Insights Into Retinal Cell Fate Determination in Vertebrates Using Transcriptomic Profiling and Genome Editing Rebecca Chowdhury Iowa State University

Iowa State University Capstones, Theses and Graduate Theses and Dissertations Dissertations

2017 Insights into retinal cell fate determination in vertebrates using transcriptomic profiling and genome editing Rebecca Chowdhury Iowa State University

Follow this and additional works at: https://lib.dr.iastate.edu/etd Part of the Developmental Biology Commons, Genetics Commons, and the Molecular Biology Commons

Recommended Citation Chowdhury, Rebecca, "Insights into retinal cell fate determination in vertebrates using transcriptomic profiling and genome editing" (2017). Graduate Theses and Dissertations. 16512. https://lib.dr.iastate.edu/etd/16512

This Dissertation is brought to you for free and open access by the Iowa State University Capstones, Theses and Dissertations at Iowa State University Digital Repository. It has been accepted for inclusion in Graduate Theses and Dissertations by an authorized administrator of Iowa State University Digital Repository. For more information, please contact [email protected].

Insights into retinal cell fate determination in vertebrates using transcriptomic profiling and genome editing

Rebecca Chowdhury

A dissertation submitted to the graduate faculty

in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

Major: Molecular, Cellular and Developmental Biology

Program of Study Committee: Jeffrey Trimarchi, Major Professor Jeffrey Essner Clark Coffman Donald Sakaguchi Basil Nikolau

Iowa State University

Ames, Iowa

2017

TABLE OF CONTENTS

ACKNOWLEDGEMENTS..………………………………………………………………………………………………..vi

ABSTRACT..…………………………………………………………………………………………………………………..vii

CHAPTER 1. GENERAL INTRODUCTION..…………………………………………………………………………1

1. Anatomy of the eye……………………………………………………………………………………………………...1

1.1 Anterior section of the eye…………………………………………………………………………………………1

1.1.1 Cornea and iris……………………………………………………………………………………………………….1

1.1.2 Lens……………………………………………………………………………………………………………………….3

1.2 Posterior section of the eye: retina………………………………………………………………………….....3

2. Development of the vertebrate eye………………………………………………………………………………5

2.1 Eye development in mammals………………………………………………………………………………...... 5

2.2 Eye development in zebrafish…………………………………………………………………………………….7

2.3 Factors important for development of the vertebrate eye……………………………………………8

3. Retinal development in vertebrates……………………………………………………………………………10

3.1 Introduction to the retina………………………………………………………………………………………...10

3.2 Competence and cell fate determination…………………………………………………………………..12

3.3 Models of cell fate determination……………………………………………………………………………..13 iii

3.4 Factors involved in retinal cell fate regulation…………………………………………………………..15

3.4.1 Atoh7…………………………………………………………………………………………………………………...15

3.4.2 The NeuroD family……………………………………………………………………...... 17

3.4.3 Dlx1/2………………………………………………………………………………………………………………….17

3.4.4 Ikzf family………………………………………………………………………………...... 18

3.4.5 Other factors………………………………………………………………………………………………………...18

3.4.6 MicroRNAs…………………………………………………………………………………………………………...19

4. Morphology and functions of the retinal neurons...... 20

4.1 Photoreceptors…………………………………………………………………………...... 20

4.2 Bipolar cells…………………………………………………………………………………………………………….22

4.3 Amacrine cells…………………………………………………………………………………………………………23

4.4 Horizontal cells……………………………………………………………………………………………………….24

4.5 Ganglion cells……………………………………………………………………………...... 24

5. Guidance of retinal axons to their target within the brain……………………………………………26

6. Conclusions...... 31

7. References...... 32

CHAPTER 2. SINGLE CELL PROFILING OF ATOH7-EXPRESSING RETINAL CELLS...... 43

Abstract...... 43 iv

Introduction...... 44

Materials and Methods...... 47

Results...... 52

Discussion...... 60

References...... 64

Figures and Legends...... 71

CHAPTER 3. TRIM GENES IN THE RETINA: EXPRESSION AND FUNCTIONAL

CHARACTERIZATION …………………………………………………………………………………………………...82

Abstract...... 82

Introduction...... 83

Materials and Methods...... 86

Results...... 94

Discussion...... 103

Acknowledgements…………………………………………………………………………………………………….106

References...... 106

Figures and Legends...... 113

CHAPTER 4. USING GENOME EDITING TO INVESTIGATE VERTEBRATE RETINAL

DEVELOPMENT IN ZEBRAFISH……...... 125

Abstract...... 125

Introduction...... 126

Materials and Methods...... 130

Results……………...... 136

Discussion…………………………………………………………………………………………………………………..145

Acknowledgements…………………………………………………………………………………………………….149

References...... 150

Figures and Legends...... 159

CHAPTER 5. GENERAL DISCUSSION...... 170

Introduction……………………………………………………………………………………………………………….170

Retinal transcriptomics in Atoh7+ cells………………………………………………………………………..172

Functional studies using Trim9 mouse mutants…………………………………………………………...173

Functional studies using zebrafish mutants………………………………………………………………….174

Conclusion………………………………………………………………………………………………………………….177

References……………………………………………………………………………..…………………………………...179

ACKNOWLEDGEMENTS

I would like to thank my committee members, Dr. Jeffrey Essner, Dr. Clark Coffman,

Dr. Don Sakaguchi, and Dr. Basil Nikolau for their insights, guidance, and support throughout the course of this research.

I am very grateful to my mentor, Dr. Jeffrey Trimarchi. His guidance every step of the way, patience and unwavering support has made this possible. He was someone I have always been able to rely on whenever I needed help. His immense knowledge, enthusiasm and motivation has made this a rewarding experience for me.

I would also like to thank members of the Trimarchi lab, who played instrumental roles in my work. Jillian Goetz taught me a lot when I first joined lab. Lauren

Laboissonniere, was an amazing colleague and friend, and someone who I have always been able to discuss anything and everything with. I would also like to thank present and past undergraduate students who were instrumental in my projects. Courtney Smith, Abbie

Burney, Bailey Mooney, Jackie Mesenbrink, Amy Stark and all the other undergraduate students in our lab have made my time here a wonderful experience.

I would like to thank my friends, Sudhanya Banerjee, Shreyosree Chatterjee and

Agnivo Gosai, and others who were always there for me, and have given me some wonderful memories. Finally, I am grateful to my parents for their constant encouragement and support throughout my years of study, and life in general. vii

ABSTRACT

Deciphering the mechanisms of development of retinal neurons is not only of immense interest to developmental biologists, but is also vital for regenerative therapeutic applications. To attain this goal, it is critical to understand how specific intrinsic factors control cell fate decisions and neuronal maturation processes. In the retina, Atoh7 is a highly conserved transcription factor that is essential for retinal ganglion cell development in the developing mouse and zebrafish. Atoh7 labels a subset of cells in the developing retina that are progressing from a progenitor to a differentiated state. To capture cells during the window when the cell fate decision is made, we performed transcriptome profiling of Atoh7+ individual cells isolated from mouse retina, thereby obtaining a wealth of information about how distinct types of retinal neurons are produced. Our efforts also led us to uncover several genes whose expression closely tracked with that of atoh7. These genes we believe will play important roles in retinal cell fate choice or early neuronal maturation. We further assayed the expression of these genes and found them to be expressed in the developing mouse and zebrafish retina by in situ hybridization. One of these genes was Trim9, an E3 ubiquitin ligase, that plays a highly significant role in axon branching and guidance in the mouse brain. Given the correlation of expression of Trim9 with Atoh7, we hypothesized that this gene may play roles in driving cell fate acquisition and neuronal differentiation in the vertebrate retina. However, examination of mice and zebrafish harboring mutations in Trim9 failed to reveal a retinal phenotype. We have also used CRISPR-Cas9 mediated genome editing in zebrafish to mutate several additional genes chosen from our single cell transcriptomic data. From our experience, single cell viii transcriptomics combined with CRISPR-Cas9 mediated genome editing, despite some potential hurdles, is a powerful approach for the study of gene networks governing cell fate decisions. 1

CHAPTER 1. GENERAL INTRODUCTION

1. Anatomy of the Eye

1.1 Anterior section of the eye

The eye is a complex structure that allows an organism to interact with its environment. It can roughly be divided into two separate structures, the anterior segment and the posterior segment, each of which plays a distinctive role. The anterior segment of the eye consists of the cornea, lens, and the iris, while the posterior segment includes the retina and the optic nerve.

1.1.1 Cornea and iris

The cornea, a dome like structure located in the front of the eye, provides 75% of the focusing power of the eye1. The structures comprising the cornea are very specifically arranged in order to support its function of transmitting and focusing light into the eye1.

Ninety percent of the cornea is made up of stroma- evenly arranged collagen fibers that provide both transparency and strength across the extent of the cornea1. The remaining

10% is composed of the epithelium and the Bowman’s layer in the front and the endothelium and Descemet’s membrane at the back1. The epithelium of the cornea acts as barrier to bacteria and other pathogens and is also responsible for maintaining the hydration of the stroma by preventing fluid entry through its tight junctions as well as by pumping out a small quantity of fluid2. The Bowman’s layer is a thin membrane of unknown function located directly beneath the epithelium2. The back of the cornea is 2 surrounded by a thin tissue known as the Descemet’s membrane, which is thought to have a protective function1. The endothelium is a single layer tissue at the posterior of the cornea that supplies nutrients from the aqueous humor to the cornea while actively pumping fluid out of the cornea, thus maintaining the hydration balance of this tissue3,4.

The aqueous humor fills the anterior chamber of the eye between the cornea and the front of the crystalline lens4. It is a transparent fluid produced by the ciliary body. One of its functions is to supply nutrients to the cornea and the crystalline lens4. It is also responsible for bending the light entering the eye through a difference in refractive index in conjunction with the cornea and the lens. The aqueous fluid needs to be constantly circulated in order to maintain the flow of nutrients to the cornea4. It flows out of the anterior chamber through the trabecular meshwork drainage system lying behind limbus between the cornea and the anterior iris4. The pressure within the eye is maintained at approximately 15mmHg through the resistance offered to the flow of aqueous fluid at the trabecular meshwork5. Too little resistance causes the eye to lose its shape whereas too much leads to an increased pressure within the eye leading to a condition known as glaucoma5. The iris, which is responsible for giving the eye its color, is a muscular pigmented structure visible through the cornea6. All irises have a dark pigmented posterior layer. However, the varying amount of pigments in the anterior or stromal layer is responsible for the characteristic color of the eye. The main role of the iris is to control the amount of light entering the eye by controlling the iris aperture, or pupil, to let in different amounts of ambient light6. The two opposing muscles, the sphincter and the dilator muscles, constrict and dilate the pupil and are innervated by the sympathetic and 3 parasympathetic nerves, respectively7. This explains why pupils dilate when danger is sensed and the sympathetic nervous system is activated during a ‘flight or fight’ response7.

1.1.2 Lens

The lens is a transparent structure that along with the cornea is responsible for focusing light entering the eye6. The lens has the capability of changing shape to modulate the refraction of the light entering the eye. The regularly spaced elongated fiber cells that comprise the lens maintain its transparency8. The fibers of the lens are made up of water soluble structural proteins called crystallins8. These crystallins give the lens its higher refractive power compared to the aqueous and vitreous humors8. The lens is surrounded by an elastic extracellular matrix known as the capsule9. This provides a smooth optical surface as well as an anchor for the suspension of the lens within the eye9. The lens connects to the ciliary muscle by a meshwork of non-elastic microfibers or zonules4,9. The contraction and relaxation of the ciliary muscles controls the adjustment of the curvature of the lens to determine the focal power of the lens9.

1.2 Posterior section of the eye

The retina has emerged as a powerful tool for studying the central nervous system and has been intensively investigated for over a century10. The retina, a relatively simple outgrowth of the developing brain, is a tissue where total control of the input signal can be achieved during study11. The different cell types of the retina are shown in Fig. 1. Light acts a stimulus for the photoreceptor cells, the rods and the cones12. The photoreceptors synapse with the interneurons, the bipolar and horizontal cells12. Further transmission to the retinal ganglion cells, modulated by the amacrine interneurons, results in the 4 processing of this information to various locations within the brain12. Retinal cells are generated sequentially with the ganglions cells (RGCs) being born first in most vertebrates.

The RGCs are followed by the cones, horizontal cells, and amacrine cells13–16. These neurons comprise the early born cells of the retina13–16. The bipolar cells and Müller glia are born later in development, while the rods continue to be generated throughout development13–16. The high degree of overlap in the generation times of the different retinal neurons rules out a simple model of development where the cells are born in a discrete sequential order13,16–19.

Figure 1. Schematic showing the organization of the vertebrate retina20. The different layers of the retina are indicated here. OS refers to the outer segments of photoreceptors, ONL to the outer nuclear layer, OPL to the outer plexiform layer, INL to the inner nuclear layer, IPL to the inner plexiform layer, and GCL to the ganglion cell layer. The photoreceptors are shown in green, bipolar 5 cells in blue, horizontal cells in orange, amacrine cells in yellow, ganglion cells in red and Müller glia in blue.

Vertebrate retinas are composed of three layers of neuronal cell bodies and two layers of synapses (Fig.1)12,21. The outer layer (ONL) is composed of the cell bodies of the photoreceptors- the rods and cones, the inner nuclear layer (INL) consists of the cell bodies of the bipolar, horizontal amacrine cells and Müller glia, and the ganglion cell layer contains cell bodies of ganglion cells and displaced amacrine cells12,21. These three layers are separated by the neuropil layers where synaptic contact occurs20. The region containing synaptic connections between the photoreceptors and the horizontal and bipolar cells is known as the outer plexiform layer (OPL)18,22. The inner plexiform layer (IPL) is the union between the processes of the interneurons and the ganglion cells18,22. The visual message from the photoreceptors is transmitted to the ganglion cells by virtue of neural processing in the IPL and then communicated to the brain through the optic nerve23.

2. Development of the Vertebrate Eye

2.1 Eye development in mammals

An organism interacts with its environment through its sensory organs, such as the eye. Soon after gastrulation, the eye primordium, or eye field containing all the progenitors of future neural-derived eye structures, is specified in the medial anterior neural plate24. At gastrulation, the involuting endoderm and mesoderm interact with the future head ectoderm to favor lens formation25. The lens forming ability of the head ectoderm is 6 activated and positioned with respect to the prospective retina by the optic vesicle25. The optic vesicle extends from the diencephalon and meets the head ectoderm to induce the formation of the lens placode, which subsequently invaginates to form the lens25. The optic vesicle itself becomes the two-walled optic cup, the layers of which develop to be functionally different. The cells of the outer layer are able to produce melanin and ultimately give rise to the pigmented retina25. The cells of the inner layer generate a collection of six different neuronal cell types and one Müller glial type; these cells constitute the neural retina. The axons of the ganglion cells meet at the base of the eye and travel down the optic stalk, which is then called the optic nerve25. During the process of developing into a lens, the lens placode rounds up and makes contact with the overlying ectoderm. The lens vesicle then induces the ectoderm to form the cornea26. The cells in the inner part of the lens vesicle elongate and, aided by the neural retina, become the lens fibers26. Intraocular fluid pressure within the cornea is important for maintaining its shape so that light passing through can be focused on the retina3. The cornea is made functionally competent through the synthesis of a transparent, lens-specific protein called crystallin by the lens fibers6. The cells in the anterior part of the lens vesicle constitute a germinal epithelium where they continue to divide. Some of the dividing cells migrate towards the equator of the lens vesicle where they begin to elongate6. Thus, the lens consists of three distinct regions, an anterior region of dividing epithelial cells, an equatorial zone of elongated cells and a central-posterior zone of crystallin containing fiber cells6. The iris, located directly in front of the lens, is a pigmented muscular tissue that is responsible for controlling the size of the pupil4. It has origins in both ectodermal and mesodermal tissues23. The optic vesicle thus gives rise to the optic cup and ultimately the retina, the 7 retinal pigmented epithelium (RPE), the ciliary epithelium and the iris23. The ciliary marginal zone that lies between the retina and the ciliary epithelium is a region of stem cells that is capable of producing new neurons and glia throughout life in some fish and amphibians23.

2.2 Eye development in zebrafish

During development, neurulation in teleosts such as the zebrafish Danio rerio occurs differently to mammals. Instead of the neural tube, the primordium in teleosts adopts a rod-like shape known as neural keel. Accordingly, instead of optic vesicles, equivalent structures named optic lobes become evident at about 11.5 hours post fertilization (hpf)27.

At about 13 hpf, the posterior portion of the optic lobe becomes detached from the brain, while the anterior portion remains attached. Over the next few hours, the optic lobe transforms into the optic cup27. At the beginning of the second day, the optic cup consists of the pseudostratified columnar neuroepithelium (rne) and the cuboidal pigmented epithelium (pe). At about 24hpf, melanin granules form in the cells of the pigmented epithelium28.

At this stage, nuclei of cells about to divide migrate to the apical surface of the neuroepithelium28,29. From here newly formed nuclei migrate to more basal locations.

Mitotic divisions are observed at the apical surface of the neuroepithelium almost exclusively until about 40 hpf, when mitotic divisions start being detected in the INL29. The ganglion cells of the retina are the first cell type to become post-mitotic at about 27hpf29.

This early onset of RGC differentiation is conserved across retinas of many vertebrate species29–31. A rudimentary RGC layer is distinguishable by 36hpf29. At this time, 8 approximately 10 hours after the onset of RGC differentiation, the cells of the INL become visible29. By 34hpf or earlier, terminal divisions of retinal progenitor cells (RPCs) give rise to RGC and PR pairs32. At around 60hpf, most of the retinal neurons are post-mitotic and the different layers of the retina are distinguishable29.

2.3 Factors important for development of the vertebrate eye

During gastrulation in mammals, the developing eye is organized into a centrally located single eye field in the forebrain, failure of which leads to phenotypes of anophthalmia. Thus, formation of the eyefield is an important aspect of development6. The eyeless mutation is a spontaneous point mutation in the homeobox gene, Rax33. The zebrafish ortholog of the Rax gene is rx3, where loss of function mutations lead to a similar phenotype where the RPE and the retina are missing, though a reduced lens still forms34.

These results suggest that Rax/rx3 is important for early eye development in mouse and zebrafish. The mouse Rax gene is expressed in the anterior neural plate at embryonic day

(E) 8.5 and restricted to the optic sulci and a narrow band of cells in the ventral forebrain35.

In zebrafish, Rx3 appears to control the segregation of telencephalic and eye-field entities in the forebrain in a cell autonomous manner34. Mutations in the human RAX gene also lead to an eyeless phenotype36.

Studies in the frog Xenopus and zebrafish have established the importance of Wnt signaling in the formation of the eye field37. The Tcf3 mutants, masterblind and headless, showed that canonical Wnt signaling inhibits eye formation. In contrast, overexpression of

Wnt pathway receptor Fzd3 in Xenopus embryos resulted in overexpression of pax6, rx and otx2 and finally in ectopic eye formation37. In Xenopus, a group of transcription factors 9 collectively named ‘eye field transcription factors’ (EFTFs) specify the eye field during the neural plate stage. This group includes Otx2, Six3, Rax, Pax6, Lhx2, Tbx3, Nr2e1 and Optx2.

Expression of the EFTF cocktail in Xenopus embryos leads to ectopic eyes outside the nervous system at high frequency. However in mammals, only mutations in Pax6 and Lhx2 show effects in early eye development6. There is considerable heterogeneity between species in terms of gene function. For example, in Drosophila, an Eya knockout leads to absence of compound eyes due to progenitor cell death38. However, the four Eya genes

(Eya1-4) have not been reported to show such drastic phenotypes in mouse. While Eya3 is associated with survival and proliferation of neural progenitor cells in the Xenopus39, no such role could be shown in mouse eyes40.

The formation of the eye field is followed by splitting of the eye field. This event occurs simultaneously with the introduction of the midline and is mainly controlled by the

Sonic hedgehog pathway. Sonic Hedgehog (Shh) along with another gene, Six3, are expressed in the single eye field6. Functional studies in the mouse indicate that Six3 represses Wnt signaling to allow the division of the eye field to occur6. In the zebrafish, cyclopic mutants arising due to mutations in the cyc gene are characterized by loss of medial floorplate and severe deficits in ventral forebrain development including cyclopia due to incomplete splitting of the eye field6. The cyc gene is thus necessary for splitting of the eye field in zebrafish.

The next step in this process is the transformation of optic vesicle to optic cup and one of the key players here is the retinoic acid (RA) signaling system41. Paracrine RA signaling generated by Retinaldehyde dehydrogenase (Raldh2) is required for both lens pit 10 and optic cup invagination. Mouse Raldh2 mutants lacking RA synthesis exhibit a failure in optic cup development42. The transcription factor, Lhx2 is also a key player in the transition from optic vesicle to the optic cup25. In mice lacking Lhx2, eye field specification and optic vesicle morphogenesis occurs but development is arrested prior to optic cup formation25. Yun et al. (2009)43 showed that Lhx2 is a central link in a pathway involving bone morphogenetic protein (BMP) signaling leading to optic cup formation.

3. Retinal Development in Vertebrates

3.1 Introduction to the retina

The neural retina is composed of six neuronal cell types (the rod and cone photoreceptors, horizontal cells, bipolar cells, amacrine cells and the retinal ganglion cells) and Müller glial cells13,19. The retinal neurons are generated sequentially at distinct, but overlapping, timepoints during embryonic development44. To establish the order of generation of cell layers, both lineage tracing studies and birthdating techniques using 3H- thymidine have been used13,15–17,45. These studies have established that the RGCs are generated first, closely followed by the cone photoreceptors, the horizontal cells and the amacrine cells13,15–17. These cells comprise the early born neuronal cells. The remaining cell types, the rod photoreceptors, the bipolar cells and Müller glia are born at later stages and comprise the late-born retinal neurons13,16,17. The schematic (Fig.2) shows the order of birth of the retinal neurons in mouse. 11

Figure 2. Schematic showing the order of generation of the six types of retinal neurons and Müller glia20. The approximate birth dates are indicated on the x-axis. RPCs are shown in grey and colors representing the different types of neuronal and glial cells are indicated below the x-axis. The degree of shading of each color roughly corresponds to the proportion of cells born at that time point. E refers to embryonic, and P to post-natal.

Lineage studies have shown that retinal progenitor cells (RPCs) are multipotent during development, that is, they are capable of giving rise to more than one cell type30,31,46.

These progenitor cells show transient biases in the proportion of cell types they generate, as seen through comparisons of the population of daughter cell types generated at either embryonic or postnatal time points20. Mouse RPCs injected with retrovirus to track their fate during early embryonic development were shown to produce all the retinal cell types31. Similar retroviral injections in rat performed at birth, labeled these RPCs as rods, bipolar cells and Müller glia, the late born cells of the retina30. The multipotency of RPCs can extend to the last the last cell division. Both during early and late stages of 12 development, RPCs can give rise to two-cell clones consisting of two different cell types30,31.

For example, in the rodent retina, RPCs can give rise to two-cell clones consisting of rod photoreceptors and Müller glia within the terminal cell division30. Thus the current understanding is that multipotent RPCs can give rise to distinctive cell types, born at about the same time20,44.

3.2 Competence and cell fate determination

The data from the experiments outlined above suggest that RPCs are capable of giving rise to the different cell types of the retina in overlapping windows of time, but in a conserved order of genesis19,44. How does this process occur? One of the hypotheses put forward was that extrinsic signals may determine the fate of these multipotent RPCs19,44.

However, while extrinsic signals can influence the proportion of each cell type generated, retinal progenitors are capable of producing only a subset of cell types during a particular window of time and extrinsic cues cannot influence RPCs to produce temporally inappropriate cell types47–49. These and other findings led to the competence model of retinal development, where RPCs pass through a series of competence states during which they become capable of giving rise to a subset of cell types19,44. Although the role of environmental signals is undeniable in determining the cell fate of a neuron within a certain competence state, the competence of an individual RPC is largely regulated at the level of gene and protein expression19. Thus, at any particular developmental stage, the capability of an RPC to generate a particular neuron is believed to be determined by intrinsic regulatory gene expression patterns20, and within a particular competence 13 window, extrinsic cues have the ability to influence the generation of a particular cell type44.

3.3 Models of cell fate determination

Exactly how cell fate determination is controlled is not precisely understood. There are two different models that have been proposed: first, a deterministic model suggests that progenitor cells are destined to produce certain subsets of retinal cell types at defined times, and this determination does not change20. For example, a particular progenitor may be fated to yield a ganglion and amacrine cell, while another may be pre-determined to produce a cone and rod photoreceptor20. The nematode, C. elegans represents the most extreme example of this type of developmental model. Studies have mapped out the specific lineages that the different neurons derive from50. While, cell fate determination in most vertebrates is known not to be as extreme, it has been presumed that cells enact specific gene expression programs which then leads to the adoption of a particular cell fate20. Progenitors are thought to be intrinsically different in terms of their competence to produce distinct cell types during different stages of development19. This was evident from heterochronic transplant experiments in chicks and rodents, where progenitor cells from certain stages of development were placed in an environment of a different age48. For example, early chick progenitors continued to produce ganglion cells, regardless of the environment they were placed in48. Similarly, rat RPCs that normally produce amacrine cells and cone photoreceptors generated the same cells when cultured in a different environment47. These experiments provide evidence for a variation of the deterministic 14 model, where the competence of the progenitor cells changes over the developmental period, but the intrinsic programs enacted by these cells would be the same for any particular cell type20.

Second, a more stochastic or probabilistic model of cell fate determination has been proposed that is a dynamic process controlled by changing transcriptional programs and differential gene expression where expression is not restricted in defined cells and times51–

53. This model has been elegantly demonstrated in the rat and zebrafish retina51,53.

Mathematical modeling combined with live microscopy to visualize late rat RPCs in cell culture showed that these RPCs randomly choose between three modes of division – two daughter progenitor cells (PP), a progenitor and a differentiating daughter cell (PD) and two differentiating daughter cells (DD). These experiments were further supported by lineage tracing experiments in zebrafish, where an inducible photoconvertible fluorescent protein was used to monitor the progression of individual RPCs through the cell cycle51.

These experiments led to a mathematical model, where RPCs choose between one of PP, PD and DD modes of division in a stochastic manner, and as development progresses, RPCs shift from a PP to PD bias51,52. The very large variety of clone types observed, their compositions, and division patterns lend credibility to a stochastic model of cell fate determination51,52. It has been further suggested that many of the molecular differences seen in RPCs54 may not be programmed but rather are the result of cycling or stochastic fluctuations in gene expression51,55. This model fits in well with the observations from transcriptional profiling of single cells isolated from the developing mouse retina54

(Chowdhury et al., in preparation, Chapter 2). The variability of gene expression in single cells, and the expression multiple of genes related to several different cell types within an 15 individual cell, would indicate that gene expression in these cells is occurring dynamically and RPCs stochastically make cell fate choices based on this variable gene expression56

(Chowdhury et al., in preparation, Chapter 2).

3.4 Factors involved in retinal cell fate regulation

3.4.1 Atoh7

Atoh7 is a bHLH transcription factor, homologous to the Drosophila atonal protein.

Atonal is essential for the specification of the initial photoreceptor (R8) during Drosophila eye development. Atoh7 and its homologues (Math5 in mouse, Xath5 in Xenopus, ath5 in zebrafish, etc) are expressed in subsets of RPCs across multiple organisms57–59. In mouse,

Atoh7 has been shown to be expressed in progenitor cells that gave rise to the early born retinal neurons60,61. Overexpression of frog Xath5 was shown to significantly increase the number of ganglion cells in the retina, while decreasing the number of amacrine cells, bipolar cells, and Müller glia59. Misexpression of either Math5 or the chicken homolog

(Cath5) in developing chicken retinas did not lead to an increase in the number of cells expressing ganglion markers62. Although there appears to be disparities between these results, it is nevertheless safe to say that Atoh7 is an important player in early retinal development with contributions to cell fate decisions as well as differentiation of RGC.s To further investigate the role of Atoh7 in retinal development, Atoh7 deficient mouse and zebrafish models were generated. It was shown that loss of Atoh7 consistently led to a loss of 80-90% of RGCs in mouse61,63–65. This phenomenon was also observed in zebrafish, where a complete loss of RGCs was found in Atoh7/lakritz mutants66. Additionally, changes in cell fate were also observed. Studies reported an increase of cone photoreceptors and 16 amacrine cells in mouse63,64, and overproduction of amacrine cells, bipolar cells and Müller glia in zebrafish66. Taken together these studies demonstrate that Atoh7 is important for the development of multiple cell types in the mouse retina, specifically pointing to the early born cell types.

Recent studies aimed at dissecting the specific role of Atoh7 in the retina showed that expression of related bHLH factors, NeuroD1 and Math3, under the Atoh7 promoter leads to a partial rescue of the ganglion cell population up to 40% and 10% respectively67.

Along the same lines, expression of Atoh7 under the NeuroD1 locus also reprograms developing amacrine cells into ganglion cells68. However, this property of Atoh7 is not absolute as it fails to convert rod precursors to ganglion cells when expressed under the rod-specific Crx promoter69. Replacement of Math5 with a different bHLH, Ascl1, failed to rescue ganglion cell production70. Thus, Atoh7 acts permissively to allow the development of the ganglion cells under certain conditions.

A recent study showed that two factors downstream of Atoh7, Pou4f2 and Isl1 are sufficient for development of mouse RGCs in the absence of Atoh771. These RGCs were essentially normal, could functionally integrate into the visual circuitry and were composed of distinct subtypes. The authors postulated a model where Atoh7 activates a core group of transcription factors including Pou4f2 and Isl1, which can then function autonomously to activate the gene expression program required for RGC differentiation71. Expression of

Pou4f2 and Isl1 in the absence of Atoh7 can activate a substantial, though incomplete part of the RGC developmental program71.

3.4.2 The NeuroD family

The NeuroD family of bHLH-transcription factors is also important in retina development. Overexpression of Neurod1 in the mouse retina leads to stimulated production of rod photoreceptors at the expense of Müller glia72. Neurod1 deficient mouse show only a slightly increased production of bipolar cells and a delay in amacrine cell production72. However, numbers of amacrine cells are restored to normal by adulthood72.

Loss of NeuroD2 in mice caused a decrease in AII amacrine cells while overexpression resulted in an increase in amacrine cells at the expense of bipolar cells and

Müller glia73. Loss of another family member, Neurod6, also led to change in cell fate, where production of glycinergic amacrine cells was stimulated at the expense of GABAergic amacrine cells74.

3.4.3 Dlx1/2

Homeodomain containing factors are also important for retinal cell fate acquisition.

Deficiency of Distal-less homeobox proteins (Dlx1/Dlx2) led to a loss of a subset of ganglion cells in the retina75. Recently, a complete loss of RGCs was observed in the

Dlx1/Dlx2/Brn3b triple knockout mouse, with a significant increase in amacrine cells in the ganglion cell layer (GCL)76. Dlx1 and Dlx2 were also found to act as transcriptional activators of Brn3b expression76. It was further suggested that Dlx1/2 may act downstream of Atoh7 in cooperative pathways to regulate Brn3b, indicating that this pathway was an important regulator of RGC fate76.

3.4.4 Ikzf family

Ikaros, or Ikzf1 is the founding member of the Kruppel zinc finger family of transcription factors. Studies in the mouse retina show that Ikzf1 is expressed in early, but not late

RPCs77. Misexpression of Ikzf1 in late mouse RPCs is sufficient to confer competence to the cells leading to generation of early born cell types such as RGCs, horizontal cells, amacrine cells and cones77. In contrast to Atoh7, which is expressed in RPCs exiting the cell cycle

(Chowdhury et al., in preparation, Chapter 2), Ikzf1 is expressed in early progenitor cells.

The observation that Atoh7 is reduced in the Ikzf1 knockout retinas suggests that this gene is expressed upstream of Atoh7. Although Ikzf1 appears to confer competence for generation of early born cell types, the cone photoreceptors are not affected by loss of this transcription factor77.

3.4.5 Other factors

Drosophila is a widely used as a model system to understand mechanisms of competence and cell fate. In the developing Drosophila embryo, the transcription factors,

Hunchback, Kruppel, Pdm, Castor and Grainyhead are expressed sequentially in an invariant order in the embryonic neuroblast78–82. These temporal competence factors together specify the neuroblast progeny to earlier and later cell fates. The parallels shared between

Drosophila and vertebrate neural progenitors, especially the sequential birth of the neuronal types, suggest that the molecular mechanisms governing eye development may be similar. Investigation into the role of Casz1, the ortholog of Castor in mouse retinal development, found that conditional deletion of Casz1 in RPCs leads to an increase of not only the early born retinal neurons but also the Müller glial cells, the last cell type produced 19 by RPCs83. Consistently, viral transfection of Casz1 in early RPCs leads to a reduction of early born neurons and Müller glia, indicating that Casz1 suppresses both early and late cell fates and promotes the production of the mid-phase cells, the rods and bipolar cells. In

Drosophila, Hunchback represses the expression of Castor and promotes early-born cell fates. Similarly, in mouse, Ikzf1 acts upstream of Casz1, and indirectly regulates its expression84. These and other observations lead to the supposition that in vertebrates, temporally expressed transcription factors bias cell fate decisions rather than rigidly specifying one cell fate, in contrast to the mechanism of cell fate determination in

Drosophila neuroblasts84.

3.4.6 MicroRNAs

MicroRNAs (miRNAs) are small, noncoding molecules of 18 to 24 nucleotides that participate in the regulation of gene expression in a wide variety of tissues and cell types85. miRNAs have been recently shown to play a role in translational control86. Numerous miRNAs are expressed in both mammalian and non-mammalian retinas85–87. The expression of Xenopus Ikaros, otx5b, vsx1, and otx2 is regulated at the translational level.

This is evidenced by the hindrance in the sequential protein expression when the miRNA processing enzyme, Dicer, is knocked down87. A set of four miRNAs were shown to control the timing of bipolar cell genesis through an effect on cell cycle length rather than the developmental stage87. Other effects of loss of miRNA function include extended production of ganglion cells beyond the normal time frame and absence of expression of late stage RPC markers, Sox9 and Mash1, among others. 20

Loss of miRNA function in the embryonic mouse retina also interferes with timing of neurogenesis86. In Dicer1 mice mutants, RGCs continue to be generated beyond the normal period of genesis compared88,89. Dicer1-deficient RPCs downregulate the late RPC markers

Sox9 and achaete-scute complex homolog 1 (Ascl1/Mash1), suggesting that DICER1 processes miRNAs that regulate particular progenitor characteristics88. However, the retinal phenotypes of mouse Dicer1 mutants are complex and eventually result in cell death and degeneration; therefore, further studies examining specific miRNAs are required to tell us more about their roles in retinal development86.

4. Morphology and Functions of Retinal Neurons

4.1 Photoreceptors

The mammalian retina is composed of two (for example, in mice90) or three (such as in humans) cone photoreceptors and one rod photoreceptor91. Non-mammalian retinas like in the zebrafish or the chick can have 4 or 5 different cone types91. The cell bodies of cone photoreceptors are located in a single row directly below the outer limiting membrane with their inner and outer segment extending in the subretinal space towards the RPE91,92.

In contrast, rods are elongated rod-like structures whose inner and outer segments fill the area between the larger cones in the subretinal space92,93. Apical processes from the RPE cloak the outer segments of both rods and cones92.

A photoreceptor is composed of a membranous stacked outer segment containing pigments like rhodopsin, an inner segment containing cell organelles which is the site of 21 assembly of the cone opsins, a cell body containing the nucleus, and a synaptic terminal for neurotransmission to the interneurons92,94. Vertebrate photoreceptors respond to light with the help of a visual pigment present in bilipid membranes of the outer segment92. This visual pigment is composed of a chromophore called 11-cis retinal, a derivative of vitamin

D, and a larger protein called opsin92. During light absorption, the energy from photons converts cis-retinal to an all-trans conformation, leading to a conformational change in the protein, opsin. The light activated opsin in turn activates transducin, a G-protein present in cytosolic surface of the membrane92. This results in the hydrolysis of cGMP and closing of a cGMP gated channel. In the dark, this channel remains open, allowing a steady flow of Na+ ions and causing the photoreceptors to remain depolarized. Depolarization of the photoreceptor leads to the release of a neurotransmitter, glutamate, at its synaptic terminals91. When light is present, the cascade of events described above ensues leading to hyperpolarization of the neurons and a stop to the release of neurotransmitter91.

Rods are more sensitive than cones and are able to detect low levels of light92.

Rhodopsin, the pigment present in rods is sensitive to blue-green light of approximate

500nm wavelength91. Cones contain cone opsins as their visual pigments and are of different types. The short wave cones, or S-cones, are sensitive to blue light, the middle wave to green (M-cones) and the long wave (L-cones) to red light91. This ability to detect different wavelengths of light is the reason for color perception92. While most mammalian species are dichromatic (containing two types of cones), primates, humans, birds, reptiles and fish may be trichromatic, tetrachromatic and even pentachromatic91,92.

4.2 Bipolar cells

Bipolar cells can either connect to rods or cones and have been named rod bipolar cells and cone bipolar cells, respectively. Although rods outnumber cones by far, the relative simplicity of the rod circuit compared to the cone circuit necessitates that there are more types of cone bipolar cells than rod bipolar cells12,95. It has recently been established that there are 11 types of cone bipolar cells in the mammalian retina compared to one type of rod bipolar cell96,97. In zebrafish, 17 types of bipolar cells have been identified98. Cone bipolar cells are fundamentally of two types, the ON-center bipolar cells and the OFF-center bipolar cells97. Werblin and Dowling99 postulated that ON-center bipolars are depolarized by stimuli positioned in the middle of the receptive field while OFF-center bipolars are hyperpolarized. Both types are depolarized by light stimulation of the peripheral receptive field99. Thus, when photoreceptors are depolarized and release glutamate, the OFF bipolar cells are active. In the presence of light, when the photoreceptors hyperpolarize and cease to release glutamate, the ON bipolar cells depolarize20. OFF and ON bipolar cells occur in almost equal numbers and can be further differentiated on the basis of their stratification in the IPL. In general, the ON bipolar cells terminate in the vitreal portion of the IPL (ON- sublamina) and the OFF bipolars terminate in the scleral half of the IPL. Even within the ON and OFF sublayer the axons of bipolar cells terminate at different levels, suggesting they synapse with different sets of neurons95.

4.3 Amacrine cells

Amacrine cells are the most diverse neuronal type in the retina20 with a wide repertoire of functions. There are about 30 types of amacrines present in the mammalian retina12. Seven types of amacrine cells have been identified so far in zebrafish100. Amacrine cells are generally inhibitory using neurotransmitters such as -amino butyric-acid (GABA) and glycine101. Amacrine cells can receive input from bipolar cells and other amacrine cells and relay that information to bipolar cells, ganglion cells or other amacrine cells102. Thus, they are actively involved in feed-forward inhibition, feedback inhibition and lateral inhibition. Many amacrine cells also carry out some degree of vertical integration, that is their presence throughout the IPL means they carry ON information to the OFF strata and vice versa101,102. Amacrine cells also create contextual effects for ganglion cells such as the classic center sound antagonism and other subtle effects such as object motion detection103. In general, object motion sensitive ganglion cells respond specifically to differential motion. These RGCs fire rapidly upon detection of movement in the receptive field center103. The trajectory of background motion is detected by amacrine cells and results in inhibitory signaling103. In case of global motion, the inhibition by amacrine cells quenches the excitatory signal from bipolar cells, causing the ganglion cells to remain silent103. When center motion differs from background motion, in the case of differential motion, the excitatory signals are not quenched and thus the ganglion cells fire leading to detection of local motion12,103.

4.4 Horizontal cells

Horizontal cells in general comprise a small population of the retina’s interneurons, approximately 5% of the INL104. In mammals, horizontal cells are of one (for example, in mice) or two morphologically distinct types in most terrestrial mammals105,106 that feedback on to both rods and cones. In the zebrafish, there are thought to be 4 types of horizontal cells107. In most species, one of the horizontal cell types possesses an axon with fine branched terminal that feeds back to rods95. Horizontal cells provide inhibitory feedback to rods and cones thus providing a mechanism to gain local control of the retina12,108. The task carried out by horizontal cells can be described as lateral integration whereby these cells can measure the level of illumination on the retinal surface by means of its wide spread12. These cells then subtract a proportionate value from the output of photoreceptors12. This serves to regulate the signal input to the inner circuitry of the retina12. The widely spreading processes of the horizontal cells allow their feedback signal to overshoot the edges of a bright object, which means the signal from neighboring darker objects is reduced as well109,110. This compensated for by mechanisms in the inner retina by created both simple and sophisticated contextual effects12.

4.5 Ganglion cells

Ganglion cells, the primary output neurons of the retina, convey visual information to the brain by means of a complex information processing system. Previous studies indicate that there are at least 25 morphologically distinct subtypes of ganglion cells in the mouse retina111,112. Ganglion cells are larger in size than most interneurons and have axons 25 with large diameters that are capable of passing the signal in the form of a transient spike over long distances. The first step in this process is the binding of neurotransmitters released by amacrine and bipolar cells to receptors in the ganglion cell dendritic membrane. This action triggers the opening of ion channels in the membrane leading to an alteration of membrane potential. A complex interplay of ion channel potentials and dendritic structures results in differential firing of the ganglion cells, thus conveying the visual message to the brain. The 8 ion channel receptors can be activated or inhibited by most common neurotransmitters such as glutamate and acetylcholine, and GABA and glycine, respectively113.

Cells classified as ganglion cells share several common features. For example, their somas are usually located in the ganglion cell layer, their dendrites arborize in the IPL, and their axons culminate in the optic nerve, which travels to the brain114. However very recently they have been divided into 25 clear subtypes based on morphology, physiology and molecular markers111. The finding that RGCs with similar physiological functions arborize in the same layer within the IPL has linked structure with function115,116. There are four types of ON-OFF direction-selective ganglion cells (DSGCs) that respond to both an increase and decrease in light intensity and depend on direction of motion117. ON-OFF

DSGCs are bistratified in the IPL and are closely linked with the starburst amacrine cells114.

A second group of ganglion cells are the ON DSGCs that are direction-selective but respond only to ON light stimulus114. Their dendrites are monostratified in the IPL closely with the

ON starburst amacrines114. The alpha retinal ganglion cells were first described in the cat retina, characterized by large somas and widely branching dendrites118,119. In the mouse, they are of three types and their common features include a large size and high levels of 26 expression of three genes: neurofilament, spp1, which encodes the secreted phosphoprotein osteopontin (OPN), and kcng4, which encodes a voltage-gated potassium channel subunit120. Experiments showed that >90% of α-RGCs were OPN+, and KCNG4 was found in all types of α-RGCs120. Another very interesting ganglion cell type is the photosensitive melanopsin containing ganglion cells. These are characterized by sensitivity to light and large dendritic arbors114. These project mainly to the suprachiasmatic nucleus and are involved in synchronizing the circadian rhythm121–123.

5. Guidance of Retinal Axons to Targets within the Brain

The vertebrate eye is the sensory organ of the visual system that transfers information from the external world to visual centers in the brain for interpretation124. To accomplish this, the incoming information must be processed in an efficient and organized fashion through a circuitry of visual projection starting at the retina passing through the optic chiasm, superior colliculus and lateral geniculate nucleus in the midbrain and ending at the visual cortex of the brain124. During this journey from the eye to the brain, there are several decisions to be made regarding the direction of movement. A combination of the identity of the neurons and several environmental factors serves to guide the axons to be able to reproduce the external signal in the form of a visual image124,125. Several transcription factors, signaling molecules and their receptors, as well as neuronal activity work in concert to establish this visual circuitry124. 27

The process of axon guidance starts with extension of the RGC axons into the optic fiber layer of the retina and then towards the optic disk through which they leave the retina124. After axons from RGCs leave the eye at the optic disc, they travel through the ventral encephalon towards the midline where they converge with axons from the contralateral eye124,125. Here, a population of axons crosses the midline at the optic chiasm to project to contralateral midbrain targets, the superior colliculus (SC), and the lateral geniculate nucleus (LGN)124. The remaining axons project away from the midline to ipsilateral midbrain targets124.

Mechanisms of axon guidance differ significantly between cold-blooded and warm- blooded vertebrates. In frogs and fish, the growth cone of the primary RGC axon extends directly to the correct topographic destination in its termination zone (TZ), and undergoes branching of the distal end of the primary axon and forms a terminal arbor directed away from the growth cone and towards the base, a phenomenon known as back-branching126–

128. However, the process of axon guidance in the chick and mouse is slightly more complex125. The primary growth cones of the RGC axons extend posteriorly past the location of the TZ. Axon branching then occurs along the length of the axon with the highest branches formed in the correct position in the TZ. The final step occurs over a few days and involves extensive remodeling in order to remove the major parts of the primary axon distal to the TZ and to trim incorrectly formed branches and arbors129–132. The elimination of the superfluous axon branches occurs by a caspase-6 dependent system, a process distinct from the caspase-3 system that leads to death of the entire cell133. 28

The Eph family of RTKs is one of the key regulators of topographic ordering of retinal projections134. It was experimentally shown that a GPI-linked protein expressed on the posterior membrane acts to repel the growth of axons135. Subsequently, these GPI- linked proteins were characterized as ephrin-A2 and ephrin-A5135. It was found that both ligands were expressed in a low to high anterior-posterior (AP) gradient in the optic tectum (OT) of chicken embryos and the corresponding tyrosine kinase receptor, EphA3, is expressed along a temporal-nasal (TN) gradient by axons of RGCs136. Targeted deletions of ephrin-A2 and ephrin-A5 in the mouse showed that they are necessary for correct topographic guidance activity in the mouse superior colliculus-analogous to the optic tectum in lower vertebrates136.

The Eph family of the largest known family of RTKs, with at least 12 members identified in vertebrates137. Eph-ephrin signaling can function to both repel and attract axonal populations depending on the developmental context136. In the chick or mouse,

EphAs are expresses in a high to low TN gradient by RGCs and ephrin-As are expressed in a low to high AP gradient in the OT/SC136,137. Since the temporal retina, expressing high levels of EphA maps to anterior OT with low levels of ephrin-A, this system must act as repellants for temporal axons more strongly than nasal axons136. Although the general pattern of signaling between the Ephrins and their receptor is similar between the chick and mouse, the type and expression of the signaling molecules and their receptors differs substantially136. For example, the mouse Ephrin-A5 is expressed in an increasing AP gradient along the AC, similar to the chick Ephrin-A2. Ephrin-A2 in the mouse, on the other hand, is expressed broadly around mid-posterior SC and gradually declines towards the far posterior136. The graded expression pattern, the demonstration of the repulsion of the 29 temporal axons, and the targeted deletion experiments firmly established the Ephrin-As as axon guidance molecules136. Thus, eph and ephrins are important molecules for guiding ventral RGC axons to the optic disc.

After reaching the optic disc, RGC axons have to exit the retina, enter the optic nerve and then project towards the midline124. The axon guidance molecule, Netrin together with its receptor, deleted in colorectal cancer (DCC) help in mediating the exit of RGC axons from the retina. Netrin-1 is expressed in glial cells at the optic disc, while DCC is expressed by the RGC axons138. Though mouse netrin-1/DCC loss of function mutants showed no defect in intra-retinal guidance, the axons failed to enter the optic stalk indicating that

NETRIN-1/DCC signaling operates at the optic disc to guide axons into the optic stalk124.

Once the ganglion cell axons enter the optic nerve, they travel in a fasciculated manner towards the brain124. The transmembrane protein Sema5A is expressed by neuroepithelial cells surrounding the optic disc and the optic nerve, and serves as an inhibitory sheath. Its function is to channel RGC axon growth within the optic nerve139. The homeodomain transcription factor, Vax1 is expressed in the optic disc and optic stalk and also later by the surrounding glial cells124. Vax1 appears to be involved in axon glia interaction within the optic nerve, possibly by regulation of the attractive molecule, netrin-1124.

The optic chiasm where RGC axons cross the ventral midline is formed at an invariant position along the antero-posterior axis of the forebrain124. The vertebrate homolog of the Drosophila gene, odd-paired, Zic2 encodes a zinc finger transcription factor expressed in uncrossed RGCs and initiates the ipsilateral pathfinding by RGC axons140. The ventral midline in both vertebrate and invertebrate organisms acts as an important 30 intermediate target for axons navigating along distinct trajectories140. The Slit family of guidance cues (Slit1-3) are among the factors that are involved in inter-retinal guidance125.

In Drosophila, the Slit receptor, Robo is diminished on crossing axons at the midline141.

After crossing the midline, Robo expression is specifically upregulated again ensuring that the axons do not recross140,141. In robo mutants, axons cross and recross the midline inappropriately. Robo is highly conserved across species both in sequence and function142.

Genetic and experimental evidence demonstrate that Slit proteins are ligands for Robo proteins in both Drosophila and vertebrates140,143. Another protein, Commissureless

(Comm) also is expressed by crossing axons and by midline glia140. It clears Robo from the cell surface and thus prevents it from signaling142. In Comm mutant embryos, the axons fail to cross because of the failure to remove Robo140. Thus, the functions of Comm are to downregulate Robo and abrogate sensitivity to Slit, allowing crossing to occur140.

The general structure of Slits and Robos is largely conserved and three vertebrate

Slits and Robos have been identified140. In the mouse, Slit1 and Robo2 are expressed in the

RGC layer by the time RGCs are generated, and slightly later in development Slit2 is also expressed144. Through loss-of-function experiments, Slit1 and Slit2 have been shown to play a role in restricting RGCs to the optic fiber layer of the retina, while Slit2 is involved in regulating the ordered growth within the optic fiber layer in the dorsal peripheral part of the retina124. An independent study showed that though Slit repellants in the floor plate act through Robo 1 and Robo 2 to guide commissural axons in the spinal cord, commissural axons are present that show no obvious axon guidance phenotype in the Slit1;Slit2;Slit3 triple mutant142. The third vertebrate Robo has been named Rig1145. Rig1 differs from the other two Robos in that it is expressed at high levels before crossing the midline and is 31 downregulated afterwards. Thus, Rig1 appears to mask the action of Slit upon the other

Robo receptors preventing them from being prematurely repelled by the midline. This function of Rig1 makes it analogous to the Drosophila Comm.

Other molecules such as Chondroitin sulfate proteoglycans (CSPGs) are extracellular matrix molecules regulating cell-cell and cell-matrix interaction124,146. Therefore, CSPGs fulfill two different functions in the course of optic chiasm formation: they are involved in patterning the early phase of RGC axonal growth across the midline and later control axon divergence at the chiasm146. Heparin sulfate (HS) is a linear polysaccharaide which also has a role in axon pathfinding across the midline147.

These results demonstrate that neuronal activity and signaling via guidance cues work together to achieve axon pathfinding and target recognition124. Many molecules involved in axon guidance in the retina, optic chiasm and optic nerve have been discovered, whereas a lot remains to be known about cues involved in axon pathfinding towards the lateral geniculate nucleus (LGN) and primary visual cortex124.

6. Conclusion

Examination of the retina carried out by numerous research groups has made it clear that the mechanisms underlying retinal cell fate acquisition constitute an intricate and dynamic process. Studies using a variety of techniques such as lineage tracing, functional studies, and transcriptomics have contributed much to our understanding of retinal development. 32

In this dissertation, I have aimed to take a step further towards the elucidation of the combinations of factors that drive cell fate decisions in the retina. To that end, I have employed single cell transcriptomics performed on developing retinal cells, as well as functional studies in two different organisms, the mouse and the zebrafish. The information extracted from transcriptome analysis of retinal cells led to a high-resolution understanding of the underlying heterogeneity in the retina, the gene expression changes occurring during development and provided new insights into the process of stochastic cell fate determination. Specific genes were identified from our single cell data that we hypothesized would play roles in driving fate of progenitors towards one cell type or another. We performed a thorough characterization for mouse mutant for the gene, Trim9, identified from our transcriptome data. To continue functional studies in a more adaptable model organism, I used CRISPR-Cas9 mediated genome editing in zebrafish to mutate several genes chosen from our single cell transcriptomic data. Taken together, our single cell transcriptomic approach coupled with genome editing is a powerful method for dissecting the precise networks of genes controlling early development of the retina.

7. References

1. DelMonte, D. W. & Kim, T. Anatomy and physiology of the cornea. J. Cataract Refract. Surg. 37, 588–598 (2011). 2. Willoughby, C. E. et al. Anatomy and physiology of the human eye: Effects of mucopolysaccharidoses disease on structure and function - a review. Clin. Exp. Ophthalmol. 38, 2–11 (2010).

3. Farjo, A., McDermott, M. & Soong, H. in Yanoff M, Duker JS, eds, Ophthalmology, 3rd ed. St. Louis, MO, Mosby 203–208 (2008). doi:10.1016/B978-0-323-04332-8.00025-1 4. McCaa, C. S. The eye and visual nervous system: anatomy, physiology and toxicology. Environmental Health Perspectives Vol. 44, 1–8 (1982). 5. Goel, M., Picciani, R. G., Lee, R. K. & Bhattacharya, S. K. Aqueous humor dynamics: a review. Open Ophthalmol. J. 4, 52–9 (2010). 6. Graw, J. Eye development. Curr. Top. Dev. Biol. 90, 343–386 (2010). 7. Hon-Vu Q Duong, M. Eye Globe Anatomy. 1 (2016). 8. Cvekl, A. & Ashery-Padan, R. The cellular and molecular mechanisms of vertebrate lens development. Development 141, 4432–4447 (2014). 9. Hagedoorn A. Comparative anatomy of the eye. Arch. Ophthalmol. 16, 783–803 (1936). 10. Seung, H. S. & Sümbül, U. Neuronal cell types and connectivity: Lessons from the retina. Neuron 83, 1262–1272 (2014). 11. Cepko, C. L., Austin, C. P., Yang, X., Alexiades, M. & Ezzeddine, D. Cell fate determination in the vertebrate retina. Proc Natl Acad Sci U S A 93, 589–595 (1996). 12. Masland, R. H. The Neuronal Organization of the Retina. Neuron 76, 266–280 (2012). 13. Young, R. W. Cell differentiation in the retina of the mouse. Anat. Rec. 212, 199–205 (1985). 14. Carter-Dawson, L. D. & LaVail, M. M. Rods and cones in the mouse retina. II. Autoradiographic analysis of cell generation using tritiated thymidine. J. Comp. Neurol. 188, 263–272 (1979). 15. Rapaport, D. H., Wong, L. L., Wood, E. D., Yasumura, D. & Lavail, M. M. Timing and topography of cell genesis in the rat retina. J. Comp. Neurol. 474, 304–324 (2004). 16. La Vail, M. M., Rapaport, D. H. & Rakic, P. Cytogenesis in the monkey retina. J. Comp. Neurol. 309, 86–114 (1991). 17. Stiemke, M. M. & Hollyfield, J. G. Cell birthdays in Xenopus laevis retina. Differentiation 58, 189–193 (1995). 18. Cepko, C. Intrinsically different retinal progenitor cells produce specific types of progeny. Nat. Rev. Neurosci. 15, 615–627 (2014). 19. Livesey, F. J. & Cepko, C. L. Vertebrate neural cell-fate determination: lessons from the retina. Nat Rev Neurosci 2, (2001). 20. R.W., R. The First Steps in Seeing. Arch. Ophthalmol. 117, 550 (1999).

21. Masland, R. H. The Neuronal Organization of the Retina. Neuron 76, 266–280 (2012). 22. Goetz, J. J., Farris, C., Chowdhury, R. & Trimarchi, J. M. Making of a retinal cell: Insights into retinal cell-fate determination. Int. Rev. Cell Mol. Biol. 308, 273–321 (2014). 23. Wohrer, A. The vertebrate retina: a functional review. (2008). 24. Chow, R. L. & Lang, R. A. Early Eye Development in Vertebrates. Annu. Rev. Cell Dev. Biol. 17, 255–296 (2001). 25. Belecky-Adams, T. et al. Pax-6, Prox 1, and Chx10 homeobox gene expression correlates with phenotypic fate of retinal precursor cells. Invest Ophthalmol Vis Sci 38, (1997). 26. Heavner, W. & Pevny, L. Eye development and retinogenesis. Cold Spring Harb. Perspect. Biol. 4, (2012). 27. Gunhaga, L. The lens: a classical model of embryonic induction providing new insights into cell determination in early development. Philos. Trans. R. Soc. B Biol. Sci. 366, 1193–1203 (2011). 28. Schmitt, E. A. & Dowling, J. E. Early Retinal Development in the Zebrafish , Danio rerio : Light and Electron Microscopic Analyses. 536, 515–536 (1999). 29. Avanesov, A. & Malicki, J. Analysis of the retina in the zebrafish model. Methods in Cell Biology 100, (2010). 30. Hu, M. & Easter, S. S. Retinal neurogenesis: the formation of the initial central patch of postmitotic cells. Dev. Biol. 207, 309–321 (1999). 31. Turner, D. L. & Cepko, C. L. A common progenitor for neurons and glia persists in rat retina late in development. Nature 328, 131–136 (1987). 32. Turner, D. L., Snyder, E. Y. & Cepko, C. L. Lineage-independent determination of cell type in the embryonic mouse retina. Neuron 4, 833–845 (1990). 33. Poggi, L., Vitorino, M., Masai, I. & Harris, W. A. Influences on neural lineage and mode of division in the zebrafish retina in vivo. J. Cell Biol. 171, 991–999 (2005). 34. Tucker, P. et al. The eyeless mouse mutation (ey1) removes an alternative start codon from the Rx/rax homeobox gene. Genesis 31, 43–53 (2001). 35. Loosli, F. et al. Loss of eyes in zebrafish caused by mutation of chokh/rx3. EMBO Rep. 4, 894–899 (2003). 36. Mathers, P. H., Grinberg, a, Mahon, K. a & Jamrich, M. The Rx homeobox gene is essential for vertebrate eye development. Nature 387, 603–607 (1997). 37. Lequeux, L. et al. Confirmation of RAX gene involvement in human anophthalmia. Clin. Genet. 74, 392–395 (2008). 35

38. Fuhrmann, S. Wnt signaling in eye organogenesis. Organogenesis 4, 60–67 (2008). 39. Bonini, N. M., Leiserson, W. M. & Benzer, S. Multiple roles of the eyes absent gene in Drosophila. Dev. Biol. 196, 42–57 (1998). 40. Kriebel, M., Muller, F. & Hollemann, T. Xeya3 regulates survival and proliferation of neural progenitor cells within the anterior neural plate of Xenopus embryos. Dev. Dyn. 236, 1526–1534 (2007). 41. Söker, T. et al. Pleiotropic effects in Eya3knockout mice. BMC Dev. Biol. 8, 118 (2008). 42. Cvekl, A. & Wang, W.-L. Retinoic acid signaling in mammalian eye development. Exp. Eye Res. 89, 280–291 (2009). 43. Mic, F. A., Molotkov, A., Molotkova, N. & Duester, G. Raldh2 expression in optic vesicle generates a retinoic acid signal needed for invagination of retina during optic cup formation. Dev. Dyn. 231, 270–277 (2004). 44. Yun, S. et al. Lhx2 links the intrinsic and extrinsic factors that control optic cup formation. Development 136, 3895–3906 (2009). 45. Livesey, F. J. & Cepko, C. L. Vertebrate neural cell-fate determination: lessons from the retina. Nat. Rev. Neurosci. 2, 109–118 (2001). 46. Cepko, C. L., Austin, C. P., Yang, X., Alexiades, M. & Ezzeddine, D. Cell fate determination in the vertebrate retina. Proc Natl Acad Sci USA 93, (1996). 47. Belecky-Adams, T., Cook, B. & Adler, R. Correlations between Terminal Mitosis and Differentiated Fate of Retinal Precursor Cellsin Vivoandin Vitro:Analysis with the ‘Window-Labeling’ Technique. Dev. Biol. 178, 304–315 (1996). 48. Goetz, J. J., Farris, C., Chowdhury, R. & Trimarchi, J. M. Making of a retinal cell: Insights into retinal cell-fate determination. International Review of Cell and Molecular Biology 308, (Elsevier Inc., 2014). 49. Holt, C. E., Bertsch, T. W., Ellis, H. M. & Harris, W. A. Cellular determination in the Xenopus retina is independent of lineage and birth date. Neuron 1, 15–26 (1988). 50. Belliveau, M. J. & Cepko, C. L. Extrinsic and intrinsic factors control the genesis of amacrine and cone cells in the rat retina. Development 126, 555–566 (1999). 51. Austin, C. P., Feldman, D. E., Ida, J. a & Cepko, C. L. Vertebrate retinal ganglion cells are selected from competent progenitors by the action of Notch. Development 121, 3637–3650 (1995). 52. Belliveau, M. J., Young, T. L. & Cepko, C. L. Late retinal progenitor cells show intrinsic limitations in the production of cell types and the kinetics of opsin synthesis. J. Neurosci. 20, 2247–2254 (2000).

53. Hobert, O. Neurogenesis in the nematode Caenorhabditis elegans. WormBook 1–24 (2010). doi:10.1895/wormbook.1.12.2 54. He, J. et al. How Variable Clones Build an Invariant Retina. Neuron 75, 786–798 (2012). 55. Chen, Z., Li, X. & Desplan, C. Deterministic or Stochastic Choices in Retinal Neuron Specification. Neuron 75, 739–742 (2012). 56. Gomes, F. L. A. F. et al. Reconstruction of rat retinal progenitor cell lineages in vitro reveals a surprising degree of stochasticity in cell fate decisions. Development 138, 227–35 (2011). 57. Trimarchi, J. M., Stadler, M. B. & Cepko, C. L. Individual retinal progenitor cells display extensive heterogeneity of gene expression. PLoS One 3, (2008). 58. Elowitz, M. Stochastic Gene Expression in a Single Cell. Science (80-. ). 297, 1183– 1186 (2013). 59. Laboissonniere, L. A. et al. Single cell transcriptome profiling of developing chick retinal cells. J. Comp. Neurol. (2017). doi:10.1002/cne.24241 60. Brown, N. L. et al. Math5 encodes a murine basic helix-loop-helix transcription factor expressed during early stages of retinal neurogenesis. Development 125, 4821–4833 (1998). 61. Masai, I. N-cadherin mediates retinal lamination, maintenance of forebrain compartments and patterning of retinal neurites. Development 130, 2479–2494 (2003). 62. Kanekar, S. et al. Xath5 Participates in a Network of bHLH Genes in the Developing Xenopus Retina. Neuron 19, 981–994 (2017). 63. Yang, Z., Ding, K., Pan, L., Deng, M. & Gan, L. Math5 determines the competence state of retinal ganglion cell progenitors. Dev. Biol. 264, 240–254 (2003). 64. Feng, L. et al. MATH5 controls the acquisition of multiple retinal cell fates. Mol Brain 3, 36 (2010). 65. Liu, W., Mo, Z. & Xiang, M. The Ath5 proneural genes function upstream of Brn3 POU domain transcription factor genes to promote retinal ganglion cell development. Proc. Natl. Acad. Sci. U. S. A. 98, 1649–1654 (2001). 66. Brown, N. L., Patel, S., Brzezinski, J. & Glaser, T. Math5 is required for retinal ganglion cell and optic nerve formation. Development 128, 2497–2508 (2001). 67. Wang, S. W. et al. Requirement for math5 in the development of retinal ganglion cells. Genes Dev. 15, 24–29 (2001).

68. Le, T. T., Wroblewski, E., Patel, S., Riesenberg, A. N. & Brown, N. L. Math5 is required for both early retinal neuron differentiation and cell cycle progression. Dev. Biol. 295, 764–778 (2006). 69. Kay, J. N. et al. Retinal Ganglion Cell Genesis Requires. 30, 725–736 (2001). 70. Mao, C.-A., Wang, S. W., Pan, P. & Klein, W. H. Rewiring the retinal ganglion cell gene regulatory network: Neurod1 promotes retinal ganglion cell fate in the absence of Math5. Development 135, 3379–3388 (2008). 71. Mao, C.-A. et al. Reprogramming amacrine and photoreceptor progenitors into retinal ganglion cells by replacing Neurod1 with Atoh7. Development 140, 541–551 (2013). 72. Prasov, L. & Glaser, T. Dynamic expression of ganglion cell markers in retinal progenitors during the terminal cell cycle. Mol. Cell. Neurosci. 50, 160–168 (2012). 73. Hufnagel, R. B. et al. Heterochronic misexpression of Ascl1 in the Atoh7 retinal cell lineage blocks cell cycle exit. Mol. Cell. Neurosci. 54, 108–120 (2013). 74. Wu, F. et al. Two transcription factors, Pou4f2 and Isl1, are sufficient to specify the retinal ganglion cell fate. Proc. Natl. Acad. Sci. U. S. A. 112, E1559-68 (2015). 75. Morrow, E. M., Furukawa, T., Lee, J. E. & Cepko, C. L. NeuroD regulates multiple functions in the developing neural retina in rodent. Development 126, 23–36 (1999). 76. Cherry, T. J. et al. NeuroD Factors Regulate Cell Fate and Neurite Stratification in the Developing Retina. J. Neurosci. 31, 7365–7379 (2011). 77. Kay, J. N., Voinescu, P. E., Chu, M. W. & Sanes, J. R. Neurod6 expression defines new retinal amacrine cell subtypes and regulates their fate. Nat. Neurosci. 14, 965–972 (2011). 78. de Melo, J. et al. Dlx1 and Dlx2 function is necessary for terminal differentiation and survival of late-born retinal ganglion cells in the developing mouse retina. Development 132, 311–322 (2005). 79. Zhang, Q. et al. Regulation of Brn 3b by Dlx 1 and Dlx 2 is required for retinal ganglion cell differentiation in the vertebrate retina. Development dev.142042 (2017). doi:10.1242/dev.142042 80. Elliott, J., Jolicoeur, C., Ramamurthy, V. & Cayouette, M. Ikaros Confers Early Temporal Competence to Mouse Retinal Progenitor Cells. Neuron 60, 26–39 (2008). 81. Brody, T. & Odenwald, W. F. Programmed Transformations in Neuroblast Gene Expression during Drosophila CNS Lineage Development. Dev. Biol. 226, 34–44 (2000). 82. Isshiki, T., Pearson, B., Holbrook, S. & Doe, C. Q. Drosophila neuroblasts sequentially express transcription factors which specify the temporal identity of their neuronal progeny. Cell 106, 511–521 (2001).

83. Kambadur, R. et al. Regulation of POU genes by castor and hunchback establishes layered compartments in the Drosophila CNS. Genes Dev. 12, 246–260 (1998). 84. Novotny, T., Eiselt, R. & Urban, J. Hunchback is required for the specification of the early sublineage of neuroblast 7-3 in the Drosophila central nervous system. Development 129, 1027–1036 (2002). 85. Pearson, B. J. & Doe, C. Q. Regulation of neuroblast competence in Drosophila. Nature 425, 624–628 (2003). 86. Mattar, P., Ericson, J., Blackshaw, S. & Cayouette, M. A conserved regulatory logic controls temporal identity in mouse neural progenitors. Neuron 85, 497–504 (2015). 87. Konstantinides, N., Rossi, A. M. & Desplan, C. Common temporal identity factors regulate neuronal diversity in fly ventral nerve cord and mouse retina. Neuron 85, 447–449 (2015). 88. Hackler, L., Wan, J., Swaroop, A., Qian, J. & Zack, D. J. MicroRNA profile of the developing mouse retina. Investig. Ophthalmol. Vis. Sci. 51, 1823–1831 (2010). 89. Bassett, E. A. & Wallace, V. A. Cell fate determination in the vertebrate retina. Trends Neurosci. 35, 565–573 (2012). 90. Decembrini, S. et al. MicroRNAs couple cell fate and developmental timing in retina. Proc. Natl. Acad. Sci. 106, 21179–21184 (2009). 91. Georgi, S. A. & Reh, T. A. Dicer is required for the transition from early to late progenitor state in the developing mouse retina. J. Neurosci. 30, 4048–4061 (2010). 92. Davis, N., Mor, E. & Ashery-Padan, R. Roles for Dicer1 in the patterning and differentiation of the optic cup neuroepithelium. Development 138, 127–138 (2011). 93. Ortín-Martínez, A. et al. Number and Distribution of Mouse Retinal Cone Photoreceptors: Differences between an Albino (Swiss) and a Pigmented (C57/BL6) Strain. PLoS One 9, e102392 (2014). 94. Kelber, A., Vorobyev, M. & Osorio, D. Animal colour vision – behavioural tests and physiological concepts. Biol. Rev. Camb. Philos. Soc. 78, S1464793102005985 (2003). 95. Helga, K. The Organization of the Retina and Visual System. (2005). 96. Steinberg, R. H., Fisher, S. K. & Anderson, D. H. Disc morphogenesis in vertebrate photoreceptors. J. Comp. Neurol. 190, 501–508 (1980). 97. Anderson, D. H. & Fisher, S. K. The photoreceptors of diurnal squirrels: outer segment structure, disc shedding, and protein renewal. J. Ultrastruct. Res. 55, 119–141 (1976). 98. Kelber, A., Vorobyev, M. & Osorio, D. Animal colour vision – behavioural tests and physiological concepts. Biol. Rev. Camb. Philos. Soc. 78, S1464793102005985 (2003).

99. Masland, R. H. Neuronal diversity in the retina. Curr. Opin. Neurobiol. 11, 431–436 (2001). 100. Kim, D. S. et al. Identification of molecular markers of bipolar cells in the murine retina. J. Comp. Neurol. 507, 1795–1810 (2008). 101. Wässle, H. Parallel processing in the mammalian retina. Nat. Rev. Neurosci. 5, 747– 757 (2004). 102. Connaughton, V. P., Graham, D. & Nelson, R. Identification and morphological classification of horizontal, bipolar, and amacrine cells within the zebrafish retina. J Comp Neurol 477, (2004). 103. Dowling, J. E. & Werblin, F. S. Organization of retina of the mudpuppy, Necturus maculosus. I. Synaptic structure. J. Neurophysiol. 32, 315–38 (1969). 104. Connaughton, V. P. & Nelson, R. Axonal stratification patterns and glutamate-gated conductance mechanisms in zebrafish retinal bipolar cells. J Physiol 524, (2000). 105. Farajian, R., Pan, F., Akopian, A., Völgyi, B. & Bloomfield, S. A. Masked excitatory crosstalk between the ON and OFF visual pathways in the mammalian retina. J. Physiol. 589, 4473–4489 (2011). 106. Demb, J. B. & Singer, J. H. Intrinsic properties and functional circuitry of the AII amacrine cell. Vis. Neurosci. 29, 51–60 (2012). 107. Gollisch, T. & Meister, M. Eye Smarter than Scientists Believed: Neural Computations in Circuits of the Retina. Neuron 65, 150–164 (2010). 108. Martin, P. R. & Gr??nert, U. Spatial density and immunoreactivity of bipolar cells in the macaque monkey retina. J. Comp. Neurol. 323, 269–287 (1992). 109. Vardi, N., Kaufman, D. L. & Sterling, P. Horizontal cells in cat and monkey retina express different isoforms of glutamic acid decarboxylase. Vis Neurosci 11, 135–142 (1994). 110. Sandmann, D., Boycott, B. B. & Peichl, L. Blue-cone horizontal cells in the retinae of horses and other equidae. J. Neurosci. 16, 3381–96 (1996). 111. Li, Y. N., Matsui, J. I. & Dowling, J. E. Specificity of the horizontal cell-photoreceptor connections in the zebrafish (Danio rerio) retina. J. Comp. Neurol. 516, 442–453 (2009). 112. Herrmann, R. et al. Rod vision is controlled by dopamine-dependent sensitization of rod bipolar cells by GABA. Neuron 72, 101–110 (2011). 113. Hartline, H. The response of single optic nerve fibers of the vertebrate eye to illumination of the retina. Am. J. Physiol. 121, 400–415 (1938). 114. Kuffler, S. W. Discharge Patterns and Functional Organization of Mammalian Retina. J. Neurophysiol. 16, 37–68 (1953). 40

115. Sanes, J. R. & Masland, R. H. The Types of Retinal Ganglion Cells: Current Status and Implications for Neuronal Classification. Annu. Rev. Neurosci. 38, 150421150146009 (2014). 116. Macosko, E. Z. et al. Highly Parallel Genome-wide Expression Profiling of Individual Cells Using Nanoliter Droplets. Cell 161, 1202–1214 (2015). 117. Cohen, E. D., Zhou, Z. J. & Fain, G. L. Ligand-gated currents of alpha and beta ganglion cells in the cat retinal slice. J. Neurophysiol. 72, 1260–9 (1994). 118. Sanes, J. R. & Masland, R. H. The Types of Retinal Ganglion Cells: Current Status and Implications for Neuronal Classification. Annu. Rev. Neurosci. 38, 150421150146009 (2015). 119. Famiglietti, E. V & Kolb, H. Structural basis for ON-and OFF-center responses in retinal ganglion cells. Science 194, 193–5 (1976). 120. Famiglietti, E. V, Kaneko, a & Tachibana, M. Neuronal architecture of on and off pathways to ganglion cells in carp retina. Science 198, 1267–9 (1977). 121. Weng, S., Sun, W. & He, S. Identification of ON-OFF direction-selective ganglion cells in the mouse retina. J. Physiol. 562, 915–923 (2005). 122. Wässle, H., Levick, W. R. & Cleland, B. G. The distribution of the alpha type of ganglion cells in the cat’s retina. J. Comp. Neurol. 159, 419–437 (1975). 123. Cleland, B. G., Levick, W. R. & Wassle, H. Physiological identification of a morphological class of cat retinal ganglion cells. J Physiol 248, 151–171 (1975). 124. Duan, X. et al. Subtype-specific regeneration of retinal ganglion cells following axotomy: effects of osteopontin and mTOR signaling. Neuron 85, 1244–1256 (2015). 125. Hattar, S. Melanopsin-Containing Retinal Ganglion Cells: Architecture, Projections, and Intrinsic Photosensitivity. Science (80-. ). 295, 1065–1070 (2002). 126. Hattar, S. et al. Melanopsin and rod-cone photoreceptive systems account for all major accessory visual functions in mice. Nature 424, 76–81 (2003). 127. Panda, S. Melanopsin (Opn4) Requirement for Normal Light-Induced Circadian Phase Shifting. Science (80-. ). 298, 2213–2216 (2002). 128. Haupt, C. How axons see their way - axonal guidance in the visual system. Front. Biosci. 13, 3136 (2008). 129. Erskine, L. & Herrera, E. The retinal ganglion cell axon’s journey: Insights into molecular mechanisms of axon guidance. Dev. Biol. 308, 1–14 (2007). 130. Haupt, C. How axons see their way - axonal guidance in the visual system. Front. Biosci. 13, 3136 (2008).

131. Fujisawa, H. Mode of growth of retinal axons within the tectum of Xenopus tadpoles, and implications in the ordered neuronal connection between the retina and the tectum. J. Comp. Neurol. 260, 127–139 (1987). 132. Harris, W. A., Holt, C. E. & Bonhoeffer, F. Retinal axons with and without their somata, growing to and arborizing in the tectum of Xenopus embryos: a time-lapse video study of single fibres in vivo. Development 101, 123–133 (1987). 133. Kaethner, R. J. & Stuermer, C. A. Dynamics of terminal arbor formation and target approach of retinotectal axons in living zebrafish embryos: a time-lapse study of single axons. J. Neurosci. 12, 3257–3271 (1992). 134. Nakamura, H. & O’Leary, D. D. Inaccuracies in initial growth and arborization of chick retinotectal axons followed by course corrections and axon remodeling to develop topographic order. J. Neurosci. 9, 3776–3795 (1989). 135. Simon, D. K. & O’Leary, D. D. Limited topographic specificity in the targeting and branching of mammalian retinal axons. Dev. Biol. 137, 125–134 (1990). 136. Simon, D. K. & O’Leary, D. D. Development of topographic order in the mammalian retinocollicular projection. J. Neurosci. 12, 1212–1232 (1992). 137. Hindges, R., McLaughlin, T., Genoud, N., Henkemeyer, M. & O’Leary, D. D. M. EphB forward signaling controls directional branch extension and arborization required for dorsal-ventral retinotopic mapping. Neuron 35, 475–487 (2002). 138. Nikolaev, A., McLaughlin, T., O’Leary, D. D. M. & Tessier-Lavigne, M. APP binds DR6 to trigger axon pruning and neuron death via distinct caspases. Nature 457, 981–989 (2009). 139. Jeffery, G. & Erskine, L. Variations in the architecture and development of the vertebrate optic chiasm. Prog. Retin. Eye Res. 24, 721–753 (2005). 140. Walter, J., Henke-Fahle, S. & Bonhoeffer, F. Avoidance of posterior tectal membranes by temporal retinal axons. Development 101, 909–913 (1987). 141. Feldheim, D. A. & O’Leary, D. D. M. Visual map development: bidirectional signaling, bifunctional guidance molecules, and competition. Cold Spring Harbor perspectives in biology 2, (2010). 142. Feldheim, D. A. & O’Leary, D. D. M. Visual map development: bidirectional signaling, bifunctional guidance molecules, and competition. Cold Spring Harb. Perspect. Biol. 2, 1–20 (2010). 143. Cheng, H. J., Nakamoto, M., Bergemann, A. D. & Flanagan, J. G. Complementary gradients in expression and binding of ELF-1 and Mek4 in development of the topographic retinotectal projection map. Cell 82, 371–381 (1995). 144. Deiner, M. S. et al. Netrin-1 and DCC mediate axon guidance locally at the optic disc: Loss of function leads to optic nerve hypoplasia. Neuron 19, 575–589 (1997).

145. Oster, S. F. Invariant Sema5A inhibition serves an ensheathing function during optic nerve development. Development 130, 775–784 (2003). 146. Brose, K. et al. Slit proteins bind Robo receptors and have an evolutionarily conserved role in repulsive axon guidance. Cell 96, 795–806 (1999). 147. Seeger, M., Tear, G., Ferres-Marco, D. & Goodman, C. S. Mutations affecting growth cone guidance in Drosophila: genes necessary for guidance toward or away from the midline. Neuron 10, 409–426 (1993). 148. Kidd, T. et al. Roundabout controls axon crossing of the CNS midline and defines a novel subfamily of evolutionarily conserved guidance receptors. Cell 92, 205–215 (1998). 149. Kidd, T., Bland, K. S. & Goodman, C. S. Slit is the midline repellent for the robo receptor in Drosophila. Cell 96, 785–794 (1999). 150. Thompson, H., Camand, O., Barker, D. & Erskine, L. Slit Proteins Regulate Distinct Aspects of Retinal Ganglion Cell Axon Guidance within Dorsal and Ventral Retina. J. Neurosci. 26, 8082–8091 (2006). 151. Yuan, W. et al. The mouse SLIT family: secreted ligands for ROBO expressed in patterns that suggest a role in morphogenesis and axon guidance. Dev. Biol. 212, 290–306 (1999). 152. Brittis, P. A., Canning, D. R. & Silver, J. Chondroitin sulfate as a regulator of neuronal patterning in the retina. Science (80-. ). 255, 733 LP-736 (1992). 153. Yamaguchi, Y. Heparan sulfate proteoglycans in the nervous system: their diverse roles in neurogenesis, axon guidance, and synaptogenesis. Semin. Cell Dev. Biol. 12, 99–106 (2001). 43

CHAPTER 2. SINGLE CELL PROFILING OF ATOH7-EXPRESSING RETINAL CELLS

Abstract

To gain insight into how cells progress from an uncommitted progenitor state to a differentiated state, it is critical to identify the factors that control fate decisions. Atoh7 is a highly conserved transcription factor whose expression coincides with the G2/M cell cycle phase in progenitor cells (i.e. mitotic cells) that may be selecting their fate or the fate of their daughter cell(s). In this study, we have isolated 19 Atoh7+ cells from the mouse retina at various developmental times and examined their transcriptomic profiles at single cell resolution. Despite the fact that the majority of Atoh7+ cells were not classified as retinal ganglion cells, they nonetheless expressed a varying number of early ganglion cell markers.

Additionally, many Atoh7+ cells expressed genes that were characteristic of other early- born retinal neurons, demonstrating that cells pass through a developmental phase where they co-express markers of multiple retinal neurons. To identify new candidate cell fate factors, we have also utilized a variety of methods to discover genes whose expression were highly correlated with that of Atoh7, and have shown, by in situ hybridization the expression of these genes within the developing retina. Thus, the single cell approach is a powerful method for querying the intricate mechanisms that accompany the development of early generated retinal cells. 44

Introduction

The exact mechanisms that control how progenitor cells in the central nervous system (CNS) generate a diverse set of cell types are not fully understood, but are believed to involve some combination of extrinsic signaling pathways and intrinsic factors19. The retina is a widely used model for studying cell fate choice in the CNS. The retina is composed of six major neuronal types (ganglion cells, amacrine cells, bipolar cells, horizontal cells, rods and cones) and Müller glia cells19. However, the retina is actually far more complex in its make-up, as many of the neuronal types can be further sub-divided12.

Lineage tracing studies have revealed that each of the retinal cell types are generated from a pool of multipotent retinal progenitor cells (RPCs) 30,31,46 (herein RPC is defined as a mitotic cell). In addition, 3[H]-thymidine based birthdating studies have established that the retinal ganglion cells (RGC) are generated first from this pool of RPCs, followed closely by cone photoreceptors, horizontal cells (HC) and then amacrine cells (AC) and rod photoreceptors13,15,148,149. The remaining cell types, the bipolar cells (BC) and the Müller glia, are produced at later timepoints and this birth order is conserved in a multitude of organisms13,15,148,149.

Overlap in the birth window of different cell types has traditionally made it challenging to determine the specific factors involved in cell fate determination and cell type diversification of particular retinal neuron classes. Although it is possible that these factors could be primarily environmental in nature, studies have strongly suggested that intrinsic factors control the production of particular retinal cells150,151. For example, cell 45 mixing experiments have shown that RPCs from younger retinas moved into older environments do not switch the cell fates they produce47,49,152. Moreover, RPCs grown in isolated cultures, where the environmental signals would be lost or minimized, mimic the cell fate decisions made by RPCs in vivo53,151. Some of these studies and others led to the proposal of the competence model of retinal development, which postulates that RPCs pass through a series of states during which they are only able to generate particular retinal cell types11. While factors that define particular competence states and their transitions have been defined in Drosophila153, the genes controlling these states in the mouse retina are only starting to come into focus. Interestingly, some of the vertebrate homologues of the

Drosophila temporal transcription factors, such as Ikaros and Casz177,83 , as well as some miRNAs154 have roles to play in the production of early vs late retinal fates.

Although it is possible that RPCs pass through different competency states, with some form of environmental influence or selection process inducing certain fate choices, there may also be a more stochastic/probabilistic component 155. Experiments performed in the rat and zebrafish retinas have demonstrated that clonal growth fits a stochastic model, particularly for clone size, and perhaps for some aspects of fate choices, which might derive from the dynamic expression of different combinations of factors that influence cell fate choices51,53. Different fate-determining genes, which are expressed at different levels over time, as proposed by the competence model, might lead a cell to produce a given fate in accord with a stochastic inheritance pattern52. Our previous assessment of single RPCs in multiple organisms showed that a variety of genes, including cell cycle regulators and transcription factors (TFs) were dynamically expressed at different levels in individual cells collected at the same developmental time54,56,156. These 46 observations are consistent with intrinsic gene expression patterns in individual cells controlling the determination of cell fate in a developing tissue, with the possibility of stochastic aspects to the inheritance pattern of factors leading to a certain distribution of fates at one time 155. Profiling the transcriptomes of single cells in a narrower window of development could reveal the gene networks that are correlated with the different cell fate outcomes produced during that window.

Atoh7 is a basic-helix-loop-helix (bHLH) TF expressed in a subset of RPCs at the time when that cell is hypothesized to be engaged in the cell fate decision-making process57,157.

Atoh7 knockout mice lose 80-90% of their RGCs63,64 and atoh7 deficient zebrafish are completely devoid of RGCs66. At the same time, the absence of Atoh7 in mice leads to changed proportions of other early-born cells, such as cone photoreceptors63. In the mouse, retinal lineage tracing studies show that photoreceptors, ACs, HCs, and a majority of RGCs have a history of Atoh7 expression60,61,158. By transcriptionally profiling Atoh7+ cells, those that are most likely transitioning from cycling progenitor to newborn neuron, we hypothesized that we would gain insight into the cell fate determination process.

Furthermore, these transcriptional profiles of single Atoh7+ RPCs will help explain the process by which distinct RPCs divide and differentiate at variable probabilities within certain competence windows.

In this study, we used single cell transcriptome profiling of Atoh7+ cells isolated at several early developmental timepoints. Our goals were twofold. First, we examined the full transcriptomes of these cells to determine what cell types they most closely resembled.

We found that many had gene expression profiles characteristic of cycling RPCs. However, a more detailed examination of the profiles also revealed the expression of several markers 47 of mature retinal neurons, with RGC genes being the most prominent. Second, we searched for genes whose expression was strongly correlated with that of Atoh7, as we predict that these genes are strong candidates for playing key roles in retinal cell fate determination.

These microarray experiments revealed several genes whose expression was highly correlated with Atoh7 at early developmental timepoints (E12.5-E16.5). In situ hybridizations (ISH) for many of these genes confirmed the correlation with Atoh7. While functional studies will be required to ascertain the role that these genes might play in cell fate, this approach has revealed a number of candidates for the determination of several early retinal cell fates. In addition, these data reveal the expression of markers of several retinal cell types within a single cell, suggesting that exiting cells and/or the RPCs that make them are competent to make or become more than early retinal cell fate.

Materials and Methods

Single cell isolation and cDNA synthesis

To isolate Atoh7-expressing cells, we employed two approaches. First, we examined our previously isolated single cells54 for the expression of Atoh7 by RT-PCR. We found 2 cells at

E12.5 (cell A1 and cell C1), 1 cell at E13.5 (cell B5) and 1 cell at E16.5 (cell D4) that had not been hybridized to microarrays and processed those 4 cells. In addition, we observed that

18 other cells from our previous study54 expressed high levels of Atoh7. Second, to accumulate more cells expressing Atoh7 for profiling, we took advantage of a mouse expressing LacZ under the Atoh7 promoter63. Retinas were dissected at E14.5, a time when

Atoh7 expressing cells would be abundant, and placed in PBS with the future ganglion cell 48 layer facing upward. The -gal+ cells were labeled by pipetting 5 µl of Fluorescein Di- -D-

Galactopyranoside (FDG) (8 mM in 10% DMSO) onto the surface of the retina159. After ~10 seconds, the solution was diluted 10-fold in fresh PBS. This resulted in the labeling of cells with a robust green fluorescence.

The remainder of the single cell transcriptome profiling was performed exactly as described previously160. Briefly, retinae were dissected from CD-1 (Charles River

Laboratories) embryos at distinct developmental timepoints [E12.5 through E16.5] and dissociated using papain (Worthington). The dissociated retinal cells were washed, pelleted and resuspended in PBS (pH 7.4) containing 0.1% BSA. Randomly selected single retinal cells were harvested using a mouth pipette and a capillary tube (Sigma) drawn into a fine glass needle. The single cells were expelled into cold lysis buffer (10 mM Tris-HCl [pH

8.3], 50 mM KCl, 1.5 mM MgCl2, 5 mM DTT, 0.5% NP-40) and reverse transcribed using a modified oligo dT primer (TATAGAATTCGCGGCCGCTCGCGAT24). The first strand cDNA was tailed using terminal deoxynucleotidyl transferase (TdT) and then PCR amplified for

35 cycles using the same oligo dT primer. The specific PCR program used was: 95C for 2 min., 37C for 5 min., 72C for 16 min., 93C for 40 sec., 67C for 1 min., 72C for 6 min with a 6 sec extension each cycle for 35 cycles. The cDNA quality was assessed by running a subset on a 1% agarose and observed the maximum and minimum sizes.

Preparation for Affymetrix array hybridization

The amplified cDNA was fragmented with DNase I in One-Phor-All Buffer (500mM

Potassium Acetate, 0.1M Tris Acetate, 0.1M Magnesium Acetate) and biotinylated with

Biotin N6-ddATP (Enzo Life Sciences) and Terminal Transferase (Roche) to prepare for 49 hybridization. Affymetrix Mouse 430 2.0 microarrays were hybridized according to standard microarray protocols160.

Detection of associated genes

Before any of the analyses were performed, the single cell data were filtered such that raw signal values less than 1000 were removed. A value of 1000 was chosen as the background level because, in general, our experience with Affymetrix mouse arrays has found that the

Affymetrix algorithm never labeled signals of 1000 and above with an absent call in our profiles. In a limited number of replicate experiments (where the same cDNA was labeled independently more than once), Affymetrix present calls were extremely reproducible in terms of signal value, while the values for the absent calls varied widely (data not shown).

Importantly, these variable signals were never labeled as present by the Affymetrix algorithm regardless of their values17.

Hierarchical clustering: Gene Cluster software was used to determine gene associations17. After the data was signal filtered, it was log transformed and normalized according to the software instructions. To produce the hierarchical clusters, average linkage clustering was utilized. The resulting heatmaps were generated using Genesis software161.

Fisher’s exact test: Probe sets were filtered such that only those that achieved at least a single value of 1000 or greater in at least one single cell were retained. The transcriptomes from the single cell microarrays in this study were combined with profiles from RPCs17, 50 developing ganglion, amacrine and photoreceptor cells160, mature bipolar cells96, mature amacrine cells162 and adult Müller glia163. Expression values were binned into 5 equally sized bins (for details on bin number choice see previous study160). All probe set pairs were then analyzed for association using the following procedure: First, a contingency table with n rows and n columns was generated that recorded the joint distribution over bins for a given probe set pair. Subsequently, a p-value for the significance of the association was calculated from this table using a Fisher’s exact test160.

To classify a particular single cell based upon its gene expression profile, genes strongly associated by the Fisher’s exact test with different known marker genes were determined. For RPCs, it was previously shown that a composite RPC classification score using four RPC-expressed genes (CyclinD1, Fgf15, Sfrp2 and -crystallin) was the most effective way to classify these cells54. Therefore, to produce an RPC character score for each single cell profile, associated genes were identified for each of these four RPC genes using the Fisher’s exact test and a cutoff p-value of 10-3. The relative expression levels were then calculated and scaled RPC scores generated just as in our previous study54. To generate scores for the other cell types, the same method was utilized to identify RGC clusters (using

Nefl), AC clusters (using Tcfap-2) and photoreceptors (using Prdm1), relative expression levels were determined and scaled scores for each cell type calculated.

Visual inspection in excel: To identify genes with higher expression in Atoh7+ versus

Atoh7- cells, the filtered single cell profiles of cell expressing and not expressing Atoh7 were compared. Since the majority of the Atoh7+ cells were isolated at E12.5 through 51

E15.5, these cells were included in the analysis. The resulting heatmap was generated using Genesis software161.

In situ hybridization

Riboprobe synthesis: A number of the in situ probes used here were from the BMAP collection, while others were cloned from cDNA. For a complete list of probes used please see Supplementary Table S1 and S2 online. Non-BMAP collection probe template sequences were synthesized using specific primers from mouse cDNA, cloned into a pGEM-

T vector and sequenced. Antisense riboprobes were synthesized using the T7 or SP6 RNA polymerase in the presence of digoxigenin (Roche) for 1-2 hours at 37°C. Riboprobes were treated with DNase I for 15 minutes and precipitated with 100% ethanol and LiCl overnight.

Section in situ hybridization: In situ hybridization (ISH) on retinal cryosections was performed as described previously160 with a few modifications. TNT (0.1 M Tris-HCl, pH

7.5, 0.15 M NaCl, 0.05% Tween-20) buffer was used to replace the MABT buffer and the blocking buffer (Roche) was left out of the pre-incubation step before antibody incubation.

After completion, the reaction was stopped by fixation in 4% paraformaldehyde/PBS and mounting the slides in Fluoromount-G (Southern Biotech).

Dissociated ISH: ISH on dissociated retinal cells was performed exactly as described previously160. Briefly, retinas were isolated, washed three times with PBS and dissociated using papain and plated on poly-D lysine (10 µg/ml in PBS; Sigma) coated slides. The cells 52 were fixed in 4% PFA/PBS for 10 minutes, washed twice with PBS and dehydrated into

100% methanol. ISH was performed using digoxigenin labeled probes (color 1) and fluorescein labeled probes (color 2). Detection was accomplished using first an anti-DIG-

POD antibody (Roche, 1:1000) along with Cy3 tyramide solution (1:50, PerkinElmer Life

Sciences). The reaction was quenched with 0.3% hydrogen peroxide in PBS. Fluorescein labeled probes were detected using an anti-fluorescein-POD antibody (Roche, 1:1000) and

Alexa 488-tyramide (1:100, Molecular Probes).

Results

Single cell isolation and classification

To gain insight into the process of retinal cell fate determination, we isolated individual cells from the developing mouse retina during the window of time when early- born retinal neurons are being produced13,149. We focused specifically on cells that were expressing the TF Atoh7, as the expression of this gene occurs during this window60,65,158.

Single cell cDNA libraries were generated and hybridized to Affymetrix 430 2.0 mouse microarrays to yield single cell transcriptomes54. Since the cells were chosen at random, we began by more extensively classifying the profiled cells as an RPC or a developing retinal neuron [RGC, AC or photoreceptor (PR)] using our previously developed post-hoc classification scheme54 that was based on the expression of specific gene clusters (Table 1).

HCs were excluded as there are no bona fide transcriptomes from these cells with which to generate our extensive gene clusters. We included two groups as controls for this analysis.

First, single cell profiles from previously characterized developing RGCs160 showed 53 significant RGC “character” scores while their RPC, AC and PR scores were much lower

(Table 1). Likewise, a set of previously identified cycling RPCs54 displayed significant RPC scores, along with low scores for RGC, AC and PR character. In the current study, 13 out of

19 Atoh7+ single cells showed a high RPC character score and failed to attain significance in scores for any other cell type. Of the remaining 6 Atoh7+ cells, 1 had high RPC and RGC scores together, 2 had significant RGC scores alone, 1 had a low, but above our threshold,

AC score and 2 failed to cross our threshold for any of the cell type scores (Table 1).

Surprisingly, given that the majority of early-born retinal cells have a history of Atoh7 expression32-34, most of the Atoh7+ cells scored minimally for AC character and PR character, indicating that most of the profiled Atoh7+ cells were either RPCs or leaning towards an RGC fate rather than an AC or PR fate. One cell, E14 A5, showed almost equally high scores for both RPC and RGC character, indicating that this cell might be transitioning from an RPC to an RGC.

Since there is evidence that some cell fate decisions in the nervous system are made during the G2 phase of the cell cycle47,157,164, we wished to establish whether the Atoh7+ cells had gene expression profiles consistent with this cell cycle phase54. To accomplish this, we analyzed Atoh7+ and Atoh7- cells for the expression of different cell cycle markers

(See Supplementary Fig. S1 online). We found that ~50% of our Atoh7+ cells expressed a broad range of cell cycle markers54,165–167 including Geminin, Cdt1, Mcms (Mcm2, 3, 5 and 6),

Rpa1, E2f1, cyclin A2, Kpna2, cyclin E1, and Rassf1, and markers more specific to the G2/M phases54,168–170 such as Aurkb, Cdc20, cyclin B1, cyclin B2, Ube2c, Racgap1, Spc25, Plk1, and

Plk4 (See Supplementary Fig. S1 online). These results suggest that many of our Atoh7+ cells still have gene expression characteristics of cycling progenitor cells with G2/M 54 markers, indicating that some of those cells may be close to a mitotic division. To visualize the expression of some of these cell cycle markers throughout the developing retina, we performed ISH on retinal cryosections. We observed the expression of Geminin, Cdt1 and

Plk1 in subsets of cells in the outer neuroblastic layer (ONBL), where RPCs reside171, at three different timepoints (E12.5, E14.5 and E16.5) (See Supplementary Fig. S1 online).

The localization of these genes by ISH further points to the fact that a majority of our

Atoh7+ cells have gene expression profiles populated by a number of cycling cell-associated transcripts.

The observation of a cell expressing RPC and RGC genes together led us to wonder if this was a more general phenomenon in our transcriptomes. Even though the majority of our cells failed to score highly for RGC “character” we examined the expression of several early-expressed ganglion cell markers in the single cell profiles (Figure 1). Interestingly, we observed that the Atoh7+ cells expressed different combinations of these markers.

Furthermore, the Atoh7+ cells could be ordered into distinct groups based upon the number of these genes present in their transcriptome, yielding a “stair step” pattern of expression, consistent with many Atoh7+ cells transitioning from an RPC to a neuron.

Group 1 cells expressed either none of the RGC markers or a few of them at low levels.

These cells were presumably still cycling RPCs. Group 2 cells (E14 B2-E12 E5) expressed

Ebf3 at high levels and few to none of the other RGC-expressed genes. It is possible that this group had just started on the RGC precursor route172. Ebf3 has been previously shown to be downregulated in Atoh7 deficient retinas and is expressed in developing neurons at

E13.5 in mouse retinas172. However, there is no direct link between Ebf3 and RGC production. Another subset of Atoh7+ cells, group 3 (E12 A4-E13 B5), expressed more early 55

RGC markers such as Ebf3, Pou4f2, Isl1, and Crmp1, indicating that these cells may have progressed farther down the route of becoming a ganglion cell. A large subset of cells, labeled group 4 (E12 A1-E15 B5), expressed all or most of the RGC markers we queried, including Gap43, Neurofilament medium (Nefm) and Neurofilament light (Nefl). This subset of cells appears to be composed of differentiating RGCs as many of these cells expressed genes similar to more differentiated RGCs that do not express Atoh7 (group 5)160. Thus, this heatmap shows that the majority of Atoh7+ cells, despite showing a great deal of RPC character, appear to be in the process of generating an RGC.

Next we were curious as to how the expression of the cell cycle markers, particularly the G2/M marker genes, related to the “stair-step” pattern of RGC marker gene expression we observed (Figure 1). To address this question we generated a heatmap combining the expression of these two sets of genes in the single cell transcriptomes (See

Supplementary Fig. S2 online). Focusing on the Atoh7+ cells that robustly expressed G2/M markers, we saw the same “stair-step” pattern of RGC marker expression in these cells (see

E12 E5-E14 E1 in Supplementary Fig. S2 online). In addition, three Atoh7+ cells failed to express G2/M markers and expressed RGC markers (E14 A3-E14 D8), much like the newborn neurons160. Perhaps most intriguing were a set of 5 cells (E14 D1-E12 C1) that expressed virtually no G2/M marker genes, but yet still displayed the “stair-step” pattern of

RGC markers, indicating that they were still potentially deciding on a cell fate. Why these cells were devoid of G2/M markers, but still not looking like more mature neurons is an open question. A closer examination of their transcriptomes revealed that they all expressed the gene Plk3, a putative cell cycle regulator173–175 (See Supplementary Fig. S2 56 online). However, the meaning of this observation is unclear, as loss of Plk3 did not have an observable effect on retinal development176.

Lineage tracing studies demonstrated that cells with a history of Atoh7 expression give rise to more cell types than just RGCs30,31,46,158. Therefore, we wished to ascertain whether the single Atoh7+ cell transcriptomes showed any hallmarks of other retinal neurons. To analyze the expression of markers of other cell fates in Atoh7+ cells, we examined the single cell transcriptomes for markers of ACs (Tcfap2b160), HCs (Onecut1 and

Onecut2177–179) and PRs (Olig2180, Otx2181, Prdm1182 and Crx183,184). While most Atoh7+ cells expressed early RGC markers, we also found many of the cells expressing at least some of these genes associated with other retinal cell types (Figure 2). In particular, we identified a few cells with interesting and complex expression patterns. Of these, E12 cell A4 expressed

Pou4f2, Onecut1, Otx2, Prdm1, Crx. E12 cell A1 expressed Onecut2 and Otx2 in addition to

Pou4f2. E13 cell B5 expressed Pou4f2, Tcfap2b, Onecut1 and Onecut2. E14 cell E1 expressed

Pou4f2, Nefl, Onecut2, Otx2, Prdm1, and Crx (Figure 2). The presence of markers of multiple retinal neurons in these Atoh7+ cells may be indicative of a period of competence to make more than one early cell fate. In addition, it may be that a preponderance of markers of a particular cell type indicate a greater the probability of producing or becoming that cell type.

Markers of more than one cell type within a single cell might be due to the isolation of doublets, i.e. two cells isolated as a “single” cell. To investigate whether this is the case, or whether the overlap in cell fate markers within the same cell is accurate, we performed fluorescent double-ISH on dissociated retinas from E14.5 mice (See Supplementary Fig. S3 online) and quantified our observations (See Supplementary Fig. S3 online). We found that 57

Atoh7 overlapped with Cdc20, a G2/M marker in 20% of cells counted (See Supplementary

Fig. S3 online). Delta-like 1 (Dll1), which has been previously correlated with production of a post-mitotic cell from a progenitor54 showed a 50% overlap with Atoh7 (See

Supplementary Fig. S3 online), thus strengthening the notion that at least several Atoh7+ cells are in the process of exiting the cell cycle and making a cell fate decision. A 30% overlap between Atoh7 and Neurod1, a marker of ACs and PRs49, was also observed (See

Supplementary Fig. S3 online). Almost no overlap was seen between Atoh7 and Tcfap2b, an early marker of ACs (See Supplementary Fig. S3 online)160. This was consistent with what we saw in the single cell profiles as only 1 of the Atoh7 cells expressed Tcfap2 (Figure 2).

Finally, Atoh7 showed a small amount of overlap with the expression of Trb2 (See

Supplementary Fig. S3 online) and Nefl (See Supplementary Fig. S3, online), markers of developing cones50 and RGCs18,51 respectively.

Genes correlated with Atoh7 expression

To gain more insight into the nature of the Atoh7+ cells and the cell fate decision process, we searched for genes whose expression patterns were highly correlated with that of Atoh7 in the single cell profiles. First, hierarchical clustering was performed on the microarray data derived from each of the 19 Atoh7+ single cells and 38 Atoh7- cells54,160.

The genes in the heatmap shown in Figure 3a clustered with Atoh7 with a correlation coefficient of 0.7. Second, the microarray signal in all of the Atoh7+ single cells was averaged and compared to the average signal from the Atoh7- single cells. For this method, we used specific criteria to identify genes to be included in our analysis. Robust expression had to be observed in at least three of the Atoh7+ cells. For the purposes of this study, 58 robust expression was defined as a signal level that was at least 3X above background (with background being defined as any Affymetrix signal less than 1000 with detection p-values

>0.05). Some of the genes identified by this method overlapped with the hierarchical clustering method and some of the genes were unique to this method (Figure 3b).

To ascertain the exact expression patterns of the genes identified above that clustered closely with Atoh7, we performed ISH on retinal cryosections (Figures 4, 5 and

See Supplementary Fig. S4 online). In Figure 4, we set out to examine the expression pattern of a subset of genes (Rassf4, Rgs16, Rcor2, Gadd45a, Jhdm1d) closely clustered with

Atoh7 by hierarchical clustering (Figure 3a). This was a diverse set of genes whose putative functions included participating in signaling pathways involved in the inhibition of cell growth185,186, controlling the state of histone methylation187,188, and being a stress inducible target of p53189. The potential functions of these genes in cellular growth, differentiation, signaling, and reprogramming of cell lines could be an indication of similar functions in retinal growth and development, but they have not been studied to date in the eye.

We found that both Atoh7 and Rassf4 were strongly expressed throughout the developing ONBL at E12.5 and E14.5 (Figure 4a, 4b, 4d, 4e), consistent with these two genes clustering closely together. While Atoh7 expression remained robust at E16.5 (Figure

4c), Rassf4 expression declined at E16.5 (Figure 4f). Rgs16 expression was observed in a subset of ONBL cells at E12.5 (Figure 4g) and shifted to the scleral side of the ONBL at

E14.5 (Figure 4h) before finally becoming undetectable at E16.5 (Figure 4i). We observed a uniform pattern of expression of Rcor2 throughout the ONBL at all three timepoints (Figure

4j – 4l). Gadd45a was expressed in subsets of cells in the ONBL at E12.5 and at E14.5

(Figure 4m, 4n) before moving to the inner neuroblastic layer (INBL), where the newborn 59 neurons reside, at E16.5 (Figure 4o). Jhdm1d was detected at low levels throughout the retina at E12.5 (Figure 4p), localized to the scleral side of the INBL at E14.5 (Figure 4q) and then was found at lower levels in the full INBL at E16.5 (Figure 4r).

While the genes assayed above showed robust expression in the developing retina, this was not true of all the genes that clustered closely with Atoh7. We also performed ISH on retinal cryosections using probes for Rio2, Ecox2, Ankrd33, Prdm16, Gusb, and Pou2f1

(For all expression patterns, see Supplementary Fig. S4 online). Rio2 was not observed at

E12.5, was detected at low levels in the INBL at E14.5 and continued to be expressed in the

INBL at E16.5. At E12.5 we observed the expression of Ecox2 weakly in the INBL and the expression gained in intensity at E14.5 and E16.5. Ankrd33 was undetectable at E12.5, began to be expressed in a small subset of cells in the scleral ONBL at E14.5 and this staining increased at E16.5. Prdm16 was only weakly detected until E16.5, at which point it was found in the INBL. Gusb starts being expressed at E14.5 in the INBL, where it remains at E16.5. Finally, Pou2f1, a gene previously shown to be involved in lens placode development190, was expressed at low intensity at E12.5 and at E14.5, before it finally localized to the INBL at E16.5. These patterns point to both the fact that Atoh7+ cells are in the process of producing neurons (INBL staining) and to the fact that the single cell transcriptomic protocol is more sensitive than ISH.

To more fully define the single Atoh7+ cells, we also examined the transcriptomes for other genes that were found in subsets of the profiled cells by visual inspection of the data (See Supplementary Fig. S5 online). In order to better understand the nature of this heterogeneity, we performed ISH for these genes (Figure 5). Wdr66 was detected in a substantial subset of the ONBL at all three timepoints (Figure 5a-5c), while Tead2 was 60 found in the ONBL at E12.5 and E14.5 (Figure 5d, 5e) and was reduced to a scattered expression through the developing retina at E16.5 (Figure 5f). Thsd7a was expressed in the developing INBL at E12.5 (Figure 5g), spreading as the number of neurons in the INBL increased at E14.5 (Figure 5h) and E16.5 (Figure 5i). Chrna, however, was expressed in a subset of the ONBL and the INBL at E12.5 (Figure 5j) and at E14.5 (Figure 5k), before becoming more INBL restricted at E16.5 (Figure 5l). Jakmip2 (Figure 5m-5o), Lss (Figure

5p-r) and Nfasc (Figure 5s-u) were all expressed in the INBL at all three timepoints.

Interestingly, Lss was found almost entirely in just the scleral portion of the INBL at E14.5

(Figure 5q), perhaps indicating a role within migrating neurons or developing ACs.

Discussion

In this study, we used the expression of Atoh7 early in retinal development to identify single RPCs that might be actively engaged in making a cell fate decision, and/or producing postmitotic daughter cells that would quickly assume a cell fate. The group of profiled Atoh7 cells also included cells that appeared to have very recently exited the cell cycle and similarly were in the process of choosing a fate. In keeping with this characterization of Atoh7 cells, the transcriptomes of 19 individual cells displayed gene expression characteristics of RPCs. However, a deeper look into the profiles revealed that while many of them did not “score” significantly as RGCs, they nonetheless expressed several genes that are commonly used as early RGC markers191–193. Furthermore, when markers of other early-born retinal cells types were added to the examination, we found that a number of these Atoh7+ cells displayed gene expression hallmarks of more than one 61 retinal neuron. These observations are consistent with the notion that the competence of an RPC to produce a temporally appropriate retinal cell type is indicated by the precocious expression of a marker gene(s) of that specific cell type. Transient expression of cell type- specific proteins within RPCs was previously reported for a marker of RGCs, as well two markers of ACs, only within RPCs from the period of genesis of RGCs and ACs, respectively.

These previous studies showed that this expression was transient as it was extinguished by the majority of daughter cells made by such RPCs. The current data set provides a more comprehensive view into such transient gene expression patterns within RPCs, extending it such that a mixture of markers of multiple cell types can be seen. This gene expression

“mixture” may indicate that a single RPC is competent to produce more than one fate, as has been shown by lineage tracing experiments wherein two distinct retinal cell types can be produced in a single terminal division. The mixture of gene expression patterns is clearly resolved at some point, as such mixtures are not detected in mature retinal neurons.

In addition to indicating the molecular underpinnings of the competence of early

RPCs to produce more than one cell type, the Atoh7+ cell profiles have yielded a number of previously unknown genes that correlate with Atoh7 expression. Using ISH, the expression of these genes has been verified in developing retinal sections. While functional studies will be needed to ascertain their exact role, these genes represent new factors potentially involved in the cell fate decision process176,178,194. The transcriptome of Atoh7 expressing cells has been previously examined either by comparing whole Atoh7 wildtype and Atoh7

KO retinas to identify genes that were differentially expressed195 or by performing RNA-

Seq on FAC-sorted Atoh7-GFP expressing cells172. In the latter study, the authors found a few genes (Isl1, Myt1, Eya2 and Ebf3)60 that were also present in our cluster of Atoh7 62 associated genes. Isl1 is of particular interest since it has been shown to be sufficient for specification of the RGC cell fate together with Pou4f271. Furthermore, finding that Ebf3 is a direct target of Atoh7 further validates our single cell data60. The expression of Myt1, Eya2 and Ebf3 preferentially in Atoh7+ cells in our single cell data strengthens both the possibility that these genes play a role in RGC development and that our cells are heading in an RGC direction. One difference between the results obtained by Gao et al., 2014172 and the present study is the differential regulation of genes belonging to axon guidance and pathfinding pathways. An increase in axon guidance signals in their data led the authors to infer that these cells were further along an RGC differentiation pathway. In contrast, we find that genes involved in axon growth and guidance are expressed in many individual cell transcriptomes harvested from the developing retina but not limited specifically to Atoh7+ cells. There are two differences in the studies that most likely account for the differences in gene expression. First, our study focuses on individual cells and not populations of cells.

This allows for finer resolution of the inner gene expression workings of cells. Second, by their own admission, the H2B-GFP used to isolate Atoh7+ cells remained visible well beyond when Atoh7 expression would normally be undetectable172. This perdurance would lead to some bias through the presence of more mature cells in their dataset.

It has been proposed that the development of RGCs can be divided into three stages: acquisition of RGC competence, cell fate specification, and cell differentiation195. The gene expression data presented here indicates that during development, the timing of Atoh7 expression is correlated with acquisition of RGC competence and the decision to adopt the

RGC fate. This raises the question of why the Atoh7+ cells in the dataset predominantly express early RGC markers even though only about half of RGCs are derived from an 63

Atoh7+ background158. Also, it has been shown that only about 11% of Atoh7+ cells go on to become RGCs158. However, we see large fractions of Atoh7+ RPCs robustly expressing markers of the RGC fate, such as Ebf3, Pou4f2, Isl1, Crmp1, Gap43, and Nefm . One possibility is that the expression of Atoh7 in developing RPCs indicates their competence to produce

RGCs, but that such a cell fate will only be achieved if the cells and/or their daughters experience a certain low level of Notch signaling. Previous studies have shown that RGC production is regulated by reduced Notch signaling, or reduced expression of its transcription partner, RBPj. Along these lines, it is interesting to note that some of the

Atoh7+ cells express Onecut TFs, markers of HCs and the RPCs that produce HCs178,179,194, along with most RGC markers. Other cells expressing Atoh7 express Otx2, a gene that controls photoreceptor and HC cell fate, in addition to one or more RGC markers. RPCs that express Olig2, Onecut1 and Otx2 as well as the early cone marker, Thrb, produce cones and

HCs194,196. The HCs produced by the Olig2+/Otx2+/Oc1+ RPCs go on to downregulate Otx2 and Olig2, while upregulating Oc1, while the cones produced by such RPCs retain Otx2 and lose Oc1 and Olig2. In contrast, later RPCs that produce the majority of rods, as well as ACs,

BPs, and MG, express Olig2 and Otx2, but not Onecut1194,196. Thus, combinatorial expression of some of the same TFs may indicate the competence of RPCs to produce different cell fates, with some of these TFs also playing a functional role in the cell fate process. These observations coupled with the single cell data suggest that we have gained a window into the dynamic gene expression patterns that occur during the time when a cell is competent to produce early cell fates11. Precisely how a cell fate “winner” is determined, however, is not yet understood.

References

1. Livesey, F. J. & Cepko, C. L. Vertebrate neural cell-fate determination: lessons from the retina. Nat. Rev. Neurosci. 2, 109–118 (2001).

2. Masland, R. H. The Neuronal Organization of the Retina. Neuron 76, 266–280 (2012).

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

4. Turner, D. L. & Cepko, C. L. A common progenitor for neurons and glia persists in rat retina late in development. Nature 328, 131–136 (1987).

5. Turner, D. L., Snyder, E. Y. & Cepko, C. L. Lineage-independent determination of cell type in the embryonic mouse retina. Neuron 4, 833–845 (1990).

6. Young, R. W. Cell differentiation in the retina of the mouse. Anat. Rec. 212, 199–205 (1985).

7. Young, R. W. Cell proliferation during postnatal development of the retina in the mouse. Brain Res. 353, 229–239 (1985).

8. Sidman, R. L. Histogenesis of mouse retina studied with thymidine-H3. Struct. Eye, Acad. Press. New York 487–506 (1961).

9. Rapaport, D. H., Wong, L. L., Wood, E. D., Yasumura, D. & Lavail, M. M. Timing and topography of cell genesis in the rat retina. J. Comp. Neurol. 474, 304–324 (2004).

10. Watanabe, T. & Raff, M. C. Rod photoreceptor development in vitro: Intrinsic properties of proliferating neuroepithelial cells change as development proceeds in the rat retina. Neuron 4, 461–467 (1990).

11. Cayouette, M., Barres, B. A. & Raff, M. Importance of intrinsic mechanisms in cell fate decisions in the developing rat retina. Neuron 40, 897–904 (2003).

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

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

14. Rapaport, D. H., Patheal, S. L. & Harris, W. A. Cellular competence plays a role in photoreceptor differentiation in the developing Xenopus retina. J. Neurobiol. 49, 129–141 (2001).

15. Gomes, F. L. et al. Reconstruction of rat retinal progenitor cell lineages in vitro reveals a surprising degree of stochasticity in cell fate decisions. Development 138, 227–35 (2011).

16. Cepko, C. L., Austin, C. P., Yang, X., Alexiades, M. & Ezzeddine, D. Cell fate determination in the vertebrate retina. Proc Natl Acad Sci U S A 93, 589–595 (1996).

17. Isshiki, T., Pearson, B., Holbrook, S. & Doe, C. Q. Drosophila neuroblasts sequentially express transcription factors which specify the temporal identity of their neuronal progeny. Cell 106, 511–521 (2001).

18. Elliott, J., Jolicoeur, C., Ramamurthy, V. & Cayouette, M. Ikaros confers early temporal competence to mouse retinal progenitor cells. Neuron 60, 26–39 (2008).

19. Mattar, P., Ericson, J., Blackshaw, S. & Cayouette, M. A conserved regulatory logic controls temporal identity in mouse neural progenitors. Neuron 85, 497–504 (2015).

20. La Torre, A., Georgi, S. & Reh, T. A. Conserved microRNA pathway regulates developmental timing of retinal neurogenesis. Proc. Natl. Acad. Sci. 110, E2362– E2370 (2013).

21. Boije, H., MacDonald, R. B. & Harris, W. A. Reconciling competence and transcriptional hierarchies with stochasticity in retinal lineages. Curr. Opin. Neurobiol. 27, 68–74 (2014).

22. He, J. et al. How Variable Clones Build an Invariant Retina. Neuron 75, 786–798 (2012).

23. Chen, Z., Li, X. & Desplan, C. Deterministic or Stochastic Choices in Retinal Neuron Specification. Neuron 75, 739–742 (2012).

24. Trimarchi, J. M., Stadler, M. B. & Cepko, C. L. Individual retinal progenitor cells display extensive heterogeneity of gene expression. PLoS One 3, (2008).

25. Mullally, M. et al. Expression Profiling of Developing Zebrafish Retinal Cells. Zebrafish 1–10 (2016). doi:10.1089/zeb.2015.1184

26. Laboissonniere, L. A. et al. Single cell transcriptome profiling of developing chick retinal cells. J. Comp. Neurol. (2017). doi:10.1002/cne.24241

27. Brown, N. L. et al. Math5 encodes a murine basic helix-loop-helix transcription factor expressed during early stages of retinal neurogenesis. Development 125, 4821–4833 (1998).

28. McConnell, S. K. & Kaznowski, C. E. Cell cycle dependence of laminar determination in developing neocortex. Science 254, 282–285 (1991).

29. Brown, N. L., Patel, S., Brzezinski, J. & Glaser, T. Math5 is required for retinal ganglion cell and optic nerve formation. Development 128, 2497–2508 (2001).

30. Wang, S. W. et al. Requirement for math5 in the development of retinal ganglion cells. Genes Dev. 15, 24–29 (2001).

31. Kay, J. N., Finger-Baier, K. C., Roeser, T., Staub, W. & Baier, H. Retinal ganglion cell genesis requires lakritz, a Zebrafish atonal Homolog. Neuron 30, (2001).

32. Feng, L. et al. MATH5 controls the acquisition of multiple retinal cell fates. Mol Brain 3, 36 (2010).

33. Yang, Z., Ding, K., Pan, L., Deng, M. & Gan, L. Math5 determines the competence state of retinal ganglion cell progenitors. Dev. Biol. 264, 240–254 (2003).

34. Brzezinski, J. A., Prasov, L. & Glaser, T. Math5 defines the ganglion cell competence state in a subpopulation of retinal progenitor cells exiting the cell cycle. Dev. Biol. 365, 395–413 (2012).

35. Le, T. T., Wroblewski, E., Patel, S., Riesenberg, A. N. & Brown, N. L. Math5 is required for both early retinal neuron differentiation and cell cycle progression. Dev. Biol. 295, 764–778 (2006).

36. Trimarchi, J. M. et al. Molecular Heterogeneity of Developing Retinal Ganglion and Amacrine Cells Revealed through Single Cell Gene Expression Profiling. J. Comp. Neurol. 504, 287–297 (2007).

37. Boije, H., Edqvist, P. H. D. & Hallböök, F. Horizontal cell progenitors arrest in G2- phase and undergo terminal mitosis on the vitreal side of the chick retina. Dev. Biol. 330, 105–113 (2009).

38. Maiorano, N. a & Hindges, R. Restricted perinatal retinal degeneration induces retina reshaping and correlated structural rearrangement of the retinotopic map. Nat. Commun. 4, 1938 (2013).

39. Shivakumar, L., Minna, J., Sakamaki, T., Pestell, R. & White, M. A. The RASSF1A Tumor Suppressor Blocks Cell Cycle Progression and Inhibits Cyclin D1 Accumulation The RASSF1A Tumor Suppressor Blocks Cell Cycle Progression and Inhibits Cyclin D1 Accumulation. Mol. Cell. Biol. 22, 4309–4318 (2002).

40. Dyer, M. A. & Cepko, C. L. Regulating proliferation during retinal development. Nat. Rev. Neurosci. 2, 333–342 (2001).

41. Yu, S.-Y. et al. A pathway of signals regulating effector and initiator caspases in the developing Drosophila eye. Development 129, 3269–3278 (2002).

42. Irniger, S. Cyclin destruction in mitosis: a crucial task of Cdc20. FEBS Lett. 532, 7–11 (2002).

43. Bischoff, J. R. & Plowman, G. D. The Aurora/Ipl1p kinase family: Regulators of chromosome segregation and cytokinesis. Trends in Cell Biology 9, 454–459 (1999). 68

44. Blackshaw, S. et al. Genomic analysis of mouse retinal development. PLoS Biol. 2, (2004) 45. Gao, Z., Mao, C. A., Pan, P., Mu, X. & Klein, W. H. Transcriptome of Atoh7 retinal progenitor cells identifies new Atoh7-dependent regulatory genes for retinal ganglion cell formation. Dev. Neurobiol. 74, 1123–1140 (2014).

46. Iida, M., Sasaki, T. & Komatani, H. Overexpression of Plk3 causes Morphological Change and Cell Growth Suppression in Ras Pathway-activated Cells. J. Biochem. 146, 501–507 (2009).

47. Archambault, V. & Glover, D. M. Polo-like kinases: conservation and divergence in their functions and regulation. Nat. Rev. Mol. Cell Biol. 10, 265–275 (2009).

48. Zimmerman, W. C. & Erikson, R. L. Polo-like kinase 3 is required for entry into S phase. Proc. Natl. Acad. Sci. U. S. A. 104, 1847–1852 (2007).

49. Goetz, J. J., Laboissonniere, L. A., Wester, A. K., Lynch, M. R. & Trimarchi, J. M. Polo- Like Kinase 3 Appears Dispensable for Normal Retinal Development Despite Robust Embryonic Expression. PLoS One 11, e0150878 (2016).

50. Sapkota, D. & Mu, X. Onecut transcription factors in retinal development and maintenance. Neural Regen. Res. 10, 899–900 (2015).

51. Goetz, J. J., Martin, G. M., Chowdhury, R. & Trimarchi, J. M. Onecut1 and Onecut2 play critical roles in the development of the mouse retina. PLoS One 9, (2014).

52. Wu, F., Sapkota, D., Li, R. & Mu, X. Onecut 1 and Onecut 2 Are Potential Regulators of Mouse Retinal Development. J. Comp. Neurol. 520, 952–969 (2012).

53. Hafler, B. P. et al. Transcription factor Olig2 defines subpopulations of retinal progenitor cells biased toward specific cell fates. Proc. Natl. Acad. Sci. U. S. A. 109, 7882–7887 (2012).

54. Nishida, A. et al. Otx2 homeobox gene controls retinal photoreceptor cell fate and pineal gland development. Nat. Neurosci. 6, 1255–1263 (2003).

55. Brzezinski, J. A., Lamba, D. A. & Reh, T. A. Blimp1 controls photoreceptor versus bipolar cell fate choice during retinal development. Development 137, 619–629 (2010).

56. Furukawa, T., Morrow, E. M. & Cepko, C. L. Crx, a Novel otx-like Homeobox Gene, Shows Photoreceptor-Specific Expression and Regulates Photoreceptor Differentiation. Cell 91, 531–541 (1997).

57. Chen, S. et al. Crx, a Novel Otx-like Paired-Homeodomain Protein, Binds to and Transactivates Photoreceptor Cell-Specific Genes. Neuron 19, 1017–1030 (1997).

58. Morrow, E. M., Furukawa, T., Lee, J. E. & Cepko, C. L. NeuroD regulates multiple functions in the developing neural retina in rodent. Development 126, 23–36 (1999).

59. Eckfeld, K. et al. RASSF4 / AD037 Is a Potential Ras Effector / Tumor Suppressor of the RASSF Family. Cancer Res. 8688–8693 (2004). doi:10.1158/0008-5472.CAN-04- 2065

60. Liang, G. Q., Bansal, G., Xie, Z. H. & Druey, K. M. RGS16 Inhibits Breast Cancer Cell Growth by Mitigating Phosphatidylinositol 3-Kinase Signaling. J. Biol. Chem. 284, 21719–21727 (2009).

61. Yang, P. et al. RCOR2 is a subunit of the LSD1 complex that regulates ESC property and substitutes for SOX2 in reprogramming somatic cells to pluripotency. Stem Cells 29, 791–801 (2011).

62. Huang, C. et al. Dual-specificity histone demethylase KIAA1718 (KDM7A) regulates neural differentiation through FGF4. Cell Res. 20, 154–165 (2010).

63. Yuan, C. et al. The GADD45A (1506T>C) Polymorphism Is Associated with Ovarian Cancer Susceptibility and Prognosis. PLoS One 10, e0138692 (2015).

64. Donner, A. L., Episkopou, V. & Maas, R. L. Sox2 and Pou2f1 interact to control lens and olfactory placode development. Dev. Biol. 303, 784–799 (2007).

65. Xiang, M. et al. Brn-3b: a POU domain gene expressed in a subset of retinal ganglion cells. Neuron 11, 689–701 (1993). 70

66. Zhang, F., Lu, C., Severin, C. & Sretavan, D. W. GAP-43 mediates retinal axon interaction with lateral diencephalon cells during optic tract formation. Development 127, 969–80 (2000).

67. Mu, X. Discrete gene sets depend on POU domain transcription factor Brn3b/Brn- 3.2/POU4f2 for their expression in the mouse embryonic retina. Development 131, 1197–1210 (2004).

68. Emerson, M., Surzenko, N., Goetz, J., Trimarchi, J. & Cepko, C. Otx2 and Onecut1 promote the fates of cone photoreceptors and horizontal cells and repress rod photoreceptors. Dev. Cell 26, 59–72 (2013).

69. Mu, X. et al. A gene network downstream of transcription factor Math5 regulates retinal progenitor cell competence and ganglion cell fate. Dev. Biol. 280, 467–481 (2005).

70. Wu, F. et al. Two transcription factors, Pou4f2 and Isl1, are sufficient to specify the retinal ganglion cell fate. Proc. Natl. Acad. Sci. U. S. A. 112, E1559-68 (2015).

71. Emerson, M. M. & Cepko, C. L. Identification of a retina-specific Otx2 enhancer element active in immature developing photoreceptors. Dev. Biol. 360, 241–255 (2011).

72. Nirenberg, S. & Cepko, C. Targeted ablation of diverse cell classes in the nervous system in vivo. J. Neurosci. 13, 3238 LP-3251 (1993).

73. Sturn, A., Quackenbush, J. & Trajanoski, Z. Genesis: cluster analysis of microarray data. Bioinformatics 18, 207–208 (2002).

74. Kim, D. S. et al. Identification of molecular markers of bipolar cells in the murine retina. J. Comp. Neurol. 507, 1795–1810 (2008).

75. Cherry, T. J., Trimarchi, J. M., Stadler, M. B. & Cepko, C. L. Development and diversification of retinal amacrine interneurons at single cell resolution. Proc Natl Acad Sci U S A 106, 9495–9500 (2009).

76. Roesch, K. et al. The transcriptome of retinal Müller glial cells. J. Comp. Neurol. 509, 225–238 (2008).

Figures and Legends

Figure 1: Single cell transcriptional profiles of selected ganglion cell genes. A heatmap was generated using Genesis software that displays the expression of the designated genes in single cells picked from retinas isolated from mice between ages E12 and E16. The intensities from the Affymetrix signals have been scaled such that a signal of 0 corresponds to a black color and a signal of 2500 corresponds to the bright red color. All intermediary signals are graded in the corresponding shades of red and black per their value. Expression of the ganglion cell markers, Ebf3, Pou4f2, Isl1, Crmp1, Nefm and Nefl is shown here.

Figure 2: Single cell transcriptional profiles of selected neuronal markers. A heatmap was generated using Genesis software that displays the expression of the designated genes in Atoh7+ single cells collected from retinas isolated from mice between ages E12 and E16.

The intensities from the Affymetrix signals have been scaled such that a signal of 0 corresponds to a black color and a signal of 2500 corresponds to the bright red color. All intermediary signals are graded in the corresponding shades of red and black per their value. Expression of RGC markers (Pou4f2 and Nefl), an AC marker (Tcfap2b), HC and cone precursor marker (Olig2), HC markers (Onecut1 and Onecut2) and PR marker genes (Otx2,

Prdm1 and Crx) are shown here. 73

Figure 3: Hierarchical clustering of Atoh7 and related genes expressed in single retinal cells. (a) Hierarchical clustering was performed on transcriptome data derived from single cells isolated between E12 and E16 using Cluster software (Eisen lab) and visualized using a heatmap generated by Genesis. 53 genes that clustered closely with

Atoh7 are shown here. (b) A Genesis generated heatmap showing the expression of closely clustered genes 32 single cells. The intensities of Affymetrix signals have been scaled such 74 that a signal of 0 corresponds to a black signal and a signal of 2500 corresponds to a bright red color.

Figure 4: Expression of genes closely associated with Atoh7 in the developing retina.

ISH was performed on retinal cryosections at E12.5, E14.5 and E16.5 using probes for

Atoh7 (4a-4c), Rassf4 (4d-4f), Rgs16 (4g-4i), Rcor2 (4j-4l), Gadd45a (4m-4o), Jhdm1d (4p-

4r). Scale bars represent 100um.

Figure 5: Expression of genes closely associated with Atoh7 in the developing retina.

ISH was performed on retinal cryosections at E12.5, E14.5 and E16.5 using probes for

Wdr66 (4a-4c), Tead2 (4d-4f), Thsd7a (4g-4i), Chrna (4j-4l), Jakmip2 (4m-4o), Lss (4p-4r) and Nfasc (4s-4u). Scale bars represent 100um.

Supplemental Table 1. The sequences of the primers used to amplify specific genes from mouse retinal cDNA for generating RNA probes used for Dissociated ISH (DISH) are shown in this table.

Probe name BMAP probe Forward primer Reverse primer collection name Atoh7/Math5 BE951657 Rassf4 gtcaaagagtggtggcgttt tccgttcttgagcagtgttg Rgs16 ctcattcttgccgtctcaca gacaatcggcacaacacaac Wdr66 ggatggccctttcctatgat aatgagctgttccgagctgt Rcor2 gtaccagccacagccctaga aaggtgaatgaggccagcta Tead2 accatcctccaggttgtgac ggtggtccaaggctcactta Glnb caacgccaaatatgatgcag cttcactccagcctctcacc Jhdm1d tcagggacctgaggtgtaactt aatgctcttaaagctccagca Gadd45a cagagcagaagaccgaaagg tgaaagtaacctggccatcc Pou2f1 actctgctgggtgctttcac cagtcatcagtgcctccaga 11989 aagggaaatgctggaaacct aagtgtcacccaggaccaag Chrna3 gctaacgtgtcccatcctgt ggcaggtagaagacgagcac Jakmip2 agcaggtggatgaggctcta tggtttcgagcttctcctgt Lss gagccttcttagccctgctt aggccagaagatgacattgg Neurofascin agtaaagccatggcaccaac ataggcaaaccaggacaacg Cdt1 ttgatgacacccaagatgga agatccatgggcaacgatag Geminin ggagcccaagagaatgtgaa gtgatagccccaggattcaa Plk1 acgtcgtaggcttccatgac tggtgaggcaggtaataggg Rio2 tttccataatttactggctttgg ataaaggatggcggcttacc Ecox2 agaagaccgcagtcagcaat tgaactggtgaagcaccttg Atat1 tccctccttccagattcctt ctctgtagccttggctggac Ankr33b tgctcttggaacagcacact tggctagcaagcgacttacc Prdm16 ttccagccattgtctgtctg acggcgatttccatagtcac Dlx1 aaggtcctcgtgtgtcggct tttaaagcgacctggaattg Dlx2 ctgctgaggtcactgctacg atcctcagggtccttggtct Plk3 AW488956 77

Supplemental Table S2. List of probes used for Dissociated ISH (DISH) are listed in this table.

Probe name BMAP probe collection name cdc20 BE988025 Delta1 AW047187 Neurod1 Morrow et al 1999 Trb2 BX513184 Ap2b BE953140 Nefl BE953485

Supplemental Figure S1: (a) Single cell transcriptional profiles of cell cycle genes. A heatmap generated using Genesis software displays the expression of markers of G1/S and 78

G2/M phases of the cell cycle in individual cells picked from retinas isolated from mice between E12 and E16. The intensities from the Affymetrix signals have been scaled such that a signal of 0 corresponds to a black color and a signal of 2500 corresponds to the bright red color. (b-j) Expression of cell cycle markers in the developing retina. ISH was performed on retinal cryosections at E12.5, E14.5 and E16.5 using probes for Cdt1 (b-d),

Geminin (e-g) and Plk1 (h-j). Scale bars represent 100um.

Supplemental Figure S2: Co-expression of G2/M markers and RGC markers. A heatmap was generated using Genesis software that displays the expression of a subset of

G2/M markers genes and RGC markers genes in single cells harvested from retinas isolated from mice between ages E12 and E16. The intensities from the Affymetrix signals have been scaled such that a signal of 0 corresponds to a black color and a signal of 2500 corresponds to the bright red color. All intermediary signals are graded in the corresponding shades of red and black per their value. 79

Supplemental Figure S3: Co-expression of cell cycle and neuronal markers with

Atoh7 using Dissociated ISH. Dissociated ISH was performed on dissociated retinal cells using fluorescently labeled probes for Cdc20 (a-c), Delta1 (d-f), Neurod1 (g-i), Tcfap2b (j-l),

Trb2 (m-o) and Nefl (p-r). The percentage of overlap of the mentioned genes with Atoh7 was quantified in (b).

Supplemental Figure S4: Expression of genes closely associated with Atoh7 in the developing retina. ISH was performed on retinal cryosections at E12.5, E14.5 and E16.5 using probes for Rio2 (a-c), Ecox2 (d-f), Ankrd33 (g-i), Prdm16 (j-l), Gusb (m-o), and Pou2f1

(p-r). Scale bars represent 100um.

Supplemental Figure S5: Genes associated with Atoh7 expression. Individual transcriptomes from Atoh7+ cells were compared to Atoh7- cells isolated between E12 and

E16. The heatmap for seven genes whose expression was enriched in Atoh7+ cells was generated using Genesis software. The intensities from the Affymetrix signals have been scaled such that a signal of 0 corresponds to a black color and a signal of 2500 corresponds to the bright red color. All intermediary signals are graded in the corresponding shades of red and black per their value.

CHAPTER 3. TRIM GENES IN THE RETINA: EXPRESSION AND FUNCTIONAL CHARACTERIZATION

Abstract

To better understand the mechanisms that govern the development of retinal neurons, it is critical to gain additional insight into how specific intrinsic factors control cell fate decisions and neuronal maturation. In the developing mouse and zebrafish retina, Atoh7, a highly conserved transcription factor, is essential for retinal ganglion cell development. We performed transcriptome profiling of Atoh7+ individual cells isolated from mouse retina.

Through the examination of these cells, we were able to gain valuable insights into the process of how cells choose their fates. One of the genes that we found correlated with

Atoh7 in our transcriptomic data was the E3 ubiquitin ligase, Trim9. The correlation between Trim9 and Atoh7 and the expression of Trim9 in the early mouse retina led us to hypothesize that this gene could play a role in the process of cell fate determination. We performed a thorough functional analysis of Trim9 in the mouse and using CRISPR-Cas9 genome editing, in zebrafish. However, we did not find any morphological changes in the retina in the absence of Trim9 in either organism. Trim9 deletion was previously shown to cause defects in axon branching in the cerebral cortex. However, the retinal ganglion cell axons in the Trim9 KO mice appeared normal. Thus, Trim9 alone does not appear to be involved in cell fate determination or early ganglion cell development in mouse and zebrafish retinas. We further hypothesize that the reason for this lack of phenotype may be 83 compensation by one of the many additional Trim family genes we find expressed in the developing retina.

Introduction

The retina has emerged as a powerful tool for studying the central nervous system and has been intensively investigated for over a century1. It is organized as a laminar tissue, comprised of six different neuronal cell types and one glial cell type. These functionally and morphologically diverse groups of cells arise from a pool of multipotent retinal progenitor cells (RPCs)2–5. In the mouse, RPCs begin generating neurons at about embryonic day (E)11.5. Birthdating studies have demonstrated that the retinal ganglion cells (RGCs) are the first retinal neurons to be born, followed closely by cone photoreceptors, horizontal cells and then amacrine cells6–9. The bipolar cells and Müller glia are born later in development, while rod photoreceptors are generated nearly throughout the developmental process6–9. One key question that arises in this context is how RPCs that are yet to choose a cell fate make the decision to generate a particular cell type. In an effort to better understand the process of cell fate determination in the retina, single cell transcriptomes of RPCs at various developmental stages were acquired and analyzed10. Mining these transcriptomes revealed a large number of new marker genes and a significant amount of gene expression heterogeneity, particularly among transcription factors10.

One such transcription factor was the well-studied Atonal homologue 7 (Atoh7), a basic helix-loop-helix (bHLH) transcription factor whose expression in RPCs is temporally 84 correlated with cell cycle exit and cell fate specification11. Furthermore, loss of Atoh7 function in the vertebrate retina leads to an almost complete loss of RGCs12–16. Lineage tracing studies have shown that other early born retinal neurons also arise from these

Atoh7+ progenitor cells, thus this transcription factor is thought to confer competence to

RPCs rather than causing differentiation into a particular neuronal type13,17,18. Moreover, studies examining the precise role of Atoh7 in RGC development remain ambiguous. For example, ectopic expression of Atoh7 in mouse retinas under the control of Crx promoter did not stimulate RGC production19. Similarly, retinal explants infected with an Atoh7 expressing retrovirus did not create an RGC bias 19. However, other studies testing the effects of Atoh7 overexpression in Müller glia or stem cells, found an increased production of RGCs20,21. Despite the ambiguity in these results, the lack of a phenotype upon overexpression of Atoh7 in vivo suggests that Atoh7 while necessary, is not sufficient to produce RGCs18. Taken together, these experiments suggest that factors other than ATOH7 are most likely involved in the process of RGC specification, at least in vivo.

In an attempt to identify some of these other factors involved in the cell fate specification of early-generated retinal cell types, the transcriptomes of single mouse RPCs were examined. Upon further examination of these transcriptomes, we have found that one of the highly represented gene families in these cells was the Tripartite motif (Trim) family.

In this current study, we have examined the expression of different Trim family genes in the developing mouse retina in more detail. Through in situ hybridization (ISH), we found at least 24 Trim family genes expressed during early retinal development in the mouse.

Since Atoh7 expression is conserved in early retinal development and associated with RGC competence17,18, we decided to focus on genes whose expression was correlated with 85

Atoh7, hypothesizing that they may play a role in deciding the fate of retinal neurons during development22–24. Of the Trim family genes, Trim9 was chosen for further study as its expression was both correlated with Atoh7 by gene clustering and was observed in subsets of Atoh7+ single cell transcriptomes. Furthermore, the heterogeneity of Trim9 expression indicated that its potential role in the retina might be in only a subset of cells.

TRIM9, a member of the tripartite motif containing (TRIM) family of E3 ubiquitin ligases, has been found in the developing and adult central nervous system25,26. TRIM9 immunoreactivity was shown to be diminished in affected brain areas in Parkinson’s disease and dementia with Lewy bodies, indicating a possible role for TRIM9 in neurodegenerative diseases25. Although TRIM9 has been shown to have ubiquitin ligase activity in vitro25, its role in axon guidance was not known until recently, when analysis of

Trim9 knockout mice established that TRIM9 mediates the response of cortical neurons to

Netrin-1 through interactions with DCC26. It was further demonstrated that TRIM9 prevents axon branching in the absence of Netrin-1 and the ligase activity of TRIM9 is required to impart sensitivity to Netrin-1 and promote axon branching. In the absence of

TRIM9, the cortical axons showed exaggerated axon branching and a reduced sensitivity to

Netrin-126. More recently, it was demonstrated to play a unique role in cortical axon guidance. TRIM9 was shown to ubiquitinate VASP, an actin regulatory protein located at the tips of filopodia, in the presence of Netrin-1 to produce a spatial gradient of filopodial stability required for axon turning toward netrin, thereby regulating axon pathfinding in the cortex27. In addition to these molecular and cellular phenotypes, severe deficits in spatial learning and memory were observed in Trim9 knockout mice28. Here, we show that

Trim9 expression, in addition to being correlated with that of Atoh7, is detected in multiple 86 stages of the developing mouse retina. Furthermore, we find that Trim9 is expressed in a subset of Atoh7+ cells analyzed from the developing zebrafish retina29. However, we observed no significant alteration of retinal morphology upon disruption of Trim9 in both a mouse knockout and a CRISPR-generated zebrafish mutant. Given the substantial number of additional Trim family genes expressed in the developing retina, it could be either that

Trim9 is not required for cell fate determination or that compensatory mechanisms exist within this gene family in the developing retina.

Materials and Methods

Ethics Statement

All procedures for the care and housing of mice conform to the U.S. Public Health

Service Policy on the Humane Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee at Iowa State University.

Zebrafish care

Zebrafish were grown on a 14 hr light/10 hr dark cycle in tanks on an Aquatic Habitat system (Aquatic Ecosystems) at 28°C. All experimental protocols were approved by the

Iowa State University Institutional Animal Care and Use Committee (IACUC).

Mouse Genotyping

The generation of Trim9 deficient mice has been described previously26. Specific primers were used to detect the KO band [F: 5’ - CTTCTGAGGGTTGGAGAAAAGC - 3’ and R:

5’ - CGTGAGAGCTGCTTTCTTATTGG- 3’] and the WT band [F: 5’ -

CTTCTGAGGGTTGGAGAAAAGC - 3’ and R: 5’ - CGACGGTATCGATAAAGCTAGCTTGG - 3’]. All three primers were used in the same reaction according to the following program: 3 min. at

95°C; then 37 cycles of 1 min. at 94°C, 1 min. at 58°C, 90 seconds at 72°C; followed by 10 min. at 72°C. The WT and KO PCR product sizes were 410 bp and 346 bp, respectively, and were separated on a 2% agarose gel.

Tissue processing for cryosectioning

Retinas were collected from adult mice as well as mice from 3 different stages of development, ranging from E12.5 to E16.5. Pregnant females were euthanized, embryos collected and heads incubated in 4% paraformaldehyde (PFA) in phosphate buffered saline

(PBS) overnight (O/N) at 4°C. Similarly, adult WT and Trim9 KO littermate pairs were euthanized and the eyes placed in 4% PFA/PBS overnight. The eyes were subjected to three 15 min. washes in PBS, after which the retinas were isolated and rocked in 30% sucrose in PBS until they sank. OCT solution (Tissue-Tek) was added at a 1:1 ratio with the

30% sucrose in PBS and rocked until the solution reached equilibration. Retinas were stored at -80°C until they were cryosectioned at 20μm and placed onto Superfrost Plus microscope slides. 88

Zebrafish were euthanized using five to six drops of 4mg/mL MS-222 Tricaine

Methanesulfonate per mL of fish water (1.25 g sea salt/20 L dH2O). Immediately upon euthanasia, fish were placed in 4% PFA/PBS and tissue processing for zebrafish was carried out exactly as described previously29.

In situ hybridization (ISH)

Riboprobe synthesis was performed as described previously22. Briefly, RNA probes

(650-800 bp in length) were synthesized by PCR amplification using primers specific for mouse cDNA, listed in Supplemental Table 1. The probe template sequences were cloned into pGEM-T vector (Promega) and sequenced. Antisense riboprobes were synthesized using either T7 or SP6 RNA polymerase (depending on clone orientation) in the presence of

Digoxigenin for 1-2 hours at 37°C. Riboprobes were treated with DNase I for 15 min. and precipitated with 100% Ethanol and LiCl overnight.

In situ hybridization on retinal cryosections was performed as described previously30. Briefly, the slides were washed with PBS three times, fixed with 4% PFA in

PBS, acetylated and the riboprobes hybridized overnight at 65°C. The next day, slides were incubated in a 1X SSC [diluted from 20X saline sodium citrate- 3M NaCl, 0.33M Sodium

Citrate, pH 7] buffer containing 50% formamide, treated with RNase A and washed with 2X and 0.2X SSC. After washing twice with TNT (0.1 M Tris-HCl, pH 7.5, 0.15 M NaCl, 0.05%

Tween-20), slides were blocked for an hour with 20% heat inactivated sheep serum (HISS) and incubated with anti-DIG-alkaline phosphatase (anti-DIG-AP) antibody (1:2500, Roche) overnight. The next day slides were washed with TNT and developed using NBT and BCIP. 89

Finally, the slides were fixed in 4% PFA/PBS and mounted with Fluoromount-G (Southern

Biotech).

Whole-mount in situ hybridization (WM-ISH)

Zebrafish WM-ISH was performed exactly as described in Mullaly et al., 201629.

Briefly, 28 hours post fertilization (hpf) zebrafish were subjected to treatment with 50%

PTU/50% fish water to ensure the lack of pigment formation. Fish were collected at 48hpf, and dehydrated through a series of increasing concentrations of ethanol in PBT

(PBS/0.1%Tween-20) washes. After an O/N incubation at -20°C, the fish were rehydrated through several washes of gradually decreasing ethanol concentrations. The remaining protocol was followed exactly as described previously29. The primers used for making the probes used are listed in Supplemental Table 1.

Section Immunohistochemistry [IHC]

Slides containing cryosections were incubated for 30 min. with blocking solution

(1% BSA, 0.01% Triton X-100, 0.004% SDS) and then placed in primary antibody, diluted according to manufacturer’s instructions in blocking solution. Slides were washed in blocking solution three times for 15 min. at room temperature. Slides were then placed in secondary antibody, diluted 1:300 with blocking solution and incubated at 4°C O/N.

Following this incubation, the slides were again washed three times for 15 min. with blocking solution and mounted with Fluoromount-G.

For adult zebrafish sections, an initial antigen retrieval step was performed. Slides containing cryosections were placed in sodium citrate buffer (10 mM Tri-sodium citrate, 90

0.05% Tween 20, pH 6.0). The bucket containing slides placed in sodium citrate buffer was microwaved for 5 min. at ‘popcorn’ settings, and then for 15 min. at the lowest power level.

This allowed the slides to be brought just to a boil and then incubated at that point for the

15 min. The slides were then cooled by placing them in water and immunohistochemistry was performed as described above.

Whole-mount IHC

The process was carried out exactly as described previously23. Briefly, mice were euthanized and the eyeballs fixed in 4% PFA/PBS overnight. After three 15 min. washes, retinas were dissected and equilibrated in increasing concentrations of sucrose (10%, 20% and 30% w/v sucrose) for 20-30 min. each. After the retinas sank in 30% w/v sucrose solution they were subjected to three freeze-thaw cycles on dry ice. The frozen retinas were stored at -80°C until ready to continue with IHC. To proceed, retinas were washed three times for 30 min. with PBS, and then rocked gently in blocking solution [3% goat serum/1% bovine serum albumin (BSA)/0.1% Triton-X100/0.02% sodium dodecyl sulfate

(SDS) in PBS] for 2 hours at room temperature. Retinas were then incubated in primary antibody in blocking solution overnight at 4°C on a rocker. The retinas were then washed with PBS three times for 30 min. each and incubated in secondary antibody at a 1:300 dilution in blocking solution overnight at 4°C on a rocker. Retinas were washed three times with PBS for 30 min. each and then flattened between two coverslips and imaged using the

Leica SP5 XMP confocal microscope at Iowa State University.

Primary antibodies used were anti-Calbindin28K (Calb) (1:2000; Swant,

Switzerland), anti-Calretinin (Calr) (1:1000; Millipore), anti-Vsx2 (1:1000;31), anti-Tcfap2α 91

(1:200; Santa Cruz Biotechnology), anti-Rhodopsin, (1:100;32), anti-Choline

Acetyltransferase (Chat, 1:100; Millipore), anti-Protein kinase C-alpha (PKC, 1:10,000;

Sigma-Aldrich), anti-Brn3a (1:500; Chemicon MAB1585), anti-Opn4 (1:1000; Advanced

Targeting System). The anti-Zn8 (1:40) and anti-Islet1 (1:20) antibodies were obtained from the Developmental Studies Hybridoma Bank (DSHB), developed under the auspices of the NICHD and maintained by the University of Iowa, Department of Biology, Iowa City.

Secondary antibodies used were AlexaFluor 488 AffiniPure donkey anti-rabbit IgG,

AlexaFluor 488 AffiniPure goat anti-mouse IgG, AlexaFluor 488 AffiniPure donkey anti-goat

IgG (Jackson ImmunoResearch laboratories).

qPCR

qPCR was performed as described previously23. RNA was isolated from the retina using Tri-reagent (Sigma) according to manufacturer’s instructions. 400ng of RNA was used to generate cDNA using random primers and SuperScript III (Life Technologies) according to standard protocols. SybrGreen MasterMix (Thermofisher) was used to perform qPCR in a Bio-Rad MiniOpticon cycler, using the following program: 15 min. at

95°C and 40 cycles of 15 sec. at 95°C, 30 sec. at 56°C, and 30 sec. at 72°C. The analysis of the qPCR data was performed exactly as described previously22. Each sample was normalized to -actin to obtain the ΔCt values. The difference in ΔCt values between experimental and control was designated as ΔΔCt. The experiments were repeated three times and the average ΔΔCt value was calculated. The base 2 analogs of the average ΔΔCt value represented the fold change. These results were plotted graphically with error bars shown. 92

qPCR primers were designed as follows: Trim9 (F: ATGGAAGAGATGGAAGA, R:

CGCACGCCTGACATAAATTG), Brn3a (F: CACAAGTACCCGTCGCTG, R:

GACACCGCGATGTCCACG), Drd2 (F: GTTATGCCCTGGGTCGTCTA, R: GATGGC

ACACAGGTTCAAGA), Drd4 (F: CTGCAGACACCCACCAACTA, R: CCATGAG CGTGTCACAGAGA

), Tbr2 (F: GACACCGCGATGTCCACG, R: TGCATGTT ATTGTCCGCTTT), MMP17 (F:

CACCACTGCTGCTTGTACTG, R: GTAGCCAAACCTGCTTAGCC ), Cdh6 (F:

GAATGAGCTGAGCCGTTCG, R: TAATGAAGAGATCGCCCGCT ), Unc5d (F:

CTCCCATGGCTAGGACTCTT, R: CCTCGATGAAATGAGGCAGC ), Jam-2 (F:

ACCACCGTCAAGAAGTCACA, R:TGTCCCACCTTCTTCCACTC ), Spig1 (F:

GAACCACTGTGAGCTTCGC, R:TTGAGACGAGCATAGCCAGC ).

Microarrays

Microarray hybridization was performed as described previously23. Briefly, RNA was isolated from retinas using Tri-reagent (Sigma) according to standard manufacturer’s protocols.

400ng of total RNA was used to generate aRNA, and 5μg was fragmented using the Ambion

MessageAmp™ II aRNA Amplification Kit according to manufacturer’s instructions. Samples were hybridized to Affymetrix GeneChip Mouse Genome 430 2.0 arrays at the Gene-Chip facility at Iowa State University.

Microarray results were analyzed using the Bioconductor Affy package in R33. RMA was employed for background adjustment and normalization34. The fold change between

WT and Trim9 KO retinas was considered significant if it was greater than 2.

CRISPR synthesis

Potential CRISPR target sites were identified using the CRISPR design program at http://zifit.partners.org/ZiFiT/ChoiceMenu.aspx. Oligos used to target the desired site were:

5’ – TAGGCCTGCGCGAGAAACATCC - 3’ and 5’ – AAACGGATGTTTCTCGCGCAGG - 3’.

For Cas9 synthesis, the pT3TS-nCas9n plasmid was used35. Guide RNA (gRNA) was synthesized as described previously36 and purified using the miRNeasy mini kit (Qiagen).

Cas9 was transcribed using the Xba1 digested pT3TS-nCas9n vector and T7 mMESSAGE mMACHINE T7 ULTRA kit (Life Technologies). It was purified using the RNeasy MinElute

Cleanup kit (Qiagen).

Microinjection of zebrafish

One nl of solution containing 25pg of gRNA and 300pg of Cas9 mRNA was injected into one-cell stage zebrafish embryos. On the next day, any deformed or dead embryos were removed. At 48hpf, 5 embryos from the injected and uninjected population were collected and genomic DNA was extracted as described below.

Genomic DNA isolation and genotyping

To extract genomic DNA from zebrafish embryos, 100 µl of 0.05M NaOH was added to each tube containing 5 zebrafish. The tubes were incubated at 95°C for 30 min., cooled and 10 µl 1M Tris, pH 8 was added. Targeted genomic DNA loci were amplified using specific primers designed around the targeted sites to form approximately 100 base-pair 94 amplicons (F: 5’ – CTGCCCTGCTCGCATAATA – 3’ and R: 5’ –

CAGGTCTAAGTAATCATAATCGGAGA – 3’). The PCR product was run on a 2.5% agarose gel to achieve distinct separation of bands corresponding to wildtype and mutated DNA.

Sequencing of mutated endogenous gene target sites

To investigate the nature of the mutation, CRISPR injected zebrafish were mated to wildtypes (WIKs) and the resulting heterozygote zebrafish analyzed for mutations as described above. The resulting PCR products were then TA cloned into the pGEM-T plasmid

(Promega). Following transformation, plasmids were isolated and sequenced at the Iowa

State University DNA facility. Single base pair insertions or deletions (in/dels) were not utilized as mutant alleles because the possibility that these alterations were either generated during the PCR reaction or were a sequencing error could not be excluded36.

Only in/dels of 4 bp or more were considered for usage as mutant alleles in our study.

Results

Retinal expression of Trim family genes

The transcriptomes of single developing RPCs were analyzed to better understand the mechanisms of cell fate determination operating within the retina. Single cells at various times of development were isolated from mouse retinas and microarrays performed in order to assess gene expression in the retina10. TRIM/RING-B-box-coiled-coil

(RBCC) proteins are involved in a wide range of developmental processes and therefore implicated in several pathological conditions from genetic diseases to cancer 95 development37. Deletion of Trim2 in mice led to a reduction in the number of retinal ganglion cells later in life, but any phenotypes associated with other Trim family genes have not been reported38. In this study, we began by assessing the expression of Trim genes in the retina. Since the gene CyclinD1 (Ccnd1) is a marker of cycling RPCs39 we first examined the transcriptomes of single RPCs expressing Ccnd1 at various stages of development (E12.5, E13.5, E14.5, E15.5, E16.5, and post-natal day (P)0. We found the Trim family of genes widely expressed in cycling progenitor cells (Figure 1A). More specifically, the family members Trim28, Trim35 and Trim44 showed the widest expression among the most RPCs (Figure 1A). Many of the Trim family genes displayed more heterogeneous expression patterns, appearing in a significant number of RPC profiles (e.g. Trim27, Trim32, and Trim41) or appearing in very few RPC profiles (e.g., Trim3, Trim8, Trim11 Trim 62 and

Trim66).

To further characterize the expression of the Trim genes in the developing retina, we next focused on their expression in subsets of single cells that appeared to be developing amacrine and ganglion cells (Figure 1B). We decided to use Neurofilament-light

(Nefl) as a marker of developing RGCs40 and Tcfap2β as a marker of developing amacrines30. Of all the genes examined, only Trim36 and Trim39 were found to be expressed in Nefl+ cells and absent from Tcfap2+ cells (Figure 1B) and even these two genes were only found in subsets of the Nefl+ profiles. The majority of the Trim family genes observed in these cells were found in subsets of both Nefl+ and Tcfap2+ cells, perhaps indicating a role for this family in general neuronal development.

Since Atoh7 has been shown to play an important role in the development and differentiation of early born retinal neurons12–16, Atoh7 positive single cells were isolated 96 from developing mouse retinas and their transcriptomes examined in an effort to identify new retinal cell fate regulators. We observed a large number of Trim family genes expressed in these Atoh7+ cells as well, similar to our results in RPCs and in

Nefl+/Tcfap2+ cells (Supplemental Figure 1). However, when hierarchical clustering was used to identify the genes most closely associated with Atoh7 expression, only one Trim family member remained, Trim9. Even though they were significantly correlated, we found that Trim9 was expressed only in a small subset of Atoh7+ cells, perhaps indicating either that Trim9 expression is quite dynamic or that this gene only plays a functional role in a subset of retinal neurons.

To further characterize the expression of the Trim genes, in situ hybridization (ISH) was performed on retinal cryosections derived from embryonic mice at various developmental stages (Figure 1C-1I’’). Consistent with our single cell transcriptomic result, we observed a number of Trim family genes robustly expressed within the layer of developing retinal neurons (inner neuroblastic layer [INBL])41 and the RPCs (outer neuroblastic layer [ONBL])41 at the developmental stages, E12.5, E14.5 and E16.5. The expression of Trim62 was found to be specific to subsets of the INBL, and absent from the

ONBL (Figure 1C, 1C’, 1C’’). The genes Trim2 and Trim46 were expressed in retinal sections throughout development specifically in the INBL (Figure 1D, 1D’, 1D’’, 1E, 1E’, 1E’’). Trim32 was found to be weakly expressed throughout the INBL and ONBL at E12.5 and E14.5, finally localizing to the INBL at E16.5 (Figure 1F, 1F’, 1F’’). Trim44 expression was also diffuse through the retina at E12.5 before localizing weakly to the INBL at E14.5, and continued to be express ed in the INBL at E16.5 (Figure 1G, 1G’, 1G’’). Trim28 expression differed from that of the remaining Trim genes in that it remained dispersed throughout 97 both ONBL and INBL at all stages of development (Figure 1H, 1H’, 1H’’). At E12.5, Trim9 was expressed in subsets of cells in the ONBL (Figure 1I). By E14.5, its expression was confined to the INBL and the outer portion of the ONBL (Figure 1I’). By E16.5, Trim9 expression was limited to the INBL (Figure 1I’’).The expression of many of the Trim genes in both Nefl+ and Tcfap2+ cells is consistent with their expression in the INBL where both

RGCs and displaced amacrine cells reside41. Of all the Trim genes, Trim9 is the only one that clustered with Atoh7, a gene known to be involved in imparting competence to RGCs, suggesting that it may play a role in early cell fate specification in the retina.

Characterization of Trim9 KO retinas

Given the expression of Trim9 in a subset of single developing Atoh7 positive progenitors and its expression in the developing retina (Figure 1), we hypothesized that

Trim9 may play a role in cell fate determination within the retina. To this end, we obtained a Trim9 knockout mouse and examined the expression of markers of ganglion cells, amacrine cells, cones and horizontal cells in mature Trim9 deficient retinas. We focused on the early-generated retinal neurons because of the expression of Trim9 and because the phenotypes of Atoh7 deficient animals are primarily associated with early born retinal neurons12–17. We used in situ hybridization on adult wildtype and Trim9 knockout retinal cryosections to assess whether the absence of Trim9 affected the development of early born retinal neurons (Figure 2). We examined populations of ganglion cells using probes to

Synuclein- (Syn-)30,42 and Cartpt43 (Figure 2A, 2A’ and 2B, 2B’), amacrine cells by expression of Tcfap244 (Figure 2C, 2C’), GABAergic amacrine cells by means of Gad145

(Figure 2D, 2D’), glycinergic amacrine cells by Slc6a9 (Figure 2E, 2E’)45, horizontal cells 98 with a probe to Septin441 (Figure 2F, 2F’), cone photoreceptors by short wavelength Opsin

(Opn1sw) (Figure 2G, 2G’), rod photoreceptors by expression of Nrl46 (Figure 2H, 2H’), bipolar cells by Vsx231 (Figure 2I, 2I’), and Müller glia by Clusterin (Clus)41 (Figure 2J, 2J’).

However, no major changes were observed in these major classes and subclasses of retinal neurons in the Trim9 deficient retinas.

Next, we wished to assess if there were any changes in broad populations of retinal neurons that we failed to detect by in situ hybridization. We thus performed antibody stains on retinal cryosections from Trim9 deficient mice and their wildtype littermates. We inspected populations of horizontal, amacrine and ganglion cells (antibodies to Calbindin-

28k45 and Calretinin45) (Figure 3A, 3A’), amacrine cells (anti-Tcfap2α44 and anti-ChAT47,

Figure 3B, 3B’; 3C, 3C’), rod photoreceptors (anti-Rhodopsin32, Figure 3D, 3D’) bipolar cells

(antibodies to Vsx231 and PKC-45) (Figure 3E, 3E’; 3F, 3F’).The Trim9 knockout mice failed to display any significant changes in their populations of retinal neurons compared to their wildtype littermates.

It is possible that any smaller differences might have been overlooked on retinal sections. Therefore, to more rigorously quantify any differences in retinal cell number between the wildtype and Trim9 knockout mice, we performed immunohistochemistry on flat-mounted retinas. For each antibody used, we separated the retina into four different quadrants and quantified using images taken from each distinct quadrant. We used anti-

Calbindin28k and anti-Calretinin antibodies to visualize a combination of amacrine, horizontal and ganglion cells, since these comprise most of the major early born cell types of the retina (Figure 4A-4C’). Brn3a has been showed to be expressed by a large fraction of

RGCs48, while Opn4 labels a subset of melanopsin positive ganglion cells49. Tcfap2α is a 99 characterized marker of amacrine cells44. Thus, we used anti-Brn3a, anti-Opn4 and anti-

Tcfap2α antibodies to visualize populations of RGCs and amacrine cells (Figure 4D-4F’).

Counting the cell numbers in three separate retinas revealed no significant differences between the cell types in wildtype and Trim9 knockout retinas (Table 1).

Since Trim9 is expressed during development and may play a role in cell fate determination, we wished to examine developing wildtype and Trim9 knockout retinas at embryonic stages. Consequently, in situ hybridization was performed on cryosections of the eye, harvested at E14.5. Probes specific for developing photoreceptors (Otx250), developing ganglion cells (Synuclein- and Ebf351), and progenitor cells (Vsx231) were used. However, no major changes were observed in each of these developing neuronal populations (data not shown).

Since our observations did not support our hypothesis that Trim9 plays a role in cell fate determination in the retina, we decided to examine other aspects of retinal development such as RGC differentiation and morphology. As it was previously reported that Trim9 plays a role in axon branching and guidance in the cerebral cortex26, we hypothesized that the retinas of Trim9 deficient mice may show defects in axon morphology and branching. In order to visualize the axons of the retinal ganglion cells, we used an antibody to SMI-32, a heavy chain non-phosphorylated neurofilament protein component of axons52 (Figure 5A, 5A’). Again, we failed to observe any notable changes in the morphology or number of ganglion cell axons.

Our heatmap data showing that Trim9 is only found in a subset of Atoh7+ cells

(Supplemental Figure 1) raised the possibility that subsets of developing ganglion cells may be affected in the retinas of Trim9 deficient mice, rather than the whole population. This 100 may have escaped detection while examining the whole retinas by in situ hybridization or immunohistochemistry because of the small number of cells involved. Thus, we used quantitative real time PCR to characterize potential differences in subsets of ganglion cells, utilizing certain specific markers of ganglion cell subsets. These included Brn3a, Drd2,

Drd4, Tbr2, Scn4, Mmp17, Cdh6, Unc5d, Jam2, and Spig1. Brn3a, a gene encoding a POU- domain containing transcription factor is a marker of ganglion cells that project to the contralateral superior colliculus and the dorsal lateral geniculate nucleus53. Drd4 expression marks the ON-OFF direction sensitive ganglion cells (DSGCs) that prefer nasal motion54. Tbr2 is a selective marker of the intrinsically photosensitive ganglion cell population expressing Opn455. Mmp17 and Cdh6 are possible markers of the ON-OFF

DSGCs43. A gene encoding a netrin receptor, Unc5d is expressed by ON RGCs55. Jam2 labels a population of OFF RGCs, also called J-RGCs56. Spig1 is a known marker of ON DSGCs43. The expression of all the genes mentioned above remained unchanged in the Trim9 KO retina

(Figure 6), indicating that loss of Trim9 does not affect the identifiable subtypes of RGCs in the retina.

Additionally, we hypothesized that a full transcriptomic analysis of Trim9 deficient mutants compared to WT retinas could reveal changes on a transcriptional level that may lead to the discovery of unpredicted phenotypes in the Trim9 deficient retinas. To that end, we performed microarrays on WT and Trim9 knockout littermates (n=3). The 20 most downregulated genes in the knockout with over a two-fold change in expression levels in the KO retinas are listed in Table 2. The genes that were downregulated in the Trim9 KO appeared to be involved in a diverse number of processes. For example, VPS25, a subunit of a protein complex named ESCRT (endosomal sorting complex required for transport) is a 101 transmembrane proteins required for sorting endosomal cargo proteins in multivesicular bodies57. SUZ12 and BANF1 are involved in histone modification and chromatin organization58,59. SPOP and STUB1 (also known as CHIP) are also E3 ubiquitin ligases involved in proteosomal degradation of target proteins60,61. Notably, Psmd11, a gene encoding a 26S proteosomal subunit62,63 was also downregulated in Trim9 knockout retinas.

Expression of Trim genes in developing zebrafish

Given the advantages of zebrafish as a model organism, we decided to analyze the expression of trim genes in the developing zebrafish eye. Although the levels of trim genes other than trim9 were generally low in the zebrafish single cells, we chose 4 genes, trim25, trim55a, trim55b and trim71 from our zebrafish single cell transcriptomic data set29 that were expressed at relatively high levels, and performed whole-mount ISH to uncover the expression of these genes in the zebrafish nervous system (Supplemental Figure 2). trim25 did not appear to be highly expressed and the retina was almost devoid of staining

(Supplemental Figure 2A, 2E). trim55a and trim55b could be detected in both the brain and the developing retina (Supplemental Figure 2B, 2C). trim71 staining, although intense in the brain, appeared to be almost negligible in the retina (Supplemental Figure 2D). The expression of the genes mentioned points to the importance of the trim gene family in the zebrafish nervous system, while also suggesting that perhaps the expression of the trim family in zebrafish retina is not as prevalent as it is in mouse.

102

Generation and analysis of trim9 CRISPR mutant zebrafish

To further analyze the expression of trim9 in zebrafish we first performed ISH on

48hpf retinal sections. In addition, we have shown the expression of atoh7 as a positive control in several large patches of cells prominently in the central retina, consistent with previous studies (Figure 7A)29. We found trim9 expression to be localized to the central retinal region above the lens, consistent with the developing ganglion cell layer (Figure 7B).

Taking advantage of the genome editing tools available in zebrafish, we decided to continue our analysis of trim9 through the generation of zebrafish mutants. We used

CRISPR-Cas9 mediated genome editing to study the function of trim9 by a loss-of-function approach. We designed a guide RNA (a short stretch of nucleotides which serves to guide

Cas9 nuclease to the desired target location) to the first exon of the gene, with the rationale that mutations in this region would cause a frameshift in the gene leading to a loss of the functionality of the protein. We injected the gRNA and the nuclease, Cas9, into zebrafish embryos at the one-cell stage. Injection at this stage should ensure that the entire organism has the mutation; however, due to incomplete diffusion of the RNA during the early stages of cell division, the end product is usually a mosaic with a certain percentage of cells carrying mutations. Upon outcrossing the injected zebrafish, we found the transmission of an 8bp deletion in exon 1 to the next generation (Figure 7C). Crossing two heterozygous zebrafish, thus generated, led to homozygous mutants in Mendelian ratios. While selecting our guide sequence, we ensured that the chosen sequence was highly specific for its target location and outcrossing the heterozygous zebrafish for 3 generations greatly reduced the likelihood of off-target mutations. 103

Comparing WT and trim9 mutants yielded similar results as we observed in the mouse retina. We performed IHC on sections obtained from WT and mutant retinas at both embryonic (48hpf) and adult stages with antibodies such as Zn864 to stain RGCs (Figure 7D,

7D’, Supplemental figure 3A, 3A’) and Islet1 to illuminate the GCL and INL65 (Figure 7E, 7E’ and Supplemental figure 3B, 3B’). Overall, there was no change in gross morphology in the neuronal layers in the retina.

Discussion

To gain a better understanding of retinal progenitor cell development, we examined the transcriptomes of 46 individual cells isolated from the developing mouse retina10,30 and the expression of 35 genes belonging to the Trim family. We found high expression of Trim genes in developing retinal progenitors and neurons between E12.5 and E16.5, which was corroborated by ISH. Of the 34 Trim genes, only Trim9, Trim36 and Trim39 were found to be expressed in Nefl+ cells and almost completely absent from Tcfap2+ cells, and therefore potential candidates to regulate RGC development. Through the use of a clustering program, we found a correlation between the expression of Atoh7 and Trim9 in our data. This indicated to us that Trim9 may be involved in the development of neurons from an Atoh7 lineage.

TRIM9, a member of the TRIM family of ubiquitin ligases, has been previously shown to play an important role in the process of axon branching and guidance in the mouse cerebral cortex26–28. After deletion of Trim9, axons in the cerebral cortex and corpus callosum were shown to exhibit exaggerated axon branching27. Despite several studies 104 investigating its function in the brain, the role of Trim9 in the retina has not been characterized. To that end, we obtained/generated Trim9 deficient mice and zebrafish and performed a thorough morphological and gene expression analysis to find a role for Trim9 in retinal development. Surprisingly, neither the production of different early-born retinal cells nor the initial formation of RGC axons was affected in the Trim9 knockout animals. qPCR analysis for different subtypes of retinal ganglion cells showed that in addition to gross morphological similarities, smaller populations of cells labeled by particular markers were also unchanged in Trim9 deficient retinas. The lack of differences between WT and

Trim9 deficient retinas was consistent across both embryonic and adult stages.

We hypothesize that the lack of phenotype in Trim9 deficient retinas could be due to redundancy through the expression of other members of the Trim family. Studies have shown that the loss of a gene in vertebrate systems is often compensated for by the function of related genes or family members. For example, it was observed that the deletion of Sox12 in mice, a member of the SoxC family highly expressed during embryogenesis particularly within the peripheral and central nervous system, does not cause gross morphological changes or loss of fertility or viability66. Upon further investigation, it was observed that two other members belonging to the SOXC family, SOX4 and SOX11 work synergistically with SOX1266. The authors concluded that one reason for the lack of a phenotype could be functional compensation by SOX4 and SOX1266. An increase of expression of these two genes in some organs was observed, which supported this theory66. A lack of phenotype upon deletion of a gene belonging to a large family is not uncommon in mouse67. Examples include knockouts of Casp12 of the caspase gene family66,

Adcy4 of the seven adenylate cyclase gene family66, Capn5 of the six calpain gene family68, 105 and Pmm1 of the two phosphomannomutase gene family69. A study further postulates that perhaps redundancy in gene families makes the absence of a phenotype more likely than for an orphan gene67. Similarly, in zebrafish, it has been shown that receptors of bone morphogenetic proteins (BMP), SMAD1 and SMAD9 function redundantly to mediate dorso-ventral patterning70. Although knockdown of one or the other does not lead to visible dorsalization, double knockdown causes a strong phenotype70. The statistic that approximately 10-15% of KO mice are shown not to have a phenotype67 makes our result less surprising given that Trim9 belongs to a large family of approximately 40 genes37 all expressed at various levels in the retina. The wide expression of Trim family members in the developing retina (Figure 1, S1), especially the expression of multiple Trim genes in the same cells as Trim9 would appear to support the possibility of compensation. The loss of

Trim9 could trigger upregulation of other Trim genes, whereby these genes could compensate for the loss of Trim9 by gain of function. Transcriptomic or proteomic analysis of individual cells isolated from developing Trim9 mutant retinas, to detect increased expression of other Trim genes could be one possible approach to verify this theory.

Targeting several of the Trim genes expressed in the retina by using multiplexed CRISPR genome editing in mouse or more conveniently in zebrafish35, could be a more successful method to completely abolishing their effect in the retina, thereby leading to more fruitful insights about the functions of these genes.

Taken together our results indicate that single knockout of Trim9, a gene correlated with Atoh7 expression and indispensable for axon branching in the mouse cerebral cortex, does not cause significant defects in normal retinal development or morphology, while not ruling out the possibility that Trim9 works in concert with other genes, that may 106 compensate for the loss of its function. This reveals a situation where the function of a gene may vary within different regions of the CNS. Investigation into the remaining members of the Trim family may reveal roles for them in retinal development and function, possibly in concert with Trim9.

Acknowledgements

We would like to thank Dr. Stephanie Gupton for the Trim9 mouse line, Dr. Marit Nilsen-

Hamilton for the use of her BioRad MiniOpticon Real-Time PCR System, Dr. Lee

Benedickson for advice about qPCR and Dr. Essner for the generous gift of pT3TS-nCas9n plasmid (Addgene plasmid # 46757) and help regarding zebrafish and the CRISPR technique.

References

1. Seung HS, Sümbül U. Neuronal cell types and connectivity: Lessons from the retina. Neuron. 2014;83(6):1262–72. 2. Cepko CL, Austin CP, Yang X, Alexiades M, Ezzeddine D. Cell fate determination in the vertebrate retina. Proc Natl Acad Sci U S A. 1996;93(January):589–95. 3. Holt CE, Bertsch TW, Ellis HM, Harris WA. Cellular determination in the Xenopus retina is independent of lineage and birth date. Neuron. 1988 Mar;1(1):15–26. 4. Turner DL, Cepko CL. A common progenitor for neurons and glia persists in rat retina late in development. Nature. 1987;328(6126):131–6. 5. Turner DL, Snyder EY, Cepko CL. Lineage-independent determination of cell type in the embryonic mouse retina. Neuron. 1990;4(6):833–45. 6. Young RW. Cell proliferation during postnatal development of the retina in the mouse. Brain Res. 1985 Aug;353(2):229–39. 107

7. Young RW. Cell differentiation in the retina of the mouse. Anat Rec [Internet]. 1985;212(2):199–205. Available from: http://www.ncbi.nlm.nih.gov/pubmed/3842042 8. Sidman RL. Histogenesis of mouse retina studied with thymidine-H3. Struct Eye, Acad Press New York. 1961;487–506. 9. Rapaport DH, Wong LL, Wood ED, Yasumura D, Lavail MM. Timing and topography of cell genesis in the rat retina. J Comp Neurol. 2004;474(2):304–24. 10. Trimarchi JM, Stadler MB, Cepko CL. Individual retinal progenitor cells display extensive heterogeneity of gene expression. PLoS One. 2008;3(2). 11. Brown NL, Kanekar S, Vetter ML, Tucker PK, Gemza DL, Glaser T. Math5 encodes a murine basic helix-loop-helix transcription factor expressed during early stages of retinal neurogenesis. Development [Internet]. 1998;125(23):4821–33. Available from: http://www.ncbi.nlm.nih.gov/cgi- bin/Entrez/referer?http://www.biologists.com/Development/125/23/dev4057.html 12. Brown NL, Patel S, Brzezinski J, Glaser T. Math5 is required for retinal ganglion cell and optic nerve formation. Development. 2001;128:2497–508. 13. Feng L, Xie ZH, Ding Q, Xie X, Libby RT, Gan L. MATH5 controls the acquisition of multiple retinal cell fates. Mol Brain [Internet]. 2010;3(1):36. Available from: http://www.ncbi.nlm.nih.gov/pubmed/21087508 14. Le TT, Wroblewski E, Patel S, Riesenberg AN, Brown NL. Math5 is required for both early retinal neuron differentiation and cell cycle progression. Dev Biol. 2006;295(2):764– 78. 15. Wang SW, Kim BS, Ding K, Wang H, Sun D, Johnson RL, et al. Requirement for math5 in the development of retinal ganglion cells. Genes Dev. 2001;15(1):24–9. 16. Kay JN, Finger-Baier KC, Roeser T, Staub W, Baier H. Retinal ganglion cell genesis requires lakritz, a Zebrafish atonal Homolog. Neuron [Internet]. 2001;30. Available from: http://dx.doi.org/10.1016/S0896-6273(01)00312-9 17. Yang Z, Ding K, Pan L, Deng M, Gan L. Math5 determines the competence state of retinal ganglion cell progenitors. Dev Biol. 2003;264(1):240–54. 18. Brzezinski JA, Prasov L, Glaser T. Math5 defines the ganglion cell competence state in a subpopulation of retinal progenitor cells exiting the cell cycle. Dev Biol. 2012;365(2):395–413. 19. Prasov L, Glaser T. Pushing the envelope of retinal ganglion cell genesis : Context dependent function of Math5 ( Atoh7 ). Dev Biol [Internet]. 2012;368(2):214–30. Available from: http://dx.doi.org/10.1016/j.ydbio.2012.05.005 20. Song W, Zhang X, Xia X. Atoh7 promotes the differentiation of Müller cells-derived retinal stem cells into retinal ganglion cells in a rat model of glaucoma. Exp Biol Med 108

[Internet]. 2015 May 4;240(5):682–90. Available from: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4935277/ 21. Chen M, Chen Q, Sun X, Shen W, Liu B, Zhong X, et al. Generation of retinal ganglion- like cells from reprogrammed mouse fibroblasts. Invest Ophthalmol Vis Sci. 2010 Nov;51(11):5970–8. 22. Goetz JJ, Martin GM, Chowdhury R, Trimarchi JM. Onecut1 and Onecut2 play critical roles in the development of the mouse retina. PLoS One. 2014;9(10). 23. Goetz JJ, Laboissonniere LA, Wester AK, Lynch MR, Trimarchi JM. Polo-Like Kinase 3 Appears Dispensable for Normal Retinal Development Despite Robust Embryonic Expression. PLoS One [Internet]. 2016;11(3):e0150878. Available from: http://dx.plos.org/10.1371/journal.pone.0150878 24. Emerson M, Surzenko N, Goetz J, Trimarchi J, Cepko C. Otx2 and Onecut1 promote the fates of cone photoreceptors and horizontal cells and repress rod photoreceptors. Dev Cell [Internet]. 2013;26(1):59–72. Available from: http://dx.doi.org/10.1016/j.devcel.2013.06.005 25. Tanji K, Kamitani T, Mori F, Kakita A, Takahashi H, Wakabayashi K. TRIM9, a novel brain-specific E3 ubiquitin ligase, is repressed in the brain of Parkinson’s disease and dementia with Lewy bodies. Neurobiol Dis. 2010;38(2):210–8. 26. Winkle CC, McClain LM, Valtschanoff JG, Park CS, Maglione C, Gupton SL. A novel netrin-1-sensitive mechanism promotes local SNARE-mediated exocytosis during axon branching. J Cell Biol. 2014;205(2):217–32. 27. Menon S, Boyer NP, Winkle CC, McClain LM, Hanlin CC, Pandey D, et al. The E3 Ubiquitin Ligase TRIM9 Is a Filopodia Off Switch Required for Netrin-Dependent Axon Guidance. Dev Cell [Internet]. 2015;35(6):698–712. Available from: http://dx.doi.org/10.1016/j.devcel.2015.11.022 28. Winkle CC, Olsen RHJ, Kim H, Moy SS, Song J, Gupton SL. Trim9 Deletion Alters the Morphogenesis of Developing and Adult-Born Hippocampal Neurons and Impairs Spatial Learning and Memory. J Neurosci [Internet]. 2016;36(18):4940–58. Available from: http://www.jneurosci.org.gate1.inist.fr/content/36/18/4940.full 29. Mullally M, Albrecht C, Horton M, Laboissonniere LA, Goetz JJ, Chowdhury R, et al. Expression Profiling of Developing Zebrafish Retinal Cells. Zebrafish [Internet]. 2016;13(4):272–80. Available from: http://online.liebertpub.com/doi/full/10.1089/zeb.2015.1184 30. Trimarchi JM, Stadler M, Roska B, Billings N, Sun B, Bartch B, et al. Molecular Heterogeneity of Developing Retinal Ganglion and Amacrine Cells Revealed through Single Cell Gene Expression Profiling. J Comp Neurol. 2007;504(3):287–97. 109

31. Morrow EM, Chen C-MA, Cepko CL. Temporal order of bipolar cell genesis in the neural retina. Neural Dev. 2008;3(January):2. 32. Molday RS. Monoclonal antibodies to rhodopsin and other proteins of rod outer segments. Prog Retin Res [Internet]. 1988;8:173–209. Available from: http://www.sciencedirect.com/science/article/pii/0278432788900259 33. Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, et al. Bioconductor: open software development for computational biology and bioinformatics. Genome Biol. 2004;5(10):R80. 34. Irizarry RA, Bolstad BM, Collin F, Cope LM, Hobbs B, Speed TP. Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Res [Internet]. 2003 Feb 15;31(4):e15–e15. Available from: http://dx.doi.org/10.1093/nar/gng015 35. Jao L-E, Wente SR, Chen W. Efficient multiplex biallelic zebrafish genome editing using a CRISPR nuclease system. Proc Natl Acad Sci U S A [Internet]. 2013;110(34):13904– 9. Available from: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3752207&tool=pmcentrez&r endertype=abstract 36. Hwang WY, Fu Y, Reyon D, Maeder ML, Tsai SQ, Sander JD, et al. Efficient genome editing in zebrafish using a CRISPR-Cas system. Nat Biotechnol [Internet]. 2013;31(3):227– 9. Available from: http://www.ncbi.nlm.nih.gov/pubmed/23360964%5Cnhttp://www.nature.com/nbt/jour nal/v31/n3/pdf/nbt.2501.pdf 37. Meroni G, Diez-Roux G. TRIM/RBCC, a novel class of “single protein RING finger” E3 ubiquitin ligases. BioEssays. 2005;27(11):1147–57. 38. Balastik M, Ferraguti F, Pires-da Silva A, Lee TH, Alvarez-Bolado G, Lu KP, et al. Deficiency in ubiquitin ligase TRIM2 causes accumulation of neurofilament light chain and neurodegeneration. Proc Natl Acad Sci U S A. 2008;105(33):12016–21. 39. Sicinski P, Donaher JL, Parker SB, Li T, Fazeli A, Gardner H, et al. Cyclin D1 provides a link between development and oncogenesis in the retina and breast. Cell. 1995 Aug;82(4):621–30. 40. Mu X, Fu X, Sun H, Beremand PD, Thomas TL, Klein WH. A gene network downstream of transcription factor Math5 regulates retinal progenitor cell competence and ganglion cell fate. Dev Biol. 2005;280(2):467–81. 41. Blackshaw S, Harpavat S, Trimarchi J, Cai L, Huang H, Kuo WP, et al. Genomic Analysis of Mouse Retinal Development. PLOS Biol [Internet]. 2004 Jun 29;2(9):e247. Available from: https://doi.org/10.1371/journal.pbio.0020247 110

42. Surgucheva I, Weisman AD, Goldberg JL, Shnyra A, Surguchov A. γ-Synuclein as a marker of retinal ganglion cells. Mol Vis [Internet]. 2008 Aug 22;14:1540–8. Available from: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2518532/ 43. Kay JN, De la Huerta I, Kim I-J, Zhang Y, Yamagata M, Chu MW, et al. Retinal Ganglion Cells with Distinct Directional Preferences Differ in Molecular Identity, Structure, and Central Projections. J Neurosci [Internet]. 2011;31(21):7753–62. Available from: http://www.jneurosci.org/cgi/doi/10.1523/JNEUROSCI.0907-11.2011 44. Bassett EA, Pontoriero GF, Feng W, Marquardt T, Fini ME, Williams T, et al. Conditional Deletion of Activating Protein 2 (AP-2 ) in the Developing Retina Demonstrates Non-Cell-Autonomous Roles for AP-2 in Optic Cup Development. Mol Cell Biol [Internet]. 2007;27(21):7497–510. Available from: http://mcb.asm.org/cgi/doi/10.1128/MCB.00687-07 45. Haverkamp S, Wassle H. Immunocytochemical analysis of the mouse retina. J Comp Neurol [Internet]. 2000;424(1):1–23. Available from: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citatio n&list_uids=10888735 46. Mears AJ, Kondo M, Swain PK, Takada Y, Bush RA, Saunders TL, et al. Nrl is required for rod photoreceptor development. Nat Genet. 2001 Dec;29(4):447–52. 47. Kondo H, Kuramoto H, Wainer BH, Yanaihara N. Discrete distribution of cholinergic and vasoactive intestinal polypeptidergic amacrine cells in the rat retina. Neurosci Lett. 1985 Mar;54(2–3):213–8. 48. Xiang M, Zhou L, Macke JP, Yoshioka T, Hendry SHC, Eddy RL, et al. The Brn-3 family of POU-domain factors: primary structure, binding specificity, and expression in subsets of retinal ganglion cells and somatosensory neurons. J Neurosci [Internet]. 1995;15(7):4762– 85. Available from: http://eutils.ncbi.nlm.nih.gov/entrez/eutils/elink.fcgi?dbfrom=pubmed&id=7623109&ret mode=ref&cmd=prlinks 49. Hughes S, Jagannath a, Rodgers J, Hankins MW, Peirson SN, Foster RG. Signalling by melanopsin (OPN4) expressing photosensitive retinal ganglion cells. Eye [Internet]. 2016;44(November 2015):1–8. Available from: http://www.nature.com/doifinder/10.1038/eye.2015.264 50. Nishida A, Furukawa A, Koike C, Tano Y, Aizawa S, Matsuo I, et al. Otx2 homeobox gene controls retinal photoreceptor cell fate and pineal gland development. Nat Neurosci [Internet]. 2003;6(12):1255–63. Available from: http://www.ncbi.nlm.nih.gov/pubmed/14625556 51. Garcia-Dominguez M, Poquet C, Garel S, Charnay P. Ebf gene function is required for coupling neuronal differentiation and cell cycle exit. Development. 2003 Dec;130(24):6013–25. 111

52. Nixon RA, Lewis SE, Dahl D, Marotta CA, Drager UC. Early posttranslational modifications of the three neurofilament subunits in mouse retinal ganglion cells: neuronal sites and time course in relation to subunit polymerization and axonal transport. Mol Brain Res [Internet]. 1989;5(2):93–108. Available from: http://www.sciencedirect.com/science/article/pii/0169328X89900016 53. Nadal-Nicolas FM, Jimenez-Lopez M, Sobrado-Calvo P, Nieto-Lopez L, Canovas- Martinez I, Salinas-Navarro M, et al. Brn3a as a marker of retinal ganglion cells: Qualitative and quantitative time course studies in naive and optic nerve-injured retinas. Investig Ophthalmol Vis Sci. 2009;50(8):3860–8. 54. Sanes JR, Masland RH. The Types of Retinal Ganglion Cells: Current Status and Implications for Neuronal Classification. Annu Rev Neurosci [Internet]. 2014;38(1):150421150146009. Available from: http://www.annualreviews.org/doi/abs/10.1146/annurev-neuro-071714-034120 55. Sweeney NT, Tierney H, Feldheim D a. Tbr2 is required to generate a neural circuit mediating the pupillary light reflex. J Neurosci. 2014;34(16):5447–53. 56. Kim I-J, Zhang Y, Yamagata M, Meister M, Sanes JR. Molecular identification of a retinal cell type that responds to upward motion. Nature [Internet]. 2008;452(7186):478– 82. Available from: http://www.nature.com/doifinder/10.1038/nature06739 57. Babst M, Katzmann DJ, Snyder WB, Wendland B, Emr SD. Endosome-Associated Complex, ESCRT-II, Recruits Transport Machinery for Protein Sorting at the Multivesicular Body. Dev Cell [Internet]. 2002 Aug;3(2):283–9. Available from: http://www.sciencedirect.com/science/article/pii/S1534580702002198 58. Cao R, Zhang Y. SUZ12 Is Required for Both the Histone Methyltransferase Activity and the Silencing Function of the EED-EZH2 Complex. Mol Cell [Internet]. 2004;15(1):57– 67. Available from: http://www.sciencedirect.com/science/article/pii/S1097276504003478 59. Kirmizis A, Bartley SM, Kuzmichev A, Margueron R, Reinberg D, Green R, et al. Silencing of human polycomb target genes is associated with methylation of histone H3 Lys 27. Genes Dev [Internet]. 2004 Jul 1;18(13):1592–605. Available from: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC443521/ 60. Li C, Ao J, Fu J, Lee D-F, Xu J, Lonard D, et al. Tumor-suppressor role for the SPOP ubiquitin ligase in signal-dependent proteolysis of the oncogenic co-activator SRC-3/AIB1. Oncogene [Internet]. 2011 Oct 20;30(42):4350–64. Available from: http://dx.doi.org/10.1038/onc.2011.151 61. Jiang J, Ballinger CA, Wu Y, Dai Q, Cyr DM, Hohfeld J, et al. CHIP is a U-box-dependent E3 ubiquitin ligase: identification of Hsc70 as a target for ubiquitylation. J Biol Chem. 2001 Nov;276(46):42938–44. 112

62. Vilchez D, Boyer L, Morantte I, Lutz M, Merkwirth C, Joyce D, et al. Increased proteasome activity in human embryonic stem cells is regulated by PSMD11. Nature. 2012 Sep;489(7415):304–8. 63. Lokireddy S, Kukushkin NV, Goldberg AL. cAMP-induced phosphorylation of 26S proteasomes on Rpn6/PSMD11 enhances their activity and the degradation of misfolded proteins. Proc Natl Acad Sci U S A. 2015 Dec;112(52):E7176-85. 64. Arenzana FJ, Clemente D, Sanchez-Gonzalez R, Porteros A, Aijon J, Arevalo R. Development of the cholinergic system in the brain and retina of the zebrafish. Brain Res Bull. 2005 Sep;66(4–6):421–5. 65. Vitorino M, Jusuf PR, Maurus D, Kimura Y, Higashijima S, Harris WA. Vsx2 in the zebrafish retina: restricted lineages through derepression. Neural Dev [Internet]. 2009;4(1):14. Available from: http://dx.doi.org/10.1186/1749-8104-4-14 66. Hoser M, Potzner MR, Koch JMC, Bo MR, Wegner M, Sock E. Sox12 Deletion in the Mouse Reveals Nonreciprocal Redundancy with the Related Sox4 and Sox11 Transcription Factors ᰔ. 2008;28(15):4675–87. 67. Barbaric I, Miller G, Dear N. Appearances can be deceiving : phenotypes of knockout mice. 2007;6(2). 68. Franz T, Winckler L, Boehm T, Dear TN. Capn5 Is Expressed in a Subset of T Cells and Is Dispensable for Development. 2004;24(4):1649–54. 69. Cromphout K, Vleugels W, Heykants L, Schollen E, Keldermans L, Sciot R, et al. The Normal Phenotype of Pmm1-Deficient Mice Suggests that Pmm1 Is Not Essential for Normal Mouse Development. 2006;26(15):5621–35. 70. Wei C, Wang H, Zhu Z, Sun Y. Transcriptional Factors Smad1 and Smad9 Act Redundantly to Mediate Zebrafish Ventral Specification Downstream. 2014;289(10):6604– 18.

113

Figures and Legends

Table 1. Quantification of neuronal populations using flatmount immunohistochemistry. Immunohistochemistry was performed on flatmounted adult retinas from WT and Trim9 knockout littermates using antibodies to CALBINDIN-28K and

CALRETININ, BRN3A, OPN4 and AP2Α. Retinal neurons were counted manually from three fields of view and numbers of each cell type are shown in Table 1. Scale bars represent 50

µm.

Antibody Average # of cells in WT Average # of cells in KO Anti-Calretinin 196.66 148 Anti-Calbindin 193.33 265.33 Anti-Brn3a 338.25 344.25 Anti-Tcfap2α 367.25 356.25 Anti-Opn4 9.75 9.66

Table 2. List of differentially expressed genes between WT and Trim9 KO retinas. A list of genes obtained by microarray analysis of WT and Trim9 mutant retinas (n=3) is shown in this table.

Probe set ID Gene symbol Gene title

receptor (calcitonin) activity modifying protein 2 /

1421051_s_at Ramp2/ Vps25 vacuolar protein sorting 25 (yeast)

114

Table 2 continued metastasis associated lung adenocarcinoma transcript 1

1418189_s_at Malat1 (non-coding RNA)

1416201_at Crk v-crk sarcoma virus CT10 oncogene homolog (avian)

1416525_at Spop speckle-type POZ protein

1419883_s_at Atp6v1b2 ATPase, H+ transporting, lysosomal V1 subunit B2

1415864_at Bpgm 2,3-bisphosphoglycerate mutase

1452364_at Suz12 suppressor of zeste 12 homolog (Drosophila)

1434540_a_at Clta clathrin, light polypeptide (Lca)

proteasome (prosome, macropain) 26S subunit, non-

1429370_a_at Psmd11 ATPase, 11

1434229_a_at Polb polymerase (DNA directed), beta

1416580_a_at Stub1 STIP1 homology and U-Box containing protein 1

1449292_at Rb1cc1 RB1-inducible coiled-coil 1

1421081_a_at Banf1 barrier to autointegration factor 1

1426621_a_at Ppp2r2b protein phosphatase 2, regulatory subunit B, beta

1432344_a_at Aplp2 amyloid beta (A4) precursor-like protein 2

1450761_s_at Rims2 regulating synaptic membrane exocytosis 2

1415817_s_at Cct7 chaperonin containing Tcp1, subunit 7 (eta)

1427640_a_at Runx1t1 runt-related transcription factor 1; translocated to, 1

(cyclin D-related)

1416180_a_at Rdx radixin

115

Figure 1: The Trim family of genes is expressed in the developing mouse retina.

Heatmaps representing microarray data for the expression of genes belonging to the Trim family in (A) single cycling retinal progenitor cells (Ccnd1+), (B) developing retinal ganglion cells (Nefl+) and amacrine cells (Tcfap2α+) (C-I’’) In situ hybridization showing the expression of Trim62, Trim2, Trim32, Trim44, Trim46, Trim28 and Trim9 genes in retinal sections at E12.5, E14.5, E16.5. Scale bars represent 100 µm.

116

Figure 2: Morphological characterization of adult WT and Trim9 deficient mouse retinas using in situ hybridization. In situ hybridization was used to identify populations of retinal neurons in WT and Trim9 knockout littermates marked by expression of

Synuclein- (A, A’), Cartpt (B, B’), Tcfap2α (C, C’), Gad1 (D, D’), Slc6a9 (E, E’), Septin4 (F, F’),

Opsin (G, G’), Nrl (H, H’), Vsx2 (I, I’), and Clusterin (J, J’). Scale bars represent 100 µm.

117

Figure 3: Morphological characterization of adult WT and Trim9 deficient mouse retinas using section immunohistochemistry. Immunohistochemistry was performed on adult retinal sections from WT and Trim9 knockout littermates to quantify the number of retinal neurons using antibodies to CALBINDIN-28K and CALRETININ (A, A’), VSX2 (B,

B’), PKCα (C, C’), TCFAP2α (D, D’), ChAT (E, E’) and RHODOPSIN (F, F’). DAPI (blue) shows nuclear staining. Scale bars represent 50 µm.

118

Figure 4: Morphological characterization of adult WT and Trim9 deficient mouse retinas using flatmount immunohistochemistry.

Immunohistochemistry was performed on flatmounted adult retinas from WT and Trim9 knockout littermates to quantify the number of retinal neurons using antibodies to

CALBINDIN-28K and CALRETININ (A, A’- C, C’), BRN3A (D, D’), OPN4 (E, E’) and TCFAP2α

(F, F’). DAPI (blue) shows nuclear staining. Scale bars represent 50 µm.

119

Figure 5: Visualization of RGC axons using IHC on adult WT and Trim9 deficient flatmounted mouse retinas. Retinas obtained from WT and Trim9 knockout mice were flatmounted and IHC was performed using an antibody to SMI-32, an axonal marker (A, A’,

B, B’). Scale bars represent 50 µm.

Figure 6: Quantification of RGC subsets in WT and Trim9 deficient using quantitative real time PCR (qPCR). RNA was isolated from retinas of adult WT and Trim9 knockout mice and qPCR was performed using primers specific for RGC subsets marked by 120 expression of the genes Brn3a, Drd2, Drd4, Tbr2, Scn4, Mmp17, Unc5d, Cdh6, Jam2 and

Spig1. Error bars indicating standard deviation were calculated in Microsoft Excel.

Figure 7: Expression and morphological characterization of trim9 in developing zebrafish retinas. In situ hybridization was used to identify populations of retinal neurons in zebrafish retinal cryosections at 48hpf using probes to atoh7 and trim9 (A, B). Scale bars represent 100 µm. Schematic showing the position of trim9 CRISPR target site with respect to the gene (C). Section immunohistochemistry was used to characterize 48hpf WT and 121 trim9 mutant zebrafish retinas using antibodies to Zn8 (D, D’) and Islet1 (E, E’). DAPI shows nuclear staining. Scale bars represent 50 µm.

Supplemental Table 1. List of probes used for ISH. The sequences of the primers used to amplify specific genes from mouse and zebrafish retinal cDNA for generating RNA probes used for section ISH are shown in this table.

Mouse gene Left Primer Right Primer Probe used Opn1sw tggtgggatccttctgtctc aactgggagaggccacataa Tcfap2-alpha acccctacagcctgaatcct ttcttgccacttgctcattg Slc6a9 tgtctgaaaggcactgaacg tcagcacatacagcctccag Gad1 ccatctcgcaagcaactaca aatgcacagtgtgggtttca Cartpt tgtggttgacctggagactg gatttgagctgtgcctttcc Jam2 cgccctggactatcataagg ccacaaccgttgctatgatg Clusterin agcttccacaacatgttcca tcctgcggtattcctgtagc Septin4 cgggtcaacattgtgcctat cagcagggatataaggcgaac Syn-gamma cagtccatagcttgcagcag cacagcagcatctgattggt Nrl Cepko lab BMAP collection BE982468 Chx10 Cepko lab BMAP collection BF461223 Trim9 Cepko lab BMAP collection AW124156 Trim62 Cepko lab BMAP collection AI838479 Trim2 cggtgaacttcccaaggtta cacatgcctttggttgacac Trim32 gaaacagcggagttctgagg atgcccactggacaggtaag Trim44 aggcagctcatctgtgtcct gctcttcccatgacctacca Trim46 tgaggtgagtggtcagcaag caggaaagcaaagggtggta Trim28 gcctctgactgaaggtcctg tggttctaccagcacagcag Zebrafish gene Left primer Right primer Trim9 gaatttgaggcgtgtttggt aagccagtgtttcccatgac Trim25 tggagaatccaaatccaagc agcaaacaccctgtcactcc Trim55a ttcgctatgcaaagtgttcg gaccggatgttacctgtgct Trim55b actggaggagacctgcagaa aaataccacgcgaacactcc Trim71 aaaccagcatgcgtctctct cgttctgttgctttggttga

122

Supplemental Table 2. Oligos and primers used for Trim9 gRNA synthesis and genotyping. The forward and reverse oligos used to generate the gRNA for targeting trim9 in zebrafish and the primers used for genotyping are shown in this table.

Forward Oligo TAGGCCTGCGCGAGAAACATCC

Reverse Oligo AAACGGATGTTTCTCGCGCAGG Forward primer CTGCCCTGCTCGCATAATA Reverse primer CAGGTCTAAGTAATCATAATCGGAGA

Supplemental Figure 1: The Trim family of genes is expressed in Atoh7+ single RPCs.

Heatmaps representing microarray data for the expression of genes belonging to the Trim family in single Atoh7+ RPCs isolated from the developing mouse retina at E12.5, E14.5 and

E16.5

123

Supplemental Figure 2: Expression of Trim genes in developing zebrafish retinas.

Whole-mount ISH was used to observe the expression of trim25 (A, E), trim55a (B, F), trim55b (C, G) and trim71 (D, H) in 48hpf zebrafish. Scale bars represent 100 µm.

124

Supplemental Figure 3: Morphological characterization of adult WT and trim9 mutant zebrafish retinas. Section immunohistochemistry was used to characterize adult trim9 deficient zebrafish retinas using antibodies to Zn8 (A, A’) and Islet1 (B, B’). DAPI

(blue) shows nuclear staining. Scale bars represent 50 µm.

125

CHAPTER 4: USING GENOME EDITING TO INVESTIGATE VERTEBRATE RETINAL DEVELOPMENT IN ZEBRAFISH

Abstract

Understanding the process by which distinct neurons are produced is an important goal of both developmental biologists and stem cell researchers. To reach this goal, it is critical to understand how specific intrinsic factors control cell fate decisions and neuronal maturation processes. In the retina, Atoh7 (also known as ath5 in zebrafish or Math5 in mouse) is a highly conserved transcription factor that is essential for retinal ganglion cell development in the developing mouse and zebrafish retina. Using single cell transcriptome profiling, we characterized individual mouse Atoh7+ cells and zebrafish atoh7+ cells and identified several genes whose expression tracks closely with Atoh7. We further assayed the expression of these genes in the zebrafish retina by section and whole-mount in situ hybridization. Given their expression patterns, we hypothesized that these candidate genes

(Plk3, Neurod6a and Ikzf5) will play important roles in either retinal cell fate determination or the differentiation of early-born retinal neurons. To gain insight into the specific functions of these genes, we have successfully generated mutations in the zebrafish homologs of these genes using the CRISPR-Cas9 genome editing system. Through a combination of in situ hybridizations, immunohistochemistry and qPCR, we are examining retinal development in these mutant fish, with a specific focus on the early-generated cell types. We have observed that CRISPR-mediated mutations in plk3 lead to changes in the expression of Islet1 and perhaps to alterations in the number of a subset of amacrine cells 126 in the zebrafish retina. Analysis of neurod6a mutants did not show a significant change in retinal cell fate. We are continuing our analysis of ikzf5 and uchl1 mutant zebrafish to uncover roles for these genes in retinal development. Taken together, our single cell transcriptomic approach coupled with genome editing is a powerful method for dissecting the precise networks of genes controlling early development of the retina.

Introduction

Teleost retinae have been studied for over a century1,2. In particular, the zebrafish,

Danio rerio, has emerged as one of the leading model organisms for the examination of the vertebrate visual system3. Features of the zebrafish such as high fecundity, transparency of embryos, ease of maintenance in large numbers, short life cycle, and external embryogenesis, make them a useful tool to study development. Furthermore, a completely sequenced genome makes the zebrafish an attractive model system for more functional applications such as genetic screens and genome editing4. Zebrafish eyes are large, neuroanatomically well defined3 and, importantly, become functional early on during development. This latter characteristic of zebrafish makes it highly suitable for genetic studies of eye development. Neurogenesis in the zebrafish retina starts at about 27 hours post fertilization (hpf) and is complete by 60 hpf5. Zebrafish have been shown to detect light starting at about 2.5 days post fertilization (dpf)6. Taken together, these qualities of the zebrafish make it an excellent organism for the study of retinal neurogenesis and gene function. 127

The cellular architecture of the vertebrate retina has been remarkably well- conserved during evolution across vertebrate species to the extent that even divergent phyla such as mammalian and teleost species share gross morphological similarities1,3.

Human, mouse and zebrafish retinas have the same major classes of retinal neurons, with the photoreceptors (PRs) located towards the apical surface of the retina, the retinal ganglion cells (RGCs) situated basally, and the interneurons, horizontal, bipolar and amacrine cells, located between the photoreceptor and ganglion cell layer3. The ganglion cells of the retina are the first cell type to become post-mitotic at about 27hpf in the zebrafish7. This early onset of RGC differentiation is conserved across retinas of many vertebrate species3. A rudimentary RGC layer is distinguishable by 36hpf7. At this time, approximately 10 hours after the onset of RGC differentiation, the cells of the INL become visible7. By 34hpf or earlier, terminal divisions of retinal progenitor cells (RPCs) give rise to

RGC and PR pairs8. At around 60hpf, most of the retinal neurons are post-mitotic and the different layers of the retina are fully distinguishable7.

The similarities between mouse and zebrafish retinas extend beyond morphology.

Several genes known to be important for mammalian retinal development have been shown to possess conserved functions in zebrafish. For example, PAX-2/Noi contributes to establishing the normal pattern of axon outgrowth from retina to the brain in both mouse and zebrafish9,10. Additionally, Chx10/vsx2 expression patterns in early retinal progenitor cells (RPCs) are very similar in zebrafish and mouse retinas9,11. Like mouse Chx10, zebrafish vsx2 plays a role in cell proliferation and bipolar cell fate11,12. atoh7 is an important genetic regulator in the zebrafish retina3. It starts being expressed in a small group of cells at 25hpf, and then spreads to the dorsal, nasal and temporal retina3. In the 128 mouse, transcription of Atoh7 begins at embryonic day (E)11, which coincides with the onset of RGC generation, in a similar fashion as zebrafish13,14. In zebrafish, mouse and many other species, the loss of Atoh7 leads to a complete or almost complete loss of RGCs15–18. In addition to loss of RGCs, Atoh7 deficiency results in increased numbers of bipolar cells, amacrines and Müller glia16,18, indicating a possible switch in cell fates from the early RGCs to later-born INL cells. Furthermore, the loss of atoh7 in zebrafish is associated with a delay in cell cycle exit18,19. While cells in wildtype retinas become post-mitotic at about 29hpf, this process is shifted to 43-48hpf in atoh7 deficient retinas19. One explanation for this phenomenon is that in the absence of atoh7, these retinas are accumulating RPCs while a wildtype retina is generating RGCs18.

Lineage tracing studies performed in mouse suggest that Atoh7+ RPCs give rise to all the early born cell types of the retina20. More recent work has shown that Atoh7+ RPCs contribute to every major cell type in the retina but are heavily skewed towards the early cell types21. This suggests that this transcription factor confers competence to RPCs in addition to promoting RGC differentiation20–22 However, the precise role of Atoh7 is still ambiguous. While ectopic expression of ATOH7 in mouse RPCs and retina explants did not lead to an increased production of RGCs23, other studies probing the misexpression of

ATOH7 in Müller glia or stem cells noted an increase in RGC genesis24,25. Even though the precise role of Atoh7 is unclear, it is doubtless an important player in early retinal development with contributions to cell fate decisions as well as differentiation of RGCs.

Given the current imprecise understanding of this important transcription factor, we wished to examine the transcriptome of cells expressing Atoh7 with the hypothesis that factors expressed simultaneously with Atoh7 may also play roles in early cell 129 determination. We postulated that extending our knowledge of the transcriptome of

Atoh7/atoh7+ cells and the identification of genes, especially those that are conserved in mouse and zebrafish, would help us achieve a better understanding of retinal cell fate determination. To that end, we collected 19 individual Atoh7+ cells from the developing mouse retina (Chowdhury et al., in preparation) and 8 atoh7+ cells from the developing zebrafish retina (50hpf)26 . Our microarray data analysis revealed a large number of genes whose expression was correlated with Atoh7 and found several genes that were correlated with Atoh7 in our mouse single cell data were also present in our zebrafish single cells. We prioritized genes for further study by both how closely associated they were with atoh7 and by their predicted functions (i.e. transcription factors). The genes chosen for further study included plk3, neurod6a, ikzf1, ikzf5 and uchl1. First, we investigated their expression in the zebrafish retina by in situ hybridization on cryosections and whole mount zebrafish.

These expression experiments demonstrated that plk3, neurod6a, and ikzf5 were expressed in the developing retina, while ikzf1 and uchl1 were expressed elsewhere.

In order to screen our candidate genes for a functional role in retinal cell fate determination in zebrafish, we used a Clustered Regularly Interspaced Palindromic Repeat

(CRISPR) - mediated genome editing approach27. We reasoned that the CRISPR-Cas9 system would allow us to create targeted mutations in the genes that we selected from our mouse and zebrafish single cell data, and, therefore, to study their role in the developing retina using a loss-of-function approach. We used the CRISPR-Cas9 system to target our genes of interest by introducing insertions/deletions (in/dels)28 that lead to frameshift mutations or stop codons. Once these mutations were characterized, we analyzed the potential loss-of-function retinal phenotypes in our mutant lines. In addition to generating 130 mutations, we have also used CRISPR-Cas9 to visualize gene expression and create a reporter line to facilitate our study of development29. We performed a pilot experiment to knock in red fluorescent protein (RFP) through the utilization of the homology directed repair (HDR) pathway29 at the ikzf1 locus to visualize the expression of the gene in vivo.

We have successfully generated mutant lines for the genes, plk3, neurod6a, ikzf5 and uchl1. We have completed the analysis of plk3 and neurod6a homozygous mutants, finding defects in subsets of cells in the inner nuclear layer (INL) and ganglion cell layer (GCL) in plk3 mutants. So far, we have seen no major changes in neurod6a mutants, and we will continue to analyze our remaining mutant zebrafish for defects in retinal development.

Materials and Methods

Zebrafish care

Zebrafish were grown on a 14 hr light/10 hr dark cycle in tanks on an Aquatic Habitat system (Aquatic Ecosystems) at 28°C. All experimental protocols were approved by the

Iowa State University Institutional Animal Care and Use Committee (IACUC).

Tissue processing for cryosections

Zebrafish were euthanized using five to six drops of 4mg/mL MS-222 Tricaine

Methanesulfonate per mL of fish water (1.25 g sea salt/20 L dH2O). Immediately upon 131 euthanasia, fish were placed in 4% PFA/PBS and tissue processing for zebrafish was carried out exactly as described previously26.

In situ hybridization

Riboprobe synthesis was performed as described previously30. Briefly, RNA probes

(650-800 bp in length) were synthesized by PCR amplification using primers specific for zebrafish cDNA, listed in Supplemental Table 1. The probe template sequences were cloned into pGEM-T vector and sequenced. Antisense riboprobes were synthesized using the T7 or

SP6 RNA polymerase in the presence of Digoxigenin for 1-2 hours at 37°C. Riboprobes were treated with DNase I for 15 minutes and precipitated with 100% ethanol and LiCl overnight.

In situ hybridization on retinal cryosections was performed as described previously31. Briefly, the slides were washed with PBS three times, fixed with 4% paraformaldehyde in PBS, acetylated and the riboprobes hybridized overnight (O/N) at

65°C. The next day, slides were incubated in a 1X SSC (diluted from 20X saline sodium citrate- 3M NaCl, 0.33M Sodium Citrate, pH 7) buffer containing 50% formamide, treated with RNase A and washed with 2X and 0.2X SSC. After washing twice with TNT (0.1 M Tris-

HCl, pH 7.5, 0.15 M NaCl, 0.05% Tween-20), slides were blocked for an hour with 20% heat inactivated sheep serum (HISS) and incubated with anti-DIG-alkaline phosphatase (anti-

DIG-AP) antibody (1:2500, Roche) overnight. The next day slides were washed with TNT and developed using NBT and BCIP. Finally, the slides were fixed in 4% 132 paraformaldehyde/PBS and mounted with Fluoromount-G (Southern Biotech). The list of primers used for probe synthesis can be found in Supplemental Table 1.

Whole-mount in situ hybridization

Zebrafish WM-ISH was performed exactly as described previously32. Briefly, 28hpf zebrafish were subjected to treatment with 50% PTU/50% fish water to ensure the lack of pigment formation. Fish were collected at 48hpf, and dehydrated through a series of increasing concentrations of ethanol in PBT (PBS/0.1%Tween-20) washes. After an O/N incubation at -20°C, the fish were rehydrated through several washes of gradually decreasing ethanol concentrations. The remaining protocol was followed exactly as described previously26. The primers used for making the probes are listed in Supplemental

Table 1.

Section Immunohistochemistry (IHC)

Slides containing cryosections were blocked for 30 minutes with blocking solution

(1% BSA, 0.01% Triton X-100, 0.004% SDS) and then placed in primary antibody, diluted according to manufacturer’s instructions in blocking solution. Slides were washed three times for 15 minutes. at room temperature. Slides were then placed in secondary antibody, diluted 1:300 with blocking solution and incubated at 4ᵒC O/N. Following this incubation, slides were again washed three times for 15 minutes with blocking solution and mounted with Fluoromount-G (Southern Biotech).

For adult zebrafish sections, an initial antigen retrieval step was performed. Slides containing cryosections were placed in sodium citrate buffer (10 mM Tri-sodium citrate, 133

0.05% Tween 20, pH 6.0). The bucket containing slides placed in sodium citrate buffer was microwaved for 5 minutes at ‘popcorn’ setting, and then for 15 minutes at the lowest power level. This allowed the slides to be brought just to a boil and then incubated at that point for the 15 minutes. The slides were then cooled by placing them in water, and immunohistochemistry was performed as described above.

Primary antibodies used were anti-Zn8 (1:40, Developmental Studies Hybridoma

Bank, DHSB), anti-Islet1 (1:20, Developmental Studies Hybridoma Bank).

CRISPR synthesis

Potential CRISPR target sites were identified using the CRISPR design program at http://zifit.partners.org/ZiFiT/ChoiceMenu.aspx. For Cas9 synthesis, the pT3TS-nCas9n plasmid was used (Addgene plasmid # 46757)28. Guide RNA (gRNA) was synthesized as described previously27 and purified using the miRNeasy mini kit (Qiagen). Cas9 was transcribed using the Xba1 digested pT3TS-nCas9n vector and T7 mMESSAGE mMACHINE

T7 ULTRA kit (Life Technologies). It was purified using the RNeasy MinElute Cleanup kit

(Qiagen).

Ikzf1 CRISPR knock-in synthesis

The process was adapted from a previously published study29. A reverse CRISPR sequence was selected for Ikzf1 using the CRISPR design program at http://www.crisprscan.org/. The sequence 48 bp upstream of the target site was added at the 5’ end. 1 bp was added as spacer between the target site and the upstream homology 134 sequence. One extra base was added to ensure that the oligos were knocked-in in frame.

The oligonucleotide sequences were:

Forward oligo: AATTCGGCGGCAAGCATAGTTGCACAGGaaaACACATCAAACTTCACTCTGGCGAGAAACCTTTC AAATGTCACCTGTGcG Reverse oligo: GATCCgCACAGGTGACATTTGAAAGGTTTCTCGCCAGAGTGAAGTTTGATGTGTtttCCTGTGCAA CTATGCTTGCCGCCG

Microinjection of zebrafish

1nl of solution containing 25pg of gRNA and 300pg of Cas9 mRNA were injected into one-cell stage zebrafish embryos. On the next day, deformed and dead embryos were removed. At 48hpf, 5 embryos from the injected and uninjected population were collected and genomic DNA was extracted as described below.

Genomic DNA isolation

To extract genomic DNA from zebrafish embryos, 100 µl of 0.05M NaOH was added to each tube containing 5 zebrafish. The tubes were incubated at 95°C for 30 minutes, cooled and 10 µl 1M Tris, pH 8 was added. Targeted genomic DNA loci were amplified using specific primers (listed in Supplemental Table 2) designed around the targeted sites to form approximately 100 bp amplicons. The PCR product was run on a 2.5% agarose gel to achieve distinct separation of bands corresponding to wild-type and mutated DNA.

135

Sequencing of mutated endogenous gene target sites To investigate the nature of the mutation, CRISPR injected zebrafish were mated to wildtypes (WIKs) and the resulting heterozygote zebrafish analyzed for mutations as described above. The resulting PCR products were then cloned into a plasmid using the pGEM-T kit (Promega). Following transformation, plasmids were isolated using the

PerfectPrep Spin Minikit (5-Prime) and sequenced at the Iowa State University DNA facility. Single base substitutions, insertions or deletions were not considered to be mutant alleles because the possibility of these alterations being generated during the PCR or sequencing process could not be excluded. Only in/dels of 4 bp or more were considered for usage as mutant alleles in our study.

Quantitative PCR

qPCR was performed as described previously33. RNA was isolated from the retina using Tri-reagent (Sigma) according to manufacturer’s instructions. 400ng of RNA was used to generate cDNA using random primers and SuperScript III (Life Technologies) according to standard protocols. SybrGreen MasterMix (Thermofisher) was used to perform qPCR in a Bio-Rad MiniOpticon cycler, using the following program: 15 minutes at

95°C and 40 cycles of 15 seconds at 95°C, 30 seconds at 56°C, and 30 seconds at 72°C. The analysis of the qPCR data was performed exactly as described previously30. Each sample was normalized to Ef1α to obtain the ΔCt values. The difference in ΔCt values between experimental and control was designated as ΔΔCt. The experiments were repeated three times and the average ΔΔCt value calculated. The base 2 analogs of the average ΔΔCt value 136 represented the fold change. These results were plotted graphically with error bars shown.

Specific primers were used for detection of Islet1 [F: 5’ – TTTGTGCGAGACGGGAAAAC – 3’ and R: 5’- CCTTCGAACGTGCTCTCATC- 3’]

Results

Expression of atoh7 correlated genes in zebrafish

Given the similarities in retinal development and morphology between mammals such as mice, and teleosts such as zebrafish, we picked eight atoh7+ single cells from the developing zebrafish retina with the expectation that we would observe the genes clustered with Atoh7 in mouse also expressed in zebrafish atoh7+ cells (Figure 1A). The genes plk3, ikzf5, neurod6a, and uchl1 were found to be correlated with the expression of

Atoh7 in developing mouse single cells thus implicating them in retinal development

(Chowdhury et al., in preparation). Although our zebrafish dataset was small, we observed the expression of our selected genes, plk3, ikzf5, neurod6a, and uchl1 in at least one of our zebrafish single cells picked at 50hpf. Interestingly, Ikzf1, one of the genes responsible for conferring temporal competence in mouse34, was found to be expressed only at very low levels in our atoh7+ zebrafish cells.

To examine the expression of our genes of interest in more detail, we performed in situ hybridization on retinal sections at 48hpf (Figure 1B-E, S1). At 48hpf, atoh7 expression was used as a control and was observed in several large patches of cells prominently in the central retina32 (Figure 1B). plk3 is localized to the region above the lens, consistent with 137 the developing ganglion cell layer35 (GCL) at 48hpf (Figure 1C) and mid-brain at 48hpf

(S1A). ikzf5 is expressed in the central retina in an area that is consistent with retinal progenitor cells and the developing GCL at 48hpf (Figure 1D). neurod6a is expressed in a pattern reminiscent of atoh7 at 48hpf, in that it is seen most predominantly in the peripheral retina and has a second domain consistent with a subset of retinal progenitor cells (Figure 1E). uchl1 expression (Supplemental Figure 1B), although negligible in the eye, is localized along the dorsal spinal cord, possibly in primary sensory neurons named

Rohon-Beard neurons36 (Supplemental Figure 1C, magnified in Supplemental Figure 1D).

F0 CRISPR injections and phenotype analysis

To further study the role of these genes in the zebrafish retina, we used CRISPR-

Cas9 mediated genome editing. The application of CRISPR/Cas9 in zebrafish genome editing has been demonstrated in various studies27,28,37. Briefly, a short CRISPR guide RNA

(gRNA), about 20 basepairs (bp) in length fused to a trans-activating RNA, directs a S. pyogenes nuclease, Cas9, for site directed cleavage of target DNA, adjacent to a NGG trinucleotide protospacer adjacent motif (PAM)37,38. This system results in efficient mutagenesis through the non-homologous end joining (NHEJ)-mediated repair pathway37.

We designed our gRNAs to target specific sequences either at the beginning of the gene or in a sequence corresponding to a conserved functional domain of the protein. Our purpose was to generate in/dels in the initial part of the coding region of the gene through the utilization of the NHEJ repair pathway39 thereby producing frameshift mutations, or to create mutations in the functional region thus rendering the protein inactive. 138

After synthesis of the guide RNA and Cas9 RNA (see methods), we injected zebrafish at the one-cell stage in order to produce mutations throughout the organism. First, we wished to establish whether the small deletions and insertions that were introduced at target sites in the zebrafish genome mutated alleles in enough cells to lead to observable phenotypes in the injected founder fish. We chose several additional genes that we identified from our mouse and zebrafish single cell data (trim9, rassf4, ikzf1, setd8a, setdb1b, setdb2, myst2, kdm7b, mettl1a, ikzf5, prdm11, ebf3) for these tests. Of these,

RASSF4 has been postulated to be a tumor suppressor belonging to the RASSF family of potential tumor suppressors and effectors40,41 and its overexpression is known to inhibit proliferation and invasion of osteosarcoma cells and affect the Wnt family of genes40. We found this gene closely clustered with Atoh7 during the examination of our mouse single cell transcriptomic data (Chowdhury et al., in preparation). TRIM9 has been shown to play roles in axon branching and guidance in the mouse brain42–44. setd8a, setdb1b, setdb2, myst2, kdm7b, and mettl1a were chosen since we noticed these genes all had in common a role in chromatin modification, and hence may influence gene expression45. SETD8A,

SETDB1B, and SETDB2 are members of the SET protein family having roles in multiple processes such as apoptosis, transcription, nucleosome assembly and histone chaperoning46. Specifically, Setdb2 is involved in dorsal organizer formation in zebrafish through negative regulation of Fgf246. KAT7/MYST2, a histone acetyl transferase has been shown to regulate chromosomal assembly47. jhdm1db/kdm7ab is found to be expressed in the central nervous system during the early stages of development in zebrafish48.

METTL11A (NRMT1) is thought to be involved in histone N-terminal methylation and knockout of the gene in mice results in impaired DNA repair49. The remaining genes were 139 selected because they are transcription factors, and therefore could potentially regulate gene expression. IKZF5 is a member of the IKAROS family of proteins involved in lymphatic development and hematopoiesis with known DNA binding activity50. PRDM11 is a PR/SET domain protein, the loss of which promoted Myc driven lymphomagenesis51. Ebf3, another tumor suppressor gene, has been shown to be required for coupling neuronal differentiation with cell cycle exit in chick52. Furthermore, we found it was expressed at high intensity in the brain and retina at 48hpf (Supplemental Figure 1B).

The expression of these genes in our zebrafish single cell data and by in situ hybridization combined with their important biological functions (or predicted functions), encouraged us to believe that these genes would play important roles in zebrafish retinal development. To functionally screen these candidate genes, we used the CRISPR-Cas9 system to generate mutations and then analyzed the injected fish with the goal of assessing retinal phenotypes in the mutant zebrafish (Figure 2). We reasoned that injecting our

CRISPRs into an atoh7:GFP line53, where green fluorescent protein (GFP) is expressed under the atoh7 promoter, would allow us to visualize morphological differences compared to uninjected control retinas. atoh7 expression begins at about 25hpf in the ventronasal retina near the choroid fissure, spreads from the nasal to the temporal and dorsal retina, and is expressed throughout the neural retina by 36hpf7. We visualized the retinas of injected fish at 48hpf, the peak of retinal development54, when both the developing INL and

GCL showed GFP expression. If our genes of interest indeed played roles in cell fate determination, the GFP expression should allow us to visualize changes in retinal morphology at 48hpf. Additionally, we injected CRISPRs to atoh7 and GFP genes as positive controls. Upon examination of the atoh7 CRISPR injected fish, we found a degree of 140 variability. While we saw significant differences in GFP expression in some of the analyzed fish, there were no changes in many others (Figure 2A, 2C). To ensure that our CRISPRs were capable of knocking down gene expression, we injected a CRISPR targeting GFP as well. We saw a robust phenotype in our GFP CRISPR injected fish, with an almost complete loss of GFP signal (Figure 2B, 2D). However, examination of the other CRISPR injected lines

(rassf4 and trim9) led to similar variability as seen with the atoh7 CRISPR injected fish

(Figure 2E-2H are representative images with relatively normal GFP expression in the mutants). We used this method to screen through all the above-mentioned genes, and obtained similar results in each case.

Because of the failure to observe distinct phenotypes, we attempted to target multiple factors belonging to the same gene family simultaneously. Our thought was that this would lead to a better chance of removing gene function by possibly eliminating compensation by family members. Co-injection of CRISPRs to more than one gene (e.g. setd8a, setdb1b and setdb2) brought about a loss of GFP signal in a fraction of injected fish; however, it still did not yield a robust and consistent phenotype (data not shown). These results led us to conclude that evaluation of CRISPR F0 injected fish was not a reliable method for screening potential gene candidates for roles in retinal development and maintenance. Additional cloning and sequencing experiments revealed much variability in the number of mutated alleles within the organisms, ranging from 25% to 80% (data not shown) despite using similar methods to build and inject the CRISPR-Cas9 RNA. Other researchers have observed a similar phenomenon of mosaicism that they attributed to either incomplete diffusion of the RNA or a lag in translation of Cas9,28,55,56.Our variable rates of mutagenesis most likely account for the general lack of observed phenotypes in our 141

F0 injected zebrafish. A new study examining the off-target effects of CRISPRs in mice has found that certain gRNAs target loci independent of their target sites57. This may confound phenotype based analysis, making the study CRIPR injected organisms challenging. A possible solution may to outcross the injected organisms for a few generations to breed out the off-target mutations.

Generation of germline CRISPR mutants

Plk3

PLK3 (Polo-like kinase-3) is a member of the family of Polo-like kinases shown to be important for regulation of cell cycle58. It was found to be correlated with Atoh7 in our mouse single cell dataset consisting of 19 Atoh7+ cells and 38 Atoh7- cells (Chowdhury et al., 2017, in preparation). Moreover, murine PLK3 recently was shown to contribute to the control and progression of the cell cycle58. To analyze the function of plk3 in zebrafish retinal development, we injected gRNA and Cas9 in one-cell zebrafish embryos. We designed our gRNA to bind a short stretch of nucleotides in the 8th exon, right before the sequence encoding the Polo-box domain important for regulation of kinase activity (NCBI

Reference Sequence: NP_958465.1) (Figure 3A). We outcrossed the injected fish for two generations to obtain fish heterozygous for the mutation, and ascertained by sequencing that this line harbored a 5bp deletion (Figure 3B). We obtained zebrafish wildtype- homozygous mutant pairs by crossing two heterozygous fish together. We then performed

IHC on retinal sections obtained from the fish both at developing (data not shown) and adult stages (Figure 4). To examine RGCs, amacrine, bipolar and horizontal cells, we stained 142 the retinal sections with an antibody to Islet112. We observed a distinct reduction of the number of Islet1+ cells in the GCL and INL (Figure 4A-4D). To confirm the results by a different assay, we performed qPCR. This method revealed a 3-fold reduction in Islet1 mRNA in the homozygous mutant retinas (Figure 4E). To further specify the subset of cells that appeared to be lost by Islet1 staining, we performed IHC using other markers. Since

Islet1 stains a subset of amacrine cells known as cholinergic amacrine cells in mouse59, we performed IHC with an antibody to Choline acetyltransferase (Chat), a marker of these cells60 (Figure 5A-5B). We also used Ap2α and Zn8, markers of amacrine cells61 and RGCs60 respectively (Figure 5C-5F). However, in each case we did not observe a significant change between WT and Plk3 homozygous mutant retinas.

Neurod6a

Next, we wished to investigate the function of Neurod6a in retinal development.

Neurod6 is a bHLH transcription factor important for regulating the fate of amacrine cell subtypes during development in mouse retinas62. Curious as to whether Neurod6a plays a similar role in cell fate determination in zebrafish, we generated neurod6a mutants with the help of CRISPR-Cas9 mediated genome editing (Figure 3). Our gRNA was targeted to the sequence encoding the bHLH domain in the second exon of the gene (NCBI Reference

Sequence: NP_571891.3) (Figure 3A). We found the presence of a 5bp deletion by sequencing (Figure 3B). To analyze the fate of the early born cells of the retina, we performed IHC on sections obtained from adult zebrafish WT and neurod6a mutant pairs using antibodies to Islet1 (Figure 6A, 6B), Zn8 (Figure 6C, 6D), and Pax6 (a marker of amacrine and ganglion cells, Figure 6E, 6F)12,61 . Given that members of the Neurod family 143 are implicated in the development of the amacrine cell classes in the mouse retina62,63, we performed IHC with ChAT61 (Figure 6G, 6H), to further examine the morphology and number of cholinergic amacrine cells in the mutant retina. However, we observed no overall changes in the mutant retinas compared to WT (Figure 6).

Ikzf5

Apart from plk3 and neurod6a, we focused on two other genes- ikzf5 and uchl1, that were hypothesized to play roles in retinal cell fate determination. IKZF5 is a member of the

IKAROS family of proteins involved in lymphatic development and hematopoiesis with known DNA binding activity50. Additionally, the related family member, IKZF1, is known to be important for the temporal specification of progenitor cells in the mouse retina34. The function of the ikzf gene family in the zebrafish retina is currently not established. We hypothesized that zebrafish ikzf5 could play a role in cell fate specification much like Ikzf1 does in mouse. We targeted the first exon of Ikzf5 (NCBI Reference Sequence:

NP_991164.1) hoping to cause frameshift mutations and to introduce premature stop codons (Figure 3A). Although we have successfully generated heterozygous mutants for ikzf5 bearing a 4bp deletion (Figure 3B), the function of the gene is still in the process of being analyzed.

Uchl1

UCHL1 (ubiquitin carboxyl-terminal esterase L1) is an unique deubiquitinating enzyme with both hydrolyzing and ligase activities64. A mouse knockout model of Uchl1 revealed early and progressive cortocospinal motor neuron degeneration64. Although 144

Uchl1 does not have a known function in the zebrafish retina, we have observed its expression in the developing mouse (data not shown) and chick retina65 by ISH. uchl1 expression has previously been reported in the brain of developing zebrafish embryo at 48 hpf, specifically in neuronal cells in diencephalon and ventral region of midbrain and hindbrain66. In our hands we identified its expression in the ventral encephalon of the brain66 and Rohon-Beard neurons36 (Supplemental Figure 1C, 1D). This neuronal pattern of expression in zebrafish could make it a suitable model for the study of Parkinson’s disease66 or Amyotrophic Lateral Sclerosis (ALS) in zebrafish67. Our CRISPR to Uchl1 was targeted to exon 3, containing the sequence for a functional peptide binding site (NCBI

Reference Sequence: NP_958885.1) (Figure 3A). We have generated a line of zebrafish carrying a 17bp deletion (Figure 3B) in hb9:GFP background, where GFP is expressed under a motor neuron specific promoter68. We believe that this model will allow us to detect changes in motor neurons in the uchl1 mutant zebrafish and provide a platform for studying motor neuron diseases in zebrafish. We are currently in the process of generating homozygous mutants, which will allow us to study defects in motor neuron development with the help of our hb9:GFP line.

Generation of IKZF1 reporter line

In addition to the generation of mutants through the introductions of in/dels, the

CRISPR-Cas9 system can also be used to knock-in a sequence of interest at a desired site within a gene through the utilization of the homology directed repair (HDR) pathway. For example, a study recently demonstrated the integration of GFP into a target locus resulting in visualization of the desired gene29. The transparency of zebrafish embryos allows real 145 time visualization of gene expression by knocking in a fluorescent construct using CRISPR-

Cas9 mediated genome editing. Ikzf1 is a gene that is known to confer early temporal identity to RPCs in mice. However, no role of Ikzf1 has been demonstrated in zebrafish retinal cell fate determination. To visualize the expression of Ikzf1 in developing zebrafish, we took advantage of the dependence of repair machinery on HDR. We generated a pair of complementary oligos bearing homology to the region around the CRISPR generated double stranded break. These oligos were cloned into a plasmid containing RFP. Injecting this plasmid DNA along with Ikzf1 gRNA and Cas9 into zebrafish embryos at the one-cell stage ensured that the RFP was knocked into the Ikzf1 locus at certain low frequencies. The

RFP expression is shown in Figure 6. Our results were consistent with previous studies which describe a major role for Ikzf1 in hematopoiesis69. From our expression data, it appears that Ikzf1 is expressed in circulating blood cells (Figure 7C, 7D). The expression in the retina seems to be confined to the vascular network as well (Figure 7A, 7B), and not, as we believed in the retinal progenitors48. This is unlike the role of Ikzf1 in mouse retinas, where it is one of the major factors conferring temporal competence to RPCs34.

Discussion

The aim of this study was to examine the functions of candidate genes we hypothesized to be involved in retinal development (Chowdhury et al., in preparation) using CRISPR mediated genome editing in a zebrafish model. We used the CRISPR-Cas9 system to target potential genes important for retinal development such as plk3, neurod6a, 146 ikzf5 and uchl1. Loss of the function of plk3 revealed a loss of Islet1 staining in the retinas of mutant IPL and GCL compared to WT retinas (Figure 4), which was verified by qPCR displaying an almost 4-fold reduction in Islet1 levels in the INL and GCL. We reasoned that this could be either due to a loss of a subset of amacrine cells or the downregulation of the

Islet1 protein itself in the absence of plk3. However, staining WT-mutant retina pairs with other amacrine or ganglion cell markers did not lead to a significant change (Figure 5).

Therefore, the exact mechanism of the loss of Islet1 remains to be elucidated. neurod6a mutants looked grossly normal with respect to WT (Figure 6). uchl1 and ikzf5 mutants are still in the process of being analyzed. Moreover, we found through the analysis of more than 12 genes that analysis of CRISPR-Cas9 injected zebrafish is not a reliable method for screening gene function in retinal development (Figure 2). We also showed that CRISPR-

Cas9 can be used to reveal the expression pattern of a gene (Ikzf1, Figure 7), which is easily visualized in developing zebrafish given the transparency and ex utero development of the embryos. We found that unexpectedly Ikzf1 is not expressed in retinal neurons, but is confined to the vascular system in stark contrast to mouse34.

CRISPR-Cas9 mediated genome editing is a powerful tool to screen gene function in zebrafish given the ease of synthesis, high efficiency, high rates of mutagenesis and relatively efficient and precise integration of exogenous DNA at desired locations. In our hands, 80-90% of CRISPRs injected were capable of introducing heritable mutations. In our experience, the CRISPR-Cas9 system was more successful than TALENs, another technology developed for genome editing70 at inducing mutagenesis. The use of CRISPRs extends beyond creation of random mutations to targeted knock-in of desired sequences into the site of interest at relatively high efficiencies29,71. We used this method to visualize the 147 expression of Ikzf1 within live zebrafish embryos (Figure 7). This technique can be extended to many other applications such as homology directed knock in of mutations associated with disease72, transcriptional modulation of genes, epigenetic modifications and functional genomic screens73,74.

Although the CRISPR-Cas9 system is a powerful tool for creating mutations in target genes, the use of CRISPRs in zebrafish may not be ideal due to the high level of regeneration in these organisms75,76. A recent study comparing phenotypes in zebrafish morphants

(zebrafish injected with antisense oligonucleotides, or morpholinos, to knock down gene function) for the gene egfl7 versus mutants found that while the mutant zebrafish do not show notable phenotypes, egfl7 morphants exhibit severe vascular defects77. This phenomenon is also observed in mice and Arabidopsis, whereby phenotypes observed in gene knockdowns are not detected in knockout organisms77. Another study performed a forward screen to generate zebrafish lines bearing mutations in more than 20 genes using zinc finger nucleases (ZFNs), TALENs and CRISPR-Cas978. The genes targeted were chosen from a list of genes expressed in endothelial cells during embryonic development with homology to genes with known functions. Moreover, it was found that 80% of phenotypes observed in morphant embryos were not present in mutants. A major percentage of the genes analyzed failed to exhibit significant developmental defects. These findings, though surprising, were in accordance with an earlier study which screened 1000 zebrafish genes and reported similar low incidence of discernible phenotypes79. These studies taken with ours suggest a high degree of redundancy in zebrafish embryos. Moreover, these findings are concerning because they point to a high false positive rate with the use of morpholinos, possibly due to off-target effects78. The previously mentioned study77 also showed 148 conclusively the upregulation of a family of genes functionally similar to the targeted gene family. This suggests that differential regulation of expression of other gene families in the absence of the targeted gene may contribute to the discrepancies observed between morphants and mutants.

This study is the first to examine the functions of large numbers of genes involved in retinal development in zebrafish through the generation of zebrafish mutant lines. Many other studies that have generated and characterized zebrafish mutants for genes thought to be involved in various biological processes have similar challenges in characterizing phenotypes. One notable example is the generation and characterization of mutants for the gene, Fus, hypothesized to contribute to neurodegeneration and ALS-like symptoms80. The approach taken here was much like ours. The researchers used the CRISPR-Cas9 system, taking advantage of NHEJ, to generate an allele with an 8 base-pair deletion resulting in frameshift and a premature stop codon. Mass spectrometry and Western Blotting showed the absence of Fus protein. However, subsequent analysis of loss-of-function mutants revealed the lack of motoneuronal defects. Transcriptome and proteome analysis failed to show major changes in alternative splicing, expression of targets of Fus, or general protein steady state levels.

Thus, in conclusion, there are many advantages offered by the zebrafish as a model system such as rapid development, relative simplicity, and accessibility to genetic tools.

However, intrinsic features of the zebrafish such as the ability to regenerate quickly75,76 could obstruct the understanding of the function of specific individual genes in the development and maintenance of the retina and its cell types. In our study, we have observed that CRISPR-mediated mutations in plk3 appear to lead to changes in the number 149 of a subset of amacrine cells in the zebrafish retina. However, only one of the many genes analyzed led to an observable phenotype. Due to redundancy in zebrafish, a better approach may be to target several genes at once through a reverse genetic screening approach. One of the approaches that we have tested preliminarily (by targeting the set genes) is to mutate multiple genes of the same family or involved in the same process, thereby disrupting the process and safeguarding against compensation by other genes at the same time. For example, Trim9 which has roles in axon branching and guidance in the mouse brain42–44, belongs to a large family of genes, members of which are conserved, at least to an extent, between mouse and zebrafish (Chapter 3). Targeting several of the trim genes expressed in the retina by using multiplexed CRISPR genome editing28, could yield more fruitful results both in injected F0 fish and homozygous mutants, thereby gaining insights into their function, by successfully abolishing their effect in the retina. Taken together, our single cell transcriptomic approach coupled with genome editing is a powerful method for dissecting the precise networks of genes controlling early development of the retina, with the potential to be even more effective when the challenges outlined above are overcome.

Acknowledgements

We would like to thank Dr. Jeffrey Essner and the Biology 423 lab for the Plk3 CRISPR design, Dr. Essner for the pT3TS-nCas9n plasmid (Addgene plasmid # 46757), Dr. Marit

Nilson Hamilton for the use of her BioRad MiniOpticon Real-Time PCR System, Dr. Lee 150

Benedickson for advice about qPCR. We would also like to thank Wes Wierson and Kevin

Natukunda for help with the design of ikzf1:GFP construct. We are also grateful to Lauren

Laboissonniere for helpful comments on this chapter.

References

1. Cajal, S. R. La retine des vertebres. Cellule 9, 17–257 (1893).

2. Dowling, J. ‘The Retina’. Harvard Univ. Press. Cambridge, MA (1987).

3. Avanesov, A. & Malicki, J. Analysis of the retina in the zebrafish model. Methods in Cell

Biology 100, (2010).

4. Howe, K. et al. The zebrafish reference genome sequence and its relationship to the

human genome. Nature 496, 498–503 (2013).

5. Nawrocki, L., BreMiller, R., Streisinger, G. & Kaplan, M. Larval and adult visual

pigments of the zebrafish, Brachydanio rerio. Vision Res. 25, 1569–1576 (1985).

6. Easter, Jr., S. S. & Nicola, G. N. The Development of Vision in the Zebrafish (Danio

rerio). Dev. Biol. 180, 646–663 (1996).

7. Hu, M. & Easter, S. S. Retinal neurogenesis: the formation of the initial central patch

of postmitotic cells. Dev. Biol. 207, 309–321 (1999).

8. Poggi, L., Vitorino, M., Masai, I. & Harris, W. A. Influences on neural lineage and mode

of division in the zebrafish retina in vivo. J. Cell Biol. 171, 991–999 (2005). 151

9. Macdonald, R. et al. The Pax protein Noi is required for commissural axon pathway

formation in the rostral forebrain. Development 124, 2397–2408 (1997).

10. Torres, M., Gomez-Pardo, E. & Gruss, P. Pax2 contributes to inner ear patterning and

optic nerve trajectory. Development 122, 3381–3391 (1996).

11. Liu, I. S. C. et al. Developmental expression of a novel murine homeobox gene

(Chx10): Evidence for roles in determination of the neuroretina and inner nuclear

layer. Neuron 13, 377–393 (1994).

12. Vitorino, M. et al. Vsx2 in the zebrafish retina: restricted lineages through

derepression. Neural Dev. 4, 14 (2009).

13. Brown, N. L. et al. Math5 encodes a murine basic helix-loop-helix transcription factor

expressed during early stages of retinal neurogenesis. Development 125, 4821–4833

(1998).

14. Masai, I., Stemple, D. L., Okamoto, H. & Wilson, S. W. Midline signals regulate retinal

neurogenesis in zebrafish. Neuron 27, 251–263 (2000).

15. Brown, N. L., Patel, S., Brzezinski, J. & Glaser, T. Math5 is required for retinal ganglion

cell and optic nerve formation. Development 128, 2497–2508 (2001).

16. Feng, L. et al. MATH5 controls the acquisition of multiple retinal cell fates. Mol Brain

3, 36 (2010).

17. Le, T. T., Wroblewski, E., Patel, S., Riesenberg, A. N. & Brown, N. L. Math5 is required

for both early retinal neuron differentiation and cell cycle progression. Dev. Biol. 295,

764–778 (2006). 152

18. Kay, J. N. et al. Retinal Ganglion Cell Genesis Requires. 30, 725–736 (2001).

19. Pujic, Z. & Malicki, J. Retinal pattern and the genetic basis of its formation in

zebrafish. Semin. Cell Dev. Biol. 15, 105–114 (2004).

20. Yang, Z., Ding, K., Pan, L., Deng, M. & Gan, L. Math5 determines the competence state

of retinal ganglion cell progenitors. Dev. Biol. 264, 240–254 (2003).

21. Brzezinski, J. A., Prasov, L. & Glaser, T. Math5 defines the ganglion cell competence

state in a subpopulation of retinal progenitor cells exiting the cell cycle. Dev. Biol.

365, 395–413 (2012).

22. Reese, B. E. Development of the Retina and Optic Pathway. Vision Res. 51, 613–632

(2010).

23. Prasov, L. & Glaser, T. Pushing the envelope of retinal ganglion cell genesis : Context

dependent function of Math5 ( Atoh7 ). Dev. Biol. 368, 214–230 (2012).

24. Song, W., Zhang, X. & Xia, X. Atoh7 promotes the differentiation of Mu. 682–690

(2015). doi:10.1177/1535370214560965

25. Chen, M. et al. Generation of Retinal Ganglion – like Cells from Reprogrammed Mouse

Fibroblasts. 5970–5978 (2017). doi:10.1167/iovs.09-4504

26. Mullally, M. et al. Expression Profiling of Developing Zebrafish Retinal Cells. Zebrafish

1–10 (2016). doi:10.1089/zeb.2015.1184

153

27. Hwang, W. Y. et al. Efficient genome editing in zebrafish using a CRISPR-Cas system.

Nat Biotechnol 31, 227–229 (2013).

28. Jao, L.-E., Wente, S. R. & Chen, W. Efficient multiplex biallelic zebrafish genome

editing using a CRISPR nuclease system. Proc. Natl. Acad. Sci. U. S. A. 110, 13904–9

(2013).

29. Hisano, Y. et al. Precise in-frame integration of exogenous DNA mediated by

CRISPR/Cas9 system in zebrafish. Sci. Rep. 5, 8841 (2015).

30. Goetz, J. J., Martin, G. M., Chowdhury, R. & Trimarchi, J. M. Onecut1 and Onecut2 play

critical roles in the development of the mouse retina. PLoS One 9, (2014).

31. Almarestani, L., Waters, S. M., Krause, J. E., Bennett, G. J. & Ribeiro-da-Silva, A.

Morphological characterization of spinal cord dorsal horn lamina I neurons

projecting to the parabrachial nucleus in the rat. J. Comp. Neurol. 504, 287–297

(2007).

32. Mullally, M. et al. Expression Profiling of Developing Zebrafish Retinal Cells. Zebrafish

13, 272–280 (2016).

33. Goetz, J. J., Laboissonniere, L. A., Wester, A. K., Lynch, M. R. & Trimarchi, J. M. Polo-

Like Kinase 3 Appears Dispensable for Normal Retinal Development Despite Robust

Embryonic Expression. PLoS One 11, e0150878 (2016).

34. Mattar, P., Ericson, J., Blackshaw, S. & Cayouette, M. A conserved regulatory logic

controls temporal identity in mouse neural progenitors. Neuron 85, 497–504 (2015).

154

35. Stenkamp, D. L. Neurogenesis in the fish retina. Int. Rev. Cytol. 259, 173–224 (2007).

36. Ackerman, K. M., Nakkula, R., Zirger, J. M., Beattie, C. E. & Boyd, R. T. Cloning and

spatiotemporal expression of zebrafish neuronal nicotinic acetylcholine receptor

alpha 6 and alpha 4 subunit RNAs. Dev. Dyn. 238, 980–992 (2009).

37. Gagnon, J. A. et al. Efficient mutagenesis by Cas9 protein-mediated oligonucleotide

insertion and large-scale assessment of single-guide RNAs. PLoS One 9, 5–12 (2014).

38. Varshney, G. K. et al. High-throughput gene targeting and phenotyping in zebrafish

using CRISPR/Cas9. Genome Res. 25, 1030–1042 (2015).

39. Zlotorynski, E. Genome engineering: NHEJ and CRISPR-Cas9 improve gene therapy.

Nat Rev Mol Cell Biol 18, 4 (2017).

40. Eckfeld, K. et al. RASSF4 / AD037 Is a Potential Ras Effector / Tumor Suppressor of

the RASSF Family. Cancer Res. 8688–8693 (2004). doi:10.1158/0008-5472.CAN-04-

2065

41. Shivakumar, L., Minna, J., Sakamaki, T., Pestell, R. & White, M. A. The RASSF1A Tumor

Suppressor Blocks Cell Cycle Progression and Inhibits Cyclin D1 Accumulation The

RASSF1A Tumor Suppressor Blocks Cell Cycle Progression and Inhibits Cyclin D1

Accumulation. Mol. Cell. Biol. 22, 4309–4318 (2002).

42. Winkle, C. C. et al. A novel netrin-1-sensitive mechanism promotes local SNARE-

mediated exocytosis during axon branching. J. Cell Biol. 205, 217–232 (2014). 155

43. Winkle, C. C. et al. Trim9 Deletion Alters the Morphogenesis of Developing and Adult-

Born Hippocampal Neurons and Impairs Spatial Learning and Memory. J. Neurosci.

36, 4940–4958 (2016).

44. Menon, S. et al. The E3 Ubiquitin Ligase TRIM9 Is a Filopodia Off Switch Required for

Netrin-Dependent Axon Guidance. Dev. Cell 35, 698–712 (2015).

45. Dong, X. & Weng, Z. The correlation between histone modifications and gene

expression. Epigenomics 5, 113–116 (2013).

46. Xu, P.-F. et al. Setdb2 restricts dorsal organizer territory and regulates left–right

asymmetry through suppressing fgf8 activity. Proc. Natl. Acad. Sci. 107, 2521–2526

(2010).

47. Ohzeki, J. et al. KAT7/HBO1/MYST2 Regulates CENP-A Chromatin Assembly by

Antagonizing Suv39h1-Mediated Centromere Inactivation. Dev. Cell 37, 413–427

(2016).

48. Thisse, C., and Thisse, B. High Throughput Expression Analysis of ZF-Models

Consortium Clones. ZFIN Direct Data Submiss. (http//zfin.org). (2005).

49. Bonsignore, L. A. et al. NRMT1 knockout mice exhibit phenotypes associated with

impaired DNA repair and premature aging. Mech. Ageing Dev. 146–148, 42–52

(2015).

50. Perdomo, J., Holmes, M., Chong, B. & Crossley, M. Eos and pegasus, two members of

the Ikaros family of proteins with distinct DNA binding activities. J. Biol. Chem. 275,

38347–38354 (2000). 156

51. Fog, C. K. et al. Loss of PRDM11 promotes MYC-driven lymphomagenesis. Blood 125,

1272–1281 (2015).

52. Garcia-Dominguez, M., Poquet, C., Garel, S. & Charnay, P. Ebf gene function is required

for coupling neuronal differentiation and cell cycle exit. Development 130, 6013–

6025 (2003).

53. Masai, I. N-cadherin mediates retinal lamination, maintenance of forebrain

compartments and patterning of retinal neurites. Development 130, 2479–2494

(2003).

54. He, J. et al. How Variable Clones Build an Invariant Retina. Neuron 75, 786–798

(2012).

55. Sung, Y., Kim, J., Kim, H., Lee, J. & Jeon, J. Highly efficient gene knockout in mice and

zebrafish with RNA-guided endonucleases. Genome Res. 24, 125–131 (2014).

56. Shah, A. N., Davey, C. F., Whitebirch, A. C., Miller, A. C. & Moens, C. B. Rapid Reverse

Genetic Screening Using CRISPR in Zebrafish. Zebrafish 13, 152–153 (2016).

57. Schaefer, K. A. et al. Unexpected mutations after CRISPR-Cas9 editing in vivo. Nat

Meth 14, 547–548 (2017).

58. Helmke, C., Becker, S. & Strebhardt, K. The role of Plk3 in oncogenesis. Oncogene 35,

135–147 (2016).

59. Elshatory, Y. et al. Islet-1 controls the differentiation of retinal bipolar and

cholinergic amacrine cells. J Neurosci 27, 12707–12720 (2007). 157

60. Arenzana, F. J. et al. Development of the cholinergic system in the brain and retina of

the zebrafish. Brain Res. Bull. 66, 421–425 (2005).

61. Farhy, C. et al. Pax6 Is Required for Normal Cell-Cycle Exit and the Differentiation

Kinetics of Retinal Progenitor Cells. PLoS One 8, e76489 (2013).

62. Kay, J. N., Voinescu, P. E., Chu, M. W. & Sanes, J. R. Neurod6 expression defines new

retinal amacrine cell subtypes and regulates their fate. Nat. Neurosci. 14, 965–972

(2011).

63. Cherry, T. J., Trimarchi, J. M., Stadler, M. B. & Cepko, C. L. Development and

diversification of retinal amacrine interneurons at single cell resolution. Proc Natl

Acad Sci U S A 106, 9495–9500 (2009).

64. Jara, J. H. et al. Corticospinal motor neurons are susceptible to increased ER stress

and display profound degeneration in the absence of UCHL1 function. Cereb. Cortex

25, 4259–4272 (2015).

65. Laboissonniere, L. A. et al. Single cell transcriptome profiling of developing chick

retinal cells. J. Comp. Neurol. (2017). doi:10.1002/cne.24241

66. Son, O. L. et al. Cloning and expression analysis of a Parkinson’s disease gene, uch-L1,

and its promoter in zebrafish. Biochem. Biophys. Res. Commun. 312, 601–607 (2003).

67. Babin, P. J., Goizet, C. & Raldua, D. Zebrafish models of human motor neuron diseases:

Advantages and limitations. Prog. Neurobiol. 118, 36–58 (2014).

158

68. Flanagan-Steet, H., Fox, M. A., Meyer, D. & Sanes, J. R. Neuromuscular synapses can

form in vivo by incorporation of initially aneural postsynaptic specializations.

Development 132, 4471–4481 (2005).

69. Wei, Y., Xu, J., Zhang, W., Wen, Z. & Liu, F. RNA polymerase III component Rpc9

regulates hematopoietic stem and progenitor cell maintenance in zebrafish.

Development 143, 2103 LP-2110 (2016).

70. Joung, J. K. & Sander, J. D. TALENs: a widely applicable technology for targeted

genome editing. Nat. Rev. Mol. Cell Biol. 14, 49–55 (2013).

71. Li, J. et al. Intron targeting-mediated and endogenous gene integrity-maintaining

knockin in zebrafish using the CRISPR/Cas9 system. Cell Res. 634–637 (2015).

doi:10.1038/cr.2015.43

72. Armstrong, G. A. B. et al. Homology directed knockin of point mutations in the

zebrafish tardbp and fus genes in ALS using the CRISPR/Cas9 system. PLoS One 11,

1–10 (2016).

73. Hsu, P. D., Lander, E. S. & Zhang, F. Development and Applications of CRISPR-Cas9 for

Genome Engineering. Cell 157, 1262–1278 (2014).

74. Barrangou, R. & Doudna, J. A. Applications of CRISPR technologies in research and

beyond. Nat Biotech 34, 933–941 (2016).

75. Wan, J. & Goldman, D. Retina regeneration in zebrafish. Curr. Opin. Genet. Dev. 40, 41–

47 (2016).

159

76. Kishimoto, N., Shimizu, K. & Sawamoto, K. Neuronal regeneration in a zebrafish

model of adult brain injury. Dis. Model. Mech. 5, 200–209 (2012).

77. Rossi, A. et al. Genetic compensation induced by deleterious mutations but not gene

knockdowns. Nature 524, 230–233 (2015).

78. Kok, F. O. et al. Reverse genetic screening reveals poor correlation between

morpholino-induced and mutant phenotypes in zebrafish. Dev. Cell 32, 97–108

(2015).

79. Kettleborough, R. N. W. et al. A systematic genome-wide analysis of zebrafish

protein-coding gene function. Nature 496, 494–497 (2013).

80. Lebedeva, S., de Jesus Domingues, A. M., Butter, F. & Ketting, R. F. Characterization of

genetic loss-of-function of Fus in zebrafish. RNA Biol. 14, 29–35 (2017).

160

Figures and Legends

Figure 1: Expression of genes correlated with atoh7 in the developing zebrafish retina. (A) A heatmap representing microarray data for the expression of atoh7 and genes identified from a mouse Atoh7 cluster, plk3, ikzf5, neurod6a, uchl1 and ikzf1 (Chapter 2) in 8 single 50hpf zebrafish single cells. (B) In situ hybridization showing the expression of plk3, ikzf5, neurod6a genes in retinal sections at 48hpf. Scale bars represent 100 µm.

161

Figure 2: Characterization of 48hpf CRISPR injected atoh7:GFP zebrafish. Zebrafish from an atoh7:GFP background, injected with CRISPRs to atoh7 (2A, 2B), GFP (2C, 2D), rassf4 (2E, 2F), and trim9 (2G, 2H), were visualized using whole-mount imaging.

162

Figure 3. Target locations of CRISPRs and mutations generated. Schematic showing the location of gRNAs targeted to exon 8 of plk3, exon 2 of neurod6a, exon 1 of ikzf5 and exon 3 uchl1 (3A). The mutation (deletion) obtained in each gene is shown in 3B. gRNA target sequences are shown in red.

163

Figure 4. Characterization of adult zebrafish plk3 mutant retinas using IHC and qPCR.

Populations of retinal neurons from both WT and plk3 mutant zebrafish were identified using antibodies to Islet1 (red) within retinal sections (4A-4D). DAPI (shown in blue) marks nuclei (4A, 4B). Scale bars represent 50 µm. islet1 gene expression change in the plk3 mutant retinas relative to WT (n=3) was quantified using qPCR (4E). The error bars shown indicate the standard deviation. 164

Figure 5. Characterization of adult zebrafish plk3 mutant retinas using

IHC.Populations of retinal neurons from both WT and plk3 mutant zebrafish were identified using antibodies (red) to Chat (5A, 5B), Ap2a (5C, 5D) and Zn8 (5E, 5F) within retinal sections. DAPI (shown in blue) marks nuclei. Scale bars represent 50 µm.

165

Figure 6. Characterization of adult zebrafish neurod6a mutant retinas using IHC.

Populations of retinal neurons from both WT and neurod6a mutant zebrafish were identified using antibodies (red) to Islet1 (6A, 6B), Zn8 (6C, 6D), Pax6 (6E, 6F) and Chat

(6G, 6H) within retinal sections. DAPI (shown in blue) marks nuclei. Scale bars represent

50 µm.

166

Figure 7. Expression analysis of ikzf1 in 72hpf zebrafish using CRISPR mediated RFP knock-in. Visualization of RFP knocked into the ikzf1 locus in retinal sections (7A, 7B) and whole mounts (7C, 7D). DAPI (shown in blue) marks nuclei (7A, 7B). Scale bars represent

50 µm.

Supplemental Table 1. List of primers used for synthesis of probes for ISH. The sequences of the primers used to amplify specific genes from zebrafish retinal cDNA for generating RNA probes used for section ISH are shown in this table.

Gene name Forward primer Reverse primer Plk3 atgctcgatcacttggcttt ttcccagtgttgacaccgta Neurod6a tacacgaatgctcgatctgg cgattgtacgcaatgtcacc Ikzf5 tgactttgcctgtcacttcg gcgatgagggagagagaatg Uchl1 ctcaggagtgacgttcacga aggatgtctggctcctctca Trim9 gaatttgaggcgtgtttggt aagccagtgtttcccatgac Ebf3 gccaccaagctaagagcatc agctgtcctttcgatctcca 167

Supplemental Table 2. List of CRISPR target sites and oligos used. The sites within the genes, plk3, neurod6a, ikzf5, uchl1, atoh7, rassf4, setd8a, setdb1b, setdb2, myst2, kdm5ba, jhdm1db, ebf3, mettl11a, and prdm11, targeted by CRISPR-mediated genome editing and the forward and reverse oligos used to generate them are shown in this table.

Gene Target site Forward oligo Reverse oligo

Plk3 GGGAGCTTAATCCACCCAGT TAGAGCTTAATCCACCCAGT AAAACTGGGTGGATTAAGCTC

Neurod6a GGTGCCTGCAGCTGAATGCA TAGGTGCCTGCAGCTGAATGCA AAATGCATTCAGCTGCAGGCA

Ikzf5 GGATTTCGTGAAGGACTTTC TAGGATTTCGTGAAGGACTTTC AAAGAAAGTCCTTCACGAAAT

Uchl1 GGTGTTGAGCAAGCTGGGTG TAGGTGTTGAGCAAGCTGGGTG AAACACCCAGCTTGCTCAACA

Atoh7 GGCCGAGCTGTGCAGACTCC TAGGCCGAGCTGTGCAGACTCC AAACGGAGTCTGCACAGCTCGG

Rassf4 GGTTTACTCAACATCTCCTG TAGGTTTACTCAACATCTCCTG AAACCAGGAGATGTTGAGTAAA

Setd8a GGCCAGAGGCGACCATGAGC TAGGCCAGAGGCGACCATGAGC AAACGCTCATGGTCGCCTCTGG

Setdb1b GGAGCTGGAACCAGAGTTGG TAGGAGCTGGAACCAGAGTTGG AAACCCAACTCTGGTTCCAGCT

Setdb2 GGTGTGTGCGGCGCCGCTGC TAGGTGTGTGCGGCGCCGCTGC AAACGCAGCGGCGCCGCACACA

Myst2 GGCTAAGGCGCAGTGATGCT TAGGCTAAGGCGCAGTGATGCT AAACAGCATCACTGCGCCTTAG

Kdm5ba GGCTCAAAGACAGGGCACTC TAGGCTCAAAGACAGGGCACTC AAACGAGTGCCCTGTCTTTGAG

Jhdm1db GGCAGCCGTACGACGTGAGC TAGGCAGCCGTACGACGTGAGC AAACGCTCACGTCGTACGGCTG

Ebf3 GGTCCGATCGTGGATGCACA TAGGTCCGATCGTGGATGCACA AAACTGTGCATCCACGATCGGA

Mettl11a GGACGTGACTCAGGAATTCC TAGGACGTGACTCAGGAATTCC AAACGGAATTCCTGAGTCACGT

Prdm11 GGCATCGAAGTGCGaAGAAA TAGGCATCGAAGTGCGaAGAAA AAATTTCTtCGCACTTCGATG

168

Supplemental Table 3. List of Primer sets used for genotyping. The primers used for genotyping the CRISPR-induced mutations in the following genes, plk3, neurod6a, ikzf5, uchl1, atoh7, rassf4, setd8a, setdb1b, setdb2, myst2, kdm5ba, jhdm1db, ebf3, mettl11a, and prdm11, are shown in this table. The exon sequences are indicated in upper case, and the intronic sequences in lower case.

Gene Forward primer Reverse primer

Plk3 GGCATTTACTTGATGCGTTG TTTCGATTTCTTCTTGCCAAA

Neurod6a gtgcaaagggctttctcagc gagaaagacgcctcaccact

Ikzf5 cctgtttgaccacagtggaa tttcacaccgatgacagagc

Uchl1 TCCTGTATTTGTGTGTAGGT TGAGGGCACTCCTGACAAAG

Atoh7 atcccaagaacatcggtgac CAAACTTCTCCGGGTCTCTG

Rassf4 ttgtttcttgcagGAAGAGG GAAAGCGCTCATTGTCATCA

Setd8a GACCGGAGAACCGCACAG cacacacacacacacacaca Setdb1b cATGGAAGTGGATGCAGGTT CTCCTTCTCACGCTGCTCAA

Setdb2 tgcatatacgcaggtgtggt tgcacctcgtcatcagacac

Myst2 AGAAGACCGACAGCTCTCAC atgacgattgtaggaagccaa

Kdm5ba tcaaacagtttactgcagtttctaca TGATGAACCCAAAAGGGTCT

Jhdm1db gacagtccgaagATGGCGA tcacatcatacggcggagaa

Ebf3 GGGGGAACCACTATGAAAGA tgatttggcaaacgtacCTC

Mettl11a CCTGCCACTCTTTCGAACAG TAGTTCTCCACCCGTTTGCT

Prdm11 ctgcgaagagtgcAAGAAGT AGACATCCACCTTATCGCCA

169

Supplemental Figure 1. Expression of plk3, ebf3, and uchl1 in developing zebrafish retinas. Whole-mount ISH was used to observe the expression of plk3 (S1A), ebf3 (S1B), uchl1 (4X, S1C) and uchl1 (10X, S1D) in 48hpf zebrafish. Scale bars represent 100 µm.

170

CHAPTER 5. GENERAL DISCUSSION

Introduction

The vertebrate retina is a highly organized tissue, functioning to transform complex information derived from a visual field into a neural message1. The cells of the retina are organized in a layered architecture, from photoreceptors that convert light signal into electrical impulses to retinal ganglion cells that relay this information to the brain2,3. This process is carried out by a repertoire of more than 60 types and subtypes of retinal cells1,4,5. Each mature retinal cell arises from the same pool of retinal progenitor cells and precisely integrates into the retinal circuit to perform a specific role in vision6. The question of how this functional diversity develops is the focus of much research1.

Moreover, the retina is a relatively simple and accessible region of the CNS3,6. Studies investigating the development of the different types of retinal neurons from a pool of retinal progenitor cells (RPCs) could provide insight into the development of the brain, an immensely complex structure containing over 109 neurons2. Not only that, the vertebrate retina exhibits a remarkable amount of conservation in structure and timing of developmental programs7. Divergent phyla such as teleosts and mammals share grossly similar retinal architecture and organization8,9. For example, humans and zebrafish retinas contain the same major classes of retinal cell types, organized and generated in a similar manner9. The conservation of gene function among different organisms suggests that the basic mechanisms of retinal function may be consistent across species10. 171

These characteristics of the retina make it a suitable system for studying neuronal development, cell fate determination and gene function. In this study, I have gained new insights into retinal development through the examination of the single cell transcriptomes of a specific set of developing retinal progenitor cells and functional analyses in mouse and zebrafish as model organisms. Specifically, one of the questions we were interested in addressing was how during development, a progenitor cell decides between various cell fate choices to ultimately generate a specific neuronal type. To answer that question, we used a combination of single cell transcriptomics and in situ hybridization to examine the influence of differential gene expression on cell fate determination (Chapter 2). We explored retinal development further by performing functional studies through the examination of a mouse mutant for a gene named Trim9 that we identified through our single cell analysis (Chapter 3). Trim9 and its approximately 30 related members of the

Trim family were seen to be widely expressed in embryonic single cell data (Chapter 3). Of these, Trim9 was clustered most closely with Atoh7, implicating this E3 ubiquitin ligase in retinal cell fate determination. Detailed analytical experiments show that Trim9 mutants are not deficient for different populations of retinal neurons (Chapter 3). To continue functional studies in a more flexible model organism, we used CRISPR-Cas9 mediated genome editing in zebrafish to mutate several genes chosen from our single cell transcriptomic data and generated mutant zebrafish lines harboring small insertions and deletions. Analysis of one of these lines of fish, the Plk3 mutants, showed a loss of cells in the INL and GCL of adult zebrafish, suggesting that Plk3 may play a role in cell fate determination or maintenance of these cell types (Chapter 4).

172

Retinal Transcriptomics in Atoh7+ Single Cells

To explore the underlying mechanisms behind the diversity that arises during retinal development, we used microarray-based transcriptome profiling of developing retinal progenitor cells. It has previously been shown that developing retinal neurons, even those isolated at the same timepoint, exhibit a large amount of molecular heterogeneity11,12. While transcriptomic profiles of ganglion, amacrine, bipolar and Müller glia have been characterized to certain degree13–15, the role of specific transcription factors in governing gene expression has not been studied extensively. We based our study on the proneural basic helix-loop-helix (bHLH) transcription factor Atoh7, whose expression is known to act permissively to produce ganglion cells, and the absence of which leads to an almost complete loss of this cell type and a concomitant increase in the other early born cells of the retina16–18. Previous studies have been valuable in terms of determining differentially expressed genes between Atoh7+ and Atoh7- populations19, or determining gene regulatory networks activated by the deletion of Atoh720. In addition, Atoh7 expression occurs late in the cell cycle, right around when a cell fate decision is believed to take place21,22. Upon studying the transcriptomes of single Atoh7+ cells, not only did we discover a great deal of heterogeneity even between individual Atoh7+ cells picked from the same timepoint, we were able to discover new genes whose expression correlated with that of Atoh7, thus implicating them in retinal cell fate determination (Chapter 2). The generation and examination of mutations in these genes will define their functions in the developing retina23,24 (Chapters 3 and 4). Moreover, we found that while the majority of 173

Atoh7+ cells were not classified as RGCs, they nonetheless expressed a number of early- expressed RGC genes. Additionally, many of these potential developing RGCs also expressed genes that were characteristic of other early-born retinal neurons. This result points to a time of “confusion” in gene expression during the time when a cell fate choice is being made and is consistent with similar results seen in the developing chicken and zebrafish retinas25,26. Taken together, these single cell transcriptomes show that cells travel through a period where they express markers of a number of cell types, suggesting that this may be the molecular correlate of the stochastic or probabilistic model of cell fate choice27–29.

Functional Studies Using Trim9 Mouse Mutants

One of the genes that was correlated to Atoh7 during development was an E3 ubiquitin ligase, Trim930,31. Although previously shown to be involved in axon branching and guidance in the cerebral cortex30,31, its role in the retina was unclear. We observed the expression of many members of the Trim family in our single cell data, of which Trim9 was one of the most closely clustered with Atoh7. Based on the correlation between Trim9 and

Atoh7 in our clustering data, we hypothesized that Trim9 could play a role in cell fate determination in the retina. However, our analysis of a Trim9 knockout mouse did not support our hypothesis. The loss of Trim9 did not lead to a significant reduction of any type of retinal neuron. One explanation for this phenomenon could be that Trim9 belongs to a big family, of which over 30 members were detected in our mouse single cell data. Some of these genes had similar expression patterns and therefore could play redundant roles in 174 the retina. Although it would have been interesting to pursue other aspects this gene’s function in the retina, such as its role in RGC axon guidance, breeding issues with mice prevented us from carrying out further experiments. We wondered if the loss of Trim9 affected the line’s ability to produce healthy pups; however, mating the mutants with wildtype mice did not solve the problem.

Functional Studies Using Zebrafish Mutants

Our single cell data had revealed to us about 50 genes that were correlated with

Atoh7, and were previously unexamined (Chapter 2). Since Atoh7 is necessary but insufficient for RGC production32,33, the entire repertoire of genes involved in early retinal cell fate determination of this cell type is still not fully understood. We decided to screen the genes correlated with Atoh7, hoping we could reveal new functions for these genes in retinal cell fate acquisition. It soon became evident that with the high cost of mouse housing and care, a relatively large scale functional screen was not practicable. Thus, we decided to switch to a more affordable and convenient model system, the zebrafish or

Danio rerio. The zebrafish genome is completely sequenced, joining human and mouse as the third high-quality complete sequenced vertebrate genome, making it particularly amenable to genome editing. We also picked 8 single Atoh7+ cells from the developing zebrafish retina to confirm the presence of our candidate genes. Therefore, we wished to study the candidate genes arising from our mouse single cell data through a loss-of- function approach. To that end, we used CRISPR-Cas9 technology34,35 to generate several mutant lines. In our experience, CRISPR-Cas9 synthesis was quicker and injections led to a 175 higher rate of success in generating mutations than TALENs36 (~80% with CRISPR-Cas9 compared to ~10% with TALENs). In general, we did not face issues with toxicity35 when we injected standard concentrations of 300 pg Cas9 and 25 pg gRNA34.

We studied CRISPR-injected zebrafish for 12 genes chosen from our transcriptomic data. However, we were not able to find robust and consistent phenotypes in the F0 injected fish, most likely due to incomplete mutagenesis leading to mosaicity and variability between individual injected zebrafish. Thus, we concluded that this ‘quick and dirty’ method for screening phenotypes with help of CRISPRs was not effective enough, and we moved on to generating homozygous mutants. Our analysis of zebrafish mutant lines revealed that in the presence of a mutation in the gene plk3, there was a loss of Islet1+ cells.

Since Islet1 stains a subset of amacrine cells in the retina37, we examined retinal populations with the help of other amacrine cell markers38,39, but did not observe any changes from the wildtype retinas. At this point, it is unclear what this means and whether there is truly a subset of amacrine cells that fail to develop in the mutant fish or if there is just a decrease in expression of Islet1 in a subset of cells. Because plk3 is a kinase by nature, further proteomic studies may be required to detect subtle changes in protein activity in the mutants and really find a clear phenotype. The other two lines analyzed, trim9, a gene encoding an E3 ubiquitin ligase30,31, for which we also studied a mouse knockout model (Chapter 3), and neurod6a, a gene encoding a bHLH transcription factor40, also did not yield significant phenotypes in our zebrafish models. Work still needs to be done to complete analysis of mutants for the genes ikzf5 (a member of the Ikaros family involved in regulation of temporal competence in mouse)41,42, and uchl1 (a gene thought be involved in motor neuron disease)43,44. 176

Redundancy and Compensation in Animal Models

One of the major challenges faced during our investigation into retinal function was the lack of phenotypic changes in the mutants generated. Despite our best efforts, we were unable to detect a significant phenotype in Trim9 mutant animals. A strategy that is often undertaken to further delve into gene function is to misexpress the gene in vivo6. For example, a study found that loss of the bHLH transcription factor, OLIG2 expressed in small subsets of RPCs in the mouse, did not lead to any observable changes in cell fate45.

However, overexpression of this gene led to premature cell cycle exit, indicating that this gene plays a distinct role in early retinal development45. Therefore, it is possible that overexpression of Trim9 in the retina could help us probe further into its function.

Unfortunately, technical issues with electroporation and in vivo transfection prevented us from carrying out this experiment. Additionally, it is entirely possible that Trim9, though implicated in cell fate acquisition, has a bigger role in axon guidance in the mouse brain.

Our strategy was to test this hypothesis by means of anterograde axonal labelling which would allow us to determine the termination points of RGC axons46, and observe changes in the path of axons from the eye to brain. However, the inability to find WT-KO breeding pairs in the Trim9 mouse line did not allow us to pursue this approach. For future studies, it may be interesting to see the results of the experiments outlined above.

We recognized that the inability to detect a phenotype is potentially one of the drawbacks of studying genes belonging to larger gene families. Trim9, for example belongs 177 to large family consisting of more than 30 members expressed in the retina (Chapter 3). It is not unprecedented for members of a gene family to functionally compensate for another’s absence47,48. In some cases, when a cell-cycle control gene is mutated, another gene from the same family is upregulated49. For example, loss of cyclin D1 in the mouse, causes levels of cyclin D3 to be upregulated to levels that approach those seen for cyclin D1 in the wild-type retina49,50. Similarly, it could be possibly that other TRIM E3 Ubiquitin ligases adopt additional roles in the absence of TRIM9. Although no such upregulation was evident in our microarray data, the abundancy of Trim family members or subtle changes in RNA expression could have made this phenomenon hard to detect. Similarly instances of genetic compensation in zebrafish are well known51,52, which could explain why we did not see drastic phenotypes in our trim9 and neurod6a mutants. To circumvent this issue, one approach could be to target multiple sets of genes at the same time, described recently though the utilization of CRISPR-Cas953. For example, it may be useful to target Trim genes expressed in similar subsets of cells, with the reasoning that these genes may all be involved in similar functions. Information about the expression of these genes in individual cells could be derived from our transcriptome data (Chapter 3). Working backwards from there, based on the phenotypes observed, one could make conclusions about the roles of individual genes.

Conclusion

Single cell transcriptomics combined with CRISPR-mediated genome editing is a powerful approach for the study of gene networks governing cell fate decisions. While 178 single-cell transcriptomics has the ability to identify new genes that could be potentially involved in biological processes, or hint at new functions for a gene11–13, CRISPR-Cas9 technology is capable of aiding in the study of large numbers of genes in shorter amounts of time through perturbation of gene function. Current technologies making use of multiplexed bi-allelic genome editing53, could resolve some of the problems we face with generating and analyzing single mutants for each gene. Targeting multiple genes53 at once would not only save time, but would allow us to study genes based on the phenotypes generated.

Technologies such as genome editing and the use of pluripotent stem cells or organoids are innovative approaches that are transforming research on neuronal development, neurodegeneration and regenerative approaches54,55. Neurons differentiated in vitro from patient-derived induced pluripotent stem cells have been used to model various degenerative diseases, including amyotrophic lateral sclerosis (ALS), Parkinson’s disease (PD), Alzheimer’s disease (PD), age-related macular degeneration (AMD), retinitis pigmentosa (RP) and others55–59. Recently, the generation of retinal organoids – large, complex, 3D retinas derived from embryonic stem cells has been reported54,59. In fact, brain and retinal organoids both appear to exhibit properties that recapitulate human organ development60. In addition to providing a model system for the study of development, the therapeutic promise of these stem cell derived tissues and organoids is endless54,59,61.

While not unexpected, the extensive heterogeneity in gene expression reported within similar types of cells comprising these organoids, and also between organoids54, could hinder the use of these organoids in applications such as clinical transplantations and drug testing. Large scale single cell gene expression analysis could be the ideal tool to resolve the 179 variation in gene expression in these organoids, and coupled with genome editing could push these towards phenotypic uniformity. For example, knowledge of key transcriptional cascades underpinning cell fate decisions revealed by single-cell profiling of neuronal tissues could be used to drive the generation of specific neuronal types60. Single cell profiling could then be used to validate the homogeneity and authenticity of the neurons generated60.

CRISPR-Cas9 is already the preferred tool of choice when it comes to disease modeling in not only in animal models, but also in stem cells and organoids55. The multiplexed genome editing ability and high efficiency of this tool makes it convenient for the induction or correction of specific mutations, which could be useful for both disease modelling and treatment53,62. This technique has been implemented for modeling diseases such as Retinitis Pigmentosa (RP)55, as well as for precise correction of a pathogenic RP mutation in patient-derived iPSCs63. These studies show that advancement in the use of

CRISPR-Cas9 technology combined with the use of iPSCs and single cell transcriptome analysis can have far-reaching clinical applications in gene therapy. Continued analysis of the retinal transcriptome and gene function to attain a better understanding of the development of retinal neurons is critical for future regenerative studies.

References

1. Cepko, C. Intrinsically different retinal progenitor cells produce specific types of

progeny. Nat. Rev. Neurosci. 15, 615–627 (2014). 180

2. Wohrer, A. The vertebrate retina: a functional review. (2008).

3. Cepko, C. L., Austin, C. P., Yang, X., Alexiades, M. & Ezzeddine, D. Cell fate

determination in the vertebrate retina. Proc Natl Acad Sci USA 93, (1996).

4. Masland, R. H. Neuronal diversity in the retina. Curr. Opin. Neurobiol. 11, 431–436

(2001).

5. Masland, R. H. The Neuronal Organization of the Retina. Neuron 76, 266–280 (2012).

6. Goetz, J. J., Farris, C., Chowdhury, R. & Trimarchi, J. M. Making of a retinal cell: Insights

into retinal cell-fate determination. International Review of Cell and Molecular Biology

308, (Elsevier Inc., 2014).

7. Livesey, F. J. & Cepko, C. L. Vertebrate neural cell-fate determination: lessons from

the retina. Nat. Rev. Neurosci. 2, 109–118 (2001).

8. Cajal, S. R. La retine des vertebres. Cellule 9, 17–257 (1893).

9. Avanesov, A. & Malicki, J. Analysis of the retina in the zebrafish model. Methods in Cell

Biology 100, (2010).

10. Demb, J. B. & Singer, J. H. Functional Circuitry of the Retina. Annu. Rev. Vis. Sci. 1, 263–

289 (2015).

11. Trimarchi, J. M., Stadler, M. B. & Cepko, C. L. Individual retinal progenitor cells display

extensive heterogeneity of gene expression. PLoS One 3, (2008).

181

12. Trimarchi, J. M. et al. Molecular Heterogeneity of Developing Retinal Ganglion and

Amacrine Cells Revealed through Single Cell Gene Expression Profiling. J. Comp.

Neurol. 504, 287–297 (2007).

13. Cherry, T. J., Trimarchi, J. M., Stadler, M. B. & Cepko, C. L. Development and

diversification of retinal amacrine interneurons at single cell resolution. Proc Natl

Acad Sci U S A 106, 9495–9500 (2009).

14. Kim, D. S. et al. Identification of molecular markers of bipolar cells in the murine

retina. J. Comp. Neurol. 507, 1795–1810 (2008).

15. Roesch, K. et al. The transcriptome of retinal Müller glial cells. J. Comp. Neurol. 509,

225–238 (2008).

16. Brown, N. L., Patel, S., Brzezinski, J. & Glaser, T. Math5 is required for retinal ganglion

cell and optic nerve formation. Development 128, 2497–2508 (2001).

17. Le, T. T., Wroblewski, E., Patel, S., Riesenberg, A. N. & Brown, N. L. Math5 is required

for both early retinal neuron differentiation and cell cycle progression. Dev. Biol. 295,

764–778 (2006).

18. Feng, L. et al. MATH5 controls the acquisition of multiple retinal cell fates. Mol Brain

3, 36 (2010).

19. Gao, Z., Mao, C. A., Pan, P., Mu, X. & Klein, W. H. Transcriptome of Atoh7 retinal

progenitor cells identifies new Atoh7-dependent regulatory genes for retinal

ganglion cell formation. Dev. Neurobiol. 74, 1123–1140 (2014).

182

20. Mu, X. et al. A gene network downstream of transcription factor Math5 regulates

retinal progenitor cell competence and ganglion cell fate. Dev. Biol. 280, 467–481

(2005).

21. Brown, N. L. et al. Math5 encodes a murine basic helix-loop-helix transcription factor

expressed during early stages of retinal neurogenesis. Development 125, 4821–4833

(1998).

22. McConnell, S. K. & Kaznowski, C. E. Cell cycle dependence of laminar determination in

developing neocortex. Science 254, 282–285 (1991).

23. Goetz, J. J., Martin, G. M., Chowdhury, R. & Trimarchi, J. M. Onecut1 and Onecut2 play

critical roles in the development of the mouse retina. PLoS One 9, (2014).

24. Goetz, J. J., Laboissonniere, L. A., Wester, A. K., Lynch, M. R. & Trimarchi, J. M. Polo-

Like Kinase 3 Appears Dispensable for Normal Retinal Development Despite Robust

Embryonic Expression. PLoS One 11, e0150878 (2016).

25. Laboissonniere, L. A. et al. Single cell transcriptome profiling of developing chick

retinal cells. J. Comp. Neurol. (2017). doi:10.1002/cne.24241

26. Mullally, M. et al. Expression Profiling of Developing Zebrafish Retinal Cells. Zebrafish

1–10 (2016). doi:10.1089/zeb.2015.1184

27. He, J. et al. How Variable Clones Build an Invariant Retina. Neuron 75, 786–798

(2012).

28. Cayouette, M., Barres, B. a & Raff, M. Importance of intrinsic mechanisms in cell fate

decisions in the developing rat retina. Neuron 40, 897–904 (2003). 183

29. Gomes, F. L. A. F. et al. Reconstruction of rat retinal progenitor cell lineages in vitro

reveals a surprising degree of stochasticity in cell fate decisions. Development 138,

227–35 (2011).

30. Winkle, C. C. et al. A novel netrin-1-sensitive mechanism promotes local SNARE-

mediated exocytosis during axon branching. J. Cell Biol. 205, 217–232 (2014).

31. Menon, S. et al. The E3 Ubiquitin Ligase TRIM9 Is a Filopodia Off Switch Required for

Netrin-Dependent Axon Guidance. Dev. Cell 35, 698–712 (2015).

32. Brzezinski, J. A., Prasov, L. & Glaser, T. Math5 defines the ganglion cell competence

state in a subpopulation of retinal progenitor cells exiting the cell cycle. Dev. Biol.

365, 395–413 (2012).

33. Prasov, L. & Glaser, T. Pushing the envelope of retinal ganglion cell genesis : Context

dependent function of Math5 ( Atoh7 ). Dev. Biol. 368, 214–230 (2012).

34. Hwang, W. Y. et al. Efficient genome editing in zebrafish using a CRISPR-Cas system.

Nat Biotechnol 31, 227–229 (2013).

35. Shah, A. N., Davey, C. F., Whitebirch, A. C., Miller, A. C. & Moens, C. B. Rapid Reverse

Genetic Screening Using CRISPR in Zebrafish. Zebrafish 13, 152–153 (2016).

36. Joung, J. K. & Sander, J. D. TALENs: a widely applicable technology for targeted

genome editing. Nat. Rev. Mol. Cell Biol. 14, 49–55 (2013).

37. Elshatory, Y. et al. Islet-1 controls the differentiation of retinal bipolar and

cholinergic amacrine cells. J Neurosci 27, 12707–12720 (2007). 184

38. Arenzana, F. J. et al. Development of the cholinergic system in the brain and retina of

the zebrafish. Brain Res. Bull. 66, 421–425 (2005).

39. Farhy, C. et al. Pax6 Is Required for Normal Cell-Cycle Exit and the Differentiation

Kinetics of Retinal Progenitor Cells. PLoS One 8, e76489 (2013).

40. Kay, J. N., Voinescu, P. E., Chu, M. W. & Sanes, J. R. Neurod6 expression defines new

retinal amacrine cell subtypes and regulates their fate. Nat. Neurosci. 14, 965–972

(2011).

41. Perdomo, J., Holmes, M., Chong, B. & Crossley, M. Eos and pegasus, two members of

the Ikaros family of proteins with distinct DNA binding activities. J. Biol. Chem. 275,

38347–38354 (2000).

42. Elliott, J., Jolicoeur, C., Ramamurthy, V. & Cayouette, M. Ikaros Confers Early

Temporal Competence to Mouse Retinal Progenitor Cells. Neuron 60, 26–39 (2008).

43. Bilguvar, K. et al. Recessive loss of function of the neuronal ubiquitin hydrolase

UCHL1 leads to early-onset progressive neurodegeneration. Proc. Natl. Acad. Sci.

110, 3489–3494 (2013).

44. Jara, J. H. et al. Corticospinal motor neurons are susceptible to increased ER stress

and display profound degeneration in the absence of UCHL1 function. Cereb. Cortex

25, 4259–4272 (2015).

45. Hafler, B. P. et al. Transcription factor Olig2 defines subpopulations of retinal

progenitor cells biased toward specific cell fates. Proc. Natl. Acad. Sci. U. S. A. 109,

7882–7887 (2012). 185

46. Yonehara, K. et al. Expression of SPIG1 reveals development of a retinal ganglion cell

subtype projecting to the medial terminal nucleus in the mouse. PLoS One 3, (2008).

47. Hoser, M. et al. Sox12 Deletion in the Mouse Reveals Nonreciprocal Redundancy with

the Related Sox4 and Sox11 Transcription Factors ᰔ. 28, 4675–4687 (2008).

48. Barbaric, I., Miller, G. & Dear, N. Appearances can be deceiving : phenotypes of

knockout mice. 6, (2007).

49. Dyer, M. A. & Cepko, C. L. Regulating proliferation during retinal development. Nat.

Rev. Neurosci. 2, 333–342 (2001).

50. Tong, W. & Pollard, J. W. Genetic Evidence for the Interactions of Cyclin D1 and

p27Kip1 in Mice. Mol. Cell. Biol. 21, 1319–1328 (2001).

51. Rossi, A. et al. Genetic compensation induced by deleterious mutations but not gene

knockdowns. Nature 524, 230–233 (2015).

52. Wei, C., Wang, H., Zhu, Z. & Sun, Y. Transcriptional Factors Smad1 and Smad9 Act

Redundantly to Mediate Zebrafish Ventral Specification Downstream. 289, 6604–

6618 (2014).

53. Jao, L.-E., Wente, S. R. & Chen, W. Efficient multiplex biallelic zebrafish genome

editing using a CRISPR nuclease system. Proc. Natl. Acad. Sci. U. S. A. 110, 13904–9

(2013).

54. Volkner, M. et al. Retinal Organoids from Pluripotent Stem Cells Efficiently

Recapitulate Retinogenesis. Stem Cell Reports 6, 525–538 (2016). 186

55. Peng, Y.-Q., Tang, L.-S., Yoshida, S. & Zhou, Y.-D. Applications of CRISPR/Cas9 in

retinal degenerative diseases. Int. J. Ophthalmol. 10, 646–651 (2017).

56. Steinbeck, J. A. & Studer, L. Moving stem cells to the clinic: Potential and limitations

for brain repair. Neuron 86, 187–206 (2015).

57. Sterneckert, J. L., Reinhardt, P. & Scholer, H. R. Investigating human disease using

stem cell models. Nat Rev Genet 15, 625–639 (2014).

58. Treutlein, B. et al. Dissecting direct reprogramming from fibroblast to neuron using

single-cell RNA-seq. Nature 534, 391–395 (2016).

59. Reichman, S. et al. Generation of Storable Retinal Organoids and Retinal Pigmented

Epithelium from Adherent Human iPS Cells in Xeno-Free and Feeder-Free

Conditions. Stem Cells 35, 1176–1188 (2017).

60. Poulin, J.-F., Tasic, B., Hjerling-Leffler, J., Trimarchi, J. M. & Awatramani, R.

Disentangling neural cell diversity using single-cell transcriptomics. Nat. Neurosci.

19, 1131–1141 (2016).

61. Lancaster, M. A. & Knoblich, J. A. Organogenesis in a dish: Modeling development and

disease using organoid technologies. Science (80-. ). 345, 1247125–1247125 (2014).

62. Pennisi, E. The CRISPR Craze. Science (80-. ). 341, 833–836 (2013).

63. Bassuk, A. G., Zheng, A., Li, Y., Tsang, S. H. & Mahajan, V. B. Precision Medicine:

Genetic Repair of Retinitis Pigmentosa in Patient-Derived Stem Cells. Sci. Rep. 6,

19969 (2016).