Regulation of Notch Activation by Lunatic Fringe During

Home , HES7 gene, Segmentation (biology), Somitogenesis

REGULATION OF NOTCH ACTIVATION BY LUNATIC FRINGE DURING

SOMITOGENESIS

DISSERTATION

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

Dustin Ray Williams

Graduate Program in Molecular Genetics

The Ohio State University

2014

Dissertation Committee:

Susan Cole, Advisor

Heithem El-Hodiri

Mark Seeger

Amanda Simcox

Dustin Ray Williams

2014

ABSTRACT

During somitogenesis, paired somites periodically bud from the presomitic mesoderm (PSM) located at the caudal end of the embryo. These somites will give rise to the axial skeleton and musculature of the back. The regulation of this process is complex and occurs at multiple levels. In the posterior PSM, Notch activity levels oscillate as part of a clock that controls the timing of somite formation. In the anterior PSM, the Notch pathway is involved in somite patterning. In the clock, cyclic Notch activation is dependent upon periodic repression by the glycosyltransferase Lunatic fringe (LFNG).

Lfng mRNA levels cycle over a two-hour period in the clock, facilitating oscillatory

Notch activity. Lfng is also expressed in the anterior PSM, where it may regulate Notch activity during somite patterning.

We previously found that mice lacking overt oscillatory Lfng expression in the posterior PSM (Lfng∆FCE) exhibit abnormal anterior development but relatively normal posterior development, suggesting distinct requirements for segmentation clock activity during the formation of the anterior skeleton compared to the posterior skeleton and tail.

To further test this idea, we created an allelic series that progressively lowers Lfng levels in the PSM. We find that further reduction of Lfng expression levels in the PSM does not increase disruption of anterior development. However tail development is increasingly compromised as Lfng levels are reduced, suggesting that primary body formation is more sensitive to Lfng dosage than is secondary body formation. Further, we find that low levels of oscillatory Lfng are expressed in the posterior PSM of Lfng∆FCE mutants. This ii

reduced expression if sufficient to support relatively normal posterior development, however, we find that the period of the segmentation clock is increased when the amplitude of Lfng oscillations are low. These data support the hypothesis that there are differential requirements for oscillatory Lfng during primary and secondary body formation and that posterior development is less sensitive to overall Lfng levels as long as some oscillatory Lfng is present.

For oscillatory Lfng transcription to result in the cyclic Notch activation that is required for somitogenesis, there must be mechanisms in place to rapidly eliminate

LFNG protein during the short “off” phase of the clock. We hypothesized that secretion of LFNG is a mechanism for terminating its activity in the clock when it is no longer needed. To test the in vivo relevance of LFNG secretion, we generated knock-in mice expressing a Golgi-tethered LFNG variant that cannot be secreted (LfngtLFNG). Mice carrying a single copy of the LfngtLFNG allele exhibit severe skeletal abnormalities, supporting our hypothesis that tethering LFNG in the Golgi generates a dominant, hyperactive fringe protein that perturbs the segmentation clock. Somites in these mutants do not exhibit proper rostro-caudal patterning and form irregular boundaries. Expression of clock genes is perturbed, and Notch activity levels no longer oscillate. These results support our hypothesis that LFNG processing and secretion play important roles in its function in the segmentation clock, and provide further evidence that post-transcriptional regulation of the segmentation clock is critical during somitogenesis.

iii

DEDICATION

This document is dedicated to my grandmother, for all her love and support.

ACKNOWLEDGEMENTS

I would like to thank my advisor, Susan Cole, for her support, enthusiasm, humor, and her patience. I could not have asked for a more positive work environment.

I want to acknowledge previous lab members for their contributions to this work.

Emily targeted the mouse mutations used in this study and was a wonderful mentor in the early stages of my graduate career. I’d like to acknowledge Jason Lather for generating the Mesp2>Lfng construct. To Maurisa, thank you for all of your help and support during my graduate studies. I am so fortunate to have been able to work with my closest friend.

To Kanu and Skye, thank you for making the lab such an exciting and fun place to work.

I wish you all the best

VITA

May 8, 1985 ...... Born – Fairview, Georgia

2007...... B.S. Cellular Biology, University of Georgia

2007 to 2008 ...... College of Biological Sciences Dean’s

F Fellowship, The Ohio State University

2008 to present ...... Graduate Research and Teaching Associate,

Department of Molecular Genetics, The

Ohio State University

PUBLICATIONS

Williams DR, Shifley ET, Lather JD, Cole SE. 2014. Posterior skeletal development and the segmentation clock period are sensitive to Lfng dosage during somitogenesis. Dev Biol. 388: 159-169

Williams DR, Shifley ET, Cole SE. Secretion of Lunatic fringe is essential for somitogenesis and segmentation clock function. (In preparation)

FIELDS OF STUDY

Major Field: Molecular Genetics

TABLE OF CONTENTS

Abstract ...... ii

Dedication ...... iv

Acknowledgements ...... v

Vita ...... vi

Table of Contents ...... vii

List of Tables ...... xi

List of Figures ...... xii

List of Abbreviatinos ...... xiv

Chapter 1: Introduction ...... 1

1.1 Introduction ...... 1

1.2 Overview of somitogenesis ...... 1

1.3 The clock and wavefront model ...... 4

1.3.1 The wavefront ...... 5

1.3.2 The segmentation clock ...... 7

1.4 The Notch signaling pathway ...... 11

vii

1.4.1 Overview of Notch signaling ...... 11

1.4.2 Oscillatory Notch activation is essential for normal segmentation ...... 12

1.4.3 Notch is required for proper somite patterning ...... 13

1.5 Post-transcriptional regulation of the segmentation clock ...... 15

1.6 Lunatic fringe is essential for oscillatory Notch activation...... 18

1.7 Overview ...... 20

1.8 Figures ...... 21

Chapter 2: Posterior development and the segmentation clock period are sensitive to Lfng dosage during somitogenesis ...... 28

2.1 Introduction ...... 28

2.2 Methods ...... 31

2.2.1 Mouse Strains ...... 31

2.2.2 Transgenic mouse production and analysis ...... 31

2.2.3 Genotyping ...... 32

2.2.4 Whole mount in situ and immunohistochemistry analysis ...... 32

2.2.5 Western blot analysis ...... 33

2.2.6 Skeletal analysis ...... 34

2.2.7 RT-PCR ...... 35

2.3 Results ...... 35

viii

2.3.1 Posterior axial skeletal development is sensitive to Lfng dosage ...... 35

2.3.2 Lfng dosage affects rostral/caudal somite patterning ...... 38

2.3.3 Exogenous expression of Lfng in the anterior PSM does not rescue posterior

skeletal development ...... 38

2.3.4 Lfng∆FCE/∆FCE mice express low levels of dynamic Lfng in the posterior PSM 40

2.3.5 The rate of segmentation is altered in Lfng∆FCE mice ...... 43

2.4 Discussion ...... 45

2.4.1 The segmentation clock period is sensitive to Lfng dosage ...... 45

2.4.2 Oscillatory Lfng expression is required during all stages of axial

development...... 45

2.5 Figures ...... 458

Chapter 3: Secretion of Lunatic fringe is essential for somitogenesis and segmentation clock function...... 59

3.1 Introduction ...... 59

3.2 Methods ...... 61

3.2.1 Targeted mutation and genotyping ...... 61

3.2.2 Skeletal preparations and histology ...... 62

3.2.3 Whole-mount RNA in situ and histochemistry analysis ...... 62

3.3 Results ...... 63

3.3.1 Prevention of LFNG secretion perturbs segmentation ...... 63

3.3.2 LfngtLFNG/+ mutants do not form epithelial somites ...... 64

3.3.3 Somitic structures are caudalized in LfngtLFNG/+ mutants ...... 64

3.3.4 Endogenous Lfng transcription is reduced in LfngtLFNG/+ embryos ...... 66

3.3.5 LFNG secretion is required for cyclic activation of Notch pathway genes ...... 67

3.3.6 FGF and WNT signaling are also affected in LfngtLFNG/+ embryos ...... 69

3.4 Discussion ...... 70

3.5 Figures ...... 70

Chapter 4: Conclusion...... 82

4.1 Overview ...... 82

4.2 Oscillatory Lfng is required during both anterior and posterior development ...... 83

4.3 The segmentation clock period is sensitive to Lfng dosage in the PSM ...... 84

4.4 Secretion of LFNG regulates its activity in the segmentation clock ...... 85

4.5 mRNA turnover may be linked to the clock ...... 86

References ...... 87

LIST OF TABLES

Table 1 Quantification and analysis of skeletal defects in mice with varying levels of

Lfng expression in the PSM ...... 58

LIST OF FIGURES

Figure 1.1 Somitogenesis in mouse...... 22

Figure 1.2 The clock and wavefront model...... 23

Figure 1.3 The Notch signaling pathway ...... 24

Figure 1.4 Notch has multiple roles in the PSM ...... 25

Figure 1.5 Notch oscillations are achieved by a delayed negative-feedback loop...... 26

Figure 1.6 A model for post-translational regulation of LFNG...... 27

Figure 2.1 Posterior skeletal development is sensitive to Lfng dosage...... 50

Figure 2.2 Rostral/caudal patterning of epithelial somites is sensitive to the dosage of

Lfng in the PSM, and correlates with tail phenotypes...... 51

Figure 2.3 Exogenous expression of Lfng in the anterior PSM does not rescue posterior skeletal development in Lfng mutant mice...... 53

Figure 2.4 RT-PCR analysis detects low levels of dynamic Lfng expression in

Lfng∆FCE/∆FCE embryos ...... 54

Figure 2.5 NICD levels are dynamic in the posterior PSM of Lfng∆FCE/∆FCE embryos. the bottom ...... 55

Figure 2.6 Somitogenesis does not terminate prematurely in Lfng-/- embryos ...... 56

Figure 2.7 Lfng∆FCE/∆FCE mutants produce fewer somites and vertebrae ...... 57

xii

Figure 3.1 Targeting the endogenous Lfng locus to express Golgi-tethered LFNG protein ...... 73

Figure 3.2 Mice expressing Golgi-tethered LFNG protein exhibit segmentation abnormalities ...... 75

Figure 3.3 Segmentation and somite patterning are perturbed in LfngtLFNG/+ embryos .... 76

Figure 3.4 Lfng expression is downregulated in LfngtLFNG/+ mutants ...... 77

Figure 3.5 Notch activity is downregulated and no longer dynamic in LfngtLFNG/+ mutants...... 78

Figure 3.6 Hes7 mRNA turnover is perturbed in LfngtLFNG/+ embryos ...... 79

Figure 3.7 Expression of Axin2 and Spry2 is not dynamic in LfngtLFNG/+ embryos ...... 80

Figure 3.8 The segmentation clock in LfngtLFNG/+ mutants...... 81

xiii

LIST OF ABBREVIATIONS

CSL CBF1, Supressor of Hairless, LAG1, Rbpjk

Dll1 Deltalike1

Dll3 Deltalike3 d.p.c. Days post-coitus

DSL Delta/Serrate/Lag2 Notch ligands

EMT Epithelial-to-mesenchymal transition

EGF Epidermal growth factor

FGF Fibroblast growth factor

Lfng Lunatic fringe gene or mRNA

LFNG Lunatic fringe protein

NICD Notch intracellular domain

NRARP Notch-related ankyrin repeat protein

PSM Presomitic mesoderm

R/C Rostral/Caudal

UTR Untranslated region

xiv

CHAPTER 1

INTRODUCTION

1.1 Introduction

The Notch pathway is a cell-signaling pathway critical for animal development.

Dysregulation of the Notch pathway leads to birth defects and cancer. In many developmental contexts, it is critical that Notch activity be controlled both spatially and temporally. This thesis focuses on the regulation of Notch during somitogenesis, the process that gives rise to the axial skeleton. During somitogenesis, Notch activity oscillates to coordinate the periodic formation of somites, the precurors of the vertebrae and ribs. These oscillations are reliant upon cyclic repression of Notch by the glycosyltransferase Lunatic fringe (LFNG). Here, I find that oscillatory Lfng transcription is required throughout somitogenesis, although at lower levels during posterior development. Additionally, I find that secretion of LFNG is a mechanism for rapidly terminating LFNG activity, facilitating the oscillatory LFNG activity required for normal somitogenesis.

1.2 Overview of somitogenesis

Somitogenesis is the process by which paired somites periodically bud from an overtly segmented structure, the presomitic mesoderm (PSM), located at the caudal tip of 1

the embryo. These somites will give rise to the segmented structures of the vertebrate body, such as the vertebrae and ribs of the axial skeleton and the associated dermis and striated musculature of the back. The process of somitogenesis can be divided into two distinct phases: the production of somites anterior to the anus (primary body formation), and the generation of somites caudal to the anus (secondary body formation). During primary body formation, cells destined to enter the PSM migrate through the primitive streak during gastrulation. However, once gastrulation has ceased (approximately 10.5 d.p.c. in mouse), a complex structure called the tailbud forms, and cells in the PSM are generated by mitotic divisions in the tail bud mesenchyme.

The regulation of somitogenesis is complex and occurs at multiple levels in the

PSM. Although overtly unsegmented, the PSM can be divided into two regions based on function and gene expression patterns (Saga and Takeda 2001) (Figure 1.1). In the posterior PSM (region I), a molecular clock acts to organize cells into somite-sized cohorts. In the anterior PSM (region II), these presomites undergo rostro/caudal patterning and begin to epithelialize, forming somitomeres. Finally, segment boundaries form between the caudal compartment of one somite and the rostral compartment of the more posterior presomite, resulting in mature, segmented somites.

Once mature somites segment from the PSM, differentiation begins as somites are exposed to signaling molecules emanating from the notochord, neural tube, and lateral plate mesoderm. Cells forming the ventral-medial portion of the somite undergo an epithelial to mesenchymal transition (EMT), and the portion of the somite giving rise to these cells is termed the sclerotome. These cells give rise to the chondrocytes that ossify

to form the vertebrae and ribs. The most dorsal cells of the sclerotome will also give rise to the tendons. This region is known as the syndetome and is not morphologically distinct from the sclerotome, but it can be visualized by gene expression patterns (Brent et al.

2003, Schweitzer et al. 2001). The remaining epithelial portion of the somite, the dermamyotome, can be divided into two functional regions. The central region, the dermatome, gives rise to the dermis of the back. The portions of the somite lateral to the central dermatome are called the myotome. The cells of the myotome divide and give rise to a layer of myoblasts ventral to the dermatome and myotome (collectively referred to as the dermamyotome). The myoblasts formed from the medial myotome will form the intercostal muscles of the rib and the musculature of the back, whereas the myoblasts formed by the lateral myotome will give rise to muscles in the body wall and limbs.

The segment boundaries of the somites in the embryo do not directly correspond to the segments of adult structures. As embryogenesis continues, the rostral and caudal compartments of the somites separate and fuse with the compartments of adjacent somites in a process termed resegmentation. Resegmentation occurs in the sclerotome compartment of a somite, but not the dermamyotome. This process occurs as spinal neurons grow from the neural tube to innervate the myotome, which is lateral to the sclerotome. This results in a muscle segment interacting with two different vertebral elements, allowing for mobility between vertebrae (Huang et al. 2000). Somites also influence the migratory pathway of neural crest cells, peripheral motor neurons, and blood vessels, and contribute to the endothelium of the dorsal aorta (Keynes and Stern

1984, Rickmann et al. 1985, Sato et al. 2008). It is therefore critical that somitogenesis is

tightly regulated to prevent disruption of skeletal development and associated segmental structures.

1.3 The clock and wavefront model

Somites bud from the PSM with a periodicity that is specific for each species (2 hours in mouse, 90 minutes in chick), although the rate of somite formation can vary during development (Tam 1981). The periodic segmentation of somites is an extremely robust process. PSM that is dissected from the embryo will continue to segment in culture, and PSM that is dissected and transplanted in the reverse orientation continues to segment in the anterior-posterior orientation of the original PSM (Christ et al. 1974,

Packard 1976). In Xenopus, reducing the size of the embryo by surgical manipulation does not lead to a change in somite number, suggesting the embryo has a counting mechanism for determining the number of somites (Cooke 1975). These experiments supported the idea that a clock mechanism in the PSM controls the timing of somite formation and the species-specific somite number. The current model for somitogenesis is known as the clock and wavefront model (Cooke and Zeeman 1976) (Figure 1.2). In this model, a hypothetical molecular clock controls the periodicity of somite formation.

This segmentation clock oscillates with a period matching the rate of somite formation, and provides temporal information on when to form a somite boundary. The position of the segment border is determined by a wavefront that regresses from anterior to posterior as segmentation proceeds. As cells move through the PSM, they experience multiple pulses of clock oscillations, but are not competent to segment until they pass the wavefront. When cells experience a specific phase of the clock after passing the

wavefront, a segment boundary is specified and segmentation occurs. Thus, the distance traveled by the wavefront during one oscillation determines the size of a somite, and the posterior movement of the wavefront, coupled to oscillations of the clock, leads to periodic segmentation of properly sized somites (Aulehla and Herrmann 2004). At the time the clock and wavefront model was proposed, there was no molecular evidence supporting the existence of either the clock or the wavefront. Since the model was proposed, the signals comprising the wavefront activity have been identified, as well as more than 50 genes that exhibit oscillatory expression in the PSM.

1.3.1 The wavefront

The oscillations of the segmentation clock are thought to interact with a wavefront, positioned by intersecting signaling gradients, that provides positional information to specify the location of the somite boundary (Aulehla and Pourquie 2010).

Currently, the wavefront is thought to be positioned by gradients of FGF and Wnt activity emanating from the posterior of the embryo, which act to keep the PSM in an immature, undifferentiated state. The FGF and Wnt gradients are antagonized by a retinoic acid gradient emanating from the anterior of the embryo, which promotes differentiation. The role of a retinoic acid gradient is based on the expression and mutant phenotypes of enzymes that synthesize or degrade this vitamin A derivative, and has remained controversial (Abu-Abed et al. 2001, Fujii et al. 1997, Niederreither et al. 1997,

Niederreither et al. 1999, Sakai et al. 2001). However, a retinoic acid gradient has recently been visualized in zebrafish embryos using fluorescent resonance energy transfer

(Shimozono et al. 2013). In addition to its role in the wavefront, retinoic acid helps

coordinate somite formation between the left and right sides of the embryo by making the somites resistant to the left-right patterning signals that break the symmetry of other embryonic structures (Vilhais-Neto et al. 2010).

The first evidence that the FGF pathway contributes to the wavefront activity came from experiments in chick and zebrafish, where manipulating FGF activity with beads containing FGF8 shifted the position of the determination front, and altered somite size (Dubrulle et al. 2001, Sawada et al. 2001). It was later observed that downstream effectors of FGF signaling (MAPK/Erk in zebrafish and chick, Akt in mouse) are activated in a gradient in the PSM, suggesting there is a gradient of FGF activity that is strongest in the posterior of the embryo (Delfini et al. 2005, Dubrulle and Pourquie 2004,

Sawada et al. 2001). Direct genetic evidence remained elusive due to the early embryonic lethality of many FGF pathway mutants. Conditional inactivation of FGFR1, the only

FGF receptor expressed in the PSM, further supported a role for FGF signaling in both clock gene oscillations and in positioning the determination front. Conditional deletion of

Fgfr1 appears to result in gradual loss of FGF activity during somitogenesis, resulting in increased somite size and eventual arrest of the segmentation clock and somite formation

(Wahl et al. 2007). This is predicted by the clock and wavefront model, as loss of FGF results in posterior displacement of the wavefront, which should lead to larger somites.

However, genetic ablation of any single FGF molecule does not terminate somitogenesis, suggesting redundant functions for FGF ligands during somitogenesis. It is now accepted that, in mouse, FGF4 and FGF8 have redundant functions in the PSM and compose the wavefront activity, as loss of both molecules abolishes FGF activity and results in

premature differentiation of the entire PSM (Naiche et al. 2011).

Similar lines of evidence indicate a role for Wnt signaling in positioning the wavefront. The critical Wnt ligand in the PSM appears to be WNT3A, which is expressed in the most posterior tailbud and is required for axial elongation (Aulehla et al. 2003). A

WNT3A protein gradient has not been visualized, but is inferred by a gradient of nuclear beta-catenin in the PSM (Aulehla et al. 2008). Experimental manipulation of this beta- catenin gradient perturbs segmentation. Conditional inactivation of beta-catenin results in lack of posterior mesoderm and somites, whereas abolishing the Wnt gradient with stabilized nuclear beta-catenin enlarges the PSM and prevents somite differentiation, consistent with anterior displacement of wavefront activity (Aulehla et al. 2008).

However, it is clear that the Wnt and FGF gradients interact in the PSM. Loss of Wnt function perturbs FGF signaling in the PSM (Aulehla et al. 2003, Aulehla et al. 2008,

Morkel et al. 2003) and loss of FGF signaling results in loss of Wnt ligand expression

(Naiche et al. 2011). Thus, it appears likely that FGF and Wnt signaling are required in parallel and enforce one another to correctly position the determination front.

1.3.2 The segmentation clock

The first molecular evidence for the segmentation clock was the discovery that the chick gene c-hairy1 exhibits oscillatory expression in the PSM (Palmeirem et al., 1997). c-hairy1 encodes a member of the Hes (Hairy/Enhancer of split) family of transcriptional repressors and is a target of Notch signaling. In embryos of the same stage, different expression patterns of c-hairy1 were observed, and expression was dynamic within a single embryo (Palmeirim et al. 1997). The wave of c-hairy1 expression is not due to cell

movement, but represents a wave of transcription progressing anteriorly through the

PSM. This is the result of cells in the anterior PSM being in a different phase of the transcriptional oscillation than cells in the posterior PSM. Soon thereafter, the second oscillatory gene to be identified was the Notch regulator Lfng (Aulehla and Johnson

1999, Forsberg et al. 1998, McGrew et al. 1998). Regulatory elements that promote oscillatory transcription of Lfng in the PSM were later identified (Cole et al. 2002,

Morales et al. 2002). Although the Notch receptor is expressed throughout the PSM and does not appear to oscillate (Saga et al. 1997), the oscillation of Notch pathway members suggested that Notch activation itself is oscillatory in the PSM. Indeed, NOTCH1 was shown to have a dynamic pattern of activation by immunohistochemistry, with a periodicity matching the rate of somite formation (Morimoto et al. 2005).

Oscillatory gene expression also links the Wnt pathway to the clock. Axin2, a negative regulator of Wnt, also oscillates in the mouse segmentation clock. The oscillations of Axin2 are out of phase with the oscillations of Lfng, such that Axin2 reaches its peak of expression when Lfng expression is low (Aulehla et al. 2003). Another

Wnt regulator, Dact1, also oscillates in phase with Axin2 (Suriben et al. 2006). The Wnt regulator Nkd1 oscillates in phase with Lfng and is negatively regulated by the Notch target HES7, suggesting cross-talk between the Notch and Wnt pathways in the clock

(Ishikawa et al. 2004). However, unlike Notch activation, Wnt activitation is manifest as a gradient of nuclear beta-catenin. It is possible that subtle oscillations that cannot be detected biochemically are nevertheless sufficient for oscillatory transcription of Wnt targets. Alternatively, because only a subset of Wnt targets oscillate in the PSM, a Wnt

cofactor may cycle under the control of another signaling pathway.

Recently, microarray analysis has uncovered over 50 oscillatory genes in mice, primarily negative regulators of the Notch, Wnt, and FGF pathways, (Dequeant et al.

2006). The FGF and Notch targets identified oscillate out of phase with Wnt targets, suggesting reciprocal inhibition of these pathways plays a role in clock oscillations.

Intriguingly, some FGF and Wnt regulators are under the control of the Notch signaling pathways, such as Sprouty4 and Nkd1, respectively. Regulation of Wnt and FGF regulators by Notch activity may link the oscillations of the segmentation clock with the positional information that regulates boundary formation, which is influenced by the Wnt and FGF pathways. Some genes appear to oscillate in phase with Notch pathway members, but are targets of Wnt and FGF signaling, suggesting more complicated crosstalk between the FGF, Wnt, and Notch pathways. For example, Snail1 in mice and

Snail2 in chick oscillate in the PSM in phase with Lfng, and are expressed normally in

Notch1 and Rbpj mutants (Dale et al. 2006). The snail genes encode transcription factors that are involved in promoting EMT in several contexts, including development and cancer (reviewed in Barrallo-Gimeno and Nieto 2005). Downregulation of these genes appears to regulate the epithelialization of somites (Dale et al. 2006), and may coordinate clock oscillations with the mesenchymal to epithelial transition that occurs as somites mature.

Interestingly, although the signaling pathways involved in somitogenesis are conserved among vertebrates, the individual components of these pathways are not.

Microarray analysis and quantitative PCR were used to identify oscillatory genes in

zebrafish, mouse, and chicken (Krol et al. 2011). Although genes in the Notch, Wnt, and

FGF pathways were identified as cyclic in chick and mouse, different genes were identified as oscillatory in the two species. In zebrafish, fewer cyclic genes were identified, with only two representatives of the Wnt and FGF pathways, providing the first evidence that these pathways oscillate in zebrafish. Surprisingly, among the three species, conservation among individual clock components was limited to the Hes/Her family of transcriptional repressors. Thus, the segmentation clock has exhibited extreme plasticity during evolution, although three signaling pathways appear to be conserved in the segmentation clock mechanism of vertebrates.

It is unclear what signal initializes the segmentation clock. It has been shown that

FGF signaling is sufficient to propagate a transcriptional wave from a region of high FGF signaling to a region of low FGF signaling (Ishimatsu et al. 2010). It was therefore suggested that FGF signaling was the key activity that initiated the clock as axial elongation began. However, loss of FGF signaling does not result in premature differentiation until 6 hours after the onset of somitogenesis, strongly suggesting that a different, currently unknown signal initiates clock oscillations (Naiche et al. 2011).

Recently, the function of the clock in generating the periodicity of somite formation has been challenged by the observation that nonsomitic mesoderm can be induced to form multiple somites of the proper shape and size simultaneously by treatment with the BMP antagonist Noggin (Dias et al. 2014). However, the somites formed were not properly patterned. The researchers concluded that segmentation can be entirely controlled by local cell-cell interactions, and hypothesized that the function of

the clock is to couple the timing of somite formation with rostro/caudal patterning such that segmental borders form at the correct location relative to cells with rostral and caudal fates. It will be interesting to determine whether the segmentation observed in these experiments is reliant upon a functional clock.

1.4 The Notch signaling pathway

1.4.1 Overview of Notch signaling

The Notch pathway is a conserved juxtacrine cell-signaling pathway involved in regulating cell-fate and proliferation decisions in multiple contexts during development.

At the protein level, Notch receptors are single-pass transmembrane proteins containing multiple EGF repeats. Upon contact with a DSL ligand on an adjacent cell, the intracellular domain of notch undergoes a conformational change that allows it to be cleaved by the gamma-secretase complex, an intramembranous protease which includes presenilin 1 or 2. Cleavage of the Notch intracellular domain (NICD) results in its translocation to the nucleus, where it associates with CSL and Mastermind like proteins to activate transcription of target genes, such as members of the Hes family of transcriptional repressors (Fortini and Artavanis-Tsakonas 1994, Struhl and Adachi 1998)

(Figure 1.3). The mouse genome encodes 4 Notch receptors (NOTCH1/2/3/4) and 5 DSL

(Delta/Serrate/LAG-2) ligands (reviewed in Bolos et al. 2007). In mouse, NOTCH1 and

NOTCH 2 are expressed throughout the PSM (Huppert et al. 2005, Reaume et al. 1992), and the critical Notch ligands during somitogenesis appear to be DLL1 and DLL3 (Hrabe de Angelis et al. 1997, Kusumi et al. 1998). Several Notch targets exhibit oscillatory expression in the PSM as part of the segmentation clock, or are expressed in the R/C

patterning region. Mutations in these genes leads to severe skeletal birth defects in mouse and human (reviewed in Shifley and Cole 2007).

Because Notch receptors are widely expressed, it is critical that their regulation be under tight spatial and temporal control. One mechanism for this regulation is modification of the EGF repeats by glycosyltransferases, including members of the fringe family, which can potentiate or inhibit the interaction with Notch and its ligands depending on the fringe, ligand, and developmental context (Bruckner et al. 2000,

Moloney et al. 2000, Panin et al. 2002) (Figure 1.3). Post-translational regulation of

Notch is critical for its oscillatory activation, which is essential for proper segmentation.

1.4.2 Oscillatory Notch activation is essential for normal segmentation

In the posterior PSM, NOTCH1 activation oscillates with a period matching the rate of somite formation, suggesting a role in the segmentation clock mechanism.

NOTCH1 is critical for normal segmentation, as Notch1 knockout mice die at 9.5 d.p.c. with severe segmentation defects (Conlon et al. 1995). Embryos expressing NICD from a non-oscillatory promoter in the PSM have severe segmentation defects and aberrant expression of clock genes, indicating that oscillatory NICD is important for segmentation

(Feller et al. 2008).

The importance of oscillatory Notch activation can also be inferred from the phenotype of oscillatory Notch targets in the PSM. Oscillations of Notch are achieved through a negative feedback loop involving Lfng and Hes7 (Figure 1.4 and 1.5). In the clock, HES7 functions to periodically inhibit Lfng transcripton to allow for periodic

Notch activation. Hes7 knockout mice have severe segmentation defects and upregulated

Lfng and Hes7 expression (Bessho et al. 2001, Bessho et al. 2003), suggesting HES7 inhibits transcription of both itself and Lfng. Lfng knockouts also have severe segmentation defects, and NICD is constitutively active in the PSM in a gradient that is strongest in the anterior PSM, strongly suggesting that LFNG protein represses Notch activity in the clock. This feedback loop clearly receives input from other pathways, as inactivation of FGF and Wnt signaling abolish oscillations of Notch pathway members

(Wahl et al. 2007). It is known that Hes7 expression is initiated by FGF signaling in the posterior PSM, where FGF signaling is strongest (Niwa et al. 2007). In this way, FGF signaling may act as a permissive signal for Notch clock oscillations.

1.4.3 Notch is required for proper somite patterning

The phenotype of Notch pathway mutants suggests a role in somite patterning. In the anterior PSM, high levels of Notch upregulate Dll1 expression and further activate

Notch. In the anterior compartment, Notch activates Mesp2, which downregulates Dll1 and upregulates Dll3 and Lfng, leading to repression of Notch signaling (Morimoto et al.

2005, Takahashi et al. 2000) (Figure 1.4). MESP2 further represses Notch signaling by promoting the degradation of Mastermind in the anterior compartment (Sasaki et al.

2011). At least in mouse, the patterning of the somites appears to be closely linked to the clock. The Notch target Mesp2 is essential for both boundary formation and specification of the anterior somite compartment (Saga et al. 1997). Mesp2 transcription is activated by

Notch, but appears to be repressed by FGF and Wnt acitivty, which is strongest in the posterior and weakest in the anterior (Dunty et al. 2008, Oginuma et al. 2008). Thus, as a wave of Notch activation moves through the PSM, the transcriptional activation of Mesp2

becomes highest in the presumptive anterior compartment, where NICD is high and

FGF/Wnt are low, and lower in the posterior compartment. Thus, the R/C polarity of the somites may be a direct output of the segmentation clock.

The role of Notch in forming a segment boundary is more controversial. Early evidence that Notch is involved in boundary formation came from experiments in chick, where electroporation of Lfng mRNA or activated NOTCH1 results in the formation of a segment boundary (Sato et al. 2002). However, in zebrafish, pharmacological disruption of Notch activity does not result in segmentation defects for many cycles (Ozbudak and

Lewis 2008). It is generally accepted that in zebrafish, Notch activation has no function in the PSM other than to maintain synchronization between cells in the PSM (Delaune et al. 2012, Ozbudak and Pourquie 2008, Ozbudak and Lewis 2008).

The function of NICD may be different in amniotes based on experiments in mouse. Mice expressing NICD throughout the PSM from a transgene form up to 13 irregularly sized somites, suggesting oscillatory NICD is not required for border formation (Feller et al. 2008). The formation of several segment boundaries in these animals may be the result of Mesp2 activation by NICD in the anterior PSM, as Mesp2 was still expressed in a stripe pattern in this region. Additionally, Notch1 knockout and

Notch1/2 double knockout mice also form several somite-like structures. However, other lines of evidence suggest NICD is essential for border formation in mice.

Pharmacological disruption of the Notch pathway in mice leads to complete absence of segmental borders and abolishes cyclic expression of clock genes in the FGF, Wnt, and

Notch pathways (Ferjentsik et al. 2009). Mice mutant for Psen1/2 presumably develop in

the complete absence of NICD and fail to form somites (Huppert et al. 2005).

Interestingly, Rbpj knockout mice also form a small number of somite-like structures, although they still express low levels of NICD in the PSM (Ferjentsik et al. 2009). It is possible that Rbpj mutants retain non-canonical Notch signaling in the PSM, similar to the Suppressor of Hairless-independent activity described in Drosophila (Hori et al.

2004). Interestingly, even the smallest amounts of NICD, below the detection threshold of immunohistochemistry, are capable of rescuing somitogenesis as long as the levels of

NICD oscillate (Huppert et al. 2005). Thus, levels of Notch3/4 below the detection threshold for in situ analysis may be present in the PSM and extremely low levels of

NOTCH3 and/or 4 activation may allow for boundary formation in the most cervical somites. Thus, it is possible that NICD functions as a permissive signal for boundary formation by activating Mesp2, but that proper timing of border formation is reliant upon oscillatory Notch activity.

1.5 Post-transcriptional regulation of the segmentation clock

According to mathematical models, transcriptional regulation alone is not sufficient to account for the robust oscillations of Notch activity. Rather, robust oscillations of the clock are critically dependent on the time it takes to synthesize clock components and by their active half-lives. For the delayed negative-feedback loop to function, there must be a delay in expression of mRNA, accumulation of protein, and then negative autorepression, and models predict that shortening these delays will dampen or abolish oscillations (Feng and Navaratna 2007, Hirata et al. 2004, Jensen et al.

2003, Lewis 2003, Monk 2003) (Figure 1.5).

One obvious factor that could contribute to the transcriptional delay is the time required to transcribe oscillatory genes. It was predicted that lack of introns would shorten the transcriptional delay necessary for robust oscillations. This was tested in vivo by targeting an allele of Hes7 that lacked introns to the endogenous locus of mice. This shortened the transcriptional delay by 19 minutes and, consistent with mathematical models, oscillations of Hes7 mRNA and protein were abolished, resulting in severe segmentation defects (Takashima et al. 2011). Interestingly, removing only a subset of

Hes7 introns does not shorten the transcriptional delay enough to abolish oscillations, but instead dampens clock oscillations and also accelerates them, resulting in more vertebral elements (Harima et al. 2013). Together, this was taken as evidence that intronic delay was an important mechanism for creating the transcriptional delay necessary for sustaining the negative-feedback loop. However, recent studies have suggested that, rather than intronic delay, mRNA processing and export are the rate-limiting steps in the expression of Hes7 and other clock genes (Hoyle and Ish-Horowicz 2013). Thus, it seems likely that removal of Hes7 introns alters Hes7 mRNA processing and transport kinetics.

This may explain why increasing the length of the Lfng transcript has no effect on the segmentation clock, as the short increase in time required to transcribe larger genes is insignificant compared to the time required to process and export the mRNA to the cytoplasm (Stauber et al. 2012).

Regulation of mRNA stability and turnover is also critical for segmentation clock function. Little is known about mechanisms that regulate mRNA turnover during somitogenesis. The role of the 3’UTR in regulating mRNA stability and translational

effeciency is well-established (Kuersten and Goodwin 2003). In Xenopus, a 25bp motif in the 3’ UTR of hairy2a (the orthologue of c-hairy1) is required to destabilize the transcript and confer the proper segmental pattern (Davis et al. 2001). The 3’ UTRs of

Lfng and Hes7 contribute to the instability of those transcripts (Hilgers et al. 2005,

Nitanda et al. 2014). Because microRNAs regulate their target by binding to the 3’UTR, they were an attractive candidate for regulating mRNA turnover and translational efficiency of clock genes. A recent model of the clock suggested miRNA regulation of clock genes could contribute to the robustness of oscillations as long as the effect of the miRNA was significant (Xie et al. 2007). The first molecular evidence that microRNAs regulate the segmentation clock came from recent experiments in chick. Blocking a conserved miRNA binding site for mir-125a-5p in the 3’ UTR of chick Lfng leads to defects in somite boundary formation, patterning, and oscillatory gene expression (Riley et al. 2013). It is unknown whether this mechanism is conserved in mouse. Conditional inactivation of Dicer in the PSM of mice resulted in relatively normal segmentation until

10.0 dpc (Zhang et al. 2011). However, microRNAs have extremely long half-lives, up to

211 hours (Gantier et al. 2011), and microRNAs processed before inactivation of Dicer may have been sufficient to sustain normal segmentation during anterior development.

Introduction to the endogenous mouse locus of Lfng lacking mir-125a-5p binding sites will help resolve this question.

Lastly, regulation of protein half-life is critical for clock function. For oscillatory transcription to result in oscillatory protein levels, there must be mechanisms in place to confer a very short protein half-life. Both HES7 and LFNG protein levels have been

observed to cycle in the clock (Dale et al. 2003, Hirata et al. 2004). The half-life of HES7 is regulated by ubiquitination and subsequent degradation by the proteosome. Mutations that prevent ubiquitination of HES7 increase the in vitro protein half-life by only 8 minutes, yet perturb the segmentation clock in vivo, demonstrating that tight regulation of clock protein half-lives is critical for clock function (Hirata et al. 2004). LFNG is also regulated at the protein level. In vitro, LFNG is cleaved by SPC proprotein convertases and secreted into the extracellular space. Mutations that prevent LFNG cleavage prolong its active half-life in the cell (Shifley and Cole 2008). We propose that cleavage and secretion of LFNG serves as a mechanism to rapidly eliminate LFNG protein when its transcription is turned off, allowing for oscillatory LFNG protein levels in the clock

(Figure 1.6) As discussed in chapter 3, expression of a LFNG variant that cannot be secreted perturbs segmentation more severely than the complete loss of LFNG function, indicating secretion of LFNG is a critical aspect of its regulation in the segmentation clock. These results demonstrate that tight regulation of mRNA and protein levels are critical for clock function.

1.6 Lunatic fringe is essential for oscillatory Notch activation

Although there is debate about the specific role of Notch signaling during segmentation, it is clear that coordinated Notch activity is essential for normal somitogenesis, as embryos lacking NOTCH1 or experiencing sustained, non-oscillatory

Notch activation do not undergo normal segmentation (Feller et al. 2008). Because the

NOTCH1 receptor is expressed throughout the PSM, most regulation of Notch signaling is at the post-transcriptional level. The glycosyltransferase Lunatic fringe plays an

important role in regulating Notch activity in mouse and chick, and regulation of Lfng is the focus of this thesis. In mouse and chick, cyclic activation of NOTCH1 is reliant upon its periodic repression by Lunatic fringe. Lfng itself is a target of Notch activity, and functions in the segmentation clock as part of a delayed negative-feedback loop with

Hes7 and NICD (Figures 1.4 and 1.5). NICD activates transcription of both Lfng and

Hes7. As LFNG protein accumulates, it modifies the Notch receptor, inhibiting its interaction with DLL1. As HES7 protein accumulates, it represses transcription of both

Lfng and itself. Repression of Lfng, coupled with rapid turnover of Lfng mRNA and protein, allows for presentation of the unmodified Notch receptor at the cell surface, which can now be activated by its ligands, repeating the cycle.

Mice lacking Lfng have severe segmentation abnormalities including fused or missing ribs, “cobblestone” vertebral phenotype, and tail truncations (Evrard et al. 1998,

Zhang and Gridley 1998). These defects are presumably due to altered patterns of Notch activation in the PSM. Indeed, Notch activity levels are elevated and ubiquitous throughout the PSM of Lfng KO mice, although they may still be dynamic (Ferjentsik et al. 2009, Shifley et al. 2008). The regulatory elements that promote oscillatory Lfng transcription have been identified and characterized (Cole et al. 2002, Morales et al.

2002). Transcriptional oscillations of Lfng are clearly critical for somitogenesis, as mice expressing transgenic Lfng from a non-oscillatory promoter have severe segmentation defects and altered expression of clock genes, and cannot be rescued by deletion of endogenous Lfng (Serth et al. 2003). Additionally, deletion of the FCE1 enhancer that

promotes oscillatory Lfng transcription in the PSM results in severe segmentation defects anterior to the tail (Shifley et al. 2008).

1.7 Overview

We now know that the segmentation clock is regulated at multiple levels. The research presented in this thesis addresses the effects of altering the transcriptional and post-translational regulation of Lfng. Chapter 2 specifically addresses the effects of reducing oscillatory Lfng levels on skeletal development. Before this research was completed, it was believed that oscillatory Lfng and Notch activation were required for segmentation of the anterior skeleton, but were largely dispensable for segmentation of the posterior skeleton (Shifley et al. 2008, Stauber et al. 2009). In contrast, we find that high levels of oscillatory Lfng are critical for segmentation of the anterior skeleton, but only low levels of Lfng oscillations are required for segmentation of the posterior skeleton. Thus, oscillatory Lfng is required during both primary and secondary body formation. Additionally, we find that reducing oscillatory Lfng levels increases the period of the segmentation clock, such that fewer vertebral elements are formed in both the anterior and posterior skeleton.

Chapter 3 specifically addresses the regulation of LFNG protein levels in the segmentation clock. For LFNG to function in the clock, there must be mechanisms to confer a rapid protein half-life. However, little is known about post-translational mechanisms that regulate protein levels during somitogenesis. The work described in chapter 3 demonstrates that secretion of LFNG regulates its turnover in the PSM, and is essential for somitogenesis and segmentation clock function. This represents a novel

mechanism for regulating Notch signaling spatially and temporally by controlling the active half-life of a key Notch modulator.

1.8 Figures

Figure 1.1 Somitogenesis in mouse. A photograph (left) showing the presomitic mesoderm and mature somites of a 10.5 d.p.c. mouse. The caudal compartment of the mature, epithelialized somites is stained for Uncx. The most posterior band of Uncx is in the unsegmented region of the PSM, illustrating that the patterning process begins before somites have segmented from the PSM. A schematic of the PSM is shown on the right. In the posterior PSM (region I), oscillations of the segmentation clock group cells into somite-sized cohorts. In the anterior PSM (region II), these unsegmented presomites undergo rostral/caudal patterning that can be visualized only by gene expression patterns. Finally, a segment boundary is established and mature somites bud on either side of the neural tube (blue). Adapted from Saga and Takeda, 2001.

Figure 1.2 The clock and wavefront model. In the clock and wavefront model, a segmentation clock oscillates with a period that matches the rate of formation of each somite pair. As cells move through the PSM, they experience multiple phases of clock oscillations, but are not competent to segment until they pass the wavefront (horizontal bar). Only when cells are in a specific phase of the clock and have passed the wavefront does a segment boundary become specified. The interaction of the clock and wavefront accounts for the periodicity of segmentation and the consistent size of the somites.

Figure 1.3 The Notch signaling pathway. The Notch receptor is modified in the Golgi by fringe proteins and other glycosyltransferases before being presented to the cell surface. Binding of the extracellular domain of Notch to DSL ligands results in cleavage of Notch, releasing the intracellular domain (NICD). NICD translocates to the nucleus, where it forms a complex with CSL and Mastermindlike proteins, activating the transcription of downstream targets, many of which cycle in the PSM.

Figure 1.4 Notch has multiple roles in the PSM. In the posterior PSM, a negative feedback loop involving Lfng and Hes7 leads to oscillatory Notch activation. Oscillations of Notch are essential for synchronizing cells into somite-sized cohorts that can be properly patterned in the anterior PSM. In the anterior PSM, MESP2 protein represses Notch signaling by activating Lfng and by degrading Mastermindlike protein. This results in NICD levels accumulating in the posterior half of the presomite, establishing the initial rostral/caudal polarity of the developing somites. Adapted from Takahashi et al, 2000.

Figure 1.5 Notch oscillations are achieved by a delayed negative-feedback loop. Lfng functions in a delayed negative-feedback loop with NICD and Hes7. NICD activates transcription of both Lfng and Hes7. The accumulation of LFNG protein leads to the repression of Notch activity. However, HES7 protein represses transcription of Lfng and itself, allowing for presentation of the unmodified Notch receptor at the cell surface. Both mathematical data and in vivo experiments indicate that stable oscillations of Notch activity are reliant upon short half-lives of clock proteins and mRNAs.

Figure 1.6 A model for post-translational regulation of LFNG. Once Lfng transcription has been repressed in the clock, LFNG protein is still present, and able to inhibit Notch signaling (left). Cleavage of LFNG by convertases (scissors) leads to its rapid secretion into the extracellular space, where it is unable to modify the extracellular domain of Notch. Notch is now able to interact with its ligands and activate gene transcription (right). Combined with the cyclic transcription of LFNG, this post- translational regulation allows for cyclic Notch activation in the segmentation clock.

CHAPTER 21

POSTERIOR DEVELOPMENT AND THE SEGMENTATION CLOCK PERIOD ARE SENSITIVE TO LFNG DOSAGE DURING SOMITOGENESIS

2.1 Introduction

As discussed in chapter 1, the process of axial skeleton formation can be divided into two distinct stages producing somites anterior and posterior to the anus (Holmdahl

1925). Early in somitogenesis, as gastrulation proceeds, cells enter the PSM via the primitive streak in a process that has been called primary body formation. In mice, gastrulation is complete at the 30-31 somite stage (around 9.5-10.0 d.p.c.), and after this point mesodermal cells in the PSM arise from a population of undifferentiated cells called the tailbud mesenchyme (TBM) during secondary body formation (Wilson and

Beddington 1996). The differences in primary and secondary body formation may be more distinct in mammals than in some other organisms (reviewed in Handrigan 2003), and the mechanisms regulating and differentiating primary and secondary body formation are poorly understood. However, the transition from primary to secondary body formation has important health implications. For example the lumbosacral junction

(defining the switch between primary and secondary body formation) is the location of

1 This chapter has been adapted from Williams et al., 2014 28

numerous caudal neural tube defects (NTDs) and malformations of the genito-urinary system (reviewed in Handrigan 2003).

Both clock function and rostral/caudal somite patterning in the anterior PSM are critical for proper development of the axial skeleton, and the Notch signaling pathway has been proposed to play critical roles in both of these stages of somitogenesis. The expression pattern of Lfng suggests possible separable roles in the segmentation clock and during rostral/caudal somite patterning (Cole et al. 2002, Morales et al. 2002).

Indeed, Lfng null mice have distinct somitic defects. Somite size, shape, and boundary formation are irregular, which could be attributed to loss of Lfng expression in the segmentation clock. Additionally, somites are abnormally patterned, which might be attributed to loss of Lfng in the patterning region. This hypothesis was supported by the phenotype of mice lacking overt oscillatory Lfng expression in the PSM. These mice were homozygous for a deletion of the FCE1 enhancer, which promotes oscillatory Lfng expression in the PSM. This allele strongly reduced the expression of Lfng in the segmentation clock, while somewhat reducing its expression during pre-somite patterning. These mice exhibited abnormal size and spacing of somites at 9.5dpc, but somites underwent relatively normal R/C patterning at 10.5dpc, suggesting proper somite patterning could occur independently of oscillatory Notch activation. Additionally, this allele had severe effects on development of the anterior axial skeleton (primary body formation), but mild effects on the posterior skeleton (secondary body development), suggesting that the requirements for Notch signaling vary during different stages of mouse segmentation (Shifley et al. 2008).

Based on these results, it was hypothesized that oscillatory Lfng expression was dispensable for caudal development, and that expression of Lfng in the patterning region was more important during this stage of development (Shifley et al. 2008). It remained unclear how posterior somites could segment relatively normally without cyclic Lfng transcription and oscillatory Notch activation. One possibility is that, during caudal development, there is information transfer between the caudal somites and the PSM such that formation of posterior somites is reliant upon the proper R/C patterning of more anterior somites. This hypothesis is supported by the observation that mice lacking

Mesp2, a critical regulator of R/C patterning, have caudal truncations (Saga et al. 1997).

Alternatively, low levels of Lfng expression, below the detection threshold for in situ analysis, could be sufficient for relatively normal posterior development.

The research discussed in this chapter further analyzes the FCE1 deletion allele to resolve this question. We find that Lfng expression is highly reduced in these mice, but remains oscillatory in the posterior PSM of Lfng∆FCE mutants. This suggests that oscillatory Lfng expression is required for all stages of somitogenesis, although at reduced levels during secondary body formation. We find that Notch activation is dynamic in the posterior PSM of Lfng∆FCE mutants, although levels of activation are overall higher than wildtype embryos, and these oscillations are able to support relatively normal posterior segmentation but not anterior segmentation. However, as oscillatory

Lfng is further reduced, posterior segmentation becomes increasingly compromised.

Additionally, as the amplitude of Lfng oscillations decrease, the period of the

segmentation clock is increased such that embryos with reduced Lfng dosage form fewer vertebral elements during both primary and secondary body formation.

2.2 Methods

2.2.1 Mouse Strains

Lfng∆FCE mice have been previously described (Lfngtm1Seco) (Shifley et al. 2008).

They were maintained on a mixed 129/SvxC57BL6/J background, or crossed one generation onto an FVB background to introduce the LfngMesp2>Lfng transgene. LfngtmRjo1/+ mice (Evrard et al. 1998) were were maintained on a mixed 129SVxC57BL6J background. Throughout this document, LfngtmRjo1/ tmRjo1 mice are referred to as Lfng null, and a '-' symbol is used for this allele. Transheterozygous mice and embryos were produced from crosses between Lfng∆FCE/+ and LfngtmRjo1/+ mice. The LfngMesp2>Lfng transgene was first crossed onto the Lfng∆FCE background and then intercrossed with Lfng tmRjo1/+ mice. Since Lfng∆FCE/- mice are viable and fertile, wild type (Lfng∆FCE/+ or Lfng+/-),

Lfng∆FCE/∆FCE, Lfng∆FCE/- and Lfng-/- mice were produced as littermates to control for background variability. Mice were maintained in an SPF facility under the care of the

Ohio State University Laboratory Animal Resources department. All procedures were conducted under protocols approved by the Ohio State University IACUC.

2.2.2 Transgenic mouse production and analysis

The previously described promoter element from Mesp2 that is sufficient to drive expression in the anterior compartment of somite S-1 (Haraguchi et al. 2001) was used to drive expression of a transcript consisting of the full length Lfng cDNA cloned upstream of an IRES.LacZ.polyA cassette (Kim et al. 1992). Transgenes were injected into

fertilized eggs from FVB/J females and transplanted into pseudopregnant female mice at the OSUCCC Transgenic Mouse Core. Founders were bred to FVB/J females to establish stable lines of transgenic mice.

2.2.3 Genotyping

Genomic DNA was prepared from tail clips via proteinase K saltout, or from yolk sac fragments via the HOTSHOT procedure and animals were genotyped by PCR.

LfngtmRjo1 mice were genotyped with the primers FNG322 (5’-gagcaccaggagacaagcc-3’)

FNG325 (5’-agagttcctgaagcgagag-3’) and PGK3(5’-cttgtgtagcgccaagtgc-3’).

FNG322+FNG325 amplify a 170bp wt product, FNG325+PGK3 amplify a 200 bp mutant product. Lfng∆FCE1 mice were genotyped with the primers SC286(5’- ttgggtctatctgggaaacg-3’) and SC287(5’- gcgactcatccagacacaga-3’) producing a 149 bp wildtype band and a 250 bp mutant band (across the LoxP site in intron 1). The presence of the LfngMesp2>Lfng transgene was assessed using primers FNG237(5’- cgacattttgcagcacag-3’) and FNG239(5’-ttcaccgatggagacgac-3’), producing a 248bp transgene band, and SC340(5’-cagaatccagacctctgcaa-3’) and SC341(5’- accaggagacaagccaacag-3’), producing a 500bp positive control band.

2.2.4 Whole mount in situ and immunohistochemistry analysis

Embryos were collected from timed pregnancies. Noon of the day of plug identification was designated as 0.5 d.p.c. In some assays, embryos were further staged according to Theiler as supplemented by http://www.emouseatlas.org/emap/ema/staging_criteria/staging_criteria.html. Embryos were fixed overnight in 4% PFA. RNA in situ hybridization using digoxigenin-labeled

probes was performed essentially as described (Shifley et al. 2008). Probes included Lfng

(Johnston et al. 1997), Mesp2 (Saga et al. 1997), Uncx (Mansouri et al. 1997), and LacZ

(Shifley et al. 2008).

For whole-mount immunohistochemistry, embryos were fixed overnight in 4%

PFA. After washing in PBT, embryos were bleached overnight in 0.1% peroxide, 10% fetal calf serum, and 1% Triton X-100. After washes in 10mM sodium citrate (pH 6.0) with .1% Tween-20, embryos were boiled for 10 minutes, transferred to PBS, and incubated for 4 days in cleaved caspase-3 antibody (Cell Signaling Technology 1:1000 dilution) After extensive washing in MABT, embryos were incubated overnight in secondary antibody, then transferred to NTMT and stained with NBT/BCIP. For double label in situ, the Mesp2 probe was labeled with fluorescein and revealed by incubation with INT/BCIP as previously described (Shifley et al. 2008).

X-gal staining of embryos was performed essentially as described (Whiting et al.

1991). After fixation in 1% formaldehyde; 0.2% glutaraldehyde; 2mM MgCl2; 5mM

EGTA; 0.02% NP-40, embryos were washed in PBS + 0.02% NP40 and stained in 5mM

K3Fe(CN)6; 5mM K4Fe(CN)6; 2mM MgCl2; 0.01% deoxycholic acid; 0.02% NP-40; 1 mg/ml X-gal overnight or longer at room temperature. Wild type littermate embryos were included to control for background X-gal staining.

2.2.5 Western blot analysis

10.5 d.p.c. embryos were dissected in cold PBS and the posterior portion of the

PSM was harvested, and boiled in 30ul of Laemmli buffer for 5 minutes. Proteins were loaded onto 10% polyacrylamide gels and transferred to PVDF membranes. Protein

levels were detected using anti-NICD1 (Cell Signaling Technology 4147 1:400 dilution) and anti-tubulin tubulin (Sigma Aldrich T5168 1:1000 dilution). Secondary antibody conjugated to HRP (1:10,000) and ProtoGlow ECL (National Diagnostics CL-300) were used to expose blots following manufacturer's instructions. Band intensities were quantified using ImageJ relative density plots. Equal amounts of a standard containing pooled whole PSMs were loaded onto each gel to permit inter-gel comparisons. Tubulin and NICD levels were normalized to this standard sample and normalized values were used to scale NICD levels among samples.

2.2.6 Skeletal analysis

For skeletal preparations, embryos were harvested at 17.5 d.p.c. Alcian Blue and

Alizarin Red staining of the fetal skeleton was performed essentially as described

(McLeod 1980), except embryos were incubated for four days in 95% ethanol followed by three days in acetone. For quantitative analyses, we assessed the number of ribs attached at the sternum, and the number of uneven rib attachment points at the sternum.

The number of rib abnormalities (fusions and bifurcations) were recorded, with each individual rib fusion or bifurcation counted as one defect. The number of normal appearing vertebral condensations in the sacrum and tail were also quantified. Statistical analysis was performed using GraphPad. Statistical significance was assessed using the

Kruskal-Wallis test with Dunn post hoc when comparing more than two genotypes, or the

Mann-Whitney test when directly comparing two genotypes or a single genotype +/- the

LfngMesp2>Lfng transgene. Nonparametric tests were chosen, as several comparisons included groups that did not exhibit normal distributions or equal standard deviations. P

values <0.05 were considered significant, and P values are indicated in the tables and figure legends.

2.2.7 RT-PCR

10.5 d.p.c. embryos were dissected in cold PBS and the posterior portion of the

PSM was harvested. Total RNA was extracted with Trizol (Invitrogen) and used as a template for first-strand cDNA synthesis with the Superscript III kit (Invitrogen). Lfng cDNA was amplified using primers SC550(5’-gatcagcgagcacaaagtga-3’) and SC551(5’- agaatggtcccttgatgtgc-3’) yielding a 335 bp band. Gapdh cDNA was amplified with primers SC280(5'-aaggtcatcccagagctgaa-3') and SC281(5'-aggagacaacctggtcctca-3'), giving a 196 bp band. Mesp2 cDNA was amplified with primers SC692(5’- ttcgaggggtcagaatccac-3’) and SC693(5’-ggaaccggacgaatctgagg-3’) producing a 317 bp product. For quantitative RT-PCR, embryos were dissected as above, but total RNA was prepared using the Norgen Total RNA Purification Micro Kit. Endogenous Lfng RNA levels were assessed at the OSUCCC core using gene specific Taqman assays (Applied

Biosystems, Mm00456128_m1*). Lfng values were normalized to values for Gapdh, and the highest level of wild type expression was set to 100 for comparison across genotypes.

2.3 Results

2.3.1 Posterior axial skeletal development is sensitive to Lfng dosage

To examine the importance of Lfng dosage during somitogenesis, we took advantage of a previously generated allele, Lfng∆FCE, that deletes an enhancer required for robust oscillatory Lfng expression in the posterior PSM. By RNA in situ analysis,

Lfng∆FCE/∆FCE embryos lack overt expression of Lfng in the posterior PSM, where the

segmentation clock is active, and express reduced levels of Lfng in the rostro-caudal patterning region in the anterior PSM (Shifley et al. 2008). These mice have disorganized anterior skeletons, but segmentation is largely rescued at the lumbar-sacral junction, suggesting that Lfng could play distinct roles during primary and secondary body formation. However, recent data (Oginuma et al. 2010, Stauber et al. 2009) suggests that low levels of Lfng in the posterior PSM may be sufficient to rescue posterior development, and that somitogenesis might be differentially sensitive to dosage of Lfng during different stages of development. To test this, we further reduced Lfng dosage in the anterior PSM by crossing Lfng∆FCE/∆FCE with Lfng+/- mice to generate Lfng∆FCE/- embryos, and assessed skeletal phenotypes.

We find that only minor changes in the level of anterior skeletal disorganization are observed as Lfng expression levels are decreased from Lfng∆FCE/∆FCE to Lfng∆FCE/- to

Lfng-/- (Figure 2.1, Table 2.1). For instance, we observed no significant change in the number of rib abnormalities, the total number of ribs, or the number of uneven rib attachments as Lfng dosage decreased (Figure. 2.1A panels a-d, i-l, and 2.1B, Table 2.1).

Although neural arch morphology became somewhat more irregular with reduction of

Lfng dosage, significant mouse to mouse variability was observed even in mice of identical genotypes (Figure 2.1, panels e-h). Taken together these results suggest that high levels of oscillatory Lfng expression are critical during primary body formation, and that below a threshold that has already been met in Lfng∆FCE/∆FCE mutants, no significant further perturbations of the anterior skeleton occur as Lfng levels are further reduced.

In contrast, Lfng mutants differed substantially in the phenotype of the posterior

skeleton (Figure 2.1A panels m-p and 2.1C, Table 2.1). In the sacral region, development recovers in Lfng∆FCE/∆FCE and Lfng∆FCE/- animals, which usually form 3-4 normal condensations (Figure. 2.1A, panels m-p). Despite developing in the absence of all LFNG activity, Lfng-/- animals still form an average of 2 normal condensations, indicating that sacral development is somewhat refractory to reduction of Lfng, but Lfng-/- animals are still more severely affected in this region of the skeleton. In the tail, Lfng∆FCE/∆FCE mice have a slight decrease in the total number of tail vertebrae compared to wild type embryos, from an average of 30 to an average of 27. Further reduction of Lfng dosage in

Lfng∆FCE/- mice results in a significant loss of posterior vertebrae to an average of 23.

Although difficult to quantify, in the absence of LFNG activity the tail is significantly truncated with an average of only 15 distinct vertebrae (Figure. 2.1C, Table 2.1). As tail length decreases, disorganization of the vertebrae increases, with only 7% of vertebrae in the posterior regions of Lfng∆FCE/∆FCE animals being abnormal, compared to 73% in

Lfng∆FCE/-, and 85% in Lfng-/- (Table 2.1). These results indicate that posterior skeletal development is less sensitive to overall Lfng dosage than is anterior skeletal development, and that the level of Lfng in Lfng∆FCE/∆FCE animals is sufficient to sustain relatively normal posterior somitogenesis, although it is not high enough to support normal anterior development. This further supports the hypothesis that primary and secondary body formation require different levels of LFNG activity. However, posterior development is still critically dependent on a threshold level of Lfng expression, based on the increased disorganization and truncation of the tail region as Lfng levels decrease in Lfng∆FCE/- and

Lfng-/- animals.

2.3.2 Lfng dosage affects rostral/caudal somite patterning

Lfng has been proposed to play roles in both the segmentation clock and in rostral/caudal patterning of the developing somites (Cole et al. 2002, Morales et al.

2002). Perturbation of either of these functions might affect development of the posterior axial skeleton. To assess the dosage sensitivity of rostral/caudal somite patterning, we examined the expression of Uncx, a marker of the caudal somite compartment, in embryos with varying levels of Lfng expression in the PSM. Interestingly, we find that the severity of the tail phenotype (Figure 2.2 a-d) correlates with loss of somite patterning as assessed by expression of Uncx in the caudal somite compartment at 10.5 d.p.c., when somites that give rise to tail vertebrae are being produced (Figure 2.2 e-h). In

Lfng∆FCE/∆FCE embryos, Uncx expression is similar to wild type, although spacing is occasionally irregular. Expression is less compartmentalized in Lfng∆FCE/- embryos, and clear borders between the rostral and caudal compartment of the somites are no longer as strongly defined. In Lfng null embryos, less compartmentalization is observed, although in tail somites clear compartments are still visible. Because Lfng wild type, Lfng∆FCE/∆FCE, and Lfng∆FCE/- embryos clearly differ in their levels of Lfng in the anterior PSM, these results could suggest that somite patterning and posterior development may be sensitive to Lfng dosage in the anterior PSM, but do they not rule out a possible role for oscillatory

Lfng in the posterior PSM, as patterning in the anterior PSM might be affected by abnormal clock function in the posterior PSM.

2.3.3 Exogenous expression of Lfng in the anterior PSM does not rescue posterior skeletal development

If oscillatory Lfng expression is dispensable during secondary body formation, exogenous expression of Lfng in the anterior somite compartment should be sufficient to rescue caudal development. To directly test the importance of Lfng expression in the anterior PSM, we developed transgenic mice specifically expressing Lfng in the anterior

PSM, in the future rostral somite compartment, where endogenous Lfng expression is observed (Cole et al. 2002). This transgene uses a previously defined Mesp2 promoter element (Haraguchi et al. 2001) to drive Lfng expression in the anterior PSM. The transgene also contains an IRES/LacZ cassette, allowing us to assess both transcription and translation from the transgene (Mesp2>Lfng transgene, Figure 2.3A). Whole mount

RNA in situ analysis for LacZ RNA sequences and comparison to Mesp2 expression demonstrates that this transgene expresses RNA in a pattern that overlaps that of endogenous Mesp2, thus recapitulating the expression of Lfng in the anterior PSM

(Figure 2.3B, panels a,b). Whole mount X-gal staining of wild type embryos indicates that LacZ protein is found in a single band in the anterior PSM, with stable protein perduring in the anterior compartment of the first few epithelial somites (Figure 2.3B, panel c). Although we cannot directly assess LFNG protein expression from the transgene due to a lack of appropriate antibodies, the translation of LacZ protein from the

IRES cassette supports the idea that LFNG protein expression in LfngMesp2>Lfng mice recapitulates the endogenous LFNG protein expression in the anterior PSM. To further examine the expression levels driven by this transgene, we directly compared Lfng expression levels in transgenic and wild type embryos, to avoid confounding differences in probe affinity between the LacZ probe and the Lfng probe. We simultaneously

performed in situ analysis using an Lfng probe on Lfng∆FCE/∆FCE embryos and Lfng-/- embryos carrying a copy of the Mesp2>Lfng transgene (where the only source of Lfng

RNA is the transgene). Although the band of Lfng RNA is diffuse in the Lfng-/-,

Mesp2>Lfng embryos (due to patterning defects associated with the loss of Lfng activity), the levels are similar to or higher than those observed in Lfng∆FCE/∆FCE embryos (Figure

2.3C). Since the Lfng∆FCE allele clearly expresses sufficient Lfng to rescue anterior skeletal development, we conclude that the Mesp2>Lfng transgene is driving sufficient

Lfng expression to have a positive effect on Lfng-related phenotypes if expression of Lfng in the anterior PSM is sufficient for rescue.

Interestingly, we find that expression of exogenous Lfng in only the anterior PSM does not notably rescue the posterior skeletal phenotypes of Lfng mutant mice. The number of tail vertebrae (Figure 2.3D), the percent of abnormal tail vertebrae (Figure

2.3E), and the number of normal sacral vertebral condensations (data not shown) are all unchanged in embryos with or without the transgene. Breeding the transgene to homozygosity did not improve the phenotypes of Lfng mutant mice, suggesting Lfng expression in the anterior PSM is not sufficient for secondary body formation.

2.3.4 Lfng∆FCE/∆FCE mice express low levels of dynamic Lfng in the posterior PSM

Our transgene data suggests that the increasing severity of posterior skeletal defects as Lfng expression levels are reduced cannot be explained by a dosage-sensitive requirement for Lfng in the anterior PSM. One possible interpretations of this result is that low levels of cyclic Lfng in Lfng∆FCE/∆FCE and Lfng∆FCE/- embryos are actually acting to rescue tail development. Recent data suggest the possibility that low levels of cyclic

Lfng expression in the posterior PSM may be able to rescue tail development in Lfng-/- animals (Oginuma et al. 2010, Stauber et al. 2009). Although Lfng expression was not detectable by whole mount in situ analysis in Lfng∆FCE/∆FCE embryos at 10.5 d.p.c.

(Shifley et al. 2008), it is possible that low levels of Lfng, below the threshold of detection for in situ hybridization, are being expressed in these embryos.

To test this possibility, we performed RT-PCR to compare Lfng expression in

Lfng+/∆FCE embryos (which exhibit completely normal skeletal development) to that in

Lfng∆FCE/∆FCE embryos. Embryos were harvested at 10.5 d.p.c. and the posterior portion of the PSM was removed for RNA extraction. As each embryo is potentially in a different stage of the clock cycle, RT-PCR was performed on RNA from 7 individual embryos of each genotype. Using semi-quantitative PCR, we observe variable levels of Lfng in the 7 control embryos, although Gapdh levels are similar (Figure 2.4A). In the Lfng∆FCE/∆FCE embryos, Lfng expression is observed, although the levels detected suggest that the amount of Lfng expressed is quite low in these embryos. To control for the possibility that our samples contained part of the anterior PSM, we examined Mesp2 expression, which co-localizes with Lfng expression in the anterior PSM. Mesp2 expression was not detected in posterior PSM samples but was strongly expressed in the anterior PSM, confirming that the posterior PSM samples are not contaminated with RNA from the anterior PSM compartment. These analyses were repeated using quantitative RT-PCR, to directly compare Lfng levels across genotypes (Figure 2.4B). We find that Lfng levels oscillate in embryos of both Lfng∆FCE/∆FCE and Lfng+/∆FCE genotypes. The peak expression of Lfng in Lfng+/∆FCE embryos was found to be about 40% of the peak expression in wild

type embryos. This indicates that the absolute level of Lfng expression in the PSM is not maintained in heterozygous embryos, and that lower absolute levels of oscillatory RNA are sufficient to support normal development. In contrast, the peak levels of Lfng expression in Lfng∆FCE/∆FCE embryos was only 10% of the highest expression observed in wild type embryos. However, even in Lfng∆FCE/∆FCE embryos, dynamic expression was observed, with Lfng mRNA levels varying over 2 fold in embryo-to-embryo comparisons. Similar analyses in Lfng null embryos indicated that Lfng mRNA levels are essentially undetectable (<0.05% of peak wild type expression), supporting the idea that the low levels observed in Lfng∆FCE/∆FCE embryos do, in fact, represent real, dynamic Lfng expression.

We then examined the functional role of the low, oscillatory Lfng expression observed in Lfng∆FCE/∆FCE embryos. Previous analysis of NICD by whole mount immunohistochemistry did not reveal overt oscillatory activation of NOTCH1 in the PSM of Lfng∆FCE/∆FCE embryos (Shifley et al. 2008). However, it is possible that during secondary body formation, low levels of Lfng oscillations may be sufficient to promote sufficient NICD oscillations to support relatively normal tail development, even if NICD levels at the trough of the cycle are not as low as is observed in wild type embryos. To test this hypothesis, we used western blot analysis to examine NICD levels in the posterior PSM of Lfng+/∆FCE and Lfng∆FCE/∆FCE embryos. Comparing NICD levels in individual PSMs, we find that NICD levels in Lfng∆FCE/∆FCE embryos are, in fact, dynamic from embryo to embryo. However, we observe that the peak of NICD expression is higher in Lfng∆FCE/∆FCE embryos than in Lfng+/∆FCE embryos, while the

trough of NICD expression is lower in Lfng+/∆FCE embryos than in Lfng∆FCE/∆FCE embryos

(Figure 2.5). Thus, our results suggest that NICD levels in Lfng∆FCE/∆FCE embryos are dynamic, but are higher in general than observed in Lfng+/∆FCE embryos, which have normal anterior body formation. Taken together, these data indicate that the Lfng∆FCE allele does, in fact, express low levels of dynamic Lfng in the posterior PSM where the clock is active in wild type embryos. Although these levels are not high enough to entirely recapitulate the normal, cyclic activation of NOTCH1 in the wild type PSM, they are sufficient to support dynamic NICD activity that coordinates relatively normal skeletal development during secondary body formation. However, as the dosage of cyclic

Lfng is reduced in Lfng∆FCE/- and Lfng null animals, posterior skeletal development is increasingly compromised.

2.3.5 The rate of segmentation is altered in Lfng∆FCE mice

As Lfng levels are reduced, the axial skeleton becomes increasingly truncated, losing from 3 posterior skeletal elements in Lfng∆FCE/∆FCE animals to 15 posterior skeletal elements in Lfng-/- animals (Figure 2.1C, Table 2.1). We were interested in examining the mechanism behind this skeletal truncation in Lfng-/- animals. One possible explanation would be that in the presence of low Lfng, somitogenesis ends prematurely in Lfng∆FCE/- and Lfng-/- animals. To address this idea, we examined the length of the PSM in wild type and Lfng-/- animals (Figure 2.6A). Because the length of the PSM shortens as somitogenesis proceeds, we expected the PSM to be reduced in size in Lfng-/- embryos at earlier embryonic stages if segmentation were ending prematurely. However, no statistically significant difference was observed between wild type and Lfng-/- embryos at

early TS19 (Theiler stage) and late TS21 (Figure 2.6B), suggesting that the duration of somitogenesis is the same in wild type and Lfng-/- embryos.

In the chick, cessation of segmentation is accompanied by localized apoptosis in the tailbud (Tenin et al. 2010). Premature apoptosis in the tailbud of Lfng mutants would be expected to lead to loss of posterior vertebrae. To determine whether apoptosis occurs earlier in Lfng-/- embryos, we performed whole-mount immunohistochemistry using a cleaved caspase-3 antibody. We assessed apoptosis in embryos at Theiler stages 22 and

23 (Figure 2.6C). At TS 23, similar levels of apoptosis were detected in the tailbud of both Lfng+/+ and Lfng-/- embryos. At TS22, we did not observe apoptosis in either Lfng+/+ or Lfng-/- embryos, suggesting that somitogenesis is not ended early in the absence of

Lfng.

An alternative explanation for reduced posterior vertebrae in Lfng mutants is that the rate of segmentation is slowed as Lfng levels are reduced, resulting in the production of fewer vertebrae. If this is the case, Lfng∆FCE/∆FCE mutants should have fewer vertebrae in the anterior skeleton. We quantified vertebral elements in Lfng∆FCE/∆FCE mutants from the cervical region to the lumbosacral junction by counting the dorsal arches of the vertebrae, which are comparatively mildly affected in most Lfng∆FCE/∆FCE animals (Figure

2.7A and 2.7B). While wild type mice (n=7) formed 26 vertebrae, Lfng∆FCE/∆FCE mice

(n=7) averaged 22 vertebrae (p<.001). We next determined whether reduction of the number of anterior vertebrae corresponded to a reduced number of somites formed during primary body formation. We counted the somites of wild type and Lfng∆FCE/∆FCE mutants at 10.5 d.p.c. as assessed by expression of Uncx (Figure 2.7C and 2.7D). Because Uncx

expression in the trunk region of Lfng∆FCE/∆FCE embryos was too disorganized to count somites reliably, we counted somites from the cervical region to the posterior boundary of the forelimb bud, as was previously done in Hes7 mutant embryos to assess segmentation clock period (Harima et al. 2013). We observed an average of 8.9 somites

(n=10 embryo halves) in wild type embryos. In contrast, Lfng∆FCE/∆FCE embryos averaged

8.0 somites (n=10 embryo halves), suggesting the clock period is longer in these mutants

(p=.001). Together, these results support the hypothesis that the rate of segmentation is slowed in Lfng mutants, leading to loss of vertebrae.

2.4 Discussion

2.4.1 The segmentation clock period is sensitive to Lfng dosage

Our data supports a recent report suggesting the clock period is sensitive to Notch activity levels. In embryos mutant for the Notch regulator Nrarp, Notch activity levels in the PSM are slightly elevated, resulting in a longer clock period and fewer vertebrae

(Kim et al. 2011). Consistent with this model, Lfng∆FCE/∆FCE and Lfng-/- animals have elevated levels of Notch activation and fewer vertebrae. The reduced number of cervical and thoracic vertebrae in Lfng-/- mutants has been suggested to be a result of homeotic transformation (Cordes et al. 2004). However, homeotic transformation alone cannot explain the loss of vertebrae in Lfng∆FCE/∆FCE and Lfng-/- animals, as the total number of somites and vertebrae formed during somitogenesis is reduced in these mutants.

However, our data is not inconsistent with a model where Hox gene expression acts on fewer segments to alter the normal pattern of vertebral identity.

2.4.2 Oscillatory Lfng expression is required during all stages of axial development

Previous data have suggested that there are differential requirements for Lfng expression and Notch pathway activity during primary and secondary body formation

(Shifley et al. 2008, Stauber et al. 2009). At least two possible models could explain the existing data. One possibility is that oscillatory Lfng is dispensable for clock function during secondary body formation, but plays a critical role in rostral/caudal somite pattering during this stage of development. Alternatively, as suggested by Oginuma et al.

(2010), it is possible that the most critical function of Lfng is its oscillatory expression linked to the segmentation clock, while its stable expression during somite patterning may be less important.

The data reported here, along with data from other studies, supports the second of these two models. We observe low levels of dynamic Lfng expression in the posterior

PSM of Lfng∆FCE/∆FCE embryos. Further, we find that increased transgenic expression of

Lfng in the anterior PSM somite patterning region is not sufficient to rescue posterior skeletal development in Lfng mutant mice. Together with data indicating that Lfng expression driven by the oscillatory Hes7 promoter can completely rescue the Lfng null phenotype (Oginuma et al. 2010), and the observation that Lfng transgenic mice that express very low levels of oscillatory expression in the posterior PSM have a phenotype very similar to that of Lfng∆FCE/∆FCE mice (Stauber et al. 2009), these data support the idea that low levels of oscillatory Lfng in the clock are sufficient to support relatively normal posterior skeleton development during secondary body formation.

These data also provide further support for the idea that primary and secondary body formation have different requirements for oscillatory Notch activity in the

segmentation clock. We observe phenotypes in the posterior skeleton that are clearly sensitive to the overall dose of Lfng, with transheterozygous Lfng∆FCE/- mice exhibiting a phenotype that is intermediate between that of homozygous Lfng∆FCE/∆FCE and Lfng-/- animals. The contrast between this observation and the finding that phenotypes in the anterior skeleton are only mildly exacerbated as Lfng dosage decreases, provide strong support for the hypothesis that somitogenesis has a more stringent requirement for clock activity during primary body formation than during secondary body formation. Our observations suggest the possibility that in Lfng∆FCE/∆FCE embryos, Notch activity levels are, in fact, dynamic, but both our previous data and that shown here indicate that the amplitude of those oscillations may be diminished and that the levels of NICD in mutant embryos does not drop as low as is observed in wild type embryos (Shifley et al. 2008)

(Figure 2.5). The fact that Lfng∆FCE/∆FCE mice form essentially normal tails despite the fact that NICD levels never reach the minimum levels seen in wild type embryos indicates that secondary body development can be coordinated by NICD oscillations of smaller amplitude and with less dramatic troughs. In contrast, NICD oscillations must be more robust to coordinate normal development of the anterior skeleton.

In contrast to the tail, we observe that the sacral region is largely resistant to loss of Lfng activity. Interestingly, Lfng-/- skeletons are also less affected in the cervical region. During primary body formation, cells in the PSM may be able to maintain phase- linked oscillations for a few cycles, but may require LFNG activity to maintain these oscillations in the long term. Because a different cell population gives rise to the posterior vertebrae, the formation of sacral vertebrae during secondary body formation may be

analogous to the production of cervical vertebrae during primary body formation. PSM progenitors in the tailbud may initially be synchronized independently of LFNG activity, but as Lfng levels are reduced, cells may be unable to maintain synchronized oscillations for as many cycles. This is consistent with our observation that in both Lfng∆FCE/- and

Lfng-/- animals the normal tail vertebrae that are observed are generally found anterior to malformed vertebrae in the more posterior tail.

It is not clear why primary and secondary body formation may have different requirements for Notch oscillations. It has been proposed that one of the functions of the

Notch pathway in the clock is to maintain synchronized gene expression within cohorts of cells as they progress through the PSM (Delaune et al. 2012, Horikawa et al. 2006,

Jiang et al. 2000, Mara et al. 2007, Ozbudak and Lewis 2008, Riedel-Kruse et al. 2007).

It is possible that during primary body formation, as migratory cells enter the PSM via the primitive streak, mobile cells may be more likely to move at different speeds, and frequently change neighbors. Indeed, FGF signaling promotes cell motility, and FGF ligands are downregulated during the last two days of somitogenesis in mouse and chick

(Olivera-Martinez et al. 2012, Stauber et al. 2009). It is interesting to speculate that during primary body formation, robust oscillations in Notch activity may be required to maintain synchrony among cells with different mobility. During secondary body formation, when many cells are proposed to arise from the static tailbud, cells in the PSM may exhibit less variability in relative mobility within the PSM, and lower amplitude oscillations of the clock may be sufficient to maintain synchronous gene expression.

2.5 Figures

Figure 2.1 Posterior skeletal development is sensitive to Lfng dosage. A. Alcian Blue/ Alizarin Red skeletal preparations of 17.5 d.p.c. embryos. Dorsal ribs (a-d), lateral view of lumbar vertebrae (e-h) ventral ribs (1-l) and dorsal views of the lumbar and sacral (m- p) regions are shown. Numerous rib fusions and bifurcations are seen in Lfng∆FCE/∆FCE (b,f), Lfng∆FCE/- (c,g) and Lfng-/- (d,h), as well as uneven rib attachment points and neural arch fusions. Though neural arch morphology generally becomes more disorganized as Lfng levels are reduced (e-h), there was significant mouse to mouse variability. In the sacral region, more normal condensations are observed in Lfng∆FCE/∆FCE (j) and Lfng∆FCE/- (k) skeletons, than are seen in Lfng-/- skeletons. B. Quantification of the total number of rib abnormalities is shown as a box and whisker plot. The number of rib abnormalities is significantly different in the Lfng mutant skeletons than wild type, but no significant differences are observed among Lfng∆FCE/∆FCE, Lfng∆FCE/- and Lfng-/- skeletons. C. Quantification of tail vertebrae is shown as a box and whisker plot. A statistically significant loss of vertebral elements is observed as Lfng dosages are decreased (see Table 1).

Figure 2.1

Figure 2.2 Rostral/caudal patterning of epithelial somites is sensitive to the dosage of Lfng in the PSM, and correlates with tail phenotypes. A. Morphological abnormalities of the tail in wild type, Lfng∆FCE/∆FCE, Lfng∆FCE/- and Lfng-/- animals exhibit increasing severity as Lfng dosage decreases. B. Uncx expression at 10.5 d.p.c. demonstrates increasing disorder in rostral/caudal patterning as Lfng dosage decreases. Clear compartment are observed in both wildtype and Lfng∆FCE/∆FCE mutants, which have only occasional defects in tail vertebrae. As tail defects become more severe in Lfng∆FCE/- and Lfng-/- mice, less compartmentalization is observed in tail somites.

Figure 2.3 Exogenous expression of Lfng in the anterior PSM does not rescue posterior skeletal development in Lfng mutant mice. A. Schematic of the LfngMesp2>Lfng transgene using an Mesp2 promoter element to drive an LfngIRESLacZ cassette. B. RNA in situ analysis for the LacZ sequences of the transgene transcript demonstrates that the transgene drives expression as a single band in the anterior PSM (a) that overlaps with endogenous Mesp2 expression (b, LacZ expression in purple, Mesp2 expression in orange). X-gal staining for LacZ protein shows expression from the LfngMesp2>Lfng transgene in the anterior compartment of somite -1, with protein perduring in the anterior compartment of the most recently formed somites (c). This indicates that the transcription and translation of the LfngMesp2>Lfng transgene recapitulates the endogenous expression of Lfng in the anterior PSM. C. Lfng RNA expression was compared in Lfng∆FCE/∆FCE embryos (a) and Lfng-/- embryos carrying the Mesp2>Lfng transgene (b, in which the only source of Lfng RNA is the transgene). Levels of Lfng RNA in the Lfng-/- embryos with the Mesp2>Lfng transgene are equal or higher than those observed in the Lfng∆FCE/∆FCE embryos, though the band of expression of broad and diffuse due to the loss of Lfng expression in the clock. D. Quantification of the number of tail vertebrae, and the percent of abnormal tail vertebrae demonstrates that the Lfng expressed in the anterior PSM from the LfngMesp2>Lfng transgene is not able to rescue posterior skeletal development in Lfng mutant embryos.

Figure 2.3

Figure 2.4 RT-PCR analysis detects low levels of dynamic Lfng expression in Lfng∆FCE/∆FCE embryos. A. RNA was isolated from the posterior portion of the PSM of 7 individual 10.5 d.p.c. Lfng∆FCE/∆FCE (right panel) and 7 Lfng+/∆FCE embryos (left panel). RNA from the anterior PSM of one wild type embryo was included as a control in the right-most lane of both panels (labeled +/+). Semi-quantitative RT-PCR was performed for Lfng (top), Mesp2 (middle) and Gapdh (bottom). 30 cycles were used to maintain reactions in the exponential stage of amplification. Although Gapdh levels are similar in all embryos, variable levels of Lfng are detected, reflecting the fact that each embryo is in a different stage of the clock. Lfng expression in Lfng∆FCE/∆FCE embryos is detectable, but much lower than that observed in wild type embryos. Mesp2 is detected in the anterior PSM sample but not posterior PSM samples, indicating that posterior samples are not contaminated with RNA from the anterior Lfng stripe. B. Quantitative PCR using Taqman assays was performed on cDNAs isolated from 7 individual 10.5 d.p.c. embryos of indicated genotypes. All dissections were performed as above, however these represent independent samples from those used in the analysis in part A. Lfng RNA levels were normalized to Gapd levels, and the highest wildtype value was set to 100. Within each genotype, values were arranged to demonstrate the cyclic nature of the RNA levels. Similar analysis of a Lfng null embryo gave Lfng RNA levels of less that 0.05 on this scale (data not shown) indicating that the levels observed in Lfng∆FCE/∆FCE embryos represent real, though low levels of Lfng expression.

Figure 2.5 NICD levels are dynamic in the posterior PSM of Lfng∆FCE/∆FCE embryos. Western blot analysis of seven individual 10.5 d.p.c. Lfng+/∆FCE embryos (left) and Lfng∆FCE/∆FCE embryos (right) are shown. NICD levels are shown at the top, and tubulin at the bottom. For inter-blot comparisons the same amount of a standard containing 4 wild type PSMs was loaded onto each gel. The lines in each gel represent individual samples that were not quantified due to gel artifacts and which have been removed from the image. Numbers below the images indicate values normalized to the loading control (above) and normalized values scaled to the highest individual value for each genotype to highlight the dynamic nature of the expression. While NICD levels in Lfng∆FCE/∆FCE embryos are higher than those observed in Lfng+/∆FCE embryos, we still detect dynamic expression.

Figure 2.6 Somitogenesis does not terminate prematurely in Lfng-/- embryos. A. The PSMs of wild type and Lfng-/- embryos were measured from the tip of the tail to the anterior expression domain of Mesp2. B. Quantification of PSM length is shown as a box and whisker plot. No significant difference was found between wild type and Lfng-/- embryos at TS19 or late TS21. C. At TS23, both wild type and Lfng-/- embryos undergo localized apoptosis in the tailbud. Apoptosis does not occur at TS22 in either wild type or Lfng-/-- embryos, suggesting segmentation does not terminate early in Lfng-/- embryos.

Figure 2.7 Lfng∆FCE/∆FCE mutants produce fewer somites and vertebrae. A, B. Quantification of vertebral number in the anterior skeleton at 17.5 dpc. Lfng∆FCE/∆FCE mutants average 22 vertebrae, whereas wild type animals have 26 vertebrae (n=7 for both genotypes). Results are plotted as mean with SD. ***p<.001, Mann-Whitney test. C, D. Quantification of somite number between the cervical region and posterior forelimb bud. Uncx marks the caudal compartment of each somite, and yellow asterisks indicate individual somites. Lfng∆FCE/∆FCE embryos average 8 somites (n=10 embryo halves) whereas wild type animals have 8.9 somites (n=10 embryo halves). Results are plotted as mean with SD. ***p<.001, Mann-Whitney test.

% uneven normal Ribs on Ribs on # rib # tail abnormal ribs at sacral right left abnormalities vertebrae tail sternum vertebrae vertebrae 7 0 +/- 0.0 29.8 +/- 0.0 7 0 +/- 0.0 Wild type 0 +/- 0.0 †(<0.01) 4 +/- 0 +/- 0.9 †(<0.05) +/- 0.0 §(<0.001) (n=10) §(<0.001) ^(<0.001) §(<0.01) §(<0.05) ^(<0.05) ^(<0.001) ^(<0.001) ^(<0.001) ^ (<0.01) 6.1 6.4 1.0 27.1 LfngΔFCE/ΔFCE 7.7 +/- 2.4 3.9 +/- 0.3 6.9 +/- 4.1 +/- 0.6 +/- 0.8 +/- 1.3 +/- 1.7 (n=10) *(<0.01) ^(<0.001) ^(<0.01) *(<0.05) ^(<0.01) 6.1 6.7 +/- 0.8 23.2 67.0 LfngΔFCE/- 8.9 +/- 1.8 3.9 +/- 0.3 +/- 0.6 0.5 +/- 1.0 +/- 1.6 +/- 9.9 (n=9) *(<0.001) ^(<0.001) *(<0.05) *(<0.01) *(<0.001) 2.1 +/- 1.2 15.1 87.5 5.7 6.3 1.0 Lfng-/- 9.2 +/- 1.8 *(<0.001) +/- 1.6 +/- 9.9 +/- 1.0 +/- 0.5 +/- 1.1 ( n=9) *(<0.001) †(<0.001) *(<0.001) *(<0.001) *(<0.01) *(<0.05) §(<0.001) †(<0.01) †(<0.01)

Table 1 Quantification and analysis of skeletal defects in mice with varying levels of Lfng expression in the PSM. 17.5 d.p.c. embryos were harvested from intercrosses that could produce all four genotypes. Skeletal preparations were analyzed for several axial skeletal phenotypes as described. Numbers represent average value +/- standard deviation. Statistical analyses was performed using the Kruskal-Wallis test with Dunn post hoc. P values <0.05 were considered significant. * = significantly different from wild type † = significantly different from Lfng∆FCE/∆FCE § = significantly different from Lfng∆FCE/- ^ = significantly different from Lfng -/-. Numbers in parentheses indicate P value.

CHAPTER 3

SECRETION OF LUNATIC FRINGE IS ESSENTIAL FOR SOMITOGENESIS AND SEGMENTATION CLOCK FUNCTION

3.1 Introduction

As discussed in chapter one, several genes have been linked to the segmentation clock based on their expression patterns in the PSM. These genes exhibit oscillatory expression, with a cyclic period that matches the rate of somite formation (Reviewed in

Kageyama et al. 2012). Many of these genes are members or targets of the Notch signaling pathway, and in mice, some targets of the FGF, and Wnt pathways also exhibit cyclic expression. Critically, not only transcript levels, but Notch activity levels oscillate as part of the clock (Morimoto et al. 2005, Shifley et al. 2008), and in both mice and chicks, this oscillatory activation requires periodic modulation of Notch signaling by the glycosyltransferase LFNG.

Lfng functions as part of a delayed negative feedback loop with the transcriptional repressor Hes7 to regulate cyclic Notch activation in the clock (Bessho et al. 2003,

Kageyama et al. 2012). In the mouse and chick, both cyclic Lfng transcription and oscillatory LFNG protein have been observed, linking cyclic LFNG activity to the clock

(Cole et al. 2002, Dale et al. 2003, Morales et al. 2002). Mutants that lack all LFNG

activity have severe skeletal abnormalities and tail truncations (Evrard et al. 1998, Zhang and Gridley 1998), and in these mutants, NOTCH1 activation is ubiquitous throughout the PSM, demonstrating the importance of tight regulation of oscillatory Notch activity during somitogenesis.

In order for cyclic Lfng transcription to result in oscillatory LFNG protein levels, there must be post-translational mechanisms that confer a short LFNG protein half-life so that LFNG activity is eradicated during the short “off” phase of the clock. The importance of this control is highlighted by findings that embryos with sustained, non- oscillatory Lfng expression have severe skeletal abnormalities, demonstrating the importance of dynamic Lfng expression in the PSM (Serth et al. 2003). However, the mechanisms that regulate LFNG protein activity in the segmentation clock are completely unknown.

Lfng encodes a glycosyltransferase that modifies Notch in the Golgi (Moloney et al. 2000, Panin et al. 2002). This glycosylation modulates ligand-receptor interactions, thus modulating pathway activity (Panin et al. 1997). The known functions of LFNG in the Notch pathway are cell-autonomous (Hicks et al. 2000, Irvine 1999, Moloney et al.

2000, Munro and Freeman 2000, Panin et al. 1997), however, both Drosophila fringe and

Lunatic fringe are cleaved by furin-like proteases and secreted into the extracellular space, where they are presumably inactive (Johnston et al. 1997). Mutations that accelerate Fringe secretion in Drosophila are hypomorphic, supporting the hypothesis that LFNG protein does not function in the extracellular space (Munro and Freeman

2000). We hypothesize that cleavage of LFNG by pro-protein convertases and its

subsequent release into the extracellular space is a mechanism for terminating LFNG activity, promoting the rapid protein turnover required for its function in the segmentation clock. This would represent a novel post-transcriptional mechanism for the spatial and temporal regulation of Notch signaling, allowing rapid and reversible modulation of Notch activity.

Previous data from our lab has shown that mutations that prevent LFNG secretion in vitro prolong its active half-life without affecting its subcellular localization or enzymatic activity in the Notch pathway (Shifley and Cole 2008). Here, we directly assess the in vivo importance of LFNG secretion by generating mice expressing a Golgi- resident LFNG variant that is not secreted (tethered LFNG or tLFNG). We find that segmentation defects in these mutants are more severe than in animals lacking all LFNG activity, demonstrating the importance of rapid protein turnover in the clock. In the presomitic mesoderm, Notch activation is no longer dynamic, and oscillations of Notch targets are abolished. Moreover, expression of clock genes outside of the Notch pathway is perturbed. We conclude that secretion of LFNG is a critical aspect of its regulation and is essential for segmentation clock function.

3.2 Methods

3.2.1 Targeted mutation and genotyping

Lfng coding sequence from the Not1-AatII sites in exon1 were replaced with the

RFNG signal sequence (amplified with primers (5'atgcggccggcggccaccatgagccgtgcgcggc gg-3’) and (5'-ggttcttccgagtggtcttg). This replaces the LFNG pre/pro region with the signal sequence and 21 amino acids of RFNG with a fusion at the first conserved amino

acid (LFNG D113). The 5’ arm contains a floxed Neo-testis Cre cassette in intron1, which is excised upon passage through the male germline (Bunting et al. 1999).

Endogenous regulatory elements and splice sites are maintained. LfngtLFNG mice were genotyped with primers SC286 (5’-ttgggtctatctgggaaacg-3’) and SC287 (5’- gcgactcatccagacacaga-3’) producing a 149 bp wild type band and a 250 bp mutant band.

LfngtmRjo1 (Lfng null) mice were genotyped as described (Shifley et al. 2008). Mice were maintained in an SPF facility under the care of the Ohio State University Laboratory

Animal Resources department. All procedures were conducted under protocols approved by the Ohio State University IACUC.

3.2.2 Skeletal preparations and histology

Embryos were harvested at 17.5 days post coitus (d.p.c.). Alcian Blue and Alizarin Red staining was performed as described (McLeod 1980), except embryos were incubated for four days in 95% ethanol and three days in acetone. For histology, embryos were fixed in

Bouin’s fixative and processed for paraffin sections. Sections were stained with hemotoxylin and eosin.

3.2.3 Whole-mount RNA in situ and histochemistry analysis

Embryos were collected from timed pregnancies (noon of the day of plug detection = 0.5 d.p.c.) and fixed overnight in 4% PFA. Hybridization using digoxigenin- labeled probes was performed essentially as described (Shifley et al. 2008) using probes for Lfng (Johnston et al. 1997), Mesp2 (Saga et al. 1997), Uncx (Mansouri et al. 1997),

Tbx18 (Kraus et al. 2001), Spry2 (Wahl et al. 2007), Axin2 (Huang and Gorman 1990),

Hes7, intronic Hes7, and Nrarp (Shifley et al. 2008). For cyclic genes, the number of

embryos in each phase is indicated in the figure legends.

Whole-mount immunohistochemistry was performed essentially as described

(Shifley et al. 2008), but NICD was detected with a rabbit monoclonal antibody (Cell

Signaling Technology 4147 1:500).

3.3 Results

3.3.1 Prevention of LFNG secretion perturbs segmentation

A chimeric LFNG construct that tethers the LFNG protein in the Golgi has previously been described. This chimeric protein is not secreted in tissue culture, but maintains the enzymatic specifity of endogenous LFNG (Shifley and Cole 2008). To express Golgi-tethered LFNG protein (tLFNG) in vivo, we generated a novel allele of

Lfng, replacing exon 1 of the endogenous locus with sequences encoding the signal sequence of Radical fringe, which contains a type II transmembrane domain, fused to

Lunatic fringe at the first conserved amino acid of the two proteins (Figure 3.1). The resulting allele, LfngtLFNG, encodes a protein with the expression pattern and enzymatic activity of LFNG, but which is not secreted. Further, this allele does not alter the endogenous transcription start site, 5' or 3'UTR, or splice sites of Lfng. If LFNG secretion provides a mechanism to inactivate LFNG activity in the clock, we predict that this mutation would result in dominant phenotypes, as preventing LFNG secretion will inhibit its inactivation, resulting in a hypermorphic protein.

As predicted, F1 heterozygotes (LfngtLFNG/+) exhibit overt skeletal abnormalities, with a shortened body axis and truncated tails, similar to Lfng null animals (Fig. 3.2A and

B). LfngtLFNG/+ mice are viable and fertile, although male mice frequently develop penile

prolapse and testicular cysts. LfngtLFNG/+ skeletons display disorganization at all levels of the axial skeleton (Figure 3.2B, panels g-i). In the anterior skeleton, LfngtLFNG/+ animals are more severely affected than Lfng nulls, with more severe fusions of the proximal ribs and neural arches. In LfngtLFNG/+ embryos, the sacral region is severely disorganized, with multiple neural arch fusions and disorganized and ectopic ossification centers. This is in contrast to Lfng null embryos, which form 1-4 sacral condensations with relatively normal neural arch morphology (Shifley et al. 2008, Stauber et al. 2009, Williams et al.

2014). In the tail region, Lfng-/- nulls and LfngtLFNG/+ mutants are similarly affected, with severely malformed vertebrae and tail truncations. These results indicate that preventing

LFNG secretion creates a dominant, gain-of-function allele that perturbs skeletal development more severely than the complete loss of LFNG function.

3.3.2 LfngtLFNG/+ mutants do not form epithelial somites

The severe axial skeleton fusions observed in LfngtLFNG/+ embryos suggested defects in boundary formation and somite patterning. We find that somitogenesis is severely disrupted in LfngtLFNG/+ embryos. At 10.5 d.p.c., borders between somites are absent or incomplete, and no distinct epithelial somites are observed at this stage (Figure

3.3A). Despite the lack of epithelial somites, mutant embryos produce somitic derivatives including myotome structures in the trunk, although these are often fused and disorganized (Figure 3.3A). Thus, LFNG secretion and protein turnover are critical for proper border formation and production of epithelial somites during somitogenesis.

3.3.3 Somitic structures are caudalized in LfngtLFNG/+ mutants

In addition to its cyclic expression in the posterior PSM, Lfng is expressed in the rostral compartment of presomites in the anterior PSM (Cole et al. 2002). It is not yet clear whether this expression pattern is important for normal somite patterning. The skeletal phenotypes of LfngtLFNG/+ mutants suggest that they exhibit somite patterning defects that are distinct from those observed after loss of Lfng activity. This could indicate that inappropriate perdurance of the LFNG protein in the caudal compartment of the presomites and/or in the mature somite region of LfngtLFNG/+ embryos might have deleterious effects on somite patterning by dysregulating Notch signaling, which has roles in somite patterning (Feller et al. 2008).

We first examined expression of Mesp2, a marker of the presumptive rostral somite compartment (Saga et al. 1997). Expression of Mesp2 in LfngtLFNG/+ mutants was similar to that seen in wild type, though slightly reduced and diffuse, suggesting rostral identity is initially specified in the presomites (Figure 3.3B, d). However, Tbx18, which marks the rostral somite compartment, is severely downregulated in mutant embryos compared to wild type, suggesting that somitic structures are caudalized (Figure 3.3B, e).

Consistent with this finding, the expression domain of a caudal somite marker, Uncx, is expanded in mutant embryos (Figure 3.3B, f). This is in contrast to the phenotypes of

Lfng null embryos, which exhibit a "salt and pepper" pattern of intermingled rostral and caudal cells (Oginuma et al. 2010). The dorsal rib fusion phenotype observed in

LfngtLFNG/+ animals is consistent with these observations, as they are similar to phenotypes observed in Tbx18 mutant mice and mice carrying a hypomorphic Mesp2 allele (Bussen et al. 2004, Nomura-Kitabayashi et al. 2002), reflecting the fact that dorsal

rib tissue arises from the caudal somite. These results suggest that cleavage and rapid secretion of LFNG from the future rostral somite compartment is important for somite patterning, and that interfering with this process caudalizes the somites.

3.3.4 Endogenous Lfng transcription is reduced in LfngtLFNG/+ embryos

Since LFNG functions in the clock by repressing Notch activity, changes in

LFNG protein turnover are anticipated to feed back on its own transcription. Lfng expression in the PSM can be divided into two functional domains, with oscillatory expression in the posterior PSM where the clock is active, and stabilized expression in the anterior PSM. Distinct cis-acting elements regulate Lfng expression in these two domains (Cole et al. 2002, Morales et al. 2002). Thus, we predicted that we might see distinct effects on expression in the anterior and posterior PSMs of mutant embryos. Wild type embryos exhibit three distinct phases of Lfng expression linked to the clock (Figure

3.4, panels a-d). In contrast, in all mutant embryos examined, Lfng mRNA expression is limited to a weak stripe in the anterior PSM after equivalent detection time. To investigate whether low levels of Lfng RNA were still being transcribed in mutant embryos, we performed long revelation steps on 8.5 d.p.c. embryos, where background staining issues are less noticeable. Again, cyclic expression of Lfng was clearly visible in wild type embryos after 6 hours, though at this time only a faint band in the anterior PSM was observed in mutant embryos (Figure 3.4, panels e-g). However, upon extended staining (48 hours), Lfng expression was observed throughout the PSM of LfngtLFNG/+ embryos but was excluded from the mature somite region (Figure 3.4A, j). This expression does not appear oscillatory, though the low levels make it difficult to exclude

the possibility of dynamic transcription. However, dynamic expression in the caudal PSM of wild type embryos was still evident, even after this prolonged staining, thus the lack of overt cycling in LfngtLFNG/+ embryos is unlikely to be an artifact of long detection times

(Figure 3.4, panels h-i). These data support the idea that increasing the stability of the

LFNG protein affects the transcription of the endogenous Lfng locus, presumably through the regulatory loops that control cyclic Notch activation in the segmentation clock. These results are consistent with a model where low levels of Lfng RNA produce a stable LFNG protein that perturbs cyclic activity of the Notch pathway in the posterior PSM.

3.3.5 LFNG secretion is required for cyclic activation of Notch pathway genes

In the segmentation clock, LFNG functions as part of a delayed negative feedback loop with HES7 to modulate oscillatory NOTCH1 activity in the PSM (Bessho et al.

2003), and LFNG has been hypothesized to repress Notch activation in this context (Dale et al. 2003, Morimoto et al. 2005, Okubo et al. 2012). To examine the effect of the tLFNG allele on Notch signaling, we examined NOTCH1 activation using whole-mount immunohistochemistry with an antibody specific to the cleaved, activated form of

NOTCH1 (NICD). In wild type embryos, distinct phases of Notch activity are observed as previously described (Morimoto et al. 2005, Shifley et al. 2008). In contrast, in

LfngtLFNG/+ mutants, NICD levels are reduced and visible only in a faint stripe in the anterior PSM after identical detection time (Figure 3.5, panels a-d). However, after long exposures of 8.5 d.p.c. embryos we observe low levels of NICD in the most posterior

PSM, which could not be observed at later stages due to background staining issues

(Figure 3.5, panels e-g). These data support the hypothesis that LFNG represses

NOTCH1 activity in the PSM, and that when the LFNG protein half-life is extended, the levels of Notch activation are severely reduced in the posterior PSM. To confirm these findings we examined expression of Nrarp, an oscillatory target of the Notch pathway that is not directly involved in the negative feedback loop with Lfng and Hes7 (Dequeant et al. 2006, Wright et al. 2009). As expected, Nrarp levels were severely reduced in all

LfngtLFNG/+ embryos examined, with expression restricted to a faint anterior stripe, similar to NICD. Thus, the reduced Notch activation in the PSM of mutant embryos has functional consequences on the activation of Notch targets (Figure 3.5, panels h-k).

We additionally examined Hes7 expression in mutant embryos. In wild type embryos, as expected, we observed that Hes7 levels oscillate in the PSM. In contrast, we observe ubiquitous expression of Hes7 mRNA in the posterior PSM of mutant embryos, although the level was slightly reduced compared to wild type Hes7 at the highest stage of its oscillation (Figure 3.6, panels a-d). To distinguish between transcriptional and post- transcriptional effects on Hes7 expression we repeated these analyses with a probe specific to the Hes7 intron, allowing us to specifically examine the levels of newly transcribed Hes7 RNA. Interestingly, while clear oscillatory expression of Hes7 was observed in wild type embryos, levels of newly transcribed Hes7 RNA in mutant embryos were significantly reduced and were restricted to the posterior PSM in all mutant embryos examined (Figure 3.6, panels e-h). This suggested that the observed expression pattern in mutant embryos is due to post-transcriptional effects on Hes7 RNA stability or turnover, as has been observed in the Hes homologues of Xenopus and

Zebrafish (Davis et al. 2001, Dill and Amacher 2005). Further, this pattern of

transcription is consistent with data suggesting that Notch is not required for the initiation of Hes7 transcription in the posterior PSM, but is required for propagation of cyclic waves through the anterior PSM (Niwa et al. 2007). More interestingly, these data suggest the existence of a post-transcriptional level of regulation of Hes7 that affects its rapid RNA turnover in the PSM. They further suggest that altering the half-life of the

LFNG protein can indirectly affect Hes7 RNA turnover, perhaps indicating that the post- transcriptional mechanisms that regulate Hes7 mRNA stability are linked to the segmentation clock or to the Notch signaling pathway.

3.3.6 FGF and WNT signaling are also affected in LfngtLFNG/+ embryos

In mouse embryos, several targets of the FGF and WNT pathways have been observed to oscillate in the PSM (Dequeant et al. 2006). Although it is not clear whether the oscillations of these targets are required for normal segmentation, it is clear that both

FGF and WNT signaling are important for normal clock function (Aulehla and Pourquie

2010). Little is known about cross-talk among the Notch, FGF, and Wnt pathways in the segmentation clock. Some findings suggest that the FGF pathway acts upstream of the

Wnt pathway, which acts upstream up Notch. For instance, pharmacological disruption of the FGF pathway rapidly disrupts Axin2 oscillations, followed by disruption of Lfng oscillations in the next cycle (Wahl et al. 2007). However, Psen1/Psen2 double knockouts, which are predicted to lack all canonical Notch activity due to loss of NICD formation, do not form somites and lack all oscillatory gene expression in the PSM, which may suggest more complicated crosstalk between the Notch, Wnt, and FGF pathways (Ferjentsik et al. 2009, Huppert et al. 2005).

To determine whether increased LFNG activity perturbs FGF signaling, we performed whole-mount immunohistochemistry with an antibody against phosphorylated

MAPK (Figure 4, panels a-c). In wild type embryos, we observed a gradient of MAPK activity in the posterior PSM. In LfngtLFNG/+ embryos, phosphorylated MAPK was absent or reduced, suggesting FGF signaling is impaired in LfngtLFNG/+ embryos. We also examined expression of the FGF regulator Spry2. In wild type embryos, we observed oscillatory expression of Spry2, however, in all LfngtLFNG/+ embryos analyzed, Spry2 expression was observed throughout the PSM, though at lower levels than the peak of expression in wild type embryos (Figure 4, panels d-f).

To determine whether increased LFNG activity also perturbs Wnt signaling, we examined expression of Axin2. Although Axin2 oscillations were evident in wild type embryos at 9.5 d.p.c., Axin2 expression in LfngtLFNG/+ embryos was limited to an anterior stripe and faint staining in the posterior tailbud (Figure 4, panels g-i). A similar pattern was observed at 10.5 d.p.c. after prolonged detection, although staining in the posterior

PSM of LfngtLFNG/+ embryos was more evident at this stage of development (Figure 4, panels j-l). Thus, the disruption of Notch signaling in LfngtLFNG/+ embryos appears to inhibit FGF signaling and prevents cyclic expression of both Wnt and Fgf pathway members in the PSM. These results provide further support of crosstalk between the

Notch, FGF, and Wnt pathways, and indicate that post-translational control of LFNG activity is essential to this regulation.

3.4 Discussion

Our results indicate that secretion of LFNG from the cell provides an essential level of post-translational control of the clock by promoting the rapid turnover of LFNG required for coordinated, oscillatory Notch activity in the PSM (Figure 3.8). The cleavage and release of the enzymatically active region of LFNG would be a novel mechanism to provide rapid temporal regulation of a critical Notch modulator. Additionally, we find that the function of other signaling pathways in the clock can be disrupted by preventing

LFNG secretion, presumably due to altered patterns of Notch activation, as has been suggested by other studies (Feller et al. 2008, Ferjentsik et al. 2009).

What role, if any, the secreted form of LFNG plays during somitogenesis is not clear. The phenotypes reported here are unlikely to be due to a lack of LFNG secretion, as LfngtLFNG/+ mice carried a single wild type copy of Lfng that produces the secreted form. Further, the increased severity of skeletal phenotypes in LfngtLFNG/+ embryos compared to Lfng null embryos, as well as the distinct effects on NICD levels do not support the idea that the phenotypes observed in mutant mice are due to a loss of LFNG activity. Mathematical models suggesting secreted LFNG plays a role in synchronizing the segmentation clock have been proposed (Cinquin 2003). However, the results described here and by others would tend to argue against such a model. Instead, they may support the recent hypothesis that LFNG synchronizes the clock by modulating Dll1 function within the LFNG-expressing cell, suggesting that secreted LFNG cannot synchronize the segmentation clock (Okubo et al. 2012). Together, these results strongly suggest secretion of LFNG from the cell functions to terminate its activity in the segmentation clock.

Little is known about post-transcriptional mechanisms that control RNA turnover in the clock. Intriguingly, we find that prolonging the half-life of LFNG activity somehow reduces Hes7 transcription while stabilizing Hes7 mature mRNA levels. This is the first mutation known to affect Hes7 mRNA turnover in mouse, although mutations that affect turnover of Hes family members in Xenopus and Zebrafish have been described (Davis et al. 2001, Dill and Amacher 2005). The exact mechanisms that regulate mRNA turnover have yet to be elucidated, although micro-RNA regulation is essential in the avian segmentation clock (Riley et al. 2013). It is exciting to speculate that these mechanisms are connected to the segmentation clock and the Notch pathway.

Further analysis of the mutant mice reported here may shed light on this important question.

3.5 Figures

Figure 3.1 Targeting the endogenous Lfng locus to express Golgi-tethered LFNG protein. A. The endogenous Lfng locus was targeted with a vector that replaces 339 bp of Lfng (aa 1-113) with 117bp of RFNG (aa 1-59). This results in replacement of the Lfng pre/pro region (gray circles) and 28 aa of the mature region (gray rectangle) with the signal sequence (hatched circle) and 21 aa (hatched rectangle) of Rfng. Exons are shown as small boxes. The 5’ splice site of intron 1 is maintained, and splices to exons 2-8. B. Southern blot of targeted colonies with 3’ flanking probes is shown (arrow=endogenous band, arrowhead=targeted band, courtesy of Emily Shifley).

Figure 3.2 Mice expressing Golgi-tethered LFNG protein exhibit segmentation abnormalities. A. Heterozygous mice (left) are viable but have a shortened body axis and truncated tail (arrow) compared to wild type littermates. B. 18.5d.p.c. LfngtLFNG/+ fetus (left) and wild type fetus (right). The shortened body axis and truncated tail (arrow) of the mutant are apparent. C. Skeletal defects in wild type (a-c), Lfng null (d-f), and LfngtLFNG/+(g-i) mice. LfngtLFNG/+ mice have more severe rib fusions (arrows in g) and more severe neural arch fusions (* in g and i) than observed in Lfng null mice. In the sacral region Lfng null mice have 1-4 relatively normal vertebrae (arrowheads in f) in contrast to LfngtLFNG/+ animals, which have disorganized sacral regions with ectopic ossifications and arch fusions (arrows in i).

Figure 3.2

Figure 3.3 Segmentation and somite patterning are perturbed in LfngtLFNG/+ embryos. A. LfngtLFNG/+ embryos have morphological segmentation defects at 10.5dpc (compare b and a), and epithelial somites are not observed in sections from LfngtLFNG/+ mutant embryos (compare d and c). In the trunk region, myotomes form, but are fused and disorganized (* in f, compare to e). B. Mesp2 expression is slightly downregulated and diffuse in LfngtLFNG/+ embryos (d) compared to wild type embryos (a). In the mature somites, Tbx18 expression is severely reduced in LfngtLFNG/+mutants (compare e to b) suggesting the somites are caudalized. This is supported by the finding that in LfngtLFNG/+ embryos, no somitic compartmentalization is observed, with Uncx expression seen throughout the mature somite region (compare f to c).

Figure 3.4 Lfng expression is downregulated in LfngtLFNG/+ mutants. Lfng expression oscillates at 10.5 d.p.c. in wild type embryos (a-c, n=4/13 phase 1, 5/13 phase 2, 4/13 phase 3), but is reduced in the posterior PSM of LfngtLFNG/+ embryos (d, n=10). Expression analysis at 8.5 d.p.c reveals cyclic Lfng expression in wild type embryos (e,f, n=4/10 phase 1, 6/10 phase 2) after 6 hours of detection, but no visible Lfng expression in the posterior PSM of LfngtLFNG/+ embryos (g, n=7). A 48 hour detection reveals low levels of Lfng in the posterior PSM of LfngtLFNG/+ embryos (j, n=4), while expression is still dynamic in wild type embryos (h,i, n=2/4 phase 1, 2/4 phase 2).

Figure 3.5 Notch activity is downregulated and no longer dynamic in LfngtLFNG/+ mutants. Whole mount immunohistochemistry for NICD indicates that Notch activity levels are severely downregulated in the posterior PSM of LfngtLFNG/+ embryos (d, n=9) compared to wild type embryos (a-c, n=3/12 phase 1, 6/12 phase 2, 3/12 phase 3) at 10.5 d.p.c.. At 8.5 d.p.c., low levels of noncyclic NICD are visible in the tailbud of LfngtLFNG/+ mutants, whereas expression is dynamic in the posterior PSM of wild type embryos (e,f, n=13) Similarly Nrarp transcript expression is downregulated in the PSM of LfngtLFNG/+embryos (k, n=9) compared to wild type embryos (h-j, n=4/11 phase 1, 3/11 phase 2, 4/11 phase 3).

Figure 3.6 Hes7 mRNA turnover is perturbed in LfngtLFNG/+ embryos. Mature Hes7 mRNA transcripts are observed throughout the PSM of mutant embryos (d, n=8) while levels cycle in wild type embryos (a-c, n=4/11 phase 1, 3/11 phase 2, 4/11 phase 3). However, Hes7 transcription, as revealed with an intron-specific probe, is reduced in the PSM of mutant embryos (h, n=7), while it is cyclic in wild type embryos (e-g, n=2/8 phase 1, 2/8 phase 2, 4/8 phase 3).

Figure 3.7 ). FGF and Wnt signaling are perturbed in LfngtLFNG/+ embryos. MAPK activity is detected as a gradient in the posterior PSM of wild type embryos (a, n=7) but is reduced (n=1/7) or undetectable (n=6/7) in LfngtLFNG/+ embryos (b,c) While Spry2 expression oscillates in wild type embryos at 10.5 dpc (d,e, n=8/18 phase 1, 10/18 phase 2), it is slightly reduced and non-cyclic in LfngtLFNG/+ embryos (f, n=9). Similarly, Axin2 expression oscillates in wild type embryos at 9.5 (g,h, n=5/12 phase 1, 7/12 phase 2) and 10.5 (j,k, n=5/11 phase 1, 6/11 phase 2) d.p.c., but is significantly reduced in the posterior of LfngtLFNG/+ embryos (i,f, n=17).

Figure 2.8 The segmentation clock in LfngtLFNG/+ mutants. A model of clock regulation in wild type (A) and LfngtLFNG/+ mutants (B). Although Lfng mRNA levels are decreased in LfngtLFNG/+ embryos, the low levels of mRNA give rise to an extremely stable protein, which constitutively represses NOTCH1, resulting in decreased activation of Hes7 and Lfng. Although Hes7 transcription is downregulated, increased LFNG activity inhibits Hes7 mRNA turnover through an unknown mechanism, resulting in ubiquitous expression of Hes7 mRNA in the PSM. Elevated levels of HES7 protein and reduced Notch activation may maintain the downregulation of Hes7 and Lfng transcription. 81

CHAPTER 4

CONCLUSION

4.1 Overview

The research presented in this study examines the transcriptional and post- transcriptional control of Lfng during somitogenesis. Somitogenesis is the process that gives rise to the segmented structures of the vertebrate body, including the vertebrae and ribs. The regulation of somitogenesis is complex. In the PSM, a segmentation clock and a wavefront of signaling gradients integrate spatial and temporal information to organize cells into somite-sized units. These presomites undergo rostro-caudal patterning in the anterior PSM. The Notch pathway plays roles in both the segmentation clock and during somite patterning. Mutations in Notch pathway members result in multiple segmentation defects that reflect the distinct roles of Notch in the PSM.

For segmentation to occur properly, Notch activity levels must be tightly regulated. This study focuses on the regulation of Notch by the glycosyltransferase

LFNG. We find that oscillatory Lfng expression is critical for all stages of somitogenesis, although posterior development tolerates a lower amplitude of Lfng oscillations. In 82

addition to oscillatory transcription, LFNG protein levels also must oscillate during somitogenesis. For protein levels to oscillate during the short period of the clock, there must be mechanisms that promote rapid LFNG protein turnover. Here, we find that secretion of LFNG terminates its function in the segmentation clock and is essential for somitogenesis.

4.2 Oscillatory Lfng is required during both anterior and posterior development.

During somitogenesis, Lfng mRNA exhibits oscillatory expression in the posterior

PSM, where the clock is active, and stable expression in the patterning region. In chapter

2, we further analyzed a mutant allele of Lfng thought to eliminate oscillatory expression of Lfng in the segmentation clock. Mice homozygous for this allele have severe segmentation defects in the anterior skeleton, but segmentation of the posterior skeleton is relatively normal (Shifley et al. 2008). This observation suggested that oscillatory Lfng expression and periodic Notch activation are essential for anterior development, but dispensable during posterior skeletal development, and that expression of Lfng in the patterning region is more important during posterior development.

Thus, our detailed analysis of Lfng∆FCE/∆FCE mice have helped clarify conflicting data regarding the requirement of Lfng during secondary body formation. We find that expression of Lfng remains oscillatory in Lfng∆FCE/∆FCE animals and that these oscillations are essential for all stages of somitogenesis. These dampened oscillations are able to promote dynamic levels of NICD that are sufficient for relatively normal posterior development. These subtle oscillations could only be detected by RT-PCR and western blotting. These results suggest caution should be used when drawing strong conclusions

about oscillatory mRNA and protein levels based on whole-mount in situ and immunohistochemical analysis.

It is unclear what role, if any, Lfng expression plays in the somite patterning region. To date, it has not been possible to eliminate endogenous expression of Lfng specifically in region II without affecting its function in the segmentation clock. It is possible that Lfng expression in this region helps refine somite patterning, but that this subtle role is overwhelmed by defects resulting from loss of Lfng function in the clock.

Lfng null mice expressing Lfng from an oscillatory transgene that lacks stable expression in the patterning region have occasional rib fusions, which may be due to lack of stable

Lfng expression in the rostral somite compartment (Oginuma et al. 2010).

4.3 The segmentation clock period is sensitive to Lfng dosage in the PSM

Here, we show that lowering the amplitude of Lfng oscillations lengthens the period of the segmentation clock, resulting in fewer vertebral elements being formed during primary and secondary body formation. However, as Lfng levels are reduced, it appears that the posterior skeleton loses more vertebral elements than the anterior skeleton. Although we do not observe premature apoptosis of the tailbud, it is interesting to speculate that segmentation clock oscillations arrest prematurely in Lfng mutants.

During normal chick development, the segmentation clock is arrested before all of the paraxial mesoderm has segmented (Tenin et al, 2010). If this arrests occured earlier than normal, caudal truncations would result. Because some genes, such as Hes7, remain oscillatory in Lfng nulls, at least at 10.5 d.p.c (Shifley et al., 2008), examining their

oscillations at later stages of development may resolve whether the segmentation clock arrests prematurely in Lfng mutants and contributes to caudal truncations.

4.4 Secretion of LFNG fringe regulates its activity in the segmentation clock

In chapter 3, we examined the mechanism of LFNG protein turnover in the segmentation clock. We hypothesized that secretion of LFNG is a mechanism to facilitate its rapid turnover and inactivation. To test this, we targeted a mutation to the Lfng locus to prevent secretion. Mice carrying a single copy of this allele exhibited skeletal defects more severe than mice lacking all LFNG activity, demonstrating that secretion of LFNG is a critical aspects of its regulation in the clock.

In the anterior PSM, the expression pattern of SPC6 suggests a role in clearing

Lfng from the anterior compartment (Shifley and Cole 2008). Because Lfng is expressed in both the clock region and the patterning region, we are unable to distinguish what segmentation defects are the result of prolonged LFNG activity in the clock versus prolonged activity in the patterning region. In the future, it will be interesting to express tLFNG protein specifically in the anterior compartment. We have generated a conditional allele of the tLFNG mutation which will allow us to conditionally express tethered LFNG protein specifically in the anterior PSM, where somites are undergoing R/C patterning, but not in the posterior PSM, where the clock is functioning. These mice will help determine whether cleavage of LFNG by SPC6 in the anterior somite compartment is important for somite patterning.

Although LFNG is cleaved at two cleavage sites in vivo, cleavage is not necessary for secretion, although cleaved LFNG is secreted more rapidly in tissue culture (Shifley

and Cole 2008). It is possible that cleaved LFNG has a reduced ability to modify the

Notch receptor in the Golgi, as soluble LFNG would be less likely to come into contact with the Notch receptor in the Golgi membrane. We have targeted a mutation in mouse embryonic stem cells that prevents cleavage of LFNG but not its secretion. Analysis of these mice will provide the first examination of the in vivo relevance of LFNG cleavage and further our understanding of the post-translational regulation of LFNG.

4.5 mRNA turnover may be linked to the clock

Interestingly, Hes7 mRNA turnvover is impaired in LfngtLFNG/+ mutants. This suggests that the mechanisms regulating mRNA half-life during somitogenesis may be linked to the segmentation clock or the Notch pathway. Further supporting this hypothesis, our preliminary data suggests that Hes7 mRNA turnover is also impaired in the Dll3 pudgy mutant, which has reduced Notch signaling and expresses low levels of

NICD in a pattern similar to LfngtLFNG/+ mutants. Because the 3’UTR of Hes7 promotes its turnover, it is attractive to speculate that miRNAs that regulate clock genes are under the control of the segmentation clock. Thus, understanding the mechanisms by which

Notch activity promotes Hes7 mRNA turnover is an exciting future direction that will further broaden our understanding of the post-transcriptional mechanisms regulating segmentation clock function.

REFERENCES

Abu-Abed S, Dolle P, Metzger D, Beckett B, Chambon P, Petkovich M. 2001. The retinoic acid-metabolizing enzyme, CYP26A1, is essential for normal hindbrain patterning, vertebral identity, and development of posterior structures. Genes Dev 15: 226-240.

Aulehla A, Herrmann BG. 2004. Segmentation in vertebrates: clock and gradient finally joined. Genes Dev 18: 2060-2067.

Aulehla A, Johnson RL. 1999. Dynamic expression of lunatic fringe suggests a link between notch signaling and an autonomous cellular oscillator driving somite segmentation. Dev Biol 207: 49-61.

Aulehla A, Pourquie O. 2010. Signaling gradients during paraxial mesoderm development. Cold Spring Harb Perspect Biol 2: a000869.

Aulehla A, Wehrle C, Brand-Saberi B, Kemler R, Gossler A, Kanzler B, Herrmann BG. 2003. Wnt3a plays a major role in the segmentation clock controlling somitogenesis. Dev Cell 4: 395-406.

Aulehla A, Wiegraebe W, Baubet V, Wahl MB, Deng C, Taketo M, Lewandoski M, Pourquie O. 2008. A beta-catenin gradient links the clock and wavefront systems in mouse embryo segmentation. Nat Cell Biol 10: 186-193.

Barrallo-Gimeno A, Nieto MA. 2005. The Snail genes as inducers of cell movement and survival: implications in development and cancer. Development 132: 3151-3161.

Bessho Y, Hirata H, Masamizu Y, Kageyama R. 2003. Periodic repression by the bHLH factor Hes7 is an essential mechanism for the somite segmentation clock. Genes & development 17: 1451-6.

Bessho Y, Miyoshi G, Sakata R, Kageyama R. 2001. Hes7: a bHLH-type repressor gene regulated by Notch and expressed in the presomitic mesoderm. Genes to cells : devoted to molecular & cellular mechanisms 6: 175-85.

Bolos V, Grego-Bessa J, de la Pompa JL. 2007. Notch signaling in development and cancer. Endocr Rev 28: 339-363. 87

Brent AE, Schweitzer R, Tabin CJ. 2003. A somitic compartment of tendon progenitors. Cell 113: 235-248.

Bruckner K, Perez L, Clausen H, Cohen S. 2000. Glycosyltransferase activity of Fringe modulates Notch-Delta interactions. Nature 406: 411-415.

Bunting M, Bernstein KE, Greer JM, Capecchi MR, Thomas KR. 1999. Targeting genes for self-excision in the germ line. Genes & development 13: 1524-8.

Bussen M, Petry M, Schuster-Gossler K, Leitges M, Gossler A, Kispert A. 2004. The T- box transcription factor Tbx18 maintains the separation of anterior and posterior somite compartments. Genes & development 18: 1209-21.

Christ B, Jacob HJ, Jacob M. 1974. Somitogenesis in the chick embryo. Determination of the segmentation direction. Verh Anat Ges 68: 573-579.

Cinquin O. 2003. Is the somitogenesis clock really cell-autonomous? A coupled- oscillator model of segmentation. Journal of theoretical biology 224: 459-68.

Cole SE, Levorse JM, Tilghman SM, Vogt TF. 2002. Clock regulatory elements control cyclic expression of Lunatic fringe during somitogenesis. Developmental cell 3: 75-84.

Conlon RA, Reaume AG, Rossant J. 1995. Notch1 is required for the coordinate segmentation of somites. Development 121: 1533-1545.

Cooke J. 1975. Control of somite number during morphogenesis of a vertebrate, Xenopus laevis. Nature 254: 196-199.

Cooke J, Zeeman EC. 1976. A clock and wavefront model for control of the number of repeated structures during animal morphogenesis. J Theor Biol 58: 455-476.

Cordes R, Schuster-Gossler K, Serth K, Gossler A. 2004. Specification of vertebral identity is coupled to Notch signalling and the segmentation clock. Development 131: 1221-1233.

Dale JK, Malapert P, Chal J, Vilhais-Neto G, Maroto M, Johnson T, Jayasinghe S, Trainor P, Herrmann B, Pourquie O. 2006. Oscillations of the snail genes in the presomitic mesoderm coordinate segmental patterning and morphogenesis in vertebrate somitogenesis. Dev Cell 10: 355-366.

Dale JK, Maroto M, Dequeant ML, Malapert P, McGrew M, Pourquie O. 2003. Periodic notch inhibition by lunatic fringe underlies the chick segmentation clock. Nature 421: 275-8.

Davis RL, Turner DL, Evans LM, Kirschner MW. 2001. Molecular targets of vertebrate segmentation: two mechanisms control segmental expression of Xenopus hairy2 during somite formation. Developmental cell 1: 553-65.

Delaune EA, Francois P, Shih NP, Amacher SL. 2012. Single-cell-resolution imaging of the impact of Notch signaling and mitosis on segmentation clock dynamics. Dev Cell 23: 995-1005.

Delfini MC, Dubrulle J, Malapert P, Chal J, Pourquie O. 2005. Control of the segmentation process by graded MAPK/ERK activation in the chick embryo. Proc Natl Acad Sci U S A 102: 11343-11348.

Dequeant ML, Glynn E, Gaudenz K, Wahl M, Chen J, Mushegian A, Pourquie O. 2006. A complex oscillating network of signaling genes underlies the mouse segmentation clock. Science 314: 1595-8.

Dias AS, de Almeida I, Belmonte JM, Glazier JA, Stern CD. 2014. Somites without a clock. Science 343: 791-795.

Dill KK, Amacher SL. 2005. tortuga refines Notch pathway gene expression in the zebrafish presomitic mesoderm at the post-transcriptional level. Developmental biology 287: 225-36.

Dubrulle J, McGrew MJ, Pourquie O. 2001. FGF signaling controls somite boundary position and regulates segmentation clock control of spatiotemporal Hox gene activation. Cell 106: 219-232.

Dubrulle J, Pourquie O. 2004. fgf8 mRNA decay establishes a gradient that couples axial elongation to patterning in the vertebrate embryo. Nature 427: 419-422.

Dunty WC,Jr, Biris KK, Chalamalasetty RB, Taketo MM, Lewandoski M, Yamaguchi TP. 2008. Wnt3a/beta-catenin signaling controls posterior body development by coordinating mesoderm formation and segmentation. Development 135: 85-94.

Evrard YA, Lun Y, Aulehla A, Gan L, Johnson RL. 1998. lunatic fringe is an essential mediator of somite segmentation and patterning. Nature 394: 377-81.

Feller J, Schneider A, Schuster-Gossler K, Gossler A. 2008. Noncyclic Notch activity in the presomitic mesoderm demonstrates uncoupling of somite compartmentalization and boundary formation. Genes & development 22: 2166-71.

Feng P, Navaratna M. 2007. Modelling periodic oscillations during somitogenesis. Math Biosci Eng 4: 661-673.

Ferjentsik Z, Hayashi S, Dale JK, Bessho Y, Herreman A, De Strooper B, del Monte G, de la Pompa JL, Maroto M. 2009. Notch is a critical component of the mouse somitogenesis oscillator and is essential for the formation of the somites. PLoS genetics 5: e1000662.

Forsberg H, Crozet F, Brown NA. 1998. Waves of mouse Lunatic fringe expression, in four-hour cycles at two-hour intervals, precede somite boundary formation. Curr Biol 8: 1027-1030.

Fortini ME, Artavanis-Tsakonas S. 1994. The suppressor of hairless protein participates in notch receptor signaling. Cell 79: 273-282.

Fujii H, Sato T, Kaneko S, Gotoh O, Fujii-Kuriyama Y, Osawa K, Kato S, Hamada H. 1997. Metabolic inactivation of retinoic acid by a novel P450 differentially expressed in developing mouse embryos. EMBO J 16: 4163-4173.

Gantier MP, McCoy CE, Rusinova I, Saulep D, Wang D, Xu D, Irving AT, Behlke MA, Hertzog PJ, Mackay F et al. 2011. Analysis of microRNA turnover in mammalian cells following Dicer1 ablation. Nucleic Acids Res 39: 5692-5703.

Handrigan GR. 2003. Concordia discors: duality in the origin of the vertebrate tail. J Anat 202: 255-267.

Haraguchi S, Kitajima S, Takagi A, Takeda H, Inoue T, Saga Y. 2001. Transcriptional regulation of Mesp1 and Mesp2 genes: differential usage of enhancers during development. Mechanisms of development 108: 59-69.

Harima Y, Takashima Y, Ueda Y, Ohtsuka T, Kageyama R. 2013. Accelerating the tempo of the segmentation clock by reducing the number of introns in the Hes7 gene. Cell Rep 3: 1-7.

Hicks C, Johnston SH, diSibio G, Collazo A, Vogt TF, Weinmaster G. 2000. Fringe differentially modulates Jagged1 and Delta1 signalling through Notch1 and Notch2. Nature cell biology 2: 515-20.

Hilgers V, Pourquie O, Dubrulle J. 2005. In vivo analysis of mRNA stability using the Tet-Off system in the chicken embryo. Dev Biol 284: 292-300.

Hirata H, Bessho Y, Kokubu H, Masamizu Y, Yamada S, Lewis J, Kageyama R. 2004. Instability of Hes7 protein is crucial for the somite segmentation clock. Nature genetics 36: 750-4.

Holmdahl DE. 1925. Experimentelle untersuchungen uber die lage der grenze zwischen primarer und sekundarere korperentwicklung beim huhn. Anaz Anat 59: 393-396. 90

Hori K, Fostier M, Ito M, Fuwa TJ, Go MJ, Okano H, Baron M, Matsuno K. 2004. Drosophila deltex mediates suppressor of Hairless-independent and late-endosomal activation of Notch signaling. Development 131: 5527-5537.

Horikawa K, Ishimatsu K, Yoshimoto E, Kondo S, Takeda H. 2006. Noise-resistant and synchronized oscillation of the segmentation clock. Nature 441: 719-723.

Hoyle NP, Ish-Horowicz D. 2013. Transcript processing and export kinetics are rate- limiting steps in expressing vertebrate segmentation clock genes. Proc Natl Acad Sci U S A 110: E4316-24.

Hrabe de Angelis M, McIntyre J,2nd, Gossler A. 1997. Maintenance of somite borders in mice requires the Delta homologue DII1. Nature 386: 717-721.

Huang MT, Gorman CM. 1990. The simian virus 40 small-t intron, present in many common expression vectors, leads to aberrant splicing. Molecular and cellular biology 10: 1805-10.

Huang R, Zhi Q, Brand-Saberi B, Christ B. 2000. New experimental evidence for somite resegmentation. Anat Embryol (Berl) 202: 195-200.

Huppert SS, Ilagan MX, De Strooper B, Kopan R. 2005. Analysis of Notch function in presomitic mesoderm suggests a gamma-secretase-independent role for presenilins in somite differentiation. Developmental cell 8: 677-88.

Irvine KD. 1999. Fringe, Notch, and making developmental boundaries. Current opinion in genetics & development 9: 434-41.

Ishikawa A, Kitajima S, Takahashi Y, Kokubo H, Kanno J, Inoue T, Saga Y. 2004. Mouse Nkd1, a Wnt antagonist, exhibits oscillatory gene expression in the PSM under the control of Notch signaling. Mech Dev 121: 1443-1453.

Ishimatsu K, Takamatsu A, Takeda H. 2010. Emergence of traveling waves in the zebrafish segmentation clock. Development 137: 1595-1599.

Jensen MH, Sneppen K, Tiana G. 2003. Sustained oscillations and time delays in gene expression of protein Hes1. FEBS Lett 541: 176-177.

Jiang YJ, Aerne BL, Smithers L, Haddon C, Ish-Horowicz D, Lewis J. 2000. Notch signalling and the synchronization of the somite segmentation clock. Nature 408: 475- 479.

Johnston SH, Rauskolb C, Wilson R, Prabhakaran B, Irvine KD, Vogt TF. 1997. A family of mammalian Fringe genes implicated in boundary determination and the Notch 91

pathway. Development 124: 2245-54.

Kageyama R, Niwa Y, Isomura A, Gonzalez A, Harima Y. 2012. Oscillatory gene expression and somitogenesis. Wiley interdisciplinary reviews Developmental biology 1: 629-41.

Keynes RJ, Stern CD. 1984. Segmentation in the vertebrate nervous system. Nature 310: 786-789.

Kim DG, Kang HM, Jang SK, Shin HS. 1992. Construction of a bifunctional mRNA in the mouse by using the internal ribosomal entry site of the encephalomyocarditis virus. Mol Cell Biol 12: 3636-3643.

Kim W, Matsui T, Yamao M, Ishibashi M, Tamada K, Takumi T, Kohno K, Oba S, Ishii S, Sakumura Y et al. 2011. The period of the somite segmentation clock is sensitive to Notch activity. Mol Biol Cell 22: 3541-3549.

Kraus F, Haenig B, Kispert A. 2001. Cloning and expression analysis of the mouse T-box gene Tbx18. Mechanisms of development 100: 83-6.

Krol AJ, Roellig D, Dequeant ML, Tassy O, Glynn E, Hattem G, Mushegian A, Oates AC, Pourquie O. 2011. Evolutionary plasticity of segmentation clock networks. Development 138: 2783-2792.

Kuersten S, Goodwin EB. 2003. The power of the 3' UTR: translational control and development. Nat Rev Genet 4: 626-637.

Kusumi K, Sun ES, Kerrebrock AW, Bronson RT, Chi DC, Bulotsky MS, Spencer JB, Birren BW, Frankel WN, Lander ES. 1998. The mouse pudgy mutation disrupts Delta homologue Dll3 and initiation of early somite boundaries. Nat Genet 19: 274-278.

Lewis J. 2003. Autoinhibition with transcriptional delay: a simple mechanism for the zebrafish somitogenesis oscillator. Curr Biol 13: 1398-1408.

Mansouri A, Yokota Y, Wehr R, Copeland NG, Jenkins NA, Gruss P. 1997. Paired- related murine homeobox gene expressed in the developing sclerotome, kidney, and nervous system. Developmental dynamics 210: 53-65.

Mara A, Schroeder J, Chalouni C, Holley SA. 2007. Priming, initiation and synchronization of the segmentation clock by deltaD and deltaC. Nat Cell Biol 9: 523- 92

530.

McGrew MJ, Dale JK, Fraboulet S, Pourquie O. 1998. The lunatic fringe gene is a target of the molecular clock linked to somite segmentation in avian embryos. Curr Biol 8: 979- 982.

McLeod MJ. 1980. Differential staining of cartilage and bone in whole mouse fetuses by alcian blue and alizarin red S. Teratology 22: 299-301.

Moloney DJ, Panin VM, Johnston SH, Chen J, Shao L, Wilson R, Wang Y, Stanley P, Irvine KD, Haltiwanger RS et al. 2000. Fringe is a glycosyltransferase that modifies Notch. Nature 406: 369-75.

Monk NA. 2003. Oscillatory expression of Hes1, p53, and NF-kappaB driven by transcriptional time delays. Curr Biol 13: 1409-1413.

Morales AV, Yasuda Y, Ish-Horowicz D. 2002. Periodic Lunatic fringe expression is controlled during segmentation by a cyclic transcriptional enhancer responsive to notch signaling. Developmental cell 3: 63-74.

Morimoto M, Takahashi Y, Endo M, Saga Y. 2005. The Mesp2 transcription factor establishes segmental borders by suppressing Notch activity. Nature 435: 354-9.

Morkel M, Huelsken J, Wakamiya M, Ding J, van de Wetering M, Clevers H, Taketo MM, Behringer RR, Shen MM, Birchmeier W. 2003. Beta-catenin regulates Cripto- and Wnt3-dependent gene expression programs in mouse axis and mesoderm formation. Development 130: 6283-6294.

Munro S, Freeman M. 2000. The notch signalling regulator fringe acts in the Golgi apparatus and requires the glycosyltransferase signature motif DXD. Curr Biol 10: 813- 20.

Naiche LA, Holder N, Lewandoski M. 2011. FGF4 and FGF8 comprise the wavefront activity that controls somitogenesis. Proc Natl Acad Sci U S A 108: 4018-4023.

Niederreither K, McCaffery P, Drager UC, Chambon P, Dolle P. 1997. Restricted expression and retinoic acid-induced downregulation of the retinaldehyde dehydrogenase 93

type 2 (RALDH-2) gene during mouse development. Mech Dev 62: 67-78.

Niederreither K, Subbarayan V, Dolle P, Chambon P. 1999. Embryonic retinoic acid synthesis is essential for early mouse post-implantation development. Nat Genet 21: 444- 448.

Nitanda Y, Matsui T, Matta T, Higami A, Kohno K, Nakahata Y, Bessho Y. 2014. 3'- UTR-dependent regulation of mRNA turnover is critical for differential distribution patterns of cyclic gene mRNAs. FEBS J 281: 146-156.

Niwa Y, Masamizu Y, Liu T, Nakayama R, Deng CX, Kageyama R. 2007. The initiation and propagation of Hes7 oscillation are cooperatively regulated by Fgf and notch signaling in the somite segmentation clock. Developmental cell 13: 298-304.

Nomura-Kitabayashi A, Takahashi Y, Kitajima S, Inoue T, Takeda H, Saga Y. 2002. Hypomorphic Mesp allele distinguishes establishment of rostrocaudal polarity and segment border formation in somitogenesis. Development 129: 2473-81.

Oginuma M, Niwa Y, Chapman DL, Saga Y. 2008. Mesp2 and Tbx6 cooperatively create periodic patterns coupled with the clock machinery during mouse somitogenesis. Development 135: 2555-62.

Oginuma M, Takahashi Y, Kitajima S, Kiso M, Kanno J, Kimura A, Saga Y. 2010. The oscillation of Notch activation, but not its boundary, is required for somite border formation and rostral-caudal patterning within a somite. Development 137: 1515-22.

Okubo Y, Sugawara T, Abe-Koduka N, Kanno J, Kimura A, Saga Y. 2012. Lfng regulates the synchronized oscillation of the mouse segmentation clock via trans- repression of Notch signalling. Nature communications 3: 1141.

Olivera-Martinez I, Harada H, Halley PA, Storey KG. 2012. Loss of FGF-dependent mesoderm identity and rise of endogenous retinoid signalling determine cessation of body axis elongation. PLoS Biol 10: e1001415.

Ozbudak EM, Lewis J. 2008. Notch signalling synchronizes the zebrafish segmentation clock but is not needed to create somite boundaries. PLoS Genet 4: e15.

Ozbudak EM, Pourquie O. 2008. The vertebrate segmentation clock: the tip of the iceberg. Curr Opin Genet Dev 18: 317-323.

Packard DS,Jr. 1976. The influence of axial structures on chick somite formation. Dev Biol 53: 36-48.

Palmeirim I, Henrique D, Ish-Horowicz D, Pourquie O. 1997. Avian hairy gene expression identifies a molecular clock linked to vertebrate segmentation and somitogenesis. Cell 91: 639-648.

Panin VM, Papayannopoulos V, Wilson R, Irvine KD. 1997. Fringe modulates Notch- ligand interactions. Nature 387: 908-12.

Panin VM, Shao L, Lei L, Moloney DJ, Irvine KD, Haltiwanger RS. 2002. Notch ligands are substrates for protein O-fucosyltransferase-1 and Fringe. The Journal of biological chemistry 277: 29945-52.

Reaume AG, Conlon RA, Zirngibl R, Yamaguchi TP, Rossant J. 1992. Expression analysis of a Notch homologue in the mouse embryo. Dev Biol 154: 377-387.

Rickmann M, Fawcett JW, Keynes RJ. 1985. The migration of neural crest cells and the growth of motor axons through the rostral half of the chick somite. J Embryol Exp Morphol 90: 437-455.

Riedel-Kruse IH, Muller C, Oates AC. 2007. Synchrony dynamics during initiation, failure, and rescue of the segmentation clock. Science 317: 1911-1915.

Riley MF, Bochter MS, Wahi K, Nuovo GJ, Cole SE. 2013. Mir-125a-5p-mediated regulation of Lfng is essential for the avian segmentation clock. Developmental cell 24: 554-61.

Saga Y, Hata N, Koseki H, Taketo MM. 1997. Mesp2: a novel mouse gene expressed in the presegmented mesoderm and essential for segmentation initiation. Genes & development 11: 1827-39.

Saga Y, Takeda H. 2001. The making of the somite: molecular events in vertebrate segmentation. Nat Rev Genet 2: 835-845.

Sakai Y, Meno C, Fujii H, Nishino J, Shiratori H, Saijoh Y, Rossant J, Hamada H. 2001. The retinoic acid-inactivating enzyme CYP26 is essential for establishing an uneven distribution of retinoic acid along the anterio-posterior axis within the mouse embryo. Genes Dev 15: 213-225.

Sasaki N, Kiso M, Kitagawa M, Saga Y. 2011. The repression of Notch signaling occurs via the destabilization of mastermind-like 1 by Mesp2 and is essential for somitogenesis. Development 138: 55-64.

Sato Y, Watanabe T, Saito D, Takahashi T, Yoshida S, Kohyama J, Ohata E, Okano H, Takahashi Y. 2008. Notch mediates the segmental specification of angioblasts in somites and their directed migration toward the dorsal aorta in avian embryos. Dev Cell 14: 890- 901.

Sato Y, Yasuda K, Takahashi Y. 2002. Morphological boundary forms by a novel inductive event mediated by Lunatic fringe and Notch during somitic segmentation. Development 129: 3633-3644.

Sawada A, Shinya M, Jiang YJ, Kawakami A, Kuroiwa A, Takeda H. 2001. Fgf/MAPK signalling is a crucial positional cue in somite boundary formation. Development 128: 4873-4880.

Schweitzer R, Chyung JH, Murtaugh LC, Brent AE, Rosen V, Olson EN, Lassar A, Tabin CJ. 2001. Analysis of the tendon cell fate using Scleraxis, a specific marker for tendons and ligaments. Development 128: 3855-3866.

Serth K, Schuster-Gossler K, Cordes R, Gossler A. 2003. Transcriptional oscillation of lunatic fringe is essential for somitogenesis. Genes & development 17: 912-25.

Shifley ET, Cole SE. 2008. Lunatic fringe protein processing by proprotein convertases may contribute to the short protein half-life in the segmentation clock. Biochimica et biophysica acta 1783: 2384-90.

Shifley ET, Cole SE. 2007. The vertebrate segmentation clock and its role in skeletal birth defects. Birth defects research Part C, Embryo today : reviews 81: 121-33.

Shifley ET, Vanhorn KM, Perez-Balaguer A, Franklin JD, Weinstein M, Cole SE. 2008. Oscillatory lunatic fringe activity is crucial for segmentation of the anterior but not posterior skeleton. Development 135: 899-908.

Shimozono S, Iimura T, Kitaguchi T, Higashijima S, Miyawaki A. 2013. Visualization of an endogenous retinoic acid gradient across embryonic development. Nature 496: 363- 366.

Stauber M, Laclef C, Vezzaro A, Page ME, Ish-Horowicz D. 2012. Modifying transcript lengths of cycling mouse segmentation genes. Mech Dev 129: 61-72.

Stauber M, Sachidanandan C, Morgenstern C, Ish-Horowicz D. 2009. Differential axial requirements for lunatic fringe and Hes7 transcription during mouse somitogenesis. PloS 96

one 4: e7996.

Struhl G, Adachi A. 1998. Nuclear access and action of notch in vivo. Cell 93: 649-660.

Suriben R, Fisher DA, Cheyette BN. 2006. Dact1 presomitic mesoderm expression oscillates in phase with Axin2 in the somitogenesis clock of mice. Dev Dyn 235: 3177- 3183.

Takahashi Y, Koizumi K, Takagi A, Kitajima S, Inoue T, Koseki H, Saga Y. 2000. Mesp2 initiates somite segmentation through the Notch signalling pathway. Nature genetics 25: 390-6.

Takashima Y, Ohtsuka T, Gonzalez A, Miyachi H, Kageyama R. 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: 3300-5.

Tam PP. 1981. The control of somitogenesis in mouse embryos. J Embryol Exp Morphol 65 Suppl: 103-128.

Tenin G, Wright D, Ferjentsik Z, Bone R, McGrew MJ, Maroto M. 2010. The chick somitogenesis oscillator is arrested before all paraxial mesoderm is segmented into somites. BMC Dev Biol 10: 24-213X-10-24.

Vilhais-Neto GC, Maruhashi M, Smith KT, Vasseur-Cognet M, Peterson AS, Workman JL, Pourquie O. 2010. Rere controls retinoic acid signalling and somite bilateral symmetry. Nature 463: 953-957.

Wahl MB, Deng C, Lewandoski M, Pourquie O. 2007. FGF signaling acts upstream of the NOTCH and WNT signaling pathways to control segmentation clock oscillations in mouse somitogenesis. Development 134: 4033-41.

Whiting J, Marshall H, Cook M, Krumlauf R, Rigby PW, Stott D, Allemann RK. 1991. Multiple spatially specific enhancers are required to reconstruct the pattern of Hox-2.6 gene expression. Genes Dev 5: 2048-2059.

Williams DR, Shifley ET, Lather JD, Cole SE. 2014. Posterior skeletal development and the segmentation clock period are sensitive to Lfng dosage during somitogenesis. Dev Biol. 388: 159-169

Wilson V, Beddington RS. 1996. Cell fate and morphogenetic movement in the late mouse primitive streak. Mech Dev 55: 79-89.

Wright D, Ferjentsik Z, Chong SW, Qiu X, Jiang YJ, Malapert P, Pourquie O, Van Hateren N, Wilson SA, Franco C et al. 2009. Cyclic Nrarp mRNA expression is regulated 97

by the somitic oscillator but Nrarp protein levels do not oscillate. Developmental dynamics : an official publication of the American Association of Anatomists 238: 3043- 55.

Xie ZR, Yang HT, Liu WC, Hwang MJ. 2007. The role of microRNA in the delayed negative feedback regulation of gene expression. Biochem Biophys Res Commun 358: 722-726.

Zhang N, Gridley T. 1998. Defects in somite formation in lunatic fringe-deficient mice. Nature 394: 374-7.

Zhang Z, O'Rourke JR, McManus MT, Lewandoski M, Harfe BD, Sun X. 2011. The microRNA-processing enzyme Dicer is dispensable for somite segmentation but essential for limb bud positioning. Dev Biol 351: 254-265.