Cyp26b1 Limits Inappropriate Activation of Rarγ by Retinoic Acid During Murine Embryogenesis

CYP26B1 LIMITS INAPPROPRIATE ACTIVATION OF RARγ BY RETINOIC ACID DURING MURINE EMBRYOGENESIS

Tracie L. Pennimpede

A thesis submitted to the Department of Pathology and Molecular Medicine In conformity with the requirements for the degree of Doctorate of Philosophy

Queen‟s University Kingston, Ontario, Canada

March, 2010

Proper embryonic patterning requires precise spatio-temporal regulation of retinoic acid (RA) activity. Morphogenesis can be regulated at the level of RA distribution, mainly via its synthesis and catabolism by the RALDH and CYP26 enzymes respectively, and at the level of RA-mediated transcription through activation of its cognate nuclear receptor, the retinoic acid receptors (RARs) α, β, and γ. Loss of Cyp26b1 leads to increased local levels of RA in tissues such as the limb and craniofacial structures, and results in neonatal lethality. Visible gross phenotypic defects in neonates include phocomelia (shortening of the limbs), adactyly (missing digits), micrognathia

(shortened lower jaw), and open eyes at birth. In addition, these embryos exhibit cleft palate and have a paucity of vibrissal (whisker) and pelage (hair) follicles. We have previously shown that ablating the gene encoding RARγ in a Cyp26a1-null background was able to rescue the caudal abnormalities associated with improper RA exposure in these embryos by limiting aberrant RA signalling, and thus rescuing expression domains of target genes involved in caudal development. I show here that ablating Rarg in a

Cyp26b1-null background is able to partially rescue the defects associated with loss of

CYP26B1. These include a reduction in the severity of limb defects, rescued vibrissae, fused eyelids, and recovered aspects of axial skeletal development. This compound-null murine model illustrates that RARγ plays a specific role in transducing the RA signal within tissues that are affected by the loss of CYP26B1. Further molecular analysis of the pathways responsible for directing limb bud outgrowth and eyelid fusion provided insight into pathways regulated by RARγ in these rescued tissues.

ii Statement of Collaboration

Generation of the mouse strain lacking Cyp26b1 by our laboratory was undertaken by Dr.

Glenn MacLean, and has been described previously (MacLean et al., 2007; Maclean et al., 2009).

Parts of this thesis concerning the limb phenotype (Chapters 2, 5 and 7) have been taken from a peer-reviewed publication (Pennimpede et al., 2010) which I prepared, but with input from my co-authors.

Parts of the Literature Review (Chapter 2) and the results for Chapter 4 were prepared for my Comprehensive Examination (December 2007) and have been updated from that work.

Don Cameron, co-author on the manuscript prepared with regards to the observed limb phenotype (Pennimpede et al., 2010), performed in situ hybridization for Cyp26b1 and

Rarg on wild type embryos (Figure 4) and eyelid sections (Figure 20).

iii Acknowledgements

This has been a great journey and I have so many people to thank for making it the most wonderful time and learning experience.

First and foremost I would like to thank my supervisor, Dr Petkovich, for being the best out of all the world‟s best supervisors, along with an incredible friend, mentor, punster, and person. Marty: I can‟t possibly thank you enough for this opportunity, for your creative vision and encouragement, and for the support that you and Patty have given to me (and all of us!) over the years. I wouldn‟t and couldn‟t have done this with anyone else.

To my labmates, past and present, for making every day spent in Botterell (especially room 359) both interesting and entertaining. Even sometimes from a science perspective.

Special thanks to Suzy, Ali, Kristen, Ana Paula, Olivier, Naomi, Hui, and Makoto for friendship and help with experiments; and the excellent summer students Helen, Katie, and Bilge. I am especially grateful to Glenn MacLean for his guidance and friendship and to Don Cameron, my compatriot – an incredible friend and benchmate. I can‟t possibly put into words how thankful I am to you both and how much I appreciated our time together.

iv Thanks also to the wonderful group of colleagues and friends that I have had the pleasure of knowing in this life. Especially Team Wicked Awesome (for the many, many good times), but also countless others on the 2nd, 3rd, and 6th floors (and at this University) for chats and advice over beer on the Grad Club patio. And to my great friends Samantha, and the Chubsies (Cory and Alisha) for being there the whole time.

To the Grad Club – my home away from lab – and all who work there. A place where many memories were forged (and some things forgotten).

I am also grateful for the technical expertise of Matt Gordon, Jeff Mewburn, Jalna Meens,

Dan Wainman, and ACS; and I would be remiss if I didn‟t express my appreciation for the many murine lives that went into making this thesis. Thanks also to the Pathology

Department and my Supervisory Committee (Drs Lebrun and Davey) for advice and encouragement, and to Barb Saunders and Jason Palmer for administrative and IT support.

Finally, thanks to my incredible family for understanding that being a lifelong student is an acceptable life choice and encouraging me as I pursued this endeavour.

“It is not birth, marriage, or death, but gastrulation which is truly

the most important time in your life.” – Professor Lewis Wolpert

v Table of Contents

Abstract ...... ii Statement of Collaboration ...... iii Acknowledgements ...... iv Table of Contents ...... vi List of Figures ...... ix List of Tables ...... x List of Abbreviations ...... xi Chapter 1: General Introduction ...... 1 Chapter 2: Literature Review ...... 4 2.1 Vertebrate Morphogenesis ...... 4 2.1.1 Setting up the embryonic body plan and morphogenesis of the axial skeleton ...... 5 2.1.2 The developing limb as a paradigm for morphogenesis ...... 8 2.1.2.1 Models of limb P-D outgrowth: Progress Zone, Early Specification, and Two- Signal ...... 10 2.1.3 Molecular mechanisms of eyelid closure ...... 12 2.2 Vitamin A metabolism, mechanism of action, and physiological roles ...... 14 2.2.1 Vitamin A in nutrition ...... 14 2.2.2 Early studies on the role of vitamin A in embryonic development ...... 15 2.2.3 Canonical vitamin A metabolism ...... 15 2.2.4 RA binding proteins ...... 19 2.2.5 Transcriptional regulation by RA ...... 20 2.3 RA and morphogenesis ...... 24 2.4 Regulation of RA signalling during embryogenesis ...... 28 2.4.1 RAR/RXR expression and function ...... 28 2.5 Regulation of RA distribution within the developing embryo ...... 30 2.5.1 Expression patterns of Raldh and Cyp26 isoforms in the developing mouse embryo .. 30 2.5.2 Raldh and Cyp26 knockout studies ...... 33 2.6 Rarg-null mice are resistant to RA-induced craniofacial and caudal defects ...... 36 2.7 Cyp26a1/Rarg double knockout mice...... 38 2.8 Hypothesis and Specific Objectives ...... 39 Chapter 3: Materials and Methods ...... 43 3.1 Creation of mouse strains and genotyping ...... 43

vi 3.2 Embryo Collection ...... 44 3.3 Skeletal analysis ...... 45 3.4 Whole-mount in situ hybridization ...... 45 3.5 Analysis of apoptosis in whole limb buds ...... 46 3.6 In situ hybridization on eyelid sections ...... 47 3.7 Immunohistochemistry on eyelid sections ...... 48 Chapter 4 : Results: Cyp26b1-/-/ Rarg-/- phenotype: gross morphology and skeletal defects ...... 49 4.1 Ablating Rarg does not rescue Cyp26b1 mutants from lethality ...... 49 4.2 Rarg expression overlaps with Cyp26a1 and Cyp26b1 in mid-gestation embryos...... 51 4.3 Analysis of gross morphology of B1-/-/γ-/- fetuses ...... 56 4.4 Skeletal malformations in B1-null and B1-/-/γ-/- embryos ...... 56 4.4.1 Ablation of Cyp26b1 does not alter skeletal defects resulting from loss of Rarg ...... 68 4.4.2 Removal of Rarg in the Cyp26b1-null background can rescue skeletal defects associated with RA treatment but which were not observed following loss of Cyp26a1 ...... 71 4.5 Expression of Hoxa3 and Hoxb3 is unchanged in Cyp26b1-null embryos at E10.5 ...... 71 Chapter 5 : Results: Analysis of limb proximal-distal outgrowth in Cyp26b1/Rarg compound-null embryos ...... 77 5.1 Expression of Rarg and Cyp26b1 overlap during limb P-D outgrowth ...... 77 5.2 Genetic removal of Rarg in a Cyp26b1-/- background partially rescues limb skeletal morphology ...... 80 5.3 Markers of the AER and ZPA are normally expressed in Cyp26b1-null embryos ...... 82 5.4 Aberrant expression of proximal-distal patterning genes remains unchanged between Cyp26b1-null and Cyp26b1/Rarg compound-null embryos ...... 85 5.5 Localized mesenchymal cell death is rescued in Cyp26b1/Rarg-null limbs ...... 97 Chapter 6 : Results: Removing Rarg in a Cyp26b1-null background is able to rescue the open- eye-at-birth phenotype ...... 101 6.1 Ablation of Rarg rescues eyelid closure in Cyp26b1-/- embryos ...... 101 6.2 Cyp26b1 and Rarg expression in the developing eyelid ...... 101 6.3 Expression of components of the FGF signalling pathway are not impaired in Cyp26b1-/- eyelids ...... 102 6.4 There are no differences in the extent of cell proliferation or cell death in Cyp26b1-null eyelids at E15.0 ...... 111 Chapter 7 : Discussion and Conclusions ...... 114 7.1 Ablation of Rarg is able to partially rescue the Cyp26b1-null phenotype ...... 114

vii 7.1.1 Improvement of craniofacial morphogenesis in Cyp26b1/Rarg compound-null embryos ...... 115 7.1.2 Vertebrothoracic defects in Cyp26b1-null embryos are rescued by removal of Rarg 117 7.1.2.1 Evidence that loss of CYP26B1 activity results in premature differentiation of osteogenic precursors ...... 120 7.1.2.2 Expression of Hoxa3 and Hoxb3 are not effected by exposure to increased levels of endogenous RA caused by loss of CYP26B1 function...... 121 7.2 Analysis of Cyp26b1/Rarg compound-null mice reveals two genetically separable effects of retinoic acid on limb outgrowth ...... 122 7.2.1 The role of RARγ in the limb...... 123 7.2.2 The nature of the P-D patterning defect caused by loss of Cyp26b1 is revealed following loss of Rarg ...... 124 7.2.3 Ablation of Cyp26b1 causes increased cell death in the limb bud which is ameliorated following removal of Rarg ...... 129 7.2.4 Perspectives...... 130 7.3 Failure of eyelid closure in Cyp26b1-null fetuses does not result from disrupted FGF signalling, decreased cell proliferation, or increased apoptosis in eyelid anlagen ...... 131 7.3.1 The role of RA in the developing eye ...... 131 7.3.2 Mechanisms of open-eyelid phenotype following ablation of Cyp26b1 ...... 133 7.3.3 Links between EOB, impaired wound healing, and defects in skin/hair development ...... 135 7.3.4 Effects on cell proliferation, migration, and/or apoptosis by RA during eyelid closure ...... 136 7.4 Perspectives and future directions ...... 137 7.4.1 Evolutionary considerations from observations of Cyp26b1/Rarg compound-null embryos ...... 139 References ...... 142

viii List of Figures

Figure 1: Overview of vitamin A metabolism...... 17 Figure 2: Canonical activation of RA-responsive genes...... 22 Figure 3: Overview of RA signalling in embryonic patterning and morphogenesis...... 26 Figure 4: Expression of Cyp26b1 and Rarg in developing embryos...... 54 Figure 5: Neonatal morphology of Cyp26b1-null and Cyp26b1/Rarg compound-null mice...... 57 Figure 6: Comparative skeletal development of forelimbs and hindlimbs in Cyp26b1-/-, Cyp26b1-/-/Rarg+/-, Cyp26b1-/-/Rarg-/-, and phenotypically normal fetuses at E18.5...... 60 Figure 7: Ossification of skull structures and the hyoid bone are rescued in Cyp26b1/Rarg compound-null mice...... 62 Figure 8: Various abnormalities of skull formation observed in Cyp26b1-null animals are rescued by ablation of Rarg...... 65 Figure 9: Vertebrothoracic abnormalities in Cyp26b1-null animals are rescued following ablation of Rarg...... 69 Figure 10: Rib phenotype of Cyp26b1-/- embryos...... 72 Figure 11: Expression of Hoxa3 and Hoxb3 at E10.5...... 75 Figure 12: Expression of Cyp26b1 and Rarg during normal limb outgrowth...... 78 Figure 13: The AER and ZPA are properly specified in Cyp26b1-/- limbs...... 83 Figure 14: Expression of Meis2 in forelimb and hindlimb buds at E10.5 and E11.5...... 86 Figure 15: Expression of Shox2 in forelimb and hindlimb buds at E10.5 and E11.5...... 89 Figure 16: Expression of Hoxa11 in forelimb and hindlimb buds at E10.5 and E11.5...... 91 Figure 17: Expression of Hoxd13 in forelimb and hindlimb buds at E10.5 and E11.5...... 93 Figure 18: Expression of Hoxa13 in forelimb and hindlimb buds at E10.5 and E11.5...... 95 Figure 19: Effects of increased RA on apoptosis in limb buds...... 98 Figure 20: Normal expression of Cyp26b1 and Rarg in eyelids at E15.0...... 103 Figure 21: Expression analysis of mediators of FGF signalling in eyelids at E15.0...... 106 Figure 22: Downstream FGF signalling is not compromised in Cyp26b1-null eyelids at E15.0. 109 Figure 23: Cell proliferation and apoptosis in eyelids at E15.0...... 112 Figure 24: Overview of the consequences following loss of CYP26B1 and RARγ on final skeletal outcome in the limb...... 126

ix List of Tables

Table 1: Expression patterns of Cyp26a1, Cyp26b1, Rara, Rarb, and Rarg in craniofacial structures, eye, and limbs during murine development ...... 42 Table 2: Resulting genotypic frequencies from B1+/-/γ+/- crosses...... 50 Table 3: Viability of pups from B1+/-/γ+/- male and B1+/+/γ+/- female crosses...... 52 Table 4: Summary of axial skeletal abnormalities observed at E18.5...... 67 Table 5: Summary of limb skeletal abnormalities observed at E18.5 ...... 81

x List of Abbreviations

ADH alcohol dehydrogenase AER apical ectodermal ridge ATP adenosine triphosphate A-P antero-posterior AP alkaline phosphatase atRA all-trans retinoic acid bp base pair(s) BMP bone morphogenic protein CBP CREB binding protein cDNA complementary DNA CRBP cellular retinol binding protein CRABP cellular retinoic acid binding protein CYP cytochrome P450 DAPI 4',6-diamidino-2-phenylindole dihydrochloride ddH2O double-distilled (Millipore™) water DEPC diethylpyrocarbonate DIG digoxigenin-11-UTP DNA deoxyribonucleic acid E embryonic day EOB open-eyes-at-birth ES embryonic stem fmol femtomoles (10-15 moles) FGF fibroblast growth factor h human HAT histone acetyltransferase HDAC histone deacetylase Hh hedgehog

xi HPLC high-pressure liquid chromatography HR homologous recombination IHC immunohistochemistry kb kilobase pair(s) lacZ β-galactosidase LCoR ligand-dependent corepressor lel lower eyelid loxP locus of cross-over (x) in P1 m mouse min minutes mL milliliters mRNA messenger RNA NCC neural crest cell NCoR nuclear receptor corepressor NGS normal goat serum OCT Optimal Cutting Temperature P post-natal day PBS phosphate-buffered saline PCR polymerase chain reaction P-D proximo-distal PFA paraformaldehyde PHPV persistence and hyperplasia of the primary vitreous body r rhombomere RA retinoic acid RALDH retinaldehyde dehydrogenase RAR retinoic acid receptor RARE retinoic acid response element RBP retinol binding protein RDH retinol dehydrogenase RNA ribonucleic acid

xii RNase ribonuclease RXR retinoid X receptor s second(s) SDR short-chain dehydrogenase/reductase Shh sonic hedgehog SMRT silencing mediator for retinoid and thyroid hormone receptor SRC steroid hormone receptor coactivator STRA stimulated by retinoic acid SWI/SNF switch/sucrose non-fermentable TGF transforming growth factor uel upper eyelid VAD vitamin A deficiency WMISH whole-mount in situ hybridization wt wild type ZPA zone of polarizing activity 4-oxo-RA 4-oxo retinoic acid 4-OH-RA 4-hydroxy retinoic acid 9-cisRA 9-cis retinoic acid 18-OH-RA 18-hydroxy retinoic acid ºC degree celsius µL microliter(s) µg microgram(s) +/+ homozygous wild type genotype +/- heterozygous genotype (one wild type and one mutant allele) -/- homozygous mutant genotype

xiii Chapter 1

General Introduction

Embryonic development is a complex and dynamic process wherein the spatio- temporal regulation of specific genes and gene products controls the sculpting and morphogenesis of tissues, resulting in embryo formation. It is remarkable that embryogenesis occurs nearly perfectly, time and time again; it is a highly plastic process.

Many of the main molecules in embryogenesis are found to be extremely well conserved across species and are also molecules found to be dysregulated in various cancers. This is not surprising, considering that the developing embryo is made up of pluripotent cells with vast differentiation potential. Determining how and when these cells are committed to a specific fate can thus provide insight into gene regulatory networks which are important not only in the context of embryonic development, but also stem cell-based therapies, cancer, and tissue engineering.

Many processes involved in embryonic development exhibit high tolerance in that the embryo can deal with certain thresholds of environmental insult and still develop normally. However, when such thresholds are exceeded, the consequences are often severe. All-trans retinoic acid (atRA or RA) is a pleiotropic metabolite of vitamin A

(retinol) able to act as both a morphogen and a teratogen during embryogenesis. Because vitamin A (and thus RA) is a nutritionally-derived molecule, classical embryonic defects characterizing both vitamin A deficiency (VAD) and RA teratogenesis can be observed

1 following changes in maternal vitamin A status. However, regulation of endogenous RA distribution within the embryo is fundamentally controlled by the expression of enzymes responsible for catalyzing its formation (members of the RALDH family) and its degradation (members of the CYP26 family) (Pennimpede et al., 2006; Duester, 2008;

Niederreither and Dolle, 2008). When the distribution or activity of these enzymes is perturbed during development, the results are characteristic of VAD and RA teratogenesis, respectively.

The effects of RA on specific cellular processes such as differentiation, proliferation, and apoptosis are mediated by its ability to act as a ligand for nuclear hormone receptor complexes comprised of retinoic acid receptor (RAR) and retinoid X receptor (RXR) heterodimeric transcription factors (Mark et al., 2009). Thus, an additional mechanism by which RA signalling can be limited during development is by genetic ablation of the receptor isoforms. These effects basically recapitulate embryonic

VAD, although there is considerable phenotypic variation and functional redundancy

(reviewed in Mark et al., 2009).

Fundamentally, the ability of RA to induce multiple effects results from precise spatio-temporal control of its distribution during embryonic development. However, other factors including tissue type, stage of development, and presence and types of receptors determine the net effect on those tissues exposed to, or devoid of, RA.

Although the signals and molecular pathways that regulate development of particular structures are often tissue-specific, it is clear that there are also some general pathways which are globally utilized during morphogenesis. For example, morphogens like 2 fibroblast growth factors (FGFs), bone morphogenic proteins (BMPs), Wnts, and RA are likely involved in setting up patterning by modulating Hox gene expression; leading to proper segmentation of the body axis (Alexander et al., 2009). This paradigm of morphogen gradients setting up gene patterning and resulting in tissue morphogenesis is re-established for various structures including the limb, hindbrain and craniofacial structures, and external genitalia. Hence, the regulation of RA signalling and metabolism in embryonic development has wide implications in a variety of tissues.

In this thesis I describe the generation and analysis of mice lacking both

CYP26B1 and RARγ. These animals provide a model to examine the effects of changes in both RA distribution and signalling, along with mechanisms for development of structures in animals exposed to lethal concentrations of RA. Overall, the ablation of

Rarg on a Cyp26b1-null background was found to ameliorate defects associated with excess RA within skeletal, limb, and craniofacial structures, including eyelid closure.

Although it was previously known that RARγ is the RAR isotype responsible for mediating the teratogenic effects of RA in the caudal embryo (Iulianella et al., 1999;

Abu-Abed et al., 2003), the current model illustrates that RARγ is also responsible for mediating effects involved in development of the limb, eyelid, bone, and hair follicles in the absence of CYP26B1. Moreover, by molecular analysis of the limb and eyelid rescue phenotypes, the RA signalling pathways involved in regulating development of these structures were further elucidated.

Chapter 2

Literature Review

2.1 Vertebrate Morphogenesis

Embryogenesis involves the generation of diverse cell types which are arranged into ordered tissues, structures, and the entire fetus. The development of an embryo from a single cell, the fertilized egg, is a dynamic and fascinating progression. This complex process of self-assembly is understandably multi-faceted and involves cell-cell contacts, cell movement, cell death, cell position and orientation, mechanical constraints, as well as genetic regulation and epigenetic modifications. One of the most important early steps in post-implantation embryogenesis is gastrulation. In mice this begins at around embryonic day (E) 6.75 (Downs and Davies, 1993) and is characterized by the epithelial tissue of the epiblast segregating into the three definitive germ layers which will form all of the tissues of the final embryo: the ectoderm (which will form the skin and nervous system), mesoderm (which will become the muscles, connective tissues, and organs), and endoderm (which contributes to the respiratory and digestive tracts) (Slack, 2006). This is also when the embryo acquires axial identity (Downs and Davies, 1993).

At around E9.5, after the embryo has turned, it moves from the formation of the body plan toward organogenesis (Slack, 2006). The mechanisms involved in the morphogenesis of various organs and tissues are complex and tissue-specific, and far beyond the scope of this thesis. However, in order to place the developmental systems

4 which were analyzed in this project into context, the morphogenesis of three specific systems will be explained in slightly more detail: the axial skeleton, limb, and eyelid.

2.1.1 Setting up the embryonic body plan and morphogenesis of the axial skeleton

In the early post-implantation stage mouse embryo (around E6.5-E7.0) the primitive streak becomes apparent and marks the antero-posterior (A-P) axis of the embryo. At this point the body plan begins to be laid out and by E7.75-E8.0 pairs of somites (epithelial blocks of mesodermal tissue which will become the vertebrae, ribs, dermis of the skin, and muscles of the body wall and limbs) start to form in a rhythmic reiterative A-P sequence. The production of somites from the anterior pre-somitic mesoderm is controlled by a molecular oscillator known as the segmentation clock

(reviewed in Pourquie, 2001; Pourquie, 2003; Dubrulle and Pourquie, 2004; Aulehla and

Pourquie, 2006). Pairs of somites are laid down at a rate of one every 1.5-2 hours until

E14, when about 65 pairs have formed (Slack, 2006).

By the primitive streak stage, positional information for the anterior and posterior parts of the embryo are marked by expression of factors including, for example, Hex,

Otx2, and Lhx1 (anterior), and Nodal, Wnt3, and T (posterior) (Rossant and Tam, 2009).

Around E8.5 the mouse embryo goes through a process known as “turning”, wherein the posterior region rotates around the long axis of the embryo. This brings the germ layers into the proper orientation and allows for the midgut and neural tube to close (Slack,

2006). Neural tube closure is initiated following embryonic turning around E8.5 at

5 several points along the neural tube, including the hindbrain/cervical boundary, and proceeds both rostrally and caudally, being completed with posterior neuropore closure around E10 (Copp et al., 2000; Ybot-Gonzalez et al., 2002). As axis extension proceeds, mice reach the tailbud stage (the phylotypic stage for mammals) around 30 somites

(E9.5-E10). It is at this stage that Hox genes, which are important for segment identity along the body axis, are maximally expressed (Wilson et al., 2009).

Hox genes are a highly conserved family of 39 helix-turn-helix transcription factors which are organized into 13 paralogous groups (1 to 13) within four chromosomal clusters (A, B, C, and D). These arose by duplications of an ancestral Hox cluster, such that the function of genes within the same paralogous group are highly redundant

(Alexander et al., 2009). Within the clusters all genes have the same transcriptional orientation, and there exists a direct correlation between the arrangement along the chromosome and the timing/boundary of expression along the A-P body axis (Duboule and Dolle, 1989; Kmita and Duboule, 2003). Thus, Hox genes exhibit both spatial and temporal colinearity in that 3‟ Hox genes (Hoxa1, Hoxb1, etc.) are expressed before and more anteriorly than 5‟Hox genes (Hoxa13, Hoxb13, etc.) (Iimura and Pourquie, 2007).

Various mouse knockout models of Hox genes have provided evidence that these genes exhibit posterior prevalence (Duboule, 1994; Duboule, 1998; Iimura and Pourquie, 2007;

Wellik, 2009), meaning that the posterior-most Hox genes expressed in a particular embryonic domain display a dominant function over more anterior Hox genes. Thus, an anterior expansion of Hox expression causes posteriorizations (for example of the hindbrain and axial skeleton), whereas loss-of-function results in anteriorizations. 6

Signalling pathways including FGF (Dubrulle et al., 2001), Wnt (Ikeya and

Takada, 2001), and RA (Kessel and Gruss, 1991) are important for establishing and maintaining the nested domains of Hox gene expression along the vertebral column termed the “Hox code” (reviewed in Iimura and Pourquie, 2007; Alexander et al., 2009) and, in general, it is the Hox code and segmental organization within a specific embryonic compartment which acts to pattern that structure. For example, although all somites are similar morphologically, they will become distinct vertebra (i.e. occipital, cervical, thoracic, lumbar, sacral, or caudal) depending on their axial identity.

Evidence from in vitro and in vivo work supports a role for RA in the early initiation of Hox gene expression both directly and indirectly. RA treatment causes Hox expansion and resultant anterior transformations, whereas reduced RA distribution or inhibition of RA signalling results in posterior transformations which directly correlate with changes in Hox gene expression (Kessel and Gruss, 1991; Niederreither et al., 2000;

Dupe and Lumsden, 2001; and reviewed in Gavalas and Krumlauf, 2000; Wendling et al.,

2001; Gavalas, 2002; Maden, 2002). It is important to note here that studies following exogenous treatment with RA (or indeed RAR antagonists) should be cautiously interpreted as they do not always correlate with the physiological role of RA; abnormally high global levels of RA may induce or repress genes which are not normally regulated by endogenous RA. Furthermore, such RA flooding may obscure some of the more subtle and transient roles of endogenous RA.

RA is also important for regulating elongation of the body axis during tail bud extension and somitogenesis. It is produced by the RALDH2 enzyme within somites and 7 the presomitic mesoderm (PSM), which acts to regulate expression of Fgf8 and controls the bilateral symmetry of somite formation in the early embryo (Vermot et al., 2005;

Sirbu and Duester, 2006). There is also evidence to suggest that there is a mutual negative feedback inhibition between RA in the PSM and FGF8 in the caudal “stem” region (Diez del Corral et al., 2003). This stem region is kept devoid of RA by expression of CYP26A1 (Abu-Abed et al., 2001). Together, this allows not only for axial elongation to proceed, but also for neurogenesis to properly occur (reviewed in Diez del Corral and

Storey, 2004).

2.1.2 The developing limb as a paradigm for morphogenesis

The limb is an important heuristic model for studying the molecular and cellular biology of pattern formation and signal transduction; ultimately leading to morphogenesis of a highly complex but systematically reproducible structure. Much of what we know about limb development has been gathered from analysis of the chick and mouse over many decades and it is, thus, one of the best-studied models of development at a cellular and molecular level.

The limb bud originates as a protrusion of mesenchymal cells encased by ectoderm and proceeds to grow until each of the characteristic limb segments (the stylopod – humerus or femur, zeugopod – radius/ulna and tibia/fibula, and autopod – wrist or ankle and digits) are established (Niswander, 2003). In the mouse, the forelimb bud emerges beginning at E9.5 adjacent to somites 8-12, with the hindlimb buds delayed

8 until E10 at the level of somites 23-28 (Slack, 2006). Within the limb bud, there are integrated signalling centers which act to direct development along the antero-posterior

(A-P; or thumb little finger, radius to ulna/ tibia to fibula), proximo-distal (P-D; or body wall to fingers/toes), and dorso-ventral (D-V; back of the hand/foot to palm/sole) axes.

These signalling centers are known as the zone of polarizing activity (ZPA), located at the posterior limb bud mesenchyme and expressing sonic hedgehog (Shh) which regulates development along the A-P axis (Harfe et al., 2004), the apical ectodermal ridge

(AER) located at the limb bud distal tip which expresses Fgf8, Fgf4, Fgf9, and Fgf17, and regulates limb P-D outgrowth (Mariani et al., 2008), and the dorsal ectoderm which expresses Wnt7a and instructs D-V development (Chen and Johnson, 1999). Interplay between these organizing centers ultimately results in the three-dimensional formation of the stylopod, zeugopod, and autopod (Laufer et al., 1994; Niswander et al., 1994;

Niswander, 2002).

The ability of the limb bud, especially that of the chick, to be manipulated in vivo, and its non-essential nature for survival, made it an ideal system for the early so-called

„cut-and-paste‟ studies of limb development. These early studies focused on physically manipulating the limb bud at different stages of outgrowth and analyzing the final skeletal pattern. Transplantations of small pieces of limb bud mesenchyme to ectopic locations determined that the ZPA was the organizing center in the posterior mesenchyme involved in A-P digit specification, and revealed that the AER was essential for P-D limb outgrowth (Saunders, 1948). If enough ZPA cells are transplanted to the anterior limb bud, digit duplications arise (Tickle, 1981), and this result can be mimicked by adding 9 exogenous RA (Tickle et al., 1982) or Shh (Riddle et al., 1993), although it is clear from loss-of-function models that Shh is the morphogen responsible for limb A-P axis specification and digit patterning (Chiang et al., 2001; Kraus et al., 2001; Harfe et al.,

2004). In terms of P-D outgrowth, removal of the AER at progressive timepoints was found to result in increasingly distal truncations (Saunders, 1948; Summerbell, 1974).

However, if an early and late AER were swapped there was no effect on the skeletal outcome, suggesting that it is the signal from the AER, rather than the AER itself, that was important for outgrowth and that this is equivalent throughout P-D limb development

(Rubin and Saunders, 1972).

2.1.2.1 Models of limb P-D outgrowth: Progress Zone, Early Specification, and Two- Signal

Condensation of the skeletal elements proceeds in a proximal to distal sequence

(i.e. with the stylopod element being formed before zeugopod, and followed by autopod)

(Mariani and Martin, 2003; Niswander, 2003). The “Progress Zone Model” of limb development states that cells in the limb bud remain undifferentiated as long as they are in the “progress zone”, located in the approximately 200µm region below the AER, and under influence of AER signals. As the limb bud grows outward cells move out of the progress zone and differentiate, thus moving in a proximal-to-distal sequence of differentiation (Summerbell et al., 1973). This model was developed to explain the results of the AER extirpation experiments described above, and was supported by later studies

10 involving X-irradiation of the chick limb bud. In these studies, irradiation-induced cell death resulted in smaller limb buds, and it was interpreted that this would allow the proximal cells to be under the control of signals from the AER for an extended period of time – causing re-specification of proximal cells to a more distal fate and resultant phocomelia (Goff, 1962; Summerbell, 1978; Wolpert et al., 1979). The mechanism postulated by these studies has, however, recently been called into question by re- examining the observations using molecular tools which were not available at the time these early experiments were performed (Galloway et al., 2009). These new studies show that irradiation causes cell death, which results in the smaller limb buds, and loss of skeletal progenitors, but that the cells are not re-specified since genes involved in patterning the limb bud segments are unaffected (Galloway et al., 2009).

The Progress Zone model prevailed for nearly 30 years, until Tabin and colleagues presented an alternative explanation for the classical experiments termed the

“Early Specification Model” (Dudley et al., 2002). At the heart of this model of limb outgrowth is the idea that the three limb segments are already specified in the early limb bud at the onset of outgrowth, and that the progenitor pools for each segment then expand as the bud grows. The best markers for each of these limb segments are Meis1 and Meis2

(stylopod), Hoxa11 (zeugopod), Hoxa13 (entire autopod) and Hoxd13 (autopod exclusive of wrist) (Tabin and Wolpert, 2007). However, neither the Progress Zone nor Early

Specification model were able to completely reconcile the compiled and emerging molecular data (Tabin and Wolpert, 2007).

The most recent model of limb P-D outgrowth is the “Two-Signal Model”

(Mariani et al., 2008). This model states that AER-FGFs are instructive for distal limb cell specification, and that an antagonizing proximal signal (which was postulated to be

RA at the time) is required for proximal cell specification. The area where these two signals overlap then specifies the zeugopod (Mariani et al., 2008). Although this model is the most unifying in terms of the available phenotypic and molecular data, it is still a matter of debate whether RA is required for normal limb development (Mackem and

Lewandoski, 2009; Zeller et al., 2009; Zhao et al., 2009), or whether there is a proximal signal at all.

Although the developing limb is one of the best-studied models of vertebrate morphogenesis, there is still much to learn with regards to how the cells within the bud interpret the molecular information that controls their fate. It is becoming clear that the interplay between spatio-temporal morphogen distribution and regulation of interconnected feedback loops, coupled with cell growth and survival, all play a role in development of the final limb structure (Tabin and Wolpert, 2007; Zeller et al., 2009).

2.1.3 Molecular mechanisms of eyelid closure

Another less well known paradigm of mouse development involves initiation and closure of the eyelid. Eyelid formation in developing mice occurs in five stages beginning at E11.5 and ending around E16.5 (Kaufman, 1992; Mine et al., 2005). In the first stage, between E11.5 and E14.5, the protruding ridges of epithelium that make up the immature

12 eyelid (the “eyelid root”) are formed, and the ectoderm begins to proliferate. Next, around E15.0, the columnar epithelial cells of the eyelid leading edge begin to extend from the tip of the root epithelium. In the third stage, between E15.0 and E16.0, both the root and the leading edge continue to extend over the cornea. Around E16.0 the upper and lower eyelid cells appose and fuse together in the center of the eye. Finally, at E16.5 and later, the roots make contact and the epithelium continues to proliferate and pile up at the contact point (Mine et al., 2005). These cells will eventually slough off after birth and the eyelids will reopen between P10-P14 (Findlater et al., 1993).

Several signal transduction pathways have been found to be important for eyelid closure including FGF (Tao et al., 2006), TGFβ/activin (acting through JNK and Jun kinases) (Kao et al., 2008), and EGF signalling at the leading edge (Mine et al., 2005). In addition, mice with mutations in planar cell polarity genes have also been shown to exhibit the open-eyes-at-birth (EOB) phenotype (Wang and Nathans, 2007). Ultimately, in each of these cases, it is the loss of actin polymerization at the leading edge and/or impaired leading edge cell migration which results in EOB. Some genes that have been shown to be essential or important for eyelid closure in the mouse include Fgf10 (Tao et al., 2005), Tgfa (Luetteke et al., 1993; Mann et al., 1993), Egfr (Miettinen et al., 1995),

Activinβb (Vassalli et al., 1994), Mekk1 (Zhang et al., 2003), Fgfr2b (De Moerlooze et al., 2000), Hb-egf (Mine et al., 2005), and Grhl3/Get1 (Yu et al., 2008).

Murine eyelid closure is considered a paradigm for wound healing, with overlapping characteristic morphogenetic steps involving cell proliferation and migration.

These include formation of a leading edge of epithelial cells with actin stress fibers, 13 directed cell migration over an open space, and epithelialization of the “wound” (Martin and Parkhurst, 2004). The involvement of RA is not well known in this process, however work described in this thesis has revealed that the metabolic control of RA is a critical step in eyelid closure, and thus may have an equally important function in the process of wound healing.

2.2 Vitamin A metabolism, mechanism of action, and physiological roles

2.2.1 Vitamin A in nutrition

Vitamin A and its active derivatives, collectively referred to as retinoids, are required for various cellular processes including growth, apoptosis, differentiation, and proliferation. Within the whole organism, regulation of these processes by retinoids has been shown to be essential for the maintenance of reproductive organs, vision, embryogenesis, and tissue homeostasis (Morriss-Kay and Ward, 1999; Mark et al., 2006).

Controlling the intake and availability of dietary vitamin A is important throughout life (Collins and Mao, 1999; Penniston and Tanumihardjo, 2006). Vitamin A deficiency (VAD) in adult mammals leads to the degeneration of many tissues including the retina and testes (Wolbach and Howe, 1978), whereas excess intake results in hypervitaminosis A, with characteristic effects on liver, skin, bone, and the central nervous system (Teelmann, 1989).

2.2.2 Early studies on the role of vitamin A in embryonic development

Maintaining proper levels of vitamin A is especially important during pregnancy for normal embryonic morphogenesis. Early studies revealed that pups from VAD rodents were often maternally resorbed (Mason, 1935) and exhibited malformations of the heart, lung, eye, and urogenital tract (Wilson and Warkany, 1950; Wilson et al.,

1953). Conversely, studies on vitamin A teratogenicity also illustrated a variety of congenital defects, including exencephaly and instances of spina bifida, cleft palate, and various limb, craniofacial, heart, and eye defects (Cohlan, 1954; Kochhar, 1973; Lammer et al., 1985). Through both VAD and teratogenic studies, it became obvious that both the time of treatment and dose of vitamin A played a major role in determining the affected tissues. These early observations led to the awareness that vitamin A plays an important role in fetal development; determining how it was able to elicit its cellular and tissue- specific effects subsequently became a major focus of interest in developmental biology.

2.2.3 Canonical vitamin A metabolism

Despite being physiologically essential, vitamin A (retinol) cannot be de novo synthesized in mammals and is obtained from the diet in the form of plant carotenoids, retinyl esters from animal products, or in fortified foods such as milk and grains

(Blomhoff and Blomhoff, 2006). Up to 80% of a vertebrate‟s total vitamin A is stored as retinyl esters within lipid droplets in hepatic stellate cells (Collins and Mao, 1999). When required, stored retinyl esters are hydrolyzed to retinol and are transported to target

15 tissues via retinol binding protein (RBP), following which retinol undergoes a series of oxidation reactions to convert it into the principle transcriptionally active metabolite all- transRA (atRA), and also the 9-cis retinoic acid (9-cisRA) stereoisomer (Figure 1). For the embryo, maternal retinol is taken up through the yolk sac and placenta.

The first step in the synthesis of RA is the conversion of retinol to retinal in a reversible and rate-limiting step which is catalyzed by the alcohol dehydrogenase and short-chain dehydrogenase/reductase (ADH and SDR) families of enzymes (Duester et al., 1997; Molotkov et al., 2002; Martras et al., 2004; Duester, 2008), perhaps the most important of which during development is the retinol dehydrogenase (RDH)10 (Sandell et al., 2007). The second step is the irreversible conversion of retinal to RA, wherein oxidation is mediated by the retinal dehydrogenase (RALDH) enzymes (Duester, 1996;

Zhao et al., 1996; Fischer et al., 1999; Niederreither et al., 2002b; Lin et al., 2003). In order to limit RA availability, tissues also have the capacity to inactivate RA through spatio-temporally regulated cytochrome P450-mediated oxidation. This step is catalyzed by the CYP26 family of enzymes (CYP26A1, B1, and C1) which specifically oxidize RA to more polar inactive metabolites such as 4-oxo-RA, 4-OH-RA, and 18-OH-RA (White et al., 2000; Chithalen et al., 2002; Taimi et al., 2004) to facilitate excretion (Collins and

Mao, 1999) (Figure 1). Although there is some debate as to whether the metabolites generated by the CYP26 enzymes are biologically active (in particular 4-oxo-RA, which is able to bind RAR), genetic evidence points to the primary effect of CYP26 activity being degradation of RA and clearance of the metabolic products (Niederreither et al.,

2002a). 16

Figure 1: Overview of vitamin A metabolism.

Retinyl esters or carotenoids from animal and plant products, respectively, are obtained from the diet and then converted to vitamin A (retinol) within the intestinal lumen. Dehydrogenation of retinol results in the formation of retinaldehyde via a reversible and rate-limiting oxidation reaction catalyzed by alcohol dehydrogenase enzymes (i.e. ADH1, ADH3 and ADH4) and short chain dehydrogenase/reductase (SDR) enzymes (such as RDH10). All-trans retinal is then converted irreversibly into all-trans retinoic acid (atRA) by the retinaldehyde dehydrogenase enzymes (primarily RALDH2). All-trans RA is the main active metabolite of vitamin A in vivo. Within cells it is able to either modulate gene transcription or is inactivated by catabolism to polar derivatives including 4-oxo, 4- OH, and 18-OH-RA – a process catalyzed by CYP26A1 and CYP26B1. Metabolism of the 9-cis isomer of retinol occurs in a manner similar to that of the all-trans isomer, and 9-cisRA can be metabolized by CYP26C1. The developmental importance of 9-cisRA remains to be proven (dashed arrow).

2.2.4 RA binding proteins

There are a variety of RA binding proteins which play a role in the absorption, storage, and mobilization of vitamin A and its derivatives throughout the metabolic process. Retinol that is secreted from the liver is transported in the blood via the retinol binding protein RBP4 (Quadro et al., 1999). Similarly, in the embryo, retinol is taken up from the placenta by RBP4 (Quadro et al., 2004). Retinol uptake into cells then occurs through the cell-surface receptor stimulated by RA (STRA)6 (Kawaguchi et al., 2007).

Within the cell, retinol is bound by the cellular retinol binding proteins (CRBPs) (Noy,

2000) which act to facilitate conversion of retinol to retinyl esters for storage or retinaldehyde for RA synthesis. It is not clear, however, whether binding by CRBP is particularly necessary for RA synthesis, since free retinol is the preferred substrate for retinaldehyde conversion (Gallego et al., 2006) and mice deficient in CRBP1 have normal levels of RA synthesis (Ghyselinck et al., 1999).

The cellular RA binding proteins (CRABPs) are thought to facilitate cellular uptake of RA and shuttling to the nucleus via a nuclear localization signal (at least for

CRABP2) (Sessler and Noy, 2005). However, these binding proteins are dispensable for embryonic development and RA synthesis/signalling both alone and in combination (de

Bruijn et al., 1994; Gorry et al., 1994; Lampron et al., 1995). It should be noted that, although the necessity of the retinol and RA binding proteins for embryonic development under conditions of vitamin A sufficiency is non-essential, this does not preclude their requirement under VAD conditions or under other environmental/genetic stresses.

2.2.5 Transcriptional regulation by RA

The biological effects of RA on developmental processes and tissue homeostasis were known for decades before the general mechanism by which it exerted these effects was uncovered. A breakthrough in the field of retinoids came in 1987 with the cloning of the first member of the retinoid nuclear receptor family, RARα (Giguere et al., 1987;

Petkovich et al., 1987). This transcription factor belongs to the nuclear hormone family of receptors and to bind RA with high specificity, a finding which established that RA could act as a transcriptional regulator. Subsequent characterization of the first RXR family member in 1990 (Mangelsdorf et al., 1990), shown later to be the heterodimeric binding partner of the RARs (Leid et al., 1992b), established a mechanism by which RA could elicit its pleiotropic cellular and biological effects.

Within each of the two receptor subfamilies (RAR and RXR) there are three isotypes (α, β, and γ), each encoded by separate genes and with multiple isoforms (α1-2,

β1-4, and γ1-2) resulting from alternative splicing and the differential use of promoters

(Leid et al., 1992a; Chambon, 1996). In addition to dimerizing with RAR, RXRs have also been shown to exhibit promiscuity by heterodimerizing with other members of the nuclear receptor superfamily including thyroid hormone receptor, vitamin D receptor, and the peroxisome proliferator-activated receptors (Yu et al., 1991; Leid et al., 1992b; Zhang et al., 1992; Bardot et al., 1993).

RARs are able to bind both atRA and 9-cisRA equivalently in vitro (Allenby et al., 1993; Chambon, 1996). Although the importance of atRA within the developing

20 embryo has been made clear by genetic and biochemical studies (Ulven et al., 2000;

Niederreither et al., 2001; Niederreither et al., 2002a), the necessity of 9-cisRA in development is still controversial. This is evidenced by the fact that 9-cisRA has not been detected in embryos (Horton and Maden, 1995), even using a very sensitive HPLC method (detection limits on the order of fmol) (Ulven et al., 2000) and recent in vivo evidence which supports the idea that RAR-RXR transactivation only requires RAR to be liganded (Mic et al., 2003). Although there is support for 9-cisRA as the presumptive natural ligand for the RXRs in vitro (Levin et al., 1992; Allenby et al., 1993), RXRs also bind oleic acid (Bourguet et al., 2000), phytanic acid (Kitareewan et al., 1996), docosahexanoic acid (de Urquiza et al., 2000) and various other unsaturated fatty acids

(Goldstein et al., 2003). These ligands bind RXRs with less affinity than 9-cisRA and their binding depends on RXRs dimerization partner.

The RAR and RXR receptors act as heterodimeric ligand-activated transcription factors by binding promoters at specific bipartite DNA motifs referred to as RA response elements (RAREs) within RA target genes (Xiao et al., 1995; Chambon, 1996;

Gronemeyer et al., 2004). The general mechanism for retinoid-regulated gene activation is that unliganded receptors bound to DNA repress transcription (Dilworth and Chambon,

2001) by interacting with corepressors such as NCoR (Horlein et al., 1995) and SMRT

(Chen and Evans, 1995), and recruiting proteins with histone deacetylase activity

(HDACs), keeping the chromatin in a compacted form (Figure 2A). It should be noted that there are also ligand-dependent corepressors such as LCoR (Fernandes et al., 2003), however, in general, RA binding induces an allosteric change in the receptor (Depoix et 21

Figure 2: Canonical activation of RA-responsive genes.

(A) In the absence of ligand the RAR/RXR heterodimer is bound to the RARE within the promoter region of an RA-responsive gene. Gene repression via chromatin compaction results from apo-receptor association with a large protein complex containing various corepressors, which can have histone deacetylase activity, such as NCoR/SMRT and RIP140. (B) Ligand binding (atRA is represented as an orange circle) causes a conformational change in the RAR which includes a repositioning of helix 12 (H12). This induces transcriptional activation through the release of corepressors and concomitant recruitment of various coactivator proteins with histone acetylase activity (e.g. members of the p160 family/SRC-1) and/or ATP-dependent chromatin remodelers like SWI/SNF. This leads to chromatin decompaction, the recruitment of the basal transcription machinery, and activation of transcription. The main histone modifications are shown in each case. (modified from Pennimpede et al., 2006 and reproduced with permission from Elsevier)

23 al., 2001) which causes the dissociation of corepressors and the concomitant recruitment of coactivators, such as SRC-1/p160 or p300/CBP (Gronemeyer et al., 2004). Many coactivators exhibit intrinsic histone acetylase or histone methyltransferase activity, while others engage protein complexes which elicit the requisite chromatin decompaction, such as ATP-dependent chromatin remodelers like SWI/SNF (Kwon et al., 1994; Bastien and

Rochette-Egly, 2004; Gronemeyer et al., 2004). This allows for the preinitiation complex to become poised, and facilitates the binding of basal transcription machinery to the promoter (Figure 2B). In vitro and in silico work has determined that more than 500 genes may be able to be regulated by RA either directly, through RARE-mediated transcription, or indirectly (Balmer and Blomhoff, 2002).

2.3 RA and morphogenesis

A classical concept in developmental biology, known as the morphogen theory, hypothesizes that positional information is conferred to nascent cell fields based on the presence of morphogens – secreted molecules which are able to set up concentration gradients from some localized “source”, and which differentially activate the transcription of genes depending on distance from that source, or until they encounter a

“sink” (Wolpert, 1969; Meinhardt, 1978; Meinhardt and Gierer, 1980; Wolpert, 1994;

Kerszberg and Wolpert, 1998; Affolter and Basler, 2007). It has recently been suggested that the interpretation of a morphogen gradient by cells/tissues/organisms is more

24 complex than originally thought (reviewed in Kerszberg and Wolpert, 2007; Tabin and

Wolpert, 2007; Zeller et al., 2009). Spatio-temporal oscillations of morphogen concentrations, cell-cell signalling, and interconnected signalling feedback loops have been shown to be involved in embryonic somite formation (Dubrulle and Pourquie, 2004;

Wahl et al., 2007) and limb development (Benazet et al., 2009); and this is likely a paradigm for many other developmental structures (Kerszberg and Wolpert, 2007).

Precise regulation of morphogens is necessary during development considering that expression of genes and resultant pattern formation is a slow process relative to the movement of morphogens between cells. In a spatially dynamic system, stable gradients can be established when morphogen production is regulated from a focal source and is coupled to a distal perimeter of facilitated degradation, forming a peripheral sink (Lander,

2007). In the case of RA, RALDH enzymes could act as the source and CYP26 enzymes as the sink within tissues and throughout development (Figure 3) (Pennimpede et al.,

2006; Niederreither and Dolle, 2008). Since RA is a small lipophilic molecule it is able to diffuse through cell membranes, and was the first candidate morphogen described in vertebrates (Thaller and Eichele, 1987) although its classification as a morphogen has since been debated. Many other candidate morphogens have been identified, including members of the FGF, Wnt, BMP, TGF, and Hh families (Jones and Trainor, 2005; Ashe and Briscoe, 2006, and references therein). RA is different from these other signalling molecules in that it is not a protein, it binds to a nuclear rather than cell-surface receptor, and it directly activates target genes as opposed to intracellular signalling cascades.

Figure 3: Overview of RA signalling in embryonic patterning and morphogenesis.

(A) Strict spatio-temporal regulation of RA is necessary in order to ensure that the normal developmental program proceeds. Within developing tissues the expression of RA synthesizing (RALDH) and RA catabolizing (CYP26) enzymes are responsible for setting up areas of RA availability. Where RALDHs are present (green shading) they convert retinal (RAH; gold hexagon) into RA (orange circle) which is able to bind and transactivate RAR/RXR heterodimers. Where CYP26s are expressed, RA is oxidized and inactivated (red shading). This creates areas within the tissue where RA is active (the activation threshold) which, in general, leads to cell differentiation or apoptosis; whereas the absence of RA can allow for proliferation to occur. (B) Loss of CYP26 activity allows RA to diffuse into cells from which it is normally excluded, and thus skews the tissue response towards premature differentiation and/or cell death as opposed to proliferation. This is a generalized paradigm and it is important to note that the type of cellular response is controlled by not only the presence or absence of RA, but also the type and availability of the receptors and target genes, and is dependent also on the tissue type and stage of development.

Moreover, it has been proven to be only permissive for the development of some tissues, as opposed to instructive (Duester, 2008; Zhao et al., 2009).

2.4 Regulation of RA signalling during embryogenesis

The various cellular processes and vast number of genes regulated by RA suggests that RA signalling needs to be tightly constrained to ensure that proper stage- and tissue-specific signalling cascades are activated during morphogenesis. One such mechanism for controlling receptor-mediated ligand-dependent activation or repression is by restricting the availability of receptor isotypes within tissues.

2.4.1 RAR/RXR expression and function

Transcripts encoding the RARs are widely distributed throughout the embryo during development. Investigations into the murine expression patterns of Rara, Rarb,

Rarg and their isoforms have been thoroughly performed between embryonic day E8.5 and E14.5. Rara shows nearly ubiquitous expression while the domains of Rarb and Rarg expression are significantly more restricted (Ruberte et al., 1991; Mollard et al., 2000;

Dolle, 2009). Major sites of Rarb expression are within the foregut endoderm, urogenital region, spinal cord neuroepithelium, in the eye and olfactory mesenchyme, and later in the limb interdigital mesenchyme (Dolle et al., 1989; Mollard et al., 2000). Rarg transcripts are highly expressed in the open caudal neural tube and tailbud starting at

E8.5, and later within frontonasal and branchial arch mesenchyme, limb buds, and somitic sclerotome (Ruberte et al., 1990). From E12.5, transcripts can be detected within precartilaginous condensations of the trachea, limb, and developing vertebrae, and within epithelia of the oral cavity and stomach (Ruberte et al., 1990; Mollard et al., 2000). Later expression of Rarg (E13.5-14.5) is found within developing teeth, whisker follicles, skin, and developing eyelids. Expression of Rarg is lost in the presomitic mesoderm when somites appear and also in precartilaginous condensations at the onset of chondrogenesis

(Dolle et al., 1990; Ruberte et al., 1990).

Gene ablation studies of particular RAR isotypes, or one isoform of a particular isotype, have been performed. It was expected that embryos would recapitulate the malformations associated with VAD. However, the single receptor knockouts, although often embryonic or neonatal lethal, did not mirror all VAD phenotypes, suggesting that functional redundancy existed between the receptors (Lohnes et al., 1993; Lufkin et al.,

1993; Kastner et al., 1996; Krezel et al., 1996; Ghyselinck et al., 1997). This was confirmed by generating animals which were compound-null for several of the receptor isoforms or isotypes, which collectively exhibited a variety of congenital malformations typifying those associated with VAD syndrome, and even sometimes presented with defects not previously observed in VAD animals, such as craniofacial defects, abnormal patterning of the spinal column, and limb skeletal defects (Lohnes et al., 1994;

Mendelsohn et al., 1994; Grondona et al., 1996; Luo et al., 1996; Ghyselinck et al., 1997, and reviewed in Lohnes et al., 1995; Mark et al., 2006; Mark et al., 2009).

2.5 Regulation of RA distribution within the developing embryo

In addition to limiting RA signalling by regulating its receptors, another overarching means by which to limit the signalling potential of RA in the embryo is to restrict RA synthesis and catabolism.

2.5.1 Expression patterns of Raldh and Cyp26 isoforms in the developing mouse embryo

There are four members of the RALDH family (RALDH1-4), but kinetic and genetic studies have illustrated that RALDH2 is the main enzyme involved in catalyzing conversion of all-trans retinal to atRA in the embryo, with RALDH3 followed by

RALDH1 involved to a lesser degree (McCaffery and Drager, 1995; Niederreither et al.,

1997; Mic et al., 2002; Niederreither et al., 2002c).

The tissue-specific expression patterns of the RALDH enzymes in the embryo closely correlate with active RA signalling. Raldh2 is the first to be expressed at E7.5 within the mesoderm in the posterior portion of the embryo, which coincides with the ability to detect RA using the RARE-lacZ reporter (Uehara et al., 2009). This reporter transgene contains 3 copies of a canonical RARE derived from the RARβ gene, upstream of hsp68-lacZ (Rossant et al., 1991). It is a widely used and highly sensitive reporter that allows for detection of all cells where RA is present and active (Uehara et al., 2009; Zhao et al., 2009). From E8.5-9.5 Raldh2 is detected within the trunk, particularly in developing somites and the cervical mesenchyme, and at later stages within discrete regions of the developing face including tooth buds and inner ear, the interdigital region

30 of the limbs, and the lower genital tract (for more detail see Niederreither et al., 1997).

During development, Raldh1 is expressed in the dorsal retina and Raldh3 in the ventral portion of the retina (McCaffery et al., 1999; Li et al., 2000; Mic et al., 2000; Suzuki et al., 2000; Wagner et al., 2000) and these isoforms are also expressed in a restricted fashion within organs during later developmental stages (Niederreither et al., 2002b). The more recent discovery of Raldh4 has shown expression of this isoform within adult mouse liver and kidney (Lin et al., 2003) but it is undetected during embryogenesis

(Niederreither and Dolle, 2006).

The expression patterns of Cyp26a1, b1, and c1 have also been studied throughout mouse embryonic development. All three Cyp26 genes are expressed during early development in extraembryonic regions surrounding the epiblast. Between E5.5 and E6.5

Cyp26a1 is detected in the extraembryonic endoderm and ectoderm, and the visceral endoderm, whereas Cyp26b1 and Cyp26c1 are found at a lower level in the ectoplacental cone (Uehara et al., 2009). As discussed above, Raldh2 expression in the embryo begins at E7.0-7.5 meaning that the Cyp26 enzymes are expressed before the onset of embryonic

RA production. They have been found to be required at this stage in order to protect the embryo from maternal RA, and for the proper expression of Nodal (Uehara et al., 2009), which is essential for establishing the embryonic body axis and left-right asymmetry.

Following gastrulation, the first detectable Cyp26 isoform in the embryo is

Cyp26a1 at E7.0 within the embryonic endoderm (Fujii et al., 1997) and then transiently in the developing headfolds and cardiac crescent by E7.0-7.5 (MacLean et al., 2001).

Transient expression of Cyp26a1 has also been observed around E8.5 within the cranial 31 neural crest and prospective hindbrain rhombomere (r)2 (Fujii et al., 1997; Moss et al.,

1998; MacLean et al., 2001). Expression is particularly strong in the tailbud region, starting around E8.0 and persisting until E11.0-11.5, during which time Cyp26a1 expression is also found in the maxillo-mandibular cleft and within the cervical mesenchyme surrounding the first and second branchial arches (Fujii et al., 1997; de

Roos et al., 1999; MacLean et al., 2001). At later stages (>E11.5), Cyp26a1 expression is restricted within the eye, mouth, and ear, as well as stomach wall, septum transversum

(developing diaphragm) and interdigital mesenchyme (de Roos et al., 1999; Abu-Abed et al., 2002). Expression of Cyp26a1 is upregulated upon RA-treatment in vivo and in cells

(Abu-Abed et al., 1998; Iulianella et al., 1999) owing to the presence of two RAREs within its promoter region, which are synergistic in their response to RA (Loudig et al.,

2000; Loudig et al., 2005).

Expression of Cyp26b1 in the embryo proper was detected starting at E8.0 within prospective r3 and r5, and intersomitically at E8.5. This expression extended to encompass r2 through r6 by E9.5, a pattern that persists until E10.5 (MacLean et al.,

2001). The Cyp26b1 isoform is also strongly expressed within the developing limbs (E9.5 through E13.5) and within the frontonasal process (MacLean et al., 2001; Abu-Abed et al., 2002; Schmidt et al., 2009). During later stages of development (after E14.5), expression is maintained within the frontonasal process, and additionally found within the brain, spleen, hair follicles, eyelid, prevertebrae, genital tubercle mesenchyme, and the developing spinal cord (Abu-Abed et al., 2002).

Expression of Cyp26c1 in the embryo has been analyzed by our lab between E7.0 and E18.5 (Tahayato et al., 2003). It is detected in the amnion starting at E7.5, and then within prospective r2 and r4 by E8.0. Between E8.5 and E10.5 expression was strongly detected in the maxillo-mandibular cleft and the lateral cervical mesenchyme, and at later stages of development (E11.5-E18.5) Cyp26c1 is localized to the mouth region, especially within tooth anlagen.

Taken together, these expression studies reveal that the local control of RA is finely established by the specific generation of domains of RA synthesis and catabolism which change over time, as dictated by the genetic programs controlling morphogenesis.

2.5.2 Raldh and Cyp26 knockout studies

Information about the biological importance of maintaining appropriate levels of

RA during development has been gathered from genetic ablation studies of the Raldh and

Cyp26 genes. Murine knockout models have been created and analyzed for Raldh1-3, and all of the Cyp26 genes.

Raldh1-null mutants were viable and had no recognizable defects in vision, even though this gene is mainly expressed in the retina, although liver production of RA was reduced which suggested that RALDH1 is able to catalyze the synthesis of RA in vivo

(Fan et al., 2003). Deletion of the Raldh3 gene results in discrete defects confined to the nasal and ocular regions, with neonates dying shortly after birth due to respiratory distress as a result of choanal atresia (Dupe et al., 2003). Raldh1/3 compound mutants 33 display much more severe ocular malformations than those found in Raldh3-null animals alone (Matt et al., 2005), and which resemble the eye defects observed following VAD or loss of RXRα or RARβ and RARγ in the neural crest cell (NCC)-derived periocular mesenchyme (Matt et al., 2005; Duester, 2009).

Ablation of Raldh2 caused the most severe abnormalities with embryos dying at midgestation (between E9.5-10.5) and failing to undergo axial rotation, exhibiting arrested development at E8.5 (Niederreither et al., 1999). Raldh2-null embryos also have a shortened antero-posterior axis and frontonasal region, they fail to grow limb buds, and their heart presents as a single, dilated cavity (Niederreither et al., 1999). Most of the fetal abnormalities in Raldh2-/- embryos could be rescued by maternal administration of exogenous RA between at least E7.5 and E8.5 (Niederreither et al., 2001; Niederreither et al., 2003). When RA signalling was examined in the Raldh2-/- embryos using the RARE- lacZ reporter transgene, Raldh2 null embryos failed to activate the reporter in nearly all tissues (Niederreither et al., 1999; Mic et al., 2002). In addition, RALDH2 activity is essential for maintaining bilateral symmetry of oscillating waves of gene expression during somite formation (Vermot et al., 2005; Sirbu and Duester, 2006). Taken together, these data reinforce the idea that Raldh2 is the main RA-synthesizing enzyme during embryogenesis.

Finely controlled RA catabolism by CYP26A1, CYP26B1, and to a lesser degree

CYP26C1 has been shown to be a requirement for proper embryonic development.

Ablation of Cyp26a1 results in embryonic lethality at late gestation, and embryos present with acute developmental defects mainly involving the lumbosacral region. The 34 phenotype is variable, with the most striking malformations including spina bifida, posterior body truncations, sirenomelia (mermaid fusion of the hind limbs), and exencephaly (Abu-Abed et al., 2001; Sakai et al., 2001). Other mutations included fused kidneys, posterior transformations of cervical vertebrae and rhombomeres, and irregular folding of the neural tube (Abu-Abed et al., 2001; Sakai et al., 2001). Cyp26a1-/- embryos expressing the RARE-lacZ transgene revealed an expansion of reporter activity into the caudal region (Sakai et al., 2001), illustrating that the posterior defects are very likely the result of ectopic RA signalling.

The more recent characterization of Cyp26b1-null embryos has revealed that these mice die shortly after birth due to respiratory distress, and present with severe limb malformations such as phocomelia of the fore and hind limbs, adactyly, micrognathia, and they lack eyelid fusion at birth (Yashiro et al., 2004; Maclean et al., 2009). In addition, Cyp26b1-null embryos have hypoplastic testes due to apoptosis of germ cells

(MacLean et al., 2007) and we have shown that CYP26B1 is a masculinizing factor in that it is required to keep male germ cells in an undifferentiated state (Li et al., 2009).

The Cyp26b1-/- embryos also display a shortened frontonasal region, cleft palate, abnormal body shape, lack vibrissal and pelage follicles, and have various defects of skeletal development (evidence from this work and Maclean et al., 2009). It is noteworthy that, although Cyp26b1-null embryos exhibit various malformations which are characteristic of defects in NCCs, markers of branchial arch patterning along with

NCC patterning and migration were found to be mainly unchanged from wild type

(Maclean et al., 2009). Also, regardless of the strong expression of Cyp26b1 within the 35 hindbrain (MacLean et al., 2001), an examination of RARE-lacZ reporter activity does not reveal an anterior expansion of RA into this tissue following loss of CYP26B1 alone

(Maclean et al., 2009), but RA expansion was detected in the limb buds using this reporter (Yashiro et al., 2004).

Genetic ablation of Cyp26c1 does not result in an overt phenotype, but targeted deletion of genes for both Cyp26a1 and Cyp26c1 results in serious defects of the developing brain, migrating NCCs and derived structures, and death by E9.5-10.5

(Uehara et al., 2007), which suggests that CYP26B1 is unable to fully compensate for the loss of CYP26A1 and CYP26C1 in the hindbrain. Targeted deletion of all three Cyp26 genes has been recently performed and was found to result in early embryonic lethality

(~E9.5), sometimes accompanied by duplications of the body axis due to aberrant Nodal expression (Uehara et al., 2009). Unfortunately, the early lethality of these mutants precludes further analysis on the morphogenesis of specific tissues lacking all CYP26 activity.

2.6 Rarg-null mice are resistant to RA-induced craniofacial and caudal defects

Examination of the expression pattern of Rarg during development (see above) suggested a role for RARγ in the formation of somites and somite-derived structures, and in chondrogenesis. Targeted disruption of all isoforms of Rarg did not lead to embryonic lethality, but only 50% of null homozygotes were found at weaning and 40% survived to 36

3 months of age (Lohnes et al., 1993). Rarg-null mice exhibited several mild phenotypic malformations including growth deficiency, male sterility, and congenital defects including homeotic transformations of cervical vertebrae and tracheal cartilage fusions

(Lohnes et al., 1993). More interestingly, although exposure to RA at E8.5-E9.0 causes homeotic vertebral transformations, craniofacial malformations, spina bifida, and near complete caudal truncations in wild type embryos (Kessel, 1992), Rarg-/- embryos (and to a lesser extent Rarg+/- embryos) were resistant to these effects. There was an especially prominent reduction in the instances of spina bifida and posterior truncations, as none of the Rarg-/- embryos exhibited these defects (Lohnes et al., 1993). This led to the discovery that RARγ was the isotype exclusively responsible for transducing the retinoid signal in the caudal region of the embryo. Subsequently it was shown that excess RA causes caudal defects by attenuating expression of Wnt3a and its downstream target

Brachyury (T), in the tailbud, through RARγ (Iulianella et al., 1999).

Further analyses of RA-treated Rarg heterozygotes and null mice revealed that they are also resistant to RA-induced resorption, as well as the craniofacial and vertebral malformations normally seen following RA treatment at E7.3 (Iulianella and Lohnes,

1997). The craniofacial defects that were rescued included anencephaly, exencephaly, micrognathia, microcephaly, and fusions of the basi and exoccipital bones of the skull.

Some vertebral transformations characteristic of Rarg ablation were seen at a lower frequency in RA-treated null embryos, while cleft palate was observed at the same frequency as in RA-treated wild type embryos (Iulianella and Lohnes, 1997).

Interestingly, although Rarg is expressed in the limbs throughout development,

Rarg-/- mice were not found to be resistant to RA-induced limb defects (Iulianella and

Lohnes, 1997), however recent work shows that Rara1-/-/Rarg-/- double knockout limbs in culture are more resistant to retinol treatment than limbs of heterozygote or wild type littermates (Galdones et al., 2006). This suggests that RARγ and RARα likely have some overlapping functions during limb outgrowth. Notably, ablation of Rxra results in null animals that exhibit complete resistance to the limb defects associated with pharmacological RA treatment (Sucov et al., 1995).

2.7 Cyp26a1/Rarg double knockout mice

The original observation that Rarg-null and Rarg+/- embryos were resistant to spina bifida and caudal truncations normally caused by exogenous RA (Kessel, 1992;

Lohnes et al., 1993), and the finding that mice lacking CYP26A1 exhibited caudal regression consistent with RA teratogenesis, suggested that removing RARγ in the absence of CYP26A1 should restore formation of the posterior embryo. The Cyp26a1-/-

/Rarg-/- (also referred to as A1-/-/γ-/-) mice were subsequently bred and analyzed in our lab

(Abu-Abed et al., 2003). At all stages examined, Cyp26a1/Rarg compound-null animals were recovered below the expected Mendelian frequency, but disruption of Rarg could rescue the embryonic lethality resulting from loss of CYP26A1 in approximately 50% of remaining cases. The loss of one allele of Rarg in A1-/- embryos led to an intermediate

38 caudal phenotype that was incompletely penetrant and did not restore viability (Abu-

Abed et al., 2003). A thorough examination of the skeletal elements indicated that some anterior vertebral abnormalities normally seen in A1 and γ knockout embryos were abolished in A1-/-/γ-/- mice, whereas more posterior homeotic transformations of the A1-/- animals could not be reversed (Abu-Abed et al., 2003).

Molecular characterization of tailbud morphogenesis in control versus A1-/-, γ-/-, and A1-/-/γ-/- embryos revealed that excess caudal RA following disruption of Cyp26a1 caused downregulation of genes known to be important for caudal development including

Wnt3a, Fgf8, and downstream targets like T, Cdx4, and Hoxd11. Expression of these genes was found to be restored in the A1-/-/γ-/- tailbud (Abu-Abed et al., 2003).

The finding that Rarg ablation in the Cyp26a1-null background could rescue not only the lethality, but also the posterior defects associated with Cyp26a1 disruption, further illustrates the importance of maintaining a fine balance of both the receptor availability (to mediate the RA signal) and RA distribution itself (by RALDH2 and

CYP26s) in proper patterning and development.

2.8 Hypothesis and Specific Objectives

We have previously shown that Rarg ablation can rescue the caudal defects associated with RA excess in Cyp26a1-null mice (described above and in Abu-Abed et al., 2003). In addition to resistance to caudal agenesis, Rarg-/- mice showed resistance to 39 various craniofacial abnormalities (see above and Iulianella and Lohnes, 1997) which are not manifested by deletion of CYP26A1 (Abu-Abed et al., 2001; Sakai et al., 2001). The recent generation of Cyp26b1 knockout mice in our lab illustrate that these mutants are subject to severe craniofacial abnormalities and reduced ossification of various structures of the skull. Therefore, we hypothesized that ablating Rarg in the Cyp26b1-null background would rescue the craniofacial and skeletal defects resulting from RA excess in a manner similar to the rescue of caudal development seen in Rarg/Cyp26a1-/- animals.

In addition to elucidating the mechanisms of RA regulation within specific structures which are mediated by CYP26B1 and RARγ, it was hoped that this model would allow for an examination of the role of CYP26B1 in adult tissues, depending on the level of rescue – an examination which is precluded by the lethality of B1-null mice.

The objectives of the current project were, therefore, to intercross B1+/-/γ+/- heterozygous mice and thoroughly examine the phenotype of the compound-mutant embryos, especially with regards to the craniofacial region. Further, lethality was monitored, and the resulting gross morphological and skeletal phenotypes were scored. These analyses revealed a partial rescue of aspects of the axial skeleton, along with rescue of vibrissae

(described in Chapter 4). Surprisingly, limb outgrowth was also recovered, especially of the proximal limb. Molecular analysis of the early limb bud revealed that this was caused by reduced cell death in compound-null embryos, and clarified the patterning defects associated with inappropriate RA exposure in this tissue ((Pennimpede et al., 2010) and described in Chapter 5). Eyelid fusion was also rescued in the compound-null neonates

(Chapter 4) and molecular analysis of genes essential for formation of the eyelid root and 40 leading edge was performed in order to uncover the mechanism of the observed EOB phenotype (Chapter 7). It was determined that the EOB phenotype does not result from impaired FGF signalling. For reference, a summary of the expression patterns of

Cyp26a1, Cyp26b1, and each Rar isotype in the limb, eye, and craniofacial structures can be found in Table 1 below.

Table 1: Expression patterns of Cyp26a1, Cyp26b1, Rara, Rarb, and Rarg in craniofacial structures, eye, and limbs during murine development.*

Cyp26a1 Cyp26b1 Rara Rarb Rarg r2-6 (E9.5- Hindbrain Rhombomere 10.5), BAs (E8.5-9.5), (r)2 (transient (E8.5-11.5), frontonasal E8.5), maxillo- nasal process mesenchyme mandibular (E11.5-13.5), (E8.5-10.5), Facial and BA cleft (E9.5), palatal shelves, tongue and mesenchyme cervical mandibular mandible (E8.5-11.5), mesenchyme bone, laryngeal (E12.5), nasal precartilaginous Craniofacial (E9.5-10.5), cartilage, cavity, inner condensations structures branchial arch tongue/ear ear (E13.5- (E11.5-E14.5), (BA)1 (E11.5), mesenchyme 14.5), tracheal mouth and (E13.5), pons, dental lamina mesenchyme tongue cerebellum, (E14.5), (E10.5-12.5) epithelium hippocampal tracheal (E13.5) field, forebrain epithelium striatal cells (E12.5) (E14.5-16.5)

Otic capsule Ubiquitous Neural retina (E12.5-13.5), Periocular Periocular at all stages (medial), prospective mesenchyme mesenchyme examined prospective RPE, (E10.5-14.5), (E10.5-14.5), Eye retinal pigment mesenchyme RPE, vitreous lids (E14.5-adult) epithelium underlying body (E12.5) (RPE) (E14.5) palpebral fissure (E14.5) Fore/hindlimb Weak buds (E9.5- Limb bud expression 11.5; no AER Proximal mesenchyme flanking apical- E11.5), limb bud (mainly Rarg2) ectodermal anterior/ mesenchyme (E9.5-E11.5), ridge (AER) posterior (E9.5-11.5), precartilaginous Limbs (E11.5), distal footplate interdigital digit limb ectoderm mesenchyme mesenchyme condensations (E12.5) (E12.5), (E12.5-14.5) (mainly Rarg1) distal digit (E13.5-14.5) mesenchyme (E14.5)

* Compiled from (Dolle et al., 1990; Ruberte et al., 1991; Mollard et al., 2000; MacLean et al., 2001; Mori et al., 2001; Abu-Abed et al., 2002; Dolle, 2009)

Chapter 3

Materials and Methods

3.1 Creation of mouse strains and genotyping Generation of the mice lacking Cyp26b1 along with genotyping methods by PCR, has been previously described (Maclean et al., 2009). Creation of mice lacking all isoforms of Rarg and PCR genotyping have also been previously described (Lohnes et al., 1993; Iulianella and Lohnes, 1997), and these animals were a kind gift from Dr.

David Lohnes (University of Ottawa). Briefly, DNA was isolated from either tail clips at

10 days of age or yolk sacs from embryos using either the HotSHOT method (tail clips)

(Truett et al., 2000), or DirectPCR Lysis Reagent (yolk sac) (Viagen Biotech Inc, Los

Angeles, CA) and proteinase K (Sigma-Aldrich, Oakville, ON, Canada). Three µl of total genomic DNA was then directly used for PCR genotyping. Standard PCR was performed using Taq DNA polymerase (2.5U per reaction; Sigma) and 3 primers for each gene.

Primer sequences for Cyp26b1 were P1: 5‟-CAGTAGATGTTTGAGTGACACAGCC,

P2: 5‟-GAGGAAGTGTCAGGAGAAGTGG and P3: 5‟-GGGCCACCAAGGAAGATG

CTGAGG, where P1 and P2 amplify a product of 223bp from the wild type allele but will not amplify a null allele. P1 and P3 will amplify a 364bp product from the null allele only. For Rarg the primers used were A1: 5‟-CAGAGCACCAGCTCGGAGGA, A2: 5‟-

CTTCACAGGAGCTGACCCCA, and A3: 5‟-GGCCGGAGAACCTGCGTGCAATCC, where A1 and A2 amplify a 120bp band from the wild type allele, and A2 and A3

43 amplify a band of approximately 700bp from the transgenic allele. For Cyp26b1 the reactions included 1.5mM MgCl2, 0.2mM dNTPs, 0.4mM P1, and 0.2mM P2 and P3 in

1x PCR Buffer. For Rarg analyses the reaction mix consisted of 1.0mM MgCl2, 0.2mM dNTPs, 0.2mM A1 and 0.2mM A2, or 0.2mM A1 and 0.2mM A3 in 1x PCR Buffer. The reactions were performed in a Mastercycler gradient thermocycler (Eppendorf Canada,

Mississauga, ON) using the following reaction conditions: 35 cycles of 95˚C for 30s,

60˚C for 45s, and 72˚C for 50s. Electrophoresis was performed in 2% agarose gel and products were visualized using ethidium bromide.

The mice used in this study were of a mixed genetic background and all embryonic genotypes were generated by intercrossing Cyp26b1+/-/Rarg+/- animals (also referred to as B1+/-/γ+/-) which are viable, fertile, and indistinguishable from wild type.

All animal experiments were performed in ethical accordance with protocols approved by the University‟s Animal Care and Use Committee.

3.2 Embryo Collection Timed matings were established by housing female mice with males overnight.

Females were checked each morning for vaginal plugs, and noon on the day that a plug was found was designated E0.5. For embryo collection on the appropriate day, pregnant females were humanely euthanized by cervical dislocation and embryos were collected into cold phosphate-buffered saline (PBS).

3.3 Skeletal analysis Fetuses at E18.5 were collected, skinned, eviscerated, and fixed overnight in

100% ethanol. Cartilage and bone were stained with Alcian blue and Alizarin red, respectively, following standard procedures (Jegalian and De Robertis, 1992). Briefly, following fixation embryos were stained for 48 hours with 0.015% alcian blue (Sigma) in

20% glacial acetic acid/80% ethanol (v/v). After dehydrating overnight in 95% ethanol, embryos were placed in 1% KOH until ribs became clearly visible (usually 8-10 hours).

Bones were then stained overnight with a solution of 0.015% alizarin red (Sigma) in 1%

KOH, and skeletons were subsequently cleared with several changes of 20% glycerol/

1% KOH until ready to be photographed. A total of 31 fetuses, including 3 B1-/-/γ-/- embryos, were scored for skeletal abnormalities.

3.4 Whole-mount in situ hybridization Staged embryos were pooled according to age and genotype and whole-mount in situ hybridization (WMISH) was performed according to standard protocol (Chotteau-

Lelievre et al., 2006). Antisense riboprobes were synthesized using the appropriate RNA polymerase and digoxigenin-11-UTP (DIG) labeling mix (Roche Diagnostics, Laval, QC,

Canada). Anti-DIG antibody conjugated with alkaline phosphatase (AP) (Roche) was incubated with embryos overnight at 4°C, and detection of AP activity was carried out using BCIP/NBT (BioShop Canada Inc, Burlington, ON, Canada) or BM Purple (Roche).

For each probe, embryos were processed concurrently and staining reaction times were maintained between genetic groups to control for variations in signal intensity. At least 3 45 control and Cyp26b1-/- (including Cyp26b1-/-/Rarg+/-), and 2 Cyp26b1-/-/Rarg-/- embryos were examined for each gene and stage (except for Hoxa3 and Hoxb3 expression where only 1 Cyp26b1-/-/Rarg-/- embryo was analyzed for each gene). The variation observed within the genetic groups was none to minimal, and results depict representative staining.

Plasmids for making antisense riboprobes were kindly provided by Drs. P. Dollé (Meis2,

Hoxa3, Hoxa13, Hoxb3, Hoxd13, Rarg), Y. Chen (Shox2), S. Potter (Hoxa11), G. Martin

(Fgf8), A. McMahon (Shh), or were made in-house as described previously (Cyp26b1)

(MacLean et al., 2001).

3.5 Analysis of apoptosis in whole limb buds Embryos were collected into warm PBS and stained with 5µM LysoTracker Red

(Invitrogen Canada Inc, Burlington, ON) in PBS at 37°C for 30 mins. Following staining, embryos were washed once with PBS and fixed in 4% paraformaldehyde for 6-18h. They were then rinsed three times with PBS, dehydrated in 100% methanol, and stored at minus 20°C until imaged. For image acquisition the embryos were re-hydrated using a methanol series (75%, 50%, 25%) and counterstained with DAPI in PBS. Then whole limbs were removed and placed onto glass slides, and overlayed with fluorescence mounting medium (Dako Canada Inc., Mississauga, ON). The entire dorso-ventral width of the limb buds were scanned for areas of apoptosis, equivalent sections were found by locating the AER, and images were captured using a Leica TCS SP2 multiphoton

46 confocal microscope. At least 8 limb buds were examined for each genotype and developmental stage, and representative images are shown.

3.6 In situ hybridization on eyelid sections E15.0 embryos were collected and fixed overnight in 4% paraformaldehyde

(PFA). Following two brief washes with 1xPBS, the embryos were cryoprotected by subjecting them to a sucrose series of 5% sucrose in PBS for 30mins, 10% sucrose for 2h, and 20% sucrose for 6h to overnight. Embryos were then embedded in OCT embedding medium (Anatomical Pathology USA, Pittsburgh, PA), frozen, and cut into 14µm sections using a Reichert-Jung Cryocut 1800 cryostat (Leica Microsystems [Canada]

Inc.). Slides were stored at -20°C until required.

Hybridization and detection of antisense DIG-labelled riboprobes was performed according to standard procedure (Chotteau-Lelievre et al., 2006). As above for WMISH, riboprobes were synthesized using the appropriate RNA polymerase and DIG-labeling mix, and detection of AP was performed using BM Purple (all from Roche Diagnostics).

Following staining, slides were cleared twice for 5min in CitriSolv clearing agent (Fisher

Scientific Canada, Ottawa, ON, Canada) and coverslips were mounted with Permount

(Fisher Scientific). Images were captured using QCapture software at 10x and 20x magnification with an Olympus BX51 microscope. Plasmids for making antisense riboprobes were gifted by Drs. H. Ohuchi (Actβb and Tgfα), S. Bellusci (Fgf10), J.

Rossant (Fgfr2), A. McMahon (Shh), and J.C. Izpisúa Belmonte (Mkp3).

3.7 Immunohistochemistry on eyelid sections Eyelid sections were prepared as above (section 3.6). Sections were brought to room temperature for 30 mins and then washed three times with 1xPBS. Slides were blocked in PBS/0.2% Triton X-100 (Fisher Scientific) containing 10% normal goat serum

(NGS) (Vector Laboratories (Canada), Inc., Burlington, ON, Canada) for 1 hr at room temperature, followed by incubation with the appropriate primary antibodies overnight in

PBS/0.2% Triton with 1% goat serum at 4°C in a humidified chamber. After three washes in PBS/0.1% Triton, slides were incubated with goat anti-rabbit IgG Alexa 488 or conjugated secondary antibody (1:500 dilution in PBS/0.2% Triton with 1% NGS;

Molecular Probes, Invitrogen Canada Inc.) in a humidified chamber for 1–2 hr at room temperature. Following three washes in PBS/0.1% Triton, coverslips were mounted with fluorescence microscopy mounting medium (Dako Canada, Inc.) and slides were imaged using an Olympus BX51 fluorescence microscope fitted with filters for TRITC and

DSRed. Images were captured using QCapture Pro software under 10x and 20x objectives. Primary antibodies used were anti-FGF10 rabbit polyclonal (1:50 dilution; sc-

7917; Santa Cruz), anti-Ki-67 rabbit polyclonal (1:100 dilution; Ab15580; Abcam Inc.,

Cambridge, MA, USA), and anti-cleaved Caspase-3 (Asp175) rabbit polyclonal (1:100 dilution; #9661S; Cell Signalling Technology, Inc., Danvers, MA, USA). At least 2 embryos of each genotype were used per marker. Specific staining was confirmed by processing control slides and omitting the primary antibodies.

Chapter 4: Results

Cyp26b1-/-/ Rarg-/- phenotype: gross morphology and skeletal defects

4.1 Ablating Rarg does not rescue Cyp26b1 mutants from lethality

An analysis of the expected Mendelian and observed ratios for all possible genotypes arising from Cyp26b1+/-/Rarg+/- (referred to as B1+/-/γ+/-) crosses was performed at E10.5-E15.0, E18.5-postnatal day (P)1, and on P10 (Table 2). A total of

673 animals were genotyped: 573 before E15.0, 61 at E18.5 to P1, and 39 at P10. All genotypes were recovered near the expected ratio at fetal stages, with the exception of

B1+/-/γ-/- which was collected at around 60% of the expected frequency regardless of stage. This suggests that the B1+/-/γ-/- genotype may be selected against and resorbed during early embryonic stages (before E10.5). Notably, no pups were recovered at P10 with null genotypes for either Rarg or Cyp26b1 or both. This suggests that the loss of

Rarg is not sufficient to rescue Cyp26b1-null embryos from early post-natal death. It cannot be excluded that a rescue may occur at a lower frequency than it is possible to note from the number of animals recovered from these breedings, however pups with knockout genotypes were almost always observed to be cannibalized by the dams shortly after birth (discussed further below). The increase in observed ratios for the remaining genotypes at P10 reflect those that are selected for at birth.

Table 2: Resulting genotypic frequencies from B1+/-/γ+/- crosses.

% (Observed) Genotype Expected (%) E10.5, 11.5, E18.5 to P1 P10 (39) Total (673) 12.5 and 15.0 (61) (573)*

Wild type 6.25 7.2 (41) 13.1 (8) 15.4 (6) 8.6 (58)

B1-/- 6.25 6.6 (38) 6.6 (4) 0 6.2 (42)

γ-/- 6.25 7.5 (43) 8.2 (5) 0 7.1 (48)

B1-/-/γ-/- 6.25 6.6 (38) 4.9 (3) 0 6.1 (41)

B1-/-/γ+/- 12.5 13.8 (79) 11.5 (7) 0 12.8 (86)

B1+/-/γ-/- 12.5 7.7 (44) 6.6 (4) 0 7.1 (48)

B1+/- 12.5 14.3 (82) 15.6 (10) 25.6 (10) 15.2 (102)

γ+/- 12.5 12.4 (71) 8.2 (5) 20.5 (8) 12.5 (84)

B1+/-/γ+/- 25 23.9 (137) 24.5 (15) 38.5 (15) 24.4 (164)

*(Total number of embryos recovered)

Breeding was also performed between B1+/-/γ+/- males and B1+/+/γ+/- females to obtain double heterozygotes. The resulting frequencies observed at P10 (Table 3) illustrate that, as in the B1+/-/γ+/- intercrosses, the surviving pups are those which are not homozygous null for Rarg. It has been previously reported that around 50% of γ-/- mice can survive to weaning (Lohnes et al., 1993), but in the current analysis only three severely growth deficient γ-/- pups were recovered at P10 out of 629 animals. The γ-/- pups that were born (presumably at the expected frequency) were easily identifiable in the days following birth by their sallow skin tone, but were almost always cannibalized within 3 days (T.P. unpublished observations).

4.2 Rarg expression overlaps with Cyp26a1 and Cyp26b1 in mid-gestation embryos

As discussed in Chapter 2, the expression patterns for the Cyp26 family of enzymes are generally spatially distinct throughout embryonic development and are thus responsible for RA metabolism in separate tissues. Whereas Cyp26a1 is strongly expressed in the tailbud and cervical mesenchyme during mid-gestation, Cyp26b1 is expressed in the frontonasal process, hindbrain, and developing limb bud throughout P-D outgrowth (MacLean et al., 2001; Abu-Abed et al., 2002). The expression patterns of these enzymes define tissues in which RA is being degraded in order to keep them RA- free and limit signalling. Notably, although the RARs (particularly RARα) are widely expressed during embryogenesis, work by our lab and others has illustrated that RARγ is

Table 3: Viability of pups from B1+/-/γ+/- male and B1+/+/γ+/- female crosses.

Expected (%) % (Observed)

Genotype P10 (629)*

Wild type 12.5 14.8 (93)

γ-/- 12.5 0.2 (1)

B1+/-/γ-/- 12.5 0.3 (2)

+/- B1 12.5 18.6 (117)

γ+/- 25 33.9 (213)

B1+/-/γ+/- 25 32.3 (203)

*(Total number of pups)

52 the isotype responsible for mediating the RA teratogenic response in tissues which are sensitive to excess RA (Iulianella and Lohnes, 1997; Abu-Abed et al., 2003).

The domains of expression of Cyp26b1 and Rarg were examined by whole-mount in situ hybridization (WMISH) in wild type embryos during mid-gestation (E9.5-E12.5)

(Figure 4). The expression pattern of Rarg between E9.5-E12.5 shows significant overlap with that of Cyp26a1 and Cyp26b1 during these stages, although receptor expression tends to extend more broadly than the domains of the metabolic enzymes.

Similarly to previously published reports, at E9.5 Cyp26b1 was detected in the hindbrain

(particularly r5 and r6) and the emerging forelimb field (Figure 4A) (MacLean et al.,

2001) whereas Rarg was strongly expressed in the forelimb field as well as in the tailbud, frontonasal process, and rostral branchial arches (Figure 4E) (Ruberte et al., 1990;

Ruberte et al., 1991) and this pattern persisted until E11.5 (Figure 4F-G). Between E10.5 and E11.5, Cyp26b1 was strongly expressed in the limb buds, although it appeared more distally restricted in comparison to the Rarg expression domain, and by E11.5 became detectable in the branchial arches and frontonasal process (Figure 4B-C). At E12.5, both

Cyp26b1 and Rarg were found in the limb, the mesenchyme of the maxillary and mandibular processes of the developing face, whisker follicles, and developing pinnae

(Figure 4D,H).

Figure 4: Expression of Cyp26b1 and Rarg in developing embryos.

Whole-mount in situ hybridization of Cyp26b1 (A-D) and Rarg (E-H) in E9.5-E12.5 wild type embryos illustrates the overlapping expression for these two genes. Both are expressed within the forelimb (fl) and hindlimb (hl) buds from E9.5-E12.5, the branchial arches (ba) and frontonasal process (fnp) at E11.5, and the external pinnae (pi) at E12.5. Cyp26b1 is also strongly expressed in the hindbrain rhombomeres (r) from E9.5-E11.5. Note that Rarg is also strongly expressed in the tailbud (tb) and craniofacial regions from E9.5-E12.5, which overlaps with expression of Cyp26a1 at these stages (MacLean et al., 2001). (flf, forelimb field; ht, heart)

4.3 Analysis of gross morphology of B1-/-/γ-/- fetuses

The phenotypic features of pups from B1+/-/γ+/- crosses at E18.5 and P1 were examined. The phenotype of B1-/- pups was similar to that previously published (Yashiro et al., 2004), exhibiting micrognathia, shortened adactylous limbs, and open-eyes-at-birth

(Figure 5B) (see also Maclean et al., 2009). These mice die soon after birth from apparent respiratory distress (Yashiro et al., 2004; Maclean et al., 2009) and are selectively cannibalized by their mothers. It was observed that Cyp26b1-null mice also have a paucity of vibrissal follicles from which no whiskers emerge (Figure 5B, middle panel, black arrow; compare with 5A).

Three B1-/-/γ-/- embryos were examined at E18.5 and, although still abnormally formed, they were found to have more complete limb development, closed eyes, and rescued vibrissae (Figure 5D). Loss of one allele of Rarg in the B1-/- background was unable to rescue eyelid closure or vibrissae; these animals resembled B1-null fetuses

(Figure 5C).

4.4 Skeletal malformations in B1-null and B1-/-/γ-/- embryos

A closer analysis of skeletal development was performed by staining with Alcian

Blue for cartilage and Alizarin Red for bone as previously described (Abu-Abed et al.,

2003). The skeletal malformations exhibited by E18.5 Rarg-/- fetuses (Lohnes et al.,

1993; Iulianella and Lohnes, 1997), along with the limb and craniofacial anomalies in

Figure 5: Neonatal morphology of Cyp26b1-null and Cyp26b1/Rarg compound-null mice.

Gross morphology of (A) wild type, (B) Cyp26b1-/-, (C) Cyp26b1-/-/Rarg+/-, and (D) Cyp26b1-/-/Rarg-/- fetuses at E18.5. The characteristic phenotype following loss of CYP26B1 includes phocomelia and adactyly (white arrows and close-ups of fore and hindlimbs in panels on the right), micrognathia (red arrowhead), open eyes (black arrowhead), and lack of vibrissae (black arrow) (compare B with A). Loss of one allele of Rarg in a Cyp26b1-null background results in the same gross phenotypic malformations as observed in Cyp26b1-null mice (compare C with B). Animals which are null homozygotes for both Cyp26b1 and Rarg have partial restoration of limb outgrowth, closed eyelids, and growth of whiskers (compare D with B and A). FL, forelimb; HL, hindlimb.

58 Cyp26b1-null fetuses (Yashiro et al., 2004; Maclean et al., 2009) have been previously characterized. An examination of 31 fetuses at E18.5, including 3 compound-null animals, revealed that B1-/-/γ-/- compound-null limbs showed rescued development of the stylopod element (humerus or femur) and some digits (autopod) in both fore and hindlimbs, along with the clavicle in forelimbs and pelvic bone in hindlimbs (Figure 6, compare B and F with D and H), whereas the zeugopod element (radius/ulna or tibia/fibula) was not rescued by ablation of Rarg. We have recently reported the details of this limb phenotype (see Pennimpede et al., 2010) which will be discussed in more detail in Chapter 5.

A lateral view of the skulls of B1-/- and B1-/-/γ-/- fetuses as compared to wild type

(Figure 7) shows the characteristic shortened mandible and maxillary process, along with reduced calvarial ossification in B1-/- animals (Figure 7B) (Maclean et al., 2009), which is partially rescued in B1-/-/γ-/- fetuses (Figure 7D, compare bracketed regions). In addition, the ectotypanic ring fails to form in B1-null fetuses (Maclean et al., 2009 and

Figure 7B), but is rescued following a loss of Rarg (Figure 7D, arrow). B1-/-/γ+/- animals displayed an intermediate phenotype (Figure 7C).

Isolation of the trachea and larynx revealed abnormalities in formation of the hyoid bone in B1-/- fetuses which was characterized by loss of ossified hyoid and apparent fusion of the abnormal hyoid-like structure to the thyroid cartilage (Maclean et al., 2009 and Figure 7F and G). The ossified hyoid structure was rescued in compound- null fetuses, and these animals also exhibited a lateral fusion of the tracheal rings (Figure

7H, asterisk) which is typical following loss of Rarg (Lohnes et al., 1993).

Figure 6: Comparative skeletal development of forelimbs and hindlimbs in Cyp26b1-/-, Cyp26b1-/-/Rarg+/-, Cyp26b1-/-/Rarg-/-, and phenotypically normal fetuses at E18.5.

Comparison of forelimbs (A-D) and hindlimbs (E-H) from E18.5 prenatal mice, genotypes as indicated. Cartilagenous elements are stained with Alcian Blue and bone with Alizarin Red. In Cyp26b1/Rarg compound-null mice both fore (D) and hindlimbs (H) exhibit full rescue of stylopod and partial rescue of autopod formation, but lack rescue of zeugopod formation. In addition, compound-null mice also show rescue of clavicle development and pelvic bone (arrows) whereas Cyp26b1-null mice (B, F), even when heterozygous for Rarg (C, G), have either absence or abnormal development of these structures (arrowheads). (A, autopod; Cl, clavicle; DT, deltoid tuberosity; Fe, femur; Fi, fibula; Hu, humerus; Rd, radius; S, stylopod; Sc, scapula; Ti, tibia; Ul, ulna; Z, zeugopod; I-V, digit number (anterior to posterior)).

Figure 7: Ossification of skull structures and the hyoid bone are rescued in Cyp26b1/Rarg compound-null mice.

(A-D) Lateral view of skull development in wild type, Cyp26b1-/-, Cyp26b1-/-/Rarg+/-, and Cyp26b1-/-/Rarg-/- E18.5 embryos. Note reduced ossification of maxilla (Mx), mandible (Md), and frontal (FP), parietal (PP), and interparietal (IP) calvarial plates, along with loss of the ectotympanic ring (Et) seen in B1-/- skulls (B) (and see also (Maclean et al., 2009)) are partially rescued in compound-null skulls (D). B1-/-/γ+/- skulls (C) have an intermediate phenotype. Brackets illustrate micrognathia, which is present in B1-null, B1- /-/γ+/-, and B1/γ compound-null animals. (E-H) Larynx and tracheal cartilage are well- formed in B1-null embryos, but the hyoid bone (black arrowhead in A-D and E&H) is unossified (F, compare with E), but is rescued in compound-null embryos (H). The asterisk denotes lateral fusion of the tracheal rings which is characteristic of and completely penetrant in Rarg-null mice. (Hy, hyoid; Pmx, premaxilla).

An examination of the B1-/- cranial base (Figure 8B) illustrates fusion of the basioccipital and exoccipital bones (red arrowhead) along with the malformed ptgerygoid process of the basisphenoid, and cleft palate (asterisk) (see also Maclean et al., 2009). All of these defects, with the exception of cleft palate are rescued in compound-null skulls

(Figure 8D and data not shown). An anterior view of the B1-/- cranial vault (Figure 8F) demonstrates the lack of ossification of the frontal, parietal, and interparietal plates, and lack of the zygomatic bone (Maclean et al., 2009). In addition, it was noted that the nasal cartilage was malformed in Cyp26b1-null animals (Figure 8F-G, nc*). The calvaria was found to be more completely developed in the B1-/-/γ-/- fetuses (Figure 8H), and the nasal cartilage more closely resembled the wild type case, although the zygomatic process was still absent (Figure 8H, black arrowhead).

Complete vertebral skeletal phenotypes were scored under a dissecting microscope for a minimum of 3 fetuses per genotype. Table 4 summarizes the main phenotypes observed for wild type, B1-/-, γ-/-, B1+/-/γ-/-, B1-/-/γ+/-, and B1-/-/γ-/- fetuses.

Since B1+/-, γ+/-, and B1+/-/γ+/- phenotypes did not differ substantially from wild type they will not be discussed further in this context. Normal wild type mice in this genetic background have a vertebral pattern of seven cervical vertebrae (C1-C7), thirteen thoracic vertebrae (T1-T13) seven of which are fused to the sternum, six lumbar vertebrae (L1-

L6), followed by sacral and caudal vertebrae (not considered here). This vertebral pattern was mainly conserved in all embryos examined (Table 4). Prominent vertebral phenotypes observed in B1-/- fetuses included fusions of the neural arches of the

Figure 8: Various abnormalities of skull formation observed in Cyp26b1-null animals are rescued by ablation of Rarg.

Comparative development of crania from wild type, Cyp26b1-/-, Cyp26b1-/-/Rarg+/-, and Cyp26b1-/-/Rarg-/- embryos at E18.5 viewed from the base of the skull (posterior) (A-D) and the skull vault (anterior) (E-H). (B) Fusion of the basioccipital (Bo) and exoccipital (Eo) bones observed following loss of CYP26B1 (red arrowhead) is rescued following removal of Rarg (D). Compound-null animals also show rescue of tympanic ring (Et) formation (red arrow) but not cleft palate (asterisk). (E-H) In B1/γ knockout mice, there is a near complete rescue of the frontal (FP), parietal (PP), and interparietal (IP) skull plates, but not the orbital bone (zygomatic process of maxilla, Zmx; black arrowhead). Nasal cartilage (nc) is also malformed in B1-null animals (F and G) and is more normally developed in the B1/γ-null fetuses (H). Also see Maclean et al., 2009 for description of B1-null cranial phenotype. (Md, mandible; Pt, palatine; Ptg, pterygoid).

Table 4: Summary of axial skeletal abnormalities observed at E18.5.

wt B1-/- γ-/-† B1+/-/γ-/- B1-/-/γ+/- B1-/-/γ-/- (n=7) (n=8) (n=3) (n=3) (n=7) (n=3) Cervical vertebral malformations Basi/exoccipital fusion 0/7 4/8 0/3 0/3 2/7 0/3 Incomplete 4/8 5/7 C1-C2 fusion 0/7 6/8 a 1/3 1/3 d 5/7 b 2/3 C2 to C1 homeotic transformation 0/7 3/8 1/3 2/3 1/7 0/3 (ectopic AAA* or thick neural arch) C6 to C5 homeotic transformation 0/7 0/8 1/3 1/3 0/7 0/3 Missing tuberculum anterior 0/7 6/8 b 0/3 0/3 1/7 0/3 Tracheal ring lateral fusion 0/7 0/8 3/3 3/3 0/7 3/3 Missing hyoid bone 0/7 8/8 0/3 0/3 7/7 0/3 Rib malformations 0/7 T1/T2 fusion 1/3 2/3 1/3e T4/T5 fusion 1/3e T5/T6 fusion 1/3 T7/T8 fusion 1/7 Misshapen „wavy‟ ribs 8/8 7/7 Number of sternebrae** Four 5/8 4/7 1/3 Five 7/7 3/8 c 3/3 3/3 3/7 2/3 Vertebral pattern C7/T12/L5 1/3 C7/T12/L6 1/8 C7/T13/L5 1/7 1/8 1/7 C7/T13/L6 6/7 6/8 3/3 2/3 5/7 2/3 C7/T14/L5 1/7 1/3

† For comprehensive description of axial defects found in Rarg-/- mice see Lohnes et al., 1993 * Anterior arch of atlas ** not including the xyphoid process a bilateral b unilateral or bilateral c including ossified manubrium d complete dorsal fusion e same fetus

67 atlas and axis (C1 and C2) in almost all cases (6/8) (Figure 9A, red arrows; and Maclean et al., 2009), sometimes accompanied by an apparent C2 to C1 homeotic transformation

(3/8) (see Figure 9A). Loss of Cyp26b1 also resulted in the absence of the tuberculi anterior (6/8), which are normally present as ventral projections from C6 (Figure 9A, asterisk). Fusions of C1 and C2 persisted in compound-null fetuses (2/3), while formation of the tuberculi anterior was rescued (3/3).

An examination of the thoracic cavity of Cyp26b1-null fetuses revealed that they have a shortened sternum, which often had only four sternebral centers of ossification

(5/8), or occasionally an ossified manubrium (2/8), suggesting possible posteriorization of the anteriormost sternebra (see Figure 9B). This phenotype could be rescued by loss of Rarg (2/3).

4.4.1 Ablation of Cyp26b1 does not alter skeletal defects resulting from loss of Rarg

The main phenotypes observed in γ-/- mice included fusions of C1 and C2 (1/3), rib fusions (T1/T2) (1/3) (see Figure 9), and lateral fusions of the tracheal cartilage (3/3)

(see Figure 7), phenotypes that agree well with previously published results following loss of RARγ (Lohnes et al., 1993; Iulianella and Lohnes, 1997; Abu-Abed et al., 2003) and which persisted in B1-/-/γ-/- mice.

Figure 9: Vertebrothoracic abnormalities in Cyp26b1-null animals are rescued following ablation of Rarg.

Comparative development of cervical vertebrae (A) and sterna (B) of wild type, Cyp26b1-/-, Cyp26b1-/-/Rarg+/-, and Cyp26b1-/-/Rarg-/- prenatal mice at E18.5. (A) Cyp26b1-null mice exhibit highly penetrant lateral fusion of the neural arches of C1 and C2 vertebrae (red arrow) which often persists in the compound-null mice. The neural arch of the axis (C2) is often thicker than normal, suggesting possible transformation to atlas (C1) identity (C1*). The tuberculi anterior (ta), which are normally present as anterior projections from C6, are aplastic in Cyp26b1-/- and some Cyp26b1-/-/Rarg+/- animals (asterisk). (B) Cyp26b1 null mice exhibit malformations of the sternum which include a reduction in number or re-specification of sternal nodes of ossification, where these are labeled 1 to 5 in wild type („1‟, „2‟, „3‟, and „4‟ refer to respecified nodes which have shifted anteriorly). B1/γ knockout animals show a rescue of sternal development, and Cyp26b1-/-/Rarg+/- animals exhibit an intermediate phenotype. The red arrowhead denotes T1/T2 rib fusions which are observed with moderate penetrance in Rarg-/- mice but not in Cyp26b1-null mice (see also Lohnes et al., 1993, and Table 4). The thoracic region from an Rarg-/- embryo is provided for comparison. (AAA, anterior arch of atlas; C, cervical vertebra; T, thoracic vertebra; ta, tuberculum anterior; xp, xiphoid process).

4.4.2 Removal of Rarg in the Cyp26b1-null background can rescue skeletal defects associated with RA treatment but which were not observed following loss of Cyp26a1

B1-null mice showed several phenotypes which were previously observed in wild type embryos that have been exposed to RA (Iulianella and Lohnes, 1997), but were not observed in A1-/- embryos (Abu-Abed et al., 2003), and were rescued in B1-/-/γ-/- mice.

These include fusion of the basioccipital and exoccipital bones of the skull (see Figure

8B) (Maclean et al., 2009) and missing tuberculi anterior on C6 (Figure 9A) (Iulianella and Lohnes, 1997). In addition these mice were observed to have dysplastic „wavy‟ ribs

(Figure 10B and C), which were rescued following loss of Rarg (Figure 10D). This rib phenotype can be observed following RA treatment of 100mg per kg maternal body weight at E11.5 in mice of the same genetic background used in this study (Iulianella and

Lohnes, 1997) but is not discussed further in the context of that manuscript. Loss of

Cyp26b1 also results in an undeveloped hyoid bone which is rescued in Cyp26b1/Rarg compound-null embryos (Figure 7) (Maclean et al., 2009), but which is not observed following maternal RA treatment. The only phenotype that was found to be shared between Cyp26a1-/- and Cyp26b1-/- mice (and was not rescued in B1-/-/γ-/- mice) was fusion of C1 and C2 (Figure 9A).

4.5 Expression of Hoxa3 and Hoxb3 is unchanged in Cyp26b1-null embryos at E10.5

Retinoids have been shown to play a key role in regulating vertebral patterning through effects on Hox gene expression, sometimes regulating Hox genes directly via

Figure 10: Rib phenotype of Cyp26b1-/- embryos.

(A-D) Misshapen „wavy‟ ribs are characteristic of Cyp26b1-/- mice (B). Typical areas of abnormal rib development are marked by red arrowheads. This phenotype was found to be rescued by complete ablation of Rarg (D), but not under conditions of Rarg heterozygosity (C). Asterisk in (D) denotes rescue of clavicle development in compound- null embryos.

RAREs within Hox gene promoters (for review see Alexander et al., 2009). Mutations in

Hox paralogous group 3 genes (Hoxa3, Hoxb3, and Hoxd3), either alone or in combination, illustrated that these animals do not have hindbrain patterning defects, but do exhibit vertebral anteriorizations, particularly a loss of the atlas, and hyoid defects following loss of multiple Hox3 genes (Manley and Capecchi, 1997). In order to determine whether the duplication of the atlas seen in B1-/- embryos corresponded to an anterior shift in expression of Hox3 genes, the expression patterns of Hoxa3 and Hoxb3 were examined in B1-/- embryos at E10.5 (Figure 11B and F). The levels and anterior border of expression for both Hoxa3 and Hoxb3 were found to resemble those of control embryos (Figure 11A and E, arrowheads). Dorsal views of the hindbrain region shows that the rostral boundaries of expression in the neural tube were also unchanged (Figures

11B and 11E, compare with 11A and 11F).

Normal expression at E10.5 for Hoxa3 in the paraxial mesoderm corresponds to the anterior border of presumptive C1 (encompassing up to somite 4), and to the r4/r5 border in the neural tube, whereas expression of Hoxb3 corresponds to the anterior boundary of C2 in the paraxial mesoderm and mid r5 in the neural tube (Figure 11A, E and Voss et al., 2009). These boundaries were found to be maintained in B1-/- embryos at

E10.5 (Figure 11B, F), and no differences were observed in either B1-/-/γ-/- (Figure 11C,

G) or γ-/- (Figure 11D, H) embryos at this stage.

Figure 11: Expression of Hoxa3 and Hoxb3 at E10.5.

Whole-mount in situ hybridization for detection of Hoxa3 (A-D) and Hoxb3 (E-H) expression in control, Cyp26b1-/-, Cyp26b1/Rarg-/-, and Rarg-/- embryos at E10.5. In all panels a lateral view of the embryo is shown on the left with a dorsal view on the right. Expression was similar for all embryos and genotypes examined. Black arrowheads mark the rostral-most point of expression into the presumptive hindbrain.

Chapter 5: Results

Analysis of limb proximal-distal outgrowth in Cyp26b1/Rarg compound-null embryos

5.1 Expression of Rarg and Cyp26b1 overlap during limb P-D outgrowth

Cyp26b1 is strongly expressed in the developing limb bud throughout proximodistal outgrowth, and is the main enzyme involved in metabolizing RA in this tissue. In the absence of CYP26 enzymes, RA is present and able to signal through RARs. Both

Rara and Rarg are expressed throughout the developing limb bud mesenchyme (Dolle et al., 1989), but RARγ is the main isotype responsible for mediating the teratogenic response in RA-sensitive tissues such as the craniofacial region and caudal embryo

(Iulianella and Lohnes, 1997; Abu-Abed et al., 2003). The domains of expression of

Cyp26b1 and Rarg were found to overlap in the developing limb throughout normal proximo-distal outgrowth between E9.5 and E12.5 (Figure 12A-H). The domain of Rarg expression is observed to be broader than that of Cyp26b1, which is progressively distalized during P-D outgrowth. Forelimb buds are shown, but the same is found for the hindlimb. Both genes are expressed in the limb bud mesenchyme during forelimb induction beginning around E9.0-9.5, with Cyp26b1 being expressed in the distal-most part of the emerging bud (Figure 12A and E). Between E10.5 and E11.5 Rarg is found throughout the limb mesenchyme (Ruberte et al., 1990, and Figure 12F and G), and

Figure 12: Expression of Cyp26b1 and Rarg during normal limb outgrowth.

Expression of Cyp26b1 (A-D) and Rarg (E-H) in E9.5-E12.5 wild type forelimb buds visualized by whole-mount in situ hybridization illustrates the overlapping expression for these two genes in the emerging limb. Dorsal views of limb buds are shown with anterior at the top.

Cyp26b1 is strongly expressed in the superficial mesenchyme and distal ectoderm but is absent from the AER (MacLean et al., 2001; Yashiro et al., 2004, Figure 12B and C). By

E12.5 Rarg is found in the mesenchyme and developing chondrogenic aspects of the limb bones (Ruberte et al., 1990, and Figure 12H), and Cyp26b1 is localized to the distal digit tips and can also be found in chondrogenic precursors (Yashiro et al., 2004, and Figure

12D).

5.2 Genetic removal of Rarg in a Cyp26b1-/- background partially rescues limb skeletal morphology

Loss of CYP26B1 causes severe truncations of both fore and hindlimbs resulting in greatly reduced stylopod and zeugopod structures, and loss of wrist elements and some digits (see Figure 6B, F; compare with Figure 6A, E). Removing RARγ in the Cyp26b1- null background was found to ameliorate these defects. Loss of one allele of Rarg partially restores digit formation and partially rescues formation of the pelvic bone

(Figure 6C and G) and compound Cyp26b1/Rarg knockout embryos were found to exhibit a complete restoration of the clavicle, pelvic girdle, and stylopod, and partial rescue of the autopod (in fore- and hindlimbs) but retain a greatly reduced zeugopod compared to wild type (Figure 6D and H). The phenotypes that were observed are summarized in Table 5.

Table 5: Summary of limb skeletal abnormalities observed at E18.5a

B1-/- γ-/- B1-/-/γ+/- B1-/-/γ-/- (n=4, (n=4) (n=8) (n=3) embryos) Forelimbs: Clavicle formed 0/4 (0%) 4/4 (100%) 0/8 (0%) 3/3b (100%) Humerus formed 0/4 (0%) 4/4 (100%) 0/8 (0%) 3/3 (100%) Radius/ulna formed 0/4 (0%) 4/4 (100%) 0/8 (0%) Partial 3/3 (100%) Digit number: One 1/4 (25%) Two 3/4 (75%) 4/8 (50%) Three 4/8 (50%) Four 3/3 (100%) Five 4/4 (100%) Hindlimbs: Pelvic bone formed 4/4 (100%) 3/3e (100%) Partial 1/4d (25%) 8/8c (100%) Femur formed 0/4 (0%) 4/4 (100%) 0/8 (0%) 3/3 (100%) Tibia/fibula formed 0/4 (0%) 4/4 (100%) 0/8 (0%) 0/3 (100%) Digit number: One 1/4f (25%) Two 3/4 (75%) 4/8 (50%) Three 3/8 (37.5%) Four 1/8 (12.5%) 3/3 (100%) Five 4/4 (100%) a all phenotypes were observed to be bilateral b smaller than normal c ilium and pubis nearly normal but no ischium d partial formation of pubis e formed normally with reduction in length of ilium f two digits but fused at terminal phalanx

5.3 Markers of the AER and ZPA are normally expressed in Cyp26b1-null embryos

Limb development is directed by coordinated signals from the AER, ZPA, and dorsal limb surface (Niswander, 2003). The main growth factors secreted from the AER are FGFs 8, 4, 9 and 17 (Mariani et al., 2008), and from the ZPA is Shh (Riddle et al.,

1993). The expression of both Fgf8 and Shh are maintained in B1-/- limb buds at E10.5 and E11.5 (Yashiro et al., 2004, and Figure 13A-F). FGF and Shh signalling were also found to be unaffected in E10.5 B1-/- limb buds as measured by pERK and Mkp3 expression (for FGF) and Ptc1 expression (for Shh) (Yashiro et al., 2004).

The AER is a strip of stratified epithelium at the distal edge of the developing limb bud (Crossley and Martin, 1995) which secretes signals that are important for limb outgrowth (Saunders, 1948; Summerbell, 1974; Sun et al., 2002). Between E12.5 and

E13.5 in the mouse, reduced mesenchymal proliferation along with negative regulation of

Fgf8 expression by BMP signalling, results in AER regression by apoptosis (Guha et al.,

2002). Analysis of the morphology of the AER in B1-/- forelimb buds as compared to wild type limb buds illustrated that the B1-/- AER was well-defined and differentiated, although thicker than the wild type AER (Figure 13G). Compaction of the AER cells into a columnar epithelium has been shown to be important for maintaining a flattened limb bud shape (Niswander, 2003). Compressed ectodermal cells were clearly visible at the distal-most edge of the ridge, and cell death within the AER as measured by

LysoTracker Red, which is a vital lysosomal fluorescent dye that marks apoptotic cells

(Zucker et al., 1999), was comparable to wild type levels (Figure 13G).

Figure 13: The AER and ZPA are properly specified in Cyp26b1-/- limbs.

Expression of Fgf8 at E10.5 (A, B) and Shh at E10.5 (C, D) and E11.5 (E, F) were examined by whole-mount in situ hybridization in control and Cyp26b1-/- limb buds (genotypes as indicated). Fgf8 expression marks the AER and Shh marks the ZPA. Dorsal views of hindlimb buds are shown with anterior at the top. (G) Confocal microscopic analysis of AER morphology at E11.5 in B1-/- and control forelimb buds shows the layered condensed epithelia which is characteristic of this structure, and the loosely- packed underlying mesenchyme. Apoptotic cells are marked by red staining (LysoTracker Red) and nuclei are shown in blue (DAPI). (A, anterior; P, posterior).

5.4 Aberrant expression of proximal-distal patterning genes remains unchanged between Cyp26b1-null and Cyp26b1/Rarg compound-null embryos

In order to examine whether the skeletal rescue phenotype observed in B1-/-/γ-/- limbs was the result of re-specification of cell fates within the developing limb bud, various markers of P-D patterning were examined during limb outgrowth. The “early specification” model of limb development argues that cells within the early limb mesenchyme are already fated to become the stylopod, zeugopod, and autopod. These limb segments can be visualized by examining the expression of genes thought to be involved in P-D specification; which include Meis1/Meis2 and Shox2 (stylopod), Hoxa11

(zeugopod), and Hoxa13 and Hoxd13 (autopod). The expression of these genes was examined in the fore and hindlimbs of control, B1-/-, B1-/-/γ+/-, B1-/-/γ-/- and γ-/- embryos during early limb development (E10.5-E11.5).

Meis2, widely accepted as a marker of the presumptive stylopod region, is an RA- responsive gene which has been found to be repressed by FGF signalling in vertebrates pharmacologically (Mercader et al., 2000) and physiologically (Mariani et al., 2008).

Meis2 encodes a homeobox transcription factor which is thought to be required for determining proximal cell fates, in that overexpression can cause proximalization of distal cells/elements (Mercader et al., 2000; Mercader et al., 2009). Meis2 expression is normally restricted to the presumptive stylopod region (Figure14A, B, K, L), but in B1-/-,

B1-/-/γ+/-, and B1-/-/γ-/- limbs at E10.5, it was found to be expressed in a distally expanded domain encompassing the zeugopod/autopod (Yashiro et al., 2004, and Figure 14C-H).

This expanded region of Meis expression was maintained into the zeugopod element in 85

Figure 14: Expression of Meis2 in forelimb and hindlimb buds at E10.5 and E11.5.

Expression of the presumptive stylopod marker Meis2 was examined by whole-mount in situ hybridization at E10.5 (A-J) and E11.5 (K-T) in control (A, B, K, L), Cyp26b1-/- (C, D, M, N), Cyp26b1-/-/Rarg+/- (E, F, O, P), Cyp26b1-/-/Rarg-/- (G, H, Q, R), and Rarg-/- (I, J, S, T) forelimb and hindlimb buds. Dorsal views of limb buds are shown with anterior at the top. Meis2 is normally expressed in the flank and proximal limb bud (A-B, I-J and K-L, S-T). At both stages of development shown, the loss of CYP26B1 results in expansion of Meis2 expression into more distal aspects of the limb bud (C-F, M-P) and expression is not returned to normal in B1/γ compound-null limb buds (G-H, Q-R).

B1-/-, B1-/-/γ+/-, and B1-/-/γ-/- limbs at E11.5 (Figure 14M-R) and this can be attributed to the concomitant expansion of RA into more distal aspects of the limb bud due to the absence of CYP26B1 in all cases. Although upregulation of Meis1and Meis2 by RA has been shown to be effected by RARγ in vitro (Galdones et al., 2006), loss of Rarg did not affect expression of Meis2 in vivo (Figure 14 I-J, and S-T).

The complete rescue of stylopod formation in B1-/-/γ-/- embryos also prompted us to examine the expression of Shox2, a transcription factor that was recently characterized as a specific determinant of stylopod formation (Cobb et al., 2006; Yu et al., 2007).

Shox2 was found to be similarly expressed in both fore and hindlimbs for all genotypes examined at E10.5 and E11.5 (Figure 15).

Expression of the zeugopod marker Hoxa11 was examined and found to be comparable in all limb buds at E10.5 and E11.5 (Figure 16). Expression of this gene has been previously found to be distally shifted in E11.5 B1-/- limbs (Yashiro et al., 2004), and this is also noted here, particularly anteriorly, in all limb buds lacking Cyp26b1

(compare bracketed areas in Figure 16K-T).

Both HOXA13 and HOXD13 are required for proper formation of the autopod, although they are able to at least partially compensate for one another (Fromental-Ramain et al., 1996). Expression of Hoxd13 in all limb buds was comparable at E10.5 (Figure

17A- J) and this expression was maintained but shifted distally in B1-/-, B1-/-/γ+/-, and B1-/-

/γ-/- limb buds at E11.5 (Figure 17K-T). In contrast, Hoxa13 was downregulated in B1-/-,

B1-/-/γ+/-, B1-/-/γ-/-, and γ-/- limbs at E10.5 (Figure 18A-J) and E11.5 (Figure 18K-T), and

Figure 15: Expression of Shox2 in forelimb and hindlimb buds at E10.5 and E11.5.

Expression of the stylopod marker Shox2 was examined by whole-mount in situ hybridization at E10.5 (A-J) and E11.5 (K-T) in control (A, B, K, L), Cyp26b1-/- (C, D, M, N), Cyp26b1-/-/Rarg+/- (E, F, O, P), Cyp26b1-/-/Rarg-/- (G, H, Q, R), and Rarg-/- (I, J, S, T) forelimb and hindlimb buds. Dorsal views of limb buds are shown with anterior at the top. Normal Shox2 expression is specific to the presumptive stylopod and zeugopod regions of the limb bud (A-B and K-L) and this domain of expression was found to be retained at both stages of development for all embryos/genotypes examined.

Figure 16: Expression of Hoxa11 in forelimb and hindlimb buds at E10.5 and E11.5.

Expression of the zeugopod marker Hoxa11 was examined by whole-mount in situ hybridization at E10.5 (A-J) and E11.5 (K-T) in control (A, B, K, L), Cyp26b1-/- (C, D, M, N), Cyp26b1-/-/Rarg+/- (E, F, O, P), Cyp26b1-/-/Rarg-/- (G, H, Q, R), and Rarg-/- (I, J, S, T) forelimb and hindlimb buds. Dorsal views of limb buds are shown with anterior at the top. Hoxa11 is generally normally detectable as a chevron-shaped stripe of expression in the middle of the limb bud, representing the presumptive zeugopod (A-B and K-L). This expression was maintained in all limb buds examined, but appeared distally shifted towards the autopod region in the forelimbs of embryos lacking CYP26B1 at E10.5 (C, E, G), and this effect was more pronounced in all B1-null limb buds at E11.5 (M-R; compare bracketed areas with K-L and S-T). Arrowheads in K-T mark the cleft of the hand and footplate.

Figure 17: Expression of Hoxd13 in forelimb and hindlimb buds at E10.5 and E11.5.

Expression of the autopod-specific marker Hoxd13 was examined by whole-mount in situ hybridization at E10.5 (A-J) and E11.5 (K-T) in control (A, B, K, L), Cyp26b1-/- (C, D, M, N), Cyp26b1-/-/Rarg+/- (E, F, O, P), Cyp26b1-/-/Rarg-/- (G, H, Q, R), and Rarg-/- (I, J, S, T) forelimb and hindlimb buds. Dorsal views of limb buds are shown with anterior at the top. Normal expression of Hoxd13 at E10.5 is seen as a crescent shaped band in the posterior margin which includes the ZPA, and was normally expressed in all embryos at this stage (A-J). By E11.5, Hoxd13 expression has expanded along the distal aspect of the limb bud (K-L and S-T). Although expression was retained in the limb buds of embryos lacking CYP26B1 function at E11.5, the domain of expression was found to be further distally shifted (M-R).

Figure 18: Expression of Hoxa13 in forelimb and hindlimb buds at E10.5 and E11.5.

Expression of Hoxa13, a marker of the autopod inclusive of wrist elements, was examined by whole-mount in situ hybridization at E10.5 (A-J) and E11.5 (K-T) in control (A, B, K, L), Cyp26b1-/- (C, D, M, N), Cyp26b1-/-/Rarg+/- (E, F, O, P), Cyp26b1-/-/Rarg-/- (G, H, Q, R), and Rarg-/- (I, J, S, T) forelimb and hindlimb buds. Dorsal views of limb buds are shown with anterior at the top. Hoxa13 expression is normally strongly detected in the autopod region at E10.5 (A, B) and E11.5 (K, L). Both the domain and intensity of expression was reduced following loss of Cyp26b1 at both stages (C-F, and M-P). This reduced expression was maintained in compound-null limb buds (G-H, Q-R), where a further reduction of intensity can be attributed to the loss of RARγ function, which appears to be important for maintaining normal Hoxa13 expression levels (I-J and S-T). Note that the domain of expression in Rarg-/- limb buds is similar to that of the control limb buds (A-B and K-L), only severely reduced in intensity.

96 was distally shifted in B1-/-, B1-/-/γ+/-, and B1-/-/γ-/- limb buds. The observed further reduction in Hoxa13 expression in B1-/-/γ-/- limb buds, especially prominent at E11.5

(Figure 18G-H and Q-R), can be explained by the loss of RARγ, since expression of

Hoxa13 was also found to be drastically downregulated in γ-/- limbs at both stages

(Figure 18I-J and S-T).

5.5 Localized mesenchymal cell death is rescued in Cyp26b1/Rarg-null limbs

Notably, B1-/- limb buds are observed to be amorphous by E11.5, although they are similar in size and shape to control limb buds at E10.5. At E11.5, Cyp26b1-null limb buds appear pointed and lack the classic paddle shape, which is found to be partially restored in B1-/-/γ-/- compound-null limb buds (for an example see Figure 14K-T). An examination of whole limb buds stained with LysoTracker Red illustrated that the cause of the loss of paddle-shape was extensive and localized cell death after E10.5. Although the amount of apoptosis was found to be similar in control, B1-/-, γ-/-, and B1-/-/γ-/- limb buds at E10.5 (Figure 19A-D), greatly expanded domains of cell death were visualized in both anterior and posterior regions of B1-/- forelimb buds at E11.5 (Figure 19G). In the

B1-/-/γ-/- limb buds the level of apoptosis is returned nearly to normal. In particular, the posterior domain of cell death was found to be completely rescued at this stage (Figure

19H). The same result is found for hindlimb buds (data not shown). A previous examination of B1-/- limb buds also noted increased cell death by TUNEL staining at

Figure 19: Effects of increased RA on apoptosis in limb buds.

Confocal images through equivalent sections of E10.5 (A-D) and E11.5 (E-H) forelimb buds from wild type, Cyp26b1-/-, Cyp26b1/Rarg-/-, and Rarg-/- embryos stained with LysoTracker Red for apoptotic cells. An equivalent basal level of cell death was observed in all limb buds at E10.5 (A-D), whereas discrete domains of increased cell death were observed in anterior and posterior areas within B1-null limb buds at E11.5 (G), and these were lessened following removal of Rarg (H). A similar result was found for hindlimb buds (not shown). Scale bar represents 300µM. (A, anterior; D, distal; P, posterior; Pr, proximal).

E11.5 and at E12.0 (Yashiro et al., 2004). In this previous study, apoptosis in B1-/- limbs was found to correspond to precartilaginous condensations of the long bones. The current analysis shows an additional, and previously uncharacterized, major area of cell death at

E11.5 in the anterior and posterior margins of the B1-/- limb bud which contributes to the change of shape observed from this time point onward.

100

Chapter 6: Results

Removing Rarg in a Cyp26b1-null background is able to rescue the open-eye-at-birth phenotype

6.1 Ablation of Rarg rescues eyelid closure in Cyp26b1-/- embryos Neonatal and late gestational Cyp26b1-/- mice were examined for gross morphological defects (Chapter 4) and, consistent with previously published observations

(Yashiro et al., 2004), it was observed that fetuses deficient in CYP26B1 exhibited an open-eyelid-at-birth (EOB) phenotype (see Figure 5 and Maclean et al., 2009). The EOB phenotype was found to be always bilateral and completely penetrant in Cyp26b1-/- and

Cyp26b1-/-/Rarg+/- fetuses, and this developmental defect was rescued in all Cyp26b1-/-

/Rarg-/- compound-null fetuses examined at E18.5 (n=3).

6.2 Cyp26b1 and Rarg expression in the developing eyelid Studies performed in our lab have previously illustrated that Cyp26b1 is expressed by E14.5 in the mesenchyme underlying the eyelid palpebral fissure, whereas

Cyp26a1 is not expressed in the eyelid (Abu-Abed et al., 2002). In addition, a comprehensive examination of retinoid receptor expression by IHC in mouse eyes from

E14.5 - adult has suggested that RARγ is the main isotype present in the lids during eyelid development (Mori et al., 2001). RARγ has been detected by IHC in lids at all

101 stages examined, whereas RARα was detected in the prenatal lid, but its expression decreased following birth, and RARβ was undetected (Mori et al., 2001).

Expression of both Cyp26b1 and Rarg was re-examined in frontal sections of wild type embryos at E15.0, during the stage of eyelid closure when epithelial migration across the cornea is beginning to occur (Figure 20A-D). Cyp26b1 was detected specifically in the mesenchyme of the eyelid root that directly underlies the epithelium in both the upper eyelid (uel) and lower eyelid (lel) (Figure 20C; higher magnification of the uel shown in Figure 20A). Rarg, on the other hand, was widely expressed throughout the eyelid, with stronger expression in the epithelium compared to the mesenchyme

(Figure 20D, higher magnification uel shown in Figure 20B). The probes that were used for detection of both Cyp26b1 and Rarg revealed specific domains of expression.

Staining of sections from E15.0 Cyp26b1-null (Figure 20E) and Rarg-null (Figure 20F) eyes, which was performed in parallel with the wild type sections, showed very low level background staining.

6.3 Expression of components of the FGF signalling pathway are not impaired in Cyp26b1-/- eyelids It is well established that FGF signalling elicited by FGF10 binding to the cognate receptor FGFR2b is essential for the epithelial cell migration and proliferation at the leading edge of the eyelid (Tao et al., 2005; Tao et al., 2006). If this signalling is disrupted through loss of either FGF10 or FGFR2 then EOB results

102

Figure 20: Normal expression of Cyp26b1 and Rarg in eyelids at E15.0.

Expression of Cyp26b1 (A, C, E) and Rarg (B, D, F) was examined by in situ hybridization on frontal sections through the eyes of E15.0 wild type fetuses, when eyelid closure is beginning to occur. Cyp26b1 was found to be expressed in the mesenchyme below the palpebral fissure (A and C), in agreement with previously published expression data at E14.5 in this tissue (Abu-Abed et al., 2002). Rarg was detected throughout the eyelid mesenchyme and more strongly within the epithelium (B and D). Background staining is shown by hybridization and detection of Cyp26b1 and Rarg on sections from Cyp26b1-/- (E) and Rarg-/- (F) embryos, respectively. (A-B) 20x magnification of the upper eyelid (uel). (C-F) 10x magnification showing upper and lower eyelids (lel).

103

104

(De Moerlooze et al., 2000; Tao et al., 2005). Antagonism between RA and FGFs has been described for other tissues during development, where RA is responsible for setting up domains of Fgf expression (Mercader et al., 2000; Diez del Corral et al., 2003; Desai et al., 2004; Goldbeter et al., 2007; Liu et al., 2008), so it seemed reasonable that the

EOB phenotype observed in the Cyp26b1-/- embryos might be a result of loss of Fgf10 expression and resultant impaired epithelial proliferation and migration.

Expression of Fgf10 and its receptor in the eyelid epithelium Fgfr2, were examined in frontal sections of control, B1-/-, B1-/-/γ-/-, and γ-/- eyelids at E15.0 (Figure

21A-H). Both Fgf10 and Fgfr2 were found to be expressed in eyelids from all animals examined. Fgfr2b is normally expressed more strongly in the eyelid epithelium and

Fgfr2c in the mesenchyme (De Moerlooze et al., 2000; Li et al., 2001). The probe used here was able to hybridize to both Fgfr2b and 2c, although only the 2b isoform binds

FGF10 in the eyelid. Expression of Fgfr2 was detected in the epithelium of eyelids from control, B1-/-, B1-/-/γ-/-, and γ-/- animals (Figure 21A-D). In all cases, expression was higher in the palpebral epidermis than conjunctival epidermis (blue and red arrows, respectively in Figure 21A, and corresponding regions in Figure 21B-D). Fgf10 has been observed to be mainly expressed in the mesenchyme during eyelid development

(Tao et al., 2005), but expression of Fgf10 in this instance was found to be weakly localized within the periocular mesenchyme and epidermis (Figure 21E-H) and was similarly expressed between control, B1-/-, B1-/-/γ-/-, and γ-/- eyelids at this stage. This result was further confirmed by immunohistochemical detection of FGF10 in control,

105

Figure 21: Expression analysis of mediators of FGF signalling in eyelids at E15.0.

In situ hybridization for expression of Fgfr2 (A-D) and Fgf10 (E-H) on frontal sections through the closing eyelids of control, Cyp26b1-/-, Cyp26b1-/-/Rarg-/-, and Rarg-/- fetuses at E15.0. Expression of each gene was detectable in comparable domains for all genotypes examined. Red and blue arrows in (A) point to the conjunctival epidermis and the palpebral epidermis, respectively. Immunohistochemistry for detection of FGF10 (I-L) confirmed similar protein expression of this growth factor for all genotypes examined.

106

107

B1-/-, B1-/-/γ-/-, and γ-/- eyelids (Figure 21I-L). Overall, these data confirm that expression of the necessary ligand and receptor required for FGF-dependent eyelid development were not sufficiently compromised in B1-/- eyelids in order to explain the EOB phenotype.

In order to examine whether FGF signalling was being disrupted downstream of the FGF10/FGFR2b interaction, the expression of both Shh and Mkp3 were examined in control versus B1-/- eyelids at E15.0, as these are direct transcriptional targets of FGF signalling in the eyelid (Tao et al., 2005; Li et al., 2007; Huang et al., 2009). Shh is normally expressed in both the uel and lel in a small, discrete band of epithelial cells below the palpebral conjunctiva (Figure 22A, arrowheads; and Motoyama et al., 1998).

Expression at this stage is adjacent to normal expression of Cyp26b1 in the mesenchyme

(see Figure 20C), and was found to be unchanged in B1-/- eyelids (Figure 22B).

In the control eyelid, low level Mkp3 expression was relatively ubiquitous

(Figure 22C), but was more intense in the epithelial cells which had migrated to the leading edge of the eyelid (arrowheads in Figure 22C). A similar level of Mkp3 expression could be detected in the B1-/- eyelid (Figure 22D), but not within the leading edge epithelial cells which were absent. However, normal levels of Mkp3 were observed in the B1-/- animals in other areas of active FGF signalling, such as the hair follicles

(arrow, Figure 22D) which also express Fgf10.

108

Figure 22: Downstream FGF signalling is not compromised in Cyp26b1-null eyelids at E15.0.

Expression of Shh (A-B) and Mkp3 (C-D), targets of FGF signalling, was examined by in situ hybridization on frontal sections through the developing eye region of E15.0 fetuses. Expression was found to be detectable in both control (A and C) and Cyp26b1-/- (B and D) eyelids at this stage for both genes.

109

110

6.4 There are no differences in the extent of cell proliferation or cell death in Cyp26b1-null eyelids at E15.0 Eyelid closure, like other morphogenetic processes, requires coordinated cell proliferation and migration, along with appropriately regulated cell death. It may be that the EOB phenotype observed in Cyp26b1-/- neonates is the result of either reduced cell proliferation or increased cell death, particularly in the migrating epithelial cells at the eyelid leading edge during closure between E15 and E16.5. Cell proliferation and apoptosis were measured in eyelids from control, B1-/-, and B1-/-/γ-/- fetuses at E15.0 by immunostaining against Ki-67 and cleaved Caspase-3, respectively (Figure 23). Discrete dispersed Caspase-3 positive cells were detected in both uel and lel in all cases and at similar levels (Figure 23A-F). Although cells undergoing apoptosis could be seen in both the epithelium and mesenchyme, they were mainly confined to the mesenchyme.

Proliferating cells were also detectable in control, B1-/-, and B1-/-/γ-/- uel and lel (Figure

23G-L) at similar levels.

111

Figure 23: Cell proliferation and apoptosis in eyelids at E15.0.

The extent of cell proliferation and apoptosis was examined by immunohistochemistry on frontal sections through control, Cyp26b1-/-, and Cyp26b1-/-/Rarg-/- eyes at E15.0. Apoptosis was measured using an antibody against cleaved Caspase-3 (A-F) and proliferation was detected using an antibody against Ki-67 (G-L). Detection was performed using Alexa fluorophore-conjugated secondary antibodies. Expression was observed to be similar for all genotypes examined. Specific staining was confirmed by control slides using secondary antibody only (not shown).

112

113

Chapter 7

Discussion and Conclusions

7.1 Ablation of Rarg is able to partially rescue the Cyp26b1-null phenotype Unlike the phenotype of Cyp26a1-null neonates, which is highly variable (Abu-

Abed et al., 2001; Sakai et al., 2001), the phenotype of Cyp26b1-null neonates is surprisingly consistent: shortened limbs and frontonasal region, open eyelids, and abnormal body curvature. Since (1) Rarg expression overlaps with the expression of both

Cyp26a1 and Cyp26b1 (Figure 4), (2) Rarg-null mice are resistant to RA treatment which normally causes embryonic caudal and some craniofacial defects (Iulianella and

Lohnes, 1997; Iulianella et al., 1999), and (3) the lethality and caudal defects found in

Cyp26a1-null animals can be rescued by ablation of Rarg (Abu-Abed et al., 2003), we reasoned that ablation of Rarg in Cyp26b1-null mice would rescue the craniofacial defects and perhaps also suppress the lethality of these mutants. Rarg/Cyp26b1 compound-null animals were neonatal lethal, similarly to B1-null animals (Table 2), but exhibited an improved phenotype with regards to various skeletal abnormalities (Chapter

4), limb outgrowth (Chapter 5) (Pennimpede et al., 2010), and eyelid closure (Chapter 6).

114

7.1.1 Improvement of craniofacial morphogenesis in Cyp26b1/Rarg compound-null embryos Tissues that make up the vertebrate head are derived from the classic germ layers

(ectoderm, mesoderm, and endoderm) along with the so-called “fourth germ layer” – the neural crest (Jones and Trainor, 2005; Szabo-Rogers et al., 2009). NCCs are a transient population of multipotent progenitors derived from the neural folds of the dorsal neural tube. As the neural folds fuse, these cells de-laminate, migrate to distant sites in the embryo, and proliferate, contributing to various structures (Chai and Maxson, 2006).

Disruption of NCC formation and migration results in severe developmental defects.

Cranial or cephalic NCCs originate from the forebrain, midbrain, and hindbrain.

NCCs originating from the hindbrain migrate ventrolaterally into the branchial arches, and contribute to formation of most of the bone, cartilage, muscle and nerves of the skull and face. These NCCs contribute to the frontal and interparietal plates, and bones of the jaw, inner ear, and hyoid. The parietal plate and basi and exoccipital bones, on the other hand, are derived exclusively from head mesodermal tissue (Chai and Maxson, 2006).

NCCs exhibit developmental plasticity which is affected by signals from their local environment. For example, studies in vertebrates have established that patterning and interactions between migrating NCCs and the pharangeal endoderm, along with cues from signalling molecules such as FGF8, Shh, Wnts, and BMP4, are important for regulating the fate of NCCs and the formation of craniofacial and other structures

(reviewed in Le Douarin et al., 2004; Jones and Trainor, 2005; Chai and Maxson, 2006;

Szabo-Rogers et al., 2009). RA is clearly required for formation of the BAs since genetic

115 inactivation of Raldh2 results in absence of all branchial arches except BA1

(Niederreither et al., 1999), and rescued Raldh2-/- embryos following RA treatment still lack posterior BAs 3-6 (Niederreither et al., 2001). However, although Cyp26b1 is strongly expressed in the developing hindbrain and BAs (MacLean et al., 2001) (Figure

4), and Cyp26b1-null embryos (and embryos isolated following maternal RA treatment

(Lammer et al., 1985)) exhibit many defects in NCC-derived structures (Figures 7 and 8)

(Maclean et al., 2009), we have shown that these defects are not mainly the result of compromised hindbrain patterning or NCC patterning and migration (Maclean et al.,

2009).

Embryos lacking both CYP26B1 and RARγ show improvements in structures derived not only from NCC-populated anterior and posterior branchial arches (i.e. the frontal and interparietal plates, maxilla, tympanic ring, and hyoid), but also structures arising from the paraxial mesoderm (including the parietal plate, basioccipital, and exoccipital bone) (Figures 7 and 8). We have observed differences in the streaming of

NCCs into the third and fourth BAs in Cyp26b1-null embryos by detection of Crabp1 expression, and this may contribute to defects in hyoid and clavicle development

(Maclean et al., 2009). However, these BAs are mainly populated by cardiac neural crest cells (Jones and Trainor, 2005), and the heart of Cyp26b1 mutants is formed normally

(Maclean et al., 2009). Therefore, although the subtle changes in NCC migration into the more posterior BAs observed in B1 mutants may be recovered in Cyp26b1/Rarg compound-null animals, and contribute to the rescue of hyoid and clavicle development, the overall rescue phenotype suggests that impaired development of the craniofacial 116 structures is instead due to an effect on skeletogenic patterning and/or bone formation as opposed to defects in tissue origin. In order to confirm this, conditional ablation of

Cyp26b1 specifically in NCCs could be performed using a Wnt1-Cre transgene to see if these mice recapitulate defects seen in Cyp26b1 total embryonic mutants.

The bones of the cranial base, along with the axial and appendicular skeleton, are formed by the process of endochondral ossification, wherein a framework of cartilage is first laid down and then replaced by bone. In contrast, the bones of the jaw and calvaria are formed by intramembranous ossification, where bone is formed without a cartilaginous precursor (Lana-Elola et al., 2007; Szabo-Rogers et al., 2009). At the time when bones are beginning to develop, Rarg is expressed within all chondrocyte condensations regardless of what type of ossification will follow (Ruberte et al., 1990).

Thus, it is possible that ablation of Cyp26b1 is causing premature cell differentiation and/or apoptosis in the presumptive bones of craniofacial structures, mediated through

RARγ; as was observed in the limb (Figure 19). Therefore, loss of Rarg would be able to protect these cells from the teratogenic/differentiating effects of increased RA, resulting in more complete ossification of the skull. This will be discussed in more detail below.

7.1.2 Vertebrothoracic defects in Cyp26b1-null embryos are rescued by removal of Rarg The vertebral defects observed in embryos lacking CYP26B1 were confined to the anterior cervical vertebrae, and included neural fusions of the atlas and axis, sometimes accompanied by apparent C2-C1 transformation, and absence of the tuberculi

117 anterior (Figure 9A). Thoracic defects consisted of shortened sterna, with absence and anteriorizations of sternal centers of ossification (Figure 9B), and defects of rib formation (Figure 10). All of the observed vertebrothoracic defects, with the exception of

C1-C2 fusion, could be rescued by genetic inactivation of Rarg (Figures 9 and 10, and summarized in Table 4).

Skeletal muscle, along with the ribs and vertebrae, arise from the somites during development. As somites mature they are compartmentalized into the sclerotome, myotome, and dermatome. The sclerotome makes up the ventral portion of the somite, which will give rise to the ribs and vertebrae, whereas the dorsolateral dermomyotome gives rise to the dermis and musculature of the back, abdomen, and limbs (Slack, 2006).

Nearly all mouse models which present with thoracic defects also have defects in somitogenesis and/or specification of the A-P axis (for examples see Weston et al., 2006).

Since we have observed that Cyp26b1 is expressed intersomitically at E8.5 (MacLean et al., 2001) it is conceivable that the vertebral defects observed following a loss of

CYP26B1 can be explained, at least partly, by deficiencies in somitogenesis or development of some somite-derived cell lineage. An interesting candidate process is myogenesis, since immunohistochemistry performed with antibodies against desmin and muscle specific actin revealed that B1-/- embryos have defects of trunk muscle development (Glenn MacLean, PhD thesis; unpublished results) and defects in myogenesis can also disrupt proper formation of the axial skeleton (Weston et al., 2006).

It would be, therefore, worthwhile to examine various markers of myoblast differentiation in order to discern whether this process is affected by the absence of 118

CYP26B1. For example, expression of transcription factors including MyoD, Myf5, and

Myogenin could be examined. These are required for differentiation of muscle progenitors and genetic inactivation of these genes results in rib and sternebral defects of varying severity (Brent and Tabin, 2002).

Thoracic abnormalities with accompanying cardiac defects have been recently, anecdotally, linked through epigenetic effects on posttranslational histone modifications by systematic analysis of the literature for genetic models and teratogens which exhibit both types of malformations (Weston et al., 2006). Of the teratogens discussed, both valproic acid and dimethadione cause wavy ribs in embryos recovered from treated dams, also observed here in Cyp26b1-/- embryos (Figure 10), and both also inhibit HDAC activity (Weston et al., 2006). Active and repressive RA signalling requires histone modifications (see Figure 2), so it is not inconceivable that the vertebrothoracic malformations observed here in Cyp26b1-/- embryos may involve improper histone modifications. However, it should be noted that background effects can also play a role in susceptibility to teratogens (Collins et al., 2006; Downing et al., 2009) so these studies should be interpreted with some caution. Because RA is a molecule with complex properties it is likely that the true nature of the observed malformations in these embryos is multifactorial; perhaps involving defects in somitogenesis coupled with epigenetic and gene dysregulation.

119

7.1.2.1 Evidence that loss of CYP26B1 activity results in premature differentiation of osteogenic precursors As discussed above, it is possible that the loss of bone in B1-/- embryos and partial rescue observed in B1-/-/γ-/- embryos can be explained by effects on osteogenesis. Recent evidence has emerged from two zebrafish mutants which result from either hypomorphic cyp26b1 (the stocksteif mutant) (Spoorendonk et al., 2008) or cyp26b1 loss-of-function

(dolphin mutant) (Laue et al., 2008). These studies showed that cyp26b1 is expressed in condensing chondrocytes and osteoblasts, and is required in zebrafish to prohibit premature ossification of osteoblasts directed by RA (Laue et al., 2008; Spoorendonk et al., 2008). Spatio-temporal analysis of bone formation showed that either loss of or reduced activity of the enzyme resulted in a hyperossification phenotype in axial and craniofacial structures. The same paradigm may hold true in the Cyp26b1-null mouse, and we are examining this possibility in collaboration with the zebrafish group which described the dolphin mutant, led by M. Hammerschmidt. If this is indeed the case, then loss of RARγ function within chondrogenic precursors of Cyp26b1-null mice could effectively limit the potential of RA to cause the premature differentiation of osteoblasts, providing an explanation for the rescue phenotype observed here.

The results presented in these zebrafish models follow our general paradigm of

RA action (Figure 3) wherein CYP26 enzymes are present to sculpt tissues; providing a barrier to RA signalling in order to limit differentiation and apoptosis until a time when it is required for the proper development of structures. In vivo, and under conditions of endogenous but mutagenic levels of RA, the effects on bone formation and/or

120 teratogenesis are mainly mediated by RARγ within the caudal embryo (Abu-Abed et al.,

2003) and eyelid (this work), but RARα and RARβ can compensate in other structures, including the limb for example.

7.1.2.2 Expression of Hoxa3 and Hoxb3 are not effected by exposure to increased levels of endogenous RA caused by loss of CYP26B1 function The expression levels of some Hox genes have been shown to be regulated by exogenous RA during embryogenesis including Hoxa1, Hoxb1, Hoxa3, Hoxd4, and

Hoxa5 (Duester, 2008). Indeed, several Hox genes contain RAREs and are thus directly regulated by RA such as Hoxa1, Hoxb1, Hoxa3, Hoxb3, Hoxa4, Hoxb4, and Hoxd4

(Marshall et al., 1994; Marshall et al., 1996; Gould et al., 1998; Packer et al., 1998;

Studer et al., 1998; Zhang et al., 2000; Voss et al., 2009).

Single and combinatorial mutations in Hoxa3, Hoxb3, and Hoxd3 genes can cause loss of the atlas in the absence of defects on hindbrain patterning (Manley and Capecchi,

1997). Further, the recent characterization of the phenotype of embryos deficient in the

HAT Moz illustrated that these embryos exhibit reduced/shifted Hox gene expression along with histone H3K9 hypoacetylation and hypermethylation at various Hox gene loci, including a3 and b3, which result in re-patterning of the embryonic axis and, more importantly, these effects can be reversed by RA treatment (Voss et al., 2009). Thus, we wanted to examine whether the opposite pattern existed in Cyp26b1-null mutants, wherein the C1-C2 fusion and/or the axis-to-atlas anterior transformation defect observed in Cyp26b1-/- embryos might correspond to expansion of Hox3 genes. However, no 121 changes in the anterior domain or expression level of Hoxa3 or Hoxb3 was observed at

E10.5 (Figure 11). This result was not entirely surprising since, even if the genes are responsive to endogenous levels of RA, rostral expression of the RARE-lacZ reporter was also not observed to be abnormally expanded into the hindbrain of B1-/- mice (Maclean et al., 2009), suggesting that activity of CYP26A1 and CYP26C1 in the hindbrain are sufficient to limit RA signalling to the r5/r6 boundary in the absence of CYP26B1. This is in agreement with previous studies showing that shifting boundaries of Cyp26a1 and c1 are all that is required for hindbrain specification rostral to the presumptive cervical vertebrae (Sirbu et al., 2005).

7.2 Analysis of Cyp26b1/Rarg compound-null mice reveals two genetically separable effects of retinoic acid on limb outgrowth The role of RA in limb morphogenesis has been highly debated. Limb defects in mice caused either by treatment with exogenous RA, or inappropriate exposure to endogenous levels of RA following the loss of CYP26B1, have revealed that controlled exclusion of RA from the limb bud is essential for proper limb morphology (Yashiro et al., 2004) (Figure 6). However, although excess RA is able to elicit P-D patterning defects it may not be essential for patterning itself to occur (Zhao et al., 2009). By modulating both RA distribution (following loss of CYP26B1) and signalling (through the additional loss of RARγ) we have uncovered two separate effects of RA in the limb which are able to reconcile these observations. The primary effect of RA in causing limb

122 defects is a “teratogenic” one, and the second effect is evidenced by P-D patterning defects caused by the expansion of the RA-responsive Meis genes.

7.2.1 The role of RARγ in the limb Mice deficient in RARγ can survive to adulthood and exhibit defects such as stunted growth, male sterility, and homeotic vertebral transformations, but have normal development of the limbs (Lohnes et al., 1993). Interestingly, following bolus treatment with a teratogenic dose of RA, caudal and craniofacial regions of Rarg-/- embryos were resistant to defects normally associated with RA treatment, whereas the limbs were not found to be resistant to teratogenesis (Iulianella and Lohnes, 1997). This is likely because

RARα can also mediate the RA signal in the limb following this supra-physiological dose. However, using a genetic model to examine the effects of modulating RA signalling following a change in the endogenous distribution of RA (following removal of CYP26B1), we have uncovered in vivo evidence that RARγ is functional in limb skeletal development in the presence of physiologically-relevant levels of RA; with the resulting compound mutant displaying an improved skeletal phenotype with a rescued stylopod and increased number of digits (Figure 6 and Table 5). This approach illustrates the limitations in using exogenously administered teratogenic doses of RA to interpret the role of RA in the development of various structures.

123

7.2.2 The nature of the P-D patterning defect caused by loss of Cyp26b1 is revealed following loss of Rarg Although it has been previously demonstrated, as well as shown here, that excess

RA can upregulate the stylopod-specific gene Meis2 (Mercader et al., 2000; Yashiro et al., 2004) (Figure 14), it should be noted that endogenous RA may not, in fact, be required for the induction of Meis2. For example, Raldh2-/- embryos at E10.5 which have been rescued with a low dose of RA retain Meis2 expression in the limb proximal mesoderm, even though RA is apparently absent from this tissue (Niederreither et al.,

2002d; Zhao et al., 2009). However, the possibility remains, albeit remote, that low levels of RA that are below the indirect methods of detection used in these studies are sufficient to trigger key patterning events. It is also noteworthy that specification of distal cells by

FGFs is not compromised in Cyp26b1-/- limb buds even though the Meis-positive domain is expanded, since Fgf8 expression was found to be normal (Yashiro et al., 2004, and

Figure 13A-B), and FGF signalling measured by Mkp3 expression and pERK levels

(Yashiro et al., 2004) is unaffected. We also showed here that the AER morphology is not compromised by loss of CYP26B1 (Figure 13G), which further illustrates that distal cell fate is unaffected.

The patterning defect caused by Meis expansion in the absence of CYP26B1 is obscured by cell death, but becomes apparent following removal of RARγ in CYP26B1 knockout mice. Our results show that the induction of Meis2 by RA found in the absence of CYP26B1 is mediated by RARs other than RARγ (presumably RARα). The Meis- negative area of the presumptive autopod in the B1-/-/γ-/- limb bud is larger than that of the

124

B1-/- limb bud, which likely contributes to the improvement in digit formation seen in the compound-null limbs. Since the stylopod is completely rescued in the B1-/-/γ-/- limbs, and

Meis2 expression was found to encompass the presumptive stylopod region along with zeugopod and part of the autopod in B1-/-/γ-/- limb buds (Figure 14G-H and Q-R; summarized in Figure 24A-B), our results also illustrate that restriction of Meis to the presumptive stylopod/flank is not essential for formation of proximal skeletal elements.

According to the prevailing notion that Meis expression is instructive for stylopod formation, along with the phenotype observed following ectopic Meis expression distally

(Mercader et al., 2009), expanded Meis expression into the presumptive zeugopod region

(as is seen here in B1-/- and B1-/-/γ-/- limbs) predicts re-specification of those cells to a more proximal fate, resulting in normal stylopod formation, but a shortened zeugopod.

This is not what is observed following removal of CYP26B1 alone (Yashiro et al., 2004, and Figure 6B and F) but is consistent with what is observed in limb buds progressively devoid of FGFs (Mariani et al., 2008). However, in each of those models the patterning effect of Meis expansion is overshadowed by significant cell death, whereas the stylopod can be observed in our model following ablation of both CYP26B1 and RARγ (Figure

6D and H), since the mediator of the apoptotic insult (RARγ) is absent.

125

Figure 24: Overview of the consequences following loss of CYP26B1 and RARγ on final skeletal outcome in the limb.

(A) RA (orange) is generated from the proximal flank mesoderm, and FGFs (green) signal from the AER. In the absence of CYP26B1, RA extends more distally, while the FGF signal remains unchanged. (B) Expression of Meis (purple) is dependent on RA, and is therefore distally expanded in B1-null and B1/γ compound-null limbs. The distal limit of expansion is equivalent in B1-/- and B1-/-/γ-/- buds, but their different shapes at E11.5 results in a larger amount of Meis-free tissue (brackets), which may contribute to increased digit number observed in compound-null limbs (D). Although Meis expression remains distalized in B1-/-/γ-/- limb buds, the most proximal limb element (the stylopod) is completely rescued in B1-/-/γ-/- limbs (D). Inappropriate expansion of RA also results in elevated apoptosis (shown in pink) within B1-/- limbs, causing gross morphological changes in bud shape, but the loss of RARγ reduces the overall cell death, particularly the posterior apoptotic domain (C). (D) Overall, the loss of Rarg in a Cyp26b1-null background rescues limb skeletal elements, particularly the stylopod (green) and autopod (yellow), as opposed to the zeugopod (blue).

126

127

The specific rescue of stylopod formation in limbs of compound-null embryos led us to look for knockout phenotypes where there is specific loss of the stylopod, especially those that have been shown to be regulated by RA, of which Shox2 was an attractive candidate gene. Shox2 expression normally extends from the flank and encompasses both the presumptive stylopod and zeugopod, and ablation of Shox2 results in absence exclusively of the stylopod (Cobb et al., 2006; Yu et al., 2007). Moreover, RA bead implantation in chicks showed that expression of Shox could be repressed by RA (Tiecke et al., 2006). However, Shox2 expression remained unchanged in the Cyp26b1-null limb buds (Figure 15). This illustrates both that Shox2 is not repressed in vivo by endogenous levels of RA, and that the lack of stylopod seen in B1-/- limbs and the corresponding rescue in B1-/-/γ-/- embryos is not the result of alterations in Shox expression.

Analysis of Hox genes which had previously been shown to be effected by loss of

CYP26B1, including Hoxa11, Hoxa13, and Hoxd13, was performed in the compound- null limbs to further examine the rescue phenotype. Expression of Hoxa11 was shifted distally in E11.5 B1-/- limbs, in agreement with previously published data (Yashiro et al.,

2004), and this shift remains in the compound-null limb buds (Figure 16K-T, brackets).

This is likely a result of the combined loss of distal tissue in B1-/-, B1-/-/γ+/-, and B1-/-/γ-/- limb buds, along with distally expanded RA and resultant Meis expression encroaching on the presumptive zeugopod/Hoxa11-expressing region; since Meis as well as RA repress the expression of distal Hox genes (Capdevila et al., 1999; Mercader et al., 1999;

Mercader et al., 2000). This is also what is observed here in terms of Hoxa13 and

Hoxd13, which were downregulated in B1-/- limb buds (Hoxa13; Figure 18C-H and M- 128

R) or distally shifted (Hoxd13; Figure 17M-R). The further reduction in Hoxa13 expression in B1-/-/γ-/- limbs caused by loss of RARγ is not sufficient to explain the defects in autopod formation in these mice because, although Hoxa13-/- mice lack formation of the anteriormost digit (Fromental-Ramain et al., 1996), Rarg-/- mice have normal limbs. This suggests that residual expression of Hoxa13 is sufficient, along with maintenance of Hoxd13 to allow for a normal autopod to be formed. Notably, expression of Hoxa11 was maintained in B1-/- and B1-/-/γ+/- limbs even though the zeugopod fails to form in these mice (Figure 16C-F and M-P) which is consistent with what has been observed following progressive disruption of limb FGFs, which also lack zeugopod formation (Mariani et al., 2008) and formed the basis for the two-signal model of limb development (see Chapter 2).

7.2.3 Ablation of Cyp26b1 causes increased cell death in the limb bud which is ameliorated following removal of Rarg We are presently unsure as to why cells located in the anterior and posterior limb mesenchyme are specifically targeted for cell death following ablation of Cyp26b1, or the mechanism resulting in increased cell survival following loss of Rarg. The result, however, is that inappropriate exposure to endogenous levels of RA causes a reduction in the amount of mesenchyme in B1-/- limb buds, affecting gross morphology. This is in agreement with previous studies showing that RA treatment can cause an expansion in areas which are normal sites of basal programmed cell death (Alles and Sulik, 1989).

129

Mesenchymal limb bud volume is likely to be a critical determinant of final skeletal pattern (Sun et al., 2002; Niswander, 2003; Tabin and Wolpert, 2007; Mariani et al., 2008). In fact, it has recently been shown that phocomelia caused by X-irradiation in chick limbs is a result of a reduction in the number of skeletal progenitors, as opposed to changes in P-D patterning (Galloway et al., 2009). Our results provide a mammalian genetic model that is consistent with this by illustrating a reduction in apoptosis in

B1-/-/γ-/- limb buds, in comparison to B1-null limbs, independently of rescued expression of classical P-D patterning genes (summarized in Figure 24). These results show that the effects of RA on cell death in the limb are mediated mainly through RARγ, which can account for the loss of digits in the B1-/- embryos, and the corresponding rescue of all but one digit in compound-null embryos. Although RA is known to interfere with chondrogenesis in vitro (Weston et al., 2002), Sox9 expression is retained in B1-null limbs (Yashiro et al., 2004). Since Rarg is expressed during later stages of embryogenesis in all chondrogenic precursors (Dolle et al., 1990; Ruberte et al., 1990), it is conceivable that at later stages the loss of RARγ in chondrocytes also aids in their survival, and the observed increase in skeletal mass.

7.2.4 Perspectives The primary role of RA in the early developing limb is to initiate limb bud outgrowth (Zhao et al., 2009), following which Cyp26b1 expression commences in order to keep the distal limb bud in an RA-free state. Although the data presented here illustrate

130 that RA is able to affect pattern in the proximal limb, this may be an effect of increased and inappropriate RA exposure in this tissue rather than a natural role for RA in limb morphogenesis, since RA is seemingly not required for early P-D limb patterning (Zhao et al., 2009). The gate-keeping function of CYP26 is critical to avoid RA-induced apoptosis, and likely also premature differentiation. This notion is supported by our results showing premature entry of germ cells into meiosis in Cyp26b1-null testes

(MacLean et al., 2007) and is also what is seen in two zebrafish models of Cyp26b1 mutation where increased RA results in premature ossification of skeletal structures

(Laue et al., 2008; Spoorendonk et al., 2008). Taken together, these results suggest that the primary role of CYP26 in the limb, and likely other tissues, is not to necessarily act as a “sink” in creating a morphogen gradient (as proposed in Yashiro et al., 2004; and see

Chapter 2, section 2.3: RA and morphogenesis), but rather to act as a barrier in order to limit cellular exposure to RA until such time as it is required for growth, differentiation, or apoptosis.

7.3 Failure of eyelid closure in Cyp26b1-null fetuses does not result from disrupted FGF signalling, decreased cell proliferation, or increased apoptosis in eyelid anlagen

7.3.1 The role of RA in the developing eye It is well-established that RA signalling is essential for proper eye development.

Developmental defects of the eye were the most prominent phenotype observed in rodents following gestational VAD (Wilson et al., 1953). Accordingly, several retinoid receptor compound mutants display various congenital ocular abnormalities that are 131 associated with VAD including retinal degeneration, persistence and hyperplasia of the primary vitreous body (also found in Rarb single mutants (Ghyselinck et al., 1997)), microphthalmia, and sclera and choroid agenesis (defects reviewed in Mark et al., 2009).

Rarb/Rarg compound-null animals have also been shown to exhibit ocular defects that resemble those from Raldh1/3 compound-null animals (Ghyselinck et al., 1997). Raldh1 and Raldh3 are expressed in the eyelid grooves and corneal ectoderm at E13.5, and loss of both enzymes results in a shortening of the ventral retina along with failure of upper eyelid groove invagination and eventual fusion of the lower eyelid, yielding a thick layer of mesenchyme over the cornea by E14.5, as opposed to proper eyelid formation and closure (Matt et al., 2005).

It has been demonstrated that RARγ is the main RAR expressed in the eyelid during development (Mori et al., 2001). RARγ may have redundant functions with other

RARs in eyelid development since Rarg-/- fetuses have normal development of the eyelids (Lohnes et al., 1993) but Rarb/g compound-null animals exhibit complete penetrance of blepharophimosis (narrow lid closure) (Grondona et al., 1996). Several studies have shown that RA is required for optic cup formation (Mic et al., 2004), and for apoptosis in the NCC-derived perioptic mesenchyme (Matt et al., 2005; Molotkov et al.,

2006), which is necessary for proper formation of the cornea and eyelids (Duester, 2009).

The effects of excess RA on eye development are less well explored in the literature, but have been found to include open eyelids and microphthalmia (Shenefelt, 1972; Sulik et al., 1995). In addition, transgenic mice which express a constitutively active RARα under

132 control of the αA-crystallin promoter for specific expression in the lens were found to exhibit cataracts and microphthalmia (Balkan et al., 1992).

The expression patterns of Raldh enzymes in the developing eye, and detection of

RA using the RARE-lacZ reporter, suggest that RA within the developing eyelid is found in the retina and eyelid grooves, and would signal through a paracrine mechanism to the eyelid epithelium (Matt et al., 2005; Molotkov et al., 2006; Duester, 2009). Specific expression of Cyp26b1 in the mesenchyme of the eyelid root, just below the conjunctival/palpebral epithelial border (Figure 20), suggests that its expression is necessary in order to limit RA signalling in the eyelid leading edge in order to ensure proper formation of the eyelids. This is illustrated genetically following loss of Cyp26b1, which always results in EOB (Figure 5).

7.3.2 Mechanisms of open-eyelid phenotype following ablation of Cyp26b1 There are various interacting cell signalling pathways which have been linked to

EOB. These include FGF10/FGFR2b signalling which is important for early eyelid formation and proliferation and migration of the leading edge cells (De Moerlooze et al.,

2000; Li et al., 2001; Tao et al., 2005). Also, EGF ligands such as HB-EGF and TGFα act through EGFR in the eyelid to activate keratinocyte migration (Luetteke et al., 1993;

Mann et al., 1993; Xia and Kao, 2004; Mine et al., 2005), and activin signalling through the MEKK/JNK/c-Jun pathway results in actin stress fiber formation and epithelial migration (Zhang et al., 2003; Xia and Kao, 2004).

133

The recent characterization of lack of eyelid formation in Fgf10-null mice revealed that Tgfα and Actβb were downregulated in the proliferating cells of the developing eyelid. This suggested that FGF10 acts upstream of TGFα and Actβb in eyelid development, and the authors purported that it is loss of these signalling molecules which ultimately resulted in the EOB phenotype (Tao et al., 2005; Tao et al., 2006).

However, although ablation of both the EGFR ligand Tgfα and the TGFβ-family growth factor Actβb result in EOB (Luetteke et al., 1993; Mann et al., 1993; Vassalli et al.,

1994), in both cases the EOB phenotype was incompletely penetrant and varied among affected animals. Occasionally, only a single eye was affected, and neonate littermates exhibited various degrees of eyelid closure, whereas EOB in Fgf10-null animals is always observed and is bilateral (Tao et al., 2005). It could be that loss of both growth factors is sufficient to always cause EOB, or that the reduction of Tgfα and Actβb expression is not the main/only cause of the EOB phenotype seen in Fgf10-null mice.

Expression of Tgfα and Actβb was examined in frontal sections of control, B1-/-, B1-/-/γ-/-, and γ-/- eyelids at E15.0 with inconclusive results (data not shown). No apparent differences in expression were noted for either Tgfα or Actβb for any of the animals examined at this stage, but the localization of staining did not appear to be specific.

Attempts at immunohistochemical localization of TGFα expression in control and B1-/- eyelids were also unsuccessful (data not shown).

Similarities in Cyp26b1-null and Fgf10-null embryos, along with evidence supporting antagonism between FGFs and RA in other tissues (Mercader et al., 2000;

Diez del Corral et al., 2003; Desai et al., 2004; Goldbeter et al., 2007; Liu et al., 2008), 134 led us to hypothesize that the EOB phenotype observed in B1-/- neonates resulted from loss of Fgf10. However, analysis of expression of Fgf10, its receptor Fgfr2 (Figure 21), and the downstream FGF signalling targets in the eyelid Shh and Mkp3 (Figure 22) illustrated that FGF10/FGFR2 signalling was not disrupted in B1-/- eyelids.

There also exist interactions between the pathways that are known to effect eyelid closure. For example, expression of c-Jun has been found to be essential for eyelid closure, and this is at least partly through its ability to regulate EGFR and HB-EGF levels in the eyelid epidermis (Li et al., 2003; Zenz et al., 2003). There are also a complement of overlapping intracellular signalling pathways activated by EGFRs and FGFRs

(Schlessinger, 2004). We acknowledge that effects on pathways other than FGF signalling have not been examined here in any detail, and it may be that the increase in local RA caused by the loss of CYP26B1 is causing disruption of EGF/ERK or

MEKK/JNK/c-Jun signalling, for example.

7.3.3 Links between EOB, impaired wound healing, and defects in skin/hair development Eyelid closure in the mouse is a paradigm for the mechanisms of wound healing and, thus, many (although not all) of the knockouts that have been shown to result in

EOB also exhibit impaired healing of skin wounds, including loss of c-Jun (Li et al.,

2003), TGFα , and inhibition of activin signalling (Wankell et al., 2001). These mice also, when they survive long enough to be able to display this, often have defects in skin/hair development, including those with loss of TGFα (Luetteke et al., 1993; Mann et

135 al., 1993), FGF10 (Ohuchi et al., 2003), FGFR2 (De Moerlooze et al., 2000; Li et al.,

2001), and animals overexpressing the activin antagonist follistatin in the epidermis

(Wankell et al., 2001). Since we have observed abnormalities in skin and hair follicle development in Cyp26b1-null embryos (G. MacLean, PhD thesis, 2007; unpublished results) it is tempting to speculate not only that these animals will exhibit deficient wound healing, but also that there is some convergent molecule(s) which we have not yet examined linking increased RA to the FGF, EGFR, or activin signalling pathways. It is likely that the target of increased RA is a transcriptional target as opposed to an effect on signalling per se. A possible target is Bmp4, since Bmp signalling has recently been implicated to be downstream of FGF signalling in eyelid development, and Bmp4 is expressed in the same cells which express Cyp26b1 in the eyelid (Huang et al., 2009). We are interested in investigating this possibility.

7.3.4 Effects on cell proliferation, migration, and/or apoptosis by RA during eyelid closure RA is pleiotropic molecule with known effects on cell proliferation, differentiation, and apoptosis (Ross et al., 2000) (Chapter 2; Figure 3). We examined the possibility that loss of CYP26B1 from the eyelid mesenchyme could be resulting in EOB through reduced proliferation or increased apoptosis in the eyelid epithelium at E15.0

(Figure 23). No noticeable differences in the amount of proliferation or apoptosis were noted between control and B1-/- eyelids at this stage, but this does not exclude the possibility that effects on cell survival may manifest earlier or later than this timepoint. 136

The mechanism through which loss of Rarg rescues limb development was found to correlate with increased domains of cell survival (Figure 19) and the rescue of eyelid development in B1-/-/γ-/- embryos might have a similar origin. In order to determine for certain whether the increased RA is resulting in changes in cell survival, we will be performing whole-mount staining on embryos during eyelid closure from E14.5-E16.5 using LysoTracker Red.

There is also some evidence to suggest that epithelial cell migration might be impaired in Cyp26b1-null eyelids. For example expression of Fgfr2 in B1-null epidermis shows a thickening away from the leading edge (Figure 21B) which is characteristic in eyelids that have defects in formation of the eyelid leading edge, such as Grhl3/Get1-/-

(Yu et al., 2008). In addition, cells of the migrating epithelial leading edge at E15.0 were observed in control lids, but never seen in B1-null eyelids (for example see Figure 22C-

D). In order to examine this possibility the expression of keratin 6, a marker of embryonic eyelid migrating keratinocytes (Mazzalupo et al., 2003; Yu et al., 2008), can be examined in control, B1-/-, and B1-/-/γ-/- eyelids to observe whether cells are able to move to the leading edge in B1-null eyelids.

7.4 Perspectives and future directions RA is a critical molecule for eliciting cellular differentiation, proliferation, and apoptosis; and whether a cell sees RA or not will determine its fate during a specific developmental process and within a particular tissue. Since RARα is ubiquitously

137 expressed throughout embryonic development (Dolle et al., 1990; Ruberte et al., 1991;

Mark et al., 2009) and detection of RA using embryos containing a highly specific and sensitive RARE-LacZ transgene has illustrated that, in embryos treated with RA, β-gal activity can be detected throughout the entire embryo (Rossant et al., 1991), every cell in a developing embryo has the ability to respond to RA, in principle. Thus, regulating RA distribution and signalling is critical for tissue morphogenesis, and determining the mechanisms by which this is accomplished provides insight into a variety of processes which are not limited only to development. Decades of research on retinoid biology has not even come close to elucidating all of the molecular events controlled by RA. It is a molecule with extremely complex actions which include combined interactions with other signalling molecules (for example FGF and SHH), induction and repression of genes important for regulating signalling pathways, coupled with the ability to elicit paracrine cellular effects; and all of this is dependent on the tissue and developmental stage.

This thesis examines the phenotypic and molecular consequences of removing a mediator of RA signalling in the embryo (RARγ) following changes to the endogenous distribution of RA. It was expected that craniofacial defects in Cyp26b1-null embryos would be rescued by loss of Rarg, and it was hoped that this compound-null model would also revert the lethality of B1-/- neonates in order to provide insight into the role of

CYP26B1 beyond embryogenesis. Unfortunately, lethality was not rescued, but the model did provide novel information about tissues where RARγ can specifically mediate

138

RA signalling, including development of the axial skeleton (Chapter 4), the limb (Chapter

5) (Pennimpede et al., 2010), and in eyelid development (Chapter 6).

The wide variety of tissues which express and are affected by removal of

Cyp26b1, not only in the embryo but also the adult, each provide a case for examining the mechanism of RA action following loss of CYP26B1. Since we have generated mice harbouring the Cyp26b1 allele flanked by loxP sites, we can use these to create various

Cyp26b1 conditional knockout animals, as we have already successfully done to examine the function of CYP26B1 in Sertoli cells (Li et al., 2009). For example, conditional knock-out models can be used to provide information into the role of this enzyme in tissues where we know it is strongly, and often differentially expressed, including brain, gonads, bones, and skin. The skin is a particularly interesting tissue because it will allow us to examine the result of loss of B1 on hair cycling and wound-healing.

7.4.1 Evolutionary considerations from observations of Cyp26b1/Rarg compound-null embryos Several excellent reviews which attempt to uncover the evolutionary origins and functional significance of RA signalling have been published (for examples see Marletaz et al., 2006; Campo-Paysaa et al., 2008; Albalat, 2009; Albalat and Canestro, 2009). RA signalling was previously thought to be a chordate invention, but the basic machinery consisting of RA binding proteins, metabolic enzymes, and receptors has more recently been found in some bilaterian animals of the protostome (e.g. molluscs, drosophila) along with deuterostome (e.g. sea cucumbers, vertebrates) lineages (Marletaz et al., 2006; 139

Albalat, 2009). Presence of the machinery does not automatically translate to a functional significance; evidence of RA regulation of Hox genes, for example, is still limited to chordates (Marletaz et al., 2006). Effects on somite segmentation, number, and symmetry appears to be limited to vertebrates (Campo-Paysaa et al., 2008), although in the cephalochordate amphioxus the position of somites along the A-P axis can be affected by

RA (Schubert et al., 2006). The role of RA on central nervous system development also appears to be vertebrate-specific (Campo-Paysaa et al., 2008).

The elaboration of roles for RA signalling in vertebrates may be correlated with the two whole genome duplication events that have occurred during vertebrate evolution

(Dehal and Boore, 2005) since, for example, amphioxus has only one rar, whereas vertebrates have three. Although a number of independent losses of the machinery have occurred over evolutionary time, notably, almost all species examined that have been found to posses a gene for Rar also have genes for Aldh and Cyp26, which may define the basic RA signalling module (Albalat and Canestro, 2009). However, very little is known about the expression patterns of these genes in animals other than in the standard vertebrate model systems.

The question as to why Rarg expression significantly overlaps with expression of the Cyp26 enzymes arises from work presented here. In other words, what evolutionary advantage/mechanism would be served by having an RA receptor present in tissues which are normally devoid of, and which are often acutely sensitive to RA? Although data presented here and elsewhere clearly shows that RA is able to actively signal through RARγ and cause changes in gene expression, there are also arguments that RARs 140 can play a role in the active repression of genes in the absence of RA (Koide et al., 2001;

Weston et al., 2002; Weston et al., 2003; Mark et al., 2009), including the recent suggestion of a role for RARγ in ligand-less repression during skeletal growth (Williams et al., 2009). From our observations in Cyp26a1/Rarg and Cyp26b1/Rarg compound-null animals we cannot rule out the possibility that RARγ plays a ligand-independent repressive role in development, and that some of the phenotypic effects seen following removal of Rarg are the result of loss of active repression. However, this is very difficult to prove and in order to definitively do this would require the generation and crossing of transgenic mice which have dominant-negative (repressive) RARγ function caused by disruptions in the activation function domains of the receptor. This may be similar to what has been done for RXRα lacking the AF domains (Mascrez et al., 1998; Mascrez et al., 2001; Mascrez et al., 2009). The argument that RARγ has some active repressor function makes evolutionary sense and would fit with the expression pattern for RARγ. It also presents another level in the very complex regulation of RA distribution and signalling during morphogenesis and adult tissue homeostasis.

141

References:

Abu-Abed S, Dolle P, Metzger D, Beckett B, Chambon P, Petkovich M. 2001. The retinoic acid- metabolizing enzyme, CYP26A1, is essential for normal hindbrain patterning, vertebral identity, and development of posterior structures. Genes Dev 15:226-240. Abu-Abed S, Dolle P, Metzger D, Wood C, MacLean G, Chambon P, Petkovich M. 2003. Developing with lethal RA levels: genetic ablation of Rarg can restore the viability of mice lacking Cyp26a1. Development 130:1449-1459. Abu-Abed S, MacLean G, Fraulob V, Chambon P, Petkovich M, Dolle P. 2002. Differential expression of the retinoic acid-metabolizing enzymes CYP26A1 and CYP26B1 during murine organogenesis. Mech Dev 110:173-177. Abu-Abed SS, Beckett BR, Chiba H, Chithalen JV, Jones G, Metzger D, Chambon P, Petkovich M. 1998. Mouse P450RAI (CYP26) expression and retinoic acid-inducible retinoic acid metabolism in F9 cells are regulated by retinoic acid receptor gamma and retinoid X receptor alpha. J Biol Chem 273:2409-2415. Affolter M, Basler K. 2007. The Decapentaplegic morphogen gradient: from pattern formation to growth regulation. Nat Rev Genet 8:663-674. Albalat R. 2009. The retinoic acid machinery in invertebrates: ancestral elements and vertebrate innovations. Mol Cell Endocrinol 313:23-35. Albalat R, Canestro C. 2009. Identification of Aldh1a, Cyp26 and RAR orthologs in protostomes pushes back the retinoic acid genetic machinery in evolutionary time to the bilaterian ancestor. Chem Biol Interact 178:188-196. Alexander T, Nolte C, Krumlauf R. 2009. Hox genes and segmentation of the hindbrain and axial skeleton. Annu Rev Cell Dev Biol 25:431-456. Allenby G, Bocquel MT, Saunders M, Kazmer S, Speck J, Rosenberger M, Lovey A, Kastner P, Grippo JF, Chambon P, et al. 1993. Retinoic acid receptors and retinoid X receptors: interactions with endogenous retinoic acids. Proc Natl Acad Sci U S A 90:30-34. Alles AJ, Sulik KK. 1989. Retinoic-acid-induced limb-reduction defects: perturbation of zones of programmed cell death as a pathogenetic mechanism. Teratology 40:163-171. Ashe HL, Briscoe J. 2006. The interpretation of morphogen gradients. Development 133:385-394. Aulehla A, Pourquie O. 2006. On periodicity and directionality of somitogenesis. Anat Embryol (Berl) 211 Suppl 1:3-8. Balkan W, Klintworth GK, Bock CB, Linney E. 1992. Transgenic mice expressing a constitutively active retinoic acid receptor in the lens exhibit ocular defects. Dev Biol 151:622-625. Balmer JE, Blomhoff R. 2002. Gene expression regulation by retinoic acid. J Lipid Res 43:1773- 1808. Bardot O, Aldridge TC, Latruffe N, Green S. 1993. PPAR-RXR heterodimer activates a peroxisome proliferator response element upstream of the bifunctional enzyme gene. Biochem Biophys Res Commun 192:37-45. Bastien J, Rochette-Egly C. 2004. Nuclear retinoid receptors and the transcription of retinoid- target genes. Gene 328:1-16. Benazet JD, Bischofberger M, Tiecke E, Goncalves A, Martin JF, Zuniga A, Naef F, Zeller R. 2009. A self-regulatory system of interlinked signaling feedback loops controls mouse limb patterning. Science 323:1050-1053.

142

Blomhoff R, Blomhoff HK. 2006. Overview of retinoid metabolism and function. J Neurobiol 66:606-630. Bourguet W, Vivat V, Wurtz JM, Chambon P, Gronemeyer H, Moras D. 2000. Crystal structure of a heterodimeric complex of RAR and RXR ligand-binding domains. Mol Cell 5:289- 298. Brent AE, Tabin CJ. 2002. Developmental regulation of somite derivatives: muscle, cartilage and tendon. Curr Opin Genet Dev 12:548-557. Campo-Paysaa F, Marletaz F, Laudet V, Schubert M. 2008. Retinoic acid signaling in development: tissue-specific functions and evolutionary origins. Genesis 46:640-656. Capdevila J, Tsukui T, Rodriquez Esteban C, Zappavigna V, Izpisua Belmonte JC. 1999. Control of vertebrate limb outgrowth by the proximal factor Meis2 and distal antagonism of BMPs by Gremlin. Mol Cell 4:839-849. Chai Y, Maxson RE, Jr. 2006. Recent advances in craniofacial morphogenesis. Dev Dyn 235:2353-2375. Chambon P. 1996. A decade of molecular biology of retinoic acid receptors. Faseb J 10:940-954. Chen H, Johnson RL. 1999. Dorsoventral patterning of the vertebrate limb: a process governed by multiple events. Cell Tissue Res 296:67-73. Chen JD, Evans RM. 1995. A transcriptional co-repressor that interacts with nuclear hormone receptors. Nature 377:454-457. Chiang C, Litingtung Y, Harris MP, Simandl BK, Li Y, Beachy PA, Fallon JF. 2001. Manifestation of the limb prepattern: limb development in the absence of sonic hedgehog function. Dev Biol 236:421-435. Chithalen JV, Luu L, Petkovich M, Jones G. 2002. HPLC-MS/MS analysis of the products generated from all-trans-retinoic acid using recombinant human CYP26A. J Lipid Res 43:1133-1142. Chotteau-Lelievre A, Dolle P, Gofflot F. 2006. Expression analysis of murine genes using in situ hybridization with radioactive and nonradioactively labeled RNA probes. Methods Mol Biol 326:61-87. Cobb J, Dierich A, Huss-Garcia Y, Duboule D. 2006. A mouse model for human short-stature syndromes identifies Shox2 as an upstream regulator of Runx2 during long-bone development. Proc Natl Acad Sci U S A 103:4511-4515. Cohlan SQ. 1954. Congenital anomalies in the rat produced by excessive intake of vitamin A during pregnancy. Pediatrics 13:556-567. Collins MD, Eckhoff C, Weiss R, Resnick E, Nau H, Scott WJ, Jr. 2006. Differential teratogenesis of all-trans-retinoic acid administered on gestational day 9.5 to SWV and C57BL/6N mice: emphasis on limb dysmorphology. Birth Defects Res A Clin Mol Teratol 76:96-106. Collins MD, Mao GE. 1999. Teratology of retinoids. Annu Rev Pharmacol Toxicol 39:399-430. Copp A, Cogram P, Fleming A, Gerrelli D, Henderson D, Hynes A, Kolatsi-Joannou M, Murdoch J, Ybot-Gonzalez P. 2000. Neurulation and neural tube closure defects. Methods Mol Biol 136:135-160. Crossley PH, Martin GR. 1995. The mouse Fgf8 gene encodes a family of polypeptides and is expressed in regions that direct outgrowth and patterning in the developing embryo. Development 121:439-451. de Bruijn DR, Oerlemans F, Hendriks W, Baats E, Ploemacher R, Wieringa B, Geurts van Kessel A. 1994. Normal development, growth and reproduction in cellular retinoic acid binding protein-I (CRABPI) null mutant mice. Differentiation 58:141-148.

143

De Moerlooze L, Spencer-Dene B, Revest JM, Hajihosseini M, Rosewell I, Dickson C. 2000. An important role for the IIIb isoform of fibroblast growth factor receptor 2 (FGFR2) in mesenchymal-epithelial signalling during mouse organogenesis. Development 127:483- 492. de Roos K, Sonneveld E, Compaan B, ten Berge D, Durston AJ, van der Saag PT. 1999. Expression of retinoic acid 4-hydroxylase (CYP26) during mouse and Xenopus laevis embryogenesis. Mech Dev 82:205-211. de Urquiza AM, Liu S, Sjoberg M, Zetterstrom RH, Griffiths W, Sjovall J, Perlmann T. 2000. Docosahexaenoic acid, a ligand for the retinoid X receptor in mouse brain. Science 290:2140-2144. Dehal P, Boore JL. 2005. Two rounds of whole genome duplication in the ancestral vertebrate. PLoS Biol 3:e314. Depoix C, Delmotte MH, Formstecher P, Lefebvre P. 2001. Control of retinoic acid receptor heterodimerization by ligand-induced structural transitions. A novel mechanism of action for retinoid antagonists. J Biol Chem 276:9452-9459. Desai TJ, Malpel S, Flentke GR, Smith SM, Cardoso WV. 2004. Retinoic acid selectively regulates Fgf10 expression and maintains cell identity in the prospective lung field of the developing foregut. Dev Biol 273:402-415. Diez del Corral R, Olivera-Martinez I, Goriely A, Gale E, Maden M, Storey K. 2003. Opposing FGF and retinoid pathways control ventral neural pattern, neuronal differentiation, and segmentation during body axis extension. Neuron 40:65-79. Diez del Corral R, Storey KG. 2004. Opposing FGF and retinoid pathways: a signalling switch that controls differentiation and patterning onset in the extending vertebrate body axis. Bioessays 26:857-869. Dilworth FJ, Chambon P. 2001. Nuclear receptors coordinate the activities of chromatin remodeling complexes and coactivators to facilitate initiation of transcription. Oncogene 20:3047-3054. Dolle P. 2009. Developmental expression of retinoic acid receptors (RARs). Nucl Recept Signal 7:e006. Dolle P, Ruberte E, Kastner P, Petkovich M, Stoner CM, Gudas LJ, Chambon P. 1989. Differential expression of genes encoding alpha, beta and gamma retinoic acid receptors and CRABP in the developing limbs of the mouse. Nature 342:702-705. Dolle P, Ruberte E, Leroy P, Morriss-Kay G, Chambon P. 1990. Retinoic acid receptors and cellular retinoid binding proteins. I. A systematic study of their differential pattern of transcription during mouse organogenesis. Development 110:1133-1151. Downing C, Balderrama-Durbin C, Broncucia H, Gilliam D, Johnson TE. 2009. Ethanol teratogenesis in five inbred strains of mice. Alcohol Clin Exp Res 33:1238-1245. Downs KM, Davies T. 1993. Staging of gastrulating mouse embryos by morphological landmarks in the dissecting microscope. Development 118:1255-1266. Duboule D. 1994. Temporal colinearity and the phylotypic progression: a basis for the stability of a vertebrate Bauplan and the evolution of morphologies through heterochrony. Dev Suppl:135-142. Duboule D. 1998. Vertebrate hox gene regulation: clustering and/or colinearity? Curr Opin Genet Dev 8:514-518. Duboule D, Dolle P. 1989. The structural and functional organization of the murine HOX gene family resembles that of Drosophila homeotic genes. Embo J 8:1497-1505.

144

Dubrulle J, McGrew MJ, Pourquie O. 2001. FGF signaling controls somite boundary position and regulates segmentation clock control of spatiotemporal Hox gene activation. Cell 106:219-232. Dubrulle J, Pourquie O. 2004. Coupling segmentation to axis formation. Development 131:5783- 5793. Dudley AT, Ros MA, Tabin CJ. 2002. A re-examination of proximodistal patterning during vertebrate limb development. Nature 418:539-544. Duester G. 1996. Involvement of alcohol dehydrogenase, short-chain dehydrogenase/reductase, aldehyde dehydrogenase, and cytochrome P450 in the control of retinoid signaling by activation of retinoic acid synthesis. Biochemistry 35:12221-12227. Duester G. 2008. Retinoic acid synthesis and signaling during early organogenesis. Cell 134:921- 931. Duester G. 2009. Keeping an eye on retinoic acid signaling during eye development. Chem Biol Interact 178:178-181. Duester G, Deltour L, Ang HL. 1997. Evidence that class IV alcohol dehydrogenase may function in embryonic retinoic acid synthesis. Adv Exp Med Biol 414:357-364. Dupe V, Lumsden A. 2001. Hindbrain patterning involves graded responses to retinoic acid signalling. Development 128:2199-2208. Dupe V, Matt N, Garnier JM, Chambon P, Mark M, Ghyselinck NB. 2003. A newborn lethal defect due to inactivation of retinaldehyde dehydrogenase type 3 is prevented by maternal retinoic acid treatment. Proc Natl Acad Sci U S A 100:14036-14041. Fan X, Molotkov A, Manabe S, Donmoyer CM, Deltour L, Foglio MH, Cuenca AE, Blaner WS, Lipton SA, Duester G. 2003. Targeted disruption of Aldh1a1 (Raldh1) provides evidence for a complex mechanism of retinoic acid synthesis in the developing retina. Mol Cell Biol 23:4637-4648. Fernandes I, Bastien Y, Wai T, Nygard K, Lin R, Cormier O, Lee HS, Eng F, Bertos NR, Pelletier N, Mader S, Han VK, Yang XJ, White JH. 2003. Ligand-dependent nuclear receptor corepressor LCoR functions by histone deacetylase-dependent and -independent mechanisms. Mol Cell 11:139-150. Findlater GS, McDougall RD, Kaufman MH. 1993. Eyelid development, fusion and subsequent reopening in the mouse. J Anat 183 ( Pt 1):121-129. Fischer AJ, Wallman J, Mertz JR, Stell WK. 1999. Localization of retinoid binding proteins, retinoid receptors, and retinaldehyde dehydrogenase in the chick eye. J Neurocytol 28:597-609. Fromental-Ramain C, Warot X, Messadecq N, LeMeur M, Dolle P, Chambon P. 1996. Hoxa-13 and Hoxd-13 play a crucial role in the patterning of the limb autopod. Development 122:2997-3011. Fujii H, Sato T, Kaneko S, Gotoh O, Fujii-Kuriyama Y, Osawa K, Kato S, Hamada H. 1997. Metabolic inactivation of retinoic acid by a novel P450 differentially expressed in developing mouse embryos. Embo J 16:4163-4173. Galdones E, Lohnes D, Hales BF. 2006. Role of retinoic acid receptors alpha1 and gamma in the response of murine limbs to retinol in vitro. Birth Defects Res A Clin Mol Teratol 76:39- 45. Gallego O, Belyaeva OV, Porte S, Ruiz FX, Stetsenko AV, Shabrova EV, Kostereva NV, Farres J, Pares X, Kedishvili NY. 2006. Comparative functional analysis of human medium-chain dehydrogenases, short-chain dehydrogenases/reductases and aldo-keto reductases with retinoids. Biochem J 399:101-109.

145

Galloway JL, Delgado I, Ros MA, Tabin CJ. 2009. A reevaluation of X-irradiation-induced phocomelia and proximodistal limb patterning. Nature 460:400-404. Gavalas A. 2002. ArRAnging the hindbrain. Trends Neurosci 25:61-64. Gavalas A, Krumlauf R. 2000. Retinoid signalling and hindbrain patterning. Curr Opin Genet Dev 10:380-386. Ghyselinck NB, Bavik C, Sapin V, Mark M, Bonnier D, Hindelang C, Dierich A, Nilsson CB, Hakansson H, Sauvant P, Azais-Braesco V, Frasson M, Picaud S, Chambon P. 1999. Cellular retinol-binding protein I is essential for vitamin A homeostasis. Embo J 18:4903- 4914. Ghyselinck NB, Dupe V, Dierich A, Messaddeq N, Garnier JM, Rochette-Egly C, Chambon P, Mark M. 1997. Role of the retinoic acid receptor beta (RARbeta) during mouse development. Int J Dev Biol 41:425-447. Giguere V, Ong ES, Segui P, Evans RM. 1987. Identification of a receptor for the morphogen retinoic acid. Nature 330:624-629. Goff RA. 1962. The relation of development status of limb formation of x-radiation sensitivity in chick embryos. I. Groso study. J Exp Zool 151:177-200. Goldbeter A, Gonze D, Pourquie O. 2007. Sharp developmental thresholds defined through bistability by antagonistic gradients of retinoic acid and FGF signaling. Dev Dyn 236:1495-1508. Goldstein JT, Dobrzyn A, Clagett-Dame M, Pike JW, DeLuca HF. 2003. Isolation and characterization of unsaturated fatty acids as natural ligands for the retinoid-X receptor. Arch Biochem Biophys 420:185-193. Gorry P, Lufkin T, Dierich A, Rochette-Egly C, Decimo D, Dolle P, Mark M, Durand B, Chambon P. 1994. The cellular retinoic acid binding protein I is dispensable. Proc Natl Acad Sci U S A 91:9032-9036. Gould A, Itasaki N, Krumlauf R. 1998. Initiation of rhombomeric Hoxb4 expression requires induction by somites and a retinoid pathway. Neuron 21:39-51. Grondona JM, Kastner P, Gansmuller A, Decimo D, Chambon P, Mark M. 1996. Retinal dysplasia and degeneration in RARbeta2/RARgamma2 compound mutant mice. Development 122:2173-2188. Gronemeyer H, Gustafsson JA, Laudet V. 2004. Principles for modulation of the nuclear receptor superfamily. Nat Rev Drug Discov 3:950-964. Guha U, Gomes WA, Kobayashi T, Pestell RG, Kessler JA. 2002. In vivo evidence that BMP signaling is necessary for apoptosis in the mouse limb. Dev Biol 249:108-120. Harfe BD, Scherz PJ, Nissim S, Tian H, McMahon AP, Tabin CJ. 2004. Evidence for an expansion-based temporal Shh gradient in specifying vertebrate digit identities. Cell 118:517-528. Horlein AJ, Naar AM, Heinzel T, Torchia J, Gloss B, Kurokawa R, Ryan A, Kamei Y, Soderstrom M, Glass CK, et al. 1995. Ligand-independent repression by the thyroid hormone receptor mediated by a nuclear receptor co-repressor. Nature 377:397-404. Horton C, Maden M. 1995. Endogenous distribution of retinoids during normal development and teratogenesis in the mouse embryo. Dev Dyn 202:312-323. Huang J, Dattilo LK, Rajagopal R, Liu Y, Kaartinen V, Mishina Y, Deng CX, Umans L, Zwijsen A, Roberts AB, Beebe DC. 2009. FGF-regulated BMP signaling is required for eyelid closure and to specify conjunctival epithelial cell fate. Development 136:1741-1750. Iimura T, Pourquie O. 2007. Hox genes in time and space during vertebrate body formation. Dev Growth Differ 49:265-275.

146

Ikeya M, Takada S. 2001. Wnt-3a is required for somite specification along the anteroposterior axis of the mouse embryo and for regulation of cdx-1 expression. Mech Dev 103:27-33. Iulianella A, Beckett B, Petkovich M, Lohnes D. 1999. A molecular basis for retinoic acid- induced axial truncation. Dev Biol 205:33-48. Iulianella A, Lohnes D. 1997. Contribution of retinoic acid receptor gamma to retinoid-induced craniofacial and axial defects. Dev Dyn 209:92-104. Jegalian BG, De Robertis EM. 1992. Homeotic transformations in the mouse induced by overexpression of a human Hox3.3 transgene. Cell 71:901-910. Jones NC, Trainor PA. 2005. Role of morphogens in neural crest cell determination. J Neurobiol 64:388-404. Kao WW, Xia Y, Liu CY, Saika S. 2008. Signaling pathways in morphogenesis of cornea and eyelid. Ocul Surf 6:9-23. Kastner P, Mark M, Leid M, Gansmuller A, Chin W, Grondona JM, Decimo D, Krezel W, Dierich A, Chambon P. 1996. Abnormal spermatogenesis in RXR beta mutant mice. Genes Dev 10:80-92. Kaufman MH. 1992. The Atlas of Mouse Development. Academic Press. 512 p. Kawaguchi R, Yu J, Honda J, Hu J, Whitelegge J, Ping P, Wiita P, Bok D, Sun H. 2007. A membrane receptor for retinol binding protein mediates cellular uptake of vitamin A. Science 315:820-825. Kerszberg M, Wolpert L. 1998. Mechanisms for positional signalling by morphogen transport: a theoretical study. J Theor Biol 191:103-114. Kerszberg M, Wolpert L. 2007. Specifying positional information in the embryo: looking beyond morphogens. Cell 130:205-209. Kessel M. 1992. Respecification of vertebral identities by retinoic acid. Development 115:487- 501. Kessel M, Gruss P. 1991. Homeotic transformations of murine vertebrae and concomitant alteration of Hox codes induced by retinoic acid. Cell 67:89-104. Kitareewan S, Burka LT, Tomer KB, Parker CE, Deterding LJ, Stevens RD, Forman BM, Mais DE, Heyman RA, McMorris T, Weinberger C. 1996. Phytol metabolites are circulating dietary factors that activate the nuclear receptor RXR. Mol Biol Cell 7:1153-1166. Kmita M, Duboule D. 2003. Organizing axes in time and space; 25 years of colinear tinkering. Science 301:331-333. Kochhar DM. 1973. Limb development in mouse embryos. I. Analysis of teratogenic effects of retinoic acid. Teratology 7:289-298. Koide T, Downes M, Chandraratna RA, Blumberg B, Umesono K. 2001. Active repression of RAR signaling is required for head formation. Genes Dev 15:2111-2121. Kraus P, Fraidenraich D, Loomis CA. 2001. Some distal limb structures develop in mice lacking Sonic hedgehog signaling. Mech Dev 100:45-58. Krezel W, Dupe V, Mark M, Dierich A, Kastner P, Chambon P. 1996. RXR gamma null mice are apparently normal and compound RXR alpha +/-/RXR beta -/-/RXR gamma -/- mutant mice are viable. Proc Natl Acad Sci U S A 93:9010-9014. Kwon H, Imbalzano AN, Khavari PA, Kingston RE, Green MR. 1994. Nucleosome disruption and enhancement of activator binding by a human SW1/SNF complex. Nature 370:477- 481. Lammer EJ, Chen DT, Hoar RM, Agnish ND, Benke PJ, Braun JT, Curry CJ, Fernhoff PM, Grix AW, Jr., Lott IT, et al. 1985. Retinoic acid embryopathy. N Engl J Med 313:837-841.

147

Lampron C, Rochette-Egly C, Gorry P, Dolle P, Mark M, Lufkin T, LeMeur M, Chambon P. 1995. Mice deficient in cellular retinoic acid binding protein II (CRABPII) or in both CRABPI and CRABPII are essentially normal. Development 121:539-548. Lana-Elola E, Rice R, Grigoriadis AE, Rice DP. 2007. Cell fate specification during calvarial bone and suture development. Dev Biol 311:335-346. Lander AD. 2007. Morpheus unbound: reimagining the morphogen gradient. Cell 128:245-256. Laue K, Janicke M, Plaster N, Sonntag C, Hammerschmidt M. 2008. Restriction of retinoic acid activity by Cyp26b1 is required for proper timing and patterning of osteogenesis during zebrafish development. Development 135:3775-3787. Laufer E, Nelson CE, Johnson RL, Morgan BA, Tabin C. 1994. Sonic hedgehog and Fgf-4 act through a signaling cascade and feedback loop to integrate growth and patterning of the developing limb bud. Cell 79:993-1003. Le Douarin NM, Creuzet S, Couly G, Dupin E. 2004. Neural crest cell plasticity and its limits. Development 131:4637-4650. Leid M, Kastner P, Chambon P. 1992a. Multiplicity generates diversity in the retinoic acid signalling pathways. Trends Biochem Sci 17:427-433. Leid M, Kastner P, Lyons R, Nakshatri H, Saunders M, Zacharewski T, Chen JY, Staub A, Garnier JM, Mader S, et al. 1992b. Purification, cloning, and RXR identity of the HeLa cell factor with which RAR or TR heterodimerizes to bind target sequences efficiently. Cell 68:377-395. Levin AA, Sturzenbecker LJ, Kazmer S, Bosakowski T, Huselton C, Allenby G, Speck J, Kratzeisen C, Rosenberger M, Lovey A, et al. 1992. 9-cis retinoic acid stereoisomer binds and activates the nuclear receptor RXR alpha. Nature 355:359-361. Li C, Guo H, Xu X, Weinberg W, Deng CX. 2001. Fibroblast growth factor receptor 2 (Fgfr2) plays an important role in eyelid and skin formation and patterning. Dev Dyn 222:471- 483. Li C, Scott DA, Hatch E, Tian X, Mansour SL. 2007. Dusp6 (Mkp3) is a negative feedback regulator of FGF-stimulated ERK signaling during mouse development. Development 134:167-176. Li G, Gustafson-Brown C, Hanks SK, Nason K, Arbeit JM, Pogliano K, Wisdom RM, Johnson RS. 2003. c-Jun is essential for organization of the epidermal leading edge. Dev Cell 4:865-877. Li H, MacLean G, Cameron D, Clagett-Dame M, Petkovich M. 2009. Cyp26b1 expression in murine Sertoli cells is required to maintain male germ cells in an undifferentiated state during embryogenesis. PLoS One 4:e7501. Li H, Wagner E, McCaffery P, Smith D, Andreadis A, Drager UC. 2000. A retinoic acid synthesizing enzyme in ventral retina and telencephalon of the embryonic mouse. Mech Dev 95:283-289. Lin M, Zhang M, Abraham M, Smith SM, Napoli JL. 2003. Mouse retinal dehydrogenase 4 (RALDH4), molecular cloning, cellular expression, and activity in 9-cis-retinoic acid biosynthesis in intact cells. J Biol Chem 278:9856-9861. Liu W, Levi G, Shanske A, Frenz DA. 2008. Retinoic acid-induced inner ear teratogenesis caused by defective Fgf3/Fgf10-dependent Dlx5 signaling. Birth Defects Res B Dev Reprod Toxicol 83:134-144. Lohnes D, Kastner P, Dierich A, Mark M, LeMeur M, Chambon P. 1993. Function of retinoic acid receptor gamma in the mouse. Cell 73:643-658.

148

Lohnes D, Mark M, Mendelsohn C, Dolle P, Decimo D, LeMeur M, Dierich A, Gorry P, Chambon P. 1995. Developmental roles of the retinoic acid receptors. J Steroid Biochem Mol Biol 53:475-486. Lohnes D, Mark M, Mendelsohn C, Dolle P, Dierich A, Gorry P, Gansmuller A, Chambon P. 1994. Function of the retinoic acid receptors (RARs) during development (I). Craniofacial and skeletal abnormalities in RAR double mutants. Development 120:2723- 2748. Loudig O, Babichuk C, White J, Abu-Abed S, Mueller C, Petkovich M. 2000. Cytochrome P450RAI(CYP26) promoter: a distinct composite retinoic acid response element underlies the complex regulation of retinoic acid metabolism [In Process Citation]. Mol Endocrinol 14:1483-1497. Loudig O, Maclean GA, Dore NL, Luu L, Petkovich M. 2005. Transcriptional co-operativity between distant retinoic acid response elements in regulation of Cyp26A1 inducibility. Biochem J 392:241-248. Luetteke NC, Qiu TH, Peiffer RL, Oliver P, Smithies O, Lee DC. 1993. TGF alpha deficiency results in hair follicle and eye abnormalities in targeted and waved-1 mice. Cell 73:263- 278. Lufkin T, Lohnes D, Mark M, Dierich A, Gorry P, Gaub MP, LeMeur M, Chambon P. 1993. High postnatal lethality and testis degeneration in retinoic acid receptor alpha mutant mice. Proc Natl Acad Sci U S A 90:7225-7229. Luo J, Sucov HM, Bader JA, Evans RM, Giguere V. 1996. Compound mutants for retinoic acid receptor (RAR) beta and RAR alpha 1 reveal developmental functions for multiple RAR beta isoforms. Mech Dev 55:33-44. Mackem S, Lewandoski M. 2009. Limb development takes a measured step toward systems analysis. Sci Signal 2:pe33. MacLean G, Abu-Abed S, Dolle P, Tahayato A, Chambon P, Petkovich M. 2001. Cloning of a novel retinoic-acid metabolizing cytochrome P450, Cyp26B1, and comparative expression analysis with Cyp26A1 during early murine development. Mech Dev 107:195-201. Maclean G, Dolle P, Petkovich M. 2009. Genetic disruption of CYP26B1 severely affects development of neural crest derived head structures, but does not compromise hindbrain patterning. Dev Dyn 238:732-745. MacLean G, Li H, Metzger D, Chambon P, Petkovich M. 2007. Apoptotic extinction of germ cells in testes of Cyp26b1 knockout mice. Endocrinology 148:4560-4567. Maden M. 2002. Retinoid signalling in the development of the central nervous system. Nat Rev Neurosci 3:843-853. Mangelsdorf DJ, Ong ES, Dyck JA, Evans RM. 1990. Nuclear receptor that identifies a novel retinoic acid response pathway. Nature 345:224-229. Manley NR, Capecchi MR. 1997. Hox group 3 paralogous genes act synergistically in the formation of somitic and neural crest-derived structures. Dev Biol 192:274-288. Mann GB, Fowler KJ, Gabriel A, Nice EC, Williams RL, Dunn AR. 1993. Mice with a null mutation of the TGF alpha gene have abnormal skin architecture, wavy hair, and curly whiskers and often develop corneal inflammation. Cell 73:249-261. Mariani FV, Ahn CP, Martin GR. 2008. Genetic evidence that FGFs have an instructive role in limb proximal-distal patterning. Nature 453:401-405. Mariani FV, Martin GR. 2003. Deciphering skeletal patterning: clues from the limb. Nature 423:319-325.

149

Mark M, Ghyselinck NB, Chambon P. 2006. Function of retinoid nuclear receptors: lessons from genetic and pharmacological dissections of the retinoic acid signaling pathway during mouse embryogenesis. Annu Rev Pharmacol Toxicol 46:451-480. Mark M, Ghyselinck NB, Chambon P. 2009. Function of retinoic acid receptors during embryonic development. Nucl Recept Signal 7:e002. Marletaz F, Holland LZ, Laudet V, Schubert M. 2006. Retinoic acid signaling and the evolution of chordates. Int J Biol Sci 2:38-47. Marshall H, Morrison A, Studer M, Popperl H, Krumlauf R. 1996. Retinoids and Hox genes. Faseb J 10:969-978. Marshall H, Studer M, Popperl H, Aparicio S, Kuroiwa A, Brenner S, Krumlauf R. 1994. A conserved retinoic acid response element required for early expression of the homeobox gene Hoxb-1. Nature 370:567-571. Martin P, Parkhurst SM. 2004. Parallels between tissue repair and embryo morphogenesis. Development 131:3021-3034. Martras S, Alvarez R, Martinez SE, Torres D, Gallego O, Duester G, Farres J, de Lera AR, Pares X. 2004. The specificity of alcohol dehydrogenase with cis-retinoids. Activity with 11- cis-retinol and localization in retina. Eur J Biochem 271:1660-1670. Mascrez B, Ghyselinck NB, Chambon P, Mark M. 2009. A transcriptionally silent RXRalpha supports early embryonic morphogenesis and heart development. Proc Natl Acad Sci U S A 106:4272-4277. Mascrez B, Mark M, Dierich A, Ghyselinck NB, Kastner P, Chambon P. 1998. The RXRalpha ligand-dependent activation function 2 (AF-2) is important for mouse development. Development 125:4691-4707. Mascrez B, Mark M, Krezel W, Dupe V, LeMeur M, Ghyselinck NB, Chambon P. 2001. Differential contributions of AF-1 and AF-2 activities to the developmental functions of RXR alpha. Development 128:2049-2062. Mason KE. 1935. Foetal death, prolonged gestation, and difficult parturition in the rat as a result of vitamin A-deficiency. Am J Anat 57:303-349. Matt N, Dupe V, Garnier JM, Dennefeld C, Chambon P, Mark M, Ghyselinck NB. 2005. Retinoic acid-dependent eye morphogenesis is orchestrated by neural crest cells. Development 132:4789-4800. Mazzalupo S, Wong P, Martin P, Coulombe PA. 2003. Role for keratins 6 and 17 during wound closure in embryonic mouse skin. Dev Dyn 226:356-365. McCaffery P, Drager UC. 1995. Retinoic acid synthesizing enzymes in the embryonic and adult vertebrate. Adv Exp Med Biol 372:173-183. McCaffery P, Wagner E, O'Neil J, Petkovich M, Drager UC. 1999. Dorsal and ventral rentinoic territories defined by retinoic acid synthesis, break-down and nuclear receptor expression. Mech Dev 85:203-214. Meinhardt H. 1978. Space-dependent cell determination under the control of morphogen gradient. J Theor Biol 74:307-321. Meinhardt H, Gierer A. 1980. Generation and regeneration of sequence of structures during morphogenesis. J Theor Biol 85:429-450. Mendelsohn C, Lohnes D, Decimo D, Lufkin T, LeMeur M, Chambon P, Mark M. 1994. Function of the retinoic acid receptors (RARs) during development (II). Multiple abnormalities at various stages of organogenesis in RAR double mutants. Development 120:2749-2771.

150

Mercader N, Leonardo E, Azpiazu N, Serrano A, Morata G, Martinez C, Torres M. 1999. Conserved regulation of proximodistal limb axis development by Meis1/Hth. Nature 402:425-429. Mercader N, Leonardo E, Piedra ME, Martinez AC, Ros MA, Torres M. 2000. Opposing RA and FGF signals control proximodistal vertebrate limb development through regulation of Meis genes. Development 127:3961-3970. Mercader N, Selleri L, Criado LM, Pallares P, Parras C, Cleary ML, Torres M. 2009. Ectopic Meis1 expression in the mouse limb bud alters P-D patterning in a Pbx1-independent manner. Int J Dev Biol 53:1483-1494. Mic FA, Haselbeck RJ, Cuenca AE, Duester G. 2002. Novel retinoic acid generating activities in the neural tube and heart identified by conditional rescue of Raldh2 null mutant mice. Development 129:2271-2282. Mic FA, Molotkov A, Benbrook DM, Duester G. 2003. Retinoid activation of retinoic acid receptor but not retinoid X receptor is sufficient to rescue lethal defect in retinoic acid synthesis. Proc Natl Acad Sci U S A 100:7135-7140. Mic FA, Molotkov A, Fan X, Cuenca AE, Duester G. 2000. RALDH3, a retinaldehyde dehydrogenase that generates retinoic acid, is expressed in the ventral retina, otic vesicle and olfactory pit during mouse development. Mech Dev 97:227-230. Mic FA, Molotkov A, Molotkova N, Duester G. 2004. Raldh2 expression in optic vesicle generates a retinoic acid signal needed for invagination of retina during optic cup formation. Dev Dyn 231:270-277. Miettinen PJ, Berger JE, Meneses J, Phung Y, Pedersen RA, Werb Z, Derynck R. 1995. Epithelial immaturity and multiorgan failure in mice lacking epidermal growth factor receptor. Nature 376:337-341. Mine N, Iwamoto R, Mekada E. 2005. HB-EGF promotes epithelial cell migration in eyelid development. Development 132:4317-4326. Mollard R, Viville S, Ward SJ, Decimo D, Chambon P, Dolle P. 2000. Tissue-specific expression of retinoic acid receptor isoform transcripts in the mouse embryo. Mech Dev 94:223-232. Molotkov A, Fan X, Duester G. 2002. Excessive vitamin A toxicity in mice genetically deficient in either alcohol dehydrogenase Adh1 or Adh3. Eur J Biochem 269:2607-2612. Molotkov A, Molotkova N, Duester G. 2006. Retinoic acid guides eye morphogenetic movements via paracrine signaling but is unnecessary for retinal dorsoventral patterning. Development 133:1901-1910. Mori M, Ghyselinck NB, Chambon P, Mark M. 2001. Systematic immunolocalization of retinoid receptors in developing and adult mouse eyes. Invest Ophthalmol Vis Sci 42:1312-1318. Morriss-Kay GM, Ward SJ. 1999. Retinoids and mammalian development. Int Rev Cytol 188:73- 131. Moss JB, Xavier-Neto J, Shapiro MD, Nayeem SM, McCaffery P, Drager UC, Rosenthal N. 1998. Dynamic patterns of retinoic acid synthesis and response in the developing mammalian heart. Dev Biol 199:55-71. Motoyama J, Heng H, Crackower MA, Takabatake T, Takeshima K, Tsui LC, Hui C. 1998. Overlapping and non-overlapping Ptch2 expression with Shh during mouse embryogenesis. Mech Dev 78:81-84. Niederreither K, Abu-Abed S, Schuhbaur B, Petkovich M, Chambon P, Dolle P. 2002a. Genetic evidence that oxidative derivatives of retinoic acid are not involved in retinoid signaling during mouse development. Nat Genet 31:84-88.

151

Niederreither K, Dolle P. 2006. Molecular mediators of retinoic acid signaling during development. In: Taneja R, editor. Nuclear Receptors in Development. San Diego: Elsevier BV. pp 105-143. Niederreither K, Dolle P. 2008. Retinoic acid in development: towards an integrated view. Nat Rev Genet 9:541-553. Niederreither K, Fraulob V, Garnier JM, Chambon P, Dolle P. 2002b. Differential expression of retinoic acid-synthesizing (RALDH) enzymes during fetal development and organ differentiation in the mouse. Mech Dev 110:165-171. Niederreither K, McCaffery P, Drager UC, Chambon P, Dolle P. 1997. Restricted expression and retinoic acid-induced downregulation of the retinaldehyde dehydrogenase type 2 (RALDH-2) gene during mouse development. Mech Dev 62:67-78. Niederreither K, Subbarayan V, Dolle P, Chambon P. 1999. Embryonic retinoic acid synthesis is essential for early mouse post- implantation development. Nat Genet 21:444-448. Niederreither K, Vermot J, Fraulob V, Chambon P, Dolle P. 2002c. Retinaldehyde dehydrogenase 2 (RALDH2)- independent patterns of retinoic acid synthesis in the mouse embryo. Proc Natl Acad Sci U S A 99:16111-16116. Niederreither K, Vermot J, Le Roux I, Schuhbaur B, Chambon P, Dolle P. 2003. The regional pattern of retinoic acid synthesis by RALDH2 is essential for the development of posterior pharyngeal arches and the enteric nervous system. Development 130:2525-2534. Niederreither K, Vermot J, Messaddeq N, Schuhbaur B, Chambon P, Dolle P. 2001. Embryonic retinoic acid synthesis is essential for heart morphogenesis in the mouse. Development 128:1019-1031. Niederreither K, Vermot J, Schuhbaur B, Chambon P, Dolle P. 2000. Retinoic acid synthesis and hindbrain patterning in the mouse embryo. Development 127:75-85. Niederreither K, Vermot J, Schuhbaur B, Chambon P, Dolle P. 2002d. Embryonic retinoic acid synthesis is required for forelimb growth and anteroposterior patterning in the mouse. Development 129:3563-3574. Niswander L. 2002. Interplay between the molecular signals that control vertebrate limb development. Int J Dev Biol 46:877-881. Niswander L. 2003. Pattern formation: old models out on a limb. Nat Rev Genet 4:133-143. Niswander L, Jeffrey S, Martin GR, Tickle C. 1994. A positive feedback loop coordinates growth and patterning in the vertebrate limb. Nature 371:609-612. Noy N. 2000. Retinoid-binding proteins: mediators of retinoid action. Biochem J 348 Pt 3:481- 495. Ohuchi H, Tao H, Ohata K, Itoh N, Kato S, Noji S, Ono K. 2003. Fibroblast growth factor 10 is required for proper development of the mouse whiskers. Biochem Biophys Res Commun 302:562-567. Packer AI, Crotty DA, Elwell VA, Wolgemuth DJ. 1998. Expression of the murine Hoxa4 gene requires both autoregulation and a conserved retinoic acid response element. Development 125:1991-1998. Pennimpede T, Cameron D, Petkovich M. 2006. Regulation of murine embryonic patterning and morphogenesis by retinoic acid signaling. In: Taneja R, editor. Nuclear Receptors in Development. San Diego: Elsevier BV. pp 65-104. Pennimpede T, Cameron DA, Maclean GA, Petkovich M. 2010. Analysis of Cyp26b1/Rarg compound-null mice reveals two genetically separable effects of retinoic acid on limb outgrowth. Dev Biol 339:179-186.

152

Penniston KL, Tanumihardjo SA. 2006. The acute and chronic toxic effects of vitamin A. Am J Clin Nutr 83:191-201. Petkovich M, Brand NJ, Krust A, Chambon P. 1987. A human retinoic acid receptor which belongs to the family of nuclear receptors. Nature 330:444-450. Pourquie O. 2001. The vertebrate segmentation clock. J Anat 199:169-175. Pourquie O. 2003. Vertebrate somitogenesis: a novel paradigm for animal segmentation? Int J Dev Biol 47:597-603. Quadro L, Blaner WS, Salchow DJ, Vogel S, Piantedosi R, Gouras P, Freeman S, Cosma MP, Colantuoni V, Gottesman ME. 1999. Impaired retinal function and vitamin A availability in mice lacking retinol-binding protein. Embo J 18:4633-4644. Quadro L, Hamberger L, Gottesman ME, Colantuoni V, Ramakrishnan R, Blaner WS. 2004. Transplacental delivery of retinoid: the role of retinol-binding protein and lipoprotein retinyl ester. Am J Physiol Endocrinol Metab 286:E844-851. Riddle RD, Johnson RL, Laufer E, Tabin C. 1993. Sonic hedgehog mediates the polarizing activity of the ZPA. Cell 75:1401-1416. Ross SA, McCaffery PJ, Drager UC, De Luca LM. 2000. Retinoids in embryonal development. Physiol Rev 80:1021-1054. Rossant J, Tam PP. 2009. Blastocyst lineage formation, early embryonic asymmetries and axis patterning in the mouse. Development 136:701-713. Rossant J, Zirngibl R, Cado D, Shago M, Giguere V. 1991. Expression of a retinoic acid response element-hsplacZ transgene defines specific domains of transcriptional activity during mouse embryogenesis. Genes Dev 5:1333-1344. Ruberte E, Dolle P, Chambon P, Morriss-Kay G. 1991. Retinoic acid receptors and cellular retinoid binding proteins. II. Their differential pattern of transcription during early morphogenesis in mouse embryos. Development 111:45-60. Ruberte E, Dolle P, Krust A, Zelent A, Morriss-Kay G, Chambon P. 1990. Specific spatial and temporal distribution of retinoic acid receptor gamma transcripts during mouse embryogenesis. Development 108:213-222. Rubin L, Saunders JW, Jr. 1972. Ectodermal-mesodermal interactions in the growth of limb buds in the chick embryo: constancy and temporal limits of the ectodermal induction. Dev Biol 28:94-112. Sakai Y, Meno C, Fujii H, Nishino J, Shiratori H, Saijoh Y, Rossant J, Hamada H. 2001. The retinoic acid-inactivating enzyme CYP26 is essential for establishing an uneven distribution of retinoic acid along the anterio- posterior axis within the mouse embryo. Genes Dev 15:213-225. Sandell LL, Sanderson BW, Moiseyev G, Johnson T, Mushegian A, Young K, Rey JP, Ma JX, Staehling-Hampton K, Trainor PA. 2007. RDH10 is essential for synthesis of embryonic retinoic acid and is required for limb, craniofacial, and organ development. Genes Dev 21:1113-1124. Saunders JW, Jr. 1948. The proximo-distal sequence of origin of the parts of the chick wing and the role of the ectoderm. J Exp Zool 108:363-403. Schlessinger J. 2004. Common and distinct elements in cellular signaling via EGF and FGF receptors. Science 306:1506-1507. Schmidt K, Hughes C, Chudek JA, Goodyear SR, Aspden RM, Talbot R, Gundersen TE, Blomhoff R, Henderson C, Wolf CR, Tickle C. 2009. Cholesterol metabolism: the main pathway acting downstream of cytochrome P450 oxidoreductase in skeletal development of the limb. Mol Cell Biol 29:2716-2729.

153

Schubert M, Holland ND, Laudet V, Holland LZ. 2006. A retinoic acid-Hox hierarchy controls both anterior/posterior patterning and neuronal specification in the developing central nervous system of the cephalochordate amphioxus. Dev Biol 296:190-202. Sessler RJ, Noy N. 2005. A ligand-activated nuclear localization signal in cellular retinoic acid binding protein-II. Mol Cell 18:343-353. Shenefelt RE. 1972. Morphogenesis of malformations in hamsters caused by retinoic acid: relation to dose and stage at treatment. Teratology 5:103-118. Sirbu IO, Duester G. 2006. Retinoic-acid signalling in node ectoderm and posterior neural plate directs left-right patterning of somitic mesoderm. Nat Cell Biol 8:271-277. Sirbu IO, Gresh L, Barra J, Duester G. 2005. Shifting boundaries of retinoic acid activity control hindbrain segmental gene expression. Development 132:2611-2622. Slack JMW. 2006. Essential Developmental Biology. Oxford, UK: Blackwell Science Ltd. 365 p. Spoorendonk KM, Peterson-Maduro J, Renn J, Trowe T, Kranenbarg S, Winkler C, Schulte- Merker S. 2008. Retinoic acid and Cyp26b1 are critical regulators of osteogenesis in the axial skeleton. Development 135:3765-3774. Studer M, Gavalas A, Marshall H, Ariza-McNaughton L, Rijli FM, Chambon P, Krumlauf R. 1998. Genetic interactions between Hoxa1 and Hoxb1 reveal new roles in regulation of early hindbrain patterning. Development 125:1025-1036. Sucov HM, Izpisua-Belmonte JC, Ganan Y, Evans RM. 1995. Mouse embryos lacking RXR alpha are resistant to retinoic-acid-induced limb defects. Development 121:3997-4003. Sulik KK, Dehart DB, Rogers JM, Chernoff N. 1995. Teratogenicity of low doses of all-trans retinoic acid in presomite mouse embryos. Teratology 51:398-403. Summerbell D. 1974. A quantitative analysis of the effect of excision of the AER from the chick limb-bud. J Embryol Exp Morphol 32:651-660. Summerbell D. 1978. Normal and experimental variations in proportions of skeleton of chick embryo wing. Nature 274:472-473. Summerbell D, Lewis JH, Wolpert L. 1973. Positional information in chick limb morphogenesis. Nature 244:492-496. Sun X, Mariani FV, Martin GR. 2002. Functions of FGF signalling from the apical ectodermal ridge in limb development. Nature 418:501-508. Suzuki R, Shintani T, Sakuta H, Kato A, Ohkawara T, Osumi N, Noda M. 2000. Identification of RALDH-3, a novel retinaldehyde dehydrogenase, expressed in the ventral region of the retina. Mech Dev 98:37-50. Szabo-Rogers HL, Smithers LE, Yakob W, Liu KJ. 2009. New directions in craniofacial morphogenesis. Dev Biol (e-pub ahead of print). Tabin C, Wolpert L. 2007. Rethinking the proximodistal axis of the vertebrate limb in the molecular era. Genes Dev 21:1433-1442. Tahayato A, Dolle P, Petkovich M. 2003. Cyp26C1 encodes a novel retinoic acid-metabolizing enzyme expressed in the hindbrain, inner ear, first branchial arch and tooth buds during murine development. Gene Expr Patterns 3:449-454. Taimi M, Helvig C, Wisniewski J, Ramshaw H, White J, Amad M, Korczak B, Petkovich M. 2004. A novel human cytochrome P450, CYP26C1, involved in metabolism of 9-cis and all-trans isomers of retinoic acid. J Biol Chem 279:77-85. Tao H, Ono K, Kurose H, Noji S, Ohuchi H. 2006. Exogenous FGF10 can rescue an eye-open at birth phenotype of Fgf10-null mice by activating activin and TGFalpha-EGFR signaling. Dev Growth Differ 48:339-346.

154

Tao H, Shimizu M, Kusumoto R, Ono K, Noji S, Ohuchi H. 2005. A dual role of FGF10 in proliferation and coordinated migration of epithelial leading edge cells during mouse eyelid development. Development 132:3217-3230. Teelmann K. 1989. Retinoids: toxicology and teratogenicity to date. Pharmacol Ther 40:29-43. Thaller C, Eichele G. 1987. Identification and spatial distribution of retinoids in the developing chick limb bud. Nature 327:625-628. Tickle C. 1981. The number of polarizing region cells required to specify additional digits in the developing chick wing. Nature 289:295-298. Tickle C, Alberts B, Wolpert L, Lee J. 1982. Local application of retinoic acid to the limb bond mimics the action of the polarizing region. Nature 296:564-566. Tiecke E, Bangs F, Blaschke R, Farrell ER, Rappold G, Tickle C. 2006. Expression of the short stature homeobox gene Shox is restricted by proximal and distal signals in chick limb buds and affects the length of skeletal elements. Dev Biol 298:585-596. Truett GE, Heeger P, Mynatt RL, Truett AA, Walker JA, Warman ML. 2000. Preparation of PCR-quality mouse genomic DNA with hot sodium hydroxide and tris (HotSHOT). Biotechniques 29:52, 54. Uehara M, Yashiro K, Mamiya S, Nishino J, Chambon P, Dolle P, Sakai Y. 2007. CYP26A1 and CYP26C1 cooperatively regulate anterior-posterior patterning of the developing brain and the production of migratory cranial neural crest cells in the mouse. Dev Biol 302:399-411. Uehara M, Yashiro K, Takaoka K, Yamamoto M, Hamada H. 2009. Removal of maternal retinoic acid by embryonic CYP26 is required for correct Nodal expression during early embryonic patterning. Genes Dev 23:1689-1698. Ulven SM, Gundersen TE, Weedon MS, Landaas VO, Sakhi AK, Fromm SH, Geronimo BA, Moskaug JO, Blomhoff R. 2000. Identification of endogenous retinoids, enzymes, binding proteins, and receptors during early postimplantation development in mouse: important role of retinal dehydrogenase type 2 in synthesis of all-trans-retinoic acid. Dev Biol 220:379-391. Vassalli A, Matzuk MM, Gardner HA, Lee KF, Jaenisch R. 1994. Activin/inhibin beta B subunit gene disruption leads to defects in eyelid development and female reproduction. Genes Dev 8:414-427. Vermot J, Llamas JG, Fraulob V, Niederreither K, Chambon P, Dolle P. 2005. Retinoic acid controls the bilateral symmetry of somite formation in the mouse embryo. Science 308:563-566. Voss AK, Collin C, Dixon MP, Thomas T. 2009. Moz and retinoic acid coordinately regulate H3K9 acetylation, Hox gene expression, and segment identity. Dev Cell 17:674-686. Wagner E, McCaffery P, Drager UC. 2000. Retinoic acid in the formation of the dorsoventral retina and its central projections. Dev Biol 222:460-470. Wahl MB, Deng C, Lewandoski M, Pourquie O. 2007. FGF signaling acts upstream of the NOTCH and WNT signaling pathways to control segmentation clock oscillations in mouse somitogenesis. Development 134:4033-4041. Wang Y, Nathans J. 2007. Tissue/planar cell polarity in vertebrates: new insights and new questions. Development 134:647-658. Wankell M, Munz B, Hubner G, Hans W, Wolf E, Goppelt A, Werner S. 2001. Impaired wound healing in transgenic mice overexpressing the activin antagonist follistatin in the epidermis. Embo J 20:5361-5372. Wellik DM. 2009. Hox genes and vertebrate axial pattern. Curr Top Dev Biol 88:257-278.

155

Wendling O, Ghyselinck NB, Chambon P, Mark M. 2001. Roles of retinoic acid receptors in early embryonic morphogenesis and hindbrain patterning. Development 128:2031-2038. Weston AD, Blumberg B, Underhill TM. 2003. Active repression by unliganded retinoid receptors in development: less is sometimes more. J Cell Biol 161:223-228. Weston AD, Chandraratna RA, Torchia J, Underhill TM. 2002. Requirement for RAR-mediated gene repression in skeletal progenitor differentiation. J Cell Biol 158:39-51. Weston AD, Ozolins TR, Brown NA. 2006. Thoracic skeletal defects and cardiac malformations: a common epigenetic link? Birth Defects Res C Embryo Today 78:354-370. White JA, Ramshaw H, Taimi M, Stangle W, Zhang A, Everingham S, Creighton S, Tam SP, Jones G, Petkovich M. 2000. Identification of the human cytochrome P450, P450RAI-2, which is predominantly expressed in the adult cerebellum and is responsible for all-trans- retinoic acid metabolism. Proc Natl Acad Sci U S A 97:6403-6408. Williams JA, Kondo N, Okabe T, Takeshita N, Pilchak DM, Koyama E, Ochiai T, Jensen D, Chu ML, Kane MA, Napoli JL, Enomoto-Iwamoto M, Ghyselinck N, Chambon P, Pacifici M, Iwamoto M. 2009. Retinoic acid receptors are required for skeletal growth, matrix homeostasis and growth plate function in postnatal mouse. Dev Biol 328:315-327. Wilson JG, Roth CB, Warkany J. 1953. An analysis of the syndrome of malformations induced by maternal vitamin A deficiency. Effects of restoration of vitamin A at various times during gestation. Am J Anat 92:189-217. Wilson JG, Warkany J. 1950. Congenital anomalies of heart and great vessels in offspring of vitamin A-deficient rats. Am J Dis Child 79:963. Wilson V, Olivera-Martinez I, Storey KG. 2009. Stem cells, signals and vertebrate body axis extension. Development 136:1591-1604. Wolbach SB, Howe PR. 1978. Nutrition Classics. The Journal of Experimental Medicine 42: 753- 77, 1925. Tissue changes following deprivation of fat-soluble A vitamin. S. Burt Wolbach and Percy R. Howe. Nutr Rev 36:16-19. Wolpert L. 1969. Positional information and the spatial pattern of cellular differentiation. J Theor Biol 25:1-47. Wolpert L. 1994. Positional information and pattern formation in development. Dev Genet 15:485-490. Wolpert L, Tickle C, Sampford M. 1979. The effect of cell killing by x-irradiation on pattern formation in the chick limb. J Embryol Exp Morphol 50:175-193. Xia Y, Kao WW. 2004. The signaling pathways in tissue morphogenesis: a lesson from mice with eye-open at birth phenotype. Biochem Pharmacol 68:997-1001. Xiao JH, Durand B, Chambon P, Voorhees JJ. 1995. Endogenous retinoic acid receptor (RAR)- retinoid X receptor (RXR) heterodimers are the major functional forms regulating retinoid-responsive elements in adult human keratinocytes. Binding of ligands to RAR only is sufficient for RAR-RXR heterodimers to confer ligand-dependent activation of hRAR beta 2/RARE (DR5). J Biol Chem 270:3001-3011. Yashiro K, Zhao X, Uehara M, Yamashita K, Nishijima M, Nishino J, Saijoh Y, Sakai Y, Hamada H. 2004. Regulation of retinoic acid distribution is required for proximodistal patterning and outgrowth of the developing mouse limb. Dev Cell 6:411-422. Ybot-Gonzalez P, Cogram P, Gerrelli D, Copp AJ. 2002. Sonic hedgehog and the molecular regulation of mouse neural tube closure. Development 129:2507-2517. Yu L, Liu H, Yan M, Yang J, Long F, Muneoka K, Chen Y. 2007. Shox2 is required for chondrocyte proliferation and maturation in proximal limb skeleton. Dev Biol 306:549- 559.

156

Yu VC, Delsert C, Andersen B, Holloway JM, Devary OV, Naar AM, Kim SY, Boutin JM, Glass CK, Rosenfeld MG. 1991. RXR beta: a coregulator that enhances binding of retinoic acid, thyroid hormone, and vitamin D receptors to their cognate response elements. Cell 67:1251-1266. Yu Z, Bhandari A, Mannik J, Pham T, Xu X, Andersen B. 2008. Grainyhead-like factor Get1/Grhl3 regulates formation of the epidermal leading edge during eyelid closure. Dev Biol 319:56-67. Zeller R, Lopez-Rios J, Zuniga A. 2009. Vertebrate limb bud development: moving towards integrative analysis of organogenesis. Nat Rev Genet 10:845-858. Zenz R, Scheuch H, Martin P, Frank C, Eferl R, Kenner L, Sibilia M, Wagner EF. 2003. c-Jun regulates eyelid closure and skin tumor development through EGFR signaling. Dev Cell 4:879-889. Zhang F, Nagy Kovacs E, Featherstone MS. 2000. Murine hoxd4 expression in the CNS requires multiple elements including a retinoic acid response element. Mech Dev 96:79-89. Zhang L, Wang W, Hayashi Y, Jester JV, Birk DE, Gao M, Liu CY, Kao WW, Karin M, Xia Y. 2003. A role for MEK kinase 1 in TGF-beta/activin-induced epithelium movement and embryonic eyelid closure. Embo J 22:4443-4454. Zhang XK, Hoffmann B, Tran PB, Graupner G, Pfahl M. 1992. Retinoid X receptor is an auxiliary protein for thyroid hormone and retinoic acid receptors. Nature 355:441-446. Zhao D, McCaffery P, Ivins KJ, Neve RL, Hogan P, Chin WW, Drager UC. 1996. Molecular identification of a major retinoic-acid-synthesizing enzyme, a retinaldehyde-specific dehydrogenase. Eur J Biochem 240:15-22. Zhao X, Sirbu IO, Mic FA, Molotkova N, Molotkov A, Kumar S, Duester G. 2009. Retinoic acid promotes limb induction through effects on body axis extension but is unnecessary for limb patterning. Curr Biol 19:1050-1057. Zucker RM, Hunter ES, 3rd, Rogers JM. 1999. Apoptosis and morphology in mouse embryos by confocal laser scanning microscopy. Methods 18:473-480.

157