Understanding the Role of Transcription Factor Regulation of Development

Understanding The Role Of Transcription Factor Regulation Of Development

Dissertation

Presented in Partial Fulfillment of the Requirements

for the Degree Doctor of Philosophy

in the Graduate School of The Ohio State University

Allison Marie Webb Chasser

The Ohio State Biochemistry Graduate Program

The Ohio State University

2019

Dissertation Committee

Dr. Helen Chamberlin, Advisor

Dr. Jane Jackman, Program Advisor

Dr. Harold Fisk

Dr. Kotaro Nakanishi

2019 Abstract

Coordinated development is the process of directing cell growth, differentiation, and proliferation. The initiation and integration of these varied events often begins directly at the level of transcription in order to regulate which gene products are available to perform temporally and spatially restricted functions. This regulation requires transcription factors to activate genes necessary for developmental outcomes and future cell phenotypic programs. In this work, I focus on the transcription factors

EGL-38/PAX and PAL-1/HOX and their involvement in cellular regulatory networks.

Pax proteins represent a dynamic method of regulation, as they can act in different cell types to perform varied functions which are all necessary for the directed organization of a single system. Hox genes, in comparison, represent the more stable gene regulatory network, which acts in singular cells to perform specific functions necessary for establishing a body plan. EGL-38/PAX and PAL-1/HOX reflect these paradigms in Caenorhabditis elegans. I have investigated how EGL-38/PAX can act in neighboring, distinct cell types of the developing egg-laying system to coordinate its own expression through the Epidermal Growth Factor signaling pathway as well as activate a cellular program of genes required for egg-laying behavioral output. I have examined how PAL-1/HOX can control expression of a range of transcription factors necessary for cell differentiation, informed by the predictions of two separate computational models of the PAL-1 regulatory network. Altogether, my results inform

i the larger knowledge of how Pax and Hox genes function, dynamically and statically, in transcriptional activation and gene regulatory networks during development.

Acknowledgements

I would like to thank Dr. Helen Chamberlin for her mentorship in graduate school, and to thank my committee members for their participation and their patience.

I would like to thank all of my worm group people, particularly my best friends Dr. Brittany Suggs, Dr. Karley Mahalak, and Kristin Balmert, without whom I would not have made it through graduate school. I want to thank Marcos Corchado, Leanne Kelley, Dr. Ryan Johnson, Monica Scanu, Kristen Navarro, and all the undergraduates who have passed through our lab, for their support, their aid, and their laughter.

Finally, I would like to show gratitude to my family for all their love and support over the years. Without them, I would not have endeavored to complete this degree and dissertation, and I sincerely thank them all for their forbearance and understanding.

To my Mom and Dad, thank you for everything you’ve ever sacrificed for me and everything you’ve ever done to aid me to becoming a PhD. Your love and encouragement have always been with me.

To my brother Eric and sister-in-law Maria, thank you for your love and for all your attempts to help me out figuring out my life over the years.

To my daughter Ella, thank you for love and understanding when I couldn’t play with you because I needed to write. Your happy laughter has helped this process along.

And to my husband Stephen, thank you for putting up with everything I’ve emotionally thrown at you while I have been in graduate school. You’ve given me so much love and understanding as I have completed this journey and I couldn’t have done it without you.

Thank you to all who have been involved.

iii

Vita

March 12, 1990…………………………...... Born, Cincinnati, OH

2008…………………………………………..Graduation, Ursuline Academy

2008-2012……………………………………B.S. Biochemistry and Molecular Biology, Marquette University, with a minor in mathematics

2012-2019……………………………………Graduate Teaching and Research Associate, The Ohio State University

Publications

Manuscript to be published:

“EGL-38/Pax coordinates development in the Caenhorhabditis elegans egg-laying system through EGF pathway dependent and independent functions” Allison Webb Chasser, Ryan Johnson, Helen Chamberlin 2019; Requested revisions submitted to Mechanisms of Development

Previously published manuscript:

“Visualizing Multidimensional Data with Glyph SPLOMs: Visualizing Multidimensional Data with Glyph SPLOMs” Andrew Yates, Allison Webb (Chasser), Michael Sharpnack, Helen Chamberlin, Ken Huang, Raghu Machiraju Published in: Computer Graphics Forum, 2014, 33:3, 301-310 DOI 10.1111/cgf.12386 ; ISSN 01677055

Fields of Study

Major Field: The Ohio State Biochemistry Program

Focus: Molecular Genetics and Development

Table of Contents Abstract ...... i Acknowledgements ...... iii Vita ...... iv List of Figures ...... ix Appendix: Tables ...... x Chapter 1 Introduction ...... 1 1.1 Transcription factors in cell differentiation and specification ...... 1 1.1.1 Pax proteins ...... 2 1.1.1.1 Pax proteins in development...... 3 1.1.1.2 Pax proteins in cancer ...... 5 1.1.1.2.1 Example of Pax protein activity in cancer ...... 7 1.1.2 Hox proteins...... 8 1.1.2.2 Hox proteins in cancer ...... 10 1.1.3 Transcription factors as cancer targets ...... 11 1.2 Inductive cell-to-cell signaling ...... 13 1.2.1 Epidermal Growth Factor Pathway ...... 14 1.2.2 EGF signaling in cancer ...... 14 1.3 Computational methods to model regulatory networks ...... 15 1.4 Research question and chapter preview ...... 16 1.5 Figures ...... 19 Chapter 2: EGL-38/Pax coordinates development in the Caenhorhabditis elegans egg- laying system through EGF pathway dependent and independent functions ...... 23 2.1 Introduction ...... 24 2.2 Materials and Methods ...... 28 2.2.1 Genetic strains and worm culture ...... 28 2.2.2 Generating tagged EGL-38:GFP ...... 28 2.2.3 Generating nlp-2 and nlp-7 reporters and deletion transgenes ...... 30 2.2.4 Microscopy...... 31 2.2.5 U0126 experiments ...... 31

2.2.6 Protein Expression and EMSA ...... 32 2.3 Results ...... 33 2.3.1 EGL-38 protein is expressed during the development of the C. elegans egg- laying system ...... 33 2.3.2 EGL-38 expression in uv1 is dependent on EGF signaling ...... 34 2.3.3 Expression of neuropeptide genes nlp-2 and nlp-7 in the uv1 cells is dependent on EGF signaling and egl-38 ...... 36 2.3.4 EGFR activation bypasses the requirement for egl-38 to promote uv1 cell placement, but not neuropeptide gene expression...... 38 2.3.5 nlp-2 is a direct target for EGL-38 in uv1 cells ...... 39 2.4 Discussion ...... 41 2.5 Figures ...... 45 Chapter 3: EGL-38 activity is partially restricted in the vulF cell and necessary in the uv1 cell for egg-laying behavior in Caenorhabditis elegans hermaphrodites ...... 59 3.1 Introduction ...... 59 3.2 EGL-38 activity in the vulF cells ...... 62 3.2.1 Introduction ...... 62 3.2.2 Materials and Methods ...... 63 3.2.2.1 Genetic strains and worm culture ...... 64 3.2.2.2 Bacterial strains ...... 64 3.2.2.3 Performing RNA interference and generating RNAi strains ...... 64 3.2.2.4 Microscopy ...... 65 3.2.2.5 Generating lin-3/egf reporter strain and isolation transgenes ...... 66 3.2.3 Results ...... 67 3.2.3.1 Expression of lin-3/egf in the vulF cells requires LIN-1/ETS and EGL-38 ...... 67 3.2.3.2 nlp-2 uv1 expression is partially dependent on LIN-1/ETS ...... 68 3.2.3.3 Expression of egl-38 in the uv1 cells is dependent on EGL-38 ...... 68 3.2.3.4 lin-3/egf vulval cell enhancer element may represent a Pax binding site ...... 69 3.2.4 Discussion ...... 70 3.3 EGL-38 activity in the uv1 cells ...... 73

3.3.1 Introduction ...... 73 3.3.2 Materials and Methods ...... 74 3.3.2.1 Genetic strains and worm culture ...... 74 3.3.2.2 Generating nlp-7 deletion transgenes ...... 74 3.3.2.3 Microscopy ...... 75 3.3.2.4 Egg-retention assay ...... 76 3.3.3 Results ...... 77 3.3.3.1 nlp-7 has a 30 bp uv1 enhancer element with a prospective Pax binding site77 3.3.3.2 Single and double mutants of nlp-2 and nlp-7 have little to no effect on egg- laying behavior ...... 78 3.3.4 Discussion ...... 79 3.4 Figures ...... 82 Chapter 4: The pal-1 regulatory network and an accompanying computational model . 90 4.1 Introduction ...... 90 4.2 Methods ...... 93 4.2.1 Genetic strains and worm culture ...... 93 4.2.2 RNA interference ...... 93 4.2.3 Microscopy...... 94 4.2.4 Summarization of data in the Glyph SPLOM ...... 95 4.2.5 Generating nhr-232 reporter ...... 95 4.3 Results ...... 96 4.3.1 pal-1 Knockdown causes significant lethality of wild-type, non-cross progeny . 96 4.3.2 Mating confers a protective factor against loss of pal-1 ...... 98 4.3.3 Glyph SPLOM has significant advantages over traditional computation methods ...... 99 4.3.4 nhr-232 expresses in the E lineage as predicted by the Glyph SPLOM ...... 101 4.4 Discussion ...... 102 4.5 Figures ...... 106 Chapter 5 Discussion ...... 113 5.1 Results overview ...... 113

vii

5.2 Pax involvement with inductive signaling ...... 115 5.3 Pax activation of cellular phenotype ...... 116 5.4 Pax role in cancer ...... 118 5.5 Significance of Glyph SPLOM computational method ...... 119 5.6 Final conclusions ...... 121 Appendix: Tables ...... 123 References ...... 126

viii

List of Figures

Figure 1: Development of the egg-laying system and the posterior muscle/skin……….19 Figure 2: Models of EGL-38/PAX and PAL-1/HOX activity in their respective systems ……………………………………………………………………..…………………………....21 Figure 3: C. elegans hermaphrodite vulval development …..………………….…….…...45 Figure 4: EGL-38::GFP is expressed during egg-laying system development.....….…..47 Figure 5: EGL-38::GFP expression is dependent on EGF signaling……………..…..….49 Figure 6: Expression of nlp-2 and nlp-7 in uv1 cells is dependent on EGF signaling and egl-38……………………………………………………………………………..………..…...51 Figure 7: Activated let-23/EGFR can bypass the egl-38 defect in uv1 placement, but not nlp gene expression…...………………………………………………………………..….....54 Figure 8: nlp-2 is a direct target for EGL-38..…………………………………………..…..56 Figure 9: lin-/3/egf and nlp-2 expression is dependent on LIN-1/ETS…..……………….82 Figure 10: lin-3/egf expression is dependent on EGL-38 and LIN-1/ETS; egl-38 expression is dependent on EGL-38………………………………………..…………….…84 Figure 11: lin-3/egf contains an anchor cell enhancer & a putative vulF cell enhancer………………………………………………………………………………………..85 Figure 12: nlp-7 contains a uv1 enhancer with a putative EGL-38 site…..….….……….87 Figure 13: Single and double nlp-2, nlp-7, and flp-11 mutants do not affect egg retention……………………………………………………………………….………………..89 Figure 14: MIM predicts direct regulatory targets of pal-1 activity………………………106 Figure 15: Expectations for embryonic survival of pal-1 KD………….…………………107 Figure 16: Loss of pal-1 contributes to significant lethality of non-cross progeny…….109 Figure 17: Mating without male-induced fertilization fails to protect embryos against pal- 1 KD………………………………………………………………………….……………..…110 Figure 18: The C. elegans 655 TF Glyph SPLOM…………..…………………….…...... 111 Figure 19: nhr-232 is expressed in the E lineage as predicted by the Glyph SPLOM……………………………………………………………………………..….……...112

Appendix: Tables

Table 1: Chapter 2 Primers……………………………………………………….………113

Table 2: Chapter 3 Primers……………………………………………………………….114

Chapter 1 Introduction

Coordination of cell growth, specification, and differentiation are required for the processes of morphogenesis and organogenesis during animal development. One of the earliest points of control for this coordination is through transcriptional activation of genes involved in regulatory pathways or networks.

1.1 Transcription factors in cell differentiation and specification

Cell differentiation is the process through which a cell becomes specialized for a particular biological function. Converting pluripotent embryonic cells to terminally differentiated cells requires restricting developmental outcomes until a specified cell fate is reached and the phenotypic output obtained (Gilbert 2000). This is accomplished through many layers of gene regulation such as inductive cell-to-cell signaling, miRNA activity, mRNA processing, protein localization, and, most integrally to this presentation, transcriptional control (K. Chen and Rajewsky 2007).

The initiation and integration of cell fate and differentiation often begins directly at the level of transcription in order to regulate which gene products are available to perform temporally and spatially restricted functions. This regulation requires transcription factors to activate genes necessary for developmental outcomes and future cell phenotypic programs (Fong and Tapscott 2013; T. I. Lee and Young 2013;

Singh, Khan, and Dinner 2014).

Terminal transcription factors required for cell fate specification include both pioneer and classical transcription factors (C.-F. Liu et al. 2017). “Pioneer” transcription factors are specialized to target previously quiescent chromatin to activate genes newly required for the differentiating cell type (Iwafuchi-Doi and Zaret 2016). “Classical” transcription factors bind to enhancers and promoters which are not shielded by chromatin for transcriptional activation or repression, including activation of genes needed to execute the terminal phenotype and repression of genes involved in maintenance of the undifferentiated phenotype (Iwafuchi-Doi and Zaret 2016; Li and

Kirschner 2014).

In the next sections, I will focus on two classical transcription factor families vital for metazoan development and specification of terminal phenotypes.

1.1.1 Pax proteins

Paired-box (Pax) proteins are a family of classical transcription factors necessary for organogenesis and are considered master regulators of development (Blake and

Ziman 2014; Kozmik et al. 2003). There are nine members of this family characterized by included structural domains. All nine members contain a Paired-DNA binding domain, which makes dual contacts to DNA; additionally, Pax 1 and 9 contain an octapeptide domain, Pax2, 5, and 8 contain the octapeptide and a partial homeodomain, Pax4 and 6 contain a full homeodomain, and Pax 3 and 7 contain the octapeptide and full homeodomain regions (Chi and Epstein 2002). The octapeptide domain is involved in transcriptional repression, and the homedomain is a conserved

2 helix-loop-helix motif involved in DNA binding to supplement the Paired domain (Frost et al. 2008).

In solution, the paired domain is largely without structure, only obtaining a fixed rigidity with the addition of target DNA (Czerny, Schaffner, and Busslinger 1993; J.

Epstein et al. 1994). The paired-domain is composed of two separate helix-turn-helix motifs, resembling homeodomains, and either N- or C-terminal contacts with DNA can predominate (J. A. Epstein et al. 1994; H. E. Xu et al. 1999; W. Xu et al. 1995). Based on the fluidity of these interactions, the paired-domain has a degenerate consensus targeting sequence which contributes to the variety of nucleotide sequences that Pax proteins may bind (Czerny, Schaffner, and Busslinger 1993). Additionally, Pax proteins are capable of interacting with other transcription factors, such as the Ets family, to contribute to target specificity (Fitzsimmons et al. 1996).

1.1.1.1 Pax proteins in development

Pax proteins were first discovered in Drosophila, but have since been characterized throughout metazoans (Bopp et al. 1986). In humans, Pax proteins are necessary for organogenesis of systems as diverse as the eye to the thyroid, the central nervous system to B lymphocytes, the pancreas to the kidney or the ear (Blake and

Ziman 2014; G. R. Dressler et al. 1990; Lang et al. 2007; Macchia et al. 1998; Magliano,

Lauro, and Zannini 2000; Noll 1993; Torres et al. 1995). Pax TF activity is prevalent during developmental processes involving cell differentiation, characterized by lineage

3 determination and restriction of cellular potential, such as establishment of the musculoskeletal system (Blake and Ziman 2014).

Single Pax TFs can act in multiple systems to direct development. For example,

PAX2 is active in organogenesis of the kidneys, urogenital tract, eyes, ears, and central nervous system (Burton et al. 2004; G. R. Dressler et al. 1990; Namm, Arend, and

Aunapuu 2014; Pichaud and Desplan 2002; Torres et al. 1995; Torres, Gomez-Pardo, and Gruss 1996). Mutations in Pax2 can, consequently, have varied effects on the body. A heterozygous frameshift mutation of Pax2 can result in renal coloboma syndrome causing both kidney and eye defects while the homozygous mutation results in early lethality (Bower, Schimmenti, and Eccles 1993; Keller et al. 1994; Sanyanusin et al. 1995).

Mutations in other Pax genes result in similarly severe defects. Pax3 mutations in mice cause a Splotch phenotype, resulting in embryonic death with significant neural crest deficiencies and congenital heart disease; in humans, Pax3 mutations can cause

Waardenburg syndrome, resulting in deafness, craniofacial abnormalities, and defective melanocyte production (Lang et al. 2000). Pax6 mutations in mice lead to a complete absence of eye structures; in humans, Pax6 mutations can cause blindness (Glaser et al. 1994; R. E. Hill et al. 1991).

The severity of phenotypes resulting from Pax mutations clearly communicates their importance in organogenesis, as mis-regulation of Pax targets leads to defects in cell specification, differentiation, and morphogenesis. The conservation of these defects between organisms points to the evolutionary significance of Pax TFs in development.

Pax proteins have adapted different functions in neighboring cell types over generations of development in order to coordinate the organogenesis of increasingly more complicated organisms. This ability, to take on varying functions in disparate cell types, reinforces the idea of Pax transcription factors as master regulators of an evolutionarily dynamic transcriptional network necessary for coordinated development (Lambert et al.

2018).

1.1.1.2 Pax proteins in cancer

Generally, Pax proteins are downregulated in differentiated tissues, and possess few activities in the adult stages (Blake and Ziman 2014; Czerny, Schaffner, and

Busslinger 1993; G. R. Dressler et al. 1990; Lang et al. 2007; Rothenpieler and Dressler

1993; Terzić et al. 1998). However, transient reactivation of Pax genes may contribute to a resurgence of the developmental phenotype which could lead to cancer (Grimley and Dressler 2018).

Pax proteins are necessary to initiate the terminally differentiated cell fate, which is disrupted in tumorigenesis. Pax proteins can trigger cell differentiation in one of two ways: 1) by being turned off when they were necessary for maintaining the undifferentiated, proliferative state; and 2) by being turned on when they are required to trigger differentiation and exit from the proliferative state (Wachtel and Schäfer 2015).

Therefore, cells rely on downregulation (first function) or upregulation (second function) of Pax during different phases of development to complete differentiation.

Consequently, both gain- and loss-of-function mutations in Pax proteins can contribute to cancer phenotype by either promoting proliferation or failing to promote differentiation, respectively (Wachtel and Schäfer 2015).

Investigation into these gain- or loss-of-function activities and whether eliminating aberrant Pax expression can ameliorate the cancer phenotype has sparked interest in several of the Pax family members and the cancer type they are involved with.

Interestingly, out of the five Pax proteins most widely expressed in various cancers, including Pax 2/3/5/7/8, Pax2 is the only one which does not exhibit chromosomal rearrangements when present in the cancer state (Robson, He, and Eccles 2006;

Wachtel and Schäfer 2015). Instead, most Pax2-cancer related phenotypes are a result of mis-regulation leading to ectopic expression in differentiated tissues where Pax2 is generally absent. These five Pax TFs are the most highly susceptible to dysregulation, as evidenced by their prevalence in cancer (Robson, He, and Eccles 2006).

Part of the impact that Pax proteins have on the incidence of cancer is that, in addition to their roles in cell differentiation and organogenesis, Pax TFs can also contribute to cell proliferation, growth, maintenance of phenotype, and apoptosis (cell death) (Mahajan, Leavey, and Galindo 2015; Robson, He, and Eccles 2006). Often, as in the case of Pax2, this wild-type, regulatory contribution to cell survival occurs concurrently with the role of Pax to specify cell lineages during development; more rarely do Pax proteins contribute to phenotypic maintenance in an adult, differentiated tissue. Consequently, dysregulated, aberrant expression of non-mutated Pax proteins

6 during the cancer state could significantly contribute to tumor survival and maintenance

(Eccles and Li 2012; Muratovska et al. 2003).

1.1.1.2.1 Example of Pax protein activity in cancer

As previously discussed, there are alternate methods of Pax proteins contributing to the cancer phenotype: mutations or ectopic expression. When mutated, a Pax protein can easily become dysregulated, expressing in a differentiated tissue where it has no typical function and possibly exhibiting new activities which support the cancer state.

Alternatively, Pax proteins can become aberrantly activated in a differentiated tissue and contribute to an undifferentiated phenotype as Pax would have done during development, which supports the growth and spread of tumor tissue. Below, I’ve presented examples that exemplify each of these methods of Pax-related cancer function.

Chromosomal translocations lead to overexpression of abnormal PAX3 and

PAX7 forms in rhabdomyosarcoma, a cancer affecting the skeletal muscle soft tissue

(Barr et al. 1999; Bernasconi et al. 1996; Tiffin et al. 2003). In alveolar rhabdomyosarcoma, both Pax3 and Pax7 are independently subject to a translocation rearrangement with the FOXO1 Forkhead TF, causing the fusion to be expressed from the Pax promoter, with the paired-DNA binding domain, and with the stronger transcriptional activation domain of the FOXO1 TF (Davis et al. 1994; Kelly et al. 1997;

Sorensen et al. 2002). This results in errors in transcriptional regulation contributing to alveolar rhabdomyosarcoma in the presence of additional cancer-contributing mutations

(Mahajan, Leavey, and Galindo 2015). The chromosomal translocation of Pax3 or Pax7 results in an entirely new function for these Pax proteins, by combining their aberrant activation with the specificity and strength of a different transcription factor.

Alternatively, abnormal expression of a wild-type Pax protein in differentiated tissue can contribute to the cancer phenotype through its functions involved in cell growth, proliferation, and maintenance of the undifferentiated state. For example, eutopic PAX2 expression must be downregulated to complete nephrogenesis, and expression is generally absent in the adult kidney (G. R. Dressler et al. 1990; Gregory

R. Dressler et al. 1993; Rothenpieler and Dressler 1993). Ectopic Pax2 is highly expressed in renal cell carcinomas and Wilm’s tumors; additionally, inhibition of Pax2 in renal cancer cell lines increases cell death and decreases proliferation (Daniel et al.

2001; G. R. Dressler and Douglass 1992; Gnarra and Dressler 1995; Hueber et al.

2006, 2008; Mazal et al. 2005; Memeo et al. 2007; Tong et al. 2006; Winyard et al.

1996). Expressing wild-type Pax2 in the cancer state contributes to the disease phenotype by maintaining the spread of undifferentiated tumor cells; eliminating this aberrant expression decreases tumor survival.

1.1.2 Hox proteins

Homebox (Hox) proteins are a family of classical transcription factors required for segmentation and organogenesis during development (Levine and Hoey 1988; Lewis

2004). The 30+ Hox gene family is far more diverse than the nine-member Pax family, but all Hox genes contain a helix-turn-helix homeodomain DNA binding region

(McGinnis et al. 1984; Scott and Weiner 1984). The homeodomain proteins, particularly

Hox TFs, all target a very similar, short, AT-rich sequence which is not sufficient to confer target specificity (Affolter, Slattery, and Mann 2008; Desplan, Theis, and

O’Farrell 1988; Ekker et al. 1991, 1994; Hoey and Levine 1988). Some Hox proteins are capable of interacting with other TFs as cofactors, while others regulate transcription in such a unique spatiotemporal cellular environment that target specificity is conferred by default, as no other suitable Hox targets are available and accessible to these classical

TFs at the time (Mann, Lelli, and Joshi 2009).

The Hox proteins can also function in DNA replication, DNA repair, protein degradation, and translation; however, I will only be addressing their roles as transcription factors in development here (Rezsohazy et al. 2015).

1.1.2.1 Hox proteins in development

Hox genes were first discovered in Drosophila as a result of dramatic homeotic transformation phenotypes such as legs growing in the place of antenna (Schneuwly,

Klemenz, and Gehring 1987). Hox TFs are required for morphogenesis, establishing the anterior-posterior axis and segmentation of a developing embryo. Interestingly, spatiotemporal expression of Hox gene clusters coincides with their chromosomal location, with the first gene along the chromosome being expressed anterior to later genes; in addition, Hox gene expression patterns are similar for all bilateral animals

(Duboule and Dollé 1989; Pearson, Lemons, and McGinnis 2005). HOX proteins are

9 evolutionarily conserved and necessary for bilateral development of metazoans, as well as features such as cell lineage commitment. Hox gene expression continues into adults.

Defects in Hox function result in homeotic transformations, causing body segment identities to be altered to a different body segment fate, as well as truncations

(or absence) of morphological elements entirely (Kmita et al. 2005; Taniguchi 2014). For example, mice lacking HoxA and HoxD function in the forelimb show arrest of limb development and truncation of the distal limb, while mice with a Hox10 or Hox11 paralog disruption result in transformation of one vertebral segment to the next, causing ribs to grow from posterior regions of the spine (Kmita et al. 2005; Wellik and Capecchi

2003).

Due to the redundancy of Hox TFs in humans, knockout of a single Hox gene produces only a mild phenotypic effect; two or more genes must be knocked out to cause a severe effect (Favier et al. 1996; McIntyre et al. 2007; Wellik 2009; Zakany and

Duboule 2007). Some of the trans-paralagous Hox genes can indeed act as functional complements of one another, such as Hoxa3 and Hoxd3 (Greer et al. 2000). There are only 10 Hox genes which have produced genotypic abnormalities in humans, and several of these have only been demonstrated in a few individuals (Quinonez and Innis

2014). Genetic abnormalities can include synpolydactyly (Hox13), disruptions to the craniofacial features (HoxA1/B1), or cleft palate (HoxA2) among other disorders (Akarsu et al. 1996; Alasti et al. 2008; Tischfield et al. 2005).

1.1.2.2 Hox proteins in cancer

Hox genes can contribute to tumorigenesis through survival of apoptosis, alterations to cell-to-cell signaling, trigger of epithelial-mesenchymal transition, and tumor cell invasion (H. Chen et al. 2007; Miao et al. 2007; Wu et al. 2006; Zhai et al.

2007). Similarly to Pax TFs, both gain- and loss-of-function Hox mutations can contribute to the cancer phenotype based on the cellular context of HOX activity (Abate-

Shen 2002). Most instances of ectopic Hox expression in cancer occurs spatially in the same body segments where the HOX proteins are active during embyrogenesis (Abate-

Shen 2002). Therefore, most cancer phenotypes attributable to Hox activity result from the aberrant re-activation of developmental processes triggered by ectopic HOX protein.

The mechanisms dysregulating Hox activity in cancer are extremely varied and nuanced, reflecting the intensely specific target regulation performed by Hox TFs themselves (Shah and Sukumar 2010). Often, the cancer state is exacerbated by a resurgence of wild-type Hox activity promoting pre-differentiation cell activity or the loss of wild-type Hox activity to trigger the differentiated cell fate, rather than by a mutation in the TF itself (Abate-Shen 2002)

1.1.3 Transcription factors as cancer targets

There are numerous methods of treating cancer that have been developed, including immunotherapy, the use of agents to strengthen the immune system to fight cancer on its own, and chemotherapy, the use of drugs which target the cancer cell to render it unable to proliferate. Often, however, these approaches have harmful effects

11 on the body by targeting both the cancer and non-cancer cell types simultaneously.

Drug inhibition of processes necessary for cell maintenance and survival, meant to prevent the growth and spread of cancer, also prevents the cell maintenance and survival of non-diseased cells. Consequently, the normal, wild-type cells are as affected as the tumor cells under these therapies. Instead, development of targeted drug therapies allows for administration of drugs meant to specifically prevent the growth, spread, and survival of cancer cells, while leaving non-cancer cells alone.

One method of targeted drug therapy is antibodies, which act by inhibiting protein function on the tumor cell surface, preventing cell-to-cell communication and proliferation. Another method is small molecule therapeutic drugs. These small molecules are meant to regulate a biological process, similar to antibodies, but have some significant advantages over the antibody approach. Small molecule agents generally have better tumor penetrance, less harmful side effects, a shorter half-life, and the ability to cross the cell membrane in comparison with traditional antibiotic reagents

(Zhu and Li 2018). Targeting the cancer phenotype using small molecules is, therefore, an attractive option amongst cancer treatments.

Small molecule inhibition can target nuclear proteins to prevent their activity, including transcription factors. Given that re-activation of developmental programs triggered by aberrant TF expression may contribute to the initiation and progression of cancer, targeting these transcription factors can specifically inhibit the cancer phenotype

(Grimley and Dressler 2018). In particular, inhibition of TFs which are active during embryogenesis and inactive in the adult, yet are activated and contributing to the cancer

12 state, would present a viable target as blocking the function of these TFs in cancer would not affect the non-tumorigenic, differentiated cells. However, targeting TFs is difficult and they have previously been considered untouchable due to their nuclear localization, charged interaction with DNA, and lack of a classical binding pocket for targeting by a small molecule (Ahmadzadeh et al. 2009; Cheng et al. 2018). Therefore, most current small molecule agents target the cell-surface receptors or ligands involved in intracellular signaling, none of which provide the same limitations listed above as do

TFs (Kerr and Chisholm 2019). Nevertheless, development of TF-targeting small molecules will be important to avoid inactivation of the wild-type cell maintenance, apoptosis, growth, and communication pathways which require intracellular signaling.

1.2 Inductive cell-to-cell signaling

Inductive signaling is the ability of one set of cells to influence another through a signaling pathway, often resulting in cell fate specification leading to embryonic patterning and organogenesis (Perrimon, Pitsouli, and Shilo 2012). An inducing cell sends a signaling ligand to a competent responder cell, initiating a pathway through its receptor which will result in transcription factor activation in the nucleus (Perrimon,

Pitsouli, and Shilo 2012). There are seven major signaling pathways: Wnt, TGF-β,

Hedgehog, receptor-tyrosine kinase, nuclear receptor, Iak/STAT, and Notch (Barolo and

Posakony 2002). These pathways result in cell non-autonomous activation of target genes that are necessary to coordinate cell specification, proliferation, migration, growth, and death (Artavanis-Tsakonas, Rand, and Lake 1999).

1.2.1 Epidermal Growth Factor Pathway

The Epidermal Growth Factor (EGF) pathway is a receptor-tyrosine kinase pathway that triggers cell proliferation, maintenance, and survival during embryogenesis and renewal of stem cell fate (Campbell and Bork 1993; Salomon et al. 1995). There are ten ligands which are EGF-related growth factors capable of binding to the receptors, including EGF (Normanno et al. 2003; Yarden and Sliwkowski 2001). The

EGF Receptor (EGFR) is one of four related receptors, all of which can respond to and amplify the same EGF signal (Yarden and Sliwkowski 2001).

Ligand binding to the EGFR causes homo- or heterodimerization of the receptor, leading to activation of the tyrosine kinase domain (Olayioye et al. 2000). This results in a phosphorylation cascade culminating in transcription factor activation, particularly of

ETS domain TFs (Klambt 1993; O’Neill et al. 1994). Additionally, once the tyrosine kinase domain of dimerized EGFR has been phosphorylated, the receptor is internalized, allowing this domain to serve as a binding site for signal transducers and activators, such as the Ras/Raf MAPK pathway, which will trigger another intracellular signal transduction cascade (Herbst 2004).

1.2.2 EGF signaling in cancer

The EGF signaling pathway has been extensively studied in cancer, due to its role in cell proliferation and survival in both the embryo and adult stem cell fate.

Exposure of tumor cell lines to ionizing radiation causes activation of EGFR, which

14 contributes to cell proliferation; this activation is increased with increasing doses of radiation (Bowers et al. 2001; Schmidt-Ullrich et al. 1994). EGFR overexpression differentiates tumor cells from non-tumor cells and presents a target for inhibitory therapeutics (Herbst 2004). This overexpression also correlates with a high incidence of metastasis, poor tumor differentiation, and high tumor growth rate; in general, highly-

EGFR expressing tumors have more aggressive growth and invasiveness (Ethier 2002,

2002; Pavelic et al. 1993).

1.3 Computational methods to model regulatory networks

One of the greatest difficulties with interpreting the effects from Pax, Hox, and

EGF genes is the inter-connectedness of the regulatory networks in which they participate. Single mutations may snowball into global, difficult to parse out defects; conversely, functional redundancy may prevent a single mutation from having any effect at all. To differentiate these outcomes, perturbations of every factor in a regulatory network would be necessary. However, large-scale perturbation experiments are difficult and costly, and may not provide a clear elucidation of function. It has therefore become important to develop computational methods to maximally extract biological information from existing data sets, particularly from the wild-type condition (Stigler and

Chamberlin 2012).

Whether from wild-type or perturbation studies, an important factor of computational methods is the ability to predict directed interactions (Marbach et al.

2010). This is particularly useful to elucidate the mechanisms of cellular differentiation.

The process of differentiation is tightly correlated to the upregulation/activation or downregulation/deactivation of different proteins, and these dynamics can be clearly modeled in directed regulatory networks (Villani, Barbieri, and Serra 2011). Additionally, the function of unknown genes can be inferred based on its position within the predicted regulatory network and its interaction with other proteins (Bodaker et al. 2014).

1.4 Research question and chapter preview

To date, much of the investigation into the transcriptional control of coordinated development has focused on the role of Pax or Hox genes in disease states such as cancer. While this has provided much information about the wild-type activity of both families, it has also left significant holes in our understanding of how Pax and Hox genes direct gene regulatory networks in development. I have undertaken my thesis research to address some of these deficiencies, using Caenorhabditis elegans as a model system.

Caenorhabditis elegans is a free-living soil nematode which presents a model organism with easily manipulated genetics, an invariant cell lineage, a transparent cuticle for cellular level observations, and simple maintenance and feeding routines

(Antoshechkin and Sternberg 2007; Brenner 1974; Sulston et al. 1983). Development of the nematode as a laboratory specimen answered a need to adapt a faster, more tractable model organism than humans or mice in order to examine trans-regulatory adaptations and transcription factor networks without destroying the model or requiring years of investigation (Lynch and Wagner 2008). Beneficially, at least 80% of the C.

16 elegans genome has been found to have homologs in the human genome, and regulatory pathways, disease genes, and transcription factors are highly conserved between species (Jones and Ashrafi 2009; Lai et al. 2000; Silverman et al. 2009).

In chapter two, I explore the role of EGL-38/PAX in cell specification of the vulval- uterine connection (Figure 1) using RNA interference, inhibitor studies, genetic mutants, and epistasis studies. I detail the expression pattern of EGL-38 using CRISPR to GFP tag the endogenous locus. My data shows that EGL-38 utilizes the EGF pathway to coordinate its own activity and expression between the vulval and uterine cells of this system, and that EGL-38 activates a cellular program of genes required for control of egg-laying behavior, including the neuropeptide genes nlp-2 and nlp-7 (Figure 2). In chapter three, I extend the exploration into the role of EGL-38 in the egg-laying system using reporter analysis, genetic mutations, RNA interference, and egg-retention assays.

I present my investigation into lin-3/egf and nlp-7 for Pax binding sites, into the LIN-

1/ETS TF as a possible co-factor for EGL-38 in the vulval cells, and into the biological effect of EGL-38 targets on egg-laying behavior. In chapter four, I investigate predictions made by two separate computational model of TF regulatory network using

RNA interference and reporter assays. In particular, I investigate the PAL-1/HOX regulatory network responsible for cell fate differentiation of C blastomere descendants in embryogenesis (Figure 1), leading to creation of posterior mesoderm or ectoderm

(Figure 2). I present my results concerning embryonic survival of PAL-1/HOX knockdown and my validation of a lineage clustering prediction from a collaboratively

17 developed computational method. Finally, in chapter five, I discuss key results and some of the new questions that have been provoked by my investigations.

1.5 Figures

Figure 1: Development of the egg-laying system and the posterior muscle/skin

Fig. 1 (A) Representation of a C. elegans hermaphrodite at the mid L4 stage. Indicated are the positions of the developing vulva and the posterior region containing C blastomere descendants. (B) The hermaphrodite vulva at the mid L4 stage, with 1° VPC

P6.p cell descendants indicated, vulF and vulE, and the newly specified uv1 cells indicated. (C) An embryo at the 8 cell stage, with 8 founder blastomeres, and two hours later at the end of gastrulation. Indicated in yellow is the C blastomere on the left, and the general position of its descendants on the right. (D-I) A mid L4 hermaphrodite expressing EGL-38::GFP protein in the vulF (D-E), vulE (F-G), and uv1 (H-I) cells of the

19 same larva. EGL-38 is the Pax2/5/8 ortholog responsible for development of the vulval- uterine connection, including expression of the EGF signal in the vulF cells and specification of the uv1 cell identity. Cells are indicated with a white arrowhead. (J-K) An embryo at the end of gastrulation expressing Ppal-1::rfp in the developing posterior mesoderm and ectoderm. PAL-1 is the Hox protein required for development of the posterior muscle and skin of mid-stage embryos.

Figure 2: Models of EGL-38/PAX and PAL-1/HOX activity in their respective systems

Fig. 2 (A) Model of EGL-38 activity in the specification of the uv1 cells of the egg-laying system. EGL-38 is required to activate expression of the lin-3/egf signal in the vulF cells, which is sent to its LET-23/EGFR receptor at the surface of presumptive uv1 cells. egl-38 expression is then activated in uv1 by the EGF pathway and will go on to activate expression nlp-2 and nlp-7, two neuropeptide proteins involved in egg-laying inhibition.

Expression of egl-38 is coordinated between vulF and uv1 through the EGF pathway.

(B) Model of PAL-1 activity in the specification of C blastomere descendants as posterior mesoderm (muscle) or ectoderm (skin), adapted from the Mathematically

Inferred Model presented in (Stigler and Chamberlin 2012). PAL-1 is responsible for activating genes involved in the initiation of fate specification, in mesoderm or ectoderm differentiation, and those with a mixed or other effect on the cell fate specification of C blastomere descendants.

Chapter 2: EGL-38/Pax coordinates development in the Caenhorhabditis elegans egg-laying system through EGF pathway dependent and independent functions

Note: This chapter is being published under the title above. I am the first author, joined by Ryan Johnson and Helen Chamberlin.

I generated the EGL-38::GFP-tagged strain, with help from Helen Chamberlin in injecting and screening for viable candidates. Helen Chamberlin performed all crosses to create double mutants and mutants with reporters. Ryan Johnson created the Pnlp-

2::gfp reporter strain and I created the Pnlp-7::gfp reporter strain. I performed all screening of fluorescence for EGL-38::GFP, Pnlp-2::gfp, and Pnlp-7::gfp, in wild-type, mutant, and drug backgrounds. I also executed all U0126 drug assays. Ryan Johnson performed the Pnlp-2::gfp deletion analysis and the nre1 EMSA in vitro.

This manuscript has been submitted to the journal Mechanisms of Development.

At the time of concluding this thesis, minor revisions have been requested and re- submitted to the editor. This manuscript is being presented here with concessions to formatting.

2.1 Introduction

Paired box (Pax) transcription factors are necessary for coordinated development and organogenesis of many mammalian systems, such as thyroid (Pax8), kidney

(Pax2/8), central nervous system (Pax2/3/5/6/7/8), B lymphocytes (Pax5), pancreas

(Pax4/6), eye (Pax6), and ear (Pax2/8) (Blake and Ziman 2014; G. R. Dressler et al.

1990; Lang et al. 2007; Macchia et al. 1998; Magliano, Lauro, and Zannini 2000; Noll

1993; Torres et al. 1995). These transcription factors contribute to the differentiation and maintenance of cell fate, as well as the initiation of cellular programs required for proper cell activity (Mansouri, Chowdhury, and Gruss 1998; Miccadei et al. 2002). It is known that Pax proteins are necessary for initiating complex developmental programs, and previous work has suggested that Pax proteins are capable of coordinating differentiation of diverse cell types due to their involvement in inductive cell-to-cell signaling (Lang et al. 2007; Püschel, Gruss, and Westerfield 1992). For example, research on Pax2, Pax6, and Pax8 has shown these proteins are primarily expressed in cells requiring inductive cell-to-cell signaling for development, arguing that it is likely that

Pax2/6/8 are targets of, or participants in, these signaling events (G. R. Dressler et al.

1990; Püschel, Gruss, and Westerfield 1992). Involvement in cell-to-cell signaling would provide a clear mechanism for Pax proteins to be able to coordinate cell fate and development among distinct, neighboring cell types; however, as of yet, Pax participation in signaling has not been well characterized in humans or mice.

To investigate the interaction of a Pax protein with signaling, and of Pax coordination of cell fate and developmental programs, we are studying the egl-38 gene

24 in Caenorhabditis elegans. EGL-38 is most similar in sequence to mammalian Pax2/5/8, and similarly possesses the paired-DNA binding domain and octapeptide sequence

(Chamberlin et al. 1997). egl-38 functions in the development of the hermaphrodite egg- laying system, hindgut, and male spicule (Chamberlin et al. 1997). In particular, the development of the egg-laying system is ideal for examining the interaction of EGL-

38/Pax with an inductive signaling pathway for establishment of cell fate and of execution of cellular programs, as this system is dependent on both EGL-38 and the

Epidermal Growth Factor (EGF) pathway for differentiation (Chang, Newman, and

Sternberg 1999; Katz et al. 1995)

The C. elegans egg-laying system results from coordinated development between vulval cells in the epidermis and mesodermal cells in the somatic gonad

(Figure 3). First, vulval cell identity is established in response to EGF signaling. The C. elegans EGF signal, LIN-3, is sent from the uterine anchor cell during the L3 larval stage to vulval precursor cells, specifying a cell (P6.p) as the 1° vulval cell (Figure 3A)

(R. J. Hill and Sternberg 1992; Katz et al. 1996). The 1° cell then signals to neighboring cells (to produce the 2° vulval cells), and divides to produce eight cells that form the upper portion of the hermaphrodite vulva. Four of these cells at the apex of the vulva

(the vulF cells) then express the lin-3/egf signal gene, which is used for reciprocal EGF signaling back to the uterus during the L4 larval stage (Figure 3E) (Chang, Newman, and Sternberg 1999). This reciprocal EGF signal is sent from the vulF cells to a subset of uterine cells surrounding the anchor cell to induce these cells to assume a uterine ventral (uv1) cell identity (Chang, Newman, and Sternberg 1999). The uv1 cells will

25 move into position closer to the vulva to serve as an anchor of the vulva to the uterus, allowing the vulF cells to separate from one another to form the passage through which eggs are laid (Figure 3F-G) (Newman and Sternberg 1996). Additionally, the uv1 cells are neurosecretory, and serve an inhibitory role during egg-laying to help control the periodicity and set inactive periods (Banerjee et al. 2017; Collins et al. 2016; Jose et al.

2007).

egl-38 plays a critical role in coordinating the connection between the vulva and the uterus. In an egl-38(lf) mutant, the lin-3/egf signal is not expressed in vulF, and the uv1 cells are not specified (Chang, Newman, and Sternberg 1999; Rajakumar and

Chamberlin 2007). Instead, the presumptive uv1 cells retain their original identity and migrate away with their sister cells to fuse with the anchor cell into the uterine seam cell syncytium (Newman, White, and Sternberg 1996). egl-38(lf) mutants also lack vulF cell separation, preventing the laying of eggs (Rajakumar and Chamberlin 2007). Previous work has shown that egl-38 has functions in both the epidermal vulF cells and the mesodermal uv1 cells to contribute to uv1 cell specification and vulF cell separation

(Rajakumar and Chamberlin 2007), but its role in coordinating development between these cell types was not clarified.

Here we show how EGL-38 functions in both autonomous (independent of the

EGF pathway) and non-autonomous (dependent on the EGF pathway) processes to influence uv1 cell fate and cellular programs necessary for egg-laying function. We have tagged the endogenous egl-38 locus with gfp, which allows for clear examination of

EGL-38::GFP localization and timing of expression. We have determined that egl-38

26 uv1 expression is dependent on the same LIN-3/EGF signal that EGL-38 activates in the vulF cells. Furthermore, our work shows that egl-38 is necessary for the expression in uv1 of two neuropeptide-encoding genes which may have roles in controlling egg- laying and thereby contributing to the functions of differentiated uv1 cells. We have discovered that the uv1 cell traits of positioning above the vulva and of expression of uv1-specific genes are separable activities, with the former being dependent on EGF signaling (and therefore indirectly on EGL-38 in the vulF cell), and the latter requiring direct, autonomous EGL-38 activity within the uv1 cell. EGL-38 coordination of cell fate, development, and its own expression through an inductive cell-to-cell signaling pathway represents a novel Pax function which may represent a concise, direct mechanism for

Pax to organize disparate cells into a single, cohesive organ system.

2.2 Materials and Methods

2.2.1 Genetic strains and worm culture

Strains were cultured under standard conditions for C. elegans (Brenner 1974;

Stiernagle 2006). All experiments were performed at 20°C unless otherwise noted. The reference wild type C. elegans strain is N2; additional C. elegans strains used were:

CM2762 egl-38(gu253[egl-38::gfp])

CM2238 unc-119(e2498); guEx1372 (Pnlp-2(0.4kb)::gfp)

CM2694 unc-119(e2498); guEx1554 (Pnlp-7:gfp)

NY2040 unc-119(e2498); ynIs40(Pflp-11:gfp)

CM2763 let-23(sy1); egl-38(gu253[egl-38::gfp])

CM2766 let-23(sy1); unc-119(e2498); guEx1372 (Pnlp-2(0.4kb)::gfp)

CM2767 let-23(sy1); unc-119(e2498); guEx1554 (Pnlp-7:gfp)

CM2760 egl-38(sy294); unc-119(e2498); guEx1372 (Pnlp-2(0.4kb)::gfp)

CM2768 egl-38(n578); unc-119(e2498); guEx1372 (Pnlp-2(0.4kb)::gfp)

CM2427 let-23(sa62); unc-119(e2498); guEx1372 (Pnlp-2(0.4kb)::gfp)

CM2444 let-23(sa62); unc-119(e2498); egl-38(sy294); guEx1372 (Pnlp-2(0.4kb)::gfp)

CM2769 egl-38(sy294); unc-119(e2498); guEx1554 (Pnlp-7:gfp)

CM2770 egl-38(n578); unc-119(e2498); guEx1554 (Pnlp-7:gfp)

CM2740 let-23(sa62); unc-119(e2498); guEx1554 (Pnlp-7:gfp)

CM2739 let-23(sa62); unc-119(e2498); egl-38(n578); guEx1554 (Pnlp-7:gfp)

2.2.2 Generating tagged EGL-38:GFP

We tagged the endogenous egl-38 locus with gfp sequences at the 3’ end using methods of (Dickinson and Goldstein 2016). Briefly, we used synthesized sgRNA and purified Cas9 (Prior et al. 2017) combined with a repair template that introduces gfp sequences and selection markers removed via self-excising cassette (Dickinson and

Goldstein 2016). A 20-bp guide RNA targeting the desired region of genomic egl-38 was chosen using the MIT CRISPR design tool (Zhang lab, MIT, http://crispr.mit.edu). The guide RNA was ordered and harnessed to the Synthego EZ Scaffold, providing a 100- mer guide RNA (Synthego, Inc.). PCR-generated homology arms, with a 3-point mutated PAM sequence (primers in Table 1) were assembled into the

GFP^SEC^3xFlag vector pDD282 (NEBuilder HiFi DNA Assembly Cloning Kit E5520S) and confirmed by PCR and sequencing (Dickinson et al. 2013). Injection mixes contained 50 ng/µL repair template, 20 ng/µL myo-2::mCherry (pCFJ90), 300 mM KCl, and 20mM HEPES, mixed with 5 µM guide RNA and 5 µM Cas9 enzyme (Frøkjaer-

Jensen et al. 2008; Prior et al. 2017). This mix was injected into the gonads of N2 adult worms; three days later, 250 µg/mL hygromycin was added to each plate to select for transformed F1 offspring. On days six-eight surviving Roller, non-myo-2::mCherry expressing F1 adults were singled out and allowed to self-cross. Multiple offspring of

Rollers were selected and allowed to self-cross, and plates with a homozygous parent were identified. These animals were heat-shocked to remove the self-excising cassette, and screened for wild-type movement and GFP expression. Correct insertion of the

GFP was confirmed by PCR and sequencing with flanking genomic primers. The tagged strain is viable, fertile, and appears wild type.

2.2.3 Generating nlp-2 and nlp-7 reporters and deletion transgenes

Reporter strains for nlp-2 and nlp-7 were constructed using pDD95.69 from the

Fire Vector kit (Addgene.org). For nlp-2, 2.6 kB of upstream genomic promoter sequence was amplified using a 5’ SphI-tagged primer and a 3’ SalI-tagged primer; the

2.6 kb amplicon includes only the start codon from the coding sequence. Similarly, for nlp-7, 2.7 kb of upstream promoter (including the start codon) was amplified using a 5’

Sal-tagged primer and a 3’ MscI-tagged primer. The amplicons of both genes were ligated into the promoter-less pPD95.69 plasmid and transformed into DH5α bacteria. nlp-2 deletion clones were generated by designing upstream primers tagged with SphI that were more proximal to the translational start site, and paired with the same 3’ primer. Site-directed mutagenesis of the nlp-2 promoter sequence in pRJ124 (0.4 kb) were performed using standard reagents, according to the Stratagene protocol

(Stratagene Quikchange II Site-Directed Mutagenesis Kit). Positive clones were verified by restriction enzyme digest and sequencing. Injection mixes containing 75 ng/µL of confirmed plasmid and 15 ng/ µL of the rescue plasmid unc-119(pTJ1043) were injected into RH10 (unc-19(e2498)) adult hermaphrodites. non-Unc F1 offspring were selected and allowed to self to identify transgenic lines. At least three lines were obtained and screened for fluorescence. The reported data represent an individual line with the most consistent expression pattern.

2.2.4 Microscopy

Morphological and fluorescent phenotypes were evaluated using a Zeiss

Axioplan microscope, under 100x magnification. Hermaphrodite animals were observed at mid-L3 larval stage through adult, depending on the experiment. Larval stage was defined on the basis of the vulval morphology in the mid-plane. Comparably-staged adult animals were evaluated by selecting L4 animals to fresh plates, and evaluating them 24 hours later. Images were taken at auto-calculated DIC and epi-fluorescent exposures, varying from 0.5-0.8 ms and 1.2-1.9 ms respectively. Worms were immobilized on agar pads consisting of 3.5% noble agar in water with a 10 mM sodium azide solution. Screening for vulval cell and uv1 phenotypes involved locating the vulval or uv1 cells in the uppermost plane of the worm; expression in uv1 cells on the side where the worm was laying was not evaluated due to variability in the data due to photo-bleaching of the GFP.

2.2.5 U0126 experiments

U0126 was dissolved in dimethyl sulfoxide (DMSO) to a stock concentration of

10mM. U0126 and DMSO control plates were made by adding 15 µM of U0126 or

DMSO to the top of 5 mL NGM in 35 mm petri plates, drying overnight, and spotting with

OP50 bacteria. Gravid worms were floated from plates using M9, dissolved with a bleach solution, and the remaining eggs were rinsed three times with M9. The eggs were hatched overnight (~18 hours) with rotation. Synchronized L1 larvae were plated onto standard NGM plates and placed at 20°C. Thirty-two hours later, the L3 larvae

31 were removed and plated to the U0126 or DMSO plates, and returned to 20°C.

Screening was performed 8 hours later, when larvae are mid-L4.

2.2.6 Protein Expression and EMSA

The EGL-38 Paired DNA-binding domain was expressed using a pET23a vector in E. coli BL21 cells (strain MC702) (G. Zhang et al. 2005). Expression and lysis were performed as previously described, with empty-vector cultures (strain MC706) as a negative control (Fitzsimmons et al. 2001). In summary, 400 µL of bacterial culture were added to 40 mL LB with 50 µg/mL Carbenicillin. This was grown at 37°C to an OD600 of

~4, before IPTG was added (final concentration 1mM) and cultures shook at room temperature overnight. Cells were pelleted and re-suspended in 1 mL cold buffer Z (25 mM HEPES, pH 7.0; 100 mM Kcl; 12.5 mM MgCl2; 20% glycerol; 0.1% NP-40; 1 mM

DTT). Cells were sonicated, lysed, and the supernatants stored on ice for use in EMSA.

EMSA was carried out as described (Roche DIG Gel Shift Kit, 2nd Generation,

#3353591910). 50 mM MgCl2 was included in the binding (Johnson et al. 2001). Protein lysates were diluted 1:50 in Buffer Z, and incubated with 2 ng labelled probes, and

50x/250x un-labelled competitors. Samples were separated on a 6% non-denaturing gel, with 0.5x TBE buffer, transferred to nylon membrane in 0.5x TBE, and performed detection following the DIG Gel Shift protocol.

2.3 Results

2.3.1 EGL-38 protein is expressed during the development of the C. elegans egg- laying system

Genetic experiments suggest that egl-38 functions in both the vulva and the uterus of C. elegans (Rajakumar and Chamberlin 2007). Previously, however, egl-38 expression was evaluated only with transcriptional reporter transgenes, which do not inform us about protein expression and localization. Therefore, we utilized CRISPR- mediated genome editing to tag the endogenous locus so that the EGL-38 protein is tagged on the C-terminus with GFP. We find that in this strain, we detect EGL-38::GFP protein fluorescence in all of the expected hermaphrodite tissues, including the vulva, uterus, and hindgut (not shown) and that the protein localizes to the nuclei of cells as predicted for a transcription factor protein. We evaluated protein expression at different time points, beginning in the early L3 stage prior to vulval precursor cell (VPC) division

(FIGURE 4). During early L3, we observed that EGL-38::GFP is absent from egg-laying system cells, but is first detected in both P6.px daughters and all four P6.pxx granddaughters following the first and second divisions of the VPCs, respectively

(FIGURE 4A-F). After the final VPC divisions, the cells vulF and vulE (derived from

P6.p) express EGL-38::GFP; this expression persists through early L4 into mid-L4

(FIGURE 4G-P). At the mid-L4 time point, expression in the uv1 cells can be detected concurrent with vulF and vulE expression (FIGURE 4M-R). EGL-38::GFP expression in vulF and vulE diminishes during mid-L4 so that by the transition to late L4, only uv1 cell expression remains (FIGURE 4S-T). This timing is consistent with the idea that this

33 expression participates in, or is responsive to, the developmental coordination between the two tissues. The uv1 cells continue to express EGL-38::GFP through the adult molt into adulthood (FIGURE 4S-X). Thus EGL-38 protein is expressed in both vulval and uterine cells consistent with the genetic functions of egl-38. EGL-38::GFP expression continues into adulthood in the uv1 cells, but not the vulval cells, indicating that EGL-38 may have a role in the function of mature uv1 cells.

2.3.2 EGL-38 expression in uv1 is dependent on EGF signaling

egl-38 mutants exhibit defects in both vulF and uv1 cell specification. In particular, egl-38 mutants are defective in vulF expression of lin-3/egf, which is used to signal to presumptive uv1 cells to promote their development (Chang, Newman, and

Sternberg 1999; Rajakumar and Chamberlin 2007). LIN-3/EGF-initiated uv1 cell specification is marked by two features: proper placement of the uv1 cells near the vulva, and expression of uv1-specific genes. If this EGF signaling is disrupted, then the presumptive uv1 cells migrate away from the vulval region and fuse with the uterine seam cell as do the other neighboring uterine (π) cells (Newman and Sternberg 1996;

Rajakumar and Chamberlin 2007). We have demonstrated that EGL-38::GFP is expressed in both the signaling and responding cells of this developmental event. To investigate the relationship between EGL-38::GFP expression and EGF signaling, we utilized both genetic disruption and chemical inhibition. For an initial genetic approach, we evaluated EGL-38::GFP expression in animals mutant for the EGF receptor gene, let-23(sy1). As this is a non-null allele, we focused solely on vulvaless let-23(sy1);egl-

38::GFP worms, to ensure that vulF cells are absent and unable to communicate with the uterine cells. These worms also lack vulval EGL-38::GFP, as there are no induced vulval cells. We find that placement of presumptive uv1 cells is eliminated in the majority of vulvaless let-23(sy1) worms, and that all lack any uv1 EGL-38::GFP expression

(FIGURE 5). Thus expression of EGL-38::GFP in the uv1 cells is dependent on let-

23/EGFR activity in the egg-laying system. However, since these animals are both mutant for let-23/EGFR and vulvaless, we sought to more directly test whether EGL-

38::GFP is dependent on vulval cell signaling using the MEK inhibitor U0126 (Duncia et al. 1998). Specifically timed application of U0126 allows blocking of the vulF to uv1 EGF signal without affecting the original EGF signal from the anchor cell that triggers the

VPCs to divide and form the vulva. We treated late L3 EGL-38::GFP worms with U0126, and find that these animals have morphologically normal vulvas while lacking uv1 cell specification. Similar to the let-23(sy1) mutants, these animals are disrupted for uv1 cell placement, and lack uv1 EGL-38::GFP expression (FIGURE 5). We note that the effect of U0126 is slightly stronger than the effect of let-23(sy1) in this experiment and the ones below. This may reflect the fact that let-23(sy1) is a non-null allele, and retains some activity and that there are non-vulval sources of diffusible lin-3/egf that may compensate for the missing vulF signaling in let-23(sy1) mutants (Aroian et al. 1990;

Aroian and Sternberg 1991; Hwang and Sternberg 2004). Nevertheless, in both let-

23(sy1) mutants and U0126-treated animals, uv1 cell placement near the vulva was disrupted, and EGL-38::GFP expression was absent. Together, these data argue that

EGL-38 expression in uv1 is dependent on EGF signaling from the vulF cells.

2.3.3 Expression of neuropeptide genes nlp-2 and nlp-7 in the uv1 cells is dependent on EGF signaling and egl-38

In the vulF cells, a critical function for egl-38 is to promote expression of lin-3/egf; however, EGL-38 targets and functions in the uv1 cells are unknown (Chang, Newman, and Sternberg 1999). The uv1 cells are neurosecretory cells and inhibit egg-laying in response to the passage of eggs out of the uterus (Banerjee et al. 2017; Collins et al.

2016; Jose et al. 2007). To determine if EGL-38 has any direct targets within the uv1 cells, we sought to identify genes which are expressed in the mature uv1 cells. Previous work led us to focus on Neuropeptide-like proteins (nlps), as one such nlp, nlp-7, has been identified as necessary for uv1 inhibition of egg-laying activity (Banerjee et al.

2017). nlps are non-insulin and non-FMRFamide related peptides, with little known about most members of this family (Nathoo et al. 2001). To determine if any additional nlps are expressed in uv1 cells, therefore representing a possible EGL-38 activation target, we tested five genes, nlp-2, 6, 7, 9, and 12. We constructed a reporter with ~2 kb of upstream promoter sequence driving the expression of gfp. Only Pnlp-2::gfp and

Pnlp-7::gfp were found to express in uv1 beginning in mid-L4, and persisting into the adult stages. The study which identified nlp-7 involvement in egg-laying also found that the FMRF-like peptide-encoding gene, flp-11, functions with nlp-7 in the uv1 cell.

However, we were not able to detect expression of Pflp-11::gfp in uv1 cells using available reporters. Consequently, we focused our studies on nlp-2 and nlp-7.

We first sought to determine if Pnlp-2::gfp and Pnlp-7::gfp expression is dependent on uv1 cell specification and if they are markers of mature uv1 cell fate. To

36 do so, we examined the interaction of Pnlp-2::gfp and Pnlp-7:gfp with the EGF pathway by first utilizing the let-23(sy1) allele (FIGURE 6A-Q). In let-23(sy1); Pnlp-2::gfp and let-

23(sy1); Pnlp-7::gfp vulvaless animals, expression of the reporter was reduced (to 50% and 25%, respectively). As with EGL-38::GFP, we also examined reporter expression when late L3 animals were treated with U0126 to specifically block signaling from vulF to the uv1 cells. For both Pnlp-2::gfp and Pnlp-7::gfp U0126-treated animals, none of the worms showed reporter expression. Neither reporter exhibits expression in non-uv1 uterine cells in any background, while expression in neuronal cells outside the egg- laying system continued unabated in all altered conditions. Thus, nlp-2 and nlp-7 exhibit a similar dependence as EGL-38 on the EGF pathway for uv1 cell expression. Given that expression from both reporters in the egg-laying system occurs only in the differentiated uv1 cells, it is suggestive that both nlps function in adult uv1 cell activity.

We next investigated the dependence of nlp-2 and nlp-7 expression on EGL-38 activity in the mature uv1 cells. We crossed the reporter transgenes into two egl-38 loss-of-function backgrounds, egl-38(n578) and egl-38(sy294) (FIGURE 6R-Z). These alleles both affect egg-laying system development (Rajakumar and Chamberlin 2007).

Expression from the reporter transgenes for both genes is affected. Pnlp-7::gfp expression is eliminated in both egl-38(n578) and egl-38(sy294) backgrounds while

Pnlp-2::gfp exhibited significantly reduced expression. These results demonstrate that both nlp-2 and nlp-7 expression is additionally dependent on egl-38 for expression in the uv1 cells.

2.3.4 EGFR activation bypasses the requirement for egl-38 to promote uv1 cell placement, but not neuropeptide gene expression.

Our EGL-38::GFP expression results and previous genetic mosaic results indicate that EGL-38 is present and can act in both vulF and uv1 cells (Rajakumar and

Chamberlin 2007). However, since egl-38 is required for expression of lin-3/egf in vulF cells, it is possible that expression of nlp-2 and nlp-7 is due indirectly to EGL-38 activity in vulF and not directly to its activity in uv1. To separate these possibilities, we examined the epistatic relationship between let-23/egfr and egl-38 in regard to nlp-2 and nlp-7. Utilizing a let-23(sa62) gain-of-function allele, which remains constitutively active in the absence of the lin-3/egf signal, we created double mutants with egl-38 (FIGURE

7). For the Pnlp-2::gfp strain, we utilized egl-38(sy294), as this mutation had a stronger effect on Pnlp-2::gfp expression, whereas for Pnlp-7::gfp, we employed egl-38(n578).

For both reporters, let-23(sa62) background predictably resulted in wild-type levels of uv1 cell placement and in reporter expression. There was no over-production of uv1-defined cells or other extraneous cells in the egg-laying system. We constructed double mutants let-23(sa62); egl-38(sy294); Pnlp-2::gfp and let-23(sa62); egl-38(n578);

Pnlp-7::gfp. For both strains, the uv1 cell placement defect found in egl-38(lf) worms was rescued by let-23(sa62), whereas the expression of Pnlp-2::gfp and Pnlp-7::gfp was not. Importantly, our data show that placement of the uv1 cells near the vulva and expression of nlp-2 and nlp-7 in the uv1 cells are distinct events with separate dependencies on egl-38 activity. Consequently, we interpret that nlp-2 and nlp-7 expression in the uv1 cells is dependent on egl-38 autonomous activity from within the

38 uv1 cells. Because both nlp-2 and nlp-7 are expressed only in the differentiated uv1 cells, their dependence on EGL-38 indicates that a function for EGL-38 within uv1 is to activate developmental programs necessary for egg-laying.

2.3.5 nlp-2 is a direct target for EGL-38 in uv1 cells

We chose nlp-2 to further examine the relationship of EGL-38 with the uv1 cell neuropeptide genes. The sequence included in the Pnlp-2 reporter consists of around

2.6 kb of upstream sequence. To identify the regulatory element responsible for uv1 cell expression, truncations were introduced beginning at the 5’ upstream end of the original

2.6 kb fragment to create successively smaller reporter genes, and the expression profile of the subsequent reporter was analyzed (FIGURE 8). Due to the abrupt loss of uv1 expression upon deletion, we interpret that a uv1 cell enhancer element exists between 0.34 and 0.38 kb from the start codon of nlp-2. We evaluated this sequence for prospective Pax binding sites using the lre2 sequence from lin-48, a known target for

EGL-38 in the hindgut cells (FIGURE 8B) (Johnson et al. 2001). We identified one site with some similarity, which we term nlp-2 response element 1 (nre1). We introduced mutations into this sequence to alter the potential binding sites for the N-, the C-, or both termini of the EGL-38 DNA-binding domain. Only transgenes with both termini mutations exhibited a strong reduction of Pnlp-2::gfp expression, although not to the extent observed when the nre1 enhancer region is deleted (FIGURE 8A). We utilized

EMSA to ask whether the DNA binding domain of EGL-38 protein can directly bind the genetically required regulatory sequence, nre1, in vitro. We found that a bacterially

39 expressed Paired Domain of EGL-38 (EGL-38 PD) caused a mobility shift in labelled nre1 probe migration (FIGURE 8C). This shift was specific to lysates expressing the

EGL-38 PD, as lysates expressing the empty vector failed to cause a similar shift.

Further, the binding of EGL-38 PD to nre1 was competed away in a dose-dependent manner by the known EGL-38 target sequence lre2, but not by a probe with partially overlapping sequence to nre1 (Johnson et al. 2001). Altogether, these data argue that nlp-2 is a direct target for EGL-38 in the uv1 cells.

2.4 Discussion

egl-38 has previously been shown to participate in the development of vulval and uterine cells necessary for egg-laying in C. elegans (Chamberlin et al. 1997; Chang,

Newman, and Sternberg 1999; Rajakumar and Chamberlin 2007). However, past work focused on its functions in the vulF cells. Here, we have explored more in depth the interaction of egl-38 with the signaling end of the EGF pathway (non-autonomous functions) and the activity of egl-38 within the responding uv1 cells for execution of cellular programs (autonomous functions). EGL-38::GFP protein is present in both the vulF and uv1 cells at developmental times when it can be functionally responsible for, and responsive to, the EGF pathway. In fact, EGL-38::GFP expression in uv1 is dependent on the EGF pathway, and egl-38 is also required for expression of the neuropeptide genes nlp-2 and nlp-7, which have a similar dependence on the EGF pathway as EGL-38. In addition, activated LET-23/EGFR can bypass the requirement for egl-38 in uv1 cell placement, but not in promoting Pnlp-2::gfp and Pnlp-7::gfp expression, showing that uv1 cell positioning and gene expression are two unique, separate aspects of uv1 cell fate. Ultimately, EGL-38 activity in uv1 is coordinated to

EGL-38 activity in vulF through the EGF signaling pathway. EGL-38 activates lin-3/egf in the vulF cells; EGF signaling from vulF directs the expression of EGL-38 in the uv1 cells; EGL-38 in uv1 activates transcription of nlp-2 and nlp-7 to mediate cellular functions.

Consistent with an active role for egl-38 in uv1 cells, our results argue that nlp-2 is a direct target for EGL-38, as EGL-38 is capable of binding in vitro to a uv1 cell

41 specific enhancer of nlp-2. We find that nlp-7 expression likewise exhibits dependence on egl-38, although a direct interaction has not been established. nlp-2 and nlp-7 likely both act during egg-laying. The uv1 cells are generally inhibitory, releasing neuropeptides to control the periodicity of egg-laying following mechano-sensory deformation (Alkema et al. 2005; Collins et al. 2016; Jose et al. 2007). uv1 cells synapse with the hermaphrodite specific neurons; the lack of a direct junction indicates that uv1 activity must be through processes such as neuropeptide vesicle release

(Banerjee et al. 2017). Consistent with this idea, nlp-7 (with flp-11) has been shown genetically to inhibit egg-laying (Banerjee et al. 2017). Thus we find that EGL-38 has uv1 cell targets important for execution of a differentiated fate, in addition to its roles in coordinating development of egg-laying system cells.

Pax proteins orthologous to EGL-38 are necessary components of cellular development and differentiation for numerous systems, such as the kidney (Pax2/8), thyroid (Pax8), and B cells (Pax5) (G. R. Dressler et al. 1990; Lang et al. 2007;

Macchia et al. 1998; Magliano, Lauro, and Zannini 2000; Torres et al. 1995). In most organs, such as the kidneys, Pax protein expression is significantly downregulated in the differentiated tissue (Blake and Ziman 2014; Czerny, Schaffner, and Busslinger

1993; G. R. Dressler et al. 1990; Lang et al. 2007; Rothenpieler and Dressler 1993;

Terzić et al. 1998). However, in some systems, such as the thyroid, Pax activity is necessary after cell specification for expression of terminally differentiated genes required for cellular function. Pax8 is necessary for establishment of the thyroid follicular cells, as well as for expression of thyroid genes including thyroglobulin and

42 thyroperoxidase (Tg and TPO) (Magliano, Lauro, and Zannini 2000; Mansouri,

Chowdhury, and Gruss 1998; Miccadei et al. 2002). Interestingly, a study found that in thyroid-derived cell lines, re-introduction of Pax8 resulted in expression of endogenous targets without fully rescuing the differentiated follicular cell phenotype, indicating that while Pax8 is necessary for both processes, only expression of terminally differentiated targets directly involves Pax8 (Fabbro et al. 1994; Magliano, Lauro, and Zannini 2000).

Similarly, we have shown in C. elegans that EGL-38 is present both in developing cells of the vulva and in the differentiating uv1 cells, but that EGL-38 also persists in uv1 cells into adulthood. EGL-38 protein becomes abruptly undetectable in the vulF and vulE cells at a time point following expression of the lin-3/egf signal in early/mid-L4 larval stage. In the uv1 cells, expression tapers off more gradually as the worms are in the adult stage, presumably accommodating the EGL-38 role in activating nlp-2 and nlp-7, which are involved in the adult egg-laying function. Both the thyroid and the uv1 cells are neurosecretory organs with important physiological signaling functions in the body

(Collins et al. 2016; Jose et al. 2007). Thus a common theme is that Pax2/5/8 activity participates in both the differentiation and function of these neurosecretory organs.

An interesting facet of Pax2/5/8 gene function is that their role in cellular differentiation, combined with temporal and spatial expression patterns, indicates that participating in inductive cell-to-cell signaling may be a common theme for proteins in this group (G. R. Dressler et al. 1990; Püschel, Gruss, and Westerfield 1992). For example, in HeLa, CaSki, and renal proximal tubule cell cultures, PAX2 expression was increased by exposure to EGF (de Graaff et al. 2012; S. Liu et al. 1997). Similarly, in

43 both eutopic and ectopic tissue of endometriosis patients, the expression of PAX2 and

EGFR was tightly correlated, indicating that PAX2 expression may be dependent on

EGFR activity (de Graaff et al. 2012). Aberrant expression of PAX proteins in cancer tissue is concomitant with high levels of EGF and hallmarks of increased EGF activity, such as the maintenance of tumor growth, proliferation, and survivability (Brand et al.

2011; Herbst 2004; Konecny et al. 2009; Muratovska et al. 2003; J. Wang et al. 2018;

Q. Wang et al. 2008). Knockdown of PAX2 in endometrial cancer lines significantly decreases cancer cell viability (Jia et al. 2016; L.-P. Zhang et al. 2011). The relationship of these PAX2/5/8 proteins with inductive signaling pathways has not yet been examined during mammalian development, given the lethal or significant loss-of- function phenotypes that result from loss of either the Pax2/5/8 gene or of signaling activity. However, our study has observed parallels between Pax2/5/8 and the EGF pathway that were highlighted by these adult disease state and cell culture experiments.

Therefore, coordination of signaling and responding within a developing organ may be a common developmental function for Pax transcription factors.

2.5 Figures

Figure 3: C. elegans hermaphrodite vulval development.

Fig. 3 (A) A representative schematic of the C. elegans hermaphrodite at the late L3 stage of development. The developing egg-laying system is found on the ventral side of the worm, mid-way down the body. (B-G) A brief representation of egg-laying system development through time. (B) early L3. The LIN-3/EGF signal is sent from the Anchor

Cell (AC) to the P6.p cell to specify it as the 1° cell. (C) late L3. P6.p divides, forming 4 granddaughters by late L3. (D) L3/L4 molt. The 1° cell descendants have divided, and

(in a single plane) there are two vulF and two vulE cells. The π cells have been

45 specified surrounding the AC. (E) early L4. The vulF cells at the apex of the vulva send a reciprocal LIN-3/EGF signal back to the nearest π cells. (F) mid L4. The now specified uv1 cells move closer to the vulva. (G) late L4. The uv1 cells are in position right next to the vulval cells to serve as an anchor of vulva to uterus.

Figure 4: EGL-38::GFP is expressed during egg-laying system development

Fig. 4 (A-X) Expression of tagged EGL-38::GFP in the mid-plane view of the developing hermaphrodite egg-laying system. Individual worms were staged and screened for a

47 single time point. Cells of interest are indicated with an arrowhead in both DIC and fluorescent panels. (A-B) Early L3. The P6.p cell lacks expression (empty arrowhead).

(C-D) Mid-L3. The two P6.p daughter cells strongly express EGL-38::GFP. (E-F) Late

L3. The four P6.p granddaughters all accumulate EGL-38::GFP (third arrowhead rotated to allow visualization of the AC immediately above). (G-J*) L3-L4 molt. Same animal, image plane shifted. vulF (G-H) and vulE (I-J) cell expression. (K-L) Early L4. Presence in vulF and vulE continues. (M-R†) Mid-L4. Same animal, image plane shifted. EGL-

38::GFP continues in vulF (M-N) and vulE (O-P) as expression begins in the presumptive uv1 cells prior to positioning (Q-R). (S-T) Mid-L4. By the end of the mid-L4 stage, uv1 expression is apparent while fluorescence in vulF and vulE has diminished to undetectable levels. (U-V) Late L4. uv1 cell expression continues. (W-X) Adult. EGL-

38::GFP persists in uv1 cells into young adulthood, but is not as consistent as the worms age. (Y) Time-course trend of EGL-38::GFP expression. Values represent average number of GFP-positive cells, with each time point corresponding to 30 animals. To limit photo bleaching, data were collected from only one side of each worm, but are extrapolated to include cells on the other side of the developing vulva to represent all of the cells of each structure (a full eight vulval cells by late L3, and four uv1 cells by mid-L4).

Work performed: I designed the primers and guide RNA needed for targeting the egl-38 locus. Thank you to Dr. Chamberlin for injecting, screening, and establishing the transformed EGL-38::GFP line. I performed all screening for uv1 cell placement and fluorescence, as well as prepared and imaged larvae on the time course.

Figure 5: EGL-38::GFP expression is dependent on EGF signaling.

Fig. 5 (A-B) Wild type late L4. EGL-38::GFP is present in uv1 cells (solid arrowhead).

(C-D) let-23(sy1). uv1 cells are absent in the loss-of-function let-23 background, and

EGL-38::GFP expression is also absent in this background (approximate position marked with an empty arrowhead. uv1 positioning can vary between worms). (E-F)

DMSO control treatment on wild-type animals. uv1 placement and expression of EGL-

38::GFP match wild type in the control treatment. (G-H) U0126 treatment. Blocking the

EGF pathway in mid L3 stage results in a lack of uv1 cells, and absent EGL-38::GFP expression. (I) Percent of worms with correctly placed uv1 cells and with expression of

EGL-38::GFP in the uv1 cells. Above each bar is shown the number of worms screened. Error bars are shown for the standard error of proportion. For each pairwise comparison of control to mutant/treatment (wild type to let-23(sy1); DMSO to U0126), the difference is significant at a p<0.05 for each pairwise comparison of experiment to control (2-tailed Z-test).

Work performed: Thank you to Dr. Chamberlin for creation of the mutant::reporter strains used here. I performed all screening for uv1 cell placement and reporter fluorescence, as well as prepared and imaged larvae. I also performed all U0126/DMSO experiments.

Figure 5: EGL-38::GFP expression is dependent on EGF signaling.

Figure 6: Expression of nlp-2 and nlp-7 in uv1 cells is dependent on EGF signaling and egl-38.

Fig. 6 (A-H) Pnlp-2::gfp expression. (A-B) Wild type. Pnlp-2::gfp expression in mid-L4

(solid arrowhead). (C-D) let-23(sy1) background. Worms lack uv1 placement and Pnlp-

2::gfp expression (approximate position of a normal uv1 cell marked with an empty arrowhead). (E-F) DMSO control treatment. uv1 cell placement and expression matches wild type in late L4. (G-H) U0126 treatment. uv1 cells are absent, as is Pnlp-2::gfp expression. (I-P) Pnlp-7::gfp expression. (I-J) Wild type. Pnlp-7::gfp expression in mid-

L4. (K-L) let-23(sy1) background. uv1 cell placement and Pnlp-7::gfp are both absent.

(M-N) DMSO control treatment. Worms in late L4 have uv1 placement and expression as in wild type. (O-P) U0126 treatment. uv1 cells and expression are absent. (Q)

Percent of worms with correctly placed uv1 cells or with expression of Pnlp-2::gfp or

Pnlp-7::gfp in the uv1 cells. Above each bar is shown the number of worms screened.

Error bars are shown for the standard error of proportion. For each pairwise comparison of control to mutant/treatment (wild type to let-23(sy1); DMSO to U0126), the difference is significant at a p<0.05 for all pairwise comparisons (2-tailed Z-test). (R-Y) Pnlp-2::gfp and Pnlp-7::gfp expression in the mid-plane of L4 hermaphrodites in wild type or loss-of- function egl-38 backgrounds. (R-U) Pnlp-2::gfp expression. (R-S) Wild type. uv1 cell placement and Pnlp-2::gfp expression in mid/late L4. (T-U) egl-38(sy294) background.

Worms lack uv1 cells and expression from the reporter in loss-of-function egl-38. (V-Y)

Pnlp-7::gfp expression. (V-W) Wild type. uv1 placement and Pnlp-7::gfp expression in mid/late L4. (X-Y) egl-38(n578) background. uv1 cells and Pnlp-7::gfp expression are both absent in the egl-38 mutant. Z) Percent of worms with correctly placed uv1 cells or

51 with expression of Pnlp-2::gfp or Pnlp-7::gfp in the uv1 cells in wild type or loss-of- function egl-38 background. Above each bar is shown the number of worms screened.

Error bars are shown for the standard error of proportion. For each pairwise comparison of control to mutant (wild type to egl-38(n578 or sy294)), the difference is significant at a p<0.05 for all pairwise comparisons (2-tailed Z-test).

Work performed: Thank you to Dr. Chamberlin for creation of the double mutants::reporter and mutant::reporter strains used here. I performed all screening for uv1 cell placement and reporter fluorescence, as well as prepared and imaged larvae.

Figure 6: Expression of nlp-2 and nlp-7 in uv1 cells is dependent on EGF signaling and egl-38.

Figure 7: Activated let-23/EGFR can bypass the egl-38 defect in uv1 placement, but not nlp gene expression.

Fig. 7 (A-L) Pnlp-2::gfp and Pnlp-7::gfp expression in the mid-plane of L4 hermaphrodites in wild type, egl-38(lf), and let-23(gf);egl-38(lf) backgrounds. uv1 cells with accurate placement are indicated with a solid arrowhead; absent uv1 cells are marked at the approximate position with an empty arrowhead. uv1 positioning can vary between worms. (A-F) Pnlp-2::gfp expression. (A-B) Wild type. uv1 cells are correctly placed and expressing Pnlp-2::gfp. (C-D) egl-38(sy294) background. Worms lack uv1 cells and expression. (E-F) let-23(sa62); egl-38(sy294) background. uv1 cell placement has been restored, but Pnlp-2::gfp expression remains absent. (G-L) Pnlp-7::gfp expression. (G-H) Wild type. Worms have uv1 cell placement and expression of reporter. (I-J) egl-38(n578) background. Both uv1 cell placement and Pnlp-7::gfp expression is absent. (K-L) let-23(sa62);egl-38(n578) background. uv1 cell placement has been restored, but Pnlp-7::gfp expression is absent. (M) Percent of worms with correctly placed uv1 cells or with expression of Pnlp-2::gfp or Pnlp-7::gfp in the uv1 cells in wild-type or mutant background. Above each bar is shown the number of worms screened. Error bars are shown for the standard error of proportion. For each pairwise comparison of control to mutant the difference is significant at a p<0.05 for all comparisons (2-tailed Z-test).

Figure 7: Activated let-23/EGFR can bypass the egl-38 defect in uv1 placement, but not nlp gene expression.

Figure 8: nlp-2 is a direct target for EGL-38.

Fig. 8 (A) Percent of worms with expression of Pnlp-2::gfp or its deletion derivatives. A full-length reporter (shown at top) was constructed and then deletions taken from the reporter to isolate a uv1 cell enhancer region. On the left is shown the length of upstream promoter sequence which was included in the reporters. Worms were

56 screened in mid-late L4 for uv1 expression of the deletion reporter. Deletion of between

0.34 and 0.38 kb upstream from the start codon significantly reduces reporter expression. We define this 40 bp region as the uv1 cell enhancer for nlp-2. Point mutations in this enhancer made to the sequences hypothesized to bind the N-, C-, and both termini (NC-) of EGL-38 significantly decrease reporter expression. Residual reporter fluorescence in the mutant construct may be due to the degenerate Pax DNA- binding domain consensus sequence (Czerny, Schaffner, and Busslinger 1993), or failure to identify the key nucleotides. Data are recovered from more than 40 worms per reporter; error bars are shown for the standard error of proportion. For each pairwise comparison of control to mutant the difference is significant at a p<0.05 (comparing 0.4 kb N-, 0. 4 kb NC-, 0.34 kb, 0.3 kb, 0.2 kb, and 0.1 kb to wild type). (B) Comparison of the Pax binding enhancer regions from lin-48 (lre2; Johnson et al., 2001) and nlp-2

(nre1; this work). The mutations introduced in the 0.4 kb reporter are indicated under nre1 (NC-). (C) EMSA showing the EGL-38 DNA-binding Paired domain interaction with the nre1 probe containing the nlp-2 uv1 cell enhancer. In lane 1, no lysate is added.

Lane 2 is lysate from bacteria bearing the empty vector. In lane 3, lysate from bacteria expressing the paired domain of EGL-38 (indicated as E38) can bind to the nre1 probe and shift mobility. In lanes 4 and 5, this binding is competed away by increasing amounts of unlabeled lre2, a sequence known to bind specifically to EGL-38. In lane 6, an unlabeled negative probe that corresponds to nlp-2 sequences that partially overlap lre2 is unable to compete away E38 binding to nre1.

Work performed: Thank you to Ryan Johnson for creation of the Pnlp-2::gfp reporter strain and the deletion analysis and EMSA performed here.

Chapter 3: EGL-38 activity is partially restricted in the vulF cell and necessary in the uv1 cell for egg-laying behavior in Caenorhabditis elegans hermaphrodites

3.1 Introduction

Paired-box proteins are extremely important transcription factors necessary for coordination of development, present in nearly every system of the human body (Blake and Ziman 2014; G. R. Dressler et al. 1990; Lang et al. 2007; Macchia et al. 1998;

Magliano, Lauro, and Zannini 2000; Noll 1993; Torres et al. 1995). Pax transcription factors help control cell specification, growth, and proliferation as well as initiate cellular programs required for cell activity (Mansouri, Chowdhury, and Gruss 1998; Miccadei et al. 2002; Noll 1993). Despite their prominence in development, facets of Pax activity are poorly understood, such as how Pax is regulated in disparate cell types, or how Pax functions after development to initiate phenotypic cellular programs. I have investigated how the activity of a Pax2/5/8 ortholog, EGL-38, is restricted to a specific cell type and how it affects biological cellular output in the egg-laying system of the nematode

Caenorhabditis elegans.

Development of the C. elegans egg-laying system relies upon the Epidermal

Growth Factor (EGF) pathway for signaling to both vulval and uterine cells to specify cell identities that serve as structural and functional components of the system (Chang,

Newman, and Sternberg 1999; Katz et al. 1996; Newman, White, and Sternberg 1996).

The EGF signal, LIN-3, is first sent from the uterine anchor cell to the vulval precursor cells at the early L3 stage of larval development to induce these vulval cells to divide (R.

J. Hill and Sternberg 1992; Katz et al. 1996). By the late L3 stage, cells at the apex of the vulva (vulF) express lin-3/egf in turn, and signal to four uterine π cells to adopt a uterine ventral (uv1) cell fate (Chang, Newman, and Sternberg 1999). These uv1 cells will migrate next to the vulF cells and elongate to form an anchor of the vulva to the uterus, allowing the vulF cells to separate and form a vulval-uterine connection

(Newman and Sternberg 1996). Additionally, the neurosecretory uv1 cells serve as inhibitors of egg-laying to set the length of inactive periods for the adult hermaphrodite

(Banerjee et al. 2017; Collins et al. 2016; Jose et al. 2007).

egl-38 functions in both the epithelial vulF cells and the mesodermal uv1 cells

(Chapter 2) (Rajakumar and Chamberlin 2007). EGL-38 is required for lin-3/egf expression, uv1 cell specification, and vulF cell separation; all factors necessary for establishment of the vulval-uterine connection (Chang, Newman, and Sternberg 1999;

Rajakumar and Chamberlin 2007). In an egl-38 mutant, the vulF cells lack lin-3/egf expression, while expression of lin-3/egf in other cells of the egg-laying system, such as the anchor cell, is unaffected (Chang, Newman, and Sternberg 1999). The lack of lin-

3/egf signal from vulF in an egl-38 mutant results in no uv1 cell specification, causing the presumptive uv1 cells to retain their π cell identity, migrate away, and fuse with the anchor cell into a syncytium (Newman, White, and Sternberg 1996; Rajakumar and

Chamberlin 2007). This prevents creation of the vulval-uterine connection necessary for egg-laying. Additionally, I have demonstrated that egl-38 is necessary for expression of two neuropeptide-encoding genes in uv1, one of which has a known function in the inhibition of egg-laying (Chapter1). nlp-2 and nlp-7 are dependent on egl-38 and the

EGF pathway for expression in the mature uv1 cells (Chapter 1), and nlp-7 is required, with flp-11, to set the periodicity of egg-laying (Banerjee et al. 2017). While EGL-38 activity is necessary for establishing the morphology of the egg-laying system and for execution of egg-laying functions, aspects of its function are unknown and represent a significant hole in clarifying Pax activity in this system.

Here, I have investigated how EGL-38 activation of lin-3/egf is restricted to the vulF cell and how EGL-38 activation of nlp-2 and nlp-7 contributes to the egg-laying phenotype. I present that LIN-1/ETS is a possible co-factor which may be responsible for confining EGL-38 activation of lin-3/egf to the vulF cells specifically. I examine lin-

3/egf for a vulF enhancer element to examine whether EGL-38 can directly activate this promoter, and I present an nlp-7 uv1 cell enhancer element. Additionally, I have utilized loss-of-function nlp-2, nlp-7, and flp-11 mutants to examine the effect of their loss upon egg-laying behavior. Overall, my results support and extend previous findings about

EGL-38 function in the egg-laying system to activate the EGF signaling pathway and uv1 cellular programs.

This chapter is separated into two sections, the first containing my investigation into EGL-38 activities in the vulF cell, and the second containing my results of EGL-38 activities in the uv1 cell.

3.2 EGL-38 activity in the vulF cells

3.2.1 Introduction

Pax proteins often behave in an exceptionally nuanced manner to control cell specification, proliferation, execution of cellular programs, and other functions in organogenesis (Schulte 2014). Constraining Pax expression and activity, both temporally and spatially, is vital to allow the proper progression of events for development. For example, egl-38 is only expressed in mature uv1 cells in response to the EGF pathway, ensuring that EGL-38 targets are activated in the specified uv1 cells at the correct time and location for egg-laying behavior (Chapter 1). In addition to regulation of Pax expression, co-factor involvement has been speculated as a concise mechanism of restricting Pax activity to particular cells (Pufall and Graves 2002; Schulte

2014). Usage of a co-factor permits discrimination of related target sites, enhancement of binding specificity, and spatiotemporal control based on availability of both partners

(Fitzsimmons et al. 2001). One co-factor family shown to cooperatively bind with Pax is the Ets family.

The Ets transcription factors are a widely expressed family with roles in regulating growth, apoptosis, differentiation, specification, and development (Oikawa and Yamada 2003). Pax5 has been shown to recruit ETS proteins for binding at the mb-

1 promoter in B-lymphocytes (Fitzsimmons et al. 1996). Ets-1 binds the suboptimal sequence mb-1 promoter weakly; when Pax5 binds first and recruits Ets-1, Ets-1 affinity for the site increases 1000-fold (Garvie, Hagman, and Wolberger 2001). Previous

62 studies investigated the ability of EGL-38 to also bind this mb-1 sequence and determined that EGL-38 binds specifically, with the ability to recruit murine Ets-1 to this site (Fitzsimmons et al. 2001). In addition, they investigated the ability of EGL-38 to recruit the C. elegans Ets ortholog LIN-1; while LIN-1 could not be recruited to the human mb-1 promoter, this does not rule out the possibility of EGL-38 interaction with

LIN-1 on genomic C. elegans targets (Fitzsimmons et al. 2001).

The LIN-1/ETS protein is a downstream target of the EGF signaling pathway, a common motif of Ets functionality (Oikawa and Yamada 2003). Prior to EGF signaling from the anchor cell to the vulval precursor cells, LIN-1/ETS acts as an inhibitor of vulval cell fate; after phosphorylation, LIN-1/ETS becomes an activator of this fate (Beitel et al.

1995; Leight et al. 2015). Based on the importance of LIN-1/ETS in the vulF cells for vulval cell fate, and the probability of EGL-38 being able to recruit LIN-1/ETS to a target,

I investigated this protein as a potential co-factor for EGL-38 in the vulF cells.

Here, I demonstrate that LIN-1/ETS is required for efficient lin-3/egf activation in the vulF cells. Additionally, LIN-1/ETS has a contradictory effect on nlp-2 expression in the uv1 cells which may be mediated by the involvement of EGL-38. In turn, I have verified that egl-38 uv1 expression requires EGL-38 activity, to confirm earlier results

(Chapter 2). An EGL-38 requirement for a co-factor would represent a functional method of restricting Pax activity to cell-specific targets at the required time; this activity can then be coordinated through cells by the EGF pathway, as we have shown to occur in the C. elegans egg-laying system.

3.2.2 Materials and Methods

3.2.2.1 Genetic strains and worm culture

Strains were cultured under standard conditions for C. elegans (Brenner 1974;

Stiernagle 2006). All experiments were performed at 20°C unless otherwise noted. The reference wild type C. elegans strain is N2; additional C. elegans strains used were:

CM2762 egl-38(gu253[egl-38::gfp])

CM2238 unc-119(e2498); guEx1372 (Pnlp-2(0.4kb)::gfp)

UL1251 (Pegl-38(uv1)::gfp) from the Ian Hope lab

MT1001 lin-1(e1777)

3.2.2.2 Bacterial strains

MC782 Plin-3 upstream sequence in pPD75.185 reporter from (Hwang and Sternberg

2004)

MC1944 Plin-3(#2) upstream sequence in pPD107.94 reporter

MC1945 Plin-3(#5) upstream sequence in pPD107.94 reporter

3.2.2.3 Performing RNA interference and generating RNAi strains

RNAi was performed by feeding as described in (Kamath et al. 2000). Briefly, a lin-1 RNAi clone (Source BioScience IV-1N03) was selected from the Ahringer RNAi library (SourceBioScience.com) and verified with sequencing. HT115 bacteria with the

L4440 vector were used as a negative empty vector (EV) control. L4 hermaphrodites were placed on 3.5 cm plates with IPTG [1 mM] and carbenicillin [25 μg/ml] seeded with

RNAi bacteria from fresh overnight culture. Hermaphrodites were removed from RNAi after 18 hours and the offspring allowed to hatch and consume RNAi bacteria for a total of approximately 30 hours until reaching the L4 stage. L4 larvae were then removed for microscopy.

To generate an egl-38 RNAi clone, PCR was used to amplify egl-38 cDNA

(primers: SacI tagged-3’ aaaaGAGCTctgttctccgacaagtccagt and KpnI tagged-5’ aaaaGGTACCcctatttggagaaattatcc) which does not overlap with UL1251 (Pegl-

38(uv1)::gfp). This amplicon was then inserted into the L4440 RNAi vector and confirmed by sequencing. The clone was then transformed into the HT115 RNAi strain

(Kamath and Ahringer 2003). All RNAi experiments were conducted at 20°C.

3.2.2.4 Microscopy

Morphological and fluorescent phenotypes were evaluated using a Zeiss

Axioplan microscope, under 100x magnification. Hermaphrodite animals were observed at mid-L4 larval stage through adult, depending on the experiment. Larval stage was defined on the basis of the vulval morphology in the mid-plane. Images were taken at auto-calculated DIC and epi-fluorescent exposures, varying from 0.5-0.8 ms and 1.2-1.9 ms respectively. Worms were immobilized on agar pads consisting of 3.5% noble agar in water with a 10 mM sodium azide solution. Screening for vulval cell and uv1 phenotypes involved locating the vulval or uv1 cells in the uppermost plane of the worm; expression in vulF or uv1 cells on the side where the worm was laying was not evaluated due to variability in the data due to photo-bleaching of the GFP.

3.2.2.5 Generating lin-3/egf reporter strain and isolation transgenes

The MC782 Plin-3::gfp reporter bacterial strain was provided by the Sternberg lab, containing 10 kb of upstream regulatory sequence from the lin-3 start codon in pPD75.185 from the Fire vector kit (Hwang and Sternberg 2004). The injection mix containing 75 ng/µL of confirmed plasmid and 15 ng/ µL of the rescue plasmid unc-

119(pTJ1043) was injected into RH10 (unc-19(e2498)) adult hermaphrodites. lin-3 isolation transgenes were created by PCR amplifying overlapping ~1 kb regions of the lin-3 upstream promoter and lin-3 genomic gene. Primers were designed as HindIII- tagged 5’ primers and NheI-tagged 3’ primers (primers in Table 2). The amplicons were ligated into the pPD107.94 vector from the Fire kit (Addgene.org) and transformed into

DH5α bacteria. Positive clones were verified by restriction enzyme digest and sequencing. Injection mixes containing 75 ng/µL of confirmed plasmid and 15 ng/µL of the rescue plasmid unc-119(pTJ1043) were injected into RH10 (unc-19(e2498)) adult hermaphrodites. non-Unc F1 offspring were selected and allowed to self to identify transgenic lines. The reported data represent an individual line with the most consistent fluorescence. Strains containing lin-3 isolation transgenes were evaluated for fluorescence and discarded.

3.2.3 Results

3.2.3.1 Expression of lin-3/egf in the vulF cells requires LIN-1/ETS and EGL-38

EGL-38 is a known requirement for vulF expression of lin-3/egf; however, it has been reported that in egl-38(lf) mutants, non-vulval expression of lin-3::LacZ is unaffected (Chang, Newman, and Sternberg 1999). Therefore, EGL-38 is required for activation of lin-3/egf in the vulF cells alone. To investigate how EGL-38 activation of lin-

3/egf has been restricted to a single cell, I looked into the interaction of lin-3/egf with

LIN-1/ETS, a potential EGL-38 co-factor.

I first examined the relationship between lin-3/egf and LIN-1/ETS using a Plin-

3::gfp reporter. In wild-type, expression of Plin-3::gfp begins in the mid-L3 stage of development and in the vulF cells at the mid through late L4 stages. In a lin-1(e1777) loss-of-function mutant, Plin-3::gfp expression was significantly decreased in the 1° vulF cells (Figure 9). These animals exhibit a multi-vulva phenotype; Plin-3::gfp expression was screened only at the primary, functional vulF cells. The decrease in expression from the Plin-3::egf reporter indicates that LIN-1/ETS affects the transcription of lin-3/egf in the vulF cells.

Next, I compared the effect of lin-1/ets and egl-38 knockdown on Plin-3::gfp expression. RNAi bacteria were fed both to the mothers and their progeny; progeny were scored at mid through late L4 stages. egl-38 knockdown (KD) results in a significant decrease in Plin-3::gfp expression, as expected, as does lin-1/ets KD (Figure

10). In fact, lin-1/ets KD had a slightly stronger effect on Plin-3::gfp expression, though

67 this could be due to the variability of Plin-3::gfp expression on the empty vector plates as well. Overall, the mutant and RNAi results together enforce that LIN-1/ETS is required in addition to EGL-38 for lin-3/egf expression in the vulF cells.

3.2.3.2 nlp-2 uv1 expression is partially dependent on LIN-1/ETS

EGL-38 is required within the uv1 cells for nlp-2 and nlp-7 expression (Chapter

2), in addition to being required in the vulF cells to activate lin-3/egf expression needed for the EGF pathway. I have shown that LIN-1/ETS is required for lin-3/egf expression in the vulF cells along with EGL-38, potentially as a co-factor for EGL-38/Pax activity and specificity. To determine if LIN-1/ETS also interacts with EGL-38 targets in the uv1 cells,

I examined the effect of loss-of-function and RNAi against lin-1/ets on the expression of

Pnlp-2::gfp. In a lin-1(e1777) mutant, uv1 expression of Pnlp-2::gfp is significantly decreased (Figure 9). However, upon lin-1/ets KD, there is no effect on Pnlp-2::gfp expression (Figure 10).

3.2.3.3 Expression of egl-38 in the uv1 cells is dependent on EGL-38

egl-38 is expressed in both the uv1 and vulF cells (Chapter 2), with functions in both pertaining to uv1 cell specification and execution of genetic programs (Rajakumar and Chamberlin 2007). We demonstrated that egl-38 uv1 expression depends on the

EGF pathway (Chapter 2); and the EGF pathway requires EGL-38 activity in the vulF cells for activation (Chang, Newman, and Sternberg 1999). A logical conclusion of this activity is that EGL-38 coordinates its own expression between these two cells in the

68 egg-laying system, which could represent an interesting motif for Pax activation as set forth in Chapter 2.

To confirm that EGL-38 activity is necessary for egl-38 uv1 expression, I utilized

RNA interference against both the egl-38 mRNA and the lin-1 mRNA to examine the effect on Pegl-38(uv1)::gfp expression. This strain contains the putative uv1 cell enhancer element for egl-38 expression which expresses gfp solely in the uv1 cells

(Hope lab, U. of Leeds). Knockdown of lin-1 resulted in no effect on Pegl-38(uv1)::gfp; however, knockdown of egl-38 mRNA resulted in complete elimination of Pegl-

38(uv1)::gfp (Figure 10). Upon egl-38 KD, the EGF pathway is not activated in vulF, which we demonstrated in Chapter 2 is necessary for egl-38 expression in the uv1 cells.

Therefore, EGL-38 activity in the egg-laying system is required to coordinate egl-38 uv1 expression.

3.2.3.4 lin-3/egf vulval cell enhancer element may represent a Pax binding site

EGL-38 is required for the expression of the lin-3/egf ligand in the vulF cells

(Chang, Newman, and Sternberg 1999). It is not known, however, if lin-3/egf is a direct target of EGL-38. To determine whether there is a Pax binding site in the lin-3/egf gene,

I performed isolation analysis of the upstream promoter region and coding region of lin-

3/egf to first identify a vulF enhancer element. Overlapping segments of the lin-3/egf upstream and coding regions were cloned into a reporter transgene, and the resulting expression profiles for each segment were analyzed (Figure 11). Two reporters, containing Plin-3(#2)::gfp and Plin-3(#5)::gfp, were found to have expression. Plin-

3(#5)::gfp contains the anchor cell enhancer element that confers expression during the

L3 stage (Hwang and Sternberg 2004). Plin-3(#2)::gfp was found to have inconsistent, low levels of expression in the vulF cells during L4. Further attempts to stabilize this reporter were put on hold until all regions of lin-3/egf could be analyzed; unfortunately, region 4 (Plin-3(#4)::gfp) was not able to be isolated and cloned into the pPD95.69 reporter plasmid. Additionally, I utilized PCR to delete the regions surrounding segment

#4 from the existing Plin-3::gfp reagent, but was unsuccessful in creating a strain which could be used to assay Plin-3(#4)::gfp. The vulF cell enhancer may exist in the region contained within Plin-3(#4)::gfp, or it may exist as a weak enhancer in the Plin-3(#5)::gfp region.

3.2.4 Discussion

egl-38 activity is required for execution of cellular programs (expression of lin-

3/egf, nlp-2, and nlp-7) and specification of cell identity (uv1) (Chapter 1)(Chang,

Newman, and Sternberg 1999; Rajakumar and Chamberlin 2007). These dual functions in neighboring cells represent typical Pax function; however, it is poorly understood how different Pax activities are restricted to the pertinent cells (Pufall and Graves 2002;

Schulte 2014). I have investigated methods by which EGL-38 activity may be directed, including the involvement of a co-factor or self-coordination of expression.

Understanding how Pax activity is regulated and coordinated would further understanding of the role of Pax in development and disease, particularly as a therapeutic target for cancer. Pax proteins are not generally expressed after

70 development is complete, yet have frequent aberrant expression in cancer tissues

(Blake and Ziman 2014; Brand et al. 2011; Czerny, Schaffner, and Busslinger 1993; G.

R. Dressler et al. 1990; Herbst 2004; Konecny et al. 2009; Lang et al. 2007; Muratovska et al. 2003; Rothenpieler and Dressler 1993; Terzić et al. 1998; J. Wang et al. 2018; Q.

Wang et al. 2008). Understanding how and when Pax proteins function would allow creation of targeted approaches to knockout Pax activity in cancerous tissue.

LIN-1/ETS is a potential EGL-38 co-factor, from a family of proteins which have been shown to interact with Pax5 in vitro to enhance target binding. A loss-of-function mutation in lin-1/ets results in decreased levels of Plin-3::gfp expression in vulF and

Pnlp-2::gfp expression in uv1. Plin-3::gfp vulF expression was also eliminated with lin-

1/ets KD, though Pnlp-2::gfp and Pegl-38(uv1)::gfp expression was unaffected. If LIN-

1/ETS is required for lin-3/egf expression, it would be expected that loss of lin-1/ets would cause defects similar to the loss of EGF signaling caused by an egl-38 mutant.

The absence of EGF signaling would result in lack of uv1 cell specification, loss of Pegl-

38(uv1)::gfp expression, and loss of Pnlp-2::gfp expression; however, none of these occur. The contradictory results of lin-1(e1777) and lin-1/ets KD on Pnlp-2::gfp presents two potential explanations: lin-1/ets KD is not efficient, or EGL-38 is activating lin-3/egf and the EGF pathway to functional levels in the absence of lin-1/ets. As lin-1/ets KD had a thorough effect on Plin-3::gfp, indicating that KD was efficient in the vulF cells, I am inclined to favor the latter possibility. Efficient elimination of lin-3/egf in the vulF cells should prevent nlp-2 expression in the uv1 cells; however, genomic lin-3/egf may be activated at a low, functional level that is not reflected in Plin-3::gfp expression (Chapter

2)(Chang, Newman, and Sternberg 1999). As a co-factor recruited by EGL-38, LIN-

1/ETS could be required to increase the level of lin-3/egf activation without being strictly necessary for transcription, allowing for the possibility that EGL-38 activates lin-3/egf to functional levels in the absence of lin-1/ets. Previous studies have found that Pax-DNA complexes will form but are significantly less stable without the ternary Pax-Ets-DNA complex, which would lead to less efficient expression in the absence of a co-factor

(Fitzsimmons et al. 1996). Therefore, it is probable that in lin-1/ets KD animals, EGL-38 is present to activate the lin-3/egf signal, leading to Pegl-38(uv1)::gfp and Pnlp-2::gfp expression in the uv1 cells.

EGL-38 is required for vulF activation of lin-3/egf and is responsive to the EGF pathway in the uv1 cells (Chapter 2)(Chang, Newman, and Sternberg 1999). In fact, egl-

38 KD results in the absence of Pegl-38(uv1)::gfp expression entirely, confirming that

EGL-38 is necessary for its own expression. EGL-38 acts by coordinating its own activity through the EGF pathway (Chapter 2), a novel Pax motif that may additionally explain how Pax activity is so tightly regulated. Pax proteins have not previously been demonstrated to synchronize their own expression between multiple cells, but they have been speculated to function in both initiation and response to inductive cell-to-cell signaling (Lang et al. 2007; Püschel, Gruss, and Westerfield 1992). Concerning EGL-38 activation of lin-3/egf expression, a clear vulF enhancer element for lin-3/egf expression was not located. However, despite not locating a Pax binding site in the lin-3/egf promoter, EGL-38 remains a requirement for lin-3/egf expression, whether directly or indirectly. The requirement of Pax for EGF expression may be recapitulated in other

72 organisms. In endometrial cancer lines, knockdown of PAX2 significantly decreases cancer cell viability, measured by tumor growth, proliferation, and survivability, all factors which are closely tied to increased EGF activity (Jia et al. 2016; L.-P. Zhang et al. 2011).

3.3 EGL-38 activity in the uv1 cells

3.3.1 Introduction

The uv1 cells are neurosecretory and serve as inhibitors to control the periodicity of egg-laying (Banerjee et al. 2017; Collins et al. 2016; Jose et al. 2007). Specification of these cells occurs at the mid L4 stage and relies upon the EGF pathway for signaling from the vulF cells to the presumptive uv1 cells, and upon EGL-38 to activate the lin-

3/egf signal in vulF (Chang, Newman, and Sternberg 1999; Rajakumar and Chamberlin

2007). Additionally, EGL-38 has autonomous functions to activate cellular programs within the uv1 cells that are necessary for egg-laying (Chapter 2). Two EGL-38 uv1 targets, nlp-2 and nlp-7, are neuropeptide-like proteins which contribute to the uv1 phenotype. NLP-7 is required, in cooperation with FLP-11, for uv1 inhibition of egg- laying; NLP-2 has no known functions (Banerjee et al. 2017). Expanding understanding of EGL-38 functions in the uv1 cells will contribute to further understanding of how Pax proteins can activate cellular programs based on phenotypic context. Here, I present that nlp-7 has a 30 bp uv1 cell enhancer with a prospective Pax binding site. I also examine the egg-laying behavior of nlp-2, nlp-7, and flp-11 mutants to determine whether the loss of function of EGL-38 targets will affect egg-laying.

3.3.2 Materials and Methods

3.3.2.1 Genetic strains and worm culture

Strains were cultured under standard conditions for C. elegans (Brenner 1974;

Stiernagle 2006). All experiments were performed at 20°C unless otherwise noted. The reference wild type C. elegans strain is N2; additional C. elegans strains used were:

CM2762 egl-38(gu253[egl-38::gfp])

CM2238 unc-119(e2498); guEx1372 (Pnlp-2(0.4kb)::gfp)

CM2694 unc-119(e2498); guEx1554 (Pnlp-7::gfp)

NY2040 unc-119(e2498); ynIs40(Pflp-11::gfp)

UL1251 (Pegl-38(uv1)::gfp) from the Ian Hope lab

FX1908 nlp-2(tm1908)

FX02984 nlp-7(tm2984)

FX02706 flp-11(tm2706)

CM2743 nlp-2(tm1908);nlp-7(tm2984)

CM2742 nlp-7(tm2984);flp-11(tm2706)

3.3.2.2 Generating nlp-7 deletion transgenes

The reporter strain for nlp-7, CM2694, was constructed (as described in Chapter

2) using pDD95.69 from the Fire Vector kit (Addgene.org). 2.7 kb of upstream promoter

(including the start codon) was amplified using a 3’ Sal-tagged primer and a 5’ MscI- tagged primer. The amplicon was ligated into the promoter-less pPD95.69 plasmid and

74 transformed into DH5α bacteria. To perform deletion analysis on this transgene, a distal

3’ facing primer and a proximal 5’ facing primer were designed (Table 6.2) to amplify away from one another around the Pnlp-7(2.7kb)::gfp reporter plasmid. This allows for elimination of the intervening sequence. This product was re-ligated, treated with DpnI to eliminate non-PCR products, and transformed into DH5α bacteria. Positive clones were verified by restriction enzyme digest and sequencing. Injections occurred as described below.

The two primers amplified away from each other in the Pnlp-7(2.7kb)::gfp reporter to eliminate the intervening sequence. The PCR product was then ligated to re- seal the reporter plasmid. Site-directed mutagenesis for the nlp-7 promoter was performed using standard reagents, according to the Stratagene protocol (Stratagene

Quikchange II Site-Directed Mutagenesis Kit). Additional mutations were introduced by purchasing a genomic block containing 0.86 kb of the nlp-7 promoter with the desired mutations, and introducing this block into pPD95.69 (IDTdna.com). Positive clones were verified by restriction enzyme digest and sequencing. Injection mixes containing 75 ng/µL of confirmed plasmid and 15 ng/ µL of the rescue plasmid unc-119(pTJ1043) were injected into RH10 (unc-19(e2498)) adult hermaphrodites. non-Unc F1 offspring were selected and allowed to self to identify transgenic lines. At least three lines were obtained and screened for fluorescence. The reported data represent an individual line with the most consistent expression pattern. Pnlp-7::gfp deletion reporter strains were analyzed for fluorescent expression and discarded.

3.3.2.3 Microscopy

Morphological and fluorescent phenotypes were evaluated using a Zeiss

3.3.2.4 Egg-retention assay

Mid-L4 hermaphrodites, determined by vulval morphology at low resolution, were transferred in groups of 10-15 to an empty NGM 10 cm plate. Animals were grown at

20°C for 18 hours until gravid adulthood. Single animals were placed into a drop of bleaching solution on a glass slide, and observed for cracking of the hermaphrodite cuticle after approximately 2 minutes. In utero eggs were separated using a platinum pick and counted at low resolution to determine the retention of eggs in the individual adult.

3.3.3 Results

3.3.3.1 nlp-7 has a 30 bp uv1 enhancer element with a prospective Pax binding site

EGL-38 is required for uv1 expression of nlp-7 (Chapter 2), but the nature of this activation (direct or indirect) is unknown. To further examine the relationship between nlp-7 and EGL-38, I searched for a Pax binding site in the nlp-7 promoter by performing deletion analysis on the full-length promoter-reporter construct. Truncations were introduced beginning at the 5’ upstream end of the original 2.7 kb fragment to create successively smaller reporter genes. The expression profile of each subsequent reporter in the uv1 cells was analyzed (Figure 12). Due to the abrupt loss of uv1 expression upon deletion, I interpret that a uv1 cell enhancer element exists between

0.83 and 0.86 kb from the start codon of nlp-7. I evaluated this 30 bp element for a PAX binding site by searching for sequence similarity to the lin-48 and nlp-2 response elements which are both bound by EGL-38 (Chapter 2 and (Johnson et al. 2001). A similar potential site was found 0.85 to 0.86 kb from the start codon in the nlp-7 uv1 response element. Mutations were introduced at the sequences predicted to interact with the N-, the C-, or both termini of the EGL-38 DNA binding domain. The level of expression of the N-, C-, and NC-termini mutation reporters were not different from the full-length reporter.

3.3.3.2 Single and double mutants of nlp-2 and nlp-7 have little to no effect on egg-laying behavior

egl-38 mutants lack a vulval-uterine connection, which prevents eggs from being laid (Chamberlin et al. 1997). To examine egg-laying behavior in the context of egl-

38(lf), therefore, I had to utilize mutant versions of EGL-38 targets to identify any effect on the egg-laying phenotype. nlp-7 has been reported to have a role in controlling the periodicity of egg-laying in conjunction with flp-11 (Banerjee et al. 2017). nlp-2 has no known functions; however, within the egg-laying system Pnlp-2::gfp expresses only in mature, specified uv1 cells, increasing the likelihood that nlp-2 has a function in egg- laying (Chapter 2). I utilized existing nlp-2, nlp-7, and flp-11 loss-of-function mutants, as well as creating double mutant strains, to examine the effect of each neuropeptide on egg-laying (Figure 13). Animals were synchronized to the same age and allowed to grow and lay eggs for the same period of time, before adults were dissolved and the number of eggs in utero counted. None of the single mutations or an nlp-2(lf);nlp-7(lf) double mutation resulted in a significant effect on the number of eggs retained compared to wild-type. Previous work in our lab confirmed that loss of nlp-2 has no effect on egg-laying (unpublished). nlp-7(lf);flp-11(lf) double mutant animals retained more eggs in utero than wild-type, though the difference does not reach statistical significance due to the large variation possible in the wild-type background. I attempted to create a triple mutant of nlp-2(lf);nlp-7(lf);flp-11(lf) but was unable to homogenize all three mutations in the same background. The results concerning the effect of each gene on egg-laying were inconclusive; however, the results do not contradict that nlp-7 and

78 flp-11 have an inhibitory role in egg-laying, or that nlp-2 has a potential role in egg- laying as well.

3.3.4 Discussion

Previously, egl-38 was confirmed to have a role in the uv1 cells, however this role was unknown (Rajakumar and Chamberlin 2007). I have identified that EGL-38 is required to activate a uv1 cellular program which contributes to the uv1 functional phenotype by activating both nlp-2 and nlp-7 expression (Chapter 2). The nlp-2 promoter possesses a uv1 cell enhancer element which contains a putative Pax binding site (Chapter 2). The nlp-7 uv1 cell enhancer is 830-860 bp from the start codon, and elimination of this 30 bp region eliminates Pnlp-7::gfp expression in the uv1 cells. A prospective Pax binding site was identified based on sequence similarity to lre2 and nre1 binding sites; however, mutations of the N-, C- and both termini do not significantly affect reporter expression. Potentially nlp-7 possesses a non-canonical Pax binding site which I did not identify for investigation; scanning mutagenesis of the uv1 cell enhancer could be used to identify the nucleotides required in the promoter, which could be verified using EMSA for binding to EGL-38. Further investigation would be required to determine if EGL-38 is capable of binding the nlp-7 promoter. Regardless, whether the interaction is direct or indirect, EGL-38 is required for expression of nlp-7 in the uv1 cells, and this activation is part of the EGL-38 regulated cellular program necessary for uv1 function (Chapter 2).

The primary biological output of the uv1 cells is inhibitory signaling that controls the periodicity of egg-laying (Alkema et al. 2005; Collins et al. 2016; Jose et al. 2007). nlp-7 contributes to this inhibitory function in conjunction with flp-11 (Banerjee et al.

2017). I investigated the role of nlp-2(tm1908), nlp-7(tm2984), flp-11(tm2706), nlp-

7(tm2984);nlp-2(tm1908) and nlp-7(tm2984); flp-11(tm2706) in egg-laying behavior; unfortunately, I could not find any significant effect on the retention of eggs in adult hermaphrodites. It is likely that the neuropeptide-like protein family is so functionally redundant that eliminating a single nlp has no phenotypic effect. In addition, my results are in contradiction of the paper which originally reported that nlp-7 and flp-11 together inhibit egg-laying; the previous paper found that nlp-7(tm2984);flp-11(tm2706) double mutants retain fewer eggs due to the loss of the nlp-7;flp-11 inhibitory effect on egg- laying (Banerjee et al. 2017). I did not observe that nlp-7(tm2984);flp-11(tm2706) mutants retained fewer eggs than did wild-type. Potentially the hermaphrodites I screened for egg retention were older than those screened by Banerjee, et al, and older adults create and retain more eggs than do young adults. There may have been another confounding variable which was causing the hermaphrodites I used to retain eggs, such as plate desiccation or crowding.

Before addressing the discrepancy between my results and the previously published results, I attempted to obtain a triple nlp-2(tm1908);nlp-7(tm2984);flp-

11(tm2706) mutant to examine whether the triple would have a more obvious phenotype. However, I could not homogenize this triple mutant background, which may indicate that the triple is lethal or that it significantly affects fitness of the worms to

80 reproduce. EGL-38 may potentially activate a battery of neuropeptide-encoding genes which are necessary to act together for an egg-laying phenotype; eliminating only one or two of these genes at a time may not be sufficient to see an effect. Despite my inability to reproduce the published nlp-7 effect on egg retention, I still theorize that

EGL-38 is involved in activating a uv1 cellular program required for egg-laying behavior in adult hermaphrodites.

3.4 Figures

Figure 9: lin-3/egf and nlp-2 expression is dependent on LIN-1/ETS

Figure 9: lin-3/egf and nlp-2 expression is dependent on LIN-1/ETS.

Fig. 9 (A-D) Plin-3::gfp expression. (A-B) Wild-type mid L4. Plin-3::gfp is present in vulF cells (solid arrowhead). (C-D) lin-1(e1777) late L4. Worms are multi-vulva, resulting in crowding of the primary, functional vulF cells marked by a solid arrowhead. Plin-3::gfp

82 expression is absent in the vulF cells (empty arrowhead). (E) Percent of worms expressing Plin-3::gfp or Pnlp-2::gfp in the vulF cells. Above each bar is shown the number of worms screened. Error bars are shown for the standard error of proportion.

For each pairwise comparison of control to mutant (wild type to lin-1(e1777), the difference is significant at a p<0.05 (2-tailed Z-test).

Work performed: I performed the RNAi experiment using RNAi from the Ahringer library.

Thank you to Ryan Johnson for creation of the Pnlp-2::gfp reporter and to Vandana

Rajakumar for establishment of the Plin-3::gfp reporter strain.

Figure 10: lin-3/egf expression is dependent on EGL-38 and LIN-1/ETS; egl-38 expression is dependent on EGL-38.

Fig. 10 (A) Percent of worms expressing Plin-3::gfp or Pnlp-2::gfp in the vulF cells or

Pegl-38(uv1)::gfp in the uv1 cells. Above each bar is shown the number of worms screened. Error bars are shown for the standard error of proportion. For each pairwise comparison of control to RNAi KD (wild type to KD) or RNAi to RNAi, the marked differences are significant at a p<0.05 (2-tailed Z-test).

Work performed: I performed the RNAi experiments shown here and developed the egl-

38 RNAi strain. Other RNAi strains came from the Ahringer library. Thank you to Ryan

Johnson for creation of the Pnlp-2::gfp reporter and to Vandana Rajakumar for establishment of the Plin-3::gfp reporter strain.

Figure 11: lin-3/egf contains an anchor cell enhancer and a putative vulF cell enhancer.

Fig. 11 (A) Isolation clones of the lin-3 upstream and coding regions contain ~1 kb of overlapping sequence in a pPD107.94 reporter vector. The region #4 clone was not

85 able to be isolated. (B) Percent of worms with expression of the transgenic reporter in vulF cells. (C-D) Plin-3(#2)::gfp is present in the anchor cell during mid-late L3 (anchor cell indicated by arrowhead). (E-F) Plin-3(#5)::gfp is inconsistently expressed in the vulF cells during early-late L4 (vulF cells indicated by arrowheads).

Work performed: Thank you to the Sternberg lab for the gift of the Plin-3::gfp reporter and to Vandana Rajakumar for establishment of the reporter strain. I performed the promoter isolation analysis design, execution, and screening.

Figure 12: nlp-7 contains a uv1 cell enhancer with a putative EGL-38 site.

Fig. 12 (A) Percent of worms with expression of Pnlp-7::gfp or its deletion derivatives. A full-length reporter (shown at top) was constructed and then deletions taken from the reporter to isolate a uv1 cell enhancer region. On the left is shown the length of upstream promoter sequence which was included in the reporters. Worms were screened in mid-late L4 for uv1 expression of the deletion reporter. Deletion of between

0.83 and 0.86 kb upstream from the start codon significantly reduces reporter expression. I define this 30 bp region as the uv1 cell enhancer for nlp-7. Point mutations in this enhancer made to the sequences hypothesized to bind the N-, C-, and both termini (NC-) of EGL-38 do not significantly decrease reporter expression. Failure to target the EGL-38 binding site may be due to the presence of a non-consensus Pax site or failure to identify key nucleotides. Data are recovered from more than 40 worms per

87 reporter; error bars are shown for the standard error of proportion. Comparisons of 0.83 kb and smaller deletion constructs to the wild-type reporter are significant at p<0.05. (B)

Comparison of the Pax binding enhancer regions from lin-48 (lre2) (Johnson et al.

2001), nlp-2 (nre1; Chapter 2), and nlp-7 (nre2, Chapter 3). The mutations introduced in the 0.4 kb reporter are indicated under nre2 NC-.

Work performed: Thank you to Abdi Jama for creating the full-length Pnlp-7::gfp reporter strain. I performed the deletion analysis of the strain, including design, execution, and screening. Thank you to Abdul Abokor for assistance in creating one of the deletions.

Figure 13: Single and double nlp-2, nlp-7, and flp-11 mutants do not affect egg retention.

Fig. 13 (A) Age-matched hermaphrodites were dissolved and the number of eggs in utero counted. The average, max, and minimum numbers of eggs amongst all hermaphrodites screened are shown. At least 50 worms were screened for each double mutant, and at least 30 worms were screened for each single mutant. Pairwise comparisons of each mutant to wild-type were not significant at p<0.05 (Mann-U

Whitney Test).

Work performed: I performed the in utero egg retention analysis. Thank you to Dr.

Chamberlin for creating the double mutant strains.

Chapter 4: The pal-1 regulatory network and an accompanying computational model

4.1 Introduction

Development of multicellular organisms requires coordination of cell growth, proliferation, specification, and differentiation. One of the first steps in controlling development occurs at the transcriptional level; however, transcriptional regulation can be a difficult phenomenon to observe through traditional genetic methods (Hasty et al.

2001; Selinger, Wright, and Church 2003). Disrupting a genetic network at the chosen node can lead to global consequences that prevent parsing out a specific effect, particularly when that node is involved in an intertwined regulatory network.

Consequently, computational methods have been developed to extract biologically relevant data from available experiments and to incorporate this data into predictions of regulatory interactions (Hasty et al. 2001; Marbach et al. 2010; Stigler and Chamberlin

2012). Unfortunately, these methods generally rely on data from expression patterns caused by genetic mutations or disruptions that can be difficult to obtain and interpret

(D’haeseleer, Liang, and Somogyi 2000; Gardner and Faith 2005; de Jong 2002; W.-P.

Lee and Tzou 2009). Developing a predictive model using wild-type data would allow for predictions of more large-scale networks that may be difficult to assess through perturbations.

A previous collaboration of Dr. Helen Chamberlin with Dr. Brandilyn Stigler

(Southern Methodist University) generated a Mathematically Inferred Model (MIM) which could mine existing data sets for biological relevance and compute predictive interaction relationships (Stigler and Chamberlin 2012). This MIM utilized data previously published comparing C. elegans embryos with wild-type or mutant C blastomere backgrounds (Baugh et al. 2005; Yanai et al. 2008). The C founder blastomere arises at the 8-cell stage of development and will differentiate into posterior mesoderm and ectoderm (Sulston et al. 1983). Differentiation of the C blastomere and its descendants depends on the Hox/Caudal-like homeodomain protein PAL-1 (Hunter and Kenyon 1996). Maternally provided pal-1 is translated in the developing embryo, and interactions with other proteins target maternal PAL-1 activity to specify and function in the C and D founder blastomeres (Bowerman et al. 1993; Hunter and

Kenyon 1996; Mello and Draper 1996; Seydoux et al. 1996). Zygotically transcribed pal-

1 is rapidly upregulated as maternal PAL-1 protein is downregulated (Edgar et al. 2001;

Hunter and Kenyon 1996). PAL-1 acts to initiate a network of transcription factors necessary for the posterior ectoderm vs mesoderm cell fate decision (Baugh et al. 2005;

Yanai et al. 2008).

The MIM developed a predictive model of PAL-1 interactions in the C blastomere for cell fate determination (Figure 14). Directed and undirected interactions produced by the MIM predict how PAL-1 is regulating a transcription factor network necessary for cell differentiation (Stigler and Chamberlin 2012). However, these predicted interactions do not differentiate between maternal or zygotic effects, an important distinction toward

91 understanding how this network functions. To investigate whether maternal or zygotic

PAL-1 is responsible for these predicted interactions, as well as to determine the validity of the predictions, I approached testing the MIM through RNA interference to knock down pal-1 and observe the effect on reporter expression of potential pal-1 targets.

While doing so, I uncovered an interesting phenomenon wherein offspring without zygotic pal-1, which should prevent posterior embryonic development, were developing into morphological normal larvae. Consequently, I began to explore whether there was any male-derived protective factor conferred by mating a C. elegans hermaphrodite and male, and I present my investigation into this occurrence.

In addition to my attempts to validate the MIM network, collaborative work has produced an additional computational method to represent this data set, termed the

Glyph SPLOM (Yates et al. 2014). The Glyph SPLOM provides a summarization of traditional scatterplot matrices as glyph representations of the measure of relatedness between variables, in this case transcription factors. This relatedness is calculated using the distance correlation (measure of dependence) and dependency class (measure of co-expression) to reveal richly detailed interactions (Sahoo et al. 2008; Székely, Rizzo, and Bakirov 2007). I briefly discuss this work, which has been published in the proceedings of the Eurographics Conference on Visualization (EuroVis) 2014 as

“Visualizing Multidimensional Data with Glyph SPLOMs” by A.Yates, A. Webb

(Chasser), M. Sharpnack, H. Chamberlin, K. Huang, and R. Machiraju (Yates et al.

2014). Additionally, I chose a transcription factor with unknown expression pattern

92 which mapped to an E cell lineage in the Glyph SPLOM, nhr-232, and created a reporter to confirm that this TF does express in the E lineage where predicted.

4.2 Methods

4.2.1 Genetic strains and worm culture

Strains were cultured under standard conditions for C. elegans (Brenner 1974;

Stiernagle 2006). All experiments were performed at 20°C unless otherwise noted. The reference wild type C. elegans strain is N2; additional C. elegans strains used were:

CM2450 unc-119(e2498); guEx1457(Pnhr-232::gfp)

JJ1247 rde-1(ne219)

RW10174 unc-119(ed3) III; zuIs178 V; stIs10024; stIs10174. (Ppal-1::rfp)

RW10177 unc-119(ed3) III; zuIs178; stIs10177. (Pcwn-1::rfp)

RW10265 unc-119(ed3) III; zuIs178 V; stIs10024; stIs10143. (Ptbx-8::rfp)

RW10346 unc-119(ed3) III; zuIs178; stIs10323. (Pelt-1::rfp)

RW10896 unc-119(ed3) III; zuIs178 V; stIs10024; stIs10808 (Pnob-1::rfp)

SL438 spe-9(eb19) I;him-5(e1490)V;ebEx126

4.2.2 RNA interference

RNAi was performed by feeding as described in (Kamath et al. 2000). Briefly, a pal-1 RNAi clone (Source BioScience III-2O04) was selected from the Ahringer RNAi library (SourceBioScience.com) and verified with sequencing. HT115 bacteria with the

L4440 vector were used as a negative empty vector (EV) control. L4 hermaphrodites

93 were placed on 3.5 cm plates with IPTG [1 mM] and carbenicillin [25 μg/ml] seeded with

RNAi bacteria from fresh overnight culture. To screen survival, parents worms were removed from RNAi after 18 hours and the number of eggs present were counted; after an additional 24 hours, all unhatched embryos or deformed larvae were scored as dead.

If a cross was performed with a transgene-bearing father, progeny were evaluated for fluorescence and only cross-progeny expressing the transgene were counted for survival. If the father did not have a transgenic reporter, cross-progeny could not be separated from self-progeny and all embryos were considered for survival.

Four existing target reporters (Pcwn-1::rfp, Pelt-1::rfp, Ptbx-8::rfp, and Pnob-

1::rfp) were initially chosen to observe whether knockdown of pal-1 by RNAi would affect reporter expression. The RNAi survival results of these four reporters and Ppal-

1::rfp were pooled for data and recorded as “wild-type bearing a transgene” (WT;rfp).

The reporter expression was not observed for pal-1 KD effect, only for cross-progeny verification as described.

4.2.3 Microscopy

Morphological and fluorescent phenotypes were evaluated using a Zeiss

Axioplan microscope, under 100x magnification. Embryos were observed from the 28- cell stage through the 2-fold comma stage, as defined on the basis of embryonic morphology in the mid-plane. Images were taken at auto-calculated DIC and epi- fluorescent exposures, varying from 0.5-0.8 ms and 1.2-1.9 ms respectively. Worms

94 were immobilized on agar pads consisting of 3.5% noble agar in water with a 10 mM sodium azide solution.

4.2.4 Summarization of data in the Glyph SPLOM

The methodology for producing the Glyph SPLOM is presented in (Yates et al.

2014). The dataset utilized for the C. elegans Glyph SPLOM represent transcripts from wild-type and mex-3(zu155) animals evaluated for abundance using gene expression microarrays (Baugh et al. 2005). These data were then filtered to select only the transcription factor (TF) genes for analysis. In brief, a comparison was made between the expression profile of every set of TFs genes to determine whether a logical, statistical dependency existed between the two; if not, an attempt was made to provide directionality to any weak interactions which may exist. A corresponding shape (glyph) and color were assigned to each class of interactions, which was represented in the

Glyph SPLOM as a visualization of, essentially, directed regulatory relationships.

4.2.5 Generating nhr-232 reporter

Reporter strain for nhr-232 was constructed using pPD95.69 from the Fire Vector kit (Addgene.org). 1.277 kb of upstream genomic sequence was amplified using a 5’

BamHI-tagged primer (aaaaGTCGACgaagaagacgagaaaggcgttg) and a 3’ Sal1-tagged primer (aaaaGGATCCcatattaaatcgaataaaaaagaggtcagtgag); this amplicon contains only the start codon from the coding sequence. The 1.277 kb amplicon was ligated into the promoter-less pPD95.69 plasmid and transformed into DH5α bacteria. The plasmid

95 was confirmed using restriction enzyme digest and sequencing. An injection mix contained 75 ng/µL of confirmed plasmid, 15 ng/ µL of the rescue plasmid unc-

119(pTJ1043), and 5 ng/µL of the pCFJ90 marker Pmyo-2::rfp were injected into RH10

(unc-19(e2498)) adult hermaphrodites. non-Unc F1 offspring were selected and allowed to self to identify transgenic lines. Two lines were established and screened for fluorescence. The reported data represent an individual line with the most consistent expression pattern.

4.3 Results

4.3.1 pal-1 Knockdown causes significant lethality of wild-type, non-cross progeny

The MIM predicts that pal-1 has several direct regulatory targets: tbx-8, tbx-9, elt-

1, lin-26, elt-3, cwn-1, unc-120, hlh-1, hnd-1, scrt-1, and nob-1 (Figure 14) (Stigler and

Chamberlin 2012). To validate these targets in vivo, I chose to knock down pal-1 using

RNAi on strains bearing a fluorescent transcriptional reporter for the target TFs to observe the effect loss of pal-1 has on expression. However, I additionally needed to separate maternal and zygotic pal-1 effects, which cannot be done using RNAi on wild- type reporter strains. In these wild-type mothers, maternal pal-1 transcripts are knocked-down in the mother and cannot be packaged into oocytes, while ingested pal-1 dsRNA is packaged into oocytes and this packaged pal-1 dsRNA triggers the RNAi response in embryos upon zygotic pal-1 activation (Fire et al. 1998; Grishok 2000).

Therefore, wild-type strains lack both maternal and zygotic pal-1 due to the heritability of

RNAi. Instead, to separate out maternal and zygotic pal-1 effects, I utilized the rde-

1(ne219) loss-of-function allele. rde-1 is an Argonaute protein necessary for the RNAi response in C. elegans; rde-1(ne219) mutants are defective in the RNAi response and possess both maternal and zygotic pal-1 even when the animals are exposed to double stranded RNA (Tabara et al. 1999). Theoretically, an rde-1(ne219/+) heterozygote would possess maternal pal-1 from its RNAi-deficient mother (rde-1(ne219)), but lack zygotic pal-1 due to the RNAi response made possible by the wild-type allele from the

RNAi-responsive father (rde-1(+/+)) (Figure 15). I utilized this crossing schematic as a method of eliminating zygotic pal-1 to observe whether maternal pal-1 is sufficient for regulating the predicted target TF genes.

I began screening RNAi by first verifying that treatment of wild-type animals with our pal-1 RNAi bacteria conferred an embryonic lethal phenotype. I plated L4 hermaphrodites to pal-1 or empty vector (EV) RNAi, allowed them to ingest bacteria and grow overnight, and removed the adults the next day. I screened for pal-1 KD survival by counting the number of embryos on the plate after adult removal and one day later: any embryos which remained unhatched or were significantly deformed after 24 hours were considered to be dead. As expected, wild-type offspring did not survive pal-1 KD, while rde-1(ne219) offspring did (Figure 16). Unexpectedly, rde-1(ne219/+) heterozygotes also survive pal-1 KD.

The survival of rde-1(ne219/+) offspring is a consistent result, whether the father is an N2 or wild-type with a TF reporter transgene (WT;rfp) (Figure 16). As these offspring were predicted to initiate an RNAi response and therefore be depleted for

97 zygotic pal-1 which leads to abnormal posterior muscle and skin development, the survival was a puzzling result. To investigate whether the transgene was conferring a level of interference to pal-1 RNAi, I crossed N2 and WT;rfp worms in both directions. I discovered that when the mother is WT;rfp bearing, the offspring had significant survival; when the father is WT;rfp bearing, the offspring did not survive RNAi or had gross morphological defects. While these two crosses contradict one another, the fact that a significant number of progeny from a wild-type cross were surviving pal-1 RNAi was a confounding result.

4.3.2 Mating confers a protective factor against loss of pal-1

The survival of heterozygous offspring from the cross between rde-1(ne219) hermaphrodites treated with pal-1 RNAi and wild-type males implies one of several outcomes: A) pal-1 RNAi is inefficiently inherited by offspring; B) the reporter transgene interferes with the efficiency of pal-1 RNAi; or C) a protective factor is conferred either by the father or D) by the mating process itself. Because pal-1 RNAi is sufficient and thorough in progeny from self-crossing mothers and because the inheritance pattern of the reporter transgene makes a significant difference in survivability, I briefly dismissed the first two possibilities to pursue the latter two. C. elegans hermaphrodites are capable of self-fertilization or mating with males; when mating, male sperm is preferentially used to the hermaphrodite sperm (Ward and Carrel 1979). However, the frequency of C. elegans mating is inefficient (Chasnov and Chow 2001; Garcia,

LeBoeuf, and Koo 2007). Therefore, I could speculate that C. elegans males are more

98 determined to see the survival of their offspring, as they have fewer opportunities to pass on their genes. It is possible that there is protective factor against RNAi provided in male sperm or triggered by the act of mating.

To investigate and separate these possibilities, I repeated pal-1 RNAi experiments using males from the spe-9(eb19) strain. spe-9(eb19) animals produce sperm that is incapable of fertilization (Singson, Hill, and L’Hernault 1999; Zannoni,

L’Hernault, and Singson 2003). By crossing spe-8(eb19) males to wild-type or rde-

1(ne219) hermaphrodites, the resulting offspring will consist of hermaphrodite self- progeny from mothers that have potentially mated. The offspring of wild-type or rde-

1(ne219) hermaphrodites and spe-9(eb19) males have similar survival as do offspring from crosses where the father is wild-type (Figure 17). Further investigation using the spe-9(eb19) strain was not pursued due to a shift of research focus.

4.3.3 Glyph SPLOM has significant advantages over traditional computation methods

Attempting to validate the pal-1 regulatory network initiated a collaborative effort to create a complementary computational model of the same data. This quickly expanded to include all 655 transcription factors present in the Baugh, et al. data set which were analyzed in the Glyph SPLOM (Baugh et al. 2005; Yates et al. 2014). The

Glyph SPLOM provides visualization of which TF genes are co-expressed, and any directed interactions that may exist can be interpreted based on the implication assigned to the visualized glyph (Figure 18). There are significant advantages to visualizing co-expression, directionality, and dependencies as a Glyph SPLOM. For

99 example, in the biological context of TF genes, the “X is necessary for Y” glyph often interprets as “X is activated before Y” in the developmental order. This type of directed inference cannot be made using distance correlation or dependency strength measures alone; however, in the Glyph SPLOM, we are able to interpret the co-expression and relatedness of two potentially non-interacting TFs. Additionally, the 655 TF genes presented in the Glyph SPLOM would be far too large for a scatterplot matrix, and fairly confusing as a heatmap presented without summarization.

Interestingly, the Glyph SPLOM is able to accurately group TF genes by the cell lineage in which they are expressed. Based on co-expression levels detected on the microarray, TF genes were grouped together if they behaved similarly, leading to TF genes expressed in cells of the same lineage grouping together. The C lineage

(posterior mesoderm and ectoderm) genes, including pal-1 and its projected targets, cluster separately from the E lineage (intestine) genes (Baugh et al. 2005; Stigler and

Chamberlin 2012; Yanai et al. 2008; Yates et al. 2014). The separation of these two clusters occurred mainly based on the dependency class, or the measure of co- expression, as both clusters are positively correlated both within the cluster and with each other. These positive correlations, measured by distance correlation and exemplified by a heatmap, would be unable to separate the C and E lineage clusters from one another due to their interrelatedness; only the addition of the co-expression measure is able to differentiate the two clusters. Therefore, combining both the dependency class and distance correlation into a single visualization provides a more richly detailed understanding of a regulatory network than does either factor alone.

100

4.3.4 nhr-232 expresses in the E lineage as predicted by the Glyph SPLOM

The Glyph SPLOM arranges TF genes which co-express together into clusters which can be used to predict the cell lineage where the TF is active. In fact, pal-1 and all of its predicted targets map to the C lineage cluster (Figure 18; annotated figure available in (Yates et al. 2014)). However, the Glyph SPLOM is comprised of microarray data from only two backgrounds, wild-type and a mutant which enriches for C blastomere identity, mex-3(zu155) (Baugh et al. 2005). To ensure that the Glyph

SPLOM was not therefore biased toward accurate predictions for C blastomere expressing-TFs, I chose an unknown TF from the E lineage cluster to investigate.

nhr-232 is a nuclear-hormone domain TF which, previously, did not have a known expression pattern available. In the Glyph SPLOM, nhr-232 clusters with TFs with a known E lineage expression pattern, such as elt-2, nhr-57, and nhr-68 (Murray et al. 2012). I created a reporter consisting of 1.3 kb of upstream nhr-232 promoter sequence driving the expression of gfp. Pnhr-232::gfp was found to express exclusively in the E lineage intestine cells (Figure 19). Reporter expression was detected as early as the 28 cell stage, soon after being laid, and as late as the 2-fold stage of embryonic development. Screening was not performed at later stages; therefore, it is possible that nhr-232 expression persists into the larval or adult stages. Verifying Pnhr-232::gfp E lineage expression allows me to conclude that the Glyph SPLOM is an accurate predictor of expression pattern by cell lineage clustering.

101

4.4 Discussion

pal-1 has been shown to be necessary for C. elegans embryonic posterior development (Hunter and Kenyon 1996; Mello and Draper 1996). Maternal PAL-1 is needed for initial C blastomere specification; PAL-1 additionally activates and regulates a network of transcription factors required for the posterior mesoderm vs ectoderm cell fate decision (Baugh et al. 2005; Bowerman et al. 1993; Edgar et al. 2001; Hunter and

Kenyon 1996; Yanai et al. 2008). Mathematical models, such as the MIM, have been developed to predict PAL-1 regulatory targets; however, this model has not been validated in vivo and cannot predict whether maternal or zygotic PAL-1 is functioning in regulation (Stigler and Chamberlin 2012). I set out to validate the direct regulatory predictions made by the MIM, but was presented instead with the unexplained phenomenon of wild-type embryos surviving pal-1 KD. At best, these embryos should lack posterior muscle and skin development leading to gross morphological defects; at worse, embryos without zygotic pal-1 would not even hatch.

Instead, I discovered that progeny from mated hermaphrodites survived pal-1

RNAi without defects at a significantly higher rate than do progeny from isolated hermaphrodites. Due to the lack of a paternal transgene in most of the matings I screened, I was unable to parse out the cross- and self-progeny from one another.

Given that C. elegans matings can also be rare events (Chasnov and Chow 2001), it is possible that on non-transgenic mated plates such as rde-1(ne219) x N2, I was only observing self-progeny possessing homozygous rde-1(ne219), leading to pal-1 KD survival. Conversely, I could have been observing heterozygous cross-progeny

102 possessing rde-1(ne219/+) which still survived pal-1 KD. In this latter case, it could be possible that: A) rde-1(ne219) is dominant over rde-1(+), preventing the RNAi response;

B) rde-1 requires two functional copies for wild-type response; or C) maternal pal-1 does not downregulate in the absence of zygotic pal-1, instead perduring to fulfill zygotic functions.

Interestingly, however, on wild-type mated plates (WT;rfp x N2), I observed a significant increase in survival that cannot be explained by the presence of an RNAi- deficient rde-1(ne219) allele, as all parents are rde-1(+). This survival on the wild-type mated plate is also significantly higher than the average survival of WT;rfp or N2 self- progeny from non-mated mothers. Therefore, I speculated that the act of mating or the addition of male sperm may confer a protective factor to protect the cross-progeny and increase their likelihood of survival. To determine whether RNAi protection is conferred by mating, I repeated my experiments using spe-9(eb19) males which can mate but not fertilize oocytes (Singson, Hill, and L’Hernault 1999; Zannoni, L’Hernault, and Singson

2003). When crossing WT;rfp x spe-9(eb19), there was embryonic lethality equal to that of WT;rfp or spe-9(eb19) self-progeny and significantly higher than WT;rfp x N2 progeny, implying that mating without male fertilization does not confer a protective factor.

In the future, additional replicates of these experiments would be required to verify my preliminary findings, as well as varied crosses to investigate each genotypic combination. Furthermore, performing RT-PCR to quantify the presence of pal-1 transcripts would allow interpretation of RNAi efficiency, to determine whether pal-1

103 persists after RNAi. Additionally, I speculate that the maternal transgene reporter could be interfering with RNAi efficiency, due to the effect that transgene-bearing mothers have progeny which survive pal-1 KD at a much higher rate than do progeny from transgene-bearing fathers. This phenomenon could be investigated by comparing the results of pal-1 RNAi on embryos from parents with a transgene to parents with an endogenously tagged gene which is stably inherited.

Further characterization of the pal-1 regulatory network resulted in the creation of the C. elegans Glyph SPLOM containing 655 transcription factors (Yates et al. 2014).

This visualization of co-expression and directedness between TF genes allows for predictions of regulatory behaviors as well as clustering of lineage-specific expression.

This clustering, in particular, is a unique aspect of the Glyph SPLOM that allows for characterization of previously unstudied TF genes. In fact, I was able to choose an E lineage (intestine) clustered gene with no previous expression data, nhr-232, and verify that it is indeed expressed in the E lineage. The predictive ability of the Glyph SPLOM is an invaluable tool for continuing to characterize the complex regulatory networks that exist in even the simple C. elegans, and can certainly be scaled and refined to investigate the interaction networks of more complex organisms.

Embryonic development in C. elegans is an interesting process, as founder blastomeres develop both based on inductive cell-to-cell signaling and autonomous factors that have been segregated within the presumptive cell (Sulston et al. 1983;

Wood and Edgar 1994). pal-1, for example, functions to specify the C blastomere; PAL-

1 can continue to contribute to the identity of C blastomere descendent cells regardless

104 of the environmental context surrounding the C blastomere (Hunter and Kenyon 1996).

The pal-1 ortholog Caudal, on the other hand, functions in the syncytial, spatially dependent development of Drosophila embryos (Boring, Weir, and Schubiger 1993).

Regardless of the differences in pal-1/Caudal function, there are insights to be gained through characterization of the C. elegans pal-1 developmental regulatory network can be applied to higher organisms. In particular, the introduction of computational methods which can interpret wild-type data to produce directed predictions, such as the MIM and

Glyph SPLOM, is a huge step forward in characterizing regulatory networks that contribute to the universal eukaryotic processes of development and organogenesis.

105

4.5 Figures

Figure 14: MIM predicts direct regulatory targets of pal-1 activity.

Fig. 14 (A) This summation of the predicted direct pal-1 targets is a representation of the data presented in the Mathematically Inferred Model of the pal-1 regulatory network

(Stigler and Chamberlin 2012). pal-1 is predicted to have direct interactions with all but two of the TFs included in this network (not shown). The colors of the targets corresponds to the associated mode of activity: blue (initiation), yellow (ectoderm differentiation), gray (mesoderm differentiation), orange (mixed effect), and green

(other).

106

Figure 15: Expectations for embryonic survival of pal-1 KD.

Fig. 15 (A) Embryonic survival of pal-1 RNAi is anticipated to be based on whether both maternal and zygotic pal-1 transcripts are present. The observed outcomes for

107 embryonic survival differed greatly from expected for offspring of crossed parents rather than isolated, self-fertilizing hermaphrodites. On the left in each cross is the hermaphrodite (in gray or pink); on the right is the male (in blue). Genotypes of parents are listed to the left and right of the outcome boxes. Circles represent unhatched embryos (lethality), bulbous larvae represent deformed progeny (ex.rde-1(ne219) x N2 expected), and smooth larvae represent morphologically normal survivors of pal-1 KD.

108

Figure 16: Loss of pal-1 contributes to significant lethality of non-cross progeny.

Fig. 16 (A) Percent of embryos which survived pal-1 KD. For each background and

RNAi combination, at least 52 embryos were screened. Error bars are shown for the standard error of proportion. The statistical significance is indicated on the graph for backgrounds of particular interest. For each pairwise comparison of EV to pal-1 RNAi, the difference is significant at p<0.05 the difference is significant at p<0.05 (2-tailed Z- test).

Work performed: I performed all crosses and RNAi experiments. I screened for all fluorescence.

109

Figure 17: Mating without male-induced fertilization fails to protect embryos against pal-1 KD.

Fig. 17 (A) Percent of embryos which survived pal-1 KD. For each background and

RNAi combination, at least 34 embryos were screened. Error bars are shown for the standard error of proportion. For each pairwise comparison of EV to pal-1 RNAi, the difference is significant at p<0.05 (2-tailed Z-test). spe-9(eb19) worms can self- propagate when maintained as Rollers; as non-Rollers, they cannot fertilize oocytes.

Non-Roller males were used in crosses to allow for mating without fertilization by male sperm.

Work performed: I performed all crosses, RNAi experiments, and screening.

110

Figure 18: The C. elegans 655 TF Glyph SPLOM.

Fig. 18 (A) The Glyph SPLOM was originally presented in (Yates et al. 2014) and represents predictions about regulatory interactions and co-expression relationships present between each pair of TFs. TF genes which are expressed within the same cell lineage cluster as evidenced by the indicated E and C lineage clusters. (B) Each glyph shape and corresponding color represents a unique regulatory implication (Yates et al.

2014). Only the first three glyphs, “Y necessary for X”, “X ⇔ Y”, and “X necessary for Y” represent directed predictions. The other five potential glyphs represent undirected interactions.

Work performed: Thank you to Andrew Yates for developing the Glyph SPLOM from previously published data.

111

Figure 19: nhr-232 is expressed in the E lineage as predicted by the Glyph SPLOM.

Figure 19: nhr-232 is expressed in the E lineage as predicted by the Glyph

SPLOM.

Fig. 19 (A-B) Early embryonic expression of Pnhr-232::gfp is localized to the intestinal cells of the E lineage. (C-D) Embryonic intestinal expression of Pnhr-232::gfp continues through at least the 2-fold stage. After elongation of the embryo, Pnhr-232::gfp is localized to the posterior of the embryo.

Work performed: I developed the Pnhr-232::gfp reporter and screened for fluorescence in vivo.

112

Chapter 5 Discussion

5.1 Results overview

In this work, I sought to explore the role of transcription factors in the coordination of development of Caenorhabditis elegans. I examined the transcription factors EGL-38 and PAL-1, as well as a large-scale prediction of transcription factor interactions, the Glyph SPLOM.

First, I examined EGL-38, a Pax2/5/8 ortholog which is required for the creation of the vulval-uterine connection in C. elegans hermaphrodites (Chamberlin et al. 1997;

Chang, Newman, and Sternberg 1999). egl-38 is expressed in both the vulval vulF and the uterine uv1 cells at the correct times to both activate and respond the inductive EGF pathway. EGL-38 is required for activation of the lin-3/egf ligand in the vulF cells

(Chang, Newman, and Sternberg 1999); in turn, egl-38 is activated by the EGF pathway in the uv1 cells. egl-38, therefore, coordinates its own expression through EGF pathway activity.

EGL-38 is required for the uv1 expression of the neuropeptide proteins nlp-2 and nlp-7. In loss-of-function egl-38(n578) and egl-38(sy294) mutants, nlp-2 and nlp-7 expression is lost; this defect cannot be rescued by activated let-23/egfr, indicating that

EGL-38 is required in the uv1 cells for nlp-2 and nlp-7 activation. The nlp-2 uv1 cell enhancer (nre1) exists upstream from the start codon and is bound by the EGL-38

113

DNA-binding domain in vitro. The nlp-7 uv1 cell enhancer exists upstream from the start codon and does not yet have an identified Pax binding site.

Second, I began to examine the role of Hox transcription factor PAL-1 in regulating a battery of transcription factors predicted by the Mathematically Inferred

Model (MIM) as targets of PAL-1 activation (Stigler and Chamberlin 2012). PAL-1 is required maternally and zygotically for specification of the C blastomere, and for differentiation of ectoderm and mesoderm from the C blastomere descendants (Edgar et al. 2001; Hunter and Kenyon 1996). Knockdown of pal-1 maternal and zygotic transcripts results in embryonic lethality of self-progeny from self-fertilizing hermaphrodites; however, cross-progeny from mated wild-type hermaphrodites survived pal-1 knockdown. I investigated whether a protective factor was being conferred by the males in these crosses to increase the survival of their offspring. Embryos from mothers mated to spe-9(eb19) males, which can mate but not fertilize, did not experience increased survival on pal-1 RNAi compared to wild-type. Therefore, there is no protection against RNAi conferred directly upon mating in the absence of fertilization.

Finally, I approached the C. elegans Glyph SPLOM, a predictor of transcription factor co-expression and interactions (Yates et al. 2014). In this computational model, transcription factors with similar predicted expression patterns group together in lineage clusters. Analysis of the E (intestine) lineage cluster revealed a transcription factor with no known expression pattern, nhr-232. A reporter analysis of this gene confirmed that nhr-232 is expressed strongly and uniquely in the intestinal cells, verifying that the

Glyph SPLOM has the power to predict lineage expression pattern clustering.

114

In the remainder of this chapter, I will discuss several of the new questions the results of my study have prompted.

5.2 Pax involvement with inductive signaling

EGL-38 is required to activate lin-3/egf expression in the vulF cells (Chang,

Newman, and Sternberg 1999); in turn, the EGF pathway is required to activate egl-38 transcription in the uv1 cells. egl-38 therefore utilizes the EGF pathway to coordinate its own expression between these neighboring cells. Is coordination of their own transcription and activity through an inductive pathway a common motif of Pax proteins?

Previously, Pax proteins have been speculated to activate and respond to inductive cell-to-cell signaling, based on their spatiotemporal expression pattern in cells which require inductive signaling (G. R. Dressler et al. 1990; Lang et al. 2007; Püschel,

Gruss, and Westerfield 1992). In Xenopus, Notch signaling has been shown to be sufficient to activate Pax6 expression in eye development (Onuma et al. 2002). In cell culture, Pax2 expression is upregulated upon addition of EGFR (de Graaff et al. 2012;

S. Liu et al. 1997). Knockdown of Pax2 in endometrial cancer lines significantly decreases cell viability, which could be an indicator that EGF expression has been consequently downregulated in the absence of Pax2 (Jia et al. 2016; L.-P. Zhang et al.

2011).

Pax proteins are also often involved in the epithelial-to-mesenchymal transition of organogenesis necessary for kidney, thyroid, nose, thymus, salivary gland, and tooth development (Dahl, Koseki, and Balling 1997; Thesleff, Vaahtokari, and Partanen

115

1995). This transition involves inductive signaling between the epithelium and presumptive mesenchyme; in fact, some Pax mutant phenotypes in these tissues can be attributed to failure of the inductive interaction between cell types (Dahl, Koseki, and

Balling 1997). Importantly, however, these examples all involve Pax as either the activator or responder to inductive signaling, not both. As of yet, there have been no other instances found where a Pax protein activates transcription of a signaling molecule, and then is in turn activated by inductive signaling of that same molecule.

EGL-38 is therefore currently unique in that regard, but given the widespread concurrence of Pax protein activity or expression with inductive signaling pathways, it is likely that more examples can be found in the future of Pax proteins coordinating their own transcription through inductive signaling pathways.

5.3 Pax activation of cellular phenotype

EGL-38 activates the neuropeptide-like genes nlp-2 and nlp-7 in the uv1 cells.

The uv1 cells are neurosecretory, mechanosensory cells which sense the passage of eggs through the vulval-uterine connection, and signals to the hermaphrodite-specific neuron to inhibit further egg-laying (Alkema et al. 2005; Collins et al. 2016; Jose et al.

2007). nlp-7 and flp-11 together are responsible for initiating the inhibitory signal sent to the hermaphrodite-specific neuron (Banerjee et al. 2017). nlp-2 has no currently known functions; however, in the egg-laying system both nlp-2 and nlp-7 are only expressed in mature, specified uv1 cells, indicating that nlp-2 may also have a role in egg-laying.

EGL-38, therefore, activates a cellular program of genes necessary for egg-laying. Do

116 other Pax proteins activate transcription of genes as part of cellular programs required for cell biology and phenotype?

In general, Pax proteins are considered “master regulators” of development because they are necessary to regulate transcription of genes required for organogenesis (Blake and Ziman 2014; Kozmik et al. 2003). Most categorized Pax targets involve cell specification, growth, and differentiation rather than involvement in biological output (Blake and Ziman 2014; Noll 1993). However, there are some examples of Pax proteins both activating cell specification and functional cellular programs. In the thyroid, Pax8 is necessary for specifying the identity of thyroid follicular cells and direct activation of its targets thyroid peroxidase and thyroglobulin (Fabbro et al. 1994; Magliano, Lauro, and Zannini 2000; Mansouri, Chowdhury, and Gruss 1998).

These two proteins are required for thyroid hormonogenesis, and therefore the biological output of the thyroid follicular cells (Di Jeso and Arvan 2016; Taurog, Dorris, and Doerge 1996). Pax6 is required for the development of pancreatic islet cells as well as transcription of pancreatic hormones such as insulin and glucagon, which have direct biological functions in regulating metabolism (Sander et al. 1997; St-Onge et al. 1997).

Interestingly, these neuroendocrine organs, the thyroid and pancreas, are among the very few organs that express Pax proteins in the adult (Lang et al. 2007). Most Pax expression is downregulated to background levels following development (Blake and

Ziman 2014; Czerny, Schaffner, and Busslinger 1993; Dressler et al. 1990; Lang et al.

2007; Rothenpieler and Dressler 1993; Terzić et al. 1998). Do Pax proteins only possess transcriptional phenotypic activation roles in neurosecretory/neuroendocrine

117 organs? It is possible and likely that Pax transcription factors can only activate biological output when they are expressed in mature, post-developmental cells, but this avenue of inquiry remains to be studied.

5.4 Pax role in cancer

EGL-38 is required both to activate and respond to the EGF pathway in the C. elegans egg-laying system (Chapter 2)(Chang, Newman, and Sternberg 1999;

Rajakumar and Chamberlin 2007). This confirmed interaction between Pax and the

EGF pathway has been speculated in other systems, due to observations such as the upregulation of Pax2 transcription upon exposure to EGFR in cancer cell lines and the concomitant expression levels of aberrant Pax2 and EGF in endometriosis ectopic tissues (de Graaff et al. 2012; S. Liu et al. 1997). Does an interaction between Pax and the EGF pathway at the transcriptional level contribute to cancer progression?

The EGF pathway is one of the most heavily investigated factors in cancer, because EGF activity contributes to the maintenance of tumor cell identity by contributing to tumor proliferation, growth, and resistance to apoptosis (Brand et al.

2011; Herbst 2004; Konecny et al. 2009; Muratovska et al. 2003; J. Wang et al. 2018;

Q. Wang et al. 2008). Pax proteins are also comprehensively studied in cancer, due to their role as master regulators of development that are necessary for enacting the cell differentiation process (Kozmik et al. 2003; Wachtel and Schäfer 2015). When mutated, aberrant PAX activity is responsible for tumorigenesis in several types of cancer, including thyroid cancer and endometrial cancer, due to un-regulated PAX keeping cells

118 in a proliferative state (Wachtel and Schäfer 2015). In addition, because Pax is not commonly expressed in adults, most PAX upregulation in the cancer state is likely contributing to malignancies. Does aberrant PAX activity upregulate EGF transcription to contribute to maintenance of tumor identity? Does overactive EGF inappropriately activate Pax transcription to maintain tumor cells in a proliferative, undifferentiated state? Do Pax and EGF feed into one another for cancer progression?

The up-regulation outcome, concomitant expression, and similar contribution to tumor proliferation would indicate that Pax and EGF are interacting in the cancer state.

EGF is a necessary component of cell maintenance that is required in healthy tissues, which makes it a poor molecular target of drug inhibition therapies as non-cancer cells would suffer in its absence. Pax, on the other hand, is rarely active in differentiated tissues, and possibly presents a directed target of cancer therapies that could halt the preservation of tumor cell identity by preventing upregulated EGF transcription or activity (Grimley and Dressler 2018).

5.5 Significance of Glyph SPLOM computational method

The Glyph SPLOM is a visualization of distance correlation (measure of dependence on one another) and dependency class (measure of co-expression levels) between 655 C. elegans transcription factors, combined into a concise summarization of interrelatedness (Yates et al. 2014). How does the Glyph SPLOM compare to existing computational methods? What is its significance in this field?

119

Essentially, the Glyph SPLOM is a combination of a highly detailed scatterplot matrix and a highly summarized heatmap; the Glyph SPLOM is more feasible to resolve for high-throughput data than the former, and more comprehensive than the latter

(Sahoo et al. 2008; Székely, Rizzo, and Bakirov 2007; Yates et al. 2014). The Glyph

SPLOM falls under the categorization of a “top down” methodology, where a model is reverse engineered from available data, often in the absence of all known features

(Stigler and Chamberlin 2012). Many reverse engineering methodologies exist; however, these often perform poorly when not provided with comprehensive genetic perturbation data (D’haeseleer, Liang, and Somogyi 2000; Gardner and Faith 2005; de

Jong 2002; W.-P. Lee and Tzou 2009; Marbach et al. 2010). The Glyph SPLOM, in contrast, was assembled from one-half wild-type microarray data and one-half genetic perturbation microarray data from a single mutation background. When comparing the

Glyph SPLOM predictions of the pal-1 14 transcription factor regulatory network to the

Gold-Standard Network derived from more intensive mutant backgrounds and formalized by (Stigler and Chamberlin 2012), the Glyph SPLOM methodology predicted pal-1 network interactions with higher accuracy, specificity, and precision (Yanai et al.

2008; Yates et al. 2014).

The ability to predict regulatory interactions with a high degree of co-localization specificity and regulatory directedness is an extremely useful feature, particularly when using primarily wild-type data. Genetic perturbation experiments are costly, time- exhaustive, and difficult to perform as well as complicated to interpret. Therefore, knowledge of some transcription factor activity will always be absent, increasing the

120 complexity of predicting regulatory networks with top down models (Selinger, Wright, and Church 2003). The Glyph SPLOM is capable of accurate predictions of directed interactions and of expression pattern clustering. The lineage clustering, in particular, is a unique feature of the Glyph SPLOM that can directly fill in gaps of knowledge about transcription factors which have not yet been studied, such as predicting the expression pattern of nhr-232 in the E lineage, which was used to verify this ability (Chapter 4).

While the Glyph SPLOM will require further enhancements to perfect its visualization of large-scale data, it does represent a significant improvement over existing methods by combining summarization and detail into a single representation

(Yates et al. 2014). Additionally, the strength of the Glyph SPLOM to predict expression patterns, directed interactions, and undirected relationships represents a substantial addition to the field of computational modeling of regulatory networks.

5.6 Final conclusions

In this study, I have examined the role of the Pax transcription factor EGL-38 in the development of the C. elegans egg-laying system. My data has shown that EGL-38 coordinates its own expression through the EGF inductive signaling pathway, and that

EGL-38 activates targets in the uv1 cells necessary for egg-laying output. In addition, I have studied the Hox transcription factor PAL-1 regulatory network, and discovered that there are unique embryonic reactions to loss of zygotic pal-1 that require further investigation. Finally, I have also verified that the large-scale representation of

121 transcription factor interactions, the Glyph SPLOM, can accurately predict lineage expression clustering of its represented proteins.

In investigating these transcription factors and their regulatory networks in C. elegans, we can gain insights and a fuller understanding of how they perform in humans to contribute to normal development as well as disease states such as cancer. I believe that the results I have discovered will provide a foundational background for pursuing the investigative questions set forth in this work.

122

Appendix: Tables

Table 1: Chapter 2 primers

EGL-38 Guide RNA sequence Guide: TTTTGTGGTAGAATAACCAT

EGL -38 Introducing homology arms to egl-38 endogenous locus Primer: scaffold HOMOLOGY PRIMER PR5128 5’- acgttgtaaaacgacggccagtcgccggcagtc ctc tgc tac gtc aca tca tc-3’ 5’- catcgatgctcctgaggctcccgatgctcccaa ttg tcc aca ata gtt ttg agg aag tat PR5132 aac -3’ 5’- cgtgattacaaggatgacgatgacaagagatag aaa tga aag tta gct tta ttg aga atc PR5133 ca-3’ PR5134 5’- ggaaacagctatgaccatgttatcgatttccct tgt tgt cac cca atg aat ctg-3’

EGL -38 Genotyping homology arm insertion Primer: Amplify endogenous locus with inserted GFP^FLAG PR5401 5’- CAACATCCGTGACGTCATCAG - 3' PR5403 5’- CTAATTAAACCAGCCATTCCAGGAC - 3' nlp -2 Pnlp-2::gfp creation Primer: SalI tagged primer at the nlp-2 start codon PR2939 5' - TAA TGT CGA CCA TTT CTC GCG TTG TTG GGA - 3' nlp-2 Pnlp-2::gfp creation (full-length and truncations) Primer: SphI tagged primer upstream of start codon Pnlp-2(2.6kb) 5' - gcg ggc atg ctc cgg att cac tag gat tgt c - 3' Pnlp-2(2 kb) 5' - gcg gca tgc atg agg agt aac ggt tct ggg - 3' Pnlp-2(1.6 kb) 5' - gcg gca tgc gcc cat aaa gtc atg cgt ggt - 3' Pnlp-2(1 kb) 5' - gcg gca tgc gac ccg gct cct aat act att cc - 3' Pnlp-2(0.5 kb) 5' - gcg gca tgc caa acc gta tat agt ttt ctg ggc - 3' Pnlp-2(0.4 kb) 5' - gcg gca tgc gct ctg cag ttt cat agc tgt tac c - 3' Pnlp-2(0.38 kb) 5' - gcg gca tgc ctt gaa tca aat cgg ttc cgc - 3' Pnlp-2(0.34 kb) 5' - gcg gca tgc caa aaa ttg tga cca ttt tcc cc - 3' Pnlp-2(0.3 kb) 5' - gcg gca tgc ctc ttc caa ttt ctg acc aac gtc - 3' Pnlp-2(0.2 kb) 5' - gcg gca tgc cct gat tcg cgt ctt gcc - 3' Pnlp-2(0.1 kb) 5' - gcg gca tgc gcc gta taa att tgg ata agc atc tac tg - 3' nlp -2 Introduction of mutated termini into Pnlp-2(0.4kb)::gfp plasmid Primer: N terminal Pnlp-2(0.4 N-) F 5' - gca cta ttc aaa aat tcg aac cat ttt ccc cat ctc - 3' Pnlp-2(0.4 N-) R 5' - gag atg ggg aaa atg gtt cga att ttt gaa tag tgc - 3' Primer: C terminal

123

Pnlp-2(0.4 C-) F 5' - caa atc ggt tcc gca cta tta gat cta ttc aaa aat tgt gac c - 3' Pnlp-2(0.4 C-) R 5' - ggt cac aat ttt tga ata gat cta ata gtg cgg aac cga ttt g - 3' Primer: N and C termini Pnlp-2(0.4 NC-) F 5' - caa atc ggt tcc gca cta tta gat cta ttc aaa aat tcg aac c - 3' Pnlp-2(0.4 NC-) R 5' - ggt tcg aat ttt tga ata gat cta ata gtg cgg aac cga ttt g - 3' nlp -7 Pnlp-7::gfp creation PR4834 5' - ata tgc atg cag tga cgt tgg tac gct gga g - 3' PR4835 5' - ata tgt cga cca tga tta cca ctg aaa taa act tta tgg tg - 3'

Table 2: Chapter 3 primers lin-3 Plin-3::gfp creation (full-length and isolations) Primer name Sequence #1 5' HindIII tagged 5' - aaaaaaagcttcggcttaccaaattgtgtctc - 3' #1 3' NheI tagged 5' -aaaaagctagccaactcgatgagcaatcactg - 3' #2 5' HindIII tagged 5' -aaaaaaagcttaggaagaaccagaagtagtgtc - 3' #2 3' NheI tagged 5' -aaaaagctagctagagaaagatcgggagaagtg - 3' #3 5' HindIII tagged 5' -aaaaaaagcttcgcatcttctctatttgctctc - 3' #3 3' NheI tagged 5' -aaaaagctagcgtgtagaacacgaacacacatg - 3' #4 5' HindIII tagged 5' -aaaaaaagcttagaataacggatcgtcggcttg - 3' #4 3' NheI tagged 5' -aaaaatctagaatgacaaaccattcaagggacc - 3' #5 5' HindIII tagged 5' -aaaaaaagcttactcttcaaaggcgctcatc - 3' #5 3' NheI tagged 5' -aaaaagctagcgtttcagtaggtgtttcaggtg - 3' #6 5' HindIII tagged 5' -aaaaaaagcttggtttcgtcaagaacgtagtg - 3' #6 3' NheI tagged 5' -aaaaagctagcgcaagaaggaacaactgctg - 3' #7 5' HindIII tagged 5' -aaaaaaagcttcgacgacatctttccatgattc - 3' #7 3' NheI tagged 5' -aaaaagctagcctattcctgcgttcgcattt - 3' #8 5' HindIII tagged 5' -aaaaaaagcttgggctttatgagagaattgtgg - 3' #8 3' NheI tagged 5' -aaaaagctagctgatacctggatggaatggtac - 3' egl -38 egl-38 RNAi construct 5' SacI tagged aaaaGAGCTCTGTTCTCCGACAAGTCCAG 3' KpnI tagged aaaaGGTACCCCTATTTGGAGAAATTATCCTTACCC nlp -7 Pnlp-7::gfp creation PR4834 5' - ata tgc atg cag tga cgt tgg tac gct gga g - 3' PR4835 5' - ata tgt cga cca tga tta cca ctg aaa taa act tta tgg tg - 3' nlp -7 Pnlp-7::gfp creation (full-length and truncations) PR4834 5' - ata tgc atg cag tga cgt tgg tac gct gga g - 3' PR4835 5' - ata tgt cga cca tga tta cca ctg aaa taa act tta tgg tg - 3'

124

Distal primer PR4834 5' - ata tgc atg cag tga cgt tgg tac gct gga g - 3' Proximal primers Pnlp -7(2 kb) 5' - tctggtgccctcttagtgatc - 3' Pnlp-7(0.86 kb) 5' - ggctccactttccgacattctgaac - 3' Pnlp-7(0.83 kb) 5' - ctgtgttctcagtgaaccaagac - 3' Pnlp-7(0.8 kb) 5' - ctgtgttctcagtgaaccaagac - 3' Pnlp-7(0.63 kb) 5' - tttattgcactctcccggtc - 3' Pnlp-7(0.5 kb) 5' - tatacctctcggtgctcctc - 3' Pnlp-7(0.38 kb) 5' -atccattggcggattacaagg - 3' nlp -7 nlp-7 introduction of N-terminal mutations into 0.86 kb construct Forward 5'- cttggaaatgaaaggctcGaGtttccgacattctgaactttcctgtcaaactttc -3' Reverse 5'- gaaagtttgacaggaaagttcagaatgtcggaaaCtCgagcctttcatttccaag -3' nlp -7 nlp-7 purchased G-Block (IDTdna) with mutations in bold acaacttggaaatgaaataagcttgcatgcggctcGAGtttccgacattctAGactttcctgtcaaactttctccgaatca aaatcaatctgtgttctcagtgaaccaagacatgtgcataaatggtcggtttaaattggaatgcaatgtgatgtgcagttttaa tctgcgaaccaatctcgtttttctttttttttttcggcagacagattctcaacggaatagtgagtgggaaaagggaagatcaaa ggaacgggcccaattttttattgcactctcccggtcttttttttcaaccttgtattcgtgatttggtttgttttggctcaagatttacga catgatgtcaatcggttttgactcggtgttctttttcatccctctctccagctatacctctcggtgctcctctatacaacggcggtc ccgcctttttttctcaacgtggtgggcacataacaagagacagaggaatcgaaatcttcagtttgacaattttcataactaat cctatccattggcggattacaaggtggtgtgaggggtggcgtggctgaagacaaccgagcaccatgagctccgccctct actagtggtcttcatttcgttttcaatcaagcccagtgtgcttatcaatcatttcctttcctgtccagctttcaacattctagaagtttt cacccagtatttccacctgtcacttttcccactcccacttgctctgatatattcaactcatgtatatatgacctgacctgatatgtg ctgagattttattttgggatgatgtttcgtgtttcatttctcacatgatctcccatcattagtattcctccttcgttaacctctctgtttag cctatcaccataaagtttatttcagtggtaatcatggtcgactctagaggatccccgggattggcc

125

References

Abate-Shen, Cory. 2002. “Deregulated Homeobox Gene Expression in Cancer: Cause or Consequence?” Nature Reviews Cancer 2(10): 777.

Affolter, Markus, Matthew Slattery, and Richard S. Mann. 2008. “A Lexicon for Homeodomain-DNA Recognition.” Cell 133(7): 1133–35.

Ahmadzadeh, Mojgan et al. 2009. “Tumor Antigen–Specific CD8 T Cells Infiltrating the Tumor Express High Levels of PD-1 and Are Functionally Impaired.” Blood 114(8): 1537–44.

Akarsu, A. Nurten et al. 1996. “Genomic Structure of HOXD13 Gene: A Nine Polyalanine Duplication Causes Synpolydactyly in Two Unrelated Families.” Human Molecular Genetics 5(7): 945–52.

Alasti, Fatemeh et al. 2008. “A Mutation in HOXA2 Is Responsible for Autosomal- Recessive Microtia in an Iranian Family.” The American Journal of Human Genetics 82(4): 982–91.

Alkema, Mark J., Melissa Hunter-Ensor, Niels Ringstad, and H. Robert Horvitz. 2005. “Tyramine Functions Independently of Octopamine in the Caenorhabditis Elegans Nervous System.” Neuron 46(2): 247–60.

Antoshechkin, Igor, and Paul W. Sternberg. 2007. “The Versatile Worm: Genetic and Genomic Resources for Caenorhabditis Elegans Research.” Nature Reviews Genetics 8(7): 518–32.

Aroian et al. 1990. “The Let-23 Gene Necessary for Caenorhabditis Elegans Vulval Induction Encodes a Tyrosine Kinase of the EGF Receptor Subfamily.” Nature 348(6303): 693.

Aroian, and P. W. Sternberg. 1991. “Multiple Functions of Let-23, a Caenorhabditis Elegans Receptor Tyrosine Kinase Gene Required for Vulval Induction.” Genetics 128(2): 251–67.

Artavanis-Tsakonas, Spyros, Matthew D. Rand, and Robert J. Lake. 1999. “Notch Signaling: Cell Fate Control and Signal Integration in Development.” Science 284(5415): 770–76.

126

Banerjee, Navonil et al. 2017. “Local Neuropeptide Signaling Modulates Serotonergic Transmission to Shape the Temporal Organization of C. Elegans Egg-Laying Behavior.” PLOS Genetics 13(4): e1006697.

Barolo, Scott, and James W. Posakony. 2002. “Three Habits of Highly Effective Signaling Pathways: Principles of Transcriptional Control by Developmental Cell Signaling.” Genes & Development 16(10): 1167–81.

Barr, Frederic G. et al. 1999. “Predominant Expression of Alternative PAX3 and PAX7 Forms in Myogenic and Neural Tumor Cell Lines.” Cancer Research 59(21): 5443–48.

Baugh et al. 2005. “The Homeodomain Protein PAL-1 Specifies a Lineage-Specific Regulatory Network in the C. Elegans Embryo.” Development 132(8): 1843–54.

Beitel, G J, S Tuck, I Greenwald, and H R Horvitz. 1995. “The Caenorhabditis Elegans Gene Lin-1 Encodes an ETS-Domain Protein and Defines a Branch of the Vulval Induction Pathway.” Genes & Development 9(24): 3149–62.

Bernasconi, Michele et al. 1996. “Induction of Apoptosis in Rhabdomyosarcoma Cells through Down-Regulation of PAX Proteins.” Proceedings of the National Academy of Sciences 93(23): 13164–69.

Blake, J. A., and M. R. Ziman. 2014. “Pax Genes: Regulators of Lineage Specification and Progenitor Cell Maintenance.” Development 141(4): 737–51.

Bodaker, Matan, Eran Meshorer, Eduardo Mitrani, and Yoram Louzoun. 2014. “Genes Related to Differentiation Are Correlated with the Gene Regulatory Network Structure.” Bioinformatics 30(3): 406–13.

Bopp, D. et al. 1986. “Conservation of a Large Protein Domain in the Segmentation Gene Paired and in Functionally Related Genes of Drosophila.” Cell 47(6): 1033– 40.

Boring, Landing, Michael Weir, and Gerold Schubiger. 1993. “Egg Ligation Alters the Bcd Protein Gradient and Segmentation Gene Expression in Embryos of Drosophila.” Mechanisms of Development 43(1): 84.

Bower, Matthew A., Lisa A. Schimmenti, and Michael R. Eccles. 1993. “PAX2-Related Disorder.” In GeneReviews®, eds. Margaret P. Adam et al. Seattle (WA): University of Washington, Seattle. http://www.ncbi.nlm.nih.gov/books/NBK1451/ (June 12, 2019).

127

Bowerman, Bruce, Bruce W. Draper, Craig C. Mello, and James R. Priess. 1993. “The Maternal Gene Skn-1 Encodes a Protein That Is Distributed Unequally in Early C. Elegans Embryos.” Cell 74(3): 443–52.

Bowers, G. et al. 2001. “The Relative Role of ErbB1–4 Receptor Tyrosine Kinases in Radiation Signal Transduction Responses of Human Carcinoma Cells.” Oncogene 20(11): 1388.

Brand, Toni M., Mari Iida, Chunrong Li, and Deric L. Wheeler. 2011. “The Nuclear Epidermal Growth Factor Receptor Signaling Network and Its Role in Cancer.” Discovery Medicine 12(66): 419–32.

Brenner, S. 1974. “The Genetics of Caenorhabditis Elegans.” Genetics 77(1): 71–94.

Burton, Quianna et al. 2004. “The Role of Pax2 in Mouse Inner Ear Development.” Developmental Biology 272(1): 161–75.

Campbell, Iain D., and Peer Bork. 1993. “Epidermal Growth Factor-like Modules.” Current Opinion in Structural Biology 3(3): 385–92.

Chamberlin, H. M. et al. 1997. “The PAX Gene Egl-38 Mediates Developmental Patterning in Caenorhabditis Elegans.” Development 124(20): 3919–28.

Chang, Chieh, Anna P. Newman, and Paul W. Sternberg. 1999. “Reciprocal EGF Signaling Back to the Uterus from the Induced C. Elegans Vulva Coordinates Morphogenesis of Epithelia.” Current Biology 9(5): 237–46.

Chasnov, J R, and King L Chow. 2001. “Why Are There Males in the Hermaphroditic Species Caenorhabditis Elegans?” : 12.

Chen, Hexin et al. 2007. “HOXA5 Acts Directly Downstream of Retinoic Acid Receptor β and Contributes to Retinoic Acid–Induced Apoptosis and Growth Inhibition.” Cancer Research 67(17): 8007–13.

Chen, Kevin, and Nikolaus Rajewsky. 2007. “The Evolution of Gene Regulation by Transcription Factors and MicroRNAs.” Nature Reviews Genetics 8(2): 93–103.

Cheng, Binbin et al. 2018. “Recent Advances in Small Molecule Based Cancer Immunotherapy.” European Journal of Medicinal Chemistry 157: 582–98.

Chi, Neil, and Jonathan A Epstein. 2002. “Getting Your Pax Straight: Pax Proteins in Development and Disease.” Trends in Genetics 18(1): 41–47.

Collins, Kevin M et al. 2016. “Activity of the C. Elegans Egg-Laying Behavior Circuit Is Controlled by Competing Activation and Feedback Inhibition.” eLife 5. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5142809/ (March 11, 2019).

128

Czerny, T., G. Schaffner, and M. Busslinger. 1993. “DNA Sequence Recognition by Pax Proteins: Bipartite Structure of the Paired Domain and Its Binding Site.” Genes & Development 7(10): 2048–61.

Dahl, Edgar, Haruhiko Koseki, and Rudi Balling. 1997. “Pax Genes and Organogenesis.” BioEssays 19(9): 755–65.

Daniel, Laurent et al. 2001. “Pax-2 Expression in Adult Renal Tumors.” Human Pathology 32(3): 282–87.

Davis, Richard J. et al. 1994. “Fusion of PAX7 to FKHR by the Variant t(1;13)(P36;Q14) Translocation in Alveolar Rhabdomyosarcoma.” Cancer Research 54(11): 2869– 72.

Desplan, Claude, Jim Theis, and Patrick H. O’Farrell. 1988. “The Sequence Specificity of Homeodomain-DNA Interaction.” Cell 54(7): 1081–90.

D’haeseleer, P., S. Liang, and R. Somogyi. 2000. “Genetic Network Inference: From Co-Expression Clustering to Reverse Engineering.” Bioinformatics 16(8): 707– 26.

Di Jeso, Bruno, and Peter Arvan. 2016. “Thyroglobulin From Molecular and Cellular Biology to Clinical Endocrinology.” Endocrine Reviews 37(1): 2–36.

Dickinson, Daniel J., and Bob Goldstein. 2016. “CRISPR-Based Methods for Caenorhabditis Elegans Genome Engineering.” Genetics 202(3): 885–901.

Dickinson, Daniel J., Jordan D. Ward, David J. Reiner, and Bob Goldstein. 2013. “Engineering the Caenorhabditis Elegans Genome Using Cas9-Triggered Homologous Recombination.” Nature Methods 10(10): 1028–34.

Dressler, G. R. et al. 1990. “Pax2, a New Murine Paired-Box-Containing Gene and Its Expression in the Developing Excretory System.” Development 109(4): 787–95.

Dressler, G. R., and E. C. Douglass. 1992. “Pax-2 Is a DNA-Binding Protein Expressed in Embryonic Kidney and Wilms Tumor.” Proceedings of the National Academy of Sciences 89(4): 1179–83.

Dressler, Gregory R. et al. 1993. “Deregulation of Pax-2 Expression in Transgenic Mice Generates Severe Kidney Abnormalities.” Nature 362(6415): 65.

Duboule, D., and P. Dollé. 1989. “The Structural and Functional Organization of the Murine HOX Gene Family Resembles That of Drosophila Homeotic Genes.” The EMBO Journal 8(5): 1497–1505.

129

Duncia, John V. et al. 1998. “MEK Inhibitors: The Chemistry and Biological Activity of U0126, Its Analogs, and Cyclization Products.” Bioorganic & Medicinal Chemistry Letters 8(20): 2839–44.

Eccles, Mike Roger, and Caiyun G. Li. 2012. “PAX Genes in Cancer; Friends or Foes?” Frontiers in Genetics 3. https://www.frontiersin.org/articles/10.3389/fgene.2012.00006/full (June 12, 2019).

Edgar, Lois G., Stephen Carr, Hong Wang, and William B. Wood. 2001. “Zygotic Expression of the Caudal Homolog Pal-1 Is Required for Posterior Patterning in Caenorhabditis Elegans Embryogenesis.” Developmental Biology 229(1): 71–88.

Ekker, S. C. et al. 1994. “The Degree of Variation in DNA Sequence Recognition among Four Drosophila Homeotic Proteins.” The EMBO Journal 13(15): 3551–60.

Ekker, S. C., K. E. Young, D. P. von Kessler, and P. A. Beachy. 1991. “Optimal DNA Sequence Recognition by the Ultrabithorax Homeodomain of Drosophila.” The EMBO Journal 10(5): 1179–86.

Epstein, J. et al. 1994. “Identification of a Pax Paired Domain Recognition Sequence and Evidence for DNA-Dependent Conformational Changes.” The Journal of Biological Chemistry 269(11): 8355–61.

Epstein, J. A. et al. 1994. “Two Independent and Interactive DNA-Binding Subdomains of the Pax6 Paired Domain Are Regulated by Alternative Splicing.” Genes & Development 8(17): 2022–34.

Ethier, Stephen P. 2002. “Signal Transduction Pathways: The Molecular Basis for Targeted Therapies.” Seminars in Radiation Oncology 12(3, Supplement): 3–10.

Fabbro, Dora et al. 1994. “Expression of Thyroid-Specific Transcription Factors TTF-1 and PAX-8 in Human Thyroid Neoplasms.” Cancer Research 54(17): 4744–49.

Favier, B. et al. 1996. “Functional Cooperation between the Non-Paralogous Genes Hoxa-10 and Hoxd-11 in the Developing Forelimb and Axial Skeleton.” Development 122(2): 449–60.

Fire, Andrew et al. 1998. “Potent and Speciﬁc Genetic Interference by Double-Stranded RNA In.” 391: 6.

Fitzsimmons, D. et al. 1996. “Pax-5 (BSAP) Recruits Ets Proto-Oncogene Family Proteins to Form Functional Ternary Complexes on a B-Cell-Specific Promoter.” Genes & Development 10(17): 2198–2211.

130

———. 2001. “Highly Conserved Amino Acids in Pax and Ets Proteins Are Required for DNA Binding and Ternary Complex Assembly.” Nucleic Acids Research 29(20): 4154–65.

Fong, Abraham P, and Stephen J Tapscott. 2013. “Skeletal Muscle Programming and Re-Programming.” Current Opinion in Genetics & Development 23(5): 568–73.

Frøkjaer-Jensen, Christian et al. 2008. “Single-Copy Insertion of Transgenes in Caenorhabditis Elegans.” Nature Genetics 40(11): 1375–83.

Frost, Victoria, Timothy Grocott, Michael Eccles, and Andrew Chantry. 2008. 43 Self- Regulated Pax Gene Expression and Modulation by the TGFβ Superfamily.

Garcia, L. Rene, Brigitte LeBoeuf, and Pamela Koo. 2007. “Diversity in Mating Behavior of Hermaphroditic and Male–Female Caenorhabditis Nematodes.” Genetics 175(4): 1761–71.

Gardner, T, and J Faith. 2005. “Reverse-Engineering Transcription Control Networks.” Physics of Life Reviews 2(1): 65–88.

Garvie, Colin W, James Hagman, and Cynthia Wolberger. 2001. “Structural Studies of Ets-1/Pax5 Complex Formation on DNA.” Molecular Cell 8(6): 1267–76.

Gilbert, Scott F. 2000. “The Developmental Mechanics of Cell Specification.” Developmental Biology. 6th edition. https://www.ncbi.nlm.nih.gov/books/NBK9968/ (June 12, 2019).

Glaser, T. et al. 1994. “PAX6 Gene Dosage Effect in a Family with Congenital Cataracts, Aniridia, Anophthalmia and Central Nervous System Defects.” Nature Genetics 7(4): 463–71.

Gnarra, James R., and Gregory R. Dressler. 1995. “Expression of Pax-2 in Human Renal Cell Carcinoma and Growth Inhibition by Antisense Oligonucleotides.” Cancer Research 55(18): 4092–98. de Graaff, A. A. et al. 2012. “Paired-Box Gene 2 Is down-Regulated in Endometriosis and Correlates with Low Epidermal Growth Factor Receptor Expression.” Human Reproduction (Oxford, England) 27(6): 1676–84.

Greer, Joy M., John Puetz, Kirk R. Thomas, and Mario R. Capecchi. 2000. “Maintenance of Functional Equivalence during Paralogous Hox Gene Evolution.” Nature 403(6770): 661.

131

Grimley, Edward, and Gregory R. Dressler. 2018. “Are Pax Proteins Potential Therapeutic Targets in Kidney Disease and Cancer?” Kidney International 94(2): 259–67.

Grishok, A. 2000. “Genetic Requirements for Inheritance of RNAi in C. Elegans.” Science 287(5462): 2494–97.

Hasty, Jeff, David McMillen, Farren Isaacs, and James J. Collins. 2001. “Computational Studies of Gene Regulatory Networks: In Numero Molecular Biology.” Nature Reviews Genetics 2(4): 268.

Herbst, Roy S. 2004. “Review of Epidermal Growth Factor Receptor Biology.” International Journal of Radiation Oncology, Biology, Physics 59(2 Suppl): 21– 26.

Hill, R. E. et al. 1991. “Mouse Small Eye Results from Mutations in a Paired-like Homeobox-Containing Gene.” Nature 354(6354): 522–25.

Hill, R. J., and P. W. Sternberg. 1992. “The Gene Lin-3 Encodes an Inductive Signal for Vulval Development in C. Elegans.” Nature 358(6386): 470–76.

Hoey, Timothy, and Michael Levine. 1988. “Divergent Homeo Box Proteins Recognize Similar DNA Sequences in Drosophila.” Nature 332(6167): 858.

Hueber et al. 2006. “PAX2 Inactivation Enhances Cisplatin-Induced Apoptosis in Renal Carcinoma Cells.” Kidney International 69(7): 1139–45.

Hueber, PA et al. 2008. “In Vivo Validation of PAX2 as a Target for Renal Cancer Therapy.” Cancer Letters 265(1): 148–55.

Hunter, Craig P, and Cynthia Kenyon. 1996. “Spatial and Temporal Controls Target Pal- 1 Blastomere-Specification Activity to a Single Blastomere Lineage in C. Elegans Embryos.” Cell 87(2): 217–26.

Hwang, Byung Joon, and Paul W. Sternberg. 2004. “A Cell-Specific Enhancer That Specifies Lin-3 Expression in the C. Elegans Anchor Cell for Vulval Development.” Development 131(1): 143–51.

Iwafuchi-Doi, Makiko, and Kenneth S. Zaret. 2016. “Cell Fate Control by Pioneer Transcription Factors.” Development 143(11): 1833–37.

Jia, Nan et al. 2016. “DNA Methylation Promotes Paired Box 2 Expression via Myeloid Zinc Finger 1 in Endometrial Cancer.” Oncotarget 7(51): 84785–97.

132

Johnson, Andrew D., Daniel Fitzsimmons, James Hagman, and Helen M. Chamberlin. 2001. “EGL-38 Pax Regulates the Ovo-Related Gene Lin-48 during Caenorhabditis Elegans Organ Development.” Development 128(15): 2857–65.

Jones, Kevin T., and Kaveh Ashrafi. 2009. “Caenorhabditis Elegans as an Emerging Model for Studying the Basic Biology of Obesity.” Disease Models & Mechanisms 2(5–6): 224–29. de Jong, Hidde. 2002. “Modeling and Simulation of Genetic Regulatory Systems: A Literature Review.” Journal of Computational Biology 9(1): 67–103.

Jose, Antony M., I. Amy Bany, Daniel L. Chase, and Michael R. Koelle. 2007. “A Specific Subset of Transient Receptor Potential Vanilloid-Type Channel Subunits in Caenorhabditis Elegans Endocrine Cells Function as Mixed Heteromers to Promote Neurotransmitter Release.” Genetics 175(1): 93–105.

Kamath, Ravi S., and Julie Ahringer. 2003. “Genome-Wide RNAi Screening in Caenorhabditis Elegans.” Methods 30(4): 313–21.

Katz, W S et al. 1996. “A Point Mutation in the Extracellular Domain Activates LET-23, the Caenorhabditis Elegans Epidermal Growth Factor Receptor Homolog.” Molecular and Cellular Biology 16(2): 529–37.

Katz, W. S., R. J. Hill, T. R. Clandinin, and P. W. Sternberg. 1995. “Different Levels of the C. Elegans Growth Factor LIN-3 Promote Distinct Vulval Precursor Fates.” Cell 82(2): 297–307.

Keller, S. A. et al. 1994. “Kidney and Retinal Defects (Krd), a Transgene-Induced Mutation with a Deletion of Mouse Chromosome 19 That Includes the Pax2 Locus.” Genomics 23(2): 309–20.

Kelly, K M et al. 1997. “Common and Variant Gene Fusions Predict Distinct Clinical Phenotypes in Rhabdomyosarcoma.” Journal of Clinical Oncology 15(5): 1831– 36.

Kerr, William G., and John D. Chisholm. 2019. “The Next Generation of Immunotherapy for Cancer: Small Molecules Could Make Big Waves.” The Journal of Immunology 202(1): 11–19.

Klambt, C. 1993. “The Drosophila Gene Pointed Encodes Two ETS-like Proteins Which Are Involved in the Development of the Midline Glial Cells.” Development 117(1): 163–76.

Kmita, Marie et al. 2005. “Early Developmental Arrest of Mammalian Limbs Lacking HoxA / HoxD Gene Function.” Nature 435(7045): 1113.

133

Konecny, G. E. et al. 2009. “HER2 Gene Amplification and EGFR Expression in a Large Cohort of Surgically Staged Patients with Nonendometrioid (Type II) Endometrial Cancer.” British Journal of Cancer 100(1): 89–95.

Kozmik, Zbynek et al. 2003. “Role of Pax Genes in Eye Evolution: A Cnidarian PaxB Gene Uniting Pax2 and Pax6 Functions.” Developmental Cell 5(5): 773–85.

Lai, Chun-Hung et al. 2000. “Identification of Novel Human Genes Evolutionarily Conserved in Caenorhabditis Elegans by Comparative Proteomics.” Genome Research 10(5): 703–13.

Lambert, Samuel A. et al. 2018. “The Human Transcription Factors.” Cell 172(4): 650– 65.

Lang, Deborah et al. 2000. “Pax3 Is Required for Enteric Ganglia Formation and Functions with Sox10 to Modulate Expression of c-Ret.” The Journal of Clinical Investigation 106(8): 963–71.

———. 2007. “PAX Genes: Roles in Development, Pathophysiology, and Cancer.” Biochemical Pharmacology 73(1): 1–14.

Lee, Tong Ihn, and Richard A. Young. 2013. “Transcriptional Regulation and Its Misregulation in Disease.” Cell 152(6): 1237–51.

Lee, W.-P., and W.-S. Tzou. 2009. “Computational Methods for Discovering Gene Networks from Expression Data.” Briefings in Bioinformatics. https://academic.oup.com/bib/article-lookup/doi/10.1093/bib/bbp028 (June 1, 2019).

Leight, Elizabeth R. et al. 2015. “Conversion of the LIN-1 ETS Protein of Caenorhabditis Elegans from a SUMOylated Transcriptional Repressor to a Phosphorylated Transcriptional Activator.” Genetics 199(3): 761–75.

Levine, Michael, and Timothy Hoey. 1988. “Homeobox Proteins as Sequence-Specific Transcription Factors.” Cell 55(4): 537–40.

Lewis, E. B. 2004. “A Gene Complex Controlling Segmentation in Drosophila.” In Genes, Development and Cancer: The Life and Work of Edward B. Lewis, ed. Howard D. Lipshitz. Boston, MA: Springer US, 205–17. https://doi.org/10.1007/978-1-4419-8981-9_13 (June 12, 2019).

Li, Victor C., and Marc W. Kirschner. 2014. “Molecular Ties between the Cell Cycle and Differentiation in Embryonic Stem Cells.” Proceedings of the National Academy of Sciences of the United States of America 111(26): 9503–8.

134

Liu, Chia-Feng, William E. Samsa, Guang Zhou, and Véronique Lefebvre. 2017. “Transcriptional Control of Chondrocyte Specification and Differentiation.” Seminars in Cell & Developmental Biology 62: 34–49.

Liu, S., D. A. Cieslinski, A. J. Funke, and H. D. Humes. 1997. “Transforming Growth Factor-Beta 1 Regulates the Expression of Pax-2, a Developmental Control Gene, in Renal Tubule Cells.” Experimental Nephrology 5(4): 295–300.

Lynch, Vincent J., and Günter P. Wagner. 2008. “Resurrecting the Role of Transcription Factor Change in Developmental Evolution.” Evolution 62(9): 2131–54.

Macchia, P. E. et al. 1998. “PAX8 Mutations Associated with Congenital Hypothyroidism Caused by Thyroid Dysgenesis.” Nature Genetics 19(1): 83–86.

Magliano, Marina Pasca di, Roberto Di Lauro, and Mariastella Zannini. 2000. “Pax8 Has a Key Role in Thyroid Cell Differentiation.” Proceedings of the National Academy of Sciences 97(24): 13144–49.

Mahajan, P., P. J. Leavey, and R. L. Galindo. 2015. “PAX Genes in Childhood Oncogenesis: Developmental Biology Gone Awry?” Oncogene 34(21): 2681–89.

Mann, Richard S., Katherine M. Lelli, and Rohit Joshi. 2009. “Chapter 3 Hox Specificity: Unique Roles for Cofactors and Collaborators.” In Current Topics in Developmental Biology, Genes, Academic Press, 63–101. http://www.sciencedirect.com/science/article/pii/S0070215309880034 (June 12, 2019).

Mansouri, Ahmed, Kamal Chowdhury, and Peter Gruss. 1998. “Follicular Cells of the Thyroid Gland Require Pax8 Gene Function.” Nature Genetics 19(1): 87.

Marbach, D. et al. 2010. “Revealing Strengths and Weaknesses of Methods for Gene Network Inference.” Proceedings of the National Academy of Sciences 107(14): 6286–91.

Mazal, Peter R. et al. 2005. “Expression of Aquaporins and PAX-2 Compared to CD10 and Cytokeratin 7 in Renal Neoplasms: A Tissue Microarray Study.” Modern Pathology 18(4): 535.

McGinnis, William et al. 1984. “A Homologous Protein-Coding Sequence in Drosophila Homeotic Genes and Its Conservation in Other Metazoans.” Cell 37(2): 403–8.

McIntyre, Daniel C. et al. 2007. “Hox Patterning of the Vertebrate Rib Cage.” Development 134(16): 2981–89.

135

Mello, Craig C, and Bruce Draper. 1996. “The PIE-1 Protein and Germline Specification in C. Elegans Embryos.” 382: 3.

Memeo, Lorenzo et al. 2007. “Immunohistochemical Analysis for Cytokeratin 7, KIT, and PAX2Value in the Differential Diagnosis of Chromophobe Cell Carcinoma.” American Journal of Clinical Pathology 127(2): 225–29.

Miao, Jiangyong et al. 2007. “HOXB13 Promotes Ovarian Cancer Progression.” Proceedings of the National Academy of Sciences 104(43): 17093–98.

Miccadei, Stefania et al. 2002. “The Synergistic Activity of Thyroid Transcription Factor 1 and Pax 8 Relies on the Promoter/Enhancer Interplay.” Molecular Endocrinology 16(4): 837–46.

Muratovska, Aleksandra et al. 2003. “Paired-Box Genes Are Frequently Expressed in Cancer and Often Required for Cancer Cell Survival.” Oncogene 22(39): 7989– 97.

Murray, John Isaac et al. 2012. “Multidimensional Regulation of Gene Expression in the C. Elegans Embryo.” Genome Research 22(7): 1282–94.

Namm, A., A. Arend, and M. Aunapuu. 2014. “Expression of Pax2 Protein during the Formation of the Central Nervous System in Human Embryos.” Folia Morphologica 73(3): 272–78.

Nathoo, Arif N., Rachael A. Moeller, Beth A. Westlund, and Anne C. Hart. 2001. “Identification of Neuropeptide-like Protein Gene Families in Caenorhabditis Elegans and Other Species.” Proceedings of the National Academy of Sciences 98(24): 14000–5.

Newman, A. P., and P. W. Sternberg. 1996. “Coordinated Morphogenesis of Epithelia during Development of the Caenorhabditis Elegans Uterine-Vulval Connection.” Proceedings of the National Academy of Sciences 93(18): 9329–33.

Newman, A. P., J. G. White, and P. W. Sternberg. 1996. “Morphogenesis of the C. Elegans Hermaphrodite Uterus.” Development 122(11): 3617–26.

Noll, Markus. 1993. “Evolution and Role of Pax Genes.” Current Opinion in Genetics & Development 3(4): 595–605.

Normanno, N. et al. 2003. “Target-Based Agents against ErbB Receptors and Their Ligands: A Novel Approach to Cancer Treatment.” Endocrine-Related Cancer 10(1): 1–21.

136

Oikawa, Tsuneyuki, and Toshiyuki Yamada. 2003. “Molecular Biology of the Ets Family of Transcription Factors.” Gene 303: 11–34.

Olayioye, Monilola A., Richard M. Neve, Heidi A. Lane, and Nancy E. Hynes. 2000. “The ErbB Signaling Network: Receptor Heterodimerization in Development and Cancer.” The EMBO Journal 19(13): 3159–67.

O’Neill, Elizabeth M., Ilaria Rebay, Robert Tjian, and Gerald M. Rubin. 1994. “The Activities of Two Ets-Related Transcription Factors Required for Drosophila Eye Development Are Modulated by the Ras/MAPK Pathway.” Cell 78(1): 137–47.

Onuma, Y. et al. 2002. “Conservation of Pax 6 Function and Upstream Activation by Notch Signaling in Eye Development of Frogs and Flies.” Proceedings of the National Academy of Sciences 99(4): 2020–25.

Pavelic, K., Z. Banjac, J. Pavelic, and S. Spaventi. 1993. “Evidence for a Role of EGF Receptor in the Progression of Human Lung Carcinoma.” Anticancer research 13(4): 1133–37.

Pearson, Joseph C., Derek Lemons, and William McGinnis. 2005. “Modulating Hox Gene Functions during Animal Body Patterning.” Nature Reviews Genetics 6(12): 893.

Perrimon, Norbert, Chrysoula Pitsouli, and Ben-Zion Shilo. 2012. “Signaling Mechanisms Controlling Cell Fate and Embryonic Patterning.” Cold Spring Harbor Perspectives in Biology 4(8): a005975.

Pichaud, Franck, and Claude Desplan. 2002. “Pax Genes and Eye Organogenesis.” Current Opinion in Genetics & Development 12(4): 430–34.

Prior, Harriet, Ali K. Jawad, Lauren MacConnachie, and Asim A. Beg. 2017. “Highly Efficient, Rapid and Co-CRISPR-Independent Genome Editing in Caenorhabditis Elegans.” G3 (Bethesda, Md.) 7(11): 3693–98.

Pufall, Miles A, and Barbara J Graves. 2002. “Ets-1 Flips for New Partner Pax-5.” Structure 10(1): 11–14.

Püschel, A. W., P. Gruss, and M. Westerfield. 1992. “Sequence and Expression Pattern of Pax-6 Are Highly Conserved between Zebrafish and Mice.” Development (Cambridge, England) 114(3): 643–51.

Quinonez, Shane C., and Jeffrey W. Innis. 2014. “Human HOX Gene Disorders.” Molecular Genetics and Metabolism 111(1): 4–15.

137

Rajakumar, Vandana, and Helen M. Chamberlin. 2007. “The Pax2/5/8 Gene Egl-38 Coordinates Organogenesis of the C. Elegans Egg-Laying System.” Developmental Biology 301(1): 240–53.

Rezsohazy, René, Andrew J. Saurin, Corinne Maurel-Zaffran, and Yacine Graba. 2015. “Cellular and Molecular Insights into Hox Protein Action.” Development 142(7): 1212–27.

Robson, Ewan J. D., Shu-Jie He, and Michael R. Eccles. 2006. “A PANorama of PAX Genes in Cancer and Development.” Nature Reviews Cancer 6(1): 52–62.

Rothenpieler, U. W., and G. R. Dressler. 1993. “Pax-2 Is Required for Mesenchyme-to- Epithelium Conversion during Kidney Development.” Development 119(3): 711– 20.

Sahoo, Debashis et al. 2008. “Boolean Implication Networks Derived from Large Scale, Whole Genome Microarray Datasets.” Genome Biology 9(10): R157.

Salomon, David S., Ralf Brandt, Fortunato Ciardiello, and Nicola Normanno. 1995. “Epidermal Growth Factor-Related Peptides and Their Receptors in Human Malignancies.” Critical Reviews in Oncology/Hematology 19(3): 183–232.

Sander, M et al. 1997. “Genetic Analysis Reveals That PAX6 Is Required for Normal Transcription of Pancreatic Hormone Genes and Islet Development.” Genes & Development 11(13): 1662–73.

Sanyanusin, Phaikasame et al. 1995. “Mutation of the PAX2 Gene in a Family with Optic Nerve Colobomas, Renal Anomalies and Vesicoureteral Reflux.” Nature Genetics 9(4): 358–64.

Schmidt-Ullrich, Rupert K., Kristoffer C. Valerie, Winnie Chan, and Dena McWilliams. 1994. “Altered Expression of Epidermal Growth Factor Receptor and Estrogen Receptor in MCF-7 Cells after Single and Repeated Radiation Exposures.” International Journal of Radiation Oncology*Biology*Physics 29(4): 813–19.

Schneuwly, Stephan, Roman Klemenz, and Walter J. Gehring. 1987. “Redesigning the Body Plan of Drosophila by Ectopic Expression of the Homoeotic Gene Antennapedia.” Nature 325(6107): 816.

Schulte, Dorothea. 2014. “Meis: New Friends of Pax.” Neurogenesis 1(1). https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4973581/ (May 26, 2019).

Scott, M P, and A J Weiner. 1984. “Structural Relationships among Genes That Control Development: Sequence Homology between the Antennapedia, Ultrabithorax,

138

and Fushi Tarazu Loci of Drosophila.” Proceedings of the National Academy of Sciences of the United States of America 81(13): 4115–19.

Selinger, Douglas W., Matthew A. Wright, and George M. Church. 2003. “On the Complete Determination of Biological Systems.” Trends in Biotechnology 21(6): 251–54.

Seydoux, Geraldine et al. 1996. “Repression of Gene Expression in the Embryonic Germ Lineage of C. Elegans.” Nature 382(6593): 713–16.

Shah, Nilay, and Saraswati Sukumar. 2010. “The Hox Genes and Their Roles in Oncogenesis.” Nature Reviews Cancer 10(5): 361–71.

Silverman, Gary A. et al. 2009. “Modeling Molecular and Cellular Aspects of Human Disease Using the Nematode Caenorhabditis Elegans.” Pediatric research 65(1): 10–18.

Singh, Harinder, Aly A. Khan, and Aaron R. Dinner. 2014. “Gene Regulatory Networks in the Immune System.” Trends in Immunology 35(5): 211–18.

Singson, Andrew, Katherine L. Hill, and Steven W. L’Hernault. 1999. “Sperm Competition in the Absence of Fertilization in Caenorhabditis Elegans.” Genetics 152(1): 201–8.

Sorensen, Poul H.B. et al. 2002. “PAX3-FKHR and PAX7-FKHR Gene Fusions Are Prognostic Indicators in Alveolar Rhabdomyosarcoma: A Report From the Children’s Oncology Group.” Journal of Clinical Oncology 20(11): 2672–79.

Stiernagle, Theresa. 2006. “Maintenance of C. Elegans.” WormBook. http://www.wormbook.org/chapters/www_strainmaintain/strainmaintain.html (April 5, 2019).

Stigler, Brandilyn, and Helen M Chamberlin. 2012. “A Regulatory Network Modeled from Wild-Type Gene Expression Data Guides Functional Predictions in Caenorhabditis Elegans Development.” BMC Systems Biology 6(1): 77.

St-Onge, Luc et al. 1997. “Pax6 Is Required for Differentiation of Glucagon-Producing α- Cells in Mouse Pancreas.” Nature 387(6631): 406.

Sulston, J.E., E. Schierenberg, J.G. White, and J.N. Thomson. 1983. “The Embryonic Cell Lineage of the Nematode Caenorhabditis Elegans.” Developmental Biology 100(1): 64–119.

139

Székely, Gábor J., Maria L. Rizzo, and Nail K. Bakirov. 2007. “Measuring and Testing Dependence by Correlation of Distances.” The Annals of Statistics 35(6): 2769– 94.

Tabara, Hiroaki et al. 1999. “The Rde-1 Gene, RNA Interference, and Transposon Silencing in C. Elegans.” Cell 99(2): 123–32.

Taniguchi, Yasushi. 2014. “Hox Transcription Factors: Modulators of Cell-Cell and Cell- Extracellular Matrix Adhesion.” BioMed Research International. https://www.hindawi.com/journals/bmri/2014/591374/ (June 12, 2019).

Taurog, Alvin, Martha L. Dorris, and Daniel R. Doerge. 1996. “Mechanism of Simultaneous Iodination and Coupling Catalyzed by Thyroid Peroxidase.” Archives of Biochemistry and Biophysics 330(1): 24–32.

Terzić, J., C. Muller, S. Gajović, and M. Saraga-Babić. 1998. “Expression of PAX2 Gene during Human Development.” The International Journal of Developmental Biology 42(5): 701–7.

Thesleff, Irma, Anne Vaahtokari, and Anna Maija Partanen. 1995. “Regulation of Organogenesis. Common Molecular Mechanisms Regulating the Development of Teeth and Other Organs.” The International journal of developmental biology 39(1): 35–50.

Tiffin, N., R. D. Williams, J. Shipley, and K. Pritchard-Jones. 2003. “PAX7 Expression in Embryonal Rhabdomyosarcoma Suggests an Origin in Muscle Satellite Cells.” British Journal of Cancer 89(2): 327.

Tischfield, Max A. et al. 2005. “Homozygous HOXA1 Mutations Disrupt Human Brainstem, Inner Ear, Cardiovascular and Cognitive Development.” Nature Genetics 37(10): 1035.

Tong, Guo-Xia et al. 2006. “PAX2: A Reliable Marker for Nephrogenic Adenoma.” Modern Pathology 19(3): 356.

Torres, M., E. Gomez-Pardo, G. R. Dressler, and P. Gruss. 1995. “Pax-2 Controls Multiple Steps of Urogenital Development.” Development 121(12): 4057–65.

Torres, M., E. Gomez-Pardo, and P. Gruss. 1996. “Pax2 Contributes to Inner Ear Patterning and Optic Nerve Trajectory.” Development 122(11): 3381–91.

Villani, M, A Barbieri, and R Serra. 2011. “A Dynamical Model of Genetic Networks for Cell Differentiation.” PLOS ONE 6(3). https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0017703 (June 10, 2019).

140

Wachtel, Marco, and Beat W. Schäfer. 2015. “Unpeaceful Roles of Mutant PAX Proteins in Cancer.” Seminars in Cell & Developmental Biology 44: 126–34.

Wang, Jieyu et al. 2018. “Paired Box 2 Promotes Progression of Endometrial Cancer via Regulating Cell Cycle Pathway.” Journal of Cancer 9(20): 3743–54.

Wang, Qiuyu et al. 2008. “Pax Genes in Embryogenesis and Oncogenesis.” Journal of Cellular and Molecular Medicine 12(6a): 2281–94.

Ward, Samuel, and John S. Carrel. 1979. “Fertilization and Sperm Competition in the NematodeCaenorhabditis Elegans.” Developmental Biology 73(2): 304–21.

Wellik, Deneen M. 2009. “Chapter 9 Hox Genes and Vertebrate Axial Pattern.” In Current Topics in Developmental Biology, Elsevier, 257–78. https://linkinghub.elsevier.com/retrieve/pii/S0070215309880095 (June 12, 2019).

Wellik, Deneen M., and Mario R. Capecchi. 2003. “Hox10 and Hox11 Genes Are Required to Globally Pattern the Mammalian Skeleton.” Science 301(5631): 363– 67.

Winyard, P. J. et al. 1996. “The PAX2 Tanscription Factor Is Expressed in Cystic and Hyperproliferative Dysplastic Epithelia in Human Kidney Malformations.” The Journal of Clinical Investigation 98(2): 451–59.

Wood, William B., and Lois G. Edgar. 1994. “Patterning in the C. Elegans Embryo.” Trends in Genetics 10(2): 49–54.

Wu, Xinyan et al. 2006. “HOXB7, a Homeodomain Protein, Is Overexpressed in Breast Cancer and Confers Epithelial-Mesenchymal Transition.” Cancer Research 66(19): 9527–34.

Xu, H. Eric et al. 1999. “Crystal Structure of the Human Pax6 Paired Domain–DNA Complex Reveals Specific Roles for the Linker Region and Carboxy-Terminal Subdomain in DNA Binding.” Genes & Development 13(10): 1263–75.

Xu, W. et al. 1995. “Crystal Structure of a Paired Domain-DNA Complex at 2.5 A Resolution Reveals Structural Basis for Pax Developmental Mutations.” Cell 80(4): 639–50.

Yanai, Itai et al. 2008. “Pairing of Competitive and Topologically Distinct Regulatory Modules Enhances Patterned Gene Expression.” Molecular Systems Biology 4. http://msb.embopress.org/cgi/doi/10.1038/msb.2008.6 (June 1, 2019).

Yarden, Yosef, and Mark X. Sliwkowski. 2001. “Untangling the ErbB Signalling Network.” Nature Reviews Molecular Cell Biology 2(2): 127.

141

Yates, A. et al. 2014. “Visualizing Multidimensional Data with Glyph SPLOMs: Visualizing Multidimensional Data with Glyph SPLOMs.” Computer Graphics Forum 33(3): 301–10.

Zakany, Jozsef, and Denis Duboule. 2007. “The Role of Hox Genes during Vertebrate Limb Development.” Current Opinion in Genetics & Development 17(4): 359–66.

Zannoni, Sonia, Steven W L’Hernault, and Andrew W Singson. 2003. “Dynamic Localization of SPE-9 in Sperm: A Protein Required for Sperm-Oocyte Interactions in Caenorhabditis Elegans.” BMC Developmental Biology 3: 10.

Zhai, Yali et al. 2007. “Gene Expression Analysis of Preinvasive and Invasive Cervical Squamous Cell Carcinomas Identifies HOXC10 as a Key Mediator of Invasion.” Cancer Research 67(21): 10163–72.

Zhang, Guojuan et al. 2005. “Alteration of the DNA Binding Domain Disrupts Distinct Functions of the C. Elegans Pax Protein EGL-38.” Mechanisms of Development 122(7–8): 887–99.

Zhang, Li-Ping et al. 2011. “RNA Interference of Pax2 Inhibits Growth of Transplanted Human Endometrial Cancer Cells in Nude Mice.” Chinese Journal of Cancer 30(6): 400–406.

Zhu, Hui-Fang, and Yan Li. 2018. “Small-Molecule Targets in Tumor Immunotherapy.” Natural Products and Bioprospecting 8(4): 297–301.

142