EXAMINING the POST-TRANSCRIPTIONAL REGULATION of LUNATIC FRINGE (Lfng) in the MOUSE SEGMENTATION CLOCK

Home , 6B (Fringe), HES7 gene, LFNG, TBX6

EXAMINING THE POST-TRANSCRIPTIONAL REGULATION OF LUNATIC FRINGE (Lfng) IN THE MOUSE SEGMENTATION CLOCK

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

Kanu Wahi

Graduate Program in Molecular Genetics

The Ohio State University

2016

Dissertation Committee:

Dr. Susan Cole, Advisor

Dr. Keith Slotkin

Dr. Robin Wharton

Dr. Dawn Chandler

Kanu Wahi

2016

Abstract

Somitogenesis is a developmental process in vertebrates involving periodic formation of somites that bud from an unsegmented region known as the pre-somitic mesoderm (PSM) and give rise to the axial skeleton and skeletal muscle in the developed organism. The process of somitogenesis is regulated by a "segmentation clock" that times somite formation and is evolutionarily conserved among vertebrates.

Genes, such as Lunatic fringe (Lfng), are required for normal clock function in chickens, mice and humans and exhibit rapid cyclic expression in the PSM with a period that matches the rate of somite formation. To maintain rapid oscillations, especially in the posterior PSM, it is hypothesized that the Lfng transcript should be promptly degraded to ensure its clearance of from cells before the next round of oscillation begins. The

3’UTR of Lfng contains number of conserved sequences that could regulate Lfng expression at the post transcriptional level. In this study we explore the post- transcriptional regulation of the Lfng transcript by the Lfng 3’UTR in the context of the segmentation clock.

The Lfng 3'UTR has binding sites for microRNAs (miRNAs), such as miR-125a and miR-200b that are known to affect transcript stability via the 3’UTR. We tested the effect of mutating the mir-125a binding sites in the mouse Lfng 3’UTR on mRNA stability

ii of Venus reporter constructs in transgenic mice and observed constitutive Venus expression in the posterior PSM, as opposed to the oscillatory Venus expression when the miR-125a binding sites are not mutated. This implies that the miR-125a binding sites in the Lfng 3'UTR influence RNA turnover in the posterior region of the mouse PSM.

Interestingly, when we generated mutant mouse lines that do not express miR-125a we found no effect on somitogenesis and skeletal formation. This indicates that either a compensatory mechanism may be in effect or that miR-125a does not play a role in the mouse segmentation clock. We also examined if another highly conserved region of the

Lfng 3'UTR, that is 120 bp in length and has a miR-200 binding site within it, can sufficiently destabilize the transcript. We examined this in two contexts. We found that blocking the miR-200b binding site in the Lfng 3’UTR in chicken embryos has no effect on somitogenesis. Similarly, in mouse myoblast cells the 120 bp sequence of the mouse

Lfng 3'UTR is not sufficient to destabilize the transcript. Our results indicate that the 120 bp sequence and the miR-200b binding site within it, are not sufficient to destabilize the transcript and it is likely that this regulation operates in a tissue specific manner.

Future work in the lab will examine the function of the endogenous miR-125a binding site in the Lfng 3'UTR on mouse segmentation and skeletal formation.

Understanding the post transcriptional regulation of clock genes like Lfng would shed light on mechanisms that maintain oscillatory gene expression in the PSM and ensure timely somite formation.

iii

Dedication

This document is dedicated to my husband, Dhrupad Siddhanta, for his endless support

and for being my cheerleader for life.

Acknowledgments

I would like to sincerely thank my advisor, Dr. Susan Cole, without whose constant guidance, motivation and support I would not have been able to make it this far. I would like to acknowledge her patience in bearing with my shortcomings and helping me overcome them. Her nurturing mentorship has made my graduate school experience truly enriching and enjoyable.

I would like to thank my lab members: Dr. Maurisa Riley, who provided the stepping stone for this project; Sophia Friesen, an extremely dedicated and hardworking undergraduate student, who has contributed immensely to the analysis of the miR-125a mutant mice; Dr. Dustin Williams, Skye Bochter and Kara Braunreiter for always supporting me and for making lab feel like a second home; Undergraduates, Madeline

Parker and Ben Schott for their assistance with lab chores.

I am grateful to my committee members: Dr. Dawn Chandler, Dr.Keith Slotkin and

Dr. Robin Wharton for sharing their expertise and for their valuable advice during committee meetings that has greatly influenced my project.

I would like to thank Dr. Jared Talbot and Dr. Sharon Amacher for sharing their expertise and equipment for the High Resolution Melt Analysis (HRMA). Dr. Vincenzo

Coppola, Director of the Mouse facility at OSU for helping us generate the Venus

v transgenic and miR-125a mice using the CRISPR/Cas9 system. I immensely thank the all members of the Dr. Harold Fisk’s for always being so helpful, especially Dr. Dwitiya

Sawant for guiding me with the use of the fluorescent microscope.

I would like to thank the Pelotonia Foundation for their financial support from

August 2014 to 2016 and the Alumni Grants Graduate Research Scholarships (AGGRS) association at OSU for funding a part of my thesis project.

Lastly, I would like to thank my family: my parents for their unconditional love and support that has helped me reach my goals, my brother and sister-in-law for their encouragement throughout graduate school and my husband for always being by my side and motivating me throughout my graduate career.

Vita

2003 ...... National Public School, INDIA

2006 ...... B.S. Biotechnology, Kasturba Medical

College, INDIA

2008 ...... M.S. Medical Biotechnology, Manipal

University, INDIA

2010 to present ...... Graduate Fellow, Department of Molecular

Genetics, The Ohio State University

Publications

1. Wahi K, Bochter MS, Cole SE. The many roles of Notch signaling during vertebrate somitogenesis. Semin Cell Dev Biol. 2016 Jan; 49:68-75. Epub 2014 Dec 4.

2. Korlimarla A, Prabhu JS, Anupama CE, Remacle J, Wahi K, Sridhar TS. Separate

Quality-Control Measures Are Necessary for Estimation of RNA and Methylated DNA from Formalin-Fixed, Paraffin-Embedded Specimens by Quantitative PCR. J Mol

Diagn. 2014 Mar; 16(2):253-60.

vii

3. Riley MF, Bochter MS, Wahi K, Nuovo GJ, & Cole SE, mir-125a-5p-mediated

Regulation of Lfng is Essential for the Avian Segmentation Clock. Dev Cell. 2013 Mar 11;

24(5):554-61.

4. Prabhu JS, Wahi K, Korlimarla A, Correa M, Manjunath S, Raman N, Srinath BS,

Sridhar TS. The epigenetic silencing of the estrogen receptor (ER) by hypermethylation of the ESR1 promoter is seen predominantly in triple-negative breast cancers in Indian women. J Tum Bio. Apr 2012; 33(2).

Fields of Study

Major Field: Molecular Genetics

viii

Table of Contents

Abstract ...... ii

Dedication ...... iv

Acknowledgments...... v

Vita ...... vii

List of Tables ...... xiii

List of Figures ...... xiv

Chapter 1: Introduction ...... 1

1.1 Segmentation in vertebrates…………………………………………………………………………………. 2

1.2 The clock and wavefront model can explain periodic somite formation …………………6

1.3 The segmentation clock operates in the vertebrate PSM ……………………………………….7

1.4 Notch signaling pathway establishes cell-to-cell communication ………………………….10

1.5 Notch pathway is required for normal segmentation clock function …………………….14

1.5.1 Role of Notch pathway in the zebrafish segmentation clock……………………….. 17

1.5.2 Role of the Notch pathway in the chicken and mouse segmentation clock…..19

1.6 Setting up the wavefront in the vertebrate PSM………………………………………………….. 25

1.7 Somite patterning in the anterior PSM is influenced by the Notch pathway…………..29

1.8 Fine tuning oscillatory Notch signaling in the segmentation clock………………………….33

Chapter 2: Regulation of Lunatic fringe by mir-125a in the mouse segmentation clock

2.1 Introduction………………………………………………………………………………………………………….. 37

2.2.1 Segmentation in vertebrates……………………………………………………………………….. 38

2.2.2 Oscillatory Lunatic fringe expression is essential for the mouse segmentation clock………………………………………………………………………………………………………………39

2.2.3 Post-transcriptional regulation of Lfng by the 3’ UTR…………………………………… 40

2.2 Materials and Methods

2.2.1 Construction of Venus reporter plasmids…………………………………………………….. 45

2.2.2 Transient transfections and generation of stable cell lines………………………..…..47

2.2.3 RT-PCR and Quantitative RT-PCR………………………………………………………………….. 49

2.2.4 Venus transgenic mice…………………………………………………………………………………. 49

2.2.5 Whole mount in situ hybridization……………………………………………………………….. 50

2.2.6 Expression profile data analysis……………………………………………………………………. 51

2.2.7 CRISPR/Cas9 plasmid and guide RNAs to target the miR-125a locus……………..52

2.2.8 Generation of miR-125a mutant mice…………………………………………………………. 53

2.2.9 High Resolution Melt Analysis (HRMA)………………………………………………………… 53

2.2.10 Sequencing and genotyping of the miR-125a mutants………………………………. 54

2.2.11 qRT-PCR for miR-125a expression in mutant mice…………………………………….. 54

2.2.12 Testes histology and analysis of sperm morphology………………………………….. 55

2.3 Results

2.3.1 miR-125a destabilizes Venus transcript with the Wt 3’UTR…………………………. 55

2.3.2 The LvLMut mouse lines exhibit stabilized Venus expression in the caudal PSM……………………………………………………………………………………………………………. 58

2.3.3 LvLMut embryos exhibit oscillatory expression in the anterior PSM………………62

2.3.4 Creating miR-125a mutant mice using the CRISPR/Cas9 system…………………… 64

2.3.5 miR-125a-5p expression is affected in all three miR-125a mutants………………. 68

2.3.6 Lfng and Uncx expression are unaffected in miR-125a mutants…………………….72

2.3.7 miR-125a-5p∆11/∆11 male mice are infertile…………………………………………………… 76

2.4 Discussion……………………………………………………………………………………………………………… 80

Chapter 3: Lfng 3’UTR sequences have complex effects on transcript stability

3.1 Introduction……………………………………………………………………………………………………………87

3.1.1 Oscillatory Lfng expression is required for the chicken and mouse segmentation clock………………………………………………………………………………………..88

3.1.2 The Lfng 3'UTR affects transcript stability………………………………………………………89

3.2 Materials and Methods

3.2.1 miRNA RT-PCR……………………………………………………………………………………………….91

3.2.2 Luciferase assays……………………………………………………………………………………………91

3.2.3 In ovo electroporation of chicken embryos……………………………………………………93

3.2.4 In situ hybridization……………………………………………………………………………………….93

3.2.5 Reporter constructs used to make stable cell lines………………………………………..94

3.2.6 Transcript half-life analysis…………………………………………………………………………… 94

3.3 Results

3.3.1 The expression of some members of the miR-200b/c/429 family of miRNAs are enriched in the mouse PSM compared to the somites……………………………………95

3.3.2 The chick and mouse Lfng 3'UTRs are targets of miR-200b…………………………….96

3.3.3 miR-200b/c/429 may play a role in chicken somitogenesis………………………….100

3.3.4 Blocking the interaction of miR-200 with Lfng 3'UTR does not perturb somitogenesis in chicken embryos……………………………………………………………….104

3.3.5 Blocking the interaction of miR-200 with Lfng 3'UTR does not dramatically affect transcript stability in the chick PSM……………………………………………………108

3.3.6 The conserved 120 bp central region of the Lfng 3'UTR is not sufficient to destabilize exogenous transcripts in C2C12 cells………………………………………….110

3.4 Discussion…………………………………………………………………………………………………………….112

Chapter 4: Conclusions………………………………………………………………………………………………. 117

4.1 miR-125a binding sites in the Lfng 3'UTR regulate transcript turnover in the tailbud region of the mouse PSM……………………………………………………………………………………..118

4.2 Loss of miR-125a in mice has no impact on Lfng oscillations and the segmentation clock……………………………………………………………………………………………………………………..119

4.3 Complete loss of miR-125a results in male infertility in mice and could be due to defective sperm morphology………………………………………………………………………………..120

4.4 The conserved 120 bp central region of the Lfng 3'UTR is not sufficient……………….121

4.5 Concluding remarks.……………………………………………………………………………………………..122

References…………………………………………………………………………………………………………………123

Appendix A: Promoter analysis of miR-125a family members……………………………………138 Appendix B: Supplemental figures…………………………………………………………………………….154

xii

List of Tables

Table 1.1: Somitogenesis in different vertebrate organisms…………………………………………..8

Table 1.2: Expression and null phenotypes of Notch pathway members in the context of the mouse segmentation clock…………………………………………………………………………………….22

Table 2. 1 Lfng 3’UTR primers……………………………………………………………………………………….47

Table 2.2 Oligos pairs used to target the miR-125a locus in the mouse genome…………..52

Table 3.1: Sequence of anti-miRs and Seedblockers used to block activity of the miR- 200b family in the chick PSM………………………………………………………………………………………..92

Table A.1 Primers for 5'RACE PCR to identify the 5’ end of the primary transcript……..141

Table A.2 Primers for 3'RACE PCR to identify the 3’ end of the primary transcript……..142

Table A.3 Primers to amplify the potential promoter region for miR-99b and 125b1….144

xiii

List of Figures

Figure 1.1: Somite formation and the clock and wavefront model…………………………………5

Figure 1.2: A simplified schematic of the canonical Notch signaling pathway……………….13

Figure 1.3: Clock-linked oscillations of hairy/enhancer of split related genes are tightly regulated during somitogenesis……………………………………………………………………………………16

Figure 1.4: Role of Notch signaling during mouse somitogenesis………………………………….32

Figure 2.1: Regulation of Lfng in the segmentation clock……………………………………………..43

Figure 2.2: Destabilized Venus reporter constructs………………………………………………………45

Figure 2.3: miR-125a destabilizes transcripts via the Lfng 3’UTR………………………………….57

Figure 2.4: Venus expression in LvLMut lines does not match endogenous Lfng expression or Venus expression in the control LvLWt lines………………………………………….60

Figure 2.5: Venus expression in the mouse PSM of LvLMut embryos does not overlap with endogenous Lfng …………………………………………………………………………………………………61

Figure 2.6: Oscillatory expression profile of transcript expression over time along the A-P axis of the mouse PSM……………………………………………………………………………………………….. 63

Figure 2.7: The CRISPR-Cas9 system utilized to make knockout mir-125a mouse……….. 65

Figure 2.8: Identification of four miR-125a mutants by High Resolution Melt Analysis (HRMA)………………………………………………………………………………………………………………………..66

Figure 2.9: Genotypes of mutant miR-125a founders…………………………………………………..67

Figure 2.10: Structures of miR-125a precursors in Del11 and Del3T mutants……………….70

xiv

Figure 2.11: miR-125a expression is lost in Delta11 (n=3) and Delta3T (n=3) mutants and reduced in insertion (n=3) mutants as compared to the Wt controls (n=3)…………………. 71

Figure 2.12: Lfng expression is unaffected by miR-125a mutants…………………………………73

Figure 2.13: Somite number and posterior somite boundaries (Uncx) are unaffected in miR-125a mutants………………………………………………………………………………………………………..74

Figure 2.14: Skeletal development is not impaired by the loss of miR-125a …………………75

Figure 2.15: The testes of infertile Del11 mutant mice do not exhibit any dramatic morphological defects………………………………………………………………………………………………… 78

Figure 2.16: Abnormal sperm morphology in Del11 homozygous mutants…………………. 79

Figure 2.17: mir-125a binding sites in the Lfng 3’UTR may affect the period of Lfng oscillations in the mouse PSM……………………………………………………………………………………..81

Figure 3.1: A conserved binding site in the Lfng 3'UTR can be directly targeted by members of the miR-200b family…………………………………………………………………………………98

Figure 3.2 miR-200b family may play role during chicken somitogenesis……………………102

Figure 3.3: Interactions between Lfng and miR-125a-5p but not miR- 200b are essential for proper vertebrate segmentation…………………………………………………………………………..106

Figure 3.4: Inhibiting interactions between Lfng and miR-200b does not affect Lfng oscillations………………………………………………………………………………………………………………….109

Figure 3.5: The conserved region containing a miR-200b/c binding site is not sufficient to destabilize an exogenous transcript…………………………………………………………………………...111

Figure A.1: miR-125a, miR-125b1 and miR-125b2 clusters in the mouse genome………139

Figure A.2: Rapid Amplification of cDNA ends (RACE)…………………………………………………143

Figure A.3: Regulation of miR-125a cluster…………………………………………………………………146

Figure A.4: Identification of miR-99b promoter………………………………………………………….148

Figure A.5: Identification of miR-125b1 primary transcript…………………………………………150 Figure A.6: Identification of miR-125b1 promoter………………………………………………………151 Figure B.1: GFP transgenic mice………………………………………………………………………………….155

CHAPTER 1

INTRODUCTION (Portions of this chapter have been published as a review, Wahi et al, 2016, and are being reproduced here with permission)

Pattern formation is observed in all multicellular organisms. From the aesthetically patterned ossicles on the aboral surface of a starfish or the arrangement of photoreceptors in the compound eye of a fruit fly to the intricately patterned structure of the vertebral column in humans; these reiterated patterns serve important functions in each of these organisms. In 1952, Alan Turing was among the first to theorize on how patterns form in a biological system, stating that, “Most of the organism, most of the time, is developing from one pattern into another, rather than from homogeneity into a pattern”(Turing 1990). This is the hallmark of how patterns form and is known as morphogenesis in biological systems(Koch & Meinhardt 1994; Reinitz 2012).

Morphogenesis occurs during embryonic development and takes the embryo from an undifferentiated ball of cells to an embryo that has a body axis and where cells are patterned and organized into distinct tissues and organs(Lecuit & Lenne 2007).

Segmentation is a morphogenetic process that sets up the body axis in all vertebrate and some invertebrate embryos and is an ideal model to study mechanisms that transform cells from an undifferentiated state to discernible structures seen in the developed organism(Saga & Takeda 2001). 1

1.1 Segmentation in vertebrates

Segmentation is observed in much of the animal kingdom particularly in annelids, arthropods, and vertebrates where the body plan is patterned into metameric segments along the anterior-posterior axis(Tautz 2004). Repetitive patterns are set up during segmentation by complex interactions between genes that are regulated temporally and spatially during the embryonic stage of development (Tautz 2004). These three groups are the most successful and diverse amongst the animal kingdom and the presence of a segmented body plan is thought to be one of the reasons for that. For instance, compartmentalization of the body by having the oral cavity in a different region as the excretory system is thought to make feeding more efficient (Roberts et al. 2000; Tautz

2004).

There is considerable dispute over the evolutionary conservation of segmentation across the phyla mentioned above(Hannibal & Patel 2013; Graham et al. 2014). This is probably due to differences in segmentation seen in these organisms such as the fact that overt segmentation is seen in most adult forms of annelids while arthropods and vertebrates are only segmented in the embryonic stage and lose the obvious segmented appearance at birth. The mode of segment formation is not completely conserved among these organisms either (Tautz 2004). The fruit fly Drosophila melanogaster along with other long germ band insects show a top-down model of segmentation where segments are formed simultaneously (Johnston et al. 1992; Pankratz, M. J. and Jackle

1993). Alternatively, all vertebrates and some arthropods, such as the spider Cupiennius 2 salei, lay out their body segments sequentially and utilize the canonical Notch signaling pathway in the process of segmentation (Tautz 2004; Stollewerk et al. 2003;

Schoppmeier & Damen 2005; Peel et al. 2005). Hence, it is debatable whether there was a single common ancestor of annelids, arthropods and vertebrates that had a segmented body plan or if a specific mode of segmentation was adopted by some organisms but lost in others and then evolved and re-adopted multiple times during the evolutionary timeline.

Within the vertebrate phylum the mode of segmentation appears to be highly conserved and gives rise to the axial skeleton and musculature. The axial skeleton and musculature consists of ribs, the vertebral column, skeletal muscles and dermis of the back (Christ et al. 2007). Precursors for the above structures are set up shortly after gastrulation is initiated by a process called somitogenesis or simply referred to as segmentation. Somitogenesis entails the periodic formation of a pair of somites from an unsegmented region of the paraxial mesoderm that is positioned at the posterior end of the embryo and referred to as the pre-somitic mesoderm (PSM). Somites were first observed in chicken embryos in 1672 by Marcello Malphigi and documented in De Ovo

Incubato following which it was rapidly appreciated that somite formation is seen in all vertebrate embryos. Hence, somite formation is critical for normal vertebral segmentation in all vertebrates (Wahi et al. 2016).

The PSM is formed initially by cells that ingress into the primitive streak during gastrulation and later by the stem cell population in the tailbud(Tajbakhsh & Spörle

1998). Cells in the unsegmented PSM have a mesenchymal identity. Like clockwork, blocks of cells at the anterior end of the PSM undergo a mesenchymal-to-epithelial transition to form a boundary between the new pair of epithelial somites that flank the neural tube and the anterior PSM as shown in Figure 1.1A. Each somite has mesenchymal cells in the center and a layer of epithelial cells on the exterior surface that forms the intersomitic boundary (Takahashi & Sato 2008). These transient somite structures will give rise to the sclerotome, myotome and dermatome which contribute to the vertebrae, striated muscles and dermis, respectively (Yusuf & Brand-Saberi 2006;

Christ et al. 2007). As cells are being lost to somites in the anterior PSM, cells in the posterior PSM are being pushed anteriorly by the ingression of new cells. Once these cells reach the boundary between the last formed somite and the anterior PSM they will become part of a new somite(Tajbakhsh & Spörle 1998). The perfectly timed formation of every new pair of somites is controlled by a genetic clock that operates in cells in the posterior PSM and is discussed in the following Section 1.2.

Disruption of somitogenesis by genetic mutations or environmental insults results in congenital abnormalities in humans such as spondylocostal dysostoses (SCD), scoliosis, spinal kyphosis or lordosis (Turnpenny et al. 2007; Giampietro et al. 2009; Eckalbar et al.

2012). SCD is the most severe of the above abnormalities where affected individuals have rib fusions and smoothened vertebrae all along the vertebral column resulting in severe back pain and disability. Thus, studying the regulation of somitogenesis can shed light on the genetic etiology of these malformations.

Figure 1.1: Somite formation and the clock and wavefront model. A. Mesenchymal to

Epithelial transition (MET) in the Pre-somitic mesoderm (PSM) forms the intersomite boundary gives rise to somites B. The clock and wavefront model. The timing of somite formation is controlled by the segmentation clock that operates in the posterior PSM, the patterning of the presumptive somite (S-1, S0) occurs in the anterior PSM. The positioning of the somite is determined by the wavefront set by the Wnt, FGF and

Retinoic acid. C. Phases of clock gene expression. Cells in the PSM (continued) 5

(Figure1.1 continued) experience a wave of clock gene oscillation and at the end of the oscillation a new pair of somites bud off from the anterior PSM (single cell represented in red).

1.2 The clock and wavefront model can explain periodic somite formation

Periodic formation of somites is regulated both spatially and temporally. In 1976,

Cooke and Zeeman proposed the clock and wavefront model to explain the mechanism behind periodic somite formation (Cooke & Zeeman 1976). While other mathematical models to describe periodic somite formation have also been proposed such as prepatterning of somites (Meinhardt 1986; Schiffmann 2004) and a cell cycle model

(Stern 1988; Primmett et al. 1988; Collier et al. 2000; McInerney et al. 2004), the clock and wavefront model has the most scientific support (Cooke & Zeeman 1976). Cooke and Zeeman proposed that pre-somitic cells in the unsegmented region of the PSM must possess oscillators (clock) which interact with a gradient (wavefront) that results in population of cells rhythmically budding off to form pairs of somites (Figure 1.1B). The clock controls the timing of somite formation and wavefront determines the position of the somite boundary. They also predicted that the oscillatory period of the clock would match the rate of somite formation and this rate would be species-specific. In 1997, the clock and wavefront model was eventually supported by experimental evidence, when

6 the first clock-related gene, c-hairy1, with oscillatory mRNA and protein expression in the chick PSM was discovered and it was also found that the oscillatory period of c- hairy1 expression matched the rate of somite formation in chicken embryos (Palmeirim et al. 1997; Cooke et al. 1998).

Before discussing the properties of the clock and wavefront model, it is important to understand the molecular basis of oscillatory expression of clock genes. A transcript or protein often displays oscillatory expression when it can directly or indirectly repress its own expression to create a negative feedback loop. The expression or repression of a gene would depend on the amount of repressor protein available in the cell at a given point of time. Hence, the competitive binding between activators and repressors at the promoter region will, respectively, result in either an ON or OFF expression of an oscillatory gene.

1.3 The segmentation clock operates in the vertebrate PSM

The segmentation clock is controlled by genes, which we refer to as clock genes, whose expression oscillates most discernibly in cells in the posterior PSM. The time taken by these genes to complete one round of oscillatory expression is defined as the period of the clock which matches the rate of somite formation (reviewed in Wahi et al.

2016). At the end of one period, cells in the anterior PSM will become part of a somite.

The number and rate of somite of formation differs between vertebrates as shown in

Table 1.1, but the basic mechanism is the same in all these species (except for zebrafish, in chickens, mice and humans the first few somites give rise to cranial structures are not 7 derived from the mesoderm but from neural crest cells)(Richardson et al. 1998; Stickney et al. 2000; Morin-Kensicki et al. 2002). Most of the oscillatory genes in the PSM are members of Notch signaling pathway but in some species such as mice they include members of the Wnt and FGF pathway as well (reviewed in Gibb et al. 2010).

*Determined based on conservation with mouse genes but not experimentally

Table 2.1: Somitogenesis in different vertebrate organisms(Richardson et al. 1998; Stickney et al. 2000; Morin-Kensicki et al. 2002).

Based on the pace of clock gene oscillations, the PSM can be divided into two functional regions (Figure 1.1B). In cells in the posterior PSM (Region I), expression of clock genes oscillates more rapidly as compared to cells in the anterior PSM (Region II), where the clock gene expression oscillates at a slower pace (Palmeirim et al. 1997; Gibb et al. 2010; Niwa et al. 2007). By mathematically modeling of the segmentation clock along the mouse PSM it is predicted that clock oscillations slowdown from a 1-segment periodicity in the posterior PSM to a 1.5-segment periodicity in the anterior PSM (Niwa et al. 2011) while in zebrafish, a 2-segment periodicity has been reported in the anterior

PSM (Shih et al. 2015). Another distinction between these two regions is that cells in

Region I remain in an undifferentiated state, cells in Region II are pre-patterned into 8 somite sized cohorts with rostral and caudal compartments, prior to forming a pair of somites(Saga 2012).

The collective network of clock genes in a cell constitutes an oscillator that functions cell autonomously. This was demonstrated previously by separating PSM cells from each other and it was found that individual cells continued to show oscillatory expression of clock genes but when oscillations were compared between cells, they were out of phase from each other (Masamizu et al. 2006). Similarly, when the PSM was divided into small segments, while oscillations between segments were synchronized initially, they went out of sync over time (Maroto et al. 2005). Oscillations were also seen in the PSM even in the absence of the caudal most region demonstrating that cell displacement from the caudal region was not responsible for phases of expression seen in the PSM and supporting the idea of an internal clock in PSM cells (Palmeirim et al. 1997; McGrew et al. 1998).

The question arises, though, of how the intracellular oscillations are coordinated and synchronized between cells from the posterior to the anterior PSM such that cells within the same region of the PSM appear to be in the same phase of the oscillation and collectively establish the period of the segmentation clock. Experimental evidence backed by mathematical modeling suggest that at the start of somitogenesis, there is a surge of clock gene expression at the posterior most region of the PSM such that initially the cells are synchronized (Jiang et al. 2000; Lewis 2003; Riedel-Kruse et al. 2007). As somitogenesis progresses, cells may go out of phase with their neighbors due to cellular

9 movement, cell division, delays in transcription, export and translation which is when intercellular coupling by Notch signaling plays an important role in synchronizing cells

(Lewis 2003; Horikawa et al. 2006; Herrgen et al. 2010; Okubo et al. 2012). Notch signaling establishes cell-to-cell contact (described in Section 1.4) which facilitates synchronization such that intracellular oscillations are coupled and in the same phase of the oscillation. If intercellular coupling is disrupted it affects both the period of the clock and the length of somites formed (Herrgen et al. 2010; Bajard et al. 2014). Thus in the

PSM intracellular oscillations are synchronized between neighboring cells in a non-cell autonomous manner by Notch signaling and tight regulation of this pathway is needed for a functional segmentation clock. The role of canonical Notch signaling in coupling oscillations between cells has mainly been studied in zebrafish and so far there is limited evidence for its role in synchronization in the mouse segmentation clock(Okubo et al.

2012). Section 1.5 focuses on the role of this pathway in the mouse segmentation clock, while highlighting similarities and differences observed between species such as zebrafish.

1.4 Notch signaling pathway establishes cell-to-cell communication

Notch signaling is an evolutionarily conserved pathway that plays numerous roles in cell differentiation and morphogenesis in both invertebrates and vertebrates(Guruharsha et al. 2012; Kopan & Ilagan 2009). Interest in Notch signaling was sparked by the discovery of a mutation in the Notch receptor in D. melanogaster that leads to a serrated or notched wing phenotype (Dexter 1914; Morgan & CB 1916; 10

Mohr 1919). In mammals there are four Notch receptors namely, Notch 1,2,3 and 4 and five canonical DSL ligands (Delta-Serrate-Lag2) three of which belong to the Delta-like family referred to as DLL1, DLL3 and DLL4 and two belong to the Jagged family, Jagged 1 and Jagged 2(Hori et al. 2013). Figure 1.2 depicts the canonical Notch signaling pathway that operates in the mouse PSM. The Notch1 receptor is a transmembrane, heterodimeric protein found on the surface of a cell designated as the signal receiving cell. Interaction of the Notch receptor in trans with a DLL1 ligand on a neighboring cell, designated as signal sending cell, activates canonical Notch signaling in the signal receiving cell. This activation occurs after the cleavage of the Notch extracellular domain

(NECD) and the Notch intracellular domain (NICD) by ADAM17 protease and Presenilin

-secretase complex, respectively. Cleaved NICD enters the nucleus and forms a complex with the DNA binding protein CSL (CBF1, Supressor of Hairless, Lag1) along with

Mastermind-like (MAML) and other co-activators to drive the expression of Notch target genes such as her1 and her7 in zebrafish, c-hairy1 in chickens and Hes7 in mice and humans (reviewed in Wahi et al. 2016).

Notch signaling is modulated in several ways to ensure context dependent activation. Fringe proteins are Golgi resident proteins and play a very important role in modulating Notch signaling either by modifying the Notch receptor or the ligands(Takeuchi & Haltiwanger 2014). The Notch receptor contains epidermal growth factor-repeats that are first extended by POFUT1 by adding an O-linked fucose and then further extended by fringe proteins in the Golgi apparatus (in mammals three fringes

11 exist: Radical, Lunatic and Manic) (Moloney et al. 2000; Munro & Freeman 2000;

Brückner et al. 2000; Irvine et al. 1997; Johnston et al. 1997; Kim et al. 1995). For instance, Lunatic fringe (LFNG), a beta 1,3-N-acetylglucosaminyltransferase, adds a N- acetylglucosamine (GlcNAc) to the O-linked fucose moieties on the Notch receptor and enables it to preferentially interact with the ligand DLL1 over the other Notch ligands in some contexts(Luther & Haltiwanger 2009; Takeuchi & Haltiwanger 2014). In other contexts such as in the case of the mouse segmentation clock, LFNG is thought to negatively regulate Notch activation but the mechanism of how this occurs is not clearly understood. There is some evidence to suggest that LFNG may act by modifying DLL1 ligand in the signal sending cell which could prevent DLL1 from activating the Notch receptor(Dale et al. 2003; Morimoto et al. 2005; Okubo et al. 2012; LeBon et al. 2014;

Irvine et al. 1997).

Figure 1.2: A simplified schematic of the canonical Notch signaling pathway(Wahi et al.

2016). Notch receptors are presented on the surface of signal receiving cells. Binding of receptor to a DSL ligand on a neighboring cell triggers a series of protein cleavage. A final presenillin-mediated event releases the active Notch intracellular domain (NICD), which translocates to the nucleus and forms a complex containing the CSL transcription factor and MAML, directly activating transcription of target genes. The Notch receptor can be modified in the Golgi through the addition of sugar moieties to the EGF repeats.

Modification by LFNG is especially relevant during mouse and chicken somitogenesis.

1.5 Notch pathway is required for normal segmentation clock function

The importance of the Notch pathway in segmentation is supported by the fact that mutating genes such as Notch1, DLL1, DLL3 and Lfng, disrupts somitogenesis in mouse embryos. Similarly in humans, individuals affected by spondylocostal dysostoses (SCD) may have mutations in either DLL3, LFNG or HES7 (Bulman et al. 2000; Sparrow et al.

2006; Sparrow et al. 2008). Hence, a better understanding of the role and regulation of the Notch pathway in somitogenesis would help decipher the mechanisms that are altered in vertebral abnormalities. This however is challenging because the Notch pathway plays sequential roles during somitogenesis. It first acts in the segmentation clock, followed by independent functions in rostral/caudal patterning of somites and finally in the formation of intersomitic boundary, thus it would be hard to interpret which aspect of somitogenesis is affected by a mutation in a Notch pathway member.

This section describes the significance of Notch signaling in the segmentation clock and what is known so far about the function of some of the members of the this pathway, particularly, Lfng which is the main focus of this dissertation.

The use of the canonical Notch pathway in somitogenesis appears to be conserved in the PSM among zebrafish, chickens and mice though there are differences in some of the proteins involved and their pattern of expression. PSM cells in all three organisms express at least one of the Notch receptors and one of the DSL ligands that interact in trans to activate the expression of basic helix-loop-helix (bHLH) transcription repressors such as Her/Hes family members whose expression oscillates in the PSM and are 14 required for the segmentation clock (Krol et al. 2011). The bHLH proteins directly repress their own transcription resulting in the formation of an autoinhibitory loop

(Figure 1.3A for zebrafish and Figure 1.3B for mice) that is responsible for the generation of their intracellular oscillations (reviewed in Wahi et al. 2014).

There are however differences in the zebrafish, chicken and mouse segmentation clock in terms of the number of somites formed, the rate of somite formation and the oscillatory Notch genes that are required for the proper functioning of the segmentation clock (briefly summarized in Table 1.1). In this section I will discuss the Notch pathway in context of the segmentation clock and differences in regulation of the Notch pathway among zebrafish, chickens and mice. Interestingly, unlike the clock in the chicken and mouse PSM, the role of the Notch pathway in the zebrafish segmentation clock is independent of Lfng and having the knowledge of these species-specific differences would provide a comprehensive understanding of Notch pathway regulation.

Figure 1.3: Clock-linked oscillations of hairy/enhancer of split related genes are tightly regulated during somitogenesis(Wahi et al. 2016). A. In the zebrafish, oscillations of her1 and her7 are important for clock function. her gene expression can be activated by both FGF and Notch signals. Her1 protein then feeds back to inhibit its own transcription allowing repeated oscillations. B. In mouse and chick, oscillations of both Notch and FGF signaling converge on Hes7 (or chairy1 in the chick). In the Notch pathway, HES7 protein inhibits transcription of Lfng, which in turn inhibits Notch activity, leading to decreased

Hes7 transcription. In the FGF pathway, HES7 protein inhibits Dusp proteins, which in turn dephosphorylate ERK to reduce Hes7 transcription. HES7 protein also feeds back to inhibit its own transcription, allowing future oscillations. Models of these oscillations suggest that the delays in production of RNA (Tm) and protein (Tp), and the (continued) 16

(Figure 1.3 continued) rapid turnover of active RNA and protein molecules are critical in the segmentation clock.

1.5.1 Role of Notch pathway in the zebrafish segmentation clock

In zebrafish PSM, deltaC/D are the ligands that interact with the Notch receptor

(particularly Notch1a) and activate Notch signaling. Notch1a receptor activation by deltaC/D results in cleavage of NICD by -secretase, resulting in transcription of bHLH l repressors, her1 and her7 (Holley et al. 2002; Henry et al. 2002). Her1 and Her7 form a negative feedback loop by repressing their own transcription by binding as homodimers or heterodimers which results in cell autonomous oscillations of the two genes (Figure

1.3A)(Henry et al. 2002). Her1 and 7 also repress the transcription of deltaC resulting in oscillatory expression of deltaC and cyclic activation of NICD (van Eeden et al. 1998;

Holley et al. 2002; Jiang et al. 2000). deltaD, on the other hand, shows non-oscillatory expression in the PSM but, like deltaC, does activate transcription of her1/7 genes via the Notch receptor (Holley et al. 2002; Mara et al. 2007).

Several studies have provided evidence for the role of Notch signaling in zebrafish segmentation. When Notch signaling is hindered by -secretase inhibitors, which prevent production of NICD, misshapen somites are formed in zebrafish embryos and this phenotype can be rescued in future somites by removing the effect of the inhibitor

17 suggesting that Notch activation is needed for normal somite formation (Jiang et al.

2000; Oates et al. 2005; Jülich et al. 2005; Mara et al. 2007; Riedel-Kruse et al. 2007;

Herrgen et al. 2010). Similar phenotypes are seen when Notch pathway members are mutated. Deleting both her1 and her7 results in larger, disorganized anterior (first five pairs) and posterior somites (remaining somites) (Henry et al. 2002). When her1 or her7 oscillations are disrupted by overexpression, anterior somites form normally but posterior somites are irregular (Giudicelli et al. 2007). Thus, the knockout or overexpression of her genes seems to affect the timing of somite formation without affecting the somite boundary.

In deltaC mutants along with formation of irregular somites in the posterior region, the expression of her1/7 is seen in a disorganized pattern indicating that while individual cells show oscillatory expression, their phases are not coordinated with adjacent cells

(Mara et al. 2007). This suggests that deltaC may be required for synchronization of the segmentation clock in zebrafish (Mara et al. 2007; Ozbudak & Lewis 2008; Soza-Ried et al. 2014). DeltaD is thought to fit into this model by initiating Notch activation in the posterior PSM since loss of deltaD disrupts oscillatory expression of Notch target genes

(her1,her7, deltaC) in the posterior region but doesn’t affect the non-oscillatory anterior stripe (Mara et al. 2007). Recent work provides concrete evidence for the role of Notch signaling in synchronizing PSM cells(Delaune et al. 2012; Soza-Ried et al. 2014). Delaune et al., found that loss of Notch family members in the zebrafish PSM does not disrupt cell autonomous oscillations but the oscillations between cells are not in phase with

18 each other. Soza-Ried et al., further tested the role of Notch signaling in the PSM by utilizing a heat-shock system that drives deltaC expression to generate pulses of Notch activation through the PSM of the deltaC mutant zebrafish. These pulses of Notch activation were able to rescue newly formed somites based on the timing and length of

Notch activation. Hence, at least in zebrafish, Notch signaling plays a role in coupling oscillations in individual cells which is required for the formation of organized and equally sized somites (Delaune et al. 2012; Soza-Ried et al. 2014).

1.5.2 Role of the Notch pathway in the chicken and mouse segmentation clock

Oscillatory Notch signaling in chickens and mice is synchronized along the PSM by the cyclic transcription of DLL1 and Notch1 mRNA and protein (Bone et al. 2014).

DLL1 ligand on the signal sending cell periodically interacts with the Notch receptor (1 and 2 in mice) on the signal receiving cell to result in pulsatile activation of Hes7 and

Lfng by NICD (Figure 1.2). HES7, which is homologous to her1 in fish, can in turn repress its own transcription as well as the transcription of Lfng (Figure 1.3B) causing intracellular oscillations of Hes7 and Lfng mRNA in the signal receiving cell, generated by this negative feedback loop (Aulehla & Johnson 1999; Bessho et al. 2001; Chen et al.

2005; Niwa et al. 2007). In the context of segmentation HES7 and LFNG are negative regulators of Notch activity which is supported by the fact that knocking out either of these genes in the mouse PSM results in upregulation of the other (Ferjentsik et al.

2009).

In the PSM, the signaling sending and receiving identities of cells are controlled temporally rather than spatially i.e. all cells present the Notch receptor on their surface

(to receive signal) and then present DLL1 on their surface (to send signal) which explains why Notch1 and DLL1 oscillate out of phase with each other(Bone et al. 2014; Shimojo et al. 2016). While Notch1 and DLL1 act in trans, LFNG appears to functions to modulate signaling activity, perhaps by modifying ligand(s) and modulating cis-interactions at specific stages of the cycle (Okubo et al. 2012; Bone et al. 2014). Thus, unlike zebrafish, oscillatory Notch signaling occurs in a LFNG dependent manner in the PSM of chickens and mice (Evrard et al. 1998; Zhang & Gridley 1998).

There is conflicting evidence on the importance of oscillatory Notch signaling in the chicken and mouse segmentation clock. Table 1.2 summaries the expression patterns and the null mutant phenotypes of some of the Notch pathway members in mouse embryos. One study proposes that Notch signaling could be the principal generator of clock oscillations in the PSM (Ferjentsik et al. 2009). This was tested by preventing Notch activation in mice by disrupting the activity of presenilin1 and 2

(Psen1, 2 -secretase complex that is required for the Notch activation. This resulted in absence of NICD expression in the PSM and in severe loss of segmentation with no somite boundaries being formed (Ferjentsik et al. 2009). These mutants showed complete loss of expression of oscillatory Notch targets such as Lfng and Nrarp, and drastic reduction in Hes7 expression in most of the PSM and to a less extent in the tailbud region where Hes7 expression is thought to be regulated by the

FGF pathway (Ferjentsik et al. 2009; Niwa et al. 2007). Psen1, 2 mutants were also defective for expression of clock genes that belong to the FGF and Wnt pathways, alluding to the idea that Notch signaling could possibly be a principal pathway of the mouse segmentation clock (Ferjentsik et al. 2009).

Table 1.2: Expression and null mutant phenotypes of Notch pathway members in the context of the mouse segmentation clock.

On the other hand, several studies suggest that Notch signaling is required for coordinating oscillations between PSM cells, as in the case of the zebrafish segmentation clock (Soza-Ried et al. 2014), which implies that the absence of oscillatory

Notch signaling would only affect the rate of somite formation (Swiatek et al. 1994;

Conlon et al. 1995; Hamada et al. 1999; Feller et al. 2008). Loss of either Notch1 or

Notch2 receptors from the mouse PSM does not prevent somite formation but results in formation of disorganized somites of varying sizes (Swiatek et al. 1994; Conlon et al.

1995; Hamada et al. 1999). These results may be attributed to redundancy of Notch receptors in the PSM that may play a compensatory role in each other’s absence and may be the cause for less severe phenotypes observed in the context of somitogenesis.

When NICD is constitutively (rather than cyclically) expressed throughout the PSM somite formation is still observed. The somites however, were irregularly shaped and differed in size from each other(Feller et al. 2008). Similarly, in Dll-1 knockout mice somite formation occurs, but somite patterning is completely disrupted and the somites lose rostral-caudal polarity. These mutants exhibit loss of expression of Lfng (Barrantes et al. 1999). These results give the impression that dynamic Notch activity may not be essential to make somite boundaries, but instead may be required to synchronize oscillations in PSM cells. However, the possibility of a compensatory role among the DSL ligands in the PSM should be acknowledged while interpreting these results.

While the mechanism by which LFNG regulates Notch signaling in the PSM of chickens and mice is not completely known, it is evident that oscillatory Lfng expression

23 in the PSM is required normal clock function and somite formation. In Lfng null mice,

Notch activity, though appears to remain dynamic, is upregulated in the PSM, and somite shape and polarity are disrupted resulting in severely fused ribs and a truncated tail in these mice (Zhang & Gridley 1998; Evrard et al. 1998; Shifley et al. 2008; Ferjentsik et al. 2009). Similarly, Hes7 is upregulated in these mice but continues to show dynamic expression as seen in the PSM of wildtype embryos(Ferjentsik et al. 2009). This supports the concept that cyclic Notch activity is controlled by oscillatory expression of Notch and

DLL1, and LFNG may be needed only to synchronize these oscillations. However, when the LFNG protein is constitutively expressed in the PSM, the skeletal phenotype is more severe than the Lfng null mice with severely perturbed somite formation and patterning and complete loss of dynamic Notch activation(Williams et al. 2016). Interestingly, constitutive expression of LFNG stabilizes mature Hes7 mRNA transcripts in the mouse

PSM which suggests that LFNG may in some way be responsible for the degradation of

Hes7 transcripts during the OFF phase of Hes7 oscillations(Williams et al. 2016). Thus, the regulation of dynamic expression of Lfng gene products are likely to be critical for oscillatory Notch activity and needed for somite formation and normal skeletal development. This indicates that apart from the production of Lfng mRNA and protein the degradation of these gene products in the PSM is equally important for the maintenance of oscillatory Notch activity.

With the evidence we have so far in mice, oscillatory Notch signaling does seem to play a role in synchronizing the segmentation clock for timely somite formation and

24 dynamic Lfng expression appears to be essential for this process. [In this context, the importance of dynamic Notch signaling in the posterior PSM in maintaining synchrony could be compared to a principal conductor in an orchestra. The musicians (cells) in the orchestra (PSM) have the ability to play their own musical instrument (individual oscillations) however the end result would be discordant in the absence of the principal conductor. The principal conductor (Notch signaling) harmonizes the music played by each member, using a baton (Lfng), to result in a synchronized and melodious symphony

(regular somites)].

While Notch activity is dynamic in the posterior PSM, in the anterior PSM Notch activity is stabilized and plays a role in pre-patterning the presumptive somites, discussed further in the section 1.7. Before discussing somite patterning, it is important to understand the role of the wavefront in the PSM and how this is related to slowing down of clock activity in the anterior PSM (region II) so that future somite boundaries can be formed. This is discussed in the following section.

1.6 Setting up the wavefront in the vertebrate PSM

Expression of clock genes, such as Lfng, oscillates at a faster pace in the posterior region and is slower in cells entering the anterior PSM. In the anterior PSM, the wavefront is thought to be responsible for the positioning of the newly formed somite boundary by interacting with clock genes(Figure 1.1B)(Oates et al. 2012). In the PSM of zebrafish, chicken and mouse embryos the wavefront is believed to be positioned by three pathways namely, the Wnt, FGF and Retinoic acid pathways (Aulehla et al. 2003; 25

Aulehla et al. 2008; Dubrulle et al. 2001; Delfini et al. 2005; Del Corral et al. 2003). The ligands of the Wnt and FGF pathways i.e Wnt3a and Fgf4/8 are expressed as gradients in a posterior to anterior direction in the PSM (Figure 1.1B) and are required to maintain

PSM cells in an undifferentiated state (Dubrulle et al. 2001; Dunty et al. 2008; Takada et al. 1994). The importance of Wnt and FGF pathways in positioning the wavefront is evident from the fact that manipulating Wnt or FGF activity in the PSM shifts the somite determination front (Takada et al. 1994; Aulehla et al. 2003; Aulehla et al. 2008;

Dubrulle et al. 2001). Retinoic acid is expressed in an opposing gradient in somites such that it is highly expressed in older, more anterior somites and its expression reduces in the posterior, more recently formed somites(Del Corral et al. 2003). Thus, the wavefront is marked in the anterior PSM by cells that show a reduction in Wnt, FGF and Retinoic acid signaling.

So far, it is unclear if and how the wavefront controls the slowing down of clock oscillations in the anterior PSM. There is, however, clear evidence of crosstalk between the Notch pathway that plays a critical role in the segmentation clock and the Wnt and

FGF pathways that are required for the formation of the wavefront suggesting a possible link between the timing (clock) and positioning (wavefront) of somite formation. Some of members of the Wnt and FGF pathways exhibit oscillatory expression and are thought to be required for proper clock function. Negative regulators of the Wnt pathway such as Axin2, Nkd1 and Dact1 exhibit oscillatory expression in the mouse PSM (Aulehla et al.

2003; Ishikawa et al. 2004; Suriben et al. 2006). A possible link between Wnt signaling

26 and the Notch pathway is supported by the evidence that upregulating Axin2 in the mouse PSM leads to formation of larger somites and cyclic expression of Lfng is impaired in the PSM of these mutant mice (Aulehla et al. 2003). Evidence of another link between the Notch and the Wnt pathway is the cyclic Notch target gene, Nrarp. The

Nrarp protein feeds back into the Notch pathway by destabilizing NICD(Lamar et al.

2001). Nrarp also regulates Wnt signaling by stabilizing LEF1, a mediator of Wnt signaling, however the significance of regulation is not fully understood in the context of segmentation (Ishitani et al. 2005). Interestingly, activated Notch and cyclic genes in the

Wnt pathway oscillate out of phase with each other. Since activated Notch and DLL1 also oscillate out of phase with each other, it can be hypothesized that Wnt signaling may be acting upstream of DLL1.

Some of the activators (phosphoERK and Dusp4/6, Figure 1.3B) and inhibitors

(Sprouty2/4) of the FGF pathway also show oscillatory expression in the mouse PSM

(Dequéant et al. 2006; Niwa et al. 2007; Hayashi et al. 2009). Of particular interest are pERK and DUSP4/6 as they regulate the expression of Hes7 in the caudal (tailbud) region of the PSM. Oscillatory pERK, a target of FGF signaling, activates expression of Hes7 and

Dusp4/6 (Niwa et al. 2011). HES7 represses its own expression as well as the expression of Dusp4/6. Dusp4/6, in turn, inactivates pERK by dephosphorylating it (Dequéant et al.

2006; Niwa et al. 2007). This negative feedback loop between pERK, Dusp4/6 and, HES7 is thought to be responsible for the weakly synchronized oscillations of Hes7 mRNA and protein in the caudal PSM that are observed even in the absence of canonical Notch

27 signaling (Niwa et al. 2007). Hence, Hes7 is thought to set the pace of clock activity in the PSM as provides link between the wavefront (FGF signaling) and the clock (Notch signaling).

While both Wnt and FGF pathways are needed for the formation of the wavefront,

FGF signaling appears to act upstream of Wnt signaling. Loss of Wnt3a expression in the

PSM caudalizes somites and reduces expression of Fgf8 (Takada et al. 1994; Aulehla et al. 2003), while loss of Fgf4 and Fgf8 from the PSM (deleting Fgf8 alone has no defect on somitogenesis) results in complete absence of somite boundaries and completely abolishes Wnt signaling (Reifers et al. 1998; Naiche et al. 2011). This suggests the FGF signaling may be the critical pathway in not only setting up the wavefront but also influencing the pace of the clock by regulating oscillatory expression of Hes7 in the caudal most region of the PSM where clock gene oscillations originate.

Most of the work discussed in this section so far used mouse or chicken as model systems to understand the interaction between the clock and wavefront. Interestingly, work done in zebrafish suggests that slowing down of clock gene oscillations in the anterior PSM such as her1 oscillations are not controlled by the wavefront (Bajard et al.

2014; Shih et al. 2015). The classical clock and wavefront model suggests that clock gene oscillations in cells in the anterior PSM are arrested by interaction with the wavefront

(Cooke & Zeeman 1976). Work done by observing activity of clock gene reporters in the zebrafish PSM in real time has found that cells in the anterior PSM continue to oscillate, though with a slower periodicity, till they become part of a somite(Shih et al. 2015).

Another study found that inhibiting Wnt signaling affects the position of the somite boundary, by lengthening the somites, but somite were formed at the same rate indicating that wavefront may not affect the pace of somite formation which is controlled by the segmentation clock(Bajard et al. 2014). So at least in the context of zebrafish somitogenesis, the wavefront does not seem to determine the pace of the segmentation clock.

In summary, Lfng oscillations slow as cells progress towards the anterior PSM. In case of the mouse segmentation clock while there is some evidence that Wnt and FGF gradients may be responsible for this difference in oscillation frequency along the PSM, there is no concrete data as yet to completely support this hypothesis. Examining the regulation of Lfng in the mouse PSM could provide insights into the relationship between the clock and the wavefront.

1.7 Somite patterning in the anterior PSM is influenced by the Notch pathway

Oscillatory expression of Lfng in the PSM is involved in the mouse segmentation clock while the constitutive anterior stripe of Lfng expression is thought to play a role in defining rostral and caudal compartments of the future somite. Thus, Notch signaling plays a dual role in somitogenesis; first by regulating the timing of somite formation and then by influencing somite patterning (reviewed in Wahi et al. 2014). Before we delve into the regulation of Lfng, it is important to understand its role in somite patterning as

29 manipulating the regulation of Lfng in the mouse PSM may have an impact not only on the segmentation clock but also on somite patterning.

In the anterior PSM or region II, the pre-patterning of somites into rostral and caudal halves, marked as pre-somite regions S0 and S-1 of future somites (Figure 1.4), is initiated in a Notch dependent manner by the activation of Mesp2 (Mesoderm posterior protein 2), throughout the pre-somite S-1 but refined in the rostral compartment of the

S0 pre-somite (Saga et al. 2000; Takahashi et al. 2000; Takahashi et al. 2007; Oginuma et al. 2008). Mesp2 is only expressed in the presence of activated NICD and TBX-6

(Yasuhiko et al. 2006; Yasuhiko et al. 2008; Oginuma et al. 2008; Oginuma et al. 2010).

Posterior to S-1, i.e below the wavefront FGF and pERK are expressed and suppress

Mesp2 expression by activated NICD. This restricts expression of Mesp2 to the pre- somite region of the anterior PSM while it is not expressed in the unsegmented posterior PSM (Saga et al. 2000; Oginuma et al. 2008; Morimoto et al. 2005; Takahashi &

Sato 2008).

Cells in the future rostral compartment of the S0 show distinct expression of Mesp2 while the future caudal compartment is distinguished by active Notch signaling. Active

Notch signaling gives caudal identity to the future somite evidenced by the fact that when NICD is overexpressed somites are caudalized while complete absence of NICD rostralizes somites (Feller et al. 2008; Koizumi et al. 2001). Tight expression of Mesp2 and NICD is achieved by multiple feedback loops (Figure 1.4). In the rostral compartment, MESP2 inhibits NICD activity by activating Lfng and by destabilizing the

30 co-activator MAML (mastermind-like) (Morimoto et al. 2005; Oginuma et al. 2010).

Further, MESP2 inhibits TBX6 in the rostral compartment by facilitating its degradation

(Oginuma et al. 2008). In the caudal compartment, Notch activation is stabilized by the coordinated role of TBX6, Wnt3a and CREB in increasing the expression of DLL1 (White

& Chapman 2005; Lopez & Fan 2013).

Both the loss and overexpression of Lfng affects somite patterning by disrupting rostral and caudal identities of somites. Lfng null mice exhibit poorly compartmentalized somites while constitutive expression of LFNG protein results in expansion of the caudal compartment attributed to the increase in NICD activity seen in this region (Zhang &

Gridley 1998; Evrard et al. 1998; Williams et al. 2016). Hence, misregulation of Lfng appears to affect somite formation and patterning.

In summary, Notch signaling plays a dual role in somitogenesis while its dynamic expression profile in the posterior PSM controls the timing of somite formation, its constitutive expression in the anterior PSM regulates boundary formation and patterning of the somites. As we have seen above that Notch activity needs to be tightly regulated, failure of which leads to defective segmentation. This is especially important in case of oscillatory Notch genes in the PSM that rapidly switch between ON and OFF phases of expression and is elaborated in the following section.

Figure 1.4: Role of Notch signaling during mouse somitogenesis(Wahi et al. 2016). The negative feedback loop in Region I maintains oscillatory Notch signaling required for timely somite formation. Notch signaling along with MESP2 in Region II influences somite patterning into rostral (blue) and caudal (white) compartments.

1.8 Fine tuning oscillatory Notch signaling in the segmentation clock

For PSM cells to sustain oscillations of clock genes it is required not only that there is a sufficient amount of delay in transcript production (Tm) and protein production and modification (Tp) but also that the mRNA and protein half-lives are extremely short. This was predicted mathematically by an autoinhibition model proposed by Julian Lewis

(Lewis 2003). It was postulated that since cyclic activity is created by negative feedback loops, the repressor protein should accumulate at a slow pace so that it takes longer to reach the threshold before it can cause its own repression.

Several experimental studies have tried to alter oscillations of clock genes, and thus the period of the clock, by manipulating the time delay by altering the time taken for transcription to occur which was done by either adding or removing introns. This was done in an attempt to change the pace of the oscillations and examine the consequence of that on somite formation. However, attempts at elongating the time delay by adding a large intron have been unsuccessful resulting in either a null or severely hypomorphic mutations by defective splicing in the case of Hes7 (Stauber et al. 2012; Fujimuro et al.

2014)or having no effect as in the case of Lfng (Stauber et al. 2012). This can be explained by the fact that transcription occurs very rapidly in vivo and that splicing and export of the transcript from the nucleus contribute most to the total time delay (Hoyle

& Ish-Horowicz 2013). Attempts at shortening the time delay by removing introns has been more successful by reducing splicing related delays. Harima et al. were able to increase the frequency of Hes7 oscillations by removing two of the three introns in mice 33

(Harima et al. 2013). Interestingly, removing all the introns in Hes7 led to the loss of oscillatory expression of the Hes7 mRNA and protein and the mice showed severe somite and skeletal defects. Due to the rapid rate of production of the intronless Hes7 transcript, the rate of degradation of the transcript and the protein is not fast enough to keep up with the production which leads to accumulation of the transcript and protein molecules in cells, eventually causing dampening of clock oscillations (Takashima et al.

2011).

Consistent with these findings, mathematical modeling predicts that for tight oscillations to be sustained in the PSM, the RNA and protein half-lives should be short(Feng & Navaratna 2007; González & Kageyama 2009). Short mRNA and protein half-lives ensure rapid degradation of gene products before the next round of oscillations begin. The 3’UTR plays a critical role in maintaining short half-lives of transcripts by either affecting RNA degradation or translation efficiency and this is especially important in the case of clock genes such as Hes7 and Lfng that require rapid

RNA turnover to maintain cyclic expression (Grzybowska et al. 2001; Chen et al. 2005;

Riley et al. 2013; Nitanda et al. 2014b). The 3’UTR of Lfng has evolutionarily conserved regions that destabilize the mRNA, and contains conserved binding sites for miR-125a-

5p(Chen et al. 2005; Riley et al. 2013). When interactions between the Lfng 3’UTR and miR-125a were disrupted in chicken embryos, Lfng transcripts were initially stabilized in the PSM, resulting in the formation of misshapen and poorly patterned somites. These

34 results suggest that rapid turnover of Lfng transcripts is essential for somitogenesis and is regulated by mir-125a during chick somitogenesis(Riley et al. 2013).

As expected, the half-lives of oscillatory proteins linked to the clock are also very short, suggesting that there are mechanisms that degrade clock-linked proteins rapidly in the PSM. In mice, increasing the half-life of the HES7 protein from 22 to 30 min, by altering a defined ubiquitination site, causes defective segmentation in homozygous mutants. These embryos exhibit constitutive low levels of Hes7 transcription, and low, stabilized expression of HES7 protein, suggesting that normal clock oscillations are lost in these mutants(Hirata et al. 2004). As discussed previously, stabilizing the LFNG protein also causes defective segmentation and patterning in mouse embryos (Williams et al. 2016). Together, these experiments emphasize the importance of tight post- transcriptional regulation of the segmentation clock, and provide interesting hints as to how the clock period may vary in a species specific manner.

The main focus of this dissertation is to examine the mechanism by which the Lfng

3'UTR regulates transcript turnover in the context of the mouse segmentation clock.

Chapter 2 examines the post-transcriptional of Lfng by miR-125a, which is known to bind to the mouse Lfng 3'UTR, while Chapter 3 investigates the role of other conserved regions in the Lfng 3'UTR in destabilizing the transcript. Chapter 4 summaries all the findings from this thesis work and discusses future directions that could be pursued to explore post-transcriptional regulation of clock genes. Appendix A discusses preliminary work done to identify promoter regions of the miR-125a family in an attempt to better

35 understand the transcriptional regulation of miR-125a which may lead to a link between miR-125a and other pathways involved in somitogenesis.

CHAPTER 2

REGULATION OF Lfng BY miR-125a IN THE MOUSE SEGMENTATION CLOCK

2.1 Introduction

Segmentation in vertebrates occurs as somites bud from the pre-somitic mesoderm

(PSM), giving rise to the axial skeleton and skeletal muscle. This process is regulated by a

"segmentation clock" that times somite formation(Wahi et al. 2016). Genes, such as

Lunatic fringe (Lfng), that are linked to the clock exhibit cyclic expression with a period that matches the rate of somite formation(Cole et al. 2002; Morales et al. 2002; Serth et al. 2003). These oscillations are rapid in cells in the posterior PSM and decelerate in cells as they enter the anterior PSM(Palmeirim et al. 1997; Gibb et al. 2010). This implies that the RNA turnover in the posterior PSM must be faster than the anterior PSM. Post- transcriptional regulation by microRNAs binding at specific sites in the 3’UTR of a transcript is an effective mechanism by which RNA turnover of the transcript can be regulated(Grzybowska et al. 2001; Riley et al. 2013; Jing et al. 2015). In the context of the segmentation clock, we have previously identified miR-125a as a regulator of Lfng mRNA turnover in the chicken PSM(Riley et al. 2013). Interestingly, mir-125a expression appears to be higher in the posterior PSM as compared to the anterior PSM of chicken and mouse embryos raising the possibility that miR-125a could influence the changes in

37 the frequency of Lfng oscillations in the PSM(Riley et al. 2013). To test effects of the mir-

125a on mRNA stability in mouse embryos, we have examined the expression of Venus reporter constructs that contain either wildtype or mutant Lfng 3'UTR sequences in cell lines and in mouse embryos. In transgenic lines expressing the reporter with the mutated miR-125a binding sites in the Lfng 3’UTR we find Venus expression to be consistently expressed in the posterior most region of the mouse PSM suggesting that the miR-125a binding sites in the Lfng 3'UTR influence RNA turnover in the posterior

PSM. To test the effects of mir-125a on mouse segmentation, we used the CRISPR-Cas9 system to create mutations that block miR-125a expression and found no effect of loss of miR-125a on somitogenesis and skeletal formation. This indicates that either a compensatory mechanism may be in effect or that miR-125a does not play a role in the mouse segmentation clock.

2.2.1 Segmentation in vertebrates

The segmented body plan in vertebrates is laid down during embryonic development by a process known as somitogenesis. During somitogenesis cohorts of cells periodically bud off from the unsegmented pre-somitic mesoderm, located at the posterior end of the embryo, to form a pairs of somites that flank the neural tube

(Figure 2.1A). At the end of somitogenesis the numerous pairs of somites that are formed (the number differs among vertebrates) will differentiate and give rise to the ribs, vertebrae, skeletal muscles, and dermis of the back (Wahi et al. 2016). Mutations in genes that are known to play a role in somite formation results in congenital vertebral

38 defects such as spondylocostal dysostoses, scoliosis or spinal kyphosis (Turnpenny et al.

2007; Eckalbar et al. 2012).

The process of periodic somite formation is controlled by a “segmentation clock” that consists of genes, known as clock genes, whose expression oscillates in the posterior region of the PSM denoted as Region I (Figure 2.1A). These oscillations are faster in cells located at the posterior region of the PSM and slower in more anteriorly placed cells (Palmeirim et al. 1997; Gibb et al. 2010). The period of clock gene oscillations in the posterior region matches the rate of somite formation, though the rate of somite formation and the oscillatory genes essential for somite formation vary among vertebrates (Cooke & Zeeman 1976; Pourquié 2011; Wahi et al. 2016).

2.2.2 Oscillatory Lunatic fringe expression is essential for the mouse segmentation clock

Lunatic fringe (Lfng) is a clock gene that is part of the Notch signaling pathway and encodes a glycosyltransferase that modifies Notch receptors and ligands by altering

Notch activity (Moloney et al. 2000; Munro & Freeman 2000; Brückner et al. 2000; Irvine et al. 1997; Hou et al. 2012; Serth et al. 2015). Expression of Lfng in Region I of the chicken and mouse PSM oscillates between three phases (Figure 2.1B) i.e. expression of oscillatory genes like Lfng is a snapshot of a specific point in the cycle and can be grouped intp three phase for convenience(Pourquié & Tam 2001). In Region II, Lfng is expressed as a stable anterior stripe. Oscillatory expression of Lfng is essential for mouse and chicken somitogenesis and loss of oscillations, either by knockout or

39 constitutive expression of Lfng in the Region I of the PSM, disrupts somitogenesis and results in vertebral defects (Zhang et al. 2015; Evrard et al. 1998; Serth et al. 2003;

Williams et al. 2016). Lfng oscillations are regulated by a negative feedback loop involving activation by Notch signaling and repression by HES7 (c-hairy1 in chickens), which is a basic helix-loop-helix (bHLH) protein and also a Notch target gene (Cole et al.

2002; Morales et al. 2002; Bessho et al. 2001; Serth et al. 2003).

Mathematical models predict that apart from transcriptional regulation, maintenance of clock gene oscillations would require additional levels of control at the post-transcriptional and post-translational level to ensure rapid RNA and protein turnover (Lewis 2003; Feng & Navaratna 2007; González & Kageyama 2009; Jing et al.

2015). In this study we have explored the mechanism by which Lfng is post- transcriptionally regulated in the context of the mouse segmentation clock.

2.2.3 Post-transcriptional regulation of Lfng by the 3’ UTR

Post-transcriptional regulation of transcripts by the 3’ untranslated region (3’UTR) occurs by altering RNA stability, RNA localization and/or translation efficiency

(Grzybowska et al. 2001). In the context of segmentation, work done in chickens and mice have found that the 3’ untranslated region (3’UTR) of cyclic genes such as Hes7 and

Lfng is essential for rapid RNA turnover in the PSM (Hilgers et al. 2005; Chen et al. 2005;

Nitanda et al. 2014; Riley et al. 2013; Fujimuro et al. 2014).

One mode by which the 3’UTR of short-lived transcripts can affect transcript stability and facilitate rapid transcript turnover is by having micro-RNA binding sites in its 40 sequence (reviewed in Filipowicz et al. 2008). Micro-RNAs (miRNAs) are short non- coding RNAs(~22nt) that repress expression of protein coding genes by binding to specific sites in their 3’UTR and causing mRNA degradation and/or translational repression (Ha & Kim 2014). Interestingly, knockout of Dicer, an enzyme required for miRNA processing, resulted in perturbed somitogenesis with irregularly formed somites in both zebrafish and mouse embryos, and the somites in the mouse embryos were caudalized (Giraldez et al. 2005; Zhang et al. 2011). Therefore, miRNAs could be potential candidates responsible for rapid RNA turnover of clock genes and may play an important role in the segmentation clock. Since clock genes oscillations are seen in the

PSM but not in the somite region, the post transcriptional regulation of clock genes is likely to differ between these two regions. For instance, Lfng mRNA expression oscillates in three phases in the PSM but is constitutively expressed in the anterior most region of the PSM in the form of a stripe (Figure 2.1B).

The role of miRNAs in the segmentation clock was first investigated using miRNA microarray to quantify the expression levels of several miRNAs in the unsegmented mouse PSM as compared to the somites (Riley et al. 2013). miR-125a-5p is one of the miRNAs that was found to be enriched in the mouse PSM and interestingly, the mouse

Lfng 3’UTR has three binding sites for this miRNA. The binding site near the 3’end of

Lfng 3'UTR is highly conserved from mouse to chicken to human (Figure 2.1C). Riley et al., studied the significance of the interaction between the Lfng 3’UTR and miR-125a in the chick segmentation clock by using anti-miR-125a or target protectors to block this

41 interaction. Blocking this interaction initially resulted in stabilized Lfng expression in the chick PSM indicating that miR-125a destabilizes Lfng mRNA. Preventing this interaction for an extended period of time resulted in reduction of Lfng expression in the PSM by perturbing the negative feedback loop and affecting Lfng transcription. Since the LFNG protein can negatively regulate Notch activation in the feedback loop, the initial stabilization of Lfng in the PSM is followed by reduction in expression of Lfng. Loss of

Lfng regulation by miR-125a also disrupted somitogenesis in chicken embryos such that the somites were irregular and their boundaries were not formed clearly. Hence, rapid turnover of Lfng transcript mediated by mir-125a is essential for the chick segmentation clock (Riley et al. 2013).

Figure 2.1: Regulation of Lfng in the segmentation clock. A. Mouse embryo showing the PSM and somites marked by Uncx (purple), a posterior boundary marker. Cartoon of

PSM showing Region I where segmentation clock genes show oscillatory expression while in Region II where clock oscillations are much slower and patterning of presumptive somites occurs. B. In situ hybridization for Lfng mRNA in the PSM of mouse embryos exhibit oscillatory expression as seen by the three phases of expression in

Region I and a non-oscillatory anterior stripe in Region II. C. Mouse Lfng 3’UTR, about

1kb in length, showing regions conserved in humans (stripes) and chickens (green). The mouse Lfng 3’UTR has three miR-125a binding sites, of which the third one (at the

3’end) is conserved in both chickens and humans.

The role of miR-125a in regulating stability of the mouse Lfng mRNA and the importance of this interaction in the mouse segmentation clock has not been studied to date. It is unclear if this regulation is species specific or conserved among vertebrates.

Computer simulations of the regulation of Lfng by miR-125a predicts that this interaction could act to fine tune Notch activity in the PSM and would affect the period and amplitude of clock gene oscillations (Jing et al. 2015). Therefore, since the period of clock gene oscillations varies among vertebrates, such as in chicks (90 mins, Palmeirim et al. 1997), mice (120 mins, Tam 1981) and humans (4-5 hours, William et al. 2007), the miR-125a mediated regulation of Lfng may differ between these species. This study investigates firstly, if the miR-125a binding sites in the Lfng 3'UTR are required for mRNA regulation in the mouse PSM and secondly, if miR-125a plays a crucial role in the mouse segmentation clock. To address the first question we used reporter transgenes with the mouse wildtype or mutated Lfng 3'UTR (Figure 2.2A and 2.2B) which is advantageous because it allows us to study the regulation of transcripts with the Lfng

3'UTR in the mouse PSM, without disrupting the negative feedback loop and affecting the segmentation clock. To address the second question we knocked out miR-125a in the mouse genome and analyzed the loss of miR-125a expression on somitogenesis. We have found that the miR-125a binding sites in the mouse Lfng 3'UTR are required for mRNA regulation, at least in the caudal PSM, however, the loss of miR-125a doesn’t seem to have an effect on mouse somitogenesis.

Figure 2.2: Destabilized Venus reporter constructs. A constitutive or oscillatory promoter drives the expression of a destabilized Venus protein that is flanked by a b- globin intron at the 5’ end and either the A. Wildtype Lfng 3’UTR or B. Mutated Lfng

3’UTR at the 3’ end. Construct also contains polyadenylation sequences from the Lfng.

2.2 Materials and Methods

2.2.1 Construction of Venus reporter plasmids

The 5’ -globin intron and Venus cassette was taken from the pRARE-Venus-Ubiq- dsRed plasmid (gift from Randall Moon) and cloned into the TOPO vector pCR2.1™

(Invitrogen) and then transferred into the pBluescript vector (Agilent Technologies). The

DNA sequence for the PEST domain, to destabilize the VENUS protein, was amplified from the pdEGFP-N1 vector (Clontech) using the forward primer,

5’TTGGCGCGCCAGCCATGGCTTCCCGCCG3’ and reverse primer,

5’TCTAGACTACACATTGATCCTAGCA3’, and inserted in frame with the Venus coding sequence in the pBluescript backbone to generate a VENUS protein with a PEST domain

(destabilized Venus or dV). The following mammalian promoters were inserted upstream of the -globin intron and destabilized Venus cassette, to drive Venus expression: Constitutive promoter, PGK amplified from the ploxp plasmid with 5’

TCAGTACTTTTCCCAAGGCAGT 3’ and 5’ ATTGGCTGCAGGTCGAAAG 3’ and inserted into a pBluescript backbone or CMV promoter in the pCDNA3.1-Neo backbone; Oscillatory promoter, 3.8kb of the mouse Lfng promoter inserted in the pBluescript backbone.

The wildtype mouse Lfng 3'UTR and polyA sequence (Wt UTR) were PCR amplified from genomic DNA with the mLfng 3'UTR primers shown in Table 2.3 and cloned into the pBluescript backbone adjacent to the stop codon of the Venus-PEST sequence. The mutated Lfng 3’UTR was created by mutating the three miR-125a binding sites in the wildtype UTR to the HindIII recognition sequence by site-directed mutagenesis using primers shown in Table 2.3 (HindIII sites shown in bold). The mutated Lfng 3’UTR and

Lfng polyA sequence (Mut UTR) were then inserted into the pBluescript backbone similar to the wildtype Lfng 3'UTR as mentioned above.

Primer Sequence (5'-3') mLfng 3'UTR F AATCTAGACAGTCGTGGTTGAAACTCTGT mLfng 3'UTR R TTATCGATTGGCTGTCCTGGAACTCACTC Mut 125a site 1 F TTCTTAAGCTTGTGGCAGGTCTGT Mut 125a site 1 R TGCCACAAGCTTAAGAAGACGACT Mut 125a site 2 F TCCACTAAGCTTTCCGTGGGTGCT Mut 125a site 2 R CACGGAAAGCTTAGTGGAGCCCCA Mut 125a site 3 F AAGCTCAAGCTTAATTGATGTGTT Mut 125a site 3 R TCAATTAAGCTTGAGCTTTTCCCT

Table 2. 3 Lfng 3’UTR primers. Primers to amplify mouse Lfng 3'UTR (mLfng) and to mutate the three miR-125a binding sites.

Plasmids that were used to make stable cell lines contained a neomycin cassette to facilitate selection when cultured in the presence of Geneticin® (Gibco). The following

Venus expression plasmids were generated at the end of the all the cloning steps:

PGK-dV-Wt UTR and PGK-dV-Mut UTR; CMV-dV-Wt UTR-Neo and CMV-dV-Mut UTR-

Neo; Lfng-dV-Wt UTR (LvLWt) and Lfng-dV-Mut UTR (LvLMut).

2.2.2 Transient transfections and generation of stable cell lines

Transient transfections

NIH3T3(ATCC® CRL-1658™), mouse fibroblasts cells, were cultured in Dulbecco’s

Minimal Eagle Media (DMEM, Corning) with 10% Fetal Bovine Serum (FBS, Sigma), 1%

Glutamine and 1% Penicillin-Streptomycin at 37°C with 5% CO2. To carry out transient transfections, cells were plated into wells of a 24-well plate at a concentration of 40,000 cells/well in antibiotic free DMEM. Transfection of 500ng PGK-dV- Wt UTR or PGK-dV-

Mut UTR plasmid with 50nM of either a Scrambled control (Ambion, negative control

NC#1 #17110) or pre-miR-125a (Ambion, pre mir #17100 PM10492) was carried out using the Lipofectamine®2000 transfection reagent (Invitrogen). Cells were incubated for 24-48 hours before isolating RNA for RT-PCR. This experiment was repeated three independent times.

Generation of stable cell lines

C2C12 (ATCC®CRL-1772™), mouse myoblast cells, were used to make stable cell lines and cultured in the same manner as 3T3 cells, mentioned above. Cells were plated at

20,000 cells/well in a 24-well plate and 500ng of the following plasmids were linearized using ClaI restriction enzyme (NEB) and transfected into the respective wells using the

Lipofectamine®2000 transfection reagent: CMV-dV-Wt UTR-Neo and CMV-dV-Mut UTR-

Neo. Cells were incubated for 24 hours and transferred into a 10 cm tissue culture plate at a 1:100 dilution. These transfected cells as well as a ‘no transfection control’ (NTC) plate were maintained under the selection of 800ug/ml Geneticin® for about 3 to 4 weeks, till most of the cells in the NTC plate were dead and colonies were visible in the four plates with transfected cells. Colonies were picked using cloning disks

(Thermofischer) and transferred to wells of a 24-well plate. After being grown to confluency and passaged twice, each of the lines were tested for Venus expression by RT

PCR. At the end of this process the cell lines generated constitutively expressed the

Venus transcript either with the Wt or Mut Lfng 3'UTR. Three clones were selected for each stable cell line.

2.2.3 RT-PCR and Quantitative RT-PCR

RT-PCR

RNA extraction was carried out using the Zymo Quick Miniprep kit. The RNA isolated was reverse transcribed to cDNA with the SuperscriptIII Reverse transcriptase kit following manufacturer instructions (Invitrogen). Expression of the Venus transcript was detected by PCR amplification using the Venus primers that go across an intron.

Expression of GAPDH was used as a control for amount of cDNA per sample. The primers used were as follows: Venus, forward 5’ ACACGCTGAACTTGTGGC 3’ and reverse 5’

CTCCGGATCGATCCTGAGAA 3’; Gapdh, forward 5’ GGTGCTGAGTATGTCGTGGAGT 3’ and reverse 5’ GGGCGGAGATGATGACCCTT 3’. The PCR products were analyzed on a 2% agarose gel.

Quantitative RT-PCR

QPCR was carried out for each cDNA sample, in triplicate, with the SYBR Green

Master Mix (Applied Biosystems) in a Step-One ABI machine using the following conditions: 50 ⁰C for 2 mins, 95 ⁰C for 10 mins and 40 cycles at 95 ⁰C for 15 secs and 60

⁰C for 1 min; followed by a melt cycle of 95 ⁰C for 15 secs, 60 ⁰C for 1 min, 95 ⁰C for 15 secs and 60 ⁰C for 15 secs. Venus and Gapdh primer concentrations were optimized to get the highest PCR efficiencies for both primer sets. The primers used were the same as the ones used for RT PCR. This assay was repeated three independent times.

2.2.4 Venus transgenic mice

The LvLWt and LvLMut was separated from the vector backbone using the restriction enzymes SpeI and ClaI (NEB) and sent to the Mouse Modeling Shared

Resource at the Ohio State University (OSU) to generate the LvLWt and LvLMut transgenic mouse lines that were maintained on the FVB/J background. The founder mice obtained after transgenesis were analyzed for the presence of Venus transgene by

PCR-genotyping using the following primers, forward 5’ CGAGGAGCTGTTCACCGG 3’ and reverse 5’ CGTGCTGCTTCATGTGGTCG 3’ which gives a band size of 200 bp. Embryos were collected from timed pregnancies, where the day of plug identification was assigned as

0.5 days post copulation (dpc). Embryos were collected from Venus positive founders and used to analyze Venus transcript expression by RT PCR and in situ hybridization. Two transgenic lines for LvLWt (LvLWt1 and LvLWt3) and two for LvLMut (LvLMut3 and

LvLMut4) that showed detectable amounts of Venus expression in the PSM were used for further analysis of transcript stability. Mice were maintained under IACUC supervision with approved animal use protocol.

[Note: We also analyzed previously made GFP mouse lines with a wildtype or mutated Lfng 3'UTR but due to a caveat in the transgene design, the results from the

GFP lines were disregarded. This is further discussed in Appendix B: Supplemental figures]

2.2.5 Whole mount in situ hybridization

For whole mount in situ hybridization 8.5 dpc and 10.5 dpc embryos were collected in cold PBS and fixed with 4% PFA overnight. In situ hybridization was performed for

Venus and Lfng (Johnston et al. 1997) on embryos from the LvLWt and LvLMut lines as mentioned previously using digoxigenin (DIG) or fluorescein (FLU) labeled dUTP mRNA probes(Shifley et al. 2008). For the Venus probe, the Venus-PEST region was amplified from the LvLWt vector using forward primer 5’ CGAGGAGCTGTTCACCGG 3’ and reverse primer 5’ TCTAGACTACACATTGATCCTAGC 3’ , and cloned into and transcribed from

TOPO vector pCR2.1™ (Invitrogen). The target mRNA-probe hybrids were detected by a chromogenic method using an anti-DIG antibody or anti-FLU conjugated to alkaline phosphatase (AP) (Shifley et al, 2008).

Expression of Lfng (Johnston et al. 1997) and Uncx (Mansouri et al. 1997) was analyzed in heterozygous and homozygous miR-125a mutants and in wildtype controls using mRNA probes as described previously. At least 10 embryos were analyzed for each mutant.

2.2.6 Expression profile data analysis

Oscillatory expression profiles were made by transforming in situ images of 10.5 dpc embryos into heat maps using the FIJI software. Each image was converted to black and white and the plot profile of one half of the PSM from the most recently formed somite to the tip of the PSM was generated and saved as a .csv file (‘comma separated values’ file) in Microsoft Excel. This file was then imported into FIJI as a text image and pseudo- colored by choosing the LUT as ‘physics’ where the maximum intensity is coded as red and minimum intensity is coded as dark blue. Similarly, expression profiles were made to depict expression patterns in the anterior PSM of LvLWt and LvLMut lines, i.e. the

51 region from the most recently formed somite till the posterior end of the neural tube.

The difference in the expression profiles of LvLWt and LvLMut lines were assessed by the Fischer exact test, with p<0.05 being considered as significant.

2.2.7 CRISPR/Cas9 plasmid and guide RNAs to target the miR-125a locus

The CRISPR design tool (http://crispr.mit.edu/) was used to design oligos for guide

RNAs that will bind to the miR-125a locus in the mouse genome(Yang et al. 2013). The following two oligo pairs were chosen because they had the lowest number of off-target hits.

Oligo pair 1 (O1) 5' caccGTCCACCATAGCTACACTGC 3' 3' CAGGTGGTATCGATGTGACGcaaa 5'

Oligo pair 2 (O2) 5' caccGGACGTCCTCACAGGTTAAA 3' 3' CCTGCAGGAGTGTCCAATTTcaaa 5'

Table 2.4 Oligos pairs used to target the miR-125a locus in the mouse genome. The BbsI restriction site is shown in small letters.

The complementary oligos with sticky ends for the BbsI enzyme were annealed to each other at equimolar concentrations of 100uM. They were mixed and placed in a heat block at 95°C for 5 mins and then the heat block was moved to the bench and allowed to gradually cool to room temperature (about 45 mins). Each annealed oligo was cloned into the CRISPR/Cas9 plasmid, pX330-U6-Chimeric_BB-CBh-hSpCas9 was a gift from Feng Zhang (Addgene plasmid # 42230), using the BbsI site. This plasmid carries the Cas9 enzyme and a mammalian U6 promoter that will drive the expression of

52 guide RNA that contains the oligo sequence or CRISPR RNA that will target the miR-125a locus.

2.2.8 Generation of miR-125a mutant mice

The plasmids pX330-O1 and pX330-O2 were sent to the Mouse Modeling Shared

Resource at OSU to generate the mutant mice. The plasmids were injected in circular form into fertilized eggs as described(Fujihara & Ikawa 2014). The founder mice generated were screened for mutations at the miR-125a locus by High Resolution Melt

Analysis discussed below. By PCR-based genotyping we also made sure that the

CRISPR/Cas9 plasmid didn’t insert anywhere in the genome. The mice were created on the C57BL6 background and successively backcrossed onto FVB/J. Analyses used embryos from the N3 generation, and no changes in phenotype were observed in crosses from the N5 generation. Mice were maintained under IACUC supervision with approved animal use protocol.

2.2.9 High Resolution Melt Analysis (HRMA)

Primers used to analyze the miR-125a locus in founders and wildtype control DNA by

HRMA were as follows, forward 5’ CCTCTGGGGAAAAGGGTTTT 3’ and reverse 5’

CTGAAATCCCTAAATTTGTGGC 3’ and the PCR reaction was set up as done previously(Talbot & Amacher 2014). The PCR was run on the CFX BioRad machine using the following program: 95˚C 3 mins, (95˚C 15 sec, 58˚C 20 sec, 70˚C 20 sec) 45x, 65˚C 30 sec, melt 65-95˚C, 95˚C 15 sec. Based on their melt curve profiles, each sample got clustered either in the same group as the wildtype control or a separate group by the 53

Precision Melt Analysis software.

2.2.10 Sequencing and genotyping of the miR-125a mutants

The miR-125a locus in three of the mutants identified by HRMA was amplified by

PCR and sequenced using the following primers: forward 5’GCTTAGGGTATCTGTTTCTG 3’ and reverse 5’GAGGAGAAGATAGTGACCTT 3’ to identify sequence changes (Talbot et al.,

2015). Genotyping of Mutant 1 (11bp deletion) and Mutant 2 (Insertion of 385 bp) embryos was done by PCR using the primers mentioned in Section 2.2.9 and analyzed on a 4% and 1% agarose gel, respectively. Mutant 3 (3bp deletion resulting in loss of MseI restriction site) was genotyped using the sequencing primers mentioned in this section followed by a digestion of the PCR product with MseI.

2.2.11 qRT-PCR for miR-125a expression in mutant mice

RNA was collected from 9.5 dpc embryos homozygous for the mutation at the miR-

125a locus. Three embryos were analyzed for each mutant and wildtype littermates were used as controls. Real time PCR was carried out using TaqMan® microRNA assays

(Applied Bioscience) with specific primers to analyze the expression of the mmu-miR-

125a-5p (2198), mmu-miR-99b (436) and mmu-let-7e (2406) and sno-135 and sno-234 were used for normalization. Results indicate mean +/- SEM from three samples for each mutant/wildtype done in triplicate and significance was calculated using Student’s

T test.

2.2.12 Testes histology and analysis of sperm morphology

Testes were isolated from wildtype and homozygous mutant males and fixed with 54

Bouin’s fixative for 24 hours. This was followed by ethanol dehydration and paraffin embedding followed by sectioning the paraffin blocks as done previously (Borg et al.

2009). For hematoxylin and eosin staining (H&E) 15µM sections were used.

Sperm from wildtype and homozygous mutant males were isolated to assess differences in sperm morphology (Yu et al. 2006). Briefly, the epididymis was isolated and placed in PBS. The tissue was macerated to release spermatozoa. Few drops of this solution were placed on a coverslip and a slide was lowered facedown onto the coverslip and flipped. The excess liquid was removed with a kimwipe and the slide and coverslip were sealed using clear coat nail polish. The spermatozoa were visualized using the 60X oil immersion objective.

2.3 Results

2.3.1 miR-125a destabilizes Venus transcript with the Wt 3’UTR

To confirm that exogenous transcripts with the Lfng 3'UTR respond to miR-125a we made Venus reporter constructs that include a -globin intron at the 5’end of the Venus coding region (Figure 2.3). These constructs are driven by a constitutive promoter (PGK or CMV) and contain the Wt or Mut Lfng 3'UTR as shown in the top panels of Figure

2.3A and 2.3B. When these constructs were transiently expressed in 3T3 cells we found that transcripts with the Wt 3’UTR exhibited a reduction in Venus expression in the presence of the pre-miR-125a. This effect was abolished when the miR-125a binding sites were mutated in the 3’UTR (Figure 2.3A, bottom panel). When we measured the

Venus expression in C2C12 cells that stably express these constructs (Figure 2.3B, top panel) we found a similar result. We found a significant difference in expression of

Venus transcripts with the Wt Lfng 3'UTR between the Scrambled control and miR-125a expressing cells. On the other hand, we found no difference in expression of Venus transcripts with the Mut Lfng 3'UTR between the Scrambled control and miR-125a expressing cells (Figure 2.3B, bottom panel). This indicates that miR-125a destabilizes transcripts directly via the miR-125a binding sites in the mouse Lfng 3'UTR. We consistently observed lower expression of Venus transcripts with the Wt 3’UTR, irrespective of the exogenous miR-125a. This could possibly be due to endogenous miR-

125a and other regulatory factors expressed in these cells.

Figure 2.3: miR-125a destabilizes transcripts via the Lfng 3’UTR. A. 3T3 cells transiently transfected with either a Scrambled control or pre-mir-125a, and Venus constructs, driven by the constitutively active PGK promoter. RT-PCR (red arrows in top cartoon depict primers used for PCR) for Venus transcript either with the wildtype (Wt) Lfng

3’UTR or mutated (Mut) 3’UTR in which the miR-125a binding sites have been mutated.

B. C2C12 cells stably expressing Venus constructs with a wildtype or mutated Lfng

3’UTR. Real time PCR for Venus transcript expression in these cells that are transiently transfected with either a Scrambled control or pre-mir-125a. The expression was normalized to GAPDH. Error bars = SEM from three independent experiments, with each sample carried out in triplicate. Statistical significance calculated using Student’s T test.

2.3.2 The LvLMut mouse lines exhibit stabilized Venus expression in the caudal PSM

After validating the functionality of the Venus constructs in cell culture, the LvLWt and LvLMut constructs, depicted in Figure 2.4, were used to make transgenic mice. We found that LvLWt lines (LvLWt1 and LvLWt3) exhibit oscillatory Venus expression patterns that matched the expression of Lfng in the mouse PSM both at 8.5 dpc (data not shown) and 10.5 dpc, as expected (Figure 2.4A and 2.4B). In LvLMut embryos Venus expression does not match the expression of endogenous Lfng and is consistently seen in the caudal most or tail bud region of the PSM in all LvLMut embryos at 10.5 dpc

(Figure 2.4C).

To further test if the exogenous Venus transcript in the LvLWt line oscillates in phase with the endogenous Lfng transcript we carried out double in situ hybridization for both transcripts in the same set of embryos. In control non-transgenic embryos Lfng expression can be visualized by the orange precipitate but no Venus expression (purple) is detected, as expected (Figure 2.5A). In embryos expressing the transgenes, Venus expression was detected by a purple precipitate and Lfng expression was detected by an orange precipitate; if the expression of the two transcripts overlapped it would be detected as a brown color. In the LvLWt3 line, we observed a complete overlap of the two colors resulting in a brown color in all three phases of oscillatory expression for

Venus and Lfng (Figure 2.5B). This indicates that the LvLWt line is an ideal proxy for oscillatory Lfng expression in the mouse PSM.

We also performed double in situs for Venus and Lfng expression in LvLMut4 lines as mentioned above (Figure 2.5C). Venus mRNA is consistently detectable in the caudal region of the PSM of the LvLMut line and does not always overlap with Lfng expression.

This is especially apparent in embryos that are in phase II of oscillatory Lfng expression

(orange) but display detectable levels of Venus expression in the tail bud (purple; Figure

2.5C,a).

Figure 2.4: Venus expression in LvLMut lines does not match endogenous Lfng expression or Venus expression in the control LvLWt lines. In situ hybridization of Lfng or Venus transcripts, with the Wildtype or Mutated Lfng 3’UTR, in the mouse PSM of

10.5 dpc embryos A. PSMs of 10.5 dpc wildtype embryos exhibit normal oscillatory expression of endogenous Lfng mRNA. B. Wildtype UTR: Venus transcript exhibits oscillatory expression in the PSM of the two wildtype transgenic lines that (continued)

(Figure 2.4 continued) matches expression of normal Lfng 3’UTR. C. Mutated UTR:

Venus expression is consistently seen in the caudal PSM of both lines of the Venus transgenic mice with the mutated Lfng 3’UTR.

Figure 2.5: Venus expression in the mouse PSM of LvLMut embryos does not overlap with endogenous Lfng. Double in situ hybridization to compare expression patterns of

Lfng and Venus mRNA in LvL mice. A. Lfng expression using a fluorescein probe (orange) in Venus negative embryos B. In Venus positive LvLWt embryos expression of Venus matches that of endogenous Lfng transcript. C. In Venus positive LvLMut embryos in contrast to Lfng expression, Venus mRNA appears to be consistently present in the posterior PSM.

2.3.3 LvLMut embryos exhibit oscillatory expression in the anterior PSM

To display the oscillatory expression profile for Lfng in wildtype embryos and Venus in transgenic embryos we created expression profiles for the PSM region of all embryos used for in situ hybridization in Section 2.3.3. This was done by selecting the region of the PSM from the most recently formed somite till the bottom most post of the PSM and then using the FIJI software to create a heat map. The oscillatory expression profile of endogenous Lfng mRNA is similar to the Venus expression profile in the LvLWt line but not to the LvLMut4 line (Figure 2.6). Using the two-tailed Fischer exact test we compared embryos in the LvLWt and LvLMut lines that display high Venus expression in the caudal one-third part of the PSM (region posterior to the neural tube) and found a statistically significant difference between the expression patterns of (with a p value of

0.0001). Since the caudal PSM of LvLMut lines exhibits high Venus expression, we were concerned that it may be obscuring oscillations in the anterior PSM. So we repeated the analysis but excluded the caudal one third region of the PSM for each expression profile.

Interestingly, we find that when expression profiles were made for the anterior PSM of

LvLWt3 and LvLMut4 lines oscillatory Venus expression was seen in both lines. These results suggest that mutating the miR-125a binding sites in the Lfng 3'UTR disrupts mRNA regulation specifically in the caudal or tailbud region of the PSM but not the rest of the PSM. This suggests that the miR-125a binding sites may be regulating transcript stability in a defined region rather than throughout the PSM.

Figure 2.6: Oscillatory expression profile of transcript expression over time along the

A-P axis of the mouse PSM. A. Expression profile of endogenous Lfng mRNA in the PSM of wildtype embryos. B. Oscillatory expression patterns of Venus transcript in LvLWt mice are comparable to the endogenous Lfng mRNA. C. Venus expression in the PSM of

LvLMut mice appears to be stabilized in the caudal region. The expression in the anterior

PSM (PSM without the caudal one-third region), however, continues to oscillate similar to Venus expression in the anterior PSM of LvLWt mice shown in B.

2.3.4 Creating miR-125a mutant mice using the CRISPR/Cas9 system

Since our results suggest a possible role for miR-125a in the mouse segmentation clock, we decided to create miR-125a knockout mice using the CRISPR/Cas9 system which is a RNA based genome editing method (Figure 2.7) (Yang et al. 2013; Fujihara &

Ikawa 2014). miR-125a is present in the mouse genome closely clustered with two other miRNAs, miR-99b and let-7e. The genome editing approach is ideal in this context because we want to make small mutations at the miR-125a locus without affecting the expression of the other two miRNAs in the cluster. We employed two strategies to do this: one was to target the region just upstream of the mature miR-125a-5p sequence using the guide RNA O1 and the other was to target the mature miR-125a using O1 and

O2 (Figure 2.8).

Using High Resolution Melt Analysis (HRMA) we screened the founder mice and identified four mutants (Figure 2.8). These mutations were sequenced and we found that Mutant 1 has an 11 bp deletion (mir-125a-5p∆11 or Del11) and Mutant 2 has a 385 bp insertion (mir-125a-5pins or INS), both just upstream of the mature miR-125a sequence. Mutant 3 has a 3 bp deletion (mir-125a-5p∆3T or Del3T) in the mature miR-

125a-5p sequence while Mutant 4 has a 2 bp substitution upstream of the mature miR-

125a sequence (not shown). We focused on the Mutants 1, 2 and 3 for the study (Figure

2.9).

Figure 2.7: The CRISPR-Cas9 system utilized to make knockout mir-125a mouse. The guide RNA contains a 20bp that binding to the genomic locus of interest. This is followed by recruitment of wildtype Cas9 nuclease to this site. Cas9 creates a double stranded break in the DNA which is then repaired by Non-homologous end joining (NHEJ). (Shao et al. 2014)

Figure 2.8: Identification of four miR-125a mutants by High Resolution Melt Analysis

(HRMA). A. Oligo1 was used to bind to the region upstream of the mature miR-125a.

Two founders were identified by HRMA. B. Oligo1 and Oligo2 were used in an attempt delete miR-125a.Mutants 3 and 4 were identified by HRMA. The melt curves for all the other founders, denoted in red (in 2.8 A and B), clustered with the wildtype control for the miR-125a locus.

Figure 2.9: Genotypes of mutant miR-125a founders. A. Wildtype miR-125a B. Mutant1 has an 11bp deletion C. Mutant2 has an insertion of 385bp D. Mutant3 has a 3bp deletion in the mature sequence.

2.3.5 miR-125a-5p expression is affected in all three miR-125a mutants

miRNAs are transcribed by RNA Pol II as capped and tailed pri-miRNAs which is cleaved into the 65 bp stem-loop pre-miRNA structure by DGCR8 and DROSHA, after which it is exported to the cytoplasm (Figure 2.10A). In the cytoplasm it is further processed by Dicer1 to result in a mature miRNA (Filipowicz et al. 2008; Ha & Kim 2014).

Since miRNA processing has predominantly been studied in vitro using synthetic miRNA precursors, the cleavage sites and the mechanism by which the proteins act is not completely understood and may not apply for all miRNAs (Zeng et al. 2005; Han et al.

2006). Due to these limitations in our knowledge of miRNA biogenesis, we were not certain if the mutations we created would affect miR-125a expression and if they would affect the expression of the other two miRNAs, miR-99b and let-7e, in the cluster as well.

We used the RNA mfold software to compare Del11 mutant and wildtype pri-miR-

125a structures and found that the stem of pri-miR-125a in Del11 mutant is shorter than the wildtype (Figure 2.1oB; Zuker 2003). While this doesn’t directly affect the pre- miR-125a structure, it may disrupt the DGCR8 and DROSHA recognition sites and affect the production of pre-miR-125a(Zeng et al. 2005; Han et al. 2006). In Mutant 3 (Del3T), since there is a 3bp deletion in the mature miR-125a-5p sequence we compared the pre-miR-125a structures in the Del3T mutant and wildtype using the RNA mfold software (Figure 2.10C; Zuker 2003). The pre-miR-125a structure is different in the Del3

68 mutant as the bulge is on the opposite side which may affect the Dicer1 directed cleavage and production of mature miR-125a-5p.

To assess if the mutations have an effect on the expression of miR-125a, quantitative PCR analysis was performed for miR-125a expression in the three 9.5 dpc homozygous mouse embryos each and compared to the wildtype control (Figure 2.11A).

The Del11 and Del3T mutants showed complete loss of miR-125a expression indicating that the altered pri-miR-125a and pre-miR-125a structures disrupted miRNA production.

In the INS mutant there was a drastic reduction in miR-125a expression possibly because all the cleavage sites for miR-125a processing are still intact in this mutant. We also tested the expression levels of miR-99b and let-7e in the mutants and found no significant difference between the mutants and wildtype controls (Figure 2.11B and

2.11C). Thus, using the CRISPR/Cas9 system we have created one miR-125a knockdown and two miR-125a knockout mutant strains.

Figure 2.10: Structures of miR-125a precursors in Del11 and Del3T mutants. A.

Overview of miRNA biogenesis B. pri-miR-125a structure differs in Del11 mutant as compared to the control C. pre-miR-125a structure differs in Del3T mutant as compared to the control.

Figure 2.11: miR-125a expression is lost in Delta11 (n=3) and Delta3T (n=3) mutants and reduced in insertion (n=3) mutants as compared to the Wt controls (n=3). A. miR-

125a expression in mutants and controls, B. and C. There is no difference in miR-99b and let-7e expression in mutants and controls. Normalization done using sno-135 or sno-

234 expression. Error bars = SEM from three samples examined in triplicate. Statistical significance calculated using Student’s T test.

2.3.6 Lfng and Uncx expression are unaffected in miR-125a mutants

The consistent expression of Venus in the caudal PSM of transgenic mice where the

Lfng 3’UTR has mutated miR-125a binding sites would suggest that loss of miR-125a would result in a similar pattern of consistent expression for Lfng in the caudal or tailbud region of the mouse PSM. To test if loss of miR-125a has an effect of Lfng oscillations in the mouse PSM, we performed in situ hybridization for Lfng on 8.5 dpc

(data not shown) and 10.5 dpc embryos (Figure 2.12). Oscillations of Lfng were unaffected in the PSM of Del11 and Del3T mutants. These results suggest that miR-125a is not required for the maintenance of Lfng oscillations in the mouse PSM.

Since Lfng expression is required for normal somite patterning, we wanted to examine if loss of miR-125a would impact the identity of somite compartments. We carried out in situ hybridization for Uncx, a marker for the posterior somite boundary. In heterozygous and homozygous mutants, we found no difference in Uncx expression between any of the homozygous mutants and the wildtype controls (Figure 2.13; heterozygous mutants, data not shown). This result does not support the idea that loss of miR-125a has negative effects on somite patterning.

A previous study that examined vertebral segmentation in double heterozygotes of

Notch1 and DLL3 found normal Uncx expression in the mutant embryos but minor axial skeletal abnormalities in the adults(Loomes et al. 2007). Similarly, we wanted to assess if the loss of miR-125a could have resulted in any subtle vertebral defects, we compared the skeletal morphology in fetuses of Del11 and Del3T mutants to that of their wildtype

72 littermates (Figure 2.14). We found no difference in the number and morphology of ribs and vertebrae between the mutants and wildtype.

Taken together these results suggest that miR-125a is dispensable in the context of the mouse segmentation clock.

Figure 2.12: Lfng expression is unaffected in miR-125a mutants. Three phases of Lfng expression (purple, top panel) and an oscillatory expression profile (bottom panl) observed in the PSM of A. Wildtype B. Homozygous Del11 C. Homozygous Del3T 10.5 dpc embryos. Oscillatory Lfng expression is unaffected by the loss of miR-125a. A similar result was seen in mutant heterozygotes.

Figure 2.13: Somite number and posterior somite boundaries (Uncx) are unaffected in miR-125a mutants. Uncx in situ on 10.5 dpc whole embryo and close-up of posterior somites for A. Wildtype control B. Homozygote Del11 mutant C. Homozygote INS mutant D. Homozygote Del3T mutant. We observed no difference in Uncx expression between the wildtype controls and the three miR-125a mutants indicating that loss of miR-125a may not have an impact on somite patterning.

Figure 2.14: Skeletal development is not impaired by the loss of miR-125a. Skeletal staining was performed on 17.5 dpc mouse fetuses of A) Wildtype B) Homozygous Del11 mutants and C) Homozygous Del3T mutants, using Alcian blue to stain cartilage and

Alzarin red to stain bone. Normal staining patterns were observed in both the Del11 and

Del3T homozygous mutants as compared to the wildtype controls. There was no difference in the number of thoracic vertebrae and ribs in the mutants as compared to the controls.

2.3.7 miR-125a-5p∆11/∆11 male mice are infertile

Since miR-125a is knocked out throughout the organism we decided to examine other tissues that may be affected by the loss of miR-125a. We observed a male sterility phenotype in the miR-125a-5p∆11/∆11 male mice. While these mice do mate based on evidence of copulatory plugs, none of the mating events resulted in any offspring.

Interestingly, a previous study found that miR-125a is enriched in male primordial germ cells but not in female primordial germ cells which justifies our finding as we only observe the infertility phenotype in male mice but not female mice (Hayashi et al. 2008;

Mciver et al. 2012).

To analyze this further, we carried out H&E staining on testes sections from the infertile male mice but found no dramatic differences between wildtype (Figure 2.15A) and Del11 mutant testes (Figure 2.15B). Next, we isolated spermatozoa from each of these mice and imaged them with a phase-contrast microscope. We found that while sperm from wildtype mice have normal morphology with a head, mid-piece and straight tails (Figure 2.16A), sperm from the Del11 mutant display a sharp bend in their tails, particularly at the site of the cytoplasmic droplet (Figure 2.16B). The cytoplasmic droplet is normally seen during the process of sperm maturation in the epididymis (Xu et al.

2013). As the sperm moves along the epididymis (from caput to corpus epididymis) the cytoplasmic droplet migrates from the head of the sperm to the annulus where it is finally shed off to result in a mature sperm(Cooper 2005; Xu et al. 2013; Ijiri et al. 2011).

In the Del11 mutant sperm the sharp bend seen in tails at the site of the cytoplasmic

76 droplet which has been documented previously in the Aquaporin3 mutant, where loss of the sperm water channel protein Aquaporin3 results in swelling of cytoplasmic droplet and hampers sperm motility (Chen et al. 2011). This indicates that similar to the

Aquaporin3 mutant, loss of miR-125a may have an effect on the osmoadaptation, which leads to defective sperm morphology and could be the cause for male sterility seen in the miR-125a-5p∆11/∆11 mutants.

Figure 2.15: The testes of infertile Del11 mutant mice do not exhibit any dramatic morphological defects. H&E staining of seminiferous tubules in section of A. Wildtype testes at 5X (a) and 20X (b) magnification. B. Del11 homozygous mutant testes at 5X (a) and 20(X) magnification (b).

Figure 2.16: Abnormal sperm morphology in Del11 homozygous mutants. A. Sperm from wildtype mice have straight tails. B. Some of the sperm from infertile Del11 mutant mice have bent tails, specifically seen at the site of cytoplasmic droplet (a and b) or abnormally shaped head (c and d), marked by the red arrow heads. C. Sperm defects seen in wildtype and Del11 mutant littermates. D. Larger percent of abnormal sperm observed in mutants as compared to the wildtype littermate.

2.4 Discussion

Our results suggest that miR-125a binding sites in the Lfng 3'UTR are required to destabilize the transcript in the caudal most or tail bud region of the mouse PSM but have no dramatic effect on transcript stability in more anteriorly placed cells in the PSM.

Despite the effects we observed on the reporter transcripts in the Venus transgenic mice, loss of miR-125a has no effect on Lfng oscillations in the PSM and does not hamper segmentation or axial skeletal formation. We also report a novel, indispensable role for miR-125a in the context of male fertility.

The model depicted in Figure 2.17, based on the Venus transgenic data, suggests a possible role for miR-125a in regulating oscillatory expression of Lfng mRNA in the mouse PSM. As discussed in Section 2.1, Lfng oscillations in cells in the posterior PSM are faster but period of oscillation slows in cells positioned more anteriorly in the PSM

(Figure 2.16A). When the miR-125a binding sites are mutated in the Lfng 3’UTR, the oscillations are not evident in the posterior region, possibly due to perturbed post- transcriptional regulation of the transcript. It is possible that Venus expression is stabilized in the caudal PSM of LvLMut embryos as it is always detectable in caudal region and hence masks the oscillations that are initiated at the transcriptional level

(Figure 2.16B). miR-125a could be controlling the pace of oscillations particularly in the caudal or tailbud region of the PSM (Figure 2.16C and 2.16D), where it is expressed and seems to affect transcript stability by binding to the Lfng 3'UTR (Riley et al. 2013).

Figure 2.17: miR-125a binding sites in the Lfng 3’UTR may affect the period of Lfng oscillations in the mouse PSM. A. In the presence of the wildtype Lfng 3’UTR, the transcript oscillates faster in the posterior PSM as compared to the anterior PSM. B.

When the miR-125a binding sites are mutated in the Lfng 3’UTR, the transcript is stabilized in the posterior PSM. C. miR-125a could regulate Notch signaling by affecting the pace of the clock in the posterior PSM. D. This is supported by the evidence that mir-

125a expression is highest in the posterior PSM.

The distinction between the posterior and anterior region is evident from the fact that the segmentation clock is faster in posterior PSM and slows down as cells move anteriorly (Palmeirim et al. 1997; Gibb et al. 2010). However, the mechanism behind this difference is unclear. Our results reinforce the difference between the two regions since transcript with the mutated miR-125a binding sites in the Lfng 3’UTR shows loss of oscillatory expression only in the posterior region. This suggests that a regulatory factor such as the miR-125a could bind to Lfng 3'UTR and ensure faster transcript turnover in the caudal region but that in anterior regions, alternative mechanisms regulate transcript turnover. This model would be consistent with previously published work that found miR-125a is expressed in the caudal most region of the mouse PSM (Riley et al.

2013; Figure 2.6D). Thus the reporter transgene analysis supports a model that miR-

125a binding sites in the Lfng 3'UTR regulate expression of Lfng in cells in the tail bud region and as cells leave this region the Lfng transcript is probably regulated by other cis-acting elements in the 3’UTR. This would result in a difference in turnover rate of

Lfng mRNA between the posterior and anterior regions and may explain why Lfng oscillations appear slower in anteriorly positioned cells in the PSM. It is also likely that regulatory factors that are expressed more anteriorly in the PSM may be targeting the

Lfng 3'UTR to maintain oscillatory expression of the transcript in cells that have exited the tail bud. While it is possible that the stabilization seen in our LvLMut mice could be an artifact of the transgene integration site, this concern is mitigated to an extent by the

82 fact that we have two LvLMut lines that show similar patterns of Venus expression in the

PSM.

Previous work has shown that miR-125a-5p is required for normal somite formation in chicken embryos(Riley et al. 2013) and our reporter transgene data suggests that the conserved miR-125a binding sites may function in post-transcriptional regulation of an exogenous transcript. Despite this, we find that loss of miR-125a-5p expression in mice does not stabilize Lfng expression in the caudal PSM and has no effect on mouse somitogenesis. There are several possible models that could reconcile our apparently contradictory findings. One possibility is that miR-125a:Lfng interactions could point towards a species specific method of regulation that is important in chickens but not mice. In this case, both miR-125a and the conserved sites are dispensable for somitogenesis in mice, and the transgene effects observed may be artifacts. This model is less convincing based on our finding that multiple independent transgene lines had similar defects in Venus expression in the caudal region. Another possibility is that the miR-125a:Lfng interactions are conserved and important but in the miR-125a knockout mice there is either a redundant mechanism or something else gets upregulated to compensate for the chronic loss of miR-125a. It is also important to note that in the study with chicken embryos loss of miR-125a was acute while in this study loss of miR-

125a in mice is chronic and could trigger upregulation of other pathways to counteract for the absence of miR-125a (Tatsumi et al. 2015). This has been documented by a previous study where complete loss of miR-125a in mice resulted in upregulation of

83 another miRNA that functionally compensated for the loss of miR-125a in the mutant mice (Tatsumi et al. 2015). miR-125b and miR-351, which have the same seed sequence as miR-125a, are expressed in the mouse PSM though at much lower levels. It is possible that in miR-125a mutant embryos there is an increase expression of these two miRNAs which could help maintain Lfng oscillations in the PSM and result in normal somite formation. This could explain the discrepancy we observe between our transgene and miR-125a mutant mouse data. A final possibility is that factors other miR-125a may be binding to the sequences that we refer to as ‘miR-125a binding sites’ and destabilizing the transcript, which would make miR-125a dispensable in the context of mouse segmentation.

Based on our results we would predict that mutation of miR-125a binding sites in the endogenous mouse Lfng 3'UTR would result in stabilization of Lfng expression in the tail bud region and this effect could be independent of miR-125a regulation. To speculate the functional consequence of mutating the endogenous Lfng 3'UTR it is important to first discuss the properties of the tail bud region of the PSM. Cells in the tail bud region are unique from the rest of the PSM due to their stem-like properties such as multipotency, as they can give rise tissues of neural and mesodermal identity

(Wolpert, Lewis ; Tickle, Cheryll; Martinez Arias 2015). Another important characteristic of the tail bud is its role in secondary body development (posterior skeleton) which includes the region from the lumbosacral vertebrae till the tip of the tail. Primary body development (anterior skeleton) which includes head and trunk is formed from the

84 primitive streak and epiblast cells with stem-like properties and antecedents of the tail bud cells (reviewed in Stern et al. 2006; Wolpert, Lewis ; Tickle, Cheryll; Martinez Arias

2015)). While there are similarities in the formation of the primary and secondary body plan, the mechanisms that govern their formation are distinct especially in mammals than in other organisms (Handrigan 2003).

It has been shown previously that primary and secondary body formation have different requirements for oscillatory Lfng expression. The anterior skeleton is found to be more sensitive to the total amount of Lfng in the mouse PSM while the posterior skeleton can form normally even at low levels of cyclic Lfng in the PSM (Shifley et al.

2008; Williams et al. 2014). This can help us speculate the functional consequence of mutating the miR-125a binding sites in the endogenous mouse Lfng 3’UTR. Due to the increase in stability of the Lfng mRNA we predict an increase in LFNG protein expression that would affect the morphology of initially formed, anterior somites. As somitogenesis progresses, the LFNG protein levels would cross the threshold to be able to inhibit

Notch signaling and in turn reduce its own expression. This reduction in Lfng expression along with other factors that may destabilize the Lfng transcript appropriately in cells exiting the tail bud could result in normal morphology of posteriorly positioned somites that are responsible for the secondary body formation.

Despite the fact that miR-125a is dispensable for normal axial skeletal development, we identified a previously unknown developmental role for mir-125a. miR-125a is expressed in male primordial germ cells and thought be involved in

85 spermatogenesis by regulating expression of LIN28, that is found to be upregulated in testicular teratomas (Hayashi et al. 2008; Zhong et al. 2010). Our study would be the first to demonstrate, using a mouse model, that loss of miR-125a has an impact on male fertility by affecting the morphology of the sperm. We observe that the male sterility phenotype is clearly associated with the Del11 mutant and we would have to examine male mice from other mutant lines (Del3T and Ins) to confirm the role of miR-125a in this context.

In conclusion, we have data from reporter transgenes supporting the idea that miR-

125a binding sites in the Lfng 3'UTR regulate transcript turnover in the caudal region of the mouse PSM. However, loss of miR-125a has no effect on Lfng oscillations and the mouse segmentation clock. These results could suggest that other factors apart from miR-125a may be destabilizing the Lfng mRNA in the mouse PSM. This is also evident from work that found several regions of conservations in the Lfng 3’UTR and binding sites for other miRNAs within these regions(Chen et al. 2005; Riley et al. 2013). In the following chapter we investigate the role of other cis regulatory regions in the mouse

Lfng 3'UTR that may be regulate transcript stability during somitogenesis.

CHAPTER 3

Lfng 3’UTR SEQUENCES HAVE COMPLEX EFFECTS ON TRANSCRIPT

STABILITY

3.1 Introduction

Somitogenesis requires oscillatory expression of genes such as Lunatic fringe (Lfng) which is essential for periodic somite formation in chickens, mice and humans(Zhang &

Gridley 1998; Evrard et al. 1998; Sparrow et al. 2006). Regulation at the transcriptional level initiates oscillations of Lfng mRNA expression but to maintain these oscillations post-transcriptional regulation is likely to play an important role(Cole et al. 2002;

Morales et al. 2002; Serth et al. 2003; Chen et al. 2005). The Lfng 3’UTR is known to destabilize the Lfng mRNA by post-transcriptional regulation(Chen et al. 2005; Hilgers et al. 2005; Riley et al. 2013; Nitanda et al. 2014a). We investigated this using the mouse and chicken Lfng 3'UTR due to presence of a highly conserved 120 bp region with a miR-

200b family binding site in both these species. We find that while exogenous miR-200b family members target mouse and chicken Lfng 3'UTR, blocking the miR-200b binding site in the Lfng 3'UTR in chicken embryos has no effect on somitogenesis. Interestingly, though preventing miR-200b from binding to any of its target mRNAs disrupts

87 somitogenesis in chicken embryos indicating that miR-200 is possibly regulating expression genes other than Lfng that are involved in segmentation. We also show that in mouse myoblast cells the 120 bp sequence of the mouse Lfng 3'UTR is not sufficient to destabilize the transcript. Our results indicate that the 120 bp sequence and the miR-

200b binding site within it, are not sufficient to destabilize the transcript and it is likely that this regulation operates in a tissue specific manner.

3.1.1 Oscillatory Lfng expression is required for the chicken and mouse segmentation clock

Somites bud at regular intervals from an unsegmented tissue in the posterior region of the embryo called the presomitic mesoderm (PSM) in pairs that flank the neural tube.

Somites are the embryonic precursors of the ribs, vertebrae and skeletal muscles in adults(Wahi et al. 2016). This process is governed by a "segmentation clock" comprising genes whose expression oscillates in the PSM with a period that matches the rate of somite formation. Evidence of clock function has been identified in all vertebrates examined to date, though the period of clock gene oscillations varies among species; 90 mins in chicken(Palmeirim et al. 1997), 120 mins in mice(Tam 1981), and 4-6 hrs in humans(William et al. 2007). Perturbations of clock gene expression result in skeletal deformities seen in adults such as spondylocostal dysostosis and kyphosis (Turnpenny et al. 2007; Eckalbar et al. 2012).

Many clock genes have been linked to the Notch signaling pathway including

Lunatic fringe (Lfng) encoding a glycosyltransferase that modifies Notch receptors and

88 ligands(Brückner et al. 2000; Moloney et al. 2000; Munro & Freeman 2000). Lfng transcript and LFNG protein levels oscillate in the mouse or chick PSM with a period that matches the rate of somite formation, suggesting that tight control of mRNA and protein expression may be important for segmentation clock function(Dale et al. 2003).

Oscillatory Lfng expression has been found to be necessary for normal clock function, with either complete loss or constitutive expression of Lfng in mice perturbing clock gene expression and resulting in defects in somites and in turn in the axial skeleton(Zhang & Gridley 1998; Evrard et al. 1998; Serth et al. 2003; Williams et al.

2016). Transcriptional regulation is known to contribute to cyclic Lfng expression (Cole et al. 2002; Morales et al. 2002), but mathematical models suggest that tight post- transcriptional regulation at the RNA and protein level is necessary for stable oscillatory

Lfng expression in the clock(Lewis 2003; Monk et al. 2003; Feng & Navaratna 2007;

González & Kageyama 2007). These models predict the existence of regulatory mechanisms that ensure rapid turnover of Lfng transcripts. In this context, functions of the Lfng 3'UTR in regulating transcript stability have been an area of considerable interest, since the binding proteins and/or small non-coding RNAs to the 3’UTR can affect mRNA export, transcript stability or translation efficiency, all of which could influence Lfng functions in the clock.

3.1.2 The Lfng 3'UTR affects transcript stability

The mouse Lfng 3'UTR has been shown to regulate the stability of exogenous transcripts in a variety of contexts including cell lines and chicken embryos (Chen et al.

2005; Hilgers et al. 2005; Riley et al. 2013; Nitanda et al. 2014a). The Lfng 3'UTRs are quite long, and exhibit relatively high levels of conservation among mouse, chickens and humans, suggesting that specific UTR sequences may be important for post- transcriptional regulation of Lfng. Indeed, a conserved binding site for the miRNA miR-

125a-5p has been shown to function in regulating Lfng stability in the chick PSM (Riley et al. 2013). However, the fact that conserved sequences are found throughout the Lfng

3'UTR (Figure 3.1A) suggests that there may be other regulatory mechanisms, and that some mechanisms may be context dependent.

Here, we examine the possible regulatory role of a highly conserved 120bp region in the middle of the Lfng 3'UTR which contains a potential binding site for miR-200b family of miRNAs. We also investigate the functional significance of the miR200b:Lfng interaction in chicken somitogenesis. The ability of miR-200 family members to target the Lfng 3'UTR has been validated in zebrafish, where this interaction has been suggested to play a role in olfactory neurogenesis (Choi et al., 2008). In this study we observe that although both the mouse and chicken Lfng 3'UTRs can be targeted by miRNAs of the miR-200b/c family, blocking the miR-200b binding site does not perturb somitogenesis in chicken embryos and that 120bp conserved region is not sufficient by itself to destabilize exogenous transcripts in a mouse cell line. These findings indicate that post-transcriptional regulation of Lfng may be context dependent during vertebrate embryogenesis.

3.2 Materials and Methods

3.2.1 miRNA RT-PCR

Tails of 9.5 d.p.c. FVB/NJ mouse embryos were dissected to separate the PSM

(containing mesoderm and ectoderm) from the five most recently segmented epithelial somites (mature somites). Total RNA was extracted from the PSM and mature somites using TRIzol per the manufacturer’s instructions (Invitrogen). RT-PCR was performed using Taqman assays (Applied Bioscience) specific for miRNAs predicted by TargetScan to target the Lfng 3’UTR: mmu-miR-200c, 2300, PN4395411, mmu-miR-200b, 2251,

PN4395362, mmu-miR-204, 508, PN4373094, mmu-miR-211, 1199, PN4373315, mmu- miR-429, 1077, PN4373355, snoRNAs 135 and 234 were used for normalization. RT-PCR was conducted in triplicate on at least three biologically independent replicates. Results indicate mean +/- SEM after normalizing expression levels of the somite samples to 1, and significance was calculated by Student’s T test.

3.2.2 Luciferase assays

Sequences of the Mouse Lfng 3'UTR (mLfng, amplified with primers forward

5’CGGAACTAGTCGTGGTTGAAACTCTGTC3’and reverse

5’CGGATCTAGAGGAAGGACGTCACC 3’) or chicken Lfng 3’UTR (cLfng amplified with primers 5’ GCTCTAGATCGTTGCTGTGGTATTGC 3’ and

5’CGTCTAGAGCTGCTTTATTGGTGACG 3’) or mouse ZEB2 3’UTR (mZEB2, amplified with

5’ GCTGGCACAGTGGCGAGATTCTGC 3’ or 5’ GCTCAGACATGCTAAGTGGTTTTCC 3’) were inserted into pMIRNA-REPORT™ vector (Ambion). PCR-based mutagenesis was used to 91 convert the miR-200 binding sites in the Lfng 3’UTRs to HindIII sites using primers

5’TTGTGAAAGCTTTTTTTTACTGTG 3’ and 5’AAAAAAAAGCTTTCACAATATGTA 3’for the mouse construct and primers for the chicken construct 5’

TTGCGCGAATTCTTTTTTTACTGTGC 3’ and 5’AAAAAAGAATTCGCGCAATAACATAT 3’.

Mutagenesis was confirmed by sequencing.

NIH3T3 mouse fibroblast cells were plated in a 24 well plate at 30,000 cells/well, and transfected with 100ng reporter, 10ng pSVRenilla (Promega). For miRNA responses, with 60nMol of either a Scrambled control (Ambion #17110) or precursor-miR-200b

(Ambion #17100) and locked nucleic acid (LNA) anti-miRs or seedblockers for miR-200b family or negative controls (Table 3.1) were co-transfected. 40 hours post-transfection, luciferase and renilla levels were assayed using the Dual Luciferase Kit (Promega) following manufacturer’s protocols. Experiments were completed three times in triplicate and normalized to expression levels from pMIR-Report. Results were assessed by ANOVA followed by Bonferroni post hoc, with p<0.05 being considered significant.

miRCURY LNA anti-miRs or Sequence Seedblockers Anti-miR-Negative control 5’ AGAGCTCCCTTCAATCCAAA-FL (anti-miR-NC) Anti-miR-200F 5’ CCATCATTACCCGGCTGTATTA-FL

SeedBlocker-NC (SB-NC) 5’ Fl-AATCCAAA

SeedBlocker-200F (SB-200F) 5’ Fl-CAGTATTA

Table 3.1: Sequence of anti-miRs(Choi et al. 2008) and Seedblockers used to block activity of the miR-200b family in the chick PSM 92

3.2.3 In ovo electroporation of chicken embryos

Fluorescein tagged long Anti-miR-200b or short seedblockers that block the miR-

200b/c/429 family or target protectors (Gene Tools, LLC.) that block the miR-200 binding site on Lfng 3'UTR were electroporated in the PSM of Hamburger- Hamilton stage 7-8 chicken embryos as described previously (Dubrulle et al. 2001; Riley et al. 2013). The embryo was visualized by injecting black ink into the yolk underneath the embryo. A piece of vitelline membrane was removed by a tungsten needle to expose the anterior primitive streak. Anti-miRs, seedblockers or target protectors were laid on the anterior primitive streak using a glass capillary. A positive electrode (platinum) was positioned under the embryo, and a minus electrode (tungsten) was put near the Target Protector solution. An electric pulse of 6V, 25 mseconds was charged three times. Embryos were incubated for 24 hrs post-electroporation followed by isolation and analysis. Only embryos that were fluorescein positive in the PSM and somite region were used for further analysis.

3.2.4 In situ hybridization

RNA in situ hybridization was carried out on 10.5dpc mouse embryos for miR-200b using a digoxigenin labeled miRCURY LNA probe (Exiqon) as described previously

(Sweetman et al., 2008, Riley et al., 2013) . In situ hybridization for Uncx or Lfng mRNA was performed on fluorescein positive chicken embryos using digoxigenin labeled mRNA probe using previously described probes and techniques (Riley et al., 2013)

3.2.5 Reporter constructs used to make stable cell lines

The pcDNA3.1 vector was used to make the reporter constructs which contain a constitutive CMV promoter (Invitrogen) and a rabbit -globin intron at the 5’ end, driving expression of a Venus reporter (gift from Dr. Randall Moon) tagged with the

PEST domain (taken from pdEGFP-N1 vector, Clontech). Full-length mLfng 3’UTR sequences were amplified using primers, forward

5’aaTCTAGACAGTCGTGGTTGAAACTCTGT 3’, and reverse 5’ ttATCGATTGGCTGTCCTGGAACTCACTC 3’, while the "cons" 3'UTR was amplified using primers, forward 5’ tctagaGTGGGGTTCAGAGCTATG 3’ and reverse 5’ aagcttCACCTCCTTTCACACCCA 3’. For the "mini" construct sequences corresponding to nucleotides 186 to 1012 were deleted from the full length sequence. Stable cell lines were created by transfection of linearized reporters followed by selections with

800ug/ml of Geneticin (Gibco) as described in Section 2.2.2. Venus expression was confirmed in subclones prior to further analysis.

3.2.6 Transcript half-life analysis

Each stable cell line was plated in 6 wells of a 24 well plate at 30,000 cells per well.

After 24 hrs, cells were treated with Actinomycin D (10ug/ml) for 1 hr and RNA was collected every 30 mins for the next 150 mins (2.5 hrs). RNA extraction (Zymo Quick

Miniprep kit) and cDNA production (Invitrogen SuperscriptIII Reverse transcriptase kit) were done per manufacturer instructions. QPCR was carried out for each sample, in

94 triplicate, using SYBR Green Master Mix in a Step-One ABI machine using the following conditions: 50 ⁰C for 2 mins, 95 ⁰C for 10 mins and 40 cycles at 95 ⁰C for 15 secs and 60

⁰C for 1 min; followed by a melt cycle of 95 ⁰C for 15 secs, 60 ⁰C for 1 min, 95 ⁰C for 15 secs and 60 ⁰C 15 secs. Venus and Gapdh primer concentrations were optimized to get the highest PCR efficiencies for both primer sets. The primers used were as follows:

Venus, forward 5’ ACACGCTGAACTTGTGGC 3’ and reverse 5’ CTCCGGATCGATCCTGAGAA

3’; Gapdh, forward 5’ GGTGCTGAGTATGTCGTGGAGT 3’ and reverse 5’

GGGCGGAGATGATGACCCTT 3’. Assays were repeated in triplicate on three independent biological samples.

3.3 Results

3.3.1 The expression of some members of the miR-200b/c/429 family of miRNAs are enriched in the mouse PSM compared to the somites

Three miRNA families were predicted to target the mouse Lfng 3'UTR by

TargetScan(Friedman et al. 2009; Grimson et al. 2007; Lewis et al. 2005): miRNA-

200bc/429, miRNA-125/351, and miRNA-204/211. Our lab previously identified that miR-125a-5p was enriched in the mouse PSM, and plays a functional role in the regulation of Lfng in the chicken segmentation clock. To ask whether these other miRNAs might target the Lfng 3'UTR in the segmentation clock, we tested if any of these miRNAs were enriched in the mouse PSM, where the clock is active. By QRT-PCR, we found that miR-200b and miR-200c were significantly enriched in the mouse PSM

95 compared to the mature somites (Figure 3.1A), while members of the miR-204/211 family were either not enriched in the PSM or enriched in the somites. Since the tissues used in the RT-PCR analyses contain both mesoderm and ectoderm from the caudal embryo, we utilized in-situ hybridization to examine expression of miR-200b in mouse embryos. In addition to the reported ectodermal expression we detect weak expression of miRNA-200b in the mesoderm, which is enriched in the most caudal regions of the

PSM (Figure 3.1B). Interestingly, the miR-200b/c family binding site is embedded in a large (120bp) region of sequence that is conserved among mouse, human and chicken

(Figure 3.1C), suggesting that miR-200:Lfng interactions might play an evolutionarily conserved role in multiple vertebrates.

3.3.2 The chick and mouse Lfng 3'UTRs are targets of miR-200b

To examine whether the Lfng 3’UTR is a target of miR-200 family members, we examined expression of luciferase reporter constructs containing the Lfng 3'UTR in the presence or absence of exogenous miR-200b. Reporter constructs containing either mouse or chicken 3'UTR sequences express less luciferase than the pMIR parent vector

(Fig. 3.1 D and E). We have previously shown that this effect is due to the activity of endogenous miRNAs in this cell line(Riley et al. 2013). When constructs carrying either the mouse (mLfng) or chicken (cLfng) 3’UTRs were co-transfected with exogenous pre- miR-200b, we observe a further reduction in luciferase levels (Fig. 3.1 D and E). Point mutations that destroy the putative miR-200 binding site abrogate this effect (Fig. 3.1 D and E). These results indicate that transcripts containing either the mouse or chicken

Lfng 3'UTR sequences can be targeted by members of the miR-200b/c family, via the conserved predicted binding site, and that this targeting affects steady state RNA levels or translational efficiency of transcripts or both.

Figure 3.1: A conserved binding site in the Lfng 3'UTR can be directly targeted by members of the miR-200b family A) QRT-PCR of total RNA from the PSM and mature somites of 9.5 d.p.c. mouse embryos shows that miRNA-200b, and miRNA-200c are significantly (*=p<0.05) enriched in the PSM compared to mature somites. Error bars =

SEM from three samples examined in triplicate B) By whole mount in situ analysis, miR-

200b expression can be detected in the posterior PSM of mouse embryos at 9.0 (a) and

10.5(b) days post coitum (d.p.c.). C) A schematic of the mouse 3'UTR shows regions that are highly conserved mouse:human as purple boxes and regions that are highly conserved from mouse:chicken as green boxes. Seed sequence targets for (continued)

(Figure 3.1, continued) the miR-200bc family in several organisms are shown below, with the binding site in bold. Numbers indicate nucleotides from the 1st nucleotide in the 3'UTR. D) Transfection of pre-miRNA-200b (200b) significantly reduces luciferase expression from transcripts containing the mouse Lfng 3'UTR (mLfng= mouse Lfng 3'UTR cloned into pMIR REPORT) compared to transfection of a negative control pre-miRNA

(Ctr). Mutations in the miRNA-200b binding site (MUT) abrogate this effect. E)

Transfection with pre-miRNA-200b significantly reduces luciferase expression from transcripts containing the chicken Lfng 3'UTR (cLfng= chicken Lfng 3'UTR cloned into pMIR REPORT) Mutations in the miRNA-200b (MUT) binding sites abrogate this effect.

Expression of exogenous miR-200b has no effects on luciferase transcripts expressed from the parental pMIR REPORT vector. Our previous work shows that the change in expression between pMIR-REPORT and mLfng or cLfng in the control transfections is due to endogenous miRNA activity in NIH3T3 cells. All results were analyzed by ANOVA,

Bonferroni post hoc; (* p<.05, error bar = s.d.). (in collaboration with Maurisa Riley)

3.3.3 miR-200b/c/429 may play a role in chicken somitogenesis

Since we found that miR-200b regulates transcripts with the Lfng 3'UTR we wanted to examine if the miR-200b family (Figure 3.2A) plays a role during chicken somitogenesis, where oscillatory Lfng expression is required. To block the miR-

200b/c/429 family from binding to its target mRNAs we used either a fluorescein tagged traditional long anti-miR oligos (Choi et al. 2008) that binds imperfectly to the mature sequences of all three miRNAs in the family or a fluorescein tagged seedblocker, which is a short oligo (8 nucleotide in length) that perfectly binds to only the seed sequence of the miR-200b/c/429 family (Figure 3.2A).

Before performing in vivo experiments to block miR-200b, we used luciferase assays

(Figure 3.2A) to validate the effectiveness of the seedblocker (SB) in preventing the miR-

200b family from binding to the ZEB2 3’UTR, which is a known target of the miR-200b family(Park et al. 2008). NIH3T3 cells were transfected with luciferase vector containing the target 3’UTR (pMIR-ZEB2 3’UTR) along with pre-miR-200b and either the SB-

Negative control (SB-NC) or SB-200F. To compare the efficacy of the seedblockers, traditional anti-miRs were used as controls: anti-miR-Negative Control (NC) that doesn’t bind to any known miRNA and anti-miR-200F that blocks the miR-200b family. We find a comparable reduction in relative luciferase activity in cells transfected with pMIR-ZEB2

3’UTR and miR-200b along with either SB-NC or anti-miR-NC. This indicates that the control SB is not blocking miR-200b activity. In cells with pMIR-ZEB 3’UTR and miR-200b along with either of SB-200F or anti-miR-200F there is an increase in relative luciferase

100 activity indicating that the SB-200F, like the anti-miR-200F is effectively blocking miR-

200b from binding to the target 3’UTR.

Next, we examined the role of miR-200b family in chicken segmentation by electroporating the PSM of chick embryos with either fluorescein tagged anti-miR-200F or SB-200F to block the miR-200b/c/429 activity. Fluorescein tagged anti-miR-NC and

SB-NC were used as negative controls. After 24 hrs, segmentation defects were assessed by observing somite morphology in fluorescein positive embryos followed by performing in situ hybridization for Uncx, a posterior somite boundary marker. In comparison to control embryos (Figure 3.2 B and D), somite morphology appears to be disrupted in embryos electroporated with anti-miR-200F or SB-200F, specifically in the regions that were fluorescein positive (Figure 3.2 C and E). These preliminary results suggest that the miR-200b family may target genes involved in chicken somitogenesis.

101

Figure 3.2 miR-200b family may play a role during chicken somitogenesis. Activity of miR-200b family was blocked using either antimiR-200F or seedblocker-200F(continued)

102

(Figure 3.2, continued) (SB-200F) and compared to anti-NC or SB-NC (Negative controls)

A) The panel on the top left lists members of the miR-200b family and the panel on the bottom left displays the interaction between the seedblocker for the miR-200b family and the seed sequence of miR-200b. Efficacy of SB-200F was assessed by luciferase assays, shown in the right panel, using the pMIR reporter with ZEB2 3’UTR, a known target of miR-200b family. Transfection with pMIR-ZEB2 3’UTR along with miR-200b and either SB-200F (red) or anti-miR-200F (green) resulted in higher relative luciferase activity compared to the controls (only miR-200b, SB-NC or anti-miR-NC). Luciferase expression was normalized to Renilla for each well and relative luciferase activity was determined with respect to luciferase expression in cells with pMIR-ZEB 3’UTR vector and a negative control (NC). Seedblockers for miR-200b family work as effectively as the traditional anti-miR-200F.

B) and D) Chicken embryos electroporated with fluorescein tagged anti-miR-NC or SB-

NC exhibit normal somite morphology and form somite boundaries as seen by Uncx expression (purple) in the posterior compartment. C) and E) Somite morphology is disrupted in chicken embryos electroporated with anti-miR-200F or SB-200F. This implies that miR-200b family may play a role in chick somitogenesis.

103

3.3.4 Blocking the interaction of miR-200 with Lfng 3'UTR does not perturb somitogenesis in chicken embryos

We previously examined the function of miR-125a-5p in chick somitogenesis, and found that interactions between this miRNA and the Lfng transcript regulate transcript stability in the PSM(Riley et al. 2013). Since we have preliminary evidence that the miR-

200b family may play a role in somitogenesis, we asked whether this role is mediated by the interaction of miR-200b with the Lfng 3’UTR. To specifically test the importance of this interaction, we utilized a fluorescein tagged target protector (TP200b/c) that specifically binds across the miR-200 recognition site in the chicken Lfng 3’UTR and thus prevents any member of the miR-200b/c family from binding to the 3’UTR, without perturbing their interactions with any other targets (Fig. 3.3A). The target protector was validated in cell culture by using the luciferase construct with the wildtype Lfng 3'UTR

(Fig 3.3B). When we transfected cells with exogenous miR-200b we found a reduction in luciferase activity mediated by miR-200b, as seen in Fig 3.1C. Co-transfection of a control target protector that binds to an unconserved region of the 3'UTR has no effect on the ability of miR-200b to regulate transcripts containing the Lfng 3'UTR. In contrast, in the presence of TP200b/c, exogenous miR-200b has no effect on luciferase activity indicating that the target protector is effectively blocking the interaction of miR-200b with the Lfng 3'UTR (Fig. 3.3B).

104

To examine the role of the miR-200b:Lfng interaction in vivo, we electroporated the

PSM of chicken embryos with fluorescein tagged TPCtrl or TP200b/c. Fluorescein positive embryos were examined after 24 hours to allow the electroporated region to be incorporated into mature somites. We observe normal morphology in both control embryos and embryos containing TP200b/c.(Fig 3.3C, panels a and c). The patterning of these somites was examined by observing the expression of Uncx, a marker of the posterior somite compartment. Again, we observe normal Uncx staining in both the control and TP200b/c electroporated chicken embryos (Fig 3.3C, panels b and d). These results suggest that interactions between members of the miR-200b/c family with Lfng

3’UTR are not required for chicken somitogenesis, and imply that the miR-200b binding site in the Lfng 3'UTR may not be critical to regulate expression of Lfng in the segmentation clock.

Since previous work indicated that interactions between miR-125a and the Lfng

3'UTR are important for normal segmentation in the chick(Riley et al. 2013), we asked whether simultaneous interference with miR-125a and miR-200 might exacerbate these phenotypes. Simultaneous electroporation of chicken embryos with TP125a (which blocks interactions between miR-125a and Lfng) and TP200b/c produced defects similar to those seen previously with electroporation of with TP125a alone(Riley et al. 2013). While the embryo-embryo variability seen with these types of experiments makes it difficult to draw strong conclusions based on these findings, our data do not provide strong support for the idea that the miR-200b binding site plays a cooperative role with the

105 miR-125a binding site in regulating Lfng expression in the context of chicken somitogenesis.

Figure 3.3: Interactions between Lfng and miR-125a-5p but not miR- 200b are essential for proper vertebrate segmentation. A) Schematic of the Target Protector, which binds across the miR-200b/c recognition site and prevents binding of endogenous (continued) 106

(Figure 3.3, continued) miRNAs B) Validation of Lfng-TPmiRNA-200b . Luciferase activity is from vectors containing the chicken Lfng 3'UTR in reduced in the presence of exogenous miR-200b. This effect is blocked by cotransfection with TP 200b/c but not by cotransfection with a control oligo (TP Ctrl), indicating that TP200b/c effectively blocks interactions between miR-200b and the Lfng 3’UTR. Results expressed as normalized firefly luciferase activity. Error bars = SD. Each experiment was performed at least times in triplicate. Results were analyzed by ANOVA followed by Bonferonni post hoc;

(*p<.05). C) 24 hours after electroporation of Lfng-TPctr (a, n 12/12) or Lfng-TPmiRNA-200b

(b, n=10/12) we observe no effect on somite morphology, but electroporation of Lfng-

TPmiRNA-125a + miRNA-200b (e, n=15/15) results in abnormal somite morphology. Similarly, electroporation of Lfng-TPctr (b,) or Lfng-TPmiRNA-200b (d,) has no effect on anterior- posterior patterning, as evidenced by normal Uncx4 expression, but electroporation of

Lfng-TPmiRNA-125a + miRNA-200b (e, n) results reduced and disorganized Uncx4.1 expression.

Fluorescein images of each embryo are shown to the left (labeled with ') to indicate the extent and localization of the electroporated region. (in collaboration with Maurisa

Riley)

107

3.3.5 Blocking the interaction of miR-200 with Lfng 3'UTR does not dramatically affect transcript stability in the chick PSM

We also examined the effects of blocking miR-200b:Lfng interactions on the expression of endogenous Lfng transcripts. After electroporation with Target Protectors, we waited 8 hours (approximately 6 oscillations of the clock), and then examined the expression of Lfng while electroporated cells were still in the PSM. As expected from our results above, we observed cyclic expression of Lfng mRNA in both the control and

TP200b/c electroporated embryos. This indicates that blocking interactions between miR-

200b/c family members and the binding site in the Lfng 3'UTR has no effect on the cyclic expression of Lfng, suggesting that this binding site is not necessary to destabilize the

Lfng mRNA in the chick PSM.

108

Figure 3.4: Inhibiting interactions between Lfng and miR-200b does not affect Lfng oscillations. Embryos were electroporated with Lfng-TPCtrl or Lfng-TPmiRNA-200b and incubated for 8 hours prior to in situ analysis to examine expression of endogenous

Lfng. Cyclic expression of endogenous Lfng is observed in embryos elestroporated with

Lfng-TPCtrl and in embryos electroporated with Lfng-TPmiRNA-200b embryos, as evidenced by the observation of three patterns of expression representing three phases of clock expression. The ratio of embryos in each phase is similar on both groups of embryos, suggesting that blocking miR-200b/c:Lfng interactions does not perturb Lfng transcript turnover in the segmentation clock. (in collaboration with Maurisa Riley)

109

3.3.6 The conserved 120 bp central region of the Lfng 3'UTR is not sufficient to destabilize exogenous transcripts in C2C12 cells

The Lfng 3'UTR has been shown to destabilize exogenous transcripts in a variety of contexts (Chen et al. 2005; Hilgers et al. 2005; Riley et al. 2013; Nitanda et al. 2014a). To examine the effects of the large region of conserved sequence containing the miR-200b binding site, we created reporter constructs containing coding sequences for Venus, as well as a rabbit beta-globin intron at the 5'end and subsequences of the Lfng 3'UTR at the 3’end and examined the effect of the 3’UTR on transcript stability using reporter constructs. We first compared the effects of the full length 3'UTR with the effects of a construct that deletes 826bp from the middle including the miR-200b and miR-125a binding site (mini-UTR). This was done by creating stable cell lines expressing each construct and examining RNA levels by QRT-PCR after blocking transcription with

Actinomycin D. We found that RNA transcripts containing the wildtype UTR exhibit rapid degradation with a half-life of about 90 mins. In contrast, deletion of the central 826 bp abrogates this effect, resulting in a transcript that is stable over the 150-minute time course (Figure 3.5 A and B). To determine whether the 120 bp conserved central region is sufficient to destabilize a transcript in this context, we performed similar experiments using a reporter construct containing only these sequences (Cons UTR). The transcripts including these sequences were found to be stable over the time course analyzed (Fig

3.5B). Taken together, these results suggest that the regions deleted in the "mini" UTR contain regulatory elements that influence RNA turnover in this context. These

110 sequences include both the miR-125a site we described previously(Riley et al. 2013), and the miR-200b binding sites we examined here. However, the highly conserved 120 bp sequence including the miR-200b binding site alone was not sufficient to destabilize the transcript in this context.

Figure 3.5: The conserved region containing a miR-200b/c binding site is not sufficient to destabilize an exogenous transcript. A) Schematic representations of the 3'UTR sequences incorporated into each reporter construct. Sequences conserved mouse:human are shown as purple boxes while sequences conserved (continued) 111

(Figure 3.5, continued) mouse:chicken are green boxes. Approximate positions of miRNA binding sites are shown. B) QRT-PCR was used to assess RNA levels from cells over a 2.5 hour time course after transcription block with Actinomycin D. Values for Venus were normalized to Gapdh and the "0" timpoint was set to 1. Values from three independent experiments are shown, with error bars depicting SEM. Trend lines were fitted to determine RNA half life.

3.4 Discussion

Sequences within transcript 3’UTRs can regulate gene expression by modulating

RNA stability, RNA localization and translation efficiency (Grzybowska et al. 2001). This post-transcriptional regulation is extremely important in the context of the vertebrate segmentation clock, where mRNAs transcribed from oscillatory genes need to have short half-lives to ensure complete clearance of gene products during the short "OFF" phase of the clock.

One key player in the mouse and chicken clock is Lfng, and there is significant evidence that rapid turnover of Lfng transcripts and proteins are critical for normal clock function(Riley et al. 2013; Williams et al. 2016). Previous work has shown that transcripts containing the Lfng 3'UTR are less stable in the PSM than the 3’UTRs of non- 112 oscillatory transcripts such as Fgf8(Hilgers et al. 2005). The Lfng 3'UTR contains several regions of high sequence conservation, that have been suggested to have different contributions to transcript stability in mouse cell lines(Chen et al. 2005). Interactions between miR-125a and one of these conserved regions has previously been shown to affect transcript stability of Lfng mRNA in the context of chicken somitogenesis(Riley et al. 2013), but it is not clear how other sequences in the 3'UTR might regulate Lfng transcripts.

In this study we have examined the possibility that a 120 bp conserved region, with a miR-200 binding site, in the central part of the Lfng 3'UTR might influence transcript stability. Although it has been shown that both the mouse and chicken Lfng

3'UTR can be targeted by members of the miR-200 family, blocking these interactions in the developing chicken PSM has no effects on cyclic expression of endogenous Lfng or on somite production. In this context it is interesting to note that miR-200b appears to be localized largely to the ectoderm and most caudal domain of the PSM in mouse embryos, thus it may not interact with Lfng in the mesoderm where the segmentation clock operates.

Though, preliminary work suggests that miR-200b may play a role during chick somitogenesis, its role is not mediated by regulating Lfng expression in the chick PSM. It should be noted that these anti-miRs and seedblockers contain locked nucleic acids

(LNA) that are known to cause moderate amount of toxicity in zebrafish, hence their use in blocking miR-200 family may need to be reconsidered(Burdick et al. 2014).

113

Looking in other contexts, we find that the full length mouse Lfng 3'UTR contains sequences that can destabilize an exogenous transcript in mouse C2C12 cells. Deletion of 826 nucleotides, from 186 to 1012, of the mouse Lfng abrogated the effect of the

3'UTR on transcript stability, suggesting that critical cis-acting sequences are found in this region (which contains all the previously identified miR-125a and miR-200b binding sites). However, the 120 bp conserved region containing the miR-200 site was not sufficient to destabilize an exogenous transcript, at least in this context. These results suggest that there are additional regions of the Lfng 3'UTR that influence transcript stability independent of or in conjunction with the 120bp region. Work from other labs suggested that the first 307 bp of the mouse Lfng 3'UTR have a moderate effect on transcript stability, while the first 700 bp or the full length 3'UTR have more dramatic effects. These findings are difficult to compare directly with ours, as they were performed in a different cell line (NIH-3T3 cells) and over an 8-hour time course (Chen et al., 2005). However, taken together with the work presented here, they support the idea that the regulation of Lfng transcript stability may have context-dependent effects, and that different regions of the 3'UTR may work together to coordinate these effects.

Previous studies have also suggested that AU-rich elements (AREs) in the Lfng

3'UTR may play a role in destabilizing the transcript(Hilgers et al. 2005; Nitanda et al.

2014a). There are two AU-rich elements (AREs) found immediately downstream of the conserved 120bp sequence in the mouse Lfng 3'UTR that are conserved in chickens and humans. There is evidence from other 3’UTRs that AREs can mediate transcript turnover 114 both through and independent of the miRNA pathway(Jing et al. 2005; Helfer et al.

2012). AREs in the Lfng 3'UTR could be acting in a similar way, either in cooperation with the 120 bp region or independent of it, to destabilize the mRNA.

In conclusion, we have identified an 826 bp region within the mouse Lfng 3'UTR that contains cis-acting elements that can destabilize the upstream mRNA. This region contains binding sites for miR-125a and miR-200b. While interactions between miR-

125a and Lfng have previously been shown to be critical for Lfng transcript instability in the chicken segmentation clock, the miR-200b binding site in the 120 bp region of the chicken Lfng 3'UTR is not necessary for normal clock function or somitogenesis. We also find that the 120bp conserved region alone had no dramatic effect on RNA stability, at least in the context of mouse myoblast cells. Our findings suggest that other regulatory regions such as the previously identified miR-125a binding site may play a more critical role in destabilizing Lfng mRNA in the context of somitogenesis.

115

CHAPTER 4

CONCLUSIONS

The main objective of this thesis is to understand the post-regulation of Lunatic fringe (Lfng), a modulator of the Notch pathway, in the mouse segmentation clock. The segmentation clock is functional during vertebrate embryogenesis and times periodic formation of somite pairs from an unsegmented pre-somitic mesoderm (PSM). Somites are transient structures that give rise to the vertebral column, rib cage, skeletal muscles and dermis of the back (reviewed in Wahi et al. 2014). Vertebral defects in humans such as spondylocostal dysostosis (SCD) is caused in some cases by mutations in Lfng

(Sparrow et al. 2006). Therefore, Lfng is required for normal vertebral development in humans but its role and regulation in this process is not completely understood.

Lfng encodes a glycosyltransferase that modifies Notch receptors and ligands causing altered Notch signaling, in a context dependent manner(Luther & Haltiwanger

2009; Takeuchi & Haltiwanger 2014). Beginning with its discovery in Drosophila melanogaster(Irvine & Wieschaus 1994), homologues of Fringe proteins have been found in several diverse organisms ranging from bacteria(Yuan et al. 1997) to mouse(Cohen et al. 1997). Lfng in mice and humans exhibits not only significant

116 sequence similarity but also conservation of function in the segmentation clock, making the mouse segmentation clock an ideal system to study the regulation of Lfng.

Lfng expression in the mouse PSM is dynamic and cycles, between ON and OFF phases, with a period that matches the rate of somite formation(Cohen et al. 1997;

Johnston et al. 1997; Evrard et al. 1998; Zhang & Gridley 1998). Oscillatory expression of

Lfng mRNA and protein in the mouse PSM is required for normal somite formation and skeletal development (Shifley et al. 2008; Williams et al. 2016). Oscillatory expression of

Lfng is controlled at the transcriptional level by a negative feedback loop, wherein Lfng is activated by Notch signaling and repressed by the HES7 protein(Aulehla & Johnson

1999; Bessho et al. 2001; Serth et al. 2003; Chen et al. 2005; Niwa et al. 2007). LFNG protein in turn negatively regulates Notch signaling, possibly by modifying Notch receptors and/or ligands(Moloney et al. 2000; Takeuchi & Haltiwanger 2014; LeBon et al. 2014; Serth et al. 2015). To sustain stable and rapid Lfng oscillations in the PSM additional levels post-transcriptional and post-translational control must be in effect to ensure degradation of Lfng transcript and protein from cells before the next round of oscillations begin(Feng & Navaratna 2007; González & Kageyama 2009; Jing et al. 2015).

In this thesis we focus on the post-transcriptional regulation of Lfng with an intent to identify regulatory regions and players that may destabilize the Lfng transcript in the mouse PSM.

The Lfng 3’UTR is found to be critical for transcript turnover during mouse and chicken somitogenesis (Hilgers et al. 2005; Chen et al. 2005; Riley et al. 2013; Nitanda et

117 al. 2014a). The Lfng 3'UTR has regions along its length that are conserved among mouse, chickens and humans; some of the regions are of particular interest as they contain binding sites for microRNAs (miRNAs). Previous work from our lab has shown that miR-125a-5p destabilizes Lfng transcripts in the chicken PSM and this interaction is required for proper functioning of the chicken segmentation clock(Riley et al. 2013). This raises the question whether this regulation is conserved in other vertebrates or specific to the chicken segmentation clock. In Chapter 2 we address this question in the context of mouse segmentation by examining the importance of the miR-125a binding sites in the mouse Lfng 3'UTR and testing the consequence of loss of miR-125a on segmentation and skeletal development in mice. In Chapter 3 we investigate the possible roles of another conserved region of the Lfng 3'UTR that contains a miR-200b binding site, in the chick segmentation clock and in mouse myoblast cells.

4.1 miR-125a binding sites in the Lfng 3'UTR regulate transcript turnover in the tailbud region of the mouse PSM

In Chapter 2, using Venus reporter constructs we tested the destabilizing ability of a mutated Lfng 3'UTR, where the miR-125a sites have been mutated. We found that mir-

125a can target the Lfng 3'UTR to reduce levels of a transcript and that the putative mir-

125a binding sites are necessary for this effect. To test the role of the miR-125a binding sites in the context of the mouse segmentation clock, we created transgenic Venus reporter mice driven by the Lfng promoter and expressing Venus transcripts with a wildtype (LvLWt) or mutated Lfng 3'UTR (LvLMut). LvLWt mice exhibit an oscillatory

118

Venus expression profile that is similar to the dynamic expression of endogenous Lfng in the mouse PSM. Venus expression in the LvLMut mice, however, is constitutive in the caudal most or tailbud region of the PSM while displaying oscillatory expression in the anterior PSM. Given that miR-125a is highly expressed in the tailbud region of the mouse PSM this could support a model that the miR-125a binding sites in the Lfng 3'UTR may be important for transcript turnover in the tailbud region of the PSM. Interestingly, the oscillatory period of clock genes is faster in cells in the posterior PSM and gradually becomes slow as cells enter the anterior PSM which may be due to a difference in RNA turnover of clock genes along the PSM(Palmeirim et al. 1997; Gibb et al. 2010; Niwa et al. 2007; Niwa et al. 2011; Shih et al. 2015). Our finding seems to corroborate the distinction in oscillatory pace of Lfng between the posterior and anterior by suggesting that miR-125a binding sites may be required for transcript turnover in the caudal most region of the posterior PSM.

4.2 Loss of miR-125a in mice has no impact on Lfng oscillations and the segmentation clock

We generated three strains of miR-125a mice by targeting the miR-125a locus using the CRISPR/Cas9 genome editing system. In two of the mutants, Del11 and Del3T, complete loss of miR-125a expression was observed while the third mutant, Ins, miR-

125a expression was reduced. Contrary to our expectation, we found that loss of miR-

125a had no effect on oscillatory expression of Lfng in the mouse PSM and segmentation or skeletal development appears to be normal in these mutants. The

119 possible explanations for this could be that 1) unlike chicken somitogenesis, miR-125a may not be required during mouse somitogenesis indicating that post-transcriptional regulation of Lfng differs in the chicken and mouse segmentation clock (however, it would not explain our Venus transgene findings), this could be addressed by mutating the endogenous miR-125a binding sites in the mouse Lfng 3'UTR and examining the effect of the mutation on Lfng expression and on mouse segmentation 2) other factors can be upregulated in the chronic absence of miR-125a and can sufficiently destabilize the Lfng transcript and compensate for the loss of miR-125a in the mouse PSM. If so, this would address the robustness of the somitogenesis process. The expression of other miR-125a family members such as miR-125b and miR-351 could be examined in the PSM of miR-125a mutants to examine the possibility of compensatory mechanisms.

4.3 Complete loss of miR-125a results in male infertility in mice and could be due to defective sperm morphology

Male mice homozygous for the Del11 mutation (miR-125a-5p∆11/∆11) failed to impregnate any wildtype or miR-125a mutant female mice. These infertile male mice have normal testes and produce sperm but the majority of the sperm display a sharp bend in their tail which could affect their motility. This finding is supported by the fact that miR-125a is highly expressed in specifically in male primordial germ cells though its role in this context is not clearly understood(Hayashi et al. 2008; Mciver et al. 2012). The discovery that miR-125a is likely to be required for sperm morphology provides a novel and indispensable role for miR-125a in the context of male fertility.

120

4.4 The conserved 120 bp central region of the Lfng 3'UTR is not sufficient to destabilize transcripts

In Chapter 3, we examined the effect of other conserved regions in the mouse Lfng

3'UTR on transcript stability. A 120 bp region in the middle of the Lfng 3'UTR is highly conserved in mice, chickens and humans and contains a miR-200b binding site.

Interestingly, expression of miR-200b family members (miR-200b/c/429) is found to be higher in the unsegmented PSM as opposed to the somites in mouse embryos. We investigated the role of miR-200b:Lfng interaction in the chicken somitogenesis and found that although miR-200b may be involved in somitogenesis, miR-200b does not interact with the Lfng 3’UTR to regulate oscillatory expression of Lfng in the chick PSM.

Hence, the miR-200b binding site in the chick Lfng 3'UTR does not seem to be required to regulate transcript expression in the PSM in the context of chick somitogenesis.

We were particularly interested in examining the destabilizing effect of the conserved 120 bp region in the middle of the 3’UTR that contains the miR-200b binding site. This region alone did not have a dramatic effect on RNA stability, at least in the context of mouse myoblast cells. We have identified an 826 bp region within the mouse

Lfng 3'UTR that contains cis-acting elements that can destabilize the upstream mRNA.

This region contains binding sites for mir-125a and mir-200b. Our findings suggest that other regulatory regions such as the previously identified mir-125a binding site may play a more critical role in destabilizing Lfng mRNA in the context of somitogenesis.

121

4.5 Concluding remarks

The findings in this thesis lay emphasis on the post-transcriptional regulation by the

Lfng 3'UTR in the context of the segmentation clock. Examining the post-transcriptional regulation of Lfng could shed light on the mechanism behind the gradual slowing of oscillatory period along the PSM which is a poorly understood aspect of somitogenesis.

Our work supports the possibility of difference in regulation of clock genes from the posterior to the anterior PSM and how the 3’UTR of these genes and the interacting factors could be critical in conferring the difference in oscillatory period along the PSM.

In the context of Lfng regulation in the mouse segmentation clock we find that miR-

125a may have a dispensable role in mediating mRNA turnover along the PSM and impacting somitogenesis. However, in other contexts such as spermatogenesis miR-

125a may have an indispensable role in ensuring normal sperm activity.

Future work could aim at mutating the miR-125a binding sites in the endogenous mouse Lfng 3'UTR to study the consequence of these mutations on oscillatory Lfng expression in the PSM and the impact on somitogenesis and skeletal development in mice. It would be interesting to explore post-transcriptional regulation by other factors that could possibly interact with the miR-125a binding sites in the Lfng 3'UTR. To address another curious aspect of post-transcriptional control, the conserved AU rich regions in the mouse Lfng 3'UTR could be investigated for their role in regulating Lfng expression during mouse segmentation.

122

REFERENCES

Aulehla, a & Johnson, R.L., 1999. Dynamic expression of lunatic fringe suggests a link between notch signaling and an autonomous cellular oscillator driving somite segmentation. Developmental biology, 207(1), pp.49–61. Available at: http://www.ncbi.nlm.nih.gov/pubmed/10049564. Aulehla, A. et al., 2008. A beta-catenin gradient links the clock and wavefront systems in mouse embryo segmentation. Nature cell biology, 10(2), pp.186–193. Aulehla, A. et al., 2003. Wnt3a plays a major role in the segmentation clock controlling somitogenesis. Developmental Cell, 4(3), pp.395–406. Bajard, L. et al., 2014. Wnt-regulated dynamics of positional information in zebrafish somitogenesis. Development (Cambridge, England), 141(6), pp.1381–91. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3943186&tool=pmcen trez&rendertype=abstract. Barrantes, I.B. et al., 1999. Interaction between Notch signalling and Lunatic fringe during somite boundary formation in the mouse. Current biology : CB, 9(9), pp.470– 80. Available at: http://www.ncbi.nlm.nih.gov/pubmed/10330372. Bessho, Y. et al., 2001. Dynamic expression and essential functions of Hes7 in somite segmentation. Genes & development, 15(20), pp.2642–7. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=312810&tool=pmcent rez&rendertype=abstract. Bone, R.A. et al., 2014. Spatiotemporal oscillations of Notch1 , Dll1 and NICD are coordinated across the mouse PSM. Development, 141, pp.4806–4816. Borg, C.L. et al., 2009. Phenotyping male infertility in the mouse: How to get the most out of a “non-performer.” Human Reproduction Update, 16(2), pp.205–224. Brückner, K. et al., 2000. Glycosyltransferase activity of Fringe modulates Notch-Delta interactions. Nature, 406(6794), pp.411–415. Bulman, M.P. et al., 2000. Mutations in the human delta homologue, DLL3, cause axial skeletal defects in spondylocostal dysostosis. Nature Genetics, 24(4), pp.438–441. Available at: http://www.ncbi.nlm.nih.gov/pubmed/10742114. Burdick, A.D. et al., 2014. Sequence motifs associated with hepatotoxicity of locked nucleic acid--modified antisense oligonucleotides. Nucleic acids research, 42(8), pp.4882–91. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=4005641&tool=pmcen trez&rendertype=abstract.

123

Chen, J., Kang, L. & Zhang, N., 2005. Negative feedback loop formed by Lunatic fringe and Hes7 controls their oscillatory expression during somitogenesis. Genesis (New York, N.Y. : 2000), 43(4), pp.196–204. Available at: http://www.ncbi.nlm.nih.gov/pubmed/16342160 [Accessed February 14, 2014]. Chen, Q. et al., 2011. Aquaporin3 is a sperm water channel essential for postcopulatory sperm osmoadaptation and migration. Cell research, 21(6), pp.922–933. Available at: http://dx.doi.org/10.1038/cr.2010.169. Choi, P.S. et al., 2008. Members of the miRNA-200 Family Regulate Olfactory Neurogenesis. Neuron, 57(1), pp.41–55. Available at: http://dx.doi.org/10.1016/j.neuron.2007.11.018. Christ, B., Huang, R. & Scaal, M., 2007. Amniote somite derivatives. Developmental Dynamics, 236(9), pp.2382–2396. Cohen, B. et al., 1997. Fringe boundaries coincide with Notch-dependent patterning centres in mammals and alter Notch-dependent development in Drosophila. Nature Genetics, 16(3), pp.283–288. Available at: http://www.nature.com/doifinder/10.1038/ng0797-283 [Accessed May 31, 2016]. Cole, S.E. et al., 2002. Clock regulatory elements control cyclic expression of Lunatic fringe during somitogenesis. Developmental Cell, 3(1), pp.75–84. Collier, J.R. et al., 2000. A cell cycle model for somitogenesis: Mathematical formulation and numerical simulation. Journal of Theoretical Biology, 207(3), pp.305–316. Conlon, R. a, Reaume, a G. & Rossant, J., 1995. Notch1 is required for the coordinate segmentation of somites. Development (Cambridge, England), 121(5), pp.1533–45. Available at: http://www.ncbi.nlm.nih.gov/pubmed/7789282. Cooke, J. et al., 1998. A gene that resuscitates a theory—somitogenesis and a molecular oscillator. Trends in Genetics, 14(3), pp.85–88. Available at: http://linkinghub.elsevier.com/retrieve/pii/S0168952598013961 [Accessed May 28, 2016]. Cooke, J. & Zeeman, E.C., 1976. A clock and wavefront model for control of the number of repeated structures during animal morphogenesis. Journal of theoretical biology, 58(2), pp.455–76. Available at: http://www.ncbi.nlm.nih.gov/pubmed/940335. Cooper, T.G., 2005. Cytoplasmic droplets: The good, the bad or just confusing? Human Reproduction, 20(1), pp.9–11. Del Corral, R.D. et al., 2003. Opposing FGF and retinoid pathways control ventral neural pattern, neuronal differentiation, and segmentation during body axis extension. Neuron, 40(1), pp.65–79. Dale, J.K. et al., 2003. Periodic notch inhibition by lunatic fringe underlies the chick segmentation clock. Nature, 421(6920), pp.275–278. Delaune, E.A. et al., 2012. Single-cell-resolution imaging of the impact of Notch signaling and mitosis on segmentation clock dynamics. Developmental cell, 23(5), pp.995– 1005. Available at: http://www.ncbi.nlm.nih.gov/pubmed/23153496 [Accessed June 2, 2016].

124

Delfini, M.-C. et al., 2005. Control of the segmentation process by graded MAPK/ERK activation in the chick embryo. Proceedings of the National Academy of Sciences of the United States of America, 102(32), pp.11343–8. Available at: http://www.ncbi.nlm.nih.gov/pubmed/16055560 [Accessed May 29, 2016]. Dequéant, M.-L. et al., 2006. A Complex Oscillating Genes Network of Signaling Underlies Clock the Mouse Segmentation. Science (New York, N.Y.), 314(5805), pp.1595–1598. Available at: http://www.mendeley.com/catalog/complex- oscillating-network-signaling-genes-underlies-mouse-segmentation-clock/ [Accessed February 11, 2016]. Dexter, J., 1914. The analysis of a case of continuous variation in Drosophila by a study of its linkage relations. The American Naturalist, 48, pp.712–758. Dubrulle, J., McGrew, M.J. & Pourquié, O., 2001. FGF signaling controls somite boundary position and regulates segmentation clock control of spatiotemporal Hox gene activation. Cell, 106(2), pp.219–232. Dunty, W.C. et al., 2008. Wnt3a/beta-catenin signaling controls posterior body development by coordinating mesoderm formation and segmentation. Development (Cambridge, England), 135(1), pp.85–94. Available at: http://www.ncbi.nlm.nih.gov/pubmed/18045842 [Accessed May 29, 2016]. Eckalbar, W.L. et al., 2012. Scoliosis and segmentation defects of the vertebrae. Wiley Interdisciplinary Reviews: Developmental Biology, 1(3), pp.401–423. van Eeden, F.J. et al., 1998. Zebrafish segmentation and pair-rule patterning. Developmental genetics, 23(1), pp.65–76. Available at: http://www.ncbi.nlm.nih.gov/pubmed/9706695 [Accessed May 28, 2016]. Emmrich, S. et al., 2014. miR-99a/100 125b tricistrons regulate hematopoietic stem and progenitor cell homeostasis by shifting the balance between TGF and Wnt signaling. Genes & Development, 28(8), pp.858–874. Available at: http://genesdev.cshlp.org/cgi/doi/10.1101/gad.233791.113 [Accessed June 1, 2016]. Evrard, Y. a et al., 1998. Lunatic Fringe Is an Essential Mediator of Somite Segmentation and Patterning. Nature, 394(6691), pp.377–381. Feller, J. et al., 2008. Noncyclic Notch activity in the presomitic mesoderm demonstrates uncoupling of somite compartmentalization and boundary formation. Genes and Development, 22, pp.2166–2171. Feng, P. & Navaratna, M., 2007. Modelling periodic oscillations during somitogenesis. Mathematical biosciences and engineering : MBE, 4(4), pp.661–673. Ferjentsik, Z. et al., 2009. Notch is a critical component of the mouse somitogenesis oscillator and is essential for the formation of the somites. PLoS genetics, 5(9), p.e1000662. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2739441&tool=pmcen trez&rendertype=abstract. Filipowicz, W., Bhattacharyya, S.N. & Sonenberg, N., 2008. Mechanisms of post-

125

transcriptional regulation by microRNAs: are the answers in sight? Nature reviews. Genetics, 9(2), pp.102–114. Friedman, R.C. et al., 2009. Most mammalian mRNAs are conserved targets of microRNAs. Genome research, 19(1), pp.92–105. Available at: http://www.ncbi.nlm.nih.gov/pubmed/18955434 [Accessed June 1, 2016]. Fujihara, Y. & Ikawa, M., 2014. Chapter Fifteen – CRISPR/Cas9-Based Genome Editing in Mice by Single Plasmid Injection. In Methods in Enzymology. pp. 319–336. Fujimuro, T. et al., 2014. Hes7 3’UTR is required for somite segmentation function. Scientific reports, 4, p.6462. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=4173035&tool=pmcen trez&rendertype=abstract. Giampietro, P.F. et al., 2009. Progress in the understanding of the genetic etiology of vertebral segmentation disorders in humans. Annals of the New York Academy of Sciences, 1151, pp.38–67. Gibb, S., Maroto, M. & Dale, J.K., 2010. The segmentation clock mechanism moves up a notch. Trends in Cell Biology, 20(10), pp.593–600. Available at: http://dx.doi.org/10.1016/j.tcb.2010.07.001. Giraldez, A.J. et al., 2005. MicroRNAs regulate brain morphogenesis in zebrafish. Science (New York, N.Y.), 308(5723), pp.833–8. Available at: http://www.ncbi.nlm.nih.gov/pubmed/15774722 [Accessed September 19, 2014]. Giudicelli, F. et al., 2007. Setting the tempo in development: An investigation of the zebrafish somite clock mechanism. PLoS Biology, 5(6), pp.1309–1323. González, A. & Kageyama, R., 2009. Hopf bifurcation in the presomitic mesoderm during the mouse segmentation. Journal of theoretical biology, 259(1), pp.176–89. Available at: http://www.ncbi.nlm.nih.gov/pubmed/19236883 [Accessed May 20, 2016]. González, A. & Kageyama, R., 2007. Practical lessons from theoretical models about the somitogenesis. Gene regulation and systems biology, 1, pp.35–42. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2759138&tool=pmcen trez&rendertype=abstract. Graham, A. et al., 2014. What can vertebrates tell us about segmentation? EvoDevo, 5(1), p.24. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=4088296&tool=pmcen trez&rendertype=abstract. Grimson, A. et al., 2007. MicroRNA targeting specificity in mammals: determinants beyond seed pairing. Molecular cell, 27(1), pp.91–105. Available at: http://www.ncbi.nlm.nih.gov/pubmed/17612493 [Accessed June 1, 2016]. Grzybowska, E.A., Wilczynska, A. & Siedlecki, J.A., 2001. Regulatory functions of 3 ’ UTRs. Biochem. Biophys. Res. Commun., 288(2), pp.291–295. Guruharsha, K.G., Kankel, M.W. & Artavanis-tsakonas, S., 2012. The Notch signalling system: recent insights into the complexity of a conserved pathway. Nature

126

Reviews. Genetics, 13(9), pp.654–666. Available at: http://search.proquest.com/docview/1040804476?accountid=41383\nhttp://resol ver.ebscohost.com/openurl?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF- 8&rfr_id=info:sid/ProQ:healthcompleteshell&rft_val_fmt=info:ofi/fmt:kev:mtx:jour nal&rft.genre=article&rft.j. Ha, M. & Kim, V.N., 2014. Regulation of microRNA biogenesis. Nature reviews. Molecular cell biology, 15(8), pp.509–524. Available at: http://www.ncbi.nlm.nih.gov/pubmed/25027649. Hamada, Y. et al., 1999. Mutation in ankyrin repeats of the mouse Notch2 gene induces early embryonic lethality. Development (Cambridge, England), 126(15), pp.3415– 24. Available at: http://www.ncbi.nlm.nih.gov/pubmed/10393120 [Accessed May 29, 2016]. Han, J. et al., 2006. Molecular Basis for the Recognition of Primary microRNAs by the Drosha-DGCR8 Complex. Cell, 125(5), pp.887–901. Handrigan, G.R., 2003. Concordia discors: duality in the origin of the vertebrate tail. Journal of anatomy, 202(Pt 3), pp.255–67. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1571085&tool=pmcen trez&rendertype=abstract. Hannibal, R.L. & Patel, N.H., 2013. What is a segment? EvoDevo, 4(1), p.35. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3880069&tool=pmcen trez&rendertype=abstract. Harima, Y. et al., 2013. Accelerating the Tempo of the Segmentation Clock by Reducing the Number of Introns in the Hes7 Gene. Cell Reports, 3(1), pp.1–7. Available at: http://dx.doi.org/10.1016/j.celrep.2012.11.012. Hayashi, K. et al., 2008. MicroRNA biogenesis is required for mouse primordial germ cell development and spermatogenesis. PLoS ONE, 3(3), pp.1–9. Hayashi, S. et al., 2009. Sprouty4, an FGF inhibitor, displays cyclic gene expression under the control of the notch segmentation clock in the mouse PSM. PloS one, 4(5), p.e5603. Available at: http://www.ncbi.nlm.nih.gov/pubmed/19440349 [Accessed May 29, 2016]. Helfer, S. et al., 2012. AU-Rich Element-Mediated mRNA Decay Can Occur Independently of the miRNA Machinery in Mouse Embryonic Fibroblasts and Drosophila S2-Cells S. Pfeffer, ed. PLoS ONE, 7(1), p.e28907. Available at: http://dx.plos.org/10.1371/journal.pone.0028907 [Accessed May 31, 2016]. Henry, C. a et al., 2002. Two linked hairy/Enhancer of split-related zebrafish genes, her1 and her7, function together to refine alternating somite boundaries. Development (Cambridge, England), 129, pp.3693–3704. Herrgen, L. et al., 2010. Intercellular Coupling Regulates the Period of the Segmentation Clock. Current Biology, 20(14), pp.1244–1253. Available at: http://linkinghub.elsevier.com/retrieve/pii/S0960982210007815. Hilgers, V., Pourquié, O. & Dubrulle, J., 2005. In vivo analysis of mRNA stability using the

127

Tet-Off system in the chicken embryo. Developmental Biology, 284(2), pp.292–300. Hirata, H. et al., 2004. Instability of Hes7 protein is crucial for the somite segmentation clock. Nature Genetics, 36(7), pp.750–754. Available at: http://www.nature.com/doifinder/10.1038/ng1372 [Accessed May 30, 2016]. Holley, S. a et al., 2002. Her1 and the Notch Pathway Function Within the Oscillator Mechanism That Regulates Zebrafish Somitogenesis. Development (Cambridge, England), 129(5), pp.1175–1183. Hori, K., Sen, A. & Artavanis-Tsakonas, S., 2013. Notch signaling at a glance. Journal of cell science, 126(Pt 10), pp.2135–40. Available at: http://jcs.biologists.org/content/126/10/2135.long. Horikawa, K. et al., 2006. Noise-resistant and synchronized oscillation of the segmentation clock. Nature, 441(7094), pp.719–723. Available at: http://www.nature.com/doifinder/10.1038/nature04861. Hou, X., Tashima, Y. & Stanley, P., 2012. Galactose differentially modulates lunatic and manic fringe effects on Delta1-induced NOTCH signaling. The Journal of biological chemistry, 287(1), pp.474–83. Available at: http://www.ncbi.nlm.nih.gov/pubmed/22081605 [Accessed May 19, 2016]. Hoyle, N.P. & Ish-Horowicz, D., 2013. Transcript processing and export kinetics are rate- limiting steps in expressing vertebrate segmentation clock genes. Proceedings of the National Academy of Sciences of the United States of America, 110, pp.E4316– 24. Available at: http://www.pnas.org/content/110/46/E4316.short. Ijiri, T.W. et al., 2011. Identification and validation of mouse sperm proteins correlated with epididymal maturation. Proteomics, 11(20), pp.4047–4062. Irvine, K.D. et al., 1997. Fringe modulates Notch|[ndash]|ligand interactions. Nature, 387(6636), pp.908–912. Available at: http://www.nature.com/doifinder/10.1038/43191 [Accessed May 19, 2016]. Irvine, K.D. & Wieschaus, E., 1994. fringe, a Boundary-specific signaling molecule, mediates interactions between dorsal and ventral cells during Drosophila wing development. Cell, 79(4), pp.595–606. Available at: http://www.ncbi.nlm.nih.gov/pubmed/7954826 [Accessed May 31, 2016]. Ishikawa, A. et al., 2004. Mouse Nkd1, a Wnt antagonist, exhibits oscillatory gene expression in the PSM under the control of Notch signaling. Mechanisms of Development, 121(12), pp.1443–1453. Ishitani, T. et al., 2005. Nrarp functions to modulate neural-crest-cell differentiation by regulating LEF1 protein stability. Nature cell biology, 7(11), pp.1106–12. Available at: http://www.ncbi.nlm.nih.gov/pubmed/16228014 [Accessed May 29, 2016]. Jiang, Y.J. et al., 2000. Notch signalling and the synchronization of the somite segmentation clock. Nature, 408(6811), pp.475–9. Available at: http://www.ncbi.nlm.nih.gov/pubmed/11100729. Jing, B. et al., 2015. Dynamic properties of the segmentation clock mediated by microRNA. International journal of clinical and experimental pathology, 8(1),

128

pp.196–206. Jing, Q. et al., 2005. Involvement of MicroRNA in AU-Rich Element-Mediated mRNA Instability. Cell, 120(5), pp.623–634. Johnston, D.S. et al., 1992. The origin of pattern and polarity in the Drosophila embryo. Cell, 68(2), pp.201–219. Available at: http://linkinghub.elsevier.com/retrieve/pii/009286749290466P [Accessed May 27, 2016]. Johnston, S.H. et al., 1997. A family of mammalian Fringe genes implicated in boundary determination and the Notch pathway. Development (Cambridge, England), 124(11), pp.2245–54. Available at: http://www.ncbi.nlm.nih.gov/pubmed/9187150. Jülich, D. et al., 2005. beamter/deltaC and the role of Notch ligands in the zebrafish somite segmentation, hindbrain neurogenesis and hypochord differentiation. Developmental biology, 286(2), pp.391–404. Available at: http://www.ncbi.nlm.nih.gov/pubmed/16125692 [Accessed May 28, 2016]. Kim, J., Irvine, K.D. & Carroll, S.B., 1995. Cell recognition, signal induction, and symmetrical gene activation at the dorsal-ventral boundary of the developing Drosophila wing. Cell, 82(5), pp.795–802. Available at: http://www.ncbi.nlm.nih.gov/pubmed/7671307 [Accessed May 28, 2016]. Koch, A.J. & Meinhardt, H., 1994. Biological Pattern Formation: form Basic Mechanisms to Complex Structures. Reviews of Modern Physics, 66, pp.1481–1508. Koizumi, K. et al., 2001. The role of presenilin 1 during somite segmentation. Development (Cambridge, England), 128(8), pp.1391–402. Available at: http://www.ncbi.nlm.nih.gov/pubmed/11262239 [Accessed May 29, 2016]. Kopan, R. & Ilagan, M.X.G., 2009. The Canonical Notch Signaling Pathway: Unfolding the Activation Mechanism. Cell, 137(2), pp.216–233. Krol, A.J. et al., 2011. Evolutionary plasticity of segmentation clock networks. Development (Cambridge, England), 138(13), pp.2783–92. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3109603&tool=pmcen trez&rendertype=abstract [Accessed January 30, 2014]. Lamar, E. et al., 2001. Nrarp is a novel intracellular component of the Notch signaling pathway. Genes & development, 15(15), pp.1885–99. Available at: http://www.ncbi.nlm.nih.gov/pubmed/11485984 [Accessed May 29, 2016]. LeBon, L. et al., 2014. Fringe proteins modulate Notch-ligand cis and trans interactions to specify signaling states. eLife, 3, p.e02950. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=4174579&tool=pmcen trez&rendertype=abstract. Lecuit, T. & Lenne, P.-F., 2007. Cell surface mechanics and the control of cell shape, tissue patterns and morphogenesis. Nature reviews. Molecular cell biology, 8(8), pp.633–644. Lewis, B.P., Burge, C.B. & Bartel, D.P., 2005. Conserved Seed Pairing, Often Flanked by

129

Adenosines, Indicates that Thousands of Human Genes are MicroRNA Targets. Cell, 120(1), pp.15–20. Lewis, J., 2003. Autoinhibition with Transcriptional Delay: A Simple Mechanism for the Zebrafish Somitogenesis Oscillator. Current biology : CB, 13(7), pp.1398–1408. Available at: http://www.ncbi.nlm.nih.gov/pubmed/12676087. Loomes, K.M. et al., 2007. Dll3 and Notch1 genetic interactions model axial segmental and craniofacial malformations of human birth defects. Developmental dynamics : an official publication of the American Association of Anatomists, 236(10), pp.2943–51. Available at: http://www.ncbi.nlm.nih.gov/pubmed/17849441 [Accessed June 2, 2016]. Lopez, T.P. & Fan, C.-M., 2013. Dynamic CREB family activity drives segmentation and posterior polarity specification in mammalian somitogenesis. Proceedings of the National Academy of Sciences, 110(22), pp.E2019–E2027. Available at: http://www.pnas.org/cgi/doi/10.1073/pnas.1222115110 [Accessed May 29, 2016]. Luther, K.B. & Haltiwanger, R.S., 2009. Role of unusual O-glycans in intercellular signaling. The International Journal of Biochemistry & Cell Biology, 41(5), pp.1011– 1024. Mansouri, A. et al., 1997. Paired-related murine homeobox gene expressed in the developing sclerotome, kidney, and nervous system. Developmental Dynamics, 210(1), pp.53–65. Mara, A. et al., 2007. Priming, initiation and synchronization of the segmentation clock by deltaD and deltaC. Nature cell biology, 9(5), pp.523–530. Maroto, M. et al., 2005. Synchronised cycling gene oscillations in presomitic mesoderm cells require cell-cell contact. , 315, pp.309–315. Masamizu, Y. et al., 2006. Real-time imaging of the somite segmentation clock: revelation of unstable oscillators in the individual presomitic mesoderm cells. Proceedings of the National Academy of Sciences of the United States of America, 103(5), pp.1313–8. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1345707&tool=pmcen trez&rendertype=abstract. McGrew, M. et al., 1998. The lunatic fringe gene is a target of the molecular clock linked to somite segmentation in avian embryos. Curr Biol, 8(17), pp.979–982. Available at: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt =Citation&list_uids=9742402. McInerney, D. et al., 2004. A mathematical formulation for the cell-cycle model in somitogenesis: analysis, parameter constraints and numerical solutions. Mathematical medicine and biology : a journal of the IMA, 21(2), pp.85–113. Available at: http://eutils.ncbi.nlm.nih.gov/entrez/eutils/elink.fcgi?dbfrom=pubmed&id=152281 01&retmode=ref&cmd=prlinks\npapers2://publication/uuid/EAAB1D9C-52A0-

130

435F-B038-9E65F8E62AF3. Mciver, S.C. et al., 2012. miRNA and mammalian male germ cells. Human Reproduction Update, 18(1), pp.44–59. Meinhardt, H., 1986. Models of Segmentation. In Somites in Developing Embryos. Boston, MA: Springer US, pp. 179–189. Available at: http://link.springer.com/10.1007/978-1-4899-2013-3_14 [Accessed May 28, 2016]. Mohr, O.L., 1919. Character Changes Caused by Mutation of an Entire Region of a Chromosome in Drosophila. Genetics, 4(3), pp.275–82. Available at: http://www.ncbi.nlm.nih.gov/pubmed/17245926 [Accessed May 28, 2016]. Moloney, D.J. et al., 2000. Fringe is a glycosyltransferase that modifies Notch. Nature, 406(6794), pp.369–75. Available at: http://www.ncbi.nlm.nih.gov/pubmed/10935626. Monk, C.E., Hutvagner, G. & Arthur, J.S.C., 2010. Regulation of miRNA transcription in macrophages in response to Candida albicans. PloS one, 5(10), p.e13669. Monk, N.A.M. et al., 2003. Oscillatory Expression of Hes1, p53, and NF-κB Driven by Transcriptional Time Delays. Current Biology, 13(16), pp.1409–1413. Available at: http://linkinghub.elsevier.com/retrieve/pii/S0960982203004949 [Accessed May 30, 2016]. Morales, A. V et al., 2002. Periodic Lunatic fringe Expression Is Controlled during Segmentation by a Cyclic Transcriptional Enhancer Responsive to Notch Signaling. Developmental cell, 3, pp.63–74. Morgan, T. & CB, B., 1916. Sex-linked inheritance in Drosophila. Publs Carnegie Instn., 237, pp.1–88. Morimoto, M. et al., 2005. The Mesp2 transcription factor establishes segmental borders by suppressing Notch activity. Nature, 435(7040), pp.354–9. Available at: http://www.ncbi.nlm.nih.gov/pubmed/15902259 [Accessed May 28, 2016]. Morin-Kensicki, E.M., Melancon, E. & Eisen, J.S., 2002. Segmental relationship between somites and vertebral column in zebrafish. Development (Cambridge, England), 129(16), pp.3851–3860. Munro, S. & Freeman, M., 2000. The Notch signalling regulator Fringe acts in the Golgi apparatus and requires the glycosyltransferase signature motif DxD. Current Biology, 10(14), pp.813–820. Naiche, L. a, Holder, N. & Lewandoski, M., 2011. FGF4 and FGF8 comprise the wavefront activity that controls somitogenesis. Proceedings of the National Academy of Sciences of the United States of America, 108(10), pp.4018–4023. Nitanda, Y. et al., 2014a. 3’-UTR-dependent regulation of mRNA turnover is critical for differential distribution patterns of cyclic gene mRNAs. The FEBS journal, 281(1), pp.146–56. Available at: http://www.ncbi.nlm.nih.gov/pubmed/24165510 [Accessed February 14, 2014]. Nitanda, Y. et al., 2014b. 3′-UTR-dependent regulation of mRNA turnover is critical for differential distribution patterns of cyclic gene mRNAs. FEBS Journal, 281(1),

131

pp.146–156. Available at: http://doi.wiley.com/10.1111/febs.12582 [Accessed October 15, 2015]. Niwa, Y. et al., 2011. Different types of oscillations in Notch and Fgf signaling regulate the spatiotemporal periodicity of somitogenesis. Genes & development, 25(11), pp.1115–20. Available at: http://www.ncbi.nlm.nih.gov/pubmed/21632822 [Accessed May 29, 2016]. Niwa, Y. et al., 2007. The Initiation and Propagation of Hes7 Oscillation Are Cooperatively Regulated by Fgf and Notch Signaling in the Somite Segmentation Clock. Developmental Cell, 13(2), pp.298–304. Oates, A.C. et al., 2012. Patterning embryos with oscillations: structure, function and dynamics of the vertebrate segmentation clock. Development (Cambridge, England), 139(4), pp.625–39. Available at: http://www.ncbi.nlm.nih.gov/pubmed/22274695 [Accessed May 29, 2016]. Oates, A.C., Mueller, C. & Ho, R.K., 2005. Cooperative function of deltaC and her7 in anterior segment formation. Developmental Biology, 280(1), pp.133–149. Available at: http://linkinghub.elsevier.com/retrieve/pii/S0012160605000370 [Accessed May 28, 2016]. Oginuma, M. et al., 2010. The oscillation of Notch activation, but not its boundary, is required for somite border formation and rostral-caudal patterning within a somite. Development (Cambridge, England), 137(9), pp.1515–1522. Oginuma, M., Hirata, T. & Saga, Y., 2008. Identification of presomitic mesoderm (PSM)- specific Mesp1 enhancer and generation of a PSM-specific Mesp1/Mesp2-null mouse using BAC-based rescue technology. Mechanisms of Development, 125(5-6), pp.432–440. Available at: http://www.ncbi.nlm.nih.gov/pubmed/18328678 [Accessed May 29, 2016]. Okubo, Y. et al., 2012. Lfng regulates the synchronized oscillation of the mouse segmentation clock via trans-repression of Notch signalling. Nature Communications, 3, p.1141. Available at: http://dx.doi.org/10.1038/ncomms2133. Ozbudak, E.M. & Lewis, J., 2008. Notch signalling synchronizes the zebrafish segmentation clock but is not needed to create somite boundaries. PLoS genetics, 4(2), p.e15. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2222926&tool=pmcen trez&rendertype=abstract [Accessed September 19, 2014]. Palmeirim, I. et al., 1997. Avian hairy gene expression identifies a molecular clock linked to vertebrate segmentation and somitogenesis. Cell, 91(5), pp.639–48. Available at: http://www.ncbi.nlm.nih.gov/pubmed/9393857. Pankratz, M. J. and Jackle, J., 1993. Blastoderm segmentation. In The Development of Drosophila melanogaster. pp. 467–516. Park, S.-M. et al., 2008. The miR-200 family determines the epithelial phenotype of cancer cells by targeting the E-cadherin repressors ZEB1 and ZEB2. Genes & Development, 22(7), pp.894–907. Available at:

132

http://www.genesdev.org/cgi/doi/10.1101/gad.1640608 [Accessed June 1, 2016]. Peel, A.D., Chipman, A.D. & Akam, M., 2005. Arthropod segmentation: beyond the Drosophila paradigm. Nature reviews. Genetics, 6(12), pp.905–16. Available at: http://www.ncbi.nlm.nih.gov/pubmed/16341071. Pourquié, O., 2011. Vertebrate segmentation: from cyclic gene networks to scoliosis. Cell, 145(5), pp.650–63. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3164975&tool=pmcen trez&rendertype=abstract [Accessed January 22, 2014]. Pourquié, O. & Tam, P.P., 2001. A nomenclature for prospective somites and phases of cyclic gene expression in the presomitic mesoderm. Developmental cell, 1(5), pp.619–20. Available at: http://www.ncbi.nlm.nih.gov/pubmed/11709182 [Accessed June 2, 2016]. Primmett, D.R., Stern, C.D. & Keynes, R.J., 1988. Heat shock causes repeated segmental anomalies in the chick embryo. Development (Cambridge, England), 104(2), pp.331–9. Available at: http://www.ncbi.nlm.nih.gov/pubmed/3254821. Reifers, F. et al., 1998. Fgf8 is mutated in zebrafish acerebellar (ace) mutants and is required for maintenance of midbrain-hindbrain boundary development and somitogenesis. Development (Cambridge, England), 125(13), pp.2381–95. Available at: http://www.ncbi.nlm.nih.gov/pubmed/9609821 [Accessed May 29, 2016]. Reinitz, J., 2012. Turing centenary: Pattern formation. Nature, 482(7386), pp.464–464. Available at: http://www.nature.com/nature/journal/v482/n7386/full/482464a.html\nhttp://w ww.nature.com/nature/journal/v482/n7386/pdf/482464a.pdf. Richardson, M.K. et al., 1998. Somite number and vertebrate evolution. …, 125(2), pp.151–160. Available at: http://dev.biologists.org/content/125/2/151.long\npapers3://publication/uuid/B0 1BD746-CC1A-4915-B07A-F21013BB70AF. Riedel-Kruse, I.H., Müller, C. & Oates, A.C., 2007. Synchrony dynamics during initiation, failure, and rescue of the segmentation clock. Science (New York, N.Y.), 317(2007), pp.1911–1915. Riley, M.F. et al., 2013. Mir-125a-5p-mediated regulation of Lfng is essential for the avian segmentation clock. Developmental cell, 24(5), pp.554–61. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3799988&tool=pmcen trez&rendertype=abstract [Accessed October 15, 2015]. Roberts, M., Reiss, M.J. & Monger, G., 2000. Advanced Biology, Saga, Y. et al., 2000. Mesp1 Expression Is the Earliest Sign of Cardiovascular Development. Trends in Cardiovascular Medicine, 10(8), pp.345–352. Available at: http://linkinghub.elsevier.com/retrieve/pii/S105017380100069X [Accessed May 29, 2016]. Saga, Y., 2012. The mechanism of somite formation in mice. Current Opinion in Genetics and Development, 22(4), pp.331–338. Available at:

133

http://dx.doi.org/10.1016/j.gde.2012.05.004. Saga, Y. & Takeda, H., 2001. The making of the somite: molecular events in vertebrate segmentation. Nature reviews. Genetics, 2(November), pp.835–845. Schiffmann, Y., 2004. Segmentation and zooid formation in animals with a posterior growing region: The case for metabolic gradients and Turing waves. Progress in Biophysics and Molecular Biology, 84(1), pp.61–84. Schoppmeier, M. & Damen, W.G.M., 2005. Suppressor of Hairless and Presenilin phenotypes imply involvement of canonical Notch-signalling in segmentation of the spider Cupiennius salei. Developmental Biology, 280(1), pp.211–224. Serth, K. et al., 2015. O-fucosylation of DLL3 Is required for its function during somitogenesis. PLoS ONE, 10(4), pp.1–21. Serth, K. et al., 2003. Transcriptional oscillation of Lunatic fringe is essential for somitogenesis. Genes and Development, 17(7), pp.912–925. Shao, Y. et al., 2014. CRISPR/Cas-mediated genome editing in the rat via direct injection of one-cell embryos. Nature Protocols, 9(10), pp.2493–2512. Available at: http://www.nature.com/doifinder/10.1038/nprot.2014.171 [Accessed May 26, 2016]. Shifley, E.T. et al., 2008. Oscillatory lunatic fringe activity is crucial for segmentation of the anterior but not posterior skeleton. Development (Cambridge, England), 135(5), pp.899–908. Available at: http://www.ncbi.nlm.nih.gov/pubmed/18234727 [Accessed February 14, 2014]. Shih, N.P. et al., 2015. Dynamics of the slowing segmentation clock reveal alternating two-segment periodicity. Development, 142(10), pp.1785–93. Available at: http://www.ncbi.nlm.nih.gov/pubmed/25968314. Shimojo, H. et al., 2016. Oscillatory control of Delta-like1 in cell interactions regulates dynamic gene expression and tissue morphogenesis. Genes and Development, 30(1), pp.102–116. Soza-Ried, C. et al., 2014. Pulses of Notch activation synchronise oscillating somite cells and entrain the zebrafish segmentation clock. Development (Cambridge, England), 141(8), pp.1780–8. Available at: http://www.ncbi.nlm.nih.gov/pubmed/24715465. Sparrow, D.B. et al., 2008. Mutation of Hairy-and-Enhancer-of-Split-7 in humans causes spondylocostal dysostosis. Human molecular genetics, 17(23), pp.3761–6. Available at: http://www.ncbi.nlm.nih.gov/pubmed/18775957 [Accessed May 28, 2016]. Sparrow, D.B. et al., 2006. Mutation of the LUNATIC FRINGE gene in humans causes spondylocostal dysostosis with a severe vertebral phenotype. American journal of human genetics, 78(1), pp.28–37. Available at: http://www.ncbi.nlm.nih.gov/pubmed/16385447 [Accessed May 28, 2016]. Stauber, M. et al., 2012. Modifying transcript lengths of cycling mouse segmentation genes. Mechanisms of Development, 129(1-4), pp.61–72. Available at: http://dx.doi.org/10.1016/j.mod.2012.01.006. Stern, C.D. et al., 2006. Head-tail patterning of the vertebrate embryo: One, two or

134

many unresolved problems? International Journal of Developmental Biology, 50(1), pp.3–15. Stern, S., 1988. A cell lineage of segmentation in the chick embryo. Development, 244, pp.231–244. Stickney, H.L., Barresi, M.J.F. & Devoto, S.H., 2000. A PEER REVIEWED FORUM Somite Development in Zebrafish. Developmental Dynamics, 303(July), pp.287–303. Stollewerk, A., Schoppmeier, M. & Damen, W.G.M., 2003. Involvement of Notch and Delta genes in spider segmentation. Nature, 423(6942), pp.863–865. Sun, Y.-M., Lin, K.-Y. & Chen, Y.-Q., 2013. Diverse functions of miR-125 family in different cell contexts. Journal of hematology & oncology, 6(1), p.6. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3566921&tool=pmcen trez&rendertype=abstract. Suriben, R., Fisher, D.A. & Cheyette, B.N.R., 2006. Dact1 presomitic mesoderm expression oscillates in phase with Axin2 in the somitogenesis clock of mice. Developmental Dynamics, 235(11), pp.3177–3183. Swiatek, P.J. et al., 1994. Notch 1 is essential for postimplantation development in mice. Genes Dev, 8(1992), pp.pp. 707–719. Tajbakhsh, S. & Spörle, R., 1998. Somite development: Constructing the vertebrate body. Cell, 92(1), pp.9–16. Takada, S. et al., 1994. Wnt-3a regulates somites and tailbud formation in the mouse embryo. Genes and Development, 8, pp.174–189. Takahashi, Y. et al., 2007. Appropriate suppression of Notch signaling by Mesp factors is essential for stripe pattern formation leading to segment boundary formation. Developmental Biology, 304(2), pp.593–603. Takahashi, Y. et al., 2000. Mesp2 initiates somite segmentation through the Notch signalling pathway. Nature genetics, 25(4), pp.390–6. Available at: http://www.ncbi.nlm.nih.gov/pubmed/10932180 [Accessed May 29, 2016]. Takahashi, Y. & Sato, Y., 2008. Somitogenesis as a model to study the formation of morphological boundaries and cell epithelialization. Development, growth & differentiation, 50 Suppl 1, pp.S149–55. Available at: http://www.ncbi.nlm.nih.gov/pubmed/18482400 [Accessed September 19, 2014]. Takashima, Y. et al., 2011. Intronic delay is essential for oscillatory expression in the segmentation clock. Proceedings of the National Academy of Sciences of the United States of America, 108(8), pp.3300–3305. Takeuchi, H. & Haltiwanger, R.S., 2014. Significance of glycosylation in Notch signaling. Biochemical and Biophysical Research Communications, 453(2), pp.235–242. Talbot, J.C. & Amacher, S.L., 2014. A streamlined CRISPR pipeline to reliably generate zebrafish frameshifting alleles. Zebrafish, 11(6), pp.583–5. Tam, P., 1981. The control of somitogenesis in mouse embryos. Journal of Embryology and Experimental Morphology, 65, pp.103–28. Tatsumi, N. et al., 2015. Deficiency in WT1-targeting microRNA-125a leads to myeloid

135

malignancies and urogenital abnormalities. Oncogene, (April), pp.1–12. Available at: http://www.nature.com/doifinder/10.1038/onc.2015.154. Tautz, D., 2004. Segmentation. Developmental Cell, 7(3), pp.301–312. Turing, A.M., 1990. The chemical basis of morphogenesis. Bulletin of Mathematical Biology, 52(1-2), pp.153–197. Turnpenny, P.D. et al., 2007. Abnormal vertebral segmentation and the notch signaling pathway in man. Developmental Dynamics, 236(6), pp.1456–1474. Wahi, K., Bochter, M.S. & Cole, S.E., 2016. The many roles of Notch signaling during vertebrate somitogenesis. Seminars in Cell and Developmental Biology, 49, pp.68– 75. White, P.H. & Chapman, D.L., 2005. Dll1 is a downstream target of Tbx6 in the paraxial mesoderm. genesis, 42(3), pp.193–202. Available at: http://doi.wiley.com/10.1002/gene.20140 [Accessed May 29, 2016]. William, D.A. et al., 2007. Identification of oscillatory genes in somitogenesis from functional genomic analysis of a human mesenchymal stem cell model. Developmental Biology, 305(1), pp.172–186. Williams, D.R. et al., 2016. A novel targeted Lunatic fringe allele predicted to reduce protein secretion is dominant and disrupts somitogenesis. Development, (January). Available at: http://dev.biologists.org/cgi/doi/10.1242/dev.128538. Williams, D.R. et al., 2014. Posterior skeletal development and the segmentation clock period are sensitive to Lfng dosage during somitogenesis. Developmental biology, 388(2), pp.159–69. Available at: http://www.sciencedirect.com/science/article/pii/S0012160614000748. Wolpert, Lewis ; Tickle, Cheryll; Martinez Arias, A. et al, 2015. Principles of Development, Xu, H. et al., 2013. The cytoplasmic droplet may be indicative of sperm motility and normal spermiogenesis. Asian journal of andrology, 15(6), pp.799–805. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3854046&tool=pmcen trez&rendertype=abstract. Yang, H. et al., 2013. One-step generation of mice carrying reporter and conditional alleles by CRISPR/Cas-mediated genome engineering. Cell, 154(6), pp.1370–9. Available at: http://www.ncbi.nlm.nih.gov/pubmed/23992847 [Accessed January 20, 2014]. Yasuhiko, Y. et al., 2008. Functional importance of evolutionally conserved Tbx6 binding sites in the presomitic mesoderm-specific enhancer of Mesp2. Development (Cambridge, England), 135(21), pp.3511–9. Available at: http://www.ncbi.nlm.nih.gov/pubmed/18849530 [Accessed May 29, 2016]. Yasuhiko, Y. et al., 2006. Tbx6-mediated Notch signaling controls somite-specific Mesp2 expression. Proceedings of the National Academy of Sciences, 103(10), pp.3651– 3656. Available at: http://www.pnas.org/cgi/doi/10.1073/pnas.0508238103 [Accessed May 29, 2016].

136

Yu, Z. et al., 2006. Abnormalities of germ cell maturation and sertoli cell cytoskeleton in androgen receptor 113 CAG knock-in mice reveal toxic effects of the mutant protein. The American journal of pathology, 168(1), pp.195–204. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1592669&tool=pmcen trez&rendertype=abstract. Yuan, Y.P. et al., 1997. Secreted fringe-like signaling molecules may be glycosyltransferases. Cell, 88(1), pp.9–11. Available at: http://www.ncbi.nlm.nih.gov/sites/entrez?db=pubmed&cmd=Retrieve&dopt=Abst ractPlus&list_uids=9019410. Yusuf, F. & Brand-Saberi, B., 2006. The eventful somite: patterning, fate determination and cell division in the somite. Brain Structure and Function, 211(S1), pp.21–30. Available at: http://link.springer.com/10.1007/s00429-006-0119-8 [Accessed May 27, 2016]. Zeng, Y., Yi, R. & Cullen, B.R., 2005. Recognition and cleavage of primary microRNA precursors by the nuclear processing enzyme Drosha. The EMBO journal, 24(1), pp.138–48. Available at: http://apps.webofknowledge.com.emedien.ub.uni- muenchen.de/full_record.do?product=WOS&search_mode=GeneralSearch&qid=1 6&SID=V2GeaYeBOKbCcKR9yck&page=1&doc=1. Zhang, N. & Gridley, T., 1998. Defects in somite formation in lunatic fringe-deficient mice. Nature, 394(6691), pp.374–377. Zhang, S., Chung, W.-C. & Xu, K., 2015. Lunatic Fringe is a potent tumor suppressor in Kras-initiated pancreatic cancer. Oncogene, (October 2014), pp.1–11. Available at: http://www.nature.com/doifinder/10.1038/onc.2015.306. Zhang, Z. et al., 2011. The microRNA-processing enzyme Dicer is dispensable for somite segmentation but essential for limb bud positioning. Developmental Biology, 351(2), pp.254–265. Zhong, X. et al., 2010. Identification of microRNAs regulating reprogramming factor LIN28 in embryonic stem cells and cancer cells. Journal of Biological Chemistry, 285(53), pp.41961–41971. Zuker, M., 2003. Mfold web server for nucleic acid folding and hybridization prediction. Nucleic Acids Research, 31(13), pp.3406–3415.

137

APPENDIX A

PROMOTER ANALYSIS OF miR-125 FAMILY MEMBERS

A.1 Introduction

As discussed previously, miR-125a mediates regulation of Lfng in the chicken PSM

(Riley et al., 2013) and possibly in the tailbud region but not the anterior region of the mouse PSM (Chapter 2). In both chicken and mouse embryos, miR-125a is expressed in the PSM; in the mouse PSM miR-125a expression appears to be higher in the tailbud region and lower in the region anterior to the tailbud(Riley et al. 2013). So far it is unknown what transcriptional regulators induce the expression of miR-125a, particularly in the context of somitogenesis. To address this question we wanted to begin by identifying the miR-125a promoter region as this would aid in the discovery of enhancers and transcriptional regulators that bind to them. We also attempted to find the promoter region of miR-125b because it has the same seed sequence, thus the same predicted target mRNAs as miR-125a, but is expressed in the reverse pattern as compared to miR-125a (miR-125b is enriched in the somites but not the PSM)(Riley et al. 2013).

Identifying the promoters of miR-125a and its family member miR-125b is challenging as mature sequences of these miRNAs are found in clusters with other

138 miRNAs (Figure A.1). miR-125a is present in the genome in a tight intergenic cluster that contains miR-99b and let-7e genes. miR-125b is transcribed from two precursor structures pre-miR-125b1 and pre-miR-125b2, located on different chromosomes, which are also part of a miRNA cluster and each cluster is embedded in the intron of a linc

RNA. One chromosomal location contains miR-125b1 along with let7-a2 and miR-100 while the other contains miR-125b2 along with let7-c1 and miR-99a. Interestingly, in humans, the miR-125b1 and miR-125b2 clusters are thought to be transcribed as a tricistronic transcript(Emmrich et al. 2014). This was done by carrying out 5’RACE to verify previously predicted TSSs that are upstream to lnc RNA, rather than specifically carrying out 5’RACE for individual miRNAs in the cluster which may be a more reliable approach(Emmrich et al. 2014).

Figure A.1: Depiction of miR-125a, miR-125b1 and miR-125b2 clusters in the mouse genome. A. miR-125a is in a tight cluster with miR-99b and let-7e. miR-125b (continued) 139

(Figure A.1, continued) is transcribed from two sites in the genome B. miR-125b1, which is in a cluster with miR-100 and let-7a C. miR-125b2, which is in a cluster with miR-99a and let-7a.

For this study we have focused on miR-125a and miR-125b1 in the mouse genome since it is not known if a single transcript is produced from each of these clusters or if each of the miRNAs is transcribed as a separate primary transcript from the cluster i.e. is there a single promoter or multiple promoters controlling transcription of the miRNAs within a cluster. Additionally, the transcriptional start sites (TSS) for miRNAs are not completely annotated and even if they are known, the putative promoters have not been tested for their transcriptional ability. In this section we examine experimentally predicted promoters for their potential to transcribe a reporter gene. We also made an attempt to determine if the miRNA cluster is transcribed as single or multiple transcripts. We successfully identified promoter regions for miR-125b1 and miR-99b, and also found that miR-125b1 is transcribed as a single transcript, independent of the other two miRNAs in its cluster.

A.2 Materials and methods

A.2.1 Rapid Amplification of cDNA ends (RACE)

Total RNA was isolated from 10.5 dpc whole embryos using TRIzol® reagent

(Invitrogen). The GeneRacerTM Kit (Invitrogen) was used for Rapid Amplification of cDNA 140 ends (RACE) using manufacturer’s instructions. Briefly, shown in Figure A.2, to obtain the 5’ end of the pri-miRNA, the total RNA was first decapped by phosphatases. An RNA oligo was then attached to the uncapped 5’ end using T4 RNA ligase. The RNA was reverse transcribed to make cDNA using either random hexamers or oligo dT primers provided with the kit. To identify the 5’ ends of the pri-miRNAs, the first PCR as well as the nested PCR were performed using the forward primer provided in the kit and gene- specific reverse primers listed below:

miRNA Primers For 5'RACE (5'-3')

Rev 99b first primer ACACGGACCCACAGACACGAGCTT

Rev 99b nested primer CAAGGTCGGTTCTACGGGTGGGT

Rev 125a first primer CCAGGCTCCCAAGAACCTCACCTG

Rev 125a nested primer TCCTCACAGGTTAAAGGGTCTCAGG

Rev let7e first primer CTGGGGAAAGCTAGGAGGCCGTAT

Rev let7e nested primer CAACTATACAACCTCCTACCTCAGC

Rev 125b-1 first primer CTCCCAAGAGCCTAACCCGTGGATT

Rev 125b-1 nested primer TCACAAGTTAGGGTCTCAGGGACTG

Table A.1 Primers for 5'RACE PCR to identify the 5’ end of the primary transcript

The PCR product was cloned into pCR®4 TOPO® vector, as per the instructions in the manual, and sequenced using the T7 and SP6 primers. RACE for the 3’end was

141 performed similarly expect it excluded the decapping and ligation steps, and cDNA was synthesized using only the oligo dT primer. The forward gene specific primers used to amplify the 3’ end of the miRNAs are listed below (the reverse primer was used from the

Generacer™ kit):

miRNA Primers For 3'RACE (5'-3')

Fwd 99b first primer GCACCCACCCGTAGAACCGACCTT

Fwd 99b nested primer CAAGCTCGTGTCTGTGGGTCCGT

Fwd 125a first primer TGGGTCCCTGAGACCCTTTAACCT

Fwd 125a nested primer TCACAGGTGAGGTTCTTGGGAGCC

Fwd let7e first primer CCGGGCTGAGGTAGGAGGTTGTATA

Fwd let7e nested primer CACTATACGGCCTCCTAGCTTTCC

Fwd 125b-1 first primer TGCGCTCCCCTCAGTCCCTGAGA

Fwd 125b-1 nested primer CACGGGTTAGGCTCTTGGGAGCT

Table A.2 Primers for 3'RACE PCR to identify the 3’ end of the primary transcript

142

Figure A.2: Rapid Amplification of cDNA ends (RACE). (A-D) 5’RACE and (C and E)

3’RACE. The mature miRNA region is shown in red. (Modified from the Generacer kit manual)

143

A.2.2 Reporter constructs driven by predicted promoters

Promoter region for miR-99b and miR-125b1 were amplified using the following primers:

Potential promoter region Primers (5'-3')

miR-99bpromoter F TGAAGCTTCATTTCCTTCTTGCCTCTGA

miR-99bpromoter R TGAAGCTTCCCTCTCTTTTTCTCCTCCC

miR-125b1 promoter F TGAAGCTTGAGCTCTTAACTGTTTCTAGTT

miR-125b1promoter R TGAAGCTTAGAATGTGAGCTCTAGGCT

Table A.3 Primers to amplify the potential promoter region for miR-99b and 125b1

The promoter region was cloned into pGL3-Enhancer Luciferase reporter vector

(Promega) in the forward direction (pGL3-Enh-Fwd) to test its transcriptional ability and in the reverse direction as a negative control (pGL3-Enh-Rev).

A.2.3 Luciferase assays

NIH3T3 cells were plated at a concentration of 40,000 cells/well of a 24-well plate.

For each experiment, transfection of 100ng luciferase reporter (pGL3-Enh-empty, pGL3-

Enh-Fwd or pGL3-Enh-Rev) and 10ng pSVRenilla (Promega) was carried out in triplicate using Lipofectamine2000 (Invitrogen) transfection reagent. The empty pGL3-Enhancer 144 vector was also used as a negative control. Cells were lysed 40 hrs post-transfection and assayed for luciferase expression using the Dual Luciferase reporter kit (Promega). This protocol was repeated for three independent experiments.

A.3 Results

A.3.1 Regulation of the miR-125a cluster

RACE has been previously used to identify pri-miRs from the miR-23b~27b~24-1 cluster, where they found that each miRNA was regulated differently in the cluster(Sun et al. 2013). Similarly, we used the RACE method to identify the primary transcripts of miR-125a and the other two miRNAs in the cluster, miR-99b and let-7e (Figure A.3A).

We successfully amplified the 5’ end of miR-99b but not its 3’ end (Figure A.3B). We cloned and sequenced the 1kb band to obtain the 5’ capped end or TSS of pri-miR-99b

(Figure A.3C). This indicates that the region upstream of the 5’cap is likely to have promoter activity. This is discussed in the following section. The 1.6kb band turned out to be a non-specific band and sequenced to a region unrelated to the miR-125a locus.

We were unable to amplify the 5’ and 3’ ends of let-7e and miR-125a. One possible explanation for this could be that the 5’ and 3’ ends of the primary transcript for both these miRNAs could be at a large distance from the mature miRNA sequence, where the gene-specific primer binds, and the PCR conditions were not ideal to amplify such a large sequence, which would be especially likely if these three miRNAs are transcribed as one long transcript.

145

Figure A.3: Regulation of miR-125a cluster. A. The miR-125a cluster, miR-99b, let-7e and miR-125a lie within a 650 bp region of chromosome 17. B. RACE PCR for 5’end for pri-miR-99b run alongside a 1kb ladder (lane 1). The bands enclosed by the red box were cloned and sequenced to identify the 5’ end of pri-miR-99b (lane 4). Negative control

PCR reactions were set up with only forward primer (lane 2) or reverse primer (lane 3).

C. 5’ capped end of the miR-99b primary transcript determined after sequencing 5’ RACE

PCR products.

146

A.3.2 Identification of miR-99b promoter

To identify the miR-99b promoter, we examined promoter activity of the 1 kb region upstream of 5’ capped end of pri-miR-99b in driving expression of the luciferin gene in

3T3 cells (Figure A.4A). Since 3T3 cells show high expression of endogenous miR-125a and miR-125b (data not shown), it is the ideal cell line to use for promoter identification.

We inserted the potential promoter region upstream of the luciferase gene in a vector that does not have promoter but does contain enhancer elements. We found a significant difference in luciferase expression in cells that had the potential miR-99b promoter driving expression of the reporter gene as compared to the empty vector

(Figure A.4B). This indicates that the selected region can initiate transcription of the downstream gene and is very likely to be the miR-99b promoter. The identified miR-99b promoter is depicted in Figure A.4C. When the promoter region is inserted in reverse

(pGL3E-99bR) there appears to be a reduction in luciferase expression in comparison to the control cells (pGL3E) which indicates the absence of a TSS.

147

Figure A.4: Identification of miR-99b promoter. A. Potential miR-99b promoter inserted in pGL3-Enhancer vector. B. Significant increase in luciferase expression in cells transfected with pGL3-Enhancer containing the miR-99b promoter in forward direction

(pGL3E-99F). Values from three independent experiments are shown, with error bars depicting SEM. Results analyzed by Student’s t test (p<0.05). C. Promoter activity of the region upstream of the pri-miR-99b has been validated.

148

A.3.3 Identification of pri-miR-125b1 sequence

Since miR-125b1 is located 3kb away from the other two miRNAs in the cluster

(Figure A.5A), we carried out 5’ and 3’ RACE only for miR-125b1 to examine if it can be transcribed independent of miR-100 and let-7a. The PCR products from the RACE PCR are shown in Figure A.5B for 5’ RACE and Figure A.5C for 3’RACE. These were cloned and sequenced to obtain the 5’ and 3’ ends of the primary transcript (Figure A.5D). These results suggest that pri-miR-125b1 is transcribed independent of miR-100 and let-7e, and helps us locate potential promoter elements that lie upstream of the 5’ capped end of the miR-125b1 transcript.

A.4.4 Identification of the miR-125b1 promoter

To identify the miR-125b1 promoter we selected 945bp upstream of the 5’capped end of the pri-miR-125b1 and cloned it into the pGL3-Enhancer vector (Figure A.6A). We found a ten-fold increase in luciferase expression in cells transfected with the pGL3E- b1Fwd, where the promoter is in the forward direction, as compared to the control cells transfected with the empty vector (Figure A.6B). This indicates the selected region does exhibit promoter activity and is likely to function as the miR-125b1 promoter. Thus, we have successfully identified the miR-125b1 promoter region used to transcribe mature miR-125b1 (Figure A.6C). When the promoter is inserted in the reverse direction we do see a moderate increase in luciferase expression which may be due to cryptic transcription sites present in this region.

149

Figure A.5: Identification of miR-125b1 primary transcript A. The miR-125b1 cluster. miR-100, let-7a2 and miR-125b1 lie within a 4kb region of chromosome 9. B. RACE PCR for 5’end for pri-miR-125b1 run alongside a 1kb ladder (lane 1). The band enclosed by the red box was cloned and sequenced to identify the 5’ end of pri-miR-125b1 (lane 4).

Negative control PCR reactions were set up with only forward primer (lane 2) or reverse primer (lane 3). C. RACE PCR for 3’end for pri-miR-125b1 run alongside a 1kb ladder

(lane 1). The band enclosed by the blue box was cloned and sequenced to identify the 3’ end of pri-miR-125b1 (lane 4). D. 5’ capped end and 3’ poly A tail of the miR-125b1 primary transcript determined after sequencing 5’ and 3’ RACE PCR products.

150

Figure A.6: Identification of miR-125b1 promoter. A. Potential miR-125b1 promoter inserted in pGL3-Enhancer vector. B. Significant increase in luciferase activity in cells transfected with pGL3-Enhancer containing the miR-125b1 promoter in forward direction (pGL3E-125b1Fwd). Values from three independent experiments are shown, with error bars depicting SEM. Results analyzed by Student’s t test (p<0.05). C.

Promoter activity of the region upstream of the pri-miR-125b1 has been validated.

151

A.4 Discussion

Transcriptional regulation of miRNAs is a poorly studied area of miRNA biology.

Understanding how miRNA expression is induced could provide insight on crosstalk between different signaling pathways or help discover feedback loops within the same pathway. In the context of the segmentation clock given that miR-125a regulates Lfng in the chicken PSM and possibly in the mouse PSM, studying the transcriptional regulation miR-125a could provide novel information on the interaction between Notch signaling and other pathways(Riley et al. 2013).

To understand miR-125a regulation we carried out promoter analysis of the miR-

125a cluster and its family member miR-125b. We identified the promoter region for miR-99b that is part of the miR-125a cluster, after locating the TSS for miR-99b. We were unable to locate the TSS for let-7e and miR-125a, possibly due to the large distance from the TSS which they may share with miR-99b. We were also unable to locate the 3’ ends of the three miRNAs and hence do not have adequate evidence to support the hypothesis that these miRNAs are transcribed as a single transcript but get alternatively spliced out. It is interesting to note that regions near the predicted promoter of miR-99b/miR-125 cluster have binding sites for Tcf/Lef, which along with - catenin, controls transcription downstream of the Wnt pathway. The expression pattern of miR-125a in the mouse and chicken PSM is similar to that of -catenin and miR-125a is found to be upregulated by ERK1/2 and p38 pathways, that act downstream of the

152

Wnt pathway(Monk et al., 2010). This makes Wnt signaling an attractive candidate for regulation of miR-125a expression in the context of the segmentation clock.

153

APPENDIX B: SUPPLEMENTAL FIGURES

B.1 GFP transgenic mice

Design of GFP transgene likely to be flawed

The Lfng promoter driven GFP constructs were designed using a common strategy used previously to make GFP transgenic mice with either a wildtype or mutated Lfng 3'UTR

(Figure B.1). The expression of the GFP transcript in the PSM of transgenic embryos from these lines was compared by in situ hybridization using a GFP mRNA probe. We found no significant difference in expression patterns between the transgenic lines with the

intron after the stop codon at the 3’end could make these transcripts ideal targets for

Nonsense mediated decay (NMD, Nagy et al., 1998; Brogna et al.; 2009). While designing transgenes introns are used to enhance the expression of the transcript

(Brinster et al., 1998). However, having an intron downstream of the stop codon at a distance greater than 55 bp is a trigger for the NMD pathway. This results in degradation of these transcripts as the normal stop codon is now recognized as a premature stop codon by members of the NMD pathway which could explain the lack of transcript stabilization we saw in the dGFP-Mut UTR line (Zhang et al., 1998).

154

Figure B.1: GFP transgenic mice carrying a Lfng promoter driven GFP construct along with a PEST domain to destabilize the protein, a Wildtype or Mutated Lfng 3’UTR and a b-globin intron at the 3’end to ensure proper splicing. In situ hybridization for GFP transcript in transgenic mice with the A. Wildtype 3’UTR: GFP transcript shows oscillatory expression similar to Lfng mRNA in the PSM, B. Mutated 3’UTR: GFP transcript shows very similar expression to the Wildtype 3’UTR. C. A possible caveat of this approach: A one kb distance between the stop codon and exon junction complex

(EJC) at the 5’ end of the intron is a prime target for Nonsense mediated decay (NMD).

This is the likely cause for destabilizing the transcript in the GFP transgenic mice with the wildtype as well as the mutated Lfng 3’UTR.

155