The Role of Iroquois 3 and 5 in Limb Bud Pattern Formation and Morphogenesis

Danyi Li

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Molecular Genetics University of Toronto

Danyi Li

Doctor of Philosophy

Graduate Department of Molecular Genetics University of Toronto

2014 Abstract

Limb skeletal pattern heavily relies on graded Sonic hedgehog (Shh) signaling. As a morphogen and growth cue, Shh regulates identities of posterior limb elements including the ulna/fibula and digits 2 through 5. In contrast, proximal and anterior structures including the humerus/femur, radius/tibia and digit 1 are regarded as Shh-independent and the mechanisms governing their pattern formation are unclear. Here I show that patterning of the proximal and anterior limb skeleton involves an early specification phase dependent on two transcription factors, Irx3 and Irx5 (Irx3/5). They are expressed in the anterior margin of the limb field to regulate expression of the key anterior prepattern gene Gli3 and specify the anterior population during limb initiation. The early specification phase is followed by a late modulation phase during which Shh signaling negatively regulates these anterior elements. In addition, Irx3/5 also play a role in limb bud morphogenesis. Irx3/5-DKO hindlimb buds are small with abnormal shape due to prolonged cell cycle time and division defects including anaphase bridge and altered cell division orientation in the anterior hindlimb field. Therefore, pattern formation of anterior limb skeletal elements requires functions of Irx3/5 during limb bud initiation for the specification and expansion of anterior population followed by negative modulation by Shh signaling.

Acknowledgments

I would like to take this opportunity to gratefully acknowledge the following people for their contributions and support throughout my graduate study.

To my supervisors Dr. Chi-chung Hui and Dr. Sevan Hopyan, who gave me the chance to work on this exciting and challenging project and provide me with continuous guidance and opportunities to help me learn to be a scientist and become a better person, thank you for all the things that you have done for me. What you have taught me will benefit me in my future career and throughout my life. To my committee members Dr. Sabine Cordes and Dr. Andrew Spence, who gave me valuable comments and suggestions on my work, thank you for your attentiveness and scientific expertise.

I am very lucky to work with a group of people who are always willing to help and make my time in the lab so enjoyable. To Mary Zhang, Rong Mo, Vijitha Puviindran and Kendra Sturgeon, thank you for your help from maintaining mouse colonies to mutant analysis and molecular experiments. I want to thank Dr. Rui Sakuma and Niki Vakili for initiating this project. To Olena Zhulyn, Wenqi Yin, Dr. Sue Li, Dr. Han Kim, Dr. Kelvin Law, Dr. Kimberly Lau, Dr. Steven Deimling, Dr. Hirotaka Tao and Laurie Wyngaarden, thank you for all the scientific and non-scientific discussions and sharing experience from experimental techniques to good food and sale discount. I especially thank Olena for helping with the flow cytometry experiment and brain storm on my manuscript development. I also want to thank Gregory Anderson from the Henkelman lab for collaboration on OPT analysis.

I am also grateful to my family and friends who are always by my side cheering for my success and accomplishment and encouraging me to get through my “dark time”.

iii

Table of Contents

Abstract ii Acknowledgment iii List of Tables vii List of Figures viii List of Videos x List of Abbreviations xi

Chapter 1 INTRODUCTION 1 1 Introduction 1 1.1 An overview of vertebrate limb development and pattern formation 1 1.1.1 Limb induction 1 1.1.2 AER and limb PD pattern formation 2 1.1.3 ZPA and limb AP pattern formation 3 1.1.4 Self-regulated feedback loop between the AER and ZPA 4 1.1.5 Proliferative expansion, determination and differentiation 4 1.2 Limb AP pattern formation 5 1.2.1 Prepatterning the limb bud and Shh-activation network 5 1.2.2 Limb AP pattern formation after Shh activation 7 1.3 Cell biology of limb bud morphogenesis 9 1.3.1 Cell proliferation and cell death 9 1.3.2 Oriented cell behavior 10 1.4 Iroquois homeobox (Irx) genes 11 1.4.1 Discovery and genomic organization 11 1.4.2 Functions of Irx genes in development 12 1.4.3 Irx3 and Irx5 have novel functions in mouse hindlimb development 14 1.5 Thesis rational and outline 16

Chapter 2 Irx3/5 interact with Shh signaling for anterior limb pattern formation 22 2 Chapter 2 23 2.1 Summary 23 2.2 Introduction 23 2.3 Materials and Methods 24 2.3.1 Mice and genotyping 24 2.3.2 Cre activity induction via tamoxifen administration 25 2.3.3 Western blot 25 2.3.4 Cartilage staining 25 2.3.5 Quantification of limb bud frontal area plane 26 2.3.6 Section immunofluorescence 26 2.3.7 Whole-mount in situ hybridization 26 2.3.8 RNA isolation and real-time quantitative PCR 27 2.3.9 Chromatin immunoprecipitation 27 2.3.10 Quantification of location of Shh domain in limb buds and percentage of limb bud frontal plane area expressing Gli1 27 2.4 Results 28 2.4.1 Irx3/5 are required prior to limb bud outgrowth 28 iv

2.4.2 Irx3/5 regulate AP prepattern to promote proximal and anterior hindlimb progenitors 29 2.4.3 Irx3 binds to Gli3 limb enhancer and regulates Gli3 expression in hindlimb buds 30 2.4.4 Early genetic interaction between Irx3/5 and Gli3 is essential for signaling center establishment 31 2.4.5 Irx3/5-dependent anterior progenitor population contributes to preaxial polydactyly 32 2.4.6 Requirement of Irx3/5 in forelimb AP pattern formation is revealed in Kif7-/- background 32 2.4.7 Forelimb bud displays lower Shh signaling activity than that of the hindlimb 33 2.4.8 Reducing Shh signaling rescues anterior skeletal formation in the Irx3/5-DKO hindlimb 33 2.5 Discussion 34 2.5.1 A biphasic model for limb anterior pattern formation 34 2.5.2 Irx3/5 are key determinants of anterior population in early hindlimb field 35

Chapter 3 Limb bud morphogenesis requires regulation of cell proliferation by Irx3/5 in the anterior hindlimb field 49 3 Chapter 3 50 3.1 Summary 50 3.2 Introduction 50 3.3 Materials and Methods 52 3.3.1 Mice 52 3.3.2 OPT and limb bud morphology analysis 52 3.3.3 Double-pulse chasing analysis and cell cycle time estimation 52 3.3.4 Flow cytometry 53 3.3.5 Live imaging 53 3.3.6 Orientation of cell division 54 3.4 Results 54 3.4.1 Irx3/5-DKO hindlimb buds are small and display abnormal shape 54 3.4.2 Prolonged cell cycle time in the anterior hindlimb field of Irx3/5-DKO embryos during limb initiation 55 3.4.3 Live imaging data revealed multiple cell division defects in initiating Irx3/5-DKO hindimb buds 56 3.5 Discussion 57

Chapter 4 Conclusion and Future Experiments 68 4 Chapter 4 68 4.1 Thesis summary 68 4.2 Irx3 and Irx5 are novel anterior limb determinants 69 4.2.1 Identifying the molecular mechanism of anterior specification by Irx3/5 70 4.2.2 Fate mapping of Irx3/5-expressing cells may reveal the origin of anterior limb structures 72 4.3 Genetic interaction between Irx3/5 and Gli3 is required for signaling center establishment and limb outgrowth 74 4.4 Early specification and progressive modulation of anterior limb elements by Irx3/5 and Shh signaling 76 v

4.5 Difference between forelimb and hindlimb development 77 4.6 Irx3/5 regulate limb bud morphogenesis in multiple aspects 78 4.7 Conclusion remarks 79

References 84

List of Tables

Table 3.1 Shape information of E10.0 forelimb buds 61 Table 3.2 Shape information of E10.0 and E10.5 hindlimb buds 61

vii

List of Figures

Figure 1.1 An overview of mouse limb bud outgrowth and AP pattern formation 18 Figure 1.2 Oriented cell behavior in limb bud morphogenesis 19 Figure 1.3 Iroquois homeobox genes 20 Figure 1.4 Expression of Irx3/5 during forelimb and hindlimb development 20 Figure 1.5 Irx3/5-DKO embryos display hindlimb specific phenotype 21 Figure 2.1 Irx3/5 are required prior to limb bud outgrowth for the formation of anterior-distal hindlimb elements 36 Figure 2.2 Morphogenesis, cell proliferation and cell death in Irx3/5-DKO hindlimb buds 37 Figure 2.3 Irx3/5 regulate hindlimb bud prepattern 38 Figure 2.4 Irx3/5-DKO hindlimb buds are primed to form Shh-responding elements 39 Figure 2.5 Irx3/5 regulate Gli3 expression in limb 40 Figure 2.6 Early genetic interaction between Irx3/5 and Gli3 is required for signaling center establishment and limb bud outgrowth 41 Figure 2.7 Irx3/5-dependent anterior progenitor population contributes to preaxial polydactyly 42 Figure 2.8 Irx3/5 are required for the formation of anterior forelimb elements in Kif7-/- background 42 Figure 2.9 Irx3/5 are involved in forelimb development 43 Figure 2.10 Shh domain is more posteriorly restricted in the forelimb bud than hindlimb 44 Figure 2.11 Forelimb bud displays lower Shh signaling activity than that of the hindlimb 45 Figure 2.12 Reducing Shh signaling in Irx3/5-DKO hindlimb buds rescues anterior pattern formation 46 Figure 2.13 Prepattern and limb bud size are not rescued in Irx3/5-DKO;Shh+/- hindlimbs 47 Figure 2.14 A biphasic model of limb anterior pattern formation 47 Figure 3.1 Surface renderings of E10.5 control and Irx3/5-DKO embryos viewed from the right 60 Figure 3.2 Dorsal and lateral views of forelimb bud isosurfaces at E10.0 61 Figure 3.3 Dorsal and lateral views of hindlimb bud isosurfaces at E10.0 and E10.5 62

viii

Figure 3.4 Anterior hindlimb mesenchymal cells of Irx3/5-DKO mutants have longer cell cycle time than those of controls during hindlimb initiation 63 Figure 3.5 Oriented cell division is altered in initiating Irx3/5-DKO hindlimb field 65 Figure 3.6 Anterior hindlimb mesenchymal cells exhibit chromosome segregation defect during mitosis in Irx3/5-DKO mutants during limb initiation 66 Figure 4.1 Overlapping and distinct functions of Irx3/5 and Gli3 in anterior pattern formation during limb development 80 Figure 4.2 Genetic fate mapping of Irx3/5-specified anterior population in limb development 81 Figure 4.3 AER-related defects in Irx3/5;Gli3-TKO hindlimb buds 82 Figure 4.4 Hand2 and Tbx2 expression in Irx3/5;Gli3-TKO hindlimb buds 83 Figure 4.5 Hhip expression level in the hindlimb is lower than that of the forelimb at comparable developmental stage 83 Figure 4.6 Clonal expansion analysis 83

List of Videos

Video 3.1 Rendered OPT of E10.5 control embryo Video 3.2 Rendered OPT of E10.5 Irx3/5-DKO embryo Video 3.3 Rendered OPT of E10.0 control hindlimb bud Video 3.4 Rendered OPT of E10.0 Irx3/5-DKO hindlimb bud Video 3.5 Rendered OPT of E10.5 control hindlimb bud Video 3.6 Rendered OPT of E10.5 Irx3/5-DKO hindlimb bud Video 3.7 Normal mesenchymal cell division in anterior hindlimb mesenchyme at E9.75 Video 3.8 Cell division with delayed chromosome segregation in Irx3/5-DKO anterior hindlimb field at E9.75 Video 3.9 Cell division with anaphase bridge in Irx3/5-DKO anterior hindlimb field at E9.75 (1) Video 3.10 Cell division with anaphase bridge in Irx3/5-DKO anterior hindlimb field at E9.75 (2) Video 3.11 Failed cell division in Irx3/5-DKO anterior hindlimb field at E9.75

List of Abbreviation

3D three dimentional 5' Hoxd 5'-located Hoxd gene AER apical ectodermal ridge AP anteroposterior cdk cyclin-dependent kinase cDNA complementary deoxyribonucleic acid CKO conditional knockout CNE conserved non-coding sequence element DEPC diethylpyrocarbonate DKO double knockout DNA deoxyribonucleic acid DV dorsoventral E embryonic day EDTA ethylene diamine tetraacetic acid EGTA ethylene glycol tetraacetic acid EtOH ethanol Fgf Fibroblast growth factor Ft Fused toes G0/G1/G2 Gap 0/1/2 phase GAPDH glyceraldehyde-3-phosphate dehydrogenase GliFL Gli full length GliR Gli repressor Irx Iroquois homeobox gene KO knockout LPM lateral plate mesoderm M-phase mitotic phase MeOH methanol OPT optical projection tomography PBS phosphate buffered saline PBT PBS with 0.1% Tween

PCR polymerase chain reaction PD proximodistal PFA Paraformaldehyde pH3 phospho-histone H3 qRT-PCR quantitative reverse-transcriptase PCR RA retinoic acid RNA ribonucleic acid s somite stage S-phase synthesis phase Shh Sonic hedgehog TM tamoxifen ZPA zone of polarizing activity ZRS ZPA regulatory sequence

xii 1

Chapter 1 Introduction 1 Introduction

The limb bud is an excellent model to understand molecular mechanisms of embryogenesis because it is 1) easily accessible, 2) highly self-regulative, 3) a three dimensional structure with clear polarity along the three axes: the anteroposterior (AP), dorsoventral (DV) and proximodistal (PD) axes, and 4) shares molecular mechanisms and pathways with development of other segmented structures (e.g. different Hox gene combinations to specify different segments) (Koussoulakos, 2004). Decades of studies have provided a wealth of information on the mechanism of limb development. By manipulating chick limb buds, in what have been called “cut and paste” experiments, early studies have defined group of cells controlling limb bud morphology. Many basic molecules and structures for limb development, which were the foundation of later genetic analysis, were discovered from these classical studies, such as the Fibroblast Growth Factors (Fgfs) from the apical ectodermal ridge (AER) and Sonic Hedgehog (Shh) expressed in the zone of polarizing activity (ZPA). With the advantage of mouse genetics, functions of particular genes and pathways in limb development and limb bud morphogenesis have been gradually revealed.

This chapter provides an overview of vertebrate limb development and current models on limb axes specification. Since my work was mainly focused on limb AP pattern formation, details will be provided with this regard. In addition, cell biology involved in limb bud morphogenesis will be reviewed. Finally, current knowledge on functions of Iroquois homeobox genes (Irx) in development will be discussed.

1.1 An overview of vertebrate limb development and pattern formation

1.1.1 Limb induction

The vertebrate limb comes from lateral plate mesoderm (LPM) in the flank of the embryo. The limb field is a group of embryonic cells that give rise to the limb bud. Long before appearance of the limb bud, the location of the limb field is already determined. The forelimb and hindlimb

2 fields are located at specific positions along the body axis under control of combinatorial Hox gene expression and other mechanisms (Burke et al., 1995; Cohn et al., 1997; McPherron et al., 1999; Marshall et al., 1996). Identities of forelimb and hindlimb are specified at this stage. The forelimb specific Tbx5 and hindlimb specific Pitx1 and Tbx4 are among the earliest markers detectable in the forelimb and hindlimb fields (Chapman et al., 1996; Gibson-Brown et al., 1998; Lanctôt et al., 1999) (Figure 1.1A). Current progress in the field suggests that while Tbx5 and Tbx4 are required for the initiation and outgrowth of forelimb and hindlimb buds respectively (Hasson et al., 2007; Naiche and Papaioannou, 2007), Pitx1 is the key determinant of the hindlimb identity as revealed in loss-of-function and gain-of-function studies (DeLaurier et al., 2006; Duboc and Logan, 2011; Marcil, 2003). Following limb field specification, canonical Wnt signaling is activated in forelimb and hindlimb field LPM, resulting in localised Fgf10 expression that is necessary for outgrowth (Min, et al., 1998; Sekine, et al., 1999; Agarwal, 2003; Kawakami et al., 2001). Meanwhile, cells in the LPM of the limb field continue to divide rapidly whereas division of flanking non-limb LPM cells slows, resulting in the formation of a noticeable limb bud (Searls and Janners, 1971).

1.1.2 AER and limb PD pattern formation

The limb bud is composed of an ectodermal pocket and inner proliferating mesenchyme. In the LPM of the limb field, Fgf10 and Bmp ligands signal to the overlying ectoderm to activate ectodermal Wnt/β-catenin signaling to initiate Fgf8 expression and AER formation (Ohuchi et al., 1997; Min et al. 1998; Sekine et al. 1999; Barrow et al., 2003; Soshnikova et al., 2003) (Figure 1.1.A). The AER is a thickened ectodermal structure that runs along the distal limb bud tip, separating the dorsal and ventral ectoderm marked by Wnt7a and En-1 expression respectively (Riddle et al., 1995; Davis et al., 1988). This structure is required to maintain limb bud elongation along the PD axis. In chick, beads soaked in Fgf2 or Fgf4 transplanted into the distal tip wwere able to restore outgrowth and patterning of a limb bud from which AER had been removed (Niswander et al., 1993; Fallon et al., 1994), indicating that Fgfs are the signaling cues mediating AER functions. In mouse, Fgfs secreted by the AER, including the sequentially expressed Fgf8, Fgf4, Fgf9 and Fgf17, are important for inhibiting apoptosis in the mesenchyme and promoting limb bud outgrowth (Lewandoski et al., 2000; Mariani et al., 2008; Sun et al., 2002).

It was believed that AER-Fgf signaling maintains cells in the distal-most limb bud mesenchyme, defined as the progress zone, in a proliferative and undetermined state. Cells along the P-D axis would be specified by a 'clock-type' mechanism as they exit the progress zone. Thus, the fate of cells is determined in a proximal-to-distal manner, and the function of AER-Fgfs is “permissive” (only required for maintenance of the progress zone) (Summerbell et al., 1973). This is known as the “progress zone” model which is challenged by recent studies, and a “two-signal” model has been proposed. Work from the Martin group provides genetic evidence supporting an “instructive” role of AER-Fgf signaling. They showed that AER-Fgfs specify PD cell fates (Mariani et al., 2008). Moreover, two recent studies using chick recombinant limb buds suggest that while Fgfs act as distal signals, retinoic acid (RA) acts as the “proximal signal” from the trunk (Cooper et al., 2011; Roselló-Díez et al., 2011). Thus, the antagonistic interaction and balance between RA and AER-Fgfs is critical for pattern formation along the PD limb axis to specify skeletal elements of the stylopod (e.g. humerus), zeugopod (e.g. radius and ulna) and autopod (e.g. wrist bones, palm bones and digit bones). However, the instructive role of RA in limb PD pattern formation is challenged by recent genetic studies in mouse which revealed that RA signaling is only required for forelimb initiation (Cunningham et al., 2011; Cunningham et al., 2013; Zhao et al., 2009). It is possible that chick limbs use RA whereas mouse limbs do not. Alternatively, RA may mimic or share redundancy with additional proximal cue(s) from the trunk.

1.1.3 ZPA and limb AP pattern formation

The ZPA is a group of cells in the limb bud posterior mesenchyme. These cells have the ability to induce mirror image duplication along the AP axis when transplanted to the anterior mesenchyme of an intact limb bud in chick (Saunders et al., 1968, Tickle, 1981). Shh was identified as a polarizing molecule secreted by the ZPA (Riddle et al., 1993). Activation of Shh in the ZPA is under tight regulation of a group of genes during limb initiation also known as the prepatterning stage (Davenport, 2003; Suzuki et al., 2004; te Welscher et al., 2002b; Zakany et al., 2004) (Figure 1.1C). A gradient of Shh is thought to be crucial for patterning along the AP axis, especially for specification of digit identities. A recent study discovered an unexpected mechanism of Shh ligand movement and reception, in which Shh particles are delivered from ZPA cells to responding cells by direct contact of extending filopodia over a long range (Sanders et al., 2013). In response to Shh signaling, downstream targets such as Gli1 and Ptc1 are

4 expressed, and the level of Gli3 repressor (Gli3R) is reduced (Jiang and Hui, 2008; Wang et al., 2000). Gli3R is a truncated form of Gli3 processed from the full length protein (Gli3FL) which functions as a transcriptional activator (Jiang and Hui, 2008). As a consequence of the Shh gradient, a counter gradient of Gli3R is established across the AP axis of the limb (Wang et al., 2000). Several studies suggest that an important function of graded Shh signaling in limb development is to counteract Gli3R-mediated repression and promote proliferation in the posterior mesenchyme thereby enabling the progression of limb outgrowth and AP pattern formation (Figure 1.1D). For example, in mice, Shh;Gli3 double knockout mutant limbs exhibit similar AP patterning defects to that of Gli3-/- and Gli3P1-4/P1-4 (no Gli3R) mutants but not Shh- null mice, supporting the notion that Gli3R is the major mediator of Shh signaling in limb AP pattern formation (Hui and Joyner, 1993; Litingtung et al., 2002; te Welscher et al., 2002b; Wang et al., 2007). Marker analysis also indicates that several genes regulating AP pattern formation including 5’ Hoxd genes (e.g. Hoxd11/12/13) are controlled by Gli3 rather than Shh (Litingtung et al., 2002; te Welscher et al., 2002b).

1.1.4 Self-regulated feedback loop between the AER and ZPA

The AER and ZPA are two major signaling centers in limb bud development. Their maintenance requires establishment of the Shh-Fgf mesenchymal-epithelial feedback loop (Figure 1.1B) (Niswander et al., 1994). Upon activation of Shh in ZPA, Gremlin 1 (Grem1), which antagonizes activity of Bmps, is upregulated in posterior limb bud mesenchyme, resulting in reduction of Bmp activity, which promotes expression of Fgf in the AER (Khokha et al., 2003; Zúñiga et al., 1999). On the other hand, high level of AER-Fgf signaling represses Grem1 expression. Thus, the positive Shh-Grem1-Fgf feedback loop drives limb bud outgrowth and upregulation of AER- Fgf signaling, which triggers the Fgf-Grem1 inhibitory loop. The inhibitory loop then operates to terminate outgrowth signals (Fgfs and Shh) (Verheyden and Sun, 2008). These studies suggest that limb bud outgrowth is regulated by a self-promoting and self-terminating circuit (Figure 1.1B).

1.1.5 Proliferative expansion, determination and differentiation

After the early patterning stage of limb bud development, coordinated proliferation along the AP and PD axes leads to expansion of progenitor pools that give rise to primordia of stylopod, zeugopod and autopod. Cell proliferation is fast and uniform at the early stage of limb

5 development, while cell death is not detectable in the limb bud (Fernandez-Teran et al., 2006). Increased cell death occurs in restricted areas during later stages. As limb bud outgrowth proceeds distally, Sox9 expression is initiated in the proximal core mesenchyme, and leads to chondrogenic differentiation (Takahashi et al., 1998). Cartilage structures appear first and develop in a proximal-to-distal sequence, and are replaced by bones at later stages.

1.2 Limb AP pattern formation

Limb AP pattern formation can be divided into two phases: an early phase polarizing the initiating limb bud along the AP axis prior to ZPA-Shh activation and a Shh-dependent phase during which identities of digits are specified (Figure 1.1C-D). In this section, I will review recent progress on limb AP pattern formation using mouse models with detailed evidence.

1.2.1 Prepatterning the limb bud and Shh-activation network

Prepatterning of the limb bud along the AP axis polarizes limb bud mesenchyme before Shh expression in the ZPA (Figure 1.1C) (Hamburger, 1938; Chiang et al., 2001; Ros et al., 1996). It is important for Shh activation in the appropriate region. Gli3 and Hand2 are the first genes shown to be involved in prepatterning (te Welscher et al., 2002a). At the prepatterning stage, Gli3 is activated in the anterior mesenchyme of the initiating limb bud (Masuya et al., 1997), pushing Hand2 expression to the posterior margin. It is proposed that the mutual genetic antagonism between these two genes polarizes the limb bud with initial AP polarity, and, together with other prepatterning genes (e.g. Tbx2/3, 5’ Hoxd genes, Etv4/5), activates and restricts ZPA-Shh in the posterior mesenchyme (Davenport, 2003; Mao et al., 2009; Suzuki et al., 2004; Zakany et al., 2004; Zhang et al., 2009) (Figure 1.1C).

Function of Gli3 in the anterior limb field is thought mainly to restrict Hand2 expression in the posterior margin. The key function of Hand2 in limb development is to induce Shh expression, as revealed by gain-of-function and loss-of-function studies (Charité et al., 2000; Fernandez- Teran et al., 2000). In agreement with this notion, Hand2 function is mainly required during prepatterning stage. Using a conditional knockout system, Galli et al. showed that when Hand2 is depleted before prepatterning, the mutant limb phenotype is similar to that of Shh-/- mutants as a consequence of failed Shh activation. In contrast, if Hand2 depletion occurs after Shh activation, mutant limb structures are generally normal (Galli et al., 2010). Aside from Hand2,

6 other genes have been suggested to induce Shh expression. Tbx2 and Tbx3 are expressed initially in the entire limb field and later downregulated in the anterior margin of the limb bud (Gibson- Brown et al., 1998; Tümpel et al., 2002). Tbx3-/- limb buds cannot activate Shh, and misexpression of Tbx2 can induce ectopic Shh expression (Davenport, 2003; Suzuki et al., 2004). 5’ Hoxd genes are expressed in posterior limb bud mesenchyme before Shh activation. Forced ectopic expression of these genes in the anterior limb mesenchyme during prepatterning stage leads to ectopic mirror-image expression pattern of Shh along limb A-P axis (Zakany et al., 2004). In addition, AER-Fgf signaling plays dual roles in regulating Shh. Besides their function in maintaining Shh expression during limb bud outgrowth, they also act at early stages to prevent inappropriate anterior expression of Shh. Recently, the function of two targets of AER-Fgf signaling, Etv4 and Etv5 belonging to the ETS family of transcription factors, in repressing Shh has been described (Mao et al., 2009; Zhang et al., 2009). Loss of Etv4/5 leads to ectopic activation of Shh in a small group of cells in the anterior limb bud mesenchyme resulting in preaxial polydactyly. These studies also suggest that Etv4/5 may function as transcriptional activators and suppress anterior Shh activation indirectly. It is established that Shh expression in the developing limb bud is regulated by a far upstream cis-regulatory element, called the ZPA regulatory sequence (ZRS) (Amano et al., 2009; Lettice, 2003; Sagai et al., 2005). And mutations in the ZRS that lead to ectopic Shh activation have been reported in preaxial polydactyly in human and other mammals (Hill, 2007). It has been shown that multiple factors mentioned above can directly associate with the ZRS (i.e. Hand2, Hoxd13 and Etv4/5), which provides molecular mechanisms by which prepatterning genes regulate Shh limb expression (Galli et al., 2010; Lettice et al., 2012).

Interestingly, the duration of prepatterning stage (from limb bud initiation to Shh activation) is about 12-18hr (E9.0-9.75 in forelimb development and E9.75-10.25 in hindlimb development). What else is happening in the limb bud during this period of time besides preparing for Shh activation? Does Gli3 have functions other than restricting Hand2 expression posteriorly during limb initiation? What are upstream regulators that promote Gli3 expression in the anterior limb field? These questions remain unexplored.

1.2.2 Limb AP pattern formation after Shh activation

Anterior and posterior limb elements have distinct features. For example, digit 1 in the anterior autopod is the only biphalangeal digit, whereas posterior digits (2 to 5) have three phalanges. Moreover, digit 1 can move in a bigger range than the posterior digits. The structural differences between the anterior and posterior elements suggest that the underlying patterning mechanisms may be distinct.

It is well characterized that formation of the posterior elements (i.e. digit 2-5 and ulna/fibula) are largely dependent on Shh signaling from the ZPA. In the absence of Shh, posterior elements do not form (Chiang et al., 2001; Kraus et al., 2001). Genetic fate mapping studies revealed that these elements (digit 5, 4, 3, part of digit 2 and ulna/fibula) are derived from Shh-responding cells (descendants of Gli1-expressing cells) during limb development (Ahn and Joyner, 2004). Harfe et al. showed that digit 5, 4, the posterior half of digit 3 and one third of the ulna/fibula are descendants of Shh-expressing cells (Figure 1.1E). They also provided evidence that cells expressing Shh for longer time progressively contribute to more posterior digits, suggesting that a temporal gradient of morphogen exposure may be involved in specifying identities of posterior digits. They speculate that while digit 2 requires paracrine Shh signaling, digit 3 to 5 are specified by autocrine Shh signaling in a concentration and time dependent mannar (Harfe et al., 2004). However, work from Ahn and Joyner suggests that shortly after Shh activation, although Shh is still highly expressed, pathway activity is greatly downregulated (reflected by Gli1 level) in ZPA cells which are shown to give rise to digit 5. Interestingly, even though Shh pathway activity is greatly decreased in Gli2-/- limb buds, digit pattern formation is unaffected (Ahn and Joyner, 2004). These data are inconsistent with the temporal gradient model. Moreover, a recent study from Zhu et al. provided further evidence that challenged the temporal gradient model. They used the tamoxifen inducible Cre-loxP system to remove Shh at different stages of early limb development. When the duration of Shh expression was reduced, digits were gradually lost in the order of 3524, which is the reverse order of digit condensation. These surprising observations cannot be explained by the temporal gradient model. To explain their findings, the authors proposed a biphasic model in which Shh signaling specifies digit identities in an early and transient phase (within ~12hr post Shh activation) while being continuously required for the expansion of digit primordia to obtain sufficient precursor cells for digit condensation (Zhu et al., 2008). The positional information specified by the early transient function of Shh signaling may

8 be inherited by secondary signals and factors, such as Bmp signaling and 5’Hoxd genes (Suzuki, 2013; Zakany and Duboule, 2007). It is unclear why condensation of posterior digits happens in a specific order (i.e. 4253). Interestingly, it seems that digit 5 and 3 branch from the base of digit 4 and 2 respectively during digit condensation, whereas digit 4 and 2 are likely extended from condensations of ulna/fibula and radius/tibia respectively (Zhu et al., 2008).

Mechanism of pattern formation of the anterior structures during limb development is less understood. A recent quantitative study of Hoxd gene expression during digit development showed that the morphology of digit 1 relies on a reduced dose of 5’Hoxd gene products (Montavon et al., 2008). However, what triggers this late digit determination event is unclear. It is known that the anterior structures (i.e. digit 1 and radius/tibia) are not derived from cells receiving Shh signaling and can form in the absence of Shh. Thus, together with the proximal element (i.e. humerus/femur) they are considered “Shh-independent” (McGlinn and Tabin, 2006). One key function of Shh signaling during limb development is to downregulate Gli3R level in the posterior margin to counteract the repression effect of Gli3R on posterior gene expression (te Welscher et al., 2002b; Wang et al., 2000). This is supported by recent work from Cao et al., who showed that in mouse limb buds expressing only Gli3R (Gli3Δ701/ Δ701), although Shh is expressed, posterior patterning genes promoted by Shh signaling cannot be activated (Cao et al., 2013). Distinct from the posterior margin, the anterior part of limb bud is enriched with Gli3R, leading to expression of a group of distinct genes. These include Alx4 that has been shown to suppress Shh activation in the anterior limb mesenchyme (Qu et al., 1997). Pax9 expression coincides with the region giving rise to digit 1 and radius/tibia, and loss of Pax9 in mice leads to digit 1 duplication (Peters et al., 1998). In Gli3-/- limb buds, Alx4 and Pax9 expression are lost (McGlinn et al., 2005; te Welscher et al., 2002a). A recent genome-wide study on Gli3 binding regions suggests that Gli3 directly regulates Alx4 and Pax9 expression in the anterior limb bud mesenchyme and suppresses posterior gene expression including Gli1, Ptc1, Hand2, Tbx2 and 5’ Hoxd genes (Vokes et al., 2008). It is speculated that high level of Gli3R in the anterior limb bud may be involved in specifying digit 1 identity (Hill et al., 2009; Hui and Joyner, 1993; Litingtung et al., 2002; Wang et al., 2007). However, Gli3 is also required to constrain digit number (Aoto et al., 2002; Hui and Joyner, 1993; Litingtung et al., 2002), which makes it difficult to dissect the molecular mechanism of its function in anterior skeletal specification. Therefore, the origin of anterior structures remains uncertain.

1.3 Cell biology of limb bud morphogenesis

The basis of tissue morphogenesis is regulated cell behaviour. During limb initiation, cells in the LPM of the limb field maintain rapid cell division, possibly under the influence of multiple mitogenic signals (e.g. Wnt and Fgf signaling) (Searls and Janners, 1971). Cell proliferation and cell death are important for limb bud morphogenesis (Boehm et al., 2010; Fernandez-Teran et al., 2006). Recent studies utilizing advanced microscopy techniques revealed directional cell movement and oriented cell behavior also play a critical role in shaping the limb bud (Boehm et al., 2010; Gros et al., 2010; Wyngaarden et al., 2010). In addition, cell-cell interaction and extracellular environment have also been implicated in limb pattern formation (Schaller et al., 2001). In this section, I’ll mainly focus on reviewing recent progress on mechanisms of cell proliferation, cell death and oriented cell behavior in limb bud morphogenesis.

1.3.1 Cell proliferation and cell death

Shortly after limb bud initiation, cell proliferation is uniform in the limb bud mesenchyme (Fernandez-Teran et al., 2006). However, during limb bud outgrowth, a gradient of proliferation rate becomes apparent (low in proximal and high in distal region underlying the AER). It is thought that this “proliferation gradient” along the PD limb axis results in limb bud elongation (Ede and Law, 1969), which is supported by series of studies in both mouse and chick limb development, as observed by mitotic index counting (Hornbruch and Wolpert, 1970), BrdU incorporation (Dudley et al., 2002), [3H]thymidine labelling (Reiter and Solursh, 1982), and phospho-Histone H3 staining (Fernandez-Teran et al., 2006). However, oriented movement and cell division underlie limb bud shape rather than spatial control of cell proliferation (Gros et al., 2010; Wyngaarden et al., 2010). In addition to cell proliferation, regional cell death is also important in limb morphogenesis. In the beginning, no obvious cell death can be detected in early limb buds. During limb bud elongation, cell death is detected in certain regions. Cell death in the center of limb bud is thought to participate in the separation of the two cartilage elements of the zeugopod, while the anterior cell death in the future autopod is involved in digit 1 formation. At later stage during digit differentiation, programmed cell death in interdigital regions is necessary for digit separation (Fernandez-Teran et al., 2006).

Fgf, Wnt and Shh signaling pathways have been shown to have mitogenic functions and regulate the cell cycle in other systems. Mutant analyses revealed that disrupting these signaling pathways

10 also causes limb outgrowth defects. For example, mutants lacking combination of AER-Fgf signaling have small and short limb bud and display increased apoptosis (Lewandoski et al., 2000; Mariani et al., 2008; Sun et al., 2002). Also, Wnt and AER-Fgf signaling act in synergy to promote cell proliferation and maintain cells in the undifferentiated state. Once limb mesenchymal cells are out of the range of Wnt and AER-Fgf signaling due to limb bud growth, they start to express Sox9 and undergo chondrogenesis (ten Berge et al., 2008). Shh-null mice also exhibit increased cell death, decreased cell proliferation and eventually loss of posterior limb elements (Chiang et al., 2001; Zhu et al., 2008). In the anterior limb bud mesenchyme, Gli3R antagonizes Shh signaling to inhibit cell proliferation. Lopez-Rios et al. found that Gli3R directly inhibits the expression of Cdk6 in the anterior region of the future autopod to suppress G1-S transition of the cell cycle and promote differentiation and chondrogenesis (Lopez-Rios et al., 2012). Interestingly, this anterior region controlled by Gli3R coincides with the anterior- distal apoptotic region. In agreement with this, in Gli3-deficient mice, cell proliferation is increased, while cell death is decreased in anterior limb bud mesenchyme, resulting in overgrowth of the anterior population giving rise to preaxial polydactyly (Aoto et al., 2002).

1.3.2 Oriented cell behavior

Although the “proliferation gradient” model can explain limb bud elongation to some extent, when applied to quantitative analysis, it turns out not to be that simple. Recent studies indicate that in addition to proliferation rate and programmed cell death, orientated cell behavior provides another major force for limb bud elongation and morphogenesis. Utilizing live imaging technique, Wyngaarden et al. showed that during limb initiation, cells from adjacent LPM elongate and move towards the limb field along the rostrocaudal direction, whereas cells in the limb field display more isotropic shape and move laterally. Similarly, cell divisions in initiating limb buds mainly occur along the PD limb axis, which is different from that of the adjacent LPM (Gros et al., 2010; Wyngaarden et al., 2010) (Figure 1.2A). At later stages during limb bud elongation, cells in the peripheral mesenchyme of the limb bud are elongated and divide towards the overlying ectoderm, while cells in the central regions display less biased elongation and directional cell division (Boehm et al., 2010; Gros et al., 2010) (Figure 1.2B). With live imaging and trajectory analysis tracking cell movement, Gros et al. revealed that during limb bud elongation, limb mesenchymal cells in general divide and move distally. They found a gradient in cell velocity, with cells located distally moving faster along the PD axis than cells located

11 more proximally (Figure 1.2B). Moreover, they suggested that after cell division, daughter cells reintercalate in a way to ensure limb bud elongation along the PD axis (Gros et al., 2010). Another feature of active cell movement is the presence of lamellipodia and filopodia among most limb bud mesenchymal cells (Boehm et al., 2010; Gros et al., 2010). Together, these studies indicate that limb mesenchymal cells undergo constant rearrangement through highly organized movements.

Molecular mechanisms underlying the oriented cell behavior and organized cell movement remain to be addressed. It was thought that the AER-Fgf signaling functions as chemoattractant in limb morphogenesis (Li and Muneoka, 1999). However, other studies suggest that function of AER-Fgf signaling is to enhance cell mobility rather than directing cell migration, thus cell movement is more active in distal limb bud mesenchyme relative to the proximal region (Gros et al., 2010; Wyngaarden et al., 2010). Instead, Wnt signaling seems to be the chemoattractant mediating directional cell movement. Multiple pieces of evidence (i.e. bead implantation, chemical drug treatment and mutant analysis) suggest that Wnt5a, which is expressed in distal limb bud, and probably other noncanonical Wnt ligands signal through the Jnk pathway to regulate oriented cell behavior (Gros et al., 2010; Wyngaarden et al., 2010). Thus it is likely that coordinated influence of Fgf signalling on cell mobility, together with the directional cue provided by noncanonical Wnt signalling, drives distal outgrowth of the limb bud (Figure 1.2B).

1.4 Iroquois homeobox (Irx) genes

1.4.1 Discovery and genomic organization

The Iroquois homeobox (Irx) genes were first discovered in Drosophila from a screen for genes involved in sensory organ development (Dambly-Chaudière. and Leyns, 1992; Gomez-Skarmeta et al., 1996; Leyns et al., 1996). The first mutant identified showed a bristle phenotype in the notum resembling the hairstyle of Iroquois American Indians - hence the name of the locus. Three Irx genes were identified within this locus: araucan (ara), caupolican (caup) and mirror (mirr) (Gomez-Skarmeta et al., 1996; McNeill et al., 1997) (Figure 1.3B). They encode highly related transcription factors, representing a new family of the TALE (Three Amino acid Loop Extension) superclass of homeodomain proteins. In addition, the Irx proteins also share a unique structure in their C-terminal region, the Iro-box (ib), consisting of thirteen conserved amino acids (Figure 1.3A).

While in C. elegans there is only one Irx gene, in mammals there are six genes organized into two clusters (IrxA and IrxB clusters) within the genome (Figure 1.3C). This is likely due to duplication of the gene cluster during evolution (Cavodeassi et al., 2001; Gómez-Skarmeta and Modolell, 2002). The IrxA cluster, containing Irx1, Irx2 and Irx4, is located on chromosome 13 in mouse or chromosome 5 in human, and the IrxB cluster, containing Irx3, Irx5 and Irx6, is positioned on mouse chromosome 8 or human chromosome 16 (Bosse et al., 2000; Gómez- Skarmeta and Modolell, 2002; Peters et al., 2000). Sequence analysis suggested that Irx1, Irx2 and Irx4 are similar to Irx3, Irx5 and Irx6, respectively. Paralogs of the family are transcribed in the same orientation (Irx1/Irx3, Irx2/Irx5 and Irx4/Irx6), with Irx1 and Irx3 in the opposite direction relative to the rest of the Irx family. Although paralogs (Irx1/Irx3 and Irx2/Irx5) are expressed identically in some tissues, in general, Irx1 expression resembles Irx2 while Irx3 and Irx5 share a similar expression pattern (Cohen et al., 2000; Houweling et al., 2001). This phenomenon is probably due to the presence of shared regulatory elements that control their expression (Cavodeassi et al., 2001; de la Calle-Mustienes et al., 2005; Tena et al., 2011).

1.4.2 Functions of Irx genes in development

Irx family members may form complexes with other transcription factors and regulate gene expression. Efforts have been put to identify conserved DNA binding site of Irx. In Drosophila, Mirr can form homodimers and heterodimers with Ara, and its binding site is identified as “ACAnnTGT” which can also be recognized by mouse Irx4 (Bilioni et al., 2005). It has been shown that Irx3 and Irx5 can bind to this sequence in the Bmp10 promoter during mouse heart development (Gaborit et al., 2012). However, other studies have found different binding sequences of Irx. By analyzing 84 homeodomains from Drosophila using a bacterial one-hybrid system, Noyes et al. identified “taACA” as the Irx binding site (Noyes et al., 2008). Another in vitro study characterized 168 mouse homeodomains using protein binding microarrays and identified “tACATGTa” as the Irx binding site (Berger et al., 2008). It is not clear whether the difference of Irx binding sequence from these studies is due to technique limitations or reflects a difference of Irx binding preference in different species/tissues or both. More studies need to be carried out to confirm these binding sites in vivo and determine their physiological significance.

During early embryogenesis, Irx genes are expressed in large domains to specify large territories. At later stages during organogenesis, they are expressed in more restricted regions to generate

13 more refined patterns (Cavodeassi et al., 2001). Studies using multiple model organisms revealed that Irx family members interact with key signaling pathways and function as transcriptional activators or repressors in development and pattern formation of tissues and organs. Recently, Bonnard et al. reported recessive point mutations in human Irx5 causing congenital disorder affecting multiple organs and tissues (i.e. face, heart, brain and blood, etc.), highlighting the significance of Irx genes in human development (Bonnard et al., 2012).

Irx genes are key players in development of the nervous system. In Drosophila neurogenesis, Ara and Caup activate expression of achaete-scute genes in presumptive lateral notum for the formation of lateral bristles of the external sensory organs (Gomez-Skarmeta et al., 1996). In Drosophila imaginal eye disc, Irx genes are expressed in the dorsal region to restrict fng expression ventrally to define the DV organizer (McNeill et al., 1997; Cavodeassi et al., 1999; Yang et al., 1999). Irx genes are also involved in vertebrate eye development by regulating retinal interneuron subtype identity (Cheng et al., 2005; Star et al., 2012). During gastrulation in Xenopus, Irx1 plays a key role in neural plate specification. It is activated by Wnt signaling and directly inhibits Bmp4 expression to define the territory of the neural ectoderm (Gómez- Skarmeta et al., 2001). At late gastrula stage, Irx1 indirectly induces expression of Gbx2 and Fgf8 for midbrain-hindbrain boundary formation (Glavic et al., 2002). In chick brain development, Irx2 activity is modulated by Fgf8a signaling and functions as a transcriptional activator in cerebellum specification (Matsumoto et al., 2004), and the repressor function of Irx3 mediates competence for thalamus induction in the forebrain (Robertshaw et al., 2013). In vertebrate ventral neural tube pattern formation, Irx3 is involved in neuronal subtype specification. It belongs to the Class I genes which are negatively regulated by Shh signaling. Combined expression of Irx3 and Nkx6.1 of the Class II genes (promoted by Shh signaling) specifies ventral 2 (V2) neurons (Briscoe et al., 2000). Studies in chick and mouse established important functions of Irx genes in heart development and homeostasis (Kim et al., 2012). In chick embryos, Irx4 displays a ventricle-restricted pattern to establish chamber-specific gene expression in the developing heart possibly through an activator function (Bao et al., 1999). It interacts with the Vitamin D receptor and retinoic acid receptor to repress expression of slow MyHC3 in cardiac ventricles (Wang et al., 2001). During mouse heart development, Irx3 and Irx5 directly suppress Bmp10 expression in the endocardium, which is required for ventricular septation (Gaborit et al., 2012). In adult heart, Irx5 represses expression of Kcnd2 (encoding

Kv4.2) in endocardium to ensure coordinated cardiac repolarization (Costantini et al., 2005), and Irx3 directly represses expression of the gap junction gene Cx43 and indirectly activates Cx40 transcription for efficient conduction in the adult heart (Zhang et al., 2011). Irx genes also function as prepattern factors during nephrogenesis. In Xenopus kidney development, RA signaling activates Irx genes in the pronephros to specify pronephric territory and define its size through their activator function, and they are subsequently required to specify intermediate tubules progenitors (Alarcón et al., 2008; Reggiani et al., 2007).

In summary, previous work, especially gain-of-function and loss-of-function studies in chick and Xenopus embryos, established that Irx transcription factors interact with multiple key signaling pathways (i.e. RA, Shh, Fgf, Wnt and Bmp) and regulate pattern formation and cell fate specification during embryogenesis and organogenesis. However, because of the redundancy of Irx genes and their genomic organization, it is difficult to study their in vivo functions using mouse genetics. Fused toes (Ft) mutation generated from transgenic insertional mutagenesis in mouse is caused by a 1.6-Mb deletion on chromosome 8 including the entire Iroquois B cluster, and Ft/Ft mutant embryos display severe defects in craniofacial structures, heart, spinal cord and limb development, suggesting potential functions of Irx B genes in development of these tissues and organs (Grotewold and Rüther, 2002; Peters et al., 2002). Nonetheless, the genomic deletion of Ft mutation also includes three other genes (Fto, Ftm and Fts), and specific functions of Irx B genes remain to be addressed.

1.4.3 Irx3 and Irx5 have novel functions in mouse hindlimb development

In our lab, mouse Irx5 has been identified and cloned (Cohen et al., 2000). Previous work from our lab and others suggests that the expression pattern of Irx5 is similar to that of Irx3 during early embryogenesis (Cohen et al., 2000; Houweling et al., 2001). For example, at E10.5, both genes are expressed in the central nervous system, heart, foregut, LPM and limb bud. Prior to limb bud outgrowth, expression of Irx3 and 5 is observed throughout the anterior part of both forelimb and hindlimb fields. During limb bud outgrowth, their expression becomes restricted to the anterior-proximal region, and this expression pattern persists throughout later stages (Figure 1.4). Their overlapping expression in developing limb buds suggests that they may have redundant functions in limb development.

Previous work from our lab revealed that inactivation of Irx3 or Irx5 results in viable and fertile animals that show mild defects in their heart and eye, with no observable limb patterning defects. Bone staining of E18.5 Irx3-/- embryos indicates that these mutants have slight defects in the scapula and pelvis. To examine whether Irx3 and Irx5 have redundant functions in limb development, an Irx3/5 double knockout mouse line (Irx3/5-DKO) was generated by our lab. Heterozygous animals are viable and fertile, but homozygous mutants die around E14.5 with defects in multiple organs. Compound mutant mouse embryos retaining at least one allele of Irx3 or Irx5 (Irx3+/-5-/- and Irx3-/-5+/-) exhibited scapular and pelvic deficiency and humeral/femoral hypoplasia as revealed by bone staining at E18.5, but no distal limb pattern defects (data not shown). However, in addition to markedly hypoplastic scapulae and pelves, Irx3/5-DKO mutant embryos exhibited a hindlimb specific phenotype including loss of the tibia and digit 1 (digits 1 and 2 in 50%) and severe femoral hypoplasia, as revealed by cartilage staining at E14.5 (Figure 1.4A). Consistent with this skeletal loss, early skeletal condensations marked by Sox9 (Bi et al., 1999) were absent in the region of the tibia and digit 1 at E12.5 (Figure 1.4A). At E11.5, Irx3/5- DKO hindlimb buds are much smaller and narrower along the AP limb axis than those of wild- type littermates. Preliminary marker analysis suggests that Irx3/5-DKO mutants lack anterior hindlimb mesenchyme during early stages of limb development. Pax9 expression, which marks the primordium of tibia and digit 1 (Peters et al., 1998), is lost in E11.5 mutant hindlimb buds (Figure 1.4B, by Dr. Rui Sakuma). In addition, expression of Hoxd13, which marks the territory of posterior digits 2 to 5 at E11.5, was expanded anteriorly (Figure 1.4B, by Dr. Rui Sakuma). Therefore, Irx3/5 are required for the formation of anterior skeletal structures prior to mesenchymal condensation in the mouse hindlimb.

As Irx3/5-DKO mutants cannot survive after E14.5, to study their limb phenotype at later stages, conditional mutants (Prx1Cre; Irx3flox5EGFP/Irx3flox5EGFP) have been generated. In these mutants, Irx3 is specifically knocked out in limb bud mesenchyme by the Prx1Cre (Logan et al., 2002) in an Irx5-null background using the Cre-loxP system (Nagy, 2000). Strikingly, bone staining in these conditional mutants at E18.5 reveals no obvious patterning defects in their hindlimbs (Rong Mo, unpublished data). Since Prx1Cre is mainly activated in hindlimb buds around E10.0 after hindlimb bud initiation (Logan et al., 2002), we conclude that the functional requirement of Irx3/5 in hindlimb development is early, prior to Prx1Cre-mediated deletion.

1.5 Thesis rational and outline

Mutant analysis and fate mapping studies have established that Shh signaling plays a central role in the pattern formation and development of posterior limb skeletal elements (i.e. digit 2-5 and ulna/fibula). In contrast, the origin of the anterior and proximal limb structures (i.e. digit 1, radius/tibia and humerus/femur), the so called “Shh-independent” elements, is unclear. The molecular mechanism underlying their specification remains to be addressed. Gli3 is implicated in anterior pattern formation. However, the polydactlous phenotype in Gli3-deficient mutant limbs makes it difficult to dissect its function in anterior specification. Our lab generated the Irx3/5-DKO mouse model and found that although Irx3/5 are expressed in both forelimb and hindlimb buds in similar patterns, embryos lacking Irx3/5 only exhibit severe defects in the hindlimb. In mutant hindlimbs, anterior distal elements (digit 1 and tibia, and digit 2 in 50%) are completely lost, and the proximal element (femur) is highly hypoplastic. Interestingly, all affected elements are thought to be “Shh-independent”. Marker analysis suggested that prior to cartilage condensation, the anterior mesenchyme is already lost in Irx3/5-DKO hindlimb buds which are much smaller than controls. Thus, we developed the hypothesis that Irx3/5 are required to specify the progenitor cells that give rise to the “Shh-independent” limb structures.

Overall my graduate study aims to address the following questions: 1) What is the mechanism underlying the anterior specification function of Irx3/5 in limb development? 2) Why does loss of Irx3/5 result in a hindlimb specific phenotype while they are expressed in both forelimb and hindlimb buds? 3) What makes Irx3/5-DKO hindlimb buds small with abnormal shape?

In Chapter 2, I investigated the roles of Irx3/5 in the early specification of anterior limb population. Using a tamoxifen inducible conditional knockout system, I showed that Irx3/5 are required early during limb initiation to specify the anterior progenitor population that gives rise to the proximal and anterior distal elements in the hindlimb in wild-type and preaxial polydactyly backgrounds. I found that Irx3/5 regulate limb AP prepattern by promoting the anterior expression of Gli3. Specifically, we performed ChIP-PCR analysis and showed that Irx3 can bind to the limb enhancer of Gli3, which provides potential molecular mechanism of Irx3/5 regulating limb AP polarity. Strikingly, we discovered an early genetic interaction between Irx3/5 and Gli3, which is necessary for establishment of signaling centers and limb bud outgrowth. Interestingly, we found that forelimb and hindlimb buds experience different levels

17 and extents of Shh signaling, and that the forelimb bud has a larger Shh-free zone in anterior mesenchyme compared to the hindlimb bud at an equivalent stage. This anterior Shh-free envirionment seems to be important for the formation of anterior distal limb elements (i.e. digit 1 and radius/tibia). In the Kif7-/- background with high Shh pathway activity, removing Irx3/5 results in activation of Shh signaling throughout distal limb bud mesenchyme and loss of anterior distal structures in the forelimb. By reducing the dosage of Shh, the anterior Shh-free zone reappears and formation of the anterior distal structures is restored in Irx3/5-DKO hindlimbs. These data suggest that anterior limb pattern formation is regulated in two phases: an Irx3/5- dependent early specification phase and a Shh-inhibited progressive modulation phase.

Given the small and abnormal shape of Irx3/5-DKO hindlimb buds, I focused on understanding the roles of Irx3/5 in limb bud morphogenesis, which is described in Chapter 3. Using OPT imaging technique, we were able to generate limb bud 3D isosurfaces and quantify limb bud shape and size. Our data suggest that the morphological defect of Irx3/5-DKO hindlimb buds arises early during limb initiation and persists at later stages. To understand cellular processes underlying the morphological defects of Irx3/5-DKO hindlimb buds, I performed double-pulse chasing experiments to estimate cell cycle time in the hindlimb field during limb initiation. I found that the cell cycle time in the mutant anterior hindlimb field is about 15% longer than that of controls during limb initiation. In addition, using live imaging technique, I showed that mesenchymal cells in the anterior hindlimb field of Irx3/5-DKO mutants displayed multiple mitotic defects, including altered cell division orientation and chromosome segregation defects (i.e. anaphase bridge). Thus, it is likely that without Irx3/5, the anterior population of the hindlimb field cannot proliferate and expand properly, resulting in defects of limb bud morphogenesis and eventually loss of anterior skeletal elements. This work provides potential mechanism at the level of cell biology underlying the early anterior specification function of Irx3/5 during limb development.

Chapter 4 provides a summary of my findings and how it contributes to our current understanding of limb AP pattern formation. In addition, outstanding questions from the studies in Chapter 2 and 3 as well as critical experiments aimed to address these lingering questions will be discussed.

Figure 1.1 | An overview of mouse limb bud outgrowth and AP pattern formation

(A) During limb induction, Fgf10 is activated by Tbx5 in the forelimb and Pitx1 and Tbx4 in the hindlimb field (yellow). Fgf10 and Bmp signaling to the overlying ectoderm triggers canonical Wnt signaling and Fgf8 expression for the formation of AER (green). (B) The self-promoting and self-terminating circuit regulate limb bud outgrowth. This system involves two feedback loops: the Shh-Grem1-Fgf positive feedback loop (blue) and the Fgf-Grem1 inhibitory loop (red). (C) The mutual antagonism between Gli3 and Hand2 prepattern the initiating limb bud along the AP axis. Hand2 and other posterior prepatterning genes (e.g. Tbx and 5’Hoxd genes) are required to activate Shh expression at later stages. (D) After Shh activation in the ZPA (blue ellipse), graded Shh signaling (blue triangle) is required to generate a countergradient of Gli3R

(red triangle) for the expression of posterior genes (e.g. Gli1, light blue) in the posterior aspect of the limb bud. (E) Genetic fate mapping studies suggest that posterior elements (digit 2-5 and the ulna) are derived from Shh-responding cells (descendants of Gli1-expressing cells, light blue), among which digit 5, 4 and half of digit 3 and one third of the ulna are from Shh-expressing cells (blue). Condensation of posterior digits happens in the order of 4253.

Figure 1.2 | Oriented cell behavior in limb bud morphogenesis (modified from Hopyan et al. 2011)

(A) Limb initiation from the lateral plate mesoderm involves the loss of longitudinal cell shapes, directional changes in cell movement (blue arrows), and biased cell division along the PD axis (directions of cell division are highlighted with red arrows). (B) During elongation of the bud, influenced by non-canonical Wnt signaling from the overlying ectoderm, cells are aligned with their long axes and processes in a radial manner. Cell division planes and cell movements are largely parallel to this orientation (red arrows and blue arrows). Lamellipodia and filopodia are highly present suggesting active cell movement. Fgf signaling from the AER enhances mobility of cells in the distal region (big blue arrows).

Figure 1.3 | Iroquois homeobox genes

(A) Schematic illustration of Iroquois protein structure. A highly conserved TALE class homeodomain (HD) is present close to the N-terminus, and the Iro-box (ib) is at the C-terminal region. (B) Genomic organization of Drosophila (Dm) Irx genes (ara, caup and mirr) forming a cluster called Iroquois complex (Iro-C). (C) Genomic organization of Irx genes in mouse (Mm) and human (Hs) forming in two clusters: the IrxA cluster (Irx1, Irx2, and Irx4) and the IrxB (Irx3, Irx5, and Irx6). The orientation and distance of the arrows depict the direction of the gene and a brief estimation of chromosomal location of Irx genes, respectively.

Figure 1.4 | Expression of Irx3/5 during forelimb and hindlimb development

Whole-mount in situ hybridization of Irx3 and Irx5 in developing forelimb and hindlimb buds at indicated stages.

Figure 1.5 | Irx3/5-DKO embryos display hindlimb specific phenotype

(A) Cartilage staining of E14.5 forelimb and hindlimbs and Sox9 in situ hybridization in E12.5 hindlimb buds of control and Irx3/5-DKO embryos. Red label marks missing or hypolastic elements. (B) In situ hybridization of Pax9 and Hoxd13 in E11.5 limb buds (data of Dr. Rui Sakuma). Black arrowheads mark the anterior extent of Hoxd13 expression. Red arrowheads highlights loss of Pax9 expression and anterior expansion of Hoxd13 expression in Irx3/5-DKO hindlimb buds. Fe, femur; Fi, fibula; Hu, humerus; Pe, pelvis; Ra, radius; Sc, scapula; Ti, tibia; Ul, ulna; 1-5, digit numbers from anterior to posterior.

Chapter 2 Irx3/5 interact with Shh signaling for anterior limb pattern formation

A version of this chapter has been published in Developmental Cell (Li, D., Sakuma, R., Vakili. N.A., Mo R, Puviindran, V., Deimling, S., Zhang, X., Hopyan, S., Hui, C.C. (2014). Formation of proximal and anterior limb skeleton requires early function of Irx3 and Irx5 and is negatively regulated by Shh signaling. Developmental cell 29, 233-40). Vijitha Puviindran generated Irx3 antibody and performed chromatin immunoprecipitation assay. Rong Mo performed cartilage staining in Irx3/5;Gli3-TKO and Irx3/5;Kif7-TKO mutants and Col2a1 in situ hybridization in Irx3/5;Kif7-TKO mutants.

2 Chapter 2 2.1 Summary

Limb skeletal pattern heavily relies on graded Sonic hedgehog (Shh) signaling. As a morphogen and growth cue, Shh regulates identities of posterior limb elements including the ulna/fibula and digits 2 through 5. In contrast, proximal and anterior structures including the humerus/femur, radius/tibia and digit 1 are regarded as Shh-independent and the mechanisms governing their specification are unclear. Here I show that patterning of the proximal and anterior limb skeleton involves an early specification phase dependent on two novel determinants, Irx3 and Irx5 (Irx3/5). They regulate expression of the key anterior prepattern gene, Gli3. They also genetically interact with Gli3 to establish signaling centers to promote limb bud outgrowth. The early specification phase is followed by a late modulation phase during which Shh signaling negatively regulates these anterior elements. Irx3/5 function is essential in the initiating limb bud to specify the anterior progenitors that give rise to the femur, tibia and digit 1. However, these skeletal elements can be restored in Irx3/5 null embryos when Shh signaling is diminished. These data provide genetic evidence supporting the concept of early specification and progressive determination of anterior limb skeletal elements.

2.2 Introduction

The pattern of skeletal condensations in the embryonic limb is derived from interaction between multiple signals (Zeller et al., 2009). Shh, a prominent regulator of skeletal pattern, is expressed by a group of cells in the posterior aspect of limb bud mesenchyme called the zone of polarizing activity (ZPA) (Riddle et al., 1993). It is a key signal for establishing the posterior aspect of the limb skeleton that includes the ulna/fibula and digits 2 through 5 (Bastida and Ros, 2008; Hill, 2007; Robert and Lallemand, 2006). In Shh-null mice, most of the femur, the tibia and digit 1 remain (Chiang et al., 2001; Chiang et al., 1996). It has been proposed that Shh functions as a morphogen by diffusing from posterior to anterior to generate a signaling gradient that regulates digit identities along the anteroposterior (AP) limb axis (Hill, 2007). Genetic fate mapping studies in mouse indicate that digits 5 and 4 as well as part of digit 3 are derived directly from cells that once expressed Shh, while digit 2 and part of digit 3 are descendants of Gli1-expressing cells, which respond to paracrine Shh signaling (Ahn and Joyner, 2004; Harfe et al., 2004). Deletion of Shh at different developmental stages revealed that Shh patterns digit 2 to 5 during a

24 transient period (~12hr) after its activation and it continues to regulate growth-expansion of digit primordia (Zhu et al., 2008). In contrast to what is known about the regulation of posterior skeletal pattern, less is known about mechanisms that govern anterior skeletal formation. It is presumed that Shh does not directly contribute to the generation of anterior elements, though the identity and temporal requirement of factors that do so remain unclear.

Gli3 is a key mediator of the Shh pathway during development (Litingtung et al., 2002; Lopez- Rios et al., 2012; Robert and Lallemand, 2006; te Welscher et al., 2002b). Shh prevents proteolysis of Gli3 into its repressor form (Gli3R) that, in turn, prevents transcription of Shh target genes (Jiang and Hui, 2008; Wen et al., 2010). A posterior to anterior gradient of Shh generates an intracellular counter-gradient of Gli3R (Wang et al., 2000). High level of Gli3R in the anterior limb bud is probably involved in specifying digit 1 identity (Hill et al., 2009; Hui and Joyner, 1993; Litingtung et al., 2002; Wang et al., 2007). However, the complex functions of Gli3 in pattern formation and digit number regulation make it difficult to dissect the molecular mechanism underlying anterior skeletal specification.

Iroquois genes encode a family of homeodomain transcription factors that are conserved from worms to vertebrates (Gómez-Skarmeta and Modolell, 2002). Six Irx genes are located in two clusters in the mouse and human (IrxA and IrxB). Irx3 and Irx5 (Irx3/5) of the IrxB cluster are expressed in the developing limb bud (Cohen et al., 2000; Houweling et al., 2001). Here we show that Irx3/5 are essential for proximal and anterior skeletal formation. They are required to establish early AP polarity and specify the anterior progenitor population. However, by reducing Shh dose, we are able to rescue skeletal elements that are lost in Irx3/5 double knockout (DKO) mutants, indicating that Shh signaling negatively regulates the formation of these anterior elements. Our data suggest a biphasic model for development of the femur, tibia and digit 1: an early Irx3/5-dependent specification phase followed by a progressive determination phase, which is inhibited by Shh signaling.

2.3 Materials and Methods

2.3.1 Mice and genotyping

Mice used in this study were housed in standard vented cages in conformity with the Toronto Center for Phenogenomics recommendations. For genotyping of the mice and embryos,

25 earnotches, tail clips, or yolk sac were digested in 300ul of 50mM NaOH at 100°C for 10 minutes. NaOH was then neutralized by adding 100ul of 0.5M Tris (pH 8.0). 3ul of DNA solution was used for PCR genotyping. Mouse lines used in this study were as follows: Cre- ERTM (Hayashi and McMahon, 2002), Prx1Cre (Logan et al., 2002), Irx3/5-DKO (Irx3-5EGFP) (Zhang et al., 2011), Irx3/5-CKO (Irx3flox5EGFP) (Zhang et al., 2011), Irx3-3Myc6His (Zhang et al., 2011), Gli3-/- (Gli3Xt) (Hui and Joyner, 1993), Gli3P1-4/+ (Wang et al., 2007), Kif7-/- (Cheung et al., 2009) and Shh+/- (Chiang et al., 1996). Mice were maintained on a mixed outbred background of CD1 and 129/Sv.

2.3.2 Cre activity induction via tamoxifen administration

To induce Cre activity in Cre-ERTM;Irx3/5-CKO embryos, tamoxifen (20 mg/ml in sesame oil, Sigma-Aldrich) was administered to pregnant females via gavage with the following doses: 0.1375mg/g body weight at E7.5 (TM E7.5), 0.1625mg/g body weight at E8.5 (TM E8.5) or 0.2mg/g at E9.5 (TM E9.5).

2.3.3 Western blot

Hindlimb buds from E10.5 (24hr post-TM), E11.0 (36hr post-TM) and E11.5 (48hr post-TM) Cre-ERTM;Irx3/5-CKO embryos and control littermates treated with tamoxifen at E9.5 were dissected in cold PBS and flash frozen in liquid nitrogen. Total cell lystes were prepared in RIPA buffer (50mM Tris pH7.4, 150mM NaCl, 5mM EDTA, 1mM EGTA, 0.1% SDS, 0.5% Doc, 1% NP-40, 25mM sodium pyrophosphate, 1mM sodium orthovadanate, 10mM NaF, 1mM β- glycerophosphate, and EDTA-free complete protease inhibitor cocktail (Roche)) followed by sonication. Proteins were separated on 8% SDS-PAGE and transferred to nitrocellulose for immunoblotting overnight at 4°C with Irx3 antibody generated by Vijitha Puviindran from our lab or Actin antibody (Oncogene). On the second day, nitrocellulose membranes were incubated with peroxidase-conjugated secondary antibodies (Oncogene) and developed using the ECL detection system (Thermo Scientific).

2.3.4 Cartilage staining

E14.5 embryos were dissected in PBS and fixed in Bouin’s solution at 4°C overnight. Embryos were washed in 70% EtOH/0.1% NH4OH for several times and then twice in 5% acetic acid 1hr each at room temperature. 0.07% alcian blue (Aldrich) in 5% acetic acid were used as staining

26 solution. After 2hr of staining, embryos were washed in 5% acetic acid and methanol several times, then cleared and stored in methyl-salicylate (Fluka).

2.3.5 Quantification of limb bud frontal area plane

Forelimb or hindlimb buds of embryos at the same somite stages were carefully photographed with the same magnification. Outlines of limb buds were drawn using the Pen tool in Adobe Photoshop and transformed to selection regions. Numbers of pixels of each selected region were recorded from information in Adobe Photoshop and standardized to those of control embryos using Microsoft Excel.

2.3.6 Section immunofluorescence

Sections were prepared for immunofluorescence following standard procedures (Li et al., 2012). Limb buds were sectioned along the coronal plane (AP-PD). Antibodies used were as follows: phospho-H3 (Cell Signaling) and cleaved Caspase-3 (Cell Signaling).

2.3.7 Whole-mount in situ hybridization

Antisense RNA digoxigenin-dUTP-labeled riboprobes were generated from linearized DNA plasmids. Probes were synthesized using a DIG-labeling kit (Roche) according to manufacturer’s instructions and purified through columns (Roche). After dilution with DEPC water, probes were stored at -80°C. In brief, embryos were dissected in cold DEPC treated PBS and fixed in 4% PFA overnight, dehydrated through a series of MeOH/PBT solutions, bleached in 6% H2O2/MeOH solution and stored in MeOH at -20°C. After rehydration, embryos were treated with proteinase K, and digestion was stopped in 2mg/ml glycine in PBT. Then embryos were refixed in 0.2% Gluteraldehyde and 4% PFA. After incubation in hybridization buffer (50% formamide, 5x SSC pH4.5, 1% SDS and 50mg/ml heparin) at 63°C for 1.5hr, probes were added, and hybridization was performed overnight at 67°C. Following a series of post-hybridization washes at 70°C, embryos were further washed in MABT (100 mM maleic acid, 150 mM NaCl, pH 7.5, 0.1% Tween-20) solution. Prior to antibody incubation, embryos were blocked in 2%RBR (Roche) and 2% RBR/20% heat-inactivated sheep serum for 2hr at room temperature. After incubation with anti-DIG antibody (1:2000, Roche) at 4°C overnight, embryos were washed with MABT extensively. BM purple AP substrate (Roche) was used to develop color in

27 dark, and the color reaction was stopped using 0.5mM EDTA in PBT solution. Embryos were then cleaned in MeOH and fixed and stored in 4% PFA.

2.3.8 RNA isolation and real-time quantitative PCR

Limb buds for qRT-PCR were dissected and snap frozen using liquid nitrogen, then stored at -80 ºC until genotypes were confirmed. Limb buds were then homogenized and pooled, and RNA was extracted using TRIzol Reagent (Invitrogen). Complementary DNA was synthesised using SuperScript First-Strand Synthesis System (Invitrogen) and analysed using the SYBR Green system (Applied Biosystems). Each reaction was repeated in technical triplicates. Relative gene expression data was analyzed using the 2-ΔΔCT method (Schmittgen and Livak, 2008). Values are normalized to GAPDH expression and are shown as means ± standard deviation. For E9.75 (30- 31 somites) embryos, two groups of HL buds were assayed. Each group was a pool of 11 or 13 pairs of mutant and control littermate HL buds. For E10.75 (40-42 somites), three groups of HL buds were assayed and each group contained two pairs of buds.

2.3.9 Chromatin immunoprecipitation

Chromatin immunoprecipitation was performed using a ChIP-IT kit (Active Motif) according to the manufacturer’s instructions as previously described (Hu et al., 2006; Pospisilik et al., 2010). Chromatin was isolated from pools of limb buds from E10.5 Irx3/5-DKO, Irx3-3Myc6His and wild-type embryos. Irx3 antibody (Vijitha Puviindran, homemade), Myc antibody and negative control IgG (Active Motif) were used to immunoprecipitate Irx3 bound-DNA complex. After sonication and reverse crosslink, Irx3 bound-DNA fragments were amplified using primer set to Gli3 CNE6 region: 5’-GTCAATCGCCAACAAAACCTT-3’ and 5’- TGGCAGCAGCTTTAATTGGTA-3’.

2.3.10 Quantification of location of Shh domain in limb buds and percentage of limb bud frontal plane area expressing Gli1

To quantify the location of Shh domain in forelimb and hindlimb buds at comparable developmental stages, whole-mount in situ hybridization of Shh was performed in wild-type embryos from E9.75 (28-somite stage) to E10.75 (40-somite stage), and images of Shh expression in forelimb and hindlimb buds were carefully taken. Images were then processed in Adobe Photoshop to position the limb bud properly (X and Y axes representing the limb PD and

AP axes respectively). The positions of anerior and posterior ends of the limb bud and the center of the Shh domain along the Y axis were recorded, and the limb bud AP length was measured. The relative position of the center of the Shh domain was calculated with anterior and posterior ends of the limb bud set to 100 and 0, respectively.

To quantify proportions of Gli1-expressing area in limb buds, saturated whole-mount in situ hybridization of Gli1 was performed in wild-type embryos from E9.75 (28-somite stage) to E10.75 (40-somite stage), and images of Gli1 expression in forelimb and hindlimb buds were carefully taken. Images were then processed in Adobe Photoshop to adjust brightness and contrast. All images were processed with the same standard. Limb bud Gli1 in situ images were then converted to greyscale mode. To select Gli1 expression domain, area with K>30% was selected automatically using the Color Range command. The limb bud region was selected carefully using the Pen tool by tracking the limb bud outline. Areas of Gli1 expression domain and limb bud were measured with pixel as unit, and the proportion of limb bud expressing Gli1 was calculated using Microsoft Excel.

2.4 Results

2.4.1 Irx3/5 are required prior to limb bud outgrowth

During limb bud initiation, expression of Irx3/5 was observed throughout the anterior region of the limb field. Then, their expression became restricted to the anterior-proximal region of the growing limb bud. However, Irx3/5-DKO mutant embryos lack anterior distal hindlimb elements. The discrepancy between the proximal expression domain in the limb bud post initiation and the distal skeletal loss led to the hypothesis that Irx3/5 are required prior to bud outgrowth when their expression is not proximally restricted.

To test the temporal requirement for Irx3/5, I employed a ubiquitously expressed tamoxifen- inducible Cre (Cre-ERTM) line (Hayashi and McMahon, 2002) to delete the floxed Irx3 gene in Irx3flox5EGFP/Irx3-5EGFP mouse embryos (Cre-ERTM;Irx3/5-CKO). Maternal delivery of a single dose of tamoxifen resulted in rapid decrease of Irx3 protein level (Figure 2.1A, left panel). At 48-hour post tamoxifen injection, no Irx3 can be detected in Cre-ERTM;Irx3/5-CKO limb buds by western blot (Figure 2.1A, right pannel). These mutant hindlimbs displayed various phenotypes that were classified into four groups based on severity, with type I similar to wild-type and type

IV phenocopying Irx3/5-DKO (Figure 2B). When tamoxifen was administered at an early time point (TM E7.5), most mutant hindlimbs exhibited a severe phenotype with loss of the femur, tibia and digit 1 that phenocopied Irx3/5-DKO mutants (Figure 2.1B (IV) and C). When tamoxifen was delivered at intermediate time points (TM E8.5), digit one and the tibia were less affected (Figure 2.1B (II, III) and C). When tamoxifen was given at the latest time point (TM E9.5), development of the femur was also less affected (Figure 2.1B (I) and 2C). Since that overt outgrowth of the hindlimb bud occurs at E9.75 (Figure 2.1D, blue arrowhead), these data suggest that Irx3/5 are required just before or during hindlimb initiation to form the affected skeletal elements (Figure 2.1D).

2.4.2 Irx3/5 regulate AP prepattern to promote proximal and anterior hindlimb progenitors

To define the early hindlimb bud defect in Irx3/5-DKO mutants, I examined their morphology. Irx3/5-DKO hindlimb buds are smaller than those of control littermates, especially along the AP axis at E10.5, and the morphological defects arise early during limb bud outgrowth (Figure 2.2A and data not shown). Proliferation, as revealed by immunostaining of the mitotic marker phosphorylated histone H3 (pH3) (Hendzel et al., 1997), was not different with respect to controls (Figure 2.2B-D). Apoptosis, assessed by section immunofluorescence staining of cleaved caspase-3 (Nicholson et al., 1995), was not detected at the initiation stage (data not shown) but increased in Irx3/5-DKO buds at E10.5 (Figure 2.2D). Interestingly, this increase was localized to the most proximal portion of the bud and is similar to that seen in Gli3;Plzf- DKO mice, which exhibit severe hypoplasia of the femur (Barna et al., 2005). The increased apoptosis in the proximal limb bud may account for femur hypoplasia, a skeletal deficiency shared by both mutants. Conversely, the distal and anterior skeletal defect of Irx3/5-DKO mutants might be attributable to other causes, such as cell fate specification.

To investigate cell fate specification in the early hindlimb bud, I examined expression of key regulators of prepattern. During limb initiation, the expression domains of Gli3 and Hand2 mark anterior and posterior limb fields, respectively (Figure 2.3A and E). A mutually antagonistic interaction between these two transcription factors establishes early AP polarity of the bud (te Welscher et al., 2002a). Similarly, Alx4 (downstream of Gli3) and Tbx2 are expressed in the anterior and posterior aspects of initiating limb bud, respectively (Suzuki et al., 2004; te Welscher et al., 2002a) (Figure 2.3C and G). Comparing to control hindlimb buds, Irx3/5-DKO

30 hindlimb buds exhibited reduced expression of both anterior markers Gli3 and Alx4. In addition, the posterior extent of their expressions was diminished (Figure 2.3A-D). On the other hand, expression of posterior markers Hand2 and Tbx2 was expanded anteriorly in mutant hindlimb buds during prepattern stage (Figure 2.3E-H). Interestingly, the anterior expression of the hindlimb field marker Tbx4 (Chapman et al., 1999) was diminished in Irx3/5-DKO mutants prior to limb bud outgrowth (Figure 2.3I and J). These data suggest that the anterior hindlimb population is not properly established in Irx3/5-DKO embryos.

Consistent with the idea that Irx3/5 are required to promote anterior hindlimb progenitors, post- initiation (E10.5) hindlimb buds in Irx3/5-DKO mutants were smaller than those of control littermates (Figure 2.2A), and the mutant limb bud was posteriorized as revealed by marker gene analysis (Figures 2.4). Signaling centers, including Shh in the ZPA and Fgf8 in the apical ectodermal ridge (AER) (Lewandoski et al., 2000), were anteriorly shifted and biased, respectively (Figures 2.4A-D). Gli1 and Ptc1 are posterior markers that are expressed in Shh- responding limb bud cells (Jiang and Hui, 2008; Zhu et al., 2008). Their expression domains were expanded to the anterior limb margin in the Irx3/5-DKO hindlimb bud (Figure 2.4E-H). Given that Gli1-expressing cells give rise to digits 2 through 5 but not digit 1 (Ahn and Joyner, 2004), this finding suggests the Irx3/5-DKO hindlimb is primed to generate only posterior digits which respond to Shh signaling. Furthermore, expression of anterior marker genes Gli3 and Alx4 (te Welscher et al., 2002a) was greatly reduced (Figure 2.4I-L). Together, these findings suggest that anterior skeletal progenitors were absent in Irx3/5-DKO hindlimbs, due to early specification defect during limb bud initiation.

2.4.3 Irx3 binds to Gli3 limb enhancer and regulates Gli3 expression in hindlimb buds

To confirm that Gli3 expression was greatly decreased in Irx3/5-DKO hindlimb buds, I purified RNA from control and mutant hindlimb buds and performed quantitative real-time PCR. As expected, Gli3 expression level was lower in Irx3/5-DKO hindlimb buds than that of control littermates (~50%) at both bud initiation and outgrowth stages (E9.75 and E10.5) (Figure 2.5A and B).

Since expression domains of Irx3/5 overlap with that of Gli3, Irx3/5 might directly influence Gli3 expression to modulate AP pattern formation. I identified a potential Irx binding site,

GCATCTGT (Bilioni et al., 2005), within the conserved non-coding sequence element 6 (CNE6) that directs limb bud expression of Gli3 (Abbasi et al., 2010 and Figure 2.5C). To test whether Irx3 binds CNE6 in vivo, we performed chromatin immunoprecipitation (ChIP) analysis in E10.5 limb buds. These experiments revealed specific signals in Irx3-bound chromatin from wild type (WT), but not Irx3/5-DKO, embryos (Figure 2.5D, left pannel). The specificity of Irx3 binding to this site was further verified by ChIP analysis using Irx3-3Myc6His embryos (Zhang et al., 2011) and anti-Myc antibody (Figure 2.5D, right pannel). These results support the notion that Irx3/5 are upstream regulators of AP polarity by directly up-regulating expression of Gli3 in the anterior hindlimb field.

2.4.4 Early genetic interaction between Irx3/5 and Gli3 is essential for signaling center establishment

Similar to Irx3/5-DKO, Gli3 mutant hindlimbs exhibited anterior deficiency, including loss of tibia and digit 1 (Figure 2.6B and C). Though Gli3 mutants display polydactyly in their limbs, the identity of those extra anterior digits is not digit 1 (Ahn and Joyner, 2004). To determine whether genetic interaction exists between Irx3/5 and Gli3 in limb development, especially anterior pattern formation, we generated Irx3/5;Gli3-triple knock out (Irx3/5;Gli3-TKO) embryos. Surprisingly, triple mutant hindlimbs exhibited total loss of skeletal elements (Figure 2.6D). In keeping with this phenotype, Fgf8 expression in limb was abolished (Figure 2.6N), suggesting that the AER is deficient. This is further supported by the evidence that the AER structure was absent in Irx3/5;Gli3-TKO hindlimb buds (see 4.3). As a result, expression of the outgrowth cue Fgf10 (Ohuchi et al., 1997; Sekine et al., 1999) was markedly downregulated by E10.5 (Figure 2.6S). Expression of Shh was delayed and decreased, and its position was shifted to the center of the distal mesenchyme (Figure 2.6I). Therefore, Irx3/5 interact with Gli3 to establish signaling centers of both PD and AP axes (AER-Fgf8 and ZPA-Shh, respectively).

To define the temporal requirement for this interaction, we used Prx1Cre to generate triple mutants at a stage coincident with hindlimb bud initiation, since hindlimb expression of Cre driven by the Prx1 enhancer initiates at E9.5 and is widespread by E10.5 (Logan et al., 2002). In conditional Prx1Cre;Irx3-5EGFP/Irx3flox5EGFP;Gli3-/- triple mutant embryos (Prx1Cre;TKO), Shh and Fgf8 expressions were robust (Figure 2.6J and O), the ZPA was posteriorly biased (Figure 2.6J) and limb outgrowth was largely restored (Figure 2.6E and T), suggesting that Irx3/5 and

Gli3 interact before limb bud outgrowth to establish and position these signaling centers to promote limb outgrowth.

2.4.5 Irx3/5-dependent anterior progenitor population contributes to preaxial polydactyly

Preaxial (anterior) polydactyly is a common congenital limb anomaly in humans. A major underlying mechanism involves ectopic Shh pathway activation in the anterior bud that leads to overgrowth of anterior mesenchyme (Hill, 2007; Lopez-Rios et al., 2012). In mice, mutants with increased Shh pathway activity, such as Gli3P1-4/+ and Kif7-/-, exhibit preaxial polydactyly (Cheung et al., 2009; Endoh-Yamagami et al., 2009; Wang et al., 2007)). To test whether polydactylous digits are derived from Irx3/5-dependent progenitors, I removed Irx3/5 function in Gli3P1-4/+ and Kif7-/- backgrounds. Consistent with the notion that Irx3/5 are required to specify progenitors that give rise to extra anterior digits in preaxial polydactyly, both Irx3/5-/-;Gli3P1-4/+ and Irx3/5;Kif7 triple KO (Irx3/5;Kif7-TKO) mutants exhibited a hindlimb phenotype similar to that of Irx3/5-DKO with neither digit 1 nor polydactyly (Figure 2.7 and Figure 2.8A-D). These data support the notion that Irx3/5 are required to specify anterior progenitors, which give rise to extra digits in preaxial polydactyly.

2.4.6 Requirement of Irx3/5 in forelimb AP pattern formation is revealed in Kif7-/- background

Surprisingly, forelimbs in Irx3/5;Kif7-TKO embryos also lacked polydactyly and digit 1. In addition, the radius was lost and the humerus was severely deficient (Figure 2.8E-H). In Irx3/5;Kif7-TKO forelimbs, early skeletal condensations marked by Col2a1 (McGlinn et al., 2005) were absent in the prospective regions of the radius and digit 1 (Figure 2.8I-L), expression of anterior markers Pax9 and Alx4 was diminished (Figure 2.8M-P and data not shown), and expression of posterior marker Gli1 was expanded anteriorly (Figure 2.8Q-T). Therefore, the phenotype of that of Irx3/5;Kif7-TKO forelimbs are analogous to Irx3/5-DKO hindlimbs.

To reassess the role of Irx3/5 in forelimb formation, we scrutinized Irx3/5-DKO forelimb buds and found that they indeed exhibited AP patterning defect though milder than that of Irx3/5- DKO hindlimb buds. The mutant forelimb buds are smaller than controls post-initiation (Figure 2.9A). Gli1 expression was slightly expanded anteriorly at E10.5 (Figure 2.8Q and R) in conjunction with diminished Alx4 expression during bud initiation (Figure 2.9B). These data

33 suggest that Irx3/5 also regulate limb bud size and AP pattern formation during the early stages of forelimb development. However, the requirement of Irx3/5 in forelimb anterior skeletal pattern can only be revealed in the Kif7-/- background with elevated Shh signaling.

2.4.7 Forelimb bud displays lower Shh signaling activity than that of the hindlimb

To test whether the forelimb is less sensitive to removal of Irx3/5 because of lower native Shh pathway activity relative to the hindlimb, I reexamined Shh and Gli1 expression at comparable stages (hindlimb is about 5 somite stage delayed than forelimb) during the Shh-dependent AP pattern time window (within ~12h post ZPA-Shh activation) (Zhu et al., 2008). Shh is first detected at the 27-28-somite stage in the forelimb bud and 32-33-somite stage in the hindlimb bud (Figure 2.10A and F). The Shh expression domain is more posteriorly restricted in the forelimb relative to the hindlimb (Figure 2.10). Consistent with this, Gli1 is also restricted more posteriorly in the forelimb than the hindlimb using a threshold value to mark the expression domains based on saturated staining of RNA in situ hybridization (Figure 2.11A and C1-D6’). Furthermore, qRT-PCR analysis confirmed that Gli1 RNA expression is indeed lower in the forelimb relative to the hindlimb at comparable stages (Figure 2.11B). Therefore, there is a broader Gli1-negative anterior domain in the developing forelimb relative to the developing hindlimb (Figure 2.11C1-D6’). This difference in the extent of Shh signaling might explain why skeletal pattern is relatively protected from the deleterious effect of Irx3/5 loss in the forelimb.

2.4.8 Reducing Shh signaling rescues anterior skeletal formation in the Irx3/5-DKO hindlimb

We next asked if reduced Shh signaling in Irx3/5-DKO hindlimb buds could rescue skeletal pattern. When one copy of the Shh gene was removed in the Irx3/5-DKO background (Irx3/5- DKO;Shh+/-), the Gli1 expression domain was reduced (Figure 2.12A-C) and Gli3 expression domain was expanded in hindlimb buds at E10.5 (Figure 2.12D-F). As a consequence, a Gli1- negative region appeared in the anterior hindlimb bud of these mutants, and Pax9 expression was rescued (Figure 2.12G-I). Consistent with this expression, although prepattern and limb bud size of Irx3/5-DKO;Shh+/- hindlimb buds were not rescued (Figure 2.13), 60% (n=12/20) of these mutants exhibited partial or complete rescue of the femur, tibia and digit 1 (Figure 2.12L-M). Thus, even anterior progenitors were not specified in the initiating Irx3/5-DKO;Shh+/- hindlimb

34 buds, by reducing Shh signaling at later stages, anterior skeletal elements can be respecified and formed. These results indicate that Shh signaling negatively regulates the formation of these anterior skeletal elements (Figure 2.14).

2.5 Discussion

In this study, we identified the novel function of Irx3/5 in the early specification of anterior population which gives rise to digit 1, tibia and femur in the hindlimb. Our data place Irx3/5 in an upstream position as regulators of limb AP prepattern by promoting Gli3 expression in the anterior hindlimb field likely in a direct manner. Surprisingly, we found an early genetic interaction between Irx3/5 and Gli3, which is essential for establishment of signaling centers. In addition, mutant analysis indicated that reducing Shh signaling in Irx3/5-DKO background rescues anterior pattern formation, suggesting that the anterior pattern can be restored if a Shh- free zone can be maintained in the anterior margin during limb bud development. And that the forelimb is more protective to the deleterious effect of Irx3/5 is likely due to its nature of a bigger Shh-free zone than in the hindlimb. With these data, we proposed a biphasic model for development of anterior limb elements, which involves early specification by Irx3/5 and progressive inhibition by Shh signaling (Figure 2.14).

2.5.1 A biphasic model for limb anterior pattern formation

Skeletal pattern is essential to limb function. Relative to posterior skeletal structures, mechanisms that specify anterior pattern are less understood. Our findings indicate that a second major group of primarily anterior progenitors is specified by Irx3/5 to generate the femur, tibia and digit 1. Without Irx3/5, the anterior population cannot be properly established during limb bud initiation, resulting in a small and posteriorized limb bud, which only gives rise to posterior skeletal elements that response to Shh signaling (descendants of Gli1-expressing cells) (Ahn and Joyner, 2004). However, the anterior structures can be restored in the Irx3/5-DKO background if a Shh signaling-free (Gli1-negative) environment can be established in the anterior limb bud during the Shh-dependent AP patterning time window (Zhu et al., 2008). Neither prepattern nor early limb bud size were rescued in Irx3/5-DKOS;Shh+/- hindlimbs, suggesting that the early specification defect remains, but by reducing Shh level, we can bypass the early requirement of Irx3/5 during limb initiation for the formation of anterior elements. These data demonstrate that anterior limb pattern formation is plastic: while anterior progenitors are specified early, their fate

35 is not committed until later. Therefore, anterior positional identity is specified by Irx3/5 during and/or prior to limb bud initiation and determined progressively by Shh signaling after limb bud outgrowth (Figure 2.14). Previous models suggested that anterior structures form independent of Shh. In contrast, our data indicate that Shh negatively regulates their formation (Figure 2.14). Digit 1 is more essential for specialized functions of the upper extremity, such as grasp, than for lower extremity function. Interestingly, development of digit 1 is apparently better protected from the deleterious effect of Irx3/5 in the forelimb than in the hindlimb of mice by virtue of a larger zone free of Shh signaling (Gli1-negative).

2.5.2 Irx3/5 are key determinants of anterior population in early hindlimb field

We also established that Irx3/5 regulate limb AP polarity upstream of the well-recognized Gli3- Hand2 dependent prepatterning. Our ChIP data suggested that Irx3/5 bind to Gli3 limb enhancer to directly promote Gli3 expression in the anterior hindlimb field. Interestingly, when Irx3/5 and Gli3 are removed together, limb bud outgrowth failed becuase of loss of AER-Fgf signaling. Both Gli3-/- and Kif7-/- mutants exhibit preaxial polydactyly limb phenotype (Cheung et al., 2009; Hui and Joyner, 1993). Then, why Irx3/5;Gli3-TKO and Irx3/5;Kif7-TKO hindlimbs displayed different phenotypes? I speculate that complete posteriorization of limb bud is inhibitory for the activation and/or maintenance of the AER-Fgf signaling center and thus limb bud outgrowth. Irx3/5 and Gli3 are two of the most important determinants of the anterior limb population, and without them the limb field may be completely posteriorized (see 4.3). Different from this, removing Kif7 only causes elevated Gli activator level, and Gli3R, despite at low level, still remains in anterior limb mesenchyme (Cheung et al., 2009), which may be sufficient to establish a small anterior limb population avoiding complete posteriorization. Previous studies using chick recombinant limb found that recombinant limbs composed of primarily posterior limb bud mesenchymal cells (especially the ZPA cells) inhibited outgrowth and the AER structure, while recombinant limbs using anterior limb tissue (without the ZPA) could generate all three limb segments (stylopod, zeugopod and autopod) with symmetrical digit pattern along the AP axis (Crosby and Fallon, 1975; Frederick and Fallon, 1983). These data indicated that the posterior limb population cannot maintain the AER and distal outgrowth by itself, which is similar to the situation of Irx3/5;Gli3-TKO hindlimbs.

Figure 2.1 | Irx3/5 are required prior to limb bud outgrowth for the formation of anterior- distal hindlimb elements

(A) (left) Irx3 protein level in Cre-ERTM;Irx3/5-CKO embryos comparing to Irx3/5-CKO embryos 24hr, 36hr and 48hr post tamoxifen (TM) administration. (right) Western blot using Irx3 antibody in an Irx3/5-CKO embryo (flox/-) and two Cre-ERTM;Irx3/5-CKO embryos (MT1 and MT2) 48 hour post TM injection. Irx3 protein cannot be detected by Western blot by 48hr post TM injection. (B) Representative hindlimb cartilage structures of type I-IV in E14.5 Cre- ERTM;Irx3/5-CKO mutants. The black asterisk and arrowhead in IV indicate loss of digit 1 and tibia, grey and light grey asterisks/arrowheads in III and II indicate different extent of rescue of

37 digit 1 and tibia. (C) Percentages of type I-IV phenotypes in the right hindlimbs (which showed consistent more severe phenotype than the left hindlimbs) of Cre-ERTM;Irx3/5-CKO embryos with tamoxifen treatment at indicated stages. The hindlimb phenotypes are less severe when TM injection was given at later stages. (D) Scheme of Cre-mediated Irx3/5-CKO hindlimb phenotypes in relation to Irx3 level in the hindlimb field after Cre activation (red arrowheads) and the stage of hindlimb bud initiation (E9.75, blue arrowhead). Complete elimination of Irx3/5 prior to hindlimb initiation is required to generate Irx3/5-DKO phenotype in the hindlimb.

Figure 2.2 | Morphogenesis, cell proliferation and cell death in Irx3/5-DKO hindlimb buds

(A) Morphology of wild-type (WT) and Irx3/5-DKO hindlimb buds 29, 32, 34 and 36-somite stages. The anterior aspect of mutant buds become flat (black arrowheads) during outgrowth. (B) Section immunofluorescence of phospho-histone H3 (pH3) and DAPI staining in 29-somite stage hindlimb bud section of control and Irx3/5-DKO, white dot lines indicate anterior border of limb buds. (C) Quantification of pH3-positive cell in the anterior half of hindlimb field in control and Irx3/5-DKO embryos at E9.75-10.0 from multiple sections of several limb buds. No obvious difference was observed between mutant and control. (D) Section immunofluorescence of pH3, cleaved Caspase-3 (Caspase3) and DAPI staining in 37-somite stage hindlimb buds of control and mutants. Apoptosis are increased in the anterior region at the base of mutant hindlimb bud.

Figure 2.3 | Irx3/5 regulate hindlimb bud prepattern

Expression of Gli3 (29-somite stage), Alx4 (29-somite stage), Hand2 (30-somite stage), Tbx2 (29-somite stage) and Tbx4 (28-somite stage, lateral view) mRNAs in control and Irx3/5-DKO hindlimb buds during limb initiation. Red markings denote expression domains of genes indicated. In initiating Irx3/5-DKO hindlimb buds, anterior genes (Gli3 and Alx4) expression is

39 reduced, expression domains of posterior genes (Hand2 and Tbx2) expanded anteriorly, and exression of hindlimb marker, Tbx4, specifically lost in the anterior hindlimb field.

Figure 2.4 | Irx3/5-DKO hindlimb buds are primed to form Shh-responding elements

Expression of Shh, Fgf8, Gli1, Ptc1, Gli3 and Alx4 mRNAs in E10.5-10.75 (36 to 39-somite stages) hindlimb buds of control and Irx3/5 DKO mutant embryos. Numbers in (A and B) indicate somite levels. Black arrowheads in (I and J) indicate the posterior boundary of Gli3 expression domain. Mutant hindlimb buds are posteriorized: ZPA-Shh domain is anteriorly shifted, AER-Fgf8 displays anterior biased exapression, posterior marker expression is expanded anteriorly (Gli1 and Ptc1), and anterior marker expression is greatly reduced (Gli3 and Alx4). The anterior Shh-free zone (Gli1-negative) has disappeared in mutant hindlimb buds, suggesting they are primed to form Shh-responding elements (descendants of Gli1-expressing cells) .

Figure 2.5 | Irx3/5 regulate Gli3 expression in limb

(A and B) Gli3 expression level in control (WT) and Irx3/5-DKO (DKO) hindlimb buds at indicated stages assayed by quantitative RT-PCR. Relative transcript levels were normalized to the expression of Gapdh. The expression levels of mutant samples were calculated in relation to wild-type controls (average set to 100%). Gli3 mRNA level is significantly decreased in mutant hindlimb buds at both initiation stage (E10.0) and during limb outgrowth (E10.75). Error bars represent standard deviations. P-values were calculated using two-tailed Student’s t-test. (C) Schematic of the mouse Gli3 locus. The Gli3 limb enhancer (CNE6) is located in intron 10. The red line indicates approximate position and size of the PCR amplicon in (D). (D) (left) ChIP-PCR using Irx3 antibody detects interaction of endogenous Irx3 protein with Gli3 CNE6 locus in E10.5 wild-type (WT) embryos but not Irx3/5-DKO embryos (negative control). IgG ChIP was performed as an additional negative control. (rightt) ChIP-PCR using Myc antibody detects interaction of Myc-tagged Irx3 protein with Gli3 CNE6 locus in E10.5 Irx3- 3MycH6His (Irx3Myc) embryos but not non-tagged Irx3 protein in wild-type (Irx3) embryos (negative control).

Figure 2.6 | Early genetic interaction between Irx3/5 and Gli3 is required for signaling center establishment and limb bud outgrowth

(A-E) Hindlimb skeletal staining in E14.5 embryos with indicated genotypes. (F-T) Expression of Shh, Fgf8 and Fgf10 in E10.5 (36-40 somites) hind buds of embryos of the genotypes indicated. The red line in (D) marks the outline of the severely hypotrophic hindlimb bud in a Irx3/5;Gli3-triple knock outembryo. In contrast, limb outgrowth is rescued in Prx1- Cre;Irx3/5;Gli3-TKO (Prx1Cre;TKO) embryos (E) as is the expression of limb organizers Shh (J) and Fgf8 (O) as well as the outgrowth cue Fgf10 (T).

Figure 2.7 | Irx3/5-dependent anterior progenitor population contributes to preaxial polydactyly

Alcian blue staining reveals cartilage structures of E14.5 hindlimbs with indicated genotypes. The extra anterior digits in Gli3P1-4/+ hindlimbs are lost together with tibia when crossing to Irx3/5-DKO background (D).

Figure 2.8 | Irx3/5 are required for the formation of anterior forelimb elements in Kif7-/- background

(A-H) Alcian Blue staining of hindlimb and forelimb cartilage elements in E13.5 embryos of indicated genotypes. Red lines in (C and G) mark the anterior extra digits in Kif7-/- mutants. Red labeling in (H) indicates the lack of cartilage condensation of radius and digit 1 in Irx3/5;Kif7- TKO forelimb. (I-L) Expression of Col2a1 marks chondrogenesis during digit formation in the forelimb at E12.5. The red line in (K) indicates extra digits forming in the anterior forelimb of Kif7-/- embryos. Red labeling in (L) indicates the lack of chondrogenesis for radius and digit 1 in Irx3/5;Kif7-TKO forelimb. (M-T) Expression of Pax9 and Gli1 at E11.5 and E10.5 (37 to 38- somite stage). Similar to Irx3/5-DKO hindlimb buds, in Irx3/5;Kif7-TKO forelimb buds, Pax9 expression is completely lost (P), and Gli1 domain displays completely anterior expansion (T). Red lines in (Q and R) mark the Gli1-negative region in the anterior forelimb bud. The Gli1- negative domain is smaller in Irx3/5-DKO forelimb buds than controls.

Figure 2.9 | Irx3/5 are involved in forelimb development

(A) Quantification of the frontal plane area of control and mutant forelimb buds at E9.75 (29- somite stage). Error bars represent standard deviations, and asterisks indicate p<0.05 (two-tailed Student’s t-test). The average area of control forelimb buds is set to 100%, and Irx3/5-DKO forelimb buds are about 20% smaller than that of controls. (B) Expression of Alx4 is decreased in Irx3/5-DKO (DKO) forelimb buds than that of controls (Ctrl) at E9.25 (23 to 24-somite stage).

Figure 2.10 | Shh domain is more posteriorly restricted in the forelimb bud than hindlimb

Expression of Shh in wild-type forelimb and hindlimb buds at indicated somite stages (at bottom right corner of each panel). Development of hindlimb buds is delayed for about 4 to 5 somite stages comparing to the forelimb buds. Black lines in indicate the extent of limb buds along the AP axis. Posterior end is set to 0 and anterior end 100 as shown in (A). Red arrowheads mark the center of Shh domain along the AP limb axis. Red numbers indicate the relative positions of the center of Shh domain from the posterior boundary of limb buds. The smaller the number is, the more posteriorly localized of the Shh domain.

Figure 2.11 | Forelimb bud displays lower Shh signaling activity than that of the hindlimb

(A) Quantification of the area of Gli1 domain relative to the limb bud using a threshold value to mark the expression domains based on saturated staining of RNA in situ hybridization (see also C1-D6’, n≥4 for each stage). X-axis indicates stages of comparable forelimb and hindlimb buds (see table below for corresponding somite stage). Error bars represent standard deviations, and asterisks indicate p<0.01. (B) Relative expression level of Gli1 (normalized to Gapdh) in wild- type hindlimb buds to forelimb buds at comparable stages assessed by qRT-PCR. The average of Gli1 level in the forelimb buds is set to 100%. Error bars represent standard deviations and the

46 asterisk indicates p<0.05. (C1-D6’) Examples of ratios of Gli1 domain to limb bud area (%) in forelimb and hindlimb buds at multiple comparable stages. Black and red circled regions mark the limb regions and Gli1 domains respectively.

Figure 2.12 | Reducing Shh signaling in Irx3/5-DKO hindlimb buds rescues anterior pattern formation

(A-I) Expressions of Gli1, Gli3 at E10.5 (36 to 37-somite stage) and Pax at E11.5 in the hindlimb buds of Irx3/5-DKO;Shh+/- (DKO;Shh+/-) embryos are largely rescued comparing to Irx3/5-DKO hindlimb buds. Red markings in (A-F) denote expression domains of Gli1 and Gli3. The anterior Gli1-negative zone is restored in Irx3/5- DKO;Shh+/- hindlimb buds (C). (J-K) Cartilage structures in E14.5 control, Irx3/5 DKO and Irx3/5 DKO;Shh+/- hindlimbs. Red labeling mark the presence of tibia and digit 1 in Irx3/5 DKO;Shh+/- mutants. 35% of Irx3/5-DKO;Shh+/- mutant hindlimbs display partial rescue in anterior pattern formation (L), and 25% showed complete rescue of digit1 and tibia (M).

Figure 2.13 | Prepattern and limb bud size are not rescued in Irx3/5-DKO;Shh+/- hindlimbs

(A) Expression of Alx4 is not rescued in Irx3/5-DKO;Shh+/- (DKO;Shh+/-) hindlimb buds during limb bud initiation (28 to 29-somite stage). (B) Quantification of the frontal plane area of control, Irx3/5-DKO and Irx3/5-DKO;Shh+/- hindlimb buds at E10.75 (40-somite stage). The average area of mutant hindlimb buds are calculated in relation to control hindlimb buds (average set to 100%). Error bars represent standard deviations, p<0.05 for both mutants comparing to controls (two-tailed Student’s t-test). Reduction of limb bud size in Irx3/5- DKO;Shh+/- hindlimb buds is similar to that of Irx3/5-DKO mutants.

Figure 2.14 | A biphasic model of limb anterior pattern formation

The anterior population in the initiating limb bud is early specified by Irx3/5, which promote the anterior expression of Gli3 (left). However, their fate is not committed until later after Shh is activated. Shh signaling is inhibitory for anterior pattern formation, and a Shh-free zone (Gli1- negative) is required for the formation of digit 1 (middle). While the posterior elements (i.e. digit 2-5) are derived from Shh-expressing (autocrine Shh signaling) and Shh-responding (paracrine Shh signaling), formation of anterior elements (i.e. digit 1) requires early function of Irx3/5 and is inhibited by Shh signaling at alter stages (right). Numbers indicate digit primordial and skeleton. Light blue and dark blue denote Gli1- and Shh-expressing cells and their descendants, respectively.

Chapter 3 Limb bud morphogenesis requires regulation of cell proliferation by Irx3/5 in the anterior hindlimb field

In this work, OPT analysis was performed together with Gregory Anderson from the Henkelman lab. Olena Zhulyn helped with DNA content analysis using flow cytometry.

3 Chapter 3 3.1 Summary

In addition to gene expression, tissue shape has a profound impact upon limb pattern in a Turing type reaction-diffusion mechanism (Sheth et al., 2012). During limb development, anteriorly biased Irx3 and Irx5 regulate AP patterning gene expression and interact with Shh signaling to regulate pattern formation. However, the function of Irx3/5 in regulating limb bud volume and shape remains unclear. Here we show that Irx3/5-DKO hindlimb buds are small and display irregular shape, as indicated by 3D microscopy using optical projection tomography (OPT) technique. Double-pulse chasing experiments suggested that the anterior mesenchymal cells in initiating mutant hindlimbs proliferate slower than those of controls. In addition, time-lapse analysis using live imaging technique revealed multiple cell division defects in the anterior hindlimb field of Irx3/5-DKO at E9.75, including reduced division frequency and altered cell division plane. Interestingly, we observed anaphase bridge in some dividing cells in the mutant anterior hindlimb mesoderm, suggesting that Irx3/5 may play a role in chromosome segregation. Our data provide evidence for novel functions of Irx3/5 in regulating cell proliferation and oriented cell division, which contribute to the morphological defect of Irx3/5-DKO hindlimb buds. We propose that Irx3/5 function is required for proper expansion of anterior limb progenitors in addition to their roles in regulating AP patterning gene expression.

3.2 Introduction

Pattern formation of organ primordia is ultimately linked to tissue size and shape (McClure and McCune, 2003; Newman and Comper, 1990; Towers and Tickle, 2009). One prominent theory is the Turing model (also known as reaction-diffusion), which describes how two interacting chemicals diffusing through space could form interacting wave patterns (Kondo and Miura, 2010; Miura, 2013). A recent study provided evidence for a Turing-type mechanism in limb skeletal pattern formation that invokes Gli3 as a modulator of the system (Sheth et al., 2012). Many mutants with limb patterning defects also exhibit limb bud morphological anomalies during development, such as limb buds lacking Gli3 (Hui and Joyner, 1993; Litingtung et al., 2002), Alx4 (Kuijper et al., 2005), Shh (Chiang et al., 2001), AER-Fgfs (Sun et al., 2002), Fgfr1

(Li et al., 2005; Verheyden et al., 2005), Raldh2 (Niederreither et al., 2002), ect. However, how these genes regulate limb bud morphology has not been well established.

Cell proliferation and cell death are important processes regulating limb bud shape (Boehm et al., 2010; Montero and Hurle, 2010). Signals such as ZPA-Shh and AER-Fgfs function as not only patterning molecules but also mitogens for cell proliferation and/or survival, without which limb buds are small and lose skeletal elements (Lewandoski et al., 2000; Sun et al., 2002; Zhu et al., 2008). Recently, Gli3 has been shown to inhibit cell proliferation by restricting the expression of regulators of G1-S transition and constraining S-phase entry (Lopez-Rios et al., 2012). In Gli3-/- limb buds, both overgrowth of anterior mesenchyme and reduction of cell death were observed (Aoto et al., 2002; Lopez-Rios et al., 2012), which contribute to a wide autopod phenotype during limb bud outgrowth and later polydactyly. Cell proliferation for organ development is usually determined by two aspects: proportion of cells within cell cycle (proliferating cells) and cell cycle time. Single marker staining, such as pH3, Ki67 and BrdU labeling, can only provide information on relative rates of proliferation (proportion of cells proliferating), but not duration of the cell cycle. A recent study using double-pulse chasing technique indicated that cells in different regions of a maturing limb bud have different cell cycle times (Boehm et al., 2010).. However, cell cycle time in early limb buds and whether it contributes to bud shape (i.e. initiation stage) have not been examined.

Another key process controlling organ shape is directional cell behavior (Hopyan et al., 2011). This includes but not limited to oriented cell division (Baena-Lopez et al., 2005), cell intercalation (Walck-Shannon and Hardin, 2013) and collective cell movement (Solnica-Krezel, 2005; Zamir et al., 2006). Planar cell polarity (PCP) through non-canonical Wnt signaling has been speculated to play a crucial role in directional cell behavior for limb bud elongation (Boehm et al., 2010; Wyngaarden et al., 2010). How directional cell behavior, especially oriented cell division, affects limb initiation remains to be addressed.

Here we show that Irx3/5-DKO hindlimb buds display abnormal shape and are small in size. Using double-pulse chasing technique, we observed prolonged cell cycle time in anterior mesenchyme of the Irx3/5-DKO hindlimb field. In addition, time-lapse analysis revealed multiple cell division defects in the anterior mutant hindlimb field, such as reduced division

52 frequency, altered division plane and anaphase bridges. Together, these data suggest a novel role of Irx3/5 in regulating cell proliferation to generate proper shape for the initiating hindlimb bud.

3.3 Materials and Methods

3.3.1 Mice

Mice used in this study were housed in standard vented cages in conformity with the Toronto Center for Phenogenomics recommendations. For genotyping of the mice and embryos, earnotches, tail clips, or yolk sac were digested in 300ul of 50mM NaOH at 100°C for 10 minutes. NaOH was then neutralized by adding 100ul of 0.5M Tris (pH 8.0). 3ul of DNA solution was used for PCR genotyping. Mouse lines used in this study were as follows: Irx3/5- DKO (Irx3-5EGFP) (Zhang et al., 2011) and H2BGFP (CAG::H2B-EGFP) (Hadjantonakis and Papaioannou, 2004). Mice were maintained on a mixed outbred background of CD1 and 129/Sv.

3.3.2 OPT and limb bud morphology analysis

Embryos were dissected in cold PBS and fixed in 4% PFA overnight at 4°C . Optical projection tomography (OPT) was then performed essentially as described (Sharpe et al., 2002; Walls et al., 2007). Visualization and manipulation of OPT data was performed with Amira software and ImageJ. Segmentation of limb buds and quantification of shape information (volume and lengths along AP, PD and DV axes) were performed automatically, or when necessary, manually with Amira.

3.3.3 Double-pulse chasing analysis and cell cycle time estimation

Pregnant females were injected first administered with CldU by IP injection at E9.75 and then with IddU after 2.5hr. 30min post second injection, embryos were dissected in cold PBS and fixed with 4% PFA overnight at 4°C. Whole-mount immunofluorescence of CldU and IddU was performed as previously described (Tkatchenko, 2006) with some modifications. Fixed embryos were washed and treated with 5ug/ml ProK in PBT for 10min. After refixation, embryos were incubated in 2N HCl for 30min at 37°C. Then, embryos were blocked in 5% donkey serum in 0.2% PBS-T for 1-3hr and incubated with primary antibodies, Rat-anti-CldU (Abcam) and Mouse-anti-IddU (BD), at 4°C for two days. After extensive washing, secondary antibodies were applied overnight at 4°C. Before imaging with confocal microscopy, embryos were stained in DAPI solution for 20min. To calculate cell cycle time, 20X confocal images of the lateral view

53 of anterior and posterior hindlimb field were acquired as z-stacks of xy images taken for 20-30 z- slices at 1um apart. 200 DAPI positive mesenchymal nuclei were counted for each region, and the number of CldU-single positive nuclei (Lcells) was recorded. 6 hindlimbs were counted per genotype. Cell cycle time (Tc) was calculated using the equation: Tc = 2.5*(200/Lcells) (see 3.4.2 for details).

3.3.4 Flow cytometry

E10.5 limb buds were dissected from embryos in cold PBS and then treated with trypsin at 37°C to dissociate limb bud cells. PBS with 2% FBS was used to resuspend cell pellets after centrifugation. Filtered cells were then fixed with cold 80% EtOH overnight at 4°C. After removal of EtOH, fixed cells were washed, resuspended and stained with DAPI solution (Sigma) for 30min in dark. Stained cells were analyzed on a FACSCanto II (Becton-Dickinson) and data acquired using CellQuest (BD Biosciences). Data were analyzed with FlowJo (Tree Star) using the cell cycle platform.

3.3.5 Live imaging

Live imaging was performed as previously described (Wyngaarden et al., 2010) with subtle modifications. E9.75 to E10.0 (27 to 31-somite stage) control and Irx3/5-DKO embryos carrying the H2BGFP marker were dissected in DMEM containing 10% fetal calf serum at 37°C . Embryos were then submerged just below the surface of optimized pre-warmed media containing 50% DMEM and 50% rat serum. Cheese cloth and FastWell were used to position the hindlimb field directly against a coverslip at the bottom of a metallic confocal well to gain the lateral view, such that the entire depth of the tissue under study could be visualized. Time-lapse imaging experiments were performed for periods of up to 3 hours (5min/frame) in a humidified chamber at 37°C in 5% CO2. Laser-scanning confocal data were acquired using a Zeiss LSM 510 META microscope system and a LiveCell culture chamber (Neue Biosciences). GFP fluorophore was excited using a 488 nm argon laser. Only movies of healthy tissue were analyzed; the presence of any pyknotic nuclei, indicating unhealthy tissue, disqualified data sets from analysis. Confocal images were acquired as z-stacks of xy images taken at 2-5 um z intervals.

3.3.6 Orientation of cell division

Oriented cell divisions were analyzed as previously described (Wyngaarden et al., 2010) with subtle modifications. Time-lapse movies of the lateral view of hindlimb field with H2BGFP marker were inspected frame-by-frame using Volocity software (Improvision). Confocal z- stacks, together encompassing the entire depth of the hindlimb field imaged, were assessed individually. Tracing paper was used to draw a line precisely representing the metaphase-to- telophase transition of every single cell division identified. Angles between these lines and the AP limb axis as derived from low-magnification images were measured from 0-180° using a compass, with 0 representing the anterior end. Polar plots for cell division planes were generated use Rose.Net software. Segment length was used to denote the proportion of divisions within that range. This methodology is capable of detecting a bias in cell division in the hindlimb field along the AP and DV axes, but not the PD axis.

3.4 Results

3.4.1 Irx3/5-DKO hindlimb buds are small and display abnormal shape

Irx3/5-DKO hindlimb buds are small and exhibit abnormal shape. To better visualize their morphological defects and quantify shape, we employed Optical Projection Tomography (OPT) - - an imaging technology ideal for generating 3D renderings for embryonic tissues in the mm range (Coultas et al., 2010; Lickert et al., 2004; Sharpe et al., 2002; Walls et al., 2008). After reconstruction using Amira software, we generated 3D intensity isosurfaces of embryos (Figure 3.1, Video 3.1 and 3.2 and data not shown). As highlighted in Figure 3.1 (lower panel), some morphological defects of Irx3/5-DKO embryos are easily revealed from their 3D images. The mesencephalon and diencephalon region in Irx3/5-DKO embryo seems smaller than that of control littermates with a bumpy superior outline. The mutant first pharyngeal arch is hypoplastic, which may be related to the cranial facial phenotype at later stages (Kim et al., unpublished data). The mutant embryo also displays morphological anomaly of the heart, consistent with their known phenotype (Gaborit et al., 2012). As expected, mutant limb buds (both forelimb and hindlimb) were smaller than those of controls (Figure 3.2 and 3.3).

To quantify the shape of Irx3/5-DKO limb buds, I segmented (virtually dissected) E10.0 (~31s) and E10.5 (~37s) limb buds from the 3D images of embryos to create meshes of limb bud

55 isosurfaces (Figure 3.2, 3.3, Video 3.3-3.6 and data not shown). Limb bud volumes and lengths along three axes (AP, PD and DV) were measured for these two early stages (Table 3.1 and 3.2). Our data suggest that both forelimb and hindlimb buds of Irx3/5-DKO are smaller in size and shorter along the AP axis than that of controls at the same somite stage, suggesting early growth defect in the mutant bud. In addition, the shape of Irx3/5-DKO hindlimb bud is abnormal. At the late stage of hindlimb initiation (E10.0), the control hindlimb bud looks like a semi-ellipse with smooth distal outline (Figure 3.3A). In contrast, the mutant limb bud displays triangular shape, and its outline is flat in the anterior region (Figure 3.3B). These features are maintained at E10.5 (Figure 3.3D). In addition, the anterior region of mutant hindlimb buds is wider (along the DV axis) than that of controls during initiation (Figure 3.3A’ and B’, red arrow heads).

3.4.2 Prolonged cell cycle time in the anterior hindlimb field of Irx3/5- DKO embryos during limb initiation

Although cell death is increased in Irx3/5-DKO hindlimb buds, it is only detected at E10.5 after the limb bud morphology defect arises (Figure 2.2 and Figure 3.3). Thus, a cell proliferation defect at limb bud initiation stage likely contributes to the small size of Irx3/5-DKO hindlimb buds. Although the proportion of pH3-positive cells in mutant seems normal (Figure 2.2), another key aspect of cell proliferation, the cell cycle time, has not been determined. Previous studies established double-pulse chasing technique to estimate regional cell cycle time (Boehm et al., 2010; Martynoga et al., 2005). As illustrated in Figure 3.4A, pregnant females were first injected with CldU to label cells in S-phase at that time point in embryos. After a 2.5hr time interval (Ti), the pregnant mice were injected again with another Thymidine analog, IddU to label S-phase cells at the second time point. At this time, the CldU single positive cells are those that left S-phase during the 2.5hr Ti. These cells are called leaving cells, the number of which is

Lcells. Therefore, the ratio of the total number of cells (Totalcells) to Lcells equals the ratio of the total cell cycle (Tc) to Ti. Among these four factors, Totalcells and Lcells can be quantified by counting, and Ti is known. Thus, the average Tc for a region of cells can be calculated by the following equation:

To examine whether mesenchymal cells in Irx3/5-DKO hindlimb field experience longer cell cycle time than those of controls during limb bud initiation, I performed double-pulse chasing

56 experiments in E9.75 embryos. After cell counting, cell cycle times in the anterior and posterior hindlimb field of mutant and controls were calculated. As we expected, the average cell cycle time in the anterior region of Irx3/5-DKO hindlimbs is about 15% longer than that of controls (11.8hr vs. 10.2hr, p<0.001, Figure 3.4C). In contrast, posterior cell cycle times are similar between mutants and controls (Figure 3.4D). These cell cycle time data are comparable to a previous study of somewhat later limb buds (Boehm et al., 2010). My data suggest that the anterior population of the mutant hindlimb field proliferates slower than that of controls, while the posterior population is unaffected. This difference at least partly contributes to the small volume and abnormal shape of Irx3/5-DKO hindlimb buds.

To examine the potential function of Irx3/5 in regulating the cell cycle, we performed DNA content analysis using flow cytometry to estimate cell populations within each cell cycle phase. Irx3/5-DKO and control E10.5 limb buds were collected. After dissociation, filtration and DAPI staining, limb bud cells were analyzed using flow cytometry. Our preliminary data suggest that about 21.3% cells in Irx3/5-DKO limb buds are in G2/M phase, as compared to about 13.4% cells in control limb buds (Figure 3.4E). These data suggest that Irx3/5-DKO limb cells are likely stuck at G2/M phase, resulting in a longer cell cycle time than controls.

3.4.3 Live imaging data revealed multiple cell division defects in initiating Irx3/5-DKO hindimb buds

To investigate potential mitotic defects in Irx3/5-DKO hindlimb field, live imaging technique was applied to record behavior during cell division in the initiating hindlimb region of controls and mutants. H2BGFP is a good nuclear marker for live imaging using confocal microscopy (Hadjantonakis and Papaioannou, 2004). I collected E9.75-10.0 Irx3/5-DKO and control embryos carrying H2BGFP marker and generated real-time movies of the lateral view of initiating hindlimb fields across 4-5 cell layers. All the movies were then analyzed frame by frame to identify mitotic cells and angles of their division orientation (Figure 3.5A).

Live imaging using a low magnification (10X) allows us to observe cell division in the whole limb field. Less cell division was observed in the anterior hindlimb field of Irx3/5-DKO than in the posterior during the same period of time (~40% less), while there was no obvious difference between division frequencies of anterior and posterior hindlimb field in control embryos (data not shown). In addition, I also observed cell division plane defect in the anterior hindlimb field

57 of Irx3/5-DKO embryos. In control hindlimb field, directions of cell division in the anterior are mainly along both AP and DV axes, while in the posterior, cell divisions happened mostly along the DV axis. In contrast, both anterior and posterior populations of the mutant hindlimb field display DV biased division orientation (Figure 3.5C-F). This altered cell division orientation may result in a wider DV axis within the anterior hindlimb field in mutants (Figure 3.3A’ and B’).

Time-lapse movies with greater magnification (20X) revealed more cell division defects in mutant hindlimb buds. More than half of mitotic cells in the anterior hindlimb field of Irx3/5- DKO mutants exhibited M-phase defects, including anaphase bridge with lagging daughter cell separation, persistent anaphase bridge, and failed cell division. All of these defects were rare in control hindlimb mesenchyme (Figure 3.6 and Video 3.7-3.11). No obvious mitotic defect was observed in the overlying ectoderm in both mutant and control (Figure 3.6D). These data suggest that Irx3/5 may regulate cell division in the mesenchyme of hindlimb field in a cell autonomous manner. Failed cell division and unseparated daughter cells may undergo apoptosis observed at a later stage (E10.5 Figure 2.2D). In addition, cell division orientation is also altered in Irx3/5- DKO hindlimb mesenchyme (more cells divided along the DV axis) but not ectoderm (Figure 3.5G-J). Collectively, these data suggest that Irx3/5 play a role in cell division by regulating mitotic frequency and orientation, and chromosome segregation. Therefore the anterior hindlimb population requires Irx3/5 function for cell proliferation to expand anterior progenitors and generate the proper shape and volume of the hindlimb bud. This function ultimately should contribute to the formation of anterior skeletal elements.

3.5 Discussion

In this study, using OPT imaging technique, we quantified basic spatial parameters (volume, limb bud length, width and thickness) of limb buds during early developmental stages (E10.0 and E10.5). We identified multiple cell proliferation and cell division defects specifically in the anterior mesenchymal cells in Irx3/5-DKO hindlimb field, including prolonged cell cycle time, reduced division frequency, altered cell division plane and anaphase bridge. Thus, Irx3/5 are involved in cell proliferation in a cell autonomous manner to regulate limb bud shape and volume during hindlimb initiation. These data support the notion that Irx3/5 are required for the expansion of anterior progenitor population in the hindlimb field to give rise to anterior skeletal elements.

Oriented cell division contributes to tissue shape formation during development (Hopyan et al., 2011). Our data suggest that in the wild-type anterior hindlimb field, cells divide along both AP and DV axes, while in the posterior margin, cells mainly divide along the DV axis (Figure 3.5C and D). This is consistent with previous findings by Wyngaarden et al., who showed that during limb bud initiation, cells in the anterior lateral plate mesoderm move towards and contribute to the anterior limb bud, generating tissue movement along the AP limb axis, which they didn’t observe in the posterior margin of limb bud (Wyngaarden et al., 2010). It is conceivable that the AP biased tissue movement in the anterior limb field is partly due to oriented cell division. Many factors have been shown to regulate division orientation through tissue polarity, including chemical cues like morphogen gradients, mechanical forces and cell-cell contact, etc (Minc and Piel, 2012; Segalen and Bellaiche, 2009). Since Irx3/5 regulate AP polarity of the limb field (see Chapter 2), it is possible that Irx3/5 are involved in generating and/or transmitting polarity information to cells within the anterior hindlimb field, which is required for the AP biased cell division. In both anterior and posterior hindlimb field, a big portion of cells display DV biased division (Figure 3.5C and E). This may be regulated by the PCP pathway of the non-canonical Wnt signaling (Segalen and Bellaiche, 2009). Wnt7a, which is expressed in the dorsal ectoderm of the limb bud (Parr and McMahon, 1995), has been shown to function as non-canonical Wnt ligand to regulate PCP (Le Grand et al., 2009). It would be interesting to see whether Wnt7a regulates the DV biased cell division in hindlimb buds during initiation.

The anaphase bridge is another interesting cell division defect observed in the anterior population of Irx3/5-DKO hindlimb field (Figure 3.6). It is a marker of aberrant chromosome segregation (Hoffelder et al., 2004). At the beginning of anaphase, the anaphase promoting complex (APC) is activated, leading to the degradation of the chromatin-bound cohesin complex to promote separation of sister chromatids (Remeseiro et al., 2013). In addition, recent studies established that the cohesin complex also functions as transcription factors and regulate gene expression (Kagey et al., 2010; Rhodes et al., 2010). It is conceivable that Irx3/5 may bind to similar regions of chromatin and interact with the cohesin complex during mitosis to facilitate dissociation of cohesin from DNA to promote chromosome segregation. Another possibility is that Irx3/5 may play indirect roles by regulating expression of key factors involved in chromosome segregation.

Shape is an essential aspect of organ pattern formation. Factors determining tissue shape include cell proliferation, cell death, orientation of cell division and cell movement, etc (Boehm et al., 2010; Hopyan et al., 2011; Montero and Hurle, 2010; Strutt, 2005; Walck-Shannon and Hardin, 2013). Previous work illustrated collective cell movement/migration in the development of early forelimb buds (Wyngaarden et al., 2010). In the future, it would be interesting to see whether cell movement is also affected in the anterior hindlimb field of Irx3/5-DKO mutants.

Figure 3.1 | Surface renderings of E10.5 control and Irx3/5-DKO embryos viewed from the right

The lateral view of rendered surfaces of E10.5 control and Irx3/5-DKO embryos. Highlighted regions in the lower panels denote morphological defects in the mutant. (Red) The mutant mesencephalon (m) and diencephalon (d) region is smaller than that of the control and has a bumpy outline on the top. (Green) The first pharyngeal arch (1st pa) in Irx3/5-DKO is hypoplastic. (Purple) The mutant heart (h) seems to be mis-positioned/oriented. (Pink) The forelimb bud (fl) is smaller in the mutant than control.

Figure 3.2 | Dorsal and lateral views of forelimb bud isosurfaces at E10.0

E10.0 forelimb buds are virtually dissected from isosurfaces of control and mutant embryos. Axes in (A) and (A’) indicate directions of limb axes for each views (AP: anteroposterior, PD: proximodistal, DV: dorsoventral). The mutant forelimb bud is smaller than that of control at this stage. See also Table 3.1.

Table 3.1 | Shape information of E10.0 forelimb buds E10.0 (~31s) volume (10-3Xmm3) AP (mm) DV (mm) PD (mm) Control FL 51.6±0.6 0.67±0.05 0.33±0.02 0.40±0.01 Irx3/5-DKO FL 40.5±1.8 0.53±0.02 0.30±0.03 0.37±0.01

Table 3.2 | Shape information of E10.0 and E10.5 hindlimb buds

volume (10-3Xmm3) AP (mm) DV (mm) PD (mm) E10.0 Control HL 29.5 0.86±0.04 0.29±0.01 0.17±0.02 (~31s) Irx3/5-DKO HL 24.8 0.76±0.02 0.29±0.01 0.16±0.02 E10.5 Control HL 118.3±21.1 0.90±0.04 0.40±0.01 0.48±0.04 (~37s) Irx3/5-DKO HL 56.8±4.5 0.80±0.04 0.31±0.01 0.35±0.01

Figure 3.3 | Dorsal and lateral views of hindlimb bud isosurfaces at E10.0 and E10.5

E10.0 and E10.5 horelimb buds are virtually dissected from isosurfaces of control and mutant embryos. Axes in (A) and (A’) indicate directions of limb axes for each views (AP: anteroposterior, PD: proximodistal, DV: dorsoventral). (A-B’) At late limb initiation stage (E10.0, 31-somite stage), the mutant hindlimb bud is smaller and shorter along the AP axis than the control bud. The distal outline of anterior mutant bud is flat (blue arrow in B), making the mutant bud a triangle-like shape (B). In addition, in the lateral view, the anterior region of the mutant bud is wider (the distance between the two red arrowheads along the DV axis) than that of the control (A’ 0.20 mm and B’ 0.26mm). (C-D’) At E10.5 (37-somite stage) during limb bud outgrowth, the mutant hindlimb bud maintains its triangle-like shape (blue arrow in D) and small size. However, at this stage length of limb bud along the DV axis in mutant is shorter than control (C’ and D’). See also Table 3.2.

Figure 3.4 | Anterior hindlimb mesenchymal cells of Irx3/5-DKO mutants have longer cell cycle time than those of controls during hindlimb initiation

(A) Illustration of double-pulse chasing experiment: Pregnant females are first injected with CldU at the beginning of experiment to label cells in S-phase (left, red bar). After time interval Ti (2.5hr), mice are injected again, this time with IddU, to label cells in S-phase at that moment. 0.5hr later, when S-phase cells were all labeled by IddU (right, green bar), embryos were dissected and fixed for immunofluorescence. During the 2.5hr time interval, some of the CldU- labeled cells leave S-phase, so that they won't be labeled by IddU later. These cells are the

"Leaving cells" (Lcells) (right, red single positive), which are counted to calculate Tc. (B) An example of cell counting for Lcells. Lcells (red/CldU single positive cells) are marked with black dots within the view, blue dots for green (IddU) single positive cells and yellow for double positive cells. (C and D) Cell cycle time (Tc) of anterior (C) and posterior (D) mesenchymal cells in the hindlimb field at E9.75. Irx3/5-DKO mutants show longer cell cycle time in the anterior hindlimb field. (E) DNA content analysis using flow cytometry to estimate proportions of cells within each cell cycle phase. More cells are stuck at G2/M phase in mutant limb bud cells at E10.5.

Figure 3.5 | Oriented cell division is altered in initiating Irx3/5-DKO hindlimb field

(A) An example of hindlimb field cell division on successive frames in a time-lapse movie with H2BGFP marking nuclei. The time interval between two adjacent frames here is 10min. White arrow heads point to a dividing cell and its two daughter cells. The angle between the white line linking the two daughter cells and the limb AP axis was measured, representing the orientation of cell division (0°: anterior, 90°: dorsal). (B) A cartoon illustration of the hindlimb field for live imaging. Numbers indicate somite levels, and limb axes (AP and DV) are as shown. (C-F) Polar

66 plots of cell division orientation in the anterior (C and D) and posterior (E and F) hindlimb field of control (WT) and Irx3/5-DKO (DKO) embryos at E9.75-10.0. More cell divisions along the DV axis than the AP axis were observed in the anterior hindlimb field of Irx3/5-DKO mutants (D). (G-J) Polar plots of cell division orientation of cell division orientation in the ectoderm (G and H) and mesoderm (I and J) of the anterior hindlimb field in control (WT) and Irx3/5-DKO (DKO) embryos at E9.75. Most cell divisions along the DV axis than the AP axis were observed in the mesoderm of Irx3/5-DKO anterior hindlimb field (J). Number of cell division analyzed: (C) 122, (D) 69, (E) 52, (F) 56, (G) 38, (H) 20, (I) 38, (J) 31. (C and D are data from low and high magnification movies pooled together, D and F are only from low magnification movies, G- J are only from high magnification movies)

Figure 3.6 | Anterior hindlimb mesenchymal cells exhibit chromosome segregation defect during mitosis in Irx3/5-DKO mutants during limb initiation

(A-C) Successive frames of normal cell division (A), cell division with anaphase bridge and delayed chromosome segregation (Lagging) (B) and cell division with anaphase bridge persistent at telophase (Persistent) (C) from metaphase to telophase. All images are from time-lapse movies with high magnification (20X) of E9.75-10.0 anterior hindlimb field in mutant embryos. The time interval between frames is 5min. White arrow heads denote anaphase bridges. (D) Percentages of dividing cells with different division phenotypes.

Chapter 4 Conclusion and Future Experiments 4 Chapter 4 4.1 Thesis summary

This thesis summarizes my work on elucidating the roles of Irx3 and Irx5 in anterior limb pattern formation from two aspects. Chapter 2 describes novel functions of Irx3/5 in regulating limb AP patterning genes for the early specification of anterior limb skeletal elements which are later modulated by Shh signaling. In Chapter 3, I examined the roles of Irx3/5 in early limb bud morphogenesis at the cellular level. I showed that multiple defects in cell cycle and cell division in the anterior limb field contribute to the morphological defect in Irx3/5-DKO hindlimb buds. Overall these studies shed lights on the molecular mechanisms of anterior limb pattern formation and how gene expression and cell behavior can be coordinately regulated to generate specific patterns.

It is well established that Shh signaling plays a central role in the pattern formation of posterior limb skeletal elements (Ahn and Joyner, 2004; Bastida and Ros, 2008; Harfe et al., 2004; Zhu et al., 2008). In contrast, mechanisms governing anterior limb pattern formation remain unclear. Using mouse genetics, my studies in Chapter 2 indicated that Irx3/5 are novel anterior determinants in limb development. They regulate limb AP prepatterning prior to limb bud outgrowth to specify the anterior limb population which gives rise to the “Shh-independent” skeletal elements as well as preaxial polydactyly. Specifically, they bind to the limb enhancer and promote the expression of key AP prepatterning gene Gli3 in the anterior hindlimb field (Figure 2.14). They also genetically interact with Gli3 to establish the AER signaling center for limb bud outgrowth. Interestingly, we found that by reducing the dosage of Shh, anterior pattern formation can be restored in Irx3/5 deficient hindlimb buds. These data suggest that anterior limb pattern formation is regulated in two steps: 1) Irx3/5-dependent early specification, and 2) progressive determination inhibited by Shh signaling (Figure 2.14). In addition, the forelimb bud experiences lower level of Shh signaling with a bigger Gli1-negative zone in the anterior margin than that of the hindlimb bud, which may explain why the forelimb is more protective to the deleterious effect of Irx3/5.

In addition to disrupted expression of AP patterning genes, Irx3/5-DKO hindlimb buds also exhibit morphological anomalies. In Chapter 3, my studies were focused on understanding the roles of Irx3/5 in limb bud morphogenesis. Using OPT imaging technique, we were able to generate limb bud 3D isosurfaces and quantify the information of limb bud shape and size. These data indicate that Irx3/5-DKO limb buds are smaller than wildtype limb buds and display irregular shape especially in the anterior part. To understand cellular processes underlying the morphological defects of Irx3/5-DKO hindlimb buds, I examined cell proliferation and cell death during limb initiation since those defects arise early. However, pH3 staining and cleaved Caspase-3 staining revealed no obvious difference between mutant and control hindlimb buds at initiation stage, although increased cell death was detected in the proximal region of mutant hindlimb buds post-initiation (Chapter 2). To further characterize cell proliferation, I performed double-pulse chasing experiments and found that the cell cycle time in the mutant anterior hindlimb field is about 15% longer than that of controls during limb initiation. In addition, mesenchymal cells in the anterior hindlimb field of Irx3/5-DKO mutants displayed multiple mitotic defects, including altered cell division plane and chromosome segregation problem (i.e. anaphase bridge). Thus, without Irx3/5, the anterior population of the hindlimb field cannot proliferate and expand properly, resulting in defects of limb bud morphogenesis and eventually loss of anterior skeletal elements.

4.2 Irx3 and Irx5 are novel anterior limb determinants

Our data indicate that, relative to the Shh-promoting posterior elements, formation of the anterior limb skeletal elements requires early functions of Irx3/5. Firstly, Irx3/5 are essential for AP prepattern during limb bud initiation to specify the anterior population. Without Irx3/5, mutant limb buds are posteriorized with greatly reduced expression of anterior marker genes (e.g. Gli3 and Alx4). Secondly, Irx3/5 are required for proper cell cycle progression and oriented cell division in the anterior mesenchyme, which is involved in limb bud morphogenesis and possibly expansion of the anterior limb population (Chapter 3). However, many questions remain to be addressed. What are the molecular mechanisms underlying the anterior specification function of Irx3/5? Do the descendants of the Irx3/5-positive anterior cells give rise to the proximal and anterior structures affected in Irx3/5-DKO limbs? In limbs exhibiting anterior deficiency, is this anterior population completely gone or respecified to form structures with more posterior identity? Do Irx3/5 play a role in directional cell movement in limb bud morphogenesis to

70 expand the anterior population? In the following sections, I will discuss these questions and suggest new directions for future experiments.

4.2.1 Identifying the molecular mechanism of anterior specification by Irx3/5

As described in Chapter 2, our ChIP data identified the key AP patterning gene, Gli3, as a potential direct downstream target of Irx3/5 during limb prepattern stage. Interestingly, a previous ChIP-on-chip study by Vokes et al. identified Irx3 as a direct downstream target of Gli3 in E11.5 limb buds (Vokes et al., 2008). I found that Irx3 expression is greatly reduced in Gli3-/- limb buds at E10.5 but not initiation stage (Figure 4.1A). These data suggest that while Irx3 is required to promote Gli3 expression in the anterior limb field at early stage, high level of Gli3 is essential to maintain Irx3, and possibly Irx5, expression during limb bud outgrowth.

Gli3 is the key mediator of the Shh signaling in limb AP pattern formation, and it also have a Shh-independent early function during limb initiation to mark the anterior territory of the limb field (Litingtung et al., 2002; te Welscher et al., 2002a; te Welscher et al., 2002b; Wang et al., 2000). However, Irx3/5 may not act simply as upstream regulators of Gli3 to regulate limb anterior pattern formation because 1) Gli3 expression is not completely lost in Irx3/5-DKO hindlimbs and 2) Irx3/5 and Gli3 mutant phenotypes are not identical (i.e. oligodactyly vs. polydactyly). It is conceivable that Irx3/5 and Gli3 may have overlapping and distinct functions in the development of anterior limb structures. I speculate that the overlapping role between Irx3/5 and Gli3 is to specify anterior identity. This is based on the similarity between Irx3/5- DKO and Gli3-/- hindlimb phenotypes. Although Gli3-/- mice display polydactyly, the extra digits do not have anterior identity (Litingtung et al., 2002). Similar to Irx3/5-DKO, Gli3-/- hindlimbs show tibial deficiency as well (from tibia hypoplasia to loss of tibia). Marker analysis data also support this concept. Irx3/5-DKO and Gli3-/- limb buds are both posteriorized with expression of anterior genes largely decreased or lost (e.g. Alx4 and Pax9) (McGlinn et al., 2005; te Welscher et al., 2002a). The major difference between Irx3/5-DKO and Gli3-/- limb buds is their size. Our data establish that Irx3/5-DKO hindlimb buds are small (~50% of the control bud volume at E10.5), which is related to cell proliferation defects in the anterior hindlimb field and increased cell death at the base of the limb bud during and post limb initiation, respectively (Chapter 3). In contrast, Gli3-/- limb buds are bigger than those of controls as a result of decreased anterior cell death and differentiation and increased cell proliferation during limb bud outgrowth and cartilage

71 condensation (Aoto et al., 2002; Lopez-Rios et al., 2012). These data imply that Irx3/5 and Gli3 have distinct functions in the expansion of anterior limb population. In Irx3/5-DKO limb buds, the anterior population is not properly expanded from the beginning, while in Gli3-/- limb buds, this population is over expanded and gains posterior identity due to loss of suppression of posterior genes. (Figure 4.1B)

The fact that Gli3 expression is not completely lost in the initiating hindlimb bud of Irx3/5-DKO mutants suggests that other factors are involved in the regulation of Gli3 limb expression. Indeed, the CNE6 limb enhancer of Gli3 contains multiple potential binding sites of other transcription factors required for limb development, such as Pitx1/2, Hoxd11/13, Hand2 and Msx1 (Abbasi et al., 2010). In addition, it has been shown that Twist, whose expression pattern is similar to that of Irx3/5 during limb initiation, can also promote Gli3 limb expression (O'Rourke et al., 2002). Therefore, Irx3/5 may function together with Twist (and possibly other factors) regulating Gli3 expression in early limb buds. To further understand the molecular mechanism controlling expression of Gli3, it is necessary to identify co-factors of Irx3/5 and examine how Irx3/5 interact with them to regulate Gli3 expression during limb development. Analyses such as EMSA and luciferase assay should also be considered to confirm that Gli3 is indeed directly regulated by Irx3/5.

Future studies should also focus on identification of other downstream targets of Irx3/5 in the anterior limb population to gain insights into molecular mechanisms of anterior specification. We can use the advantage of the GFP signals in control (Irx3+5+/Irx3-5EGFP) and Irx3/5-DKO (Irx3-5EGFP/Irx3-5EGFP) embryos to collect the anterior limb bud cells by FACS. Microarray or RNASeq analyses can be performed to determine differentially expressed genes. ChIP- sequencing analysis can be used to reveal genes and regions that are directly bound by Irx3 and Irx5 in the anterior population, and together with the transcriptome data, direct targets of Irx3/5 can be determined. These experiments can also provide information on whether Irx3/5 function as transcriptional activators or repressors.

It has been speculated that the limb AP polarity might be transferred from the body axis (rostrocaudal) to the limb field (Zeller et al., 2009). It is tempting to hypothesize that Irx3/5 may be involved in this process. Hox genes are well established in controlling body plan along the rostrocaudal axis (Mallo et al., 2010). What are upstream regulators of Irx3/5? Is the anterior-

72 biased expression of Irx3/5 in the limb field regulated by some of the Hox genes? These questions remain to be addressed.

4.2.2 Fate mapping of Irx3/5-expressing cells may reveal the origin of anterior limb structures

As described in Chapter 3, Irx3/5 are required to promote expansion of the anterior population during limb initiation. Without Irx3/5, the anterior hindlimb cells experience less cell divisions, prolonged cell cycle time and disrupted cell division plane. It is conceivable that the Irx3/5- expressing anterior progenitors at limb initiation stage give rise to the proximal and anterior hindlimb structures, which are lost in Irx3/5-DKO mutants. To test this hypthesis, genetic fate mapping of descendants of Irx3/5-expressing cells in wild-type and mutant backgrounds is necessary.

In my Master study, I have generated Irx5CreERT2 gene targeting vectors, in which a tamoxifen inducible Cre sequence (CreERT2) with a polyA tail was knocked in to replace the sequence after the star codon in the first exon of Irx5. With this targeting vector, we can generate the mouse line expressing CreERT2 under the control of Irx5 promoter, and the resulting allele is Irx5-null at the same time. In addition, we can also use this vector to retarget the Irx3-null mouse ES cells to obtain the Irx3-Irx5CreERT2 allele, which is Irx3/5-null. Similar to previous fate mapping studies (Ahn and Joyner, 2004; Harfe et al., 2004), these mouse lines can be crossed to the Rosa26LacZ reporter line (Soriano, 1999) to track the descendants of Irx5-expressing cells at different developmental stages with tamoxifen injection. Since Irx3 and Irx5 show overlapping expression during limb development, we can use this strategy to fate-map the Irx3/5-expressing population.

It has been shown that the activity of CreERT2 can be detected in the nucleus within 6hr and persists for about 36hr with the peak activity at ~20-24hr following single tamoxifen injection (Ahn and Joyner, 2004; Hayashi and McMahon, 2002). Thus, tamoxifen treatment could be performed ~22hr prior to the stage of interest. The hindlimb phenotypes in Cre-ERTM;Irx3/5- CKO embryos suggest that Irx3/5 function is required early (during hindlimb initiation, before E10.5) for the formation of distal anterior skeletal elements and late (before E11.5) for the femur (see 2.4.1), and their function is still essential after E11.5 for development of the pelvis. Thus, I speculate that tamoxifen injection at early stage (i.e. E8.75) to label Irx3/5-expressing cells during hindlimb initiation will result in LacZ expression in digit1, tibia, femur and pelvis at

E14.5. When tamoxifen is given at a later stage (i.e. E9.5), part of the tibia, entire femur and pelvis may be labeled, but not digit 1. Injection at even later stages (i.e. after E10.5) probably only labels the pelvis. (Figure 4.2A) Alternatively, part of posterior limb elements may be also labeled as descendants of early Irx3/5-expressing cells (i.e. tamoxifen injection at E8.75), since Irx3/5 are expressed in a majority of limb bud cells during limb initiation. It is possible that at later stages some posterior Irx3/5-expressing cells turn off Irx3/5 expression and receive Shh signaling and contribute to posterior element formation. These experiments will answer two questions. First, which elements are derived from Irx3/5-expressing cells? (Do Irx3/5-expressing cells in the anterior hindlimb field directly give rise to the “Shh-independent” limb elements or even some posterior elements?) Second, are distal elements derived from early Irx3/5-expressing cells and do late Irx3/5-expressing cells only give rise to proximal elements? Whether the limb PD axis is specified early or progressively is debating (Tabin and Wolpert, 2007). If our expectation is correct, these fate mapping data together with the Cre-ERTM;Irx3/5-CKO hindlimb phenotypes will provide strong evidence supporting the notion that PD identity is specified in the early limb field, at least for Shh-independent progenitors of the hindlimb.

We could also use this fate-mapping strategy to examine where the Irx3/5-specified anterior population ends up with in mutant backgrounds. Firstly, since Irx3/5-dependent anterior population is required for the formation of preaxial polydactyly (see 2.4.4), I speculate that Irx3/5-expressing lineage during limb initiation gives rise to the anterior extra digits in polydactyly backgrounds (e.g. Gli3-/-, Kif7-/- and Gli3P1-4/+). Secondly, since the pattern of GFP signals, which reflects Irx5 promoter activity, is similar between Irx3/5-DKO (Irx3-5EGFP/Irx3- 5EGFP) and control (Irx3-5EGFP/+) embryos during hindlimb initiation (data not shown), we can generate Irx3-5EGFP/Irx3-5CreERT2;Rosa26LacZ embryos to track the anterior population (Irx3/5- expressing lineage) in Irx3/5-DKO limbs using LacZ staining post tamoxifen administration. Our data suggest that this anterior population is not expanded properly in early Irx3/5-DKO hindlimb buds (Chapter 3), since cells within this population have proliferation defects and increased apoptosis. I expect to observe LacZ positive cells overlapping with the increased apoptotic cells in the anterior-proximal region of the mutant hindlimb bud at E10.5. However, it is still unclear whether this anterior population is completely gone during hindlimb development in Irx3/5-DKO background. If some of these progenitors remain, what structures will they give rise to? There are at least three possibilities: 1) the anterior population is completely eliminated prior to

74 cartilage condensation, then no LacZ cells should be detected in the hindlimb of Irx3-5EGFP/Irx3- 5CreERT2;Rosa26LacZ at E14.5; 2) a small portion of the anterior population remains proximally, then we expect to detect some LacZ cells in the hypoplastic proximal elements (i.e. femur and pelvis); and 3) if the remaining anterior cells can still expand distally and are respecified to gain posterior fate, we will possibly see LacZ cells distributed distally and contributing to posterior elements (e.g. part of the fibula and digit 2). Furthermore, our biphasic model of limb anterior pattern formation suggests that in Irx3/5-DKO;Shh+/- hindlimbs, the anterior elements are respecified as a consequence of reduced Shh signaling after ZPA-Shh activation, even though the initial anterior population is not established properly (Chapter 2). The origin of these anterior distal structures is unknown. We can generate Irx3-5EGFP/Irx3-5CreERT2;Shh+/-;Rosa26LacZ embryos and use a similar fate mapping strategy to try to answer this question. One potential outcome is that LacZ cells are not detected in the restored digit 1 and tibia in these mutants, indicating that the rescued elements are from initial posterior population which gains anterior identity in a Shh-free environment. Alternatively, if the restored anterior elements are resulting from rescued expansion of the remaining anterior population, we expect to observe LacZ cells in the digit 1 and tibia. (Figure 4.2B)

4.3 Genetic interaction between Irx3/5 and Gli3 is required for signaling center establishment and limb outgrowth

In addition to their roles in anterior pattern formation, Irx3/5 and Gli3 have synergistic functions in the establishment of signaling centers (i.e. AER and ZPA). Our genetic data indicate that without Irx3/5 and Gli3, signaling centers cannot be established properly in the hindlimb, resulting in failed limb bud outgrowth. The AER signaling center is especially disrupted in Irx3/5;Gli3-TKO hindlimb buds with complete loss of Fgf8 expression (see 2.4.4). Other AER- Fgfs (i.e. Fgf4/9/17) are also compromised, since expression of Spry1, as a readout of total AER- Fgf signaling (Liu et al., 2003), is almost completely gone in the mesenchyme of Irx3/5;Gli3- TKO hindlimb buds (Figure 4.3B). Consistently, the stratified AER structure is not formed in Irx3/5;Gli3-TKO hindlimb buds (Figure 4.3A).

Since Irx3/5 and Gli3 are expressed in the limb bud mesenchyme, an immediate question is how the mesenchymal transcription factors regulate AER formation and expression of Fgfs in the overlying ectoderm. One possibility is that Irx3/5 and Gli3 regulate expression of a signalling

75 molecule in the limb mesenchyme which signals to the ectoderm to trigger AER formation. Fgf10 secreted from the mesoderm of limb field acts through its receptor, FgfR2IIIb, in the overlying ectoderm to trigger Wnt3/β-catenin cascade to activate Fgf8 expression and AER formation (Kawakami et al., 2001; Xu et al., 1999; Xu et al., 1998). In addition, Bmp4 signaling from the limb field LPM is also necessary for AER induction (Benazet et al., 2009), possibly through its receptor BmpRIA in the limb ectoderm (Barrow et al., 2003). However, despite largely compromised at E10.5, Fgf10 and Bmp4 expressions are still expressed in Irx3/5;Gli3- TKO hindlimb field during limb initiation (Figure 4.3C and data not shown). Expression of the downstream target of Bmp signaling, Msx2, also appears normal in the TKO hindlimb field (data not shown), suggesting that the limb ectoderm receives and responds to Bmp signaling. Although Fgf10 expression is present in the mesoderm of limb field, it is unclear whether the limb ectoderm receives Fgf10 signaling properly (e.g. Fgf10 protein level is disrupted / Fgf10 is not secreted to the ectoderm). Wnt3 expression could be examined as readout of the Fgf10 signaling cascade in the limb ectoderm to determine whether the signal is transmitted to the ectoderm.

Classic chick experiments using recombinant limb discovered that posterior limb bud mesenchymal cells (especially the ZPA cells) inhibit outgrowth and the AER structure (Crosby and Fallon, 1975; Frederick and Fallon, 1983). A recent study from Zhulyn et al. established that the anterior limb progenitors are required for induction of limb organizers (i.e. ZPA and AER) (Zhulyn et al., 2014). This might explain the hindlimb phenotype of Irx3/5;Gli3-TKO. Irx3/5 and Gli3 are key anterior determinants during limb initiation, but Irx3/5-DKO and Gli3-/- limb buds are not completely posteriorized, and there is still some AP polarity in those mutant limbs. For example, in Irx3/5-DKO hindlimb buds, anterior markers (i.e. Gli3 and Alx4) are still expressed despite at low level, and posterior genes (i.e. Hand2 and Tbx2) are not completely expanded (see 2.4.2). In Gli3-/- limbs, Hand2 and Tbx2 still display posterior-biased expression (Figure 4.4), and ectopic Gli1 expression in the anterior limb bud is activated later and in a much smaller domain than its posterior expression induced by the ZPA (Lopez-Rios et al., 2012). Indeed, Irx3/5;Gli3-TKO hindlimb buds appear to be more posteriorized than those of Irx3/5-DKO and Gli3-/- during limb bud initiation. The hindlimb bud of Irx3/5;Gli3-TKO embryos show widespread expression of the posterior markers Hand2 and Tbx2 (Figure 4.4). In addition, the ZPA-Shh signaling center is established in the middle of the distal limb bud mesenchyme (see 2.4.3). These data suggest that the AP polarity is lost in Irx3/5;Gli3-TKO hindlimb buds, and the

76 mutant limb bud is completely posteriorized during limb initiation. It is possible that expression of other anterior marker genes, such as Alx4, is completely missing in the TKO hindlimb bud. If this is true, then without the key anterior determinants (Irx3/5 and Gli3), the entire anterior progenitor population is missing, which is inhibitory for signaling center induction. To test this hypothesis, additional marker analysis of AP patterning genes needs to be done in Irx3/5;Gli3- TKO mutants.

Key questions still remain on the nature of the interaction between Irx3/5 and Gli3 in controlling signaling center establishment. Do Irx3/5 form complexes with Gli3 to regulate gene expression? What are their downstream targets involved during limb initiation? Why complete posteriorization is inhibitory for signaling center establishment, especially AER induction? Answers to these questions will provide new knowledge on how limb mesoderm influences the overlying ectoderm to trigger formation of the AER.

4.4 Early specification and progressive modulation of anterior limb elements by Irx3/5 and Shh signaling

It has been long debating that whether positional information along the PD axis is specified early or progressively over time during limb development (Tabin and Wolpert, 2007). Extensive studies have been carried out, and the “two-signal” and “progress zone” models have been proposed. It is less known about when limb elements are specified along the AP axis. Work from Zhu et al. suggests that primordia of posterior digits (digit 2 to 5) are specified transiently by Shh signaling (within ~12h post Shh activation) which promotes their expansion at later stages (Zhu et al., 2008). However, how digit 1 and other anterior limb elements are specified is unclear. Our data suggest that the anterior limb elements (digit 1 and radius/tibia) are specified early by Irx3/5 and determined progressively by Shh signaling (Figure 2.14). Without Irx3/5, limb AP prepattern is disrupted with reduced anterior gene expression (see 2.4.2), and the Irx3/5-expressing anterior progenitor cells cannot expand properly (Chapter 3). As a consequence, after Shh activation, the entire mutant limb bud mesenchyme displays Shh pathway activity as revealed by Gli1 expression (see 2.4.2). It is conceivable that high level of Shh signaling is inhibitory for limb anterior pattern formation, since Shh pathway activation inhibits anterior gene expression and Gli3 repressor formation, and also promotes expression of posterior patterning gene such as 5’ Hoxd genes (Litingtung et al., 2002; te Welscher et al., 2002b). In agreement with this concept,

77 ectopic Shh activation in the anterior margin of limb bud could result in conversion of the biphalangial thumb to a triphalangial posterior finger (Anderson et al., 2012). The hindlimb phenotype in our Irx3/5-DKO;Shh+/- mouse model provides genetic evidence supporting the concept that Shh signaling inhibits anterior identity and formation of anterior structures. Furthermore, these data suggest that anterior pattern formation is plastic. Even the anterior progenitor population is not properly specified and expanded during limb initiation, it can be re- established in a Shh-free environment (Gli1-negative) at later stages as long as sufficient Gli3 repressor is present.

However, there are still questions on how the anterior elements are rescued in Irx3/5-DKO;Shh+/- hindlimbs. Similar to Irx3/5-DKO hindlimb buds, Irx3/5-DKO;Shh+/- buds also exhibit morphological defects with small size and abnormal shape at E10.5, suggesting less limb progenitor cells (see 2.4.8). One possibility is that proliferation in the anterior Gli1-negative region is increased at later stages (e.g. after E10.5) probably controlled by rescued expression of anterior genes. Thus, it is important to obtain shape information and examine cell proliferation in Irx3/5-DKO;Shh+/- hindlimb buds. Another possibility is that the amount of limb bud progenitor cells is not rescued, but the digit period (distance between each digit) is decreased. Work from Seth et al. suggested that digit period along the AP axis is controlled by a self-organizing Turing- type mechanism (Sheth et al., 2012). It is possible that altering Shh dosage in a sensitive background (i.e. Irx3/5-DKO) may result in change of digit period. To test this, we could use Sox9 staining to mark digit condensation and measure digit period.

4.5 Difference between forelimb and hindlimb development

Forelimb and hindlimb are different in many aspects such as function, structure and gene expression. During development, some genes are differentially expressed in forelimb and hindlimb buds, including forelimb-specific Tbx5 as well as hindlimb-specific Tbx4 and Pitx1, which are thought to promote limb initiation and determine limb identities (Agarwal, 2003; DeLaurier et al., 2006; Marcil, 2003; Naiche and Papaioannou, 2007; Takeuchi, 2003). In addition, some genes involved in both forelimb and hindlimb development display different expression levels (Shou et al., 2005). Our Gli1 in situ and qRT-PCR data suggest that Shh pathway is regulated differently in forelimb and hindlimb buds during comparable stages of limb bud development (see 2.4.7). The bigger Shh-free zone (Gli1-negative) observed in the anterior

78 margin of the forelimb bud relative to the hindlimb may explain why the forelimb is more protective to the deleterious effect of Irx3/5. We can perform genetic fate mapping experiments (as described in 4.2.2) to test this hypothesis. I speculate that the fate mapping result of Irx3/5- DKO forelimb is similar to that of Irx3/5-DKO;Shh+/- hindlimb (Figure 4.2B). Previous study by Zhu et al. indicates that Shh is required in an early transient time window (~12hr) to specify identities of digit 2 to 5 (Zhu et al., 2008). Interestingly, our data revealed that the forelimb and hindlimb difference in Shh pathway activity can only be detected in that time window and seems lost at later stages (Figure 2.11 and data not shown), suggesting that this difference is mainly involved in pattern formation rather than progenitor expansion and outgrowth. The key question now is why the Shh pathway is differentially regulated in forelimb and hindlimb development. We showed that the position of ZPA-Shh domain is more posteriorly restricted in the forelimb at early stages than that of the hindlimb (see 2.4.7). Posterior prepattern genes, including Hand2, Tbx2 and 5’ Hox genes, have been shown to regulate ZPA-Shh. Whether expression of these genes is more posteriorly restricted in initiating forelimb bud relative to the hindlimb needs to be determined. Moreover, we also observed higher Hhip level in the forelimb relative to the hindlimb using qRT-PCR (Figure 4.5). This is consistent with a previous gene profile analysis in autopod development at later stage (Shou et al., 2005). Hhip is a potent negative regulator of the Hh pathway by binding to Hh ligands at the cell membrane (Jiang and Hui, 2008). High level of Hhip expression may be responsible for the low Shh pathway activity in the forelimb. However, why Hhip expression is differently regulated during forelimb and hindlimb development remains to be addressed.

4.6 Irx3/5 regulate limb bud morphogenesis in multiple aspects

As described in Chapter 3, Irx3/5 regulate multiple aspects of cell division during limb initiation. I showed that anterior progenitor cells in the hindlimb field of Irx3/5-DKO exhibit reduced division frequency, prolonged cell cycle time, altered cell division plane and chromosome segregation defects. In addition, we also observed increased cell death in the anterior proximal region of limb bud mesenchyme in Irx3/5-DKO hindlimb at later stages. Because of these defects, Irx3/5-DKO limb buds are small with abnormal shape resulting in loss of limb structures. Previous work identified Tbx4 as a key regulator to promote hindlimb initiation (Naiche and Papaioannou, 2007). Interestingly, Tbx4 expression is downregulated in the anterior region of the hindlimb field in Irx3/5-DKO embryos (see 2.4.2), which may contribute to the

79 anterior outgrowth defect. However, molecular mechanisms underlying Irx3/5 regulating cell division remain to be addressed. Our DNA content analysis suggests that Irx3/5-DKO limb bud cells are stuck at G2/M phase. To confirm this and better understand the cell cycle phenotype, it is necessary to examine other cell cycle markers in Irx3/5-DKO limb buds. Different types of Cyclins bind to and activate corresponding Cdk kinases at different cell cycle phases to promote cell cycle progression (Lim and Kaldis, 2013; Malumbres and Barbacid, 2009). For example, Cyclin B can be used as a G2/M marker. It forms complex with Cdk1 and is accumulated during G2 and degraded as cells exit from mitosis (at the end of the M-phase). I expect that Cyclin B protein level is higher in Irx3/5-DKO limb buds than controls during early stages of limb development. Moreover, the live imaging data revealed chromosome segregation defects during mitosis in the anterior region of Irx3/5-DKO hindlimb field. It is important in future study to understand molecular mechanism for Irx3/5 regulating chromosome segregation. The cohesin complex and its regulated factors have been shown to participate in segregation of sister chromatids during anaphase (Remeseiro et al., 2013). One possibility is that Irx3/5 may interact with these proteins to facilitate chromosome segregation.

Our data suggest that expansion of the anterior progenitor population is affected in Irx3/5-DKO limb buds. We can test this hypothesis using clonal analysis as described previously (Arques et al., 2007). In brief, Cre-ERTM can be induced using a low dose of tamoxifen to randomly label cells at single cell resolution in embryos carrying the Rosa26LacZ or other reporter allele. According to our hypothesis, we would expect that the distribution of single-cell descendants in the anterior mesenchyme of Irx3/5-DKO limb buds is less spread out during limb initiation than in the posterior part and control background (Figure 4.6). This experiment can also provide information on cell mobility, which reflects cell movement. In addition, it is shown that directions of Golgi body (relative to nucleus) and extension of filophodia are related to oriented cell migration (Boehm et al., 2010). We can also examine these factors during early stages of limb development to determine whether Irx3/5 regulate oriented cell movement.

4.7 Conclusion remarks

Work presented in this thesis has provided new insight into how anterior pattern formation is regulated in limb development. Analysis using mouse genetics identified Irx3/5 as early determinants of the anterior population by regulating AP parepattern in the limb field, whereas

80 later Shh signaling gradually modulates the anterior pattern. In addition, expansion of the early anterior population requires Irx3/5 function for proper cell cycle progression and oriented cell division. These findings highlight key events in early limb pattern formation and morphogenesis, which will be helpful for future regenerative studies.

Figure 4.1 | Overlapping and distinct functions of Irx3/5 and Gli3 in anterior pattern formation during limb development

(A) Irx3 expression in Gli3-/- hindlimb buds is not greatly altered during initiation (31-somite stage) but decreased in the anterior mesenchyme at E10.5 (37-somite stage). (B) Irx3/5 and Gli3 are both required for anterior identity, whereas their functions in the expansion of anterior population are different.

Figure 4.2 | Genetic fate mapping of Irx3/5-specified anterior population in limb development

(A) Schematic illustration of Irx3/5 expression pattern (red) and prediction of fate-mapping of descendants of Irx3/5-expressing cells (blue) labeled upon tamoxifen administration at indicated stages. (B) Potential outcomes (1-3) of fate mapping study to track descendants of Irx3/5- specified anterior population (blue) in Irx3/5-DKO background (Irx3-5EGFP/Irx3- 5CreERT2;Rosa26LacZ) and Irx3/5-DKO;Shh+/- background (Irx3-5EGFP/Irx3-5CreERT2;Shh+/- ;Rosa26LacZ) with tamoxifen treatment at E8.75. Red dots represent increased cell death at E10.5.

Figure 4.3 | AER-related defects in Irx3/5;Gli3-TKO hindlimb buds

(A) Eosin staining of E11.5 hindlimb bud sections. Black arrowheads indicate the stratified AER structure, which is lost in Irx3/5;Gli3-TKO. (B) Spry1 expression in E10.5 (~38s) hindlimb buds. (C) Fgf10 expression in the hindlimb field at E9.75 (28-29s).

Figure 4.4 | Hand2 and Tbx2 expression in Irx3/5;Gli3-TKO hindlimb buds

Hand2 (37-38s) and Tbx2 (33-34s) expression are more anteriorly expanded in Irx3/5;Gli3-TKO (TKO) hindlimb buds relative to Irx3/5-DKO (DKO) and Gli3-/-.

Figure 4.5 | Hhip expression level in the hindlimb is lower than that of the forelimb at comparable developmental stage qRT-PCR of Hhip in 31-32s forelimb buds and 36-37s hindlimb buds. Expression data are normalized to Gapdh. The relative level of Hhip in forelimb is set as 100%. Asterisk indicates p<0.02.

Figure 4.6 | Clonal expansion analysis

Using low dose of tamoxifen treatment, single cell clone can be tracked in Cre-ERTM;R26R embryos. Blue and red dots represent cells from a single clone in the anterior and posterior limb bud mesenchyme respectively.

References

Abbasi, A.A., Paparidis, Z., Malik, S., Bangs, F., Schmidt, A., Koch, S., Lopez-Rios, J., and Grzeschik, K.H. (2010). Human intronic enhancers control distinct sub-domains of Gli3 expression during mouse CNS and limb development. BMC developmental biology 10, 44.

Agarwal, P. (2003). Tbx5 is essential for forelimb bud initiation following patterning of the limb field in the mouse embryo. Development 130, 623-633.

Ahn, S., and Joyner, A.L. (2004). Dynamic changes in the response of cells to positive hedgehog signaling during mouse limb patterning. Cell 118, 505-516.

Alarcón, P., Rodríguez-Seguel, E., Fernández-González, A., Rubio, R., and Gómez-Skarmeta, J.L. (2008). A dual requirement for Iroquois genes during Xenopus kidney development. Development 135, 3197-3207.

Amano, T., Sagai, T., Tanabe, H., Mizushina, Y., Nakazawa, H., and Shiroishi, T. (2009). Chromosomal dynamics at the Shh locus: limb bud-specific differential regulation of competence and active transcription. Developmental cell 16, 47-57.

Anderson, E., Peluso, S., Lettice, L.A., and Hill, R.E. (2012). Human limb abnormalities caused by disruption of hedgehog signaling. Trends in genetics : TIG 28, 364-373.

Aoto, K., Nishimura, T., Eto, K., and Motoyama, J. (2002). Mouse GLI3 Regulates Fgf8 Expression and Apoptosis in the Developing Neural Tube, Face, and Limb Bud. Developmental biology 251, 320-332.

Arques, C.G., Doohan, R., Sharpe, J., and Torres, M. (2007). Cell tracing reveals a dorsoventral lineage restriction plane in the mouse limb bud mesenchyme. Development 134, 3713-3722.

Baena-Lopez, L.A., Baonza, A., and Garcia-Bellido, A. (2005). The orientation of cell divisions determines the shape of Drosophila organs. Current biology : CB 15, 1640-1644.

Bao, Z.Z., Bruneau, B.G., Seidman, J.G., Seidman, C.E., and Cepko, C.L. (1999). Regulation of Chamber-Specific Gene Expression in the Developing Heart by Irx4. Science 283, 1161-1164.

Barna, M., Pandolfi, P.P., and Niswander, L. (2005). Gli3 and Plzf cooperate in proximal limb patterning at early stages of limb development. Nature 436, 277-281.

Barrow, J.R., Thomas, K.R., Boussadia-Zahui, O., Moore, R., Kemler, R., Capecchi, M.R., and McMahon, A.P. (2003). Ectodermal Wnt3/beta-catenin signaling is required for the establishment and maintenance of the apical ectodermal ridge. Genes & development 17, 394- 409.

Bastida, M.F., and Ros, M.A. (2008). How do we get a perfect complement of digits? Current opinion in genetics & development 18, 374-380.

Benazet, J.D., Bischofberger, M., Tiecke, E., Goncalves, A., Martin, J.F., Zuniga, A., Naef, F., and Zeller, R. (2009). A self-regulatory system of interlinked signaling feedback loops controls mouse limb patterning. Science 323, 1050-1053.

Berger, M.F., Badis, G., Gehrke, A.R., Talukder, S., Philippakis, A.A., Pena-Castillo, L., Alleyne, T.M., Mnaimneh, S., Botvinnik, O.B., Chan, E.T., et al. (2008). Variation in homeodomain DNA binding revealed by high-resolution analysis of sequence preferences. Cell 133, 1266-1276.

Bi, W., Deng, J.M., Zhang, Z., Behringer, R.R., and de Crombrugghe, B. (1999). Sox9 is required for cartilage formation. Nat Genet 22, 85-89.

Bilioni, A., Craig, G., Hill, C., and McNeill, H. (2005). Iroquois transcription factors recognize a unique motif to mediate transcriptional repression in vivo. Proc Natl Acad Sci U S A 102, 14671-14676.

Boehm, B., Westerberg, H., Lesnicar-Pucko, G., Raja, S., Rautschka, M., Cotterell, J., Swoger, J., and Sharpe, J. (2010). The Role of Spatially Controlled Cell Proliferation in Limb Bud Morphogenesis. PLoS Biol 8, e1000420.

Bonnard, C., Strobl, A.C., Shboul, M., Lee, H., Merriman, B., Nelson, S.F., Ababneh, O.H., Uz, E., Guran, T., Kayserili, H., et al. (2012). Mutations in IRX5 impair craniofacial development and germ cell migration via SDF1. Nat Genet 44, 709-713.

Bosse, A., Stoykova, A., Nieselt-Struwe, K., Chowdhury, K., Copeland, N.G., Jenkins, N.A., and Gruss, P. (2000). Identification of a novel mouse Iroquois homeobox gene, Irx5, and chromosomal localization of all members of the mouse Iroquois gene family. Developmental dynamics : an official publication of the American Association of Anatomists 218, 160-174.

Briscoe, J., Pierani, A., Jessell, T.M., and Ericson, J. (2000). A homeodomain protein code specifies progenitor cell identity and neuronal fate in the ventral neural tube. Cell 101, 435- 445.6.

Burke, A. C., Nelson, C. E., Morgan, B. A., Tabin, C. (1995). Hox genes and the evolution of vertebrate axial morphology. Development. 121, 333–46.

Cao, T., Wang, C., Yang, M., Wu, C., and Wang, B. (2013). Mouse limbs expressing only the Gli3 repressor resemble those of Sonic hedgehog mutants. Developmental biology 379, 221-228.

Cavodeassi, F., Diez del Corral, R., Campuzano, S. and Domínguez, M. (1999). Compartments and organising boundaries in the Drosophila eye: the role of the homeodomain Iroquois proteins. Development 126, 4933-4942.

Cavodeassi, F., Modolell, J., and Gómez-Skarmeta, J.L. (2001). The Iroquois family of genes: from body building to neural patterning. Development 128, 2847-2855.

Chapman, D.L., Garvey, N., Hancock, S., Alexiou, M., Agulnik, S.I., Gibson-Brown, J.J., Cebra- Thomas, J., Bollag, R.J., Silver, L.M., and Papaioannou, V.E. (1996). Expression of the T-box

86 family genes, Tbx1-Tbx5, during early mouse development. Developmental dynamics : an official publication of the American Association of Anatomists 206, 379-390.

Chapman, D.L., Garvey, N., Hancock, S., Alexiou, M., Agulnik, S.I., Gibson-Brown, J.J., Cebra- Thomas, J., Bollag, R.J., Silver, L.M., and Papaioannou, V.E. (1999). Expression of the T-box family genes, Tbx1-Tbx5, during early mouse development. Developmental dynamics : an official publication of the American Association of Anatomists 206, 379-390.

Charité, J., McFadden, D.G., and Olson, E.N. (2000). The bHLH transcription factor dHAND controls Sonic hedgehog expression and establishment of the zone of polarizing activity during limb development. Development 127, 2461-2470.

Cheng, C.W., Chow, R.L., Lebel, M., Sakuma, R., Cheung, H.O., Thanabalasingham, V., Zhang, X., Bruneau, B.G., Birch, D.G., Hui, C.C., et al. (2005). The Iroquois homeobox gene, Irx5, is required for retinal cone bipolar cell development. Developmental biology 287, 48-60.

Cheung, H.O., Zhang, X., Ribeiro, A., Mo, R., Makino, S., Puviindran, V., Law, K.K., Briscoe, J., and Hui, C.C. (2009). The kinesin protein Kif7 is a critical regulator of Gli transcription factors in mammalian hedgehog signaling. Science signaling 2, ra29.

Chiang, C., Litingtung, Y., Harris, M.P., Simandl, B.K., Li, Y., Beachy, P.A., and Fallon, J.F. (2001). Manifestation of the limb prepattern: limb development in the absence of sonic hedgehog function. Developmental biology 236, 421-435.

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

Cohn, M. J., Patel, K., Krumlauf, R., Wilkinson, D. G., Clarke, J. D., Tickle, C. (1997). Hox9 genes and vertebrate limb specification. Nature. 387, 97–101.

Cohen, D.R., Cheng, C.W., Cheng, S.H., and Hui, C.C. (2000). Expression of two novel mouse Iroquois homeobox genes during neurogenesis. Mech Dev 91, 317-321.

Cooper, K.L., Hu, J.K., ten Berge, D., Fernandez-Teran, M., Ros, M.A., and Tabin, C.J. (2011). Initiation of proximal-distal patterning in the vertebrate limb by signals and growth. Science 332, 1083-1086.

Costantini, D.L., Arruda, E.P., Agarwal, P., Kim, K.H., Zhu, Y., Zhu, W., Lebel, M., Cheng, C.W., Park, C.Y., Pierce, S.A., et al. (2005). The homeodomain transcription factor Irx5 establishes the mouse cardiac ventricular repolarization gradient. Cell 123, 347-358.

Coultas, L., Nieuwenhuis, E., Anderson, G.A., Cabezas, J., Nagy, A., Henkelman, R.M., Hui, C.C., and Rossant, J. (2010). Hedgehog regulates distinct vascular patterning events through VEGF-dependent and -independent mechanisms. Blood 116, 653-660.

Crosby, G.M., and Fallon, J.F. (1975). Inhibitory effect on limb morphogenesisi by cells of the polarizing zone coaggregated with pre- or postaxial wing bud mesoderm. Developmental biology 46, 28-39.

Crossley, P. H., Minowada, G., MacArthur, C. A., Martin, G. R. (1996). Roles for FGF8 in the induction, initiation, and maintenance of chick limb development. Cell. 84, 127–36.

Cunningham, T.J., Chatzi, C., Sandell, L.L., Trainor, P.A., and Duester, G. (2011). Rdh10 mutants deficient in limb field retinoic acid signaling exhibit normal limb patterning but display interdigital webbing. Developmental dynamics : an official publication of the American Association of Anatomists 240, 1142-1150.

Cunningham, T.J., Zhao, X., Sandell, L.L., Evans, S.M., Trainor, P.A., and Duester, G. (2013). Antagonism between retinoic acid and fibroblast growth factor signaling during limb development. Cell reports 3, 1503-1511.

Dambly-Chaudière., C., and Leyns, L. (1992). The determination of sense organs in Drosophila a search for interacting genes. The International journal of developmental biology 36, 85-91.

Davenport, T.G. (2003). Mammary gland, limb and yolk sac defects in mice lacking Tbx3, the gene mutated in human ulnar mammary syndrome. Development 130, 2263-2273.

Davis, C. A., Joyner, A. L.iverge during mouse development. Genes Dev. 2, 1736–44. de la Calle-Mustienes, E., Feijoo, C.G., Manzanares, M., Tena, J.J., Rodriguez-Seguel, E., Letizia, A., Allende, M.L., and Gomez-Skarmeta, J.L. (2005). A functional survey of the enhancer activity of conserved non-coding sequences from vertebrate Iroquois cluster gene deserts. Genome research 15, 1061-1072.

DeLaurier, A., Schweitzer, R., and Logan, M. (2006). Pitx1 determines the morphology of muscle, tendon, and bones of the hindlimb. Developmental biology 299, 22-34.

Duboc, V., and Logan, M.P. (2011). Regulation of limb bud initiation and limb-type morphology. Developmental dynamics : an official publication of the American Association of Anatomists 240, 1017-1027.

Dudley, A.T., Ros, M.A., and Tabin, C.J. (2002). A re-examination of proximodistal patterning during vertebrate limb development. Nature 418, 539-544.

Ede, D. A., Law, J. T. (1969). Computer simulation of vertebrate limb morphogenesis. Nature 221, 244–248.

Endoh-Yamagami, S., Evangelista, M., Wilson, D., Wen, X., Theunissen, J.W., Phamluong, K., Davis, M., Scales, S.J., Solloway, M.J., de Sauvage, F.J., et al. (2009). The mammalian Cos2 homolog Kif7 plays an essential role in modulating Hh signal transduction during development. Current biology : CB 19, 1320-1326.

Fallon, J. F., López, A., Ros, M. A., Savage, M. P., Olwin, B. B., Simandl, B. K. (1994). FGF-2: apical ectodermal ridge growth factor for chick limb development. Science. 264, 104–107.

Fernandez-Teran, M., Piedra, M.E., Kathiriya, I.S., Srivastava, D., Rodriguez-Rey, J.C., and Ros, M.A. (2000). Role of dHAND in the anterior-posterior polarization of the limb bud: implications for the Sonic hedgehog pathway. Development 127, 2133-2142.

Fernandez-Teran, M.A., Hinchliffe, J.R., and Ros, M.A. (2006). Birth and death of cells in limb development: a mapping study. Developmental dynamics : an official publication of the American Association of Anatomists 235, 2521-2537.

Frederick, J.M., and Fallon, J.F. (1983). The proportion and distribution of polarizing zone cells causing morphogenetic inhibition when coaggregated with anterior half wing mesoderm in recombinant limbs. J Embryol exp Morph 67.

Gaborit, N., Sakuma, R., Wylie, J.N., Kim, K.H., Zhang, S.S., Hui, C.C., and Bruneau, B.G. (2012). Cooperative and antagonistic roles for Irx3 and Irx5 in cardiac morphogenesis and postnatal physiology. Development 139, 4007-4019.

Galli, A., Robay, D., Osterwalder, M., Bao, X., Benazet, J.D., Tariq, M., Paro, R., Mackem, S., and Zeller, R. (2010). Distinct roles of Hand2 in initiating polarity and posterior Shh expression during the onset of mouse limb bud development. PLoS genetics 6, e1000901.

Gibson-Brown, J.J., Agulnik, S.I., Silver, L.M., Niswander, L., and Papaioannou, V.E. (1998). Involvement of T-box genes Tbx2-Tbx5 in vertebrate limb specification and development. Development 125, 2499-2509.

Glavic, A., Gómez-Skarmeta, J.L., and Mayor, R. (2002). The homeoprotein Xiro1 is required for midbrain-hindbrain boundary formation. Development 129, 1609-1621.

Gómez-Skarmeta, J.L., de La Calle-Mustienes, E., and Modolell, J. (2001). The Wnt-activated Xiro1 gene encodes a repressor that is essential for neural development and downregulates Bmp4. Development 128, 551-560.

Gomez-Skarmeta, J.L., Diez del Corral, R., de la Calle-Mustienes, E., Ferré-Marcó, D., and Modolell, J. (1996). Araucan and caupolican, two members of the novel iroquois complex, encode homeoproteins that control proneural and vein-forming genes. Cell 85, 95-105.

Gómez-Skarmeta, J.L., and Modolell, J. (2002). Iroquois genes: genomic organization and function in vertebrate neural development. Current opinion in genetics & development 12, 403- 408.

Gros, J., Hu, J.K., Vinegoni, C., Feruglio, P.F., Weissleder, R., and Tabin, C.J. (2010). WNT5A/JNK and FGF/MAPK pathways regulate the cellular events shaping the vertebrate limb bud. Current biology : CB 20, 1993-2002.

Grotewold, L., and Rüther, U. (2002). The Fused toes (Ft) Mouse Mutation Causes Anteroposterior and Dorsoventral Polydactyly. Developmental biology 251, 129-141.

Hadjantonakis, A.K., and Papaioannou, V.E. (2004). Dynamic in vivo imaging and cell tracking using a histone fluorescent protein fusion in mice. BMC biotechnology 4, 33.

Harfe, B.D., Scherz, P.J., Nissim, S., Tian, H., McMahon, A.P., and Tabin, C.J. (2004). Evidence for an expansion-based temporal Shh gradient in specifying vertebrate digit identities. Cell 118, 517-528.

Harrison, RG. (1918). Experiments on the development of the forelimb of Ambystoma, a self- differentiating equipotential system. J. Exp. Zool. 25, 413–461.

Hasson, P., Del Buono, J., and Logan, M.P. (2007). Tbx5 is dispensable for forelimb outgrowth. Development 134, 85-92.

Hayashi, S., and McMahon, A.P. (2002). Efficient recombination in diverse tissues by a tamoxifen-inducible form of Cre: a tool for temporally regulated gene activation/inactivation in the mouse. Developmental biology 244, 305-318.

Hendzel, M.J., Wei, Y., Mancini, M.A., Van Hooser, A., Ranalli, T., Brinkley, B.R., Bazett- Jones, D.P., and Allis, C.D. (1997). Mitosis-specific phosphorylation of histone H3 initiates primarily within pericentromeric heterochromatin during G2 and spreads in an ordered fashion coincident with mitotic chromosome condensation. Chromosoma 106, 348-360.

Hill, P., Gotz, K., and Ruther, U. (2009). A SHH-independent regulation of Gli3 is a significant determinant of anteroposterior patterning of the limb bud. Developmental biology 328, 506-516.

Hill, R.E. (2007). How to make a zone of polarizing activity: insights into limb development via the abnormality preaxial polydactyly. Development, growth & differentiation 49, 439-448.

Hoffelder, D.R., Luo, L., Burke, N.A., Watkins, S.C., Gollin, S.M., and Saunders, W.S. (2004). Resolution of anaphase bridges in cancer cells. Chromosoma 112, 389-397.

Hopyan, S., Sharpe, J., and Yang, Y. (2011). Budding behaviors: Growth of the limb as a model of morphogenesis. Developmental dynamics : an official publication of the American Association of Anatomists 240, 1054-1062.

Hornbruch, A., Wolpert, L. (1970). Cell division in the early growth and morphogenesis of the chick limb. Nature 226, 764-6.

Houweling, A.C., Dildrop, R., Peters, T., Mummenhoff, J., Moorman, A.F., Rüther, U., and Christoffels, V.M. (2001). Gene and cluster-specific expression of the Iroquois family members during mouse development. Mech Dev 107, 169-174.

Hui, C.C., and Joyner, A.L. (1993). A mouse model of greig cephalopolysyndactyly syndrome: the extra-toesJ mutation contains an intragenic deletion of the Gli3 gene. Nat Genet 3, 241-246.

Itoh, M., Kudoh, T., Dedekian, M., Kim, C.H., and Chitnis, A.B. (2002). iro1 and iro7_AP patterning_ectoderm adjacent to the midbrain-hindbrain boundary. Development, 2317-2327.

Jiang, J., and Hui, C.C. (2008). Hedgehog signaling in development and cancer. Developmental cell 15, 801-812.

Kagey, M.H., Newman, J.J., Bilodeau, S., Zhan, Y., Orlando, D.A., van Berkum, N.L., Ebmeier, C.C., Goossens, J., Rahl, P.B., Levine, S.S., et al. (2010). Mediator and cohesin connect gene expression and chromatin architecture. Nature 467, 430-435.

Kawakami, Y., Capdevila, J., Büscher, D., Itoh, T., Rodríguez Esteban, C., and Izpisúa Belmonte, J.C. (2001). WNT Signals Control FGF-Dependent Limb Initiation and AER Induction in the Chick Embryo. Cell 104, 891-900.

Khokha, M.K., Hsu, D., Brunet, L.J., Dionne, M.S., and Harland, R.M. (2003). Gremlin is the BMP antagonist required for maintenance of Shh and Fgf signals during limb patterning. Nat Genet 34, 303-307.

Kim, K.H., Rosen, A., Bruneau, B.G., Hui, C.C., and Backx, P.H. (2012). Iroquois homeodomain transcription factors in heart development and function. Circulation research 110, 1513-1524.

Kondo, S., and Miura, T. (2010). Reaction-diffusion model as a framework for understanding biological pattern formation. Science 329, 1616-1620.

Koussoulakos, S. (2004). Vertebrate limb development: from Harrison's limb disk transplantations to targeted disruption of Hox genes. Anatomy and embryology 209, 93-105.

Kraus, P., Fraidenraich, D., and Loomis, C.A. (2001). Some distal limb structures develop in mice lacking Sonic hedgehog signaling. Mech Dev 100, 45-58.

Kuijper, S., Feitsma, H., Sheth, R., Korving, J., Reijnen, M., and Meijlink, F. (2005). Function and regulation of Alx4 in limb development: complex genetic interactions with Gli3 and Shh. Developmental biology 285, 533-544.

Lanctôt, C., Moreau, A., Chamberland, M., Tremblay, M.L., and Drouin, J. (1999). Hindlimb patterning and mandible development require the Ptx1 gene. Development 126, 1805-1810.

Le Grand, F., Jones, A.E., Seale, V., Scime, A., and Rudnicki, M.A. (2009). Wnt7a activates the planar cell polarity pathway to drive the symmetric expansion of satellite stem cells. Cell stem cell 4, 535-547.

Lettice, L.A. (2003). A long-range Shh enhancer regulates expression in the developing limb and fin and is associated with preaxial polydactyly. Human Molecular Genetics 12, 1725-1735.

Lettice, L.A., Williamson, I., Wiltshire, J.H., Peluso, S., Devenney, P.S., Hill, A.E., Essafi, A., Hagman, J., Mort, R., Grimes, G., et al. (2012). Opposing functions of the ETS factor family define Shh spatial expression in limb buds and underlie polydactyly. Developmental cell 22, 459-467.

Lewandoski, M., Sun, X., and Martin, G.R. (2000). Fgf8 signalling from the AER is essential for normal limb development. Nature 26, 460-463.

Leyns, L., Gómez-Skarmeta, J.L., and Dambly-Chaudière, C. (1996). iroquois: a prepattern gene that controls the formation of bristles on the thorax of Drosophila. Mech Dev 59, 63-72.

Li, C., Xu, X., Nelson, D.K., Williams, T., Kuehn, M.R., and Deng, C.X. (2005). FGFR1 function at the earliest stages of mouse limb development plays an indispensable role in subsequent autopod morphogenesis. Development 132, 4755-4764.

Li, D., Sakuma, R., Vakili. N.A., Mo R, Puviindran, V., Deimling, S., Zhang, X., Hopyan, S., Hui, C.C. (2014). Formation of proximal and anterior limb skeleton requires early function of Irx3 and Irx5 and is negatively regulated by Shh signaling. Developmental cell 29, 233-40

Li, S., and Muneoka, K. (1999). Cell migration and chick limb development: chemotactic action of FGF-4 and the AER. Developmental biology 211, 335-347.

Li, Z.J., Nieuwenhuis, E., Nien, W., Zhang, X., Zhang, J., Puviindran, V., Wainwright, B.J., Kim, P.C., and Hui, C.C. (2012). Kif7 regulates Gli2 through Sufu-dependent and -independent functions during skin development and tumorigenesis. Development 139, 4152-4161.

Lickert, H., Takeuchi, J.K., Von Both, I., Walls, J.R., McAuliffe, F., Adamson, S.L., Henkelman, R.M., Wrana, J.L., Rossant, J., and Bruneau, B.G. (2004). Baf60c is essential for function of BAF chromatin remodelling complexes in heart development. Nature 432, 107-112.

Lim, S., and Kaldis, P. (2013). Cdks, cyclins and CKIs: roles beyond cell cycle regulation. Development 140, 3079-3093.

Litingtung, Y., Dahn, R.D., Li, Y., Fallon, J.F., and Chiang, C. (2002). Shh and Gli3 are dispensable for limb skeleton formation but regulate digit number and identity. Nature 418, 979- 983.

Liu, A., Li, J.Y., Bromleigh, C., Lao, Z., Niswander, L.A., and Joyner, A.L. (2003). FGF17b and FGF18 have different midbrain regulatory properties from FGF8b or activated FGF receptors. Development 130, 6175-6185.

Logan, M., Martin, J.F., Nagy, A., Lobe, C., Olson, E.N., and Tabin, C.J. (2002). Expression of Cre Recombinase in the developing mouse limb bud driven by a Prxl enhancer. Genesis 33, 77- 80.

Lopez-Rios, J., Speziale, D., Robay, D., Scotti, M., Osterwalder, M., Nusspaumer, G., Galli, A., Hollander, G.A., Kmita, M., and Zeller, R. (2012). GLI3 constrains digit number by controlling both progenitor proliferation and BMP-dependent exit to chondrogenesis. Developmental cell 22, 837-848.

Mallo, M., Wellik, D.M., and Deschamps, J. (2010). Hox genes and regional patterning of the vertebrate body plan. Developmental biology 344, 7-15.

Malumbres, M., and Barbacid, M. (2009). Cell cycle, CDKs and cancer: a changing paradigm. Nature reviews Cancer 9, 153-166.

Mao, J., McGlinn, E., Huang, P., Tabin, C.J., and McMahon, A.P. (2009). Fgf-dependent Etv4/5 activity is required for posterior restriction of Sonic Hedgehog and promoting outgrowth of the vertebrate limb. Developmental cell 16, 600-606.

Marcil, A. (2003). Pitx1 and Pitx2 are required for development of hindlimb buds. Development 130, 45-55.

Mariani, F.V., Ahn, C.P., and Martin, G.R. (2008). Genetic evidence that FGFs have an instructive role in limb proximal-distal patterning. Nature 453, 401-405.

Marshall, H., Morrison. A,, Studer. M,, Pöpperl, H., Krumlauf, R. (1996). Retinoids and Hox genes. FASEB J. 10, 969–78.

Martynoga, B., Morrison, H., Price, D.J., and Mason, J.O. (2005). Foxg1 is required for specification of ventral telencephalon and region-specific regulation of dorsal telencephalic precursor proliferation and apoptosis. Developmental biology 283, 113-127.

Masuya, H., Sagai, T., Moriwaki, K., and Shiroishi, T. (1997). Multigenic control of the localization of the zone of polarizing activity in limb morphogenesis in the mouse. Developmental biology 182, 42-51.

Matsumoto, K., Nishihara, S., Kamimura, M., Shiraishi, T., Otoguro, T., Uehara, M., Maeda, Y., Ogura, K., Lumsden, A., and Ogura, T. (2004). The prepattern transcription factor Irx2, a target of the FGF8/MAP kinase cascade, is involved in cerebellum formation. Nature neuroscience 7, 605-612.

McClure, M., and McCune, A.R. (2003). Evidence for Developmental Linkage of Pigment Patterns with Body Size and Shape in Danios (Teleostei: Cyprinidae). Evolution 57, 1863-1875.

McGlinn, E., and Tabin, C.J. (2006). Mechanistic insight into how Shh patterns the vertebrate limb. Current opinion in genetics & development 16, 426-432.

McGlinn, E., van Bueren, K.L., Fiorenza, S., Mo, R., Poh, A.M., Forrest, A., Soares, M.B., Bonaldo, M.F., Grimmond, S., Hui, C.C., et al. (2005). Pax9 and Jagged1 act downstream of Gli3 in vertebrate limb development. Mech Dev 122, 1218-1233.

McNeill, H., Yang, C.H., Brodsky, M., Ungos, J., and Simon, M.A. (1997). mirror encodes a novel PBX-class homeoprotein that functions in the definition of the dorsal-ventral border in the Drosophila eye. Genes & development 11, 1073-1082.

McPherron, A. C., Lawler, A. M., Lee, S. J. (1999). Regulation of anterior/posterior patterning of the axial skeleton by growth/differentiation factor 11. Nat. Genet. 22, 260–64.

Mic, F.A., Sirbu, I.O., and Duester, G. (2004). Retinoic acid synthesis controlled by Raldh2 is required early for limb bud initiation and then later as a proximodistal signal during apical ectodermal ridge formation. The Journal of biological chemistry 279, 26698-26706.

Min, H., Danilenko, D. M., Scully, S. A., Bolon, B., Ring, B. D., Tarpley, J. E., DeRose, M., Simonet, W. S. (1998). Fgf-10 is required for both limb and lung development and exhibits striking functional similarity to Drosophila branchless. Genes Dev. 12, 3156–61.

Minc, N., and Piel, M. (2012). Predicting division plane position and orientation. Trends in cell biology 22, 193-200.

Miura, T. (2013). Turing and Wolpert work together during limb development. Science signaling 6, pe14.

Montavon, T., Le Garrec, J.F., Kerszberg, M., and Duboule, D. (2008). Modeling Hox gene regulation in digits: reverse collinearity and the molecular origin of thumbness. Genes & development 22, 346-359.

Montero, J.A., and Hurle, J.M. (2010). Sculpturing digit shape by cell death. Apoptosis : an international journal on programmed cell death 15, 365-375.

Nagy, A. (2000). Cre Recombinase: The Universal Reagent for Genome Tailoring. Genesis 26, 99-109.

Naiche, L.A., and Papaioannou, V.E. (2007). Tbx4 is not required for hindlimb identity or post- bud hindlimb outgrowth. Development 134, 93-103.

Newman, S.A., and Comper, W.D. (1990). 'Generic' physical mechanisms of morphogenesis and pattern formation. Development 110, 1-18.

Nicholson, D.W., Ali, A., Thornberry, N.A., Vaillancourt, J.P., Ding, C.K., Gallant, M., Gareau, Y., Griffin, P., Labelle, M., Lazebnik, Y.A., et al. (1995). Identification and inhibition of the ICE/CED-3 protease necessary form mammalian apoptosis. Nature 376, 37-43.

Niederreither, K., Vermot, J., Schuhbaur, B., Chambon, P., and Dollé, P. (2002). Embryonic retinoic acid synthesis is required for forelimb growth and anteroposterior patterning in the mouse. Development 129, 3563-3574.

Niswander, L., Tickle, C., Vogel, A., Booth, I., Martin, G. R. (1993). FGF-4 replaces the apical ectodermal ridge and directs outgrowth and patterning of the limb. Cell. 75, 579–587.

Niswander, L., Jeffrey, S., Martin, G. R., Tickle, C. (1994). A positive feedback loop coordinates growth and patterning in the vertebrate limb. Nature. 371, 609–612.

Noyes, M.B., Christensen, R.G., Wakabayashi, A., Stormo, G.D., Brodsky, M.H., and Wolfe, S.A. (2008). Analysis of homeodomain specificities allows the family-wide prediction of preferred recognition sites. Cell 133, 1277-1289.

O'Rourke, M.P., Soo, K., Behringer, R.R., Hui, C.-C., and Tam, P.P.L. (2002). Twist Plays an Essential Role in FGF and SHH Signal Transduction during Mouse Limb Development. Developmental biology 248, 143-156.

Ohuchi, H., Nakagawa, T., Yamamoto, A., Araga, A., Ohata, T., Ishimaru, Y., Yoshioka, H., Kuwana, T., Nohno, T., Yamasaki, M., et al. (1997). The mesenchymal factor, FGF10, initiates and maintains the outgrowth of the chick limb bud through interaction with FGF8, an apical ectodermal factor. Development 124, 2235-2244.

Parr, B.A., and McMahon, A.P. (1995). Dorsalizing signal Wnt-7a required for normal polarity of D-V and A-P axes of mouse limb. Nature 374, 350-353.

Peters, H., Neubuser, A., Kratochwil, K., and Balling, R. (1998). Pax9-deficient mice lack pharyngeal pouch derivatives and teeth and exhibit craniofacial and limb abnormalities. Genes & development 12, 2735-2747.

Peters, T., Ausmeier, K., Dildrop, R., and Ruther, U. (2002). The mouse Fused toes (Ft) mutation is the result of a 1.6-Mb deletion including the entire Iroquois B gene cluster. Mammalian genome : official journal of the International Mammalian Genome Society 13, 186-188.

Peters, T., Dildrop, R., Ausmeier, K., and Rüther, U. (2000). Organization of Mouse Iroquois Homeobox Genes in Two Clusters Suggests a Conserved Regulation and Function in Vertebrate Development. Genome research 10, 1453-1462.

Qu, S., Niswender, K.D., Ji, Q., van der Meer, R., Keeney, D., Magnuson, M.A., and Wisdom, R. (1997). Polydactyly and ectopic ZPA formation in Alx-4 mutant mice. Development 124, 3999-4008.

Reggiani, L., Raciti, D., Airik, R., Kispert, A., and Brandli, A.W. (2007). The prepattern transcription factor Irx3 directs nephron segment identity. Genes & development 21, 2358-2370.

Reiter, R. S, Solursh, M. (1982). Mitogenic property of the apical ectodermal ridge. Dev. Biol. 93, 28–35.

Remeseiro, S., Cuadrado, A., and Losada, A. (2013). Cohesin in development and disease. Development 140, 3715-3718.

Rhodes, J.M., Bentley, F.K., Print, C.G., Dorsett, D., Misulovin, Z., Dickinson, E.J., Crosier, K.E., Crosier, P.S., and Horsfield, J.A. (2010). Positive regulation of c-Myc by cohesin is direct, and evolutionarily conserved. Developmental biology 344, 637-649.

Riddle, R.D., Johnson, R.L., Laufer, E., and Tabin, C. (1993). Sonic hedgehog mediates the polarizing activity of the ZPA. Cell 75, 1401-1416.

Riddle, R. D., Ensini, M., Nelson, C., Tsuchida, T., Jessell, T. M., Tabin, C. (1995). Induction of the LIM homeobox gene Lmx1 by WNT7a establishes dorsoventral pattern in the vertebrate limb. Cell. 83, 631-640.

Robert, B., and Lallemand, Y. (2006). Anteroposterior patterning in the limb and digit specification: contribution of mouse genetics. Developmental dynamics : an official publication of the American Association of Anatomists 235, 2337-2352.

Robertshaw, E., Matsumoto, K., Lumsden, A., and Kiecker, C. (2013). Irx3 and Pax6 establish differential competence for Shh-mediated induction of GABAergic and glutamatergic neurons of the thalamus. Proc Natl Acad Sci U S A 110, E3919-3926.

Roselló-Díez, A., Ros, M.A., and Torres, M. (2011). Diffusible signals, not autonomous mechanisms, determine the main proximodistal limb subdivision. Science 332, 1086-1088.

Rosenquist GC. (1971). The origin and movement of the limb-bud epithelium and mesenchyme in the chick embryo as determined by radioautographic mapping. J. Embryol. Exp. Morphol. 25, 85–96.

Sagai, T., Hosoya, M., Mizushina, Y., Tamura, M., and Shiroishi, T. (2005). Elimination of a long-range cis-regulatory module causes complete loss of limb-specific Shh expression and truncation of the mouse limb. Development 132, 797-803.

Sanders, T.A., Llagostera, E., and Barna, M. (2013). Specialized filopodia direct long-range transport of SHH during vertebrate tissue patterning. Nature 497, 628-632.

Saunders JW, et al. (1968). Ectoderm-mesenchymal interaction in the origins of wing symmetry. In Epithelial-Mesenchymal Interactions, ed. R Fleischmajer, RE Billingham, pp. 289–314. Baltimore: Williams & Wilkins.

Schaller, S.A., Li, S., Ngo-Muller, V., Han, M.J., Omi, M., Anderson, R., and Muneoka, K. (2001). Cell Biology of Limb Patterning. Int Rev Cytol 203, 483-517.

Searls, R. L., Janners, M. Y. (1971). The initiation of limb bud outgrowth in the embryonic chick. Dev. Biol. 24, 198–213.

Segalen, M., and Bellaiche, Y. (2009). Cell division orientation and planar cell polarity pathways. Seminars in cell & developmental biology 20, 972-977.

Sekine, K., Ohuchi, H., Fujiwara, M., Yamasaki, M., Yoshizawa, T., Sato, T., Yagishita, N., Matsui, D., Koga, Y., Itoh, N., et al. (1999). Fgf10 is essential for limb and lung formation. Nat Genet 21, 138-141.

Sharpe, J., Ahlgren, U., Perry, P., Hill, B., Ross, A., Hecksher-Sorensen, J., Baldock, R., and Davidson, D. (2002). Optical projection tomography as a tool for 3D microscopy and gene expression studies. Science 296, 541-545.

Sheth, R., Marcon, L., Bastida, M.F., Junco, M., Quintana, L., Dahn, R., Kmita, M., Sharpe, J., and Ros, M.A. (2012). Hox genes regulate digit patterning by controlling the wavelength of a Turing-type mechanism. Science 338, 1476-1480.

Shou, S., Scott, V., Reed, C., Hitzemann, R., and Stadler, H.S. (2005). Transcriptome analysis of the murine forelimb and hindlimb autopod. Developmental dynamics : an official publication of the American Association of Anatomists 234, 74-89.

Solnica-Krezel, L. (2005). Conserved patterns of cell movements during vertebrate gastrulation. Current biology : CB 15, R213-228.

Soriano, P. (1999). Generalized lacZ expression with the ROSA26 Cre reporter strain. Nat Genet 21, 70-71.

Soshnikova, N., Zechner, D., Huelsken, J., Mishina, Y., Behringer, R.R., Taketo, M.M., Crenshaw, E.B., 3rd, and Birchmeier, W. (2003). Genetic interaction between Wnt/beta-catenin and BMP receptor signaling during formation of the AER and the dorsal-ventral axis in the limb. Genes & development 17, 1963-1968.

Star, E.N., Zhu, M., Shi, Z., Liu, H., Pashmforoush, M., Sauve, Y., Bruneau, B.G., and Chow, R.L. (2012). Regulation of retinal interneuron subtype identity by the Iroquois homeobox gene Irx6. Development 139, 4644-4655.

Strutt, D. (2005). Organ shape: controlling oriented cell division. Current biology : CB 15, R758- 759.

Summerbell, D., Lewis, J. H., Wolpert, L. (1973). Positional information in chick limb morphogenesis. Nature. 244, 492–496.

Sun, X., Mariani, F.V., and Martin, G.R. (2002). Functions of FGF signalling from the apical ectodermal ridge in limb development. Nature 418, 501-508.

Suzuki, T. (2013). How is digit identity determined during limb development? Development, growth & differentiation 55, 130-138.

Suzuki, T., Takeuchi, J., Koshiba-Takeuchi, K., and Ogura, T. (2004). Tbx Genes Specify Posterior Digit Identity through Shh and BMP Signaling. Developmental cell 6, 43–53.

Tabin, C., and Wolpert, L. (2007). Rethinking the proximodistal axis of the vertebrate limb in the molecular era. Genes & development 21, 1433-1442.

Takahashi, I., Nuckolls, G. H., Takahashi, K., Tanaka, O., Semba, I., Dashner, R., Shum, L., Slavkin, H. C. (1998). Compressive force promotes sox9, type II collagen and aggrecan and inhibits IL-1beta expression resulting in chondrogenesis in mouse embryonic limb bud mesenchymal cells. J Cell Sci. 111, 2067-2076.

Takeuchi, J.K. (2003). Tbx5 and Tbx4 trigger limb initiation through activation of the Wnt/Fgf signaling cascade. Development 130, 2729-2739. te Welscher, P., Fernandez-Teran, M., Ros, M.A., and Zeller, R. (2002a). Mutual genetic antagonism involving GLI3 and dHAND prepatterns the vertebrate limb bud mesenchyme prior to SHH signaling. Genes & development 16, 421-426. te Welscher, P., Zuniga, A., Kuijper, S., Drenth, T., Goedemans, H.J., Meijlink, F., and Zeller, R. (2002b). Progression of vertebrate limb development through SHH-mediated counteraction of GLI3. Science 298, 827-830. ten Berge, D., Brugmann, S.A., Helms, J.A., and Nusse, R. (2008). Wnt and FGF signals interact to coordinate growth with cell fate specification during limb development. Development 135, 3247-3257.

Tena, J.J., Alonso, M.E., de la Calle-Mustienes, E., Splinter, E., de Laat, W., Manzanares, M., and Gomez-Skarmeta, J.L. (2011). An evolutionarily conserved three-dimensional structure in the vertebrate Irx clusters facilitates enhancer sharing and coregulation. Nature communications 2, 310.

Tickle C. (1981). The number of polarizing region cells required to specifiy additional digits in the developing chick wing. Nature. 289, 295–298.

Tkatchenko, A.V. (2006). Whole-mount BrdU staining of proliferating cells by DNase treatment: application to postnatal mammalian retina. Biotechniques 40, 29-30.

Towers, M., and Tickle, C. (2009). Generation of pattern and form in the developing limb. The International journal of developmental biology 53, 805-812.

Tümpel, S., Sanz-Ezquerro, J.J., Isaac, A., Eblaghie, M.C., Dobson, J., and Tickle, C. (2002). Regulation of Tbx3 Expression by Anteroposterior Signalling in Vertebrate Limb Development. Developmental biology 250, 251-262.

Verheyden, J.M., Lewandoski, M., Deng, C., Harfe, B.D., and Sun, X. (2005). Conditional inactivation of Fgfr1 in mouse defines its role in limb bud establishment, outgrowth and digit patterning. Development 132, 4235-4245.

Verheyden, J.M., and Sun, X. (2008). An Fgf/Gremlin inhibitory feedback loop triggers termination of limb bud outgrowth. Nature 454, 638-641.

Vogel, A., Rodriguez, C., Izpisúa-Belmonte, J. C. (1996). Involvement of FGF-8 in initiation, outgrowth and patterning of the vertebrate limb. Development. 122, 1737–50.

Vokes, S.A., Ji, H., Wong, W.H., and McMahon, A.P. (2008). A genome-scale analysis of the cis-regulatory circuitry underlying sonic hedgehog-mediated patterning of the mammalian limb. Genes & development 22, 2651-2663.

Walck-Shannon, E., and Hardin, J. (2013). Cell intercalation from top to bottom. Nature reviews Molecular cell biology 15, 34-48.

Walls, J.R., Coultas, L., Rossant, J., and Henkelman, R.M. (2008). Three-Dimensional Analysis of Vascular Development in the Mouse Embryo. PLoS One 3, e2853.

Walls, J.R., Sled, J.G., Sharpe, J., and Henkelman, R.M. (2007). Resolution improvement in emission optical projection tomography. Physics in medicine and biology 52, 2775-2790.

Wang, B., Fallon, J.F., and Beachy, P.A. (2000). Hedgehog-Regulated Processing of Gli3 Produces an Anterior/Posterior Repressor Gradient in the Developing Vertebrate Limb. Cell 100, 423-434.

Wang, C., Ruther, U., and Wang, B. (2007). The Shh-independent activator function of the full- length Gli3 protein and its role in vertebrate limb digit patterning. Developmental biology 305, 460-469.

Wang, G.F., Nikovits, W., Jr., Bao, Z.Z., and Stockdale, F.E. (2001). Irx4 forms an inhibitory complex with the vitamin D and retinoic X receptors to regulate cardiac chamber-specific slow MyHC3 expression. The Journal of biological chemistry 276, 28835-28841.

Wen, X., Lai, C.K., Evangelista, M., Hongo, J.A., de Sauvage, F.J., and Scales, S.J. (2010). Kinetics of hedgehog-dependent full-length Gli3 accumulation in primary cilia and subsequent degradation. Molecular and cellular biology 30, 1910-1922.

Wyngaarden, L.A., Vogeli, K.M., Ciruna, B.G., Wells, M., Hadjantonakis, A.K., and Hopyan, S. (2010). Oriented cell motility and division underlie early limb bud morphogenesis. Development 137, 2551-2558.

Xu, X., Weinstein, M., Li, C., and Deng, C. (1999). Fibroblast growth factor receptors (FGFRs) and their roles in limb development. Cell Tissue Res 296, 33-43.

Xu, X., Weinstein, M., Li, C., Naski, M., Cohen, R.I., Ornitz, D.M., Leder, P., and Deng, C. (1998). Fibroblast growth factor receptor 2 (FGFR2)-mediated reciprocal regulation loop between FGF8 and FGF10 is essential for limb induction. Development 125, 753-765.

Yang, C. H., Simon, M. A. and McNeill, H. (1999). mirror controls planar polarity and equator formation through repression of fringe expression and through control of cell affinities. Development 126, 5857-5866.

Zakany, J., and Duboule, D. (2007). The role of Hox genes during vertebrate limb development. Current opinion in genetics & development 17, 359-366.

Zakany, J., Kmita, M., and Duboule, D. (2004). A dual role for Hox genes in limb anterior- posterior asymmetry. Science 304, 1669-1672.

Zamir, E.A., Czirok, A., Cui, C., Little, C.D., and Rongish, B.J. (2006). Mesodermal cell displacements during avian gastrulation are due to both individual cell-autonomous and convective tissue movements. Proc Natl Acad Sci U S A 103, 19806-19811.

Zeller, R., Lopez-Rios, J., and Zuniga, A. (2009). Vertebrate limb bud development: moving towards integrative analysis of organogenesis. Nature reviews Genetics 10, 845-858.

Zhang, S.S., Kim, K.H., Rosen, A., Smyth, J.W., Sakuma, R., Delgado-Olguín, P., Davis, M., Chi, N.C.P., V.

Gaborit, N., Sukonnik, T., Wylie, J.N., et al. (2011). Iroquois homeobox gene 3 establishes fast conduction in the cardiac His-Purkinje network. Proc Natl Acad Sci U S A 108, 13576-13581.

Zhang, Z., Verheyden, J.M., Hassell, J.A., and Sun, X. (2009). FGF-regulated Etv genes are essential for repressing Shh expression in mouse limb buds. Developmental cell 16, 607-613.

Zhao, X., Sirbu, I.O., Mic, F.A., Molotkova, N., Molotkov, A., Kumar, S., and Duester, G. (2009). Retinoic acid promotes limb induction through effects on body axis extension but is unnecessary for limb patterning. Current biology : CB 19, 1050-1057.

Zhu, J., Nakamura, E., Nguyen, M.T., Bao, X., Akiyama, H., and Mackem, S. (2008). Uncoupling Sonic hedgehog control of pattern and expansion of the developing limb bud. Developmental cell 14, 624-632.

Zhulyn, O., Li, D., Deimling, S., Vakili, N.A., Mo, R., Puviindran, V., Chen, M.H., Chuang, P.T., Hopyan, S., Hui, C.C. (2008). A Switch from Low to High Shh Activity Regulates Establishment of Limb Progenitors and Signaling Centers. Developmental cell 29, 241-9.

Zúñiga, A., Haramis, A.P., McMahon, A.P., and Zeller, R. (1999). Signal relay by BMP antagonism controls the SHH/FGF4 feedback loop in vertebrate limb buds. Nature 401, 598-602.