Characterizing Retinal Stem Cell Proliferation and Differentiation Potentials

Kenneth N. Grisé

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Department of Molecular Genetics University of Toronto

Kenneth N. Grisé

Doctor of Philosophy

Department of Molecular Genetics

University of Toronto

2021 Abstract

A quiescent population of retinal stem cells (RSCs) in the adult mammalian eye retains the ability to proliferate and generate all retinal cell types when cultured in vitro. Thus, understanding the biological mechanisms that regulate RSC quiescence, proliferation and differentiation could lead to new possibilities for retinal endogenous repair or cell therapy.

Previous in vitro experiments in our lab have implicated two proteins, BMP and sFRP2, as potential mediators of RSC quiescence. Herein, I show that antagonism of BMP or sFRP2 proteins in the adult eye can induce the proliferation and expansion of RSCs in vivo and potentiate the effects of exogenous growth factors. Using genetic lineage tracing, I found that

RSCs express the gene Msx1. Moreover, in response to BMP/sFRP2 antagonism with growth factor stimulation, photoreceptor degeneration, or all combined, Msx1-lineage cells migrate from the RSC niche into the retina and express markers of mature retinal neurons, including photoreceptors.

To identify novel regulators of RSC proliferation, I developed a phenotypic drug screening platform and tested a library of small molecules with well-characterized molecular targets.

Screening identified synthetic glucocorticoid (GC) agonists as compounds that increase RSC

ii self-renewal and retinal stem and progenitor cell (RSPC) proliferation. GC agonists did not affect the differentiation profile of RSPCs, suggesting proliferation and differentiation are dissociable processes. In addition, injection of GC agonist, dexamethasone, into the adult mouse eye resulted in proliferation in the RSC niche.

Intrinsic and extrinsic signals can influence RSPC fate specification. Here we used a multifunctional antagonist of TGFβ, BMP and Wnt signaling – COCO -- to direct RSC progenitors to produce up to 100% cone photoreceptor progeny. Transcriptome profiling of endogenous cones and RSC-derived cones showed that both cell types are genetically similar and distinct from undifferentiated RSC progenitors. Also, we found COCO is required during the entire differentiation period to specify the cone fate. This suggests the cone fate is the default differentiation program of retinal progenitors, which is instructed via the absence of extrinsic cues.

Together, these results provide novel insights into the mechanisms regulating RSC quiescence, proliferation and differentiation in vitro and in vivo.

iii

Acknowledgements

I dedicate this thesis to the memory of my mom, Joan AKA Jzen, who I lost recently due to complications of ALS. I know she would have been very proud to see me complete this journey, during which she has been a limitless source of comfort and resolve. My mom radiated a profound and doting love for me and my brother that always has, and always will, suffuse my life with an unfaltering sense of gratitude. Her parting wisdom to us was to “always be true to yourself” and those were truly words she lived by. She had an unapologetically unique and indominable spirit, yet an extraordinarily gentle and kind heart. She also instilled in me an open mind, an understanding that not everything has an answer but to always keep asking questions. Thank you, mom. I will forever hold dear the love and light you brought to my life.

My PhD has been as much about learning from the trial and error of experimentation and reading literature, as it has been about learning from my exceptionally inspiring and insightful colleagues, friends and mentors along the way.

Foremost, I would like to thank my supervisor, Derek van der Kooy. Derek’s enthusiasm and aptitude for science has been a constant source of inspiration and motivation. His own passion and curiosity translate into his mentorship style, which encourages independent thought and exploration while also offering earnest critique and guidance. And when things have gotten tough, Derek has afforded me a tremendous amount of patience and understanding that has helped me push through. Derek has also made my time in his lab a wellspring of opportunity to develop into a well-rounded scientist through experiences outside of the lab – from enabling me to attend international conferences and engage with the global scientific community, being supportive of my pursuits as an educator when seeking TAships and guest lectures, to trusting me to represent the lab at meetings among CEOs in order to establish an academic-industry partnership. Derek, thank you for making my PhD such a rewarding experience, all about exploring open questions and open doors.

I would also like to thank my advisory committee members Vince Tropepe, Rod Bremner and David Kaplan. You provided me with an extremely effective balance of scientific challenge and guidance that has always made me feel encouraged and eager to redouble my efforts toward achieving my goals. Thank you for your mentorship and support over all these years.

To Matthias Steger and the Endogena crew: It has been a rare opportunity as a graduate student to participate in a biotech company as it is built from the ground up. I thoroughly enjoyed

iv working with all of you, even when balancing my PhD work and our collaboration had me a little overwhelmed at times. I hope the future holds many exciting achievements for you all and great success for the company.

To all the van der Kooy labbies that I have worked with day in and day out over the years, and all of my MoGen colleagues-cum-dear friends, you have been the single greatest source of moral support and encouragement (and perhaps on very rare occasions commiseration) that has kept me going. To my lab sis, Samantha Yammine, thanks for always looking out for me and making everyday lab life so much fun. We ended up pursuing so many endeavours/adventures together over the years, it really does feel like I had a sibling there supporting me along the way. To Brenda Coles, there would have been far less sanity and far more catastrophic failures if you weren’t always there to provide the guidance to keep things on track. Also, thanks for always being so enthusiastic to blast some tunes in the tissue culture room and help me complete an innumerable number of experiments. To my retinoids Saeed Khalili, Tahani Baakdhah, Justin Belair-Hickey and Brian Ballios: I have learned so much from you. You each bring your own unique approach and perspective to research that has truly expanded my… vision as a scientist. Justin, I can’t imagine more capable hands to be carrying the retinoid torch forward. Ahmed Fahmy, you’ve been a lovely labmate, a terrific travel buddy and I hear you’re pretty OK at being a father or something too. Krystal Jacques, it’s been fun collaborating on experiments, and I appreciate you for making me feel like I have at least some competence in the mentorship department. My main Moties, Geith Maal-Bared, Lyla El-Fayomi, and Reza Amirzadeh. Geith, whether in the lab, standing in front of an entire class explaining what they did wrong on their exam, or showing the world that science is a drag, you’ve truly enriched my experience these past years. Lyla, I was thoroughly impressed and enjoyed your debut novel and I look forward to reading all your future NYT bestselling Terminum series books. Hopefully we can soundboard more sci-fi story ideas in the future… but outside of the surgery room. Reza, it has been great fun diving into so many philosophical wormholes with you. Daniel Merritt, Glenn Wolfe and Isabel Mackay-Clackett: you didn’t think you’d worm your way out of this one, did you? Daniel, I don’t think there is a subject on which I wouldn’t be rapt to hear your thoughts. However, whether it is you or the whiskey talking I’m not entirely certain. Glenn, thanks for always giving your honest and insightful advice as the elder statesmen of the lab. Also, please give me your dog. Isabel, thanks for being a wonderful lab neighbour. I’ve really enjoyed all our chats, even if you probably haven’t enjoyed all my desk drum solos. I’d also like to thank all the undergrads who have enriched my time in the lab, especially Nelson

Bautista, whose thoughtfulness and work ethic made him a terrific collaborator and helped experiments run smoothly and successfully. To my extended lab fam – the former Zandstra crew – I cherish the time we’ve had together and your continuing friendships.

To my dear friends, of both the Penetanguishene and post-Penetanguishene era, who I consider my chosen family: I have been ridiculously fortunate to have so many wonderful human beings in my life. You are a huge part of the ups in my life and I wouldn’t have the resolve to get through the downs without you – including the challenges this PhD has entailed. I’m fortunate there are so many of you, but unfortunately that means I cannot list you all. Please know that your friendship and support has meant the world to me. I look forward to all that the future has in store for us. On that note, a special shoutout to the Gruvi Institute: I’m excited to see where our open minds might take us.

Jeff, Ryan and Kyle of Phantom Atlantic fame: these past few years have been a gift. This is not one of those endeavours that you look back on and realize how good you had it – I am continually thrilled by it and I get a sense of appreciation every single time we play music together. Thank you for all your hard work. I know full well that I owe the opportunity before me to how dedicated you all have been to see this dream through. And in that vein, thank you for your understanding, patience and support as I’ve worked away at this here thesis. I can’t tell you how optimistic I am for the opportunities that lay ahead. Also, to our unofficial tour manager/my roommate Saunders: it has been a crazy year, but we’ve managed to get through the great pandemic of 2020 with both our physical and mental health intact (so far). Thanks for all the supply runs and being so supportive as I’ve been toiling away.

And finally, the BIGGEST thanks to my amazing and supportive family. I constantly regale old friends and new acquaintances with stories of the family antics and wholesome fun that molded me into who I am today. Dad and Debbie, every now and then I try to convey how unbelievably fortunate I am to have you as parents, and then I get stuck in a paradox because the only word that seems to suffice is “ineffable”. Dad, if this is being read aloud to you, don’t worry I didn’t say that I’m stuck in a pair of docks. You have both shown unwavering belief in me and have always given me the freedom and support to follow my passions. Thank you from the bottom of my heart for all the sacrifices you’ve made to afford me such incredible opportunities in life. This PhD is only possible because of you and I love you dearly. Gordon, we’ve been through a lot together. In particular, I don’t know how I would have gotten through the last few years without you (PhD or otherwise). One silver lining is I feel we’re closer now than ever. But just know that I

vi plan to make life more difficult for you and Ayako by spoiling Ren at every turn. Heather, Melanie and Simon, I bet you knew you had something to do with the ‘family antics’ part, didn’t you? Heather and Melanie, I can reflect on several life lessons that I learned from you, which meant I got to benefit from not having to learn some things the hard way, while still growing as person. I really do tell people often how you helped me grow up to be a more well-rounded individual. And I’m sure I have a lot more to learn from you (and Brendan) now that I’ve seen what awesome individuals Sedona, Sam and Liam have grown up to be. Simon, I cherish that I was bestowed a best friend as a brother. You always made me want to be a better person and I’m pretty sure some of the strange things we did to entertain ourselves in our small town can technically be classified as scientific experiments (turnip bombs?). I think the sense of adventure, curiosity and creativity we shared growing up has a lot to do with why we both ended up pursuing science.

vii

Table of Contents

Acknowledgements………………………………..……………………………………….…………….iv

Table of Contents……………….…………………..………………………………….……………….viii

List of Tables……………….………………………………..…………………………………………..xv

List of Figures………………………………….………………………………………..……………....xvi

List of Appendices………………………………………………………………………….………..….xx

List of Abbreviations………………………………………………………………………….…………xxi

Chapter 1 General Introduction………………………………….…………………………………...... 1

1.1 Eye Anatomy & Function……………………………………………………………………...……3

1.1.1 Gross eye anatomy and function…………………………..……………………………3

1.1.2 The Mammalian Retina..………………………………………….………………….…..7

1.1.3 Photoreceptors and Phototransduction…………..………………………………..…...9

1.2 Vertebrate Eye Development & Retinogenesis………………………………………………....11

1.2.1 Establishing the eye field………………………………..…………..……….…………12

1.2.2 Splitting the eye field..……….…………………………….….……………………...…14

1.2.3 Optic vesicle formation……………………………………….…………………………15

1.2.4 Lens and cornea development...…………………….…………………….…………..15

1.2.5 Optic cup formation & patterning…………………………………………..…………..16

1.2.6 Ciliary body & iris development (the origin of retinal stem cells)……………………19

1.2.7 Retinogenesis………………………………………………………………………….…20

viii

1.2.8 Photoreceptor Development…………………………………………………………..22

1.3 The ciliary marginal zone and retinal stem cells of non-mammalian vertebrates……………23

1.4 Retinal stem cells in the adult mammalian ciliary epithelium…………. ………………………25

1.4.1 Differential potential of retinal stem cells………………………………………..…….27

1.4.2 Proliferation potential of retinal stem cells…………………………………….………29

1.4.3 Regenerative potential of retinal stem cells in vivo…………………………….…….31

1.4.4 Other potential adult stem cells in the mammalian retina……….....……………..…34

1.6 Thesis Hypotheses & Aims….……………………………………………………………….……36

Chapter 2 Reactivation of proliferation and neurogenesis in the adult mammalian retinal stem cell niche via antagonism of BMP and sFRP2……………………………………………………….39

2.1 Abstract…………………………………………………………………...……………………..…..39

2.2 Introduction…………………………………………………………………………….……………40

2.3 Materials & Methods………………………………….………………..…………………………..41

2.3.1 Mice………………………………………………………………..…………..…………41

2.3.2 Drug and Protein Preparations……………………………………….…………..……42

2.3.3 Intravitreal Injections…………………………………………………………..………..42

2.3.4 Immunohistochemistry………………………………………………………….………43

2.3.5 Isolation of Retinal Stem Cells from the Ciliary Epithelium of the Adult Eye and Primary Clonal Sphere Assay………………………..………………………….……………44

2.3.6 Fluorescence Assisted Cell Sorting…………………..…………………….…………45

2.3.7 Statistical Analysis…………………………………………………....…………………46

2.4 Results……………………………………………………………………………………………….46

2.4.1 Intravitreal injection of Noggin or anti-sFRP2 stimulates ciliary body-specific proliferation and expands the retinal stem cell population………………..…………….….46

2.4.2 Combinatorial injections of Noggin, anti-sFRP2, FGF2 and Insulin have differential effects on CE proliferation versus retinal stem cell expansion…...... ….52

2.4.3 Inducible Msx1-CreERT2 mouse lineage labeling marks the adult ciliary epithelium and retinal stem cells…………………………………………………………………………..56

2.4.4 FINS-mediated CE proliferation is potentiated by photoreceptor degeneration in Msx1-CreERT2;Rosa26-tdTomato mice………………………………….……………………60

2.4.5 Photoreceptor degeneration and FINS treatment induce CE cell migration into the neural retina………………………………………..……………………………………………62

2.4.6 Ciliary epithelium derived Msx1-TdTomato+ cells in the neural retina express photoreceptor or retinal ganglion cell markers………..…………….……………...……….65

2.5 Discussion………………………………………………..………………………………………….68

Chapter 3: Glucocorticoid agonists enhance retinal stem cell self-renewal and proliferation in vitro and stimulate proliferation of the ciliary epithelium in the mouse eye in vivo……………….72

3.1 Abstract………………………………………………………………………………………………72

3.2 Introduction………………………………….………………………………………………………73

3.3 Materials & Methods……………………………………………………...………………………..74

3.3.1 Mice……………………………………………………………………………………….74

3.3.2 Isolation of Retinal Stem Cells from the Ciliary Epithelium of the Adult Eye and Primary Clonal Sphere Assay……………………………………………………………….75

3.3.3 Human Eye Procurement and Dissection………………….…………………………76

3.3.4 Mouse pancreatic multipotent progenitor isolation and sphere assay…...... …76

3.3.5 Sphere Passaging…………………………...…………………………….…………….77

3.3.6 Medium-Throughput Screening Pipeline………………………………………………………….…………….…………………77

3.3.7 Medium-throughput and Content Imaging…………………………………………………………….……………………….……78

3.3.8 Proliferation and Cell Death Assays…………………….…………..…………………79

3.3.9 Differentiation Assay………………………………….………………………...……….79

3.3.10 Immunohistochemistry and immunocytochemistry…………….…………...………80

3.3.11 Intravitreal Injections…………………………………….…….……………………….80

3.3.12 Statistical Analysis………………………….………………………………………….81

3.4 Results……………………………………………………………………………………………….82

3.4.1 Medium-throughput screening identifies several unique compound classes that expand retinal stem and progenitor cells in culture……………...…………………………82

3.4.2 Glucocorticoid agonists stimulate mouse retinal stem and progenitor cell proliferation in vitro via both glucocorticoid receptor and mineralocorticoid receptor signaling………………………………………………………………………………...….……85

3.4.3 Glucocorticoid agonism has differential effects on proliferation and self-renewal of adult stem and progenitor cells from different tissues……………...... ….88

3.4.4 Glucocorticoid agonism does not change the differentiation profile of retinal stem cell progeny…………………...…………………..……………………………………….……91

3.4.5 Glucocorticoid agonism in vivo induces proliferation in the ciliary epithelium of the mouse eye but does not expand the retinal stem cell population…………………………94

3.5 Discussion……………………………………………………………………………...……………97

Chapter 4: Induction of rod versus cone photoreceptor-specific precursors from retinal stem and progenitor cells…...…...………………………………………………..…………………...…………102

4.1 Abstract…………………………………………….………………………………………………102

4.2 Introduction………………………………...………………………………………………………103

4.3 Materials & Methods………………………………………………………………………………105

4.3.1 Animals………………………….………………………………………………………105

4.3.2 Cell Culture………………………...……………………………………………………105

4.3.3 Differentiation…………………...………………………………………………………105

4.3.4 Immunostaining…………………………………………………………………………106

4.3.5 Retroviral Clonal Labeling……………..………………………………………………106

4.3.6 Flow cytometry, sorting and labeling cells ……………..……………………………107

4.3.7 Quantitative RT-PCR cells………………….…………………………………………107

4.3.8 RNA sequencing……………………..…………………………………………………107

4.3.9 Correlation at pathway activity level …………………………………………………108

4.3.10 Determining differentially expressed (DE) genes and ranks…………..…………108

4.3.11 Lists of differentially expressed genes…………..…………………………………108

4.3.12 Cell counts and statistics…………………………………….………………………109

4.4 Results……………………………………………………………………………………………..109

4.4.1 RSC progeny can be biased towards rod or cone fates……….……..……………109

4.4.2 T+RA instruct RSC progeny to rods and COCO to cones…………………………112

xii

4.4.3 Fetal and adult progenitors respond identically……………………………..………116

4.4.4 Sonic hedgehog regulates proliferation in the early progenitor expansion phase of RSC progeny differentiation…………………………………………………………….……118

4.4.5 T+ RA act on early retinal progenitors, while COCO is required throughout the differentiation………………………………………………………….…………………….…120

4.4.6 RSC-derived cones exhibit similar transcriptomes to those of endogenous cones………………………………………………………..…………………………….……122

4.5 Discussion………………………………………………………………………………………126

Chapter 5 Discussion…………………………..………..……………………………………………129

5.1 Summary & Conclusions………………….………………………………………..…………….129

5.2 Discussion & Future Directions……………….…………………………………………………132

5.2.1 Chapter 2……………………………………………..…………………………………132

5.2.1.1 Downstream mechanisms of BMP & sFRP2 antagonism………...…….132

5.2.1.2 Does Msx1 Regulate RSC Quiescence?...... 134

5.2.1.3 Assessing Migration & Differentiation of CE-derived Cells in the Retina……………………………………………………………………….…………136

5.2.1.4 Other Variables and Open Questions toward Activating Endogenous RSCs……………………………………..…………………………………..………..137

5.2.2 Chapter 3………………………………………………………..………………………138

5.2.2.1 Glucocorticoid Mechanisms and Molecular Pathways……………..……138

5.2.2.2 Investigating Other Hit Compounds and Future Screens……………….139

5.2.2.3 Effect of Glucocorticoids on Human RSCs/Ciliary Epithelium………….140

xiii

5.2.3 Chapter 4………………………………………………………………………..………141

5.2.3.1 The molecular mechanisms of COCO in cone specification…….……..141

5.2.3.2 Investigating Novel Cone Genes and Future “omics” Approaches..….142

5.2.3.3 Functional Characterization of RSC-derived Cones………….………...143

5.2.4 Combined Insights and Future Directions…………………………………………..145

References…………………………………….……………………………………………………….147

Appendix A – Chapter 2 Supplementary Information………………………………………….….182

Appendix B – Chapter 3 Supplementary Information……………………………………………..211

Appendix C – Chapter 4 Supplementary Information……………………………………………..220

xiv

List of Tables

Table A1. Genotyping primer list…………………………………………………….……………….210

Table B1. Compounds that met hit criteria in at least one of two screens………………………211

Table B2. Screening quality metrics……………………………………...…………………….……213

Table C1. The top 20 differentially express genes in RSC spheres vs. RSC-derived cones….220

Table C2. The top 20 differentially express genes in RSC spheres vs. endogenous cones….222

Table C3. The top 20 differentially express genes in endogenous cones vs. RSC-derived cones. ………………………………………………………….……………………………………..…………224

List of Figures

Figure 1.1 Anatomy of the human and mouse eye………………………………………….………..6

Figure 1.2 Anatomical organization of the vertebrate retina………….…………….………………10

Figure 1.3 Schematic view of a developing vertebrate eye: from late gastrula to the optic cup..12

Figure 1.4 Optic cup formation and patterning……………………………………………………….18

Figure 1.5 Chronological order and transcriptional regulation of retinal cell birth………………..21

Figure 2.1 Intravitreal injection of Noggin or anti-sFRP2 stimulates proliferation in the ciliary epithelium and increases primary sphere-forming retinal stem cell number…………………...…50

Figure 2.2 Combinatorial injection of Noggin and anti-sFRP2 with and without growth factors have differential effects on CE proliferation and retinal stem cell expansion…………………..…54

Figure 2.3 Tamoxifen induction of reporter expression in Msx1-CreERT2;Rosa26-TdTomato mice labels the ciliary epithelium and CE-derived RSC spheres……….………………………………..58

Figure 2.4 FINS-mediated CE proliferation is augmented by photoreceptor degeneration in Msx1-CreERT2;Rosa26-TdTomato mice……………………………………………………………….61

Figure 2.5 Figure 2.5. Photoreceptor degeneration or FINS treatment induces ciliary epithelial cell migration into the neural retina………………………………………….…….…………………..64

Figure 2.6 Ciliary epithelium derived TdTomato+ cells in the neural retina express photoreceptor or retinal ganglion cell markers……………………………………………………………..…………67

Figure 3.1 Medium-throughput screening identifies several unique compound classes that increase retinal stem and progenitor cell number……………………………………………….…..84

Figure 3.2 Glucocorticoid agonists increase retinal stem and progenitor cell proliferation through glucocorticoid receptor and mineralocorticoid receptor signaling in mouse………………………87

Figure 3.3 Dexamethasone increases retinal stem cell sphere size, number and self-renewal but inhibits growth and insulin expression of pancreatic multipotent progenitor spheres……………90

xvi

Figure 3.4 Dexamethasone does not affect the differentiation profile of retinal stem cell progeny……………..………………………………………………………………………………...….93

Figure 3.5 Intravitreal dexamethasone injection induces ciliary epithelium proliferation but does not expand the retinal stem cell population in vivo……………………………………………….….96

Figure 4.1 Taurine and retinoic acid do not affect RSC sphere derivation from adult ciliary epithelium but do shift baseline rod differentiation potential. COCO increases cone differentiation from RSC progeny………………………………………………………………..…..111

Figure 4.2 Distribution of retinal progenitor clones in retroviral lineage tracing shows a bias towards mixed clones of high percentage rhodopsin+cells when differentiated in taurine and retinoic acid…………………………………………………………………………………………….115

Figure 4.3 Taurine and retinoic acid act instructively to generate lineage-restricted rod-specific progenitors while COCO suppresses other retinal cell fates and permits only cone photoreceptor differentiation…...………………………………………………………………………………………117

Figure 4.4 E14 RSC or neural retinal progenitor spheres exhibit similar differentiation and rod- lineage priming effects of taurine and retinoic acid, and E14 neural retina progenitor spheres show similar cone differentiation patterns in COCO to RSC progeny……………………………119

Figure 4.5 Sonic hedgehog regulates proliferation in the early progenitor expansion phase of RSC progeny differentiation in T+RA…………………………………………………………….….121

Figure 4.6Taurine/retinoic acid act instructively on early retinal progenitors to bias rod differentiation, while COCO is required throughout the differentiation period in order for RSC progeny to develop cone phenotypes with similar transcriptomes to endogenous cones. Rod differentiation outcompetes cone differentiation…………………………………………………....124

Figure 4.7 Alternative models of directed photoreceptor differentiation from RSC progeny…..126

Appendices

Figure A1 Some EdU+ cells in the ciliary epithelium and neural retina co-labeled for endothelial cell marker ERG…………………….……………………………………………………...………….183

xvii

Figure A2 Some EdU+ cells in the ciliary epithelium and neural retina co-labeled for microglia/macrophage cell marker CD68………………………………………..………………….185

Figure A3 Modulation of downstream BMP and Wnt signaling mediates CE proliferation……………………………………………………………………………………..………187

Figure A4 Day 4 multichannel images of the representative IHC images in Figure 2………....189

Figure A5 Day 31 multichannel images of the representative IHC images in Figure 2………..191

Figure A6 Inducible Msx1-CreERT2 mouse lineage labeling in mice with and without tamoxifen exposure and in numerous tissues in the adult mouse eye……………………………….……...193

Figure A7 Initial FACS gating to select live cells and the stop codon genotype of post-FACS RSC spheres…………………………………….…………………………………………………..…195

Figure A8 N-methyl-N-nitrosourea induces tunable photoreceptor degeneration………………197

Figure A9 PBS injected eyes with MNU injury occasionally developed gross phthisis bulbi pathology whereas FINS injected eyes with MNU injury did not………………………………....199

Figure A10 Multichannel images of the representative Msx1 and Pax6 IHC images in Figure 6……………………………………………………………………………….…………………………200

Figure A11 The Msx1-TdTomato expression domain of naïve eyes at 1 day and 45 days after tamoxifen induction……………………….……………………………………………..…………….202

Figure A12 Multichannel images of the representative IHC images in Figure 6……………………………………………………………………………………….…………………204

Figure A13 Multichannel images of the representative IHC images in Figure 6…………………………………………………………………………………………….……………206

Figure A14 Msx1-TdTomato+ cells were detected in the RPE layer at very low frequency……………………………………………………………………………………………….208

Figure B1 Visual confirmation that glucocorticoid agonists enhanced retinal stem and progenitor yield and was not due to artifacts……………………………………….…………………………...214

xviii

Figure B2 Representative staining of cell death marker ethidium homodimer (EthD-1) and thymidine analog EdU………………………………………….……………………………………..215

Figure B3 Glucocorticoid agonists increase retinal stem and progenitor cell proliferation only through mineralocorticoid receptor signaling in human………………………………………..….216

Figure B4 Intravitreal dexamethasone injection induces ciliary epithelium proliferation……………………………………………………………………………………………..217

Figure B5 EdU-positive cells co-label with endothelial and microglia/macrophage markers…………………………………………………………………………………………………218

Figure B6 Glucocorticoid receptor, Mineralocorticoid receptor, and 11-β-HSD1 & 2 RNA expression in RSC spheres…………………………………………………….…………………….219

Figure C1 COCO makes cones but not other cell types………………………………………..…226

Figure C2 Flow cytometry and sorting of pigmented and non-pigmented progenitors, as well as endogenous cones and RSC derived cones…………………………………………….………….228

Figure C3 Correlation of RSC-derived cones with reference cones at pathway activity level……………………………………………………………………………………………………..229

xix

List of Appendices

Appendix A. Chapter 2 Supplementary Information……………………………………………….182

Appendix B. Chapter 3 Supplementary Information……………………………………………….211

Appendix C. Chapter 4 Supplementary Information……………………………………………….220

List of Abbreviations

ATP Adenosine triphosphate bHLH Basic helix-loop-helix KO Knockout BMP Bone morphogenetic protein MC Mineralocorticoid BrdU 5-bromo-2'-deoxyuridine mESC Mouse embryonic stem cell CB Ciliary body MG Muller glia CE Ciliary epithelium MIP Mouse indulin promoter cGMP Cyclic guanosine monophosphate MNU N-methyl-N-nitrosourea CM Conditioned media MR Mineralocorticoid receptor CMZ Ciliary marginal zone MTS Medium-throughput screening CNGC Cyclic nucleotide gated channel NC Naïve control CNS Central nervous system NR Neural retina DEX Dexamethasone OICR Ontario Institiue for Cancer DMSO Dimethylsulfoxide Research EdU 5-ethynyl-2’-deoxyuridine ONL Outer nuclear layer EFTF Eye field transcription factor OPL Outer plexiform layer EGF Epidermal growth factor OS Outer segment ERG Electroretinogram PBS Phosphate buffered saline EthD-1 Ethidium homodimer pCE Presumptive ciliary epithelium EtOH Ethanol PCR Polymerase chainreaction FACS Fluorescence-activated cell sorting PDE Phosphodiesterase FBS Fetal bovine serum PEDF Pigment epithelium-derived factor FGF Fibroblast growth factor PMP Pancreatic multipotent progenitor FH FGF+Heparin PND Postnatal day FINS FGF+Insulin+Noggin+anti-sFRP2 POM Periocular mesenchyme GC Glucocorticoid PRED Prednisolone GCL Ganglion cell layer RA Retinoic acid GFP Green fluorescent protein RGC Retinal ganglion cell GR Glucocorticoid receptor RPC Retinal progenitor cell GTP Guanosine triphosphate RPE Retinal pigmented epithelium HC Hydroxycortisone RSC Retinal stem cell HDAC Histone deacetylase RSPC Retinal stem/progenitor cell hESC Human embryonic stem cell scRNA-seq single cell RNA sequencing Hh Hedgehog SFM Serum-free media HSD Hydroxysteroid dehydrogenase sFRP Secreted frizzled-related protein ICC Immunocytochemistry Shh Sonic hedgehog IHC Immunohistochemistry T Taurine INL Inner nuclear layer T3 Triiodothyronine IPE Iris pigmented epithelium TCL Tool compound library IPL Inner plexiform layer TGF Transforming growth factor ipRGC Intrinsically photoresponsive retinal TRβ2 Thyroid hormone β2 ganglion cell YFP Yellow fluorescent protein

xxi

Chapter 1 Introduction

1. General Introduction

The relative simplicity and accessibility of the retina has made it one of the most studied neural tissues across a breadth of disciplines, such as developmental biology, anatomy, physiology and metabolism (Cepko et al., 1996). For these same reasons, the retina has long been considered a good model system to study neural diseases. Yet, despite the longstanding and in- depth study of retinal biology, degenerative diseases of the retina currently afflict hundreds of millions of people and represent a significant, and growing, global health and economic burden (Gasparini et al., 2019). In fact, two of the most common forms of retinal degeneration, age- related macular degeneration (AMD) and retinitis pigmentosa (RP), are presently estimated to affect 196 million and 1.5 million people worldwide, respectively (O’Neal and Luther, 2020; Wong et al., 2014). Both diseases are characterized by the progressive dysfunction and loss of photoreceptors, while AMD can also involve retinal pigmented epithelial (RPE) cell death (Fritsche et al., 2014; Hartong et al., 2006). The ensuing vision loss is permanent as there is scant evidence that the mammalian retina can undergo intrinsic regeneration or functional recovery (Berry et al., 2008; Kubota et al., 2002; Wohl et al., 2012). However, while some remodelling and gliosis does occur in the remaining retinal layers after photoreceptor or RPE degeneration, the majority of the circuitry and architecture remains intact (Jones et al., 2003). This led to the hypothesis that transplantation of healthy photoreceptors and/or RPE cells into the degenerated retina could restore visual function.

In recent years, stem cells have become the most promising source of retinal cells for transplantation therapies. This is due to the two cardinal properties of stem cells: self-renewal and differentiation. In this thesis, stem cells, progenitor cells and precursor cells describe distinct stages of lineage progression that have discrete functional definitions. Both stem cells and progenitor cells can proliferate. When stem cells divide, they can generate stem cells (self- renewal) or progenitor cells (differentiation) as progeny. When progenitor cells divide, they can generate progenitor cells (self-renewal) or post-mitotic precursor cells (differentiation). Proliferating stem cells will always self-renew, and thus, undifferentiated stem cells can persist in a tissue for the entire lifetime of the organism (Seaberg and van der Kooy, 2003). In contrast, if progenitor cells are allowed to proliferate continuously, they will eventually transition into their

1 downstream lineal cell type, precursor cells, or die. Thus, stem cells have greater self-renewal capacity than progenitor cells. Precursor cells, which cannot proliferate, differentiate into the specialized cell types that make up a tissue, such as photoreceptors and RPE. How and when cell fate is specified during the progression from stem cells to specialized cell types is discussed in Chapter 1.2.7-8, Chapter 1.4. and Chapter 4.

The ability to continuously self-renew makes stem cells a potentially unlimited cell source, while various differentiation potencies (totipotent, pluripotent, multipotent, etc.) renders them capable of generating specialized cell types of various lineages and tissues (Rao et al., 2017; Shen, 2020; West et al., 2020). Pluripotent stem cells and fetal retinal progenitors have shown promise as sources of photoreceptors and RPE for retinal cell therapy, with some clinical trials even reporting modest improvements in visual function in humans (Liu et al., 2017; Mandai et al., 2017; Schwartz et al., 2015; Singh et al., 2020). However, relatively few therapies have actually made it into the clinic and most blinding diseases still have no cure (Roska and Sahel, 2018). Furthermore, there are serious drawbacks that complicate the clinical application of these cell sources, such as the limited supply of fetal donors cells, ethical concerns associated with the use of embryonic stem cells (ESCs) and fetal cells, the potential oncogenicity of ESCs and induced pluripotent stem cells (iPSCs), as well as genetic aberrations caused by reprogramming of iPSCs (Ballios and van der Kooy, 2010; Locker et al., 2010; Ramsden et al., 2013). However, many of these limitations do not apply to tissue-specific, adult retinal stem cells (RSCs), which have shown promise in pre-clinical experiments as a source of retinal cells for transplantation (Ballios et al., 2010, 2015; Coles et al., 2004).

Adult RSCs have significant self-renewal capacity and multipotential differentiation capacity that is restricted to the retinal lineage. In fact, RSCs can differentiate into all the cell types of the retina, including Müller glia and RPE (Coles et al., 2004; Inoue et al., 2010). Thus, not only did the discovery of RSCs raise new questions about mammalian retinal developmental, it presented a new source of retinal cells for transplantation therapy and introduced the possibility of endogenous retinal regeneration in the mammalian eye (Ahmad et al., 2000; Tropepe et al., 2000). However, adult RSCs are quiescent in vivo and are not believed to contribute new cells to the retina under homeostatic conditions or in response to injury (Kubota et al., 2002). Thus, so far, RSCs have mainly been studied in vitro as a means to investigate retinal development and for their potential as a source of retinal progenitors and photoreceptors for cell transplant therapies. In this thesis, I will discuss the groundwork that has led up to our current knowledge of RSC biology and present new research that aims to further our understanding of the

2 mechanisms that regulate RSC quiescence, proliferation and differentiation – and the implications thereof for retinal cell therapy and endogenous regeneration.

1.1 Anatomy and Function of the Mammalian Eye

1.1.1 Gross Eye Anatomy & Function

The mammalian eye has three gross anatomical layers: the external layer, the intermediate layer and the inner layer. The external layer consists of the cornea and the sclera, the intermediate layer (uveal tract) is divided into anterior (iris, ciliary body) and posterior (choroid vasculature) sections, while the inner layer is the sensory tissue of the eye, the retina (Figure 1.1). There are also three fluid chambers within the eye: the anterior chamber (between the cornea and the iris), the posterior chamber (between the iris and the anterior lens and zonula fibers that connect the lens to the ciliary body), and the vitreous chamber (between the lens and the retina). Thus, the lens acts as a barrier that separates the aqueous and vitreous chambers (Kolb et al., 2020)

The sclera, or “white of the eye”, is dense connective tissue composed of mainly type 1 collagen fibers, oriented in different directions. The lack of parallel orientation of collagen fibers gives the sclera its white appearance (Pradeep et al., 2020). The sclera provides the “globe” shape and forms the supporting wall that gives resistance to the eyeball. Inserted into the sclera are three pairs of muscles. The two pairs of rectus muscles run straight to the orbit of the skull, orthogonal to each other (the superior rectus, the inferior rectus, the lateral rectus, and the medial rectus muscles). Another pair of muscles, the oblique muscles (superior oblique and inferior oblique), are angled, as the name implies, obliquely. Together, these extraocular muscles, rotate the eyeball in the orbits and allow the light entering the eye to remain focused on the fovea of the central retina (Kolb et al., 2020).

The processing of light in the vertebrate visual system begins with the anterior-most tissue of the eye, the cornea. The cornea consists of type I collagen fibers oriented in a uniform, parallel direction, which results in its transparency and provides the first refractive surface for focusing light onto the retina (Pradeep et al., 2020). Functionally, the cornea accounts for about 80% of

3 the eye’s focusing power and serves as an outer layer of protection (Goldstein and Brockmole, 2017; Miesfeld and Brown, 2019).

Next, light travels through the aqueous humor, which is a transparent liquid produced by the ciliary epithelium that fills the anterior and posterior chambers of the eye. The aqueous humor is 99.9% water and the remaining 0.1% consists of organic and inorganic ions, carbohydrates, glutathione, urea, amino acids, proteins, oxygen and carbon dioxide. While the air in front of the cornea has a refractive index of 1, the aqueous humor has an index of 1.336. This difference, due to the density of air versus aqueous humor, is responsible for the refractive power of the cornea (Ott, 2006).

As light passes through the aqueous humor, it also travels through the pupil; or rather, the opening in the centre of the iris. The iris is composed of an inner and outer pigmented epithelium, stroma, and iridial muscles (sphincter and dilator pupillae) (Davis-Silberman and Ashery-Padan, 2008). These components form a thin, contractile disk that is located between the lens and the cornea and regulates the amount (or intensity) of light that passes through the lens and falls on the retina – thus the iris acts as the aperture of the eye. The iris also helps focus the lens on nearby objects and aids in regulating intraocular pressure through the circulation of aqueous humor (Davis-Silberman and Ashery-Padan, 2008).

The vertebrate lens resides posterior to the iris and functions as the second refractive surface in the eye. The lens has an even higher refractive index than that of the surrounding fluids and is responsible for the remaining 20% of the focusing power of the eye (Goldstein and Brockmole, 2017). The refractive powers of the cornea and lens complement one another so that the total focal length is precisely aligned to the distance of the photoreceptor plane in the emmetropic (normal sighted) eye (Ott, 2006). Structurally, the lens is comprised of the capsule, epithelium, and densely packed and elongated fiber cells.

The Ciliary body (CB) lies between the neural retina (NR) and the iris in a ring-like structure. The lumen/stroma of the CB consists mainly of connective tissue, ciliary muscle and blood vessels. On the inner surface of the CB is an epithelial bilayer of cuboidal cells called the ciliary epithelium (CE). The outer pigmented layer of the CE is continuous with the retinal pigmented epithelium (RPE) and the inner non-pigmented layer is continuous with the NR. The two layers cooperate to produce the aqueous humor and some components of the vitreous humor, however the secretory function is carried out by the inner CE (Bishop et al., 2002; Coca-Prados

4 and Escribano, 2007). The pigmented CE layer harbours the adult retinal stem cell (RSC) population, which is discussed at length in Chapter 1.4. Depending on the species of vertebrate, the ciliary body can be subdivided into three regions: the ora serrata, the pars plana and the pars plicata. The ora seratta is a serrated junction between the retina and the CE. The pars plana is flat and immediately adjacent to and continuous with the peripheral edge of the retina. The pars plicata, which has a folded, ruffled appearance, is distal to the pars plana and is continuous with the iris epithelium. The CB can be a large structure, as in the eyes of newts, birds, cats, rabbits, dogs and primates; or relatively small as in the eyes of rodents. In the eyes of zebrafish or frogs, the CB is so small and rudimentary there is little anatomical distinction between the pars plana and pars plicata (Fischer et al., 2013).

The CB is also responsible for facilitating lens accommodation. In order to visualize objects clearly at various distances, the lens changes its shape to adjust the focal length to correspond to the distances of objects in the visual field and focus light onto the retina. This process is called accommodation, and it is usually achieved by a dynamic change of the total refractive power of the eye (Ott, 2006). To accomplish accommodation, the lens is stretched and flattened by the tension of the anterior zonula fibers, which are attached to the ciliary body. During contraction, the smooth, ring-like ciliary muscle of the CB shortens, bulges inward toward the lens, and releases the tension of the anterior zonula fibers (Ott, 2006). Under less tension the inherent elasticity of the lens causes it to form into a more spherical shape with higher refractive power (Croft et al., 2001). If focal length is adjusted incorrectly, the image either comes into focus in front of the photoreceptor plane (myopia or short-sightedness) or behind the photoreceptor plane (hyperopia or far-sightedness); also called presbyopia when it results from stiffening of the lens due to aging (Croft et al., 2001).

After passing through the lens, light must traverse the vitreous humor, which fills the space between the lens and the retina. The vitreous is a mass of transparent extracellular matrix containing 98–99% water (De Smet et al., 2013). However, its viscosity is two to four times greater than water, which gives it a gelatinous consistency (Locke and Morton, 1965). Its gel structure is maintained by a network of randomly spaced, nonbranching collagen fibrils primarily held apart by hyaluronan (HA). This network of collagen fibers provides the vitreous with mechanical strength, allowing it to sustain impacts, and transmit force to the retinal surface, while HA-collagen interactions ensure proper hydration and spacing, which functionally decreases light scattering and enhances transparency (Bishop, 2000; De Smet et al., 2013). Though the vitreous is more viscous than the aqueous humor, it has a similar refractive index of

~1.335. Since the lens has a refractive index of 1.44, the refractive power through the lens/vitreous is not as strong as through the cornea (del Amo et al., 2017; De Smet et al., 2013). The vitreous contains both structural proteins (in addition to collagen and HA), such as fibrillin and cartilage oligomeric matrix protein, and non-structural proteins, such as albumin, immunoglobulin, complement proteins, globulins and transferrin; and visual cycle proteins (del Amo et al., 2017; Bishop, 2000; Murthy et al., 2014). Most important to this thesis work, the vitreous also contains proteins related to signaling. As discussed in Chapter 1.4 and Chapter 2, signaling proteins in the vitreous are key regulators of adult mammalian retinal stem cell proliferation.

Figure 1.1 Anatomy of the human and mouse eye

Schematic cross-sections of the human and mouse eye. Light is focused by optical elements (such as cornea and lens) on the neural retina at the back of the eye. The central cone-only region of the human retina, called the fovea, is responsible for high resolution vision. The mouse retina lacks a distinct fovea and/or macula. The retinal pigment epithelium (RPE) monolayer separates the choroidal blood supply from the photoreceptors and is crucial for visual function. The lens is much larger in mouse than humans relative to eye size. Adapted from (Veleri et al., 2015).

1.1.2 The Mammalian Retina

Once light reaches the retina, it passes through the transparent retinal tissue to reach the photoreceptors, where absorption of the electromagnetic radiation of light and its phototransduction into electrochemical neural signals occurs (Palczewski, 2012). The simplest light detecting organs are composed of two cell types: the light sensitive photoreceptor cell and the non-neural retinal pigmented epithelial (RPE) cell. Both cell types are found in conjunction in every eye of the animal kingdom from insects and mollusca to higher vertebrates (Lamb et al., 2007). All vertebrate neural retinae are composed of three layers of cell bodies and two layers of synaptic neuropil (Figure 1.2). The outer nuclear layer (ONL) contains the cell bodies of rod and cone photoreceptors, the inner nuclear layer (INL) contains the cell bodies of bipolar, horizontal and amacrine cells, and the ganglion cell layer (GCL) contains cell bodies of retinal ganglion cells (RGCs) and displaced amacrine cells. The first area of synaptic neuropil in the retina is called the outer plexiform layer (OPL), where connections between rods and cones occur with vertically running bipolar cells and horizontally oriented horizontal cells. The second neuropil of the retina is called the inner plexiform layer (IPL), which functions as a relay station for bipolar cells to connect with RGCs. In addition, different varieties of horizontally- and vertically-directed amacrine cells (largely inhibitory neurons), interact to influence and integrate ganglion cell signals (Kolb et al., 2020). RGC axons run along the inner surface of the RGC layer and then exit together, through the retina via the optic nerve and project to the brain for higher order visual processing (Chalupa and Lia, 1991; Jeffery, 2001). Of note, the retina contains a small subset of ganglion cells that are sensitive to light, called intrinsically photosensitive retinal ganglion cells (ipRGCs) (Chen et al., 2011). ipRGCs express a rhodopsin-

7 like molecule, melanopsin, and innervate distinct regions of the brain to control circadian photoentrainment and the pupillary light response (Chen et al., 2011).

The INL also contains the cell bodies of Müller Glia (Kolb et al., 2020). Müller glial cells (MGs) span the entire thickness of the retinal tissue and ensheath all neurons therein. They form the outer limiting membrane (OLM) via adherens junctions with photoreceptor cell inner segments. MGs also form the inner limiting membrane (ILM) via lateral contacts between Müller cell end feet, which form a diffusion barrier between the inner retina and the vitreous. Müller cells perform many functions, including regulation of the homeostasis of the retinal extracellular environment (ions, water, neurotransmitter molecules, and pH), further metabolic regulation and are involved in the processing of visual information (including neurotransmitter recycling) (Bringmann et al., 2006; Kolb et al., 2020). Though Müller Glia are the most abundant and functionally important glial cell in the retina, there are two other types of glial cells resident in most vertebrate retinae: microglia and astrocytes. In contrast to Müller glia, which develop in situ from the retinal neuroepithelium, the other glial cells migrate into the retina during development.

The RPE is located between the light-sensitive outer segments of photoreceptors and blood supply of the choroid. The RPE consists of a hexagonally packed, tight-junction-connected, monolayer of cells that contain pigment granules. The apical membrane and processes face the subretinal space, which is occupied by an extracellular matrix specialized to enable interaction between RPE cells and outer segments of photoreceptors (Hollyfield, 1999; Strauss, 2005). The basolateral membrane of the RPE is in contact with the highly specialized multilayered Bruch’s membrane, which functions as an interaction matrix between the RPE and blood in the fenestrated vessels of the choroid (the choriocapillaris) (Guymer et al., 1999). Due to its pigmentation, the RPE functions to absorb light and reduce light scattering, which improves optical quality. It also forms a part of the blood-retina barrier, which is important for the immune privilege of the eye and regulating transport between the blood and the subretinal space. This epithelial transport function serves to supply nutrients to the photoreceptors, control the ion homeostasis in the subretinal space and to eliminate water and metabolites from retinal tissue (Steinberg, 1985; Strauss, 2005). Also, the RPE contains organelles for digestion of photoreceptor outer segment (OS) membranes and enzymes for processing visual proteins, including the conversion of of all-trans-retinal back to its cis configuration (which is important to phototransduction, as described in the following section) (Palczewski, 2012).

1.1.3 Photoreceptors and Phototransduction

Most vertebrates have two types of light-sensing photoreceptors – rods and cones – which were initially named based on their morphological appearance. Rods have long, thin, cylindrical outer segments, whereas cones have shorter conical outer segments (Michaelides et al., 2006). In addition to their structural differences, rods and cones have distinct functional properties (including light sensitivity, response kinetics, and adaptation range) that make them suitable for dim-light and bright-light functions, respectively (Fu and Yau, 2007). Rods are so sensitive to light stimuli they can detect a single photon, though kinetically rod responses are slow. This makes rods optimal for scotopic (dim light) conditions. While rods function well under dim light, in even moderately bright conditions they are saturated easily, have a long refractory period and lose their light-sensing abilities. In contrast to rods, cones are less sensitive, have fast responses to light, have a wide dynamic range, and recover rapidly following exposure, which makes them optimal for photopic (bright light) vision (Wang and Kefalov, 2011).

Each individual photoreceptor cell usually contains one type of opsin (the protein moiety of visual pigment) which determines the absorption spectrum of its related pigment and the spectral sensitivity of the photoreceptor. Rods contain only rhodopsin, and therefore, mediate monochromatic vision. Cones have either short-wave (S-opsin; peak sensitivity to blue light), medium-wave (M-opsin; peak sensitivity to green light) or long-wave (L-opsin; peak sensitivity to red light) opsins and mediate colour vision. However, mouse cones are unusual, as S-opsin and M-opsin are co-expressed in most cones (Applebury et al., 2000). In humans, cones express opsins in a mosaic-like pattern, whereas in mice, cones express S-opsins and M-opsins in opposing distribution gradients along the superior (M-opsin-high) to inferior (S-opsin-high) axis. In mice and humans, photoreceptors constitute over 70% of retinal cells, but rods outnumber cones by 30:1 in mice and 18–20:1 in humans (Swaroop et al., 2010). A major difference between humans and mice (and most other mammals) is the presence in humans of a thin, concave, cone-only region in the centre of the retina, called the fovea, which is responsible for the highest visual acuity. In humans, rod density is highest adjacent to the fovea and continually decreases toward the periphery of the retina (Swaroop et al., 2010).

Proteins involved in visual phototransduction are located predominantly in the photoreceptor OS of rods and cones, which are integrally associated with the RPE (both functionally and physically). The central foundation of our vision is the photochemical isomerization of the vitamin A-derived visual chromophore 11-cis-retinal from its cis- to all-trans-configuration. A

9 single photon of light isomerizes a single 11-cis-retinal bound to rod or cone opsins. For the retina to remain responsive to light and maintain vision, 11-cis-retinal, must be continuously and efficiently regenerated. RPE cells perform chromophore regeneration for both rods and cones, through a process known as the retinoid cycle (Palczewski, 2012). Isomerization of retinal and activation of opsins is the first step in the phototransduction cascade, however three other proteins are essential in the photoreceptor response to light: 1) the G-proteins, called transducins, present ~1:10 relative to rhodopsin; 2) the cyclic nucleotide phosphodiesterase (PDE), present at ~1:100 relative to rhodopsin; and 3) the cyclic nucleotide-gated channels (CNGCs), which control the flow of electrical current in the OS (Lamb and Pugh, 2006). Briefly, activated opsins transiently interact with transducins, which enables them to bind with GTP and become activated. Activated transducins bind to a catalytic subunit on the PDE and increase its rate of cGMP hydrolysis. Then, the cGMP concentration drops causing cGMP to unbind from CNGCs, leading the channels to close, and thereby generating the cell’s electrical response – a reduction in the circulating current and a consequent hyperpolarization. Thus, Rods and cones hyperpolarize in response to light, which ceases the tonic synaptic transmission of glutamate to second-order neurons, known as the “dark current”. (Lamb and Pugh, 2006; Wang and Kefalov, 2011; Wässle et al., 2009).

Figure 1.2 Anatomical organization of the vertebrate retina

Histological image overlaid with schematic diagram of the mature vertebrate neural retina structure, using the mouse retina as an example. The photoreceptor outer segments associate with the retinal pigmented epithelium (RPE), whereas their cell bodies reside in the outer nuclear layer (ONL). The inner nuclear layer (INL) contains the cell bodies of horizontal, bipolar, and amacrine cells, as well as Müller glia. The ganglion cell layer (GCL) contains the cell bodies of both ganglion and displaced amacrine cells. Connections between the photoreceptor, bipolar, and horizontal cells are found in the outer plexiform layer (OPL), whereas synapses between bipolar, ganglion, and amacrine cells occur in the inner plexiform layer (IPL). The ganglion cell (GC) axons make up the nerve fiber (NF) layer. Adapted from (Bassett and Wallace, 2012).

1.2 Vertebrate Eye Development & Retinogenesis

Vertebrate ocular development begins during gastrulation with the specification of a singular eye field located centrally in the developing forebrain (Graw, 2010; Miesfeld and Brown, 2019). More specifically, the eye field forms in the anterior neuroectoderm, adjacent to the telencephalic/diencephalic boundary (Figure 1.3). The iris epithelium, CE, NR and RPE derive from the neural tube eye field. Surface ectoderm gives rise to the lens and corneal epithelium, while neural crest cells from the periocular mesenchyme (POM) form the stroma and endothelium of the cornea, iris stroma, and ciliary body stroma. The timely action of transcription factors and inductive signals ensure the correct development of the different eye components (Figure 1.4). The earliest specification of the eye anlage is sub-divided into three major steps: neural induction in the presumptive ectoderm, anterior–posterior subdivision of the neural plate and specification of the eye field in the diencephalic territories (Graw, 2010; Miesfeld and Brown, 2019; Sinn and Wittbrodt, 2013). Eye field specification begins when signaling pathways and transcription factors in the presumptive telencephalon neuroectodermal cells initiate expression of the eye-field transcription factors (EFTFs) (that include among, others: ET, Rax/Rx, Six3, Pax6, Lhx2, Optx2/Six6, Tll and Ath5/Atoh7), which can act independently during subsequent eye field specification and optic vesicle formation, but also share a set of downstream genes (Heavner and Pevny, 2012; Zuber et al., 2003). Many of these EFTFs are also expressed in adult RSCs, as discussed in Chapter 1.4.

Figure 1.3 Schematic view of a developing vertebrate eye: from late gastrula to the optic cup

The most important stages form the gastrula to optic cup stage are shown. The first main step occurs when the single central eye field splits into two lateral parts to form the optic vesicles. At the same time (at embryonic day E9.5 in the mouse and 28 days of gestation in the human) the lens placode forms (placode stage). Invagination of the lens placode occurs at E10.5 in the mouse (lens pit stage). By the optic cup stage (E11.5 in the mouse and 31– 35 days of gestation in the human), the lens pit closes to form the lens vesicle, the future cornea becomes visible, and the retina begins to differentiate. Adapted from (Graw, 2010).

1.2.1 Establishing the Eye Field

Many EFTFs are required for eye field specification and act recursively throughout eye development in the specification of many tissues and retinal cell types. Retina and anterior neural fold homeobox (Rx/Rax) is crucially involved in optic cup morphogenesis and photoreceptor specification. In Rx null mouse embryos, or embryos with targeted inactivation of Rx, there is no evidence of eye development; both eye-specific gene expression and morphology is absent (Mathers et al., 1997; Zhang et al., 2000). This indicates an early role of Rx genes in the specification of retinal progenitor cells (RPCs). In fact, many homozygous null Rx mutants die neonatally with severe brain defects including absence of forebrain/midbrain structures, and fail to form eye structures, while homozygous hypomorph mutants are viable, but lack eyes and optic tracts and have hypothalamic defects (Graw, 2010).

The gene most commonly referred to as the “master control gene” of eye development is paired box 6 (Pax6), a transcription factor containing two DNA binding domains: a homeodomain and a paired-type domain (Czerny and Busslinger, 1995). Pax6 is highly conserved between vertebrates and invertebrates. Loss of Pax6 function leads to the eyeless phenotype in

Drosophila, the small eye phenotype in mouse and aniridia in humans (Glaser et al., 1994; Quiring et al., 1994). The pioneering work of Halder et al. (1995) showed that ectopic expression of the mouse Pax6 induces functional ommatidal eyes in Drosophila antennae or legs. This result suggested that the same genetic/signaling cascade set in motion by Pax6 expression leads to the development of the ommatidal eye in insects and the lens eye in mammals (Halder et al., 1995). A corresponding experiment was done in the frog, where injection of Pax6 mRNA led to the formation of ectopic eyes (Chow et al., 1999). These results, together with the early expression of vertebrate Pax6 genes in the anterior neural plate, led to the master control gene hypothesis for Pax6 as key regulator of eye development conserved throughout bilaterian evolution (Halder et al., 1995; Quiring et al., 1994). However, Pax6 mutant mice or rats initially develop eyes that fail to be maintained, whereas, in absence of homeobox- containing transcription factor six/sine oculis 3 (Six3), eyes do not form at all, suggesting it may be higher in the genetic hierarchy (Miesfeld and Brown, 2019).

Studies on Xenopus and zebrafish have established the role of Wnt-signaling in the formation of the eye field (Iongh et al., 2006). Wnt needs to be repressed in the eye anlage to allow the subsequent development of the optic vesicles. Six3 performs multiple functions in the eye field, including antagonizing Wnt ligand transcription (Lagutin et al., 2003; Liu et al., 2010). In Six3 knockout mice, the absence of Six3 results in the anterior expansion of posterior markers, pointing to a direct role for Six3 in shaping the Wnt-activity gradient in the anterior neural plate (Lagutin et al., 2003; Liu et al., 2010). Since Pax6 can activate the expression of Six3, it has been suggested that ectopic retinal structures induced by Pax6 in vertebrates are, in fact, due to the activation of Six3 expression in response to ectopic Pax6 (Chow et al., 1999; Loosli et al., 1999). Indeed, ectopic overexpression of Six3 results in the formation of ectopic eye cups in fish (Loosli et al., 1999). However, the authors note it only occurs in a competence domain established by Orthodenticle homeobox 2 (Otx2) expression.

Otx2 is expressed throughout the anterior forebrain early in embryogenesis and precedes any other eye field marker (Miesfeld and Brown, 2019). Also, Otx2 is required for the establishment of the eye field specific transcription factor network formed by Six3, Rx and Pax6, which subsequently act to down-regulate Otx2 expression (Chow and Lang, 2001; Sinn and Wittbrodt, 2013; Zhang et al., 2000). Similar to Six3, the potency of a cocktail of EFTF genes to induce ectopic eye structures in Xenopus is strongly enhanced by the addition of Otx2 (Zuber et al., 2003) Thus, Otx2 appears to play a permissive role in the initiation of the eye field and confers

13 eye field competence to the neural plate. However, Otx2 must eventually be downregulated during eye field specification to establish retinal identity.

Alongside Otx2, sex determining region Y (SRY)-box 2 (Sox2) is broadly expressed in the presumptive telencephalon where, together, they are required for Rx activation (Danno et al., 2008). Furthermore, Sox2 and Otx2 positive neuroepithelial cells secrete BMPs to shape the adjacent Rx-expression domain, thereby establishing a presumptive telencephalon/eye field boundary required for proper eye development (Bielen and Houart, 2012; Danno et al., 2008). Indeed, in humans, it was shown that Sox2 mutations frequently lead to anophthalmia (Ragge et al., 2005).

1.2.2 Splitting the Eye Field

After retinal identity is established in the eye field of the anterior neural plate, and before eye vesicle morphogenesis initiates, secreted factors of the TGFβ, FGF and sonic hedgehog (Shh) families are released from the underlying mesoderm, which splits the eye anlage into two bilateral symmetric retinal primordia (around E8.5 in mouse). Shh and Six3 both are critical regulators of eye field bifurcation, and their removal results in failure to split the eye field, and ultimately, cyclopia (Geng et al., 2008; Jeong et al., 2009). Furthermore, Geng et al. (2008) showed that Six3 regulates Shh expression in the underlying mesendoderm of the rostral diencephalon ventral midline. This domain is sensitised and rendered competent to respond to Shh by midline secreted FGF (Carl and Wittbrodt, 1999).

Cell migration is also a pivotal morphogenetic process in splitting the eye field. Fate mapping in wild-type zebrafish showed that, in response to secreted TGFβ/nodal, prospective hypothalamic cells migrate anteriorly along the midline from a position posterior to the eye field. The anterior movement of these median cells pushes the eye field cells laterally and separates the single, central eye field and establishes two eye primordia (England et al., 2006; Varga et al., 1999). However, in cyclops TGFβ/nodal mutants, this cell migration fails, resulting in failure to split the eye field.

1.2.3 Optic Vesicle Formation

After bilateral eye primordia are established, the optic vesicle precursor cells continue migrating, which initiates subsequent vesicle evagination. During this period, the eye field expands massively in size. Even though morphogenesis and proliferation are intimately linked, live imaging of zebrafish embryos demonstrated that cell shape changes and migration, rather than proliferation, are the driving forces of optic vesicle evagination (Kwan et al., 2012). Interestingly, moue ESC-derived retinal organoid cultures also produce optic vesicles via similar cell shape changes, however cell migration contributes less to the process (Eiraku et al., 2011a). As well, a subset of the eye field transcription factors controls optic vesicle evagination and formation. Loss of both Pax6 and Tll results in underdeveloped and irregular vesicles while the loss of Rx completely blocks optic vesicle development (Hollemann et al., 1998; Zhang et al., 2000) One proposed mechanism for Rx-dependent control of evagination is through repression of the cell adhesion molecule Nlcam and activation of the chemokine CXCR4, which influences cell shape and cell movement (Bielen and Houart, 2012; Brown et al., 2010). In a similar manner, the extracellular matrix protein Laminin has also been shown to impact cell shape changes during evagination (Ivanovitch et al., 2013). Thus, at E8.5–9.0 of mouse development, the walls of the diencephalon evaginate to form the optic vesicles (Heavner and Pevny, 2012).

1.2.4 Lens and Cornea Development

Lens development is controlled by signals originating from the optic vesicle and the periocular mesenchyme (POM). Initially, TGFβ and Wnt signaling from the POM to the ectoderm suppresses lens formation by repressing Pax6 expression. However, once the optic vesicle evaginates, it physically pushes the POM away from the ectoderm where the lens placode forms. This process relieves Pax6 repression and brings the optic vesicle close enough to the lens placode to induce lens morphogenesis through BMP signaling (Figure 1.4) (Furuta and Hogan, 1998). Upon induction, the lens placode thickens and undergoes morphogenetic movements that cause it to invaginate into the lens pit; which occurs concurrently with optic cup invagination (Figures 1.3 and 1.4). This occurs at ~day 33 of gestation in humans and ~E9.5 in mouse (Graw, 2010). After detachment, the lens pit fuses to form the lens vesicle, wherein FGF and BMP signaling control subsequent lens cell proliferation, differentiation and polarization (Jarrin et al., 2012). Differentiated lens fiber cells produce Crystallin proteins and lose their

15 mitochondria and cell nuclei to facilitate lens transparency (Vrensen et al., 1991). In the epithelial layer of the lens, progenitor cells remain proliferative and generate new fiber cells to maintain the lens throughout life (Graw, 2010).

The cornea originates from both the surface ectoderm (epithelium) and the neural crest cells of the POM (endothelium and stroma). After the lens detaches from the surface ectoderm, the space between is filled by invading cells from the POM. This wave of neural crest cells arrives in the mouse at ~E12 (Cvekl and Tamm, 2004). In the same period, the Pax6 positive cells of the surface ectoderm fuse to create a contiguous presumptive corneal epithelium. Pax6 expression is maintained throughout differentiation of the corneal epithelium and if it is not maintained at this stage, several corneal defects can occur (Davis et al., 2003). The POM cells express BMP inhibitors which are believed to prevent the lens fate in the presumptive cornea (Gerhart et al., 2009). Likewise, to restrict the epidermal fate, Dkk2 antagonizes WNT signals emanating from the forming corneal limbus (Mukhopadhyay et al., 2006). Another molecular differentiation event that occurs during corneal development involves cytokeratin proteins. The surface ectoderm first express K5/K14, but the pCE switches to K8/K18, and then again, to cornea-specific keratins K3/K12, which is dependent on continued Pax6 expression (Wolosin et al., 2004). At the end of corneal differentiation, Pax6 expression is maintained in the limbal stem cells, which continue to proliferate and replenish the epithelium throughout life.

1.2.5 Optic Cup Formation & Patterning

When the optic vesicles evaginate, they come in close contact with the surface ectoderm. These two tissues form a highly interactive system with several mutual interactions leading to the formation of the optic cup (Figure 1.4). One of the key players in the transformation of the optic vesicle to the optic cup is the retinoic acid (RA) signaling system. RA signaling generated by Raldh2 (retinaldehyde dehydrogenase) from the temporal periocular mesenchyme reaches the optic vesicle and is required for both, the lens pit and optic cup invagination (Figures 1.3 and

1.4). In particular, mouse Raldh2 knockout embryos lacking RA synthesis in the optic vesicle exhibit a failure in optic vesicle invagination, which is the first step in optic cup development (Mic et al., 2004). Another important factor in the transition from optic vesicle to optic cup is the transcription factor LIM homeobox protein 2 (Lhx2). In Lhx2 knockout mouse embryos, eye field specification and optic vesicle morphogenesis occur, but development arrests prior to optic cup

16 formation in both the optic neuroepithelium and lens ectoderm. This is because Lhx2 is required for optic vesicle patterning and lens formation (at least in part) by regulating BMP signaling in an autocrine manner in the optic neuroepithelium, and in a paracrine manner in the lens ectoderm (Yun et al., 2009). As with the optic vesicle, cell movement, not cell proliferation, is the main driver of optic cup formation (Kwan et al., 2012). Thus, at ~E10.5 in the mouse, the optic vesicle invaginates centrally to create a bilayered cup, the inner layer being the presumptive NR and the outer layer the presumptive RPE (Figure 1.4). Also notable, although there is coordinated morphogenesis between the lens and optic cup, a normally laminated NR develops in the absence of a lens in vivo (Harrington et al., 1991; Hyer et al., 2003) and in vitro in retinal organoids (which lack any surrounding tissue) (Eiraku et al., 2011a).

Intrinsic and extrinsic factors give positional and patterning information to the dividing optic vesicle/cup progenitor cells (Figure 1.4). The surrounding periocular mesenchyme, surface ectoderm, and optic neuroepithelium send signals to each other that define the regional boundaries between the RPE, NR and optic stalk (the presumptive optic nerve) and influence cell fate specification. For instance, FGF signaling from both the surface ectoderm and NR help define the boundary between RPE and NR by restricting visual system homeobox 2 (Vsx2) expression to the NR (Vsx2 was formerly called Chx10 and this nomenclature is still used throughout this thesis) (Nguyen and Arnheiter, 2000). In contrast, TGFβ family proteins secreted from the extraocular mesenchyme induce RPE cell fate through expression of Mitf (Fuhrmann et al., 2000). Shh from the ventral midline now promotes proximal-distal and dorsal-ventral patterning through the induction of Pax2 expression. Pax2 is required for optic stalk formation and represses Pax6, which in turn represses Pax2, thereby establishing a boundary between the developing optic cup and stalk (Patel and Sowden, 2019).

Activities of these early factors include cross-regulation at the genetic level, pointing to an early ocular transcriptional regulatory hierarchy (Miesfeld and Brown, 2019). For instance, Lhx2, Pax6, Pax2, Otx2, Mitf, and Vsx2/Chx10 are initially expressed uniformly within the optic vesicle, but their expression is eventually restricted to the NR (Pax6, Lhx2, Rx, Vsx2/Chx10), optic stalk (Pax2) and RPE (Otx2, Lhx2, Mitf), respectively (Heavner and Pevny, 2012; Patel and Sowden, 2019). Intrinsically, Mitf and Vsx2/Chx10 repress each other’s expression and activity through protein-protein and protein-DNA interactions to help define the RPE/NR domain (Zou and Levine, 2012). Further, Lhx2 appears to act upstream of Mitf, Vsx2/Chx10 and Pax2, as Lhx2 mutants are missing Mitf and Vsx2/Chx10 expression, while Pax2 is downregulated but

17 not completely abolished (Yun et al., 2009). Once patterning of the NR, RPE, optic stalk and other presumptive eye tissues is achieved, cell fate specification begins.

Figure 1.4 Optic cup formation and patterning

(A) Gradients of signalling molecules pattern the developing optic cup. NR: light gray, RPE: dark gray. The single pseudo stratified neuroepithelium of the optic vesicle folds to form the optic cup; the basal lamina faces the overlying surface ectoderm and the apical surface faces the lumen of the developing forebrain. (B) Differential expression of key transcription factors in the optic cup. Adapted from (Patel and Sowden, 2019).

1.2.6 Ciliary Body/Iris Development (the origin of retinal stem cells)

The outer lip of the optic cup (i.e. the ciliary margin of the optic cup), where the developing NR and RPE meet, will ultimately form into the iris epithelium and ciliary epithelium (CE). This is also the developmental origin of CE-RSCs. RSCs can be derived from the optic cup margin as early as E11.5-E12, but can be more consistently isolated by E14 (Coles et al., 2006; Tropepe et al., 2000). It has been suggested that the derivation of RSCs from RPE-NR border results in their potential to generate both NR and RPE progeny (Ahmad et al., 2000; Coles et al., 2006; Tropepe et al., 2000). This may also explain why RSCs are regulated by factors important in CE development (compare this section with Chapter 1.4 and Chapter 2). Like the corneal stroma and endoderm, the ciliary body and iris stroma derive from migrating POM cells (Cvekl and Tamm, 2004). The optic cup/ciliary margin is close to the lens vesicle, which releases signaling proteins that influence iris and CE development. Indeed, a classic experiment demonstrated that, in mice, when the entire lens is ablated by lens-specific expression of the cytotoxic diphtheria toxin A the iris and CE fail to differentiate (Harrington et al., 1991). Dias da Silva et al. (2007) proposed a model where anti-parallel gradients of BMP (from the POM and within the optic cup) and FGF (from the lens) define the region where the ciliary epithelium/body will develop. By forcing RPE cells to express FGF4 in the embryonic chick eye, they found the FGF- expressing cells and their immediate neighbours transformed into NR. At a distance from the FGF4-expressing cells, the tissue transitioned back to pigmented epithelium and CE tissue was found in the transition zone (Dias da Silva et al., 2007). This model referred to previous work showing that BMP signaling is critical for CE development. BMPs are dynamically expressed in the iris, CE and RPE throughout development, and Bmp4+/- mice develop a malformed iris and CB (Chang et al., 2001). Further, Zhao et al. (2002) found that the ciliary body was completely absent in transgenic mice engineered to overexpress the BMP antagonist, Noggin, under a lens-specific promoter, which could be rescued by co-expression of Bmp7 (Zhao et al., 2002a). More recently, it has been proposed that Notch signaling might repress BMP inhibitors in the CE to facilitate both BMP signaling and ciliary body development (Zhou et al., 2013). Also, canonical Wnt signaling is crucial for specification and development of the iris and CE. Wnt signaling is progressively and particularly active in the ciliary margin during development (Liu et al., 2003) and perturbations lead to significant changes in tissue specification. For example, Cho and Cepko (2006) showed that Wnt signal activation in the central retina of the chick embryo inhibited retinal progenitor proliferation and differentiation and induced the expression of markers for the iris and CE. Furthermore, expression of dominant-negative Lef1 (a Wnt target

19 gene) in the optic cup led to the suppression of peripheral genes and iris hypoplasia (Cho and Cepko, 2006). Likewise, Liu et al. (2007) found that stabilization of β-catenin expression caused the transdifferentiation of NR tissue into ciliary margin tissue, which led to an expanded ciliary body-like region later in development. Conversely, inactivation of β-catenin reduced ciliary margin marker expression and reduced the size of the subsequent ciliary body and iris (Liu et al., 2007). In addition, the Hippo signaling pathway is also important for proper CE and iris epithelium development. Neurofibromin 2 (NF2), an upstream Hippo signaling component, regulates proper proliferation of progenitors in the ciliary margin through the transcriptional co- activators Yap/Taz. If NF2 is knocked out in the embryonic neuroepithelium, the ciliary margin becomes hyperplastic and malformed (Moon et al., 2018).

Pax6 is expressed in a gradient throughout the optic cup, with the highest expression in the peripheral ciliary margin and progressively decreasing centrally (Davis-Silberman et al., 2005). Pax6 is required for the formation of both the iris and ciliary body and mediates their development through cell-autonomous downstream targets, including Foxc1, BMP4, and TGFβ2 (Wang et al., 2017). The transcription factors Otx1 and Msh homeobox 1 (Msx1) are expressed in the early ciliary margin (beginning around E12.5 in mouse) and delineate the presumptive CE and iris from the neural retina (Bélanger et al., 2017; Trimarchi et al., 2009). Coincidentally, this is around the earliest time that RSCs can be derived from the ciliary margin. Further, Zhao et al. (2002) found both Msx1 and Otx1 expression was suppressed by Noggin expression in the lens (in the mice that failed to form a ciliary body). However, only Otx1 appears essential for CE and iris epithelium development. In Otx1-null mice, the CE does not develop and the iris epithelium is underdeveloped (Acampora et al., 1996), whereas Msx1-null mice do not exhibit any abnormalities in the ciliary body or iris (Satokata and Maas, 1994).

1.2.7 Retinogenesis

All the cell types of the retina derive from multipotent retinal progenitor cells (RPC) within the optic cup in a sequence that is remarkably conserved across all vertebrates. During retinogenesis in mammals, RGCs are generated first, followed by the production of cone photoreceptors, horizontal cells and most of the amacrine cells. Bipolar cells, Müller glia, the remaining amacrine cells and most rod photoreceptors are generated postnatally. However there is considerable temporal overlap in the production of retinal cell types at any given time

(Figure 1.5) (Bassett and Wallace, 2012; Cepko et al., 1996). Also, notable, retinogenesis proceeds in a central to peripheral pattern, such that the last active RPCs are near the ciliary margin where the RSCs reside (as discussed in Chapter 1.4) (Moshiri et al., 2004). Many transcription factors, such as Rx, Vsx2/Chx10, Pax6, Sufu, Six6, and Sox2 regulate RPC proliferation and/or multipotency (Bassett and Wallace, 2012). Furthermore, an increasing number of transcription factors have emerged as key intrinsic regulators of retinal cell fate, including members of the basic helix-loop-helix (bHLH), homeodomain, and forkhead families (Ohsawa and Kageyama, 2008). However, several extrinsic cell signaling pathways, such as Notch, Hedgehog (Hh), Wnt, TGFβ/BMP, receptor tyrosine kinase (RTK) and Jak/ STAT have been shown to regulate RPC proliferation and cell fate choice (Yang, 2004). These findings have led to ongoing questions about whether environmental cues versus intrinsic cell competencies are the main determinants of retinal cell fate (Livesey and Cepko, 2001). However, it is currently held that the environment and the intrinsic state of the cell are both likely to be important factors in determining retinal cell fate (Bassett and Wallace, 2012). Potential models of fate determination are discussed further in Chapter 4.

Figure 1.5 Chronological order and transcriptional regulation of retinal cell birth

Embryonic and postnatal times based on mouse development. This provides a detailed and concise description of the temporal window and overlap of each cell type and the genes involved in cell fate determination. Adapted from (Bassett and Wallace, 2012).

1.2.8 Photoreceptor Development

Chapter 4 of this thesis focuses on the use of exogenous signals to direct the specification of RSC progeny to the photoreceptor fate. How these extrinsic signals influence the known genetic regulators of rod and cone development is of great interest. Thus, genetic regulators of photoreceptor fate are mainly discussed here. For discussion of extrinsic factors that influence the fate specification of RSC progeny, see Chapter 1.4. and Chapter 4.

For cones, rods, and bipolar cells, Otx2-expression is critical for their differentiation. Otx2 loss- of-function blocks the formation of rods, cones and bipolar cells, whereas overexpression stimulates the overproduction of all three cell types (Nishida et al., 2003; Wang et al., 2014). Immediately downstream of Otx2 are the activities of Vsx2/Chx10, Prdm1 (also know as Blimp1) and Cone-rod homeobox (Crx), which influence whether an Otx2 positive cell will differentiate into a bipolar cell (Vsx2/Chx10) or photoreceptor (Prdm1 and Crx) (Swaroop et al., 2010). Vsx2/Chx10 is initially expressed by every RPC, but is later restricted to the bipolar cell lineage and represses the photoreceptor lineage (Kim et al., 2008), whereas Prdm1 controls the binary fate decision between rod and bipolar cells by promoting the rod fate and repressing the bipolar fate (Wang et al., 2014). Crx is also downstream of Otx2 where it acts to specify rods and cones, however, Crx cannot induce cone or rod development on its own. Crx must act with the factors Neural leucine zipper (Nrl), which is essential for the rod fate, and Retinoid-related orphan receptor β (RORβ), which regulates the development of rods and cones (Swaroop et al., 2010). In fact, it is thought that retinal precursors differentiate to a default S-cone state under the control of Otx2, Crx and RORβ unless diverted into a rod or M-cone state by additional signals. This default cone state has important implications for the findings in Chapter 4 and is discussed further therein. This idea is corroborated by the finding that, in Nrl knockout mice, all rods default to an S-cone fate and develop relatively normal morphological, molecular and functional cone-like properties (Daniele et al., 2005; Mears et al., 2001). Indeed, Nrl and its target, photoreceptor-specific nuclear receptor (Nr2e3), actively suppress cone genes in order to induce the rod fate (Oh et al., 2008). However, while Nr2e3 expression can directly repress the cone fate, it alone is not sufficient to induce functioning rods, and therefore, it is considered a co-activator of the rod specifying genes alongside Crx and Nrl (Cheng et al., 2006).

Other factors involved in M-opsin and S-opsin patterning in cones include thyroid hormone β2 (TRβ2) and its ligand triiodothyronine (T3), retinoid X receptor-γ (RXRγ), COUP transcription factors (COUP-TFs), RORα and Neurod1. TRβ2 requires T3 at appropriate developmental

22 stages to promote expression of M-opsin and suppress the expression S-opsin. In fact, in mice with congenital hypothyroidism, M-opsin expression is impaired, whereas treatment with excessive T3 suppresses the induction of S-opsin in early cones (Lu et al., 2009; Roberts et al., 2006). However, since TRβ2 is expressed in all cones throughout the retina, other factors contribute to regional differences in opsin expression. For example, RXRγ has been shown to repress S-opsin specifically in the superior retina to establish the M- and S-cone gradient in the mouse (Roberts et al., 2005). The expression of COUP-TFs, which is regulated by BMP signaling, are expressed in specific domains throughout the retina and establish proper dorsoventral patterning of opsins via the suppression of M- and S-opsin (Satoh et al., 2009). Neurod1 enhances TRβ2 expression, and thereby, indirectly promotes M-opsin expression (Liu et al., 2008), whereas the activation of cone genes by RORα appears to be involved in cone maturation rather than specification (Fujieda et al., 2009).

All together, this section on Eye Development & Retinogenesis demonstrates that much is known about key extrinsic and intrinsic factors that regulate the complex processes of morphogenesis and RPC fate specification in the eye. However, fundamental questions still loom about how the interplay between these factors enables the ordered yet concurrent development of such astounding cell-type diversity. Indeed, it has recently been suggested that the mouse retina contains ~130 neuronal cell types and is therefore comparable in complexity to the brain (Yan et al., 2020).

1.3 The ciliary marginal zone and retinal stem cells of non- mammalian vertebrates

The structure, cell types, and the function of the retina is largely conserved among vertebrates, yet the regenerative competency of different vertebrate classes varies tremendously and appears to have diminished with evolutionary time. Furthermore, even species that do regenerate the retina display differences in the cellular and molecular mechanisms that are used to do so. Below, I discuss two vertebrate classes that have been investigated most thoroughly for their profound regenerative abilities, fish and amphibians.

In some fish and amphibians, there is a ring of cells at the periphery of the maturing and adult neural retina that contains retinal stem and progenitor cells (RSCs and RPCs, respectively).

This is known as the ciliary marginal zone (CMZ) and is often compared as a homologous region to the CE, where mammalian RSCs reside (as discussed in the following section Chapter 1.4). Retinal neurons are added to the periphery of the retina by differentiation of these CMZ RSCs and RPCs and the retina grows via the addition and integration of concentric rings of newly generated cells at the CMZ. In fact, a significant portion of the retina is formed from the CMZ after the initial differentiation of the retinal neuroepithelium. This is most dramatically observed in teleost fish – from the time of their hatching to when they reach their mature size, the eye of a teleost fish can grow 100‐fold (Ail and Perron, 2017; Fischer et al., 2013; Moshiri et al., 2004). The Zebrafish and Xenopus CMZ is ordered spatially by level of differentiation with the most peripheral cells being the retinal stem cells followed by the retinal progenitors, which become more differentiated with increasing distance into the neural retina (Ail and Perron, 2017). CMZ-RSCs express the eye field transcription factors XSix3, Xrx1 and Pax6, the latter two being necessary for eye development (Zuber et al., 2003). As the peripheral RSCs slowly proliferate, they will divide asymmetrically, where some of the daughter cells will maintain RSC identity, while others develop into RPCs and migrate inward (Wan et al., 2016). Both RSCs and RPCs throughout the CMZ express the transcription factor profile of retinal progenitors, including Rx, Vsx2/Chx10, and Pax6 (Perron et al., 1998; Zuber et al., 2003). Also, CMZ cells express proneural transcription factors, such as Ngn2 and Ascl1 (Harris and Perron, 1998; Zuber et al., 2003). Lineage‐tracing studies of the Xenopus CMZ cells have shown that they can give rise to clones that contain all retinal cell types (Wetts and Fraser, 1988; Wetts et al., 1989). In zebrafish, the CMZ is not believed to give rise to rod photoreceptors directly (Raymond et al., 2006; Wilson et al., 2016), instead Müller cells are the source of newly born rods throughout the lifetime of the zebrafish retina (Lenkowski and Raymond, 2014; Stenkamp, 2011). Yet, the CMZ produces Müller glia that become dedicated rod lineage progenitors (Stenkamp, 2011). In Xenopus, the amount of proliferation in the CMZ and therefore the size of the CMZ is controlled through a balance of Wnt and hedgehog signaling. Wnt signaling delays cell cycle withdrawal so that gain of Wnt function results in an increase in RSC/RPC proliferation, leading to a larger CMZ. Conversely, Hh signaling has the opposite effect, promoting cell cycle exit (Borday et al., 2012). Wnt and Hh signaling do not appear to have the same interplay in zebrafish as they do in Xenopus. As in Xenopus, the canonical Wnt pathway is needed to maintain the proliferative RPCs in the zebrafish CMZ, as well as to inhibit differentiation. However, in the zebrafish retina, Hh signaling is important for both cellular proliferation and cell cycle exit. This was revealed in sonic‐you (syu) Hh mutants, where in some eyes, retinal proliferation is reduced during

24 neurogenesis, while in others many retinal cells fail to exit the cell cycle and remain in a neuroepithelial state (Stenkamp et al., 2002).

In both frogs and fish, the CMZ contributes to regeneration of the damaged retina. It has long been known that retinal injury in fish causes increased cell proliferation in the margin, suggesting that the CMZ contributes to retinal regeneration (Maier and Wolburg, 1979). However, in fish, the CMZ is the major source of the regenerated peripheral retina but not of the central retina (Stenkamp et al., 2001). Müller glia are the source of central regeneration in zebrafish retina following injury, and as in development, the only direct source of new rod cells (Lenkowski and Raymond, 2014; Stenkamp et al., 2001; Wilson et al., 2016). The contribution of the CMZ to retinal regeneration appears to be variable among Xenopus species. In Xenopus tropicalis, the entire retina regenerates from the CMZ (Miyake and Araki, 2014). In Rana pipiens, the CMZ is able to replenish specific cell types that were ablated by chemical treatment (Reh, 1987). In Xenopus laevis, retinal regeneration occurs primarily through the transdifferentiation of RPE cells. Thus, the CMZ is capable of profound regeneration of the retina in both frogs and fish. Whether a direct evolutionary link between CMZ-RSCs and the CE- RSCs of mammals remains to be elucidated.

1.4 Retinal stem cells in the adult mammalian ciliary epithelium

Unlike fish and amphibians, birth-dating studies have indicated that mouse retinal histogenesis is normally completed by the end of the second postnatal week (around P11-P14) and proliferation in the adult retina or ciliary epithelium (CE) does not continue beyond that point (Dhomen et al., 2006; Kubota et al., 2002; Moshiri and Reh, 2004; Young, 1985b, 1985a). Furthermore, it had long been held that retinal damage and disease in mammals causes irreversible vision loss with no evidence of proliferation, regeneration or visual recovery. This led researchers to conclude there is no CMZ-like zone in the mammalian eye (Aladdad and Kador, 2019; Moshiri et al., 2004; Wohl et al., 2012).

However, in the year 2000, two independent groups reported the discovery of retinal stem cells in adult mouse (Tropepe et al., 2000) and rat (Ahmad et al., 2000) eyes. These studies used clonogenic sphere assays with cells from the neural retina, RPE and CE and observed that clonal spheres arose solely from the pigmented layer of the CE. The spheres could be

25 passaged, evincing that the sphere-initiating cells within could self-renew. They also had the capacity for multipotential differentiation, and have since been shown to be capable of producing all of the cell types of the retina and RPE cells (Coles et al., 2004; Inoue et al., 2010). Thus, these sphere-forming CE cells demonstrated both hallmarks of a stem cell, leading the authors to dub them “retinal stem cells” (RSCs). Furthermore, both studies considered the anatomical location of RSCs in the CE to be evidence of a region homologous with the CMZ of fish and amphibians in the mammalian eye.

In culture, CE-RSCs lose their pigmentation and proliferate with or without exogenous growth factor stimulation (Ahmad et al., 2000; Das et al., 2005; Engelhardt et al., 2004, 2005; Giordano et al., 2007; Tropepe et al., 2000). However, pigmentation itself does not appear to influence the stem/progenitor cell properties of RSCs, as shown in albino animals (Ahmad et al., 2000; Tropepe et al., 2000). RSCs are a rare subset of cells among the adult CE, with ~0.1-0.2% of cells being capable of forming clonal spheres (Coles et al., 2004; Das et al., 2004; Inoue et al., 2005; Tropepe et al., 2000). However, certain features, such as pigmentation, cell size P- cadherin expression and Pax6 expression have been used to prospectively enrich for sphere- forming RSCs within the CE by up to 10-fold; with RSCs being characterized as relatively heavily pigmented, large cells with low P-cadherin expression and high Pax6 expression (Ballios et al., 2012; Xu et al., 2007b). To date, RSCs have been derived from the CE of the mouse, rat, rabbit, pig, cow, primate and human eye, demonstrating their conservation across mammalian species (Coles et al., 2004; Gu et al., 2007; Inoue et al., 2005; MacNeil et al., 2007; Tropepe et al., 2000; Wohl et al., 2012). Also, RSCs have been isolated from fetal human CE (Abburi et al., 2013), human embryonic stem cells (hESCs) (Clarke et al., 2012), and hESC-derived retinal organoids with a ciliary margin-like stem cell niche (Kuwahara et al., 2015).

The protein expression and gene expression profiles reported for CE-derived retinal stem/progenitor colonies include many genes and proteins expressed in the early eye field that are important for retinal identity, and many that are shared with embryonic retinal progenitors, CMZ-RSCs and other stem cells. For example, CE-retinal stem/progenitor cells (CE-RSPCs) have been reported to express Pax6, Vsx2/Chx10, Rx, Six3, Sox2, Lhx2, Nestin, Vimentin, Notch and Nanog (Das et al., 2005; Del Debbio et al., 2013; Demontis et al., 2012; Lord- Grignon et al., 2006; MacNeil et al., 2007; Tropepe et al., 2000; Xu et al., 2007a). In fact, they have been characterized to share ~80% of their expressed genes with embryonic RPCs, with more genes in common with early retinal progenitors in particular (Das et al., 2005). However, isolation and direct characterization of RSCs is still hindered by the lack of specific and selective

26 markers, as most of the genes and proteins profiled are also expressed in RSC progeny and embryonic retinal progenitors; while a subset are expressed in mature CE, retinal neurons and glia. For example, Pax6 is expressed in all RSC progeny, all embryonic retinal progenitors and in mature CE cells and amacrine cells (Das and James, 2005; Marquardt et al., 2001; Xu et al., 2007b); Vsx2/Chx10 is expressed in mature bipolar cells (Kim et al., 2008); Sox2 is expressed in Muller glia and retinal astrocytes (Fischer et al., 2010); while the intermediate filaments vimentin and nestin are both expressed in glial cells (Fischer et al., 2010; Wohl et al., 2011), and nestin is also expressed in endothelial cells and pericytes (Kim et al., 2016a). Thus, the search for markers that can uniquely identify RSCs and distinguish them from progenitor cells, precursor cells, and mature cell types is ongoing. Notably, some reports have indicated that CE- RSCs possess only limited proliferative and neurogenic potential in comparison to neural stem cells (NSCs) of the subventricular zone (SVZ) in rats and humans (Engelhardt et al., 2004; Moe et al., 2009), rat embryonic forebrain-derived progenitors (Yanagi et al., 2006) or neonatal mouse RPCs (Wohl et al., 2012). This has led these authors to question the ‘stem cell’ designation of CE-RSCs. These criticisms are noted throughout Chapters 2, 3 & 4. However, the differentiation and proliferative capacities of RSCs are discussed below and should also address those criticisms.

1.4.1 Differentiation Potential of Retinal Stem Cells

CE-RSC progeny can spontaneously differentiate into all cell types of the retina and RPE in vitro. Further, it has been demonstrated that, following transplantation into the eye, RSPCs integrate into the retina and spontaneously differentiate into photoreceptors, interneurons, Muller glia and RPE in vivo (Chacko et al., 2003; Coles et al., 2004; Guduric-Fuchs et al., 2011; Inoue et al., 2010). However, much like embryonic RPCs (Cepko, 1999), both intrinsic and extrinsic factors seem to strongly influence the neurogenic capacity of CE-RSCs and the specification of their progeny. For instance, RSPC cultures induced to express Otx2 (Akagi et al., 2004) or Crx (Akagi et al., 2004; Haruta et al., 2001) generate over 90% rhodopsin- expressing rod progeny. In contrast, overexpression of Vsx2/Chx10 maintains RSPCs in undifferentiated state and enhances progenitor proliferation (Inoue et al., 2010). However, expression of a construct designed to reverse the repressive effects of Vsx2/Chx10 (Chx10VP16) decreased proliferation and enhanced photoreceptor differentiation. Likewise, Crx, Otx2 or their co-expression each increased photoreceptor differentiation, while Otx2, Crx

27 and Chx10VP16 together showed the greatest increase in rod and cone specification. What’s more, transplantation of Chx10VP16/Otx2/Crx-transduced RSPCs showed greater survival, integration and differentiation in vivo and produced functional improvements in electroretinogram (ERG) b-wave response in Transducin-1 mutant mice (which do no have functional rods) (Inoue et al., 2010).

Exogenous factors can also influence the differentiation of RSC progeny toward functional photoreceptors. Demontis et al. (2012) showed that factors known to favour embryonic rod differentiation – taurine, retinoic acid (RA), sodium butyrate and T3 – influenced the specification of RSC progeny toward a rod fate. Furthermore, they extensively characterized the expression of phototransduction components, electrophysiological function and light responsiveness in the RSC-derived rod progeny (Demontis et al., 2012). Similarly, Ballios et al. (2012) screened factors known to be important in embryonic rod fate specification for their ability to influence the differentiation of RSC progeny. Individually, FGF2, Shh, taurine and retinoic acid (RA) each improved rod differentiation. However, combinations including taurine and RA together produced over 90% rhodopsin-expressing rod progeny from mouse and human RSCs. In a later experiment, transplantation of those RSC-derived rods were shown to improve the pupillary light response of genetically blind mice (Ballios et al., 2015). Del Debbio et al. (2013) found that retinal stem cell progeny could generate different functional neurons in response to conditioned media (CM) from different stages of embryonic retinal development. When cultured in CM from the E14 retina, RSPCs generated functional RGC progeny, while CM from the PND1 retina generated functional rod photoreceptors (Del Debbio et al., 2013). Thus, it appeared that exogenous factors in the CM influenced the specification of RSC progeny toward cells normally born during the corresponding developmental stage. This strategy had previously been used to successfully generate photoreceptors from mESCs (Zhao et al., 2002c). Recently, Zhou et al. (2015) reported that treating hESCs with an exogenous inhibitor of TGFβ, BMP and Wnt signaling (COCO) induces their differentiation into ~60-80% functional cone photoreceptors in 4-5 weeks (Zhou et al., 2015). In Chapter 4, we employ this same approach to investigate if exogenous COCO can also influence the differentiation of adult RSPCs to generate cone photoreceptor progeny.

1.4.2 Proliferative Potential of Retinal Stem Cells

Similar to RSC differentiation, both intrinsic and extrinsic factors have been shown to regulate the proliferation of RSCs. In culture, RSCs appear to be inherently proliferative as they form clonal spheres without exogenous growth factor stimulation (Giordano et al., 2007; Tropepe et al., 2000), though both FGF and EGF enhance RSC proliferation and self-renewal (FGF having greater effect than EGF) (Giordano et al., 2007). However, RSPCs have been shown to produce their own endogenous FGF (Balenci and van der Kooy, 2014; Tropepe et al., 2000). Further, FGF has been shown to be required for RSC-derived progenitor proliferation and RSC self- renewal (Balenci and van der Kooy, 2014). In fact, both FGF and Notch signaling are required for RSC self-renewal to occur, and exogenous Notch ligands alone can modulate the extent of progenitor proliferation and RSC self-renewal (Balenci and van der Kooy, 2014). Consistent with that finding, Alexson et al. (2006) had previously found that presenilin-1 mutant mice, which have reduced Notch signaling, had a decreased pool of RSCs in vivo and that isolated RSCs had impaired self-renewal (Alexson et al., 2006). Further, Das et al. (2004) found that both Notch signaling and c-kit signaling promoted RSC sphere proliferation and passaging at the expense of differentiation (Das et al., 2004). Also, pigment epithelium-derived factor (PEDF), which as the name suggests is synthesized and secreted by RPE cells, has been shown to enhance RSPC proliferation and RSC expansion. (De Marzo et al., 2010).

Both the Wnt and BMP signaling pathways have been shown to regulate RSPC proliferation in vitro and have even been posited to be pivotal regulators of RSC quiescence in vivo – this will be discussed further in Chapter 2. Briefly, Wnt activation, via Wnt ligands or GSK3B inhibitors, enhances RSC proliferation and self-renewal (Inoue et al., 2006). Wnt signaling is active in RSC spheres and its antagonism has been shown to inhibit RSPC proliferation (Balenci et al., 2013). Furthermore, genetic Wnt activation in the adult mouse has been reported to increase the number of RSCs in vivo (Fang et al., 2013). In contrast, activation of BMP signaling inhibits RSPC proliferation in vitro and has been shown to mediate the quiescence of adult mammalian neural stem cells in the brain (Johnston and Lim, 2010). Further, in P4 rat retinal explants, addition of exogenous TGFβ1/2 inhibits proliferation in the retinal margin. Conversely, inhibition of TGFβ signaling in P6 explants and in vivo extends the period of postnatal proliferation (Close et al., 2005).

Some of the transcription factors used to characterize the early retinal stem/progenitor identity of RSCs also regulate RSC proliferation. For instance, not only do RSCs maintain high

29 expression of Pax6, it is required for RSC proliferation and expansion (Xu et al., 2007b). As mentioned above, overexpression of Vsx2/Chx10 maintains RSPCs in undifferentiated state and enhances progenitor proliferation (Inoue et al., 2010). However, in surprising contrast, in Vsx2/Chx10-null mice, it was found that the pigmented ciliary margin was expanded compared to the wild type and the number of clonal RSC spheres derived from the CE was significantly increased (Coles et al., 2006; Tropepe et al., 2000). Similar results were obtained with CE cells derived from Mitf knockout mice, which exhibit a complete loss of the differentiated RPE. Coles et al. (2006) proposed that this may not have been a direct intrinsic effect, and that the number of RSCs in the CE may be regulated by cell-extrinsic feedback signals from retinal and RPE progenitor cells during development. Both β-catenin knockout and overexpression in vivo was shown to decrease the number of RSC spheres in subsequent sphere assays (Liu et al., 2007). This is surprising given that Wnt signaling has been shown to enhance RSC proliferation and self-renewal in culture (as described above). It is unclear if this could have resulted from genetic interactions with other factors, such as Pax6, or may have involved extrinsic feedback as proposed by Coles et al. (2006). Another report showed that P-Cadherin expression in RSCs (mentioned in 1.4 as a marker to enrich for RSCs) is important for the formation and proliferation of RSC spheres in vitro but is not required for maintenance of RSCs in vivo (Coles and van der Kooy, 2017). Thus, cell adhesion could be another factor influencing the lack of RSC sphere formation in vitro following β-catenin perturbations.

Cell culture conditions also impact retinal stem cell proliferation and self-renewal. Coles et al. (2004) showed that, while free-floating RSC spheres do not increase in size after 7 days and have significantly reduced self-renewal/passaging ability after the 3rd passage, when RSC spheres were transferred to monolayer culture they had indefinite proliferation and expansion capacity (Coles et al., 2004). Recently, Baakdhah and van der Kooy (2019) used a suspension bioreactor system to reveal that many different parameters influence growth characteristics of RSCs in culture. For instance, the agitation rate, oxygen content, cell density, and the use of cell microcarriers each impacted one or more of RSC survival, proliferation or self-renewal

(Baakdhah and van der Kooy, 2019). In particular, low 02 (5%) increased RSC symmetric expansion, lower agitation increased RSPC survival, high cell density increased RSC yield, and one particular cell microcarrier enhanced RSPC growth and RSC expansion.

These experiments have significantly advanced our understanding of how RSC and RSC- derived progenitor proliferation is regulated and uncovered mechanistic commonalities and differences to other stem cell populations. Furthermore, the characterization of some of these

30 mechanisms in vitro has led to follow-up studies examining whether those same mechanisms control RSC quiescence and proliferation in vivo.

1.4.3 Regenerative potential of retinal stem cells in vivo

As mentioned above, retinogenesis in mice is concluded within the first two postnatal weeks and no further proliferation in the peripheral retina/CE is typically observed beyond that period, suggesting that mammals do not have a CMZ-like region in the adult eye (Dhomen et al., 2006; Kubota et al., 2002; Moshiri and Reh, 2004; Young, 1985b, 1985a). However, the discovery of RSCs in the adult mammalian CE prompted researchers to revisit whether some latent regenerative potential might indeed exist in the adult mammalian eye in vivo. This re- examination led researchers to investigate injury models, genetic manipulations and injection of exogenous factors in the eye. However, perhaps most surprising are reports of CE proliferation and neurogenesis in normal mammalian eyes with no intervention. For example, Fischer et al. (2001) described markers of all retinal cell types located in cysts within the primate pars plana. These cysts were not found in monkeys aged 1 or 2 years old, but were found in those 6 years or older, suggesting that neurogenesis may occur in the normal adult primate CE (Fischer et al., 2001). Likewise, Martinez-Navarrete et al. (2008) found evidence of retinal progenitor cells that can generate retinal neurons, including photoreceptors, in the peripheral retina/CE of adult monkeys and humans (Martínez-Navarrete et al., 2008). In the mouse, Nishiguchi et al. (2008) described proliferating photoreceptor progenitors in the CE during development and found that cells with immature cone morphology persisted even at PND120. Furthermore, chemically- or genetically-induced retinal degeneration significantly increased the number of photoreceptor progenitor cells detected in the adult CE (Nishiguchi et al., 2008, 2009).

Indeed, there have been other reports of at least some proliferation and neurogenesis in the CE following retinal injury. For instance, Johnsen et al. 2012 found that human eyes with vitreoretinopathy had ki67 staining in the CE and produced more RSC spheres in culture than healthy controls (Johnsen et al., 2012). In two studies, optic nerve lesion was reported to induce proliferation in the adult mouse CE (Nickerson et al., 2007; Wohl et al., 2009). In both studies, the proliferating cells expressed progenitor markers (such as Nestin), but there was no evidence of cell migration into the retina. Further, only Nickerson et al. (2007) reported photoreceptor marker expression in the CE, but there was no evidence those cells were derived from

31 proliferating progenitors, as they were not labeled with BrdU. Similarly, retinal ischemia has been described to induce the proliferation of Nestin+ cells in the CE (Wohl et al., 2012). However, using Nestin expression as an RSC marker is problematic as it is expressed in other cell types in the CE (discussed in Chapter 2.5). A study by Jian et al. (2009) detected a small number of proliferating progenitors in the retinal margin/CE of wildtype rats up to PND60. However, in Royal College of Surgeons (RCS) rats, a genetic model for retinitis pigmentosa, there was a ~4-fold increase in BrdU positive cells. Interestingly, the authors also found there was a concomitant upregulation of Hh signaling genes in the degenerating rat retina (Jian et al., 2009).

Several gene knockout experiments have also produced evidence of retinal stem/progenitor proliferation and RSC expansion in vivo. In fact, in line with increased Hh expression after injury, Moshiri and Reh (2004) found that ptc+/− mutant mice – which have a single functional allele of the Shh receptor patched and therefore, constitutively activated Shh signaling – have a greater number of proliferative cells in the retinal margin and in the CE that persist up to PND15. Even greater postnatal proliferation was observed at that timepoint when these mice were crossed with pro23his (rhodopsin mutant) mice, which undergo severe retinal degeneration. The ptc+/− mice were also more responsive to intravitreal injection of EGF and FGF growth factors but that was only evaluated at PND10 (Moshiri and Reh, 2004). Thus, Hh signaling appears to enhance postnatal progenitor proliferation. However, proliferating cells were very rare in adult ptc+/− mice, indicating the effect does not persist. In an ephrin-A3 knockout model, mice were found to have significantly increased BrdU incorporation in the adult CE and peripheral retina, and also produced 2-fold more RSC spheres in culture than wildtype mice (Fang et al., 2013). Interestingly, when ephrin-A3-/- RSC spheres were differentiated, they produced more progeny expressing markers of retinal neurons and photoreceptors than wildtype spheres. The authors conclude that ephrin-A3 signaling normally suppresses Wnt/β-catenin signaling in the adult eye, and thus in the KO mice, elevated Wnt signaling in the CE causes increased proliferation, RSC expansion, and the subsequent shift in differentiation. Consistent with the findings of Coles et al. (2006) (discussed in 1.4.2), where Vsx2/Chx10 mutants had an expanded CE and increased RSC population, Dhomen et al. (2006) used repeated BrdU injections to show that the embryonic retinal progenitor pool was markedly reduced in Vsx2/Chx10 null mice, while the number of proliferating cells in the CE was increased and persisted into the adult (Dhomen et al., 2006). Also, congruent with the idea that cellular feedback affects RSCs in vivo, Kiyama et al. (2012) found that Atoh7 mutants, with varying degrees of RGC loss, resulted in cell

32 proliferation and neurogenesis in the pars plana that continued into adulthood (Kiyama et al., 2012).

In the early postnatal mouse retina, before retinogenesis is completed, intravitreal delivery of FGF2 and insulin, or PEDF, significantly increases the number of proliferating CE and retinal progenitor cells (Das et al., 2004; Zhao et al., 2005), and results in the migration and differentiation of these cells into retinal neurons and Müller glia (Zhao et al., 2005). However, this effect continuously decreases with postnatal age until it becomes negligible (by PND21), suggesting inhibitory signals in the RSC niche annul the effects of growth factors after the conclusion of retinogenesis. However, some studies have reported limited stimulation of CE proliferation/differentiation by exogenous factors in the adult mammalian eye. Abdouh and Bernier (2006) found that FGF2 and insulin injection into the adult rat eye induced CE cell proliferation. Yet, they reported there was no evidence of migration or differentiation into retinal cells (Abdouh and Bernier, 2006). Another strategy that has been explored to stimulate endogenous RSCs is to use exogenous factors known for their regulation of other classes of stem cells. For example, due to its effectiveness in stimulating the proliferation of NSCs, Wang et al. (2010) tested systemic injections of the selective serotonin reuptake inhibitor, paroxetine, and detected a modest increase in BrdU-positive cells in the adult rat CE compared to control (~3.5% vs ~2.5%, respectively) (Wang et al., 2010). Likewise, because Rho GTPase inhibitors promote pluripotent stem cell maintenance and regulate various processes in other stem/progenitor cells, Del Debbio et al. (2014) performed intravitreal injections of general Rho GTPase inhibitor Toxin A in the adult mouse eye and showed that it increased ki67 labeling in the CE. Further, Toxin A had an additive effect with growth factors (FGF and insulin), which on their own had a very small effect (Del Debbio et al., 2014). However, cell type markers were not assessed, and thus, it is possible other cell types in the CE, such as endothelial cells, microglia or macrophages could be included in the ki67+ population (discussed further in Chapter 2).

Despite these findings, to date, the adult mammalian retina is still not considered to have any relevant regenerative capacity (Aladdad and Kador, 2019; Rao et al., 2017; Shen, 2020; Singh et al., 2020). This is likely, in part, because none of the paradigms used to stimulate endogenous RSCs has demonstrated any improvements in visual function. Most often, the inhibitory environment in the RSC niche is cited as the fundamental barrier to activating endogenous RSCs, and has even been described as “insurmountable” (Balenci et al., 2013; Wohl et al., 2012). However, work toward identifying the inhibitory factors that mediate RSC quiescence in vivo may lead the way to overcoming this regeneration roadblock (Balenci et al.,

2013). Another recent and significant advancement, two lineage tracing studies have conclusively demonstrated that the CE participates in retinogenesis of the peripheral NR during development (Bélanger et al., 2017; Marcucci et al., 2016). Although it remains unclear to what extent the peripheral retina is derived from the CE in mammals, this shows that RSC-mediated neurogenesis is a normal developmental process and gives credence to the hypothesis that this capacity can be reactivated in the adult mammalian eye.

1.4.4 Other potential adult stem cells in the mammalian retina

Though CE-RSCs are the focus of this thesis, it is notable that other cell types in the adult mammalian retina have been characterized to have stem cell-like and/or neurogenic potential. These putative adult retinal stem cells include Müller glia, RPE, Iris epithelium and Lgr5+ amacrine cells.

As mentioned, Müller glia (MG) are well characterized for their role as proliferating progenitors in the zebrafish. Therefore, they have also been of interest as a potential source of regeneration in the mammalian retina. In retinal injury models in the adult rodent retina, MG have been shown to proliferate, express RSC markers (such as Pax6, Vsx2/Chx10 and Sox2) and can produce progeny that express retinal neuron markers when stimulated with EGF or FGF and insulin (Bernardos et al., 2007; Bhatia et al., 2011; Karl et al., 2008; Webster et al., 2017). Several in vitro studies have identified a population of human retina-derived MG cells that are able to differentiate into RPE cells and retinal neurons, including those with photoreceptor markers (Giannelli et al., 2011; Johnsen et al., 2012, 2018; Roesch et al., 2008). Transplantation of human MG-derived photoreceptors into a rat model of photoreceptor degeneration showed these cells were able to migrate and integrate into the host ONL and improve visual function (Jayaram et al., 2014). Thus, these studies indicate that MG cells have stem cell-like properties and may be a potential source of neural regeneration for retinal degenerative diseases in adult mammals, including humans.

Several studies in different vertebrate species, including mammals, have indicated that embryonic RPE is capable of transdifferentiation and generating new neurons (Fischer and Reh, 2001). During transdifferentiation, ~10% of RPE-derived cells lose their pigmentation, expand in culture and express stem cell markers. These so-called RPE-derived stem cells (RPESCs) are able to generate retinal neurons, glial cells, mesenchymal cells, and of course, RPE (Luo and

Chen, 2018; Reh et al., 1987; Salero et al., 2012; Stanzel et al., 2014). Although several in vitro studies have demonstrated the neuronal transdifferentiation capability of RPE in adult rats and humans (Amemiya et al., 2004; Engelhardt et al., 2005; Salero et al., 2012), their function as RPCs has not been clearly documented in vivo (Chichagova et al., 2018); though subretinal transplantation of RPESC-derived RPE have been shown to prevent photoreceptor degeneration (Davis et al., 2016; Stanzel et al., 2014). In vivo, Adult rat RPE demonstrated the capacity to proliferate in situ but not to transdifferentiate into other retinal cell types (Al-Hussaini et al., 2008). Together, these studies indicate that RPE cells, or perhaps a subset of RPE cells with stem-cell like properties, may be capable of repair and regeneration of RPE and retinal neurons.

Some studies have demonstrated that iris epithelial cells from rodents exhibit RPC properties, can form clonal spheres, and have the potential to generate retinal-specific cells in vitro (Arnhold et al., 2004; Asami et al., 2007; Sun et al., 2006). Only the inner pigmented epithelial (IPE) layer contain these nestin-positive “RPCs”, which can proliferate and differentiate into diverse retinal cell types in the presence of growth factors (Asami et al., 2007). Adult pig IPE cells, in both sphere and monolayer cultures, were able to differentiate into neurons and glia (MacNeil et al., 2007). Likewise, iris tissue derived from adult rat eyes showed the ability to generate cells expressing differentiated neuronal markers (Arnhold et al., 2004). More recently, a tissue culture protocol involving matrigel embedding of adult porcine IPE-derived cells reported significant differentiation to neuronal and rod photoreceptor-like cells without the need for serum or growth factors in the media (Royall et al., 2017). Yet, in many of these studies, IPE cells maintain properties of differentiated epithelial cells and lack the capability to differentiate into retinal neurons (Frøen et al., 2011). Therefore, further studies are needed to characterize the true potential of IPE cells and determine if their reported neurogenic ability will have relevance clinically or remain a biologically interesting investigation.

A single study has described leucine-rich repeat containing G protein receptor 5 (Lgr5) expressing amacrine cells as a potential source of retinal progenitor cells in the adult mouse retina (Chen et al., 2015). Lgr5 – which is involved in regulation of the Wnt pathway (Carmon et al., 2012) and has been used as a marker for adult stem cells in other tissues, such as the intestine (Schuijers and Clevers, 2012) – had previously been shown to be expressed in glycinergic amacrine cells in the adult mouse retina (Sukhdeo et al., 2014). Using Lgr5-eGFP mice and EdU labeling, Chen et al. (2015) observed that Lgr5+ cells spontaneously proliferate in the adult retina under naïve conditions. Further, they found evidence that Lgr5+ cells could

35 generate Müller glia, bipolar cells and photoreceptor progeny, with or without retinal injury. However, a recent study isolated Lgr5+ and Lgr5- cells from PND5 and adult mouse retinas and compared their transcriptomic profiles. They determined that Lgr5+ cells do not show any expression of stem cell genes. They also determined that Wnt signaling is not active in the Lgr5+ population and that these cells, in fact, had the expression profiles of differentiated glycinergic amacrine cells (Trepp et al., 2019). Thus, it is still unclear whether Lgr5+ amacrine cells represent a true subpopulation of cells with regenerative capacity in the adult retina. As the study by Trepp et al. did not investigate any injury models, it is still possible that retinal injury might induce stem cell properties/gene expression in retinal Lgr5+ cells.

Thus, although the adult mammalian retina is considered to be incapable of regeneration, multiple endogenous cell types have been reported to have stem/progenitor-like properties and are being investigated for their potential to generate new retinal cells in vivo. Perhaps future work will look to activating multiple cell types in concert to regenerate the mammalian retina. Yet, to date, CE-RSCs have been the most thoroughly characterized adult stem cell in the mammalian eye.

1.6 Thesis Hypotheses & Aims

Chapter 2

A previous study by our lab discovered that Bone Morphogenic Proteins (BMP) 2 & 4 and secreted frizzled-related protein 2 (sFRP2) are secreted by the adult lens and cornea and are capable of suppressing RSC proliferation and clonal RSC-derived sphere growth in vitro (Balenci et al., 2013). Moreover, when BMP antagonist, Noggin, or an anti-sFRP2 function blocking antibody was added to the media, RSC proliferation inhibition was reversed. BMP2 and BMP4 proteins are expressed in the developing CE and are required for normal CE development and morphogenesis (Zhao et al., 2002b; Zhou et al., 2013). sFRP proteins are well characterized as Wnt antagonists (Satoh et al., 2006, 2008; Tran and Zheng, 2017) and Wnt signaling is required for specifying the CE during development (Cho and Cepko, 2006; Lad et al., 2009; Liu et al., 2003) and can modulate the number of RSCs in vivo (Liu et al., 2007). In addition, BMP and Wnt signaling are known to regulate retinal stem and progenitor cell proliferation and differentiation in the non-mammalian CMZ (Denayer et al., 2008; Haynes et al.,

2007). Thus, in Chapter 2, I investigated the hypothesis that BMP and sFRP2 proteins in the RSC niche maintain RSCs in a quiescent state in the adult mouse eye. This work aimed to determine if counteracting BMP and sFRP2 signaling can induce endogenous RSC proliferation in vivo, and/or render endogenous RSCs amenable to stimulation by exogenous growth factors. Further, I hypothesized that if endogenous RSC proliferation can be facilitated by abrogating inhibitory niche signals, RSCs may then be able to respond to retinal injury and contribute newborn retinal neurons to the injured retina. Thus, I also aimed to determine if BMP and sFRP2 antagonism would enable RSC-derived neurogenesis in vivo and confer regenerative capacity to the adult mammalian retina in response to injury.

Chapter 3

In Chapter 2, I followed up on evidence from developmental biology and in vitro experiments to investigate specific factors for their role in regulating RSC proliferation. However, it is likely that there are molecular pathways regulating RSC proliferation that have not yet been implicated by observational or experimental data. Thus, in Chapter 3, I aimed to discover entirely novel molecular mechanisms that regulate RSC proliferation. To do so, I investigated an unbiased phenotypic screening approach that measured indices of proliferation as a strategy to identify positive regulators of RSC proliferation. Further, by screening small molecules with well characterized molecular targets, I hypothesized that specific signaling pathways and mechanisms that regulate RSC proliferation could be identified. Further, I hypothesized that any mechanism that regulates RSCs in culture may also regulate RSCs in vivo. Thus, this work aimed to identify molecular pathways that regulate RSC proliferation in vitro and in vivo. Concurrently, by virtue of screening drugs, this work aimed to identify small molecules that could be applied in culture to scale up retinal progenitor production or in vivo to stimulate endogenous RSC proliferation.

Chapter 4

Although several studies have suggested that the cone fate may be the default differentiation program of retinal progenitors and precursors (Akimoto et al., 2006; Brzezinski and Reh, 2015; Mears et al., 2001; Swaroop et al., 2010; Szél et al., 1994), RSC progeny produce less than 1% cone photoreceptors under standard differentiation conditions. This suggests either the intrinsic properties of RSC progeny or extrinsic factors in culture are preventing cone specification. Recently, COCO, a Cerberus/Dan family protein that inhibits TGFβ, BMP and Wnt signaling,

37 was demonstrated to induce the differentiation of human embryonic stem cells into cone photoreceptors (Zhou et al., 2015). This led us to hypothesize that exogenous signals are the predominant factors influencing cone fate specification. Thus, in Chapter 4, I investigated whether COCO-mediated TGFβ, BMP and Wnt antagonism can direct the differentiation of RSC progeny to the cone photoreceptor fate and aimed to elucidate the cell biological mechanisms involved in cone specification. Further, by analyzing the transcriptomes of endogenous cones and RSC-derived cones, I aimed to gain insight into the developmental maturity of RSC-derived cones and identify conserved genetic pathways and novel genes involved in cone specification

General Hypothesis

Overall, my general hypothesis is that environmental factors have a profound influence on RSC proliferation and differentiation, and therefore environmental manipulations will be able to activate quiescent RSCs to proliferate and differentiate in vivo, as well as influence the fate specification of RSC progeny in vitro.

Chapter 2

Reactivation of proliferation and neurogenesis in the adult mammalian retinal stem cell niche via antagonism of BMP and sFRP2

*This chapter was submitted to Cell Reports

Kenneth N. Grisé, Brenda L.K. Coles1, Nelson X. Bautista1, Derek van der Kooy. Reactivation of proliferation and neurogenesis in the adult mammalian retinal stem cell niche via antagonism of BMP and sFRP2. Submitted August 2020.

K.N.G designed, executed and analyzed all experiments and wrote the manuscript with the assistance of the other authors as indicated. B.L.K.C. helped perform sphere assay dissections in Fig. 2.1, Fig. 2.2 and primary dissections for FACS experiments in Fig. 2.3/Fig A7. N.X.B genotyped Msx1-Cre and td-Tomato transgenic mouse lines, assisted with tissue processing and sectioning for Fig. 2.4-2.6/Fig. A10-14, and quantified outer nuclear layer thickness in Fig. A8. D.v.d.K. contributed to experimental design, analyses, and edited the manuscript.

2.1 Abstract

The adult mammalian retina does not have the capacity to regenerate cells lost due to damage or disease. However, retinal stem cells (RSCs) in the ciliary epithelium (CE), which participate in retinogenesis during development, persist in a quiescent state in the adult eye and retain the ability to generate all retinal cell types, including photoreceptors, when cultured in vitro. Here, we show that in vivo antagonism of two proposed regulators of RSC quiescence, BMP and sFRP2 proteins, enables the RSC niche to proliferate, respond to exogenous growth factors and injury, and expand the RSC pool. Furthermore, using genetic lineage tracing, we show that CE cells can migrate into the retina and express markers of mature photoreceptors and retinal ganglion cells. Together, these results indicate that endogenous adult mammalian RSCs have latent regenerative potential that can be reactivated by modulating the RSC niche.

2.2 Introduction

The pigmented ciliary epithelium (CE) in the adult mammalian eye harbours a rare subpopulation of retinal stem cells (RSCs) that are capable of clonal expansion, self-renewal, and differentiation into all the cell types of the retina when isolated in vitro. (Ahmad et al., 2000; Ballios et al., 2012; Coles et al., 2004; Inoue et al., 2010; Khalili et al., 2018; Tropepe et al., 2000). The CE is an epithelial bilayer that overlays the ciliary body and is contiguous with the peripheral retina and retinal pigmented epithelium (RPE) (Delamere, 2005). The anatomical location and properties of CE-RSCs has led to their comparison with the stem cells in the ciliary marginal zone (CMZ) of non-mammalian vertebrates, which have the capacity for retinal neurogenesis and regeneration in response to injury throughout the lifespan of the organism

(Aladdad and Kador, 2019; Fischer et al., 2013). However, unlike stem cells in the CMZ, CE-RSCs become quiescent in the early postnatal period and remain dormant in vivo throughout maturity (Kubota et al., 2002; Moshiri et al., 2004; Perron and Harris, 2000; Reh and Fischer, 2006). Indeed, it is typically held that the adult mammalian RSC niche does not respond to exogenous stimulation or injury, and does not have neurogenic potential in vivo (Aladdad and Kador, 2019; Kubota et al., 2002; Moshiri et al., 2004; Perron and Harris, 2000; Reh and Fischer, 2006; Wohl et al., 2012). Further, two studies reported that CE-RSCs do not differentiate into mature retinal neurons in vitro, leading the authors to challenge previous findings and the designation of RSCs as stem cells (Cicero et al., 2009; Gualdoni et al., 2010). Yet, more recent evidence that adult RSC progeny can generate functional photoreceptors in vitro (Del Debbio et al., 2013; Demontis et al., 2012) and in vivo following transplantation (Ballios et al., 2015; Inoue et al., 2010) corroborates the neurogenic capacity of adult RSCs and gives credence to the premise that such potential could be evoked in endogenous RSCs. Furthermore, in vivo lineage tracing studies have revealed that the mammalian CE contributes to retinogenesis during development, when CE cells migrate into the retina and generate all seven major retinal cell types (Bélanger et al., 2017; Marcucci et al., 2016). Together, these findings have led to the hypothesis that the in vivo quiescence of RSCs in the adult mammalian CE, and their insensitivity to stimulation, may be mediated by inhibitory factors in the adult RSC niche (Balenci et al., 2013; Perron and Harris, 2000; Wohl et al., 2012).

A previous study by our lab discovered that Bone Morphogenetic Proteins (BMP) 2 & 4 and secreted frizzled-related protein 2 (sFRP2) are secreted by the adult lens and cornea and are capable of suppressing RSC proliferation and clonal RSC-derived sphere growth in vitro

(Balenci et al., 2013). Moreover, when BMP antagonist, Noggin, or an anti-sFRP2 function blocking antibody was added to the media, RSC proliferation inhibition was reversed. BMP2 and BMP4 proteins are expressed in the developing CE and are required for normal CE development and morphogenesis (Zhao et al., 2002b; Zhou et al., 2013). sFRP proteins are well-known as Wnt antagonists (Satoh et al., 2006, 2008; Tran and Zheng, 2017) and Wnt signaling is required for specifying the CE during development (Cho and Cepko, 2006; Lad et al., 2009; Liu et al., 2003) and can modulate the number of RSCs in vivo (Liu et al., 2007). In addition, BMP and Wnt signaling are known to regulate retinal stem and progenitor cell proliferation and differentiation in the non-mammalian CMZ (Denayer et al., 2008; Haynes et al., 2007). Thus, we hypothesized that BMP2/4 and sFRP2 may be the main inhibitory factors in the adult eye that mediate RSC quiescence (Balenci et al., 2013).

In this study we show that intravitreal injection of Noggin or anti-sFRP2 induces proliferation of CE cells and expands the RSC population in vivo. Using the inducible Msx1-CreERT2 mouse line, we were able to lineage label the adult CE and revealed that RSCs and their progeny were also labeled. By lineage labeling the CE, we show that combinatorial injection of noggin, anti-sfrp2, FGF2 and insulin, with or without retinal injury, induces CE cells to migrate into the retina and differentiate into retinal neurons. These findings demonstrate that BMP and sFRP2 proteins are regulators of RSC quiescence in the adult mouse eye and provide evidence that modulating the RSC niche microenvironment can reactivate the proliferative and neurogenic potential of the adult mammalian CE.

2.3 Materials & Methods

2.3.1 Mice

All mouse protocols were approved by the Animal Care Committee at the University of Toronto, which operates in accordance with the Canadian Council on Animal Care. Mice were kept on a 12-hour light dark/light cycle. Food was available ad libitum. Water was supplied ad libitum except during EdU delivery (see below). Adult mice used in this study were a minimum of 8 weeks old and included: CD1 mice (022, Charles River), C57/BL6J mice (000664, Jackson Laboratories), B6.Cg-Gt(Rosa26)SorTm14(CAG-TdTomato)Hze mice (Madisen et al., 2010) (007914, Jackson Laboratories) and Msx1-CreERT2 mice (Lallemand et al., 2013). The Msx1-CreERT2 mouse line was a gift from Dr. Michel Cayouette and Dr. Benoît Robert and the CreERT2 construct originated from the IGBMC via Pierre Chambon. Msx1-CreERT2 is a transgenic line

41 where the CreERT2 fusion protein is expressed in place of the endogenous Msx1 protein, as it has been knocked in, in frame, in the first exon of Msx1. Msx1-CreERT2 mice were crossed with B6.Cg-Gt(Rosa26)SorTm14(CAG-TdTomato)Hze mice to generate Msx1-CreERT2;B6.Cg-Gt

(Rosa26)SorTm14(CAG-TdTomato)Hze mice, which were used for lineage tracing experiments. Each mouse used was genotyped by PCR amplification. For full list of genotyping primers see Table A1.

2.3.2 Drug and Protein Preparations

EdU (Thermo; A10044) was dissolved at 0.2mg/mL in 1% sucrose (BioShop; SUC507.1) in ddH2O and placed into 50mL Falcon tubes fitted with standard rubber water stoppers/sipper tubes for ad libitum access during intravitreal injection periods. The EdU water level was topped up daily.

Tamoxifen (Sigma; T5648) was dissolved in 1:10 anhydrous EtOH (Sigma; 676829) to sunflower seed oil (Sigma; S5007), vortexed and incubated for 15min in a 37°C water bath. It was injected i.p. at 180mg/kg, once a day for 4 days, in order to induce reporter expression in Msx1-CreERT2;B6.Cg-Gt (Rosa26)SorTm14(CAG-TdTomato)Hze mice.

CHIR99021 (Sigma; SML1046) was dissolved in DMSO (Sigma; 276855) and then diluted to 0.525µM or 5.25µM to arrive at 1.75% DMSO in PBS, which was then injected into the 7µL intravitreal space (Schlichtenbrede et al., 2009) to arrive at a final in vivo concentration estimated at 0.15µM or 1.5µM, respectively, in 0.5% DMSO. LDN-193189 (SelleckChem; S2618) was dissolved in DMSO and then diluted to 0.35µM or 3.5µM to arrive at 1.75% DMSO in PBS, which was then injected into the 7µL intravitreal space to arrive at a final in vivo concentration estimated at 0.1µM or 1µM, respectively, in 0.5% DMSO.

Noggin (R&D Systems; 719-NG-050), anti-sFRP2 (R&D Systems; MAB1169), FGF2 (100ng/eye; R&D Systems; 3139-FB-025/CF) and Insulin (2µg/eye; Sigma; 16634) were dissolved directly in PBS. All concentrations listed for proteins in the manuscript are the final in vivo concentrations accounting for the 7µL intravitreal space.

2.3.3 Intravitreal Injections

Intravitreal injections were carried out using a 10µL WPI Nanofil® Injector System with micro- machined 34-gauge beveled needle (World Precision Instruments, Sarasota, FL), a dissecting

42 microscope or surgical scope (Moller Hi-R 900C), a mouse stereotaxic apparatus and heat pad. Mice were brought to a surgical plane of anesthesia via 5% isoflurane and placed on the heat pad in the mouse stereotaxic apparatus (without head stabilization with the ear bars). Once anesthetized, isoflurane was reduced to 3% for maintenance. Mice were injected with 2mg/Kg meloxicam for analgesia. One drop of anticholinergic mydratic (Mydriacyl®) was applied to each mouse eye to dilate pupils. Mice were positioned on one side, so that the eye to be operated on was facing upward, directly under the surgical microscope. A small rubber washer was placed over the eye, so that the washer surrounds the eye like a monocle. A single drop of 3% methylcellulose (MC) solution (in saline) into the monocle, which allows clear visualization of the posterior segment of the eye by the surgeon. The mouse head was stabilized with the non- dominant hand and the needle was controlled with the dominant hand. With the needle, a trans- scleral puncture was made (at a perpendicular angle to the globe) approximately 1 mm posterior of the limbus, in the nasal (anterior) aspect of the eye. The needle passed through the sclera, choroid and retina to enter the retrolental vitreous. The needle was inserted as far as the central area of the retina, taking care to avoid striking the lens, retina or the hyaloid canal. A 2µL bolus of fluid was then injected at an approximate rate of 4µL/min. The adult mouse vitreous space can accommodate up to 3 ul of total fluid because it replaces fluid initially lost from pre-injection vitreous outflow. Thus, the final vitreous volume in the eye is the same as the standard vitreous volume of the mouse eye (7µL) (Schlichtenbrede et al., 2009). Once the injection was completed, the needle remained in the retrolental vitreous for an additional 10-15 seconds. This allows for pressure equilibration and works to prevent significant backflow following withdrawal of the needle. Next, the needle was removed, the monocle was removed, and the mouse is rotated to position the other eye for surgery. Once the surgery on both eyes was completed, the mouse was left to recover alone in a recovery cage with a heat lamp, and then reunited with its original cage-mates.

2.3.4 Immunohistochemistry

Mice were euthanized by cervical dislocation while under isoflurane anesthesia. Eyeballs were enucleated from adult mouse skulls, post-fixed in 4% PFA for 4 hours at 4°C, then transferred to a cryoprotectant 30% sucrose solution for a minimum of 24 hours at 4°C. Next, eyes were embedded in Tissue Tek, frozen at -80°C and then sectioned at 10 µm using a cryostat. Fixed frozen eye slides were permeabilized with 0.3% Triton X-100 (Sigma; T8787) in PBS for 20 min. Then, they were blocked in 10% normal goat serum (NGS) or 10% normal donkey serum (NDS)

43 for 1 hour. Primary antibodies were diluted in 1% serum from the species used for blocking (to the dilutions indicated below) and incubated overnight at 4°C. After washing, secondary antibodies were diluted in 1% serum of the same species at 1:400 (Alexa fluor; ThermoFisher) and incubated for 1 hour. After washing, nuclei were stained with Hoechst 33258 (10μg/mL; ThermoFisher; H1399) or DRAQ5 (1uM; ThermoFisher; 62251) for 20 min before a final wash. Mounting medium was added to wells or slides and slides were then coverslipped. Primary antibodies used in this study were used at the following dilutions: rabbit anti-Ki67 (1:100; Ab15580; Abcam), rabbit anti-Recoverin (1:1000; AB5585; Millipore), goat anti-Chx10 (1:1000; Sc-21690; Santa Cruz), rabbit anti-cone arrestin (1:1000; AB15282; Millipore), mouse anti- rhodopsin (1:500; MAB5316; Millipore), goat anti-Brn3a (1:500; SC-31984; Santa Cruz) rabbit anti-Pax6 (1:1000; AB2237; Millipore), rabbit anti-ERG (1:250; AB92513; Abcam) and rat anti- CD68 (1:500; MCA1957; BioRad). For EdU detection, Click-iT™ EdU detection kits were used according to the manufacturer’s instuctions (ThermoFisher: 488 dye [C10337], 555 dye [C10638], 647 dye [C10640]). Fluorescence images were acquired using a Zeiss AxioObserver D1 inverted fluorescence microscope (Zeiss) equipped with Zeiss Axiovision software (v4.8.2) or Olympus FV1000 laser scanning confocal microscope (Olympus Life Science) using Olympus Fluoview software (FV10-ASW v4.2b).

2.3.5 Isolation of Retinal Stem Cells from the Ciliary Epithelium of the Adult Eye and Primary Clonal Sphere Assay

44 was inserted between the CE and the iris, using pressure to slice the iris/cornea side off from the rest of the shell. Next, the scalpel was run along the border between the CE and the RPE to obtain the CE isolated as a thin strip of tissue. The CE strips were then transferred to a 35mm dish containing 2mL of dispase solution (Sigma; T1005) and incubated for 10 minutes at 37°C. Next, the strips were transferred from dispase into a 35mm dish containing 2mL of sterile filtered kynurenic acid (02.mg/mL; Sigma), trypsin (1.33mg/mL; Sigma) and hyaluronidase (0.67mg/mL; Sigma) in high magnesium/low calcium artificial cerebral spinal fluid (hi/lo aCSF) and incubated at 37°C for 10 minutes. After incubation, the dish was returned to the dissecting scope, and the CE strips were pinned down with straight, non-serrated forceps, while non-serated curved forceps were used to scrape the CE off from the underlying sclera. The bare scleral strips were then discarded, such that only the CE cells remained in the enzyme solution. Using a fire- polished, cotton-plugged glass pipette, the cells and enzyme solution were transferred to a 15mL tube and triturated approximately 45 times to break apart the tissue. The 15mL tube with cell suspension was centrifuged for 5 minutes at 1500 rpm. The supernatant was gently aspirated from the resulting pellet using a fire-polished, cotton-plugged glass pipette and 2mL of sterile-filtered ovomucoid trypsin inhibitor (1mg/mL; Sigma) in serum-free media (SFM) was added to the pellet. Using a small borehole, fire-polished, cotton-plugged glass pipette, the sample was triturated approximately 45 times to generate a single-cell suspension. The 15mL tube with cell suspension was centrifuged for 5 minutes at 1500 rpm. The supernatant was gently aspirated from the resulting pellet and 1-2mL of SFM with fibroblast growth factor 2 (FGF2, 10ng/mL; Sigma) and heparin (2μg/mL; Sigma) (SFM+FH) was added. The cells and media were mixed to ensure a uniform cell suspension and a 10µL sample was taken for cell density determination. The cells were then seeded and cultured at 10 cells/μL in culture-treated plates or flasks and incubated in a humidified incubator at 37°C in 5% CO2 and ambient room

O2. After one week, roughly 1 in 500 cells are expected to proliferate to form free-floating, clonal spheres greater than 80μm in diameter.

2.3.6 Fluorescence Assisted Cell Sorting

The ciliary epithelium was dissected dissociated to single cells as outlined in the above section “Isolation of Retinal Stem Cells from the Ciliary Epithelium of the Adult Eye and Primary Clonal Sphere Assay”. The single cell suspension was filtered through a 40µm cell strainer. Cells were counterstained with DAPI (0.1µg/mL; Invitrogen; D1306) to assess viability. The CE from from C57/BL6J mice was used to set the gating for pigmented and non-pigmented CE, based on

45 forward and side scatter, and to gate negative TdTomato (red) fluorescence using a FACS Aria II (BD Biosciences). The single cell suspension from the CE of tamoxifen-treated Msx1- CreERT2;B6.Cg-Gt (Rosa26)SorTm14(CAG-TdTomato)Hze mice was sorted into pigmented and non- pigmented fractions, and also fluorescent and non-fluorescent fractions using endogenous TdTomato expression. After sorting, cells were plated at 1 cell/μL in 500μL of SFM+FH per well in 24-well plates for a 7-day sphere assay. Flow cytometry analysis was performed using BD FACS Diva Software V6.1.2.

2.3.7 Statistical Analysis

Data are presented as mean ± standard error (SE) unless otherwise noted. Statistical analyses were run using Sigmaplot 12 (Systat Software Inc.) or GraphPad Prism 6 (GraphPad Software Inc.). Student’s t-test (two-tailed) was performed for statistical analysis between two groups. A one-way ANOVA or two-way ANOVA with a Holm-Sidak post-hoc test (pairwise or versus control comparison) was used when three or more groups were compared. Sample size (N) and p-values are provided in the figure legends. Statistical significance was set at p < 0.05.

2.4 Results

2.4.1 Intravitreal injection of Noggin or anti-sFRP2 stimulates ciliary body- specific proliferation and expands the retinal stem cell population

To determine if blocking endogenous BMP proteins or sFRP2 proteins present in the adult mouse eye can modulate the proliferation of cells within the CE or the neural retina (NR), we delivered the BMP antagonist, Noggin, or a function blocking anti-sFRP2 antibody in vivo via intravitreal injection (Figure 2.1A). Albino CD1 mice were used to more easily visualize cells that might otherwise be obscured by the pigment of the outer CE. Mice received one injection per eye, per day, for 3 consecutive days (2µL volume per injection). For the first four days of the experiment the thymidine analog 5-ethynyl-2’-deoxyuridine (EdU), which is incorporated into the DNA of cells during the S phase of the cell cycle (Zeng et al., 2010), was delivered via the drinking water (0.2mg/mL). Following the injection period, eyes were fixed, frozen and sectioned for analysis of EdU labeling. At Day 4 and Day 31, anti-sFRP2 treatment resulted in a discernible increase in EdU+ cells in the CE (Figure 2.1C,F) relative to PBS control (Figure 2.1B,E). At Day 4, EdU+ cells were located throughout the CE, usually as single or doublet cells, in both the inner and outer CE layers. At Day 31, larger clusters of EdU+ cells were evident. To

46 quantify differences between conditions, the number of EdU+ cells was counted and normalized by regional area of the CE or NR. Here, naïve, un-injected eyes were also examined to assess any effects of PBS vehicle injection on EdU labeling. At Day 4, the 2.5µg/mL dose of anti-sFRP2 increased EdU labeling in the CE nearly 2-fold relative to PBS (Figure 2.1L). In contrast, neither dose of anti-sFRP2 increased EdU+ cell number in the NR, though a small effect of PBS injection versus naïve control was observed. At Day 31, anti-sFRP2 resulted in a ~4-fold increase in EdU+ cells relative to PBS control (Figure 2.1M). These results suggest anti-sFRP2 antibody binding and antagonism of sFRP2 proteins in the vitreous results in dose-dependent, CE-specific proliferation.

Intravitreal Noggin treatment resulted in a discernible increase in EdU+ cells in the CE at both Day 4 and Day 31 (Figure 2.1D,G). The pattern of EdU+ cell localization in the CE after Noggin treatment was similar to anti-sFRP2 treatment, with single and doublet cells evident in both layers of the CE at Day 4, and larger cell clusters observed at Day 31. Quantitatively, at Day 4, both doses of Noggin increased EdU labeling in the CE relative to PBS control, by roughly 3-fold and 2-fold, respectively (Figure 2.1N). There also was an effect of PBS injection versus naïve control in the CE. In contrast, there was no effect of PBS or Noggin on EdU labeling in the NR. At Day 31, though there was a trend toward increased EdU+ cell number in the CE or NR with Noggin injection, it was not statistically significant (Figure 2.1O). These results indicate that Noggin binding and antagonism of BMP proteins induces CE-specific proliferation. The non- significant trend of Noggin treatment at Day 31 could have resulted from intermittent EdU+ cell death, cell migration out of the CE, or possibly due to dilution of the EdU label by persistent proliferation (Bonhoeffer et al., 2000).

To examine the proportion of EdU+ cells that continue to proliferate following the intravitreal injection period, we immunostained for the proliferation marker Ki67. At Day 4, both Ki67+ cells and EdU+Ki67+ co-labeled cells were detected in anti-sFRP2 and Noggin treated eyes, but not in PBS controls (Figure 2.1H-J). This demonstrates that a subset of CE cells were still proliferating 24 hours after the injection period. However, there was a low proportion of EdU+Ki67+ cells (~18-20%), indicating the majority of EdU+ cells labeled during the injection period are no longer dividing 24 hours later (Figure 2.1K). At Day 31, no Ki67 was detectable in any condition; thus, proliferation in the CE does not persist one month after intravitreal injection.

To determine if antagonism of sFRP2 and BMPs has an additive effect on EdU labeling, we performed combinatorial intravitreal injections of sFRP2 and Noggin. Once again, anti-sFRP2 or

Noggin treatment alone resulted in increased EdU labeling in the CE relative to PBS control (Figure 2.1P). Anti-sFRP2 and Noggin combined had a greater effect than Noggin alone, but not anti-sFRP2 alone, suggesting there is not an additive effect of antagonism of BMP proteins with sFRP2 antagonism. Once again, no effect on EdU labeling was observed in the NR, suggesting that sFRP2 and BMP blockade have specific effects in the CE.

We observed that some EdU+ cells in the CE and NR co-labelled for endothelial cell and microglia/marcrophage markers (Figure A1, A2). Therefore, we sought to refine our EdU labeling analysis to specifically analyze CE cells. There are currently no markers that uniquely identify RSCs in vivo. However, Pax6 marks retinal progenitor cells during eye development and has been shown to be highly expressed and functionally required in RSCs (Xu et al., 2007b). Also, Pax6 is known to label both layers of the CE and amacrine cells in the adult mouse eye (Das et al., 2005; Marquardt et al., 2001). Yet, amacrine cells and CE cells marked by Pax6 are easily distinguished based on anatomical location in the retina vs the CB, respectively (Figure A3E). Henceforth, we quantified the number of EdU+ cells that were co-labeled with Pax6 to determine the level of CE-specific proliferation. Also, anti-sFRP2 and Noggin interfere with sFRP2 and BMP ligand-receptor interactions at the extracellular level, but they are expected to mediate their effects by modulating downstream Wnt and BMP signaling, respectively. To investigate the effects of modulating downstream Wnt and BMP signaling more directly, we injected adult mouse eyes with two ATP-competitive small molecule inhibitors that act at the intracellular level (Figure A3A). GSK3β inhibitor CHIR99021 disrupts the β-catenin destruction complex, enabling β-catenin to enter the nucleus and activate canonical Wnt target genes (Tran and Zheng, 2017). BMP inhibitor LDN-193189 selectively inhibits BMP receptor kinases, preventing SMAD phosphorylation and translocation to BMP target genes (Mohedas et al., 2013). Using Pax6 EdU analysis, we found that CHIR99021 and LDN-193189 each significantly increased the number of Pax6+EdU+ CE cells relative to DMSO control (Figure A3C). These findings confirm that modulating the BMP and Wnt signaling pathways can induce CE cell proliferation.

Not all stem cell proliferation results in an expansion in stem cell number (Morrison and Kimble, 2006; Post and Clevers, 2019). RSCs can divide symmetrically to generate two stem cells and increase RSC number, or divide asymmetrically to generate one stem cell and one progenitor cell and maintain the RSC pool (Baakdhah and van der Kooy, 2019; Balenci and van der Kooy, 2014; Coles et al., 2004; Tropepe et al., 2000). In vitro, single RSCs proliferate to form clonal spheres of cells. Thus, the number of RSC spheres is a measure of the number of endogenous

RSCs and can be used to detect changes in mode of division (Balenci and van der Kooy, 2014; Tropepe et al., 2000). To determine if injection of anti-sFRP2 or Noggin modulates the numbers of RSCs in vivo, we injected a broad dose-range of Noggin or anti-sFRP2 followed by a clonal sphere-forming assay. Seven days following the intravitreal injection period, the CE was dissected, dissociated to single cells, then plated at clonal density (10 cells/µL) for a 7-day sphere forming assay (Figure 2.1Q). For anti-sFRP2, only the 2.5µg/mL dose resulted in a significant increase in RSC sphere number, with a ~ 2.2-fold increase relative to PBS control (Figure 2.1R). For Noggin, only the 2µg/mL dose resulted in a significant increase in RSC sphere number, with a ~2.4-fold increase relative to PBS control (Figure 2.1S). Thus, intravitreal injection of anti-sFRP2 or Noggin induce expansion of the endogenous RSC population in a dose-specific manner. Together, these results demonstrate that antagonism of sFRP2 or BMP proteins in the adult mouse vitreous induces CE-specific proliferation and can expand the RSC pool in vivo.

Figure 2.1. Intravitreal injection of Noggin or anti-sFRP2 stimulates proliferation in the ciliary epithelium and increases primary sphere-forming retinal stem cell number

(A) Schematic of the intravitreal injection paradigm followed by endpoint IHC. Mice received one intravitreal injection per day for three days while EdU was delivered via the drinking water continuously until Day 4. Mice were euthanized for histochemical analysis on Day 4 or Day 31. Injections consisted of PBS control, Noggin or anti-sFRP2 at indicated concentrations.

(B-G) Fluorescence images of EdU labeling in the CE of eye sections from mice treated with PBS, anti-sFRP2 or Noggin at Day 4 and Day 31. (B, E) EdU labeling in PBS-treated eyes. (C, F) EdU labeling in Noggin-treated eyes. (D, G) EdU labeling in and anti-sFRP2-treated eyes. Hoechst stain was used to label all nuclei. White arrows indicate EdU+ cells. Straight dashed line indicates the border between the ciliary epithelium and the retina.

(H-J) Ki67 immunostaining and EdU labeling at the Day 4 timepoint in the CE of eye sections from mice treated with PBS, anti-sFRP2 or Noggin. (H) PBS-treated eyes. (I) Anti-sFRP2- treated eyes. (J) Noggin-treated eyes. White arrows indicate EdU+ cells. Green arrows indicate Ki67+ cells. Yellow arrows indicate EdU+Ki67 co-labeled cells. Hoechst stain was used to label all nuclei. Dashed line box indicates high magnification inset.

(K) Quantification of the proportion of EdU+Ki67 co-labeled cells relative to the total number of EdU-labeled cells in the CE at Day4 and Day 31. N=3-5 mice per group.

(L-O) Quantification of EdU-labeled cells in the CE and NR normalized by regional area in eye sections from mice treated with the indicated conditions. (L) Day 4 anti-sFRP2 (one-way ANOVA F(3,11)=8.0, p=0.004; N=3-6 mice per group). (M) Day 31 anti-sFRP2. (one-way ANOVA F(2,6)=9.27, p=0.015; N=3 mice per group). (N) Day 4 Noggin (one-way ANOVA F(3,11)=24.61, p=<0.001; N=3-6 mice). (O) Day 31 Noggin; N=3 mice per group.

(P) Quantification of EdU-labeled cells in the CE and NR normalized by area in eyes treated with PBS, Noggin, anti-sFRP2 or both Noggin+anti-sFRP2 at Day 4 (one-way ANOVA F(4,20)=21.76, p<0.001; N=3-6 eyes per group).

(Q) Schematic of the intravitreal injection paradigm followed by endpoint clonal RSC sphere forming assay from primary CE. Mice received one intravitreal injection per day for three days. On Day 10, mice were enucleated, and the CE was dissected for a subsequent 7-day clonal sphere growth assay in vitro.

(R-S) Quantification of RSC sphere frequency normalized to naïve un-injected control. (R) anti- sFRP2 dose-response (one-way ANOVA F(8,62)=7.49, p<0.001; * = p<0.05; N=5-16 eyes per group). (S) Noggin dose-response (one-way ANOVA F(8,67)=5.83, p<0.001; * = p<0.05; N=5-16 eyes per group). Each data point represents a single eye.

Statistics: One-way ANOVAs with Holm-Sidak posthoc tests. Data are means ± SEMs. * = p<0.05.

2.4.2 Combinatorial injections of Noggin, anti-sFRP2, FGF2 and Insulin have differential effects on CE proliferation versus retinal stem cell expansion

Given the evidence that BMP and sFRP2 are negative regulators of RSC proliferation in vivo, we hypothesized that antagonism of BMPs and sFRP2 may render RSCs more amenable to growth factor stimulation. We chose FGF2 and Insulin as our growth factors of interest due to the importance of FGF and Insulin/IGF signaling in retinogenesis (Agathocleous and Harris, 2009), evidence that FGF is imperative for adult RSC proliferation in vitro (Balenci and van der Kooy, 2014; Tropepe et al., 2000), and because FGF2 and Insulin have been reported to stimulate CE proliferation and neurogenesis in the perinatal CE/CMZ (Fischer and Reh, 2003a; Zhao et al., 2005). We examined two outcome measures in this experiment: 1) immunohistochemistry (IHC) to determine the number of Pax6+EdU+ co-labeled CE cells, and 2) a clonal, sphere-forming assay as a proxy for the number of RSCs in vivo (Figure 2.2A). Intravitreal injections included 6 different conditions: PBS control (C), Noggin (N), anti-sFRP2 (S), Noggin+anti-SFRP2 (NS), FGF2+Insulin (FI) and all factors combined (FINS). The primary sphere assays were normalized to naïve un-injected controls (NC). At Day 4, the number of Pax6+EdU+ CE cells was greater in all treatment conditions relative to PBS control (Figure 2.2B, Figure A4). FI alone increased CE proliferation to a similar extent as Noggin and anti-sFRP2 combined, revealing that growth factors can stimulate CE cell proliferation even with the RSC quiescence factors present. However, the largest effect was observed in the FINS treated eyes, which resulted in a ~21-fold increase in Pax6+EdU+ cells compared to PBS, and a ~2-fold increase compared to NS, which had the second highest effect. FINS also resulted in the largest clusters of Pax6+EdU+ cells evident in eye sections (Figure 2.2E; Figure A4F). Thus, sFRP2 and BMP antagonism and growth factor stimulation have an additive effect on CE proliferation. At Day 31, only the FINS group retained a significantly greater number of Pax6+EdU+ cells relative to PBS control (Figure 2.2C,G; Figure A5).

In order to determine to what extent off-target (non-CE) cell types in the CE are stimulated to proliferate, we quantified the proportion of total EdU+ cells in the CE that were co-labeled with Pax6 (CE cells), ERG (a nuclear marker for endothelial cells), or CD68 (a marker of microglia/macrophages). All conditions significantly increased total EdU labeling in the CE compared to PBS control, but FI and FINS had roughly twice the effect of the other treatment groups (Figure 2.2H). Despite a similar total number of EdU+ cells in the CE after FI and FINS

52 treatment, only ~8% of EdU+ cells co-labeled with Pax6 in FI treated eyes, similar to PBS control (~7%) (Figure 2.2I). In contrast, FINS resulted in the highest proportion of EdU+Pax6+ cells at ~25%. However, FINS induced proportionally more ERG+ endothelial cell proliferation (48%) than Noggin or anti-sFRP2 (22% and 35%, respectively) (Figure 2.2J). Yet, all three of those conditions had a lower proportion of EdU+ERG+ cells than PBS control (80%). In addition, a significant proportion of EdU+ cells co-labeled for the microglia/macrophage marker CD68 in all conditions (Figure 2.2K). However, all conditions were proportionally the same, indicating that none of the injected factors increased microglia activation or macrophage infiltration. Thus, antagonism of sFRP2 or BMP proteins induces proliferation of CE cells and enhances the potency and specificity of growth factor stimulation of CE cells.

Next, we performed combinatorial intravitreal injections of Noggin, anti-sFRP2 and growth factors, followed by clonal sphere-forming assays to assess the effects on the number of RSCs in vivo (Figure 2.2A). At Day 4, no treatment condition had any effect on RSC sphere number (Figure 2.2L). At Day 10, only anti-sFRP2 or Noggin alone increased the number of RSC spheres (Figure 2.2M). This was the same timepoint and doses that increased RSC sphere number previously. Surprisingly, the combined NS condition did not have a significant effect at the Day 10 timepoint, nor did FI, or all factors combined (FINS). At Day 31, once again, only individual anti-sFRP2 or Noggin treatments resulted in increased RSC sphere number (Figure 2.2N). Therefore, although sFRP2 and BMP antagonism combined with growth factor stimulation resulted in the largest stimulation of CE cell proliferation, it did not result in expansion of the RSC pool. Likewise, Noggin and anti-sFRP2 combined did not cause expansion of RSCs. Thus, if CE proliferation is indicative of RSC division, RSCs appear to divide asymmetrically under most conditions, while symmetric expansion of the RSC population occurs only with discrete antagonism of BMPs or sFRP2 at specific concentrations. In addition, these three time-points indicate that symmetric expansion of endogenous RSCs in response to BMP or sFRP2 antagonism does not occur immediately following the injection period but does ensue within seven days. Furthermore, the expanded endogenous RSC pool persists up to one month later but does not continue to expand beyond the extent observed at Day 10.

Figure 2.2. Combinatorial injection of Noggin and anti-sFRP2 with and without growth factors have differential effects on CE proliferation and retinal stem cell expansion

(A) Schematic of the intravitreal injection paradigm followed by endpoint clonal retinal stem cell (RSC) sphere forming assay from primary ciliary epithelium (CE) and/or immunohistochemistry (IHC). Mice received one intravitreal injection per day for three days followed by endpoint analysis at Day 4, 10 or 31. Injections consisted of: C=PBS control, N=Noggin, S=anti-sFRP2, NS=Noggin+anti-sFRP2 combined, FI=FGF2+Insulin combined, FINS=FGF2+Insulin+Noggin+anti-sFRP2 combined.

(B-C) Quantification of Pax6+EdU+ co-labeled cells relative to total CE area in eyes treated with + + PBS vehicle or indicated factors. (B) Day 4 Pax6 EdU cells (one-way ANOVA F(5,28)=8.76, + + p<0.001; N=5-6 eyes per group). (C) Day 31 Pax6 EdU cells (one-way ANOVA F(5,26)=2.81, p=0.037; N=3-6 eyes per group).

(D-G) Pax6 immunostaining and EdU labeling in the ciliary epithelium and peripheral retina of eyes injected with PBS vehicle or indicated factors at Day 4 and Day 31. Hoechst stain was used to label all nuclei. White arrows indicate Pax6+EdU+ double-positive cells. Straight dashed line indicates the border between the ciliary epithelium (CE) and the neural retina (NR). Dashed line box indicates high magnification inset.

(H) Quantification of EdU cell number in the CE normalized by area in eyes treated with the indicated conditions (one-way ANOVA F(5,28)=10.63, p<0.001; N=5-6 eyes per group).

(I-K) Percent of total EdU-positive cells in the CE that co-labeled for cell-type-specific markers in eyes treated with the indicated conditions. (I) % Pax6+EdU+ CE cells (one-way ANOVA + + F(5,28)=6.11, p<0.001; N=5-6 eyes per group). (J) % ERG EdU endothelial cells. (one-way + + ANOVA F(5,29)=6.94, p<0.001; N=5-6 eyes per group). (K) % CD68 EdU microglia/macrophage cells. N=5-6 eyes per group.

(L-N) Quantification of RSC sphere frequency normalized to naïve un-injected control following intravitreal injection of the indicated factors. (L) Day 4. N= 5-12 eyes per group. (M) Day 10 (one-way ANOVA F(6,53)=3.92, p=0.003; N=5-12 eyes per group). (N) Day 31 (one-way ANOVA F(6,31)=4.78, p=0.001; N=4-6 eyes per group). Each data point represents a single eye.

Statistics: One-way ANOVAs with Holm-Sidak posthoc tests. Data are means ± SEMs. * = p<0.05.

2.4.3 Inducible Msx1-CreERT2 mouse lineage labeling marks the adult ciliary epithelium and retinal stem cells

Given that EdU+ cells are detectable in the retina soon after intravitreal injection (see: IHC in previous figures and quantification in Figure 2.1), along with the above-mentioned potential for EdU label dilution, assessing CE cell migration or neurogenesis by profiling EdU+ cells in the retina is not conclusive. Thus, we sought to determine if an in vivo genetic lineage tracing model could be used to label CE cells in adult mouse eyes. We used the previously generated tamoxifen-inducible Msx1-CreERT2 knock-in mouse line (Lallemand et al., 2013) and crossed it with the B6.Cg-Gt(Rosa26)SorTm14(CAG-TdTomato)Hze TdTomato reporter mouse line (Madisen et al., 2010) (Figure 2.3A).

To determine the specificity of reporter expression, littermate adult mice that were Msx1-CreERT2 positive and negative crossed with Rosa-TdTomato were injected i.p. with tamoxifen (180mg/kg) for 4 consecutive days (Figure 2.3A). Cre+ mice had Msx1-TdTomato reporter expression in the entire (proximal and distal) CE (Figure 2.3B), although there were varying degrees of penetrance. In contrast, Cre- mice did not show any Msx1-TdTomato reporter labeling in the CE (Figure 2.3C). Also, Cre+ mice that did not receive tamoxifen did not show Msx1-TdTomato reporter labeling in the CE (Figure A6A,B). Some other eye tissues do express Msx1 and thereby are labeled with TdTomato, however, they are not in the CE nor retina and we concluded that they would not confound CE lineage tracing (Figure A6C-F). Additionally, we found that RSC spheres derived from Cre+ mice expressed the Msx1-TdTomato reporter (Figure 2.3D). As single, Msx1-TdTomato+ RSCs would be required for TdTomato+ clonal RSC spheres to arise, this is the first evidence we are aware of that adult RSCs express Msx1. Further, roughly 84% of all RSC spheres had Msx1-TdTomato+ cells indicating the majority of RSCs are labeled (Figure 2.3F). The clonal spheres from Cre+ mice without Msx1-TdTomato labeling may have arisen due to variable efficiency of tamoxifen induction. None of the RSC spheres derived from the eyes of Cre- mice were labeled with Msx1-TdTomato (Figure 2.3E). In order to further characterize reporter expression within the CE, we performed FACS on primary CE cells from mice that had received tamoxifen induction 1-2 weeks prior (Figure 2.3G). The CE was sorted into 4 populations of cells: pigmented TdTomato-negative cells (P-TdT-), pigmented TdTomato-positive cells (P-TdT+), non-pigmented TdTomato-negative cells (NP-TdT- ), and non-pigmented TdTomato-positive cells (P-TdT+) (Figure 2.3H-J). After sorting, each population was plated at 1 cell/µL for a 7-day clonal sphere-forming assay. As expected,

56 pigmented and non-pigmented CE were nearly equivalent proportions of the CE (46.6% pigmented, 53.4% non-pigmented) (Figure 2.3K). However, on average 71% of pigmented CE cells were Msx1-TdTomato+ (Figure 2.3L), whereas only 20% of non-pigmented CE cells were Msx1-TdTomato+ (Figure 2.3M). The post-FACS sphere assay revealed that only P-TdT+ CE cells could generate RSC spheres (Figure 2.3N-S). This is concordant with previous reports that sphere-forming RSCs are pigmented cells in the outer CE (Ballios et al., 2012; Tropepe et al., 2000). However, this also suggests that RSCs can be further characterized as Msx1- expressing pigmented CE cells. Although there was a trend toward increased sphere frequency in the P-TdT+ population relative to the unsorted control, sorting did not lead to a significant enrichment of RSCs (Figure 2.3N). However, only ~25% live cells were recovered during FACS, therefore RSC sphere frequency after sorting likely underrepresents the actual RSC number (Fig A7A-E).

In some RSC spheres, not all cells were Msx1-TdTomato+. This was true for spheres formed using the regular sphere assay paradigm with non-sorted cells (Figure 2.3D, bottom sphere), as well as spheres formed post-FACS assay (Figure 2.3P, white arrowhead). As the sphere- forming assay is clonal at 10 cells/µL (Coles-Takabe et al., 2008; Tropepe et al., 2000), and the post-FACS assay was carried out at 1 cell/µL, we hypothesized that non-labelled cells in spheres would more likely indicate transgene silencing (Martin and Whitelaw, 1996) rather than non-clonal mixing of TdTomato positive and negative cells. To investigate if transgene silencing was occurring, we extracted DNA from single spheres for PCR analysis using primers spanning the floxed stop sequence of the Rosa-TdTomato reporter construct. In clonal spheres, a single band at 2Kb (stop sequence present) or 1Kb (stop sequence excised) is expected. If, however, there was non-clonal mixing of cells with both genotypes, then both bands would be expected to appear in the PCR gel. Out of 21 spheres analyzed, only one had bands at both the 1Kb and 2Kb molecular weights. Thus, in 95% of spheres, all cells had a single genotype (Figure A7F). Therefore, it is likely that spheres with incomplete reporter labeling had heterogeneous transgene silencing in a subset of cells in the clonal RSC spheres. These results indicate that the Msx1-CreERT2 mouse line can be used to lineage label the adult mouse CE and reveals that adult RSCs and their progeny are labeled by this paradigm. However, we predict it will underrepresent the extent of CE cell migration and differentiation due to incomplete tamoxifen penetrance and transgene silencing.

Figure 2.3. Tamoxifen induction of reporter expression in Msx1-CreERT2;Rosa26- TdTomato mice labels the ciliary epithelium and CE-derived RSC spheres

(A) Schematic of Msx1-CreERT2;Rosa26-TdTomato mouse lines and transgene induction paradigm. Four consecutive daily intraperitoneal tamoxifen injections were followed by endpoint clonal RSC sphere assay and immunohistochemistry (IHC) two weeks after the injection period.

(B-E) Fluorescence microscopy of eye sections and RSC spheres from the eyes of Msx1- CreERT2;Rosa26-TdTomato mice two weeks post-tamoxifen injection. (B-D) Msx1-Cre+ mice. (C- E) Littermate control Msx1-Cre-mice. CE = ciliary epithelium, NR = neural retina.

(F) The Percentage of RSC spheres with Msx1-TdTomato expression derived from tamoxifen- treated Msx1-CreERT2;Rosa26-TdTomato mice. Each data point represents the average % labeled RSC spheres per well for a single mouse. Red dashed line indicates between mouse average. N=4 mice.

(G) Schematic of Msx1-CreERT2;Rosa26-TdTomato transgene induction followed by fluorescence activated cell sorting (FACS) and subsequent clonal RSC sphere assay.

(H-J) Representative FACS gating plots for CE cells derived from Msx1-CreERT2;Rosa26- TdTomato mice following TdTomato reporter induction. NP=non-pigmented; P=pigmented; TdT+=TdTomato positive; TdT-=TdTomato negative.

(K-M) The proportions of each FACS subpopulation. (K) The proportion of pigmented and non- pigmented cells. (L) The proportion of Msx1-TdTomato+ and Msx1-TdTomato- cells within the pigmented CE population. (two-tailed t-test t(6) = 14.28, p<0.001; N=4 FACS experiments). (M) The proportion of Msx1-TdTomato+ and Msx1-TdTomato- cells within the non-pigmented CE population. (two-tailed t-test t(6) = -4.25, p=0.005; N=4 FACS experiments).

(N) The number of RSC spheres that formed 7 days after sorting for each FACS subpopulation. N=3 FACS experiments.

(O-S) Representative images of RSC sphere assay output wells for each sorted subpopulation 7 days following FACS. (O) Unsorted CE cells. Two images stitched together (indicated by the dashed line). (P) Pigmented Msx1-TdTomato+ CE cells. Two images stitched together (indicated by the dashed line). The white arrowhead indicates a sphere that has partial expression of TdTomato. (Q) Pigmented Msx1-TdTomato- CE cells. (R) Non-pigmented Msx1-TdTomato+ CE cells. (S) Non-pigmented Msx1-TdTomato- CE cells. White arrows indicate single Msx1- TdTomato expressing cells. Dashed line box indicates high magnification inset.

Statistics: t-tests. Data are means ± SEMs. * = p<0.05.

2.4.4 FINS-mediated CE proliferation is potentiated by photoreceptor degeneration in Msx1-CreERT2;Rosa26-tdTomato mice

Thus far, our analyses of CE proliferation had been in mice with healthy retinas (with the caveat that intravitreal injection results in minor retinal damage). Therefore, in addition to examining the effects of FINS treatment, we used the Msx1-CreERT2 mouse line to assess if retinal injury would influence CE proliferation, migration or differentiation. To induce photoreceptor degeneration, we used the N-methyl-N-nitrosourea (MNU) injury model. MNU is a DNA alkylating agent that selectively induces 7-methyldeoxyguanosine DNA adducts in photoreceptor nuclei in the eye and results in photoreceptor-specific degeneration via multiple cell death pathways (Reisenhofer et al., 2015; Tsubura et al., 2011). A single, 45mg/kg i.p. injection of MNU results in a ~90% depletion of the photoreceptor outer nuclear layer (ONL) in two weeks (Figure A8). All Msx1- CreERT2;Rosa26-TdTomato mice first underwent the tamoxifen injection paradigm prior to MNU injection and/or intravitreal injection (Figure 2.4A). A separate naïve un-injected group was used to control for the effects of intravitreal injection. Therefore, 5 groups were included in this experiment: Naïve control, PBS, PBS+MNU, FINS and FINS+MNU. All mice received EdU in their drinking water during the injection period. Some PBS+MNU eyes had pathology indicative of phthisis bulbi and were excluded from further analyses (Figure A9). Otherwise, no difference in ONL thickness between PBS+MNU mice and FINS+MNU mice was observed (Figure 2.4B). Both PBS+MNU and FINS+MNU eyes had significantly reduced ONL thickness compared to the un-injured groups, indicating FINS treatment did not ameliorate photoreceptor degeneration at this gross level (Figure 2.4B-G). Due to the incomplete penetrance of Msx1-TdTomato expression in the CE, we again used Pax6 to quantify EdU labeling in the CE (Figure 2.4H). PBS+MNU did not significantly increase proliferation compared to PBS alone. Yet, there was a pronounced increase in the effect of FINS injection in MNU injured eyes. Indeed, the FINS+MNU group had the highest level of Pax6+EdU+ labeling in the CE of all conditions tested in this study, suggesting the CE had a greater proliferative response to FINS when injury was present. Although not used for quantification, co-labeled EdU+/Msx1-TdTomato+ CE cells were readily apparent, and the extent of visible EdU labeling correlated well with the Pax6+EdU+ cell quantification across conditions (Figure 2.4I-M; Figure A10). Thus, Msx1-TdTomato served as a second CE-specific marker to validate that CE cells proliferate in response to FINS. These results confirm that MNU causes reliable photoreceptor degeneration and revealed that MNU injury potentiates FINS stimulation of CE proliferation.

Figure 2.4. FINS-mediated CE proliferation is augmented by photoreceptor degeneration in Msx1-CreERT2;Rosa26-TdTomato mice

(A) Schematic of the lineage tracing paradigm in tamoxifen-treated Msx1-CreERT2;Rosa26- TdTomato mice used to investigate the effects of MNU injury and/or FINS (FGF2+Insulin+Noggin+anti-sFRP2) treatment on ciliary epithelium proliferation, migration and differentiation. The injection period was followed by tissue fixation and IHC analysis at Day 31. Groups included: Naïve control = no injection; PBS = intravitreal PBS; PBS+MNU = intravitreal PBS and i.p. MNU; FINS = intravitreal FINS; FINS+MNU = intravitreal FINS and i.p. MNU

(B) Quantification of outer nuclear layer (ONL) thickness in eyes from mice of the indicated conditions (two-way ANOVA F(1,20)=184.53, p<0.001; N=6 eyes per group). Each data point represents a single eye.

(C-G) Representative images of Hoechst nuclear-stained eye cross-sections from mice of the indicated conditions. RGCL = retinal ganglion cell layer; yellow indicators. INL = inner nuclear layer; blue indicators. ONL = outer nuclear layer; red indicators.

(H) Quantification of Pax6+EdU+ co-labeled cells relative to total CE area in eyes from mice of the indicated conditions. There was a significant effect of treatment (two-way ANOVA F(1,20)=34.58, p<0.001) and a significant effect of MNU injury (two-way ANOVA F(1,20)=5.12, p=0.035). N=6 eyes per group. Each data point represents a single eye.

(I-M) Brightfield and fluorescence overlay images of Msx1-TdTomato expression and EdU labeling in eyes from mice of the indicated conditions. White arrows indicate Msx1- TdTomato+EdU+ co-labeled cells. Straight dashed line indicates the border between the ciliary epithelium (CE) and the neural retina (NR). Dashed line box indicates high magnification inset.

Statistics: Two-way ANOVAs with Holm-Sidak posthoc tests except where stated otherwise. Means ± SEMs indicated. * = p<0.05.

2.4.5 Photoreceptor degeneration and FINS treatment induce CE cell migration into the neural retina

In order to qualify what constitutes CE migration into the peripheral neural retina, we analyzed the domain of Msx1-mediated TdTomato expression in naïve mouse eyes 1 day and 45 days after the tamoxifen injection paradigm (Figure A11A-E). The farthest into the retina an Msx1- TdTomato+ cell was detected at Day 1 was 10µm. Therefore, an Msx1-TdTomato+ cell needed to be beyond 10µm into the retina to be included in the experimental migration analyses below (Figure A11B). Of note, only 7 cells total were observed beyond 10µm in Day 45 naïve eyes, after analyzing a combined total of 120 sections from 6 eyes. However, that represented a significant increase in migration distance at Day 45 compared to Day 1 (Figure A11C). This

62 suggests that, even in naïve eyes, there may be cell migration into the retina at extremely low frequency.

PBS injection alone did not increase the frequency of cell migration, cell migration distance, or the number of cells that migrated per section compared to naïve eyes (Figure 2.5A-E). FINS injection alone significantly increased migration frequency, with an average of ~39% of eye sections showing CE cell migration per eye (Figure 2.5A). However, unlike proliferation, there was not an increased migration effect in the FINS+MNU injury group compared to FINS alone. Although the mean migration frequency was higher in the PBS+MNU group versus PBS alone, this change was not statistically significant. In contrast, cell migration distance was markedly increased by both MNU injury and FINS treatment, as average migration distance in PBS eyes was ~20µm, while PBS+MNU, FINS and FINS+MNU all averaged around 500µm (Figure 2.5B- C). Likewise, PBS+MNU, FINS and FINS+MNU all had similar cell number averages of ~1.5-1.8 cells per section (Figure 2.5D-E). Both, the increased migration distance and cell number were evident in retinal sections (Figure 2.5F-I). Also, some Msx1-TdTomato+ cells in the retina had neuroepithelium-like morphology while others appeared to extend processes (Figue 2.5J-M). These results reveal that photoreceptor degeneration and FINS stimulation can each induce CE migration into the retina, but that FINS stimulation has a greater effect.

Figure 2.5. Photoreceptor degeneration or FINS treatment induces ciliary epithelial cell migration into the neural retina

(A-I) Analyses of eye slides from the endpoint of the experimental paradigm outlined in Figure 4A.

(A) Percent of eye sections with Msx1-TdTomato+ cells present in retina for mice of the indicated conditions. There was a significant interaction between injury and treatment on migration frequency (two-way ANOVA F(1,23)=6.13, p=0.021; N=6-8 eyes per group. Each data point represents a single eye.

(B-C) Quantification of the migration distance of Msx1-TdTomato+ cells into the retina for mice of the indicated conditions. (B) The migration distance recorded for individual cells. Each data point represents a single cell, N=9-228 cells from 6-8 eyes per group. (C) Average migration distance per eye. There was a significant effect of treatment (two-way ANOVA F(1,23)=5.44, p=0.029) and injury (two-way ANOVA F(1,25)=5.52, p=0.028). N=6-8 eyes per group. Each data point represents a single eye.

(D-E) Quantification of the number of Msx1-TdTomato+ cells in the retina for mice of the indicated conditions. (D) The total number of cells recorded in individual retina sections. Each data point represents a single retina section with at least one Msx1-TdTomato+ cell detected. (E) Average number of Msx1-TdTomato+ cells in retina sections per eye. There was a significant interaction between injury and treatment on the number of Msx1-TdTomato+ cells in the retina (two-way ANOVA F(1,23)=8.63, p=0.007; N=6-8 eyes per group). Each data point represents a single eye.

(F-I) Brightfield and fluorescence overlay images of Msx1-TdTomato+ cells in retinas from mice treated with the indicated conditions. White arrows indicate Msx1-TdTomato+ cells in the retina.

(J-M) Msx1-TdTomato+ CE cells in the retina with neuroepithelial or neurite morphology at Day 31. DRAQ5 stain was used to label all nuclei.

Straight dashed line indicates the border between the ciliary epithelium (CE) and the neural retina (NR). Dashed line box indicates high magnification inset.

Statistics: Two-way ANOVAs with Holm-Sidak posthoc tests. Means ± SEMs indicated. * = p<0.05.

2.4.6 Ciliary epithelium derived Msx1-TdTomato+ cells in the neural retina express photoreceptor or retinal ganglion cell markers

We next sought to determine if the Msx1-TdTomato+ CE cells that migrated into the retina showed evidence of differentiation into retinal neurons by co-expressing retinal cell type markers. In the MNU injury model, photoreceptors are selectively degenerated. Furthermore, using Msx1-CreERT2 lineage tracing, Belanger et al. (2017) reported that photoreceptors were the

65 most abundant cell type produced by the embryonic CE during retinogenesis. Therefore, we examined whether Msx1-TdTomato+ cells in the retina co-expressed the mature photoreceptor marker Recoverin. We were not able to locate enough Msx1-TdTomato+ cells in the Naïve control sections to quantify co-expression for Recoverin. All other conditions exhibited Msx1- TdTomato+Recoverin+ co-labeled cells in the retina (Figure 2.6A-F; Figure A12). Both MNU injury and FINS treatment resulted in a greater proportion (~50-55%) of Msx1-TdTomato+ cells in the retina that co-labeled for Recoverin relative to the PBS control (~7%) (Figure 2.6A). In fluorescence images, Msx1-TdTomato+Recoverin+ cells were usually located apposed to the ONL between the ONL and RPE (Figure 2.6C,E; Figure A12C,D) or embedded within the ONL (Figure 2.6F; Figure A12E). Of note, the Msx1-TdTomato+Recoverin+ cells in the retina did not appear to have mature rod or cone photoreceptor morphology with inner and outer segments.

A study by Marcucci et al. (2016) reported that a subpopulation of proliferative, Cyclin D2+ embryonic CE cells migrate into the retina and generate retinal ganglion cells (RGCs) during retinogenesis. Therefore, we examined whether Msx1-TdTomato+ cells in the retina co- expressed the RGC marker Brn3a. Once again, we were not able to locate enough Msx1- TdTomato+ cells in the Naïve control sections to quantify co-expression for Brn3a. In PBS injected eyes, ~20% of Msx1-TdTomato+ cells detectable in the retina co-labeled for Brn3a (Figure 2.6G). In both PBS+MNU and FINS eyes there was a greater proportion of Msx1- TdTomato+Brn3a+ co-labeled cells in the retina relative to PBS (60% and 65%, respectively). However, there was a smaller proportion of Msx1-TdTomato+ cells that co-labeled for Brn3a+ in the retinas of FINS+MNU mice than in FINS mice. In fluorescence images (Figure 2.6H-L; Figure A13), Msx1-TdTomato+Brn3a+ cells were usually located adjacent to the CE in the peripheral retina (Figure 2.6H,K; Figure A13B,D) or embedded within the ONL (Figure 2.6I,L; Figure A13C,E). Of note, Msx1-TdTomato+Brn3a+ co-labeled cells did not extend neurites and were not observed in the RGC layer. However, in general, Msx1-TdTomato+ cells were rarely detected in the RGC layer in our migration analyses. These results provide evidence that FINS treatment and photoreceptor degeneration can induce adult CE cells to migrate into the retina and express retinal neuron markers. Also notable, though extremely infrequent, some Msx1- TdTomato+ cells were detected in the RPE layer (Figure A14).

Figure 2.6. Ciliary epithelium derived TdTomato+ cells in the neural retina express photoreceptor or retinal ganglion cell markers

(A-L) Analyses of retinal cell type markers in eye slides from the endpoint of the experimental paradigm outlined in Figure 4A.

(A) Quantification of the percentage of Msx1-TdTomato+ cells in the retina that co-labeled for the photoreceptor marker Recoverin in eye sections from the indicated conditions. There was a significant effect of treatment (two-way ANOVA F(1,124)=4.96, p=0.028) and injury (two-way + + ANOVA F(1,124)=4.45, p=0.037) on the proportion of Msx1-TdTomato Recoverin cells. N=10-76 cells from 3-6 eyes per group.

(B-F) Brightfield and fluorescence overlay images of Msx1-Cre driven TdTomato expression and immunostaining for photoreceptor marker Recoverin in eye sections from the indicated conditions. DRAQ5 stain was used to label all nuclei. White arrows indicate Msx1- TdTomato+Recoverin+ co-labeled cells.

(G) Quantification of the percentage of Msx1-TdTomato+ cells in the retina that co-labeled for the retinal ganglion cell marker Brn3a in eye sections from the indicated conditions. There was a significant interaction between treatment and injury on the proportion of Msx1-TdTomato+Brn3a+ cells (two-way ANOVA F(1,130)=9.04, p=0.003; N=10-78 cells from 3-7 eyes per group).

(H-L) Brightfield and fluorescence overlay images of Msx1-Cre driven TdTomato expression and immunostaining for retinal ganglion cell marker Brn3a in eye sections from the indicated conditions. DRAQ5 stain was used to label all nuclei. White arrows indicate Msx1- TdTomato+Brn3a+ co-labeled cells.

Straight dashed line indicates the border between the ciliary epithelium (CE) and the neural retina (NR). Dashed line box indicates high magnification inset.

Statistics: Two-way ANOVAs with Holm-Sidak posthoc tests. Data are means ± SEMs.* = p<0.05.

2.5 Discussion

In this study we found that intravitreal delivery of BMP antagonist, Noggin, or a function blocking antibody against sFRP2, can each induce proliferation of the adult CE in vivo. This is consistent with the evidence that endogenous BMP and sFRP2 proteins regulate the proliferation of RSCs in vitro (Balenci et al., 2013). We also found that inhibiting BMP and sFRP2 proteins made the RSC niche more responsive to stimulation with exogenous growth factors. Previous attempts to reactivate RSCs in the adult mammalian eye with exogenous factors have resulted in reports of some limited proliferation in the CE (Abdouh and Bernier, 2006; Del Debbio et al., 2014; Wang et al., 2010). However, those studies did not use CE cell type markers to quantify proliferation,

68 and instead, included all proliferating cells in the region in their analyses. Also, some of those studies evaluated nestin expression as a putative retinal stem cell marker, however nestin also marks endothelial cells (Kim et al., 2016a) and microglia (Wohl et al., 2011) in the retina and CE. Thus, it is unknown to what extent proliferation in non-CE cells, such as endothelial cells and microglia, may have been included in previous reports. By quantifying EdU+ cells that co- labeled for Pax6 we increased the specificity and sensitivity to changes in proliferation in CE cells, which includes the RSC population. In addition, we observed a significant number of EdU+CD68+ microglia/macrophages in the CE and showed that different combinations of Noggin, anti-sFRP2, and growth factors had differential off-target effects on the proliferation of endothelial cells, reinforcing the importance of evaluating cell-type-specific markers in the RSC niche.

No change in the number of RSC spheres derived from the CE was observed after FINS treatment, indicative of an asymmetric mode of RSC division (Balenci and van der Kooy, 2014; Morrison and Kimble, 2006; Post and Clevers, 2019). In contrast, individual anti-sFRP2 or Noggin each increased in vivo RSC number, signifying symmetric expansion of RSCs. This is concordant with previous studies that have characterized various factors with different influences on RSC mode of division. For example, in vitro, RSC symmetric expansion can be promoted by exogenous Wnt ligands (Inoue et al., 2006), Notch ligands (Balenci and van der Kooy, 2014) and PEDF (De Marzo et al., 2010), while asymmetric self-renewal of RSCs is dependent on FGF and Notch signaling (Balenci and van der Kooy, 2014). In vivo, an expanded RSC population has been demonstrated in mutant mice with reduced populations of NR progenitors (Chx10orj/orj) or RPE progenitors (Mitfmi/mi), suggesting non-cell autonomous signals can regulate the size of the RSC pool (Coles et al., 2006). Another study observed only asymmetric cell divisions in the CE of rats in response to intravitreal injections of FGF and Insulin (Abdouh and Bernier, 2006), which also is the preferential mode of division of RSCs in the Zebrafish CMZ (Centanin et al., 2014). Furthermore, both Wnt signaling and BMP signaling are known to crosstalk with one another, as well as with other stem cell regulating pathways, including Notch and Hedgehog (Baker et al., 1999; Borday et al., 2012; Katoh, 2007; Zhou et al., 2013). Thus, it is likely the different effects on RSC number of the various doses and combinations of exogenous factors used in this study is determined by which signaling pathways predominate under each condition and their particular influences on RSC mode of division. Alternatively, it is possible that a subset of RSCs remain quiescent and do not generate spheres in vitro unless first stimulated in vivo. Indeed, other endogenous stem cells, such as

69 neural stem cells in the brain, contain subpopulations in quiescent and activated states which have different colony forming ability in vitro (Codega et al., 2014; Reeve et al., 2016, 2017; Wang et al., 2011). Recent experiments in our lab have generated evidence that endogenous RSCs may also include a subpopulation that are primed to proliferate with a greater propensity for sphere formation in vitro (submitted). However, resolving the active signaling pathways in RSCs or distinct RSC subpopulations in vivo remains challenging given that RSCs are rare cells with no known unique molecular markers.

Since a specific marker for RSCs is not yet known, we investigated the possibility of lineage labeling the entire CE, which includes the RSC population. We observed that Msx1-mediated reporter expression labels the entire adult mouse CE. This corresponds with previous studies showing Msx1 expression throughout the CE during embryonic and early postnatal development (Monaghan et al., 1991; Zhao et al., 2002b). In contrast, Belanger et al (2017) reported that late embryonic Msx1 is restricted only to the proximal CE that is continuous with the peripheral neural retina. However, they successfully used the Msx1-CreERT2 mouse line for lineage tracing in the embryonic mouse eye and demonstrated retinal neurogenesis from the CE during development (Bélanger et al., 2017). Contrary to our results, the authors stated that their Msx1 reporter did not label the pigmented CE, and thus, likely did not label CE-RSCs. Here, we report the first evidence Msx1 is expressed in adult mouse RSCs via the lineage labeling of clonal RSC-derived spheres. Furthermore, FACS revealed that only Msx1-labeled pigmented CE cells give rise to clonal RSC spheres. Thus, these results suggest Msx1 is not only expressed in the pigmented CE and RSCs but may even regulate RSC function. Indeed, Msx1 is known to regulate proliferation, differentiation and regenerative processes in other adult stem cells and tissues (Beck et al., 2003; Bhatt et al., 2013; Ding et al., 2017; Taghiyar et al., 2017). Also notable, BMPs and Wnts have been proposed to cooperate to pattern the CE by promoting Msx1 expression (Liu et al., 2007; Zhao et al., 2002b). Thus, better resolution of the relationship between BMP, Wnt and Msx1 in RSCs and the adult CE could provide further insight into the regulation of RSC quiescence in the adult mammalian eye.

Some studies have reported proliferation, and even expression of retinal neuron markers, in the adult mammalian CE following injury (Ducournau et al., 2012; Johnsen et al., 2012; Nickerson et al., 2007; Nishiguchi et al., 2008, 2009). We found only minor evidence of CE proliferation in response to MNU injury and did not find any evidence of retinal neuron markers in the CE. However, MNU injury did induce some CE cell migration into the retina where, remarkably, the majority of cells expressed photoreceptor or RGC markers. Furthermore, FINS with MNU injury

70 resulted in similar migration and fewer RGCs than FINS injection alone. In the xenopus and zebrafish CMZ, the influence of injury on retinal stem and progenitor cell proliferation, migration and differentiation arises from damage-liberated trophic factors and cytokines (Agathocleous and Harris, 2009; Ail and Perron, 2017; Ting et al., 2008). Thus, damage-liberated factors could modulate proliferation and differentiation in the reactivated adult mammalian CE as well. Another consideration is that injury-induced reactive gliosis in the retina, which is caused by MNU-mediated retinal degeneration (Reisenhofer et al., 2016), has been shown to decrease migration, maturation and survival of transplanted photoreceptors (Gasparini et al., 2019). Thus, a less severe injury model than MNU may also change migration and differentiation outcomes.

It remains to be investigated whether the level of retinal neurogenesis observed in this study might have a functional impact toward restoring vision. Relatively few cells per eye may be needed to improve visual function, as a previous study found the survival of ~350 RSC-derived rod photoreceptors in the retina after transplantation was sufficient to increase the pupillary light response in blind mice (Ballios et al., 2015), while a survival rate of as few as 10-200 cells in the retina has been shown to improve visual function after retinal precursor transplantation (MacLaren et al., 2006). Yet, it is likely the strategy employed in this study can be improved to generate an even greater level of neurogenesis, with an even higher likelihood of restoring visual function. This study employed an acute, three-day intravitreal injection paradigm and there was evidence the effects of Noggin and anti-sFRP2 largely subsided 24 hours after injection (Figure 2.1H-K). Furthermore, we used a variety of factors that likely have different pharmacokinetics in the vitreous (del Amo et al., 2017). Therefore, follow up studies using biomaterials, such as polymeric microparticles or hydrogels, to achieve controlled and sustained release of FINS components into the vitreous may elicit more robust and consistent proliferation, migration and neurogenesis in the RSC niche (Delplace et al., 2019; Hettiaratchi and Shoichet, 2019).

Chapter 3

Glucocorticoid agonists enhance retinal stem cell self-renewal and proliferation in vitro and stimulate proliferation of the ciliary epithelium in the mouse eye in vivo

*This chapter was submitted to Stem Cells

Kenneth N. Grisé, Nelson X. Bautista, Krystal Jacques, Brenda L.K. Coles, Derek van der Kooy. Glucocorticoid agonists enhance retinal stem cell self-renewal and proliferation in vitro and stimulate proliferation of the ciliary epithelium in the mouse eye in vivo. Submitted June 2020.

K.N.G designed and executed all experiments except the pancreatic progenitor experiments in Fig. 3.3F-L, which were designed, executed and analyzed by K.D.J. K.N.G quantified and analyzed all other experiments except for the IHC data in Fig. 3.5, which was quantified by N.X.B. N.X.B. also helped process tissue and perform IHC for Fig. 3.5/Fig. B4-5 and helped execute the differentiation experiments in Fig. 3.4. B.L.K.C. helped with primary dissections for screening in Fig. 3.1 and sphere assays in Fig. 3.5G. K.N.G. wrote the manuscript. D.v.d.K. contributed to experimental design, analyses, and edited the manuscript.

3.1 Abstract

Adult mammalian retinal stem cells (RSCs) readily proliferate, self-renew and generate progeny that differentiate into all retinal cell types in vitro. RSC progeny can be induced to differentiate into photoreceptors, making them a potential source for retinal cell transplant therapies. Despite their proliferative capacity in vitro, RSCs in the adult mammalian eye do not proliferate and do not have a regenerative response to injury. Here, we used medium-throughput screening to identify small molecules that can expand the number of RSCs and their progeny in culture. We discovered high-affinity synthetic glucocorticoid (GC) agonists increase RSC self-renewal and increase retinal progenitor proliferation up to 6-fold without influencing their differentiation in vitro. Intravitreal injection of synthetic GC agonist dexamethasone induced proliferation in the ciliary epithelium – the niche in which adult RSCs reside. Together, our results identify GCs as novel regulators of retinal stem and progenitor cell proliferation in culture and provide evidence that GCs may activate endogenous RSCs.

3.2 Introduction

Retinal degenerative diseases cause permanent vision loss in mammals because the retinal neurons that are lost, such as photoreceptors and retinal ganglion cells (RGCs), are not replaced and their axons do not regenerate after damage (Berry et al., 2008; Kubota et al., 2002). Stem cells and embryonic retinal progenitors have been used as exogenous cell sources for retinal transplantation therapies with varying degrees of efficacy depending on the cell source, the degree of pre-transplant cell type maturity, and the number of cells transplanted (Gasparini et al., 2019). However, to what degree the functional improvement from transplanted cells is due to the secretion of trophic factors or material transfer of phototransduction machinery is still being actively investigated (Gasparini et al., 2019). Advances in stem cell biology have heralded activation of resident adult stem cells as a promising strategy for tissue regeneration (Aladdad and Kador, 2019; Miller and Kaplan, 2012; Post and Clevers, 2019). In the mammalian eye, Müller glia (MG) (Bringmann et al., 2006; Fischer and Reh, 2003b), the retinal pigmented epithelium (RPE) (Al-Hussaini et al., 2008; Salero et al., 2012) and the ciliary epithelium (CE) (Ahmad et al., 2000; Tropepe et al., 2000) have been identified as tissues containing a subset of cells with stem/progenitor-like properties and retinal neurogenic potential. Thus, whether these cells can be sources for endogenous retinal regeneration is actively being investigated.

Retinal stem cells (RSCs) are a rare and quiescent subpopulation of cells in the pigmented layer of the CE of the mammalian eye that are capable of clonal expansion, self-renewal, and differentiation into all the cell types of the retina when isolated in vitro (Ahmad et al., 2000; Ballios et al., 2012; Coles et al., 2004; Inoue et al., 2010; Tropepe et al., 2000). RSCs in the mammalian CE have been compared to the proliferative ciliary marginal zone (CMZ) of non- mammalian vertebrates which harbour stem cells that have neurogenic and regenerative potential in the adult eye (Fischer et al., 2013). Recent in vivo lineage tracing studies have shown that CE progenitor cells migrate into the peripheral retina and generate all seven major retinal cell types during eye development (Bélanger et al., 2017; Marcucci et al., 2016), similar to CMZ progenitors (Aladdad and Kador, 2019; Fischer et al., 2013). However, unlike the CMZ, that process arrests postnatally and no further generation of retinal neurons by the CE is observed. Despite the expression of stem cell and retinal progenitor genes in CE-RSCs and their progeny, some studies have suggested that they are not true stem cells based on the observation of a limited in vitro proliferative/self-renewal ability, maintenance of features of

73 epithelial cells in RSC progeny, and described only ectopic expression of mature retinal cell markers after differentiation (Cicero et al., 2009; Gualdoni et al., 2010); suggesting that CE cells might have general proliferative competency and plasticity as opposed to containing rare stem cells. However, the ability to prospectively identify and sort RSCs indicates a pre-existing rare cell type within the CE with proliferative competency (Ballios et al., 2012), while in vitro growth and self-renewal of RSCs and their progeny can be profoundly enhanced based on cell culture conditions (Baakdhah and van der Kooy, 2019; Coles et al., 2004). Further, RSC-derived photoreceptors have been shown to be functional pre-transplant in vitro (Del Debbio et al., 2013; Demontis et al., 2012), as well as post-transplant in vivo (Ballios et al., 2015). Thus, CE RSCs continue to be investigated as an exogenous source for cell replacement therapy and a potential source of endogenous retinal regeneration (Aladdad and Kador, 2019; Stern and Temple, 2014) However, attempts to activate adult mammalian RSCs in vivo have not been effective; an outcome that is attributed to the presence of quiescence factors in the RSC niche (Balenci et al., 2013; Wohl et al., 2012).

In this study, we sought to further elucidate the signaling pathways that regulate RSC quiescence and proliferation, identify small molecules that could be applied to scale up retinal progenitor production in vitro, and further assess those compounds for the potential to stimulate RSC proliferation in vivo. We discovered that synthetic glucocorticoid (GC) agonists enhance the proliferation of retinal progenitors and increase the symmetric self-renewal of RSCs in culture. Furthermore, intravitreal injection of the synthetic GC agonist dexamethasone (Dex) into the adult mouse eye induced proliferation in the CE, the niche in which endoegenous RSCs reside, indicating glucocorticoid signaling may stimulate RSC proliferation in vivo.

3.3. Materials & Methods

3.3.1 Mice

All mouse protocols were approved by the Animal Care Committee at the University of Toronto, which operates in accordance with the Canadian Council on Animal Care. Adult mice used in this study were a minimum of 8-10 weeks old and included: CD1 mice (022, Charles River), C57/BL6J mice (000664, Jackson Laboratories), Actin-GFP mice [FVB.Cg-Tg(CAG- EGFP)B5Nagy/J; 003516, Jackson Laboratories], mouse insulin promoter (MIP)-GFP mice [CD1/Tg(Ins1-EGFP/GH1); 006864, Jackson Laboratories]. Mice were kept on a 12-hour light

74 dark/light cycle. Food was available ad libitum. Water was supplied ad libitum except during EdU delivery.

3.3.2 Isolation of Retinal Stem Cells from the Ciliary Epithelium of the Adult Eye and Primary Clonal Sphere Assay

A dissecting microscope, cold light source, and sterile surgical instruments were set up inside of a sterile biological safety cabinet (BSC). Mouse eyes were enucleated immediately prior to beginning the dissection protocol. Mouse or human eyes were placed in a petri dish containing cold, sterile regular aCSF. Under the dissecting microscope, hair, connective tissue and the dorsal and ventral oblique muscles were cleared from the scleral/corneal border with two sets of forceps. Next, curved or angled micro-dissecting scissors were used to cleave off any remaining extraocular muscle tissue and the optic nerve and cut the eyeball into symmetrical halves, beginning and finishing the cut from the hole left by the optic nerve. Using two sets of forceps to grasp the cornea, the two eye halves were peeled apart. The lens, retina, and vitreous were separated from the eye shells and the eye shells were transferred into a new petri dish (also containing cold, sterile regular aCSF). To isolate the ciliary epithelium (CE), eye shells were oriented with the cornea on the right and retinal pigmented epithelium (RPE) on the left. A pair of straight forceps were used to pin down the eye shell on the RPE side while a scalpel blade was inserted between the CE and the iris, using pressure to slice the iris/cornea side off from the rest of the shell. Next, the scalpel was run along the border between the CE and the RPE to obtain the CE isolated as a thin strip of tissue. The CE strips were then transferred to a 35mm dish containing 2mL of dispase solution (Sigma; T1005) and incubated for 10 minutes at 37°C. Next, the strips were transferred from dispase into a 35mm dish containing 2mL of sterile filtered kynurenic acid (02.mg/mL; Sigma), trypsin (1.33mg/mL; Sigma) and hyaluronidase (0.67mg/mL; Sigma) in high magnesium/low calcium artificial cerebral spinal fluid (hi/lo aCSF) and incubated at 37°C for 10 minutes. After incubation, the dish was returned to the dissecting scope, and the CE strips were pinned down with straight, non-serrated forceps, while non-serated curved forceps were used to scrape the CE off from the underlying sclera. The bare scleral strips were then discarded, such that only the CE cells remained in the enzyme solution. Using a fire- polished, cotton-plugged glass pipette, the cells and enzyme solution were transferred to a 15mL tube and triturated approximately 45 times to break apart the tissue. The 15mL tube with cell suspension was centrifuged for 5 minutes at 1500 rpm. The supernatant was gently aspirated from the resulting pellet using a fire-polished, cotton-plugged glass pipette and 2mL of

75 sterile-filtered ovomucoid trypsin inhibitor (1mg/mL; Sigma) in SFM was added to the pellet. Using a small borehole, fire-polished, cotton-plugged glass pipette, the sample was triturated approximately 45 times to generate a single-cell suspension. The 15mL tube with cell suspension was centrifuged for 5 minutes at 1500 rpm. The supernatant was gently aspirated from the resulting pellet and 1-2mL of SFM with fibroblast growth factor 2 (FGF2, 10ng/mL; Sigma) and heparin (2μg/mL; Sigma) was added. The cells and media were mixed to ensure a uniform cell suspension and a 10µL sample was taken for cell density determination. The cells were then seeded and cultured at 10 cells/μL in culture-treated plates or flasks and incubated in a humidified incubator at 37°C in 5% CO2 and ambient room O2. After one week, roughly 1 in 500 cells are expected to proliferate to form free-floating, clonal spheres greater than 80μm in diameter.

3.3.3 Human Eye Procurement and Dissection

Human eye protocols were approved by the University of Toronto Research Ethics Board. Human eyes were procured from the Eye Bank of Canada (Toronto, ON) within 24h post- mortem. Human eye dissection was performed using the same procedure as the mouse eye dissection with the following exceptions: Once the CE strip was cut away from the rest of the eye tissue, a pair of curved scissors and forceps were used to blunt dissect and peel the CE off from the underlying sclera. The Dispase enzyme digestion and the kynurenic acid/trypsin/hyaluronidase enzyme digestion were each 20 minutes at 37°C.

3.3.4 Mouse pancreatic multipotent progenitor isolation and sphere assay

A modified version of our previously described mouse pancreatic islets isolation protocol was performed (Seaberg et al., 2004). Briefly, mice were anesthetized using sodium pentobarbital prior to terminal dissections. 1mg/mL of collagenase type V (Sigma) dissolved in 1X HBSS (Gibco) was perfused through the bile duct. Perfused pancreas was incubated in a 37 °C water bath to digest the pancreas. Islets were immediately hand-picked out of the total pancreatic tissue. A pure population of islets were incubated with trypsin ETDA (Sigma) at 37 °C and triturated with a small-borehole siliconized pipette into a single cell suspension. Viable cells were counted using Trypan Blue (Sigma) exclusion and the PMP sphere formation assay was performed as previously described (Seaberg et al., 2004), with the addition of conditions containing dexamethasone at 0.1µM, 1 µM or 10 µM. PMP spheres were obtained from adult mice ranging in age from 4 weeks to >18 months from pooled sexes. PMP spheres derived from

76 mouse insulin promoter (MIP)-GFP mice [CD1/Tg(Ins1-EGFP/GH1); 006864, Jackson Laboratories] were used for live sphere quantification of insulin-GFP intensity. Similar sized spheres from SFM, DMSO and 1µM Dex conditions were pulsed with Hoechst for 30 mins, and 8-10 confocal z-stack images were taken with GFP, and DAPI channel power remaining constant for all images. Projection images were created and analyzed on Image J (https://imagej.nih.gov/ij/) by finding the average pixel intensity value for each sphere and comparing the means of spheres in each condition. More specifically, TIF images of only the GFP channel of each sphere was imported into Image J and converted to grayscale (16 bit). Individual spheres were isolated by tracing around the sphere border and the mean gray value was extracted for quantitative analysis.

3.3.5 Sphere Passaging

Human-derived spheres were passaged using the kynurenic acid, trypsin, hyaluronidase enzyme solution with the addition of collagenase I (0.5mg/mL), collagenase II (0.5mg/mL) and elastase (0.1mg/mL). Mouse-derived spheres were passaged using hyaluronidase (0.67mg/mL), collagenase I (0.5mg/mL), and collagenase II (0.5mg/mL) dissolved in Accustase solution (Sigma; SCR005). Spheres were collected en masse from culture plates or flasks, transferred into one or more 50mL tubes and centrifuged for 5 minutes at 1500rpm. The supernatant was gently aspirated from the pellet and 2-5mL of enzyme solution was added to the pellet and mixed thoroughly. The 2-5mL enzyme and sphere suspension was transferred to a 15mL tube and laid horizontally on an automated rocker at 37°C for 45 minutes. After incubation, the enzyme solution with spheres was triturated approximately 45 times to mechanically dissociate the spheres. The cell suspension was centrifuged for 5 minutes at 1500 rpm. The supernatant was gently aspirated and 1-2mL of trypsin inhibitor solution was added to the pellet and triturated approximately 45 times. The cell suspension was centrifuged for 5 minutes at 1500 rpm. The supernatant was gently aspirated from the resulting pellet and 1-2mL of SFM with FGF2 and heparin (plating media) was added. The cells and media were mixed to ensure a uniform cell suspension and a 10µL sample was taken and cell density was determined from that sample. The main cell pellet was then diluted to 10c/μL.

3.3.6 Medium-Throughput Screening Pipeline

The 400-compound Ontario Institute for Cancer Research (OICR) Tool Compound Library (TCL) 1mM stock plate, screening plate preparation and cell seeding apparatus were provided by the

Toronto Hospital for Sick Children SPARC Biocentre (Toronto, ON). An Echo acoustic dispenser (Labcyte) was used to seed 100nL of DMSO into all vehicle control wells of five 96-well assay plates. Next, 100nL of the 1mM OICR TCL plate drugs were transferred according to the predetermined plate map layouts into the assay plates. Actin-GFP mouse primary RSC spheres were grown and then passaged into a single cell suspension of secondary retinal stem and progenitor cells at a density of 10 cells/µL, according to the sphere growth and passaging methods detailed above. Next, 100µL of cells were seeded into each well of the prepared assay plates using the Bravo liquid handler (Agilent) to result in 1000 cells/well, with drug wells at 1uM, and all wells at 0.1% DMSO. After seeding, another 100µL of cells with 0.1% DMSO were added to the 2x pseudo-positive control wells to achieve a final density of 2000 cells/well. The assay plates were then incubated at 37°C for 7 days. On day 7 of the MTS assay, Hoechst 33342 (10μg/mL; ThermoFisher) was added directly to each well of the 96-well plates and cells were imaged (according to Medium-throughput and content Imaging detailed below) a minimum of 10 minutes afterward.

3.3.7 Medium-throughput and Content Imaging

Primary screening, 96-well plate imaging (2D culture) was performed using a Celigo imaging cytometer, equipped with 4x F-theta lens and 2024x2024 CCD camera (Nexcelom Bioscience). The Celigo software suite was used to extract cell counts and area quantifications for cells that were grown for one week and then live stained for nuclei (Hoechst 33258; 10μg/mL). For mouse tissue, an actin-GFP transgenic mouse strain was used and cell number quantifications were made based on individual nuclei count in DAPI channel or total area of GFP-expressing cells. For human tissue, the green fluorescent cell permeant dye, calcein AM (2μM; ThermoFisher C3100MP) was added at least 15 minutes prior to imaging. Quantifications were made based on nuclei number or total area of calcein-positive cells.

Spheres assays in 24-well plates (3D culture) were imaged using IN Cell Analyzer 6000 (GE Healthcare) equipped with Nikon Plan Apo 4x/NA 0.2 objective and 2048x2048 sCMOS camera. 3D datasets (5 z-planes, 15μm spacing) were acquired using the FITC channel for a total of 12 fields per well. Z-stack was collapsed using maximum-intensity projection and analyzed using a custom image analysis routine for MATLAB 2015b (Mathworks).

Live-dead assays and EdU proliferation assays in 24-well plates (2D culture) were also imaged using IN Cell Analyzer 6000 using the same objective and camera as for spheres assay

78 described above. Fixed cells were stained with Hoechst 33258 to label all nuclei and compared to EthD-1 or EdU-positive nuclei. Image analysis to extract total number of cells and number of EthD-1 or EdU-positive cells was performed in MATLAB 2015b (Mathworks) using custom image analysis routine.

3.3.8 Proliferation and Cell Death Assays

24-well plates were coated with laminin (300µL/well; Sigma) and incubated at least 4 hours at 37°C, then washed with SFM or PBS prior to cell seeding. Secondary retinal stem and progenitor cells in a single cell suspension were seeded manually at 2500 cells/well (5 cells/µL). For cell death assays, 2µM ethidium homodimer (EthD-1; Abcam ab145323) was added to cells and incubated for 15 minutes at 37°C. EthD-1 was then washed out with 3 successive PBS rinses and then cells were fixed using 4% paraformaldehyde (PFA, Sigma). For cell proliferation assays, 10µM 5-ethynyl-2’-deoxyuridine (EdU; Sigma) was added to cells and incubated for 3 hours at 37°C. EdU was then washed out with 3 successive PBS rinses and then cells were fixed using 4% PFA. Fluorescent EdU labeling was achieved using the EdU Click-iT detection kit (ThermoFisher). Hoechst (10μg/mL) was added and 24-well plates were imaged using the IN Cell Analyzer 6000 (GE Healthcare) to determine total nuclei number vs number of nuclei labeled by EthD-1 or EdU (see “Medium-throughput and content imaging”).

3.3.9 Differentiation Assay

24-well plates were coated with laminin (300µL/well; Sigma) and incubated at least 4 hours at 37°C, then washed with SFM or PBS prior to sphere plating. Primary RSC spheres were picked using a 200µL pipet, from plates on an inverted microscope with an external cold light source. Spheres were plated into wells pre-filled with 500µL one of 3 treatment conditions: SFM+1%FBS, SFM+1%FBS+0.1% DMSO or SFM+1%FBS+0.1% DMSO+1µM Dex. Two spheres per well were plated to prevent loss of wells as spheres occasionally do not fully adhere prior to the first media change. Media changes were performed every 4 days by aspirating old media and then refilling wells with 500µL of the same treatment conditions. After 6 weeks, wells were washed with PBS, fixed with 4% PFA and ICC was performed for retinal cell-type markers.

3.3.10 Immunohistochemistry and immunocytochemistry

Mice were euthanized by cervical dislocation while under isoflurane anesthesia. Eyeballs were enucleated from adult mouse skulls, postfixed in 4% PFA for 4 hours at 4°C, then transferred to a cryoprotectant 30% sucrose solution for a minimum of 24 hours. Next, eyes were embedded in Tissue Tek, frozen at -80°C and then sectioned at 10 µm using a cryostat. Fixed frozen eye slides or fixed cells in wells were permeabilized with 0.3% Triton X-100 (Sigma) in PBS for 20 min. Then, they were blocked in 10% normal goat serum (NGS) or 10% normal donkey serum (NDS) for 1 hour. Primary antibodies were diluted in 1% serum from the species used for blocking (to the dilution indicated below) and incubated overnight at 4°C. After washing, secondary antibodies were diluted in 1% serum of the same species at 1:400 (Alexa fluor, ThermoFisher) and incubated for 1 hour. After washing, nuclei were stained with Hoechst 33258 (10μg/mL) for 20 min before a final wash. Mounting medium was added to wells or slides and slides were then coverslipped. Primary antibodies used in this study include: rabbit anti-cone arrestin (1:1000; AB15282, Millipore), mouse anti-rhodopsin (1:500; MAB5316, Millipore), goat anti-calbindin (1:500; SC-7691, Santa Cruz), rabbit anti-PKCα (1:1000; P4334, Sigma), mouse anti-syntaxin (1:100; AB3265, Abcam), goat anti-Brn3a (1:500; SC-31984, Santa Cruz), mouse anti-GFAP (1:500; G3893, Sigma), mouse anti-RPE65 (1:250; MAB5428, Millipore), rabbit anti- Pax6 (1:1000; AB2237, Millipore), rabbit anti-ERG (1:250; AB92513, Abcam) and rat anti-CD68 (1:500; MCA1957, BioRad).

3.3.11 Intravitreal Injections

Intravitreal injections were carried out using a 10µL WPI Nanofil® Injector System with micro- machined 34-gauge beveled needle (World Precision Instruments, Sarasota, FL), a dissecting microscope or surgical scope (Moller Hi-R 900C), a mouse stereotaxic apparatus and heat pad. Mice were brought to a surgical plane of anesthesia via 5% isoflurane and placed on the heat pad in the mouse stereotaxic apparatus (without head stabilization with the ear bars). Once anesthetized, isoflurane was reduced to 3% for maintenance. Mice were injected with 2mg/Kg meloxicam for analgesia. One drop of anticholinergic mydratic (Mydriacyl®) was applied to each mouse eye to dilate pupils. Mice were positioned on one side, so that the eye to be operated on was facing upward, directly under the surgical microscope. A small rubber washer was placed over the eye, so that the washer surrounds the eye like a monocle. A single drop of 3% methylcellulose (MC) solution (in saline) into the monocle, which allows clear visualization of the posterior segment of the eye by the surgeon. The mouse head was stabilized with the non-

80 dominant hand and the needle was controlled with the dominant hand. With the needle, a trans- scleral puncture was made (at a perpendicular angle to the globe) approximately 1 mm posterior of the limbus, in the nasal (anterior) aspect of the eye. The needle passed through the sclera, choroid and retina to enter the retrolental vitreous. The needle was inserted as far as the central area of the retina, taking care to avoid striking the lens, retina or the hyaloid canal. A 2µL bolus of fluid was then injected at an approximate rate of 4µL/min. The adult mouse vitreous space can accommodate up to 3 ul of total fluid because it replaces fluid initially lost from pre-injection vitreous outflow. Thus, the final vitreous volume in the eye is the same as the standard vitreous volume of the mouse eye (7µL) (Schlichtenbrede et al., 2009). Due to the expected final volume of 7µL in the vitreous, Dex concentrations of 0.35µM 3.5µM and 35µM were injected and in order to achieve a final concentration of 0.1µM, 1µM and 10µM in vivo. Once the injection was completed, the needle remained in the retrolental vitreous for an additional 10-15 seconds. This allows for pressure equilibration and works to prevent significant backflow following withdrawal of the needle. Next, the needle was removed, the monocle was removed, and the mouse is rotated to position the other eye for surgery. Once the surgery on both eyes was completed, the mouse was left to recover alone in a recovery cage with a heat lamp, and then reunited with its original cage-mates.

3.3.12 Statistical Analysis

Data are presented as mean ± standard error (SE) unless otherwise noted. Microsoft Excel was used to compute the strictly standardized mean difference (SSMD), signal-to-noise ratio, coefficient of variation, and hypergeometric test. All other statistical analyses were run using Sigmaplot 12 (Systat Software Inc.) or GraphPad Prism 6 (GraphPad Software Inc.). Student’s t-test (two-tailed) was performed for statistical analysis between two groups. One-way ANOVA or a two-way ANOVA (for factor comparisons) with a Holm-Sidak or Fisher’s LSD multiple comparison post-hoc test was used when three or more groups were compared. Sample size (N) values are provided in the figure legends. Statistical significance was set at p < 0.05.

3.4 Results

3.4.1 Medium-throughput screening identifies several unique compound classes that expand retinal stem and progenitor cells in culture

To identify compounds that can expand retinal stem and progenitor cell number, we developed a medium-throughput screening (MTS) pipeline that combined a method for the generation and seeding of retinal stem and progenitor cells with medium-throughput image analysis. Because RSC spheres contain varying proportions of pigmented cells that can obfuscate fluorescence, to help facilitate imaging we used an albino actin-GFP mouse strain for primary RSC dissection and clonal sphere expansion. RSC primary spheres – which contain <1% RSCs and over 99% retinal progenitor cells – were generated using a clonal sphere forming assay (Balenci et al., 2013; Tropepe et al., 2000). Cultures derived from RSC spheres are referred to as retinal stem/progenitor cell (RSPC) cultures. The RSC primary spheres were collected and dissociated into a single cell suspension (10cells/µL, 1000 cells/well) in serum-free media containing FGF2 (10ng/mL) and Heparin (2µg/mL) and were then seeded via automated liquid handler into 96- well plates. Since we did not have a positive reference compound, we used a pseudo-positive “2x control” by seeding some wells at 2000 cells/well (20 cells/µL) rather than 1000 cells/well (10 cells/µL). Each well contained 0.1% DMSO vehicle control, or 0.1% DMSO + 1µM of a single compound from the Ontario Institute for Cancer Research (OICR) tool compound library (TCL). The OICR TCL consists of 400 small molecule agonists and inhibitors, the majority of which are either clinical trial-phase or approved therapeutics. RSPCs were incubated with molecules for 7 days in a monolayer culture and were then live-cell imaged to determine the number of Hoechst-positive nuclei and total area of GFP expression in each well. Data were normalized as percent of control and a hit was defined as a compound that resulted in an increase in both nuclei number and GFP area that were each 3 standard deviations (SD) above the control mean (Figure 3.1, Table B1). Hits also were validated visually to ensure quantification was based on enumeration of Hoechst-stained nuclei and the area of GFP- expressing cells in the well and not due to compound precipitation or other artifacts (Figure B1) and the GFP area-to-nuclei number ratio was calculated to assess compounds for cell hypertrophy effects (Table B1). Control conditions showed low variability and large signal-to- noise ratios that enabled hit identification with statistical confidence (Table B2). We performed two full-library screens that independently resulted in 7 hits (Screen 1) and 6 hits (Screen 2) for a total of 12 unique hits identified (Figure 3.1B-E; Table B1).

Only the synthetic glucocorticoid (GC) agonist prednisolone (Pred) met the hit criteria in both screens. Another synthetic GC agonist, Dex, was a hit in Screen 2. Both Pred and Dex had GFP:Nuclei ratios close to 1 (Table B1), indicating they did not cause cell enlargement relative to 0.1% DMSO control. The OICR TCL contains two other drugs classified as corticosteroids: Hydrocortisone (HC) and Prednisone. However, given that the activity of Prednisolone is 4x that of HC and the activity of Dexamethasone is 25x that of HC, it may not be surprising HC was not a hit at the same 1µM concentration (Page and Barnes, 2017). Likewise, Prednisone is an inactive pro-drug/metabolite that requires conversion to Prednisolone by 11β-hydroxysteroid dehydrogenase (11β-HSD) to be able to cross the cell membrane and have pharmacological effects (Bergmann et al., 2012). Nonetheless, a hypergeometric statistical test still found that 2 hits out of 4 GC-class compounds in the library was a greater hit rate than would be expected by chance (Table B1). As GCs previously have been found to modulate neural progenitor proliferation and differentiation (Anacker et al., 2013; Jeong and Mangelsdorf, 2009), and are known to be important in retinal development and maturation (Gallina et al., 2014), the synthetic GC agonists became our lead hits of interest.

Other hit compounds identified during screening that target molecular signaling pathways known to regulate various stem and progenitor cell types, included: the Rho/ROCK inhibitor, Thiazovivin; the dipeptidyl peptidase IV inhibitor, MK-0431; the indoleamine dioxygenase inhibitor, INCB024360; the a2-adrenergic receptor agonist, Guanabenz; the cathepsin K inhibitor, Odanacatib; and the HDAC inhibitor, Sodium Butyrate (Table B1). However, some of those compounds did have GFP:Nuclei ratios much greater than 1, indicating increased cell number and cell size versus control. A 5-point dose-response assay verified that Pred and Dex resulted in the greatest increase in RSPCs (2.32-fold and 1.96-fold of control, respectively) at the 1µM dose (Figure 3.1F). The compounds with the next highest effects were Thiazovivin (1.92-fold at 1µM) and Guanabenz acetate (1.67-fold at 0.1µM). Therefore, MTS successfully identified several putative hit compound classes, with synthetic GC agonists as the lead hit compound class.

Figure 3.1. Medium-throughput screening identifies several unique compound classes that increase retinal stem and progenitor cell number

(A) Schematic overview of the MTS pipeline to detect RSPC expanding compounds.

(B-E) Quantification as percent of control (POC) of all compounds that met the hit criteria of having a relative number of nuclei or total area of actin-GFP expression over 3xSD above the 1x Control mean. Hits that were identified in multiple criteria and/or screens are colour-coded. (B- C) Seven compounds in Screen 1 were found to be hits in both the number of nuclei and actin- GFP area and were selected as lead compounds. 1x Control: N=80 technical replicates. 2x Control: N=12 technical replicates. Compounds: N=1 technical replicate. (D-E) Eight compounds in Screen 2 were found to be hits in both the number of nuclei and actin-GFP area and were selected as lead compounds. 1x Control: N=80 technical replicates. 2x Control: N=15 technical replicates. Compounds: N=1 technical replicate.

(F) A dose-response assay of the hit compounds identified via MTS. N=3 for all compounds at each concentration.

Data are mean ± SEM.

3.4.2 Glucocorticoid agonists stimulate mouse retinal stem and progenitor cell proliferation in vitro via both glucocorticoid receptor and mineralocorticoid receptor signaling

Here we investigated the cell-biological effects that underpinned the ability of the synthetic GC agonists and other hit molecules to effect the increases in RSPC number detected via MTS. We dissociated primary RSC spheres to single cell suspensions of RSPCs and then performed a 6- day monolayer cell culture assay, in laminin-coated 24-well plates (4 cells/µL, 2000cells/well), to assess survival and proliferation of cells. We analyzed 3 timepoints during the assay: Day 2, Day 4 and Day 6. To assess whether any hit compounds affected cell death, we compared the total cell number at each timepoint with the number of cells that were positive for the cell- membrane-impermeable nucleic acid stain ethidium homodimer (EthD-1), which was pulsed into the culture 15 minutes prior to each timepoint (Figure B2A-B). None of the hit compounds had any effect on cell survival at any timepoint (Figure 3.2A). There was a remarkable amount of cell death at Day 2, as only ~40%-45% of nuclei remained unlabeled for EthD-1 in all conditions. However, the proportion of cell death decreased progressively as ~80% of nuclei were unlabeled by EthD-1 at Day 4 and that increased to ~90% or above by Day 6. Such significant cell death in the first days of the assay likely explains why an increase in total cell number was not observed until Day 6 of the assay, when only the Pred and Dex conditions showed a

85 significant increase compared to control (Figure 3.2C). To assess whether any hit compounds affected cell proliferation, we compared the proportion of total nuclei that had incorporated the thymidine analog 5-ethynyl-2’-deoxyuridine (EdU), which was pulsed for 3 hours prior to fixation at each timepoint (Figure B2C-D). EdU is incorporated into the DNA of cells during the DNA synthesis phase (S phase) of the cell cycle and is subsequently labeled by an azide-containing fluorescent dye to enable detection of proliferating cells (Zeng et al., 2010). In the 0.1% DMSO control condition, EdU-labeling was highest at Day 4 and was almost completely absent by Day 6. Only Pred and Dex significantly increased the proportion of EdU labeled cells compared to control. At Day 4, Dex had a ~5.6-fold increase in EdU labeling while Pred had a ~3-fold increase. Thus, both synthetic GC agonists increased the maximum proportion of RSPCs in S phase of the cell cycle at the peak of proliferation on Day 4 and increased the total cell number at the end of the assay, whereas no other hit compound resulted in a significant increase in EdU labeling or cell number.

To further resolve the molecular signaling pathways mediating the proliferative effect of synthetic GC agonists on RSPCs, we investigated whether a chemical antagonist to mineralocorticoid receptor (MR) (spironolactone) or glucocorticoid receptor (GR) (RU486) could abolish the increased cell number induced by Dex treatment. We focused on Dex because it has much higher potency than Pred (as mentioned above) and a longer duration of action (36- 72hrs for Dex vs 12-36hrs for Pred) (Page and Barnes, 2017). In mouse cells, Dex treatment increased cell number by 2.45-fold compared to control (Figure 3.2D). When MR was blocked, the Dex effect was reduced to 1.78-fold of control, indicating that 46.2% of the increase in cell number was mediated through MR signaling. When GR was blocked, the Dex effect was reduced to 1.55-fold of control, indicating that 62.1% of the increase in cell number was mediated through GR signaling. This indicates that both MR and GR pathways are activated by Dex and contribute to its proliferative effect on mouse RSPCs. This contrasts with other neural progenitors, such as human hippocampal neural progenitors, for which MR signaling has been shown to have a proliferative effect, whereas GR signaling inhibits progenitor proliferation (Anacker et al., 2013). For human RSPCs, MR antagonism abolished the entire Dex effect, whereas blocking GR had no effect (Figure B3). Thus, despite that synthetic GC agonists have greater affinity for the GR, their activation of the MR is potent enough to mediate significant proliferative effects in RSPCs.

Figure 3.2. Glucocorticoid agonists increase retinal stem and progenitor cell proliferation through glucocorticoid receptor and mineralocorticoid receptor signaling in mouse

(A-C) Quantification of growth parameters for RSPCs at Day 2, Day 4 and Day 6 of a monolayer growth assay. * = significantly different from control within that timepoint. (A) The proportion of live cells was not significantly different across drug treatments (p=0.55) and there was no interaction between drug and time (p=0.99). There was a significant effect of time on the proportion of live cells and all times were significantly different (two-way ANOVA F(2,195)=261.9, p<0.001; Holm-Sidak posthoc test, * = p<0.05). N=6 for all conditions at each timepoint. (B) The proportion of proliferating EdU-labeled cells relative to control. There was a significant interaction between drug and time (two-way ANOVA F(24,195)=2.59, p<0.001; Holm-Sidak posthoc test, * = p<0.05). N=6 for all conditions at each timepoint. (C) The total number of cells relative to control. There was a significant interaction between drug and time (two-way ANOVA F(24,429)=3.79, p<0.001; Holm-Sidak posthoc test, * = p<0.05). N=12 for all conditions at each timepoint.

(D) Total RSPC number at the end of a 7-day monolayer growth assay with 0.1% DMSO and 0.1% EtOH in all conditions. Control was significantly different from all Dex treatment conditions but not Spironolactone (Spiro) or RU486 alone (one-way ANOVA F(5,12)=67.88, p<0.001; Holm- Sidak posthoc test, * = p<0.05). N=3 for RU486 alone. N=4 for all other groups.

Data are mean ± SEM.

3.4.3 Glucocorticoid agonism has differential effects on proliferation and self-renewal of adult stem and progenitor cells from different tissues

To further characterize the differential effects of GC signaling on different adult stem cell populations, we treated adult stem cells from two different germ layers with Dex during 7-day clonal sphere forming assays – RSCs and pancreatic multipotent progenitor cells (PMPs). For RSC spheres, a threshold of 80µm in diameter is used to distinguish between spheres that arise from an RSC (≥80µm) and those that arise from a progenitor cell (<80µm), as determined by the diminished passaging ability of spheres below 80µm in our hands. Dex treatment increased the total number of RSC spheres greater than 80µm in diameter (Figure 3.3A-B) and increased the maximum diameter of these RSC spheres (Figure 3.3C). Thus, Dex likely increased progenitor proliferation, which increased sphere size. The increase in sphere number in response to Dex may be due to enhanced progenitor proliferation, resulting in more spheres exceeding 80µm in diameter, or could be due to a subpopulation of quiescent RSCs that normally remain dormant in culture becoming stimulated to proliferate and form spheres. To assess if synthetic GC agonists influence RSC self-renewal, spheres ≥80µm in diameter that were initially grown in 0.1% DMSO, Dex or Pred were passaged into single cell suspensions of tertiary cells and re-

88 plated for a subsequent clonal sphere forming assay in serum-free media + FGF2 and heparin (with no compounds or DMSO present). Both Dex and Pred-treated spheres resulted in a similar increases in the number of tertiary spheres compared to the 0.1% DMSO control (~2-fold increase at 1µM and over 4-fold increase at 10µM), indicating GC agonism can stimulate symmetric RSC self-renewal and expansion (Figure 3.3D-E). Therefore, it appears the synthetic GC agonists increase retinal progenitor proliferation and RSC self-renewal.

For PMPs spheres, ≥30µm in diameter was set as the threshold for quantification, as they are typically smaller than RSC spheres and objects less than 30µm in diameter tend to form due to aggregation given the higher seeding density (20 cells/µL vs 10 cells/µL). Another important difference is that PMP spheres have very low passage efficiency and rarely form secondary spheres (Seaberg et al., 2004). In contrast to RSC spheres, PMP sphere growth was suppressed by Dex treatment. The total number of spheres was significantly reduced, which could be due to fewer sphere colonies reaching the 30µm threshold resultant of attenuated proliferation, or alternatively, some sphere-initiating pancreatic progenitors may remain quiescent in culture in response to GC agonism (Figure 3.3F). However, while the number of spheres between 30-49µm in diameter did not change (Figure 3.3G), there was a significant reduction in spheres 50µm or above in diameter, indicating GC agonism likely inhibits pancreatic progenitor proliferation rather than sphere initiation (Figure 3.3H). This contrasting outcome to that observed for RSC spheres demonstrates that adult stem and progenitor cells from different tissues are regulated by GC signaling in a cell-type-specific manner. Next, using a mouse insulin promoter (MIP)-GFP mouse line, we examined whether Dex treatment would influence the early fate specification of PMPs. PMPs themselves are known to express insulin and the MIP-GFP reporter at a low level, whereas their differentiated beta cell progeny expresses high levels of insulin and the MIP-GFP reporter (Razavi et al., 2015; Smukler et al., 2011). During sphere growth, an increase in reporter expression relative to control would denote the differentiation of PMPs toward beta cells, whereas a relative decrease in reporter expression would denote fate specification toward non-insulin expressing pancreatic progeny. Treatment of PMPs with Dex during 7 days of sphere growth resulted in decreased MIP-GFP reporter expression relative to control, indicating that GC agonism likely has a differentiation effect, directing PMPs toward non-beta cell progeny (Figure 3.3I-L). Furthermore, a differentiation effect could explain the decrease in PMP proliferation caused by Dex. Thus, we have identified that GC signaling also can regulate adult pancreatic progenitor cell differentiation and proliferation, but with different effects compared to retinal stem and progenitor cells.

Figure 3.3. Dexamethasone increases retinal stem cell sphere size, number and self- renewal but inhibits growth and insulin expression of pancreatic multipotent progenitor spheres

(A-C) Quantification of sphere diameter after a 7-day free-floating clonal sphere assay with actin-GFP mouse-derived secondary RSPCs in 0.1% DMSO vehicle or 1µM Dex. (A) The number of cell colonies less than 80µm in diameter was not significantly different between 0.1% DMSO vehicle control and 1µM Dex treatment. N=8 per condition. (B) Cell colonies greater than 80µm in diameter increased in number by 1.63-fold with Dex treatment. (t-test t(14) = -5, p<.001; N=8 per condition). (C) RSC sphere colonies 80µm or above in diameter demonstrated an overall increase in size with Dex treatment (t-test t(14) = 2.56, p<.05; N=8 per condition). * indicates significantly different from 0.1% DMSO control.

(D-E) Quantification of the number of tertiary RSC sphere colonies grown in drug-free media after prior exposure of secondary cells to the indicated compounds at 1µM or 10µM during a 7- day free-floating clonal sphere assay. Exposure to Dex and Pred increased the number of RSC spheres after passaging by 2.2-fold and 2-fold, respectively, at 1µM (D; one-way ANOVA F(2,15)=4.02, p=0.04; Fisher LSD posthoc test, * = p<0.05; N=6 per group) and increased RSC sphere number by 5.2-fold and 4-fold, respectively, at 10µM (E; one-way ANOVA F(2,6)=9.6, p=0.014; Fisher LSD posthoc test, * = p<0.05; N=3 per group). * = indicates significantly different from 0.1% DMSO control.

(F-H) Quantification of adult PMP spheres after a 7-day free-floating clonal sphere assay. (F) The total number of spheres ≥30µm was significantly reduced by all concentrations of Dex tested (one-way ANOVA F(4,10)=4.76, p=0.02; posthoc test, Holm-Sidak posthoc test, * = p<0.05; N=3 experiments). (G) The number of spheres 30-49µm in diameter was not influenced by Dex treatment at any concentration tested. N=3 experiments. (H) The number of spheres ≥50µm was significantly reduced by all concentrations of Dex tested (one-way ANOVA F(4,10)=16.12, p<0.001; Holm-Sidak posthoc test, * = p<0.05; N=3 experiments). * = indicates significantly different from indicated conditions.

(I-L) The intensity of MIP-GFP expression in PMP spheres. (I) Quantification of the intensity of MIP-GFP expression in PMP spheres (one-way ANOVA F(2,6)=18.35, p=0.0028; Holm-Sidak posthoc test, * = p<0.05; N=3 replicates per condition. (J-L) Representative confocal projection images of MIP-GFP expression in PMP spheres.

Data are mean ± SEM.

3.4.4 Glucocorticoid agonism does not change the differentiation profile of retinal stem cell progeny

To investigate whether GC signaling influences the differentiation profile of RSC progeny, we added 1µM Dex for the entire duration of a 6-week differentiation protocol, where whole clonal RSC spheres derived from C57/BL6J mice were plated in laminin-coated 24-well plates (Figure 3.4A). We used these mice due to the potential for retinal cell defects caused by mutations in

91 albino mouse strains and other C57 strains (Mattapallil et al., 2012). The proliferative effect of Dex was evident as the Dex-treated wells had an average 2.3-fold increase in cell number compared to the 0.1% DMSO vehicle at the end of the 6-week differentiation (Figure 3.4B). Markers for all differentiated retinal cell types were assessed, which included: cones (cone arrestin), rods (rhodopsin), horizontal cells (calbindin), bipolar cells (PKCα), amacrine cells (syntaxin), retinal ganglion cells (Brn3a), Müller glia (GFAP) and retinal pigmented epithelium (RPE65). However, no difference in the proportion of any cell type was detected across all conditions (Figure 3.4C-J). On average, bipolar cells appeared to be the most frequent cell type produced at ~57%-84% of progeny across groups, whereas RPE cells were nearly undetected ranging from 0%-0.46% of progeny. Across the 3 treatment conditions, rods were detected at an average range of 4%-28%, cones ranged from 0.8%-14% and RGCs were 2%-15%. Similar to some previous reports using 2D culture (Coles et al., 2004; Del Debbio et al., 2013; Demontis et al., 2012; Gualdoni et al., 2010), RSC progeny did not take on morphological features of mature retinal cell types, such as photoreceptor outer segments. However, this appears to be dependent on culture conditions, as retinal precursors more readily acquire mature morphology in 3D/co-cultures cultures (Akhtar et al., 2019; Eiraku et al., 2011b) and RSC progeny develop mature morphology after transplantation into the retina (Ballios et al., 2015; Inoue et al., 2010). Despite the range of cell type proportions observed between conditions, cell type output was highly variable across biological replicates within conditions, and thus, there were no statistically significant differences in cell type proportions between treatment groups. Nonetheless, due to the 3.58-fold Dex-mediated increase in cell number compared to the 1% FBS condition, the absolute number of cell types produced (such as photoreceptors) is greater with Dex treatment. In that regard, it is notable that, in a separate 6-week assay where RSPCs were seeded as a monolayer instead of plating whole spheres, the proliferative effect of Dex was even greater (6.67-fold increase in cell number versus 0.1% DMSO control; Figure 3.4K-M). However, that experiment also used actin-GFP mice, not C57/BL6J mice, so both assay format and strain- specific differences may influence the degree of GC-mediated proliferation of RSPCs. In sum, GC agonism resulted in a pronounced increase in RSPC proliferation but did not influence cell- type specification during differentiation.

Figure 3.4. Dexamethasone does not affect the differentiation profile of retinal stem cell progeny

(A) Schematic overview of the RSPC differentiation assay.

(B) Quantification of cell number per well after 6 weeks in the indicated differentiation conditions (one-way ANOVA F(2,8)=16.46, p=0.001; Holm-Sidak posthoc test, * = p<0.05; N=3-4 per group). * indicates significantly different between indicated conditions.

(C-J) IHC images and quantification of mature retinal cell type markers following 6 weeks of differentiation across the indicated conditions. N=3-4 per group. Nuclei labeled with Hoechst stain (blue). White arrows indicate cells positive for cell type markers.

(K-M) Images and quantification of secondary RSPCs from actin-GFP mice grown as monolayers for 6 weeks in 1% FBS differentiation media (t-test t(14) = -9.54, p<.001; N=8 wells per condition). * = indicates significantly different between indicated conditions.

Data are mean ± SEM.

3.4.5 Glucocorticoid agonism in vivo induces proliferation in the ciliary epithelium of the mouse eye but does not expand the retinal stem cell population

GCs, including dexamethasone, are commonly used clinically for anti-inflammatory therapy in the eye (Gallina et al., 2014; Nuzzi et al., 2012). Yet, despite this wide-spread use, whether GC agonism stimulates adult RSC or CE proliferation has not been examined to our knowledge. Here, we investigated whether in vivo delivery of Dex can overcome the inhibitory signals in the RSC niche and induce proliferation in the adult mouse eye. We used an intravitreal injection paradigm where each eye received one injection per day for 3 days and was then collected and fixed 24 hours after the final injection for subsequent immunohistochemical (IHC) analyses (Figure 3.5A). We delivered 3 different concentrations of Dex (to achieve final concentrations of 0.1µM, 1µM or 10µM in vivo) or 0.5% DMSO as a vehicle control. For the whole duration of the experiment the thymidine analog EdU was delivered via the drinking water so it would be incorporated by, and label, any cell that entered S phase of the cell cycle. There are currently no exclusive markers for RSCs. Pax6, which is a marker of retinal progenitor cells during development, has been shown to be highly expressed and functionally required in RSCs (Xu et al., 2007b). Pax6 is also known to label amacrine cells, and both layers of the ciliary epithelium in the adult mouse eye (Das et al., 2005; Marquardt et al., 2001). However, the amacrine cells

94 and CE cells marked by Pax6 are easily distinguished based on anatomical location in the retina vs the CE. Therefore, we quantified the proportion of Pax6 stained cells in the CE that were co- labeled with EdU to determine the level of CE proliferation in vivo, which may be an indication of RSC proliferation. The 10µM dose of Dex resulted in a significant increase in Pax6-positive CE cells co-labeled with EdU (Figure 3.5B, E-F; Figure B4), indicating GC agonism can induce CE proliferation in vivo, and thus, potentially stimulate RSCs as well. It also was evident that there were Pax6-negative cells that were labeled with EdU in the ciliary body. To determine what are the non-CE cell types labeled by EdU, we co-stained for the endothelial cell nuclear marker, ERG, and activated microglia/macrophage marker, CD68. There was a consistent proportion of ERG + EdU co-labeled cells across all conditions, indicating there is a basal population of proliferating endothelial cells labeled by EdU that is not enhanced by GC agonism (Figure 3.5C; Figure B5A-B). In contrast, the 10µM dose of Dex increased the proportion of CD68 + EdU co- labeled cells compared to the 0.5% DMSO control (Figure 3.5D; Figure B5C-D). This result could be an indication of microglia/macrophage EdU incorporation due to GC-mediated cell death/DNA repair activity (Zeng et al., 2017). However, no evidence of co-localization of activated caspase 3 with CD68 or EdU was found (data not shown). To investigate whether in vivo GC agonism not only causes CE proliferation but increases the number of RSCs, we performed a similar intravitreal injection paradigm, but this time waited 7 days following the final injection and then dissected the primary ciliary epithelium to perform 7-day clonal sphere assays (Figure 3.5G). We then quantified the number of primary RSC spheres generated from eyes of each treatment condition as a proxy for the number of RSCs present in vivo. There were no significant differences in the number of spheres produced across treatment conditions (Figure 3.5H). Thus, whereas GC agonism in vivo does induce CE cells to proliferate, which may include stimulation of RSC proliferation, it does not result in increased RSC symmetric self- renewal and expansion. Therefore, if RSCs are being stimulated to exit quiescence and proliferate, they are most likely proliferating asymmetrically.

Figure 3.5. Intravitreal dexamethasone injection induces ciliary epithelium proliferation but does not expand the retinal stem cell population in vivo

(A) Schematic of the intravitreal injection paradigm followed by endpoint IHC.

(B) Quantification of Pax6+EdU co-labeled cells relative to total CE area in eyes treated with 0.5% DMSO vehicle or indicated Dex concentrations (one-way ANOVA F(3,15)=6.21, p=0.006; Holm-Sidak posthoc test, * = p<0.05; N=3-6 eyes per condition). * = indicates significantly different between indicated conditions.

(C) The proportion of ERG + EdU co-labeled endothelial cells relative to the total number of EdU-labeled cells in the ciliary body for each indicated condition. N=3-6 eyes per condition.

(D) The proportion of CD68 + EdU co-labeled microglia/macrophages relative to the total number of EdU-labeled cells in the ciliary body for each indicated condition (one-way ANOVA F(3,14)=6.89, p=0.004; Fisher Holm-Sidak posthoc test, * = p<0.05; N=3-6 eyes per condition). * = indicates significantly different between indicated conditions.

(E-F) Representative images of Pax6 IHC and EdU labeling in the ciliary body and peripheral retina of mouse eyes exposed to 0.5% DMSO vehicle or indicated Dex concentrations. Nuclei labeled with Hoechst stain. White arrows indicate Pax6 + EdU co-labeled cells. 10 µm-thick sections. 50µm scale bars.

(G) Schematic of intravitreal injection paradigm followed by endpoint primary sphere-forming assay.

(H) Quantification of RSC sphere frequency relative to naïve un-injected control after a seven- day clonal sphere growth assay following intravitreal injection of indicated conditions. N=6 eyes per group.

Data are mean ± SEM.

3.5 Discussion

In this study we developed an MTS screening method to identify small molecules that expand the number of retinal stem and progenitor cells in culture. We demonstrated that our lead hit molecules, synthetic glucocorticoid agonists, enhance RSPC proliferation and RSC self-renewal in vitro, and have the capacity to induce CE proliferation in vivo. Synthetic glucocorticoids signal through the glucocorticoid and mineralocorticoid receptors and also are regulated at the prereceptor level by the isozymes 11-β-HSD1 and 11-β-HSD2 (Oyimba et al., 2006; Ramamoorthy and Cidlowski, 2013). The GR, MR, and the 11β-HSD1 and 11β-HSD2 isozymes are known to be expressed in the CE of humans and rodents (Oyimba et al., 2006). In a recent study, our lab performed RNAseq on primary RSC spheres derived from two different strains of mice (Khalili et

97 al., 2018). The transcriptomic data set shows that mouse RSC-derived clonal spheres express the GR gene (Nr3c1), the MR gene (Nr3c2) and 11β-HSD isozymes 1 and 2 (hsd11b1, hsd11b2), with Nr3c1 showing particularly high expression (Figure B6). Given that the expression level of GR protein determines the magnitude of glucocorticoid response, these data support the findings of this study that RSPC cultures are responsive to GR agonists dexamethasone and prednisolone (Ramamoorthy and Cidlowski, 2013). By inhibiting GR or MR signaling, we found that both pathways act in mouse RSPCs to produce the observed increase in cell number caused by dexamethasone treatment. However, the increase in the number of human RSPCs caused by dexamethasone treatment was completely abolished by the addition of MR antagonist spironolactone, indicating MR but not GR has an influence on human RSPC proliferation. Further, the potency of the increase in cell number was much less in human compared to mouse (1.22-fold vs 2.45-fold), which may be due to the absence of an effect of GR. Indeed, when GR was blocked in mouse RSPCs using RU486, the Dex effect in mouse was reduced to a similar magnitude (1.55-fold) observed for human. Expression data for GR, MR and 11β-HSD isozymes in human RSPCs could help explain differences in their GR sensitivity, but these studies have not yet been explored. However, this apparent species- specific difference for the cell-biological effects of glucocorticoid signaling on retinal stem and progenitor cells could result from differences at the pre-receptor/receptor level or be due to downstream factors such as co-activator regulation, chromatin landscape, posttranslational modifications and interactions with other signaling pathways (Ramamoorthy and Cidlowski, 2013).

Another factor that could influence the outcome of GR and MR agonism is the dose. Our screen was designed to identify compounds that increased the number of RSPCs in culture at a 1µM concentration. The highest dose tested in this study in vitro was 10µM: cell expansion was tested with a dose-response assay and self-renewal was tested with a sphere passaging/self- renewal assay. In both cases, 10µM Dex and Pred produced significant increases. Compared to 1µM, 10µM Dex and Pred had slightly lesser effects on cell number but greater effects on self- renewal. Generally, GCs are thought to decrease proliferation and neurogenesis of NSCs and their progeny (Odaka et al., 2017). However, an in vitro study on human hippocampal progenitors by Anacker et al. (2013) found that low-dose cortisol (100nM) operated through MR to enhance progenitor proliferation, whereas high-dose cortisol (100 µM) acted via GR to inhibit progenitor proliferation. That observation is in line with the higher affinity of cortisol for MR than GR (Pariante and Miller, 2001). Also notable, at both high and low doses of cortisol, they found

98 a reduction in neurogenesis and differential influence on astrogliogenesis. This contrasts with our finding that Dex had no effect on RSC progeny differentiation into any specific retinal cell types. Yet, in our study only 1µM Dex was tested during differentiation, so the possibility remains that there could be a dose-dependent effect of GR agonism on RSC progeny differentiation. However, since here we are discussing stem/progenitor cells of different species and different tissues, and glucocorticoids are well-known to have cell type-specific effects, these differences may occur regardless of dose (Ramamoorthy and Cidlowski, 2013; Sulaiman et al., 2018). Indeed, our experiments with mouse PMPs resulted in findings that directly contrast our RSPC results. Dex treatment mediated a reduction in clonal PMP sphere size and number, as well as decreased insulin expression, which is indicative of decreased proliferation and altered differentiation (Razavi et al., 2015; Smukler et al., 2011). These results were achieved at the same 1µM and 10µM Dex doses that had no effect on RSC progeny differentiation and increased RSPC proliferation confirming cell-specific differences in the effect of glucocorticoids on adult progenitor cells from different tissues.

It has been demonstrated previously that drugs which increase stem cell proliferation and self- renewal in vitro can have regenerative effects when applied in vivo (Dadwal et al., 2015; Naska et al., 2016; Reeve et al., 2016). We hypothesized this may also be true for RSCs, and therefore, we tested several concentrations of Dex in vivo in the mouse eye via a series of intravitreal injections. We found that a 10µM dose of Dex could induce a significant increase in EdU labeling in the CE, indicative of CE proliferation. Since we used Pax6 to label the CE, and Pax6 is known to be expressed in retinal progenitors and RSCs, it is possible that proliferating RSCs and RSC-derived progenitors were the cells being labeled by EdU in the CE (Xu et al., 2007b). Further, while the frequency of Pax6 + EdU labeling was very low, that is concordant with previous reports of RSCs being very rare at ~1 in 500 CE cells (Coles et al., 2004). However, without a distinctive molecular marker for RSCs or newborn progenitors, this interpretation cannot be determined directly via immunohistochemistry.

We also found that Dex increased the proportion of CD68 + EdU co-labeled cells in the CE and did not find any evidence this was due to DNA repair activity. This finding could be due to increased immune cell infiltration as, although GCs are well-known for their anti-inflammatory properties, recent studies have demonstrated that GCs can also mediate pro-inflammatory responses (Cruz-Topete and Cidlowski, 2015; Gallina et al., 2014). Alternatively, GCs have been shown to increase microglia proliferation in the CNS in vivo (Nair and Bonneau, 2006). Furthermore, Dex-mediated immunomodulation in the zebrafish eye can both delay or

99 accelerate neuronal regeneration by MG cells depending on whether it is delivered pre- or post- injury (White et al., 2017). Thus, it is possible that Dex may mediate CE proliferation (and potentially RSC activation) indirectly via immune modulation.

We sought to ascertain if there was an in vivo RSC self-renewal effect of Dex by performing the same intravitreal injection paradigm with a follow-up primary sphere assay 7 days after the injection period. However, no difference in sphere number resulted from Dex treatment, indicating no in vivo effect of Dex on symmetric self-renewal and expansion of RSCs. This is in line with previously reported findings of exclusive asymmetric division of proliferating CE/RSCs in vivo (Abdouh and Bernier, 2006; Coles et al., 2006). Furthermore, asymmetric division is known to be the exclusive mode of division for several other adult stem and progenitor cells in vivo (Post and Clevers, 2019). Thus, until specific markers of RSCs are elucidated, it is inconclusive whether the GR agonist-mediated proliferation in the CE demonstrated herein is due to the stimulation of RSCs. It is also possible that a higher dose or more prolonged exposure could lead to RSC expansion in vivo.

Many stem cell signaling pathways have been demonstrated to regulate RSPC proliferation and RSC self-renewal. For instance, Wnt activation and Notch activation have each been shown to increase RSPC proliferation and symmetric self-renewal of RSCs (Balenci and van der Kooy, 2014; Inoue et al., 2006). Also, Hedgehog signaling blockade has been shown to decrease the proliferation of RSPCs in culture (Khalili et al., 2018), and mice with a Ptc+/- mutation have an extended period of postnatal retinal progenitor proliferation in vivo (Moshiri and Reh, 2004). However, since the discovery of RSCs onward, it has been postulated that the in vivo quiescence of RSCs in the adult mammalian eye is mediated by inhibitory factors in the RSC niche that impede the ability of exogenous factors to stimulate endogenous RSCs (Ahmad et al., 2000; Tropepe et al., 2000; Wohl et al., 2012). Notably, Balenci et al. (2013) reported that lens and cornea-secreted BMP and sFRP proteins might be responsible for the quiescence of RSCs in vivo based on their ability to reversibly suppress RSC sphere growth in vitro. Coincidentally, glucocorticoid signaling has been shown to regulate several molecular signaling pathways, including Wnt signaling, Notch signaling, BMP signaling and Hedgehog signaling in various progenitor populations and tissues, including neural progenitors (Anacker et al., 2013). Thus, it will be important to investigate the effect of glucocorticoid signaling on the regulation of these canonical stem cell signaling pathways in retinal stem and progenitor cells to determine if modulation of these pathways explains the proliferative effect of Dex on the CE in vivo. It also may be possible that concurrent blockade of BMP and/or sFRP proteins will enhance the

100 proliferative effect of Dex in vivo and lead to greater therapeutic potential for endogenous retinal repair.

All together, our findings suggest that compound screening can identify novel regulators of RSPC proliferation and self-renewal, which may have efficacy in the activation of endogenous RSCs. Further, as synthetic glucocorticoid agonists are commonly used clinically for the treatment of ocular diseases (Sulaiman et al., 2018), this study raises the possibility that these drugs, which are already known to be safe in humans for ocular use, could be adapted for retinal regenerative therapy. And, more speculatively, it may be that RSC-mediated retinal regeneration is an as-of-yet unexamined outcome of ocular glucocorticoid administration in humans.

101

Chapter 4

Induction of rod versus cone photoreceptor-specific precursors from retinal stem and progenitor cells

*This chapter was adapted from the following paper published in Stem Cell Research:

Saeed Khalili, Brian G. Ballios, Justin Belair-Hickey, Laura Donaldson, Jeff Liu, Brenda L.K. Coles, Kenneth N. Grisé, Tahani Baakdhah, Gary D. Bader, Valerie A. Wallace, Gilbert Bernier, Molly S. Shoichet, Derek van der Kooy. Induction of rod versus cone photoreceptor-specific progenitors from retinal precursor cells. Stem Cell Research 33 (2018) 215–227. https://doi.org/10.1016/j.scr.2018.11.005 This is an open access article under the CC BY-NC- ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).

S.K and B.G.B contributed equally to this work and were the team leads for the cone experiments and rod experiments, respectively.

K.N.G primarily contributed to cone experiments and minorly to rod experiments, however the rod and cone experiments are highly integrated in the results, so both are included for completeness and clarity. K.N.G assisted with primary cell isolation, differentiation and IHC in Fig. 4.1D-E; post-FACS cell culture/quantification in Fig. 4.2D; primary cell isolation and cell culture in Fig. 4.3; the experimental design, primary cell isolation and cell culture for Fig. 4.6C- D; the design, implementation and cell isolation/culture of the FACS-RNAseq experiment that produced Fig. 4.6D/Fig. C2-C3. However, S.K. performed the FACS, RNA isolation and cDNA library prep, and J.L. performed the sequencing data analysis that generated the PCA/pathway analyses and Tables C1-C3. K.N.G also assisted with editing of the manuscript.

4.1 Abstract

During development, multipotent progenitors undergo temporally-restricted differentiation into post-mitotic retinal cells; however, the mechanisms of progenitor division that occurs during retinogenesis remain controversial. Using clonal analyses (lineage tracing and single cell cultures), we identify rod versus cone lineage- specific progenitors derived from both adult retinal stem cells and embryonic neural retinal progenitors. Taurine and retinoic acid are shown to act in an instructive and lineage-restricted manner early in the progenitor lineage hierarchy to

102 produce rod-restricted progeny from RSCs. We also identify an instructive, but lineage- independent, mechanism for the specification of cone-restricted progeny through the suppression of multiple differentiation signaling pathways. These data indicate that exogenous signals play critical roles in directing lineage decisions and resulting in fate-restricted rod or cone photoreceptor progeny in culture. Additional factors may be involved in governing photoreceptor fates in vivo.

4.2 Introduction

Retinal cells are born in a prescribed and sequential manner during development (Turner and Cepko, 1987). Previous studies (Holt et al., 1988; Turner et al., 1990; Wetts and Fraser, 1988) showed that retinal progenitor cells can give rise to heterogeneous clones, but it was still unclear whether multipotency was a common feature of all retinal progenitors, or whether this potency became restricted with continuing progenitor divisions. Recent live-cell imaging techniques have shown that some late retinal progenitors may be programmed to produce specific combinations of retinal cells (Cohen et al., 2010; Gomes et al., 2011).

Adult retinal stem cells (RSCs) are rare pigmented cells in the ciliary epithelium at the retinal periphery of mice (Tropepe et al., 2000) and humans (Coles et al., 2004). RSCs proliferate in vitro to give rise to spheres of stem/progenitor cells and can differentiate into all retinal neural lineages, including photoreceptors, as well as retinal pigment epithelial (RPE) cells (Tropepe et al., 2000). While some labs have challenged the stem cell nature of the RSC (Cicero et al., 2009; Gualdoni et al., 2010), others have confirmed RSCs as stem cells (Abdouh and Bernier, 2006; Ahmad et al., 2000; Coles et al., 2004; Del Debbio et al., 2013; Demontis et al., 2012; Fang et al., 2013; Inoue et al., 2010). One lab has argued a transdifferentiation origin for RSCs (Cicero et al., 2009); however, this is unlikely given the ability to prospectively enrich a rare clonal population from the ciliary epithelium with stem cell properties (Ballios et al., 2012). Transdifferentiation is defined as the conversion of one differentiated, non-stem cell type directly to another differentiated cell type, and has been demonstrated in the retina of amphibians as well as embryonic chick and rat by manipulation of exogenous growth factors (Opas and Dziak, 1994; Park and Hollenberg, 1989). Prospective enrichment of a clonally proliferative and multipotent retinal stem cell based on the criteria of size, pigmentation and P-cadherin levels (Ballios et al., 2012) suggests that transdifferentiation is an unlikely mechanism to describe these results. Additionally, in both Chx10-null and Mitf-null mice with reduced neural retinal and RPE progenitor populations, respectively, a 3–8-fold increase of RSCs was observed (Coles et al.,

103

2006). If this was due to ciliary epithelium transdifferentiation, then in the Mitf-mutant, fewer pigmented epithelial cells should be available for transdifferentiation into RSCs.

We previously found that combinations of taurine (T), retinoic acid (RA), Fibroblast Growth Factor 2 (F) and heparin (H) (T+RA+FH) added to differentiating clonal RSC colonies increases the number of rods to 90% of all progeny (Bassett and Wallace, 2012; Demontis et al., 2012). These terminally-differentiated cells show no evidence of pigmentation by electron microscopy and display multiple markers of mature rod photoreceptors (Ballios et al., 2012). The time courses for the expression of immature (Neural retina leucine zipper, Nrl+) and mature (Rhodopsin+) rod markers by RSC-derived rods closely follow the profile of Nrl/Rhodopsin expression during rod development in vivo (Akimoto et al., 2006), suggesting adult RSC-derived rods in vitro may pass through a similar intrinsic differentiation program as newborn rods in vivo. In this study, we investigated the hypothesis that T+RA acts directly on early RSC progeny in an instructive, rather than permissive manner, to bias photoreceptor differentiation through the enrichment of rod-specific progenitors.

Several studies have suggested that the cone fate may be a default pathway of photoreceptor development (Akimoto et al., 2006; Brzezinski and Reh, 2015; Mears et al., 2001; Szél et al., 1994). For example, deletion of Nrl leads to the complete loss of rod function with normal cone function, mediated by short-wavelength or S-cones (Mears et al., 2001). Likewise, retinoid- related orphan receptor β (RORβ, a rod and cone differentiation regulator) knockout mice lack rods and show an excess of S cone-like photoreceptors, recapitulating the effects seen in Nrl−/− mice (Swaroop et al., 2010). Finally, studies on the ontogeny and evolution of photoreceptors suggest that the S-cone represents an evolutionarily older form of photoreceptors (Kim et al., 2016b; Szél et al., 2000). To study whether S-cone is a default pathway for retinal progenitors and precursors, we used COCO, a Cerberus/Dan family member that inhibits BMP, Nodal/TGFβ and Wnt signaling pathways to block other retinal cell fates. COCO has been previously used to induce the differentiation of human embryonic stem cells into cone photoreceptors and has been shown to be a powerful neural and photoreceptor inducer (Zhou et al., 2015). The cell biological mechanisms underlying the production of cone photoreceptors by COCO are not known, but we hypothesize that it may instruct lineage decisions in retinal progenitors and precursors by suppressing alternative non-cone cell fates.

104

4.3 Materials & Methods

4.3.1 Animals

The mice used in these studies for isolation and characterization of RSCs and RSC-derived rod and cone photoreceptors include C57BL/6, Actin.gfp (This transgenic mouse line, with an “enhanced” GFP (EGFP) cDNA under the control of a chicken beta-actin promoter and cytomegalovirus enhancer, has widespread EGFP fluorescence, with the exception of erythrocytes and hair), Actin.yfp (constitutively express YFP in all cells) and CCDC 136−/− mice (Smiley et al., 2016). Experimental procedures were performed in accordance with the Guide to the Care and Use of Experimental Animals and approved by the Animal Care Committee at the University of Toronto.

4.3.2 Cell Culture

RSCs were derived from the ciliary epithelium of adult mice (male and female, minimum 6 weeks old), or from E14 presumptive ciliary marginal zone epithelium, as described previously (Coles et al., 2006). Neural retina progenitors were derived from the neural retina tissue of the E14. Cells were plated in serum free media (SFM) on non-adherent tissue culture plates (Nunc; Thermo Fisher Scientific) at a clonal density of 10–20 cell/μL+Fibroblast Growth Factor and Heparin.

4.3.3 Differentiation

For differentiation, individual RSC spheres were selected after 7 days of primary culture. Spheres were derived from Actin.gfp mice to confirm appropriate sub-cellular localization of protein products on immunofluorescence and spheres from C57BL/6 were picked on average 80–100 μm. Spheres (1–2/well) were plated on laminin (50 ng ml−1, Sigma)-coated 24-well plates (Nunc). Following four days of culture in SFM plus FGF2 (10 ng ml−1, human recombinant; Sigma) and heparin (2 ng ml−1; Sigma) to encourage sphere adhesion and spreading, the media was replaced with the rod-induction media (refreshed every four days): SFM plus taurine (100 μM; Sigma), retinoic acid (RA) (500 nM; Sigma), FGF2/heparin (FH), as described previously (Ballios et al., 2012). Pan-retinal differentiation media includes 1% FBS (Invitrogen, Burlington, ON) and FGF2/heparin. For cone differentiation: Pan-retinal conditions + COCO 50 ng/mL (R&D Systems 3356- CC) were used for 28 or 45 days. COCO was added

105 from day 0 of the differentiation. For experiments involving sonic hedgehog signaling blockade, cyclopamine (1 μM; Toronto Research Chemicals, Inc.) was added beginning at either 0 days or 14 days of differentiation according to the schema outlined in Fig. 4.5A. Cyclopamine dose was based on dose-response for pan-retinal cell survival at 14 days of differentiation (data not shown).

4.3.4 Immunostaining

Cells were rinsed with PBS, followed by 4% paraformaldehyde for 10 min at room temperature. Cells were permeabilized with 0.3% Triton X-100 for 10 min and pre-blocked with 5% normal goat serum for 1 h at room temperature. Primary antibodies, rabbit anti-cone arrestin (AB15282, 1:2000; Millipore), rabbit anti-S-opsin (Ab81017, 1:100; Abcam), mouse anti rhodopsin (MAB5316, RetP1, 1:250; Millipore), mouse anti RPE65 (MAB5428, 1:250; Millipore), mouse anti-Pax6 (1:400; Developmental Studies Hybridoma Bank, Iowa City, IA), PKC (ab19031, 5 μg/mL), MITF (ab12039, 1 μg/mL), mouse anti- CRALBP (1:500; Abcam, Cambridge, MA), mouse anti-calbindin (1:500; Sigma) and Rhodamine peanut agglutinin (RL-1072, Vector laboratories) were reacted for 1 h at room temperature or overnight at 4 °C. Samples then were incubated with Alexa Fluor secondary antibodies conjugated with Alexa488, Alexa568, or Alexa647 (all 1:400) (Life Technologies) for 1 h at room temperature. Nuclei were stained with Hoechst dye 33258 (Sigma) (Bazhulina et al., 2009). Cell staining was examined under a fluorescence microscope (Axio Observer D1; Carl Zeiss) with AxioVision 4.8 software (Carl Zeiss).

4.3.5 Retroviral Clonal Labeling

Proliferating RSC progeny were labeled using a GFP-expressing retrovirus 12 h after plating in rod differentiation or pan-retinal differentiation conditions. Retrovirus was prepared as previously described (Holowacz et al., 2011). Concentrated aliquots of virus from a single stock were serially diluted in SFM immediately before use. A dilution of approximately one viral particle per well reliable produced 0-to-1 labeled clone. Media was changed after 48 h. Cells were fixed and clones analyzed by immunostaining at 28 days of differentiation (a time sufficient to produce mature, post-mitotic, RSC-derived rods (Ballios et al., 2012). Clone survival rates were calculated taking into account the plating efficiency in terms of the number of wells with single cells at 16 h post-FACS, and the survival of those clones at the end of differentiation.

106

4.3.6 Flow cytometry, sorting and labeling cells

For pigmentation sorts, RSCs spheres were dissociated into single cell and cells were sorted based on forward and side scatter without the use of surface markers using a FACS Aria (BD Biosciences) (Fig. C2 A–D) (Ballios et al., 2012). Cells were counterstained with propidium iodide (0.9 μg μl−1, Invitrogen) to assess viability. Analysis was performed using BD FACS Diva Software V6.1.2. Single cell/well sorts were performed into 96 well clear-bottom plates (Nunc), coated with laminin. 14% of single cells survived dissociating and sorting, and 50–70% of those clones were present by the end of the experiment, showing that clones in approximately 7–10% of wells initially plated survived until the end of the experiment. When defining a pigmented population of cells from the sphere for sorting, a population of cells with high SSC was chosen by comparison to control neural retina samples (which do not contain pigmented cells). This sort was displayed by plotting against a second, empty channel (FITC) as is standard practice. Cells exhibit some autofluorescence in the FITC spectrum, but were not specifically stained with any fluorescent-tagged antibodies, or sorted based on this. For cone photoreceptor isolation and sort, neural retina tissue was harvested and dissociated into single cell using a Papain kit (Worthington #LK003150). Cone photoreceptors first labeled with peanut agglutinin (PA), a cone specific marker, and then doubled sorted for both GFP and Rhodamine Red.

4.3.7 Quantitative RT-PCR cells

RNA was extracted using NORGEN BIOTEK RNA extraction kit (Cat# 35300) with DNase to remove genomic DNA contamination. RNA was quantified using Nanodrop and a specified amount of cDNA was reverse-transcribed using Superscript III (Invitrogen#18080-051). PCR was carried out using standardized TaqMan Gene Expression Assays in a 7900HT Fast Real- Time PCR System (Applied Biosystems). Quantification was performed using the delta Ct method with Hprt, Beta-actin or Gapdh as an endogenous control, and neural retina tissue as calibrator.

4.3.8 RNA sequencing

The transcriptome of 3 groups (and 3 independent biological experiment in each group): CCDC- RSCs, CCDC-COCO-cones, and CCDC- endogenous cones were compared. CCDC 136−/− mice express GFP in their cone photoreceptors (Smiley et al., 2016). We took advantage of GFP marker to purify endogenous cones and RSC-derived cones from CCDC 136−/− mice for

107

FACS (Fig. C2 B and C). High quality total RNA (RIN: 9–10) was subjected to directional RNA- sequencing library construction from three independent biological replicates. Sequencing was performed using GAIIx (Illumina, Inc., San Diego, CA; www.illumina. com). FASTQ files were generated from reads passing Chastity filter and analyzed for differential expression and PCA analysis.

4.3.9 Correlation at pathway activity level

We employed the BioConductor package of Gene Set Variation Analysis (GSVA) with the ability to perform pathway analysis on individual samples. GSVA was performed using the CPM- normalized and log2-transformed gene expression values of all samples with the following pathway gene set database: Mouse_GOBP_AllPathways_no_ GO_iea_October_01_2016_symbol.gmt (Hänzelmann et al., 2013; Merico et al., 2010). GeneSet size was limited to a range of 10–200 genes. GeneSet size was limited to range 10– 200. 2000 permutations were carried out. Using the gmt file and same parameters as for GSEA the pathway activities of each cone samples in our study was calculated. The rank file generated for the reference dataset was used for GSEA calculation and served as reference in correlation analysis.

4.3.10 Determining differentially expressed (DE) genes and ranks

The standard method in the EdgeR software, Quasi-likelihood F-test, was used for DE determination in EdgeR, because we have the minimal number of samples required (minimum 4 samples total and at least 2 per group) and it is more stringent than the classical and likelihood ratio methods. The ranking score for each gene is generated by p-values and fold changes from the analysis with the following formula: Sign(logFC) x −log10(p value). Sign(logFC) determines the direction of the change with +ve as up-regulation and −ve as down. −log10(p-value) determine the scale of ranking, the lower the p-value, the higher the score. The genes are ordered from top up-regulated to down-regulated ones as rank files.

4.3.11 Lists of differentially expressed genes

DE genes from all 3 comparisons with false discovery rate or FDR q-value<0.05 are listed and ranked by logFC (log 2 of Fold Change). The tables include logFC, rank scores, p-values, and FDR q-values. Top 20 regulated genes are listed here for each comparison. In each comparison

108 we compared two types of cells, the genes are labeled in black for one cell type and blue for the other. Total 3 comparisons are: 1) Endogenous Cone vs COCO Cone (Table C1); 2) CCDC- RSC vs COCO Cone (Table C2) and 3) CCDC-RSC vs Endogenous Cone (Table C3).

4.3.12 Cell counts and statistics

All cell counts and pooled data are presented as averages with standard errors of the mean (SEMs). Statistics were performed using R (V 2.15.0) and Prism 5. Significance is noted using Student's t-test to compare two groups, or ANOVA when comparing three or more groups, with Tukey-Kramer post-hoc analysis (Bonferroni-adjusted p-values) for pairwise comparisons, where appropriate. Significance was noted for p- values<0.05.

4.4 Results

4.4.1 RSC progeny can be biased towards rod or cone fates

In order to understand if early exposure to exogenous factors could bias RSC progeny fate, primary cultures of dissociated ciliary epithelium from adult mice were treated with T+RA+FH or FH-only during the standard 7-day sphere forming assay (14, 21). Our previous work showed that FH alone does not affect rod differentiation or pigmentation (Ballios et al., 2012). There was no difference in the number of clonal RSC sphere colonies (Fig. 4.1A), which were of similar size (99 ± 13 μm in FH-only vs. 103 ± 11 μm in T+RA+FH). Clonal RSC spheres are composed of both neural retinal (non-pigmented) and RPE (pigmented) progenitors (Coles et al., 2006). The pigmented progenitors lose their pigments in culture conditions while maintaining the expression of RPE markers such as MITF (Fig. C1B and C). Next, we used Actin.yfp mice, a transgenic mouse line with an “enhanced” YFP (EYFP) with widespread YFP fluorescence. The advantage of using Actin.yfp cells is easier visualization of pigmented cells, as pigment absorbs YFP fluorescence. Thus, the degree of pigmentation in mixed spheres can be easily visualized. Spheres derived in T+RA+FH exhibited less pigmentation than FH-derived spheres, and thus may contain more non- pigmented progenitors than pigmented progenitors (Fig. 4.1C). When T+RA+FH-derived spheres (rod “lineage-primed” spheres) were differentiated in 1%FBS+FH for 40 days, the percentage of cells expressing Rhodopsin (a rod photoreceptor marker) was increased and RPE65 (an RPE marker) decreased compared to FH-derived spheres differentiated in 1%FBS+FH (Fig. 4.1B for quantitative data and Fig. 4.1E for immunocytochemical images). We and other groups have shown that RSC derived progeny

109 treated with T+ RA express the Nrl and Rhodopsin genes (Ballios et al., 2012; Demontis et al., 2012), but do not express other retinal cell type specific genes (Ballios et al., 2012).There were no differences in total cell numbers of the minor populations expressing Pax6 (retinal progenitors and small numbers of differentiated amacrine cells), calbindin (interneurons), or CRALBP (Müller glia) (Fig. 4.1B).The increased number of rods arising from progenitors primed in T+RA at the expense of RPE differentiation suggests that these factors are instructive for the production of neural retinal progenitors during clonal RSC sphere formation, or at least critical for directing early lineage decisions between fate-restricted progenitors in vitro.

To evaluate the effect of COCO on adult RSC differentiation, we first tested different concentrations of COCO during the differentiation of clonal spheres derived from RSCs (Ballios et al., 2012; Tropepe et al., 2000). We found that a 50 ng/mL or higher concentration of COCO (+FH+1%FBS) induced cone differentiation to approximately 60% of RSC progeny during a 28- day differentiation period, measured by cone arrestin expression, a mature marker of cone photoreceptors (Fig. C1A). Furthermore, protein kinase (PKC), a bipolar cell marker, was not expressed or detected in cells cultured in COCO conditions (Fig. C1D and E). In pan-retinal differentiation conditions (1% FBS+FH) alone, RSC progeny produced<1% cone arrestin positive cells (Fig. 4.1D). In contrast, following 28 days of COCO treatment (COCO +1%FBS+FH), 56% of RSC progeny were positive for cone arrestin (Fig. 4.1D) and 46% were positive for S-opsin (Fig. 4.1D). Given that the commercial primary antibodies that were available to co-stain the COCO-derived cells for both of these markers were from the same species, we instead tested whether there were any differences in the expression levels of cone arrestin versus S-opsin (Fig. 4.1E). Immunofluorescence staining of S-opsin and cone arrestin expressing cells are shown in Fig. 4.1E. None of the cells positive for these cone markers co- stained for Rhodopsin or RPE65 (data not shown). Important, similar to previous studies (Demontis et al., 2012; Sparrow et al., 1990), we observed both in vitro rod and cone like-cells in two-dimensional cultures develop processes, but do not elaborate an outer-segment like structures (Ballios et al., 2012, 2015; Demontis et al., 2012; McUsic et al., 2012; Sparrow et al., 1990). However, scaffolds in three-dimensional cultures help photoreceptors develop outer- segment like structures (McUsic et al., 2012).

110

111

Figure 4.1. Taurine and retinoic acid do not affect RSC sphere derivation from adult ciliary epithelium but do shift baseline rod differentiation potential. COCO increases cone differentiation from RSC progeny.

(A) No difference was found in the number of clonal RSC spheres derived from adult ciliary epithelium in standard growth media (FH) or with the addition of T+RA (t-test, p=0.19).

(B) Priming RSC progeny in T+RA shows a significant effect on neurogenic potential in pan- retinal differentiation conditions (1%FBS) assayed at 40 days, marked by a proportional shift from RPE (RPE65+) to rod photoreceptors (Rhodopsin+) (two-way ANOVA, interaction effect of cell type and differentiation condition on expression levels F (8,50) =12.81, p=0.0001; Tukey- Kramer post-hoc, p=0.0001). Markers include Pax6 (retinal progenitors), Rhodopsin (rod photoreceptors), RPE65 (retinal pigment epithelium), calbindin (horizontal/off-bipolar cells) and CRALBP (Müller glia). The protocol for investigating neurogenic potential of RSC progeny “primed” in T+RA during RSC sphere derivation is illustrated. Scale bars represent 50 μm. Mean ± s.e.m. of n=3 independent biological replicates.

(C) Clonal RSC spheres derived in FH are a mixture of pigmented RPE progenitors and non- pigmented neural retinal progenitors. When derived in T+RA, the spheres show less pigmentation (indicated by less absorption of YFP signal in Actin.yfp cells) but a similar size (i.e., cell number), suggesting an asymmetric shift towards production of neural retinal progeny during RSC division.

(D) Most RSC progeny treated with COCO throughout sphere growth and differentiation (and added to the 1% FBS+FH pan-retinal differentiation conditions) were positive for cone arrestin and S-opsin. In contrast, in pan-retinal differentiation conditions alone, RSC progeny produced<1% cones.

(E) Representative images of rhodopsin, S-opsin and cone arrestin (yellow arrows) expressing cells. Negative cells are shown with white arrows. The bisBenzimide H 33258, Hoechst stain, was used to visualize nuclei (blue) (39) (One-way ANOVA, F=10.23, p=0.0016). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

4.4.2 T+RA instruct RSC progeny to rods and COCO to cones

Two clonal approaches were used to investigate the possibility that T+RA+FH can encourage the specification of rod-specific progeny from RSCs. A fluorescent retroviral construct at limiting dilutions in vitro, allowed visualization of RSC progeny clones derived from single RSC- derived progenitors. Clonality was achieved using limiting dilutions to give a virus titer that infected<1 cell/well. Clone composition was divided into three categories: clones in which there were all Rhodopsin + cells (rod-only clones), mixed clones, and no Rhodopsin+ cells (non- rod clones). Comparisons of the frequency and character of clones fated for Rhodopsin-expression revealed enrichment in the percentage of rod- only clones between 1%FBS (13% of all clones) and T+RA

112

(over 70% of all clones), without any effects on the survival of clones, the numbers of cells per clone or the total numbers of cells per well (Fig. 4.2A and B and C) (> 40 clones in each condition, n=3 independent biological replicates). This strongly argues against selective survival effects on rod progenitors, rod precursors or differentiated rods within a clone as mechanisms for producing the increase in rod-only clones. Most interesting, when analyzing the composition of the individual mixed clones in both conditions (Fig. 4.3B and C), we found that mixed clones in T+RA+FH conditions were predominantly Rhodopsin+ (> 80% of cells in the clones), while those in 1%FBS+FH conditions were only 10–20% Rhodopsin+, consistent with those values seen at the population level (Ballios et al., 2012).

A second approach to analyze the cell biological mechanisms underlying rod specification made use of fluorescence-activated cell sorting (FACS) of dissociated, undifferentiated clonal RSC progeny. We sorted single non-pigmented or single pigmented retinal progenitors per well (see Methods), which were then treated with 1% FBS, T+RA, or T+RA+1%FBS for 28 days of differentiation (Fig. 4.2D). In pan-retinal differentiation conditions (1% FBS), clones derived from non-pigmented progenitors were distributed between non-rod and mixed clones, with a minority of rod-only clones (100% Rhodopsin-positive; n=4 of 28 clones) (Fig. 4.2D). Clones derived from pigmented cells in pan-retinal differentiation conditions never gave rise to rod-only clones, and only a minority of the clones were mixed. In T+RA conditions, all clones derived from non- pigmented progenitors (n=34) were rod-only clones (100% Rhodopsin-positive), while those derived from pigmented progenitors (n=47 of 48 clones) were almost all non-rod clones (Fig. 4.3A). Of note, one rod-only clone was derived from a single pigmented cell in T+RA conditions (the largest of the pigmented cell-derived clones in T+RA conditions with 20 cells), suggesting potential neural lineage plasticity in very early pigmented progenitors (which may explain the decreased pigmentation within T+RA treated clonal RSC spheres – Fig. 4.1C). Similar to T+RA treatment, in T+RA+1%FBS all clones derived from non-pigmented progenitors (n=34) were rod-only clones, while all those derived from pigmented progenitors (n=48) were no-rod clones (Fig. 4.3A). Clone sizes were similar among the groups, although the addition of FBS did produce a non-significant trend to bigger clones (Fig. 4.3A). Survival rates of single cell clones one day after plating were similar between non-pigmented single cell clones in T+RA (7.0%), 1%FBS (8.9%) and T+RA+1%FBS (10.2%); these single cells then proliferated to produce clones. These findings suggest that T+RA may instruct the neural retinal progeny of RSCs to become rod cells in vitro.

113

Next, we explored the cell biological mechanism underlying cone specification. We again sorted undifferentiated clonal RSC progeny into single non-pigmented or pigmented progenitors, plated them at single-cell-per-well density and differentiated the clones in COCO or pan-retinal control conditions for 28 days. We found that only clones from non-pigmented progenitors (all 47 clones) in COCO contained cone arrestin positive cells. Interestingly, smaller sized clones (< 100 cells, n=10 clones) were 100% cone arrestin positive. The larger clones were between 86%–96% cone arrestin positive (n=37) after 28-days of differentiation (Fig. 4.3D; only clones<600 cells are shown). All four groups exhibited similar average clone sizes, indicating that COCO does not cause differences in non-pigmented intra-clonal survival (Fig. 4.3E). Similar percentages of non-pigmented clones present at day 1 of plating survived to the end of the 28-day differentiation period in pan-retinal and COCO conditions, suggesting that COCO does not cause differences in inter-clonal survival (Fig. 4.3F). We hypothesize that during 28- day incubation with COCO, larger clones may still contain a majority of immature progenitors, which do not yet express cone arrestin, or alternatively larger clones might be earlier progenitors that already have committed their earliest progeny to other retinal cell fates such as retinal ganglion cells and thus COCO may not have inhibitory effects on those specific cells. These data suggest that COCO may act only on non-pigmented RSC progeny to specify the cone fate in vitro.

114

Fig. 4.2. Distribution of retinal progenitor clones in retroviral lineage tracing shows a bias towards mixed clones of high percentage rhodopsin+cells when differentiated in taurine and retinoic acid.

(A) Clonal lineage analysis was performed in vitro by infecting differentiating cultures with a limiting dilution of replication-incompetent GFP retrovirus, such that only one cell per well was labeled at day 1. The composition of clones was analyzed after 44 d of differentiation. There was a large enrichment of rod-only clones present in T+RA cultures compared to pan-retinal differentiation conditions (1%FBS+FH).

(B and C) The average clone size (t-test, p=0.31) and the absolute number of cells/well (t-test, p=0.24) in both conditions at 44 d of differentiation was similar. Means±SEMs of>40 clones in each condition across n=3 independent biological replicates.

115

(D) Single-cell per well analysis revealed that clones derived from non-pigmented progenitors in 1%FBS+FH were composed of mixed and no-rod clones, with a minority of rod-only (100% Rhodopsin+) clones. Those derived from pigmented cells were mostly no-rod clones, with only a minority of mixed clones. In T+RA, all clones derived from non-pigmented progenitors were rod- only clones, while those derived from pigmented progenitors were predominantly no-rod clones (no Rhodopsin+ cells).

4.4.3 Fetal and adult progenitors respond identically

To determine whether the rod lineage induction effects of T+RA observed with adult RSC- derived progenitors are applicable to the multipotent progenitor cells that build the retina during development, we isolated clonal RSCs from the presumptive ciliary margin of 14 day- old embryos (E14), as well as neural retinal progenitor cells from the developing retina. This early embryonic time point was chosen before rods are normally produced in the developing retina, to test the hypothesis that exogenous factors also could instruct early embryonic retinal progenitors born in vivo. Cultures of clonal non-pigmented E14 neural retinal progenitor cells or E14 RSCs in T+RA for 28 days resulted in rod differentiation to>90% of progeny, similar to adult RSC cultures (Fig. 4.4A). This reinforces the similarity between newborn progenitors from adult RSCs and early embryonic RSC-derived retinal progenitors. E14 RSC progeny was also subjected to a pulse of T+RA early in differentiation (a “lineage-priming” regime) similar to experiments performed on early adult RSC progeny (Fig. 4.1B). In keeping with these results, E14 RSC progeny showed more biased rod differentiation when primary cultures were primed with T+RA (over days 0–3 only of 7-day sphere growth) and then differentiated in 1% FBS+FH for 28 days (Fig. 4.4B) compared to cultures not primed before differentiation. This propensity for lineage priming suggests that early retinal progenitors from developing retina show a similar readiness for instruction to a rod fate as those progenitors derived from the early asymmetric divisions of adult RSCs.

A similar result was obtained when clonal neural retinal progenitor spheres derived from E14 embryos were exposed to COCO during differentiation. Over 90% of E14 neural retinal progenitor cell progeny expressed cone arrestin and S-opsin when treated with COCO (Fig. 4.4C and D). whereas <5% of neural retinal progeny expressed cone arrestin in pan-retinal conditions (Fig. 4.4C and D). These data suggest that the effects of COCO on adult RSC- derived progenitors are similar to those on embryonic neural retina progenitors. Taken together,

116 these data reinforce the similarity between newborn progenitors from adult RSCs and early embryonic retinal progenitors in culture conditions.

Fig. 4.3. Taurine and retinoic acid act instructively to generate rod progeny while COCO suppresses other retinal cell fates and permits only cone photoreceptor differentiation.

(A) The distribution of Rhodopsin+ cells in clones of varying size in each of the analyzed clonal growth conditions is shown. A small number of clones>1000 cells in size were excluded from this graph. These include 4 clones (> 1000 cells) from the 1%FBS+FH non-pigmented fraction (1 rod-only clone, 3 mixed with %Rhodopsin<50%), and 3 clones (> 1000 cells) from the 1%FBS+FH pigmented fraction (1 non-rod clone, and 3 mixed with % Rhodopsin<10%). The numbers placed above the counts represent the number of rod-only clones in that cluster derived from non-pigmented progenitors in T+RA (blue) and T+RA+1%FBS (red), and non- pigmented progenitors in 1%FBS+FH (purple). Clone sizes were similar among the groups, as two-way ANOVA revealed no significant main effects of differentiation condition or pigmentation on clone size, and no significant interaction effect (F(2,234)=2.41, p=0.09).

(B) Representative images of terminally differentiated clones labeled with GFP-retrovirus. Cells are stained for rhodopsin, showing rod-only, mixed and no-rod clones. Note, the morphology of

117

RSC-derived progeny in vitro is diverse and does not correlate with the expression of rhodopsin in these cells. Hoechst stain is used to visual nuclei (blue). Scale bars represent 100 μm.

(C) Mixed clones that underwent differentiation in pan-retinal conditions show a shift towards lower percentages of rhodopsin+ cells (multiple mixed clones are 10–20% rhodopsin+) compared to clones differentiated in T+RA (multiple mixed clones are>80% rhodopsin+). Data presented from>40 clones in each condition across n=3 biological experiments.

(D) When single cells were exposed to COCO for 28 days, the large majority of non-pigmented clonal progeny were cone arrestin positive (all of the clones of<75 cells were 100% cone arrestin positive). Pigmented progenitors did not produce any cones under COCO or pan-retinal conditions (One-way ANOVA, F=0.8287, Tukey post-hoc test, p < 0.0001).

(E) Clone size among the 4 different groups. Similar average clone sizes and thus survival between clones suggest that COCO may not cause differences in non-pigmented intra-clonal survival (F3, 8=0.828, p=0.5143).

(F) Similar percentages of non-pigmented clones present at day 1 of plating survived to the end of differentiation period in pan-retinal and COCO conditions suggesting that COCO does not cause differences in inter-clonal survival. Data represents means±SEMs across n=3 independent biological experiments (t-test, p=0.321). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

4.4.4 Sonic hedgehog regulates proliferation in the early progenitor expansion phase of RSC progeny differentiation

The sonic hedgehog (Shh) pathway increases the proliferation of retinal progenitors, while inactivation of Shh decreases the number of progenitors (Wall et al., 2009; Wang et al., 2005). We hypothesized that, similar to retinal development, Shh signaling would be critical in the early progenitor expansion phase of RSC differentiation. Cyclopamine was used to antagonize hedgehog signaling in differentiating cultures (Fig. 4.5A). Blockade of Shh reduced the expansion of RSC-derived progenitors (decreased absolute numbers of cells and Ki67 staining by 50%, Fig. 4.5B and C, n=3 independent biological replicates) specifically in the first two weeks in differentiation culture, with no effect on the percentages of cell phenotypes assayed at 2 or 6 weeks (Fig. 4.5D–G). This suggests that the composition of the progenitor pool, as well as rod fate specification, was unaffected. Taken together, these results demonstrate the double dissociation of two distinct processes: progenitor expansion by Shh and fate change by T+RA signaling.

118

Figure. 4.4. E14 RSC or neural retinal progenitor spheres exhibit similar differentiation and rod-lineage priming effects of taurine and retinoic acid, and E14 neural retina progenitor spheres show similar cone differentiation patterns in COCO to RSC progeny.

(A) Clonal E14 neural retinal progenitor cell (RPC) spheres were subjected to standard sphere growth conditions before differentiation for 28 days in pan-retinal or rod-inducing media. These neural retinal progenitors also showed enrichment for rod differentiation when subjected to T+RA compared to pan-retinal differentiation media. Two-way ANOVA revealed a significant interaction of cell type and differentiation conditions on Rhodopsin expression levels F (1, 17)=6.95, p=0.0003; Tukey-Kramer post-hoc, p=0.0001. Means±SEMs of n=3 independent biological experiments.

(B) RSC clonal spheres derived from E14 presumptive ciliary marginal zone epithelium were subjected to standard sphere growth conditions (0-7d FH) before differentiation for 28 days in either pan-retinal (1%FBS+FH) or rod-inducing (T+RA+FH) media. The enrichment of E14 RSC- derived rods demonstrates that T+RA also has a rod-inducing effect on progenitors derived from RSCs that are isolated during embryonic development from the growing eye. Cultures also were subjected to priming in T+RA (0–3d) at the start of clonal RSC sphere growth, and demonstrated rod enrichment in subsequent differentiation in pan-retinal conditions (1%FBS+FH), relative to RSC spheres that lacked priming. Two-way ANOVA revealed a significant interaction effect of sphere growth conditions (i.e., primed with T+RA, days 0–3 of initial sphere growth or not) and post-sphere differentiation conditions on percentages of cell expressing rhodopsin F (1,15)=7.33, p=0.0002; Tukey-Kramer post-hoc, p=0.0001. Data represent means±SEMs of n=5 independent experiments.

119

(C) Clonal non-pigmented spheres derived from E14 neural retina tissue were exposed to COCO or pan-retinal control differentiation conditions for 45 days. The vast majority of progeny in COCO were cone arrestin positive compared to 5% positive in pan-retinal conditions (t-test, p=0.0001).

(D) Neural retina spheres treated with COCO were largely positive for S-opsin, while<2% were positive in pan-retinal control (t-test, p=0.0001). Data represents means±SEMs (* p < 0.05).

4.4.5 T+ RA act on early retinal progenitors, while COCO is required throughout the differentiation

To confirm our hypothesis that the effect of T+RA on rod fate bias was specifically on early progenitors, we subjected adult RSC progeny to pulses of T+RA in culture at progressively later time points in differentiation (Fig. 4.6A). Pulses of T+RA on late progenitors (11–14 days of differentiation versus 7–10 days of differentiation) in culture showed progressively decreased bias for rod differentiation with the later pulses (Fig. 4.6B), suggesting T+RA does not act instructively on late retinal progenitors or precursors to bias rod differentiation. Taken together with our earlier T+RA priming experiments (Fig. 4.1B) these data strongly argue against the selective survival of a late progenitor or post- mitotic cells as a mechanism for rod-enrichment within clones.

To investigate the temporal window during which COCO influences cell fate during differentiation, we treated RSC derived spheres with or without COCO at different time points throughout the differentiation period (first two weeks of differentiation, last two weeks of differentiation or the entire period of differentiation). Greater than 50% of RSC progeny expressed cone arrestin and S-opsin only when cells were treated with COCO throughout the entire differentiation period (Fig. 4.6C). However, these cone markers were never detected when COCO was present only at the beginning or late in the differentiation period. Thus, COCO must be present throughout the 28-day differentiation period to bias cone production.

120

121

Fig. 4.5. Sonic hedgehog regulates proliferation in the early progenitor expansion phase of RSC progeny differentiation in T+RA.

(A) Cyclopamine (Cyclo) was used to antagonize sonic hedgehog (Shh) signaling in differentiating cultures of RSC progeny according to this experimental schema. The effect of hedgehog blockade was evaluated in terms of absolute numbers of cells/well, expression of mature and progenitor cell markers (Rhodopsin and Pax6, respectively), and cell proliferation markers (Ki67). The first two weeks of differentiation are marked by significant proliferation under the influence of 1%FBS. * Represents time points for experimental analysis.

Quantification of the absolute numbers of cells/well and the Rhodopsin expression in cultures at 14 (B and C and D) and 44 (C and E) days of differentiation. The overall proliferation in these cultures was decreased when cyclopamine was present in the first two weeks, specifically. This effect can be observed directly at 14 days in cultures treated with cyclopamine in (B), (Two-way ANOVA, main effect of sphere growth condition (F(3,45)=17.93, p=0.001), and main effect of cyclopamine (F(1,45)=14.91, p=0.0003), but no significant interaction effect (F(3,45)=0.62, p=0.09) on absolute numbers of cells/well) and C, (One-way ANOVA, shows an effect of condition (F(6,31)=30.08, p=0.002) on absolute numbers of cells/well; Bonferroni's post-hoc, p=0.02). (D and E) Treatment with cyclopamine had no significant effect on rod phenotype in the immature (14 days) and mature cultures (44 days) (ANOVA analysis).

(F and G) Quantification of Pax6 and Ki67 expression at 14 days showed a decrease in cell proliferation when cultures were treated with cyclopamine; in G Two-way ANOVA, main effect of cyclopamine, F (1, 23) =21.23, p=0.003. Means±SEMs of n=3 independent experiments.

4.4.6 RSC-derived cones exhibit similar transcriptomes to those of endogenous cones

To examine the overall gene expression similarities between adult RSC-derived cones and endogenous cones, we used adult CCDC136−/− mice, which express GFP specifically in cones and a sub-population of bipolar cells (Smiley et al., 2016). To isolate cones, we sorted both RSC-derived and endogenous cones for GFP and peanut agglutinin (a mature cone marker) (Fig. C2 B and C, see Materials and Methods) (Smiley et al., 2016). Principal component analysis of RNA sequencing data revealed very similar transcriptomes of the adult RSC-derived cones to their endogenous counterparts (the adult stem cell derived cone transcriptome samples encircle the endogenous cone gene expression samples - Fig. 4.6D). Analyses of gene pathway activities are thought to provide a more robust measure of correlation between samples in RNA sequencing experiments. Therefore, using GSEA (gene set enrichment analysis), we performed pathway analysis on our endogenous and RSC-derived cones and correlated this with sequencing data of endogenous photoreceptors from a separate and

122 independent reference database (Mo et al., 2016). These data show that both our endogenous and RSC-derived cones are highly correlated at the gene pathway level with the reference cones and not with reference rod photoreceptors (Fig. C3). 11 out of the top 20 differentially expressed genes were identical in endogenous and RSC-derived cones when compared to the starting RSC-derived sphere colonies, supporting the transcriptomic similarities between endogenous and RSC-derived cones (Tables C1 and C2). Additionally, the top 20 differentially expressed cone genes in Tables C1 and C2 had an average of 9-fold change (up-regulation), whereas the top 20 differentially expressed genes in Table C3 between endogenous and RSC- derived cones had an average of 4-fold change. Given the proposed similarities between endogenous and RSC derived cones, we would predict there to be smaller fold changes in their differential expression. The cone transcriptomes are clearly separated from the RSC-derived sphere transcriptomes (Fig. 4.6D). Moreover, the transcript of rod specific genes such as Rho, Nrl, Nr2E3, Rod arrestin, PDE beta and CNGB1 was not detected by RNA sequencing, suggesting specificity of cone photoreceptor enrichment during differentiation with COCO.

Due to sensitivity and specificity, we used qRT-PCR to validate the RNA Sequencing result. Retinal stem cell cone progeny showed high gene expression of Crx, cone arrestin (Arr 3), S- opsin and M-opsin only when cells were exposed to COCO throughout the entire differentiation period (Fig. 4.6E). Furthermore, Rho, Nrl, NR2E3 and GNGT1 (rod specific genes) were not detected in RSC-derived cones (Fig. 4.6E), while rod genes were present in T/RA induced rods (Ballios et al., 2012). Pax6 and Vsx2 were down regulated by the end of the differentiation periods, suggesting a loss of retinal multipotency (Fig. 4.6E). The effects of COCO on the differentiation of RSC progeny suggest that it act instructively to suppress alternate retinal fates throughout the differentiation period. Our results suggest that cones might be the default pathway for non-pigmented RSC progeny, consistent with previous studies (Brzezinski and Reh, 2015; Mears et al., 2001; Szél et al., 1994), and that the continued suppression of instructive signals for other non-cone retinal fates may allow the production of large cone-only clones with similar transcriptomes to those of endogenous cones.

Finally, we exposed non-pigmented RSC progeny to both T+RA and COCO simultaneously during differentiation (Fig. 4.6F). The vast majority of post-mitotic cells expressed rod rhodopsin rather than cone arrestin. This reinforces our observation that T+RA acts in an instructive manner to bias rod fate and does so despite the presence of COCO.

123

124

Fig. 4.6. Taurine/retinoic acid act instructively on early retinal progenitors to bias rod differentiation, while COCO is required throughout the differentiation period in order for RSC progeny to develop cone phenotypes with similar transcriptomes to endogenous cones. Rod differentiation outcompetes cone differentiation.

(A) To test the effect of late pulses of T+RA exposure on rod lineage priming in RSC-derived progenitors, the schematic shows RSC-derived progenitors treated in pan-retinal conditions (1%FBS+FH) with 4 days of pulses of T+RA at 7–10 days and 11–14 days of differentiation.

(B) T+RA is unable to bias rod fate when pulsed at later time points, demonstrating that the effect of T+RA on lineage priming of rod-restricted progenitors is likely an effect on relatively early progenitors, Means±SEMs of n > 6 wells in each condition across n=3 independent biological replicates (t-test, p=0.001).

(C) RSC progeny expressed cone arrestin and S-opsin only when cells were treated with COCO throughout the entire differentiation period (One-way ANOVA p < 0.0001).

(D) Principal component analyses (PCA) of whole transcriptome data from undifferentiated clonal RSC colonies, RSC-derived cones and endogenous cones. The transcriptomes of adult RSC-derived cones clustered around their endogenous counterparts, and are distinct from the RSC-derived sphere transcriptomes.

(E) RNA expression revealed that Crx (t-test, p=0.003) and cone arrestin (Arr3) expression increased significantly (t-test, p=0.0027) when RSC progeny were treated with COCO throughout the entire differentiation period compared to pan-retinal differentiation conditions. The expression of M and S opsin markedly increased under COCO differentiation conditions (t- test, p=0.007). On the other hand, Pax6 and Vsx2, retinal progenitor genes, were decreased at end of differentiation. Data represent means±SEMs of n=3 independent biological replicates. (* p < 0.05).

(F) Non-pigmented RSC progeny were differentiated in pan-retinal conditions, COCO, or a combination of COCO and T+RA simultaneously for 45 days. Cells exposed to COCO alone expressed cone arrestin. On the other hand, cells differentiated under the combination of COCO + T+RA expressed rhodopsin. We suggest that COCO may serve to instruct a cone fate by suppressing instructive signals for other non-cone retinal fates, and is overwhelmed by the additional presence of exogenous T+RA. However, it also may be that T+RA act intracellularly, and thus the effect of T+RA on promoting rod fate in the presence of COCO inhibition may simply be the result of T+RA action downstream of COCO inhibition at the cell membrane receptor. Nevertheless, these findings emphasize that COCO must be active continuously to block other extracellular instructive differentiation signals.

125

4.5 Discussion

It may be informative to conceptualize rod and cone differentiation as possessing both “mechanisms” (instructive or permissive) and “modes” (lineage-restricted or lineage independent). A priori, three alternative models might explain how directed RSC progeny differentiation can produce rod or cone-enriched cultures (Fig. 4.7). Our results with clonal analyses of rod and cone differentiation after T+RA or COCO, respectively, argue for an instructive, rather than a permissive (i.e., selective survival) mechanism of differentiation, as the overall absolute cell survival as well as inter- and intra-clone survival did not vary among differentiation conditions. This makes the permissive Model 1 unlikely. Model 2 suggests that T+RA and COCO instruct the fate of multipotent progenitors at every division to produce rod or cone precursors, respectively. Indeed, this mechanism is consistent with the finding that COCO must be present throughout the differentiation period in a lineage independent fashion to inhibit more instructive signals from biasing differentiation towards alternative non-cone retinal fates and allowing the default to a cone cell fate (and thus support Model 2 for cone differentiation in COCO).

126

Figure. 4.7. Alternative models of directed photoreceptor differentiation from RSC progeny.

Model 1 represents a permissive mechanism and lineage-restricted mode of differentiation; Model 2 represents an instructive mechanism and lineage-independent mode of differentiation; while Model 3 represents an instructive mechanism and lineage-restricted mode of differentiation. Our results with retroviral clonal analysis and single cell per well analyses support an instructive, rather than a permissive model for rod differentiation. Our data fit Model 3 suggesting an early effect of T+RA in instructing the production of rod-fate-restricted proliferative progenitors, which then divide symmetrically to produce rod-only clones. In the case of cones, our clonal analyses show that there is no difference in survival between non- pigmented progenitors in COCO or pan-retinal conditions, suggesting that COCO does not cause differences in inter- or intra-clonal survival. Moreover, COCO must be continuously present throughout the differentiation period, suggesting that it may act to promote the cone fate of the non-pigmented progeny through suppression of alternate retinal fates. These findings support an instructive mechanism and lineage-independent mode of differentiation for cone photoreceptors (Model 2).

However, the results of the lineage priming experiments on rod differentiation suggest that the effect of T+RA is on early, not late, progenitors or precursors. If multipotent progenitors were maintained throughout T+RA-induced differentiation, then we would expect clones with at least a small number of Rhodopsin-negative cells in the single-cell-per-well differentiation assays. These results make Model 2 unlikely for T+RA induced rod differentiation. Our data satisfy all of the criteria for Model 3, suggesting an early effect of T+RA in instructing the rod fate in early retinal progenitor cells, which produce rod-only clones. The single large clone consisting of all rod photoreceptors that came from a single pigmented cell may indicate an instructive effect on retinal stem cells themselves. However, the lack of a remaining undifferentiated cell (the stem cell) in this clone and the low frequency of RSCs compared to RPE progenitors might suggest instead that an early RPE progenitor was instructed to produce a large rod-only clone in T+RA.

Our results indicate that exogenous factors can influence retinal progenitor fate restriction and, moreover, that rod and cone lineage- restricted cells may exist during retinal stem cell differentiation in vivo. Moreover, there is a striking similarity between adult RSC-derived photoreceptor differentiation, and photoreceptor differentiation from embryonic neural retinal progenitor cells in vitro, in terms of their differentiation potential (Altshuler et al., 1993; Kelley et al., 1994) and factors governing progenitor proliferation (Wang et al., 2005). The early in vivo birth of cone photoreceptors compared to other retinal cell types might be explained by a default

127 of some retinal progenitors to cone fate before the instructive signals for other retinal lineages are turned on (Akimoto et al., 2006; Brzezinski and Reh, 2015; Mears et al., 2001).

Much of our analyses of cell type composition in varying differentiation conditions utilize immunofluorescence and qPCR-based methods, which limit one to a small number of specific retinal markers. Therefore, we carried out RNA-sequencing to allow for a more robust and complete evaluation of how transcriptionally similar the cones we make from stem cells are to their endogenous counterparts in the mature retina. These data show that, at both the level of individual genes and genetic pathways, the RSC-derived cones are highly correlated with cones isolated from the mature mammalian retina (Fig 4.6D, Fig C3). Furthermore, through comparison to the initiating and undifferentiated retinal progenitors, RNA-seq may enable us to identify novel markers involved in cone development and function. It is noteworthy that of the top 20 differentially expressed genes (compared to RSC spheres), 11 are shared between RSC- derived and endogenous cones. While none of these appear known to be involved in cone development or function, the large overlap warrants further investigation. Furthermore, the sequencing data from cones encourages similar analysis on rods derived from RSCs.

Overall, these results also contribute to the resolution of conflicting models of cellular determination in the retina, by showing that there may exist numerous pre-programmed (lineage-restricted) progenitors for various retinal lineages. In this study, we demonstrate instructive environmental cues influencing specific fates among retinal progenitors. Our results are consistent with the model developed by (Cepko et al., 1996), as we demonstrate that it is early, rather than late, progenitors in vitro that are able to respond to environmental cues to undergo lineage restriction to rod-specific retinal progenitors (Cepko et al., 1996). Taken together, analyses of RSC and neural retinal progenitor differentiation represent a tractable system for studying the response of retinal progenitors to environmental cues at the clonal level, and the molecular mechanisms by which uncommitted progenitors make the decisions between proliferation, survival, and fate selection.

128

Chapter 5

Discussion

5.1 Summary & Conclusions

In Chapter 2, I investigated whether antagonism of signaling proteins posited to mediate endogenous RSC quiescence would enable activation of RSC proliferation, migration and differentiation in vivo. This approach had not been attempted previously as very little was known about the identity of inhibitory factors in the RSC niche until recent work in our lab implicated BMP and sFRP2 proteins. I demonstrated that BMP and sFRP2 antagonists, or small molecules that modulate BMP and Wnt signaling, can stimulate CE cell proliferation in vivo. Furthermore, I found BMP or sFRP2 antagonism expands the RSC population in vivo. These results represent the first evidence that BMP and sFRP2 regulate endogenous CE/RSC proliferation and endogenous RSC self-renewal. Further, I showed that BMP/sFRP2 antagonism enhances the stimulation of CE proliferation by exogenous growth factors (GFs). Yet, because the effect of GF stimulation and BMP/sFRP2 antagonism was roughly additive, the result can be interpreted in two ways: 1) endogenous BMP and sFPR2 normally prohibit GFs from stimulating CE proliferation to their highest potential, or; 2) blocking endogenous BMP and sFRP2 induces positive stimulation of CE proliferation and, when combined with GFs, there is an additive positive effect on CE proliferation. My initial hypothesis favours the former interpretation, but this is yet to be resolved. Notably, I found that GFs and BMP/sFRP2 antagonism combined (called ‘FINS’) does not result in RSC expansion. Thus, if RSCs are among the proliferating CE population as I predict, they appear to be undergoing only asymmetric divisions in response to the combined FINS factors. Using a chemically induced retinal degeneration paradigm, I found that retinal injury alone does not induce significant CE cell proliferation. However, combined injury and FINS treatment demonstrated that injury potentiates the effects of FINS and increases total CE proliferation. Thus, it appears CE cells/RSCs become responsive to damage when stimulated by FINS treatment. In contrast, CE lineage tracing revealed that injury alone promotes cell migration from the CE into the retina and that combined injury and FINS treatment does not augment cell migration beyond the effects of FINS alone. Thus, it appears the mechanisms mediating cell migration in response to FINS treatment and retinal injury are not additive. Strikingly, lineage tracing also revealed that a significant proportion of CE cells that

129 migrate into the retina come to express markers of mature photoreceptors or RGCs. This provides evidence for latent neurogenic and regenerative potential of the adult CE and/or the RSCs therein. Overall, this study demonstrates for the first time that BMP and sFRP2 proteins are key regulators of RSC quiescence in the adult mouse eye and that modulating inhibitory signals in the RSC niche can reactivate the proliferative and neurogenic potential of the adult mammalian CE. Further, this work may be a first step towards an endogenous regeneration strategy to treat retinal injury and disease.

In Chapter 3, I developed a medium-throughput phenotypic screening platform to identify small molecules for their ability to influence retinal stem/progenitor cell proliferation. By using an unbiased screening approach, along with small molecules with well characterized molecular targets, I sought to discover molecular pathways that were previously unknown to regulate RSC function. Using this approach, I identified high-affinity synthetic glucocorticoid (GC) agonists as the primary hit compound class, which implicated GC signaling as a novel positive regulator of RSPC proliferation. Assays to assess the cell biological mechanisms influenced by GC agonism confirmed that it enhances RSPC proliferation, revealed that it increases RSC symmetric self- renewal divisions, and showed that it has no influence on cell death/survival. Furthermore, pathway-specific antagonists indicated that both GC receptor and mineralocorticoid (MC) receptor signaling mediated the proliferative effects of GC agonism in mouse RSPCs. However, this appears to be species-specific, as only MC receptor antagonism attenuated the proliferative effect of GC agonism in human RSPCs. Thus, GC receptor signaling is either not active or not involved in regulating proliferation in human RSPCs under GC agonism at the dose tested. Likewise, the influence of GC signaling on progenitor proliferation appears to be cell-type- specific, as it suppressed the proliferation of pancreatic multipotent progenitor (PMP) cells. Interestingly, differentiation assays with RSPCs showed no effect of GC agonism on retinal cell type specification. As proliferation and differentiation often occur at one another’s expense in stem and progenitor cells, this was a noteworthy finding. Another notable finding was that intravitreal injection of the synthetic GC agonist dexamethasone into the adult mouse eye induced proliferation of CE cells in vivo. Very few cells in the CE were affected, yet, given that RSCs are also rare cells in the CE, this may not be unexpected if GC agonism stimulates only RSCs. However, if RSCs are stimulated to proliferate in vivo by GC agonism, they are likely dividing asymmetrically, as no expansion in RSC number following injection was evident. Altogether, this study reveals the previously unknown mechanistic and functional role of

130 glucocorticoid signaling in retinal stem and progenitor cell proliferation, which may have implications for retinal cell transplant therapies or even endogenous activation of RSCs.

In chapter 4, I investigated whether TGFβ, BMP and Wnt antagonism could direct the differentiation of RSC progeny to a cone photoreceptor fate and the cell biological mechanisms involved in cone specification. This was predicated on evidence that the cone fate is a default differentiation pathway of retinal progenitors and precursors, and in particular, on evidence that COCO, a Cerberus/Dan family protein that inhibits BMP, Nodal/TGFβ and Wnt signaling pathways, can direct hESCs to a cone fate. Using whole RSC sphere differentiation assays, we found that COCO increases the proportion of cone progeny from <1% to ~60% without affecting cell survival, indicating inhibition of TGFβ, BMP and Wnt does influence the specification of RSC progeny toward cones. Using FACS and subsequent clonal assays, we showed that individual non-pigmented NR progenitors can generate ~85-100% cone progeny. In contrast, pigmented, presumptive RPE progenitors were incapable of producing cone progeny. This indicates fate restriction in these two classes of early RSC-derived progenitors and likely explains why mixed progenitor spheres do not exceed 60% cone progeny. In support of this interpretation, embryonic RPCs isolated from the E14 NR differentiate into over 90% cone progeny in the presence of COCO. This result suggests that embryonic RPCs and RSC-derived NR progenitors have similar differentiation potentials and respond to the same environmental cues with the same fate choice (which is further corroborated by similar responses to rod instructive factors in RPCs and RSC progenitors). Another important finding was that COCO must be present during the entire differentiation period to specify the cone fate. Thus, if TGFβ, BMP, and Wnt are not suppressed in early and late phases of differentiation, RSC progeny can be instructed into non-cone fates by those signals during either period. This indicates that cone specification is not due to lineage restriction of progenitor potency but rather inhibition of positive signals that instruct other retinal cell fates. Combining COCO and small molecules instructive for rod photoreceptors supports this interpretation as ~90% of cells expressed rod markers and did not express cone markers. Additionally, we compared the gene expression profiles of RSC-derived cones and endogenous cones. RNA sequencing revealed both cone types had very similar transcriptomes and are highly correlated at the gene pathway level. Further, RSC-derived cones do not express rod genes and are genetically very different from RSC sphere progenitors. These data indicate that exogenous signals play critical roles in directing lineage decisions during the differentiation of RSC progeny and indicate cones may be the default pathway for non-pigmented RSC-derived progenitors and precursors, consistent with

131 previous studies on RPCs. These results also contribute to the resolution of conflicting models of cellular determination in the retina, by showing that there may exist both lineage-restricted and lineage-independent progenitors that produce various retinal cell types. Also, by characterizing a method to generate a high yield of cone photoreceptors from RSCs in vitro, these findings may have translational potential for cell therapy.

All together, this thesis work shows for the first time that endogenous RSC can be activated by the removal of in vivo inhibitory signals, establishes that glucocorticoid signaling can enhance RSC proliferation and self-renewal, and demonstrates that RSCs can be instructed to the cone fate by blocking exogenous differentiation signals.

5.2 Discussion and Future Directions

5.2.1 Chapter 2

5.2.1.1 Downstream mechanisms of BMP & sFRP2 antagonism

In Chapter 2, I discussed the potential signaling pathways that could be mediating RSC activation/proliferation by the various injected factors tested. Not discussed in Chapter 2 are the methods to assess those mechanisms. So, far I have attempted three different methods to examine the molecular pathways mediating CE/RSC proliferation and RSC expansion following intravitreal injection of Noggin and/or anti-sFRP2. I have used IHC to examine downstream proteins that mediate BMP and Wnt signaling, e.g. phospho-SMAD1/5/8 (BMP signaling) and nuclear translocation of β-catenin (Wnt signaling). I have used in situ hybridization (ISH) to examine expression and localization of various stem cell signaling pathway target genes. And I have isolated the CE post-injection and run qPCR on a panel of target genes for stem cell signaling pathways and cell proliferation. Unfortunately, none of these methods has been able to clearly demonstrate changes in signaling pathways in response to intravitreal injections. The rarity of RSCs within the CE may be a general limitation as to why these approaches were not definitive. For instance, bulk qPCR of the whole CE is likely not sensitive to changes in gene expression in such a small subset of cells. Likewise, IHC and ISH entails looking for changes in rare cells within a whole tissue section, which is difficult with no markers to distinguish RSCs

132 from the rest of the CE. I attempted to multiplex IHC and ISH with EdU labeling to more selectively examine proliferating cells in the CE, but I was unable to find consistent evidence of alterations to BMP and Wnt signaling, nor co-localization with up-regulated stem cell signaling target genes. Further, the need for additional markers to ensure EdU-labeled cells were not endothelial cells or microglia made interpretation more difficult. In addition, it’s possible this strategy would only be successful for a short window of time, when molecular signaling changes induced by the injected factors would be still underway, but sufficient time passed for CE/RSCs to enter the cell cycle and incorporate EdU. Thus, direct evidence that BMP and Wnt signaling are modulated in response to BMP/sFRP2 antagonism in vivo remains to be established. So far, the efficacy of small molecules, CHIR99021 (GSK3β inhibitor/Wnt activator) and LDN-193189 (BMP inhibitor), to induce CE proliferation provides some confirmatory evidence that BMP and Wnt signaling regulate in vivo CE/RSC proliferation. Yet, whether these molecules work additively or induce RSC expansion remains to be tested.

Approaches to more directly evaluate the effects of BMP and Wnt signaling in the CE/RSCs can be pursued. For example, transgenic mice with reporter constructs for Wnt or BMP signaling exist and could provide direct readouts for each pathway following intravitreal injections (Currier et al., 2010; Doan et al., 2012). Likewise, there are transgenic mouse strains in which one can inducibly knockout or activate Wnt or BMP signaling (Aoki and Taketo, 2008; Yang and Mishina, 2019). These mice could be used to observe direct effects of genetically modulating signaling and examine resultant changes in RSC function, as well as changes in other signaling pathways (likely via scRNA-seq, as described in the next paragraph). Alternatively, vectors carrying BMP/Wnt reporter constructs, overexpression constructs, or RNAi constructs, could be delivered via electroporation or intravitreal injections (Nickerson et al., 2012).

An especially informative strategy to evaluate CE/RSC proliferation in vivo would be to cross the R26-FUCCI 2aR mouse with the Msx1-CreERT2 mouse and then perform intravitreal injections or direct modulation of BMP/Wnt signaling in those mice. As elaborated below, one would then be able to specifically isolate proliferating CE/RSCs via FACS and assess transcriptional pathway changes via single cell RNA sequencing (scRNA-seq). The fluorescent ubiquitination-based cell cycle indicator (FUCCI) system enables monitoring of cell cycle phasing in live cells by labeling two components of the DNA replication control system which oscillate inversely – Ctd1 and Geminin. Cdt1 peaks in G1 phase just before the onset of DNA replication and declines abruptly after the initiation of S phase. Geminin levels are high during S and G2 phase, but low during late mitosis and G1 phase (Zielke and Edgar, 2015). The R26-FUCCI 2aR mouse enables

133 conditional expression of both FUCCI probes after excision of a Cre-removable stop cassette in vivo (Mort et al., 2014). Therefore, in Msx1-CreERT2;R26-FUCCI 2aR mice, only Msx1- expressing cells (i.e. CE cells) would express the FUCCI reporters and allow live, real-time cell cycle tracking. The FUCCI+ and FUCCI- CE cells could then be sorted by FACS and undergo scRNA-seq. I believe this experiment could lead to 3 key insights: 1) One could determine in vivo proliferation dynamics of activated RSCs by assessing FUCCI reporter expression via FACS analysis and corresponding cell cycle-related gene expression via scRNA-seq. 2) Comparing differentially expressed genes between the FUCCI+ and FUCCI- CE cells (presumably active RSCs vs regular CE/inactive RSCs, respectively) could reveal the molecular signature of endogenous RSCs and identify novel, unique markers. 3) The most up- and downregulated genes and subsequent pathway analyses of the FUCCI+ cell transcriptomes would indicate which molecular pathways are being modulated and are likely mediating endogenous RSC proliferation.

5.2.1.2 Does Msx1 Regulate RSC Quiescence?

In Chapter 2, I found that Msx1 is not only expressed in the pigmented CE but that only Msx1- labeled pigmented CE cells give rise to clonal RSC spheres, indicating all RSCs express Msx1. This is consistent with recent experiments in our lab using microfluidic sorting of primary CE to enrich for RSCs followed by RNA-seq (unpublished, led by Brenda Coles). This data indicates there may be two populations of endogenous RSCs (quiescent & primed to proliferate) and both were found to express Msx1. Likewise, our RNA-seq data from Chapter 4 shows that Msx1 is expressed in RSC spheres (data not shown). Msx1 is expressed as early as E12.5 in the distal tip of the eyecup where it distinguishes the ciliary margin from the NR (Martinez-Morales et al., 2001; Monaghan et al., 1991) and has recently been shown to label embryonic CE stem/progenitor cells that contribute to NR development (Bélanger et al., 2017). However, Bélanger et al. did not investigate if loss-of-function might affect the proliferation or neurogenic potential of embryonic CE progenitors, and furthermore, Msx1-null mice have not been reported to have developmental abnormalities in the CB (Satokata and Maas, 1994). Thus, whether Msx1 might have a functional role in RSCs is still uncertain. However, Msx1 functionally regulates many other adult stem cells. For instance, Msx1 has been shown to promote digit tip regeneration by mesenchymal stem cells (Taghiyar et al., 2017), promote Xenopus tadpole tail regeneration (including spinal cord and muscle) (Beck et al., 2003), regulate differentiation of

134 neural crest stem cells (Bhatt et al., 2013) and maintain muscle stem cells in an undifferentiated state (and even de-differentiate committed muscle progenitors) (Ding et al., 2017). Further, RNA profiling has shown that Msx1 is highly expressed in quiescent brain NSCs (Llorens-Bobadilla et al., 2015) and in putative quiescent spinal cord NSCs (Ghazale et al., 2019). Consistent with that data, the Drosophila Msx1/2/3 ortholog (known as muscle segment homeobox (Msh) or Drop/Dr in flies) was recently proposed to be a crucial regulator of Drosophila NSC quiescence during development (Otsuki and Brand, 2019). Moreover, Msx1 is a target of both BMP and Wnt signaling (Timmer et al., 2002; Willert et al., 2002) and both pathways have been shown to induce Msx1 expression in the developing ciliary margin (Liu et al., 2007; Zhao et al., 2002a). Indeed, it has been proposed that BMPs and Wnts cooperate to pattern the CMZ by promoting Msx1 expression, which in turn represses the neurogenic potential of CMZ progenitors and promotes CE fate (Liu et al., 2007). A similar Wnt/BMP-Msx1 signaling axis has been proposed to regulate proliferation and differentiation of dorsal spinal cord epithelial cells in mice (Ille et al., 2007) and in neuroblastoma cells (Szemes et al., 2020). Therefore, it is not only possible that Msx1 has a functional role in RSCs but may integrate and mediate the effects of BMP and Wnt signaling on RSC quiescence/proliferation and differentiation.

To determine if Msx1 has direct functional influence on RSCs, both in vitro and in vivo experiments using inducible knockdown/knockout or overexpression of Msx1 followed by assessment of RSC self-renewal, proliferation and differentiation could be carried out. If Msx1 has functional effects, profiling the subsequent transcriptional and epigenetic changes would also be of interest. To determine if Msx1 is modulated in response to BMP/Wnt signaling, Msx1 expression levels in RSPCs following in vitro inhibition or activation of the BMP or Wnt pathways can be assessed. Further, if the Msx1-CreERT2;R26-FUCCI 2aR mouse experiments proposed in the preceding section (5.2.1.1) are carried out, any changes in Msx1 gene expression in endogenous RSCs in response to BMP/Wnt modulation would be detected. Conversely, it may be the case Msx1 regulates the effects of BMP or Wnt signaling on RSCs. Knockdown/knockout or overexpression of Msx1 followed by modulation of BMP/Wnt signaling in RSPCs or in the CE could reveal such an interaction.

135

5.2.1.3 Assessing Migration & Differentiation of CE-derived Cells in the Retina

In Chapter 2, Msx1 lineage labeled cells in the NR were found to co-express either an RGC marker or photoreceptor marker, but this does not exclude the possibility that other cell types are also generated. Likewise, further characterization of these apparent newborn neurons beyond a single marker is warranted to verify their classification as such. Only a small number of Msx1-labeled cells (often one or none) was present in any given slide. This made quantification of a whole panel of NR cell markers, or multiple markers for a single cell type, across all the conditions challenging. Thus, more robust methods to measure the degree of endogenous CE/RSC activation, migration into the NR, and retinal differentiation achieved by the current activation paradigm -- as well as future paradigms that aim to enhance these outcomes -- are needed.

One method to assess global cell migration is to use tissue clarification to render the whole, intact eye transparent, and then use lightsheet fluorescence microscopy to construct a 3D rendering of the whole eye. This method can be used to observe fluorescent reporters and IHC labeling throughout an entire intact tissue (Richardson and Lichtman, 2015). Thus, instead of examining 10µm-thick sections for proliferating cells in the CE or Msx1 lineage-labeled cells in the NR, the entire CE and NR of the eye can be assessed at once for those markers, with both cell and tissue morphology intact. This possibility has already been discussed with our collaborators in Valerie Wallace’s lab, who have been refining clarification and imaging protocols for mouse eyes. The Wallace lab has even developed a bleaching protocol to remove pigmentation so mice on a pigmented background can be imaged, removing limitations on which genetic mouse strains can be used.

To gain both in-depth and holistic insight into the molecular identities of the Msx1-labeled cells present in the NR, once again, scRNA-seq may be the optimal approach. For example, following our intravitreal injection paradigm in Msx1-CreERT2;R26-tdTomato mice, the entire NR could be dissociated and all labeled cells resident in the NR could be sorted via FACS and then sequenced. Though we would lose data about morphology and location, we would gain incredible insight into each individual cell’s gene expression profile, which could inform not only what cell types are present but even their developmental maturity/lineage based on pseudotime analyses.

136

5.2.1.4 Other Variables and Open Questions Toward Activating Endogenous RSCs

As discussed in Chapter 2.5, there are a number of experimental variables in the endogenous RSC activation experiments that leave open questions for future consideration. For instance, these experiments involved the intravitreal injection of proteins, antibodies and small molecules, all of which are likely to have inherent differences in pharmacokinetics and clearance from the vitreous. Thus, the controlled and sustained delivery of factors, and the measurement of their clearance from the eye, would likely improve experimental consistency and efficacy going forward. Toward this aim, a collaboration with Molly Shoichet’s lab is underway that aims to employ hydrogels and microbeads to tune the controlled release and pharmacodynamics of individual and combined factors for intravitreal delivery.

Another variable that might influence endogenous RSC activation is the severity and type of injury used. There are many models of induced and genetic retinal injury that operate via distinct mechanisms and affect different cell types (Collin et al., 2020; Wohl et al., 2012). For example, genetic mutations that lead to photoreceptor degeneration do so by causing dysregulation in many biological processes, such as visual transduction, the visual/retinoid cycle, ciliary function and trafficking, metabolism, synapse formation, adhesion molecules, membrane channels, signaling, transcription factors, DNA repair, and so on. Further, these different mutations lead to broad differences in the onset and rate of cell loss (Collin et al., 2020). Likewise, there are many forms of induced injury, including retinal ischemia, optic nerve crush, excitotoxic injury, neurotoxic/chemical injury, light/laser injury, ocular hypertension, mechanical injury, etc. Some examples of cell-type-specific effects of injury include: colchicine, which causes very selective cell death, only killing RGCs; excitotoxic injury via NMDA or kainite, which induces RGC, bipolar, amacrine, photoreceptor and glial cell death (Wohl et al., 2012); and chemical injury via sodium iodate, which causes RPE cell death and subsequent photoreceptor degeneration (Carido et al., 2014). Each of these degeneration paradigms may have profoundly different effects on the RSC niche microenvironment and therefore RSC functional responses (or lack thereof). In Chapter 2, I employed a single dose of MNU (45mg/kg) that caused near complete photoreceptor-specific degeneration over a period of 2 weeks. Even increasing the severity of this injury with a higher dose could have led to different results. Further, the timing of intravitreal injection following injury could also impact whether there are any protective effects of the

137 injected factors against retinal degeneration, as well as influence the degree or timing of RSC activation, expansion, migration and differentiation outcomes.

Another question that warrants investigation is whether MNU injury influences the symmetric expansion of RSCs in vivo. This could be determined using sphere assays following MNU injury – with and without Noggin, anti-sFRP2, NS, FI or FINS treatment – to assay for an expansion in primary RSCs. In particular, it would be interesting to examine if Noggin or anti-sFRP2 would impact injury-mediated RSC expansion since they already increase RSC symmetric division. Likewise, only the combined FINS factors were tested in conjunction with injury when evaluating CE proliferation, migration and differentiation in vivo. Thus, it would be informative to test noggin and/or anti-sFRP2 injection without growth factor stimulation to determine if BMP/sFRP2 antagonism (or direct genetic modulation of BMP and Wnt signaling) is sufficient to mediate CE migration and neurogenesis.

5.2.2 Chapter 3

5.2.2.1 Glucocorticoid Mechanisms and Molecular Pathways

As discussed in Chapter 3, GC signaling is mediated through the glucocorticoid receptor (GR) or mineralocorticoid receptor (MR). Whether one or both receptors is activated depends on many factors, such as the species, sex, cell-type, GC agonist activity and concentration (Ramamoorthy and Cidlowski, 2013). Although which receptor(s) facilitated the proliferative effects of GC agonists on RSPCs was examined, the downstream signaling mechanisms that mediate their effects were not. Glucocorticoid signaling has been shown to regulate several molecular signaling pathways known to influence adult RSPCs and RPCs during retinal development, including Wnt signaling, Notch signaling, BMP signaling and Hedgehog signaling (Anacker et al., 2013). Also, the highest concentration of GC agonists tested in RSPC proliferation/self-renewal assays was 10µM, while only one concentration of dexamethasone (1µM) was tested during RSC progeny differentiation. Thus, further characterization is warranted to determine: 1) if broader concentrations of GC agonists might have different influences on RSPC proliferation, self-renewal and differentiation and, 2) what molecular pathways are mediating the cell biological effects of GCs on those processes. To determine the latter, gene expression data (via microarrays or RNA sequencing) and subsequent pathway

138 analysis of RSPCs exposed to various GC concentrations could establish what pathways are up- or down-regulated dynamically in response to GC agonism. Once individual pathways have been implicated, their direct effects on RSPC function can be tested.

So far, our data indicates that GC agonism enhances RSPC proliferation and self-renewal without influencing differentiation. Therefore, in addition to molecular signaling pathways, how GCs modulate genes related to cell cycle progression would also be valuable to glean from gene expression profiling. Likewise, determining the influence of GCs on RSPC cell cycle kinetics will more precisely resolve how GCs enhance proliferation. This could be done using nucleoside pulse-chase experiments, cell DNA content analysis via FACS, or reporter assays such as the FUCCI cell cycle sensor system.

5.2.2.2 Investigating Other Hit Compounds and Future Screens

Investigating other hits from the OICR library screens in Chapter 3 or using our established screening pipeline to screen other libraries may reveal further novel regulators of RSPCs. Though the other hit compounds in our screens did not appear to have the consistency or potency of GCs, they may still be worth further investigation. Some of the molecular targets of the hit compounds included signaling pathways known to regulate various stem and progenitor cell types. For example, we identified as a hit the Rho/ROCK inhibitor, Thiazovivin. As mentioned in Chapter 1.4.3, Del Debbio et al. (2014) reported that Rho GTPase inhibitors can induce proliferation in the mouse CE in vivo. This points to the likelihood that further in vitro assays with this compound class might lead to insights into how they regulate RSPC proliferation; and it may have led us to independently arrive at testing Rho GTPase inhibitors in vivo. Another interesting hit compound is the dipeptidyl peptidase IV (DPP-4) inhibitor, MK- 0431. DPP-4 is believed to cleave SDF-1 and thereby impair its capacity to trigger CXCR4 signaling (Fadini and Avogaro, 2013). This signaling axis is well characterized for its role in facilitating hematopoietic stem cell homing. However, CXCR4 expression is maintained by Rx in the early eye field and is crucial for adhesion and morphogenetic movements of cells in the eye field (Bielen and Houart, 2012). Therefore, inhibiting DPP-4 could regulate similar properties within RSPCs in a manner that augments their proliferative potential. Additionally, the histone deacetylase (HDAC) inhibitor, sodium butyrate, was identified as a hit in our screen. More recent experiments in our lab (led by Brenda Coles) using microfluidic sorting to enrich for RSCs

139 followed by scRNA-seq have found that HDAC is highly expressed in quiescent RSCs but not RSCs that are “primed” to proliferate (unpublished). Also, siRNA knockdown of a specific Hdac resulted in an increase in RSC spheres from the more quiescent population. Thus, inhibition of HDACs, which are known to mediate transcriptional repression and regulate cell cycle progression/differentiation in retinal cells (Zhang et al., 2015), could potentially be promoting RSPC proliferation by modulating epigenetic marks that maintain RSCs in quiescence. Follow- up ATAC-seq experiments to investigate this hypothesis could reveal epigenetic loci and genes important in switching RSCs between quiescence and proliferation.

Moreover, having developed an effective drug screening platform for identifying RSPC proliferation enhancing molecules, further screening with other drug libraries to identify more as- of-yet unknown regulators of RSCs could be performed. In fact, I have been involved in an academic-industry partnership with a biotechnology company where my screening platform has been used to successfully identify, and iterate on, proprietary compounds that may have application for stimulating RSC proliferation in vivo.

5.2.2.3 Effect of Glucocorticoids on Human RSCs/Ciliary Epithelium

In Chapter 3, I found that dexamethasone increased the proliferation of adult human RSPCs in culture and there appeared to be a modest proliferation effect of dexamethasone in vivo in the mouse CE. Given that GCs are used commonly in the human eye for their anti-inflammatory, immunosuppressive and anti-VEGF properties (Sulaiman et al., 2018), the feasibility of a clinical collaboration to receive donor eyes from humans that were treated with GC implants or injections prior to death could be explored. A contralateral control eye or other age/sex-matched controls could be used to assess whether there are differences in RSC number via sphere assays or evaluate tissue sections for differences in readily detectable proliferation markers in the CE, such as ki67 or PCNA. Alternatively, GCs are routinely used systemically for various disease indications and this treatment population could be a broader pool of donor tissue for consideration. However, further characterization of the effects of GCs on human RSCs and progenitors in vitro is warranted as only RSPCs from a single donor have been examined so far.

140

5.2.3 Chapter 4

5.2.3.1 The molecular mechanisms of COCO in cone specification

To direct RSC progeny to the cone fate we used COCO, a single protein that is a multifunctional antagonist of TGFβ, BMP and Wnt signaling ligands. This revealed that RSC- derived non-pigmented progenitors and their progeny differentiate along the cone lineage in the absence of all three of those signals. However, exactly what mechanistic influence each of those signaling pathways has in directing RSC progeny to non-cone fates, and what cell types they specify instead, was not investigated. Thus, it may be pertinent to: 1) confirm COCO is effectively blocking signaling through all of the expected pathways and, 2) determine the influence of each signaling pathway antagonized by COCO on the fate of RSC progeny. The activity of COCO can be assessed through downstream signaling readouts: phosphorylated- SMAD2/3 (TGFβ signaling), p-SMAD1/5/8 (BMP signaling) and p-β-catenin (which represents the β-catenin pool targeted for proteosomal degredation when Wnt signaling is inactive) can be measured in the presence and absence of COCO to demonstrate its pathway-specific activity (Zhou et al., 2015). To assess the effect of each pathway on the fate of RSC progeny, individual or various combinations of TGFβ, BMP and Wnt ligands (or small molecule agonists) could be added during COCO differentiation to overcome the antagonism of each pathway and determine its effect on fate specification. Alternatively, individual inhibitors of TGFβ, BMP and Wnt could be used combinatorially to examine the effects of specific pathway inhibition during differentiation.

Another fascinating mechanistic insight would be the examination of RSC-derived cone subtype specification. This could be investigated by adding T3 to cultures at different stages of differentiation, assessing the cultures for Trβ2 expression, and ultimately, determine if the typically S-opsin expressing RSC cone progeny begin to specify to M-opsin expressing cones, and if there is a critical period for this switch.

Two other exogenous factors that may influence cone fate are IGF-1 and Hh. IGF-1 has been shown to enhance cone differentiation from pluripotent stem cells (Mellough et al., 2015); in fact, it has even specifically been shown to do so in concert with COCO (Zhou et al., 2015). Thus, whether IGF-1 influences the temporal development, yield, survival or maturity of RSC-derived cones is of interest. In Chapter 4, Hh signaling was shown to be important for the proliferative expansion of RSC-derived progenitors, without having any influence on rod fate specification.

141

Given that finding and the importance of Hh signaling in retinal development, it would be interesting to determine if Hh signaling affects progenitors in the same capacity during COCO- mediated cone differentiation.

5.2.3.2 Investigating Novel Cone Genes and Future “omics” Approaches

From our RNA-seq data, we have attained a list of genes that are highly expressed in both RSC-derived and endogenous cones. Some of these genes are not currently known to be associated with cone development or function. Furthermore, we have now constructed a transcriptional time course across COCO-mediated cone differentiation of E14 RPCs (timepoints include: Day 0, 5, 10, 21 and 35) (work led by Saeed Khalili). Thus, novel genes we have now identified that are associated with fully differentiated cones, or that appear to be dynamically regulated during differentiation, are prime candidates for follow-up studies investigating the relevance of those genes/gene products in cone differentiation and/or function. However, it may be the case that our use of bulk sequencing has made identification of developmentally dynamic genes at particular timepoints more difficult.

To reduce the heterogeneity of our bulk cell seq comparisons, we sorted using FACS to enrich for cones. However, single-cell sequencing would likely further resolve whether RSC cones vs endogenous cones have discernible heterogeneity. For instance, if RSC-derived cones are heterogenous in level of maturity at any given timepoint, it would convolute bulk gene expression interpretations at a population level. In contrast, single-cell data could resolve such heterogeneity, identify rare cell types, and potentially lead to insights about cone maturation by establishing developmental trajectories over pseudotime.

A further limitation is that transcriptomic data can only give insight into which genes are actively being transcribed into RNA but does not provide insight into whether those genes are being translated into functional protein. Recent studies have emphasized the importance of translational regulation and miRNA activity in the intrinsic regulation of RPC temporal competence. Growing evidence suggests the genes ikaros and castor interact to regulate the shift in RPC competence over time, with ikaros conferring competence for early retinal fates and castor conferring competence for late-born fates (Elliott et al., 2008; Mattar et al., 2015). However, ikaros mRNA is expressed throughout all layers of the developing retina, from early embryonic to late postnatal stages, whereas the protein is present only in early RPCs.

142

Therefore, the timing of ikaros translation may be crucial to its regulation of temporal competence. This sort of dynamic translational regulation would not be revealed by RNA-seq alone. Further, miRNAs are known regulators of translation and have also been shown to be important in regulating RPC temporal competence (La Torre et al., 2013). Thus, both proteomics and miRNA sequencing may add important layers of insight into how RPC competence is regulated. In fact, the combination of data from many different sequencing or “omics” techniques – such as RNA-seq, ChIP-seq and ATAC-seq – has already led to new, systems-biology-level insights into the regulatory networks involved in retinal development (Buono and Martinez-Morales, 2020). Furthermore, single-cell multi-omics techniques are being developed – such as single-cell miRNA-RNA-seq (Wang et al., 2019), single-cell transcriptome- proteome profiling, and single-cell transcriptome-epigenome profiling (Hu et al., 2018). These approaches can not only address cell heterogeneity but provide broader insights into the dynamic interrelation between these molecular layers. However, there are still many technical challenges limiting the application of these techniques, from depth and throughput to bioinformatics. Yet, these could be powerful tools to elucidate key molecular factors involved in RSC fate specification via exogenous factors. Ideally, these single-cell multi-omic profiling approaches could also be applied to determining the regulation of RSC quiescence and proliferation covered in section 5.2 of this discussion.

5.2.3.3 Functional Characterization of RSC-derived Cones

I have already attempted two different strategies to assay RSC-derived cones for functional photosensitivity in vitro: 1) a cGMP enzyme immunoassay (EIA); and 2) a multi-electrode array (MEA). Unfortunately, neither strategy was successful but could warrant further exploration as it is unclear whether they did not work for technical or biological reasons. Thus, both assays could be revisited in the future for discerning RSC-derived cone functional properties.

The EIA strategy aimed to assess whether RSC-derived cones degrade cGMP in response to light. This normally occurs during phototransduction when activated PDEs hydrolyze cGMP and reduce its concentration (see Chapter 1.1.3) (Lamb and Pugh, 2006; Michaelides et al., 2006). At the end of the COCO differentiation protocol, I kept one set of plates in the dark for ~2 days and performed all subsequent steps in the dark (cell lysing, supernatant collection, EIA plating). The other set of plates were exposed to light for a defined time and intensity before collection. A

143 subset of wells in each condition was exposed to PDE-inhibitor IBMX in order to compare total PDE activity to light-responsive PDE activity. Unfortunately, I did not see cGMP changes reflective of photosensitive degradation.

The MEA strategy involved specialized multi-well plates with microelectrodes embedded on the bottom of each well and a detector apparatus that both records electrical currents/spikes from each well and is capable of computer-controlled light flashes of set intensity, duration and colour. For this experiment, I performed COCO differentiation protocols on E14 retinal progenitors or RSC sphere-derived progenitors. Each week, I flashed/measured all control and experimental conditions across the plates for changes in electrical activity in response to light stimulus. Some wells were supplemented with exogenous 9-cis-retinal (physiologically active, more stable and more easily synthesized than 11-cis-retinal) in case endogenous retinal levels might not be sufficient for a photo-response. After the final timepoint, I spiked all wells with IBMX (PDE inhibitor) and measured again for changes/abrogation of electrical activity/light responsiveness. There was no convincing evidence of a light response. However, cell survival issues in the MEA plate may have been a confounding factor.

Patch-clamp electrophysiology and calcium imaging are two other methods to evaluate in vitro neuronal function and both have previously been used to demonstrate the function of RSC- derived rods in culture (Del Debbio et al., 2013; Demontis et al., 2012). Thus, these may be worthwhile methods to determine whether RSC-derived cones possess ion channel dynamics typical of photoreceptor electrophysiological and electrochemical function. One potential caveat to all of these in vitro strategies is that stem cell-derived photoreceptors in 2D culture do not typically form outer segments, which may be a limitation for development of photosensitive functional properties.

However, the best benchmark of RSC-derived cone function would be transplantation and subsequent functional integration into the retina. Our lab has a triple knockout (TKO) mouse model in which rod, cone and ipRGC function has been molecularly disabled (Opn4-/-, Gnat1-/-, Cnga3-/-) (Hattar et al., 2003). The cells are still healthy and intact but non-functional, which means transplanted RSC-derived cones can be assessed for functional integration into the healthy retina, or in an induced injury paradigm. Previous studies have shown that the stage of development of photoreceptor precursor cells impacts survival and integration (Gasparini et al., 2019). Thus, future studies could be carried out where RSC-derived cones from various differentiation timepoints are transplanted into healthy and injured TKO mouse retinas, followed

144 by functional assessment via ERG, pupillary light response or optokinetic assessment. The strength of this mouse model is that the baseline for each functional measurement is effectively zero, making it incredibly sensitive to any improvement in visual function.

5.2.4 Combined Insights and Future Directions

When considering the results of these three thesis chapters all together, some collective hypotheses arise. Given that GC agonism induces RSC expansion and RSPC proliferation in vitro without influencing the differentiation profile of RSPCs, it should follow that including a GC agonist, like dexamethasone, during COCO-mediated cone differentiation will significantly increase the yield of RSC-derived cones in vitro. Thus, combining our novel insights into cone differentiation with the newly identified class of RSPC proliferation enhancing compounds could increase the scalability of RSCs as a source for cones photoreceptors and improve the feasibility of using RSC-cones for cell therapies.

Likewise, I found that intravitreal injection of Noggin and anti-sFRP2 counteracts inhibitory BMP/sFRP2 proteins in the RSC niche and induces the proliferation and self-renewal of endogenous RSCs. Given the evidence that dexamethasone has a significant effect on RSPC proliferation in vitro but only a limited capacity to induce CE proliferation in vivo, it is possible combinatorial injections of Noggin, anti-sFRP2 and Dex may act additively or synergistically to stimulate endogenous RSC activation. If effective, it would be particularly interesting to compare the most effective combination of Dex, Noggin and anti-sFRP2 head-to-head with FINS and evaluate its effects in a retinal degeneration paradigm.

Perhaps a more challenging hypothesis is that the combination of FINS-mediated endogenous activation of RSCs, followed by subsequent in vivo delivery or overexpression of COCO, may lead to a higher proportion of endogenous cone generation in vivo. Foremost, a more robust induction of endogenous RSC/CE proliferation, migration and differentiation is likely required before this would be worth investigating. Further, heterochronic experiments have shown that early- or late-born RPCs do not seem to change their fate output when placed in the opposite environment (Belliveau and Cepko, 1999; Belliveau et al., 2000; Watanabe and Raff, 1990). However, our experiments in Chapter 4 suggest that newborn RSC progenitors do have competency for both early- and late-born retinal neurons and that their differentiation into those

145 cell types is profoundly influenced by extrinsic signals. Thus, if sufficient numbers of RSC- derived progenitors can be produced in vivo, it could be examined if altering the endogenous environment influences their fate specification. In fact, it has been suggested that different exogenous growth factors injected into the chick eye produce different cell types from CMZ progenitors due to the varied microenvironment (Fischer et al., 2013).

One final future direction, less consequent of this thesis work, could be the development of a surgical technique to extract CE-RSCs from a living donor, so that the cells can then be cultured and subsequently used for autologous transplantation into the same donor/eye. This procedure could also facilitate experimentation with ex vivo gene editing/correction in CE-RSCs extracted from donors with genetic rod-cone dystrophies. These cells could be used for disease modeling in vitro, or genetically corrected and used as a source of donor-specific functional photoreceptors for transplantation. To my knowledge, the proof-of-principal that this is possible has not been established for adult CE-RSCs. However, the surgical procedure should be relatively straightforward to develop given there are already established biopsy techniques for extracting tissue from the ciliary body, typically used for the histopathological evaluation of intraocular tumors (Rishi et al., 2016). Also, it would be beneficial to use of a large animal model, like the pig, to develop this technique. The pig is an enticing model for surgical interventions and retinal cell therapy experiments as they have many similarities to human ocular morphology (including size and a cone-rich, macula-like area centralis) and because there are existing transgenic cone and rod dystrophy models (Kostic and Arsenijevic, 2016). If successful, this approach could significantly increase the clinical relevance of CE-RSCs as a source of cells for transplantation.

146

References

Abburi, C., Prabhakar, S., Kalra, J., Huria, A., and Anand, A. (2013). Vascular endothelial growth factor (VEGF) induced proliferation ofhuman fetal derived ciliary epithelium stem cells is mediated by jagged-N cadherin pathway. Curr Neurovasc Res 10, 93–102.

Abdouh, M., and Bernier, G. (2006). In vivo reactivation of a quiescent cell population located in the ocular ciliary body of adult mammals. Exp. Eye Res. 83, 153–164.

Acampora, D., Mazan, S., Avantaggiato, V., Barone, P., Tuorto, F., Lallemand, Y., Brulet, P., and Simeone, A. (1996). Epilepsy and brain abnormalities in mice lacking the Otx1 gene. Nat. Genet. 14, 218–222.

Agathocleous, M., and Harris, W. a (2009). From progenitors to differentiated cells in the vertebrate retina. Annu. Rev. Cell Dev. Biol. 25, 45–69.

Ahmad, I., Tang, L., and Pham, H. (2000). Identification of neural progenitors in the adult mammalian eye. Biochem. Biophys. Res. Commun. 270, 517–521.

Ail, D., and Perron, M. (2017). Retinal Degeneration and Regeneration—Lessons From Fishes and Amphibians. Curr. Pathobiol. Rep. 5, 67–78.

Akagi, T., Mandai, M., Ooto, S., Hirami, Y., Osakada, F., Kageyama, R., Yoshimura, N., and Takahashi, M. (2004). Otx2 homeobox gene induces photoreceptor-specific phenotypes in cells derived from adult iris and ciliary tissue. Investig. Ophthalmol. Vis. Sci. 45, 4570–4575.

Akhtar, T., Xie, H., Khan, M.I., Zhao, H., Bao, J., Zhang, M., and Xue, T. (2019). Accelerated photoreceptor differentiation of hiPSC-derived retinal organoids by contact co-culture with retinal pigment epithelium. Stem Cell Res. 39, 101491.

Akimoto, M., Cheng, H., Zhu, D., Brzezinski, J.A., Khanna, R., Filippova, E., Oh, E.C.T., Jing, Y., Linares, J.L., Brooks, M., et al. (2006). Targeting of GFP to newborn rods by Nrl promoter and temporal expression profiling of flow-sorted photoreceptors. Proc. Natl. Acad. Sci. U. S. A. 103, 3890–3895.

Al-Hussaini, H., Kam, J.H., Vugler, A., Semo, M., and Jeffery, G. (2008). Mature retinal pigment epithelium cells are retained in the cell cycle and proliferate in vivo. Mol. Vis. 14, 1784–1791.

147

Aladdad, A.M., and Kador, K.E. (2019). Adult Stem Cells, Tools for Repairing the Retina. Curr. Ophthalmol. Rep. 7, 21–29.

Alexson, T.O., Hitoshi, S., Coles, B.L., Bernstein, A., and van der Kooy, D. (2006). Notch signaling is required to maintain all neural stem cell populations--irrespective of spatial or temporal niche. Dev. Neurosci. 28, 34–48.

Altshuler, D., Turco, J.J.L., Rush, J., and Cepko, C. (1993). Taurine promotes the differentiation of a vertebrate retinal cell type in vitro. Development 119, 1317–1328.

Amemiya, K., Haruta, M., Takahashi, M., Kosaka, M., and Eguchi, G. (2004). Adult human retinal pigment epithelial cells capable of differentiating into neurons. Biochem. Biophys. Res. Commun. 316, 1–5. del Amo, E.M., Rimpelä, A.K., Heikkinen, E., Kari, O.K., Ramsay, E., Lajunen, T., Schmitt, M., Pelkonen, L., Bhattacharya, M., Richardson, D., et al. (2017). Pharmacokinetic aspects of retinal drug delivery. Prog. Retin. Eye Res. 57, 134–185.

Anacker, C., Cattaneo, A., Luoni, A., Musaelyan, K., Zunszain, P.A., Milanesi, E., Rybka, J., Berry, A., Cirulli, F., Thuret, S., et al. (2013). Glucocorticoid-related molecular signaling pathways regulating hippocampal neurogenesis. Neuropsychopharmacology 38, 872–883.

Aoki, K., and Taketo, M.M. (2008). Tissue-Specific Transgenic, Conditional Knockout and Knock-In Mice of Genes in the Canonical Wnt Signaling Pathway. In Wnt Signaling: Pathway Methods and Mammalian Models, E. Vincan, ed. (Totowa, NJ: Humana Press), pp. 307–331.

Applebury, M.L., Antoch, M.P., Baxter, L.C., Chun, L.L.Y., Falk, J.D., Farhangfar, F., Kage, K., Krzystolik, M.G., Lyass, L.A., and Robbins, J.T. (2000). The murine cone photoreceptor: A single cone type expresses both S and M opsins with retinal spatial patterning. Neuron 27, 513– 523.

Arnhold, S., Semkova, I., Andressen, C., Lenartz, D., Meissner, G., Sturm, V., Kochanek, S., Addicks, K., and Schraermeyer, U. (2004). Iris pigment epithelial cells: A possible cell source for the future treatment of neurodegenerative diseases. Exp. Neurol. 187, 410–417.

Asami, M., Sun, G., Yamaguchi, M., and Kosaka, M. (2007). Multipotent cells from mammalian iris pigment epithelium. Dev. Biol. 304, 433–446.

148

Baakdhah, T., and van der Kooy, D. (2019). Expansion of retinal stem cells and their progeny using cell microcarriers in a bioreactor. Biotechnol. Prog. 35, e2800.

Baker, J.C., Beddington, R.S.P., and Harland, R.M. (1999). Wnt signaling in Xenopus embryos inhibits Bmp4 expression and activates neural development. Genes Dev. 13, 3149–3159.

Balenci, L., and van der Kooy, D. (2014). Notch signaling induces retinal stem-like properties in perinatal neural retina progenitors and promotes symmetric divisions in adult retinal stem cells. Stem Cells Dev. 23, 230–244.

Balenci, L., Wonders, C., Coles, B.L.K., Clarke, L., and van der Kooy, D. (2013). Bone morphogenetic proteins and secreted frizzled related protein 2 maintain the quiescence of adult mammalian retinal stem cells. Stem Cells 31, 2218–2230.

Ballios, B.G., and van der Kooy, D. (2010). Biology and therapeutic potential of adult retinal stem cells. Can. J. Ophthalmol. 45, 342–351.

Ballios, B.G., Cooke, M.J., van der Kooy, D., and Shoichet, M.S. (2010). A hydrogel-based stem cell delivery system to treat retinal degenerative diseases. Biomaterials 31, 2555–2564.

Ballios, B.G., Clarke, L., Coles, B.L.K., Shoichet, M.S., and van der Kooy, D. (2012). The adult retinal stem cell is a rare cell in the ciliary epithelium whose progeny can differentiate into photoreceptors. Biol. Open 1, 237–246.

Ballios, B.G., Cooke, M.J., Donaldson, L., Coles, B.L.K., Morshead, C.M., van der Kooy, D., and Shoichet, M.S. (2015). A Hyaluronan-Based Injectable Hydrogel Improves the Survival and Integration of Stem Cell Progeny following Transplantation. Stem Cell Reports 4, 1031–1045.

Bassett, E. a, and Wallace, V. a (2012). Cell fate determination in the vertebrate retina. Trends Neurosci. 35, 565–573.

Bazhulina, N.P., Nikitin, A.M., Rodin, S.A., Surovaya, A.N., Kravatsky, Y. V., Pismensky, V.F., Archipova, V.S., Martin, R., and Gursky, G. V. (2009). Binding of hoechst 33258 and its derivatives to dna. J. Biomol. Struct. Dyn. 26, 701–718.

Beck, C.W., Christen, B., and Slack, J.M.W. (2003). Molecular pathways needed for regeneration of spinal cord and muscle in a vertebrate. Dev. Cell 5, 429–439.

149

Bélanger, M.-C., Robert, B., and Cayouette, M. (2017). Msx1-Positive Progenitors in the Retinal Ciliary Margin Give Rise to Both Neural and Non-neural Progenies in Mammals. Dev. Cell 40, 1–14.

Belliveau, M.J., and Cepko, C.L. (1999). Extrinsic and intrinsic factors control the genesis of amacrine and cone cells in the rat retina. Development 126, 555–566.

Belliveau, M.J., Young, T.L., and Cepko, C.L. (2000). Late retinal progenitor cells show intrinsic limitations in the production of cell types and the kinetics of opsin synthesis. J. Neurosci. 20, 2247–2254.

Bergmann, T.K., Barraclough, K.A., Lee, K.J., and Staatz, C.E. (2012). Clinical pharmacokinetics and pharmacodynamics of prednisolone and prednisone in solid organ transplantation. Clin. Pharmacokinet. 51, 711–741.

Bernardos, R.L., Barthel, L.K., Meyers, J.R., and Raymond, P.A. (2007). Late-stage neuronal progenitors in the retina are radial Müller glia that function as retinal stem cells. J. Neurosci. 27, 7028–7040.

Berry, M., Ahmed, Z., Lorber, B., Douglas, M., and Logan, A. (2008). Regeneration of axons in the visual system. Restor. Neurol. Neurosci. 26, 147–174.

Bhatia, B., Jayaram, H., Singhal, S., Jones, M.F., and Limb, G.A. (2011). Differences between the neurogenic and proliferative abilities of Müller glia with stem cell characteristics and the ciliary epithelium from the adult human eye. Exp. Eye Res. 93, 852–861.

Bhatt, S., Diaz, R., and Trainor, P.A. (2013). Signals and switches in mammalian neural crest cell differentiation. Cold Spring Harb. Perspect. Biol. 5, 1–20.

Bielen, H., and Houart, C. (2012). BMP Signaling Protects Telencephalic Fate by Repressing Eye Identity and Its Cxcr4-Dependent Morphogenesis. Dev. Cell 23, 812–822.

Bishop, P.N. (2000). Structural Macromolecules and Supramolecular Organisation of the Vitreous Gel. Prog. Retin. Eye Res. 19, 323–344.

Bishop, P.N., Takanosu, M., Le Goff, M., and Mayne, R. (2002). The role of the posterior ciliary body in the biosynthesis of vitreous humour. Eye (Lond). 16, 454–460.

150

Bonhoeffer, S., Mohri, H., Ho, D., and Perelson, A.S. (2000). Quantification of Cell Turnover Kinetics Using 5-Bromo-2′-deoxyuridine 1 . J. Immunol. 164, 5049–5054.

Borday, C., Cabochette, P., Parain, K., Mazurier, N., Janssens, S., Tran, H.T., Sekkali, B., Bronchain, O., Vleminckx, K., Locker, M., et al. (2012). Antagonistic cross-regulation between Wnt and Hedgehog signalling pathways controls post-embryonic retinal proliferation. Dev. 139, 3499–3509.

Bringmann, A., Pannicke, T., Grosche, J., Francke, M., Wiedemann, P., Skatchkov, S.N., Osborne, N.N., and Reichenbach, A. (2006). Müller cells in the healthy and diseased retina. Prog. Retin. Eye Res. 25, 397–424.

Brown, K.E., Keller, P.J., Ramialison, M., Rembold, M., Stelzer, E.H.K., Loosli, F., and Wittbrodt, J. (2010). Nlcam modulates midline convergence during anterior neural plate morphogenesis. Dev. Biol. 339, 14–25.

Brzezinski, J.A., and Reh, T.A. (2015). Photoreceptor cell fate specification in vertebrates. Dev. 142, 3263–3273.

Buono, L., and Martinez-Morales, J.R. (2020). Retina Development in Vertebrates: Systems Biology Approaches to Understanding Genetic Programs: On the Contribution of Next- Generation Sequencing Methods to the Characterization of the Regulatory Networks Controlling Vertebrate Eye Development. BioEssays 42, 1–9.

Carido, M., Zhu, Y., Postel, K., Benkner, B., Cimalla, P., Karl, M.O., Kurth, T., Paquet-Durand, F., Koch, E., Münch, T.A., et al. (2014). Characterization of a mouse model with complete RPE loss and its use for RPE cell transplantation. Invest. Ophthalmol. Vis. Sci. 55, 5431–5444.

Carl, M., and Wittbrodt, J. (1999). Graded interference with FGF signalling reveals its dorsoventral asymmetry at the mid-hindbrain boundary. Development 126, 5659–5667.

Carmon, K.S., Lin, Q., Gong, X., Thomas, A., and Liu, Q. (2012). LGR5 Interacts and Cointernalizes with Wnt Receptors To Modulate Wnt/ -Catenin Signaling. Mol. Cell. Biol. 32, 2054–2064.

Centanin, L., Ander, J.-J., Hoeckendorf, B., Lust, K., Kellner, T., Kraemer, I., Urbany, C., Hasel, E., Harris, W. a, Simons, B.D., et al. (2014). Exclusive multipotency and preferential asymmetric

151 divisions in post-embryonic neural stem cells of the fish retina. Development 141, 3472–3482.

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.

Chacko, D.M., Das, A. V., Zhao, X., James, J., Bhattacharya, S., and Ahmad, I. (2003). Transplantation of ocular stem cells: the role of injury in incorporation and differentiation of grafted cells in the retina. Vision Res. 43, 937–946.

Chalupa, L.M., and Lia, B. (1991). The nasotemporal division of retinal ganglion cells with crossed and uncrossed projections in the fetal rhesus monkey. J. Neurosci. 11, 191–202.

Chang, B., Smith, R.S., Peters, M., Savinova, O. V., Hawes, N.L., Zabaleta, A., Nusinowitz, S., Martin, J.E., Davisson, M.L., Cepko, C.L., et al. (2001). Haploinsufficient Bmp4 ocular phenotypes include anterior segment dysgenesis with elevated intraocular pressure. BMC Genet. 2.

Chen, M., Tian, S., Glasgow, N.G., Gibson, G., Yang, X., Shiber, C.E., Funderburgh, J., Watkins, S., Johnson, J.W., Schuman, J.S., et al. (2015). Lgr5+ amacrine cells possess regenerative potential in the retina of adult mice. Aging Cell 14, 635–643.

Chen, S.-K., Badea, T.C., and Hattar, S. (2011). Photoentrainment and pupillary light reflex are mediated by distinct populations of ipRGCs. Nature 476, 92–95.

Cheng, H., Aleman, T.S., Cideciyan, A. V, Khanna, R., Samuel, G., and Swaroop, A. (2006). In vivo function of the orphan nuclear receptor NR2E3 in establishing photoreceptor identity during mammalian retinal development. Hum Mol Genet 15, 2588–2602.

Chichagova, V., Hallam, D., Collin, J., Zerti, D., Dorgau, B., Felemban, M., Lako, M., and Steel, D.H. (2018). Cellular regeneration strategies for macular degeneration: Past, present and future. Eye 32, 946–971.

Cho, S.H., and Cepko, C.L. (2006). Wnt2b/β-catenin-mediated canonical Wnt signaling determines the peripheral fates of the chick eye. Development 133, 3167–3177.

152

Chow, R.L., and Lang, R.A. (2001). Early eye development in vertebrates. Annu. Rev. Cell Dev. Biol. 17, 255–296.

Chow, R.L., Altmann, C.R., Lang, R.A., and Hemmati-Brivanlou, A. (1999). Pax6 induces ectopic eyes in a vertebrate. Development 126, 4213–4222.

Cicero, S.A., Johnson, D., Reyntjens, S., Frase, S., Connell, S., Chow, L.M.L., Baker, S.J., Sorrentino, B.P., and Dyer, M.A. (2009). Cells previously identified as retinal stem cells are pigmented ciliary epithelial cells. Proc. Natl. Acad. Sci. U. S. A. 106, 6685–6690.

Clarke, L., Ballios, B.G., and van der Kooy, D. (2012). Generation and clonal isolation of retinal stem cells from human embryonic stem cells. Eur. J. Neurosci. 36, 1951–1959.

Close, J.L., Gumuscu, B., and Reh, T.A. (2005). Retinal neurons regulate proliferation of postnatal progenitors and Müller glia in the rat retina via TGFβ signaling. Development 132, 3015–3026.

Coca-Prados, M., and Escribano, J. (2007). New perspectives in aqueous humor secretion and in glaucoma: the ciliary body as a multifunctional neuroendocrine gland. Prog. Retin. Eye Res. 26, 239–262.

Codega, P., Silva-Vargas, V., Paul, A., Maldonado-Soto, A., DeLeo, A., Pastrana, E., and Doetcsch, F. (2014). Prospective identification and purification of quiescent adult neural stem cells from their in vivo niche. Neuron 82, 545–559.

Cohen, A.R., Gomes, F.L.A.F., Roysam, B., and Cayouette, M. (2010). Computational prediction of neural progenitor cell fates. Nat. Methods 7, 213–218.

Coles-Takabe, B.L.K., Brain, I., Purpura, K. a, Karpowicz, P., Zandstra, P.W., Morshead, C.M., and van der Kooy, D. (2008). Don’t look: growing clonal versus nonclonal neural stem cell colonies. Stem Cells 26, 2938–2944.

Coles, B.L.K., and van der Kooy, D. (2017). P-Cadherin is necessary for retinal stem cell behavior in vitro, but not in vivo. Stem Cell Res. 21, 141–147.

Coles, B.L.K., Angénieux, B., Inoue, T., Del Rio-Tsonis, K., Spence, J.R., McInnes, R.R., Arsenijevic, Y., and van der Kooy, D. (2004). Facile isolation and the characterization of human retinal stem cells. Proc. Natl. Acad. Sci. U. S. A. 101, 15772–15777.

153

Coles, B.L.K., Horsford, D.J., McInnes, R.R., and van der Kooy, D. (2006). Loss of retinal progenitor cells leads to an increase in the retinal stem cell population in vivo. Eur. J. Neurosci. 23, 75–82.

Collin, G.B., Gogna, N., Chang, B., Damkham, N., Pinkney, J., Hyde, L.F., Stone, L., Naggert, J.K., Nishina, P.M., and Krebs, M.P. (2020). Mouse Models of Inherited Retinal Degeneration with Photoreceptor Cell Loss. Cells 9, 1–67.

Croft, M.A., Glasser, A., and Kaufman, P.L. (2001). Accommodation and presbyopia. Int. Ophthalmol. Clin. 41, 33–46.

Cruz-Topete, D., and Cidlowski, J.A. (2015). One Hormone Two Actions: Anti- and Pro- inflammatory Effects of Glucocorticoids. Neuroimmunomodulation 22, 20–32.

Currier, N., Chea, K., Hlavacova, M., Sussman, D., Seldin, D., and Dominguez, I. (2010). Dynamic Expression of a LEF-EGFP WNT Reporter in Mouse Development and Cancer. Genesis 48, 183–194.

Cvekl, A., and Tamm, E.R. (2004). Anterior eye development and ocular mesenchyme: New insights from mouse models and human diseases. BioEssays 26, 374–386.

Czerny, T., and Busslinger, M. (1995). DNA-binding and transactivation properties of Pax-6: three amino acids in the paired domain are responsible for the different sequence recognition of Pax-6 and BSAP (Pax-5). Mol. Cell. Biol. 15, 2858–2871.

Dadwal, P., Mahmud, N., Sinai, L., Azimi, A., Fatt, M., Wondisford, F.E., Miller, F.D., and Morshead, C.M. (2015). Activating Endogenous Neural Precursor Cells Using Metformin Leads to Neural Repair and Functional Recovery in a Model of Childhood Brain Injury. Stem Cell Reports 5, 166–173.

Daniele, L.L., Lillo, C., Lyubarsky, A.L., Nikonov, S.S., Philp, N., Mears, A.J., Swaroop, A., Williams, D.S., and Pugh, E.N. (2005). Cone-like morphological, molecular, and electrophysiological features of the photoreceptors of the Nrl knockout mouse. Invest. Ophthalmol. Vis. Sci. 46, 2156–2167.

Danno, H., Michiue, T., Hitachi, K., Yukita, A., Ishiura, S., and Asashima, M. (2008). Molecular links among the causative genes for ocular malformation: Otx2 and sox2 coregulate rax

154 expression. PNAS 105, 5408–5413.

Das, A. V, and James, J. (2005). Retinal properties and potential of the adult mammalian ciliary epithelium stem cells. 45, 1653–1666.

Das, A. V., James, J., Zhao, X., Rahnenführer, J., and Ahmad, I. (2004). Identification of c-Kit receptor as a regulator of adult neural stem cells in the mammalian eye: Interactions with Notch signaling. Dev. Biol. 273, 87–105.

Das, A. V, James, J., Rahnenführer, J., Thoreson, W.B., Bhattacharya, S., Zhao, X., and Ahmad, I. (2005). Retinal properties and potential of the adult mammalian ciliary epithelium stem cells. Vision Res. 45, 1653–1666.

Davis-Silberman, N., and Ashery-Padan, R. (2008). Iris development in vertebrates; genetic and molecular considerations. Brain Res. 1192, 17–28.

Davis-Silberman, N., Kalich, T., Oron-Karni, V., Marquardt, T., Kroeber, M., Tamm, E.R., and Ashery-Padan, R. (2005). Genetic dissection of Pax6 dosage requirements in the developing mouse eye. Hum. Mol. Genet. 14, 2265–2276.

Davis, J., Duncan, M.K., Robison, W.G., and Piatigorsky, J. (2003). Requirement for Pax6 in corneal morphogenesis: A role in adhesion. J. Cell Sci. 116, 2157–2167.

Davis, R.J., Blenkinsop, T.A., Campbell, M., Borden, S.M., Charniga, C.J., Lederman, P.L., Frye, A.M., Aguilar, V., Zhao, C., Naimark, M., et al. (2016). Human RPE stem cell-derived RPE preserves photoreceptors in the royal college of surgeons rat: Method for quantifying the area of photoreceptor sparing. J. Ocul. Pharmacol. Ther. 32, 304–309.

Del Debbio, C.B., Peng, X., Xiong, H., and Ahmad, I. (2013). Adult ciliary epithelial stem cells generate functional neurons and differentiate into both early and late born retinal neurons under non-cell autonomous influences. BMC Neurosci. 14, 130.

Del Debbio, C.B., Santos, M.F., Yan, C.Y.I., Ahmad, I., and Hamassaki, D.E. (2014). Rho GTPases control ciliary epithelium cells proliferation and progenitor profile induction in vivo. Invest. Ophthalmol. Vis. Sci. 55, 2631–2641.

Delamere, N.A. (2005). Ciliary Body and Ciliary Epithelium. Adv Organ Biol. 10, 127–148.

155

Delplace, V., Ortin-Martinez, A., Tsai, E.L.S., Amin, A.N., Wallace, V., and Shoichet, M.S. (2019). Controlled release strategy designed for intravitreal protein delivery to the retina. J. Control. Release 293, 10–20.

Demontis, G.C., Aruta, C., Comitato, A., de Marzo, A., and Marigo, V. (2012). Functional and molecular characterization of rod-like cells from retinal stem cells derived from the adult ciliary epithelium. PLoS One 7, e33338.

Denayer, T., Locker, M., Borday, C., Deroo, T., Janssens, S., Hecht, A., van Roy, F., Perron, M., and Vleminckx, K. (2008). Canonical Wnt Signaling Controls Proliferation of Retinal Stem/Progenitor Cells in Postembryonic Xenopus Eyes . Stem Cells 26, 2063–2074.

Dhomen, N.S., Balaggan, K.S., Pearson, R.A., Bainbridge, J.W., Levine, E.M., Ali, R.R., and Sowden, J.C. (2006). Absence of Chx10 Causes Neural Progenitors to Persist in the Adult Retina. Invest Ophthalmol Vis Sci 47, 386–396.

Dias da Silva, M.R., Tiffin, N., Mima, T., Mikawa, T., and Hyer, J. (2007). FGF-mediated induction of ciliary body tissue in the chick eye. Dev. Biol. 304, 272–285.

Ding, K., Liu, W. ying, Zeng, Q., Hou, F., Xu, J. zhong, and Yang, Z. (2017). Msx1-modulated muscle satellite cells retain a primitive state and exhibit an enhanced capacity for osteogenic differentiation. Exp. Cell Res. 352, 84–94.

Doan, L.T., Javier, A.L., Furr, N.M., Nguyen, K.L., Cho, K.W., and Monuki, E.S. (2012). A Bmp Reporter with Ultrasensitive Characteristics Reveals That High Bmp Signaling Is Not Required for Cortical Hem Fate. PLoS One 7, e44009.

Ducournau, Y., Boscher, C., Adelman, R.A., Guillaubey, C., Schmidt-Morand, D., Mosnier, J.F., and Ducournau, D. (2012). Proliferation of the ciliary epithelium with retinal neuronal and photoreceptor cell differentiation in human eyes with retinal detachment and proliferative vitreoretinopathy. Graefe’s Arch. Clin. Exp. Ophthalmol. 250, 409–423.

Eiraku, M., Takata, N., Ishibashi, H., Kawada, M., Sakakura, E., Okuda, S., Sekiguchi, K., Adachi, T., and Sasai, Y. (2011a). Self-organizing optic-cup morphogenesis in three- dimensional culture. Nature 472, 51–56.

Eiraku, M., Takata, N., Ishibashi, H., Kawada, M., Sakakura, E., Okuda, S., Sekiguchi, K.,

156

Adachi, T., and Sasai, Y. (2011b). Self-organizing optic-cup morphogenesis in three- dimensional culture. Nature 472, 51–56.

Elliott, J., Jolicoeur, C., Ramamurthy, V., and Cayouette, M. (2008). Ikaros Confers Early Temporal Competence to Mouse Retinal Progenitor Cells. Neuron 60, 26–39.

Engelhardt, M., Wachs, F.P., Couillard-Despres, S., and Aigner, L. (2004). The neurogenic competence of progenitors from the postnatal rat retina in vitro. Exp. Eye Res. 78, 1025–1036.

Engelhardt, M., Bogdahn, U., and Aigner, L. (2005). Adult retinal pigment epithelium cells express neural progenitor properties and the neuronal precursor protein doublecortin. Brain Res. 1040, 98–111.

England, S.J., Blanchard, G.B., Mahadevan, L., and Adams, R.J. (2006). A dynamic fate map of the forebrain shows how vertebrate eyes form and explains two cause of cyclopia. Development 133, 4613–4617.

Fadini, G.P., and Avogaro, A. (2013). Dipeptidyl peptidase-4 inhibition and vascular repair by mobilization of endogenous stem cells in diabetes and beyond. Atherosclerosis 229, 23–29.

Fang, Y., Cho, K.-S., Tchedre, K., Lee, S.W., Guo, C., Kinouchi, H., Fried, S., Sun, X., and Chen, D.F. (2013). Ephrin-A3 suppresses Wnt signaling to control retinal stem cell potency. Stem Cells 31, 349–359.

Fischer, A.J., and Reh, T. a (2003a). Growth factors induce neurogenesis in the ciliary body. Dev. Biol. 259, 225–240.

Fischer, A.J., and Reh, T. a (2003b). Potential of Müller glia to become neurogenic retinal progenitor cells. Glia 43, 70–76.

Fischer, A.J., and Reh, T.A. (2001). Transdifferentiation of pigmented epithelial cells: A source of retinal stem cells? Dev. Neurosci. 23, 268–276.

Fischer, A.J., Hendrickson, A., and Reh, T.A. (2001). Immunocytochemical characterization of cysts in the peripheral retina and pars plana of the adult primate. Investig. Ophthalmol. Vis. Sci. 42, 3256–3263.

Fischer, A.J., Zelinka, C., and Scott, M.A. (2010). Heterogeneity of glia in the retina and optic

157 nerve of birds and mammals. PLoS One 5, e10774.

Fischer, A.J., Bosse, J.L., and El-Hodiri, H.M. (2013). The ciliary marginal zone (CMZ) in development and regeneration of the vertebrate eye. Exp. Eye Res. 116, 199–204.

Fritsche, L.G., Fariss, R.N., Stambolian, D., Abecasis, G.R., Curcio, C.A., and Swaroop, A. (2014). Age-related macular degeneration: Genetics and biology coming together. Annu. Rev. Genomics Hum. Genet. 15, 151–171.

Frøen, R.C., Johnsen, E.O., Petrovski, G., Berényi, E., Facskõ, A., Berta, A., Nicolaissen, B., and Moe, M.C. (2011). Pigment epithelial cells isolated from human peripheral iridectomies have limited properties of retinal stem cells. Acta Ophthalmol. 89, 635–644.

Fu, Y., and Yau, K.-W. (2007). Phototransduction in mouse rods and cones. Pflugers Arch 454, 805–819.

Fuhrmann, S., Levine, E.M., and Reh, T.A. (2000). Extraocular mesenchyme patterns the optic vesicle during early eye development in the embryonic chick. Development 127, 4599–4609.

Fujieda, H., Bremner, R., Mears, A.J., and Sasaki, H. (2009). Retinoic acid receptor-related orphan receptor α regulates a subset of cone genes during mouse retinal development. J. Neurochem. 108, 91–101.

Furuta, Y., and Hogan, B.L.M. (1998). BMP4 is essential for lens induction in the mouse embryo. Genes Dev. 12, 3764–3775.

Gallina, D., Zelinka, C., and Fischer, A.J. (2014). Glucocorticoid receptors in the retina, Müller glia and the formation of Müller glia-derived progenitors. Dev. 141, 3340–3351.

Gasparini, S.J., Llonch, S., Borsch, O., and Ader, M. (2019). Transplantation of photoreceptors into the degenerative retina: Current state and future perspectives. Prog. Retin. Eye Res. 69, 1– 37.

Geng, X., Speirs, C., Lagutin, O., Inbal, A., Liu, W., Solnica-Krezel, L., Jeong, Y., Epstein, D.J., and Oliver, G. (2008). Haploinsufficiency of Six3 Fails to Activate Sonic hedgehog Expression in the Ventral Forebrain and Causes Holoprosencephaly. Dev. Cell 15, 236–247.

Gerhart, J., Pfautz, J., Neely, C., Elder, J., DuPrey, K., Menko, A.S., Knudsen, K., and George-

158

Weinstein, M. (2009). Noggin producing, MyoD-positive cells are crucial for eye development. Dev. Biol. 336, 30–41.

Ghazale, H., Ripoll, C., Leventoux, N., Jacob, L., Azar, S., Mamaeva, D., Glasson, Y., Calvo, C.F., Thomas, J.L., Meneceur, S., et al. (2019). RNA Profiling of the Human and Mouse Spinal Cord Stem Cell Niches Reveals an Embryonic-like Regionalization with MSX1+ Roof-Plate- Derived Cells. Stem Cell Reports 12, 1159–1177.

Giannelli, S.G., Demontis, G.C., Pertile, G., Rama, P., and Broccoli, V. (2011). Adult human Müller glia cells are a highly efficient source of rod photoreceptors. Stem Cells 29, 344–356.

Giordano, F., De Marzo, A., Vetrini, F., and Marigo, V. (2007). Fibroblast growth factor and epidermal growth factor differently affect differentiation of murine retinal stem cells in vitro. Mol. Vis. 13, 1842–1850.

Glaser, T., Ton, C., Mueller, R., Petzl-Erler, M., Oliver, C., Nevin, N., Housman, D., and Maas, R. (1994). Absence of PAX6 Gene Mutations in Gillespie Syndrome (Partial Aniridia, Cerebellar Ataxia, and Mental Retardation). Genomics 19, 145–148.

Goldstein, E.B., and Brockmole, J.R. (2017). Sensation & Perception, 10th Edition.

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. Development 138, 227–235.

Graw, J. (2010). Eye development. Curr. Top. Dev. Biol. 90, 343–386.

Gu, P., Harwood, L.J., Zhang, X., Wylie, M., Curry, W.J., and Cogliati, T. (2007). Isolation of retinal progenitor and stem cells from the porcine eye. Mol. Vis. 13, 1045–1057.

Gualdoni, S., Baron, M., Lakowski, J., Decembrini, S., Smith, A.J., Pearson, R.A., Ali, R.R., and Sowden, J.C. (2010). Adult ciliary epithelial cells, previously identified as retinal stem cells with potential for retinal repair, fail to differentiate into new rod photoreceptors. Stem Cells 28, 1048– 1059.

Guduric-Fuchs, J., Chen, W., Price, H., Archer, D.B., and Cogliati, T. (2011). RPE and neuronal differentiation of allotransplantated porcine ciliary epithelium-derived cells. Mol. Vis. 17, 2580– 2595.

159

Guymer, R., Luthert, P., and Bird, A. (1999). Changes in Bruch’s membrane and related structures with age. Prog. Retin. Eye Res. 18, 59–90.

Halder, G., Callaerts, P., and Gehring, W.J. (1995). Induction of ectopic eyes by targeted expression of the eyeless gene in Drosophila. Science (80-. ). 267, 1788–1792.

Hänzelmann, S., Castelo, R., and Guinney, J. (2013). GSVA: Gene set variation analysis for microarray and RNA-Seq data. BMC Bioinformatics 14.

Harrington, L., Klintworth, G.K., Secor, T.E., and Breitman, M.L. (1991). Developmental analysis of ocular morphogenesis in αA-crystallin/diphtheria toxin transgenic mice undergoing ablation of the lens. Dev. Biol. 148, 508–516.

Harris, W.A., and Perron, M. (1998). Molecular recapitulation: The growth of the vertebrate retina. Int. J. Dev. Biol. 42, 299–304.

Hartong, D., Berson, E., and Dryja, T. (2006). Retinitis pigmentosa Prevalence and inheritance patterns. Lancet 368, 1795–1809.

Haruta, M., Kosaka, M., Kanegae, Y., Saito, I., Inoue, T., Kageyama, R., Nishida, A., Honda, Y., and Takahashi, M. (2001). Induction of photoreceptor-specific phenotypes in adult mammalian iris tissue. Nat. Neurosci. 4, 1163–1164.

Hattar, S., Lucas, R.J., Mrosovsky, N., Thompson, S., Douglas, R.H., Hankins, M.W., Lem, J., Biel, M., Hofmann, F., Foster, R.G., et al. (2003). Melanopsin and rod—cone photoreceptive systems account for all major accessory visual functions in mice. Nature 424, 76–81.

Haynes, T., Gutierrez, C., Aycinena, J.C., Tsonis, P.A., and Del Rio-Tsonis, K. (2007). BMP signaling mediates stem/progenitor cell-induced retina regeneration. Proc. Natl. Acad. Sci. U. S. A. 104, 20380–20385.

Heavner, W., and Pevny, L. (2012). Eye development and retinogenesis. Cold Spring Harb. Perspect. Biol. 4, a008391.

Hettiaratchi, M.H., and Shoichet, M.S. (2019). Modulated Protein Delivery to Engineer Tissue Repair. Tissue Eng. - Part A 25, 925–930.

Hollemann, T., Bellefroid, E., and Pieler, T. (1998). The Xenopus homologue of the Drosophila

160 gene tailless has a function in early eye development. Development 125, 2425–2432.

Hollyfield, J.G. (1999). Hyaluronan and the functional organization of the interphotoreceptor matrix. Investig. Ophthalmol. Vis. Sci. 40, 2767–2769.

Holowacz, T., Huelsken, J., Dufort, D., and van der Kooy, D. (2011). Neural stem cells are increased after loss of β-catenin, but neural progenitors undergo cell death. Eur. J. Neurosci. 33, 1366–1375.

Holt, C.E., Bertsch, T.W., Ellis, H.M., and Harris, W.A. (1988). Cellular determination in the xenopus retina is independent of lineage and birth date. Neuron 1, 15–26.

Hu, Y., An, Q., Sheu, K., Trejo, B., Fan, S., and Guo, Y. (2018). Single cell multi-omics technology: Methodology and application. Front. Cell Dev. Biol. 6.

Hyer, J., Kuhlman, J., Afif, E., and Mikawa, T. (2003). Optic cup morphogenesis requires pre- lens ectoderm but not lens differentiation. Dev. Biol. 259, 351–363.

Ille, F., Atanasoski, S., Falk, S., Ittner, L.M., Märki, D., Büchmann-Møller, S., Wurdak, H., Suter, U., Taketo, M.M., and Sommer, L. (2007). Wnt/BMP signal integration regulates the balance between proliferation and differentiation of neuroepithelial cells in the dorsal spinal cord. Dev. Biol. 304, 394–408.

Inoue, T., Kagawa, T., Fukushima, M., Shimizu, T., Yoshinaga, Y., Takada, S., Tanihara, H., and Taga, T. (2006). Activation of canonical Wnt pathway promotes proliferation of retinal stem cells derived from adult mouse ciliary margin. Stem Cells 24, 95–104.

Inoue, T., Coles, B.L.K., Dorval, K., Bremner, R., Bessho, Y., Kageyama, R., Hino, S., Matsuoka, M., Craft, C.M., McInnes, R.R., et al. (2010). Maximizing functional photoreceptor differentiation from adult human retinal stem cells. Stem Cells 28, 489–500.

Inoue, Y., Yanagi, Y., Tamaki, Y., Uchida, S., Kawase, Y., Araie, M., and Okochi, H. (2005). Clonogenic analysis of ciliary epithelial derived retinal progenitor cells in rabbits. Exp. Eye Res. 81, 437–445.

Iongh, R.U. De, Abud, H.E., and Hime, G.R. (2006). WNT/Frizzled signaling in eye development and disease. Front. Biosci. 11, 2442–2464.

161

Ivanovitch, K., Cavodeassi, F., and Wilson, S.W. (2013). Precocious Acquisition of Neuroepithelial Character in the Eye Field Underlies the Onset of Eye Morphogenesis. Dev. Cell 27, 293–305.

Jarrin, M., Pandit, T., and Gunhaga, L. (2012). A balance of FGF and BMP signals regulates cell cycle exit and Equarin expression in lens cells. Mol. Biol. Cell 23, 3266–3274.

Jayaram, H., Jones, M.., Eastlake, K., Cottrill, P.B., Becker, S., Wiseman, J., Khaw, P.T., and Limb, G.A. (2014). Transplantation of photoreceptors derived from human Muller glia restore rod function in the P23H rat. Stem Cells Transl. Med. 3, 323–333.

Jeffery, G. (2001). Architecture of the optic chiasm and the mechanisms that sculpt its development. Physiol. Rev. 81, 1393–1414.

Jeong, Y., and Mangelsdorf, D.J. (2009). Nuclear receptor regulation of stemness and stem cell differentiation. Exp. Mol. Med. 41, 525–537.

Jeong, Y., Leskow, F.C., El-jaick, K., Roessler, E., Muenke, M., Yocum, A., Dubourg, C., Li, X., Geng, X., and Epstein, D.J. (2009). Regulation of a remote Sonic hedgehog forebrain enhancer by the Six3 homeoprotein Yongsu. Nat. Genet. 40, 1348–1353.

Jian, Q., Xu, H., Xie, H., Tian, C., Zhao, T., and Yin, Z.Q. (2009). Activation of retinal stem cells in the proliferating marginal region of RCS rats during development of retinitis pigmentosa. Neurosci. Lett. 465, 41–44.

Johnsen, E.O., Frøen, R.C., Albert, R., Omdal, B.K., Sarang, Z., Berta, A., Nicolaissen, B., Petrovski, G., and Moe, M.C. (2012). Activation of neural progenitor cells in human eyes with proliferative vitreoretinopathy. Exp. Eye Res. 98, 28–36.

Johnsen, E.O., Frøen, R.C., Olstad, O.K., Nicolaissen, B., Petrovski, G., Moe, M.C., and Noer, A. (2018). Proliferative Cells Isolated from the Adult Human Peripheral Retina only Transiently Upregulate Key Retinal Markers upon Induced Differentiation. Curr. Eye Res. 43, 340–349.

Johnston, M.A., and Lim, D.A. (2010). Keeping them quiet: BMPs maintain adult neural stem cell quiescence. Cell Stem Cell 7, 9–10.

Jones, B.W., Watt, C.B., Frederick, J.M., Baehr, W., Chen, C.K., Levine, E.M., Milam, A.H., Lavail, M.M., and Marc, R.E. (2003). Retinal remodeling triggered by photoreceptor

162 degenerations. J. Comp. Neurol. 464, 1–16.

Karl, M.O., Hayes, S., Nelson, B.R., Tan, K., Buckingham, B., and Reh, T.A. (2008). Stimulation of neural regeneration in the mouse retina. Proc. Natl. Acad. Sci. U. S. A. 105, 19508–19513.

Katoh, M. (2007). Networking of WNT, FGF, Notch, BMP, and Hedgehog signaling pathways during carcinogenesis. Stem Cell Rev. 3, 30–38.

Kelley, M.W., Turner, J.K., and Reh, T. a (1994). Retinoic acid promotes differentiation of photoreceptors in vitro. Development 120, 2091–2102.

Khalili, S., Ballios, B.G., Belair-Hickey, J., Donaldson, L., Liu, J., Coles, B.L.K., Grisé, K.N., Baakdhah, T., Bader, G.D., Wallace, V.A., et al. (2018). Induction of rod versus cone photoreceptor-specific progenitors from retinal precursor cells. Stem Cell Res. 33, 215–227.

Kim, D.S., Matsuda, T., and Cepko, C.L. (2008). A core paired-type and POU homeodomain- containing transcription factor program drives retinal bipolar cell gene expression. J. Neurosci. 28, 7748–7764.

Kim, J.S., Park, S.W., Hwang, I.Y., Kim, Y.W., Kim, J.H., and Kim, J.H. (2016a). Expression of Nestin on Endothelial Cells and Pericytes During Retinal Vascular Development in Mouse. J. Korean Ophthalmol. Soc. 57, 499.

Kim, J.W., Yang, H.J., Oel, A.P., Brooks, M.J., Jia, L., Plachetzki, D.C., Li, W., Allison, W.T., and Swaroop, A. (2016b). Recruitment of Rod Photoreceptors from Short-Wavelength-Sensitive Cones during the Evolution of Nocturnal Vision in Mammals. Dev. Cell 37, 520–532.

Kiyama, T., Li, H., Gupta, M., Lin, Y.P., Chuang, A.Z., Otteson, D.C., and Wang, S.W. (2012). Distinct neurogenic potential in the retinal margin and the pars plana of mammalian eye. J. Neurosci. 32, 12797–12807.

Kolb, H., Ralph, N., and Eduardo, F. (2020). Webvision: The Organization of the Retina and Visual System [Internet]. 1995.

Kostic, C., and Arsenijevic, Y. (2016). Animal modelling for inherited central vision loss. J. Pathol. 238, 300–310.

Kubota, R., Hokoc, J.N., Moshiri, A., McGuire, C., and Reh, T. a (2002). A comparative study of

163 neurogenesis in the retinal ciliary marginal zone of homeothermic vertebrates. Brain Res. Dev. Brain Res. 134, 31–41.

Kuwahara, A., Ozone, C., Nakano, T., Saito, K., Eiraku, M., and Sasai, Y. (2015). Generation of a ciliary margin-like stem cell niche from self-organizing human retinal tissue. Nat Commun 6.

Kwan, K.M., Otsuna, H., Kidokoro, H., Carney, K.R., Saijoh, Y., and Chien, C. Bin (2012). A complex choreography of cell movements shapes the vertebrate eye. Development 139, 359– 372.

Lad, E.M., Cheshier, S.H., and Kalani, M.Y.S. (2009). Wnt-signaling in retinal development and disease. Stem Cells Dev. 18, 7–16.

Lagutin, O., Zhu, C., Kobayashi, D., Topczewski, J., Shimamura, K., Puelles, L., Russell, H., McKinnon, P., Solnica-Krezel, L., and Oliver, G. (2003). Six3 repression of Wnt signaling in the anterior neuroectoderm is essential for vertebrate forebrain development. Genes Dev. 17, 368– 379.

Lallemand, Y., Moreau, J., Cloment, C. Saint, Vives, F.L., and Robert, B. (2013). Generation and characterization of a tamoxifen inducible Msx1CreERT2 knock-in allele. Genesis 51, 110– 119.

Lamb, T.D., and Pugh, E.N. (2006). Phototransduction, dark adaption, and rhodopsin regeneration: The proctor lecture. Investig. Ophthalmol. Vis. Sci. 47, 5138–5152.

Lamb, T.D., Collin, S.P., and Pugh, E.N. (2007). Evolution of the vertebrate eye: Opsins, photoreceptors, retina and eye cup. Nat. Rev. Neurosci. 8, 960–976.

Lenkowski, J.R., and Raymond, P.A. (2014). Müller glia: Stem cells for generation and regeneration of retinal neurons in teleost fish. Prog Retin Eye Res 0, 94–123.

Liu, H., Mohamed, O., Dufort, D., and Wallace, V.A. (2003). Characterization of Wnt signaling components and activation of the Wnt canonical pathway in the murine retina. Dev. Dyn. 227, 323–334.

Liu, H., Xu, S., Wang, Y., Mazerolle, C., Thurig, S., Coles, B.L.K., Ren, J.C., Taketo, M.M., van der Kooy, D., and Wallace, V.A. (2007). Ciliary margin transdifferentiation from neural retina is controlled by canonical Wnt signaling. Dev. Biol. 308, 54–67.

164

Liu, H., Etter, P., Hayes, S., Jones, I., Nelson, B., Hartman, B., Forrest, D., and Reh, T.A. (2008). NeuroD1 regulates expression of thyroid hormone receptor β2 and cone opsins in the developing mouse retina. J. Neurosci. 28, 749–756.

Liu, W., Lagutin, O., Swindell, E., Jamrich, M., and Oliver, G. (2010). Neuroretina specification in mouse embryos requires Six3-mediated suppression of Wnt8b in the anterior neural plate. J. Clin. Invest. 120, 3568–3577.

Liu, Y., Chen, S.J., Li, S.Y., Qu, L.H., Meng, X.H., Wang, Y., Xu, H.W., Liang, Z.Q., and Yin, Z.Q. (2017). Long-Term safety of human retinal progenitor cell transplantation in retinitis pigmentosa patients. Stem Cell Res. Ther. 8, 1–12.

Livesey, F.J., and Cepko, C.L. (2001). Vertebrate neural cell-fate determination: Lessons from the retina. Nat. Rev. Neurosci. 2, 109–118.

Llorens-Bobadilla, E., Zhao, S., Baser, A., Saiz-Castro, G., Zwadlo, K., and Martin-Villalba, A. (2015). Single-Cell Transcriptomics Reveals a Population of Dormant Neural Stem Cells that Become Activated upon Brain Injury. Cell Stem Cell 17, 329–340.

Locke, J.C., and Morton, W.R. (1965). Further studies of the viscosity of aspirated human vitreous fluid: with special reference to its use in retinal detachment surgery. Trans. Am. Ophthalmol. Soc. 63, 129–145.

Locker, M., El Yakoubi, W., Mazurier, N., Dullin, J.P., and Perron, M. (2010). A decade of mammalian retinal stem cell research. Arch. Ital. Biol. 148, 59–72.

Loosli, F., Winkler, S., and Wittbrodt, J. (1999). Six3 overexpression initiates the formation of ectopic retina. Genes Dev. 13, 649–654.

Lord-Grignon, J., Abdouh, M., and Bernier, G. (2006). Identification of genes expressed in retinal progenitor/stem cell colonies isolated from the ocular ciliary body of adult mice. Gene Expr. Patterns 6, 992–999.

Lu, A., Ng, L., Ma, M., Kefas, B., Davies, T.F., Hernandez, A., Chan, C.C., and Forrest, D. (2009). Retarded developmental expression and patterning of retinal cone opsins in hypothyroid mice. Endocrinology 150, 1536–1544.

Luo, M., and Chen, Y. (2018). Application of stem cell-derived retinal pigmented epithelium in

165 retinal degenerative diseases: Present and future. Int. J. Ophthalmol. 11, 150–159.

MacLaren, R.E., Pearson, R. a, MacNeil, a, Douglas, R.H., Salt, T.E., Akimoto, M., Swaroop, a, Sowden, J.C., and Ali, R.R. (2006). Retinal repair by transplantation of photoreceptor precursors. Nature 444, 203–207.

MacNeil, A., Pearson, R. a, MacLaren, R.E., Smith, A.J., Sowden, J.C., and Ali, R.R. (2007). Comparative analysis of progenitor cells isolated from the iris, pars plana, and ciliary body of the adult porcine eye. Stem Cells 25, 2430–2438.

Madisen, L., Zwingman, T.A., Sunkin, S.M., Oh, S.W., Zariwala, H.A., Gu, H., Ng, L.L., Palmiter, R.D., Hawrylycz, M.J., Jones, A.R., et al. (2010). A robust and high-throughput Cre Repooting and characterization. Nat Neurosci 13, 133–140.

Maier, W., and Wolburg, H. (1979). Regeneration of the goldfish retina after exposure to different doses of ouabain. Cell Tissue Res. 202, 99–118.

Mandai, M., Watanabe, A., Kurimoto, Y., Hirami, Y., Morinaga, C., Daimon, T., Fujihara, M., Akimaru, H., Sakai, N., Shibata, Y., et al. (2017). Autologous induced stem-cell-derived retinal cells for macular degeneration. N. Engl. J. Med. 376, 1038–1046.

Marcucci, F., Murcia-Belmonte, V., Coca, Y., Ferreiro-Galve, S., Wang, Q., Kuwajima, T., Khalid, S., Ross, M., Herrera, E., and Mason, C. (2016). The ciliary margin zone of the mammalian retina generates retinal ganglion cells. Cell Rep. 17, 3153–3164.

Marquardt, T., Ashery-Padan, R., Andrejewski, N., Scardigli, R., Guillemot, F., and Gruss, P. (2001). Pax6 Is Required for the Multipotent State of Retinal Progenitor Cells. Cell 104, 205– 215.

Martin, D.K., and Whitelaw, E. (1996). The vagaries of variegating transgenes. BioEssays 18, 919–923.

Martinez-Morales, J.R., Signore, M., Acampora, D., Simeone, a, and Bovolenta, P. (2001). Otx genes are required for tissue specification in the developing eye. Development 128, 2019–2030.

Martínez-Navarrete, G.C., Angulo, A., Martín-Nieto, J., and Cuenca, N. (2008). Gradual morphogenesis of retinal neurons in the peripheral retinal margin of adult monkeys and humans. J. Comp. Neurol. 511, 557–580.

166

De Marzo, A., Aruta, C., and Marigo, V. (2010). PEDF Promotes Retinal Neurosphere Formation and Expansion In Vitro. In Retinal Degenerative Diseases: Laboratory and Therapeutic Investigations, R.E. Anderson, J.G. Hollyfield, and M.M. LaVail, eds. (New York, NY: Springer New York), pp. 621–630.

Mathers, P.H., Grinberg, A., Mahon, K.A., and Jamrich, M. (1997). The Rx homeobox gene is essential for vertebrate eye development. Nature 387, 603–607.

Mattapallil, M.J., Wawrousek, E.F., Chan, C.C., Zhao, H., Roychoudhury, J., Ferguson, T.A., and Caspi, R.R. (2012). The Rd8 mutation of the Crb1 gene is present in vendor lines of C57BL/6N mice and embryonic stem cells, and confounds ocular induced mutant phenotypes. Invest. Ophthalmol. Vis. Sci. 53, 2921–2927.

Mattar, P., Ericson, J., Blackshaw, S., and Cayouette, M. (2015). A conserved regulatory logic controls temporal identity in mouse neural progenitors. Neuron 85, 497–504.

McUsic, A.C., Lamba, D.A., and Reh, T.A. (2012). Guiding the morphogenesis of dissociated newborn mouse retinal cells and hES cell-derived retinal cells by soft lithography-patterned microchannel PLGA scaffolds. Biomaterials 33, 1396–1405.

Mears, A.J., Kondo, M., Swain, P.K., Takada, Y., Bush, R.A., Saunders, T.L., Sieving, P.A., and Swaroop, A. (2001). Nrl is required for rod photoreceptor development. Nat. Genet. 29, 447– 452.

Mellough, C.B., Collin, J., Khazim, M., White, K., Sernagor, E., Steel, D.H.W., and Lako, M. (2015). IGF-1 Signaling Plays an Important Role in the Formation of Three-Dimensional Laminated Neural Retina and Other Ocular Structures from Human Embryonic Stem Cells. Stem Cells 33, 2416–2430.

Merico, D., Isserlin, R., Stueker, O., Emili, A., and Bader, G.D. (2010). Enrichment map: A network-based method for gene-set enrichment visualization and interpretation. PLoS One 5, e13984.

Mic, F.A., Molotkov, A., Molotkova, N., and Duester, G. (2004). Raldh2 expression in optic vesicle generates a retinoic acid signal needed for invagination of retina during optic cup formation. Dev. Dyn. 231, 270–277.

167

Michaelides, M., Hardcastle, A.J., Hunt, D.M., and Moore, A.T. (2006). Progressive Cone and Cone-Rod Dystrophies: Phenotypes and Underlying Molecular Genetic Basis. Surv. Ophthalmol. 51, 232–258.

Miesfeld, J.B., and Brown, N.L. (2019). Eye organogenesis: A hierarchical view of ocular development (Elsevier Inc.).

Miller, F.D., and Kaplan, D.R. (2012). Mobilizing endogenous stem cells for repair and regeneration: Are we there yet? Cell Stem Cell 10, 650–652.

Miyake, A., and Araki, M. (2014). Retinal stem/progenitor cells in the ciliary marginal zone complete retinal regeneration: A study of retinal regeneration in a novel animal model. Dev. Neurobiol. 74, 739–756.

Mo, A., Luo, C., Davis, F.P., Mukamel, E.A., Henry, G.L., Nery, J.R., Urich, M.A., Picard, S., Lister, R., Eddy, S.R., et al. (2016). Epigenomic landscapes of retinal rods and cones. Elife 5, 1–29.

Moe, M.C., Kolberg, R.S., Sandberg, C., Vik-Mo, E., Olstorn, H., Varghese, M., Langmoen, I.A., and Nicolaissen, B. (2009). A comparison of epithelial and neural properties in progenitor cells derived from the adult human ciliary body and brain. Exp. Eye Res. 88, 30–38.

Mohedas, A., Xing, X., Armstrong, K., Bullock, A., Cuny, G., and Yu, P. (2013). Development of an ALK2-biased BMP type I receptor kinase inhibitor. 8, 1291–1302.

Monaghan, A.P., Davidson, D.R., Sime, C., Graham, E., Baldock, R., Bhattacharya, S.S., and Hill, R.E. (1991). The Msh-like homeobox genes define domains in the developing vertebrate eye. Development 112, 1053–1061.

Moon, K.H., Kim, H.T., Lee, D., Rao, M.B., Levine, E.M., Lim, D.S., and Kim, J.W. (2018). Differential Expression of NF2 in Neuroepithelial Compartments Is Necessary for Mammalian Eye Development. Dev. Cell 44, 13-28.e3.

Morrison, S., and Kimble, J. (2006). Asymmetric and symmteric division in cancer. Nat. Rev. 441, 1068–1074.

Mort, R.L., Ford, M.J., Sakaue-Sawano, A., Lindstrom, N.O., Casadio, A., Douglas, A.T., Keighren, M.A., Hohenstein, P., Miyawaki, A., and Jackson, I.J. (2014). Fucci2a: A bicistronic

168 cell cycle reporter that allows Cre mediated tissue specific expression in mice. Cell Cycle 13, 2681–2696.

Moshiri, A., and Reh, T. a (2004). Persistent progenitors at the retinal margin of ptc+/- mice. J. Neurosci. 24, 229–237.

Moshiri, A., Close, J., and Reh, T. aM (2004). Retinal stem cells and regeneration. Int. J. Dev. Biol. 48, 1003–1014.

Mukhopadhyay, M., Gorivodsky, M., Shtrom, S., Grinberg, A., Niehrs, C., Morasso, M.I., and Westphal, H. (2006). Dkk2 plays an essential role in the corneal fate of the ocular surface epithelium. Development 133, 2595.

Murthy, K.R., Goel, R., Subbannayya, Y., Jacob, H.K.C., Murthy, P.R., Manda, S.S., Patil, A.H., Sharma, R., Sahasrabuddhe, N.A., Parashar, A., et al. (2014). Proteomic analysis of human vitreous humor. Clin. Proteomics 11, 1–11.

Nair, A., and Bonneau, R.H. (2006). Stress-induced elevation of glucocorticoids increases microglia proliferation through NMDA receptor activation. J. Neuroimmunol. 171, 72–85.

Naska, S., Yuzwa, S.A., Johnston, A.P.W., Paul, S., Smith, K.M., Paris, M., Sefton, M. V., Datti, A., Miller, F.D., and Kaplan, D.R. (2016). Identification of Drugs that Regulate Dermal Stem Cells and Enhance Skin Repair. Stem Cell Reports 6, 74–84.

Nguyen, M., and Arnheiter, H. (2000). Signaling and transcriptional regulation in early mammalian eye development: a link between FGF and MITF. Development 127, 3581–3591.

Nickerson, J., Goodman, P., Chrenek, M., Johnson, C., Berglin, L., Redmond, T., and Boatright, J. (2012). Subretinal delivery and electroporation in pigmented and nonpigmented adult mouse eyes. Methods Mol Biol. 884, 53–69.

Nickerson, P.E.B., Emsley, J.G., Myers, T., and Clarke, D.B. (2007). Proliferation and expression of progenitor and mature retinal phenotypes in the adult mammalian ciliary body after retinal ganglion cell injury. Investig. Ophthalmol. Vis. Sci. 48, 5266–5275.

Nishida, A., Furukawa, A., Koike, C., Tano, Y., Aizawa, S., Matsuo, I., and Furukawa, T. (2003). Otx2 homeobox gene controls retinal photoreceptor cell fate and pineal gland development. Nat. Neurosci. 6, 1255–1263.

169

Nishiguchi, K.M., Kaneko, H., Nakamura, M., Kachi, S., and Terasaki, H. (2008). Identification of photoreceptor precursors in the pars plana during ocular development and after retinal injury. Investig. Ophthalmol. Vis. Sci. 49, 422–428.

Nishiguchi, K.M., Kaneko, H., Nakamura, M., Kachi, S., and Terasaki, H. (2009). Generation of immature retinal neurons from proliferating cells in the pars plana after retinal histogenesis in mice with retinal degeneration. Mol. Vis. 15, 187–199.

Nuzzi, R., Gunetti, M., Rustichelli, D., Roagna, B., Fronticelli Bardelli, F., Fagioli, F., and Ferrero, I. (2012). Effect of in vitro exposure of corticosteroid drugs, conventionally used in AMD treatment, on mesenchymal stem cells. Stem Cells Int. 2012, 946090.

O’Neal, T., and Luther, E. (2020). Retinitis Pigmentosa. StatPearls [Internet]. Treasure Isl. StatPearls Publ.

Odaka, H., Adachi, N., and Numakawa, T. (2017). Impact of glucocorticoid on neurogenesis. Neural Regen. Res. 12, 1028–1035.

Oh, E.C.T., Cheng, H., Hao, H., Jia, L., Khan, N.W., and Swaroop, A. (2008). Rod differentiation factor NRL activates the expression of nuclear receptor NR2E3 to suppress the development of cone photoreceptors. Brain Res. 1236, 16–29.

Ohsawa, R., and Kageyama, R. (2008). Regulation of retinal cell fate specification by multiple transcription factors. Brain Res. 1192, 90–98.

Opas, M., and Dziak, E. (1994). bFGF-induced transdifferentiation of RPE to neuronal progenitors is regulated by the mechanical properties of the substratum. Dev. Biol. 161, 440– 454.

Otsuki, L., and Brand, A.H. (2019). Dorsal-Ventral Differences in Neural Stem Cell Quiescence Are Induced by p57KIP2/Dacapo. Dev. Cell 49, 293-300.e3.

Ott, M. (2006). Visual accommodation in vertebrates: Mechanisms, physiological response and stimuli. J. Comp. Physiol. A Neuroethol. Sensory, Neural, Behav. Physiol. 192, 97–111.

Oyimba, C.U., Vijapurapu, N., Curnow, S.J., Khosla, P., Stewart, P.M., Murray, P.I., Walker, E.A., and Rauz, S. (2006). Characterisation of the prereceptor regulation of glucocorticoids in the anterior segment of the rabbit eye. J. Endocrinol. 190, 483–493.

170

Page, C.P., and Barnes, P.J. (2017). Pharmacology and Therapeutics of Asthma and COPD.

Palczewski, K. (2012). Chemistry and biology of vision. J. Biol. Chem. 287, 1612–1619.

Pariante, C.M., and Miller, A.H. (2001). Glucocorticoid receptors in major depression: Relevance to pathophysiology and treatment. Biol. Psychiatry 49, 391–404.

Park, C.M., and Hollenberg, M.J. (1989). Basic Fibroblast Growth Induces Retinal Regeneration in vivo. 134, 201–205.

Patel, A., and Sowden, J.C. (2019). Genes and pathways in optic fissure closure. Semin. Cell Dev. Biol. 91, 55–65.

Perron, M., and Harris, W.A. (2000). Retinal stem cells in vertebrates. BioEssays 22, 685–688.

Perron, M., Kanekar, S., Vetter, M.L., and Harris, W.A. (1998). The genetic sequence of retinal development in the ciliary margin of the Xenopus eye. Dev. Biol. 199, 185–200.

Post, Y., and Clevers, H. (2019). Defining Adult Stem Cell Function at Its Simplest: The Ability to Replace Lost Cells through Mitosis. Cell Stem Cell 25, 174–183.

Pradeep, T., Mehra, D., and Le, P.H. (2020). Histology, Eye.

Quiring, R., Walldorf, U., Kloter, U., and Gehring, W.J. (1994). Homology of the eyeless gene of drosophila to the small eye gene in mice and aniridia in humans. Science (80-. ). 265, 785–789.

Ragge, N.K., Lorenz, B., Schneider, A., Bushby, K., De Sanctis, L., De Sanctis, U., Salt, A., Collin, J.R.O., Vivian, A.J., Free, S.L., et al. (2005). SOX2 anophthalmia syndrome. Am. J. Med. Genet. 135 A, 1–7.

Ramamoorthy, S., and Cidlowski, J.A. (2013). Exploring the Molecular Mechanisms of Glucocorticoid Receptor Action from Sensitivity to Resistance. Endocr. Dev. 24, 41–56.

Ramsden, C.M., Powner, M.B., Carr, A.-J.F., Smart, M.J.K., da Cruz, L., and Coffey, P.J. (2013). Stem cells in retinal regeneration: past, present and future. Development 140, 2576– 2585.

Rao, R.C., Dedania, V.S., and Johnson, M.W. (2017). Stem Cells for Retinal Disease: A Perspective on the Promise and Perils. Am. J. Ophthalmol. 179, 32–38.

171

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.

Razavi, R., Najafabadi, H.S., Abdullah, S., Smukler, S., Arntfield, M., and van der Kooy, D. (2015). Diabetes enhances the proliferation of adult pancreatic multipotent progenitor cells and biases their differentiation to more β-cell production. Diabetes 64, 1311–1323.

Reeve, R.L., Yammine, S.Z., Deveale, B., and van der Kooy, D. (2016). Targeted activation of primitive neural stem cells in the mouse brain. Eur. J. Neurosci. 43, 1474–1485.

Reeve, R.L., Yammine, S.Z., Morshead, C.M., and van der Kooy, D. (2017). Quiescent Oct4+ Neural Stem Cells (NSCs) Repopulate Ablated Glial Fibrillary Acidic Protein+ NSCs in the Adult Mouse Brain. Stem Cells 35, 2071–2082.

Reh, T.A. (1987). Cell-specific regulation of neuronal production in the larval frog retina. J. Neurosci. 7, 3317–3324.

Reh, T.A., and Fischer, A.J. (2006). Retinal Stem Cells. Methods Enzymol. 419, 52–73.

Reh, T.A., Nagy, T., and Gretton, H. (1987). Retinal pigmented epithelial cells induced to transdifferentiate to neurons by laminin. Nature 330, 68–71.

Reisenhofer, M., Balmer, J., and Zulliger, R. (2015). Multiple programmed cell death pathways are involved in N-methyl-N-nitrosourea-induced photoreceptor degeneration. Graefe’s Arch. Clin. Exp. Ophthalmol. 253, 721–731.

Reisenhofer, M., Pannicke, T., Reichenbach, A., and Enzmann, V. (2016). Characteristics of Müller glial cells in MNU-induced retinal degeneration. Vis. Neurosci. 33, 1–8.

Richardson, D., and Lichtman, J. (2015). Clarifying Tissue Clearing. Cell 162, 246–257.

Rishi, P., Dhami, A., and Biswas, J. (2016). Biopsy techniques for intraocular tumors. Indian J. Ophthalmol. 64, 415–421.

Roberts, M.R., Hendrickson, A., McGuire, C.R., and Reh, T.A. (2005). Retinoid X receptor (gamma) is necessary to establish the S-opsin gradient in cone photoreceptors of the developing mouse retina. Invest. Ophthalmol. Vis. Sci. 46, 2897–2904.

172

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.

Roesch, K., Jadhav, A.P., Trimarchi, J.M., Stadler, M.B., Roska, B., Sun, B.B., and Cepko, C.L. (2008). The transcriptome of retinal Müller glial cells. J. Comp. Neurol. 509, 225–238.

Roska, B., and Sahel, J.-A. (2018). Restoring vision. Nature 557, 359–367.

Royall, L.N., Lea, D., Matsushita, T., Takeda, T.A., Taketani, S., and Araki, M. (2017). A novel culture method reveals unique neural stem/progenitors in mature porcine iris tissues that differentiate into neuronal and rod photoreceptor-like cells. Brain Res. 1675, 51–60.

Salero, E., Blenkinsop, T.A., Corneo, B., Harris, A., Rabin, D., Stern, J.H., and Temple, S. (2012). Adult human RPE can be activated into a multipotent stem cell that produces mesenchymal derivatives. Cell Stem Cell 10, 88–95.

Satoh, S., Tang, K., Iida, A., Inoue, M., Kodama, T., Tsai, S.Y., Tsai, M.J., Furuta, Y., and Watanabe, S. (2009). The spatial patterning of mouse cone opsin expression is regulated by bone morphogenetic protein signaling through downstream effector COUP-TF nuclear receptors. J. Neurosci. 29, 12401–12411.

Satoh, W., Gotoh, T., Tsunematsu, Y., Aizawa, S., and Shimon, A. (2006). Sfrp 1 and Sfrp2 regulate anteroposterior axis elongation and somite segmentation during mouse embryogenesis. Development 133, 989–999.

Satoh, W., Matsuyama, M., Takemura, H., Aizawa, S., and Shimono, A. (2008). Sfrp1, Sfrp2, and Sfrp5 regulate the Wnt/β-catenin and the planar cell polarity pathways during early trunk formation in mouse. Genesis 46, 92–103.

Satokata, I., and Maas, R. (1994). Msx1 deficient mice exhibit cleft palate and abnormalities of craniofacial and tooth development. Nat. Genet. 6, 348–356.

Schlichtenbrede, F.C., Mittmann, W., Rensch, F., Vom Hagen, F., Jonas, J.B., and Euler, T. (2009). Toxicity assessment of intravitreal triamcinolone and bevacizumab in a retinal explant mouse model using two-photon microscopy. Invest. Ophthalmol. Vis. Sci. 50, 5880–5887.

Schuijers, J., and Clevers, H. (2012). Adult mammalian stem cells: The role of Wnt, Lgr5 and R-

173 spondins. EMBO J. 31, 2685–2696.

Schwartz, S.D., Regillo, C.D., Lam, B.L., Eliott, D., Rosenfeld, P.J., Gregori, N.Z., Hubschman, J.P., Davis, J.L., Heilwell, G., Spirn, M., et al. (2015). Human embryonic stem cell-derived retinal pigment epithelium in patients with age-related macular degeneration and Stargardt’s macular dystrophy: Follow-up of two open-label phase 1/2 studies. Lancet 385, 509–516.

Seaberg, R.M., and van der Kooy, D. (2003). Stem and progenitor cells: The premature desertion of rigorous definitions. Trends Neurosci. 26, 125–131.

Seaberg, R.M., Smukler, S.R., Kieffer, T.J., Enikolopov, G., Asghar, Z., Wheeler, M.B., Korbutt, G., and van der Kooy, D. (2004). Clonal identification of multipotent precursors from adult mouse pancreas that generate neural and pancreatic lineages. Nat. Biotechnol. 22, 1115–1124.

Shen, Y. (2020). Stem cell therapies for retinal diseases : from bench to bedside. J. Mol. Med. 98, 1347–1368.

Singh, M.S., Park, S.S., Albini, T.A., Canto-Soler, M.V., Klassen, H., MacLaren, R.E., Takahashi, M., Nagiel, A., Schwartz, S.D., and Bharti, K. (2020). Retinal stem cell transplantation: Balancing safety and potential. Prog. Retin. Eye Res. 75, 100779.

Sinn, R., and Wittbrodt, J. (2013). An eye on eye development. Mech. Dev. 130, 347–358.

De Smet, M.D., Gad Elkareem, A.M., and Zwinderman, A.H. (2013). The vitreous, the retinal interface in ocular health and disease. Ophthalmologica 230, 165–178.

Smiley, S., Nickerson, P.E., Comanita, L., Daftarian, N., El-Sehemy, A., Tsai, E.L.S., Matan- Lithwick, S., Yan, K., Thurig, S., Touahri, Y., et al. (2016). Establishment of a cone photoreceptor transplantation platform based on a novel cone-GFP reporter mouse line. Sci. Rep. 6, 1–15.

Smukler, S.R., Arntfield, M.E., Razavi, R., Bikopoulos, G., Karpowicz, P., Seaberg, R., Dai, F., Lee, S., Ahrens, R., Fraser, P.E., et al. (2011). The adult mouse and human pancreas contain rare multipotent stem cells that express insulin. Cell Stem Cell 8, 281–293.

Sparrow, J.R., Hicks, D., and Barnstable, C.J. (1990). Cell commitment and differentiation in explants of embryonic rat neural retina. Comparison with the developmental potential of dissociated retina. Dev. Brain Res. 51, 69–84.

174

Stanzel, B. V., Liu, Z., Somboonthanakij, S., Wongsawad, W., Brinken, R., Eter, N., Corneo, B., Holz, F.G., Temple, S., Stern, J.H., et al. (2014). Human RPE stem cells grown into polarized RPE monolayers on a polyester matrix are maintained after grafting into rabbit subretinal space. Stem Cell Reports 2, 64–77.

Steinberg, R.H. (1985). Interactions between the retinal pigment epithelium and the neural retina. Doc. Ophthalmol. 60, 327–346.

Stenkamp, D.L. (2011). The rod photoreceptor lineage of teleost fish. Prog Retin Eye Res 30, 395–404.

Stenkamp, D.L., Powers, M.K., Carney, L.H., and Cameron, D.A. (2001). Evidence for two distinct mechanisms evidence for neurogenesis and cellular pattern formation in regenerated goldfish retinas. J. Comp. Neurol. 431, 363–381.

Stenkamp, D.L., Frey, R.A., Mallory, D.E., and Shupe, E.E. (2002). Embryonic retinal gene expression in sonic-you mutant zebrafish. Dev. Dyn. 225, 344–350.

Stern, J., and Temple, S. (2014). Stem cells for retinal repair. Cell-Based Ther. Retin. Degener. Dis. 53, 70–80.

Strauss, O. (2005). The retinal pigment epithelium in visual function. Physiol. Rev. 85, 845–881.

Sukhdeo, K., Koch, C.E., Miller, T.E., Zhou, H., Rivera, M., Yan, K., Cepko, C.L., Lathia, J.D., and Rich, J.N. (2014). The Lgr5 Transgene is Expressed Specifically in Glycinergic Amacrine Cells in the Mouse Retina. Exp. Eye Res. 119, 106–110.

Sulaiman, R., Kadmiel, M., and Cidlowski, J.A. (2018). Glucocorticoid receptor signaling in the eye. Steroids 133, 60–66.

Sun, G., Asami, M., Ohta, H., Kosaka, J., and Kosaka, M. (2006). Retinal stem/progenitor properties of iris pigment epithelial cells. Dev. Biol. 289, 243–252.

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.

Szél, Á., Van Veen, T., and Röhlich, P. (1994). Retinal cone differentiation. Nature 370, 336.

175

Szél, Á., Lukáts, Á., Fekete, T., Szepessy, Z., and Röhlich, P. (2000). Photoreceptor distribution in the retinas of subprimate mammals. J. Opt. Soc. Am. A 17, 568.

Szemes, M., Melegh, Z., Bellamy, J., Greenhough, A., Kollareddy, M., Catchpoole, D., and Malik, K. (2020). A Wnt-BMP4 Signaling Axis Induces MSX and NOTCH Proteins and Promotes Growth Suppression and Differentiation in Neuroblastoma. Cells 9.

Taghiyar, L., Hesaraki, M., Sayahpour, F.A., Satarian, L., Hosseini, S., Aghdami, N., and Eslaminejad, M.B. (2017). Msh homeobox 1 (Msx1)- and Msx2-overexpressing bone marrow- derived mesenchymal stem cells resemble blastema cells and enhance regeneration in mice. J. Biol. Chem. 292, 10520–10533.

Timmer, J.R., Wang, C., and Niswander, L. (2002). BMP signaling patterns the dorsal and intermediate neural tube via regulation of homeobox and helix-loop-helix transcription factors. Development 129, 2459–2472.

Ting, A.E., Mays, R.W., Frey, M.R., Hof, W.V., Medicetty, S., and Deans, R. (2008). Therapeutic pathways of adult stem cell repair. Crit. Rev. Oncol. Hematol. 65, 81–93.

La Torre, A., Georgi, S., and Reh, T.A. (2013). Conserved microRNA pathway regulates developmental timing of retinal neurogenesis. Proc. Natl. Acad. Sci. U. S. A. 110, E2362– E2370.

Tran, F.H., and Zheng, J.J. (2017). Modulating the wnt signaling pathway with small molecules. Protein Sci. 26, 650–661.

Trepp, C., Quintela Pousa, A.M., and Enzmann, V. (2019). Transcriptome Analysis Did Not Show Endogenous Stem Cell Characteristics in Murine Lgr5+ Retinal Cells. Int. J. Mol. Sci. 20.

Trimarchi, J.M., Cho, S.-H., and Cepko, C.L. (2009). Identification of genes expressed preferentially in the developing peripheral margin of the optic cup. Dev. Dyn. 238, 2327–2329.

Tropepe, V., Coles, B.L.K., Chiasson, B.J., Horsford, D.J., Elia, A.J., McInnes, R.R., and van der Kooy, D. (2000). Retinal Stem Cells in the Adult Mammalian Eye. Science (80-. ). 287, 2032–2036.

Tsubura, A., Lai, Y., Miki, H., and Sasaki, T. (2011). Animal Models of N- Methyl- N - nitrosourea-induced Mammary Cancer and Retinal Degeneration with Special Emphasis on

176

Therapeutic Trials. 22, 11–22.

Turner, D., and Cepko, C. (1987). A common progenitor for neurons and glia persists in rat retina late in development. 328, 131–136.

Turner, D.L., Snyder, E.Y., and Cepko, C.L. (1990). Lineage-independent determination of cell type in the embryonic mouse retina. Neuron 4, 833–845.

Varga, Z.M., Wegner, J., and Westerfield, M. (1999). Anterior movement of ventral diencephalic precursors separates the primordial eye field in the neural plate and requires cyclops. Development 126, 5533–5546.

Veleri, S., Lazar, C.H., Chang, B., Sieving, P.A., Banin, E., and Swaroop, A. (2015). Biology and therapy of inherited retinal degenerative disease: Insights from mouse models. DMM Dis. Model. Mech. 8, 109–129.

Vrensen, G.F.J.M., Graw, J., and De Wolf, A. (1991). Nuclear breakdown during terminal differentiation of primary lens fibres in mice: A transmission electron microscopic study. Exp. Eye Res. 52, 647–659.

Wall, D.S., Mears, A.J., McNeill, B., Mazerolle, C., Thurig, S., Wang, Y., Kageyama, R., and Wallace, V.A. (2009). Progenitor cell proliferation in the retina is dependent on Notch- independent Sonic hedgehog/Hes1 activity. J. Cell Biol. 184, 101–112.

Wan, Y., Almeida, A.D., Rulands, S., Chalour, N., Muresan, L., Wu, Y., Simons, B.D., He, J., and Harris, W. (2016). The ciliary marginal zone of the zebrafish retina: clonal and time-lapse analysis of a continuously growing tissue. Development 143, 1099–1107.

Wang, J.-S., and Kefalov, V.J. (2011). The Cone-specific Visual Cycle. Prog. Retin. Eye Res. 30, 115–128.

Wang, H., Lau, B.W.M., Yau, S.Y., Li, S.Y., Leung, N., Wang, N.L., Tang, S.W., Lee, T., and So, K.F. (2010). Roles of paroxetine and corticosterone on adult mammalian ciliary body cell proliferation. Chin. Med. J. (Engl). 123, 1305–1310.

Wang, N., Zheng, J., Chen, Z., Liu, Y., Dura, B., Kwak, M., Xavier-Ferrucio, J., Lu, Y.C., Zhang, M., Roden, C., et al. (2019). Single-cell microRNA-mRNA co-sequencing reveals non-genetic heterogeneity and mechanisms of microRNA regulation. Nat. Commun. 10, 1–12.

177

Wang, S., Sengel, C., Emerson, M.M., and Cepko, C.L. (2014). A gene regulatory network controls the binary fate decision of rod and bipolar cells in the vertebrate retina. Dev. Cell 30, 513–527.

Wang, X., Shan, X., and Gregory-Evans, C.Y. (2017). A mouse model of aniridia reveals the in vivo downstream targets of Pax6 driving iris and ciliary body development in the eye. Biochim. Biophys. Acta - Mol. Basis Dis. 1863, 60–67.

Wang, Y., Dakubo, G.D., Thurig, S., Mazerolle, C.J., and Wallace, V.A. (2005). Retinal ganglion cell-derived sonic hedgehog locally controls proliferation and the timing of RGC development in the embryonic mouse retina. Development 132, 5103–5113.

Wang, Y.A.Z., Plane, J.M., Jiang, P., Zhou, C.J., and Deng, W. (2011). Concise review: Quiescent and active states of endogenous adult neural stem cells: Identification and characterization. Stem Cells 29, 907–912.

Wässle, H., Puller, C., Müller, F., and Haverkamp, S. (2009). Cone contacts, mosaics, and territories of bipolar cells in the mouse retina. J. Neurosci. 29, 106–117.

Watanabe, T., and Raff, M.C. (1990). Rod photoreceptor development in vitro: Intrinsic properties of proliferating neuroepithelial cells change as development proceeds in the rat retina. Neuron 4, 461–467.

Webster, M.K., Cooley-Themm, C.A., Barnett, J.D., Bach, H.B., Vainner, J.M., Webster, S.E., and Linn, C.L. (2017). Evidence of BrdU-positive retinal neurons after application of an Alpha7 nicotinic acetylcholine receptor agonist. Neuroscience. 346, 437–446.

West, E.L., Ribeiro, J., and Ali, R.R. (2020). Development of Stem Cell Therapies for Retinal Degeneration. Cold Spring Harb. Perspect. Biol. 12, a035683.

Wetts, R., and Fraser, S.E. (1988). Multipotent precursors can give rise to all major cell types of the frog retina. Science (80-. ). 239, 1142–1145.

Wetts, R., Serbedzija, G.N., and Fraser, S.E. (1989). Cell lineage analysis reveals multipotent precursors in the ciliary margin of the frog retina. Dev. Biol. 136, 254–263.

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

178 photoreceptor cell ablation in the zebrafish retina. Proc. Natl. Acad. Sci. U. S. A. 114, E3719– E3728.

Willert, J., Epping, M., Pollack, J.R., Brown, P.O., and Nusse, R. (2002). A transcriptional response to Wnt protein in human embryonic carcinomacells. BMC Dev. Biol. 2, 1–7.

Wilson, S.G., Wen, W., Pillai-Kastoori, L., and Morris, A.C. (2016). Tracking the fate of her4 expressing cells in the regenerating retina using her4:Kaede zebrafish. Exp Eye Res 145, 75– 87.

Wohl, S.G., Schmeer, C.W., Kretz, A., Witte, O.W., and Isenmann, S. (2009). Optic nerve lesion increases cell proliferation and nestin expression in the adult mouse eye in vivo. Exp. Neurol. 219, 175–186.

Wohl, S.G., Schmeer, C.W., Friese, T., Witte, O.W., and Isenmann, S. (2011). In Situ dividing and phagocytosing retinal microglia express Nestin, Vimentin, and NG2 In Vivo. PLoS One 6, e22408.

Wohl, S.G., Schmeer, C.W., and Isenmann, S. (2012). Neurogenic potential of stem/progenitor- like cells in the adult mammalian eye. Prog. Retin. Eye Res. 31, 213–242.

Wolosin, J.M., Budak, M.T., and Akinci, M.A.M. (2004). Ocular surface epithelial and stem cell development. Int. J. Dev. Biol. 48, 981–991.

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 2040: A systematic review and meta-analysis. Lancet Glob. Heal. 2, e106–e116.

Xu, H., Sta Iglesia, D.D., Kielczewski, J.L., Valenta, D.F., Pease, M.E., Zack, D.J., and Quigley, H.A. (2007a). Characteristics of progenitor cells derived from adult ciliary body in mouse, rat, and human eyes. Invest. Ophthalmol. Vis. Sci. 48, 1674–1682.

Xu, S., Sunderland, M.E., Coles, B.L.K., Kam, A., Holowacz, T., Ashery-Padan, R., Marquardt, T., McInnes, R.R., and van der Kooy, D. (2007b). The proliferation and expansion of retinal stem cells require functional Pax6. Dev. Biol. 304, 713–721.

Yan, W., Laboulaye, M.A., Tran, N.M., Whitney, I.E., Benhar, I., and Sanes, J.R. (2020). Mouse Retinal Cell Atlas: Molecular Identification of Over Sixty Amacrine Cell Types. J. Neurosci. 40,

179

5177–5195.

Yanagi, Y., Inoue, Y., Kawase, Y., Uchida, S., Tamaki, Y., Araie, M., and Okochi, H. (2006). Properties of growth and molecular profiles of rat progenitor cells from ciliary epithelium. Exp. Eye Res. 82, 471–478.

Yang, X.-J. (2004). Roles of cell-extrinsic growth factors in vertebrate eye pattern formation and retinogenesis. Physiol. Behav. 15, 91–103.

Yang, J., and Mishina, Y. (2019). Generation and Identification of Genetically Modified Mice for BMP Receptors. Methods Mol Biol. 1891, 165–177.

Young, R.W. (1985a). Cell Proliferation During Postnatal Development of the Retina in the Mouse Preparation of autoradiograms Analysis of autoradiograms. 21, 229–239.

Young, R.W. (1985b). Cell differentiation in the retina of the mouse. Anat. Rec. 212, 199–205.

Yun, S., Saijoh, Y., Hirokawa, K.E., Kopinke, D., Murtaugh, L.C., Monukï, E.S., and Levine, E.M. (2009). Lhx2 links the intrinsic and extrinsic factors that control optic cup formation. Development 136, 3895–3906.

Zeng, C., Pan, F., Jones, L.A., Lim, M.M., Griffin, E.A., Sheline, Y.I., Mintun, M.A., Holtzman, D.M., and Mach, R.H. (2010). Evaluation of 5-ethynyl-2′-deoxyuridine staining as a sensitive and reliable method for studying cell proliferation in the adult nervous system. Brain Res. 1319, 21– 32.

Zeng, S., Qiao, H., Lv, X.W., Fan, D., Liu, T., and Xie, D. (2017). High-dose dexamethasone induced LPS-stimulated rat alveolar macrophages apoptosis. Drug Des. Devel. Ther. 11, 3097– 3104.

Zhang, H., Dai, X., Qi, Y., He, Y., Du, W., and Pang, J.J. (2015). Histone deacetylases inhibitors in the treatment of retinal degenerative diseases: Overview and perspectives. J. Ophthalmol. 2015.

Zhang, L., Mathers, P., and Jamrich, M. (2000). Function of Rx, but not Pax6, is essential for the formation of retinal progenitor cells in mice. Genesis 28, 135–142.

Zhao, S., Chen, Q., Hung, F.C., and Overbeek, P.A. (2002a). BMP signaling is required for

180 development of the ciliary body. Development 129, 4435–4442.

Zhao, S., Chen, Q., Hung, F., and Overbeek, P.A. (2002b). BMP signaling is required for development of the ciliary body. Development 129, 4435–4442.

Zhao, X., Liu, J., and Ahmad, I. (2002c). Differentiation of embryonic stem cells into retinal neurons. Biochem. Biophys. Res. Commun. 297, 177–184.

Zhao, X., Das, A. V, Soto-Leon, F., and Ahmad, I. (2005). Growth factor-responsive progenitors in the postnatal mammalian retina. Dev. Dyn. 232, 349–358.

Zhou, S., Flamier, A., Abdouh, M., Tétreault, N., Barabino, A., Wadhwa, S., and Bernier, G. (2015). Differentiation of human embryonic stem cells into cone photoreceptors through simultaneous inhibition of BMP, TGFβ and Wnt signaling. Development 142, 3294–3306.

Zhou, Y., Tanzie, C., Yan, Z., Chen, S., Duncan, M., Gaudenz, K., Li, H., Seidel, C., Lewis, B., Moran, A., et al. (2013). Notch2 regulates BMP signaling and epithelial morphogenesis in the ciliary body of the mouse eye. Proc. Natl. Acad. Sci. U. S. A. 110, 8966–8971.

Zielke, N., and Edgar, B.A. (2015). FUCCI sensors: Powerful new tools for analysis of cell proliferation. Wiley Interdiscip. Rev. Dev. Biol. 4, 469–487.

Zou, C., and Levine, E.M. (2012). Vsx2 Controls Eye Organogenesis and Retinal Progenitor Identity Via Homeodomain and Non-Homeodomain Residues Required for High Affinity DNA Binding. PLoS Genet. 8, e1002924.

Zuber, M.E., Gestri, G., Viczian, A.S., Barsacchi, G., and Harris, W.A. (2003). Specification of the vertebrate eye by a network of eye field transcription factors. Development 130, 5155–5167.

181

Appendices Appendix A. Chapter 2 Supplementary Information

182

183

Figure A1. Some EdU+ cells in the ciliary epithelium and neural retina co-labeled for endothelial cell marker ERG (A-B) EdU and ERG labeling in the (A) neural retina and (B) ciliary epithelium of PBS treated eyes. (C-D) EdU and ERG labeling in the (C) neural retina and (D) ciliary epithelium of FINS treated eyes. Hoechst was used to label all nuclei. Dashed line box indicates high magnification inset. Straight dashed line indicates ciliary epithelium (CE) neural retina (NR) border. 10µm thick sections.

184

185

Figure A2. Some EdU+ cells in the ciliary epithelium and neural retina co-labeled for microglia/macrophage cell marker CD68 (A-B) )EdU and CD68 labeling in the (A) neural retina and (B) ciliary epithelium of PBS treated eyes. (C-D) EdU and CD68 labeling in the (C) neural retina and (D) ciliary epithelium of FINS treated eyes. Hoechst was used to label all nuclei. Dashed line box indicates high magnification inset. Straight dashed line indicates ciliary epithelium (CE) neural retina (NR) border. 10µm thick sections.

‘

186

187

Figure A3. Modulation of downstream BMP and Wnt signaling mediates CE proliferation (A) Schematic of the intravitreal injection paradigm followed by endpoint IHC. Mice received one intravitreal injection per day for three days while EdU was delivered via the drinking water continuously until the Day 4 endpoint. Injections consisted of 0.5% DMSO control, CHIR99021 or LDN-193189. (B) Quantification of EdU cell number in the CB normalized by area in eyes treated with the indicated conditions. N=5-6 eyes per group. Data are Mean ± SEM. (C) Quantification of Pax6+EdU+ co-labeled cells relative to total CE area in eyes treated with PBS vehicle or indicated factors (one-way ANOVA F(4,23)=3.23, p=0.031; N=5-6 eyes per group). Holm-Sidak posthoc test, * = p<0.05. Data are Mean ± SEM. (D) Percent of total EdU-positive cells in the CB that co-labeled for CE marker Pax6. (one-way ANOVA F(4,22)=10.58, p<0.001; N=5-6 eyes per group). Holm-Sidak posthoc test, * = p<0.05. Data are Mean ± SEM. (E-I) Pax6 immunostaining and EdU labeling in the ciliary body. Hoechst stain was used to label all nuclei. White arrows indicate Pax6+EdU+ double-positive cells. Red arrows indicate EdU+ only cells. Dashed line box indicates high magnification inset. Straight dashed line indicates ciliary epithelium (CE) neural retina (NR) border. 10µm-thick sections.

188

189

Figure A4. Day 4 multichannel images of the representative IHC images in Figure 2 (A-F). Pax6 immunostaining and EdU labeling in the ciliary epithelium and peripheral retina of eyes injected with PBS vehicle or indicated factors at Day 4. Hoechst stain was used to label all nuclei. White arrows indicate Pax6+EdU+ double-positive cells. Dashed line box indicates high magnification inset. Straight dashed line indicates ciliary epithelium (CE) neural retina (NR) border. 10µm-thick sections.

190

191

Figure A5. Day 31 multichannel images of the representative IHC images in Figure 2 (A-F). Pax6 immunostaining and EdU labeling in the ciliary epithelium and peripheral retina of eyes injected with PBS vehicle or indicated factors at Day 31. Hoechst stain was used to label all nuclei. White arrows indicate Pax6+EdU+ double-positive cells. Dashed line box indicates high magnification inset. Straight dashed line indicates ciliary epithelium (CE) neural retina (NR) border. 10µm-thick sections.

192

193

Figure A6. Inducible Msx1-CreERT2 mouse lineage labeling in mice with and without tamoxifen exposure and in numerous tissues in the adult mouse eye. (A) TdTomato expression is not apparent in the CE of Cre+ mice that did not receive tamoxifen. (B) TdTomato expression is present in the CE of Cre+ mice that did receive tamoxifen. (C) TdTomato expression in the ciliary epithelium and the trabecular meshwork (white arrow). (D) TdTomato expression in the choroid vasculature (white arrows). (E) TdTomato expression in the corneal endothelium (up arrows) and iris vasculature (down arrows). (F) TdTomato expression in epithelial cells of the lens (white arrows). 10µm-thick sections. DRAQ5 was used to label all nuclei.

194

195

Figure A7. Initial FACS gating to select live cells and the stop codon genotype of post- FACS RSC spheres (A-D) Representative FACS gating plots for CE cells derived from Msx1-CreERT2;Rosa26- TdTomato mice (A-C) Forward and side scatter gating. (D) DAPI based live-dead cell gating. (E) Average live cell recovery from as a percentage of total events detected. N=3. Data is mean ± SEM. (F) Cropped, PCR gel for DNA extracted from single RSC spheres, which was amplified using primers spanning the floxed stop codon region of the Rosa26-TdTomato reporter construct. A 2Kb band indicates the presence of the stop codon. A 1Kb band indicates excision of the stop codon. For 20/21 of spheres tested, a single 1Kb or 2Kb DNA band was detected, indicating the presence of a single genotype for all cells in the sphere.

196

197

Figure A8. N-methyl-N-nitrosourea induces tunable photoreceptor degeneration (A) Schematic of intraperitoneal injection of 45mg/kg or 60mg/kg MNU followed by IHC of eyes 7 days and 14 days after injection. (B) Outer nuclear layer thickness from mice treated with MNU or 10% DMSO. (two-way ANOVA F(2,29)=12.15, p<0.001; N=5-6 eyes per group). Holm-Sidak posthoc test, * = p<0.05. Data are mean ± SEM. (C) Hoechst nuclear staining of retinas from mice treated with the indicated conditions at Day 7 and Day 14. RGC = retinal ganglion cell layer, INL = inner nuclear layer, ONL = outer nuclear layer 10µm-thick sections. (D) Immunostaining for cones (cone arrestin), bipolar cells (Chx10) and rod outer segments (RET-P1) in retinas from mice treated with the indicated conditions at Day 14. Hoechst was used to label all nuclei. 10µm-thick sections.

198

Figure A9. PBS injected eyes with MNU injury occasionally developed gross phthisis bulbi pathology whereas FINS injected eyes with MNU injury did not (A) Quantification of mid-sagittal eye diameter. N=5-12 eyes per group. Each data point represents a single eye and mean ± SEM is indicated. (B-C) Brightfield and fluorescence overlay images of Msx1-Cre driven TdTomato expression in PBS+MNU treated eyes from the same mouse with (B) whole-eye phthisis bulbi with internal disorganization and (C) normal morphology. 10µm-thick sections.

199

200

Figure A10. Multichannel images of the representative Msx1 and Pax6 IHC images in Figure 6 (A-D) Brightfield and fluorescence overlay images of Pax6 immunostaining and Msx1-Cre driven TdTomato expression in eye sections from the indicated conditions. White arrows indicate Msx1-TdTomato+Pax6+ co-labeled cells. Dashed line box indicates high magnification inset. Dashed line box indicates high magnification inset. Straight dashed line indicates ciliary epithelium (CE) neural retina (NR) border. 10µm-thick sections.

201

202

Figure A11. The Msx1-TdTomato expression domain of naïve eyes at 1 day and 45 days after tamoxifen induction (A) Percent of eye sections with Msx1-TdTomato+ cells present in retina for naïve Msx1- CreERT2;Rosa26-TdTomato mice immediately following tamoxifen induction (Day 1) or 45 days after tamoxifen induction (Day 45). The two timepoints were not significantly different (t-test t(8) = 0.53, p=0.61; Naïve Day 1, N=4, Naïve Day 45, N=6. Each data point represents a single eye and mean ± SEM is indicated. (B-C) Quantification of the migration distance of Msx1-TdTomato+ cells into the retina for for naïve Msx1-CreERT2;Rosa26-TdTomato mice immediately following tamoxifen induction (Day 1) or 45 days after tamoxifen induction (Day 45). (B) The migration distance recorded for individual cells. Each data point represents a single cell and mean ± SEM is indicated. Naïve Day 1, N=4 cells from 4 eyes; Naïve Day 45, N=10 cells from 6 eyes. (C) Average migration distance per eye. At Day 45, cells were a significantly greater distance into the retina than at Day 1 (t-test t(8) = 0.91, p=0.0086. Naïve Day 1, N=4, Naïve Day 45, N=6. Each data point represents a single eye. * = p<0.05. (D-E) Brightfield and fluorescence overlay images of Day1 (D1-D2) and Day 45 (E1-E2) Naïve Msx1-CreERT2;Rosa26-TdTomato mouse eyes with TdTomato+ cells in the retina. B1 and B2 are the same retinal margin at different magnifications. .C1 and C2 are the same retinal margin at different magnifications. Straight dashed line indicates ciliary epithelium (CE) neural retina (NR) border. White arrows indicate Msx1-TdTomato+ cells in the retina. 10µm-thick sections.

203

204

Figure A12. Multichannel images of the representative IHC images in Figure 6 (A-D) Brightfield and fluorescence overlay images of Recoverin immunostaining and Msx1-Cre driven TdTomato expression in eye sections from the indicated conditions. DRAQ5 stain was used to label all nuclei. White arrows indicate Msx1-TdTomato+Recoverin+ co-labeled cells. Dashed line box indicates high magnification inset. Straight dashed line indicates ciliary epithelium (CE) neural retina (NR) border. 10µm-thick sections.

205

206

Figure A13. Multichannel images of the representative IHC images in Figure 6 (A-D) Brightfield and fluorescence overlay images of Brn3a immunostaining and Msx1-Cre driven TdTomato expression in eye sections from the indicated conditions. DRAQ5 stain was used to label all nuclei. White arrows indicate Msx1-TdTomato+Brn3a+ co-labeled cells. Dashed line box indicates high magnification inset. Straight dashed line indicates ciliary epithelium (CE) neural retina (NR) border. 10µm-thick sections.

207

208

Figure A14. Msx1-TdTomato+ cells were detected in the RPE layer at very low frequency (A-B) Msx1-TdTomato+ CE cells in the RPE layer of MNU injured eyes treated with FINS. DRAQ5 was used to label nuclei. Dashed line box indicates high magnification inset. Dashed line box indicates high magnification inset. 10µm thick sections. RGC = retinal ganglion cell layer, INL = inner nuclear layer, ONL = outer nuclear layer, RPE = retinal pigmented epithelium layer.

209

Table A1. Detailed primer sequence list Primer Sequence (5’→3’) Msx1- F GGCTGTCTCGAGCTGCGGCTGGAGGG Msx1- R CCATGGCGGTTGCGGTGGCCGCAGC Cre-R GCTGGATAGTTTTACTGCCAGACCGCGCGCC pCAG-F GCAACGTGCTGGTTATTGTG (stop sequence evaluation) TdTomato-R TCTTTGATGACGGCCATGT (stop sequence evaluation) WT tdTomato-F AAGGGAGCTGCAGTGGAGTA WT tdTomato-R CCGAAAATCTGTGGGAAGTC Mut tdTomato-F GGCATTAAAGCAGCGTATCC Mut tdTomato-R CTGTTCCTGTACGGCATGG

210

Appendix B. Chapter 3 Supplementary Information

Table B1. Compounds that met hit criteria in at least one of two screens

Mechanism Hit Screen 1 Hit Screen 2 Hit Frequency

Compound Mechanism Freq. P<0.05 Nuclei GFP Ratio Nuclei GFP Ratio

Prednisolone Glucocorticoid 2/4 √ 145% 160% 1.10 158% 162% 1.02 receptors alpha and beta agonist

Dexamethasone Glucocorticoid 2/4 √ 142% 150% 1.05 receptors alpha and beta agonist

MK-0431 Dipeptidyl 1/1 √ 150% 137% 1.09 peptidase IV (CD26; DPP- IV) inhibitor

MDV-3100 Androgen 1/5 139% 189% 1.36 * * recepptor antagonist

Abbvie Mcl-1 Cpd Mcl-1 inhibitor 1/5 140% 148% 1.06 30b

Sodium Butyrate HDAC inhibitor 1/47 134% 208% 1.55 * *

Odanacatib Cathepsin K 1/1 √ 136% 240% 1.76 inhibitor

INCB024360 Indoleamine 1/1 √ 143 139% 0.97 2,3- dioxygenase inhibitor

Thiazovivin Rho/ROCK 1/3 140% 132% 0.94 inhibitor

Lopinavir HIV protease 1/1 √ 129% 254% 1.97 inhibitor

Guanabenz A2-adrenergic 1/1 √ 129% 142% 1.10 Acetate receptor agonist

Irrestatin 9389 IRE1 inhibitor 1/3 137% 143% 1.04

211

* = seeding error: compound not in second screen List of hit compounds and their known signaling pathways/mechanistic targets. Mechanism hit frequency indicates the number of hit compounds with the indicated mechanistic target relative to the total number of compounds with that target in the library. P-values are indicated for any target pathway with a significant enrichment within the library, as determined by the hypergeometric test. Percentages represent the number of nuclei or total actin-GFP area relative to 0.1% DMSO control (set as 100%). Percent of control is only indicated if a compound was found to be greater than three standard deviations above the control mean for both nuclei number and GFP area within a single screen. Ratio indicates the GFP area relative to the number of nuclei. A large GFP/Nuclei ratio indicates a disproportionate increase in cell area vs cell number and may signify compounds that caused cell hypertrophy.

212

Table B2. Screening quality metrics

Screen 1 Screen 2 Nuclei GFP Nuclei GFP SSMD -4.30 -2.83 -3.75 -2.55 (Strong) (Moderate) (Strong) (Moderate) Coefficient of 13.92 11.51 7.97 10.21 Variation Signal:Noise 8.54 10.9 18.08 4.16 Statistical evaluation of the variability and sensitivity of the MTS assay. Controls were determined to meet a threshold of statistical confidence to reliably determine hit compounds by assessing the variability within the 1x Control (coefficient of variation) and comparing the 1x Control with the pseudo-positive 2x Control (signal:noise ratio; strictly standardized mean difference (SSMD)). If SSMD is ≤ -2 and > -3 it is considered a moderate control. If SSMD is ≤ - 3 and > -5 it is considered a strong control.

213

Figure B1. Visual confirmation that glucocorticoid agonists enhanced retinal stem and progenitor yield and was not due to artifacts Images from the Celigo imaging cytometer showing 96-well plate wells at the end of a 7-day growth assay. Wells were treated with the indicated glucocorticoid agonist compounds. The nuclear channel, the GFP channel and merge demonstrate the ability to differentiate Hoechst and actin-GFP double-positive objects from debris and other artifacts that appear only in the nuclear channel. Visually, it is apparent dexamethasone and prednisolone have greater signal than the other conditions. Red arrows indicate artifacts that fluoresce in the blue nuclear channel that do not fluoresce in the GFP channel. 4x magnification.

214

Figure B2. Representative staining of cell death marker ethidium homodimer (EthD-1) and thymidine analog EdU (A-B) EthD-1 labeling at Day 2 in cells treated with (A) 0.1% DMSO, or (B) 1µM Dexamethasone. (C-D) EdU labeling at Day 4 in cells treated with (A) 0.1% DMSO, or (B) 1µM Dexamethasone.

215

Figure B3. Glucocorticoid agonists increase retinal stem and progenitor cell proliferation only through mineralocorticoid receptor signaling in human Total retinal precursor cell number at the end of a 7-day monolayer growth assay with 0.1% DMSO and 0.1% EtOH in all conditions. Dex treatment increased cell number by 1.22-fold, whereas Dex + Spiro abolished the effect (one-way ANOVA F(5,42)=16.48, p<0.001; Fisher LSD posthoc test, * = p<0.05). N=8 technical replicates. * = significantly different from indicated conditions.

216

Figure B4. Intravitreal dexamethasone injection induces ciliary epithelium proliferation (A-B) Representative images of Pax6 IHC and EdU labeling in the ciliary body of mouse eyes exposed to (A) 0.5% DMSO vehicle, or (B) 1µM Dexamethasone. Nuclei are labeled via Hoechst staining. White arrows indicate Pax6 + EdU co-labeled cells. Dashed line indicates inset. 10 µm-thick sections.

217

Figure B5. EdU-positive cells co-label with endothelial and microglia/macrophage markers (A-B) Representative images of ERG IHC and EdU labeling in the ciliary body of mouse eyes exposed to (A) 0.5% DMSO vehicle, or (B) 1µM Dexamethasone. (C-D) Representative images of CD68 IHC and EdU labeling in the ciliary body of mouse eyes exposed to (C) 0.5% DMSO vehicle, or (D) 1µM Dexamethasone. Nuclei are labeled via Hoechst staining. White arrows indicate co-labeled cells. 10 µm-thick sections.

218

Figure B6. Glucocorticoid receptor, Mineralocorticoid receptor, and 11-β-HSD1 & 2 RNA expression in RSC spheres Transcriptomic data showing the expression of the glucocorticoid receptor (Nr3c1), mineralocorticoid receptor (Nr3c2) and the two 11-β-HSD isozymes in RSC spheres, supporting the finding of retinal precursor sensitivity to GR agonists. This graph was created from RNAseq data collected in Khalili et al. 2018. Two different mouse strains were used to generate RSC spheres that were lysed and high-quality total RNA (RIN: 9–10) was subjected to directional RNA-sequencing library construction from three independent biological replicates per mouse strain. Sequencing was performed using GAIIx (Illumina, Inc., San Diego, CA; www.illumina. com).

219

Appendix C. Chapter 4 Supplementary Information

Table C1. The top 20 differentially express genes in RSC spheres vs. RSC-derived cones.

Genes LogFC Rankscore PValue FDR

Dct 11.59 4.49 3.27E-05 2.00E-03

Slc26a4 9.74 8.1 8.01E-09 3.26E-05

Tmem132d 9.65 7.71 1.94E-08 4.74E-05

Rgr 9.59 7.28 5.30E-08 7.20E-05

Tspan10 9.43 5.25 5.63E-06 9.17E-04

Lypd6b 9.42 7.53 2.95E-08 5.15E-05

Gsta2 9.34 5.37 4.27E-06 8.16E-04

Cyp1a1 9.02 6.01 9.82E-07 3.43E-04

Dpp4 8.76 6.17 6.72E-07 3.04E-04

Lgi1 8.69 6.65 2.25E-07 1.37E-04

Calcb 8.65 6.96 1.09E-07 9.81E-05

Tie1 8.57 6.81 1.54E-07 1.01E-04

Myrip 8.27 6.05 9.01E-07 3.42E-04

Mlana 8.26 8.16 6.97E-09 3.26E-05

Dmp1 8.22 4.38 4.17E-05 2.26E-03

Sv2b 8.05 8.77 1.69E-09 2.06E-05

Krt8 8.01 4.68 2.09E-05 1.68E-03

Ttr 7.99 4.59 2.55E-05 1.86E-03

Ctss 7.96 4.9 1.27E-05 1.37E-03

220

Genes LogFC Rankscore PValue FDR

Tmem27 7.67 5.04 9.10E-06 1.18E-03

Cyp2f2 −12.17 −2.93 1.17E-03 1.32E-02

Kera −12.12 −2.68 2.09E-03 1.90E-02

Dpt −11.82 −3.13 7.41E-04 9.92E-03

Ddx3y −11.54 −3.85 1.42E-04 4.05E-03

Pi15 −10.97 −2.65 2.25E-03 1.98E-02

Eif2s3y −10.57 −3.59 2.56E-04 5.63E-03

Fgf10 −10.17 −2.97 1.08E-03 1.25E-02

Uty −9.82 −3.92 1.20E-04 3.71E-03

Kdm5d −9.78 −3.73 1.85E-04 4.70E-03

Xpnpep2 −9.43 −3.4 3.99E-04 6.92E-03

Sfrp2 −8.73 −2.07 8.57E-03 4.70E-02

Tnn −8.65 −2.37 4.23E-03 2.99E-02

Sfrp4 −8.39 −2.78 1.66E-03 1.67E-02

F830016B08Rik −8.12 −3.22 5.99E-04 8.70E-03

Apod −8.03 −2.68 2.07E-03 1.88E-02

Myh15 −8.03 −2.73 1.85E-03 1.78E-02

Osr2 −7.66 −2.7 1.98E-03 1.83E-02

Rxfp1 −7.63 −2.56 2.73E-03 2.25E-02

Omd −7.53 −2.53 2.93E-03 2.37E-02

Chodl −7.44 −3.05 8.99E-04 1.12E-02

221

Table C2. The top 20 differentially express genes in RSC spheres vs. endogenous cones.

Genes LogFC Rankscore PValue FDR

Tspan10 10.4432427 16.1586498 6.94E-17 5.58E-15

Ttr 10.3087045 15.776313 1.67E-16 1.13E-14

Slc26a4 10.1827813 15.5438996 2.86E-16 1.74E-14

Dct 10.0358862 19.2012938 6.29E-20 2.85E-17

Tie1 9.01111536 12.710142 1.95E-13 4.42E-12

Ctss 8.58186018 13.0004954 9.99E-14 2.47E-12

Rgr 8.56649697 12.4708177 3.38E-13 7.09E-12

Sv2b 8.51907534 18.1884909 6.48E-19 1.57E-16

C1qa 8.45905406 11.4799513 3.31E-12 4.90E-11

Gsta2 8.43424768 17.3848417 4.12E-18 5.86E-16

Krt8 8.13444342 14.5603696 2.75E-15 1.18E-13

Slc4a5 8.00016679 20.2036869 6.26E-21 7.96E-18

Pld5 7.93656486 17.0157094 9.64E-18 1.11E-15

C1qb 7.87732224 10.0862462 8.20E-11 8.03E-10

Lgi1 7.51740836 10.5860699 2.59E-11 2.89E-10

Hkdc1 7.42516697 16.4772783 3.33E-17 2.97E-15

Myrip 7.41011002 9.99642575 1.01E-10 9.61E-10

Gucy2e 7.35547054 10.1974048 6.35E-11 6.42E-10

Tmem27 7.16000309 12.198696 6.33E-13 1.17E-11

Stxbp5l 7.09891972 13.4572497 3.49E-14 1.01E-12

Cyp2f2 −12.093933 −19.015194 9.66E-20 3.69E-17

222

Genes LogFC Rankscore PValue FDR

Dpt −11.991097 −19.943594 1.14E-20 1.07E-17

Pi15 −11.343474 −20.451256 3.54E-21 6.66E-18

Kera −11.291784 −18.549153 2.82E-19 8.42E-17

Xpnpep2 −10.417263 −16.330386 4.67E-17 4.02E-15

Tnn −10.103648 −16.727952 1.87E-17 1.84E-15

Fgf10 −9.8992156 −15.443936 3.60E-16 2.07E-14

Myh15 −8.7061324 −12.840425 1.44E-13 3.38E-12

Osr2 −8.6266181 −13.925633 1.19E-14 4.18E-13

Gdf6 −8.4556046 −13.952177 1.12E-14 4.00E-13

Fmo3 −8.387632 −11.998747 1.00E-12 1.74E-11

Tnfsf11 −8.3208991 −14.327896 4.70E-15 1.88E-13

Apod −8.0655817 −19.535029 2.92E-20 1.88E-17

Sfrp4 −7.9567052 −12.612337 2.44E-13 5.32E-12

4833403I15Rik −7.8724397 −12.459182 3.47E-13 7.25E-12

Bst1 −7.7909023 −12.008915 9.80E-13 1.71E-11

Kcnj15 −7.6961924 −13.740742 1.82E-14 5.81E-13

Crct1 −7.6528833 −11.076532 8.38E-12 1.09E-10

Itga11 −7.5503402 −13.473774 3.36E-14 9.85E-13

Gdpd2 −7.5291817 −13.521224 3.01E-14 9.04E-13

223

Table C3. The top 20 differentially express genes in endogenous cones vs. RSC-derived cones.

Genes logFC Rank score PValue FDR

Kdm5d 4.65 2.61 2.45E-03 1.26E-01

Eif2s3y 4.57 2.42 3.84E-03 1.46E-01

Uty 4.55 2.8 1.59E-03 1.10E-01

Ddx3y 4.13 2.41 3.90E-03 1.46E-01

Wnt9b 3.95 2.05 8.92E-03 1.91E-01

Dlk1 3.63 4.44 3.63E-05 4.62E-02

C1qb 3.6 1.72 1.92E-02 2.56E-01

Bex4 3.44 2.04 9.02E-03 1.92E-01

Sfrp2 3.41 1.08 8.25E-02 4.23E-01

Gm10800 3.06 2.03 9.24E-03 1.93E-01

Megf10 2.96 1.14 7.24E-02 4.06E-01

Gm21738 2.88 1.52 3.00E-02 2.94E-01

Erbb3 2.84 2.06 8.63E-03 1.89E-01

Cntn1 2.83 1.38 4.15E-02 3.35E-01

Ryr2 2.82 1.91 1.23E-02 2.17E-01

BC064078 2.79 1.96 1.09E-02 2.07E-01

Gm10801 2.78 2.21 6.22E-03 1.66E-01

Cwc22 2.78 1.76 1.74E-02 2.48E-01

Alpl 2.7 1.12 7.60E-02 4.12E-01

Khdrbs2 2.65 1.16 6.90E-02 4.00E-01

Crct1 −6.69 −3.65 2.23E-04 6.41E-02

224

Genes logFC Rank score PValue FDR

Nefl −5.99 −4.26 5.51E-05 4.62E-02

Rab3c −5.81 −3.51 3.07E-04 6.70E-02

Col10a1 −5.56 −2.97 1.07E-03 9.94E-02

Xirp2 −5.22 −3.75 1.80E-04 6.41E-02

Aqp5 −4.75 −3.32 4.83E-04 8.43E-02

Myh2 −4.54 −3.57 2.70E-04 6.43E-02

Car8 −4.22 −2.94 1.16E-03 1.01E-01

Lrat −4.21 −1.71 1.93E-02 2.57E-01

Sostdc1 −4.15 −2.78 1.65E-03 1.10E-01

Ivl −4.06 −4.75 1.77E-05 4.62E-02

Cpa4 −4.04 −2.53 2.98E-03 1.35E-01

Myh11 −3.91 −2.98 1.04E-03 9.94E-02

Asb5 −3.79 −3.63 2.35E-04 6.41E-02

Bst1 −3.73 −1.64 2.29E-02 2.71E-01

Gm25911 −3.73 −3.15 7.02E-04 9.01E-02

Lce1h −3.67 −2 9.93E-03 1.98E-01

Dgkk −3.66 −3.37 4.22E-04 8.06E-02

Serpinb9b −3.55 −2.47 3.39E-03 1.38E-01

Cyp1a1 −3.51 −2.58 2.65E-03 1.32E-01

225

226

Figure C1. COCO makes cones but not other cell types (A) COCO dose response curve on RSC and characterization of RSC progeny. 50 ng/mL of COCO produced a plateau level of cells expressing cone arrestin. Data represent means ± SEMs (one-way ANOVA, p<0.001). (B and C) Pigmented progenitors in culture shed their pigment (bright field); however, they express retinal pigment epithelium markers like MITF. (D and E) PKC (bipolar cell) expression in COCO conditions and pan-retinal control (yellow arrow). PKC is expressed in cells in pan-retinal conditions at < 1%, but PKC could not be detected in COCO conditions.

227

Figure C2. Flow cytometry and sorting of pigmented and non-pigmented progenitors, as well as endogenous cones and RSC derived cones. (A) RSC progeny are composed of pigmented progenitors with high side scatter and non- pigmented progenitors with lower side scatter. (B) Isolated cells from neural retina tissue of adult CCDC 136-/- GFP mice were stained for PNA conjugated with Rhodamine red, a specific cone marker, gated and sorted for GFP and Rhodamine. Double positive GFP+PNA+ were approximately 3% and represent cones. (C) Similarly, RSC derived cone progeny (COCO cones) were sorted for GFP and Rhodamine PNA; 89% of the total population were cones (the large cluster to the left contains primarily debris and dead cells). (D) Finally, CCDC 136-/- GFP RSC progeny treated in pan-retinal conditions were sorted as the baseline control; approximately 0.3% of these cells expressed PNA and GFP.

228

Figure C3. Correlation of RSC-derived cones with reference cones at pathway activity level. Using pathway activity analysis (please see methods), we obtained robust correlation results between RSC-derived cones and endogenous cones in our study with a positive and consistent correlation with cone photoreceptors from a reference database.

229