) 100

µm and C-D) 50 µm.

Figure 3.6. Retinal specification transcription factors are downregulated in sema3fa mutants. Expression of atoh7 by in situ hybridization in wholemount (A,B) and in sections (C,D) is reduced in the retinas of sema3fa mutants (B,D N=3, n=14/18) as compared to wildtype embryos, where atoh7 is expressed throughout the entire differentiating retina (A,C N=3, n=13/15) at 36 hpf. Additionally, reduced expression of neurod4 was evident in mutant retinas (F, N=3, n=14/20) compared with wildtype embryos (E, N=3, n=19/20) at 36 hpf. N: nasal; T: temporal.

Scale bar: 50 µm.

retinal progenitors begin to generate specific cell types, the normal patterns of expression of specification genes is disrupted.

3.2.3 Sema3fa is involved in RGC and inner nuclear layer specification and differentiation

3.2.3.1 Sema3fa deficient embryos have smaller temporal retinas during mid-retinogenesis

Due to the clear reduction in cell type specific transcription factor expression with Sema3fa loss, I next asked if the genesis of retinal neurons was impacted. atoh7 and neurod4 are required for the specification and differentiation of RGCs and amacrine cells, respectively (Jin, 2017).

These two factors also participate in the genesis of all retinal cell types through transcriptional hierarchies (Jin, 2017; Sinn et al., 2014). Thus, multiple retinal cell types could be affected by the expression changes I observed in atoh7 and neurod4 at 36 hpf. As a first assessment of retinal cell genesis, 52 hpf retinas were stained by hematoxylin and eosin to label all cells and plexiform layers within the retina. We used transverse sections where the optic nerve head was present in order to distinguish nasal and temporal retinal sectors (Figure 3.7A-B). While no significant differences in eye diameter were observed with Sema3fa loss when measured from lateral views of the eyes in whole embryos (Figure 3.3C), the area of the temporal retina was slightly, but significantly, smaller by 20% in mutant (N=3, n=10) vs. wildtype (N=3, n=9) embryos (p=0.035, Figure 3.7D).

Of note, the area of the nasal retina was unchanged (p=0.549, Figure 3.7C). Further, while nascent inner- and outer- plexiform layers were present in both the nasal and temporal retina of sections from all wildtype embryos (N=3, n=8) and the nasal retinas of mutants (N=3, 9/9), an inner plexiform layer was unidentifiable in the majority (56%; N=3, n=5/9) of temporal retinas of mutant eyes. These data suggest a change in the neurogenesis of retinal neurons specific to the temporal retina, where sema3fa mRNA was expressed from 18-36 hpf. Of note, polarity of the retinal neuroepithelium was not disrupted in the mutant embryos in that there were no differences in the

Figure 3.7. Temporal retinal tissue is significantly smaller in sema3fa mutants. Representative sections of hematoxylin and eosin stained wildtype (A) and mutant (B) retinas at 52 hpf. C,D)

Area measurements of retina on either side of the optic nerve were measured. Nasal areas are unchanged between wildtype and mutant (p=0.55) embryos. However, temporal tissue is significantly smaller in the mutants (p=0.035). Error bars are standard error of the mean (SEM) over 3 independent experiments (n=9 wildtype, n=10 mutant). Statistics represent the non- parametric Mann-Whitney U test. inl: inner nuclear layer; ipl: inner plexiform layer; l: lens; N: nasal; onl: outer nuclear layer; opl: outer plexiform layer; rgcl: retinal ganglion cell layer; T: temporal. Scale bar is 50 µm.

apical surface immunolabelling of mutant and wildtype retinas with the Zs4 antibody (Figure

3.11E-F).

3.2.3.2 Retinal cell type differentiation is largely spared in mid-retinogenesis in sema3fa mutants

I next used the expression of markers of specification, differentiation and/or cell cycle progression to analyze the mutant retinas at 52 hpf. Interestingly, despite the dramatic decrease in atoh7 at 36 hpf in sema3fa mutants, expression of atoh7 recovered and was unchanged between wildtype and mutant embryos (Figure 3.8A-B). Yet, expression of chemokine receptor 4b

(cxcr4b), a receptor expressed by RGCs and necessary for the guidance of their axons (Li et al.,

2005), was reduced in 90% of mutant (n=9/10) embryos as compared to wildtype (n=10, Figure

3.8C-D). Additionally, expression of irx1a, a homeobox gene required for propagation of the Sonic hedgehog (Shh) signal which drives RGC generation in the retina (Cheng et al., 2006; Masai et al.,

2000) was also reduced in mutants (n=3/5) as compared to wildtype embryos (n=5) (Figure 3.8E-

F). Nonetheless, immunolabeling of cryosections revealed that similar numbers of ELAV like

RNA binding protein (HuC) (Marusich et al., 1994) positive differentiated RGCs were present in retinal sections from wildtype and mutant embryos (Figure 3.8G-H). Together, these data suggest that with the loss of Sema3fa RGCs differentiate, but exhibit altered transcriptional profiles.

I next assessed cells in the INL, using markers that label one or more cell types of this layer. By 52 hpf, amacrine and horizontal cells are present in the INL (Jusuf and Harris, 2009).

Expression of the transcription factor pax6a, which is necessary for timing of amacrine cell differentiation (Farhy et al., 2013), was unchanged across genotypes at both the transcript (Figure

3.9A-B) and protein (Figure 3.9C-D) levels. Nor was expression of the bHLH protein neurod4, which is necessary for amacrine cell genesis (Inoue et al., 2002) (Figure 3.9G-H). Yet, pancreas

Figure 3.8. Terminal retinal ganglion cell differentiation occurs in sema3fa mutants in mid retinogenesis. Lateral whole-mount imaged embryos at 52 hpf processed by in situ hybridization

(A-F). Expression of the transcription factor atoh7 (A-B) necessary for RGC fate is now unchanged between wildtype (N=1) and mutant (N=1) embryos. However, the differentiating marker cxcr4b (C-D) is decreased in mutants (N=2) as compared to wildtype (N=2), as is irx1a

(N=1, wildtype, C, and N=1, mutant, D). Post mitotic expression of HuC reveals no differences in the extent of differentiation of the RGC layer between wildtype (N=3, G) and mutant (N=3, H). inl: inner nuclear layer; ipl: inner plexiform layer; l: lens; N: nasal; onl: outer nuclear layer; rgcl: retinal ganglion cell layer; T: temporal. Scale bar are 50 µm.

Figure 3.9. Amacrine and horizontal cell specification occurs in sema3fa mutants during mid retinogenesis. Lateral whole-mount imaged embryos at 52 hpf processed by in situ hybridization

(A-B, E-H). Expression of the mRNA (N=1, A-B) and protein (N=3, C-D) for transcription factor

Pax6a (A-B) necessary for amacrine and RGC differentiation is unchanged between wildtype and mutant embryos. Expression of ptf1a, necessary for amacrine and horizontal cell fates, is increased

in mutant retinas (N=2) compared to wildtype (N=2). Cell fate specification transcription factor neurod4 is unchanged in wildtype (N=1) and mutant (N=1) retinas. inl: inner nuclear layer; ipl: inner plexiform layer; l: lens; N: nasal; onl: outer nuclear layer; opl: outer plexiform layer; rgcl: retinal ganglion cell layer; T: temporal. Scale bars are 50 µm.

transcription factor 1a (ptf1a), which is necessary for amacrine and horizontal cell genesis (Jusuf and Harris, 2009), was increased in 70% of mutant retinas (n=7/10) as compared to wildtype (n=8,

Figure 3.9E-F). Overall, these data suggest amacrine cells are specified and differentiate in the mutant retinas, but like RGCs exhibit a somewhat altered transcriptional profile with respects to wildtype embryos.

The transcription factors vsx2 and crx, which generate bipolar neurons and photoreceptor cells, respectively, were used as markers to assess the genesis of these two cell types. Additionally,

RPE specification and differentiation was assessed by expression of the RPE marker, retinaldehyde binding protein 1 b (rlbp1b) (Collery et al., 2008). Of note, both Vsx2 and Crx function within progenitor cells to maintain an undifferentiated and proliferative state (Shen and Raymond, 2004;

Wong et al., 2015), though it is unknown whether they function in parallel. While vsx2 expression was upregulated in 63% of mutant (n=5/8) as compared to wildtype (n=8 Figure 3.10A-B) 52 hpf retinas, crx expression appeared unaltered (Figure 3.10C-D). These data suggest preliminarily that bipolar cell genesis could be precocious in the absence of Sema3fa. No difference in expression was observed between genotypes, however, for the RPE marker rlbp1b between controls and mutants (Figure 3.10E-F).

In order to differentiate, neurons must exit the cell cycle. Cycling progenitors were detected by expression of cyclin d1 (ccnd1), while cells exiting the cell cycle were identified by p57kip2

(cdkn1c) (Dyer and Cepko, 2001). Expression of ccnd1 was unchanged between genotypes (Figure

3.11A-B), suggesting that similar numbers of progenitors were present in mutant and wildtype retinas at 52 hpf. cdkn1c expression, however, was reduced in 60% of mutant retinas (n=11/18) as compared to the wildtype controls (n=12, Figure 3.11C-D). These data argue there was a reduction

Figure 3.10. Specification of retinal pigment epithelium, and inner and outer nuclear cells occurs normally in sema3fa mutants during mid retinogenesis. Lateral whole-mount imaged embryos at 52 hpf processed by in situ hybridization. Expression of INL marker vsx2 (N=2), inner and outer nuclear layer marker crx (N=2) and retinal pigment epithelium marker rlbp1b (N=2) are present in wildtype and mutant embryos. vsx2 expression is increased (B) in the mutants as compared to wildtype (A). N: nasal; T: temporal. Scale bars are 50 µm.

Figure 3.11. Retinal progenitors undergo cell cycle progression. Lateral whole-mount images of 52 hpf eyes processed by in situ hybridization (A-D). Expression of a cell cycle progression cyclin (ccnd1) is unchanged between wildtype (N=2) and mutant (N=2). Cell cycle exit marker is decreased in mutant (N=2, n=11/18) as compared to wildtype embryos (N=2). Polarity of the retinal neuroepithelium (zs4) is unperturbed in the mutants (N=1). a: apical; b: basal; N: nasal; T: temporal. Scale bars are 50 µm.

in the numbers of cells leaving the cell cycle and becoming post-mitotic in the mutant retinas.

3.2.4 Sema3fa deficient embryos have reduced numbers of amacrine cells in the temporal retina

3.2.4.1 Temporal amacrinogenesis is reduced with sema3fa deficiency

At 52 hpf we found that the temporal retina was significantly smaller in mutant vs. wildtype retinas. To determine which specific retinal cell populations were ultimately impacted, we first assessed by hematoxylin and eosin staining retinal sections at 72 hpf, when retinogenesis and lamination is largely complete (Amini et al., 2018; Stuermer, 1988). Retinas were comparable in size between genotypes (Figure 3.12A-B), and did not significantly differ in medial-lateral width in the nasal retina (p>0.99, Figure 3.12C), temporal retina (p=0.096, Figure 3.12D) or in the nasal- temporal axis (p>0.99, Figure 3.12E). I assessed the laminar structure of the retina (Figure 3.13A-

D) and found the INL, which encompasses amacrine and bipolar cells, was reduced in thickness in mutants in the temporal, but not nasal, retina (Figure 3.13E). Counting cells within the distinct amacrine cell layer of the INL between the nasal and temporal aspects of the retinas revealed that while RGC and bipolar cell numbers were unchanged (p=0.90; p=0.31 nasal, p=0.87; p=0.71 temporal, respectively) between mutants and wildtype retinas, the total number of amacrine cells was reduced significantly in the temporal (p=0.0002), and not the nasal (p=0.94), mutant retinas

(Figure 3.13F). One possible explanation for the reduced number of amacrine cells in sema3fa mutants is that cells are lost through apoptosis. There was no significant difference, however, in the number of TUNEL positive apoptotic nuclei between genotypes (p=0.52, Figure 3.13G).

3.2.4.2 Retinal cell type differentiation is largely spared in late-retinogenesis in sema3fa mutants

To verify the reduced numbers of cells within the inner portion of the INL where amacrine cells reside, I assessed the expression of amacrine cell markers at 72 hpf. At this time, nuclear

Figure 3.12. sema3fa mutant eyes are the same size as wildtype eyes. Representative sections of hematoxylin and eosin stained wildtype (A) and mutant (B) retinas at 72 hpf. C,D) Width measurements of retina on either side of the optic nerve were measured. Nasal (C, p>0.99) and temporal (D, p=0.096) widths are unchanged between wildtype and mutant embryos. Additionally, the nasal-temporal axis length is unchanged in wildtype and mutant embryos (p>0.99) (E). Error bars are standard error of the mean (SEM) over 3 independent experiments (n=9 wildtype, n=12 mutant). Statistics represent the non-parametric Mann-Whitney U test. inl: inner nuclear layer; ipl: inner plexiform layer; L: lateral, l: lens; M: medial; N: nasal; onl: outer nuclear layer; opl: outer plexiform layer; rgcl: retinal ganglion cell layer; T: temporal. Scale bar: 50 µm.

Figure 3.13. The temporal amacrine cell layer has fewer cells in sema3fa mutants.

Representative sections of hematoxylin and eosin stained wildtype (A,C) and mutant (B,D) nasal

(A-B) and temporal (C-D) retinas at 72 hpf. E. Width measurements for INL in mutants vs. wildtype are not significantly different in the nasal retina (p=0.51), but significantly thinner in the temporal retina (p<0.0001). Similarly, in mutants vs. wildtype eyes, nasal differences were not significant for the amacrine cell layer (ACL, p=0.72), while temporal thickness of the ACL is significantly thinner (p=0.012). Error bars are standard error of the mean (SEM) over 3 independent experiments (n=9 wildtype, n=12 mutant). F) Quantification of the number of HuC+ retinal ganglion cells (RGC), Pax6+ amacrine cells (AC) and Pax6- INL cells for bipolar cells

(BC). Mutants have significantly fewer amacrine cells within the temporal retina (***, p=0.0002) while no significant differences were found in the nasal retina (p=0.94, N=3, n=11 wildtype; N=3, n=12 mutant). No significant differences were found in the RGCs in both nasal (p=0.90) and temporal (p=0.87, N=3, n=12 wildtype; N=3, n=15 mutant) retina or in the BCs nasally (p=0.31) and temporally (p=0.71, N=3, n=9 wildtype; N=3, n=9 mutant). Quantification of apoptotic nuclei in the eye. No significant differences were observed in TUNEL+ cells in the eyes of homozygotes

(N=2, n=10, p=0.52) compared to wildtype controls (N=2, n=10). Statistics represent the non- parametric Mann-Whitney U test. acl: amacrine cell layer; bcl: bipolar cell layer; L: lateral, M: medial; onl: outer nuclear layer; opl: outer plexiform layer; rgcl: retinal ganglion cell layer. Scale bar: 50 µm.

layers are laminated and most cells in the retina are post-mitotic. The transcription factor pax6a is expressed strongly by amacrine cells and more weakly by a subset of RGCs. pax6a expression was unchanged across genotypes at the transcript level (Figure 3.14A-B), however, there appeared to be a reduction in Pax6 protein as assessed by immunohistochemistry in the temporal retina (Figure

3.14C-D) in over half of the mutant embryos (n=8/13). Similar results were obtained using HuC immunolabel (n=8/15, Figure 3.14K-L). In agreement, zn12 label of the cell-surface carbohydrate

L2/HNK-1 (Metcalfe et al., 1990) revealed that the inner plexiform layer of the temporal retina was thinner in 80% of retinas relative to wildtype (n=9/11, Figure 3.14O-P). While amacrine cell numbers are reduced in the sema3fa mutant retinas, the remaining cells continue to express the appropriate cell specific differentiation genes in that expression of ptf1a did not appear different between genotypes (Figure 3.14E-F), nor did expression of amacrine precursor genes neurod1

(Figure 3.14G-H) and neurod4 (Figure 3.14I-J).

To assess whether other retinal cell types in sema3fa mutant retinas are affected, I analyzed the expression profiles of cell type markers at 72 hpf. RGCs were detected by expression of irx1a

(Figure 3.15A-B), the cell adhesion molecule DM-GRASP as detected by the antibody zn8

(Fashena and Westerfield, 1999) (Figure 3.15C-D), and by immunolabelling with the HuC antibody (Figure 3.15E-F). In all cases, there were no discernible differences between genotypes.

In agreement, the number of HuC positive RGCs in the nasal and temporal retina was similar in wildtype and mutant larvae (Figure 3.13F). Müller glia, detected using antibodies for glutamine synthetase (Figure 3.14M-N) and zn12 (Figure 3.14O-P), were present at similar levels within mutant and wildtype retinas. Expression of vsx2 in bipolar cells in the outer INL, and progenitor cells of the CMZ, was also not obviously different between genotypes (Figure 3.16A-D). crx mRNA is restricted to the periphery of the INL and photoreceptor layer, and expression did not

Figure 3.14. Temporal amacrinogenesis and inner plexiform layer development is reduced in sema3fa mutants. Lateral whole-mount imaged embryos at 72 hpf processed by in situ hybridization (A-B, E-J) or cross-sectional immunohistochemistry (C-D, K-T). Expression of the mRNA (N=3, A-B) and protein (N=3, C-D) for transcription factor Pax6a necessary for amacrine and RGC differentiation is present, although reduced temporally, between wildtype (A,C) and mutant (B,D) embryos. Expression of ptf1a (E,F), in horizontal cells (N=2) is unchanged. 94

Amacrine specific expression of neurod1 (G-H, N=2) and neurod4 (I-J, N=1), and protein presence of HuC (K-L, N=3), are unchanged between genotypes, except for temporal reduction in HuC number. Additionally, the presence of Glutamine-synthase (GS, M-N, N=2) and zn12 (O-P, N=3) positive Muller glia appears normal with respects to their radial spokes traversing the entire retina

(white arrows). Additionally, inner plexiform layer thickness (zn12) is qualitatively reduced (red asterisk) in the mutant (N=3) compared to the wildtype control (N=3). inl: inner nuclear layer; ipl: inner plexiform layer; l: lens; N: nasal; onl: outer nuclear layer; opl: outer plexiform layer; rgcl: retinal ganglion cell layer; T: temporal. Scale bars are 50 µm.

Figure 3.15. RGC layer forms normally in sema3fa mutants. Lateral whole-mount imaged embryos at 72 hpf processed by in situ hybridization for irx1a (N=1) is unchanged (A-B).

Immunohistochemistry for RGCs (zn8, C-D, N=1) and differentiated neurons (HuC, E-F, N=3), show no major differences between genotypes. Post mitotic expression of HuC reveals no obvious differences in the extent of differentiation in the RGC layer between wildtype (N=3) and mutant

(N=3). inl: inner nuclear layer; ipl: inner plexiform layer; l: lens; N: nasal; onl: outer nuclear layer; opl: outer plexiform layer; rgcl: retinal ganglion cell layer; T: temporal. Scale bar are 50 µm.

Figure 3.16. Presence of differentiation markers of the retinal pigment epithelium, and inner and outer nuclear cells is normal in sema3fa mutants. Lateral whole-mount imaged embryos

(A-B, E-F, M-N) and respective sections (C-D, G-H) at 72 hpf processed by in situ hybridization.

Expression of cmz and INL marker vsx2 (N=2), inner and outer nuclear layer marker crx (N=2) and retinal pigment epithelium marker rlbp1b (N=2) are all present in wildtype and mutant eyes.

Cross sectional immunohistochemistry for outer segments of rods (zpr1, N=2), and rod and short double cones (zpr3, N=1) is unchanged across genotypes. inl: inner nuclear layer; ipl: inner plexiform layer; l: lens; N: nasal; onl: outer nuclear layer; opl: outer plexiform layer; rgcl: retinal ganglion cell layer; T: temporal. Scale bars are 50 µm.

appear dramatically different between genotypes (Figure 3.16E-H). To further assess the photoreceptors, I immunolabelled for green/red double cones (zpr1 antibody) and rod and short double cones (zpr3 antibody) and found no differences between genotypes (Figure 3.16I-L).

Together these data suggest that genesis of Müeller glia, bipolar cells, and photoreceptors are unaffected by loss of Sema3fa. In contrast, expression of rlbp1b appeared to be reduced in all

(n=8/8) mutant RPE (Figure 3.16M-N), which may reflect the expression of sema3fa by RPE cells.

3.2.4.3 Light perception is not robustly impaired with loss of Sema3fa

Because of a reduction in the numbers of amacrine cells in the temporal retina, I asked whether the change resulted in deficits in visual function. Two assays were conducted. First, embryos were dark-adapted for 24 hours and then exposed to thirty minutes of light to elicit a circuit response which can be detected by a surrogate marker of neural activity, c-fos (Bertolesi et al., 2014). No obvious differences were detected within either the nasal or temporal retina in terms of the presence of c-fos mRNA expressing cells (Figure 3.17A-B). Additionally, I analyzed the embryo startle responses to bright light at 72 hpf (Figure 3.17C) and 5 dpf (Figure 3.17D) using

Zebrabox. I found no significant difference between genotypes in their ability to respond to a light pulse (p=0.95), and while at 5 dpf, mutants responded less, this was not statistically significant

(p=0.20). Together, I conclude that loss of Sema3fa does not result in robust visual impairment.

3.2.5 Sema3fa deficient embryos have changed neurogenic potentials in the ciliary marginal zone

3.2.5.1 The ciliary marginal zone undergoes transcriptional changes with loss of Sema3fa

Given the presence of sema3fa mRNA in the CMZ at 72 hpf, I examined the expression of

CMZ markers in retinal sections from 72 hpf wildtype and mutant embryos. I first asked whether cells of the CMZ were proliferative (ccnd1 positive) and capable of exiting the cell cycle (cdkn1c

Figure 3.17. sema3fa mutants retain visual function. Similar numbers and distribution of neutrally active c-fos+ cells in response to a light stimulus in wildtype (A) and mutant (B) retinae

(N=1). C) Quantification of a light-induced startle response at 72 hpf indicates no significant difference between wildtype and mutant larvae (N=3, n=144, p=0.94). D) Quantification of startle response at 5 dpf show no significant differences (N=3, n=144, p=0.20). N: nasal; T: temporal.

Scale bar: 50 µm.

100

positive). By 72 hpf, cycling progenitors are expected to only exist within the CMZ (Kubota et al.,

2002), and this was true of ccnd1 (Figure 3.18A-D) and cdkn1c (Figure 3.18E-H) in both wildtype and mutant embryos. However, ccnd1 presented with increased (Figure 3.17B) and expanded

(Figure 3.18D) expression in the CMZ of the mutants. In contrast, cdkn1c mRNA was decreased

(Figure 3.18F), but still localized in the CMZ (Figure 3.18H). Overall, these data support a potential ongoing role for Sema3fa on neural progenitor cells residing in the CMZ.

Cells of the CMZ are organized with respects to their stemness, whereby the most stem- like progenitor is found most peripheral and those that are proliferating and specified to exit the cell cycle most central (Cerveny et al., 2010). By 72 hpf, atoh7 is restricted in its expression to the proliferative zone within the CMZ (Cerveny et al., 2010). However, in 92% of sema3fa mutants atoh7 was expanded in the CMZ (n=11/12) (Figure 3.19A-B,N). A subset (42%, n=5/12) of embryos showed total expansion throughout the entirety of the CMZ (Figure 3.19B). I next asked whether the undifferentiated state of the CMZ was impaired with expansion of atoh7 by looking at pax6a (Figure 3.19C-D) and vsx2 (Figure 3.19E-F). While pax6a was unchanged, the domain of expression of vsx2 was expanded. Cycling marker ccnd1 was expanded, as mentioned prior, in over half (n=10/18) of embryos (Figure 3.19G-H), as was immunolabel for S phase proliferating cells by PCNA (Figure 3.19K-L). cdkn1c expression appeared relatively unchanged in spatial organization within the CMZ (Figure 3.19I-J), although was notably decreased in wholemount

(Figure 3.18F). Taken together, these data support a role for sema3fa in regulating progenitor cells transcriptional profiles within the CMZ.

101

Figure 3.18. Retinal progenitor cells of the ciliary marginal zone maintain cell cycle markers in sema3fa mutants. Lateral whole-mount imaged embryos at 72 hpf processed by in situ hybridization (A-B,E-F) and their respective sections (C-D, G-H). Expression of a cell cycle progression cyclin (N=3 whole mount and section, ccnd1) and cell cycle exit marker (N=3 whole mount and N=1 section, cdkn1c) are both expressed by the cmz of wildtype and mutants. However,

102

expansion of ccnd1 is found in 10/18 mutant retinas in section and a reduction of cdkn1c in 10/19 mutant eyes. cmz: ciliary marginal zone; inl: inner nuclear layer; ipl: inner plexiform layer; l: lens;

N: nasal; onl: outer nuclear layer; opl: outer plexiform layer; rgcl: retinal ganglion cell layer; T: temporal. Scale bars are 50 µm.

103

Figure 3.19. The CMZ of the sema3fa mutants expands spatially restricted transcriptional profiles. Expression of neurogenesis related genes in the ciliary marginal zone. Expansion of atoh7

(B, N=2), ccdn1 (H, N=3), vsx2 (F,N=2), and PCNA (L, N=3) expression in mutants is observed.

104

However, the extent of pax6a (D, N=2) and cdkn1c (J, N=1) label are unaffected. M) Percentage of embryos with a CMZ with expanded atoh7 expression in mutant embryos (N=2, n=11/12). Of these, atoh7 expanded into the entirety of the CMZ in 42% of embryos (N=2, n=5/12) (N). Scale bar: 25 µm.

105

3.2.6 sema3fa signalling is eye tissue autonomous

3.2.6.1 Retinal cell type specification deficit is independent of circulating factor changes

sema3fa is expressed in multiple tissues throughout the embryo, including neuroendocrine regions of the brain (Callander et al., 2007; Sahay et al., 2003). Therefore, it is possible that transcriptional changes observed during retinogenesis are a consequence of altered neuroendocrine signalling, which could potentially affect eye development. To address this possibility, I cultured retinal explants independent of the circulatory system and most brain tissue, removing eye progenitors from non-eye sources of Sema3fa-dependent signals. I then asked whether the dramatic decrease in the RGC specification marker atoh7 we observed at 36 hpf in zebrafish sema3fa-/- eyes was recapitulated in culture. I explanted and cultured retinas of wildtype and mutant 24 hpf embryos for 24 or 48 hours. Of note, morphologically the wildtype eye explants after 24 hours resembled a 36 hpf eye, while those 48 hours in culture resembled a 48 hpf eye.

Explants were fixed and assessed by ISH for atoh7 (Figure 3.20). Expression of atoh7 was reduced in all retinas (n=3) compared to wildtype 24 hours post dissection (Figure 3.20B), but no change in expression was observed 48 hours post dissection (Figure 3.20D). These data corroborate those found in vivo and suggest that retinogenesis defects present in the sema3fa mutants occur independently of a more global change in the embryo or brain with loss of Sema3fa.

3.2.6.2 Sema3fa may act in a cell autonomous manner to control progenitor biology within the CMZ

Sema3s signal canonically through Nrp and Plxn receptors (Fujisawa, 2004). In sema3fa mutants, the change in the expression profile of the CMZ in the sema3fa mutants argues that

Sema3f functions in a cell-autonomous manner to control progenitor behaviour within the CMZ.

This model would require that cells within the CMZ express receptors for Sema3fa. In a screen of

106

Figure 3.20. Change in retinogenesis are not due to loss of Sema3fa in tissues outside of the eye. Explanted eyes were cultured for 24 hours post dissection (hpd) or 48 hpd. Eyes were processed by whole-mount in situ hybridization for atoh7 expression. By 24 hpd, mutant explant retinas (N=1) have reduced expression of atoh7 as compared to wildtype explants. At 48 hpd, these differences are resolved (N=1). N: nasal; T: temporal. Scale bars are 50 µm.

107

various plxna and nrp genes, I found that nrp2b and plxna3 were the only receptor genes whose

mRNA was confined to the CMZ at 72 hpf (Figure 3.21B-C). These data support the idea that

Sema3fa functions in a cell autonomous manner within the CMZ.

3.3 Specific Discussion

My data support a role for Sema3fa in regulating the transcriptome of progenitor cells undergoing neurogenesis. I find sema3fa mRNA present in temporal retinal progenitor cells throughout retinogenesis, and interestingly this expression persists within the stem-cell niche of the CMZ to adulthood. Using the genetic mutant of sema3fa, I show that loss of Sema3fa results in a decrease in the expression of markers of the specification of distinct retinal cell types, but no change to retinal progenitor cell specification. There are three main downstream consequences of the altered expression of specification markers. First, on a morphological level, the size of the temporal retina is smaller during retinogenesis and manifests ultimately in mutants as a smaller amacrine cell population in this sector of the retina. Second, I show that transcriptional profiles of differentiating neurons within the RGC and amacrine cell layers differ to what is observed in wildtype embryos, however, cells ultimately differentiate. Finally, I demonstrate that the progenitors of the CMZ require ongoing Sema3fa signalling within their niche to maintain normal transcriptional profiles. Overall, my work implicates for the first time a Sema as a regulator of progenitor cell dynamics and transcriptional differentiation within the retina.

The most robust phenotype in sema3fa mutants is a spatially restricted deficiency in amacrine cells within the temporal retina. In mutants, the temporal retina is significantly smaller during mid-retinogenesis (52 hpf) and ultimately the inner portion of the INL of the post-mitotic retina (72 hpf) that contains the amacrine cells remains thin. In agreement, cell counts indicate fewer numbers of amacrine cells in the temporal retinas of mutants as compared to wildtype, while

108

Figure 3.21. Nrp2b and Plxna3 are likely receptors for Sema3fa signalling. Whole mount expression of sema3fa (A), and receptors nrp2b (B) and plxna3 (C) at 72 hpf, as viewed dorsally.

The ciliary marginal zone (yellow chevron) expresses sema3fa and both receptors. Additionally, both receptors are expressed throughout the RGC layer (white arrow). N: nasal; T: temporal. Scale bar: 50 µm.

109

the RGC and bipolar cell layer and nasal amacrine cell populations remain unaffected. Based on these data I propose a model whereby Sema3fa signalling acts early (24-36 hpf) on progenitor cells of the temporal retina to control the production of amacrine cells of the temporal retina, ultimately influencing their proliferation and/or differentiation (Figure 3.22A). In support for the timing of

Sema3fa action, while the amacrine cell phenotype observed in the sema3fa mutants at 72 hpf is restricted to the temporal retina, sema3fa is expressed by temporal retinal progenitor cells early, but subsequently shows no temporal restriction in its expression. Thus, we propose early spatially restricted Sema3fa signalling influences future amacrine cell genesis in a topographic manner.

Sema3fa signalling appears to be an extrinsic factor that regulates the transcriptome of progenitors. Retinal neurogenesis is a careful interplay between intrinsic factors, such as transcription factors (Cepko, 1999), and extrinsic factors, such as FGF, BMP, and Shh (Martinez-

Morales et al., 2005; Murali et al., 2005; Zhang and Yang, 2001). While extrinsic factors influence the expression of intrinsic molecules, intrinsic factors are also capable of modulating the response to extrinsic signalling pathways (Luo et al., 2012). Current literature supports the idea that complex gene regulatory networks exist between cells and their environment. This is true of mitosis and differentiation of progenitor cells as well. Classic models assume a developmental sequence consisting of neural progenitor growth, migration, cellular division and finally differentiation.

Single cell lineage tracing in vivo in the zebrafish retina, however, has revealed that progenitors can differentiate in the absence of cell division, and that progenitors already express differentiated transcriptomes prior to their final neurogenic division (Engerer et al., 2017). Thus, it is evident that neurogenesis cannot simply be described as a discrete sequence of events, as it overlooks the complexity at play within neural progenitor cells. My data supports this complexity as the influence

110

Figure 3.22. Summary of sema3fa roles in the neural retina. A) During early retinogenesis, sema3fa is expressed by the temporal neural retina (red) and regulates the spatial identity of these cells. By the end of retinogenesis, temporal RPCs that lose Sema3fa signalling early in development produce fewer amacrines cell (green) in the temporal retina. B) By late retinogenesis, sema3fa is expressed by progenitors of the ciliary marginal zone (cmz) and maintains the spatial organization of progenitors in the niche whereby atoh7 (differentiating progenitor) and ccnd1

(proliferating progenitor) populations remain discrete from one another. Loss of Sema3fa signalling disrupts the environmental niche and the populations of RPCs lose their spatially discrete expression profiles.

111

of Sema3fa in the temporal retina appears restricted to progenitors that produce amacrine cells, with other retinal cell types not impacted in numbers by the loss of Sema3fa. In the future, it will be interesting to determine whether a specific subtype of amacrine cell is preferentially affected.

There are at least 28 different subtypes of amacrine cells within the zebrafish retina which can be identified by their morphologies and molecular markers (Cederlund et al., 2011). Determining which subtype, if any, by an immunolabelling approach, will aid our understanding in the molecular mechanism behind Sema3fa signalling within the RPCs as the differentiation pathways between amacrine subtypes can be analyzed.

The expression of atoh7 and neurod4 early during retinogenesis support their location at the top of a hierarchy of a transcription factors involved in the genesis of retinal neurons by RPCs

(Inoue et al., 2002; Jin, 2017; Sinn et al., 2014). Loss of either factor would be anticipated to produce a gross loss to RGCs and amacrine cells, as these transcription factors are thought to be necessary to form precursors of both neural types (Jin, 2017). Yet, while there was a clear reduction in transcript of both genes in the sema3fa mutant retina at 36 hpf, gross changes in the genesis of either neural population was not observed with Sema3fa loss. A likely explanation is that the reduction in transcripts are transient. In support, atoh7 expression by 48 hpf was comparable in wildtype and sema3fa mutant embryos. Zebrafish are a regenerative model (Gemberling et al.,

2013), and so it is not surprising that compensation may occur to allow recovery from loss of a key factor (White et al., 2017). Of note, sema3fa is expressed by the RGCs at 48 hpf and bipolar cells at 72 hpf, and yet their numbers are not altered by loss of Sema3fa signalling. This is presumably because the ligand is playing other roles, possibly to do with synaptogenesis, process outgrowth or cell polarity, which are independent of the cellular genesis roles I observe with respects to temporal amacrine cells.

112

Compensatory mechanisms may also explain why despite increases at 52 hpf in the transcriptome of progenitors and bipolar cells, with vsx2 (Wong et al., 2015), and global amacrine/horizontal cell genesis with ptf1a (Jusuf and Harris, 2009) (Jin, 2017), an increase in undifferentiated progenitors within the central retina (ccnd1+), or increased numbers of bipolar cells (vsx2+) and amacrine cells (HuC+, Pax6+) were not observed. Instead, markers of all INL cell types are expressed normally, except for a decrease rather than an increase in the number of amacrine cells – only within the temporal retina.

The expression of sema3fa within the CMZ suggests a second, and likely ongoing

mechanism of Sema3fa action, whereby the local signalling environment generated by Sema3fa in

the CMZ is necessary for the establishment of normal transcriptional profiles within the progenitor

cells (Figure 3.22B). The CMZ is organized spatially, whereby quiescent stem cells reside in the

peripheral CMZ and the cells towards the central retina are proliferative and progressively more

restricted in their differentiation capacity (Cerveny et al., 2010). As the CMZ does not show axial

differences in sema3fa expression, it is likely that post 72 hpf any further changes in retinal cell

genesis in the absence of Sema3fa would not be spatially restricted. The decrease in amacrine cell

number within the temporal retina is likely independent of the CMZ, and future projects can

explore the role of Sema3fa signalling in the CMZ.

The literature proposes a model of the CMZ whereby cycling progenitors integrate intrinsic cues and extrinsic feedback cues from surrounding cells to transition from a proliferative to a differentiating state (Cerveny et al., 2010; Kubota et al., 2002). In support, in the zebrafish flotte lotte (flo) mutant, which carries a mutation in the elys (ahctf1) gene, the CMZ is able to proliferate but not enter a final neurogenic division due to loss of atoh7 from the CMZ (Cerveny et al., 2010).

When atoh7 mutant progenitors are transplanted into a wildtype CMZ, however, they can

113

differentiate. These data suggest that a loss of an extrinsic cue in the mutant is responsible for the phenotype. Interestingly, in the sema3fa mutants an expansion of atoh7 occurs within the CMZ in

92% of mutant embryos, such that a factor known to control neurogenic divisions (Sinn et al.,

2014) is now located within the all progenitors of the CMZ. Presumably, this expansion influences the ability of progenitors to differentiate into specific neural cell types. These data suggest that in mutants there is a loss of the spatial organization of the environmental niche of the CMZ.

Additionally, I find an expansion of ccnd1, vsx2 and PCNA within the CMZ, which suggests more

RPCs are proliferative in mutants as compared to wildtype. In support, vsx2 is suggested to potentially act upstream of ccnd1 to maintain progenitor proliferation (Wong et al., 2015). A feedback model in the CMZ proposes the tissue is self-limiting, such that when cells with an intrinsic capacity to proliferate reach the central margins of the CMZ, environmental cues from the cells of the post-mitotic neural retina instruct them to differentiate (Cerveny et al., 2010). All neurons post 48 hpf are CMZ derived, and so I would have liked to have determined the consequence of an increase in vsx2/ccnd1 and atoh7 to the growing retina post 72 hpf.

Unfortunately, for technical reasons I could not perform birth dating analyses or lineage tracing using CMZ transgenics which would have aided in this regard. Intriguingly, the CMZ in the

Sema3fa mutants likely expresses markers of proliferation (ccnd1/vsx2/PCNA) and of differentiation (atoh7) in the same RPCs instead of in spatially discreet bands where proliferative and undifferentiated cells are localized more to the peripheral CMZ and differentiating cells located closer to the neural retina. It is possible that Sema3fa signalling provides a spatial cue for a subset of CMZ progenitors, possibly the proliferative and undifferentiated cells, to remain in such a state, and when signalling is lost, the cells become more responsive to their environment and turn on differentiation genes prematurely (atoh7).

114

Finally, my expression data argue that Sema3fa has additional roles as an extrinsic regulator. Its expression in post-mitotic neurons that are differentiating, specifically RGCs at 52 hpf, and at 72 hpf in cells of the outer INL, likely bipolar cells, suggests functions of Sema3fa in controlling differentiation. In support, differentiated HuC RGCs are present in similar numbers in sema3fa mutants to those found in wildtype embryos, but the expression of differentiated markers for RGCs, specifically cxcr4b and irx1a, are reduced. Cxcr4b in zebrafish RGCs is required for axon guidance (Li et al., 2005), and Sema3f may control the polarity of Xenopus RGCs (Kita et al., 2013), so Sema3fa in the post-mitotic retina could function to regulate neuronal morphogenesis, neuronal polarity, dendrite formation and survival. The latter seems unlikely that there was no change in the number of apoptotic cells in sema3f-/- retina as compared to control at

72 hpf.

To describe transcriptome differences between wildtype and sema3fa mutant progenitors,

I performed an RNAseq analysis (described in Chapter 2) by using whole embryos at 36 hpf

(Figure 3.23). This embryonic stage was chosen because of the robustness of the reduction in atoh7, and because at this time point sema3fa is still expressed by temporal retinal progenitor cells

(Figure 3.1). A total of 15,455 expressed genes were identified, of which 1,149 were differentially expressed in mutant compared to wildtype embryos. Considering a global RNAseq approach was used, I decided fold changes of >1.2 (with a p-value <0.05) represent genes of interest, as any retinal changes may be “washed out” in the global RNAseq with more stringent cut-offs. Of these differentially expressed genes, 508 were upregulated and 641 were downregulated (Figure 3.23A-

115

Figure 3.23. Summary of RNAseq differential gene expression analysis on sema3fa mutants.

A) Number of genes differentially expressed between genotypes. Upregulated genes (green) and downregulated genes (red) are presented with total number of genes in each category within the bar. B) Representative volcano plot of differentially expressed genes between mutant and wildtype embryos. C) Principle component analysis (PCA) of the nine samples (3 wildtype, 3 heterozygous,

3 mutant). Scatter plots between two wildtype samples (D) and two mutant samples (E) shows little variation in gene expression between sample sets. F) Jensen-Shannon Divergence (jsd) values calculated as similarity of the probability distributions of the expression levels (FPKM) for 15,455 total transcripts. G) Functional pathway analysis for downregulated genes in mutant embryos

116

compared to wildtype. Number of genes involved in each category expressed on the right. H) Table highlighting fold change and significance (q-value) of notable retinogenesis related genes.

117

B) in mutants compared to wildtype. As expected, a 1.7 and 3 fold reduction in sema3fa transcript in the heterozygous and homozygous embryos, respectively, was observed (Table 6.1). To further ensure data sets generated were of high quality, I used principle component analysis (PCA) to determine similarities and differences between data sets to identify outliers. As evident in Figure

3.23C, HET1 was an outlier and was eliminated from further analysis. The PCA clearly showed

WT and MT samples clustering independently of one another suggesting they define two independent sets or pools of data. Variances between samples were plotted (Figure 3.23D-E) between data sets. These plots support that samples within each genotype were highly correlated with one another, and argue that the data sets generated of differentially expressed genes between the genotypes are of high quality and reproducibility. To further ensure the data sets between groups were similar, a Jensen-Shannon Divergence plot was constructed that showed similarities within genotypes, but dissimilarities existing between genotypes (Figure 3.23F).

Considering the large number of genes changed between mutant and wildtype embryos, I annotated the gene lists into 15 subgroups by using what is known about gene function and available expression data from the ZFIN database. Subgroups ranged from genes with unknown functions, metabolism and ECM formation, to cell cycle, cell survival, retina specific and CNS specific genes. Using these subdivisions, 9% (46 genes) of upregulated genes and 25% (160 genes) of downregulated genes were specific to the retina. Additionally, genes directly associated with cell cycle and survival made up 4% (20 genes) of the total upregulated genes and 5% (32 genes) of the downregulated genes. Additionally, I graphed Gene Ontology (GO) analysis bioinformatic data, which revealed biological processes that were different between genotypes. The pathways with the biggest changes are Muscular System Development and Function, Hematological System

Development and Function, DNA Replication Recombination and Repair, Carbohydrate

118

Metabolism, Cellular Growth, Proliferation and Pluripotency, and Cell Death and Survival. For the purpose of my work, only those related to the project are graphed (Figure 3.23G). Many of the

GO terms relate to developmental processes such as proliferation and differentiation, which further supports my functional data that suggest a role for Sema3fa signalling in developmental processes outside of axon guidance. I have only just started to explore the database of genes collected by

RNAseq in order to fully understand the phenotype present within the mutant.

Interestingly, the first defect in retinal cell genesis presents as retinal neurons are being specified (36 hpf), and not early when progenitors are being specified and patterned (13 ss), despite expression of sema3fa in the temporal eye vesicle at these earlier time points. Potentially, the orthologue, sema3fb, which is expressed in the same domain during early retinogenesis, compensates for this loss at these early time points. Functional upregulation of compensatory molecules, usually orthologues, is observed in many zebrafish mutants (Anderson et al., 2017; El-

Brolosy et al., 2018; Rossi et al., 2015). A phenotype manifests in the sema3fa mutants at a time point when sema3fb mRNA is no longer present within the neural retina. Of note, I was fortunate in my analysis using zebrafish that with the exception of the early eye, the domains of expression between sema3fa and sema3fb are nearly entirely exclusive to one another, thereby avoiding the issue of redundancy. Certainly, at a later stage (36 hpf) Sema3fb does not appear to compensate for Sema3fa loss in that the RNAseq analysis did not identify any change in expression of sema3fb transcript across genotypes.

Visual function is not impaired in the mutants with the two assays performed. However, I cannot rule out the possibility that neural circuit function is disrupted within the retina of the mutants. The startle response measures a behavioural response to a visual stimulus (Scott et al.,

2016). Differences in such a robust response are likely only observable with a major loss of retinal

119

cells, for instance photoreceptors which convert the visual stimulus into neural signals. As the main phenotype was a reduction of amacrine cells in the temporal retina, the unchanged responsiveness between genotypes is not surprising. The more sensitive approach was the use of c-fos expression as a surrogate measure of neural connectivity, though this assay still did not identify any striking differences between the genotypes following light exposure. An analysis of the cell types that turn on c-fos expression could be carried out, as it is possible that the reduction of amacrine cells in the

INL of the temporal retina sufficiently disrupts the pattern of cell activation of INL interneurons and their RGC targets induced by a visual stimulus.

My data support a spatially restricted role of Sema3fa on progenitor cells of the temporal retina, and of progenitors within the CMZ. Spatial restriction of Sema3fa signalling in the CMZ is likely mediated by cell autonomous expression of both nrp2b and plxna3, which are the only nrp and plxna I could find mRNA for within the CMZ. I did not identify a receptor expressed by the temporal RPCs, however, presumably because nrp and plxn receptors are expressed generally at low levels. Of note, it is unlikely that all Sema3fa signalling within the retina is mediated through the same receptors. Sema3f is a secreted ligand, and between 48 hpf and 72 hpf mRNA is expressed first by cells of the RGC and then cells of the outer INL. Considering this expression profile and the secreted nature of Sema3fa, it is likely that multiple cells are exposed to an environment rich in Sema3fa at some point during embryogenesis. To this point, we have previously published that

Sema3f signalling from the INL biases apical dendrite formation, via Nrp1 and Plxna1, of RGCs in Xenopus (Kita et al., 2013). Further, it is possible that the defects observed in retinal neurogenesis arise because of a role of Sema3fa outside of the eye, given the expression of sema3fa in the brain (Callander et al., 2007), and a suggested role of SEMA3F in the neuroendocrine system in mouse (Sahay et al., 2003). The fact that the reduction in atoh7 in the mutant embryos relative

120

to wildtype eyes was observed when eye tissue was explanted and grown in culture argues against this possibility. Instead, the data argue that Sema3fa provided by eye progenitors drives the retinal cell type specification program, or more appropriately the cell type specification program of amacrine cells.

3.4 Significance

In this chapter, I provide the first evidence that Sema3fa is a novel extrinsic regulator of

retinogenesis. Semas are known in their roles for guidance, however, non-guidance functions, such

as in the control of proliferation and cellular polarity, are now being explored (Alto and Terman,

2017; Arbeille et al., 2015; Berndt and Halloran, 2006; Jongbloets and Pasterkamp, 2014). Here I

add to the growing body of literature on overlooked roles of Semas outside of their guidance

capabilities. Interestingly, my data suggests progenitor cells have an additional level of polarity

outside of what has been previously described with respects to retinotopic mapping of RGCs in

the tectum (Stuermer, 1988). Instead, it appears that temporal retinal progenitor cells must “know”

they reside within the temporal retina, mediated by Sema3fa spatial signalling, in order to produce

neurons of the inner nuclear layer within the same temporal retina. This raises many interesting

biological questions with respects to progenitor competence and dynamics in their spatial

organization.

121

Chapter Four: Sema3fa signalling in ocular vascularization

4.1 Introduction

The retina is a highly metabolic structure of the central nervous system, with the highest

ATP demand coming from photoreceptors and RPE (Country, 2017). It is necessary to match vascularization to metabolic demand – a process which is only just being appreciated. The embryonic retina becomes perfused by two vessel beds – the hyaloid and choroidal vasculatures

(Campochiaro, 2015). In mouse and humans, the optic artery enters the eye at the site of the optic nerve head and forms branching arterioles along the inner limiting membrane of the retina that penetrate and vascularize the inner 2/3 of the retina. This intraretinal plexus network establishes the hyaloid, or retinal, vasculature. On the back surface of the eye, a dense plexus of fenestrated choroidal vessels supplies nutrient/waste and oxygen transfer, via diffusion, from the systemic vasculature across the retinal pigment epithelium (RPE) and into the outer 1/3 of the retina where photoreceptors reside. Avascularity of this outer retina is necessary to avoid light absorption from hemoglobin containing blood cells traveling through capillaries (Campochiaro, 2015; Saint-

Geniez and D’Amore, 2004), which results in visual function impairment. The zebrafish retina is nourished by choroidal vasculature and hyaloid vasculature, but unlike described in mouse and primates (Fruttiger, 2007; Provis, 2001), no intraretinal plexus formations are fashioned (Alvarez et al., 2007). Additionally, it remains unclear how the choroid plexus of zebrafish forms, and whether the embryonic retina surface vasculature is the foundational plexus from which it is derived (Kaufman et al., 2015). Nonetheless, the choroid plexus as an anatomical structure in the zebrafish model has been identified, and its growth in response to genetic manipulations analyzed

(Hashiura et al., 2017; Isogai et al., 2001; Rezzola et al., 2016; Rooijen et al., 2010). Overall, it is

122

the absolute physiologic avascularity of the zebrafish neural retina which makes it an excellent model to study vascular retinopathies.

Due to the high energy demand of the retina, vulnerability to disease and dysfunction is large. In homeostasis, the RPE generates a unique environment surrounding the retina, as its apical surface (towards the photoreceptors) is avascular, and basal surface sits on a extracellular matrix rich layer called Bruch’s membrane, which provides physical separation from the highly fenestrated choroid vasculature (Campochiaro, 2015). Diffusion of nutrients, waste and oxygen occur across this blood-RPE barrier. If compromised, blood vessels penetrate through the physical barrier of the epithelium. Pathologies which manifest within the choroidal vessel bed are encompassed by the broad term choroidal retinopathy, and it is neovascularization or pathological angiogenesis of the outer retina that drives visual impairment progression.

Age-related macular degeneration (AMD) is one such choroidal retinopathy and is the primary cause of blindness in individuals over the age of 65 in industrialized nations (Saint-Geniez and D’Amore, 2004). AMD exists in two forms, a non-exudative (dry) type, and exudative (wet) type. Wet AMD is characterized by pathologic outgrowth of vessels from the choroid, termed choroidal neovascularization (CNV), which breach Bruch’s membrane and enter the subretinal space. Ultimately, these vessels leak, which results in macular edemas and detachment of either the RPE or retina, and functional vision loss. Development of CNV is thought to require both increased expressions of VEGF and tears to Bruch’s membrane (Baffi et al., 2000; Spilsbury et al.,

2000). Currently, anti-VEGF therapy is the only treatment available for wet AMD, but is limited in its effectiveness (Cui and Lu, 2017). Understanding the etiology of the disease will aid in exploring new therapeutic avenues.

123

Normally, vessels of the orbit are quiescent in growth once established. However, some retinal pathologies exist in which vessel proliferation is seen. These diseases are broadly termed proliferative retinopathies, and include proliferative diabetic retinopathy, venous occlusion, and retinopathy of prematurity (Saint-Geniez and D’Amore, 2004). Interestingly, these diseases are characterized by regions of ischemic retina which triggers neovascularization to allow for reperfusion, a process that is similar to that of choroidal retinopathies in being VEGF or hypoxia driven.

The chemorepellent SEMA3F has garnered translational interest with the finding of reduced levels of transcript in the RPE of patients with wet AMD (Buehler et al., 2013). The anti- angiogenic potential of SEMA3F has been studied both in vitro and in vivo using gain-of-function approaches in mouse models of CNV (Buehler et al., 2013; Sun et al., 2017a). These seminal papers are the first evidence that SEMA3F behaves as an retinal anti-angiogenic, with SEMA3F expressed in the outer retina preventing the erroneous invasion of choroidal vessels. What remains to be explored is whether endogenous SEMA3F, expressed by the murine and human outer nuclear layer and RPE (Buehler et al., 2013), is necessary to maintain the physiologic avascularity of the outer retina.

In this chapter, I take advantage of the sema3fa mutant zebrafish line I generated. I show the expression of sema3fa in the RPE of the zebrafish retina, in the vicinity of the overlying choroid plexus. With the loss of Sema3fa, I find that the intraocular hyaloid vasculature is overgrown in the embryo, a feature that does not resolve into adulthood. Additionally, I demonstrate that

Sema3fa is necessary in the retinal environment to curb blood vessel infiltration from the choroid.

I find that without any additional manipulations to the embryo, CNV occurs and blood vessels are leaky. Further, I show this infiltration is not mediated by an ectopic increase in vegfa, suggesting

124

that it is not only growth inducing molecules that are necessary for pathological infiltration, but also the loss of anti-angiogenic ones. I establish the sema3fa mutant zebrafish as the first model of vascular retinopathy not due to genetic manipulations of the blood vessels or artificially-induced vascularization, but instead due to localized changes in the cellular environment (RPE) of the blood vessels.

4.2 Results

4.2.1 semaphorin3fa is expressed by the retinal pigment epithelium in development and adulthood

I characterized the neural retina expression of sema3fa in Chapter 3 during embryogenesis.

In addition to the neural retina expression (Figure 4.1A), I found sema3fa was expressed in non- neural tissue as well, including the RPE and the surrounding periocular mesenchyme (Figure

4.1B). Expression was maintained into adulthood throughout the RPE, the three nuclear layers of the retina and the ciliary marginal zone (Figure 4.1C). Previous reports in mouse have shown expression of Sema3f in RPE and the outer nuclear layer (Buehler et al., 2013; Sun et al., 2017a).

The similar expression patterns of Sema3f in the mouse and zebrafish RPE support a conserved role.

4.2.2 sema3fa mutants present with increased intraocular vascularization

4.2.2.1 Establishing a sema3fa mutant transgenic reporter line

Previous reports suggest that SEMA3F is a novel repellent of CNV (Sun et al., 2017a), in that overexpression of SEMA3F is protective against the pathologic vascularization observed in the Vldlr-/- embryo mouse model, and that induced by laser damage to the RPE and Bruch’s membrane. While these data provide evidence of the anti-angiogenic potential of Sema3f on pathologic vessels, the models do not appropriately recapitulate human wet AMD as mutant Vldlr retinas have biochemical alterations ranging from graded inflammation, ectopic expression of 125

Figure 4.1. sema3fa is expressed by the retinal pigment epithelium. RNA in situ hybridization of whole-mount (A) and retinal sections (B-C). sema3fa is detected throughout the retinal pigment epithelium (RPE) during development and into adulthood (A-C). Additional expression is detected within the retinal ganglion cell layer (RGC), inner nuclear layer (INL), outer nuclear layer (ONL) and progenitors of the ciliary marginal zone (CMZ) (C). dpf: days post-fertilization; m: months.

Scale bar: 100 µm in A, 50 µm B-C

126

CNV does not reflect the same angiogenic pathways (Shah et al., 2015). It is important to note that clinical analysis of post-mortem RPE lysates found 9/15 patients with wet AMD have reduced levels of SEMA3F transcript (Buehler et al., 2013), suggesting an environmental loss of SEMA3F is associated with the disease. As such, I used the previously described sema3fa mutant (Chapter

3) as an appropriate model to investigate orbital vasculature behaviour in an environment lacking

Sema3fa. To do this, I outcrossed a female sema3fa homozygous mutant to a male

Tg(kdrl:mCherry) in which all endothelial cells are labelled by mCherry (Proulx et al., 2010).

Resulting heterozygotes were raised to adulthood, and then incrossed to generate wildtype and homozygous mutant lines. All subsequent analyses in this Chapter are a mix of embryos generated from either heterozygous incross matings, or from the established monogenic transgenic lines.

4.2.2.2 Early intraocular vascularization of the retina is normal

The hyaloid vessel bed forms a network of vessels which covers the posterior aspect of the lens - forming the intraocular vascular system (Alvarez et al., 2007; Hartsock et al., 2014; Isogai et al., 2001; Saint-Geniez and D’Amore, 2004). Beginning at 15 dpf, the hyaloid detaches from the lens and adheres to the inner limiting membrane, forming the adult retinal vasculature (Alvarez et al., 2007). This process contrasts to that in mouse, where the hyaloid vessel regresses completely and regrows to form the retinal vasculature through an angiogenic program (Saint-Geniez and

D’Amore, 2004). Given that sema3fa is expressed in the 18-24 hpf optic cup (Chapter 3), I first asked whether the normal hyaloid entry of the hyaloid vessel into the retina (Hartsock et al., 2014)

(Figure 4.2A) is affected in the sema3fa (Chapter 3). By confocal live-imaging at 24 hpf, I found no difference in the placement of the hyaloid vessel or its contact with the lens in wildtype (N=1, n=3) and sema3fa mutant (N=1, n=4/4) Tg(kdrl:mcherry) embryos (Figure 4.2B-C). Additionally, at 72 hpf the branching structure of the hyaloid around the entirety of the lens

127

Figure 4.2. Initial entry of the hyaloid vessel into the eye in normal in sema3fa mutants. The hyaloid artery (labelled by mCherry) of Tg(kdrl:mcherry) embryos normally makes contact with the lens (asterisk) of the retina (labelled by GFP) by 24 hpf. This is true in both wildtype and mutant embryos: Live imaging of 24 hpf sema3fa wildtype (N=1, n=3/3) (B) and mutant (N=1,

4/4 (C) embryos reveal no differences in the location or timing of hyaloid vessel entry.

128

(Hartsock et al., 2014) formed normally in mutant (N=1, n=4/4) and wildtype (N=1, n=4/4) embryos (Figure 4.3). Thus, I conclude that Sema3fa is not necessary for early retinal vascularization.

4.2.2.3 Hyaloid vessel is compact, and adult retinal vasculature more dense

Hyaloid vessel development is largely non-stereotypic (Kaufman et al., 2015), however, refinement to an organized stereotypic pattern of vessels on the lens occurs between 4-5 dpf, whereby vessel complexity is markedly reduced by retraction of extraneous vascular connections

(Hartsock et al., 2014). Considering the expression of sema3fa in the CMZ and nuclear layers of the 72 hpf retina (Figure 3.1E,E’), close to the hyaloid vessel bed, I next asked whether refinement of the hyaloid network was affected by loss of Sema3fa. From confocal projections of live embryos at 4 dpf, I observed that while 91% of wildtype (N=4, n=10/11) embryos had refined their networks to the simplified stereotypic pattern, all the heterozygous (N=1, n=2/2) and homozygous (N=4, n=12/12) sema3fa-/- larvae retained complex hyaloid vessel networks with interconnections throughout the plexus (Figure 4.4). Overall, these data support a role for Sema3fa in embryonic hyaloid vasculature development.

To determine if refinement of the hyaloid vessels eventually occurred in the absence of

Sema3fa, I looked at the hyaloid vasculature in the retinas of wildtype and mutant adults. Of note, the hyaloid network detaches from the lens and adheres to the inner limiting membrane of the retina beginning at 15 dpf until 60 dpf (Alvarez et al., 2007), and so the state of complexity of the network would be reflected on the inner limiting membrane. I performed retinal flat mounts of transgenic sema3fa+/+ and sema3fa-/- 5-month-old adult Tg(kdrl:mcherry) fish (Figure 4.5A,B), and measured 3 key parameters of vessel growth; vessel length and width, and spacing

129

Figure 4.3. Initial intraocular vascular development is normal during embryonic development. The hyaloid artery (arrowhead) forms a vascular plexus network around the lens by

72 hpf. Maximal projections of stacks of ventral views of 72 hpf sema3fa wildtype (N=1, n=4/4)

(A) and mutant (N=1, n=4/4) (B) Tg(kdrl:mCherry) eyes acquired by confocal live imaging reveal no differences in the hyaloid vessel morphology. Scale bar: 100 µm.

130

Figure 4.4. sema3fa deficient embryos present with abnormal intraocular blood vessels. Live

Tg(kdrl:mCherry) embryos of different sema3fa genotypes imaged from a ventral view on a confocal microscope and z-stacks maximally projected. The intraocular hyaloid vessel (yellow chevron) undergoes remodelling between 4-5 dpf to simplify the network (Hartsock et al., 2014).

The density of vessel crossovers is increased in heterozygous (B, N=1, n=2) and homozygous mutant (C, N=4, n=12) embryos compared to wildtype siblings (A, N=4, n=11). Respective insets of hyaloid vasculature (A’-C’). dpf: days post fertilization. Scale bar: 100 µm and 50 µm for insets.

131

Figure 4.5. Increased vascularization of the retina is observed into adulthood. Retinal flatmounts of Tg(kdrl:mCherry) sema3fa wildtype siblings (A) and mutant (B) 5 month old fish.

Primary vessel length (optic nerve head (onh) to first branching event) is significantly shorter in mutants (C, N=2, n=6, **p=0.0016) as compared to wildtype siblings (N=2, n=6). These width of these same vessels is significantly reduced in mutants (D, N=2, n=6, **p=0.0078) as compared to wildtype (N=1, n=6). Capillaries furthest from the onh in the peripheral retina (E,F) and the intercapillary distances (G). Mutant capillaries are less spaced apart (G, N=1, n=4, ****p<0.0001) as compared to wildtype (N=1, n=4). Additionally, a greater number of tip cell interconnections between capillaries are present in mutant retina (yellow chevron). Error bars represent standard error of the mean (SEM). Statistics represent the non-parametric Mann-Whitney U test. Scale bar:

200 µm (A) and 50 µm (E).

132

(Bozic et al., 2018). Primary vessel length was measured from the optic nerve head to the first branching event, and was shorter by 25% (N=2, n=6, p=0.0016) in mutants as compared to wildtypes (N=2, n=6) (Figure 4.5C). Additionally, the width of these primary vessels was significantly thinner by 20% (N=2, n=6, p=0.0078) in mutants as compared to wildtype vessels

(N=1, n=6) (Figure 4.5D). Finally, the spacing of capillaries in the retinal periphery, where arterial capillaries anastomose with circumferential vein capillaries (Cao et al., 2008) (Figure 4.5E,F), was considerably shorter in the mutant vessel bed (48.0± 1.6 µm; N=1, n=4, p<0.0001) as compared to that observed in wildtype retinas (92.7 ± 2.3 µm; N=1, n=4) (Figure 4.5G). Additionally, I found a qualitative increase in the number of intercapillary connections in the mutants as compared to the wildtypes (Figure 4.5E,F). These data further support a role for Sema3fa in the network formation of the hyaloid and retinal vasculature throughout life.

4.2.3 sema3fa mutants have pathologic, leaky vessels that enter the eye

4.2.3.1 Choroid plexus undergoes neovascularization in the sema3fa mutant retina

The choroid plexus surrounds the back of the retina, providing nourishment from outside of the RPE (Alvarez et al., 2007; Isogai et al., 2001; Saint-Geniez and D’Amore, 2004). In zebrafish, plexus formation is first observed at the back of the eye by 5.75 dpf (Rooijen et al.,

2010). At this time point, and through to the adult, sema3fa is expressed in the RPE underlying the choroid plexus (Figure 4.1B,C). Thus, I asked if a loss of sema3fa would impact the adjacent choroidal vessel network. Using live imaging of wildtype and mutant Tg(kdrl:mCherry) embryos,

I found that while wildtype embryos showed little or no choroid plexus at 6 dpf (N=1, n=0/3), sema3fa-/- embryos exhibited precocious growth of the choroid plexus (N=1, n=3/3) (Figure 4.6).

This time point, however, was highly variable for capturing plexus formation faithfully, and

133

Figure 4.6. sema3fa deficient embryos present with choroid plexus overgrowth. Live imaging of 6 dpf Tg(kdrl:mCherry) wildtype (A, N=1, n=3) and mutant (B, N=1, n=3) embryos on a confocal microscope with maximal projection z-stacks shown. At this stage the extraocular choroid plexus (yellow arrowhead) begins to branch over the back of the retina in mutant, but not wildtype eyes. Scale bar: 100 µm.

134

instead I found that 7 dpf was the most reliable stage to use, as wildtypes all have an obviously visible choroid plexus. At 7 dpf, the choroid plexus was just forming in 88% of wildtype embryos

(N=6, n=15/17, Figure 4.7A), but was largely overgrown, or advanced in 69% of heterozygous

(N=4, n=9/13, Figure 4.7B) and 71% of homozygous (N=6, n=12/17, Figure 4.7C) sema3fa-/- embryos. These data suggest that precocious choroid plexus growth occurs when Sema3fa signalling is removed from the environment.

4.2.3.2 sema3fa -/- orbital vasculature is leaky

An overgrowth of the choroid plexus suggests neovascularization occurs over top of the

RPE. To determine if the physiologic avascularity of the retina was compromised as a secondary consequence, blood vessels were examined in histological retinal sections. While the

Tg(kdrl:mCherry) is a reporter line for endothelial cells that express Vegfr2 (kdrl) (Proulx et al.,

2010), there is a possibility that not all arterial and venous vessels are labelled. Thus, to label the entire vasculature I performed fluorescent angiography by using 2,000,000 MW rhodamine- dextran, a molecule normally too large to pass through any vessel wall. I injected the hearts of 7 and 9 dpf non transgenic wildtype and mutant embryos with the 2,000,000 MW rhodamine- dextran, fixed the embryos 24 hours post injection, and sectioned retinas for analysis using the confocal (Figure 4.8A,D,G,J). Vessels were detected within the outer nuclear layer in all of the 8 dpf (N=2, n=4/4) and 10 dpf (N=3, n=10/10) (Figure 4.8F,L) embryos, but were absent from all of the wildtype embryos (N=2, n=4/4 for 8 dpf; N=2, n=10/10 for 10 dpf) (Figure 4.8C,I). In addition to the ectopic vessels, dextran accumulated throughout the nuclear layers of the retina in all of the 8 dpf ((N=2, n=4/4) and 80% of 10 dpf (N=3, n=8/10) (Figure 4.8E,K) embryos, but was largely absent from the nuclear layers of wildtypes (N=2, n=4/4 for 8 dpf; N=3, n=9/10 10 dpf)

135

Figure 4.7. sema3fa deficient embryos present with abnormal extraocular blood vessels. Live

Tg(kdrl:mCherry) embryos of different sema3fa genotypes imaged from a ventral view on a confocal microscope and z-stacks maximally projected. By 7 dpf, the extraocular choroid plexus

(yellow arrowhead), begins to branch over the back of the retina. Heterozygous (E, N=4, n=13) and homozygous mutant (F, N=6, n=17) reveal a dramatic increase in the choroid vasculature compared to wildtype siblings (D, N=6, n=17). dpf: days post fertilization. Scale bar: 100µm.

136

Figure 4.8. Sema3fa deficiency permits entry of leaky vessels into the neural retina. Embryos injected at 7 dpf (A-F) or 9 dpf (G-L) with 2,000,000MW rhodamine dextran and fixed 24 hours later. Thick (40µm) eye cryosections counterstained with nuclear label (Hoechst) reveal infiltrating blood vessels (yellow arrowhead) in homozygous embryos (N=2,n=4/4) at 8 dpf (F) and 10 dpf

(N=3, n=10/10) (L) as compared to wildtype (C,I, N=2, n=4 and N=3, n=0/10). Leakage of dye is also present in the neural retina at 10 dpf in homozygous embryos (N=3, n=8/10, K), as compared to wildtypes, which show almost no leakage (N=3, n=9/10, H). Dashed line represents boundary between the retinal pigment epithelium and neural retina. Scale bar: 100 µm and of inset 20 µm. 137

(Figure 4.8B,H). Taken together, these data indicate that in the absence of Sema3fa blood vessels invade the normally avascular outer retina and leak into the neural retina proper.

To address whether the entry of leaky vessels into the eye was solely a larval phenotype or a progressive phenotype, I analyzed the retinas of wildtype and mutant adult (5 months)

Tg(kdrl:mCherry) fish (Figure 4.9). In retinal sections, mCherry positive cells were present in 75%

(N=1, n=3/4, Figure 4.9F,G) of the mutants and none of the wildtype siblings (N=1, n=4, Figure

4.9B,C). To determine whether the vessels remained leaky in the adult, I injected the fluorescent angiography dye into aged (12 month) adult non-transgenic wildtype and mutant fish. As was the case for the larval preparations, dextran was present within the nuclear layers of the retina (Figure

4.9H) in 60% of homozygous mutants (N=1, n=3/5) and none of the wildtype siblings (N=1, n=3,

Figure 4.9D), indicating that in the mutants the ectopic vessels remain leaky.

4.2.4 sema3fa mutants have apparently normal retinal pigment epitheliums

4.2.4.1 Vessel infiltration in sema3fa mutants is not due to an increase in Vegfa

Increased expression of VEGF is thought to drive CNV development in AMD (Baffi et al.,

2000; Spilsbury et al., 2000), as anti-VEGF therapy limits disease progression (Cui and Lu, 2017).

In zebrafish, Vegfa is a potent endothelial cell mitogen necessary for vascular development

(Coultas et al., 2005), promotes hypoxia driven angiogenesis (Thomas, 1996), and drives retinal and choroidal vascularization (Cheung et al., 2014). As such I asked whether the sema3fa mutant

CNV phenotype is due to an upregulation of vegfa expression in the outer retina. Zebrafish have a duplicated vegfa, with vegfaa and vegfab isoforms (Bahary et al., 2007), and so I analyzed sections of retinas from 10 dpf embryos processed by whole mount in situ hybridization for both orthologs.

Neither vegfaa or vegfab mRNAs were evident in the retinas of wildtype (N=2, n=10/10, both

138

Figure 4.9. Vessel infiltration persists in the adult sema3fa-/- zebrafish. Cryosections made through the whole retina of Tg(kdrl:mCherry) sema3fa wildtype (A-C) and mutant (E-G) 5 month old siblings. Nuclei are labelled by Hoechst (grey). In 75% of mutants, a blood vessel in the area of the outer segments (os) of the photoreceptors was detected (N=1, n=3/4), while no blood vessels were detected within the neural retina in the wildtype fish (N=1, 4/4). One year old adult sema3fa+/+ (D) and sema3fa-/- (H) fish were injected with 2,000,000MW rhodamine dextran

(purple) and fixed 4 hours later. Thick (40µm) eye cryosections counterstained with nuclear label

(Hoechst) reveals leakage of dye (purple, yellow chevron) is also present in the neural retina of sema3fa-/- fish (N=1, n=3/5, H), as compared to wildtype siblings that showed no leakage (N=1, n=3, D). inl: inner nuclear layer; onl: outer nuclear layer; RGC: retinal ganglion cell layer; RPE/C: retinal pigment epithelium/choroid. Scale bar: 50 µm. Scale bar of inset: 10 µm.

139

assays) or mutant (N=2, n=10/10, both assays) embryos (Figure 4.10). These data suggest vessel infiltration into the neural retina is not due to an increase in vegfa expression.

4.2.4.2 Retinal pigment epithelium is unperturbed in sema3fa mutants

The extent of the relationship between the RPE and overlaying choroidal plexus is not entirely appreciated (Campochiaro, 2015; Nickla and Wallman, 2010; Saint-Geniez and D’Amore,

2004). Tears in the RPE are generally associated with the presence of subretinal fluid build up and pathological vessel presence (Ersoz et al., 2017). Whether an RPE tear permits vessel entry, or tearing is the result of vessel entry, is debated. Nonetheless, disruptions in the continuity of the

RPE would provide a conduit for blood vessel sprouts from the choroid plexus to enter the neural retina. To analyze the intactness of RPE I first looked at histological sections stained by hematoxylin and eosin at 72 hpf (Figure 4.11A,B) and 5 dpf (Figure 4.11C,D). The RPE appeared continuous, with no gaps, at the two time points in both wildtypes and mutants (N=3, n=9 at 72 hpf for both genotypes, N=2, n=5 at 5 dpf for both genotypes). Further, neither genotype exhibited significant death of retinal cells, as assessed by TUNEL. There was no significant difference in the number of apoptotic nuclei in eyes of wildtype (N=2, n=10), heterozygous (N=2, n=10, p>0.99) and homozygous (N=2, n=10, p>0.99) at 7 dpf (Figure 4.11E). Together, these data suggest that at a gross level the physical RPE barrier is not disturbed in the mutant background, and cannot explain the ectopic presence of vessels in the outer retina.

To assess whether RPE differentiation was impacted by the loss of Sema3fa, I assessed the expression of two maturation markers of the RPE: retinal pigment epithelium-specific protein 65a

(rpe65a), which encodes an isomerase necessary to convert all-trans-retinyl ester to 11-cis-retinol for visual function (Jin et al., 2005), and immunolabel for Zpr2, which specifically labels mature

RPE (Zou et al., 2008). By whole mount in situ hybridization, I found no difference in the extent

140

Figure 4.10. vegfa is not induced in the retina of sema3fa mutants. Sections through the retina of larvae processed for whole mount in situ hybridization for vegfaa (A-B) and vegfab (C-D) present with an apparent absence of vegfa in both the wildtype (N=2, n=10 A, C) and mutant (N=2, n=10 B, D) retinas at 10 dpf. INL: inner nuclear layer; ONL: outer nuclear layer; RGC: retinal ganglion cell layer. Scale bar: 20 µm.

141

Figure 4.11. The RPE appears physically intact in the sema3fa mutant. Hematoxylin and eosin stained sections of 72 hpf (A-B) and 5 dpf (C-D) embryos. RPE (black) was indistinguishable between wildtype (N=3, n=9 for 72 hpf and N=2, n=5 for 5 dpf) and mutant (N=3, n=9 for 72 hpf and N=2, n=5 for 5 dpf) embryos. (E) Quantitation of the number of apoptotic nuclei in wholemount eyes. No significant differences are observed in TUNEL+ cells in the eyes of heterozygous (N=2, n=10, p>0.99) and homozygotes (N=2, n=10, p>0.99) as compared to wildtype controls (N=2, n=10) at 7 dpf. Scale bar: 100 µm

142

and intensity of the rpe65a label between wildtype (N=1, n=7) and mutant (N=1, n=7) embryos at

4 dpf (Figure 4.12A,B). Similarly, Zpr2 immunolabel at 10 dpf was indistinguishable between wildtype (N=1, n=5) and mutant (N=1, n=5) embryos (Figure 4.12C,D). Ultimately these results suggest that the ectopic retinal entry of blood vessels with the loss of Sema3fa likely arises from the loss of an endogenous anti-angiogenic, rather than an underlying RPE dysfunction.

4.2.5 Sema3fa signalling on endothelial cells is likely mediated by Nrp1b

Sema3s signal canonically through Nrp and Plxn receptors (Fujisawa, 2004). I find that sema3fa is expressed by the RPE and underlying neural retina (Figure 4.1), and loss of this known secreted chemorepellent results in the overgrowth of both intraocular and extraocular vessel beds

(Figure 4.4 & 4.7). As such, it seems likely that Sema3fa acts directly via Nrp/Plxn receptors expressed by vessel endothelial cells. A screen of identified Nrp and Plxn receptors by wholemount in situ hybridization identified nrp1b as a likely candidate to mediate the anti-angiogenic effects of Sema3fa, as mRNA was expressed within the cranial vasculature (Figure 4.13). Specifically, at

18 hpf, nrp1b was expressed by the hyaloid vasculature (Figure 4.13A) and at 28 hpf, the embryonic surface vasculature of the retina of which the choroid plexus is derived (Figure 4.13B).

Taken together, the ligand (sema3fa) expression in the retina and RPE, and receptor (nrp1b) expression in the vasculature, supports a model whereby Sema3fa acts in a paracrine manner to impact vessel growth.

4.3 Specific Discussion

My data is the first report to demonstrate the necessity of Sema3fa in maintaining the physiologic avascularity of the retina and network patterning of the retinal vasculature.

Specifically, I find I find that both the retinal and choroid vessel beds of the eye are disrupted with the loss of Sema3fa. Interestingly, the impact of Sema3fa loss is different in the two beds. I

143

Figure 4.12. The RPE of wildtype and sema3fa mutant eyes express RPE specific genes.

Whole mount in situ hybridization for the visual cycle gene rpe65a viewed dorsally (A,B), and immunolabeling with the RPE specific antibody zpr2 in retinal sections (C,D). The extent and density of rpe65a label was unchanged between wildtype (N=1, n=7, A) and mutant (N=1, n=7,

B) embryos at 4 dpf. Additionally, zpr2 immunostaining was detected in the apical (a) and basal

(b) aspects of the RPE in both wildtype (C, N=1, n=5) and mutant (D, N=1, n=5) embryos at 10 dpf. Scale bar: 100 µm

144

Figure 4.13. Sema3fa likely signals through Nrp1b. Whole mount in situ hybridization for nrp1b mRNA. A) Lateral image of an 18 hpf embryo shows expression of mRNA for the nrp1b receptor in the hyaloid (yellow arrowhead) as it enters the invaginating eye (black dashed line). B) At 28 hpf, in a dorsal view nrp1b remains expressed in the hyaloid (yellow arrowhead) and likely by the extraocular vasculature (black arrow). A: anterior; P: posterior. Scale bar: 50 µm.

145

demonstrate that intraocular vasculature does not undergo normal refinement embryonically and shows increased vessel presence in adulthood. In contrast, the absence of Sema3fa permits precocious growth of the extraocular vasculature, coincident with aberrant infiltration of leaky vessels into the embryonic neural retina in a non-Vegfa dependent manner that remain present throughout the lifetime of the fish. My work provides sema3fa knockout zebrafish as a novel model for understanding the cellular and molecular mechanisms of choroidal retinopathies due to environmental changes, and to test novel anti-angiogenics via high throughput drug screening.

Both intra and extra-ocular vascular beds are significantly affected in the sema3fa mutant embryo and adults. Based on my data, I propose a model whereby localized Sema3fa signalling negatively controls vascular growth (Figure 4.14). In the case of the choroid vessels, I suggest that

Sema3fa secreted from the RPE inhibits vessel sprouting, while it is Sema3fa from the inner neural retina and CMZ that helps first refine and maintain a simplified hyaloid and, later retinal vessel network. In support, sema3fa mRNA is expressed by the RPE, inner nuclear layers and CMZ in the larval and adult zebrafish retina. In addition, I find that in the absence of Sema3fa, growth of the choroid vessel bed occurs both precociously and inappropriately through the RPE and into the outer retina. Further, loss of Sema3fa disrupts physiologic hyaloid/retinal vessel refinement resulting in an increased vascularity within the retinal vasculature. Expression of nrp1b in the orbital vasculature supports a paracrine signalling modality, and the role of Sema3fa as anti- angiogenic in the eye. In agreement, unpublished endothelial cell specific RNAseq data from Dr.

Sarah Childs (Whitesell, 2018) identified nrp1b, plxnd1 and plxnb3 transcript enrichment at 5 dpf.

In support of its role as an anti-angiogenic in the eye, AAV-mediated expression of Sema3f in mouse models is sufficient to inhibit sprouting from retinal and choroidal vessels

146

Figure 4.14. Summary and working model of roles for Sema3fa in regulating vascularization of the neural retina. (A) In wildtype, the choroid plexus (C) remains outside of the RPE and

Bruch’s membrane (bm) through the release of Sema3fa (purple hexagon) by the RPE. (B) If sema3fa signalling is perturbed, vessels pass through the RPE and leak into the nuclear layer containing the photoreceptors (cones, C, and rods, R) and INL containing the horizontal cells (HC) and bipolar cells (BC). In the hyaloid/retinal vascular bed, Sema3fa (purple hexagon) maintains spacing and reduced density of blood vessel networks in wildtype retinas (C), but in the mutant environment, blood vessels overgrow and attempt to make aberrant connections with each other

(D).

147

(Buehler et al., 2013; Sun et al., 2017a). While the original description of the phenotypes of the

Sema3f null mouse did not reveal vascular abnormalities (Sahay et al., 2003), it is unclear how careful an analysis was undertaken. While my data and that in mouse argues that Sema3f functions to inhibit eye vessel growth, in other systems, such as the placenta (Regano et al., 2017), Sema3f may act in a pro-angiogenic fashion (Kessler et al., 2004). Thus, Sema3f likely functions in a tissue specific manner.

With the loss of Sema3fa, the choroid plexus forms precociously and coincides with pathologic vascularization into the neural retina. Plexus overgrowth can be due to a number of reasons – increased endothelial cell proliferation, increased stability of vessels, and/or proangiogenic cellular guidance. VEGFA acting through FLK-1 is necessary for endothelial cell proliferation (West et al., 2012). Through several intracellular signalling pathways, this pair induce the recruitment and activation of Src kinases (Mahabeleshwar et al., 2007), which mediate

VEGFA mitogenic signalling (Werdich and Penn, 2005) by activating the MAPK intracellular signalling pathway (Soldi et al., 1999). Once new vessels are established, the endothelial cells become quiescent and resistant to many exogenous factors. In the mouse retina, Müller glia- derived VEGF contributes to pathologic neovascularization (Bai et al., 2009). At larval stages, where CNV is present in the mutant embryos, I find no evidence of expression of vegfaa or vegfab within the neural retina by ISH. These data suggest that the expansion overgrowth of the choroid plexus and presence of infiltrating vessels in sema3fa mutants is not Vegfa-dependent. However, without RT-qPCR data to validate, I cannot completely rule out a Vegfa role. Given that Sema3f signalling through Nrp2 inhibits Vegf-induced ERK1/2 phosphorylation and proliferation of endothelial cells in vitro (Kessler et al., 2004), it will be important to determine whether the plexus is in fact overgrown due to a loss of inhibitory control of endothelial cell proliferation.

148

Vegf is not only an endothelial cell mitogen (West et al., 2012), but a pro-angiogenic guidance cue (Gerhardt, 2008). And while it is an important and likely major factor in the pathogenesis of neovascularization, there exist other proinflammatory and angiogenic factors, including Fibroblast growth factor (FGF)s, Insulin-like growth factor 1 (IGF-1), Placental growth factor (PIGF), Angiogenin (Ang), Platelet-derived growth factor (PDGF), Erythropoietin (Epo), interleukins and chemokines (Gariano and Gardner, 2005; Sherris, 2007). If we consider cellular guidance is an interplay between “stop” and “go” signals, the literature is saturated with proangiogenic mechanisms, with anti-angiogenics only relevant when pathological vessels are present

(Serini and Tamagnone, 2015; Sun et al., 2017a). I propose that Sema3fa signalling is a necessary, and active process that maintains vessel quiescence. In support, I show that when Sema3fa is lost in the environment, CNV ensues in a non-Vegf dependent manner. With no identification of a proangiogenic guidance cue (Vegfa) in the retina driving pathological growth, I propose that the loss of the anti-angiogenic cue (Sema3fa) removes the “brakes” on vessel growth and permits vessels to wander into the retina.

Choroidal vessels that invade the neural retina are generally immature and lack proper tight-junctions and smooth muscle/pericyte coverage, and are therefore prone to leakage

(Törnquist et al., 1990). Leakage and subsequent fluid accumulation in the macula of the human retina is what accelerates vision loss in wet AMD (Saint-Geniez and D’Amore, 2004). Leakage is likely due to disruption of the barrier between the vasculature and retina. The blood retina barrier

(BRB) is a selective barrier which facilitates and regulates the transportation of substances in between the neural retina and blood (Cunha-Vaz et al., 2011). The BRB has two components; the outer BRB consists of tight junctions between RPE cells, which prevent passage of substances, while the inner BRB is comprised of tight junctions between endothelial cells. Dysfunction in

149

maintaining these junctions results in vascular leakage and subsequent edema (Bharadwaj et al.,

2013). The nearly 100% penetrance of vascular leakage (though no obvious edema was present) in sema3fa mutants suggests a disruption to the BRB. Is the leakiness directly attributed to the loss of Sema3fa, or is it a secondary consequence of the formation of small capillaries that navigate into the subretinal space? Currently, literature suggests that the BRB breakdown is due to an accumulation of reactive oxygen species (El-Remessy et al., 2003). I did not explore this in my own work, however RNAseq data from Chapter 2 suggests an increase in DNA damage and reduced cell viability – hallmarks of aberrant ROS activity.

My preliminary analysis of the physical and physiological state of the RPE suggests no abnormalities in the epithelium. Certainly, this investigation was not extensive, and the RPE has many functions that Sema3fa might impact, though this analysis is beyond the scope of my thesis.

One function that is worth mentioning is that of RPE phagocytosis of shed photoreceptor outer segments – a physiological function necessary in the visual cycle (Simó et al., 2010). While not quantified in my work, the outer segments of the photoreceptors do appear shorter in the mutant at larval stages (Figure 4.10), and disorganized in the adult fish (Figure 4.9). Thus, it is possible that Sema3fa, in addition to being an anti-angiogenic, functions to maintain the health of the RPE and underlying photoreceptors, and that their dysfunction could facilitate blood vessel entry as the epithelium is “unhealthy”. To test RPE phagocytosis, I could use Oil Red O staining of cryosections to label lipid (therefore photoreceptor outersegment membranes) containing vesicles within the RPE (Schonthaler et al., 2008). An accumulation of stain would indicate an impairment in lysosomal degradation of phagocytosed membranes and/or excessive phagocytosis.

To determine the physical intactness of the RPE, I used pigment presence and the RPE specific immunolabel Zpr2. Although both assays argue that there are no gaps in the epithelium, I

150

did not directly assay RPE barrier ability. The RPE is held together by three intercellular junctions: tight, gap and adherens. Tight junctions prevent passage of unregulated substances and prevent fluid leakage from the overlying fenestrated choroidal capillaries, acting as part of the blood-retina barrier (Bharadwaj et al., 2013), gap junctions allow cell-to-cell communication, and adherens junctions provide mechanical attachment between RPE cells (Obert et al., 2017). In AMD, pathological levels of VEGF impair the expression of the tight junction structural protein zona- occludens 1 (ZO-1), increasing layer permeability (Fischer et al., 2002; González-Mariscal et al.,

2003). Unfortunately, a different lot of the ZO-1 antibody that was effective in zebrafish (Hehr et al., 2018) provided unreliable staining of both the mutant and wildtype RPE at later stages in larval development (4 dpf+). Potentially, I could source the antibody from a different company and test whether the integrity of the RPE as a barrier is compromised in the Sema3fa mutant. Additionally, the role of Müller glia in maintaining the BRB was recently described, with pharmacological destruction of the glia producing CNV in mouse (Li et al., 2018). In Chapter 3, I found no obvious difference in glia between mutants and wildtype retinas, however, my analysis ends at 72 hpf.

Considering the pathogenesis of the infiltrating vessels is a much later phenotype (7 dpf), an assessment of Müeller glia in the presence and absence of Sema3fa needs to be performed to determine whether glial architecture changes, in addition to a Sema3fa anti-angiogenic function, contribute to the pathogenesis of the disease and dissolution of the BRB.

Dysregulation of crosstalk between the neural retina and vasculature is recognized as a major contributor to the pathogenesis of vascular retinopathies (Sun et al., 2017b). The adult retinal vasculature on the inner limiting membrane of the retina was particularly dense in the mutant compared to their wildtype siblings. In mammals, retinal vasculature is thought to associate with a glial scaffold made of astrocytes which influence branching pattern (Fruttiger, 2007). This

151

influence is in part mediated by VEGF secretion, as astrocyte specific deletion of Vegf reduces artery branch numbers, with the weak penetrance of the phenotype likely because the neural retina itself can provide VEGF (Scott et al., 2010). In fact, whether the scaffold is that of a mature astrocytic network or one of immature differentiating astrocytes is still in question (Duan et al.,

2017), and so it is unclear whether the retinal vasculature is patterned by the underlying astrocytic networks. Interestingly, no astrocytic scaffold has been identified in zebrafish, though Müller glia appear to be tightly associated with the retinal vessels, whereby glial end feet form intimate connections with the endothelium following vascular dissection (Alvarez et al., 2007). It is possible that this glial population plays some role in vascular patterning in teleosts. If patterning and stability of the retinal vasculature is mediated in part by Vegf, then it is logical that an opposing system refines these networks to prevent overgrowth. I propose that Sema3fa from the neural retina is involved in “spacing” and calming vessel growth along the inner limiting membrane (Figure

4.14C,D). Since Sema3fa is secreted, this source could come from any of the nuclear layers which express (retinal ganglion cell layer, inner nuclear layer, ciliary marginal zone). One interesting idea is that Müller glia, whose cell bodies reside in the inner nuclear layer, secrete Sema3fa preferentially from their end feet in contact with the inner limiting membrane and overlying endothelium. To investigate this possibility, a myc or HA Sema3fa fusion protein would need to be generated to determine protein localization, alongside immunohistochemistry for Müller glia.

It is clear that the zebrafish mutant I generated serves as an invaluable model for choroidal and vascular retinopathies – but does it truly model any specific retinal disease? I have two pieces of evidence that suggest that the retinopathy described above is most similar to that of wet AMD.

Crystallin proteins, first annotated as structural proteins of the transparent lens (Wistow and

Piatigorsky, 1988), are of new focus in their role in retinal degeneration (Fort and Lampi, 2011).

152

Three major classes exist, alpha-, beta-, and gamma-crystallin and their upregulation, downregulation or bioaccumulation are described in various retinal diseases (Fort and Lampi,

2011). AMD pathogenesis implicates crystallin - where crystallin protein accumulates in the fatty deposits of drusen (Kannan et al., 2016). In my RNAseq analysis of wildtype and sema3fa mutant zebrafish at 36 hpf (Chapter 3), 13 crystallin isoforms (beta and gamma) are upregulated 2-15 fold in the mutant compared to wildtype (Table 6.1). The functions of these Crystallins are of ongoing research, however, it is clear they play cytoprotective roles and modulate angiogenesis in a VEGF- dependent manner (Kase et al., 2010). It is possible that the induction of Crystallins in the Sema3fa mutant eye promotes angiogenesis. Through both ISH and immunohistochemistry of the various

Crystallin isoforms, I will start to address this point.

My second line of support for the value of the Sema3fa mutants as a model of AMD also comes from the RNAseq data, in which I identified a 1.5 fold induction of the complement system

(Table 6.1). Clinical and genetic data support that AMD is burdened with low-grade inflammation and complement factor H, a key regulatory glycoprotein of the complement mediated immune system. Complement H is the first significantly associated gene with AMD (Edwards et al., 2005;

Hageman et al., 2005; Haines et al., 2005). While the precise role for complement in AMD pathogenesis is not fully defined, it is clear that its activation is contributory to the disease (Kawa et al., 2014). The expression of complement factors in the larval and adult eyes of sema3fa mutants could be assessed by ISH and RT-qPCR. The low-grade inflammation in AMD is driven in part by macrophage, and microglial accumulation in the choroid capillaries (McLeod et al., 2016). To assess whether the sema3fa mutant presents with a change in phagocytic cell accumulation, live embryos across stages previously defined with CNV in my model, could be incubated in solution containing 5 µg/ml Neutral Red (Demy et al., 2017). Neutral Red is taken up by endocytosis,

153

accumulating in the lysosomes of macrophages and microglia (Herbomel et al., 2001), and will allow for the direct visual comparison of the distribution and number of phagocytic cells within the retina. Taken together, while I have not explored either Crystallins or immune/complement factors in my model, the differential gene expression data from the RNAseq guide towards application of this model for AMD. It is interesting that these “disease related” genes are expressed embryonically in the mutant but not wildtype retinas (36 hpf for RNAseq data), and suggests that the pathogenesis of AMD is an accumulation of multiple insults, and while the embryo is far from being “aged”, the biochemical signalling pathways involved in the disease may already have manifested.

4.4 Significance

In this chapter, I provide biologically relevant evidence that Sema3fa maintains physiologic avascularity of the zebrafish retina. While the role of Semas in mediating blood vessel dynamics is known (Sakurai et al., 2012), I successfully demonstrate that the loss of a Sema3 anti- angiogenic molecular barrier, with no further insults, makes the RPE receptive to blood vessel infiltration. To my knowledge, this has never been shown before. Additionally, I show the absolute robustness of the zebrafish model in mitigating such a pathology. While blood vessels are present aberrantly in the retina throughout the lifetime of the mutants, I do not see evidence for retinal detachments, suggesting that the regenerative and restorative capacity of the organism continually aims to correct the insult. This feature is of particular interest, as understanding the molecular mechanisms that drive recovery can be important clinically for the treatment of retinal disease and degeneration. Of note, while we have just begun to explore the RNAseq data from Chapter 3, the induction of genes associated with retinal neurodegenerative diseases further supports continued

154

exploration of the sema3fa mutant model and more in-depth analysis of the biological role(s) of

Sema3fa in eye physiology and disease.

155

Chapter Five: Sema3fb signalling in heart development

5.1 Introduction

The heart is the first functional organ to form in an embryo. In vertebrates this begins as a

single tubular structure which twists, expands and undergoes septation to form a two-, three- or

four-chambered heart (DeRuiter et al., 1992). These cardiac chambers, the atrium and ventricle,

differ in their biochemical properties, electrophysiological and contractile capacities, as well as in

their transcriptional profiles (Chien et al., 1993; Lyons et al., 1990). What establishes chamber

specific development is an area of active research.

Discrete differentiation of the atrium and ventricles is essential to the function of the heart,

whereby loss of chamber specific myosins result in fatal cardiomyopathies (Chen et al., 1998).

Large bodies of work have identified intrinsic mechanisms, such as transcription factors, necessary

for cardiogenesis and chamber-specific cardiomyocyte development. For example, ventricle

specific expression of Irx4 is necessary for the expression of ventricular myosin heavy chain I

(MHC1), and suppression of atrial MHC1, in chick (Bao et al., 1999). Further, Pitx2 expression

within the left atrium supresses ectopic expression of Shox2, which drives formation of the

pacemaker cells of the sinoatrial node in mouse (Wang et al., 2010a). However, two aspects remain

unclear: what are the extrinsic molecules that provide the spatial information to naïve

cardiomyocytes to drive chamber-specific differentiation, and are there functional consequences

if this localized signalling is disrupted.

Paracrine and juxtacrine extrinsic signalling pathways involving Nodal, Notch, Bone

morphogenetic protein (Bmp), Fibroblast growth factor (Fgf) and retinoic acid are each necessary

for certain aspects of cardiovascular development. For example, Bmp and Notch signalling are

required in the patterning and differentiation of the valves (Garside et al., 2013), while retinoic

156

acid acts as a morphogen to induce expression of Tbx5 in a posterior to anterior fashion (Bruneau et al., 1999). Additionally, Fgfs act bifunctionally to first regulate chamber proportionality, and later ventricular cell number (Marques et al., 2008). How these signals are interpreted to produce restricted expression domains of differentiation within specific chambers or valves is unclear. It is likely that these molecules grossly pattern the heart tube, and establishment of a secondary localized signalling is necessary to drive specific differentiation pathways.

Class 3 Semaphorins (Semas) belong to the family of secreted Semas used as spatial guidance cues for vessels, axons and neural crest cells (Alto and Terman, 2017; Fujisawa, 2004;

Jongbloets and Pasterkamp, 2014; Mecollari et al., 2014). Sema signalling is increasingly implicated in cardiovascular development, both physiologically and in pathology (Epstein et al.,

2015). For instance, in mice SEMA3C (Feiner et al., 2001), SEMA3D (Degenhardt et al., 2013),

SEMA3E (Gu et al., 2005), and SEMA3G (Kutschera et al., 2011) function in embryonic vascular patterning, SEMA3A in sympathetic innervation of the heart (Ieda et al., 2007), and SEMA3C in neural crest cell migration into the heart (Feiner et al., 2001). Sema3 signalling in these systems is mediated through the canonical Plexin (Plxn) receptors and the coreceptor, Neuropilin (Nrp)

(Epstein et al., 2015). No cardiac phenotype has been described in Sema3f knockouts, despite evidence for its expression in the embryonic heart of several vertebrate species (Jin et al., 2006).

In this chapter I take advantage of the zebrafish model as a powerful tool to study the development of the heart, as the embryos do not require a functional cardiovascular system during embryogenesis (Pelster and Burggren, 1996), which allows for the extended characterization of normally fatal manipulations. I provide evidence that sema3fb is a novel extrinsic regulator of cardiac chamber differentiation by using a CRISPR/Cas9 generated sema3fb mutant. I show that sema3fb is expressed early in cardiomyocytes during their specification and differentiation. Loss

157

of sema3fb signalling results in disruption of chamber cardiomyocyte differentiation, whereby

chamber specific gene expression is either lost or expanded outside of the chamber border. The

consequence is significantly impaired cardiac output and function of the larval heart. To our

knowledge, this is the first Sema reported to impact spatialized cardiac chamber differentiation in

a cell autonomous manner.

5.2 Results

5.2.1 sema3fb is expressed by the cardiomyocyte progenitor cells of the developing heart

In zebrafish, cardiac chamber cells of the atrium and ventricle are bilaterally specified by

5 hours post fertilization (hpf) (Stainier et al., 1993). By 18 hpf, these progenitors condense at the

midline and undergo a series of morphological movements to produce a linear heart tube of

differentiating myocardium surrounding the underlying endocardium (Fishman and Chien, 1997;

Stainier et al., 1993). By 48 hpf, the two-chambered is discernable by the constriction of the

atrioventricular canal (AVC) (Beis et al., 2005; Stainier et al., 1993). To elucidate the spatial and

temporal expression of sema3fb in the heart, I performed wholemount RNA in situ hybridization

(ISH) on zebrafish embryos from the 3 somite stage (ss) to 7 dpf, using a previously published

probe (Table 2.2) (Yu and Moens, 2005). sema3fb transcript was detected at the 3 ss to 18 hpf (not

shown) in a bilateral fashion in the presumptive cardiac progenitor fields (Figure 5.1A). By 24 hpf,

sema3fb was expressed by cardiomyocyte progenitor cells throughout the entire linear heart tube

(Figure 5.1B). As the heart undergoes rightward looping and chamber morphogenesis (24-48hpf),

sema3fb label was present throughout both the ventricle and atrium (Figure 5.1C). Sparse

expression persisted throughout the ventricular and atrial myocardium of the 72 hpf heart (Figure

5.1D), while the bulbus arteriosus highly expressed sema3fb mRNA through 7 dpf (data not

shown).

158

Figure 5.1. sema3fb is expressed by the developing heart. Whole mount in situ hybridization viewed dorsally (A), laterally (B) and ventrally (C-D) with antisense riboprobe for sema3fb mRNA. mRNA is detected throughout the developing myocardium at all stages analyzed. a: atrium; ba: bulbus arteriosus; fp: floor plate; ht: heart tube; l: lens; v: ventricle. Scale bar: 50 µm.

159

Of note, no commercial antibody exists for zebrafish Sema3fb, so I tested instead an antibody against mouse SEMA3F. As mentioned in Chapter 3, this antibody failed to recognize any protein by Western blot, and promiscuously labelled (likely) many Sema3s by immunohistochemistry of zebrafish cryostat sections.

5.2.2 Sema3fb is necessary for spatially restricted chamber differentiation

5.2.2.1 Generation of the sema3fb genetic mutant

Through CRISPR/Cas9 gene editing technology (Gagnon et al., 2014), I generated two allelic variants to assess Sema3fb loss of function in vivo. I generated one sgRNA (Table 2.1) to target exon 1 (Figure 5.2A), which produced a single founder that housed both allelic variants (19 bp deletion and 10bp insertion). Both lines were propagated from this founder to generate stable homozygous viable generations, alongside wildtype siblings. The 19 bp deletion (hereafter Δ19) and 10 bp insertion (hereafter +10) alleles are predicted to produce premature truncations (32aa and 58aa in length, respectively) within the 500aa SEMA domain that is necessary to elicit intracellular signalling (Tamagnone et al., 1999) (Figure 5.2B). Since the mouse SEMA3F antibody described earlier did not appear to be specific to Sema3f, I instead employed RT-qPCR to detect relative transcript levels of mRNA isolated from 48 hpf wildtype and Δ19 embryos. There was a clear reduction in sema3fb mRNA levels in the mutant as compared to wildtype (Figure

5.2C), suggestive of nonsense mediated decay (Hentze and Kulozik, 1999).

5.2.2.2 Embryos that lack Sema3fb exhibit pericardial edema

F1 heterozygous adults were incrossed and progeny assessed throughout early embryonic development. Interestingly, half of the progeny presented with obvious but varying degrees of pericardial edema at 48 hpf. Genotyping of embryos with edema vs. non-edema phenotypes revealed afflicted embryos were either heterozygotes or homozygotes (data not shown). To

160

Figure 5.2. Generation of the sema3fb genetic mutant. (A) Chromosomal overview of the sema3fb locus targeted by CRISPR/Cas9 mutagenesis to exon 1 (guide RNA: B1) and primers used to identify the mutation. UTR: untranslated region; F: forward primer; R: reverse primer. (B)

Schematic representation of wildtype and premature stop codon mutant proteins. sema3fb +10 mutants have a 10 bp insertion (red sequence) to produce a predicted protein product of 58 amino acids (aa) while sema3fb Δ19 have a 19 bp deletion (dashes) to produce a predicted product of 32 aa. Striped patterns represent missense amino acids (miss.aa). (C) RT-qPCR of sema3fb mRNA levels in wildtype and sema3fb Δ19 embryos at 48 hpf suggest nonsense mediated decay of mRNA transcript (N=2). Error bars represent standard error of the mean (SEM).

161

determine the penetrance of this phenotype in the mutant, lines were bred to homozygosity and incrossed within their respective genotype (sema3fb +/+, sema3fb Δ19/Δ19, sema3fb +10/+10).

Embryos were indistinguishable by gross size at 72 hpf (Figure 5.3A-C), but homozygous embryos presented with edema preferentially in both allelic variants at 48 hpf (8.3% +/+, 91.7% Δ19, 76.7%

+10, N=3, n=60 each genotype, Figure 5.3D) and 72 hpf (5.0% +/+, 91.7% Δ19, 41.7% +10, N=3, n=60 each genotype, Figure 5.3E) as compared to wildtype. The degree of severity of the edema fell on a spectrum, with most embryos showcasing visible edema. While most embryos with edema developed to adulthood, a small proportion (<10%) of homozygous embryos exhibited severe cardiac edema and expired by 5-7 dpf. All future analyses were carried out on embryos in which the pericardial edema was of the average expressivity, and those with lethal phenotypes were excluded so as not to confound any of the data collected herein.

5.2.2.3 sema3fb mutants have smaller hearts that are only delayed slightly in their development

Pericardial edema is generally indicative of a cardiac defect, which may arise due to problems in cardiac progenitor cell specification and differentiation, morphogenesis of the heart tube, chambers and AVC, or the establishment of normal cardiac function (Miura and Yelon,

2012). Because sema3fb is expressed early in the heart fields, I first analyzed whether the heart tube and chambers developed normally, as these would be the most obvious causative candidates for the edema. In zebrafish, cardiogenesis occurs rapidly, and so precise staging and timing of sample fixation is of utmost importance. I ensured all samples processed would be of comparable ages (as described in the Methods). At 24 hpf, the heart is expected to elongate through tube extension (Brown et al., 2016; Grant et al., 2017). Using a cardiomyocyte specific antisense

162

Figure 5.3. sema3fb mutants have cardiac edema. Representative brightfield images of 72 hpf wildtype (A), sema3fb Δ19bp (B) and sema3fb +10 (C) embryos. Dotted line outlines the heart region. Quantification of the percentage of embryos of the different genotypes that present with cardiac edema at 48 hpf (N=3, n=60 for each genotype) (D) and 72 hpf (N=3, n=60 for each genotype) (E). Most mutant embryos of the two alleles present with edema at 48 hpf, with some recovery by 72 hpf. Scale bar: 1 mm.

163

riboprobe to cardiac troponin t2a (tnnt2a) (Hsiao et al., 2003), I found that 88% of sema3fb +/+

(N=3, n=15/17) embryos were in tube extension, as compared to 0% of the Δ19 and +10 embryos

(Figure 5.4A-C). Instead, mutant embryos appeared to be in either the cone (67% for Δ19, N=3, n=14/21) or pre-tubulogenesis/condensed state (71% for +10, N=3, n=15/21), as quantified in

Figure 5.4G. With the heart affected in mutants at a stage (24 hpf) when the heart tube is elongating, I next asked what the consequence of such a defect was on development of the cardiac chambers. Cardiac chambers are most appropriately studied following the twisting and bulging of the heart tube to form the single atrium and ventricle at around 48 hpf (Brown et al., 2016). Using the tnnt2a riboprobe, I found no abnormal looping, and a ventricle and an atrium were present across all genotypes (Figure 5.4D-F). Hearts were smaller on average in mutants as compared to wildtypes, as assessed by quantifying the linear A-P length of the labelled heart (Figure 5.4H).

Both the Δ19 (N=3, n=16, p<0.0001) and +10 (N=3, n=16, p=0.0023) mutant embryos were significantly shorter than controls (N=3, n=18) by at least 25%.

It was clear that the heart as a whole was affected in the mutant, and so I next asked whether the defect was restricted to one or both of the cardiac chambers. I first used wholemount ISH for myosin heavy chain 7 (myh7), which specifically labels ventricular cardiomyocytes (Yelon et al.,

1999). At 24 hpf, the hearts of the majority of Δ19 (63%; N=3, n=12/19) and +10 (68%; N=3, n=13/19) embryos were in a cone state, while those of wildtype controls (85%; N=3, n=17/20) were undergoing tube extension (Figure 5.5A-C,G). Therefore, ventricular identity of cardiomyocytes is at least specified with respect to myosin expression. At 48 hpf (Figure 5.5D-F), the ventricle appeared smaller in mutants as compared to wildtype. Indeed, the long axis of the ventricle of the Δ19 (N=3, n=27, p=0.0085) and +10 (N=3, n=26, p=0.0133) embryos was significantly shorter than controls (N=3, n=28) by at least 15% (Figure 5.5H). This size reduction

164

Figure 5.4. Heart tube extension is disrupted in sema3fb mutant embryos. Heart development was visualized by whole mount in situ hybridization using the cardiomyocyte specific tnnt2a riboprobe. (A-C) Dorsal views of 24 hpf embryos whereby the heart elongates in the wildtype to form a tube (N=3, n=15/17) (A), but progression into the extension stage is defective in both sema3fb mutant alleles (N=3, 14/21 Δ19; N=3, 15/21 +10) (B-C). (G) Quantification of cardiac morphologies observed. (D-F) Ventral views of 48 hpf embryos reveal total heart morphology following chamber development. By 48 hpf, the hearts of wildtype embryos loop normally with large chambers present (N=2, n=16/18) (D), while both sema3fb mutant alleles present with normally looped, but smaller hearts (N=2, 14/16; N=2, 15/16) (E-F). (H) Quantification of the anterior to posterior length (red line) of the heart (see schematic) reveals significantly smaller hearts in the sema3fb Δ19 (N=3, n=16, ***p<0.0001) and sema3fb +10 (N=3, n=16, **p=0.0023) hearts as compared to controls (N=3, n=18). Error bars are standard error of the mean (SEM).

Statistics represent Kruskal Wallis One Way ANOVA, followed by Dunn’s method for multiple comparisons. A: anterior; a: atrium; P: posterior; v: ventricle. Scale bar: 50 µm.

165

Figure 5.5. Ventricle chamber extension is disrupted in sema3fb mutant embryos. Ventricle chamber development was visualized by whole mount in situ hybridization using the ventricular cardiomyocyte specific myh7 riboprobe. (A-C) Dorsal views of 24 hpf embryos show the heart elongates in wildtypes (N=3, n=17/20) (A), but that progression into the extension stage is defective in both sema3fb mutant alleles (N=3, 12/19 Δ19; N=3, 13/19 +10) (B-C). (G)

Quantification of cardiac morphologies observed. (D-F) Ventral views of 48 hpf embryos reveal ventricle morphology following chamber development. By 48 hpf, wildtype embryos present with an elongated ventricle chamber (N=2, n=19/20) (D) while both sema3fb mutant alleles present with observably smaller ventricles (N=2, 17/20 Δ19; N=2, 15/18 +10) (E-F). (H) Quantification of the long axis (red line) ventricle length reveals significantly shorter ventricles in sema3fb Δ19

(N=3, n=27, *p=0.0085) and sema3fb +10 (N=3, n=26, p=0.0133) embryos as compared to controls (N=3, n=28). Error bars are standard error of the mean (SEM). Statistics represent Kruskal

Wallis One Way ANOVA, followed by Dunn’s method for multiple comparisons. A: anterior; P: posterior. Scale bar: 50 µm.

166

was still observed at 72 hpf, suggesting that the phenotype did not recover with further embryonic development (Figure 5.6).

To assess atrial development, I used a riboprobe for myosin heavy 6 (myh6) (Berdougo et al., 2003). At 24 hpf, 95% of Δ19 (N=3, n=19/20) and 61% of +10 (N=3, n=11/18) embryonic hearts were in a cone state, while those of the vast majority of wildtype controls (85%; N=3, n=17/20) had undergone tube extension (Figure 5.7A-C,G). Therefore, atrial identity of cardiomyocytes is at least specified with respect to myosin expression. By 48 hpf, while a myh6 positive atrium was present in mutant embryos (Figure 5.7D-F), it was obviously smaller than in wildtypes. I measured the area of the atrium and found that on average the Δ19 (N=3, n=19, p<0.0002) and +10 (N=3, n=10, p=0.0001) hearts were significantly smaller than controls (N=3, n=11) by at least 35% (Figure 5.7H). Ultimately, these data suggest that while both ventricle and atrium are specified, the chambers are affected equivalently in size in both mutant allele backgrounds. As the phenotypes between the two allelic variants was highly comparable, I characterized only the sema3fb Δ19 mutant in all subsequent analyses.

5.2.2.4 Chamber specific expression patterns are disrupted in the sema3fb mutant

Atrial (myh6) and ventricular (myh7) specific myosins distinguish the two populations of myocardial precursors from the earliest stages of heart development (13 hpf), and through the development and life of zebrafish (Berdougo et al., 2003; Yelon et al., 1999). As such, I expected to see restricted expression of myh7 and myh6 within their respective chambers, separated by the

AVC. Instead, in the majority (59%) of sema3fb Δ19 embryos (N=4, n=17/29), myh7+ cells were detected at 48 hpf in the most anterior atrium as compared to in 13% of sema3fb +/+ embryos

(N=4, 4/30) (Figure 5.8A-B). Further, I found that 62% of sema3fb Δ19 embryos (N=5, n=16/26), and only 15% of sema3fb +/+ embryos (N=5, 6/39), showed expansion of the myh6 atrial marker

167

Figure 5.6. Ventricles are smaller in sema3fb mutants. Ventricle measurements were made from still images taken during video microscopy of wildtype and mutant hearts during diastole (A,C) and systole (B,D) at 72 hpf. Quantification of the ventricle diameter during diastole is significantly smaller (A,**p=0.0017) and during systole (B, **p=0.0063). In mutants as compared to wildtype ventricle length is significantly shorter at diastole (C, *p=0.0482) and systole (D, *p=0.0309).

Error bars are standard error of the mean (SEM) over 3 independent experiments (n=12 wildtype, n=20 sema3fb Δ19). Statistics represent the non-parametric Mann-Whitney U test.

168

Figure 5.7. Atrial chamber development is disrupted in sema3fb mutant embryos. Atrial chamber development was visualized by whole mount in situ hybridization using the atrial cardiomyocyte specific myh6 riboprobe. (A-C) Ventral views of 24 hpf embryos where atrial cardiomyocytes are tubularized in wildtype (N=3, n=17/20) (A), but are primarily at the cone stage in both sema3fb mutant alleles (N=3, 19/20 Δ19; N=3, 11/18 +10) (B-C). (G) Quantification of cardiac morphologies observed. (D-F) Ventral views of 48 hpf embryos reveal atrial morphology following chamber development. By 48 hpf, wildtype embryos present with an enlarged chamber

(N=2, n=19/19) (D) while both sema3fb mutant alleles present with observably smaller atria (N=2,

18/21 Δ19; N=2, 15/17 +10) (E-F). (H) Quantification of surface area (red fill) reveals significantly smaller atria in sema3fb Δ19 (N=3, n=19, ***p=0.0002) and sema3fb +10 (N=3, n=10,

***p<0.0001) hearts as compared to wildtype (N=3, n=11). Error bars are standard error of the mean (SEM). Statistics represent Kruskal Wallis One Way ANOVA, followed by Dunn’s method for multiple comparisons. A: anterior; P: posterior. Scale bar: 50 µm.

169

Figure 5.8. Cardiac chamber specific gene expression is disrupted in sema3fb mutants.

Images were taken ventrally of wholemount 48 hpf embryos processed for in situ hybridization.

Wildtype expression of the ventricle marker myh7 (A, N=4, n=26/30) is restricted to the ventricle in wildtype embryos (dashed white line), but expands into the atrium in mutants (B, N=4, n=17/29). Expression of atrial marker myh6 in wildtype embryos (N=5, n=33/39) is confined to the atrium (dashed white line), but extends into the ventricle of the mutants (N=5, n=16/26). A: anterior; P: posterior. Scale bar: 50 µm

170

throughout the entire ventricular myocardium (Figure 5.8C-D). These data argue that in the absence of sema3fb there is a failure to restrict chamber specific myosins, and support the idea that the chamber-specific differentiation of cardiomyocytes in the sema3fb mutant is not spatially discrete.

To understand whether the failure in restriction of a key atrial marker from the ventricle affected the ventricle differentiation program, I assayed markers of ventricle differentiation at 48 hpf. The homeobox containing protein iroquois1a (irx1a) is restricted to the cardiac ventricle by

48 hpf, when chamber morphogenesis occurs (Cheng et al., 2001). The majority (73%) of homozygous sema3fb mutants lacked any detectable irx1a in their ventricles (N=2, n=8/11), as compared to their wildtype counterparts (N=2, 6/7) (Figure 5.9A-B). I next looked at the expression of the T-box transcription factor tbx5a, which is expressed in a high to low gradient in the ventricle and atrium of embryonic zebrafish (Garrity et al., 2002). Interestingly, the expression pattern was unchanged between wildtype (N=1, n=9) and sema3fb Δ19 embryos (N=1, n=8)

(Figure 5.9C-D).

To determine whether the disruptions to chamber specific myosin and irx1a expression could be due to changes in signalling pathways critical for heart development, I investigated the expression of bmp4a, which is expressed initially throughout the antero-posterior length of the myocardium, and becomes restricted to the AVC region by 37 hpf (Walsh and Stainier, 2001).

While bmp4a expression was restricted to the valves in 90% of wildtype embryos (N=2, n=9/10), all mutant embryos (N=2, n=11/11) showed expansion of the bmp4a AVC expression domain through the entire ventricular myocardium (Figure 5.9E-F). To determine whether these global changes to the ventricle differentiation program were transient or sustained, I analyzed embryos at

72 hpf. At this stage cardiac function is considered rhythmic, the circulatory system mature, and

171

Figure 5.9. Ventricle identity is disrupted in sema3fb mutants. Images were taken ventrally of wholemount 48 hpf embryos processed for in situ hybridization. Expression of the ventricle specific marker irx1a is diminished (asterisk) in the mutant (B, N=2, n=8/11) as compared to wildtype (A, N=2, n=6/7). Expression of tbx5a, another ventricle marker, is present in both wildtype (C, N=1, n=9/9) and mutant (D, N=1, n=8/8) ventricles. Expression of the atrioventricular valve marker, bmp4, is restricted to the valves (white arrowhead) in wildtype hearts (E, N=2, n=9/10), but expands through the entire ventricle in the mutant (F, N=2, n=11/11). A: anterior; P: posterior. Scale bar: 50 µm.

172

the atrium has assumed a dorsal position (Denvir et al., 2008). Expression of irx1a was still diminished in 63% of mutants (N=1, n=5/8) as compared to wildtype embryos (N=1, n=4/4)

(Figure 5.10A-B). Additionally, ventricular expansion of bmp4a was present in 67% of sema3fb homozygotes (N=1, n=8/12), but bmp4a expression was valve-restricted in all wildtype embryos

(N=1, n=10/10) (Figure 5.10C-D). Overall, these changes support the idea that in the absence of

Sema3fb the clear separation of cardiac cells into different chambers with distinct differentiated properties is disrupted.

5.2.2.5 sema3fb mutant hearts do not present sharp boundaries between chambers

I next asked what effect the disruption in the differentiation program of the cardiac chambers had on the expression of proteins necessary for cardiac function. MF20, labels striated myocardium expressing sarcomeric protein Myh1e (Bader et al., 1982), and is expressed more highly in the ventricle than in the atrium, as the antibody labels striated myocardium that predominates in the ventricle (Grimes et al., 2006). The cell-surface adhesion molecule DM-

GRASP (Fashena and Westerfield, 1999), is also expressed at higher levels in the ventricular than atrial myocardium, and by 55 hpf begins to be expressed by differentiated atrioventricular endocardial cells making up the endocardial cushions (Beis et al., 2005). Immunostaining with antibodies for both proteins revealed a clear distinction between the ventricle and atrium of 48 hpf hearts of wildtype embryos (N=1, n=5/5 MF20, N=3, n=11/12 DM-GRASP, Figure 5.11A,C) as assessed in confocal maximal projections. In contrast, this border was muddled in at least 83% of mutants (N=1, n=8/9 MF20, N=3, n=10/12 DM-GRASP, Figure 5.11B,D). These data provide further support to the idea that chamber restricted differentiation, on the protein level, is perturbed in the absence of Sema3fb.

173

Figure 5.10. Disruption of ventricle identity is sustained in sema3fb mutants. Images were taken ventrally of wholemount 72 hpf embryos processed for in situ hybridization. Expression of the ventricle specific marker irx1a is still diminished (asterisk) in the mutant (B, N=1, n=5/8) as compared to wildtype (A, N=1, n=4/4). Expression of bmp4a remains expressed throughout the heart of the mutant (D, N-1, n=8/12) as compared to the wildtype embryos (C, N=1, n=10/10) A: anterior; P: posterior. Scale bar: 50 µm.

174

Figure 5.11. Disruption of chamber specific protein expression with Sema3fb loss. Confocal images of wholemount immunohistochemistry for MF20 (A,B) and DM-GRASP(C-D) of 48 hpf hearts. MF20 and DM-GRASP are more highly expressed in the ventricle, and exhibit a fairly sharp expression border at the atrioventricular constriction (yellow arrowheads) of wildtype hearts

(A, N=1, n=5/5, and C, N=3, n=11/12). In contrast, expression of both proteins expands into the atrium of the mutant hearts (B, N=1, n=8/9 and D, N=3, n=10/12). A: anterior; P: posterior. Scale bar: 50 µm.

175

5.2.3 Small heart phenotype is due to a reduction in cardiomyocyte size

5.2.3.1 Cardiomyocytes are specified and develop normally in sema3fb mutants

Since sema3fb mRNA is expressed early in development within the presumptive heart field, I asked whether early events of heart development were impacted by the loss of Sema3fb.

RNA in situ hybridization on 13 ss embryos was performed to assess the presence and specification of heart precursors. fgf8a mRNA was assessed first as Fgf8a is necessary for the induction and differentiation of cardiac precursors (Reifers et al., 2000). I found the presumptive cardiac precursors expressed fgf8a at similar levels and distribution across the genotypes, suggesting no defects in this extracellular signalling pathway (Figure 5.12A-C). Importantly, fgf8a was expressed in a similar fashion throughout the embryo – in the anterior forebrain, midbrain-hindbrain boundary and within the somites – in wildtypes and mutants (Figure 5.12D-F), arguing that gross embryonic development occurs normally in mutants. Next, I analyzed the expression of tbx5a within the anterior lateral plate mesoderm (Begemann and Ingham, 2000), as tbx5 is essential for the differential establishment of atrial and ventricular cardiomyocyte fates in vertebrates (Bruneau et al., 2001; Liberatore et al., 2000). tbx5a in situ label was similar across the genotypes (Figure

5.12G-H), arguing that the specification and differentiation of the cardiac progenitor cells were unaffected at the 13ss in sema3fb mutants.

The expression of fgf8a and tbx5a also argue that embryogenesis occurs on a similar time course between genotypes. Nonetheless, to address whether the failure of the heart to elongate in sema3fb mutants at 24 hpf was due to a general developmental delay I took two approaches. First,

I used tnnt2a expression to determine when the embryonic heart of the mutants underwent tube formation, which occurs in wildtype embryos by 23 hpf. As presented in Figure 5.13(A-C,F) tube formation does occur in sema3fb mutants, with around 80% of Δ19 (N=1, n=8/10) and 90% of +10

176

Figure 5.12. Gross patterning of the anterior lateral plate mesoderm is not disrupted in sema3fb mutants. Dorsal (A-C, G-H) and lateral (D-F) views of 13 ss wildtype (A,D,G), sema3fb

Δ19 (B, E, H), and sema3fb +10 (C, F, I) embryos processed by wholemount in situ hybridization for fgf8a (A-F, N=2, n=10) and tbx5a (G-H, N=1, n=8) expression. Anterior lateral plate mesoderm is indicated by arrowheads. Scale bar: 50 µm for A-C, G-H. Scale bar: 100 µm for D-F.

177

Figure 5.13. Heart tube extension defect is not due to global developmental delay. Dorsal images of embryos processed for wholemount in situ hybridization for the ventricle specific marker myl7. Heart tube elongation was scored for wildtype embryos at 23 hpf (N=1, n=9/10), and at 26 hpf for sema3fb Δ19 (N=1, n=8/10) and sema3fb +10 (N=1, n=9/10) embryos. (F) Percentage occurence of heart morphologies between genotypes at their respective age. DM-GRASP wholemount immunostaining of embryos at 24 hpf and viewed ventrally show the stereotypic presence of a cranial ganglia (red asterisk) that has migrated normally despite the heart developmental delay (yellow arrowhead).

178

(N=1, n=9/10) mutants exhibiting elongated heart tubes at 26 hpf, comparable to the numbers observed in control (N=1, n=9/10). These data suggest, however, that mutant embryos are at least

3 hours behind with respect to heart development. The second approach involved use of the antibody to DM-GRASP at 24 hpf. At this stage, DM-GRASP labels cardiomyocytes, as well as the trigeminal ganglion (Liu et al., 2011), which projects fibers posteriorly to the spinal cord and anteriorly along the surface of the head (Zhao et al., 2013). While the heart of the sema3fb mutant was visibly in a cone state (N=1, n=4/4, Figure 5.13E), unlike the elongated tube in the wildtype

(N=1, n=4/4, Figure 5.13D), the trigeminal nerve cell bodies and axon processes were similar in the two genotypes (Figure 5.13D,E). Overall, these data support the idea that the defect observed in sema3fb mutant hearts at 48 hpf is independent of any significant gross developmental delay.

5.2.3.2 Cardiomyocyte contributions from the second heart field and neural crest cell

populations do not explain the sema3fb mutant heart phenotype

The smaller size of the 48 hpf heart in mutants vs. wildtype could be due to a decrease in the number of cardiomyocytes. While cardiomyocyte progenitor cells are specified and differentiate in the absence of Sema3fb, changes in proliferation and/or apoptosis could account for the observable difference in heart size. An antibody against phosphorylated histone H3 (pHH3), to label mitotically active cardiomyocytes, revealed no significant differences (p=0.29 and p=0.71, respectively) between 24 and 48 hpf wildtype (N=3, n=6 at 24 hpf, N=3, n=16 at 48 hpf) and mutant (N=3, n=5 at 24 hpf, N=3, n=27 at 48 hpf) hearts (Figure 5.14A-B). Additionally, no significant changes were observed in cardiomyocyte apoptosis (p=>0.99) between wildtype (N=2, n=8) and mutant (N=2, n=8) 48 hpf embryos, as assessed by TUNEL (Figure 5.14C).

Heart tube elongation is largely driven by the additional integration of cardiomyocytes from the second heart field (de Pater et al., 2009) and cardiac neural crest (CNC)

179

Figure 5.14. sema3fb mutant hearts show no change in proliferation or apoptosis. Embryos were co-immunolabelled for cardiomyocyte markers (DM-GRASP or MF20) and mitotically active cells using phosphohistone H3 (pHH3) (A-B) or apoptotic cells by TUNEL (C). (A)

Quantification of actively mitotic cells are not significantly different between wildtype (N=3, n=6) and mutant (N=3, n=5) embryos (p=0.29) at 24 hpf, or between wildtype (N=3, n=16) and mutant

(N=3, n=27) embryos (p=0.71) at 48 hpf. (C) The average number of apoptotic cells is not significantly different between wildtype (N=2, n=8) and mutant (N=2, n=8) embryos (p=0.99) at

48 hpf. Error bars represent standard error of the mean (SEM). Statistics represent the non- parametric Mann-Whitney U test.

180

(Sato and Yost, 2003). To determine whether the smaller mutant heart size was due to a failure of cardiomyocyte integration from these sources, I analyzed embryos by ISH at 24 hpf and 48 hpf, using markers of the second heart field (latent tgfβ binding protein 3a (ltbp3a)) (Guner-Ataman et al., 2013; Zhou et al., 2011) and CNC (cysteine-rich intestinal protein 2 (crip2) (Sun et al., 2008).

At 24 hpf, comparable crip2 CNC ISH label was present at the arterial pole of the ventricle in wildtype and mutant embryos (Figure 5.15A-B). Of note, the differences in the appearance of the arterial aggregation of crip2 positive CNC at 24 hpf between mutant and wildtype embryos simply reflects the morphogenetic state of the mutant heart (cone vs. tube), rather than a defect in CNC integration into the heart. Indeed, at 48 hpf, where the hearts of wildtype and mutant embryos are grossly comparable in morphology, CNC-derived (Figure 5.15E-F) and second heart field (Figure

5.15C-D) derived cardiomyocytes were observed to have integrated into both the control and mutant hearts. Taken together, these data argue that the smaller stature of the sema3fb-/- heart is not due to changes in proliferation and apoptosis, or the contribution of cardiomyocytes known to elongate the tube.

5.2.3.3 Cardiomyocytes are significantly smaller in sema3fb mutants

Interestingly, the sizes of ventricular cardiomyocytes were strikingly different between the wildtype and sema3fb Δ19 mutant embryos when immunolabelled by DM-GRASP (Figure 5.16A-

B). To quantitate this change, I took confocal projections of DM-GRASP labelled ventricles, and measured circularity, area and perimeter of the cardiomyocytes. While circularity was not significantly different in mutant as compared to wildtype controls (p=0.66, Figure 5.16C), both the area (p=0.016, Figure 5.16D) and perimeter (p=0.03, Figure 5.16E) of ventricular cardiomyocytes were significantly reduced in the mutant embryos by at least 30% and 10%, respectively. Considering the 3D nature of a cell, I next considered whether a smaller

181

Figure 5.15. Cardiac neural crest cell and second heart field cardiomyocyte contributions are unaffected in sema3fb mutants. Embryos were processed by wholemount in situ hybridization to visualize the second heart field with ltbp3a, and the cardiac neural crest with crip2.

Ventral views of 24 hpf embryos show no difference in crip2 expression between wildtype (A,

N=2, n=10/10) and sema3fb mutants (B, N=2, n=10/10). At 48 hpf, cells expressing both markers have integrated into the heart and no difference is observed in the presence of expression for ltbp3a

(C-D, N=1, n=7 wildtype and N=1, n=7 mutant) or crip2 (E-F, N=1, n=11 wildtype and N=1, n=11 mutant). A: anterior; P: posterior. Scale bar: 50 µm.

182

Figure 5.16. Ventricular cardiomyocytes are smaller in sema3fb mutants. Representative confocal projections of 48 hpf wildtype (A) and sema3fb Δ19 (B) ventricles immunostained with an antibody against DM-GRASP to label cellular boundaries. Measurements were made using

N=3, n=5 wildtype and N=3, n=6 mutant embryos. (C) Quantification of cardiomyocyte circularity reveals that the shape of cardiomyocytes is not significantly different between wildtype and mutant cells (p=0.66). (D) The area of ventricular cardiomyocytes is significantly (p=0.016) reduced in sema3fb mutants as compared to wildtype. Quantification of cardiomyocyte perimeter (E) reveals a similar significant reduction (p=0.030) for mutant cardiomyocytes as compared to wildtype.

Error bars represent standard error of the mean (SEM). Statistics represent the non-parametric

Mann-Whitney U test. Scale bar: 5 µm.

183

cardiomyocyte would impact ventricular myocardium thickness by staining histological sections of 72 hpf wildtype and mutant hearts (Figure 5.17A-B) with hematoxylin and eosin. I found a significant 20% reduction (p=0.024) in the thickness of the ventricular myocardium (Figure

5.17C). In summary, these data argue that the small sema3fb-/- heart phenotype is due to a failure of cardiomyocytes to enlarge.

5.2.4 sema3fb mutants have impaired cardiac function

5.2.4.1 Cardiac output is significantly reduced in mutants

Hemodynamic forces, such as those associated with blood flow, are suggested to contribute to cardiomyocyte size. In zebrafish, cardiomyocytes begin to enlarge during chamber emergence

(36 hpf), and more than half of ventricular myocytes have increased in size between 40 – 45 hpf.

Physiological hypertrophy of cardiomyocytes cells is largely dependent on normal levels of blood flow (Auman et al., 2007; Lin et al., 2012). To determine whether the defects in sema3fb mutants impact cardiac function, hearts of 72 hpf embryos were imaged by live video microscopy, and measurements of the ventricle made during end-diastole and end-systole (Figure 5.18A). Cardiac function was assessed through three independent experiments comprising a total of 12 wildtype and 20 sema3fb Δ19 embryos. As expected by the ventricle measurements at 72 hpf, ventricle volume was reduced significantly during diastole (p=0.0027) and systole (p=0.0063) in sema3fb mutants as compared to wildtype controls (Figure 5.18B-C). Additionally, mutants had a significantly slower heart rate (p=0.0008) (Figure 5.18D). Cardiac output and stroke volumes

(Figure 5.18E,G) were also impaired significantly in mutants as compared to control (p=0.026 and p=0.033, respectively), while ejection fraction and fractional shortening (Figure 5.18F,H) were unchanged (p=0.45 and p=0.17, respectively). Overall, the embryonic heart of the sema3fb mutant was impaired significantly in function.

184

Figure 5.17. Ventricular wall thickness is reduced in sema3fb mutants. Histological sections stained by hematoxylin and eosin were used to visualize the ventricle walls (arrowhead) of wildtype (A) and sema3fb Δ19 embryos (B) at 72 hpf. (C) Quantification of the width of the ventricular wall reveals that mutant ventricle walls are significantly thinner (*p=0.024) than those of wildtype hearts. Error bars represent standard error of the mean (SEM). Statistics represent the non-parametric Mann-Whitney U test. Scale bar: 50 µm.

185

Figure 5.18. sema3fb mutant embryos have reduced cardiac function. (A) Still image from a sample video at 72 hpf. Ventricle measurements were made as indicated, in both the long axis (red line, L: length) and short axis (black line, W: width) to determine ventricle volumes. ba: bulbus arteriosus. Quantification of the average ventricle volume at end diastole (B) and end systole (C) reveals a decreased capacity in the sema3fb mutant (**p=0.0027 and **p=0.0063, respectively).

(D) Measurements of heart rate (***p=0.0008), (E) cardiac output (*p=0.026), (F) ejection fraction

(p=0.45), (G) stroke volume (*p=0.033) and (H) fractional shortening (p=0.17). Error bars are

186

standard error of the mean (SEM) over 3 independent experiments (n=12 wildtype, n=20 sema3fb

Δ19). Statistics represent the non-parametric Mann-Whitney U test. Scale bar: 100 µm.

187

5.2.4.2 Valve development is apparently normal in mutants

The hemodynamic forces present within the developing heart are proposed to be necessary for the normal formation of the atrioventricular valve (Hove et al., 2003). Conversely, defects in valve formation impair cardiac function due to a retrograde flow of blood from ventricle to atrium

(Scherz et al., 2008). Given the expansion of bmp4a AV marker expression, I asked whether altered valve development might account for defects in cardiac function. I found the presence of an atrioventricular cushion in hematoxylin and eosin stained hearts of 72 hpf wildtype (N=1, n=3) and mutant (N=1, n=6) embryos (Figure 5.19A-B). Additionally, I took advantage of the sema3fb

+/Δ19 Tg(kdrl:mCherry) fish line I generated, in which all endothelial, and therefore endocardial cells, are labelled by mCherry. I incrossed adult sema3fb Δ19 heterozygous transgenic fish, and genotyped embryos post analysis. The embryos were immunolabelled with MF20 to mark the myocardium. Using confocal maximum projections, I identified the presence of an atrioventricular cushion (comprised of both myocardium and endocardium) in wildtype (N=1, n=5) and mutant

(N=1, n=6) embryos (Figure 5.19C-D). Thus, the cardiac functional deficits in sema3fb mutants are unlikely to be due to an absent atrioventricular valve.

5.2.5 sema3fb signalling is tissue autonomous

5.2.5.1 Atrial chamber size deficit is independent of flow

As discussed earlier, cardiomyocyte size, and therefore chamber size, are intimately linked to the hemodynamic forces of flow within the heart (Auman et al., 2007; Lin et al., 2012).

Therefore, small heart chambers (such as the atrium) could arise from a deficit in the shear forces that contribute to chamber enlargement. To address this possibility, I cultured heart explants independent of the circulatory system, and asked whether chamber size was still decreased in mutants as compared to wildtypes. I explanted hearts of wildtype and mutant 24 hpf embryos, and

188

Figure 5.19. Atrial-ventricular valves develop in the sema3fb mutant. Histological sections stained by hematoxylin and eosin reveal the presence of endocardial cushions (black arrowheads) in sema3fb wildtype (A, N=1, n=3) and sema3fb Δ19 hearts (B, N=1, n=3) at 72 hpf. Using the

Tg(kdrl:mCherry) background, endocardial cells (EC, purple) and myocardial cells (MF20, grey) are shown to make up the cushion (outline) in wildtype (C, N=1, n=5) and mutant (D, N=1, n=6) hearts at 72hpf. Scale bar in A-B: 50 µm. Scale bar in C-D: 20 µm.

189

cultured them for 24 hours. Explants were fixed at 48 hpf and assessed by ISH for the expression

of atrial myosin, myh6 (Figure 5.20A-B). The myh6+ mutant atrium (N=2, n=12) was decreased

significantly (p=0.025) in overall size as compared to wildtype (N=2, n=8) by approximately 25%

(Figure 5.20C). This finding suggests that the chamber defects present in the Sema3fb mutants

occur independently of hemodynamic forces.

5.2.5.2 Nrp2b and Plxna3 are potential receptors mediating Sema3fb signalling

Sema3s signal canonically through Nrp and Plxn receptors (Fujisawa, 2004). In sema3fb

mutants, the chamber specific marker analysis indicates a more significant disruption of the

ventricles than atria, suggesting that there is a spatial component to the phenotype. sema3fb is

expressed throughout the ventricle and atrial myocardium (Figure 5.1), however, so I next asked

whether Sema3fb signalling could be localized predominantly to ventricular cardiomyocytes

through spatially localized receptor expression. In a screen of various plxna and nrp receptor genes,

I found that nrp2b and plxna3 were expressed in the bilateral heart field (Figure 5.21A-C) and later

confined to the ventricular myocardium at 48 hpf (Figure 5.21D-F). No other plxna or nrp were

expressed in the heart by my analysis. Thus, while sema3fb is expressed throughout the developing

heart, receptor expression suggests a signalling role restricted mainly to the ventricular

myocardium in a likely cell autonomous fashion.

5.1 Specific Discussion

My data support a role for Sema3fb secreted from cardiomyocytes in modulating chamber- specific gene expression in the developing cardiovascular system. I find sema3fb mRNA is expressed by cardiomyocytes throughout cardiogenesis – from cardiomyocyte progenitor specification and into chamber morphogenesis. With a genetic mutant of sema3fb, I show that a loss of Sema3fb signalling produces a number of specific heart defects. First, cardiac tube

190

Figure 5.20. Atrial size deficit in sema3fb mutants occurs independently of decreased cardiac output. Dissected hearts were explanted from 24-48 hpf and processed by wholemount in situ hybridization for the atrial cardiomyocyte marker myh6. The atria of mutant hearts (N=2, n=11/13) are observationally smaller in size than wildtype atria (N=2, n=8/8). (C) Quantification of atrial area shows a significant reduction in size in mutants (p=0.025) as compared to wildtype. Error bars represent standard error of the mean (SEM). Statistics represent the non-parametric Mann-

Whitney U test. Scale bar: 50 µm.

191

Figure 5.21. Sema3fb likely signals through Nrp2b and Plxna3. Wholemount in situ hybridization with sema3fb (A,D) nrp2b (B,E) and plxna3 (C,F) riboprobes. Dorsal images at 18 hpf (A-C) reveal overlapping expression domains (arrowhead) between the sema3fb ligand (A) and the potential receptors (B-C) in the region of myocardial cells. By 48 hf (D-F), ventral photomicrographs reveal that while sema3fb is expressed throughout the heart (D), and nrp2b (E) and plxna3 within the ventricle (F). Scale bar: 50 µm.

192

elongation is specifically delayed. Second, once chambers emerge both the atria and the ventricles are significantly smaller in size, a feature that is maintained through development and is likely attributable to a decrease in cardiomyocyte size. I then go on to show that while the two heart chambers are present, their transcriptional profiles are disrupted in a manner that suggests a failure to spatially restrict differentiation. Finally, I demonstrate that the defects in cardiac development result in impaired cardiac function. My work implicates for the first time a heart specific Sema that is necessary for regional differentiation of cardiomyocytes (Figure 5.22).

The most striking phenotype in the sema3fb mutants was the loss of chamber restricted

mRNA expression of myh7 and myh6 myosins (Figure 5.8). In mutants, expression of ventricle

myh7 spills over across the AVC into the atrium to some degree, however, more prominent is the

expansion of myh6 from the atrium into the whole ventricle. Based on these data I propose a model

whereby Sema3fb signalling is necessary for the full expression establishment of a ventricle fate,

in part through the concurrent inhibition of atrial fate (Figure 5.23). In support, I find irx1a

expression is severely downregulated in the sema3fb-/- ventricle, a transcription factor important

in chick for supressing expression of atrial myosin heavy chain I and driving expression of

ventricle myosin heavy chain 1 (Bao et al., 1999). Bmp signalling may play a role here, in that

atrial size decreases in the lost-a-fin (laf) Type I BMP receptor alk8 mutant (Marques and Yelon,

2009), suggesting a need for BMP signalling in atrial differentiation. It is possible that in sema3fb

mutants, ectopic bmp4a in the ventricle drives inappropriate atrial-related gene expression in the

ventricle. Interestingly, myh7 expression is not lost in mutant ventricles, suggesting that additional

mechanisms drive ventricle identity in zebrafish. Indeed, the expression of tbx5a is sustained in

the mutant (Figure 5.9C-D), which indicates that not all transcriptional identities are perturbed.

193

Figure 5.22. Summary of sema3fb mutant expression analyses. In wildtype embryos, chamber specific expression is maintained discretely whereby MF20, DM-GRASP, myh7 and irx1a are restricted to the ventricle, myh6 to the atrium, and bmp4 to the atrioventricular valve. In the sema3fb mutant, ventricle markers expand into the atrium, the atrial marker into the ventricle, bmp4 expands into both the atrium and ventricle, and ventricular irx1a is lost.

194

Figure 5.23. Working model of Sema3fb signalling in the heart. In wildtype embryos, Fgf8 induces expression of sema3fb throughout the myocardium, but signalling is restricted to the ventricular cardiomyocytes that express Plxna3. Sema3fb signalling is necessary for the expression of ventricle specific irx1a, and the downstream inhibition of both bmp4a and myh6 within the ventricle. Loss of sema3fb signalling in the ventricle results in a loss of irx1a and no spatial restriction of the expression of bmp4 and atrial specific genes in ventricular cardiomyocytes.

195

Interestingly, the expression of tbx5a does expand somewhat into the atrium. However, what

regulates tbx5a expression is currently unknown.

The fact that in sema3fb mutants the atrial myosin expands into the ventricle to a much greater extent than the converse expansion of ventricle markers into atrium, argues that the cells of the ventricle are not fully expressing a ventricle fate identity. Considering this, I would have only expected to find a phenotype localized to the ventricle. Yet, the atrium is affected in its development, as evidenced by its decreased size, and misexpression of ventricular markers.

Possibly, the atrial phenotype reflects ventricle differentiation and dysfunction consequently impacting atrial size. This idea of reciprocal dysfunction is best illustrated by the zebrafish mutant weak atrium (wea) where myh6 is lost. These mutants lack atrial contractility, but their ventricle myocardium is overly thick – likely in response to an atrial dysfunction (Berdougo et al., 2003).

Conversely, in the sema3fb-/- embryos, partial loss of ventricle identity (loss of irx1a) or expansion of myocardial signalling (bmp4a) may perturb atrial development. Thus, it is possible that a functional mechanism explains why the atrium is impacted in size. Dysfunction, however, must arise secondary to disruptions in loss of chamber specific gene expression, which is most prominent for the ventricle.

The chamber marker analysis showing dramatic changes in gene expression within the ventricles of sema3fb-/- hearts, raises the possibility that Sema3fb signalling occurs mainly in ventricular cardiomyocytes, even though sema3fb is expressed throughout the myocardium.

Interestingly, spatial restriction of Sema3fb signalling may arise through ventricular specific expression of its receptors. I identified nrp2b and plxna3 as the only nrp and plxna obviously expressed within the ventricle. In the future, this could be tested. But how does a Sema3 influence differentiation? There is precedence for non-canonical roles of the secreted chemorepellents, and

196

my work reinforces this body of literature. For instance, Sema3f negatively regulates expression of the transcription factor Ascl2 within colorectal cancer cells in vitro (Zhou et al., 2015), and

Sema3a drives Wnt/β-catenin signalling in osteoblast differentiation in mouse (Hayashi et al.,

2012).

In addition to the changes in chamber specific gene expression, the ventricles and atria of the sema3fb mutants are smaller than their wildtype controls. While the small heart phenotype of the sema3fb mutant could arise from a failure to integrate cardiomyocytes from the second heart field progenitors (Guner-Ataman et al., 2013) or CNC (Cavanaugh et al., 2015), analysis of markers of these two populations at 24 and 48 hpf suggest that these two populations are unaffected

(Figure 5.15). Nor do my data support a model whereby the small hearts are caused by increased cardiomyocyte/progenitor cell death and/or failure to proliferate, in that these measures were unaltered in the mutant fish as compared to their wildtype counterparts (Figure 5.14). Instead, my data support the idea that decreased cardiomyocyte size is a contributor to the smaller heart size

(Figure 5.16). Since these measurements were only generated for cardiomyocytes of the ventricle, it remains a question as to whether atrial cardiomyocyte size is also changed by the loss of

Sema3fb. Atrial cardiomyocte area measurements were not made in the absence of a marker to faithfully label atrial cardiomyocyte boundaries. Possibly ventricular cardiomyocytes that fail to differentiate properly, because of misexpression of key genes, result in a dysfunctional heart with disrupted shear stresses and the cardiomyocytes fail to enlarge as a result (Auman et al., 2007). As discussed above, size change in the atrium could arise as a result of ventricular dysfunction.

Certainly, cardiac function is significantly impaired in the mutants, in almost all assays performed (Figure 5.18). Given the chamber size analysis, it is not surprising that ventricle volumes, and as such, stroke volumes, are significantly smaller. Of specific interest are heart rate

197

and cardiac output, the latter being dependent on heart rate and total stroke volume. Heart rates in the mutants were significantly slower, suggesting they have bradycardia. This slowing could be due to several factors. Heart rate is established by pacemaker cells of the sinoatrial node which produce rhythmic action potentials allowing for rhythmic heart beating (Larsson, 2010).

Sympathetic innervation of these pacemaker cells allows for modulation of heart rate. In mouse,

Sema3a is expressed in the Purkinje fibers of the cardiac conduction system, and loss of Sema3a results in a failure to form a patterned sympathetic innervation gradient and results in bradycardic mice (Ieda et al., 2007). While it is possible that the conduction system in zebrafish is patterned by the molecularly similar Sema3fb, my focus has been on stages where sympathetic innervation has not yet been properly established and thus the phenotypes seen here can be considered independent of any innervation defects (Poon et al., 2016). Instead, it is more likely that the impaired differentiation of the cardiomyocytes would manifest in electrophysiological dysfunctions of the cardiomyocytes, which would consequently negatively influence heart rate.

Given its expression in the bilateral heart fields, Sema3fb could have early effects in heart development, when cardiomyocyte progenitors are first specified (Reifers et al., 2000) and migrate to produce a single heart tube. Indeed, there is a slight delay in tube morphogenesis from a cone structure to an elongated tube at 24 hpf in the sema3fb mutants. These data might suggest that

Sema3fb is important for cellular migration during the morphogenetic events leading to heart tube formation, however, no instances of cardia bifida, otherwise known as a failure of the bilateral heart field to fuse, were noted. Cardiac fusion requires a number of events to occur normally, such as formation of a normal endoderm (Reiter et al., 1999), midline signalling to drive myocardial migration (Kupperman et al., 2000), normal epithelial polarity (Yelon et al., 1999), and myocardial differentiation (Yelon et al., 1999). Interestingly, problems with myocardial differentiation do not

198

generally seem to result in fusion defects, but rather delays in tube formation, suggesting that cardiomyocytes need to be at a certain point in their differentiation program to be competent and express the genes necessary for the morphogenesis process to occur. It is possible that the defects in ventricular and atrial chamber patterning I defined at 48 hpf in the sema3fb mutants are already present in cardiomyocytes before the heart enters the linear tube state. Unfortunately, there are no additionally identified markers of the presumptive chambers at 24 hpf, with the exception of the two myosins used here.

Preliminary analysis also suggests that despite sema3fb expression in the early embryonic zebrafish heart, specification is unperturbed in the mutants, though it is possible that transcriptional pathways that were not assessed are affected. Nonetheless, I find that fgf8a, a gene encoding a growth factor necessary for cardiomyocyte induction and specification (Reifers et al., 2000), is similarly expressed in wildtype and mutant embryos (Figure 5.12A-F), suggesting that at least this key cue for induction is present at the right time and place. I also looked at tbx5a (Figure 5.12G-

H), that produces Holt-Oram Syndrome when mutated in humans – a syndrome characterized by heart septal defects. Tbx5 is expressed throughout the heart field, with Tbx5 loss in both Xenopus

(Horb and Thomsen, 1999) and zebrafish (Garrity et al., 2002) resulting in hearts with gross morphological abnormalities. Tbx5 homozygous null mice are E10.5 lethal and present with hypoplastic atria (Bruneau et al., 2001). Considering the chamber specificity of the murine phenotype, it is possible that in zebrafish, tbx5a represents a subpopulation of cardiomyocytes whom are spared in the mutant. Indeed, expression of tbx5a in the sema3fb mutant ventricle is unchanged at 48 hpf (Figure 5.9C-D), despite a loss of ventricle specific irx1a (Figure 5.9A-B).

These data either suggest subpopulations of cardiomyocytes exist or that the transcriptional profile of the cardiomyocytes has been altered in the sema3fb mutants, but not in such a way that they

199

cannot express ‘normal’ chamber restricted genes. This is certainly true of ventricular myh7, which is downstream in the ventricle differentiation pathway. Of note, given that the expression of fgf8a and tbx5a at the 13ss between wildtype and sema3fb -/- embryos is relatively comparable (Figure

5.12), it seems unlikely that a smaller primary heart field accounts for the small hearts seen in the mutants at 48 hpf (of note, heart size could not be assessed at 24 hpf, because of the differences in heart morphology between wildtype and sema3fb -/- mutant embryos).

Extrinsic signalling molecules are known to modulate heart development, and often these are morphogens. For instance, Fgf8a is necessary for heart development, and its signalling is thought to be tittered by an opposing gradient of retinoic acid (Sorrell and Waxman, 2011).

Interestingly, in mouse Sema3f expression in the midbrain is regulated by the expression of FGF8 from the midbrain-hindbrain boundary (Yamauchi et al., 2009). Thus, it is possible that a similar regulatory pathway controls sema3fb expression in the heart: fgf8a expression in the zebrafish heart induces expression of sema3fb, and Sema3fb exerts its downstream effects on cardiomyocyte differentiation. This model is testable by using a pharmacological inhibitor of Fgf signalling,

SU54012, applied between 20-26 hpf, and 36-48 hpf, followed by ISH for sema3fb. The working model I have proposed is modular and testable (Figure 5.23). Whether Fgf8a is upstream of

Sema3fb can be readily assessed.

5.2 Significance

In this chapter, I provide evidence for secreted Sema functioning as an extrinsic regulator of regional differentiation in the cardiovascular system. While Semas are known to be important for development of the cardiovascular system (Epstein et al., 2015), with roles in vascular and neural crest cell patterning, here I report a novel cardiomyocyte autonomous and non-canonical function of Sema3fb in governing the molecular differentiation of the heart. As with the data I

200

present in Chapter 3, the roles of Sema3 in progenitor cell dynamics have been historically overlooked and underappreciated. Interestingly, while multiple Sema3s are expressed in the cardiovascular system of mice and chick (Epstein et al., 2015), my own ISH screen indicate

Sema3fb is the sole Sema3 expressed by cells of the developing zebrafish heart. Thus, it is likely that in the mammalian heart other Sema3s compensate functionally for the loss of a specific

Sema3. I was fortunate in the zebrafish model to not have such explicit redundancy and was therefore able to take advantage of the mutant and characterize the biological function of this secreted molecule.

201

Chapter Six: General discussion

In this thesis I explore the spatial roles of secreted Sema3f signalling in different organ systems. In Chapter 3, I present the first report of a Sema3 involved in neurogenesis of the retina.

I characterize the expression of sema3fa across development and find persistent expression within

RPCs, suggesting an uncharacterized role in progenitor biology. I present that early in development

Sema3fa signalling in the temporal retina is necessary for the genesis of amacrine cells in only this retinal domain, and that subsequently Sema3fa within the CMZ regulates the spatial organization of the CMZ transcriptome. In Chapter 4, I provide evidence for the endogenous role of Sema3fa in maintaining avascularity of the neural retina. I demonstrate that spatially localized Sema3fa signalling is necessary in both intra- and extra-ocular vessel development, and, that in the absence of Sema3fa, vessels are overgrown and dense, or infiltrate the neural retina, respectively.

Additionally, I present that vessels entering the neural retina are leaky – reminiscent of what occurs with CNV during AMD. Finally, in Chapter 5 I demonstrate that Sema3fb signalling is spatially restricted in the developing heart and is necessary for normal ventricle development. To my knowledge, this is the first report of direct involvement of a Sema3 in primary heart development, not related to an influence on vascular and/or neural crest development. Together, I present a novel role of Sema3f to provide spatial information not only as a prototypical guidance cue, as appears to be the case for blood vessels, but as a regulator of gene expression of progenitor cells (RPC and heart) to influence their cellular differentiation.

In this chapter, I discuss my findings, and future experiments to answer questions which stem from my work.

202

6.1 Sema3fa signalling primes temporal RPCs for future amacrinogenesis

My data support the idea that Sema3fa provides a signal to retinal progenitors that remains localized so as to control cell genesis in the temporal neural retina. sema3fa mRNA is initially expressed by progenitors within the temporal retina until 36 hpf, but subsequently shows no nasotemporal differences in expression: sema3fa mRNA is expressed throughout the RGC layer at 52 hpf, and in the INL and CMZ at 72 hpf. Interestingly, however, my analysis of 72 hpf sema3fa mutant retinas - when retinogenesis is largely complete - identified cellular deficits in only amacrine cells of the temporal, and not, nasal retina. These data argue strongly that the effects on temporal retina amacrine cell numbers in the mutants result from a loss of Sema3fa signalling in the early optic cup, when the signals themselves would have been restricted mostly to temporal neural retina. This idea is intriguing as it suggests temporal RPCs are primed for cell genesis early, and that when the Sema3fa priming signal is lost, the RPCs are unable to generate enough amacrine cells. Since cellular apoptosis is unchanged in sema3fa mutant retinas, a reduction in the proliferation of RPCs is likely involved. To confirm this model, I will need to assess the mitotic profile and/or cell cycle length in my mutants.

If proliferative deficits of temporal RPCs are the main driver behind the spatial phenotype with Sema3fa loss, however, I would have expected to see a reduction in other cell types within the temporal eye. In fact, I found RGC numbers did not differ between the nasal and temporal retina in either genotype, nor did the numbers of cells within the outer portion of the INL (bipolar and horizontal cells). These data suggest that only a subset of temporal RPCs require, or are competent to respond to, Sema3fa signalling for cell type specific differentiation (to an amacrine cell), while others work independently of Sema3fa. Traditional models of RPCs suggest that progenitors lose competence during development – whereby cells are generated in a sequential and

203

temporal manner (Cepko, 2014). A recent review suggests instead that progenitor cells are transitional in nature, existing both as multipotent progenitors capable of giving rise to all retinal neurons, and more lineage-restricted progenitors, with both populations present within the developing eye (Jin, 2017). My work supports the transitional RPC model, as only amacrine cells are affected negatively with Sema3fa loss, arguing that the RPCs must be heterogeneous, with regards both their intrinsic properties and competence to respond to extrinsic cues, in their ability to give rise to specific cell types. In support of the idea of heterogeneous RPCs, a subset of Olig2+

RPCs present in the early mouse retina undergo neurogenic divisions to generate horizontal cells and cones, exclusively (Hafler et al., 2012). This is reminiscent of ganglion mother cells in the ventral nerve cord of Drosophila melanogaster, which also are heterogeneous and exhibit different potencies (Homem and Knoblich, 2012). In this system, neurons and glia are generated by proliferative neuroblasts that undergo asymmetric divisions to produce a neuroblast daughter and a lineage restricted ganglion mother cell. The ganglion mother cell divides only once to produce two differentiated neurons whose fates are determined by factors expressed in the prior cell division. My data suggest that an RPC subtype is present in the temporal retina that expresses a specific transcriptome that makes it competent to respond to Sema3fa signalling by activating downstream effectors that dictate the genesis of amacrine cell types over others.

This model raises an interesting question – which intracellular pathway(s) confers amacrine cell identity downstream of Sema3fa signalling? Interestingly, SEMA3F inhibits metastasis of cultured human colorectal cancer cells through a PI3K-AKT dependent downregulation of the

ASCL2-CXCR4 axis (Zhou et al., 2015). Specifically, downstream of SEMA3F, PI3K signalling phosphorylates the transcription factor ASCL2 to prevent it from inducing expression of the pro- migratory and metastatic receptor, CXCR4. This was the first report of a link between SEMA3F

204

and transcriptional regulation. The RNAseq data I acquired for 36 hpf sema3fa mutant and wildtype embryos reveal a number of differentially regulated genes related to amacrine cell differentiation (Table 6.1). Interestingly, these genes are all upregulated, including amacrine specification factor neurod4 (Inoue et al., 2002) that I showed in Chapter 3 is absent largely from the retinas of mutants at the same developmental stage. The disagreement between the whole embryo RNAseq and in situ hybridization data is not surprising, as all the retina-expressed genes, except for foxn4, are also present in other CNS tissue. Likely, neurod4 was upregulated in another tissue of the mutant embryo, as its retina presented with clear downregulation of this gene by in situ hybridization. To validate the RNAseq data for the eye specifically, I will need to analyze by

RT-qPCR the cDNA of dissected eyes. The possibility remains, however, that certain amacrine cell-related genes are upregulated in the sema3fa mutant retina. With an increase in transcripts of genes known to drive amacrine cell genesis, it would be logical to assume an increase in amacrinogenesis with Sema3fa loss – although this is not the case. The upregulation of amacrine genes may indicate an inadequate compensatory mechanism to push more RPCs towards an amacrine fate.

While a significant decrease in the number of amacrine cells was observed in the temporal retina, the retinas were still capable of making amacrine cells. These data raise the possibility that a specific amacrine cell subtype(s) is lost. While amacrine cells make up only 8% of cells of the zebrafish retina (Jeon et al., 1998), there are at least 28 different subtypes of amacrine cells, identifiable morphologically and by molecular markers (Cederlund et al., 2011). It will be interesting to determine if an amacrine cell subtype is affected preferentially by Sema3fa loss, as it will aid our understanding of the molecular mechanism behind Sema3fa signalling. The literature suggests that subtype fate choice of amacrine cells is progressive, whereby specification to the

205

Table 6.1. Differentially expressed genes in mutant Sema3fa versus wildtype embryos.

Gene Name Association Fold Change q-value c3b.2 complement immune system +1.52 7.9E-03 cfh complement immune system +1.56 9.44E-05 crygmx1 γ crystallin +14.32 1.69E-08 crybb1l3 β crystallin +3.53 5.55E-06 foxn4 amacrine cell development +1.48 1.05E-03 ism2b complement immune system +13.85 4.03E-02 neurod4 amacrine cell development +1.52 1.60E-05 pou3f1 amacrine cell development +1.51 3.24E-05 prom1b amacrine cell development +2.43 3.75E-06 sema3fa -2.95 5.46E-06 tfap2d amacrine cell development +2.48 1.13E-05

206

amacrine lineage occurs first, followed by molecular transitions towards specific subtypes with unique expression of neurotransmitter-related genes (Goodson et al., 2018; Kay et al., 2011).

Amacrine cell subtypes in the murine retina can be grouped as follows: 43% are Glutamate

Decarboxylase (GAD67+) and GABAergic, 43% are Glycine Transporter 1 (GlyT1+) and glycinergic, with the remaining 15% of cells expressing Neurod6, but neither neurotransmitter.

My analysis used general markers of amacrine cells, and therefore future work, with for example analysis of GAD67, GlyT1 and Neurod6 expression, could ask if a specific amacrine cell subtype is lost in sema3fa mutant retinas.

My data suggest that the microenvironment of the temporal retina is inherently different than that of the nasal retina, and Sema3fa is a player in regulating a subset of temporal RPCs fated to the amacrine lineage. RGC differentiation in the ventro-nasal retina is induced by Shh secreted from the optic stalk, and moves in a clockwise manner towards the temporal retina (Masai et al.,

2000). Yet, progenitors in both the nasal and temporal retina would be expected to have equal access to Shh from the optic stalk. Instead, temporal RPCs are the last to respond to Shh. This observation argues for a unique extrinsic environment in the temporal retina that delays or decreases competence to Shh signalling at the time of the ventro-nasal neurogenic wave. To further investigate these domain-specific microenvironments, I could generate a foxg1a (nasal) and foxd1

(temporal) fluorescent reporter line on the sema3fa background. Fluorescently-labelled cells from wildtype and mutant retinas could then be FACS analyzed, and sent for single cell RNAseq. In this way, individual temporal and/or nasal RPCs could be assessed, with the single cell RNAseq transcriptome data identifying subgroups of RPCs. I attempted to generate a foxd1 transgenic, with the assistance of Jonathan Yang, but to no avail. Likely foxd1 is regulated by an enhancer region that we were unable to identify. With RNAseq data from nasal and temporal RPCs of wildtype and

207

mutant backgrounds, we could further assess downstream transcriptome changes. Such analysis would also explore for the first time the distinct transcriptomes of the temporal vs. nasal retina in a wildtype background to really begin to ask what other factors, in addition to Sema3fa, regulate the spatial environment.

6.2 Sema3fa signalling as a novel regulator of the stem cell niche

Interestingly, sema3fa is also expressed within the CMZ (72 hpf) – a zone of the peripheral retina which generates neurons throughout the lifetime of the fish (Kubota et al., 2002). As discussed above, my data argue that the decrease in temporal amacrine cells of sema3fa mutant retinas is not likely to result from a CMZ defect, in that expression is not spatially restricted to the temporal CMZ, rather sema3fa mRNA is present throughout the entire 72 hpf CMZ. If CMZ related, one would have expected both temporal and nasal amacrine cell populations to be impacted in the mutants. A change in the sema3fa mutants potentially more related to CMZ function is suggested by the ectopic expression of the cellular adhesion molecular DM-GRASP (Laessing and

Stuermer, 1996) in the photoreceptor layer of mutant retinas (Figure 6.1) that is not restricted spatially to temporal vs. nasal retina.

What could be going on here? The CMZ is spatially organized with respect to progenitor cell transcriptomes that relates to their respective potencies, with multipotent progenitors sitting peripherally and unipotent progenitors centrally (Cerveny et al., 2010) (Figure 1.3). The expansion of both ccnd1 and atoh7 within the CMZ of sema3fa mutants argues that this organization is disrupted, and could affect the ongoing normal genesis of neurons.

Several extrinsic factors are known to modulate the CMZ, and include Wnts, Notch and

Shh (Raymond et al., 2006). Together these signalling pathways regulate progenitor proliferation, cell cycle progression and neurogenesis. In my mutants, however, the proliferative potential of

208

Figure 6.1. Ectopic expression of DM-GRASP in sema3fa mutant retinas. A)

Immunohistochemistry of 72 hpf retinal sections using the Zn8 antibody detects DM-GRASP in the retinal ganglion cells (and optic nerve) of the retina in wildtype. B-C) Mutant retina present with ectopic immunolabel with the Zn8 antibody in the outer nuclear layer, where normally photoreceptors reside. inl: inner nuclear layer; ipl: inner plexiform layer; l: lens; N: nasal; onl: outer nuclear layer; opl: outer plexiform layer; rgcl: retinal ganglion cell layer; T: temporal. Scale bar: 50 µm and 10 µm for inset (A’-C’).

209

CMZ progenitors does not appear to be affected grossly as ccnd1 mRNA and PCNA are both still present, although expanded in their domain of expression, and the cell cycle exit marker cdkn1c is still expressed normally. Indeed, the most striking CMZ phenotype of sema3fa mutants is that all progenitors express the differentiating factor atoh7, whose expression is normally restricted to the more central domain of specified progenitors. atoh7, while necessary for RGC fate, is expressed by all medaka RPCs prior to their terminal division (Sinn et al., 2014). Thus, with Sema3fa loss it appears that cells within the more peripheral, immature CMZ progenitor domain express atoh7 prematurely, and so may be poised to undergo a terminal division before they are “ready”. As such, the differentiation potential of progenitors could be altered. This paradigm suggests a role for

Sema3fa signalling in restricting the competence of neural progenitors. I propose that normally an unknown extrinsic factor induces expression of atoh7 in RPCs located more centrally in the CMZ, and when Sema3fa signalling is attenuated, like in the mutant, this factor can act on more peripheral multipotent RPCs and cause them to ectopically express atoh7. Extrinsic regulation in this manner is reminiscent of neuroectoderm formation during gastrulation (Bond et al., 2012).

Here, BMP secreted from ectodermal cells induces epidermal fates, with secretion of BMP antagonists, such as Noggin, necessary for the formation of neuroectoderm. When BMP antagonists are lost, spatial differentiation of the neuroectoderm is lost. Exploring the expression of extrinsic factors and their downstream targets within the CMZ of wildtype vs. sema3fa mutants will aid in our understanding of the role of Sema3fa in the CMZ. Of note, the CMZ helps repopulate all layers of the retina following injury, except for the rod photoreceptor lineage which is derived from dedifferentiated Müller glia (Stenkamp, 2011). Thus, two big questions related to

Sema3fa in the CMZ are – what is its role in retinal growth, both ongoing (72 hpf +) and during injury and repair?

210

To address how Sema3fa in the CMZ affects cell genesis in the mature retina I would follow newly generated cells from the CMZ. To do so, I would Edu pulse treat embryos at 72 hpf, and detect by immunohistochemistry where the Edu positive nuclei reside at least 24 hours later.

This would give insight into the developmental capacity of the CMZ under naïve conditions and could be repeated over development, and in sema3fa mutants. While transcriptional changes are clearly evident in the CMZ of sema3fa mutants, preliminary analysis of the retinas indicate that

CMZ disruption does not result in striking deficits in the generation of retinal cell types, retinal lamination, or vision. However, the neural retina contains 6 cell types, and each neural type has multiple members. For example, as mentioned previously, there exists at least 28 subtypes of amacrine cells characterized by their morphologies, stratification and neurotransmitter functionality. Thus, it is possible that with Sema3fa loss, neurons continue to be generated from the CMZ post 72 hpf, but that the transcriptome changes in the RPCs means that genesis of specific cellular subtypes is altered. In the future, we could also address a role for CMZ Sema3fa in regulating the replacement of cells lost due to retinal damage, by damaging the mature wildtype and mutant retinas. To trigger primarily CMZ-mediated regeneration of neurons, we could damage the retina by stab wounding, or inject the neurotoxin oubain (Fimbel et al., 2007). Following injury, wildtype and sema3fa mutant larvae would be pulsed with Edu and assessed at least 24 hours later to observe the amounts and types of cells that are generated in response to neural damage.

While gross visual function is not impaired by Sema3fa loss, my analysis of visual function was likely not sensitive enough to detect changes in circuits due to a relatively small loss in amacrine cells. The startle response is a visual reflex that is lost primarily when light perception is lost (e.g. photoreceptor loss). With a loss of inhibitory amacrine cells, one would predict that in the temporal neural retina the visual circuit from the ONL to the RGCs is less inhibited in terms

211

of signal transmission. My c-fos analysis indicates that cells are communicating. To confirm that circuit formation is grossly normal, I will need to repeat the c-fos experiments with additional time- points (post 72 hpf), and co-label, by immunohistochemistry, different retinal cell types. It is possible that while cells can communicate in the mutant, the specific cell types communicating are changed. Additional vision functional tests could be performed to determine physiological responses to light and vision dependent behaviors. Electroretinograms (ERG) can be used to determine electrical activity generated by zebrafish retinal cells in response to light (Chrispell et al., 2015). Amacrine cells strongly influence RGC receptive fields (Lukasiewicz, 2005), and spatiotemporal processing of visual information. Thus, an optokinetic reflex (OKR) or optomotor response (OMR), which measure visual acuity, could be performed with zebrafish (Cameron et al.,

2013; Portugues et al., 2015). Altogether, multiple visual function assays must be carried out to define how the loss of Sema3fa affects vision and light perception.

6.3 Mining the Sema3fa mutant

The Sema3fa RNAseq I performed at 36 hpf identified a number of genes unrelated to retina or cellular differentiation related processes. Interestingly, a large number of these differentially expressed genes are associated with the optic tectum, cerebellum, or skeletal muscle.

Can the Sema3fa mutant model other diseases? Interestingly, I observed that adult mutants develop scoliotic body shapes around 6-8 months of age, and thus Sema3fa may regulate muscle homeostasis. Additionally, RNAseq found a number of developmental genes expressed in the embryonic tectum, but also found changes in synaptic function proteins, such as the homotypic synaptic binding molecules – the protocadherins. What could these changes indicate? Sema3f mouse knockouts exhibit miswiring within a number of brain regions such as the olfactory, limbic and cortico-thalamic systems (Matsuda et al., 2016; Sahay et al., 2003; Takeuchi et al., 2010).

212

Interestingly, 6-8 month old homozygous mutants exhibit transient bouts of ataxia; characterized by a sequence of poor swimming, laying on the tank floor, and sudden upward jerking movements.

These behaviours are reminiscent of seizure-like behaviour described in fish (Kalueff et al., 2013), and could suggest that sema3fa zebrafish mutants exhibit brain miswiring, recapitulating the mouse model. A benefit of the zebrafish model, however, is that zebrafish can be used in high- throughput drug screens to identify compounds that limit seizing-like behaviours. To move in this direction, we need to confirm whether the behavioural abnormalities I observed are due to CNS- related issues, or due to the potential skeletal muscle changes that link to scoliosis.

6.4 Sema3fa signalling in CNV

I showed that loss of Sema3fa signalling results in defective eye vascular beds. First, the intraocular vessels are thin and branch more often, suggestive of proliferative angiogenesis.

Additionally, the extraocular choroid develops precociously in the mutants, and can aberrantly invade the neural retina. It is interesting that loss of a single extrinsic factor produces any vascular phenotype, as many developmental processes are controlled by multiple redundant pathways.

Vessel entry and leakage present in mutant fish in a relatively limited fashion, and I do not find evidence of retinal detachment following leakage that would be expected with prolonged edema.

Thus, Sema3fa may not be the only mechanism to ensure an avascular retina. Alternatively, I analyzed primarily the retinas of relatively immature wildtype and mutant embryos, and may have missed a phenotype that exacerbates with time. Indeed, the vessel bed continually grows as zebrafish age, and, as suggested by human data (Campochiaro, 2015), is likely more susceptible to pathogenesis with age. Finally, the retina may have mechanisms to “fix” any vessel pathology, given that zebrafish is a regenerative model. In this regard, it was interesting that experimentally I found that during my initial trial of injecting dextran into adult mutants, “blood spots” appeared

213

minutes later on the surface of the eye, suggesting some sort of leakage and/or hemorrhage.

Following a 24-hour recovery period, however, dextran was not detected within the retina. As such, subsequently I assessed leakage on the same day as the injection, to mitigate potential masking by recovery mechanisms. Of note, with this paradigm I found vessel leakage was present in the adult mutant retinas.

An interesting approach to further reveal roles for Sema3fa in the eye would be to stress the visual system. Humans encounter varying light levels throughout the day, for instance as they move from daylight to indoors. As such, our retinas are forced to adapt to different light settings in order to provide undisrupted vision. This in turns puts strain on the RPE to both renew the shed outer segments of the photoreceptors, and to convert the all-trans-retinol back to 11-cis-retinal to be used for photon detection. Yet, all of my fish were housed on a 14 hour light/ 10 hour dark lighting regimen, with non-fluctuating intensities of light. To challenge the mutant model, we could employ the light box used to induce c-fos expression (described in Chapter 2). In this paradigm, embryos would be exposed to 1, 2, 4, 8, 12 and 24 hours of light prior to analysis for

CNV. Of note, 30 minutes of light exposure is sufficient to drive neural activity as evidenced by c-fos expression. In doing this assay, we assume the RPE has to work harder with longer light exposure to maintain photoreceptor outer segment turnover. This natural strain may be sufficient to drive an increased pathogenic response in the mutant, as assayed by dextran injection or by confocal analysis of kdrl transgenic animals.

The RPE, through ZO-1 tight junctions, prevents leakage from fenestrated capillaries of the choriocapillaris into the neural retina. My work suggests that the RPE additionally acts as a molecular barrier to the choroidal vasculature. In support, I find that with loss of Sema3fa leaky blood vessels invade into the neural retina, despite there being no obvious discontinuities to the

214

RPE. I did not assess ZO-1, however, in the mutant, and therefore it is possible that a loss of ZO-

1 and disruption of the BRB is the driver for vessel invasion. I have correlative evidence to support the idea of a molecular barrier function for Sema3fa of the RPE. The regenerative capacity of the

RPE was studied in vivo by using zebrafish with specific and patchy ablation of the RPE (Hanovice et al., 2018), with adjacent RPE cells proliferating and migrating to the site of injury. One would have also expected that physical loss of the RPE would allow blood vessels to enter and disrupt the nuclear retina. Yet, published histological sections showed no gross disruption of the retinal nuclear layers, suggesting that a molecular barrier, such as one provided by Sema3fa, is produced by the surrounding RPE to maintain the antiangiogenic signal to the overlying choroid. Ideally, establishing the RPE ablation model in the wildtype and mutant sema3fa embryos would allow me to determine whether physical ablation of RPE in a wildtype background is spared from CNV because of the presence of Sema3fa.

Additional analysis must be carried out to determine the full extent of any retinal phenotype presented by the CNV afflicted fish. AMD is characterized by photoreceptor degeneration, retinal detachment and inflammatory processes (Coleman et al., 2008; Kawa et al., 2014; McLeod et al.,

2009; Mitchell et al., 2018). While not analyzed quantitatively, the ONL in sema3fa mutants is more thin than in wildtype fish post 10 dpf, and the outer segments of adult photoreceptors appear less organized. These observations suggest a new project to ask whether photoreceptor dysfunction precedes CNV. Currently the literature is undecided as to the order of events that trigger CNV.

Which comes first? RPE dysfunction, blood vessel abnormalities or photoreceptor degeneration.

To identify photoreceptor abnormalities over the time period that CNV emerges in the mutants, I would immunolabel the inner and outer segments of the photoreceptors, and co-label with apoptotic markers. My preliminary analysis of the RPE suggests no underlying dysfunction, as

215

Zpr2 and visual cycle genes are expressed. The RPE, however, has additional functions that include ion homeostasis, lipid membrane synthesis, and glucose transport. I did not determine whether these functions are normal in the mutants that experience CNV. A side-by-side analysis of the RPE and photoreceptors across development into larval stages in wildtype and mutant fish will allow us to determine which system presents “first” with a pathology, and whether that coincides with CNV, as visualized by blood vessel entry into the eye.

The RNAseq data suggest a possible inflammatory response upregulation in the mutant fish at 36 hpf – well before any CNV was detected. This data poses an interesting question as to why expression of a potentially pathogenic pathway is present during early embryogenesis and warrants further investigation. The complement system is known to be involved in pathological progression of AMD, and is looked at currently as a promising venture to develop therapeutics, as the immune system can potentially be modulated (Kawa et al., 2014). The expression of components of the complement system is induced in sema3fa mutants, which could either drive

CNV or possibly be part of an unrelated process. Increasingly, Semas have been implicated in immune system pathologies, and though they serve as prognostic markers there is generally little understanding of their biological relevance (Takamatsu and Kumanogoh, 2012). Additionally, some Semas have been implicated in immunity. For instance, mouse dendritic cells secrete

SEMA3E to recruit natural killer cells to sites of infection (Alamri et al., 2018). Therefore, it is possible that Sema3f also directly regulates cells of the immune system, and the upregulation of complement factor mRNAs observed by RNAseq does not reflect the earliest onset of a pathology in the retina. In this regard, sema3fa is expressed by blood islands in the tail of the zebrafish embryo – a site where hematopoietic lineages are specified and differentiate (Paik and Zon, 2010).

Given that Sema3f signalling likely controls some aspects of retinal cell genesis, it is conceivable

216

that Sema3fa also plays a role in immune system development. To identify the reason for complement factor upregulation with Sema3fa loss, I would first verify that changes in mRNA occurred within the retina by in situ hybridization. Second, I would use neutral red, a vital dye which is phagocytosed by microglia and macrophages and makes them visible under light microscopy, to analyze whether macrophages congregate in the retina before, during and throughout CNV. Methodically mining the CNV phenotype will allow the lab to further develop the AMD animal model I have generated.

The hyaloid vessel was also affected by the loss of Sema3fa signalling. My data suggest that in this system neural retina-secreted Sema3fa acts as an antiangiogenic to ensure adequate spacing and spread of vessels by establish no-grow domains forcing vessels to branch towards the peripheral retina and cover the inner limiting membrane. Loss of Sema3fa signalling results in thin, highly branched vessels from the optic nerve head, which terminate as densely compact capillaries in the retinal periphery. No area of the retina was obviously avascular in the sema3fa mutants, suggesting that loss of the molecule was not necessary for vascularization as a whole. My data argue that Sema3fa controls the pattern of branching of retinal vessels, but prevents choroid capillaries from entering the neural retina. What effect a highly vascularized hyaloid might have on retinal and vessel-associated cell types in the sema3fa mutants remains to be determined.

6.5 Sema3fb in the heart

I was fortunate that sema3fb and sema3fa have generally non-overlapping expression patterns, which allowed me to use mutant alleles to distinguish readily their roles within the different organ systems. Interestingly, the one region of overlap of expression of the two orthologues is the early embryonic eye, where both genes are expressed in the temporal neural retina (though sema3fb for a shorter period of time), and at later time points in different cell types

217

of the eye. I have not explored the role of Sema3fb signalling in the eye, although the biological roles are likely to be interesting. The co-expression of sema3fa and sema3fb in the temporal retina at 18 hpf may mitigate any early retinal cell genesis phenotypes that could present in the sema3fa mutant. It will be interesting to generate the double homozygote for both genes, and reassess the retinal phenotype to determine whether compensation by the orthologue did occur. The RNAseq data suggests that there is no transcriptional compensation by sema3fb at 36 hpf, so I do not expect the phenotypes to change grossly in the double mutant fish, aside from unveiling a possibly earlier phenotype that was indistinguishable in the single mutant. In the older 72 hpf retina sema3fb shows spotted expression within the INL, while sema3fa is expressed throughout the inner portion of the

INL. Given that Sema3fa is a secreted molecule, and Sema3fa and Sema3fb have highly similar structures, environmental compensation from the sema3fa would likely mask any late phenotypes in the sema3fb mutant. Nonetheless, determining whether the double homozygote phenotype is different is of interest as we move forward to fully understand the role of a Sema3f in retinal development.

Instead, my analysis of Sema3fb signalling focused entirely on heart development, where I identify Sema3fb as a novel regulator of ventricle chamber differentiation. Loss of Sema3fb resulted in smaller ventricles, loss of chamber specific marker (irx1a) and misexpressed atrial and endocardial cushion markers (myh6, bmp4a, respectively). On a cellular level, ventricular cardiomyocytes are smaller in size than their wildtype counterparts, and ultimately produce thinner chamber walls. Multiple, linked, phenotypes exist, and my analysis primarily focused on the ventricle due to the availability of assays and techniques, and more so spatial expression of putative

Sema3fb receptors in the ventricle itself. While the atrial chamber was smaller and misexpressed ventricle marker proteins in the mutants, I proposed (discussed in Chapter 5) that the atrial defects

218

are likely a consequence of ventricular dysfunction, as Sema3fb ligand and receptors were only coexpressed within the ventricle. Because I lacked antibodies to label the membranes of atrial cardiomyocytes, I was unable to assess atrial cardiomyocyte size, and with no atrial markers other than myh6 I could not determine whether transcriptional differentiation was perturbed like in the ventricle. Regardless, the protein data supports the gene expression data in that DM-GRASP and

MF20 spread into the atrium from the ventricle, supporting the idea atrial differentiation is also perturbed. Interestingly, sema3fb mRNA is present in the early embryo within the presumptive heart field. While fgf8a and tbx5a, factors important for heart induction and heart specification, respectively, were expressed similarly in wildtype and mutant embryos, the early presence of sema3fb could suggest some early, overlooked role. It will be important to determine, by ISH at the bilateral heart field stage, whether sema3fb and its receptors are expressed by both atrial and ventricular cardiomyocyte progenitor pools, or only co-expressed in the ventricle, as seen in the hearts of older embryos. Bilateral heart fields move and fuse at the midline (Yelon et al., 1999), and while I saw no evidence of failure to fuse in the sema3fb mutants, it will be interesting to determine by cell counts and spacing between cells whether Sema3fb plays a role in this process.

Chamber identities are plastic and therefore constant maintenance of ventricle or atrial identity is necessary following the initial specification during early gastrulation (Bao et al., 1999;

Bruneau et al., 2001; Pradhan et al., 2017). What remains unclear is how chamber-specific expression of transcription factors is established and maintained. As I proposed in Chapter 5,

Sema3fb is one such factor necessary for the development of complete ventricle identity by way of supressing atrial identity. I identified irx1a as a likely downstream transcriptional target of

Sema3fb signalling, as expression was largely downregulated in the ventricle in the mutant. In turn

Irx1a likely inhibits atrial marker gene expression (Bao et al., 1999), explaining why myh6 atrial

219

expression is expanded into the ventricle of mutant embryos. The model I propose will allow us to dissect spatial chamber identity acquisition in a methodical manner, as the ventricle appears to be directly affected by disruptions of Sema3fb signalling, while changes to the atrium are consequential. To delve further we will need to perform RNAseq analysis to get a fuller picture of the: 1) Transcriptome of the ventricle versus atrium in wildtype embryos, and 2) Chamber specific changes between wildtype and mutant embryos. Through ventricle dissection and RNA extraction, we can use differential expression of genes between wildtype and mutant embryos to define the

Sema3fb signalling cascade within 48 hpf ventricular cardiomyocytes. Additionally, we can explore atrial changes in mutants to identify differentially expressed genes from wildtype embryos which may reveal chamber regulatory pathways that are disrupted. While such changes are likely not due to direct actions of Sema3fb, they will inform our understanding of maintenance of the atrial fate. Altogether, RNAseq data will permit deeper exploration of the downstream regulation by Sema3fb, similar to what I achieved with the Sema3fa analysis of Chapter 3.

6.6 Importance of spatial signalling in development

My work focused on a single signalling molecule, Sema3f, and its ability to provide spatial information to distinct cell types with different downstream effects during development. Sema3fa in the early optic cup informs RPCs that they are in the temporal retina, and within the CMZ and

RPE tells the choroid and hyaloid vessels where not to grow. In the heart, Sema3fb acts on ventricular cardiomyocyte progenitor cells to regulate ventricle chamber differentiation and development. Collectively my work raises two big questions: why Sema3f and how can one molecule play a role in so many distinct developing systems? Ultimately, what is unique about

Sema3f that it is used repeatedly during development in various tissues?

220

To understand the functions of Sema3f we need to first consider what type of protein it is, and how Sema3f would be presented to cells in its environment. Secreted molecules generally diffuse away from their source - such a feature is essential in morphogen patterning as the concentration of a morphogen in the extracellular environment impacts cellular development.

Sema3f is also secreted, and to confer spatial identity, the protein would need to stay relatively localized. While I have no protein localization data, the known functions of secreted chemotactic molecules such as the Sema3s do tend to argue that the proteins are localized. For instance, during optic tract outgrowth in Xenopus laevis, RGC axons make a caudal turn at the mid-diencephalon towards the optic tectum (Atkinson-Leadbeater et al., 2010) that depends on the expression of two chemorepulsive molecules, sema3a and slit1, within close by neighbouring forebrain regions.

Thus, I propose that secreted Sema3fs are spatially localized to a high degree. But what gives

Sema3f the flexibility to act with spatially restricted downstream effects? Below I discuss the possibility that distinct Sema3f functions require specific receptor expression.

In my thesis I looked at two types of cells: progenitor cells (retina and heart) and differentiated cells (endothelial vasculature). Inherently these cell populations are at distinct points in their developmental trajectory, as progenitors are immature and still acquiring identity, while the vessel endothelial cells have established their identity. Thus, the signals acting downstream of

Sema3fa in these two cases may differ. How can this be brought about? My data argue that this may happen through select receptor mediated signalling. I find that both plxna3 and nrp2b are expressed in the progenitors of the retina and the heart ventricle, while nrp1b is expressed by the vasculature. I still need to verify the role of these receptors in the different Sema3f-dependent processes, but it does suggest an interesting biology whereby Plxna3 and Nrp2b may interpret

Sema3f signalling for downstream transcriptional changes, while Nrp1b interprets Sema3f

221

signalling in a more canonical, chemorepulsive manner that acts via the cytoskeleton. Literature supports this dual biology - SEMA3F signalling is proposed to downregulate transcriptional expression of CXCR4 in tumor cells in vitro (Zhou et al., 2015) and repulse blood vessels through cytoskeletal collapse and induction of apoptosis in an in vitro model of angiogenesis (Guttmann-

Raviv et al., 2007). Which paired receptors (PLXN and NRP) SEMA3F signals through in these different cases, however, is unknown. What is also interesting, and yet to be determined, is whether the localization of the receptor and ligand plays a role in determining the downstream consequences. Progenitors apparently express both the receptor(s) and the ligand, while the endothelial cells of vessels express the receptor but other cell types, the RPE and RGCs/CMZ, secrete the ligand. Is cell autonomy (progenitors) vs. non-autonomy (vessels) an additional mechanism to maintain spatial sensitivity and elicit specific signalling pathways? What is clear is that Sema3f signalling is clearly indispensable for normal embryogenesis and that its ability to confer spatial information is necessary for the genesis and development of multiple organ systems.

6.7 Concluding Remarks

Cells are continually exposed to extrinsic factors in their environment which in turn regulates their intrinsic properties. It is essential that spatial and temporal information is conferred to cells by way of extrinsic mechanisms to guide differentiation, migration and even cellular functioning. Studies of spatial regulators of development have focused primarily on the actions of morphogens and their presentation in gradients. Here, for the first time I showed novel roles for a single secreted Sema, Sema3f, in RPC development, choroidal and hyaloid vessel growth, and heart development. Together my data adds to and supports literature suggesting non-guidance related functions of Semas. And further, my data provides the cellular importance of spatial

222

information conferred by a single signalling molecule, Sema3f, to elicit downstream changes to development and differentiation and cellular biology.

223

References

Adams, R.H., and Eichmann, A. (2010). Axon guidance molecules in vascular patterning. Cold Spring Harb. Perspect. Biol. 2, a001875. Alamri, A., Rahman, R., Zhang, M., Alamri, A., Gounni, A.S., and Kung, S.K.P. (2018). Semaphorin-3E Produced by Immature Dendritic Cells Regulates Activated Natural Killer Cells Migration. Front. Immunol. 9, 1005. Alsan, B.H., and Schultheiss, T.M. (2002). Regulation of avian cardiogenesis by Fgf8 signalling. Dev. Camb. Engl. 129, 1935–1943. Alto, L.T., and Terman, J.R. (2017). Semaphorins and their Signalling Mechanisms. Methods Mol. Biol. Clifton NJ 1493, 1–25. Alvarez, Y., Cederlund, M.L., Cottell, D.C., Bill, B.R., Ekker, S.C., Torres-Vazquez, J., Weinstein, B.M., Hyde, D.R., Vihtelic, T.S., and Kennedy, B.N. (2007). Genetic determinants of hyaloid and retinal vasculature in zebrafish. BMC Dev. Biol. 7, 114. Amini, R., Rocha-Martins, M., and Norden, C. (2018). Neuronal Migration and Lamination in the Vertebrate Retina. Front. Neurosci. 11. Anderson, J.L., Mulligan, T.S., Shen, M.-C., Wang, H., Scahill, C.M., Tan, F.J., Du, S.J., Busch- Nentwich, E.M., and Farber, S.A. (2017). mRNA processing in mutant zebrafish lines generated by chemical and CRISPR-mediated mutagenesis produces unexpected transcripts that escape nonsense-mediated decay. PLoS Genet. 13. Arbeille, E., Reynaud, F., Sanyas, I., Bozon, M., Kindbeiter, K., Causeret, F., Pierani, A., Falk, J., Moret, F., and Castellani, V. (2015). Cerebrospinal fluid-derived Semaphorin3B orients neuroepithelial cell divisions in the apicobasal axis. Nat. Commun. 6, 6366. Atkinson-Leadbeater, K., Bertolesi, G.E., Hehr, C.L., Webber, C.A., Cechmanek, P.B., and McFarlane, S. (2010). Dynamic expression of axon guidance cues required for optic tract development is controlled by fibroblast growth factor signalling. J. Neurosci. Off. J. Soc. Neurosci. 30, 685–693. Auman, H.J., Coleman, H., Riley, H.E., Olale, F., Tsai, H.-J., and Yelon, D. (2007). Functional Modulation of Cardiac Form through Regionally Confined Cell Shape Changes. PLOS Biol. 5, e53. Avanesov, A., and Malicki, J. (2010). Analysis of the Retina in the Zebrafish Model. Methods Cell

224

Biol. 100, 153–204. Bader, D., Masaki, T., and Fischman, D.A. (1982). Immunochemical analysis of myosin heavy chain during avian myogenesis in vivo and in vitro. J. Cell Biol. 95, 763–770. Baffi, J., Byrnes, G., Chan, C.-C., and Csaky, K.G. (2000). Choroidal Neovascularization in the Rat Induced by Adenovirus Mediated Expression of Vascular Endothelial Growth Factor. Invest. Ophthalmol. Vis. Sci. 41, 3582–3589. Bahary, N., Goishi, K., Stuckenholz, C., Weber, G., LeBlanc, J., Schafer, C.A., Berman, S.S., Klagsbrun, M., and Zon, L.I. (2007). Duplicate VegfA genes and orthologues of the KDR receptor tyrosine kinase family mediate vascular development in the zebrafish. Blood 110, 3627–3636. Bai, Y., Ma, J., Guo, J., Wang, J., Zhu, M., Chen, Y., and Le, Y.-Z. (2009). Müller cell-derived VEGF is a significant contributor to retinal neovascularization. J. Pathol. 219, 446–454. Bai, Y., Liang, S., Yu, W., Zhao, M., Huang, L., Zhao, M., and Li, X. (2014). Semaphorin 3A blocks the formation of pathologic choroidal neovascularization induced by transforming growth factor beta. Mol. Vis. 20, 1258–1270. Bakkers, J. (2011). Zebrafish as a model to study cardiac development and human cardiac disease. Cardiovasc. Res. 91, 279–288. Bao, Z.Z., Bruneau, B.G., Seidman, J.G., Seidman, C.E., and Cepko, C.L. (1999). Regulation of chamber-specific gene expression in the developing heart by Irx4. Science 283, 1161–1164. Barbieri, A.M., Lupo, G., Bulfone, A., Andreazzoli, M., Mariani, M., Fougerousse, F., Consalez, G.G., Borsani, G., Beckmann, J.S., Barsacchi, G., et al. (1999). A homeobox gene, vax2, controls the patterning of the eye dorsoventral axis. Proc. Natl. Acad. Sci. U. S. A. 96, 10729–10734. Bedell, V.M., Yeo, S.Y., Park, K.W., Chung, J., Seth, P., Shivalingappa, V., Zhao, J., Obara, T., Sukhatme, V.P., Drummond, I.A., et al. (2005). Roundabout4 is Essential for Angiogenesis in Vivo. Proc. Natl. Acad. Sci. U. S. A. 102, 6373–6378. Begemann, G., and Ingham, P.W. (2000). Developmental regulation of Tbx5 in zebrafish embryogenesis. Mech. Dev. 90, 299–304. Beis, D., Bartman, T., Jin, S.-W., Scott, I.C., D’Amico, L.A., Ober, E.A., Verkade, H., Frantsve, J., Field, H.A., Wehman, A., et al. (2005). Genetic and cellular analyses of zebrafish atrioventricular cushion and valve development. Development 132, 4193–4204. Berdougo, E., Coleman, H., Lee, D.H., Stainier, D.Y.R., and Yelon, D. (2003). Mutation of weak

225

atrium/atrial myosin heavy chain disrupts atrial function and influences ventricular morphogenesis in zebrafish. Development 130, 6121–6129. Berndt, J.D., and Halloran, M.C. (2006). Semaphorin 3d promotes cell proliferation and neural crest cell development downstream of TCF in the zebrafish hindbrain. Dev. Camb. Engl. 133, 3983–3992. Bertolesi, G.E., Hehr, C.L., and McFarlane, S. (2014). Wiring the retinal circuits activated by light during early development. Neural Develop. 9, 3. Bharadwaj, A.S., Appukuttan, B., Wilmarth, P.A., Pan, Y., Stempel, A.J., Chipps, T.J., Benedetti, E.E., Zamora, D.O., Choi, D., David, L.L., et al. (2013). Role of the retinal vascular endothelial cell in ocular disease. Prog. Retin. Eye Res. 32, 102–180. Bill, A., Sperber, G., and Ujiie, K. (1983). Physiology of the choroidal vascular bed. Int. Ophthalmol. 6, 101–107. Blackshaw, S., Harpavat, S., Trimarchi, J., Cai, L., Huang, H., Kuo, W.P., Weber, G., Lee, K., Fraioli, R.E., Cho, S.-H., et al. (2004). Genomic analysis of mouse retinal development. PLoS Biol. 2, E247. Bloomekatz, J., Singh, R., Prall, O.W., Dunn, A.C., Vaughan, M., Loo, C.-S., Harvey, R.P., and Yelon, D. (2017). Platelet-derived growth factor (PDGF) signalling directs cardiomyocyte movement toward the midline during heart tube assembly. ELife 6, e21172. Boije, H., Rulands, S., Dudczig, S., Simons, B.D., and Harris, W.A. (2015). The Independent Probabilistic Firing of Transcription Factors: A Paradigm for Clonal Variability in the Zebrafish Retina. Dev. Cell 34, 532–543. Bond, A.M., Bhalala, O.G., and Kessler, J.A. (2012). The Dynamic Role of Bone Morphogenetic Proteins in Neural Stem Cell Fate and Maturation. Dev. Neurobiol. 72, 1068–1084. Bozic, I., Li, X., and Tao, Y. (2018). Quantitative biometry of zebrafish retinal vasculature using optical coherence tomographic angiography. Biomed. Opt. Express 9, 1244–1255. Brose, K., and Tessier-Lavigne, M. (2000). Slit proteins: key regulators of axon guidance, axonal branching, and cell migration. Curr. Opin. Neurobiol. 10, 95–102. Brose, K., Bland, K.S., Wang, K.H., Arnott, D., Henzel, W., Goodman, C.S., Tessier-Lavigne, M., and Kidd, T. (1999). Slit proteins bind Robo receptors and have an evolutionarily conserved role in repulsive axon guidance. Cell 96, 795–806.

226

Brown, D.R., Samsa, L.A., Qian, L., and Liu, J. (2016). Advances in the Study of Heart Development and Disease Using Zebrafish. J. Cardiovasc. Dev. Dis. 3. Bruneau, B.G., Logan, M., Davis, N., Levi, T., Tabin, C.J., Seidman, J.G., and Seidman, C.E. (1999). Chamber-specific cardiac expression of Tbx5 and heart defects in Holt-Oram syndrome. Dev. Biol. 211, 100–108. Bruneau, B.G., Nemer, G., Schmitt, J.P., Charron, F., Robitaille, L., Caron, S., Conner, D.A., Gessler, M., Nemer, M., Seidman, C.E., et al. (2001). A murine model of Holt-Oram syndrome defines roles of the T-box transcription factor Tbx5 in cardiogenesis and disease. Cell 106, 709– 721. Buehler, A., Sitaras, N., Favret, S., Bucher, F., Berger, S., Pielen, A., Joyal, J.S., Juan, A.M., Martin, G., Schlunck, G., et al. (2013). Semaphorin 3F forms an anti-angiogenic barrier in outer retina. FEBS Lett. 587, 1650–1655. Callander, D.C., Lamont, R.E., Childs, S.J., and McFarlane, S. (2007). Expression of multiple class three semaphorins in the retina and along the path of zebrafish retinal axons. Dev. Dyn. Off. Publ. Am. Assoc. Anat. 236, 2918–2924. Cameron, D.J., Rassamdana, F., Tam, P., Dang, K., Yanez, C., Ghaemmaghami, S., and Dehkordi, M.I. (2013). The Optokinetic Response as a Quantitative Measure of Visual Acuity in Zebrafish. J. Vis. Exp. JoVE. Campochiaro, P.A. (2015). Molecular Pathogenesis of Retinal and Choroidal Vascular Diseases. Prog. Retin. Eye Res. 49, 67. Cao, R., Jensen, L.D.E., Söll, I., Hauptmann, G., and Cao, Y. (2008). Hypoxia-Induced Retinal Angiogenesis in Zebrafish as a Model to Study Retinopathy. PLOS ONE 3, e2748. Cavanaugh, A.M., Huang, J., and Chen, J.-N. (2015). Two developmentally distinct populations of neural crest cells contribute to the zebrafish heart. Dev. Biol. 404, 103–112. Cechmanek, P. (2018). “Sema6d Signaling in the Developing Visual System”. Doctoral Thesis, Neuroscience, University of Calgary, Calgary AB. Cederlund, M.L., Morrissey, M.E., Baden, T., Scholz, D., Vendrell, V., Lagnado, L., Connaughton, V.P., and Kennedy, B.N. (2011). Zebrafish Tg(7.2mab21l2:EGFP)ucd2 Transgenics Reveal a Unique Population of Retinal Amacrine Cells. Invest. Ophthalmol. Vis. Sci. 52, 1613–1621.

227

Cepko, C. (2014). Intrinsically different retinal progenitor cells produce specific types of progeny. Nat. Rev. Neurosci. 15, 615–627. Cepko, C.L. (1999). The roles of intrinsic and extrinsic cues and bHLH genes in the determination of retinal cell fates. Curr. Opin. Neurobiol. 9, 37–46. Cepko, C.L., Austin, C.P., Yang, X., Alexiades, M., and Ezzeddine, D. (1996). Cell fate determination in the vertebrate retina. Proc. Natl. Acad. Sci. U. S. A. 93, 589–595. Cerani, A., Tetreault, N., Menard, C., Lapalme, E., Patel, C., Sitaras, N., Beaudoin, F., Leboeuf, D., Guire, V.D., Binet, F., et al. (2013). Neuron-derived semaphorin 3A is an early inducer of vascular permeability in diabetic retinopathy via neuropilin-1. Cell Metab. 18, 505–518. Cerveny, K.L., Cavodeassi, F., Turner, K.J., de Jong-Curtain, T.A., Heath, J.K., and Wilson, S.W. (2010). The zebrafish flotte lotte mutant reveals that the local retinal environment promotes the differentiation of proliferating precursors emerging from their stem cell niche. Dev. Camb. Engl. 137, 2107–2115. Chapman, D.L., Garvey, N., Hancock, S., Alexiou, M., Agulnik, S.I., Gibson-Brown, J.J., Cebra- Thomas, J., Bollag, R.J., Silver, L.M., and Papaioannou, V.E. (1996). Expression of the T-box family genes, Tbx1-Tbx5, during early mouse development. Dev. Dyn. Off. Publ. Am. Assoc. Anat. 206, 379–390. Chen, J., Kubalak, S.W., Minamisawa, S., Price, R.L., Becker, K.D., Hickey, R., Ross, J., and Chien, K.R. (1998). Selective Requirement of Myosin Light Chain 2v in Embryonic Heart Function. J. Biol. Chem. 273, 1252–1256. Chen, Y., Hu, Y., Lu, K., Flannery, J.G., and Ma, J.-X. (2007). Very low density lipoprotein receptor, a negative regulator of the wnt signalling pathway and choroidal neovascularization. J. Biol. Chem. 282, 34420–34428. Chen, Y., Hu, Y., Moiseyev, G., Zhou, K.K., Chen, D., and Ma, J. (2009). Photoreceptor degeneration and retinal inflammation induced by very low-density lipoprotein receptor deficiency. Microvasc. Res. 78, 119–127. Chen, Z., Li, X., and Desplan, C. (2012). Deterministic or stochastic choices in retinal neuron specification. Neuron 75, 739–742. Cheng, C.W., Hui, C., Strähle, U., and Cheng, S.H. (2001). Identification and expression of zebrafish Iroquois homeobox gene irx1. Dev. Genes Evol. 211, 442–444.

228

Cheng, C.W., Yan, C.H.M., Hui, C., Strähle, U., and Cheng, S.H. (2006). The homeobox gene irx1a is required for the propagation of the neurogenic waves in the zebrafish retina. Mech. Dev. 123, 252–263. Cheung, N., Wong, I.Y., and Wong, T.Y. (2014). Ocular anti-VEGF therapy for diabetic retinopathy: overview of clinical efficacy and evolving applications. Diabetes Care 37, 900–905. Chien, K.R., Zhu, H., Knowlton, K.U., Miller-Hance, W., van-Bilsen, M., O’Brien, T.X., and Evans, S.M. (1993). Transcriptional regulation during cardiac growth and development. Annu. Rev. Physiol. 55, 77–95. Choi, W., Mohler, K.J., Potsaid, B., Lu, C.D., Liu, J.J., Jayaraman, V., Cable, A.E., Duker, J.S., Huber, R., and Fujimoto, J.G. (2013). Choriocapillaris and Choroidal Microvasculature Imaging with Ultrahigh Speed OCT Angiography. PLoS ONE 8. Chrispell, J.D., Rebrik, T.I., and Weiss, E.R. (2015). Electroretinogram analysis of the visual response in zebrafish larvae. J. Vis. Exp. JoVE. Christoffels, V.M., Burch, J.B.E., and Moorman, A.F.M. (2004). Architectural plan for the heart: early patterning and delineation of the chambers and the nodes. Trends Cardiovasc. Med. 14, 301– 307. Coleman, H.R., Chan, C.-C., Ferris, F.L., and Chew, E.Y. (2008). Age-related macular degeneration. The Lancet 372, 1835–1845. Collery, R., McLoughlin, S., Vendrell, V., Finnegan, J., Crabb, J.W., Saari, J.C., and Kennedy, B.N. (2008). Duplication and divergence of zebrafish CRALBP genes uncovers novel role for RPE- and Muller-CRALBP in cone vision. Invest. Ophthalmol. Vis. Sci. 49, 3812–3820. Coultas, L., Chawengsaksophak, K., and Rossant, J. (2005). Endothelial cells and VEGF in vascular development. Country, M. (2017). Retinal metabolism: A comparative look at energetics in the retina. Brain Res. 1672, 50–57. Cui, C., and Lu, H. (2017). Clinical observations on the use of new anti-VEGF drug, conbercept, in age-related macular degeneration therapy: a meta-analysis. Cunha-Vaz, J., Bernardes, R., and Lobo, C. (2011). Blood-Retinal Barrier. Eur. J. Ophthalmol. 21, 3–9. D. Lukasiewicz, P. (2005). Synaptic mechanisms that shape visual signalling at the inner retina.

229

In Progress in Brain Research, (Elsevier), pp. 205–218. Degenhardt, K., Singh, M.K., Aghajanian, H., Massera, D., Wang, Q., Li, J., Li, L., Choi, C., Yzaguirre, A.D., Francey, L.J., et al. (2013). Semaphorin 3d signalling defects are associated with anomalous pulmonary venous connections. Nat. Med. 19, 760–765. Deiner, M.S., Kennedy, T.E., Fazeli, A., Serafini, T., Tessier-Lavigne, M., and Sretavan, D.W. (1997). Netrin-1 and DCC mediate axon guidance locally at the optic disc: loss of function leads to optic nerve hypoplasia. Neuron 19, 575–589. Demy, D.L., Tauzin, M., Lancino, M., Le Cabec, V., Redd, M., Murayama, E., Maridonneau- Parini, I., Trede, N., and Herbomel, P. (2017). Trim33 is essential for macrophage and neutrophil mobilization to developmental or inflammatory cues. J. Cell Sci. 130, 2797–2807. Denvir, M.A., Tucker, C.S., and Mullins, J.J. (2008). Systolic and diastolic ventricular function in zebrafish embryos: Influence of norepenephrine, MS-222 and temperature. BMC Biotechnol. 8, 21. DeRuiter, M.C., Poelmann, R.E., VanderPlas-de Vries, I., Mentink, M.M.T., and Gittenberger-de Groot, A.C. (1992). The development of the myocardium and endocardium in mouse embryos. Anat. Embryol. (Berl.) 185, 461–473. Dooley, K., and Zon, L.I. (2000). Zebrafish: a model system for the study of human disease. Curr. Opin. Genet. Dev. 10, 252–256. Duan, L.-J., Pan, S.J., Sato, T.N., and Fong, G.-H. (2017). Retinal Angiogenesis Regulates Astrocytic Differentiation in Neonatal Mouse Retinas by Oxygen Dependent Mechanisms. Sci. Rep. 7, 17608. Durston, A.J., Jansen, H.J., and Wacker, S.A. (2010). Time-space translation: a developmental principle. ScientificWorldJournal 10, 2207–2214. Dyer, M.A., and Cepko, C.L. (2001). Regulating proliferation during retinal development. Nat. Rev. Neurosci. 2, 333–342. Edwards, A.O., Iii, R.R., Abel, K.J., Manning, A., Panhuysen, C., and Farrer, L.A. (2005). Complement Factor H Polymorphism and Age-Related Macular Degeneration. 308, 5. El-Brolosy, M., Rossi, A., Kontarakis, Z., Kuenne, C., Guenther, S., Fukuda, N., Takacs, C., Lai, S.-L., Fukuda, R., Gerri, C., et al. (2018). Genetic compensation is triggered by mutant mRNA degradation. BioRxiv 328153.

230

El-Remessy, A.B., Behzadian, M.A., Abou-Mohamed, G., Franklin, T., Caldwell, R.W., and Caldwell, R.B. (2003). Experimental Diabetes Causes Breakdown of the Blood-Retina Barrier by a Mechanism Involving Tyrosine Nitration and Increases in Expression of Vascular Endothelial Growth Factor and Urokinase Plasminogen Activator Receptor. Am. J. Pathol. 162, 1995–2004. Engerer, P., Suzuki, S.C., Yoshimatsu, T., Chapouton, P., Obeng, N., Odermatt, B., Williams, P.R., Misgeld, T., and Godinho, L. (2017). Uncoupling of neurogenesis and differentiation during retinal development. EMBO J. 36, 1134–1146. Epstein, J.A., Aghajanian, H., and Singh, M.K. (2015). Semaphorin Signalling in Cardiovascular Development. Cell Metab. 21, 163–173. Erskine, L., Reijntjes, S., Pratt, T., Denti, L., Schwarz, Q., Vieira, J.M., Alakakone, B., Shewan, D., and Ruhrberg, C. (2011). VEGF signalling through neuropilin 1 guides commissural axon crossing at the optic chiasm. Neuron 70, 951–965. Ersoz, M.G., Karacorlu, M., Arf, S., Sayman Muslubas, I., and Hocaoglu, M. (2017). Retinal pigment epithelium tears: Classification, pathogenesis, predictors, and management. Surv. Ophthalmol. 62, 493–505. Farhy, C., Elgart, M., Shapira, Z., Oron-Karni, V., Yaron, O., Menuchin, Y., Rechavi, G., and Ashery-Padan, R. (2013). Pax6 Is Required for Normal Cell-Cycle Exit and the Differentiation Kinetics of Retinal Progenitor Cells. PLOS ONE 8, e76489. Fashena, D., and Westerfield, M. (1999). Secondary motoneuron axons localize DM-GRASP on their fasciculated segments. J. Comp. Neurol. 406, 415–424. Feiner, L., Webber, A.L., Brown, C.B., Lu, M.M., Jia, L., Feinstein, P., Mombaerts, P., Epstein, J.A., and Raper, J.A. (2001). Targeted disruption of semaphorin 3C leads to persistent truncus arteriosus and aortic arch interruption. Dev. Camb. Engl. 128, 3061–3070. Feldheim, D.A., and O’Leary, D.D. (2010). Visual map development: bidirectional signalling, bifunctional guidance molecules, and competition. Cold Spring Harb. Perspect. Biol. 2, a001768. Feng, L., Xie, Z., Ding, Q., Xie, X., Libby, R.T., and Gan, L. (2010). MATH5 controls the acquisition of multiple retinal cell fates. Mol. Brain 3, 36. Fimbel, S.M., Montgomery, J.E., Burket, C.T., and Hyde, D.R. (2007). Regeneration of inner retinal neurons after intravitreal injection of ouabain in zebrafish. J. Neurosci. Off. J. Soc. Neurosci. 27, 1712–1724.

231

Fischer, S., Wobben, M., Marti, H.H., Renz, D., and Schaper, W. (2002). Hypoxia-induced hyperpermeability in brain microvessel endothelial cells involves VEGF-mediated changes in the expression of zonula occludens-1. Microvasc. Res. 63, 70–80. Fishman, M.C., and Chien, K.R. (1997). Fashioning the vertebrate heart: earliest embryonic decisions. Dev. Camb. Engl. 124, 2099–2117. Fort, P.E., and Lampi, K.J. (2011). New focus on alpha-crystallins in retinal neurodegenerative diseases. Exp. Eye Res. 92, 98. Fruttiger, M. (2007). Development of the retinal vasculature. Angiogenesis 10, 77–88. Fujisawa, H. (2004). Discovery of semaphorin receptors, neuropilin and plexin, and their functions in neural development. J. Neurobiol. 59, 24–33. Gagnon, J.A., Valen, E., Thyme, S.B., Huang, P., Ahkmetova, L., Pauli, A., Montague, T.G., Zimmerman, S., Richter, C., and Schier, A.F. (2014). Efficient mutagenesis by Cas9 protein- mediated oligonucleotide insertion and large-scale assessment of single-guide RNAs. PloS One 9, e98186. Gammill, L.S., Gonzalez, C., Gu, C., and Bronner-Fraser, M. (2006). Guidance of trunk neural crest migration requires neuropilin 2/semaphorin 3F signalling. Dev. Camb. Engl. 133, 99–106. Gammill, L.S., Gonzalez, C., and Bronner-Fraser, M. (2007). Neuropilin 2/semaphorin 3F signalling is essential for cranial neural crest migration and trigeminal ganglion condensation. Dev. Neurobiol. 67, 47–56. Garavito-Aguilar, Z.V., Riley, H.E., and Yelon, D. (2010). Hand2 ensures an appropriate environment for cardiac fusion by limiting Fibronectin function. Dev. Camb. Engl. 137, 3215– 3220. Gariano, R.F., and Gardner, T.W. (2005). Retinal angiogenesis in development and disease. Nature 438, 960–966. Garrity, D.M., Childs, S., and Fishman, M.C. (2002). The heartstrings mutation in zebrafish causes heart/fin Tbx5 deficiency syndrome. Development 129, 4635–4645. Garside, V.C., Chang, A.C., Karsan, A., and Hoodless, P.A. (2013). Co-ordinating Notch, BMP, and TGF-β signalling during heart valve development. Cell. Mol. Life Sci. CMLS 70, 2899–2917. Gemberling, M., Bailey, T.J., Hyde, D.R., and Poss, K.D. (2013). The zebrafish as a model for complex tissue regeneration. Trends Genet. 29, 611–620.

232

Georgala, P.A., Carr, C.B., and Price, D.J. (2011). The role of Pax6 in forebrain development. Dev. Neurobiol. 71, 690–709. Gerhardt, H. (2008). VEGF and endothelial guidance in angiogenic sprouting. Organogenesis 4, 241–246. Gerhardt, H., Golding, M., Fruttiger, M., Ruhrberg, C., Lundkvist, A., Abramsson, A., Jeltsch, M., Mitchell, C., Alitalo, K., Shima, D., et al. (2003). VEGF guides angiogenic sprouting utilizing endothelial tip cell filopodia. J. Cell Biol. 161, 1163–1177. Giacobini, P., and Prevot, V. (2013). Semaphorins in the development, homeostasis and disease of hormone systems. Semin. Cell Dev. Biol. 24, 190–198. Godinho, L., Williams, P.R., Claassen, Y., Provost, E., Leach, S.D., Kamermans, M., and Wong, R.O.L. (2007). Nonapical symmetric divisions underlie horizontal cell layer formation in the developing retina in vivo. Neuron 56, 597–603. Gomes, F.L.A.F., Zhang, G., Carbonell, F., Correa, J.A., Harris, W.A., Simons, B.D., and Cayouette, M. (2011). Reconstruction of rat retinal progenitor cell lineages in vitro reveals a surprising degree of stochasticity in cell fate decisions. Dev. Camb. Engl. 138, 227–235. González-Mariscal, L., Betanzos, A., Nava, P., and Jaramillo, B.E. (2003). Tight junction proteins. Prog. Biophys. Mol. Biol. 81, 1–44. Goodson, N.B., Nahreini, J., Randazzo, G., Uruena, A., Johnson, J.E., and Brzezinski, J.A. (2018). Prdm13 is required for Ebf3+ amacrine cell formation in the retina. Dev. Biol. 434, 149–163. Grant, M.G., Patterson, V.L., Grimes, D.T., and Burdine, R.D. (2017). Chapter One - Modeling Syndromic Congenital Heart Defects in Zebrafish. In Current Topics in Developmental Biology, K.C. Sadler, ed. (Academic Press), pp. 1–40. Grimes, A.C., Stadt, H.A., Shepherd, I.T., and Kirby, M.L. (2006). Solving an enigma: Arterial pole development in the zebrafish heart. Dev. Biol. 290, 265–276. Gu, C., Yoshida, Y., Livet, J., Reimert, D.V., Mann, F., Merte, J., Henderson, C.E., Jessell, T.M., Kolodkin, A.L., and Ginty, D.D. (2005). Semaphorin 3E and plexin-D1 control vascular pattern independently of neuropilins. Science 307, 265–268. Guner-Ataman, B., Paffett-Lugassy, N., Adams, M.S., Nevis, K.R., Jahangiri, L., Obregon, P., Kikuchi, K., Poss, K.D., Burns, C.E., and Burns, C.G. (2013). Zebrafish second heart field development relies on progenitor specification in anterior lateral plate mesoderm and nkx2.5

233

function. Development 140, 1353–1363. Guttmann-Raviv, N., Shraga-Heled, N., Varshavsky, A., Guimaraes-Sternberg, C., Kessler, O., and Neufeld, G. (2007). Semaphorin-3A and semaphorin-3F work together to repel endothelial cells and to inhibit their survival by induction of apoptosis. J. Biol. Chem. 282, 26294–26305. Hafler, B.P., Surzenko, N., Beier, K.T., Punzo, C., Trimarchi, J.M., Kong, J.H., and Cepko, C.L. (2012). Transcription factor Olig2 defines subpopulations of retinal progenitor cells biased toward specific cell fates. Proc. Natl. Acad. Sci. 109, 7882–7887. Hageman, G.S., Anderson, D.H., Johnson, L.V., Hancox, L.S., Taiber, A.J., Hardisty, L.I., Hageman, J.L., Stockman, H.A., Borchardt, J.D., Gehrs, K.M., et al. (2005). A common haplotype in the complement regulatory gene factor H (HF1͞CFH) predisposes individuals to age-related macular degeneration. 6. Haines, J.L., Hauser, M.A., Schmidt, S., Scott, W.K., Olson, L.M., Gallins, P., Spencer, K.L., Kwan, S.Y., Noureddine, M., Gilbert, J.R., et al. (2005). Complement Factor H Variant Increases the Risk of Age-Related Macular Degeneration. 308, 4. Hanovice, N., Leach, L., Slater, K., Gabriel, A., Romanovicz, D., Shao, E., Collery, R., Burton, E., Lathrop, K., Link, B., et al. (2018). Regeneration of the zebrafish retinal pigment epithelium after widespread genetic ablation. BioRxiv 372151. Hartsock, A., Lee, C., Arnold, V., and Gross, J.M. (2014). In vivo analysis of hyaloid vasculature morphogenesis in zebrafish: A role for the lens in maturation and maintenance of the hyaloid. Dev. Biol. Hashiura, T., Kimura, E., Fujisawa, S., Oikawa, S., Nonaka, S., Kurosaka, D., and Hitomi, J. (2017). Live imaging of primary ocular vasculature formation in zebrafish. PLOS ONE 12, e0176456. Hayashi, M., Nakashima, T., Taniguchi, M., Kodama, T., Kumanogoh, A., and Takayanagi, H. (2012). Osteoprotection by semaphorin 3A. Nature 485, 69–74. Hehr, C.L., Halabi, R., and McFarlane, S. (2018). Polarity and morphogenesis of the eye epithelium requires the adhesion junction associated adaptor protein Traf4. Cell Adhes. Migr. 1– 14. Hentze, M.W., and Kulozik, A.E. (1999). A perfect message: RNA surveillance and nonsense- mediated decay. Cell 96, 307–310.

234

Herbomel, P., Thisse, B., and Thisse, C. (2001). Zebrafish Early Macrophages Colonize Cephalic Mesenchyme and Developing Brain, Retina, and Epidermis through a M-CSF Receptor- Dependent Invasive Process. Dev. Biol. 238, 274–288. Hernández-Bejarano, M., Gestri, G., Spawls, L., Nieto-López, F., Picker, A., Tada, M., Brand, M., Bovolenta, P., Wilson, S.W., and Cavodeassi, F. (2015). Opposing Shh and Fgf signals initiate nasotemporal patterning of the zebrafish retina. Dev. Camb. Engl. 142, 3933–3942. Hillenbrand, P., Gerland, U., and Tkačik, G. (2016). Beyond the French Flag Model: Exploiting Spatial and Gene Regulatory Interactions for Positional Information. PloS One 11, e0163628. Hoage, T., Ding, Y., and Xu, X. (2012). Quantifying Cardiac Functions in Embryonic and Adult Zebrafish. In Cardiovascular Development: Methods and Protocols, X. Peng, and M. Antonyak, eds. (Totowa, NJ: Humana Press), pp. 11–20. Holtzman, N.G., Schoenebeck, J.J., Tsai, H.-J., and Yelon, D. (2007). Endocardium is necessary for cardiomyocyte movement during heart tube assembly. Dev. Camb. Engl. 134, 2379–2386. Homem, C.C.F., and Knoblich, J.A. (2012). Drosophila neuroblasts: a model for stem cell biology. Development 139, 4297–4310. Horb, M.E., and Thomsen, G.H. (1999). Tbx5 is essential for heart development. Development 126, 1739–1751. Hove, J.R., Köster, R.W., Forouhar, A.S., Acevedo-Bolton, G., Fraser, S.E., and Gharib, M. (2003). Intracardiac fluid forces are an essential epigenetic factor for embryonic cardiogenesis. Nature 421, 172–177. Hsiao, C.-D., Tsai, W.-Y., Horng, L.-S., and Tsai, H.-J. (2003). Molecular structure and developmental expression of three muscle-type troponin T genes in zebrafish. Dev. Dyn. 227, 266– 279. Hu, M., and Easter, S.S. (1999). Retinal neurogenesis: the formation of the initial central patch of postmitotic cells. Dev. Biol. 207, 309–321. Ieda, M., Kanazawa, H., Kimura, K., Hattori, F., Ieda, Y., Taniguchi, M., Lee, J.-K., Matsumura, K., Tomita, Y., Miyoshi, S., et al. (2007). Sema3a maintains normal heart rhythm through sympathetic innervation patterning. Nat. Med. 13, 604–612. Inoue, T., Hojo, M., Bessho, Y., Tano, Y., Lee, J.E., and Kageyama, R. (2002). Math3 and NeuroD regulate amacrine cell fate specification in the retina. Development 129, 831–842.

235

Isogai, S., Horiguchi, M., and Weinstein, B.M. (2001). The vascular anatomy of the developing zebrafish: an atlas of embryonic and early larval development. Dev. Biol. 230, 278–301. Ito, K., Kawasaki, T., Takashima, S., Matsuda, I., Aiba, A., and Hirata, T. (2008). Semaphorin 3F confines ventral tangential migration of lateral olfactory tract neurons onto the telencephalon surface. J. Neurosci. Off. J. Soc. Neurosci. 28, 4414–4422. Jeon, C.J., Strettoi, E., and Masland, R.H. (1998). The major cell populations of the mouse retina. J. Neurosci. Off. J. Soc. Neurosci. 18, 8936–8946. Jin, K. (2017). Transitional Progenitors during Vertebrate Retinogenesis. Mol. Neurobiol. 54, 3565–3576. Jin, M., Li, S., Moghrabi, W.N., Sun, H., and Travis, G.H. (2005). Rpe65 is the retinoid isomerase in bovine retinal pigment epithelium. Cell 122, 449–459. Jin, Z., Chau, M.D., and Bao, Z.-Z. (2006). Sema3D, Sema3F, and Sema5A Are Expressed in Overlapping and Distinct Patterns in Chick Embryonic Heart. Dev. Dyn. Off. Publ. Am. Assoc. Anat. 235, 163–169. Jones, C.A., London, N.R., Chen, H., Park, K.W., Sauvaget, D., Stockton, R.A., Wythe, J.D., Suh, W., Larrieu-Lahargue, F., Mukouyama, Y.S., et al. (2008). Robo4 stabilizes the vascular network by inhibiting pathologic angiogenesis and endothelial hyperpermeability. Nat. Med. 14, 448–453. Jongbloets, B.C., and Pasterkamp, R.J. (2014). Semaphorin signalling during development. Dev. Camb. Engl. 141, 3292–3297. Joyal, J.S., Sitaras, N., Binet, F., Rivera, J.C., Stahl, A., Zaniolo, K., Shao, Z., Polosa, A., Zhu, T., Hamel, D., et al. (2011). Ischemic neurons prevent vascular regeneration of neural tissue by secreting semaphorin 3A. Blood 117, 6024–6035. Jusuf, P.R., and Harris, W.A. (2009). Ptf1a is expressed transiently in all types of amacrine cells in the embryonic zebrafish retina. Neural Develop. 4, 34. Kalueff, A.V., Gebhardt, M., Stewart, A.M., Cachat, J.M., Brimmer, M., Chawla, J.S., Craddock, C., Kyzar, E.J., Roth, A., Landsman, S., et al. (2013). Towards a Comprehensive Catalog of Zebrafish Behavior 1.0 and Beyond. Zebrafish 10, 70–86. Kang, S., and Kumanogoh, A. (2013). Semaphorins in bone development, homeostasis, and disease. Semin. Cell Dev. Biol. 24, 163–171. Kannan, R., Sreekumar, P.G., and Hinton, D.R. (2016). Alpha Crystallins in the Retinal Pigment

236

Epithelium and Implications for the Pathogenesis and Treatment of Age-Related Macular Degeneration. Biochim. Biophys. Acta 1860, 258. Kase, S., He, S., Sonoda, S., Kitamura, M., Spee, C., Wawrousek, E., Ryan, S.J., Kannan, R., and Hinton, D.R. (2010). αB-crystallin regulation of angiogenesis by modulation of VEGF. Blood 115, 3398. Kaufman, R., Weiss, O., Sebbagh, M., Ravid, R., Gibbs-Bar, L., Yaniv, K., and Inbal, A. (2015). Development and origins of Zebrafish ocular vasculature. BMC Dev. Biol. 15, 18. Kawa, M.P., Machalinska, A., Roginska, D., and Machalinski, B. (2014). Complement System in Pathogenesis of AMD: Dual Player in Degeneration and Protection of Retinal Tissue. J. Immunol. Res. 2014. Kay, J.N., Finger-Baier, K.C., Roeser, T., Staub, W., and Baier, H. (2001). Retinal ganglion cell genesis requires lakritz, a Zebrafish atonal Homolog. Neuron 30, 725–736. Kay, J.N., Voinescu, P.E., Chu, M.W., and Sanes, J.R. (2011). NEUROD6 EXPRESSION DEFINES NOVEL RETINAL AMACRINE CELL SUBTYPES AND REGULATES THEIR FATE. Nat. Neurosci. 14, 965–972. Keegan, B.R., Meyer, D., and Yelon, D. (2004). Organization of cardiac chamber progenitors in the zebrafish blastula. Dev. Camb. Engl. 131, 3081–3091. Keegan, B.R., Feldman, J.L., Begemann, G., Ingham, P.W., and Yelon, D. (2005). Retinoic acid signalling restricts the cardiac progenitor pool. Science 307, 247–249. Kessler, O., Shraga-Heled, N., Lange, T., Gutmann-Raviv, N., Sabo, E., Baruch, L., Machluf, M., and Neufeld, G. (2004). Semaphorin-3F is an inhibitor of tumor angiogenesis. Cancer Res. 64, 1008–1015. Kholodenko, B.N. (2006). CELL SIGNALLING DYNAMICS IN TIME AND SPACE. Nat. Rev. Mol. Cell Biol. 7, 165–176. Kidd, T., Bland, K.S., and Goodman, C.S. (1999). Slit is the midline repellent for the robo receptor in Drosophila. Cell 96, 785–794. Kita, E.M., Bertolesi, G.E., Hehr, C.L., Johnston, J., and McFarlane, S. (2013). Neuropilin-1 biases dendrite polarization in the retina. Development 140, 2933–2941. Ko, J.-A., Minamoto, A., Sugimoto, Y., and Kiuchi, Y. (2016). Down-regulation of semaphorin 3F in rat retinal ganglion cells in response to optic nerve crush. Cell Biochem. Funct. 34, 378–384.

237

Krebs, W., and Krebs, I.P. (1988). Ultrastructural evidence for lymphatic capillaries in the primate choroid. Arch. Ophthalmol. Chic. Ill 1960 106, 1615–1616. Kruger, R.P., Aurandt, J., and Guan, K.-L. (2005). Semaphorins command cells to move. Nat. Rev. Mol. Cell Biol. 6, 789–800. Kubota, R., Hokoc, J.N., Moshiri, A., McGuire, C., and Reh, T.A. (2002). A comparative study of neurogenesis in the retinal ciliary marginal zone of homeothermic vertebrates. Brain Res. Dev. Brain Res. 134, 31–41. Kullander, K., and Klein, R. (2002). Mechanisms and functions of Eph and ephrin signalling. Nat. Rev. Cell Biol. 3, 475–486. Kumanogoh, A., and Kikutani, H. (2013). Immunological functions of the neuropilins and plexins as receptors for semaphorins. Nat. Rev. 13, 802–814. Kung, F., Wang, W., Tran, T.S., and Townes-Anderson, E. (2017). Sema3A Reduces Sprouting of Adult Rod Photoreceptors In Vitro. Invest. Ophthalmol. Vis. Sci. 58, 4318–4331. Kupperman, E., An, S., Osborne, N., Waldron, S., and Stainier, D.Y.R. (2000). A sphingosine-1- phosphate receptor regulates cell migration during vertebrate heart development. Nature 406, 192– 195. Kurihara, T., Westenskow, P.D., Bravo, S., Aguilar, E., and Friedlander, M. (2012). Targeted deletion of Vegfa in adult mice induces vision loss. J. Clin. Invest. 122, 4213–4217. Kutschera, S., Weber, H., Weick, A., De Smet, F., Genove, G., Takemoto, M., Prahst, C., Riedel, M., Mikelis, C., Baulande, S., et al. (2011). Differential endothelial transcriptomics identifies semaphorin 3G as a vascular class 3 semaphorin. Arterioscler. Thromb. Vasc. Biol. 31, 151–159. Laessing, U., and Stuermer, C.A. (1996). Spatiotemporal pattern of retinal ganglion cell differentiation revealed by the expression of neurolin in embryonic zebrafish. J. Neurobiol. 29, 65–74. Larsson, H.P. (2010). How is the heart rate regulated in the sinoatrial node? Another piece to the puzzle. J. Gen. Physiol. 136, 237–241. Lee, J., Cox, B.D., Daly, C.M.S., Lee, C., Nuckels, R.J., Tittle, R.K., Uribe, R.A., and Gross, J.M. (2012). An ENU mutagenesis screen in zebrafish for visual system mutants identifies a novel splice-acceptor site mutation in patched2 that results in Colobomas. Invest. Ophthalmol. Vis. Sci. 53, 8214–8221.

238

Legg, J.A., Herbert, J.M., Clissold, P., and Bicknell, R. (2008). Slits and Roundabouts in cancer, tumour angiogenesis and endothelial cell migration. Angiogenesis 11, 13–21. Lemke, G., and Reber, M. (2005). Retinotectal mapping: new insights from molecular genetics. Annu. Rev. Cell Dev. Biol. 21, 551–580. Leung, D.W., Cachianes, G., Kuang, W.J., Goeddel, D.V., and Ferrara, N. (1989). Vascular endothelial growth factor is a secreted angiogenic mitogen. Science 246, 1306–1309. Li, C., Huang, Z., Kingsley, R., Zhou, X., Li, F., Parke, D.W., and Cao, W. (2007). Biochemical alterations in the retinas of very low-density lipoprotein receptor knockout mice: an animal model of retinal angiomatous proliferation. Arch. Ophthalmol. Chic. Ill 1960 125, 795–803. Li, Q., Shirabe, K., Thisse, C., Thisse, B., Okamoto, H., Masai, I., and Kuwada, J.Y. (2005). Chemokine signalling guides axons within the retina in zebrafish. J. Neurosci. Off. J. Soc. Neurosci. 25, 1711–1717. Li, Y., Busoy, J.M., Zaman, B.A.A., Tan, Q.S.W., Tan, G.S.W., Barathi, V.A., Cheung, N., Wei, J.J.-Y., Hunziker, W., Hong, W., et al. (2018). A novel model of persistent retinal neovascularization for the development of sustained anti-VEGF therapies. Exp. Eye Res. 174, 98– 106. Li, Z., Hao, J., Duan, X., Wu, N., Zhou, Z., Yang, F., Li, J., Zhao, Z., and Huang, S. (2017). The Role of Semaphorin 3A in Bone Remodeling. Front. Cell. Neurosci. 11, 40. Liberatore, C.M., Searcy-Schrick, R.D., and Yutzey, K.E. (2000). Ventricular expression of tbx5 inhibits normal heart chamber development. Dev. Biol. 223, 169–180. Lin, Y.-F., Swinburne, I., and Yelon, D. (2012). Multiple influences of blood flow on cardiomyocyte hypertrophy in the embryonic zebrafish heart. Dev. Biol. 362, 242–253. Liu, J., and Stainier, D.Y.R. (2012). Zebrafish in the Study of Early Cardiac Development. Circ. Res. 110, 870–874. Liu, Q., Dalman, M.R., Sarmah, S., Chen, S., Chen, Y., Hurlbut, A.K., Spencer, M.A., Pancoe, L., and Marrs, J.A. (2011). CELL ADHESION MOLECULE CADHERIN-6 FUNCTION IN ZEBRAFISH CRANIAL AND LATERAL LINE GANGLIA DEVELOPMENT. Dev. Dyn. Off. Publ. Am. Assoc. Anat. 240, 1716–1726. Lu, X., Noble, F.L., Yuan, L., Jiang, Q., Lafarge, B.D., Sugiyama, D., Breant, C., Claes, F., Smet, F.D., Thomas, J.L., et al. (2004). The netrin receptor UNC5B mediates guidance events controlling

239

morphogenesis of the vascular system. Nature 432, 179–186. Luo, H., Jin, K., Xie, Z., Qiu, F., Li, S., Zou, M., Cai, L., Hozumi, K., Shima, D.T., and Xiang, M. (2012). Forkhead box N4 (Foxn4) activates Dll4-Notch signalling to suppress photoreceptor cell fates of early retinal progenitors. Proc. Natl. Acad. Sci. U. S. A. 109, E553-562. Lupo, G., Harris, W.A., and Lewis, K.E. (2006). Mechanisms of ventral patterning in the vertebrate nervous system. Nat. Rev. Neurosci. 7, 103–114. Lyons, G.E., Schiaffino, S., Sassoon, D., Barton, P., and Buckingham, M. (1990). Developmental regulation of myosin gene expression in mouse cardiac muscle. J. Cell Biol. 111, 2427–2436. Mahabeleshwar, G.H., Feng, W., Reddy, K., Plow, E.F., and Byzova, T.V. (2007). Mechanisms of integrin-vascular endothelial growth factor receptor cross-activation in angiogenesis. Circ. Res. 101, 570–580. Malicki, J. (1999). Development of the retina. Methods Cell Biol. 59, 273–299. Manuel, D.G., Leung, M., Nguyen, K., Tanuseputro, P., Johansen, H., and Canadian Cardiovascular Outcomes Research Team (2003). Burden of cardiovascular disease in Canada. Can. J. Cardiol. 19, 997–1004. Marques, S.R., and Yelon, D. (2009). Differential requirement for BMP signalling in atrial and ventricular lineages establishes cardiac chamber proportionality. Dev. Biol. 328, 472–482. Marques, S.R., Lee, Y., Poss, K.D., and Yelon, D. (2008). Reiterative roles for FGF signalling in the establishment of size and proportion of the zebrafish heart. Dev. Biol. 321, 397–406. Martinez-Morales, J.-R., Del Bene, F., Nica, G., Hammerschmidt, M., Bovolenta, P., and Wittbrodt, J. (2005). Differentiation of the vertebrate retina is coordinated by an FGF signalling center. Dev. Cell 8, 565–574. Marusich, M.F., Furneaux, H.M., Henion, P.D., and Weston, J.A. (1994). Hu neuronal proteins are expressed in proliferating neurogenic cells. J. Neurobiol. 25, 143–155. Marvin, M.J., Di Rocco, G., Gardiner, A., Bush, S.M., and Lassar, A.B. (2001). Inhibition of Wnt activity induces heart formation from posterior mesoderm. Genes Dev. 15, 316–327. Masai, I., Stemple, D.L., Okamoto, H., and Wilson, S.W. (2000). Midline signals regulate retinal neurogenesis in zebrafish. Neuron 27, 251–263. Masai, I., Yamaguchi, M., Tonou-Fujimori, N., Komori, A., and Okamoto, H. (2005). The hedgehog-PKA pathway regulates two distinct steps of the differentiation of retinal ganglion cells:

240

the cell-cycle exit of retinoblasts and their neuronal maturation. Development 132, 1539–1553. Matsuda, I., Fukaya, M., Nakao, H., Nakao, K., Matsumoto, H., Mori, K., Watanabe, M., and Aiba, A. (2010). Development of the somatosensory cortex, the cerebellum, and the main olfactory system in Semaphorin 3F knockout mice. Neurosci. Res. 66, 321–329. Matsuda, I., Shoji, H., Yamasaki, N., Miyakawa, T., and Aiba, A. (2016). Comprehensive behavioral phenotyping of a new Semaphorin 3 F mutant mouse. Mol. Brain 9, 15. McLeod, D.S., Grebe, R., Bhutto, I., Merges, C., Baba, T., and Lutty, G.A. (2009). Relationship between RPE and choriocapillaris in age-related macular degeneration. Invest. Ophthalmol. Vis. Sci. 50, 4982–4991. McLeod, D.S., Bhutto, I., Edwards, M.M., Silver, R.E., Seddon, J.M., and Lutty, G.A. (2016). Distribution and Quantification of Choroidal Macrophages in Human Eyes With Age-Related Macular Degeneration. Invest. Ophthalmol. Vis. Sci. 57, 5843–5855. Mecollari, V., Nieuwenhuis, B., and Verhaagen, J. (2014). A perspective on the role of class III semaphorin signalling in central nervous system trauma. Front. Cell. Neurosci. 8. Meeker, N.D., Hutchinson, S.A., Ho, L., and Trede, N.S. (2007). Method for isolation of PCR- ready genomic DNA from zebrafish tissues. BioTechniques 43, 610, 612, 614. Metcalfe, W.K., Myers, P.Z., Trevarrow, B., Bass, M.B., and Kimmel, C.B. (1990). Primary neurons that express the L2/HNK-1 carbohydrate during early development in the zebrafish. Dev. Camb. Engl. 110, 491–504. Mitchell, C.A., Rutland, C.S., Walker, M., Nasir, M., Foss, A.J., Stewart, C., Gerhardt, H., Konerding, M.A., Risau, W., and Drexler, H.C. (2006). Unique vascular phenotypes following over-expression of individual VEGFA isoforms from the developing lens. Angiogenesis 9, 209– 224. Mitchell, P., Liew, G., Gopinath, B., and Wong, T.Y. (2018). Age-related macular degeneration. The Lancet 392, 1147–1159. Molofsky, A.V., Kelley, K.W., Tsai, H.-H., Redmond, S.A., Chang, S.M., Madireddy, L., Chan, J.R., Baranzini, S.E., Ullian, E.M., and Rowitch, D.H. (2014). Astrocyte-encoded positional cues maintain sensorimotor circuit integrity. Nature 509, 189–194. Morlot, C., Thielens, N.M., Ravelli, R.B., Hemrika, W., Romijn, R.A., Gros, P., Cusack, S., and McCarthy, A.A. (2007). Structural insights into the Slit-Robo complex. Proc. Natl. Acad. Sci. U.

241

S. A. 104, 14923–14928. Murali, D., Yoshikawa, S., Corrigan, R.R., Plas, D.J., Crair, M.C., Oliver, G., Lyons, K.M., Mishina, Y., and Furuta, Y. (2005). Distinct developmental programs require different levels of Bmp signalling during mouse retinal development. Dev. Camb. Engl. 132, 913–923. Nakhai, H., Sel, S., Favor, J., Mendoza-Torres, L., Paulsen, F., Duncker, G.I.W., and Schmid, R.M. (2007). Ptf1a is essential for the differentiation of GABAergic and glycinergic amacrine cells and horizontal cells in the mouse retina. Dev. Camb. Engl. 134, 1151–1160. Neufeld, G., Sabag, A.D., Rabinovicz, N., and Kessler, O. (2012). Semaphorins in angiogenesis and tumor progression. Cold Spring Harb. Perspect. Med. 2, a006718. Neumann, C.J., and Nuesslein-Volhard, C. (2000). Patterning of the zebrafish retina by a wave of sonic hedgehog activity. Science 289, 2137–2139. Nguyen-Ba-Charvet, K.T., and Chedotal, A. (2002). Role of Slit proteins in the vertebrate brain. J. Physiol. Paris 96, 91–98. Nickla, D.L., and Wallman, J. (2010). THE MULTIFUNCTIONAL CHOROID. Prog. Retin. Eye Res. 29, 144–168. Obert, E., Strauss, R., Brandon, C., Grek, C., Ghatnekar, G., Gourdie, R., and Rohrer, B. (2017). Targeting the tight junction protein, zonula occludens-1, with the connexin43 mimetic peptide, αCT1, reduces VEGF-dependent RPE pathophysiology. J. Mol. Med. 95, 535–552. Oshima, Y., Oshima, S., Nambu, H., Kachi, S., Hackett, S.F., Melia, M., Kaleko, M., Connelly, S., Esumi, N., Zack, D.J., et al. (2004). Increased expression of VEGF in retinal pigmented epithelial cells is not sufficient to cause choroidal neovascularization. J. Cell. Physiol. 201, 393– 400. Paik, E.J., and Zon, L.I. (2010). Hematopoietic development in the zebrafish. Int. J. Dev. Biol. 54, 1127–1137. Park, K.W., Crouse, D., Lee, M., Karnik, S.K., Sorensen, L.K., Murphy, K.J., Kuo, C.J., and Li, D.Y. (2004). The axonal attractant Netrin-1 is an angiogenic factor. Proc. Natl. Acad. Sci. U. S. A. 101, 16210–16215. Park, K.W., Urness, L.D., Senchuk, M.M., Colvin, C.J., Wythe, J.D., Chien, C.B., and Li, D.Y. (2005). Identification of new netrin family members in zebrafish: developmental expression of netrin 2 and netrin 4. Dev. Dyn. Off. Publ. Am. Assoc. Anat. 234, 726–731.

242

de Pater, E., Clijsters, L., Marques, S.R., Lin, Y.-F., Garavito-Aguilar, Z.V., Yelon, D., and Bakkers, J. (2009). Distinct phases of cardiomyocyte differentiation regulate growth of the zebrafish heart. Dev. Camb. Engl. 136, 1633–1641. Pelster, B., and Burggren, W.W. (1996). Disruption of hemoglobin oxygen transport does not impact oxygen-dependent physiological processes in developing embryos of zebra fish (Danio rerio). Circ. Res. 79, 358–362. Pera, M.F., and Tam, P.P.L. (2010). Extrinsic regulation of pluripotent stem cells. Nature 465, 713–720. Peterson, S.M., and Freeman, J.L. (2009). RNA Isolation from Embryonic Zebrafish and cDNA Synthesis for Gene Expression Analysis. J. Vis. Exp. JoVE. Pitulescu, M.E., and Adams, R.H. (2010). Eph/ephrin molecules–a hub for signalling and endocytosis. Genes Dev. 24, 2480–2492. Plein, A., Calmont, A., Fantin, A., Denti, L., Anderson, N.A., Scambler, P.J., and Ruhrberg, C. (2015). Neural crest-derived SEMA3C activates endothelial NRP1 for cardiac outflow tract septation. J. Clin. Invest. 125, 2661–2676. Poon, K.-L., Liebling, M., Kondrychyn, I., Brand, T., and Korzh, V. (2016). Development of the cardiac conduction system in zebrafish. Gene Expr. Patterns 21, 89–96. Portugues, R., Haesemeyer, M., Blum, M.L., and Engert, F. (2015). Whole-field visual motion drives swimming in larval zebrafish via a stochastic process. J. Exp. Biol. 218, 1433–1443. Pradhan, A., Zeng, X.-X.I., Sidhwani, P., Marques, S.R., George, V., Targoff, K.L., Chi, N.C., and Yelon, D. (2017). FGF signalling enforces cardiac chamber identity in the developing ventricle. Development 144, 1328–1338. Proulx, K., Lu, A., and Sumanas, S. (2010). Cranial vasculature in zebrafish forms by angioblast cluster-derived angiogenesis. Dev. Biol. 348, 34–46. Provis, J.M. (2001). Development of the primate retinal vasculature. Prog. Retin. Eye Res. 20, 799–821. Raymond, P.A., Barthel, L.K., Bernardos, R.L., and Perkowski, J.J. (2006). Molecular characterization of retinal stem cells and their niches in adult zebrafish. BMC Dev. Biol. 6, 36. Raymond, P.A., Colvin, S.M., Jabeen, Z., Nagashima, M., Barthel, L.K., Hadidjojo, J., Popova, L., Pejaver, V.R., and Lubensky, D.K. (2014). Patterning the cone mosaic array in zebrafish retina

243

requires specification of ultraviolet-sensitive cones. PloS One 9, e85325. Regano, D., Visintin, A., Clapero, F., Bussolino, F., Valdembri, D., Maione, F., Serini, G., and Giraudo, E. (2017). Sema3F (Semaphorin 3F) Selectively Drives an Extraembryonic Proangiogenic Program. Arterioscler. Thromb. Vasc. Biol. 37, 1710–1721. Reifers, F., Walsh, E.C., Léger, S., Stainier, D.Y., and Brand, M. (2000). Induction and differentiation of the zebrafish heart requires fibroblast growth factor 8 (fgf8/acerebellar). Dev. Camb. Engl. 127, 225–235. Reiter, J.F., Alexander, J., Rodaway, A., Yelon, D., Patient, R., Holder, N., and Stainier, D.Y. (1999). Gata5 is required for the development of the heart and endoderm in zebrafish. Genes Dev. 13, 2983–2995. Reiter, J.F., Verkade, H., and Stainier, D.Y. (2001). Bmp2b and Oep promote early myocardial differentiation through their regulation of gata5. Dev. Biol. 234, 330–338. Reuer, T., Schneider, A.-C., Cakir, B., Bühler, A.D., Walz, J.M., Lapp, T., Lange, C., Agostini, H., Schlunck, G., Cursiefen, C., et al. (2018). Semaphorin 3F Modulates Corneal Lymphangiogenesis and Promotes Corneal Graft Survival. Invest. Ophthalmol. Vis. Sci. 59, 5277– 5284. Rezzola, S., Paganini, G., Semeraro, F., Presta, M., and Tobia, C. (2016). Zebrafish ( Danio rerio ) embryo as a platform for the identification of novel angiogenesis inhibitors of retinal vascular diseases. Biochim. Biophys. Acta BBA - Mol. Basis Dis. 1862, 1291–1296. Roberts, M.R., Srinivas, M., Forrest, D., Morreale de Escobar, G., and Reh, T.A. (2006). Making the gradient: thyroid hormone regulates cone opsin expression in the developing mouse retina. Proc. Natl. Acad. Sci. U. S. A. 103, 6218–6223. Rohr, S., Bit-Avragim, N., and Abdelilah-Seyfried, S. (2006). Heart and soul/PRKCi and nagie oko/Mpp5 regulate myocardial coherence and remodeling during cardiac morphogenesis. Development 133, 107–115. Rooijen, E. van, Voest, E.E., Logister, I., Bussmann, J., Korving, J., Eeden, F.J. van, Giles, R.H., and Schulte-Merker, S. (2010). von Hippel-Lindau tumor suppressor mutants faithfully model pathological hypoxia-driven angiogenesis and vascular retinopathies in zebrafish. Dis. Model. Mech. 3, 343–353. Rossi, A., Kontarakis, Z., Gerri, C., Nolte, H., Hölper, S., Krüger, M., and Stainier, D.Y.R. (2015).

244

Genetic compensation induced by deleterious mutations but not gene knockdowns. Nature 524, 230–233. Roth, L., Koncina, E., Satkauskas, S., Cremel, G., Aunis, D., and Bagnard, D. (2009). The many faces of semaphorins: from development to pathology. Cell. Mol. Life Sci. CMLS 66, 649–666. Rousseau, B., Dubayle, D., Sennlaub, F., Jeanny, J.C., Costet, P., Bikfalvi, A., and Javerzat, S. (2000). Neural and angiogenic defects in eyes of transgenic mice expressing a dominant-negative FGF receptor in the pigmented cells. Exp. Eye Res. 71, 395–404. Rousseau, B., Larrieu-Lahargue, F., Bikfalvi, A., and Javerzat, S. (2003). Involvement of fibroblast growth factors in choroidal angiogenesis and retinal vascularization. Exp. Eye Res. 77, 147–156. Sahay, A., Molliver, M.E., Ginty, D.D., and Kolodkin, A.L. (2003). Semaphorin 3F is critical for development of limbic system circuitry and is required in neurons for selective CNS axon guidance events. J. Neurosci. Off. J. Soc. Neurosci. 23, 6671–6680. Saint-Geniez, M., and D’Amore, P.A. (2004). Development and pathology of the hyaloid, choroidal and retinal vasculature. Int. J. Dev. Biol. 48, 1045–1058. Sakai, J.A., and Halloran, M.C. (2006). Semaphorin 3d guides laterality of retinal ganglion cell projections in zebrafish. Dev. Camb. Engl. 133, 1035–1044. Sakurai, A., Doci, C., and Gutkind, J.S. (2012). Semaphorin signalling in angiogenesis, lymphangiogenesis and cancer. Cell Res. 22, 23. Sapieha, P. (2012). Eyeing central neurons in vascular growth and reparative angiogenesis. Blood 120, 2182–2194. Sapieha, P., Sirinyan, M., Hamel, D., Zaniolo, K., Joyal, J.S., Cho, J.H., Honore, J.C., Kermorvant- Duchemin, E., Varma, D.R., Tremblay, S., et al. (2008). The succinate receptor GPR91 in neurons has a major role in retinal angiogenesis. Nat. Med. 14, 1067–1076. Sato, M., and Yost, H.J. (2003). Cardiac neural crest contributes to cardiomyogenesis in zebrafish. Dev. Biol. 257, 127–139. Sawamiphak, S., Seidel, S., Essmann, C.L., Wilkinson, G.A., Pitulescu, M.E., Acker, T., and Acker-Palmer, A. (2010). Ephrin-B2 regulates VEGFR2 function in developmental and tumour angiogenesis. Nature 465, 487–491. Scherz, P.J., Huisken, J., Sahai-Hernandez, P., and Stainier, D.Y.R. (2008). High-speed imaging

245

of developing heart valves reveals interplay of morphogenesis and function. Development 135, 1179–1187. Schmitt, E.A., and Dowling, J.E. (1994). Early eye morphogenesis in the zebrafish, Brachydanio rerio. J. Comp. Neurol. 344, 532–542. Schonthaler, H.B., Fleisch, V.C., Biehlmaier, O., Makhankov, Y., Rinner, O., Bahadori, R., Geisler, R., Schwarz, H., Neuhauss, S.C.F., and Dahm, R. (2008). The zebrafish mutant lbk/vam6 resembles human multisystemic disorders caused by aberrant trafficking of endosomal vesicles. Development 135, 387–399. Schultheiss, T.M., Burch, J.B., and Lassar, A.B. (1997). A role for bone morphogenetic proteins in the induction of cardiac myogenesis. Genes Dev. 11, 451–462. Scott, A., Powner, M.B., Gandhi, P., Clarkin, C., Gutmann, D.H., Johnson, R.S., Ferrara, N., and Fruttiger, M. (2010). Astrocyte-Derived Vascular Endothelial Growth Factor Stabilizes Vessels in the Developing Retinal Vasculature. PLOS ONE 5, e11863. Scott, C.A., Marsden, A.N., and Slusarski, D.C. (2016). Automated, high-throughput, in vivo analysis of visual function using the zebrafish. Dev. Dyn. Off. Publ. Am. Assoc. Anat. 245, 605– 613. Serini, G., and Tamagnone, L. (2015). Bad vessels beware! Semaphorins will sort you out! EMBO Mol. Med. 7, 1251–1253. Shah, R.S., Soetikno, B.T., Lajko, M., and Fawzi, A.A. (2015). A Mouse Model for Laser-induced Choroidal Neovascularization. J. Vis. Exp. JoVE. Shakib, M., and Cunha-Vaz, J.G. (1966). Studies on the permeability of the blood-retinal barrier. IV. Junctional complexes of the retinal vessels and their role in the permeability of the blood- retinal barrier. Exp. Eye Res. 5, 229–234. Sheldon, H., Andre, M., Legg, J.A., Heal, P., Herbert, J.M., Sainson, R., Sharma, A.S., Kitajewski, J.K., Heath, V.L., and Bicknell, R. (2009). Active involvement of Robo1 and Robo4 in filopodia formation and endothelial cell motility mediated via WASP and other actin nucleation-promoting factors. FASEB J. Off. Publ. Fed. Am. Soc. Exp. Biol. 23, 513–522. Shen, Y., and Raymond, P.A. (2004). Zebrafish cone-rod (crx) homeobox gene promotes retinogenesis. Dev. Biol. 269, 237–251. Sherris, D. (2007). Ocular drug development--future directions. Angiogenesis 10, 71–76.

246

Shui, Y.-B., Wang, X., Hu, J.S., Wang, S.-P., Garcia, C.M., Potts, J.D., Sharma, Y., and Beebe, D.C. (2003). Vascular endothelial growth factor expression and signalling in the lens. Invest. Ophthalmol. Vis. Sci. 44, 3911–3919. Simó, R., Villarroel, M., Corraliza, L., Hernández, C., and Garcia-Ramírez, M. (2010). The Retinal Pigment Epithelium: Something More than a Constituent of the Blood-Retinal Barrier— Implications for the Pathogenesis of Diabetic Retinopathy. Sinn, R., Peravali, R., Heermann, S., and Wittbrodt, J. (2014). Differential responsiveness of distinct retinal domains to Atoh7. Mech. Dev. 133, 218–229. Soldi, R., Mitola, S., Strasly, M., Defilippi, P., Tarone, G., and Bussolino, F. (1999). Role of alphavbeta3 integrin in the activation of vascular endothelial growth factor receptor-2. EMBO J. 18, 882–892. Sorrell, M.R.J., and Waxman, J.S. (2011). Restraint of Fgf8 signalling by retinoic acid signalling is required for proper heart and forelimb formation. Dev. Biol. 358, 44–55. Spilsbury, K., Garrett, K.L., Shen, W.-Y., Constable, I.J., and Rakoczy, P.E. (2000). Overexpression of Vascular Endothelial Growth Factor (VEGF) in the Retinal Pigment Epithelium Leads to the Development of Choroidal Neovascularization. Am. J. Pathol. 157, 135–144. Stainier, D.Y., Lee, R.K., and Fishman, M.C. (1993). Cardiovascular development in the zebrafish. I. Myocardial fate map and heart tube formation. Development 119, 31–40. Stenkamp, D.L. (2011). The Rod Photoreceptor Lineage of Teleost Fish. Prog. Retin. Eye Res. 30, 395–404. Stenkamp, D.L. (2015). Chapter Twenty-Three - Development of the Vertebrate Eye and Retina. In Progress in Molecular Biology and Translational Science, J.F. Hejtmancik, and J.M. Nickerson, eds. (Academic Press), pp. 397–414. Strasser, G.A., Kaminker, J.S., and Tessier-Lavigne, M. (2010). Microarray analysis of retinal endothelial tip cells identifies CXCR4 as a mediator of tip cell morphology and branching. Blood 115, 5102–5110. Stuermer, C.A. (1988). Retinotopic organization of the developing retinotectal projection in the zebrafish embryo. J. Neurosci. Off. J. Soc. Neurosci. 8, 4513–4530. Suetterlin, P., Marler, K.M., and Drescher, U. (2012). Axonal ephrinA/EphA interactions, and the emergence of order in topographic projections. Semin. Cell Dev. Biol. 23, 1–6.

247

Sun, X., Zhang, R., Lin, X., and Xu, X. (2008). Wnt3a regulates the development of cardiac neural crest cells by modulating expression of cysteine-rich intestinal protein 2 in rhombomere 6. Circ. Res. 102, 831–839. Sun, Y., Liegl, R., Gong, Y., Bühler, A., Cakir, B., Meng, S.S., Burnim, S.B., Liu, C.-H., Reuer, T., Zhang, P., et al. (2017a). Sema3f Protects Against Subretinal Neovascularization In Vivo. EBioMedicine 18, 281–287. Sun, Y., Liu, C.-H., Wang, Z., Meng, S.S., Burnim, S.B., SanGiovanni, J.P., Kamenecka, T.M., Solt, L.A., and Chen, J. (2017b). RORα modulates semaphorin 3E transcription and neurovascular interaction in pathological retinal angiogenesis. FASEB J. Off. Publ. Fed. Am. Soc. Exp. Biol. 31, 4492–4502. Suzuki, S.C., Bleckert, A., Williams, P.R., Takechi, M., Kawamura, S., and Wong, R.O.L. (2013). Cone photoreceptor types in zebrafish are generated by symmetric terminal divisions of dedicated precursors. Proc. Natl. Acad. Sci. U. S. A. 110, 15109–15114. Suzuki-Hirano, A., and Shimogori, T. (2009). The role of Fgf8 in telencephalic and diencephalic patterning. Semin. Cell Dev. Biol. 20, 719–725. Swaroop, A., Kim, D., and Forrest, D. (2010). Transcriptional regulation of photoreceptor development and homeostasis in the mammalian retina. Nat. Rev. Neurosci. 11, 563–576. Takahashi, H., Sakuta, H., Shintani, T., and Noda, M. (2009). Functional mode of FoxD1/CBF2 for the establishment of temporal retinal specificity in the developing chick retina. Dev. Biol. 331, 300–310. Takamatsu, H., and Kumanogoh, A. (2012). Diverse roles for semaphorin−plexin signalling in the immune system. Trends Immunol. 33, 127–135. Takeuchi, H., Inokuchi, K., Aoki, M., Suto, F., Tsuboi, A., Matsuda, I., Suzuki, M., Aiba, A., Serizawa, S., Yoshihara, Y., et al. (2010). Sequential arrival and graded secretion of Sema3F by olfactory neuron axons specify map topography at the bulb. Cell 141, 1056–1067. Tamagnone, L. (2012). Emerging role of semaphorins as major regulatory signals and potential therapeutic targets in cancer. Cancer Cell 22, 145–152. Tamagnone, L., Artigiani, S., Chen, H., He, Z., Ming, G.I., Song, H., Chedotal, A., Winberg, M.L., Goodman, C.S., Poo, M., et al. (1999). Plexins are a large family of receptors for transmembrane, secreted, and GPI-anchored semaphorins in vertebrates. Cell 99, 71–80.

248

Tata, M., Ruhrberg, C., and Fantin, A. (2015). Vascularisation of the central nervous system. Mech. Dev. 138, 26–36. Thomas, K.A. (1996). Vascular Endothelial Growth Factor, a Potent and Selective Angiogenic Agent. J. Biol. Chem. 271, 603–606. Tischer, E., Mitchell, R., Hartman, T., Silva, M., Gospodarowicz, D., Fiddes, J.C., and Abraham, J.A. (1991). The human gene for vascular endothelial growth factor. Multiple protein forms are encoded through alternative exon splicing. J. Biol. Chem. 266, 11947–11954. Toledano, S., Lu, H., Palacio, A., Ziv, K., Kessler, O., Schaal, S., Neufeld, G., and Barak, Y. (2016). A Sema3C Mutant Resistant to Cleavage by Furin (FR-Sema3C) Inhibits Choroidal Neovascularization. PloS One 11, e0168122. Törnquist, P., Alm, A., and Bill, A. (1990). Permeability of ocular vessels and transport across the blood-retinal-barrier. Eye Lond. Engl. 4 ( Pt 2), 303–309. Trinh, L.A., and Stainier, D.Y.R. (2004). Fibronectin regulates epithelial organization during myocardial migration in zebrafish. Dev. Cell 6, 371–382. Unoki, N., Murakami, T., Nishijima, K., Ogino, K., Rooijen, N. van, and Yoshimura, N. (2010). SDF-1/CXCR4 contributes to the activation of tip cells and microglia in retinal angiogenesis. Invest. Ophthalmol. Vis. Sci. 51, 3362–3371. Valdembri, D., Regano, D., Maione, F., Giraudo, E., and Serini, G. (2016). Class 3 semaphorins in cardiovascular development. Cell Adhes. Migr. 10, 641–651. Voronov, D.A., Alford, P.W., Xu, G., and Taber, L.A. (2004). The role of mechanical forces in dextral rotation during cardiac looping in the chick embryo. Dev. Biol. 272, 339–350. Walsh, E.C., and Stainier, D.Y.R. (2001). UDP-Glucose Dehydrogenase Required for Cardiac Valve Formation in Zebrafish. Science 293, 1670–1673. Walz, A., Feinstein, P., Khan, M., and Mombaerts, P. (2007). Axonal wiring of guanylate cyclase- D-expressing olfactory neurons is dependent on neuropilin 2 and semaphorin 3F. Dev. Camb. Engl. 134, 4063–4072. Wang, B., Xiao, Y., Ding, B.B., Zhang, N., Yuan, X., Gui, L., Qian, K.X., Duan, S., Chen, Z., Rao, Y., et al. (2003). Induction of tumor angiogenesis by Slit-Robo signalling and inhibition of cancer growth by blocking Robo activity. Cancer Cell 4, 19–29. Wang, J., Klysik, E., Sood, S., Johnson, R.L., Wehrens, X.H.T., and Martin, J.F. (2010a). Pitx2

249

prevents susceptibility to atrial arrhythmias by inhibiting left-sided pacemaker specification. Proc. Natl. Acad. Sci. U. S. A. 107, 9753–9758. Wang, Y., Nakayama, M., Pitulescu, M.E., Schmidt, T.S., Bochenek, M.L., Sakakibara, A., Adams, S., Davy, A., Deutsch, U., Luthi, U., et al. (2010b). Ephrin-B2 controls VEGF-induced angiogenesis and lymphangiogenesis. Nature 465, 483–486. Weber, I.P., Ramos, A.P., Strzyz, P.J., Leung, L.C., Young, S., and Norden, C. (2014). Mitotic position and morphology of committed precursor cells in the zebrafish retina adapt to architectural changes upon tissue maturation. Cell Rep. 7, 386–397. Werdich, X.Q., and Penn, J.S. (2005). Src, Fyn and Yes play differential roles in VEGF-mediated endothelial cell events. Angiogenesis 8, 315–326. West, X.Z., Meller, N., Malinin, N.L., Deshmukh, L., Meller, J., Mahabeleshwar, G.H., Weber, M.E., Kerr, B.A., Vinogradova, O., and Byzova, T.V. (2012). Integrin β3 crosstalk with VEGFR accommodating tyrosine phosphorylation as a regulatory switch. PloS One 7, e31071. White, D.T., Sengupta, S., Saxena, M.T., Xu, Q., Hanes, J., Ding, D., Ji, H., and Mumm, J.S. (2017). Immunomodulation-accelerated neuronal regeneration following selective rod photoreceptor cell ablation in the zebrafish retina. Proc. Natl. Acad. Sci. U. S. A. 114, E3719– E3728. Wilson, B.D., Ii, M., Park, K.W., Suli, A., Sorensen, L.K., Larrieu-Lahargue, F., Urness, L.D., Suh, W., Asai, J., Kock, G.A., et al. (2006). Netrins promote developmental and therapeutic angiogenesis. Science 313, 640–644. de Winter, F., Cui, Q., Symons, N., Verhaagen, J., and Harvey, A.R. (2004). Expression of class- 3 semaphorins and their receptors in the neonatal and adult rat retina. Invest. Ophthalmol. Vis. Sci. 45, 4554–4562. Wistow, G.J., and Piatigorsky, J. (1988). Lens Crystallins: The Evolution and Expression of Proteins for a Highly Specialized Tissue. Annu. Rev. Biochem. 57, 479–504. Wong, L., Power, N., Miles, A., and Tropepe, V. (2015). Mutual antagonism of the paired-type homeobox genes, vsx2 and dmbx1, regulates retinal progenitor cell cycle exit upstream of ccnd1 expression. Dev. Biol. 402, 216–228. Wong, W.L., Su, X., Li, X., Cheung, C.M.G., Klein, R., Cheng, C.-Y., and Wong, T.Y. (2014). Global prevalence of age-related macular degeneration and disease burden projection for 2020 and

250

2040: a systematic review and meta-analysis. Lancet Glob. Health 2, e106-116. Xia, J., Swiercz, J.M., Banon-Rodriguez, I., Matkovic, I., Federico, G., Sun, T., Franz, T., Brakebusch, C.H., Kumanogoh, A., Friedel, R.H., et al. (2015). Semaphorin-Plexin Signalling Controls Mitotic Spindle Orientation during Epithelial Morphogenesis and Repair. Dev. Cell 33, 299–313. Xuan, S., Baptista, C.A., Balas, G., Tao, W., Soares, V.C., and Lai, E. (1995). Winged helix transcription factor BF-1 is essential for the development of the cerebral hemispheres. Neuron 14, 1141–1152. Yamauchi, K., Mizushima, S., Tamada, A., Yamamoto, N., Takashima, S., and Murakami, F. (2009). FGF8 signalling regulates growth of midbrain dopaminergic axons by inducing semaphorin 3F. J. Neurosci. Off. J. Soc. Neurosci. 29, 4044–4055. Yelon, D., Horne, S.A., and Stainier, D.Y. (1999). Restricted expression of cardiac myosin genes reveals regulated aspects of heart tube assembly in zebrafish. Dev. Biol. 214, 23–37. Yelon, D., Ticho, B., Halpern, M.E., Ruvinsky, I., Ho, R.K., Silver, L.M., and Stainier, D.Y. (2000). The bHLH transcription factor hand2 plays parallel roles in zebrafish heart and pectoral fin development. Dev. Camb. Engl. 127, 2573–2582. Yu, H.H., and Moens, C.B. (2005). Semaphorin signalling guides cranial neural crest cell migration in zebrafish. Dev. Biol. 280, 373–385. Zacchigna, S., Almodovar, C.R. de, and Carmeliet, P. (2008). Similarities between angiogenesis and neural development: what small animal models can tell us. Curr. Top. Dev. Biol. 80, 1–55. Zhang, X.M., and Yang, X.J. (2001). Regulation of retinal ganglion cell production by Sonic hedgehog. Dev. Camb. Engl. 128, 943–957. Zhang, Y., Yang, Y., Trujillo, C., Zhong, W., and Leung, Y.F. (2012). The Expression of irx7 in the Inner Nuclear Layer of Zebrafish Retina Is Essential for a Proper Retinal Development and Lamination. PLOS ONE 7, e36145. Zhao, Y., Lin, M.-C., Farajzadeh, M., and Wayne, N. (2013). Early Development of the Gonadotropin-Releasing Hormone Neuronal Network in Transgenic Zebrafish. Front. Endocrinol. 4. Zhou, Y., Cashman, T.J., Nevis, K.R., Obregon, P., Carney, S.A., Liu, Y., Gu, A., Mosimann, C., Sondalle, S., Peterson, R.E., et al. (2011). Latent TGF-β binding protein 3 identifies a second heart

251

field in zebrafish. Nature 474, 645–648. Zhou, Z., Rao, J., Yang, J., Wu, F., Tan, J., Xu, S., Ding, Y., Zhan, N., Hu, X., Cui, Y., et al. (2015). SEMA3F prevents metastasis of colorectal cancer by PI3K–AKT-dependent downregulation of the ASCL2–CXCR4 axis. J. Pathol. 236, 467–478. Zou, J., Lathrop, K.L., Sun, M., and Wei, X. (2008). Intact RPE maintained by Nok is essential for retinal epithelial polarity and cellular patterning in zebrafish. J. Neurosci. Off. J. Soc. Neurosci. 28, 13684–13695.

252