Copyright by Alfire Sidik 2019

The Dissertation Committee for Alfire Sidik Certifies that this is the approved version of the following Dissertation:

Genetic and Bioinformatic Approaches to Characterize Ethanol Teratogenesis.

Committee:

Johann K. Eberhart, Supervisor

R. Adron Harris

Vishwanath R. Iyer

Christopher S. Sullivan

John B. Wallingford

Genetic and Bioinformatic Approaches to Characterize Ethanol Teratogenesis.

by

Alfire Sidik

Dissertation

Presented to the Faculty of the Graduate School of The University of Texas at Austin in Partial Fulfillment of the Requirements for the Degree of

Doctor of Philosophy

The University of Texas at Austin August 2019

Acknowledgements

I first and foremost want to extend my deepest gratitude to my mentor, Dr. Johann K. Eberhart, without whom this dissertation would not be possible. Johann, thank you for your invaluable insight, patience, guidance, and unwavering encouragement. I am also extremely grateful to Mary Swartz. Mary, you truly are the glue that holds the lab together. I joined the lab with little-to-no wet lab and embryology experience and attribute a lot of what I learned to you. Thank you for always being receptive to my questions and for your valuable scientific advice. I would like to express my appreciation to all of my committee members for their constructive advice, suggestions, and professional guidance. I would especially like to thank Dr. Adron Harris. Adron, thank you for your compassion and support of me throughout the years. I’d like to thank all members of the lab, both past and present, who I’ve had the pleasure of working with. Thank you Neil McCarthy, Patrick McGurk, Ben Lovely, Anna Percy, Angie Martinez, Tim Kuka, Ranjeet Kar, Desire Buckley, Yohaan Fernandes, Scott Tucker, and Josh Everson. To my undergraduate and high school mentees, Elaine Avshman, Jenna Beam, Jennyly Nguyen, Hannah Kirby, Sruthi Ramaswamy, Maitreyi Ramaswamy, and Natasha Paul. Thank you for making the mentorship experience so enjoyable. I’d also like to thank all of my graduate school friends for their constant support and encouragement. Many thanks to Groves Dixon for his bioinformatic expertise and contributions to the project. Lastly, I want to extend my deepest appreciation to my parents and sister, Almire Sidik. I would not be where I am today if it weren’t for your unconditional love and support. iv

Abstract

Genetic and Bioinformatic Approaches to Characterize Ethanol Teratogenesis.

Alfire Sidik, PhD The University of Texas at Austin, 2019

Supervisor: Johann K. Eberhart

Alcohol consumption during pregnancy is the most preventable cause of birth defects, yet approximately 2-5% of children are afflicted with Fetal Alcohol Spectrum Disorders (FASD). FASD describes the complex and highly variable deleterious phenotypes caused by prenatal alcohol exposure. Twin studies suggest a genetic predisposition, contributing to the variation in risk for FASD. Despite this, we lack a basic understanding of 1) the factors that protect or predispose an individual to FASD and 2) how these genetic factors interact in ethanol teratogenesis. Results from a genetic “shelf” screen revealed vangl2, a member of the Wnt/planar cell polarity (PCP) pathway that mediates convergent extension movements that narrow and elongate the body axis, as an ethanol-sensitive genetic locus. Untreated vangl2 mutants displayed a relatively intact craniofacial skeleton. Ethanol-exposed vangl2 heterozygotes and mutants, displayed cyclopean and midfacial defects. To assess the relative level of variation of the transcriptional response to ethanol, I performed single embryo RNA-seq during early embryonic stages. Individual zebrafish v embryos were exposed to a subteratogenic dose of 1% ethanol in embryo media. My data suggests that the effect of ethanol is subtle; time is the most important variable driving variation in fold coverage across all samples. Despite this, I find a number of differentially expressed genes in response to ethanol. Transcriptional changes due to ethanol are indicative of increased oxidative stress and ion transport and reduced DNA replication and cell division. Using a bioinformatic approach, I find cyclopamine, a Hedgehog pathway inhibitor, interacts with ethanol. Further genetic analyses shows that ethanol disrupts convergent extension of the mesoderm, which in turn disrupts localization of shh in the axial mesoderm, a signal necessary to separate the eye field. I find this effect to be further exacerbated in the vangl2 mutant background. Together these data yield important insight necessary to advance understanding and treatment for FASD.

vi Table of Contents

List of Tables ...... xi

List of Figures ...... xii

Chapter 1: General Introduction and Significance ...... 1

1.1. Fetal Alcohol Spectrum Disorders (FASD) ...... 2

1.1.1. Holoprosencephaly ...... 4

1.2 Gene-Ethanol Interactions ...... 5

I.2.1. The vangl2-ethanol interaction ...... 6

I.2.2. The hedgehog signaling pathway ...... 7

1.3 Gastrulation ...... 8

I.3.1. Germ layer formation in humans ...... 8

I.3.2. Germ layer formation in zebrafish ...... 9

1.4 Eye Field Formation ...... 10

I.4.1. Optic vesicle morphogenesis ...... 11

I.4.2. Transcription factors specify the eye field ...... 11

I.4.3. The Wnt signaling pathway in forebrain patterning ...... 12

1.5 The Wnt Signaling Pathway ...... 12

1.5.1. The non-canonical Wnt/PCP pathway ...... 13

1.5.2. Convergent extension ...... 15

Chapter 2: Convergent extension defects underlie susceptibility to midfacial hypoplasia in an ethanol-sensitive mutant, vangl2...... 16

2.1. Abstract ...... 16

2.2. Introduction ...... 17

vii 2.3. Results ...... 19

2.3.1. vangl2 mutants and heterozygotes are sensitive to 1% ethanol during early embryogenesis ...... 19

2.3.2 The effect of ethanol on the early zebrafish transcriptome is subtle relative to developmental time ...... 20

2.3.3. Ethanol has effects on transcription that are largely distinct between different developmental timepoints ...... 21

2.3.4. Ethanol does not affect the Wnt/PCP pathway at the transcriptional level ...... 22

2.3.5. Transcriptional changes due to ethanol are indicative of increased oxidative stress and ion transport and reduced DNA replication and cell division ...... 22

2.3.6. Modules of co-regulated genes related to ethanol exposure ...... 23

2.3.7. Cyclopamine is predicted to mimic the effects of ethanol ...... 24

2.3.8. Ethanol indirectly attenuates Shh signaling ...... 26

2.3.9. Ethanol disrupts convergent extension ...... 27

2.3.10. Ethanol alters six3 and rx3 expression in the eye field ...... 29

2.3.11. Mutation in gpc4 enhances cyclopia in a dose-dependent manner ...29

2.4. Discussion ...... 30

2.5. Materials and Methods ...... 32

2.5.1. Zebrafish (Danio rerio) Care and Use ...... 32

2.5.2. Sample collection and RNA extraction ...... 33

2.5.3. RNA-seq data processing ...... 33

2.5.4. Differential expression analysis ...... 34

2.5.5. GO enrichment ...... 35

2.5.6. Weighted Gene Correlation Network Analysis (WGCNA) ...... 35

viii 2.5.7. Quantitative Real-Time qRT-PCR (qRT-PCR) ...... 36

2.5.8. Cartilage and Bone Staining and Measurements ...... 36

2.5.9. In Situ Hybridization ...... 36

2.5.10. TUNEL Staining ...... 36

2.6. Acknowledgements ...... 37

2.7. Figures ...... 38

Chapter 3: RNA-seq analysis of vangl2...... 53

3.1. Introduction ...... 53

3.2. Results ...... 54

3.2.1. A transcriptional profile of untreated and ethanol-treated vangl2 embryos ...... 54

3.2.2. Differentially expressed genes in response to genotype, temperature, and ethanol ...... 55

3.2.3. GO and KEGG pathway analysis ...... 56

3.2.4. Weighted Gene Correlation Network Analysis (WGCNA) ...... 57

3.2.5. Gene-ethanol interactions ...... 60

3.3. Discussion ...... 60

3.4. Materials and Methods ...... 62

3.4.1. Zebrafish (Danio rerio) Care and Ethanol Treatment ...... 62

3.4.2. Sample collection and RNA extraction ...... 62

3.4.3. RNA-seq data processing ...... 63

3.4.4. Differential expression analysis ...... 64

3.4.5. GO enrichment ...... 64

3.4.6. Weighted Gene Correlation Network Analysis (WGCNA) ...... 64

ix 3.5. Figures ...... 65

3.6. Tables ...... 74

Chapter 4: Summary and future directions ...... 75

4.1. Future directions ...... 77

4.1.1. microRNAs and ethanol ...... 77

4.1.2. Protein localization and levels ...... 78

4.2. Tables ...... 80

4.3. Figures ...... 81

Appendix: List of Acronyms and Abbreviations ...... 82

Bibliography ...... 84

Vita ...... 97

x List of Tables

Table 3.1: Significant GO terms describing biological processes for the cyan

module ...... 87

Table 4.1: microRNAs upregulated by ethanol are predicted to target Wnt/PCP

pathway members ...... 93

xi List of Figures

Figure 2.1: vangl2 mutants are sensitive to ethanol during early embryogenesis ...... 51

Figure 2.2: The effect of ethanol on the early zebrafish transcriptome is subtle ...... 52

Figure 2.3: Time is the biggest driver of variation in the dataset ...... 53

Figure 2.4: Effects on transcription are largely distinct between developmental time

points...... 54

Figure 2.5: There are more upregulated than downregulated genes among ethanol-

treated individuals...... 55

Figure 2.6: Trees illustrate the hierarchical organization of the enriched GO terms ...... 56

Figure 2.7: WGCNA identifies two modules that are significantly correlated with ethanol exposure ...... 57

Figure 2.8: Quantitative Real-Time qRT-PCR (qRT-PCR) validates RNA-seq ...... 58

Figure 2.9: Expression of hub genes for development-related modules indicate ethanol did not retard developmental progression ...... 59

Figure 2.10: Ethanol indirectly attenuates Shh signaling ...... 60

Figure 2.11: Cell death by TUNEL at 11 hpf in untreated and ethanol-treated vangl2 mutants ...... 62

Figure 2.12: Ethanol disrupts convergent extension ...... 63

Figure 2.13: Ethanol alters six3 and rx3 expression in the eye field ...... 64

Figure 2.14: gpc4 mutation enhances cyclopia in a dose-dependent manner ...... 65

Figure 3.1: Schematic representation of the vangl2 RNA-seq experimental design ...... 78

Figure 3.2: Summary of sample read counts through the data processing pipeline ...... 79

Figure 3.3: Ethanol treatment correlates to PC1 and PC2 ...... 80

Figure 3.4: Ethanol and temperature illicit a robust transcriptional response ...... 81

xii Figure 3.5: Trees illustrate the hierarchical organization of the enriched GO terms ...... 82

Figure 3.6: The Neuroactive Ligand Receptor Interaction pathway is differentially expressed due to ethanol treatment ...... 83

Figure 3.7: WGCNA identifies modules that are significantly correlated with ethanol exposure, genotype, and temperature ...... 84

Figure 3.8: Differences in ethanol-response based on genotype ...... 86

xiii Chapter 1: General Introduction and Significance

Birth defects, or congenital malformations, are the leading cause of infant mortality in the US (Xu et al., 2018). Although the multifactorial etiologies of birth defects are not well understood, many are thought to derive from a complex interplay between genetic and environmental perturbations. The adverse effects of teratogens, or environmental agents that cause irreversible developmental defects, were first recognized in the 1950s and 1960s (Shepard, 1982). It was then that investigators began to observe developmental anomalies in infants exposed to methylmercury and thalidomide in utero (Harada, 1995; Kim and Scialli, 2011). Since then, other environmental teratogens have come to light, including but not limited to, pollutants, pharmaceuticals, and chemicals. Ethanol is likely the most common teratogen to date. Prenatal alcohol exposure (PAE) is known to cause a spectrum of ailments known as Fetal Alcohol Spectrum Disorders (FASD). Studies in human and animal models strongly suggest a genetic component but identifying and understanding how these genetic components interact with ethanol to cause phenotypic variability has been an ongoing challenge. Ethical considerations preclude experimental testing of ethanol and other teratogens in humans; for this and other reasons, animal models have emerged as a valuable asset to researchers. Common model organisms utilized to study the harmful effects of PAE include rodents (Sulik and Johnston, 1983; Sulik et al., 1981), chick (Berres et al., 2017; Flentke and Smith, 2018; Garic et al., 2014), and zebrafish (Lovely et al., 2016; McCarthy et al., 2013; Swartz et al., 2014). The zebrafish (Danio rerio) has been particularly advantageous in the study of vertebrate development and toxicology (including PAE) due to their rapid development, small size, external fertilization, transparency as embryos, and genetic tractability. For these reasons, my lab uses 1 zebrafish to better understand the early effects of ethanol exposure on the developing embryo. Identifying and characterizing these genetic modifiers are essential for understanding variability in FASD and could aid in earlier detection and improved clinical management and prevention.

1.1. FETAL ALCOHOL SPECTRUM DISORDERS (FASD)

Fetal Alcohol Spectrum Disorders (FASD) describes a highly variable continuum of birth defects that result from PAE, affecting an estimated 2-5% of children born in the US (May et al., 2009). Despite being the most preventable cause of birth defects, a reported 11.5% of pregnant women consume alcohol during pregnancy and 3.9% engage in binge drinking (Denny et al., 2019). FASD is not a clinical diagnosis in and of itself, but rather encompasses a range of diagnostic subgroups, listed in order of severity: Fetal Alcohol Syndrome (FAS), Partial Fetal Alcohol Syndrome (pFAS), Alcohol Related Neurodevelopmental Disorder (ARND) and Neurobehavioral Disorder Associated with Prenatal Alcohol Exposure (ND-PAE) (Hoyme et al., 2016). The teratogenic effects of PAE were first documented approximately 50 years ago (1968), when researchers were struck by the phenotypic similarity among a group of 127 children born to alcoholic mothers (Lemoine et al., 1968). A few years later (1973), Jones and colleagues observed similar phenotypes in 8 unrelated children exposed to alcohol prenatally, drawing a clear association between PAE and developmental abnormalities (Jones et al., 1973). In the same year, Jones and Smith coined the term “Fetal Alcohol Syndrome”, which is characterized by craniofacial anomalies, neurocognitive, and growth deficits (Jones and Smith, 1973). The diagnostic trio of craniofacial anomalies required for a clinical diagnosis of FAS include a thin upper lip, smooth philtrum, and short palpebral fissures (Hoyme et al., 2005). These craniofacial defects are commonly

2 associated with midfacial hypoplasia, flattening of the nasal bridge, malar region, and reduced midfacial depth (Moore et al., 2002; Muggli et al., 2017; Sulik et al., 1981). Development of the face and brain are highly interconnected and many craniofacial defects associate with neurocognitive or behavioral problems. Numerous studies have examined how the forebrain provides both structural support and molecular instructions to influence morphogenesis of the face. For instance, incomplete cleavage of the forebrain hemispheres, a congenital malformation known as holoprosencephaly, subsequently causes loss of midfacial tissues and results in craniofacial defects like cleft lip and palate or cyclopia (Sulik et al., 1981). Additionally, Sonic hedgehog (Shh) signaling in the forebrain patterns the ventral neural tube and craniofacial skeleton (discussed further below) (Hu and Marcucio, 2009; Johnson and Rasmussen, 2010). However, the difficulty in diagnosing FASD lies in the fact that the converse is not always true; children born with pFAS, ARND, or ND-PAE, may present with neurocognitive or behavioral problems without salient craniofacial abnormalities. However, not all children exposed to alcohol in utero develop a form of FASD, suggesting that there are factors that influence the susceptibility to ethanol teratogenesis. Two well-known factors that influence outcomes of PAE include dosage and timing of ethanol exposure (Sulik et al., 1981). Work in animal models suggest gastrulation, a period in embryogenesis that results in the formation of the three germ layers and clearly defined body axes, is particularly sensitive to the adverse effects of ethanol (Blader and Strahle, 1998; Sulik et al., 1981). Acute ethanol exposure during gastrulation in a mouse model phenocopies defects seen in human FAS (Sulik et al., 1981). In humans, gastrulation occurs during the third week of pregnancy, 15 to 28 days post fertilization; a time when most women do not realize they are pregnant (Webster et al., 1988). 3 Growing evidence supports the importance of genetics in FASD. In human twin studies, monozygotic twins, who inherit an identical set of genes and develop simultaneously in the womb, were found to be 100% concordant for FAS, whereas dizygotic twins, who share roughly half their genes, were only 63% concordant (Eberhart and Parnell, 2016; Streissguth and Dehaene, 1993). Thus, genetic risk and resiliency are important in the genesis of FASD.

1.1.1. Holoprosencephaly

Midfacial hypoplasia and other midline defects associated with FAS are believed to be part of a wider spectrum of holoprosencephaly (O'Leary-Moore et al., 2011; Sulik and Johnston, 1983). Holoprosencephaly, which translates to “single cavity forebrain,” is a congenital defect of the forebrain (prosencephalon) and midface that results when the forebrain fails to partition into two distinct hemispheres (Johnson and Rasmussen, 2010; Johnston and Bronsky, 2002). This abnormality is relatively common in conception, accounting for 1 in every 250 pregnancies, but as these defects are typically lethal, only about 1 in 10,000 come to term (Roessler et al., 2018). The holoprosencephaly spectrum is divided into three categories, from complete to incomplete failure to partition the forebrain: alobar, semilobar, and lobar (Krauss, 2007). The related facial abnormalities are variable, ranging from apparently unaffected, hypotelorism, cleft lip, and/or a central maxillary incisor, to cyclopia (Cohen and Sulik, 1992). There are a few known genetic and environmental risk factors that contribute to the etiology of the holoprosencephaly spectrum. Genetic risk factors for humans include mutations in SHH, SIX3, ZIC2, and FGFR1, with the most common being in SHH (Roessler et al., 2018). Heterozygous mutations in members of the hedgehog signaling pathway (e.g. PTCH1, CDON, GAS1) are also implicated in the pathogenesis of HPE

4 (Hong and Krauss, 2012). A non-genetic risk factor based on epidemiological studies in humans, point to acute maternal alcohol use during the first trimester (Bönnemann and Meinecke, 1990; Ronen and Andrews, 1991).

1.2 GENE-ETHANOL INTERACTIONS

Studies in animal models support the idea that genetic risk factors underlie susceptibility for FASD. In a study of different inbred mouse strains, the C57BL/6J strain were most sensitive to the adverse effects of ethanol, while other strains were resistant to malformations with the same acute dose of ethanol (Downing et al., 2009). When environmental variables such as dose, timing, and concentration of ethanol, are tightly controlled, genetic risk factors modulate susceptibility for craniofacial, neurodevelopmental, and cardiovascular defects (Smith et al., 2014). Maternal and fetal genetics are dually implicated in risk for FASD. Alcohol is metabolized into a toxic byproduct, acetaldehyde, by the enzyme, alcohol dehydrogenase (ADH), and to a lesser extent, cytochrome P450 (CYP2E1) and catalase (Eberhart and Parnell, 2016; McCarthy and Eberhart, 2014). Acetaldehyde is further metabolized to acetate via aldehyde dehydrogenase 2 (ALDH2) (Eberhart and Parnell, 2016; McCarthy and Eberhart, 2014). This process leads to teratogenesis in part by the generation of damaging reactive oxygen species (ROS), which alter gene expression and cause oxidative stress and DNA damage (Bhatia et al., 2019). Polymorphisms in alcohol metabolizing enzymes can increase or reduce risk for FASD. For instance, the ADH1B*2 allele is associated with increased efficiency in alcohol metabolism, imparting protection against the teratogenic effects of alcohol exposure (Green and Stoler, 2007; Viljoen et al., 2001). In murine model studies, zygotic Aldh2 also confers protection against ethanol teratogenesis (McCarthy and Eberhart, 2014). Ethanol metabolizing enzymes are clear

5 candidates for gene-ethanol interactions, however mutations in other genes also modulate susceptibility to FASD. Identifying these genetic risk factors is challengeing, due to the nature of ethanol. Ethanol is a small molecule with pleiotropic effects. Although its mechanism of action is incompletely understood, studies have revealed ethanol-binding pockets in various receptors and proteins (Harris et al., 2008). Its broad interactions lead to alterations in various developmental signaling pathways. Although we have an incomplete understanding of these genetic factors, genetic screens using zebrafish have revealed several susceptibility loci including, pdgfra, vangl2, mars, hinfp, plk1, and foxd1 (McCarthy et al., 2013; Swartz et al., 2014).

I.2.1. The vangl2-ethanol interaction

In a screen to identify genetic modifiers that influence sensitivity to ethanol teratogenesis, my lab identified a strong interaction between vangl2 and ethanol (Swartz et al., 2014). Twenty mutant lines from the Zebrafish International Resource Center were screened for morphological defects using a subteratogenic dose of ethanol. Embryos were exposed to ethanol at the initiation of gastrulation (6 hours post fertilization, hpf) until fixation at 4 days post fertilization (dpf). Embryos were subsequently stained with Alcian Blue and Alizarin Red to assess defects to the cartilage and bone. A mutation in vangl2 synergistically enhanced the teratogenic effects of ethanol on craniofacial morphogenesis and axonal migration. Vangl2 is a member of the non-canonical/Wnt planar cell polarity (Wnt/PCP) pathway that mediates convergent extension movements of the mesoderm and neuroectoderm. In control conditions, homozygous vangl2 mutants displayed a shortened-broadened body axis due to a reduction in convergent extension, but their

6 craniofacial skeleton is largely intact. Ethanol-exposed mutants were invariably synophthalmic and exhibited defects in midfacial cartilage elements (Swartz et al., 2014). One heterozygote phenocopied an ethanol-sensitive mutant, providing further evidence for a strong genetic interaction.

1.2.2. The hedgehog signaling pathway

The signaling molecule and morphogen, sonic hedgehog (SHH), plays a critical role in various developmental processes affected by ethanol exposure (Eberhart and Parnell, 2016). In vertebrates, the co-receptors, CDON and brother of CDO (BOC), facilitate the binding of SHH to the transmembrane receptor, Patched (PTCH) (Ingham and Placzek, 2006). In conditions where SHH is absent, PTCH inhibits the seven-pass- transmembrane protein, Smoothened (SMO), whose activity is necessary for pathway signal transduction. Activation of SMO allows for nuclear transport and transcription of the Gli family of transcription factors. GLI2 and GLI3 can both activate and repress transcription of Shh target genes through distinct domains (Ingham and Placzek, 2006). Their repressive activities are abolished upon binding of SHH to PTCH. GLI1 supports GLI2 in activating transcription. Several members of Shh signaling are themselves target genes (i.e. GLI1, SMO, and PTCH2) and thus function in a positive-feedback-loop. Shh in the prechordal plate mesoderm is necessary for proper midfacial development. It induces expression of Shh target genes (including SHH itself) in both the ventral diencephalon and telencephalon, subdivisions of the forebrain (Hong and Krauss, 2013; Marcucio et al., 2005). In addition to providing structural support for the developing midface, the forebrain also serves as an important Shh signaling center. In chick, disruption of Shh in the forebrain neuroectoderm causes midfacial hypoplasia (Marcucio et al., 2005). In mice, loss of Shh causes severe midfacial defects and cyclopia

7 (Muenke and Beachy, 2000). Since both suppression of Shh signaling and high concentrations of ethanol predispose an embryo to midline patterning defects, researchers hypothesized that ethanol disrupts Shh signaling (Eberhart and Parnell, 2016; McCarthy and Eberhart, 2014). The first study providing evidence for a genetic interaction between ethanol and the Shh pathway was published only 7 years ago (Hong and Krauss, 2012). The Shh co- receptor CDON positively regulates Shh signaling by binding directly to SHH and PTCH1. Mice deficient in Cdon (Cdon-/-) and ethanol-exposed 129S6 mice did not exhibit defects in midline patterning alone, but ethanol-exposed Cdon-/- presented with holoprosencephaly-related defects (Hong and Krauss, 2012). Furthermore, additional loss of one copy of the negative regulator, Ptch1, significantly improved outcomes (Hong and Krauss, 2013; Kietzman et al., 2014). These studies collectively demonstrate the importance of Shh signaling in protection against the teratogenic effects of ethanol exposure.

1.3 GASTRULATION

1.3.1. Germ layer formation in humans

Gastrulation in humans converts the bilaminar blastocyst into a trilaminar embryo with three primary germ layers (Webster et al., 1988). Prior to gastrulation, the bilaminar disc consists of the epiblast, which contributes to embryonic tissues, and the hypoblast, which contributes to extraembryonic tissues (Johnston and Bronsky, 2002; Webster et al., 1988). At the onset of gastrulation, a transient thickening of cells known as the primitive streak, appears in the epiblast, establishing the anterior-posterior axis (Johnston and Bronsky, 2002). From here, cells undergo a series of morphogenetic movements including invagination and ingression to create the underlying endoderm and mesoderm. 8 The chordamesoderm subsequently forms in the mesodermal layer and gives rise to the notochord, a defining characteristic of all chordates (Johnston and Bronsky, 2002). The notochord provides both structural axial support and serves as an important signaling center for the overlying ectoderm. The prechordal mesendoderm (PCM), the precursor to the prechordal plate, lies anterior to the chordamesoderm and ventral to the neural plate (Johnston and Bronsky, 2002). The PCM and chordamesoderm together are responsible for patterning the neural plate, a precursor to the neural tube.

1.3.2. Germ layer formation in zebrafish

The process of germ layer formation during gastrulation is highly conserved across vertebrates. In zebrafish, cleavage precedes gastrulation, generating the blastoderm, a spherical ball of cells that sit atop the yolk. The yolk provides lipids, proteins, and nutrients to the embryo, until it develops into a free-feeding larva. Gastrulation commences when the blastoderm thins and spreads over half the yolk (50% epiboly) towards the vegetal pole (Warga and Kimmel, 1990). At this stage, the germ ring forms at the perimeter of the blastula. Blastoderm cells involute at the blastoderm margin, equivalent to the primitive streak in humans, and converge towards a thickening of cells known as the embryonic shield (6 hours post fertilization, hpf) (Solnica-Krezel and Sepich, 2012). Cells that involute at shield stage migrate beneath the epiblast

(ectoderm) and move towards the animal pole to form the axial hypoblast (mesoderm and endoderm) (Warga and Kimmel, 1990). During gastrulation, cells also undergo convergence and extension (discussed in detail later) to narrow and elongate the body axis, respectively. Gastrulation concludes at 100% epiboly or bud stage (10 hpf), when gastrula cells have completely enveloped the yolk. The embryo is now clearly defined with a head and tailbud.

9 The PCM in zebrafish differentiates into head and eye muscles and induces anterior neural tissue (Kiecker and Niehrs, 2001). Accordingly, ablation of the prechordal plate results in cyclopic embryos with a single eye field (Kiecker and Niehrs, 2001). This phenotype is also seen in embryos deficient in Shh signaling and some Wnt/PCP pathway mutants. The axial mesendoderm patterns the ventral neuroectoderm and shield ablation results in a reduction of shh in the floorplate and ventral brain and causes cyclopia (Saúde et al., 2000).

1.4 EYE FIELD FORMATION

The formation of the vertebrate eye is a complicated process involving various morphogenetic movements that are intimately tied to the development of the midline (Martinez-Morales and Wittbrodt, 2009). Hypotheses regarding eye field formation were proposed as far back as the early 19th century in an effort to understand the developmental origin of cyclopia. Speer (1819) and Meckel (1826) argued cyclopia resulted from fusion of two bilateral domains, whereas Von Baer (1828) proposed cyclopia resulted from failure to split a single eye field into two (Adelmann, 1929). Fate mapping studies later confirmed that early in embryogenesis, the bilateral eyes arise from a single mass of cells in the anterior neural plate (Graw, 2010). The eye field in the anterior neural plate is flanked anteriorly and posteriorly by the prospective telencephalon and diencephalon, respectively (Chuang and Raymond, 2001). It is now understood that signals that drive separation of the eye field derive from the prechordal plate underlying the neuroectoderm or anterior neural plate (Bailey et al., 2004). Xenopus explant studies demonstrate that the anterior neural plate and underlying prechordal mesoderm are necessary to form two retinas (Li et al., 1997). Furthermore, transplanted prechordal plate is sufficient to rescue cyclopia in a chick model, while other mesodermal

10 tissues are not (Li et al., 1997). In zebrafish, diencephalic precursor cells from the ventral midline additionally function in separation of the eye field (Varga et al., 1999).

1.4.1 Optic vesicle morphogenesis

During neurulation, the optic vesicles appear as evaginations of the forebrain (i.e. diencephalon), in the most anterior region of the neural plate (Chuang and Raymond, 2001). Two bilateral eyes develop from the single eye field by way of optic vesicle evagination. In zebrafish, this process begins at the end of gastrulation (~10–11 hpf) and two optic primordia are distinguishable by 14 hpf (Santos-Ledo et al., 2013). These bilateral optic primordia further differentiate into the lens, retina, optic nerve, and related epithelia, to form the complex eye (Chuang and Raymond, 2001).

1.4.2 Transcription factors specify the eye field

The prospective eye field in the neuroectoderm of the anterior neural plate, expresses a set of conserved homeobox-containing transcription factors that are critical for eye field specification (Chuang and Raymond, 2001). These transcription factors (i.e. rx3, six3, pax2, otx2, and zic2) are first activated during gastrula stages (Bailey et al., 2004; Santos-Ledo et al., 2013). The importance of the specification of the eye field is evidenced by rx null zebrafish, who lack eyes altogether (anophthalmia) (Mathers et al., 1997). Conversely, gain-of-function experiments in zebrafish demonstrate that Rx is sufficient to induce retinal pigmented epithelium (RPE) (Chuang and Raymond, 2001). In the most severe cases, these embryos possessed an ectopic third medial eye (Chuang and Raymond, 2001). Zebrafish have 3 rx paralogs (rx1/2/3) that are all activated in the anterior neural plate but later show preferential expression in either the presumptive retina (rx1/2) or anterior forebrain (rx3) (Mathers et al., 1997). Lastly, relevant to the

11 studies herein, SIX3, a transcription factor critical in optic field specification is also among the genes that when disrupted are most commonly implicated in HPE.

1.4.3 The Wnt signaling pathway in forebrain patterning

The Wnt family of signaling pathways (discussed in detail below) plays a critical role in forebrain patterning (Chuang and Raymond, 2001). The canonical pathway promotes posterior neural fates (i.e. posterior neural ectoderm) while inhibiting anterior neural fates (i.e. the forebrain and axial mesoderm) (Park and Moon, 2002). Vangl2, which functions in the non-canonical pathway, promotes anterior neural fates through inhibition of the canonical pathway (Park and Moon, 2002).

1.5 THE WNT SIGNALING PATHWAY

Wnt proteins are evolutionarily conserved signaling molecules that control various embryonic developmental processes (Logan and Nusse, 2004). These ligands modulate their various functions by activating either the canonical or non-canonical signaling pathways. The canonical pathway has been extensively studied and is known to mediate its effects, such as cell fate specification and determination, via nuclear localization of b-catenin and subsequent transcriptional activation of its target genes (Clevers, 2006). The noncanonical Wnt/PCP pathway is unique in that it is b-catenin independent, utilizes different cellular components, and activates a distinct intracellular signaling response that mediates cell polarity and migration (Sokol, 2015; van Amerongen et al., 2012). Unlike the canonical pathway, molecular and genetic interactions among its various components are not well understood. Studies in vertebrates suggest that there is cross-talk between the two pathways, whereby the non-canonical

12 antagonizes the canonical (Li et al., 2011; Mentink et al., 2018; Topczewski et al., 2001; Westfall et al., 2003).

1.5.1 The non-canonical Wnt/PCP pathway

“Planar polarity” was first coined in 1987 to describe the coordinated orientation of cellular behaviors and structures in the plane of the epithelium, orthogonal to the apical-basal axis (Gray et al., 2011; Nubler-Jung, 1987). Studies on this phenomenon began in 1975, when investigators recognized the importance of epidermal polarity for coordinated orientation of ommatidia in the compound eye in the large milkweed bug, Oncopeltus fasciatus (Lawrence and Shelton, 1975; Seifert and Mlodzik, 2007). The establishment of cell polarity was later extensively characterized in the ommatidia, cuticle, hairs, and sensory bristles of Drosophila melanogaster (Gubb and Garcia- Bellido, 1982; Nubler-Jung et al., 1987). Subsequent genetic analyses revealed genetic components whose mutation disrupt polarity of these structures (Adler, 1992; Gubb, 1993). Later, studies in vertebrates elucidated a critical role of PCP in regulating processes such as convergent extension movements during gastrulation, patterning of sensory hair cells in the inner ear, polarization of ciliary structures, and neural tube closure (Curtin et al., 2003; Heisenberg et al., 2000; Tada and Smith, 2000; Wallingford and Harland, 2001; Wallingford et al., 2000). Interestingly, these effects were not only in epithelial cells but also mesenchymal cells, which are a migratory cell type derived from the mesoderm and ectoderm that lack apical-basal polarity and give rise to a variety of morphological structures (Seifert and Mlodzik, 2007). In vertebrates, the six evolutionary conserved “core PCP components” consist of a seven-pass-transmembrane receptor, Frizzled; a four-pass-transmembrane protein, Vangl2; a seven-pass transmembrane protein, Celsr; and the cytoplasmic proteins

13 Dishevelled, Prickle, and Inversin (Devenport, 2014; Gray et al., 2011). Asymmetric localization of these core components is crucial for proper PCP signaling and mutation in any one component disrupts this asymmetry (Curtin et al., 2003; Devenport, 2014; Gray et al., 2011). Two homologs (Wnt5a and Wnt11) of the Wnt family of lipoglycoproteins are known to function in the non-canonical pathway, but it is unclear whether these ligands play an instructive or merely permissive role for PCP signaling (Heisenberg et al., 2000; Klein and Mlodzik, 2005; Tada and Smith, 2000). The binding of Wnt to the Frizzled receptor, recruits Dishevelled to the cell membrane, whereas Vangl promotes Prickle (Gray et al., 2011; Klein and Mlodzik, 2005). Planar polarity is established by asymmetric localization of these complexes to opposing sides of the cell and studies in Drosophila suggest an antagonistic relationship between these complexes (Klein and Mlodzik, 2005). Celsr is not asymmetrically localized and is believed to interact with both complexes to propagate the Wnt/PCP signal between cells (Gray et al., 2011). The heparan sulfate proteoglycan, Glypican4, serves as a co-factor and positive modulator of Wnt/PCP signaling (Topczewski et al., 2001). This signal is further transduced intracellularly through Dishevelled and the Rho small GTPases, Rho, Rac, and Cdc42, that enable actin cytoskeletal rearrangements (Schlessinger et al., 2009). Studies suggest the Wnt/PCP pathway does not act as a simple linear pathway as mechanical forces and epigenetic modulators are also known to affect pathway function (Ossipova et al., 2015). Collectively, a delicate balance of Wnt/PCP components is necessary, and gain-of-function mutants often phenocopy loss-of-function mutants.

14 1.5.2 Convergent extension

Wnt/PCP signaling is best known for its role in mediating convergent extension movements during gastrulation. Convergent extension results in the elongation and narrowing of the anterior-posterior and mediolateral axes, respectively. Mesenchymal cells form polarized projections (i.e. lamellipodia and filopodia) which mediate this type of cellular rearrangement (Seifert and Mlodzik, 2007). Work in xenopus and zebrafish has revealed critical roles for wnt11, wnt5, vangl2, and gpc4, in mediating convergent extension movements. Disruptions to these genetic components inhibits convergent extension and results in a widened body axis that fails to elongate (Heisenberg et al., 2000; Rauch et al., 1997; Tada and Smith, 2000). In zebrafish, at the start of convergent extension, lateral gastrula cells migrate and converge to the dorsal region of the developing embryo, where they intercalate between neighboring cells to drive extension of the body axis (Topczewski et al., 2001). As these cells begin their dorsal migration, they elongate along their mediolateral axis and polarize their actin-based cytoskeletal processes medially and laterally to drive intercalation (Glickman et al., 2003). Convergent extension movements during gastrulation also alter the shape of the eye field, whereby diencephalic precursor cells articulate both medially and caudally (Chuang and Raymond, 2001).

15 Chapter 2: Convergent extension defects underlie susceptibility to midfacial hypoplasia in an ethanol-sensitive mutant, vangl2.

2.1. ABSTRACT

Fetal Alcohol Spectrum Disorders (FASD) describes the full range of defects that result from prenatal alcohol exposure. The clinical diagnosis of Fetal Alcohol Syndrome (FAS) is the most severe outcome, characterized by facial dysmorphism and neurological deficits. Gene-ethanol interactions underlie susceptibility to FASD but we lack a clear understanding of the genetic risk factors. Here, I leverage the genetic tractability of zebrafish to address this problem. I first show that vangl2, a member of the Wnt/planar cell polarity (Wnt/PCP) pathway that mediates convergent extension movements, strongly interacts with ethanol during late blastula and early gastrula stages. Embryos mutant or heterozygous for vangl2 are sensitized to ethanol-induced midfacial hypoplasia. However, we have no understanding of the potential mechanism of this sensitization. I performed single-embryo RNA-Seq during early embryonic stages, to assess individual variation to the transcriptional response of ethanol and determine the mechanism of the vangl2-ethanol interaction. Transcriptional changes due to ethanol are indicative of increased oxidative stress and ion transport as well as reduced DNA replication and cell division. To identify the pathway(s) that are disrupted by ethanol I used these global changes in gene expression to identify molecules, often pathway inhibitors, that mimic the effects of ethanol via the LINCS L1000 dataset. This dataset predicted that the Sonic Hedgehog (Shh) pathway inhibitor, cyclopamine, would mimic the effects of ethanol, despite the fact that ethanol does not alter the expression levels of direct targets of Shh signaling. Indeed, I find that ethanol and cyclopamine strongly interact to disrupt midfacial development. Collectively, these results suggest that the

16 midfacial defects in ethanol-exposed vangl2 mutants is due to an indirect interaction between ethanol and the Shh pathway. Vangl2 functions as part of a signaling pathway that regulates coordinated cell movements during midfacial development. Consistent with an indirect model, I find that a critical source of Shh signaling that separates the developing eye field into bilateral eyes, allowing the expansion of the midface, becomes mispositioned in ethanol-exposed vangl2 mutants.

2.2. INTRODUCTION

Prenatal alcohol exposure (PAE) is the most preventable cause of birth defects, yet the prevalence of fetal alcohol spectrum disorders (FASD) in the US is as high as 2-

5% (May et al., 2009). Fetal alcohol syndrome (FAS) is the most severe outcome and is characterized by midfacial hypoplasia as well as growth and neural deficits (Jones and Smith, 1973). While PAE is required for the development of FASD, the teratogenic effects of ethanol are modulated by genetics (McCarthy et al., 2016). Genetics play a significant role as evidenced by human twin studies. Monozygotic twins were 100% concordant for FAS, whereas dizygotic twins were only 63% concordant (Eberhart and Parnell, 2016; Streissguth and Dehaene, 1993). Furthermore, animal models of FAS show strain-specific differences after controlling for environmental variables such as dose and timing (Downing et al., 2009; Loucks and Carvan, 2004). Despite this, the genetic factors that protect or predispose an individual to FASD are poorly understood. Moreover, we still lack a basic understanding of the mechanism of ethanol teratogenesis. Zebrafish are well suited to address this problem due to their genetic tractability and ease of embryological manipulation (Lovely et al., 2016). In a screen to identify genetic modifiers of ethanol teratogenicity, vangl2, a member of the non-canonical Wnt/PCP pathway, emerged as an ethanol sensitive locus

17 (Swartz et al., 2014). In zebrafish, mutations in the vangl2 locus disrupt convergent extension movements during gastrulation, resulting in a shortened, broadened, body axis (Solnica-Krezel et al., 1996). These mutants infrequently present with synopthalmia, incomplete fusion of the eyes, but the phenotypic expressivity of this trait varies with factors such as temperature and genetic background (Marlow et al., 1998). In the screen, all untreated vangl2 mutants displayed proper separation of the eyes and craniofacial skeletal elements were intact (Swartz et al., 2014). Upon exposure to a subteratogenic dose of ethanol, vangl2 mutants were fully penetrant for synopthalmia and displayed profound defects to the midfacial skeleton (Swartz et al., 2014). Ethanol-treated vangl2 heterozygotes were largely indistinguishable from their wild-type siblings, with the exception of a single synopthalmic ethanol-treated individual, providing evidence for latent haploinsufficiency. These data suggest a synergistic interaction between vangl2 and ethanol. We know vangl2 plays a critical role in mediating convergent extension movements as evidenced by their body axis defect, however the early effects of ethanol exposure have not been investigated in zebrafish. To better understand how ethanol interacts with vangl2 to alter phenotypic outcomes, I took an unbiased approach to assess the transcriptional response to ethanol. I performed single embryo RNA-sequencing (RNA-seq) on control and ethanol-treated wild-type embryos across gastrulation and early segmentation stages. Bioinformatic and functional analyses indicate that midfacial defects in ethanol- exposed vangl2 mutants is due to an indirect interaction between ethanol and the Shh pathway. While there is no alteration in the level of expression of direct Shh targets, a critical source of Shh signaling that separates the developing eye field into bilateral eyes becomes mispositioned in ethanol-exposed vangl2 mutants. 18 2.3. RESULTS

2.3.1. vangl2 mutants and heterozygotes are sensitive to 1% ethanol during early embryogenesis

The vangl2m209 allele (formerly known as trim209) was identified as a splice site mutation containing an intronic insertion that results in a frameshift, terminating in a premature stop codon (Jessen et al., 2002). Using PCR analysis of genomic DNA from fin clips, I identified the splice acceptor mutation that alters the invariant 3’ consensus sequence, AG, of intron 7, to TG. Zebrafish homozygous for vangl2m209 are phenotypically indistinguishable from vangl2 null alleles, suggesting vangl2m209 to be an amorph (Jessen et al., 2002; Marlow et al., 1998). To determine the critical time window of sensitivity in the vangl2 mutants, I initiated ethanol treatment at various stages comprising late blastula to early gastrula for 24 hours, using inner lens-to-lens width as a measure of cyclopia. In control conditions, homozygotes exhibited midline defects ranging in severity from synopthalmia to cyclopia, complete fusion of the eyes, across all time points but were fully penetrant for cyclopia (100% fused; n=5/5) when ethanol was applied at shield stage (6 hpf) (Figure 2.1A, B). Heterozygotes showed increased sensitivities when ethanol was applied at high stage (3.3 hpf) (22% fused; n=4/18), a time when treating wild-type embryos with higher concentrations of ethanol causes similar defects (Figure 2.1B) (Blader and Strahle, 1998). Thus, heterozygotes and homozygotes are equally sensitive at early stages of development but subsequently diverge in their sensitivity. This may be due to a compensatory genetic mechanism in vangl2 heterozygotes because zygotic gene expression initiates after high stage, 4 hpf in zebrafish.

19 2.3.2. The effect of ethanol on the early zebrafish transcriptome is subtle relative to developmental time

To determine if ethanol caused transcriptional changes that could underlie the interaction with vangl2, I designed two RNA-Seq experiments that largely overlapped in design. Embryos were exposed to a subteratogenic dose of 1% ethanol in embryo media (171 mM), which equilibrates to approximately 50 mM tissue concentration (Lovely et al., 2014). For the first experiment, embryos were treated at 6 hours post fertilization (hpf) (the onset of gastrulation) and collected at mid-gastrulation (8 hpf) and the end of gastrulation (10 hpf) (Figure 2A). Each sample consisted of an individual zebrafish embryo with five replicates per timepoint and treatment. A second experiment was performed to increase power. Embryos were similarly exposed with the addition of an eight-hour exposure window (6-14 hpf), when the eye fields have completely separated. I also omitted the 6 hpf control samples (Figure 2B) since they lacked ethanol-treated samples. The data from both experiments were combined for subsequent analyses, controlling for batch effects. To assess the effect of time, batch (i.e. experiment 1 and 2), and ethanol, on the early zebrafish transcriptome, I performed principal component analysis (PCA). Individuals from each timepoint clustered tightly along PC1, which explained 38% of the variation in the dataset. This strong effect of time on the zebrafish transcriptome is in agreement with previous studies (Farrell et al., 2018). Clustering of samples by time regardless of ethanol treatment suggested that control and ethanol-treated samples were accurately staged and that ethanol did not appear to delay developmentally regulated transcriptome patterns. There was greater discrimination of 14 hpf samples relative to earlier timepoints (6, 8, and 10 hpf), which is consistent with greater distinction of this

20 timepoint in terms of developmental time and morphology (Figure 2.2C). PC2 largely captured batch effects between experiment 1 and 2 (Figure 2.2D). The majority of variation between samples did not appear to be due to treatment, with control and ethanol-treated samples randomly interspersed along PC1 and PC2 (Figure 2.2E). Separation by treatment was observed along PC8 and PC9, which accounts for 2% of the variation in the data (Fig. 2.2F). Hierarchical clustering of samples based on correlation, further corroborated this finding. The 6 and 14 hpf samples showed the greatest dissimilarity, whereas the 8 and 10 hpf samples showed the greatest similarity, irrespective of treatment (Figure 2.3). In summary, the transcriptional effect of ethanol on the early zebrafish transcriptome is subtle, while time provides the strongest transcriptional fingerprint.

2.3.3. Ethanol has effects on transcription that are largely distinct between different developmental timepoints

Although developmental age was a stronger source of transcriptional variation, I still detected substantial variation in transcription in response to ethanol. There were 1,414 differentially expressed genes (DEGs), with a False-discovery rate (FDR) less than 0.1 (Figure 2.4A; Benjamini–Hochberg procedure). There were more upregulated than downregulated DEGs among ethanol-treated individuals across timepoints (Figure 2.4A, 2.5A) and these DEGs were distinct between developmental timepoints (Figure 2.4B). The 14 hpf samples were most distinct (Figure 2.5D), with considerably lower variation due to ethanol treatment and fewer DEGs compared to the 8 and 10 hpf samples (Fig. 2.5B,C). One explanation for the decreased effect of ethanol is that in contrast to the other timepoints, the 14 hpf samples were exclusively in experiment 2, with only half the number of samples. An alternative explanation is that cells become more restricted in

21 their potency through developmental time and thus more dissimilar in their transcriptional profiles, allowing for little overlap. In summary, while some genes are generally affected by ethanol early in development, ethanol elicits largely distinct responses at different developmental stages.

2.3.4. Ethanol does not affect the Wnt/PCP pathway at the transcriptional level

Transcription of Wnt/PCP pathway members was largely unaffected by ethanol exposure. No Wnt/PCP pathway members were among the 15 shared DEGs across all timepoints (Figure 2.4C). KEGG pathway enrichment analysis further confirmed that ethanol has little effect on the Wnt/PCP pathway at the level of transcription (Figure

2.4D). Only two Wnt/PCP pathway members were moderately affected: ethanol exposure moderately decreased expression of the cofactor, glypican4 (gpc4), (log2 fold=-0.237; p- value=0.036) across all timepoints (Marlow et al., 1998; Topczewski et al., 2001) and upregulated expression of rac3a (log2 fold=0.737; p-value=8.89E-06), a member of the Rho family of small GTPases (White et al., 2018).

2.3.5. Transcriptional changes due to ethanol are indicative of increased oxidative stress and ion transport and reduced DNA replication and cell division

To summarize the functions of ethanol-responsive genes, I tested for enrichment of Gene Ontology (GO) terms for up- and downregulation due to ethanol treatment. This was done separately for GO terms describing molecular function (Figure 2.6), biological processes (Figure 2.6B), and cellular components (Figure 2.6C). Transcriptional changes due to ethanol are indicative of increased oxidative stress and reduced DNA replication and cell division. Highly enriched GO terms indicative of oxidative stress include oxidoreductase, oxidation-reduction process, ion transport, and endoplasmic reticulum (Figure 2.6A-C). GO terms indicative of cell division and DNA replication include 22 helicase, ATP-dependent helicase, ATP-dependent DNA helicase, chromatin modification, chromosome organization, DNA packaging complex, and nucleolus (Figure 2.6A-C). The metabolism of alcohol to acetate involves the production of damaging reactive oxygen species, which contribute to increased oxidative stress (G. I. Henderson, 1995). My GO analysis results support that ethanol exposure generates oxidative stress, downregulating genes involved in morphogenesis.

2.3.6. Modules of co-regulated genes related to ethanol exposure

To determine if ethanol disrupted networks of genes that could explain the vangl2-ethanol interaction, I next performed Weighted Gene Co-expression Network

Analysis (WGCNA) (Langfelder and Horvath, 2008). This unsupervised network analysis identifies groups of genes, termed modules, based on correlated expression patterns across the samples. Modules are summarized by the first principal component for the expression estimates of the included genes, termed the module eigengene, which can be correlated with sample traits to identify biological significance. The cluster dendrogram generated in this analysis illustrates the presence of highly distinct and clustered modules (Figure 2.7A). Merging of similar modules produced eleven total modules (Figure 2.7B). The module eigengenes were correlated with treatment, time, and sequencing job (i.e. batch). As a negative control, I also included a randomized set of ethanol treatment factors. Consistent with my PCA results, nearly all modules showed a strong correlation with time, but two modules (mediumpurple4 and darkolivegreen4) also significantly correlated with ethanol treatment (Figure 2.7B-E). The purple and green modules positively and negatively correlated with ethanol treatment, respectively (Figure 2.7E). A significant differentially expressed gene from the purple (Figure 2.8A) and green module (Figure 2.8B) were selected for independent validation on independent

23 biological replicate samples derived from the same wild-type line using quantitative real- time qRT-PCR (qRT-PCR). These results indicate that my RNA-seq faithfully represents transcript levels. The purple module revealed GO enrichment for three transmembrane transporters (i.e. solute carrier family 16a) (Figure 2.8F). The green module revealed GO enrichment for genes encoding zinc finger (ZnF) proteins (Figure 2.8F). Previous work has shown that ethanol delays development in a dose-dependent manner at concentrations equal to or greater than 1.5% ethanol (Loucks and Ahlgren, 2009). For my experiments, ethanol-treated samples were morphologically stage-matched to control samples to exclude differences due to developmental age. To confirm that the ethanol samples were indeed time-matched to the control samples at the transcriptional level, I compared expression patterns of developmentally regulated genes between the ethanol and control samples from each timepoint. For developmentally regulated genes, I used the gene with the highest module membership (i.e. the hub gene) from each of the WGCNA modules that was associated with time (p<0.05) (Figure 2.9). Consistent with the results from the PCA (Figure 2.2C), expression levels for each gene were similar in control and ethanol-treated samples across time. Despite increased expression of slc16a9a in the mediumpurple4 module in ethanol-treated samples, directionality remained consistent with control samples (Figure 2.9A). Similarly, I observed a modest decrease in expression of the hub gene (ENSDARG00000101103) for the darkolivegreen4 module in ethanol-treated samples (Figure 2.9F). Altogether, these data indicate samples were accurately stage-matched based on morphology.

2.3.7. Cyclopamine is predicted to mimic the effects of ethanol

One challenge in RNA-seq analysis is inferring the mechanism underlying a diseased or environmentally-perturbed state from a set of differentially expressed genes.

24 Individual functional analyses of significant gene-ethanol interactions are time consuming and inefficient. To circumvent this problem, I adopted a bioinformatics approach, utilizing the Library of Integrated Network-Based Cellular Signatures (LINCS L1000) (Subramanian et al., 2017). This publicly available dataset is an extension of the Connectivity Map (CMap), a resource of transcriptomic microarray data of four human cancer cell lines perturbed by 164 chemicals (Lamb et al., 2006). CMap was created as a tool to uncover associations between disease states and drugs that might go unnoticed from traditional high-throughput sequencing analyses. The NIH-funded LINCS project greatly expanded on CMap by the hybridization-based L1000 assay, which directly measured changes to gene expression for 978 landmark genes by various chemicals to infer changes for an additional 11,350 genes (Subramanian et al., 2017). I queried the top 100 and 150 up- and down-regulated genes induced by ethanol exposure against the LINCS L1000 dataset using the clue.io platform (https://clue.io) (Figure 2.10A). The query generated a list of chemicals predicted to have a positive or negative correlation to the input signature. I was particularly interested in those chemicals with positive correlation to ethanol as they would give insight into the mechanism of ethanol teratogenicity. Interestingly, cyclopamine, a Shh pathway inhibitor, was predicted to positively correlate with the ethanol signature. The Shh pathway is critical for midfacial development and mice deficient in Sonic Hedgehog (Shh) have holoprosencephaly and a single medial eye (Chiang et al., 1996). Furthermore, reduction of shh or loss of smo in zebrafish, results in severe loss of craniofacial midline structures (i.e. the anterior neurocranium) (Eberhart et al., 2006). Ethanol is another environmental risk factor for holoprosencephaly, resulting in a characteristic set of midfacial defects, a hypomorphic forebrain, and cyclopia (Blader and Strahle, 1998; Cohen and Sulik, 1992; Hong and 25 Krauss, 2017; Sulik et al., 1981). Thus, attenuation of the Shh pathway could mechanistically explain the vangl2-ethanol interaction.

2.3.8. Ethanol indirectly attenuates Shh signaling

To investigate the interaction of cyclopamine and ethanol, I first exposed wild- type zebrafish embryos to cyclopamine (50 µM) at shield stage (6 hpf) for 24 hours, mimicking the ethanol exposure window for the vangl2 mutants. Embryos were fixed at 4 dpf and the cartilage and bone were stained with Alcian blue and Alizarin red, respectively. I observed a range of midfacial defects with the most severe phenotype resulting in a complete loss of the anterior neurocranium and reduced spacing between the eyes (Figure 2.10B). Since cyclopamine was predicted to mimic the effects of ethanol, I next combined this low dose of cyclopamine with the subteratogenic dose of ethanol. Strikingly, all embryos presented with synopthalmia or cyclopia with significant reduction to the neuro- and viscerocranium (Figure 2.10B). To quantify this combinatorial effect, I measured the distance between the lens and observed a significant reduction relative to both control and cyclopamine alone (p<0.0001), suggesting a strong synergistic interaction (Figure 2.10C). In chick, ethanol exposure during somitogenesis has been proposed to suppress Shh signaling and induce apoptosis in cranial neural crest cells that make up the craniofacial skeleton (Ahlgren et al., 2002). However, work in zebrafish shows only a modest increase in cell death within the eye field at a much higher dose of 2% ethanol (Santos-Ledo et al., 2013). To ensure that cells in the eye field are not simply undergoing apoptosis, I performed a TUNEL cell death assay in ethanol-treated vangl2 mutants at 11 hpf, prior to optic vesicle evagination. As expected, I failed to detect an increase in apoptotic cells within the eye field in vangl2 mutants or their wild-type/heterozygote

26 siblings (Figure 2.11). Thus, Shh-mediated pro-survival signals do not appear to be disrupted by ethanol. To test for a direct effect of ethanol on Shh signaling, I quantified the relative level of ptch2, a canonical read-out of Shh pathway activity, at bud stage (10 hpf). Ethanol exposure had no effect on the expression levels of ptch2 (Figure 2.10D). While cyclopamine significantly reduced Shh signaling (p=0.0006), ethanol did not further reduce ptch2 levels (p=0.1115) (Figure 2.10D). KEGG analysis from the RNA-seq confirmed that ethanol does not reduce the levels of Shh target genes (i.e. gli, ptch) (Figure 2.10E). Thus, at normally subteratogenic doses, ethanol does not appear to directly attenuate Shh signaling.

2.3.9. Ethanol disrupts convergent extension

The ethanol-induced vangl2 mutant phenotype closely mirrors those in compound mutants between vangl2 and other Wnt/PCP pathway members (Heisenberg et al., 2000; Marlow et al., 1998). In addition, these double mutants display a further reduction in convergent extension, as evidenced by a shorter and broader body axis (Marlow et al., 1998). Based on these data, I hypothesized that ethanol disrupts convergent extension, which would mislocalize the Shh signal and result in eye defects. I performed in situ hybridization on untreated and ethanol-treated vangl2 embryos to examine convergent extension. I analyzed expression of shh in the axial mesoderm and paired box 2a (pax2a) (Krauss et al., 1991) in the midbrain-hindbrain boundary at bud stage (10 hpf). Ethanol was initiated at high stage (3.3 hpf), when vangl2 heterozygotes and homozygotes are equally sensitive to the effects of ethanol. I observed a gene and ethanol-dose dependent reduction in the length of the shh expression domain (extension)

27 and an increase in the width of the pax2a expression domain (convergence) (Figure 2.12A). To quantify the effect of ethanol on convergent extension, I plotted the normalized expression values of shh/pax2 (Figure 2.12B). Post-hoc analyses (Tukey’s) revealed significant differences between untreated embryos across vangl2 genotypes, indicating that gene dosage affects convergent extension (Figure 2.12B). Similarly, I found significant differences between ethanol-treated mutants and their heterozygous (p = 0.017) and wild-type siblings (p = 0.017). This observation confirms that loss of vangl2m209 results in reduced convergent extension movements, as evidenced by their short body axis. This phenotype was exacerbated with ethanol treatment, resulting in reduced convergent extension for vangl2 heterozygotes and homozygotes, compared to their untreated counterparts. While there was a trend, I did not observe a difference in convergent extension between ethanol-treated vangl2 heterozygotes and their wild-type siblings in their shh/pax2 ratio. However, synophthalmia occurred in ethanol-treated heterozygotes but not wild-type embryos, demonstrating that the reduced convergent extension in the heterozygotes can have a phenotypic consequence. Previous work in Wnt/PCP mutants suggests an indirect relationship between convergent extension defects and cyclopia (Marlow et al., 1998). For instance, a shorter expression domain for shh was noted for zebrafish homozygous for gpc4m119 relative to vangl2m209, despite the fact that vangl2 mutants had higher incidences of cyclopia (Marlow et al., 1998). The cyclopia phenotype was correlated with reduced anterior movement of the axial mesoderm expressing shh (Marlow et al., 1998). Thus, quantifying the distance between dlx3, which labels the neural plate border, and the anterior tip of shh, may be better suited to capture the heterozygous effect.

28 2.3.10. Ethanol alters six3 and rx3 expression in the eye field

Transcription factors involved in the specification of the eye field have also been implicated in the mechanism of eye field separation. The expression of three of these transcription factors, six3, rx3, and rx1, are altered by ethanol exposure (Santos-Ledo et al., 2013). I examined the effect of ethanol on six3a and rx3 in ethanol-treated vangl2 mutants at the initiation of optic vesicle evagination. In situ hybridization of six3a in 11 hpf embryos shows a heart-shaped expression pattern in the prospective forebrain. The caudal indentation marks the splitting of the eye field into bilateral domains (Fig. 2.13A). This expression pattern becomes more diffuse with loss of vangl2. In untreated homozygous mutants, I observed a shortening along the anterior-posterior (AP) axis and a broadening along the mediolateral axis, consistent with the convergent extension defect. This expression pattern was further exacerbated in ethanol-treated mutants with complete loss of the caudal indentation. I observed a similar effect of genotype and ethanol on rx3 expression at mid-evagination (12 hpf) (Figure 2.13B). At this stage, rx3 is localized to the prospective forebrain and retina (Mathers et al., 1997). Ethanol-treated homozygous mutants exhibit a compressed expression domain, clearly displaying a reduction in convergent extension. My expression analyses suggest the eye field is specified but mis- localized likely due to defects in mesodermal migration.

2.3.11. Mutation in gpc4 enhances cyclopia in a dose-dependent manner

My data support a model in which ethanol interacts with vangl2 via a combinatorial disruption of convergent extension. One possibility is that ethanol disrupts the transcription of Wnt/PCP pathway members. KEGG pathway analysis showed that the expression of gpc4 was modestly reduced in ethanol-treated embryos across the RNA-seq dataset (Fig. 2.4D). I plotted the normalized read counts across different time

29 points from the RNA-seq to further investigate the potential alteration to gpc4 levels. Ethanol exposure consistently downregulated gpc4 at 8 and 10 hpf, but the magnitude of the downregulation was relatively modest (Figure 2.14A). To statistically compare ethanol-treated and control embryos, expression of gpc4 at 10 hpf was investigated using qRT-PCR. This result demonstrated that gpc4 was not significantly affected by ethanol exposure (Figure 2.14B). Thus, transcriptional alteration to the Wnt/PCP pathway is unlikely to cause the ethanol-induced phenotypes in vangl2 mutants and heterozygotes. If ethanol disrupts convergent extension which leads to interactions with vangl2, then further genetic disruption to convergent extension should exacerbate the ethanol- vangl2 phenotype. Previous work in zebrafish has shown a functional interaction between vangl2 and gpc4, where vangl2;gpc4 double mutants were invariably cyclopic (Marlow et al., 1998). I conducted additional functional analyses to further examine the relationship between these two genes in the context of ethanol exposure (Figure 2.14D). Consistent with Marlow et al., double mutants were fully penetrant for cyclopia with or without ethanol. While ethanol did appear to alter the face of gpc4 mutants, it did not cause cyclopia. However, embryos carrying 3 mutant alleles (either vangl2-/-;gpc4+/- or vangl2+/-;gpc4-/-) were greatly sensitized to ethanol-induced cyclopia. Collectively these data strongly suggest that ethanol alters Wnt/PCP activity post-transcriptionally and that, in sensitized genotypes, this mispositions a source of Shh that is essential for separation of the eye field into bilateral primordia.

2.4. DISCUSSION

Despite its prevalence, the mechanism of ethanol-induced teratogenesis remains uncertain. Here I analyzed gene expression in the context of ethanol teratogenesis in wild-type embryos to shed light upon this mechanism. My subteratogenic dose does not

30 elicit physical malformations in a wild-type background but can in certain sensitized backgrounds (Swartz et al., 2014). This dose mimics an acute (binge-like) alcohol exposure roughly equivalent to a blood alcohol concentration of 0.19 in humans. My data demonstrates that the effect of ethanol on the early transcriptome is mild; nonetheless I were able to detect a transcriptional response using a single embryo assay. My study is unique from other transcriptome profiling studies, which often use a higher dose of ethanol to induce craniofacial and neurological malformations, resulting in large-scale transcriptional changes (Berres et al., 2017; Green et al., 2007). Phenotypic dysmorphology results in a wild-type background with high enough concentrations of ethanol (Blader and Strahle, 1998; Joya et al., 2014), which can easily alter gene expression. By using a subteratogenic dose, I identified dominant modifiers of ethanol teratogenicity that co-vary with treatment across individuals. WGCNA analysis identified two modules (mediumpurple4 and darkolivegreen4) of genes that are coordinately altered by ethanol exposure. The vast majority of genes that negatively correlate with ethanol exposure (membership in darkolivegreen4) encode zinc finger proteins located on chromosome 4. Ethanol has similarly been shown to downregulate Zinc finger protein, subfamily 1A, 4 in mice fetuses exposed to ethanol during early development (Da Lee et al., 2004). Despite this, the functional significance of these ZnF proteins remains unknown and they often have diverse binding affinities and functions. The long arm of zebrafish chromosome 4 (Chr4q) is typically heterochromatin and thus condensed, lacking in protein-coding genes (Howe et al., 2013). However, it is not late replicating until the end of gastrulation or bud stage (10 hpf) (Siefert et al., 2017). Previous work in zebrafish has found ZnF proteins on chromosome 4 to undergo robust expression following the initiation of zygotic transcription until mid-gastrula stage (White et al., 2017). Since many chromosome 4 genes are downregulated across ethanol- 31 treated individuals during gastrulation, ethanol may interfere with replication timing, blocking the early-to-late replication switch, or chromatin remodeling. Based on these findings, the ethanol-sensitive ZnF proteins most likely play a role in transcriptional regulation. As Vangl2 is a core member of the Wnt/PCP pathway, and the vangl2m209 mutant phenotype indicates dysfunction of this pathway, a simple explanation for ethanol- sensitivity in these mutants is that ethanol itself dysregulates the Wnt/PCP pathway. However, RNA-seq results indicated that, at least at the transcriptional level, this is not the case. Although ethanol treatment induced differential expression of hundreds of genes with functional enrichment for oxidative stress and reduced cell division, this transcriptional response showed no overlap with the Wnt/PCP pathway. Transcriptional patterns did however resemble those induced by cyclopamine, a drug known to cause teratogenic phenotypes resulting from high or subteratogenic levels of ethanol exposure in wild-type and vangl2 mutants, respectively. Hence, although Wnt/PCP is not transcriptionally dysregulated by ethanol, differential expression was linked with both cellular stress and teratogenesis. To explain these results, I propose that ethanol leads to midfacial defects and cyclopia, not by dysregulating Wnt/PCP directly, but indirectly through inhibition of convergent extension. In support of this hypothesis, I demonstrate that ethanol does indeed disrupt convergent extension, with greater disruption observed in vangl2 mutants.

2.5. MATERIALS AND METHODS

2.5.1. Zebrafish (Danio rerio) Care and Use

Zebrafish were cared for using standard IACUC-approved protocols at the University of Texas at Austin. The wild-type AB strain was used for RNA-seq analysis. The 32 chemically-induced vangl2m209 allele, originally described as trim209 (Jessen et al., 2002), was obtained from the Zebrafish International Resource Center (ZIRC) as reported (Swartz et al., 2014). The gpc4fr6 line was provided by Dr. Lila Solnica-Krezel. Adult fish were maintained on a 14h/10h light-dark cycle at 28.5°C. Embryos were collected and staged according to morphology (Kimmel et al., 1995). AB, vangl2m209, and gpc4fr6 embryos were treated with 1% ethanol diluted in embryo media.

2.5.2. Sample collection and RNA extraction

A single embryo was manually dechorionated and collected in a 1.75 mL microcentrifuge tube with 500 mL of TRIzol reagent (Life Technologies, 15596-026). Embryos were homogenized with a motorized pestle (VWR, 47747-370) and stored at -80°C until RNA extraction. Total RNA was processed according to the TRIzol RNA isolation protocol. Samples were re-suspended with 50 uL of nuclease-free water and subsequently purified using the RNA Clean & Concentrator kit (Zymo, R1018). The concentration of each sample was determined using a Nanodrop spectrophotometer. The quality of total RNA was analyzed with the Agilent BioAnalyzer to ensure that the RNA Integrity Number (RIN) was ≥ 8. Samples were submitted to the Genomic Sequencing and Analysis Facility (GSAF) at the University of Texas at Austin. The GSAF performed standard RNA-Seq library preparations with poly-A mRNA capture.

2.5.3. RNA-seq data processing

Sequencing on the NextSeq 500 platform produced an average of 40.8 million ± 1.4 million (SE) raw paired end reads per sample. Adapter trimming was performed using Cutadapt with a minimum length of 25 bp (Martin, 2011). Following adapter trimming, we retained an average of 40.0 ± 1.3 million (SE) reads per sample. Genome Reference

33 Consortium Zebrafish Build 10 (GRCz10) for D. rerio was downloaded from Ensembl (Aken et al., 2017). Trimmed reads were mapped to the reference using STAR (Dobin et al., 2013). Mean mapping efficiency was 78.2% ± 0.8% (SE). Following mapping PCR duplicates were removed using Picard (https://broadinstitute.github.io/picard/).

Duplication rate was estimated at 85% ± 0.8% (SE). Sorting and conversion between SAM and BAM files was performed using samtools (Li et al., 2009a). Reads mapping to annotated genes were counted with HTseq version 0.6.1p1 using the intersection nonempty mode (Anders et al., 2015). The final number of reads mapped to annotated genes was on average 3.9 ± 0.2 million reads per sample. Detailed instructions and example commands for implementing the data processing steps described above are available on Github (https://github.com/grovesdixon/Drerio_early_ethanol_RNAseq).

2.5.4. Differential expression analysis

Normalization and statistical analysis of read counts was performed using DESeq2 (Love et al., 2014). Factors included in the differential expression models were ethanol treatment (control, treated), developmental timepoint (8 hpf, 10 hpf, and 14 hpf), and sequencing batch (experiment 1 or experiment 2). Because none of the 6 hpf samples were treated with ethanol, these samples were not included. We tested for differential expression associated with ethanol treatment using likelihood ratio tests—comparing the model including all three factors to a reduced model that did not include ethanol treatment. To further examine timepoint-specific ethanol effects we split the samples by developmental timepoint and tested for ethanol effects within each.

34 2.5.5. GO enrichment

Enrichment of Gene Ontology for ethanol responsiveness was tested using two-tailed Mann-Whitney U-tests (Dixon et al., 2015) followed by Benjamini-Hochberg procedure for false discovery correction (Benjamini and Hochberg, 1995). The results were plotted as a dendrogram tracing hierarchical relationships between significant GO terms. The direction of enrichment (for upregulation or downregulation) was indicated by text color and significance of enrichment by font type. An advantage of this approach is that it does not require an arbitrary cutoff to provide counts of “significant” and “non-significant” genes as in typical enrichment tests.

2.5.6. Weighted Gene Correlation Network Analysis (WGCNA)

Gene expression data were further analyzed with Weighted Gene Correlation Network Analysis (WGCNA) (Langfelder and Horvath, 2008). For input into the analysis we used variance stabilized counts generated using the rlog function in DESeq2 (Love et al., 2014). Genes that were not sequenced across sufficient samples (4320 in total) were removed using the goodSamplesGenes function in the WGCNA package. Because there were no ethanol treated samples for the six hour timepoint, the five samples from this timepoint were removed before further analysis. To ensure sufficient expression for correlation detection, genes were further filtered based on a base mean expression cutoff of 5. We controlled for batch effects using the ComBat function from the R package sva (Leek and Storey, 2005). We selected a soft threshold of 15, where the scale free topology model fit surpassed 0.8. WGCNA was run with a minimum module size of 10. Following network analysis, we tested for GO enrichment within modules using Fisher’s exact tests.

35 2.5.7. Quantitative Real-Time qRT-PCR (qRT-PCR)

To validate our RNA-seq data, we selected two genes to test using qRT-PCR. Total RNA was reverse transcribed using SuperScript™ First-Strand Synthesis System for RT-PCR (Invitrogen) with oligo-d(T) primers. qRT-PCR was performed with Power Sybr Green PCR Master Mix (Thermo Fisher Scientific, 4367659) on the Applied Biosystems ViiA™ 7 Real-Time PCR System. QuantStudio Real-Time PCR Software was used for data analysis. Two endogenous controls were selected based on their stable expression profiles across treatment in the RNA-seq data: eef1a1l1 and mob4.

2.5.8. Cartilage and Bone Staining and Measurements

Embryos were fixed at 4 days post fertilization (dpf) and stained with Alcian blue for cartilage and Alizarin red for mineralized bone (Walker and Kimmel, 2007). Whole- mounts of vangl2m209 embryos were captured using a Zeiss Axio Imager A1 microscope. To assess the degree of cyclopia, the distance between the medial edges of the lenses was measured using the AxiovisionLE software.

2.5.9. In Situ Hybridization

Antisense riboprobes for shh, pax2, dlx3, six3a (gift from Dr. Steve Wilson), and rx3 (gift from Dr. Steve Wilson) were used. Whole-mount in situ hybridization was performed as described {Miller:2000tt}. Images were captured using the Zeiss Axio Imager A1 and expression domains were measured using the AxiovisionLE software. An ANOVA and post-hoc Tukey’s test were used for statistical analyses.

2.5.10. TUNEL Staining

Whole-mount TUNEL staining was modified from (Lovely et al., 2016). Samples were fixed overnight in 4% paraformaldehyde in PBS (PFA) at 4°C. Samples were dehydrated

36 in methanol and subsequently rehydrated in phosphate-buffered saline containing 0.5% Triton X-100 (PBTx). Samples were permeabilized with 25 ug/mL proteinase K (1mg/ml) in PBT for 30 min. After two, 5 min washes with PBTx, samples were fixed with 4% PFA for 20 min at room temperature. Residual PFA was removed with four, 5 min washes of 4% PFA. Samples were incubated with 50 ul of 1:10 Enzyme:TUNEL reagent (TdT and fluorescein-dUTP) (Roche, Cat No. 11684795910) at 37°C for 3 h in the dark. The reaction was stopped with two, 5 min washes of PBTx. Confocal images were captured with a Zeiss LSM 710.

2.6 ACKNOWLEDGEMENTS

We thank Dr. Steve Wilson for his kind contributions of the six3a and rx3 riboprobes. We also thank Dr. Lila Solnica-Krezel for providing the gpc4fr6 line. We are grateful to Angie Martinez for maintenance and care of all zebrafish lines.

37 2.7 FIGURES

Figure 2.1: vangl2 mutants are sensitive to ethanol during early embryogenesis. (A) Treatment of vangl2 embryos with 1% ethanol at four different stages comprising late blastula to early gastrula [3.3, 4, 4.5, and 6 hours post fertilization (hpf)] for 24 hours. Homozygous mutants showed a significant decrease in the inner lens-to-lens width when treated from 6 to 30 hpf. (B) Heterozygous mutants were most sensitive to ethanol when exposed from 3.3 to 27.3 hpf (n=4/18 cyclopic). Homozygous mutants were most sensitive when exposed from 6 to 30 hpf (n=5/5 cyclopic).

38

Figure 2.2: The effect of ethanol on the early zebrafish transcriptome is subtle. (A) Schematic representation of the RNA-seq experimental design. Wild-type AB embryos were exposed to a subteratogenic dose of 1% ethanol in the embryo media at shield stage (6 hpf). Embryos were subsequently collected at 8 and 10 hpf for experiment 1 and (B) 8, 10, and 14 hpf, for experiment 2. Each sample consisted of a single zebrafish embryo with 5 control and 5 ethanol-treated samples per timepoint with the exception of the 6 hpf timepoint, which only had 5 control samples. (C-F) Principal components analysis of top 25,000 most variable genes. The percentage of variance explained is given on each axis label. (C) PC1 and PC2 color coded by developmental timepoint. (D) PC1 and PC2 color coded by RNA-seq experiment (batch). (E) PC1 and PC2 color coded by ethanol treatment. (F) PC8 and PC9 showing separation of ethanol treated and control samples.

39

Figure 2.3: Time is the biggest driver of variation in the dataset. Heatmap showing overall correlation of gene expression among samples. Samples were hierarchically clustering based on similarity. Treatment [control or ethanol treated (C or E)] and developmental age [hours post fertilization (6, 10, 8, or 14)] are indicated in the row and column labels.

40

Figure 2.4: Effects on transcription are largely distinct between developmental time points. (A) Volcano plot showing differential expression due to ethanol treatment across all samples. Significant genes (FDR < 0.1)* are indicated in red. Genes that were significant for each time point individually are indicated in blue. (B) Venn-diagram showing overlap of significant genes between the three individual timepoints. (C) Table of genes that were significant for each timepoint individually (D) KEGG pathway schematic illustrating differential expression due to ethanol treatment in the Wnt/planar cell polarity (Wnt/PCP) pathway. Color coding indicates log2 fold differences due to ethanol treatment across all samples, red indicates upregulation due to ethanol, blue indicates downregulation due to ethanol. Only pathway members with significant genes were color coded. In the event that multiple genes from the dataset were annotated with the same pathway member, the log2 fold difference for the gene with the greatest absolute value for the difference was used.

41

Figure 2.5: There are more upregulated than downregulated genes among ethanol- treated individuals. Volcano plot showing variation in the transcriptional response to ethanol treatment across developmental timepoints. Significant genes (FDR < 0.1) are indicated in red. For each subset, the names of the topmost significant genes next to the gene’s data point (A) All timepoints together (B) 8 hpf only (C) 10 hpf only (D) 14 hpf only.

42

Figure 2.6: Trees illustrate the hierarchical organization of the enriched GO terms. Red GO terms are enriched for genes upregulated by ethanol. Blue GO terms are enriched for genes downregulated by ethanol. (A) Molecular Function (B) Biological Process (C) Cellular Component.

43

Figure 2.7: WGCNA identifies two modules that are significantly correlated with ethanol exposure. (A) Dendrogram illustrating the hierarchical clustering of the genes and with their corresponding modules colors. The top layer of colors indicates the Dynamic Tree Cutoff, including all assigned modules, before merging by module similarity. The bottom layer indicates the module colors after merging. These were the modules used for further analysis. (B) Heatmap of module-trait correlations. The eigengene for each module was correlated with ethanol treatment (ethanol), hours post fertilization (time), experimental batch (seqjob), and as a negative control, a randomly shuffled version of the ethanol treatments (rand.eth). Intensity of the color in each cell indicates the strength of correlation between the module (row labels) and the sample trait (column labels). Two modules, (mediumpurple4 and darkolivegreen4) significantly correlated with ethanol treatment (p < 0.05). (C-D) Scatterplots of correlation with ethanol treatment against module membership. Each datapoint is a gene assigned to the indicated module. Ethanol correlation is the Pearson correlation between the gene’s expression level and ethanol treatment. Module membership is the correlation between the genes expression level and the module eigengene and describes how well the gene matches the overall patterns of the module. (E) Boxplot of log2 fold differences due to ethanol for the two significant modules. (F) Gene ontology enrichment tree for Molecular Function for the mediumpurple4 module. (G) Gene ontology enrichment tree for Molecular Function for the darkolivegreen4 module. 44

Figure 2.8: Quantitative Real-Time qRT-PCR (qRT-PCR) validates RNA-seq. Changes in gene expression from the RNA-seq were validated using wild- type embryos at 10 hpf. (A) slc16a9a was selected from the mediumpurple4 module and (B) znf1015 was selected from the darkolivegreen4 module. Fold change indicates the degree of change between untreated control and ethanol-treated stage-matched embryos. lsm12b (like-Sm protein 12 homolog b) was used as a normalization control.

45

Figure 2.9: Expression of hub genes for development-related modules indicate ethanol did not retard developmental progression. (A-F) Development- related WGCNA modules were identified as those with significant relationship to hours post fertilization (Pearson correlation; p<0.05). The hub gene for each of these modules was identified as the gene assigned to that module with the highest module membership (defined as the correlation of the gene’s expression level with the module eigengene). Normalized expression levels for the hub genes were plotted against hours post fertilization. The relationship between expression and time is largely consistent between the ethanol treated and controls samples. Even for the two modules associated with ethanol treatment (mediumpurple4 and darkolivegreen4), the slopes of the lines are very similar, indicating a similar relationship to time despite a consistent increase or decrease in expression due to ethanol exposure.

46

Figure 2.10: Ethanol indirectly attenuates Shh signaling. (A) Schematic representation of the LINCS L1000 query. Gene expression signatures, such as the top differentially expressed genes (DEGs) for a diseased or experimental state, can be queried against LINCS L1000. The query generates a rank-ordered list of experimental conditions, including small molecule compounds, that have similar or opposite signatures to the input signature. 47 Figure 2.10, cont.: (B) Alcian blue and Alizarin red whole mount staining of untreated (control), cyclopamine-treated (50 µM), and ethanol- and cyclopamine- treated (1% ethanol and 50 µM cyclopamine) wild-type embryos from 6 hpf to 4 dpf. Different panels represent the spectrum of phenotypes observed for the treatment groups. Embryos were fixed at 4 dpf. Dorsal view, anterior to the left. (C) Quantification of the effect of ethanol and cyclopamine on the eye field. Inner lens-to-lens width was used as a morphometric measure of cyclopia. Both cyclopamine alone and cyclopamine and ethanol, were significantly different from controls (p<0.0001). (D) qRT-PCR of ptch2 in 10 hpf wild-type embryos. slc25a5 (solute carrier family 25) was used as a normalization control. (E) KEGG pathway schematic illustrating differential expression due to ethanol treatment in the sonic hedgehog (Shh) signaling pathway. Color coding indicates log2 fold differences due to ethanol treatment across all samples; red indicates upregulation due to ethanol, blue indicates downregulation due to ethanol. In the event that multiple genes from the dataset were annotated with the same pathway member, the log2 fold difference for the gene with the greatest absolute value for the difference was used.

48

Figure 2.11: Cell death by TUNEL at 11 hpf in untreated and ethanol-treated vangl2 mutants. The number of positive cells in the eye field (indicated by dashed circle) were not higher in homozygous or ethanol-treated mutants. ant = anterior; nc = notochord

49

Figure 2.12: Ethanol disrupts convergent extension. (A) vangl2 embryos were treated with 1% ethanol from 3.3 to 10 hpf. Embryos were subsequently harvested at 10 hpf for in situ hybridization with shh (midline) and pax2a (midbrain- hindbrain boundary) probes. Dorsal views, anterior to the left. (B) Quantification for the normalized shh/pax2a expression domains were calculated using the AxiovisionLE software. Two-way ANOVA followed by Tukey’s post-hoc test was used to analyze the results. Differences were observed between control genotypes and ethanol-treated homozygous mutants.

50

Figure 2.13: Ethanol alters six3 and rx3 expression in the eye field. Expression pattern of transcriptions factors that specify the eye field stained using whole mount in situ hybridization (WISH) with (A) six3a at 11 hpf and (B) rx3 at 12 hpf . Dorsal views, anterior to the left.

51

Fig 2.14: gpc4 mutation enhances cyclopia in a dose-dependent manner. (A) Normalized read counts indicating expression of gpc4 across different subsets of the dataset. (B) qRT-PCR of gpc4 in 10 hpf wild-type embryos. actb1 (actin, beta 1), lsm12b (like-Sm protein 12 homolog b), and slc25a5 (solute carrier family 25) were used for normalization. (C) Alcian blue and Alizarin red whole- and flat-mount staining of untreated and ethanol-treated (1% ethanol) gpc4 homozygous mutants from 6 hpf to 4 dpf. Embryos were fixed at 4 dpf. Dorsal view, anterior to the left. (D) Functional analyses of vangl2;gpc4 double mutants. Enhanced cyclopia was observed in ethanol- treated vangl2;gpc4 double mutants with loss of at least one copy of either gene.

52 Chapter 3: RNA-seq analysis of vangl2

3.1. INTRODUCTION

Birth defects manifest as structural or functional malformations at birth and are a leading cause of infant mortality in the US (Czeizel, 2005). Genetics, environmental factors, or complex multifactorial interactions are implicated in the etiology of birth defects. A gene-environment interaction can occur when a genetic predisposition sensitizes an individual to the teratogenic effects of an environmental agent (Czeizel, 2005). The leading cause of preventable birth defects in the US is maternal alcohol consumption and it can cause a spectrum of defects collectively referred to as fetal alcohol spectrum disorders (FASD). Although the underlying mechanisms are not well understood, studies in humans and animal models provide strong evidence for genetic susceptibility. Genetic screens are one method to identify genetic modifiers of ethanol teratogenicity. Using zebrafish, my lab identified 6 candidate loci that interact with ethanol to cause craniofacial anomalies (McCarthy et al., 2013; Swartz et al., 2014). This screen revealed a strong interaction with vangl2, a member of the non-canonical Wnt/planar cell polarity (PCP) pathway, and ethanol (Swartz et al., 2014). In control conditions, homozygous mutants infrequently present with cyclopia, however, mutants exposed to a normally subteratogenic dose of ethanol at gastrulation have fully penetrant midfacial defects, including cyclopia (Swartz et al., 2014). To determine the early effects of ethanol exposure on the transcriptome, I recently performed RNA-seq in ethanol- treated wild-type embryos (described in Chapter 2). I did not detect broad attenuation of the Wnt/PCP pathway at the level of transcription in a wild-type background. Rather, this study implicated an indirect effect of ethanol exposure on the Sonic hedgehog (Shh)

53 pathway. Consistent with this model, I find that ethanol disrupts convergent extension movements of the mesoderm, which in turn mispositions the Shh signal. In this follow-up study, I extend my RNA-seq analyses to examine the early effects of ethanol exposure in a vangl2 background. The vangl2m209 allele was previously reported to be cold-sensitive. Embryos maintained at 23oC phenotypically resemble ethanol-treated embryos normally maintained at 28.5oC (Marlow et al., 1998). For this reason, I included untreated embryos raised at 23oC, to determine if cold-reared mutants share a similar gene expression profile as ethanol-treated mutants. I hypothesize that this unbiased transcriptomic approach would reveal mechanistic insights behind the synergistic vangl2-ethanol interaction.

3.2. RESULTS

3.2.1. A transcriptional profile of untreated and ethanol-treated vangl2 embryos

I raised clutches of embryos generated from an incross of vangl2m209 heterozygotes at two temperatures, 23°C (room temperature) and 28.5°C (incubator). Both ethanol and incubation at 23°C increase the penetrance and expressivity of the cyclopia phenotype in homozygous mutants (Marlow et al., 1998; Swartz et al., 2014). To analyze the effect of ethanol and temperature on the developing eye field, I sampled embryos before eye field separation (10 hours post fertilization, hpf), using previously established morphological criteria. I treated a subset of embryos raised at 28.5°C with a normally subteratogenic dose of 1% ethanol in embryo media during gastrulation (6–10 hpf) (Figure 3.1). Offspring were phenotypically separated into two pools consisting of homozygous mutants with reduced convergent extension of the body axis (vangl2m209/m209) or siblings exhibiting a wild-type body axis (vangl2m209/?), for each treatment group (control and ethanol). Six biological replicates of pools of 3 embryos 54 were collected for each genotypic group and I obtained an average of 25.7 million reads per sample following adapter trimming. Samples 1-3 and 4-6 were collected on separate days and later corrected for batch effects. I performed two runs (approximately 330 million reads/run) on a NextSeq 500 platform for 36 samples. Mapping efficiency to the Danio rerio genome assembly GRCz10 with appropriate paired-end orientation, was approximately 95%. PCR duplication rate was approximately 21%. After removal of PCR duplicates, an average of 7.5 million reads per sample (~39%) were mapped to annotated gene regions (Figure 3.2). I first applied principal components analysis (PCA), a linear analysis that reduces a multidimensional dataset to two axes (principal components) that capture the largest drivers of variation. I then post-hoc correlated these axes with my known experimental and technical variables (treatment, temperature, genotype, and batch). PCA shows close clustering of samples based on treatment (Figure 3.3A). The first and second principal components explain about 46% of the variance in the data. I find a strong effect of treatment as it corresponds to both PC1 and PC2. Samples secondarily clustered by temperature despite the lack of ethanol-treated samples at 23°C (Figure 3.3B). I did not find genotype (Figure 3.3C) or batch (Figure 3.3D) to correlate with either PC1 or PC2.

3.2.2. Differentially expressed genes in response to genotype, temperature, and ethanol

I detected extensive differential gene expression in response to both ethanol and temperature treatments. Of the 16,913 genes with an average of at least 5 reads per sample, 3,035 and 2,647 genes were differentially expressed in response to ethanol and temperature, respectively (false-discovery rate (FDR) < 0.1, Benjamini–Hochberg

55 procedure). A plot of negative log10 transformed p-values against log2 fold differences between ethanol treatment groups shows nearly even distribution of DEGs (1,610 downregulated vs. 1,425 upregulated) (Figure 3.4A). I find a similar distribution due to temperature (1,368 downregulated vs 1,279 upregulated) (Figure 3.4B). I was not able to detect a strong effect of genotype (Figure 3.4C). Only 737 genes showed a tendency towards differential expression (raw p-value <0.05) and only nine met the FDR significance cut-off of < 0.1 (Figure 3.4E). Three of the nine are currently uncharacterized (CABZ01113812.1, si:ch211-74f19.2, si:dkey-229d11.5). CABZ01113812.1 is significantly upregulated in homozygous mutants and BLAST analysis identified 98.18% sequence homology with emc4, a subunit of the ER membrane protein complex (EMC). Previous studies suggest a critical role of the EMC in protein trafficking and accumulation on the plasma membrane (Chitwood and Hegde, 2019). Another upregulated gene, furinA, is an enzyme that cleaves and activates proproteins involved in various pathways (Thomas, 2002). It plays a critical role in proper development of the ventral jaw by activating the endothelin pathway in zebrafish via cleavage of Edn1 (Walker et al., 2006). I also detected upregulation of intu (inturned), a gene that encodes a Wnt/PCP effector necessary for formation of primary cilia (Park et al., 2006; Zeng et al., 2010). Despite the striking phenotypic effect of ethanol in vangl2m209 mutants, no genes met the FDR cut-off of < 0.1 for a gene-ethanol interaction (Figure 3.4D).

3.2.3. GO and KEGG pathway analysis

To gain insight on the molecular functions, biological processes, and cellular components of the 3,035 DEGs due to ethanol treatment, I performed GO enrichment analysis (Figure 3.5). GO terms enriched for upregulation were RNA metabolic process,

56 RNA processing, mRNA/RNA binding, DNA packaging complex, and protein-DNA complex. GO terms enriched for downregulated genes were regulation of cell development, multicellular organismal development, calcium ion binding, extracellular region part. Additionally, I performed KEGG analysis to identify ethanol-induced alterations of genes involved in different pathways. The neuroactive ligand receptor interaction was the only statistically significant pathway identified, indicating dysregulation of genes involved in neural activity (Figure 3.6).

3.2.4. Weighted Gene Correlation Network Analysis (WGCNA)

Next, I performed Weighted Gene Correlation Network Analysis (WGCNA) to uncover groups of genes (modules) that are coordinately altered in my samples. The cluster dendrogram is organized like a hierarchical tree, with each twig on a branch representing an individual gene. Branches of genes clustered into 25 colored modules using a height cut-off of 0.25 (Figure 3.7A). I examined correlations to treatment, genotype, temperature, and batch. To examine the effects of these variables on molecular functions, I performed GO enrichment analysis for each module. Homozygous mutants raised at 23°C phenocopy an ethanol-treated mutant raised at 28.5°C. Both sets of mutants resulted in a similar phenotype despite different environmental exposures. I identified several modules with the same directional change in both sets of mutants (Figure 3.7B). The cyan module represents genes upregulated in both ethanol and cold-reared mutants. These genes were enriched for several significant GO terms, including but not limited to, eye morphogenesis, tissue morphogenesis, and the non-canonical Wnt signaling pathway (Table 3.1). The dark orange module was conversely downregulated in both mutants. These downregulated genes were associated

57 with embryonic morphogenesis, multicellular organismal development, neuron development and differentiation. From these GO analyses, I identified alteration of several notable genes specifically in ethanol-treated embryos. The seven-pass transmembrane and Wnt receptor, fzd8b (frizzled class receptor 8b), was significantly downregulated in ethanol- treated mutants, but not cold-reared mutants. Previous work in zebrafish has shown expression in the prospective prechordal plate during gastrulation (Kim et al., 1998). A similar reduction in fzd8b expression is seen in cyclopic one-eyed pinhead (oep) mutants that lack a prechordal plate (Kim et al., 1998). I previously showed that ethanol disrupts convergent extension most likely by reducing anterior extension of the axial mesoderm and prechordal plate (Blader and Strahle, 1998). From this, I hypothesize that ethanol and temperature have distinct effects on the prechordal plate. Similarly, the eye specification markers, six3b (SIX homeobox 3b) and rx3 (retinal homeobox gene 3), were significantly downregulated in ethanol-treated mutants. Using in situ hybridization, I and others have shown ethanol alters expression of six3b and rx3 during early segmentation (Santos-Ledo et al., 2013). The chaperonin gene, cct3 (chaperonin containing TCP1, subunit 3), was slightly downregulated and upregulated in ethanol-treated and cold-reared mutants, respectively. This gene has robust expression at the end of gastrulation and is known to facilitate cytoplasmic protein folding (Matsuda and Mishina, 2004). I also detected a significant decrease of cyp11a1 (cytochrome P450, family 11, subfamily A, polypeptide 1) in ethanol-treated mutants. Zebrafish Cyp11A1 is known to synthesize pregnenolone from cholesterol, to facilitate the spreading of the blastoderm over the yolk during epiboly, via stabilization of yolk microtubules (Hsu et al., 2006). Embryos deficient in cyp11a1 fail to complete epiboly and result in a shortened axis (Goldstone et al., 2010). 58 I also identified modules that show opposing correlations in the two sets of mutants. The blue2 module was significantly upregulated in ethanol-treated mutants and significantly downregulated in cold-reared mutants. GO analysis revealed enrichment of genes involved in ion transport and oxidation-reduction process, suggesting ethanol, but not temperature, increases oxidative stress in homozygous mutants. The antiquewhite2 module was slightly downregulated in ethanol-treated mutants and upregulated in cold- reared mutants. These genes were enriched for cell migration involved in gastrulation, mesoderm morphogenesis, dorsal convergence, regulation of embryonic development, and the non-canonical Wnt signaling pathway. Lastly, I identified one module (lavendarblush2) that was upregulated in ethanol- treated wild-types and heterozygotes and downregulated in ethanol-treated mutants. These genes were associated with endoplasmic reticulum organization. Although most genes did not show significant differential expression at the gene level, they were part of a module of co-regulated genes that correlated with genotype in ethanol-treated embryos. The endoplasmic reticulum (ER) is critical for proper modification and folding of proteins before they are transported to the Golgi apparatus for eventual secretion to the plasma membrane or extracellular space. Proper ER organization is necessary to prevent the unfolded protein response (UPR) (Sapir et al., 2013). The metabolism of alcohol to acetate generates damaging reactive oxygen species (ROS), leading to oxidative stress (McCarthy and Eberhart, 2014). Oxidative stress can compromise the ER’s ability to properly and efficiently fold proteins, thereby leading to ER stress. The UPR is an adaptive mechanism to restore homeostasis by inhibiting translation of nascent polypeptides, facilitating proper folding, and assisting in the degradation of misfolded proteins (Bravo et al., 2013). This result may suggest that the balance of protein load and misfolded proteins is not efficiently maintained in ethanol-treated mutants. 59 The brown module was significantly associated with treatment and temperature with a positive correlation in cold-reared control embryos and a negative correlation in ethanol-treated embryos raised at 28.5°C. The genes in this module were enriched for RNA processing/splicing and ribosome biogenesis. Ethanol has previously been shown to reduce expression of genes involved in ribosome biogenesis (Berres et al., 2017). Furthermore, hypomorphic mutations in these genes correlate with increased susceptibility for ethanol-induced craniofacial defects (Berres et al., 2017).

3.2.5. Gene-ethanol interactions

In order to identify genes with distinct expression in the genotypic groups, I calculated the log2 fold difference in expression between ethanol treated and control samples for the vangl2m209/? and vangl2m209/m209 genotypes. From these, I identified 49 genes (Figure 3.8A, red dots) for which the log2 fold difference between ethanol and control samples differed by at least 1 between vangl2m209/? and vangl2m209/m209. Many of these genes were also found to have module membership in a WGCNA module that show correlated expression across samples (Figure 3.8B). These genes were nearly twice as frequent in the navajowhite2 module as other modules. Furthermore, navajowhite2 was more correlated with ethanol among vangl2m209/m209 (r=0.36) than vangl2m209/? (r=0.13), so these genes are a group of potentially co-related genes that are particularly responsive to ethanol among mutants.

3.3. DISCUSSION

I adopted a transcriptomic approach to better understand the molecular mechanisms of ethanol teratogenesis underlying a gene-ethanol interaction. Ethanol- exposed vangl2 mutants raised at 28.5oC display midline defects and cyclopia. Untreated

60 mutants raised at 23oC (room temperature) phenocopy ethanol-exposed embryos, but the mechanism of this interaction is unclear. Using whole-embryo RNA-seq, I examined differences between these groups to identify synergistic effects due to ethanol, versus phenotypic outcomes alone. I find ethanol to be the main driver of variation in the dataset. According to my PCA analysis, the effect of ethanol and temperature segregate in distinct manners along PC1 and PC2. Surprisingly, genotype does not appear to be a strong driver of variance. In line with these results, I find robust differential expression of genes in response to both ethanol and temperature alone, but not genotype. Furthermore, I fail to identify synergistic gene-ethanol effects at the level of transcription. Nonetheless, WGCNA analysis uncovered modules of genes that correlate with my experimental variables. The cyan module positively correlated with all experimental variables but showed a slightly higher correlation at room temperature. These genes showed functional enrichment for biological processes implicated in eye field morphogenesis and separation. From these analyses, I find that shha (log2 fold=-0.551, p-value=0.0015) and six3b (log2 fold=-0.384, p-value=4.50E-05), two genes that are implicated in holoprosencephaly, are significantly downregulated in ethanol-treated mutants, but not cold-reared mutants. As discussed in chapters 1 and 2, shha plays a critical role in forebrain development, as Shh-null mice develop cyclopia (Chiang et al., 1996). Furthermore, high concentrations of ethanol in chick reduce Shh and downstream transcriptional targets (Ahlgren et al., 2002). I also find a modest decrease in ptch2 (log2 fold= -0.151, p-value= 0.0771), the canonical read-out of Shh pathway activity. In a previous study in zebrafish, ethanol was demonstrated to impede anterior migration of the prechordal plate mesoderm (Blader and Strahle, 1998). In chapter 2, I show evidence for an indirect effect of Shh signaling in ethanol-treated mutants, via a reduction of 61 convergent extension. When a sensitized vangl2 mutant is exposed to ethanol, I find a more robust effect on Shh signaling, likely due to synergistic prechordal mesoderm defects. My analyses suggest ethanol induces cyclopia in vangl2 mutants in part by reducing expression of shha and six3b.

3.4. MATERIALS AND METHODS

3.4.1. Zebrafish (Danio rerio) Care and Ethanol Treatment

Zebrafish were cared for using IACUC-approved protocols at the University of Texas at Austin. The vangl2m209 allele was obtained from stocks originating from the Zebrafish International Resource Center (ZIRC, Eugene, OR). Adult fish were maintained on a 14h/10h light-dark cycle at 28.5°C. Embryos were collected and staged according to morphology (Kimmel et al., 1995). Clutches of embryos obtained from an incross of vangl2m209 heterozygotes were divided before sphere stage and maintained at either 23°C or 28.5°C until time of collection (10 hpf). Embryos at 28.5°C receiving ethanol treatment were bathed in 1% ethanol in embryo media (171 mM) at 6 hours post fertilization (hpf) until 10 hpf. The embryonic tissue level of ethanol rapidly equilibrates to approximately one-third of the embryo media concentration (50 mM).

3.4.2. Sample collection and RNA extraction

Embryos were divided into genotypic groups, wildtype and heterozygote (+/?) and homozygote (-/-), based on phenotypic morphology at 10 hpf. Homozygotes exhibited an increase in the width of their mediolateral axis and a reduction in the length of their anterior-posterior axis. We sequenced 6 biological replicates for each treatment and genotypic group at 28.5°C and 6 biological replicates for each genotypic group at 23°C. Samples maintained at 23°C did not receive ethanol treatment. Each sample consisted of

62 a pool of 3 embryos. Samples 1-3 for every treatment and genotypic group were collected on one day. Samples 4-6 were collected on a separate day. RNA extraction, purification, and quality assessment were performed as previously described (Chapter 2.5.2). Samples were submitted to the Genomic Sequencing and Analysis Facility (GSAF) at the University of Texas at Austin. The GSAF performed standard RNA-Seq library preparations from total RNA with poly-A mRNA capture.

3.4.3. RNA-seq data processing

Sequencing on the NextSeq 500 platform produced an average of 26.0 million (SE) raw paired end reads per sample. Adapter trimming was performed using Cutadapt with a minimum length of 25 bp (Martin, 2011). Following adapter trimming, we retained an average of 25.7 million paired-end reads per sample. Version 10 of the reference genome for D. rerio was downloaded from Ensemble. Trimmed reads were mapped to the reference using STAR (Dobin et al., 2013). Mean mapping efficiency was 95%. Following mapping PCR duplicates were removed using Picard (https://broadinstitute.github.io/picard/). Duplication rate was estimated at 21%. Sorting and conversion between SAM and BAM files was performed using samtools (Li et al., 2009a). Reads mapping to annotated genes were counted with FeatureCounts. The final number of reads mapped to annotated genes was on average 7.5 million reads per sample.

Detailed instructions and example commands for implementing the data processing steps described above are available on Github (https://github.com/grovesdixon/vangl2_gene_expression).

63 3.4.4. Differential expression analysis

Normalization and statistical analysis of read counts was performed using DESeq2 as previously described (Chapter 2.5.4). Factors included in the differential expression models were ethanol treatment (control, treated), temperature (23°C or 28.5°C), and sequencing batch (experiment 1 or experiment 2).

3.4.5. GO enrichment

Enrichment of Gene Ontology for ethanol responsiveness was tested as previously described (Sidik et al., unpublished). The results were plotted as a dendrogram tracing hierarchical relationships between significant GO terms. The direction of enrichment (for upregulation or downregulation) was indicated by text color and significance of enrichment by font type.

3.4.6. Weighted Gene Correlation Network Analysis (WGCNA)

Gene expression data were analyzed with Weighted Gene Correlation Network Analysis (WGCNA) as previously described (Sidik et al., unpublished). We selected a soft threshold of 10, where the scale free topology model fit surpassed 0.8. WGCNA was run with a minimum module size of 20. Following network analysis, we tested for GO enrichment within modules using Fisher’s exact tests.

64 3.5 FIGURES

Figure 3.1: Schematic representation of the vangl2 RNA-seq experimental design. Embryos derived from a vangl2m209/+ cross were maintained at either room temperature (23oC) or the incubator (28.5oC). A subset of embryos maintained in the incubator were exposed to a subteratogenic dose of 1% ethanol in the embryo media at shield stage (6 hpf). Right before collection at bud stage (10 hpf), embryos were grouped by phenotype into two genotypic groups, wild-type and heterozygote (+/?) or homozygote (-/-). Each sample consisted of a three zebrafish embryos with 6 biological replicates per genotypic group, with a total of 36 samples. Samples 1-3 and 4-6 were collected on separate days (batch 1 or 2).

65

Figure 3.2: Summary of sample read counts through the data processing pipeline. (A) Absolute read counts: rawCounts = the total number of paired end reads acquired for the sample; trimmedCounts = the number of paired end reads remaining after adapter trimming; predupPropPaired = the number of paired-end reads that aligned in proper orientation to the reference genome; dedupPropPair = the number of paired-end reads retained after PCR duplicate removal; geneCounted = the total number of read pairs counted for annotated genes. The number of reads included for the sample in the count matrix input into DESeq2. (B) the same as in A, only proportional to the initial raw read count. (C) Histogram of total read pairs counted on annotated genes. (D) Ordered barplot of read paired counted on genes.

66

Figure 3.3: Ethanol treatment correlates to PC1 and PC2. Principal components analysis of the top 10,000 most variable genes. The percentage of variance explained is given on each axis label. (A) PC1 and PC2 color coded by (A) treatment (control or ethanol), (B) temperature (incubator or room temperature), (C) genotype (wild-type/heterozygote or homozygote), and (D) RNA-seq experiment (batch).

67

Figure 3.4: Ethanol and temperature elicit a robust transcriptional response. Volcano plot showing variation in the transcriptional response to (A) ethanol treatment, (B) temperature, (C) genotype, and (D) gene:ethanol. Significant genes (FDR < 0.1) are indicated in red (A-C). The nine significant genes in the genotype plot are listed with their log2 fold change (E).

68

Figure 3.5: Trees illustrate the hierarchical organization of the enriched GO terms. Red GO terms are enriched for genes upregulated by ethanol. Blue GO terms are enriched for genes downregulated by ethanol. (A) Biological Process (B) Molecular Function (C) Cellular Component. 69

Figure 3.6: The Neuroactive Ligand Receptor Interaction pathway is differentially expressed due to ethanol treatment. KEGG pathway schematic illustrating differential expression due to ethanol treatment in the Neuroactive Ligand Receptor Interaction pathway. Color coding indicates log2 fold differences due to ethanol treatment across all samples, red indicates upregulation due to ethanol, blue indicates downregulation due to ethanol. Only pathway members with significant genes were color coded. In the event that multiple genes from the dataset were annotated with the same pathway member, the log2 fold difference for the gene with the greatest absolute value for the difference was used.

70

Figure 3.7: WGCNA identifies modules that are significantly correlated with ethanol exposure, genotype, and temperature. (A) Dendrogram illustrating the hierarchical clustering of the genes and with their corresponding modules colors. The top layer of colors indicates the Dynamic Tree Cutoff, including all assigned modules, before merging by module similarity. 71 Figure 3.7, con’t: The bottom layer indicates the module colors after merging. These were the modules used for further analysis. (B) Heatmap of module-trait correlations. The eigengene for each module was correlated with temperature (room temperature), ethanol treatment (ethanol), genotype (mutant), and batch. Genotype for temperature and treatment are also shown. Intensity of the color in each cell indicates the strength of correlation between the module (row labels) and the sample trait (column labels).

72

Figure 3.8: Differences in ethanol-response based on genotype. (A) log2 fold differences for wild-type/heterozygote (wt/het) vs. mutant in response to ethanol exposure. Genes colored in red indicate a difference in log2 change >1 between wt/het and mutant. (B) Bar plot shows the percentage of total genes in a WCGNA module that were also red in (A).

73 3.6 TABLES n name p.adj 43 morphogenesis of an epithelial sheet 0.002 24 hindbrain development 0.002 612 anatomical structure morphogenesis 0.005 656 single-multicellular organism process 0.007 34 epiboly 0.010 210 tissue morphogenesis 0.015 9 regulation of retinoic acid receptor signaling pathway 0.015 301 embryonic morphogenesis 0.016 270 multicellular organismal development 0.026 20 non-canonical Wnt signaling pathway 0.026 21 eye morphogenesis 0.031 156 regionalization 0.048 202 G-protein coupled receptor signaling pathway 0.049 51 positive regulation of organelle organization 0.049 37 response to cytokine 0.049 15 embryonic eye morphogenesis 0.059 239 pattern specification process 0.085 8 positive regulation of Notch signaling pathway 0.090

Table 3.1: Significant GO terms describing biological processes for the cyan module. Rows highlighted in yellow indicate GO terms of implicated in convergent extension and eye morphogenesis. n=number of genes for each GO term

74 Chapter 4: Summary and future directions

Fetal alcohol spectrum disorders (FASD) describes a highly variable continuum of birth defects where the single known environmental cause is prenatal alcohol exposure. Not all fetuses exposed to chronic alcohol develop FASD, signifying the importance of additional predisposing factors (Abel, 1988; Eberhart and Parnell, 2016). Human twin and animal model studies provide strong evidence for genetic susceptibility. Despite this, our understanding of these genetic factors and underlying molecular mechanisms are incomplete. My work here provides important insight into early ethanol- induced alterations to the transcriptome that associate with increased risk for midfacial defects in an ethanol-sensitive mutant, vangl2. Using a genetic “shelf screen,” my lab previously identified vangl2 as an ethanol- sensitive locus, where ethanol-treated mutants have fully penetrant eye fusions (Swartz et al., 2014). In chapter 2, I extended this finding and showed that vangl2 mutants display fully penetrant eye fusions when exposed to ethanol at the initiation of gastrulation. I further uncovered evidence for haploinsufficiency, where a significant number of ethanol-exposed heterozygotes display synophthalmia when exposed to ethanol prior to the initiation of zygotic transcription. Since numerous studies have already elucidated the role of vangl2, I went on to characterize the effects of ethanol during stages where vangl2 embryos are sensitive to the deleterious effects of ethanol. I designed a single-embryo RNA-seq experiment using a subteratogenic dose of ethanol at 3 developmental stages. My experimental design differed from previous studies that use a much higher dose of ethanol known to cause developmental anomalies (Bilotta et al., 2004; Kashyap et al., 2014). Using a subteratogenic dose of ethanol at biologically relevant tissue concentrations, my RNA- 75 seq analyses demonstrated a subtle effect of ethanol on the transcriptome (Lovely et al., 2014). Samples primarily clustered by developmental age and within a given time, I detected individual variation between samples. Across all 3 developmental times, ethanol predominately upregulated gene expression. I did not detect a broad attenuation of the Wnt/PCP pathway at the transcriptional level. The differentially expressed genes that mapped to GO terms associated with increased oxidative stress and decreased embryonic development and morphogenesis. Using a bioinformatic approach, I found that the Shh pathway inhibitor, cyclopamine, which also induces cyclopia, shares a similar gene expression profile with ethanol. Despite this finding, ethanol did not lead to downregulation of Shh target genes. Instead, I showed that ethanol significantly reduces convergent extension of tissues that express Shh signaling. The mechanism of this synergistic interaction is due to mispositioning of the Shh signal, which is necessary to separate the eye fields. In chapter 3, I extended the RNA-seq analyses to examine the synergistic effects of ethanol on vangl2. Sub-optimal cold temperatures produce a similar phenotypic outcome as ethanol exposure in vangl2 mutants. I included this treatment group in my analyses to identify transcriptional differences between these groups. I eliminated the developmental time variable by collecting embryos at a single stage (10 hpf) before eye field separation. With these alterations to the RNA-seq experimental design, I found that ethanol is the main driver of variation in the dataset. Embryos raised at cold temperatures clustered together and away from ethanol-treated embryos raised at the optimal incubation temperature, suggesting inherent differences on the effects of these experimental variables. I found many differentially expressed genes in response to ethanol and temperature but a limited number in response to genotype. Furthermore, I did not detect significant interaction between ethanol and genotype. Despite this, I found 76 several modules that correlate with ethanol and genotype using WGCNA. Some of these modules are enriched for genes implicated in convergent extension and eye field development. Two genes that are implicated in the etiology of holoprosencephaly, shh and six3, were significantly downregulated in ethanol-treated mutants. Together, my work suggests a subteratogenic dose of ethanol reduces convergent extension, which is further exacerbated in the sensitized vangl2 background. These mesodermal defects affect the localization of eye field markers and shh, which is critical for eye field separation. Downstream readouts of Wnt/PCP activity are tissue dependent (Mlodzik, 2002; Veeman et al., 2003). Ethanol does not appear to broadly alter expression of the Wnt/PCP pathway at the transcriptional level. Since the ethanol- induced vangl2 mutant phenotype phenocopies midfacial defects in other vangl2 compound mutants that fail to separate the eye field, I hypothesize that ethanol broadly attenuates the Wnt/PCP pathway at the post-transcriptional level.

4.1 FUTURE DIRECTIONS

4.1.1 microRNAs and ethanol

One potential mechanism for post-transcriptional alterations are microRNAs (miRNAs), highly conserved small non-coding RNA molecules. These small molecules are about 22 nucleotides in length and can alter protein levels via translation inhibition (Bartel, 2004). Due to their small size and ability to regulate hundreds of genes, their effects are pervasive. It is estimated that nearly 30% of protein-coding genes in the human genome are regulated by miRNAs (Li et al., 2009b). These small RNAs are initially transcribed by RNA polymerase II or III as double-stranded transcripts (i.e. pri- miRNA) with a stem-loop, flanked by single-stranded segments (Miranda, 2014). The enzymes Drosha and DGCR8 process the pri-miRNA into pre-miRNA, a shorter stem- 77 loop structure (Miranda, 2014). Dicer, an endonuclease, further processes the pre-miRNA into a mature RNA duplex. The double-stranded duplex is unwound by helicase and one strand is preferentially bound to the RNA-induced silencing complex (RISC) (Miranda, 2014). A complementary region of the miRNA known as the “seed region,” directs and allows for specificity to target mRNAs. Previous work in zebrafish shows 1% and 1.5% ethanol upregulates expression of several miRNAs (Soares et al., 2012). Many of these miRNAs are predicted to target Wnt/PCP pathway members (Table 4.1). Furthermore, injection of a dicer morpholino rescues eye fusions in ethanol-treated vangl2 mutants [(n=18/19), suggesting an upregulated miRNA is involved in the vangl2-ethanol interaction (Figure 4.1). For future experiments, I would first validate the microarray data from Soares et al. using Taqman qRT-PCR. If these miRNAs are mechanistically involved in ethanol teratogenesis, then loss- and gain-of-function, using an inhibitor and mimic, should block and recapitulate the effects of ethanol, respectively. Furthermore, I could assay localization of these miRNAs in mesodermal regions that undergo convergent extension using locked-nucleic acid detection riboprobes. I could then quantify protein levels of predicted Wnt/PCP targets using immunoblots.

4.1.2 Protein localization and levels

The asymmetric localization of core Wnt/PCP components on the cell membrane is a hallmark of proper Wnt/PCP signaling. Disruption of a single component can destabilize the entire asymmetrical complex (Devenport, 2014). With the advent of CRISPR/Cas9 technology, we now have the ability to endogenously knock-in genes by harnessing the power of homology-directed repair (HDR). This technique has successfully been used in zebrafish and my lab is currently testing these methods (Auer et

78 al., 2014). I could knock-in fluorescently tagged Wnt/PCP components to not only assess protein localization but also relative levels after ethanol exposure.

79 4.2 TABLES microRNA predicted targets miR-153 ptk7 miR-216a celsr3, dvl1, fat1, fzd3, fzd7, intu, prickle2, ptk7, vangl1, vangl2 miR-731 gpc4, vangl2 miR-183 ptk7, vangl2 miR-130b fuz, fzd2, fzd3, prickle1, ptk7 miR-107 dchs1, fzd3, intu, prickle2 , ror2, vangl1, vangl2 miR-203 ptk7, wnt11

Table 4.1: microRNAs upregulated by ethanol are predicted to target Wnt/PCP pathway members. List of microRNAs upregulated by 1 and 1.5% ethanol adapted from Soares et al., 2012. Predicted targets were obtained from the miRNA target prediction algorithm, TargetScan6.2.

80 4.3 FIGURES 50 40 30 20 10 % % mutants fused 0

Control EtOH (6-30 hpf) Dicer MO Dicer MO + EtOH

treatments

Figure 4.1: Preliminary data suggests dicer morpholino (MO) rescues cyclopia in ethanol-treated vangl2 mutants. Embryos injected with 1mg/mL dicer MO at 2-4 cell stage and subsequently treated with 1% ethanol from 6 – 30 hpf.

81 Appendix: List of Acronyms and Abbreviations

PAE: Prenatal Alcohol Exposure FASD: Fetal Alcohol Spectrum Disorders FAS: Fetal Alcohol Syndrome pFAS: Partial Fetal Alcohol Syndrome ARND: Alcohol Related Neurodevelopmental Disorder ND-PAE: Neurobehavioral Disorder Associated with Prenatal Alcohol Exposure ADH: Alcohol dehydrogenase ALDH2: Aldehyde dehydrogenase 2 ROS: Reactive oxygen species HPF: Hours post fertilization DPF: Days post fertilization Wnt/PCP: Wnt planar cell polarity SHH: Sonic Hedgehog PTCH: Patched SMO: Smoothened PCM: Prechordal mesendoderm RPE: Retinal pigmented epithelium PCA: Principal component analysis DEGs: Differentially expressed genes FDR: False-discovery rate GO: Gene ontology WGCNA: Weighted Gene Co-expression Network Analysis qRT-PCR: Quantitative real-time PCR

82 LINCS: Library of Integrated Network-Based Cellular Signatures CMap: Connectivity Map AP: Anterior-posterior ER: Endoplasmic reticulum UPR: Unfolded protein response ROS: Reactive oxygen species

83 Bibliography

Abel, E.L., 1988. Fetal alcohol syndrome in families. Neurotoxicol Teratol 10, 1-2.

Adelmann, H.B., 1929. Experimental studies on the development of the eye. I. The effect of the removal of median and lateral areas of the anterior end of the urodelan neural plate on the development of the eyes (Triton teniatus and Amblystoma punctatum). Journal of Experimental Zoology 54, 249 - 290.

Adler, P.N., 1992. The Genetic Control of Tissue Polarity in Drosophila. BioEssays 14, 735-741.

Ahlgren, S.C., Thakur, V., Bronner-Fraser, M., 2002. Sonic hedgehog rescues cranial neural crest from cell death induced by ethanol exposure. Proc Natl Acad Sci U S A 99, 10476-10481.

Aken, B.L., Achuthan, P., Akanni, W., Amode, M.R., Bernsdorff, F., Bhai, J., Billis, K., Carvalho-Silva, D., Cummins, C., Clapham, P., Gil, L., Giron, C.G., Gordon, L., Hourlier, T., Hunt, S.E., Janacek, S.H., Juettemann, T., Keenan, S., Laird, M.R., Lavidas, I., Maurel, T., McLaren, W., Moore, B., Murphy, D.N., Nag, R., Newman, V., Nuhn, M., Ong, C.K., Parker, A., Patricio, M., Riat, H.S., Sheppard, D., Sparrow, H., Taylor, K., Thormann, A., Vullo, A., Walts, B., Wilder, S.P., Zadissa, A., Kostadima, M., Martin, F.J., Muffato, M., Perry, E., Ruffier, M., Staines, D.M., Trevanion, S.J., Cunningham, F., Yates, A., Zerbino, D.R., Flicek, P., 2017. Ensembl 2017. Nucleic Acids Res 45, D635- D642.

Anders, S., Pyl, P.T., Huber, W., 2015. HTSeq--a Python framework to work with high- throughput sequencing data. Bioinformatics 31, 166-169.

Auer, T.O., Duroure, K., De Cian, A., Concordet, J.P., Del Bene, F., 2014. Highly efficient CRISPR/Cas9-mediated knock-in in zebrafish by homology-independent DNA repair. Genome Res 24, 142-153.

Bailey, T.J., El-Hodiri, H., Zhang, L., Shah, R., Mathers, P.H., Jamrich, M., 2004. Regulation of vertebrate eye development by Rx genes. Int J Dev Biol 48, 761-770.

Bartel, D.P., 2004. MicroRNAs: genomics, biogenesis, mechanism, and function. Cell 116, 281-297.

Benjamini, Y., Hochberg, Y., 1995. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society. Series B (Methodological) 57, 289-300.

84 Berres, M.E., Garic, A., Flentke, G.R., Smith, S.M., 2017. Transcriptome Profiling Identifies Ribosome Biogenesis as a Target of Alcohol Teratogenicity and Vulnerability during Early Embryogenesis. PLoS One 12, e0169351.

Bhatia, S., Drake, D.M., Miller, L., Wells, P.G., 2019. Oxidative stress and DNA damage in the mechanism of fetal alcohol spectrum disorders. Birth Defects Research.

Bilotta, J., Barnett, J.A., Hancock, L., Saszik, S., 2004. Ethanol exposure alters zebrafish development: a novel model of fetal alcohol syndrome. Neurotoxicol Teratol 26, 737- 743.

Blader, P., Strahle, U., 1998. Ethanol Impairs Migration of the Prechordal Plate in the Zebrafish Embryo. Developmental Biology 201, 185-201.

Bönnemann, C., Meinecke, P., 1990. Holoprosencephaly as a Possibly Embyronic Alcohol Effect: Another Observation. American Journal of Medical Genetics 37, 431- 432.

Bravo, R., Parra, V., Gatica, D., Rodriguez, A.E., Torrealba, N., Paredes, F., Wang, Z.V., Zorzano, A., Hill, J.A., Jaimovich, E., Quest, A.F., Lavandero, S., 2013. Endoplasmic reticulum and the unfolded protein response: dynamics and metabolic integration. Int Rev Cell Mol Biol 301, 215-290.

Chiang, C., Litingtung, Y., Lee, E., Young, K.E., Corden, J.L., Westphal, H., Beachy, P.A., 1996. Cyclopia and defective axial patterning in mice lacking Sonic hedgehog gene function. Nature 383, 407-413.

Chitwood, P.J., Hegde, R.S., 2019. The Role of EMC during Membrane Protein Biogenesis. Trends Cell Biol 29, 371-384.

Chuang, J.C., Raymond, P.A., 2001. Zebrafish genes rx1 and rx2 help define the region of forebrain that gives rise to retina. Dev Biol 231, 13-30.

Clevers, H., 2006. Wnt/beta-catenin signaling in development and disease. Cell 127, 469- 480.

Cohen, M.M., Sulik, K.K., 1992. Perspectives on holoprosencephaly: Part II. Central nervous system, craniofacial anatomy, syndrome commentary, diagnostic approach, and experimental studies. Journal of craniofacial genetics and developmental biology 12, 196- 244.

Curtin, J.A., Quint, E., Tsipouri, V., Arkell, R.M., Cattanach, B., Copp, A.J., Henderson, D.J., Spurr, N., Stanier, P., Fisher, E.M., Nolan, P.M., Steel, K.P., Brown, S.D.M., Gray, I.C., Murdoch, J.N., 2003. Mutation of Celsr1 Disrupts Planar Polarity of Inner Ear Hair

85 Cells and Causes Severe Neural Tube Defects in the Mouse. Current Biology 13, 1129- 1133.

Czeizel, A.E., 2005. Birth defects are preventable. Int J Med Sci 2, 91-92.

Da Lee, R., Rhee, G.S., An, S.M., Kim, S.S., Kwack, S.J., Seok, J.H., Chae, S.Y., Park, C.H., Yoon, H.J., Cho, D.H., Kim, H.S., Park, K.L., 2004. Differential gene profiles in developing embryo and fetus after in utero exposure to ethanol. J Toxicol Environ Health A 67, 2073-2084.

Denny, C.H., Acero, C.S., Naimi, T.S., Kim, S.Y., 2019. Consumption of Alcohol Beverages and Binge Drinking Among Pregnant Women Aged 18–44 Years — United States, 2015–2017. MMWR Morb Mortal Wkly Rep 68, 365-368.

Devenport, D., 2014. The cell biology of planar cell polarity. J Cell Biol 207, 171-179.

Dixon, G.B., Davies, S.W., Aglyamova, G.V., Meyer, E., Bay, L.K., Matz, M.V., 2015. Genomic determinants of coral heat tolerance across latitudes. Science 348

Dobin, A., Davis, C.A., Schlesinger, F., Drenkow, J., Zaleski, C., Jha, S., Batut, P., Chaisson, M., Gingeras, T.R., 2013. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15-21.

Downing, C., Balderrama-Durbin, C., Broncucia, H., Gilliam, D., Johnson, T.E., 2009. Ethanol teratogenesis in five inbred strains of mice. Alcohol Clin Exp Res 33, 1238-1245.

Eberhart, J.K., Parnell, S.E., 2016. The Genetics of Fetal Alcohol Spectrum Disorders. Alcohol Clin Exp Res 40, 1154-1165.

Eberhart, J.K., Swartz, M.E., Crump, J.G., Kimmel, C.B., 2006. Early Hedgehog signaling from neural to oral epithelium organizes anterior craniofacial development. Development 133, 1069-1077.

Farrell, J.A., Wang, Y., Riesenfeld, S.J., Shekhar, K., Regev, A., Schier, A.F., 2018. Single-cell reconstruction of developmental trajectories during zebrafish embryogenesis. Science 360.

Flentke, G.R., Smith, S.M., 2018. The avian embryo as a model for fetal alcohol spectrum disorder. Biochem Cell Biol 96, 98-106.

G. I. Henderson, B.G.D., A. Perez, and S. Schenker, 1995. In Utero Ethanol Exposure Elicits Oxidative Stress in the Rat Fetus. Alcohol Clm Exp Res 19, 714-720.

86 Garic, A., Berres, M.E., Smith, S.M., 2014. High-throughput transcriptome sequencing identifies candidate genetic modifiers of vulnerability to fetal alcohol spectrum disorders. Alcohol Clin Exp Res 38, 1874-1882.

Glickman, N.S., Kimmel, C.B., Jones, M.A., Adams, R.J., 2003. Shaping the zebrafish notochord. Development 130, 873-887.

Goldstone, J.V., McArthur, A.G., Kubota, A., Zanette, J., Parente, T., Jonsson, M.E., Nelson, D.R., Stegeman, J.J., 2010. Identification and developmental expression of the full complement of Cytochrome P450 genes in Zebrafish. BMC Genomics 11, 643.

Graw, J., 2010. Eye Development, Organogenesis in Development, pp. 343-386.

Gray, R.S., Roszko, I., Solnica-Krezel, L., 2011. Planar cell polarity: coordinating morphogenetic cell behaviors with embryonic polarity. Dev Cell 21, 120-133.

Green, M.L., Singh, A.V., Zhang, Y., Nemeth, K.A., Sulik, K.K., Knudsen, T.B., 2007. Reprogramming of genetic networks during initiation of the Fetal Alcohol Syndrome. Dev Dyn 236, 613-631.

Green, R.F., Stoler, J.M., 2007. Alcohol dehydrogenase 1B genotype and fetal alcohol syndrome: a HuGE minireview. Am J Obstet Gynecol 197, 12-25.

Gubb, D., 1993. Genes controlling cellular polarity in Drosophila. Development, 269- 277.

Gubb, D., Garcia-Bellido, A., 1982. A genetic analysis of the determination of cuticular polarity during development in Drosophila melanogaster. J. Embryol. exp. Morph 68, 37-57.

Harada, M., 1995. Minamata Disease: Methylmercury Poisoning in Japan Caused by Environmental Pollution. Critical Reviews in Toxicology 25, 1-24.

Harris, R.A., Trudell, J.R., Mihic, S.J., 2008. Ethanol's molecular targets. Sci Signal 1, re7.

Heisenberg, C.-P., Tada, M., Rauch, G.-J.r., Saude, L., Concha, M.L., Geisler, R., Stemple, D.L., Smith, J.C., Wilson, S.W., 2000. Silberblick/Wnt11 mediates convergent extension movements during zebrafish gastrulation. Nature 405, 76-81.

Hong, M., Krauss, R.S., 2012. Cdon mutation and fetal ethanol exposure synergize to produce midline signaling defects and holoprosencephaly spectrum disorders in mice. PLoS Genet 8, e1002999.

87 Hong, M., Krauss, R.S., 2013. Rescue of holoprosencephaly in fetal alcohol-exposed Cdon mutant mice by reduced gene dosage of Ptch1. PLoS One 8, e79269.

Hong, M., Krauss, R.S., 2017. Ethanol itself is a holoprosencephaly-inducing teratogen. PLoS One 12, e0176440.

Howe, K., Clark, M.D., Torroja, C.F., Torrance, J., Berthelot, C., Muffato, M., Collins, J.E., Humphray, S., McLaren, K., Matthews, L., McLaren, S., Sealy, I., Caccamo, M., Churcher, C., Scott, C., Barrett, J.C., Koch, R., Rauch, G.J., White, S., Chow, W., Kilian, B., Quintais, L.T., Guerra-Assuncao, J.A., Zhou, Y., Gu, Y., Yen, J., Vogel, J.H., Eyre, T., Redmond, S., Banerjee, R., Chi, J., Fu, B., Langley, E., Maguire, S.F., Laird, G.K., Lloyd, D., Kenyon, E., Donaldson, S., Sehra, H., Almeida-King, J., Loveland, J., Trevanion, S., Jones, M., Quail, M., Willey, D., Hunt, A., Burton, J., Sims, S., McLay, K., Plumb, B., Davis, J., Clee, C., Oliver, K., Clark, R., Riddle, C., Elliot, D., Threadgold, G., Harden, G., Ware, D., Begum, S., Mortimore, B., Kerry, G., Heath, P., Phillimore, B., Tracey, A., Corby, N., Dunn, M., Johnson, C., Wood, J., Clark, S., Pelan, S., Griffiths, G., Smith, M., Glithero, R., Howden, P., Barker, N., Lloyd, C., Stevens, C., Harley, J., Holt, K., Panagiotidis, G., Lovell, J., Beasley, H., Henderson, C., Gordon, D., Auger, K., Wright, D., Collins, J., Raisen, C., Dyer, L., Leung, K., Robertson, L., Ambridge, K., Leongamornlert, D., McGuire, S., Gilderthorp, R., Griffiths, C., Manthravadi, D., Nichol, S., Barker, G., Whitehead, S., Kay, M., Brown, J., Murnane, C., Gray, E., Humphries, M., Sycamore, N., Barker, D., Saunders, D., Wallis, J., Babbage, A., Hammond, S., Mashreghi-Mohammadi, M., Barr, L., Martin, S., Wray, P., Ellington, A., Matthews, N., Ellwood, M., Woodmansey, R., Clark, G., Cooper, J., Tromans, A., Grafham, D., Skuce, C., Pandian, R., Andrews, R., Harrison, E., Kimberley, A., Garnett, J., Fosker, N., Hall, R., Garner, P., Kelly, D., Bird, C., Palmer, S., Gehring, I., Berger, A., Dooley, C.M., Ersan-Urun, Z., Eser, C., Geiger, H., Geisler, M., Karotki, L., Kirn, A., Konantz, J., Konantz, M., Oberlander, M., Rudolph-Geiger, S., Teucke, M., Lanz, C., Raddatz, G., Osoegawa, K., Zhu, B., Rapp, A., Widaa, S., Langford, C., Yang, F., Schuster, S.C., Carter, N.P., Harrow, J., Ning, Z., Herrero, J., Searle, S.M., Enright, A., Geisler, R., Plasterk, R.H., Lee, C., Westerfield, M., de Jong, P.J., Zon, L.I., Postlethwait, J.H., Nusslein-Volhard, C., Hubbard, T.J., Roest Crollius, H., Rogers, J., Stemple, D.L., 2013. The zebrafish reference genome sequence and its relationship to the human genome. Nature 496, 498-503.

Hoyme, H.E., Kalberg, W.O., Elliott, A.J., Blankenship, J., Buckley, D., Marais, A.-S., Manning, M.A., Robinson, L.K., Adam, M.P., Abdul-Rahman, O., Jewett, T., Coles, C.D., Chambers, C., Jones, K.L., Adnams, C.M., Shah, P.E., Riley, E.P., Charness, M.E., Warren, K.R., May, P.A., 2016. Updated Clinical Guidelines for Diagnosing Fetal Alcohol Spectrum Disorders. Pediatrics 138.

Hoyme, H.E., May, P.A., Kalberg, W.O., Kodituwakku, P., Gossage, J.P., Trujillo, P.M., Buckley, D.G., Miller, J.H., Aragon, A.S., Khaole, N., Viljoen, D.L., Jones, K.L., Robinson, L.K., 2005. A Practical Clinical Approach to Diagnosis of Fetal Alcohol 88 Spectrum Disorders: Clarification of the 1996 Institute of Medicine Criteria. Pediatrics 115, 39–47.

Hsu, H.-J., Liang, M.-R., Chen, C.-T., Chung, B.-c., 2006. Pregnenolone stabilizes microtubules and promotes zebrafish embryonic cell movement. Nature 439, 480-483.

Hu, D., Marcucio, R.S., 2009. A SHH-responsive signaling center in the forebrain regulates craniofacial morphogenesis via the facial ectoderm. Development 136, 107-116.

Ingham, P.W., Placzek, M., 2006. Orchestrating ontogenesis: variations on a theme by sonic hedgehog. Nat Rev Genet 7, 841-850.

Jessen, J.R., Topczewski, J., Bingham, S., Sepich, D.S., Marlow, F., Chandrasekhar, A., Solnica-Krezel, L., 2002. Zebrafish trilobite identifies new roles for Strabismus in gastrulation and neuronal movements. Nat Cell Biol 4, 610-615.

Johnson, C.Y., Rasmussen, S.A., 2010. Non-genetic risk factors for holoprosencephaly. Am J Med Genet C Semin Med Genet 154C, 73-85.

Johnston, M.C., Bronsky, P.T., 2002. Chapter 5: Craniofacial Embryogenesis: Abnormal Developmental Mechanisms.

Jones, K.L., Smith, D.W., 1973. Recognition of the Fetal Alcohol Syndrome in Early Infancy. The Lancet, 999-1001.

Jones, K.L., Smith, D.W., Ulleland, C.N., Streissguth, P., 1973. Pattern of malformation in offspring of chronic alcoholic mothers. Lancet 1, 1267-1271.

Joya, X., Garcia-Algar, O., Vall, O., Pujades, C., 2014. Transient exposure to ethanol during zebrafish embryogenesis results in defects in neuronal differentiation: an alternative model system to study FASD. PLoS One 9, e112851.

Kashyap, B., Pegorsch, L., Frey, R.A., Sun, C., Shelden, E.A., Stenkamp, D.L., 2014. Eye-specific gene expression following embryonic ethanol exposure in zebrafish: roles for heat shock factor 1. Reprod Toxicol 43, 111-124.

Kiecker, C., Niehrs, C., 2001. The role of prechordal mesendoderm in neural patterning. Current Opinion in Neurobiology 11, 27–33.

Kietzman, H.W., Everson, J.L., Sulik, K.K., Lipinski, R.J., 2014. The teratogenic effects of prenatal ethanol exposure are exacerbated by Sonic Hedgehog or GLI2 haploinsufficiency in the mouse. PLoS One 9, e89448.

Kim, J.H., Scialli, A.R., 2011. Thalidomide: the tragedy of birth defects and the effective treatment of disease. Toxicol Sci 122, 1-6. 89 Kim, S.-H., Park, H.-C., Sang-Yeo’, Yeo, Hong, S.-K., Choi, J.-W., Kim, C.-H., Weinstein, B.M., Huh, T.-L., 1998. Characterization of two frizzled8 homologues expressed in the embryonic shield and prechordal plate of zebrafish embryos. Mechanisms of Development 78, 193–198.

Kimmel, C.B., Ballard, W.W., Kimmel, S.R., Ullman, B., Schilling, T.F., 1995. Stages of Embryonic Development of the Zebrafish. Developmental Dynamics 203, 253-310.

Klein, T.J., Mlodzik, M., 2005. Planar Cell Polarization: An Emerging Model Points in the Right Direction. Annu Rev Cell Dev Biol 21, 155-176.

Krauss, R.S., 2007. Holoprosencephaly: new models, new insights. Expert Rev Mol Med 9, 1-17.

Krauss, S., Johansen, T., Korzh, V., Fjose, A., 1991. Expression of the zebrafish paired box gene pax[zf-b] during early neurogenesis. Development 113, 1193-1206.

Lamb, J., Crawford, E.D., Peck, D., Modell, J.W., Blat, I.C., Wrobel, M.J., Lerner, J., Brunet, J.-P., Subramanian, A., Ross, K.N., Reich, M., Hieronymus, H., Wei, G., Armstrong, S.A., Haggarty, S.J., Clemons, P.A., Wei, R., Carr, S.A., Lander, E.S., Golub, T.R., 2006. The Connectivity Map: Using Gene-Expression Signatures to Connect Small Molecules, Genes, and Disease. Science 313.

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

Lawrence, P.A., Shelton, P.M.J., 1975. The determination of polarity in the developing insect retina. J. Embryol. exp. Morph 33, 471-486.

Leek, J.T., Storey, J.D., 2005. Capturing Heterogeneity in Gene Expression Studies by "Surrogate Variable Analysis". PLoS Genetics preprint.

Lemoine, P., Harousseau, H., Borteyru, J.P., Menuet, J.C., 1968. Children of Alcoholic Parents—Observed Anomalies: Discussion of 127 Cases. Ouest Med-ical 21, 476–482.

Li, H., Handsaker, B., Wysoker, A., Fennell, T., Ruan, J., Homer, N., Marth, G., Abecasis, G., Durbin, R., Genome Project Data Processing, S., 2009a. The Sequence Alignment/Map format and SAMtools. Bioinformatics 25, 2078-2079.

Li, H., Tierney, C., Wen, L., Wu, J.Y., Rao, Y., 1997. A single morphogenetic field gives rise to two retina primordia under the influence of the prechordal plate. Development 124, 603-615.

90 Li, M., Marin-Muller, C., Bharadwaj, U., Chow, K.H., Yao, Q., Chen, C., 2009b. MicroRNAs: control and loss of control in human physiology and disease. World J Surg 33, 667-684.

Li, S., Esterberg, R., Lachance, V., Ren, D., Radde-Gallwitz, K., Chi, F., Parent, J.L., Fritz, A., Chen, P., 2011. Rack1 is required for Vangl2 membrane localization and planar cell polarity signaling while attenuating canonical Wnt activity. Proc Natl Acad Sci U S A 108, 2264-2269.

Logan, C.Y., Nusse, R., 2004. The Wnt Signaling Pathway in Development and Disease. Annu Rev Cell Dev Biol 20, 781-810.

Loucks, E., Carvan, M.J., 3rd, 2004. Strain-dependent effects of developmental ethanol exposure in zebrafish. Neurotoxicol Teratol 26, 745-755.

Loucks, E.J., Ahlgren, S.C., 2009. Deciphering the role of Shh signaling in axial defects produced by ethanol exposure. Birth Defects Res A Clin Mol Teratol 85, 556-567.

Love, M.I., Huber, W., Anders, S., 2014. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 15, 550.

Lovely, C.B., Fernandes, Y., Eberhart, J.K., 2016. Fishing for Fetal Alcohol Spectrum Disorders: Zebrafish as a Model for Ethanol Teratogenesis. Zebrafish 13, 391-398.

Lovely, C.B., Nobles, R.D., Eberhart, J.K., 2014. Developmental age strengthens barriers to ethanol accumulation in zebrafish. Alcohol 48, 595-602.

Marcucio, R.S., Cordero, D.R., Hu, D., Helms, J.A., 2005. Molecular interactions coordinating the development of the forebrain and face. Dev Biol 284, 48-61.

Marlow, F., Zwartkruis, F., Malicki, J., Neuhauss, S.C.F., Abbas, L., Weaver, M., Driever, W., Solnica-Krezel, L., 1998. Functional Interactions of Genes Mediating Convergent Extension, knypek and trilobite, during the Partitioning of the Eye Primordium in Zebrafish. Developmental Biology 203, 382-399.

Martin, M., 2011. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.journal 17, 10-12.

Martinez-Morales, J.R., Wittbrodt, J., 2009. Shaping the vertebrate eye. Curr Opin Genet Dev 19, 511-517.

Mathers, P.H., Grinberg, A., Mahon, K.A., Jamrich, M., 1997. The Rx homeobox gene is essential for vertebrate eye development. Nature 387, 603-607.

91 Matsuda, N., Mishina, M., 2004. Identification of chaperonin CCT gamma subunit as a determinant of retinotectal development by whole-genome subtraction cloning from zebrafish no tectal neuron mutant. Development 131, 1913-1925.

May, P.A., Gossage, J.P., Kalberg, W.O., Robinson, L.K., Buckley, D., Manning, M., Hoyme, H.E., 2009. Prevalence and Epidemiologic Characteristics of FASD from Various Research Methods with an Emphasis on Recent Inschool Studies Developmental Disabilities 15, 176-192.

McCarthy, N., Eberhart, J.K., 2014. Gene-ethanol interactions underlying fetal alcohol spectrum disorders. Cell Mol Life Sci 71, 2699-2706.

McCarthy, N., Sidik, A., Bertrand, J.Y., Eberhart, J.K., 2016. An Fgf-Shh signaling hierarchy regulates early specification of the zebrafish skull. Dev Biol 415, 261-277.

McCarthy, N., Wetherill, L., Lovely, C.B., Swartz, M.E., Foroud, T.M., Eberhart, J.K., 2013. Pdgfra protects against ethanol-induced craniofacial defects in a zebrafish model of FASD. Development 140, 3254-3265.

Mentink, R.A., Rella, L., Radaszkiewicz, T.W., Gybel, T., Betist, M.C., Bryja, V., Korswagen, H.C., 2018. The planar cell polarity protein VANG-1/Vangl negatively regulates Wnt/beta-catenin signaling through a Dvl dependent mechanism. PLoS Genet 14, e1007840.

Miranda, R.C., 2014. MicroRNAs and ethanol toxicity. Int Rev Neurobiol 115, 245-284.

Mlodzik, M., 2002. Planar cell polarization: do the same mechanisms regulate Drosophila tissue polarity and vertebrate gastrulation? Trends Genet. 11, 564-571.

Moore, E.S., Ward, R.E., Jamison, P.L., Morris, C.A., Bader, P.I., Hall, B.D., 2002. New perspectives on the face in fetal alcohol syndrome: what anthropometry tells us. Am J Med Genet 109, 249-260.

Muenke, M., Beachy, P.A., 2000. Genetics of ventral forebrain development and holoprosencephaly. Curr Opin Genet Dev 10, 262-269.

Muggli, E., Matthews, H., Penington, A., Claes, P., O’Leary, C., Forster, D., Donath, S., Anderson, P.J., Lewis, S., Nagle, C., Craig, J.M., White, S.M., Elliott, E.J., Halliday, J., 2017. Association Between Prenatal Alcohol Exposure and Craniofacial Shape of Children at 12 Months of Age. JAMA Pediatrics 171.

Nubler-Jung, K., 1987. Insect epidermis: disturbance of supracellular tissue polarity does not prevent the expression of cell polarity. Roux’s Arch Dev Biol 196, 286-289.

92 Nubler-Jung, K., Bonitz, R., Sonnenschein, M., 1987. Cell polarity during wound healing in an insect epidermis. Development 100, 163-170.

O'Leary-Moore, S.K., Parnell, S.E., Lipinski, R.J., Sulik, K.K., 2011. Magnetic resonance-based imaging in animal models of fetal alcohol spectrum disorder. Neuropsychol Rev 21, 167-185.

Ossipova, O., Kim, K., Sokol, S.Y., 2015. Planar polarization of Vangl2 in the vertebrate neural plate is controlled by Wnt and Myosin II signaling. Biol Open 4, 722-730.

Park, M., Moon, R.T., 2002. The planar cell-polarity gene stbm regulates cell behaviour and cell fate in vertebrate embryos. Nat Cell Biol 4, 20-25.

Park, T.J., Haigo, S.L., Wallingford, J.B., 2006. Ciliogenesis defects in embryos lacking inturned or fuzzy function are associated with failure of planar cell polarity and Hedgehog signaling. Nat Genet 38, 303-311.

Rauch, G.-J., Hammerschmidt, M., Blader, P., Schauerte, H.E., Strahle, U., Ingham, P.W., McMahon, A.P., Haffter, P., 1997. Wnt5 is Required for Tail Formation in the Zebrafish Embryo. Cold Spring Harbor Symposia on Quantitative Biology LXII.

Roessler, E., Hu, P., Muenke, M., 2018. Holoprosencephaly in the genomics era. Am J Med Genet C Semin Med Genet 178, 165-174.

Ronen, G.M., Andrews, W.L., 1991. Holoprosencephaly as a Possible Embryonic Alcohol Effect. American Journal of Medical Genetics 40, 151-154.

Santos-Ledo, A., Cavodeassi, F., Carreno, H., Aijon, J., Arevalo, R., 2013. Ethanol alters gene expression and cell organization during optic vesicle evagination. Neuroscience 250, 493-506.

Sapir, T., Levy, T., Sakakibara, A., Rabinkov, A., Miyata, T., Reiner, O., 2013. Shootin1 acts in concert with KIF20B to promote polarization of migrating neurons. J Neurosci 33, 11932-11948.

Saúde, L., Woolley, K., Martin, P., Driever, W., Stemple, D.L., 2000. Axis-inducing activities and cell fates of the zebrafish organizer. Development 127, 3407-3417.

Schlessinger, K., Hall, A., Tolwinski, N., 2009. Wnt signaling pathways meet Rho GTPases. Genes Dev 23, 265-277.

Seifert, J.R., Mlodzik, M., 2007. Frizzled/PCP signalling: a conserved mechanism regulating cell polarity and directed motility. Nat Rev Genet 8, 126-138.

93 Shepard, T.H., 1982. Detection of human teratogenic agents. The Journal of Pediatrics 101, 810-815.

Siefert, J.C., Georgescu, C., Wren, J.D., Koren, A., Sansam, C.L., 2017. DNA replication timing during development anticipates transcriptional programs and parallels enhancer activation. Genome Res 27, 1406-1416.

Smith, S.M., Garic, A., Berres, M.E., Flentke, G.R., 2014. Genomic factors that shape craniofacial outcome and neural crest vulnerability in FASD. Front Genet 5, 224.

Soares, A.R., Pereira, P.M., Ferreira, V., Reverendo, M., Simoes, J., Bezerra, A.R., Moura, G.R., Santos, M.A., 2012. Ethanol exposure induces upregulation of specific microRNAs in zebrafish embryos. Toxicol Sci 127, 18-28.

Sokol, S.Y., 2015. Spatial and temporal aspects of Wnt signaling and planar cell polarity during vertebrate embryonic development. Semin Cell Dev Biol 42, 78-85.

Solnica-Krezel, L., Sepich, D.S., 2012. Gastrulation: making and shaping germ layers. Annu Rev Cell Dev Biol 28, 687-717.

Solnica-Krezel, L., Stemple, D.L., Mountcastle-Shah, E., Rangini, Z., Neuhauss, S.C.F., Malicki, J., Schier, A.F., Stainier, D.Y.R., Zwartkruis, F., Abdelilah, S., Driever, W., 1996. Mutations affecting cell fates and cellular rearrangements during gastrulation in zebrafish. Development 123, 67-80.

Streissguth, A.P., Dehaene, P., 1993. Fetal alcohol syndrome in twins of alcoholic mothers: concordance of diagnosis and IQ. Am J Med Genet 47, 857-861.

Subramanian, A., Narayan, R., Corsello, S.M., Peck, D.D., Natoli, T.E., Lu, X., Gould, J., Davis, J.F., Tubelli, A.A., Asiedu, J.K., Lahr, D.L., Hirschman, J.E., Liu, Z., Donahue, M., Julian, B., Khan, M., Wadden, D., Smith, I.C., Lam, D., Liberzon, A., Toder, C., Bagul, M., Orzechowski, M., Enache, O.M., Piccioni, F., Johnson, S.A., Lyons, N.J., Berger, A.H., Shamji, A.F., Brooks, A.N., Vrcic, A., Flynn, C., Rosains, J., Takeda, D.Y., Hu, R., Davison, D., Lamb, J., Ardlie, K., Hogstrom, L., Greenside, P., Gray, N.S., Clemons, P.A., Silver, S., Wu, X., Zhao, W.N., Read-Button, W., Wu, X., Haggarty, S.J., Ronco, L.V., Boehm, J.S., Schreiber, S.L., Doench, J.G., Bittker, J.A., Root, D.E., Wong, B., Golub, T.R., 2017. A Next Generation Connectivity Map: L1000 Platform and the First 1,000,000 Profiles. Cell 171, 1437-1452 e1417.

Sulik, K.K., Johnston, M.C., 1983. Sequence of Developmental Alterations Following Acute Ethanol Exposure in Mice: Craniofacial Features of the Fetal Alcohol Syndrome. The American Journal of Anatomy 166, 257-269.

94 Sulik, K.K., Johnston, M.C., Webb, M.A., 1981. Fetal Alcohol Syndrome: Embryogenesis in a Mouse Model. Science 214, 936-938.

Swartz, M.E., Wells, M.B., Griffin, M., McCarthy, N., Lovely, C.B., McGurk, P., Rozacky, J., Eberhart, J.K., 2014. A screen of zebrafish mutants identifies ethanol- sensitive genetic loci. Alcohol Clin Exp Res 38, 694-703.

Tada, M., Smith, J.C., 2000. Xwnt11 is a target of Xenopus Brachyury: regulation of gastrulation movements via Dishevelled, but not through the canonical Wnt pathway. Development 127, 2227-2238.

Thomas, G., 2002. Furin at the cutting edge: from protein traffic to embryogenesis and disease. Nat Rev Mol Cell Biol 3, 753-766.

Topczewski, J., Sepich, D.S., Myers, D.C., Walker, C., Amores, A., Lele, Z., Hammerschmidt, M., Postlethwait, J., Solnica-Krezel, L., 2001. The Zebrafish Glypican Knypek Controls Cell Polarity during Gastrulation Movements of Convergent Extension. Developmental Cell 1, 251-264. van Amerongen, R., Fuerer, C., Mizutani, M., Nusse, R., 2012. Wnt5a can both activate and repress Wnt/beta-catenin signaling during mouse embryonic development. Dev Biol 369, 101-114.

Varga, Z.M., Wegner, J., Westerfield, M., 1999. Anterior movement of ventral diencephalic precursors separates the primordial eye field in the neural plate and requires cyclops. Development 126, 5533-5546.

Veeman, M.T., Axelrod, J.D., Moon, R.T., 2003. A second canon. Functions and mechanisms of beta-catenin-independent Wnt signaling. Dev Cell 5, 367-377.

Viljoen, D.L., Carr, L.G., Foroud, T.M., Brooke, L., Ramsay, M., L, T.K., 2001. Alcohol Dehydrogenase-2*2 Allele is Associated WithDecreased Prevalence of Fetal Alcohol Syndrome in theMixed-Ancestry Population of the Western CapeProvince, South Africa. Alcoholism: Clinical and Experimental Research 25, 1719-1722.

Walker, M.B., Kimmel, C.B., 2007. A two-color acid-free cartilage and bone stain for zebrafish larvae. Biotech Histochem 82, 23-28.

Walker, M.B., Miller, C.T., Coffin Talbot, J., Stock, D.W., Kimmel, C.B., 2006. Zebrafish furin mutants reveal intricacies in regulating Endothelin1 signaling in craniofacial patterning. Dev Biol 295, 194-205.

Wallingford, J.B., Harland, R.M., 2001. Xenopus Dishevelled signaling regulates both neural and mesodermal convergent extension: parallel forces elongating the body axis. Development 128, 2581-2592. 95 Wallingford, J.B., Rowning, B.A., Vogeli, K.M., Rothbacher, U., Fraser, S.E., Harland, R.M., 2000. Dishevelled controls cell polarity during Xenopus gastrulation. Nature 405, 81-85.

Warga, R.M., Kimmel, C.B., 1990. Cell movements during epiboly and gastrulation in zebrafish. Development 108, 569-580.

Webster, W.S., Lipson, A.H., Sulik, K.K., 1988. Interference With Gastrulation During the Third Week of Pregnancy as a Cause of Some Facial Abnormalities and CNS Defects. American Journal of Medical Genetics 31, 505-512.

Westfall, T.A., Brimeyer, R., Twedt, J., Gladon, J., Olberding, A., Furutani-Seiki, M., Slusarski, D.C., 2003. Wnt-5/pipetail functions in vertebrate axis formation as a negative regulator of Wnt/beta-catenin activity. J Cell Biol 162, 889-898.

White, J.J., Mazzeu, J.F., Coban-Akdemir, Z., Bayram, Y., Bahrambeigi, V., Hoischen, A., van Bon, B.W.M., Gezdirici, A., Gulec, E.Y., Ramond, F., Touraine, R., Thevenon, J., Shinawi, M., Beaver, E., Heeley, J., Hoover-Fong, J., Durmaz, C.D., Karabulut, H.G., Marzioglu-Ozdemir, E., Cayir, A., Duz, M.B., Seven, M., Price, S., Ferreira, B.M., Vianna-Morgante, A.M., Ellard, S., Parrish, A., Stals, K., Flores-Daboub, J., Jhangiani, S.N., Gibbs, R.A., Baylor-Hopkins Center for Mendelian, G., Brunner, H.G., Sutton, V.R., Lupski, J.R., Carvalho, C.M.B., 2018. WNT Signaling Perturbations Underlie the Genetic Heterogeneity of Robinow Syndrome. Am J Hum Genet 102, 27-43.

White, R.J., Collins, J.E., Sealy, I.M., Wali, N., Dooley, C.M., Digby, Z., Stemple, D.L., Murphy, D.N., Billis, K., Hourlier, T., Fullgrabe, A., Davis, M.P., Enright, A.J., Busch- Nentwich, E.M., 2017. A high-resolution mRNA expression time course of embryonic development in zebrafish. Elife 6.

Xu, J., Murphy, S.L., Kochanek, K.D., Bastian, B., Arias, E., 2018. Deaths: Final Data for 2016. Natl Vital Stat Rep. 67.

Zeng, H., Hoover, A.N., Liu, A., 2010. PCP effector gene Inturned is an important regulator of cilia formation and embryonic development in mammals. Dev Biol 339, 418- 428.

96 Vita

Alfire graduated from West Windsor Plainsboro High School South in West Windsor, New Jersey. After high school, she moved to Sewanee, Tennessee, to pursue a liberal arts education at The University of the South. Her interest in scientific research began freshman year, as a work-study student in the Sewanee Herbarium. She continued research in the Herbarium and Landscape Analysis Lab throughout her undergraduate career, where she helped catalog all the vascular plants on the Sewanee Domain, the University’s land holdings. She graduated with a B.S. in biology in 2009 and continued research in the lab as a post-baccalaureate fellow. From there she joined the lab of Dr. Daniel Promislow at the University of Georgia as a research technician, to gain more wet lab experience. She began her graduate work at the University of Texas at Austin in 2012. After her PhD, she hopes to pursue a career in clinical or translational research.

Permanent email: [email protected]

This dissertation was typed by Alfire Sidik.

97