Wesleyan University. By Frank Joseph Tulenko. Faculty Advisor: Dr. Ann C. Burke

Transcription

1 Wesleyan University INVESTIGATING THE LATERAL SOMITIC FRONTIER: INSIGHTS INTO THE EVOLUTION AND DEVELOPMENT OF THE VERTEBRATE POST-CRANIAL MUSCULOSKELETAL SYSTEM By Frank Joseph Tulenko Faculty Advisor: Dr. Ann C. Burke A Dissertation submitted to the Faculty of Wesleyan University in partial fulfillment of the requirements for the Degree of Doctor of Philosophy Middletown, Connecticut July, 2013

2 Frank J. Tulenko, 2013 ii

3 ABSTRACT The post-cranial musculoskeletal system of vertebrates derives from somitic and lateral plate mesoderm. The somites form axial structures such as the vertebrae and ribs, as well as all skeletal muscle fibers. In contrast, paired fins and limbs form as outgrowths of the somatopleure, a composite tissue consisting of somatic lateral plate and ectoderm. Recently, these mesodermal lineages have been used to define two embryonic domains for cell differentiation. In the primaxial domain, somitic cells differentiate in a somitic environment, whereas in the abaxial domain somitic and lateral plate cells differentiate in a lateral plate environment. The interface between these domains is termed the lateral somitic frontier. Notably, studies of gene regulation, mutant phenotypes, and heterotopic challenge experiments fuel the hypothesis that the primaxial and abaxial domains are developmental modules. Importantly, any degree of patterning independence between these modules has the potential to facilitate independent morphological evolution. Herein, I use the conceptual framework of the lateral somitic frontier to explore the relationship between the development and evolution of the vertebrate musculoskeletal system. In chapter two of this thesis, I address a long-standing question in comparative anatomy, the origin of paired fins. I present new experimental data on the distribution of somitic and lateral plate mesoderm in a representative agnathan (lamprey), chondrichthyan (catshark), and amphibian (axolotl), and compare these data with published observations from amniotes (mouse and chick). In lamprey, the somatopleure is present early in development, but is iii

4 eliminated during ventral myotome growth, as cells of the lateral plate are displaced to the inner body wall only. Thus, the muscular body wall of lamprey is entirely primaxial. In contrast, cells of the somatic lateral plate are retained in the lateral body wall of catshark, axolotl, and amniotes at fin and inter-fin levels. I hypothesize that retention of somatic lateral plate in the lateral body wall arose along the lineage leading to jawed vertebrates, and the evolution of a persistent somatopleure was a critical early step in the origin of the abaxial module and paired fins. In limbed vertebrates, paired appendages are anchored to the axial system by distinct girdle bones, which thus represent critical sites of anatomical transition. In chapter three of this thesis, I use an approach that integrates surgical transplantation with highly detailed three-dimensional reconstructions of grafted tissues. These reconstructions reveal that the majority of the avian scapula derives from lateral plate mesoderm and is abaxial. Somitic mesoderm contributes only to the distal tip of the scapular blade, which appears to be displaced by more rapid growth of the abaxial domain. Importantly, these data provide a view of mesodermal segregation in the avian shoulder that is radically different from previously published observations, and call for a reassessment of hypotheses on the development and evolution of the vertebrate scapula. In chapter four of this thesis I provide a detailed method for isotopic transplantation of lateral plate mesoderm in axolotls, complete with images of donors and hosts at each surgery step. Additionally, I use data from these surgeries to identify the position of the lateral somitic frontier in the pectoral and inter-limb iv

5 region. These fate maps reveal that muscles of the primaxial domain do not grow beyond at the dorsal margin of the body cavity, and all hypaxial body wall muscles ventral to this point are abaxial. The abaxial nature of these muscles strongly suggests that they derive from migratory somitic cells, similar to the muscles of the limbs and shoulder. Future work examining the distribution of lateral plate mesoderm at earlier stages of body wall formation, in combination with the use of molecular markers for migratory myoblasts, will test this hypothesis. v

6 ACKNOWLEDGEMENTS To Ann C. Burke, my advisor and mentor, for sharing her passion for evolutionary morphology, creative thinking, and insights into science. Her patience and unwavering confidence in me were a great source of support over the years, and I consider myself lucky; To my committee members, Stephen H. Devoto, Laura B. Grabel, and Barry Chernoff, for their invaluable guidance and technical expertise, and for making my time at Wesleyan a wonderful experience; To Rebecca M. Shearman, for helping me get my bearings in the beginning, and for being a great friend and colleague; To Rosemarie A. Doris and Patricia Thompson, for putting up with all of my questions and keeping me pointed in the right direction; To Ethan L. MacKenzie for his dedication, skill, and enthusiasm; To all members of the Burke lab, past and present, for their help over the years; To David W. McCauley, for sharing his lab, expertise, and home during lamprey season; To Shigeru Kuratani, and the members of his lab including Fumiaki Sugahara, Masaki Takechi, Hiroshi Nagashima, Kenya Ota, and Rie Kusakabe for their help, advice, and friendship; To Jeff Gilarde, for his technical expertise in microscopy; To Blanche Meslin, Marjorie Fitzgibbons, and Susan Lastrina, for their support; To Peri Mason, for our conversations about science and life in general; To my fellow Wesleyan graduate students, past and present, I couldn t have asked for a better group; To the NSF EAPSI program, Sigma Xi, and the Wesleyan University Department of Biology for funding this work; To my parents, Verona and Joseph, and my brother Andrew, for constantly supporting my education. vi

7 TABLE OF CONTENTS ABSTRACT iii ACKNOWLEDGEMENTS vi TABLE OF CONTENTS vii LIST OF FIGURES viii LIST OF TABLES xi CHAPTER 1: CHAPTER 2: Introduction 1 A. The persistence of body wall somatopleure in gnathostomes and a new perspective on the origin of vertebrate paired fins 14 B. The origin of the vertebrate appendicular system: genetic redeployment, tissue redistribution, and the birth of the abaxial domains 59 CHAPTER 3: CHAPTER 4: 3D Reconstructions of quail-chick chimeras provide a new fate map of the avian scapula 94 A detailed method for isotopic transplantation of lateral plate mesoderm in axolotl and insights into the position of the lateral somitic frontier _138 CONCLUDING REMARKS 169 Appendix A: Heterotopic transplantation experiments in axolotls provide preliminary data that trunk somites are competent to form abaxial appendicular musculature 173 REFERENCES 185 vii

8 LIST OF FIGURES Chapter 1: Figure 1. Illustration of the primaxial and abaxial domains of a developing mouse embryo 10 Figure 2. Simplified vertebrate phylogeny 12 Chapter 2A: Figure 1. Distribution of lateral plate mesoderm in Stage 23/24 lamprey embryos 30 Figure 2. Plastic sections and DP312 labeling of Petromyzon embryos and larvae 32 Figure 3. DiI labeling of somites in Petromyzon 34 Figure 4. DiI labeling of LPM in Petromyzon 36 Figure 5. DP312 labeling in Scyliorhinus 38 Figure 6. GFP to WT Isotopic transplants of LPM/Pronephros/Ect in Ambystoma 40 Figure 7. Evolution of the persistent somatopleure 42 Figure S1. Plastic sections of Lethenteron embryos and larvae 44 Figure S2. Myotome formation in Petromyzon 46 Figure S3. LjHandA expression in Lethenteron 48 Figure S4. DP312 labeling in Petromyzon 50 Figure S5. DiI labeling of LPM in Petromyzon 52 Figure S6. GFP to WT isotopic transplants of LPM/Pronephros/Ect in Axolotl one day post-surgery 54 Chapter 2B: Figure 1. Classical hypotheses for the origin of paired fins in verterbates 88 Figure 2. Illustrations of cross sections of Lampetra fluviatilis from Goette (1890) 90 Figure. 3. Illustration of Scyliorhinus canicula from Balfour (1878) 92 viii

9 Chapter 3: Figure 1. Illustrations of the primaxial and abaxial domains of a developing chick embryo and surgical manipulations performed 119 Figure 2. Cross sections and three dimensional reconstruction of a SP chimera at Day Figure 3.Three dimensional reconstructions of chimeras at Day Figure 4. Surface models of the vertebrae, ribs, and scapula of chimeras at Day Figure 5. Surface models of the vertebrae, ribs, scapula, and shoulder muscles of chimeras at Day Figure 6. Sox9 or QcPN labeling of alternate sections of a Day 5 chimera 129 Figure 7. Sequence of cross sections through the scapula of chimeras progressing from anterior to posterior 131 Figure 8. QcPN labeled sections of an SP chimera counterstained with DAPI _133 Figure 9. Illustrations of fate maps of the avian scapula 135 Chapter 4: Figure 1. Transgenic Axolotl ubiquitously expressing Green Fluorescent Protein (GFP) 153 Figure 2. Isotopic transplantation of Lateral Plate Mesoderm, Ectoderm, and Pronephros from GFP Donor to Wild Type host 155 Figure 3. Distribution of GFP-Positive Donor Cells in Whole Mount Axolotl Chimeras 157 Figure 4. Evaluation of graft incorporation approximately one-day postsurgery 159 Figure 5. Distribution of graft in the pectoral region of axolotl chimeras following isotopic transplantation of LPM/Pronephros/Ect from GFP donor to WT host 161 Figure 6. Distribution of graft in the pectoral region and trunk of an axolotl chimera following isotopic transplantation of LPM/Pronephros/Ect from GFP donor to WT host 163 ix

10 Figure 7. Cross sections of the spinal nerves of a Stage 57 WT axolotl and a chimera following isotopic transplantation of LPM/Pronephros/Ect from GFP donor to WT host 167 Appendix A: Figure 1. Axolotl chimeras following isotopic transplantation of somitic mesoderm from GFP donors to WT hosts 178 Figure 2. Axolotl chimeras following heterotopic transplantation of somitic mesoderm from GFP donors to WT hosts 180 x

11 LIST OF TABLES Chapter 2: Table 1. Total number of DiI injections in Petromyzon embryos 56 Table 2. Distribution of DiI in Sectioned Petromyzon embryos 57 Table 3. Surgeries in Axolotl: number of LPM/Ectoderm and LPM/Ect/Pronephros isotopictransplants 58 Chapter 3: Table 1. Number of avian segmental plate and lateral plate transplant surgeries 137 Appendix A: Table 1. Axial levels that contribute to limb musculature in Axolotl 182 Table 2. Total number of Anterior-to-Posterior heterotopic surgeries in Axolotls 183 Table 3. Total number of Posterior-to-Anterior heterotopic surgeries in Axolotls 184 xi

12 CHAPTER 1: INTRODUCTION A near endless amount of morphological diversity has evolved since the birth of the vertebrate lineage over 500 million years ago. This natural history is rich with anatomical invention, inspiring the imagination and scientific curiosity of comparative anatomists for centuries. Changes in the musculoskeletal system provide some of the most striking examples of morphological evolution in vertebrates, including the origin of jaws, paired fins, digits, and the turtle shell, to name just a few. Since adult form is generated during ontogeny, it is changes in ontogenetic processes that underlie these innovations. This simple idea provides the motivation for this thesis, which in the broadest sense explores the relationship between the development and evolution of the vertebrate musculoskeletal system. Much of what is known about vertebrate development draws from studies of a few select model organisms, including chick, mouse, and more recently, zebrafish. Structures of the post-cranial musculoskeletal system derive from two populations of embryonic mesoderm, the paraxial mesoderm and lateral plate mesoderm (Fig. 1A). Early in development, the paraxial mesoderm flanks the neural tube and notochord laterally, and segments into epithelial spheres of cells, the somites, which establish the initial metameric body-plan of vertebrates (Christ and Ordahl, 1995). Shortly after forming, each somite re-organizes into compartments that give rise to distinct cell lineages (Christ et al., 2007). The medio-ventral face of each somite deepithelializes and forms the mesenchymal sclerotome, which then re-segments into 1

13 anterior and posterior streams that contribute to adjacent vertebrae and ribs (Huang et al., 2000a; Evans, 2003; Christ et al., 2004; Christ et al., 2007). The dorsolateral face of each somite is retained as the dermomyotome, which forms the dermis of the dorsal body wall, and all skeletal muscle precursors (Christ and Ordahl, 1995; Christ and Brand-Saberi, 2002; Scaal and Christ, 2004; but see Theis et al., 2010 for skeletal muscle that derives from occipital lateral plate ). The lateral plate mesoderm flanks the paraxial mesoderm laterally, and initiates as a single sheet of cells before splitting into a dorsal somatic and ventral splanchnic layer (Funayama et al., 1999). The somatic layer and overlying ectoderm together comprise the somatopleure, which forms the limb buds, sternum, connective tissue of the ventro-lateral body wall, and parietal coelomic lining. The splanchnic layer and endoderm comprise the splanchnopleure, which contributes to the heart, the walls of the gastrointestinal tract, and the visceral mesentery. The space between the somatic and splanchnic layers is retained in adults as the coelomic cavity. Although many studies have explored the molecular underpinnings of somite compartmentalization in amniotes (reviewed in Stockdale et al., 2000; Scaal and Christ, 2004; Yusuf and Brand-Saberi, 2006, Christ, 2007), less work has been done to characterize the development of the lateral plate, or how mesodermal derivatives are globally organized into a properly functioning musculoskeletal system (Burke and Nowicki, 2003; Winslow et al., 2007). The amniote body is regionalized along the antero-posterior axis into cervical, thoracic, lumbar, sacral and caudal portions based largely on skeletal morphology and limb position. These distinct anatomical regions are established during development 2

14 by the differential dorso-ventral patterning of somitic and lateral plate mesoderm at different axial levels (Winslow et al., 2007). One dramatic example of this is provided at limb-forming levels, where migratory muscles precursors delaminate from the dermomyotome and enter the limb bud mesenchyme to form appendicular musculature (Buckingham et al., 2003; Vasyutina and Birchmeier, 2006). Nowicki and Burke (2003) and Durland et al. (2008) used detailed lineage analyses to map the interface of somitic and lateral plate populations in chick and mouse, respectively, and described two embryonic domains based on mesodermal context for cell differentiation (Fig. 1b,c). The primaxial domain consists exclusively of somitic cells and includes the vertebrae, ribs, peri-verterbral and intercostal muscles, and dorsal dermis. In contrast, the abaxial domain derives from lateral plate cells as well as migratory somitic cells that mix with and differentiate in the context of lateral plate connective tissue. This domain includes the sternum, the bones and muscles of the limbs, the muscles of the abdominal body wall, and the ventro-lateral dermis of the inter-limb flank. The lateral somitic frontier (LSF) is the interface between primaxial and abaxial domains. Early in development, the LSF is linear reflecting the initial arrangement of somites and lateral plate mesoderm (the incipient frontier of Burke and Nowicki, 2003), but then shifts dramatically in older embryos due to regional differences in cell behavior and morphogenesis the development of the avian pectoral region, presented in chapter three of this thesis, provides an example. Although the topography of the frontier appears generally smooth in cross-section, discontinuities occur at structures that bridge the axial and appendicular systems, such 3

15 as spinal nerves contributing to the brachial plexus, and muscles and bones that bridge the axial and appendicular systems (Durland et al., 2008). These discontinuities likely reflect the early interaction of tissues across the frontier, prior to differential growth of primaxial and abaxial morphologies. Burke and Nowicki (2003) hypothesize that the primaxial and abaxial domains are developmental modules, i.e., exhibit some degree of independent patterning. One line of evidence supporting this hypothesis draws from studies of the role of lateral plate connective tissue in patterning somitic muscle precursors that cross the frontier. For example, experiments using classical embryonic manipulations have shown that in the absence of muscles cells, lateral plate-derived connective tissue forms a pre-pattern of limb muscle organization (Jacob and Christ, 1980; Kardon et al., 2003). Moreover, somites from any axial level can provide myofibers to a normal complement of limb muscles (Chevallier et al., 1977; Christ et al., 1977; Gumpel-Pinot, 1984; Hayashi and Ozawa, 1995). Functional studies further support the role of lateral plate-derived connective tissue in organizing limb muscles, as the conditional loss of Tcf4 (a component of the Wnt signaling pathway) or the transcription factor Tbx5 in limb-bud lateral plate causes muscle patterning defects (Kardon et al., 2003; Hasson et al., 2010; Hasson, 2011; Mathew et al., 2011). Together, these data indicate somitic migratory cells are non-autonomously patterned into abaxial derivatives by extrinsic cues from their new environment. Additional evidence for the developmental modularity of the primaxial and abaxial domains draws from studies of gene regulation, and the use of distinct cis- 4

16 regulatory elements to drive expression of the same patterning genes independently in different embryonic tissues (reviewed in Burke and Nowicki, 2003). The Hox family of transcription factors provides an important example of this. Hox genes are arranged in highly ordered genomic clusters (Kmita and Duboule, 2003; Duboule, 2007), and have been well-studied for their role in specifying regional morphology in the trunk and limbs of vertebrates (Burke, 2000; Wellik and Capecchi, 2003; McIntyre et al., 2007; Wellik, 2007; Zakany and Duboule, 2007; Mallo et al., 2010). Remarkably, the 3 to 5 position of Hox genes within a cluster mirrors their spatial and temporal expression domains in the embryo, a phenomenon termed co-linearity (reviewed in Tshcopp and Duboule, 2011). Detailed analyses of HoxD paralogues have shown that cis-elements within the cluster direct trunk expression, whereas elements both upstream and downstream of the cluster direct limb expression (Kmita and Duboule, 2003; Spitz et al., 2003; Deschamps, 2007; Tschopp and Duboule, 2011). Thus, discreet genomic sequences regulate the same HoxD paralogues independently during the formation of primaxial (axial) and abaxial (appendicular) derivatives. Data from molecular and tissue manipulation experiments also support this model. For example, the exogenous application of FGF signaling ligands to the flank of chick embryos alters Hox expression in the lateral plate but leaves somitic expression unaffected (Cohn et al., 1997). Morphologies also appear to correlate with independent Hox regulation. When paraxial mesoderm is heterotopically transplanted from one axial level to another, it maintains its original Hox expression profile 5

17 (Nowicki and Burke, 2003) and gives rise to primaxial structures characteristic of its original location (Kieny et al., 1972; Murakami and Nakamura, 1991; Nowicki and Burke, 2000). In contrast, grafted cells that migrate into the lateral plate do not maintain the Hox identity of their parent somites. Instead, these cells appear to conform to local lateral plate Hox expression and form abaxial muscles typical of their new environment (Nowicki and Burke, 2003). Together, the examples described above support the developmental modularity of the primaxial and abaxial domains (for complete reviews see Burke and Nowicki, 2003 and Shearman and Burke, 2009). Importantly, any degree of patterning independence between these domains has the potential to facilitate independent morphological evolution. Although the lateral somitic frontier has not been mapped in anamniotes, several studies suggest that cell behavior at the frontier varies during fin development in bony and cartilaginous fishes (Fig. 2). In actinopterygians, the pectoral and pelvic fin muscles derive from migratory precursor cells that dissociate from the dermomyotome and enter the fin bud mesenchyme where they differentiate (Neyt et al., 2000; Cole et al., 2005; Cole et al., 2011). These muscles, like the limb muscles of amniotes, are abaxial. Interestingly, this also appears to be the mechanism for the formation of certain hypaxial body wall muscles in the teleost pearlfish (Windner et al., 2011). The fin muscles of chondrichthyans, on the other hand, differentiate from epithelial projections of the dermomyotome, which extend into fin buds as cohesive sheets (Goodrich, 1930; Neyt et al., 2000; 6

18 Cole and Currie, 2007). These muscles are likely primaxial; however, any potential role of the lateral plate in their formation is unknown. In chapter two of this thesis I use the conceptual framework of the lateral somitic frontier to provide a new perspective on a long standing issue in comparative anatomy, embryology, and paleontology: the origin of the vertebrate appendicular system. I present new experimental data on the distribution of somitic and lateral plate mesoderm in a representative agnathan (lamprey), chondrichthyan (catshark), and amphibian (axolotl), and use these data to anchor comparisons with welldescribed amniote model systems (mouse and chick). In lamprey, the somatopleure is present early in development, but is eliminated during muscularization of the body wall, as cells of the lateral plate are segregated to the coelomic linings. Thus, the muscular body wall of lamprey is entirely primaxial. In contrast, cells of the somatic lateral plate are retained in the lateral body wall of catshark, axolotl, and amniotes at fin and inter-fin levels, a developmental feature I term somatopleure persistence. I propose that somatopleure persistence is a shared derived feature of jawed vertebrates (or possibly the group jawed vertebrates plus osteostracans, see Sanson, 2009), and was a key first step in the origin of the abaxial module and paired fins. In this chapter I also summarize published observations supporting the redeployment of gene networks during the early evolution of the appendicular system. These data are consistent with the idea that morphological innovation is facilitated by the recycling of pre-existing genetic circuits rather than the de novo assembly of new networks from scratch. 7

19 Chapter three of this thesis presents work done in collaboration with R. M. Shearman, a post-doc in the laboratory of A.C. Burke from The pectoral girdle bridges the somite-derived axial column with the lateral plate-derived appendicular system, and is thus a site of anatomical transition in the gnathostome body plan. Previous fate maps of the avian scapula indicate its dual embryonic origin from somitic and lateral plate mesoderm. Notably, these fate maps have been used to generate a model of scapular development in which somitic mesoderm maintains is metameric arrangement in both the scapular blade and the muscle groups that link the axial and appendicular systems. In this chapter I use an approach that combines surgical transplantation with highly detailed three-dimensional reconstructions to map the primaxial and abaxial domains in the avian pectoral region. These reconstructions provide a very different view of scapula development than previously thought. The majority of the avian scapula derives from lateral plate mesoderm and is abaxial. The somitic contribution to the scapular blade is modest, and differential growth of the abaxial scapula displaces the primaxial region posteriorly. Graft contributions to primaxial muscles attaching to the scapula also reflect this differential growth, and, in contrast to previous reports, insert on both primaxial and abaxial regions of the blade. Amphibians are an important group for studying the evolution and development of the vertebrate musculoskeletal system because of their phylogenetic position at the base of the tetrapod radiation. In chapter four of this thesis I provide a detailed method for isotopic transplantation of lateral plate mesoderm in axolotls, and extend the fate maps presented in chapter two to describe the position of the lateral 8

20 somitic frontier. Data from these surgeries provide evidence that the majority of the axolotl pectoral girdle and its associated musculature derive from lateral plate mesoderm and are abaxial. The scapula, however, incorporates the lateral somitic frontier, and is primaxial along its vertebral border. Interestingly, the hypaxial muscles of the lateral body wall are abaxial, similar to the muscles of the limb and shoulder. This observation strongly suggests that the body wall musculature of axolotls derives from migratory somitic cells that differentiate in the context of lateral plate connective tissue. 9

21 Figure 1. Illustration of the primaxial and abaxial domains of a developing mouse embryo. A. Cross-section through an embryo at approximately 9 days postfertilization. The somites (som) are pink, the somatic lateral plate (slp) purple, and the splanchnic lateral plate (splp) grey. The somatic lateral plate and overlying ectoderm comprise the somatopleure. The splanchnic lateral plate and endoderm comprise the splanchnopleure. B,C. Cross sections through the levels of the forelimb (left) and inter-limb trunk (right) at approximately thirteen days post-fertilization. The primaxial domain is pink, the abaxial domain is purple. Intercostal muscles (ic) are primaxial whereas the muscles of the limbs and ventro-lateral body wall are abaxial (purple dots at limb and trunk levels, respectively). The intermediate mesoderm and splanchnopleure derivatives are not shown for simplicity. Figure modified from Durland et al (2008). nc, notochord; nt, neural tube. 10

22 11

23 Figure 2. Simplified vertebrate phylogeny. Left to right: Tunicata (Ciona intestinalis, post-metamorphic), Cyclostomata (Petromyzon marinus), Chondrichthyes (Scyliorhinus canicula), Actinopterygii (Danio rerio), Lissamphibia (Ambystoma tigrinum), Mammalia (Mus musculus), Testudinata (Chelydra serpentina), Lepidosauria (Python curtus), Archosauria (Gallus gallus). Red asterisks indicate the model and non-model systems examined in this thesis. Figure modified from Shearman and Burke,

24 * * * * 13

25 CHAPTER 2: Part A. The persistence of body wall somatopleure in gnathostomes and a new perspective on the origin of vertebrate paired fins. The majority of Part A has been accepted for publication in Proceedings of the National Academy of Sciences (Tulenko et al., in press) SUMMARY Classical hypotheses regarding the evolutionary origin of paired appendages propose transformation of precursor structures (gill arches, lateral fin-folds) into paired fins. During development, gnathostome paired appendages form as outgrowths of the body-wall somatopleure, a tissue composed of somatic lateral plate mesoderm (LPM) and overlying ectoderm. In amniotes, LPM contributes connective tissue to abaxial musculature and forms the ventro-lateral dermis of the inter-limb body-wall. The phylogenetic distribution of this character is uncertain, however, as lineage analyses of LPM have not been generated in anamniotes. In this work I focus on the evolutionary history of the somatopleure to gain insight into the tissue context in which paired fins first appeared. Lampreys diverged from other vertebrates prior to the acquisition of paired fins and provide a model for investigating the preappendicular condition. Herein, I use vital-dye fate maps to provide evidence that the somatopleure is eliminated in lamprey as LPM is separated from the ectoderm and 14

26 sequestered to the coelomic linings during myotome extension. I also examine the distribution of post-cranial mesoderm in catshark and axolotl. In contrast to lamprey, findings from these experiments support an LPM contribution to the trunk body-wall of these taxa, similar to published data for amniotes. I propose that the innovation of a persistent somatopleure in the lateral body wall is a synapomorphy of gnathostomes and may be a feature of the clade gnathostomes plus osteostracans, the earliest jawless fishes that bore pectoral fins (see Sanson, 2009). Furthermore, I propose that redistribution of LPM was a key step in generating the novel developmental module that ultimately produced paired fins. These new embryological criteria can refocus arguments on the origin of paired fins and generate hypotheses testable by comparative studies on the source, sequence, and extent of genetic re-deployments. INTRODUCTION Paired fins were a key novelty that arose early in the radiation of vertebrates, changing locomotor ability and ecological opportunity. Historically, two hypotheses for the evolutionary origin of paired fins have generated the most discussion: the gill arch hypothesis (Gegenbaur, 1876; Gegenbaur, 1878) and the lateral fin-fold hypothesis (Thacher, 1877; Balfour, 1878; Mivart, 1879; Balfour, 1881). The gill arch hypothesis posits that paired fins arose through transformation of the posterior gill skeleton. The lateral fin-fold hypothesis maintains that paired fins evolved as retained portions of a continuous lateral fin structurally similar to the median fin observed in anamniote embryos. Neither hypothesis in its original formulation is well-supported 15

27 by either the fin morphologies of stem gnathostomes or the developmental morphologies of extant taxa (Coates, 1994; Coates and Cohn, 1998; Bemis and Grande, 1999; Tanaka et al., 2002; Coates, 2003). Recent studies have explored generative homologies to gain insight into how paired appendages evolved, and have shown that several of the genes that pattern paired fins/limbs also function during the development of gill arches [Shh and FGF8: (Gillis et al, 2009)], median fins [Hox9 13, Tbx18 and FGF s (Freitas et al, 2006; Yonei-Tamura et al, 2008; Abe et al, 2007)] and the heart field [Tbx4/5 (Horton et al, 2008; Kokubo et al, 2010; Onimaru et al, 2011)]. Notably, these data are consistent with aspects of both the gill arch and lateral fin-fold hypotheses, and support the argument that the evolution of paired fins involved the redeployment of pre-existing patterning programs into a new embryonic context, i.e., the fin-forming fields (see also pp of this thesis) (Shubin et al., 1997; Shubin et al., 2009). Implicit in both classical and recent discussions of the origin of the appendicular system is the presence of undifferentiated precursor tissue in which evolution produced paired fins. We address a neglected aspect of this discussion by focusing on the nature of the ancestral body wall in which paired fins evolved. The musculoskeletal body plan of ancestral vertebrates consisted of branchial and axial structures only, including gills supported by skeletal arches, segmental myotomes and median fins (Janvier, 1996; Donoghue et al., 2000). In extant gnathostomes paired appendages appear as additions to an embryo that has already developed an axial body plan. Vertebrae, ribs, and segmental myotomes comprise the 16

28 axial musculoskeletal system and derive from somitic mesoderm. Fin/limb buds appear at pectoral and pelvic levels as outgrowths of the somatopleure, and signaling between the ectoderm and somatic lateral plate mesoderm (LPM) is critical for their formation (Capdevila and Izpisua Belmonte, 2001). Somitic myoblasts infiltrate these buds to form appendicular musculature (Haines and Currie, 2001). Lampreys are agnathan vertebrates that split from the lineage leading to gnathostomes prior to the origin of paired fins (Coates, 1994; Osorio and Retaux, 2008). Although their suitability as a proxy for the ancestral condition of more derived vertebrates is a matter of continuing debate (Janvier, 2011; Shimeld and Donoghue, 2012) lamprey embryos provide the best option for exploring the embryological context in which paired fins arose, and for testing hypotheses about the distribution of LPM in early gnathostomes. Recent studies have shown that the lamprey myotome is dorso-ventrally compartmentalized despite lacking a horizontal septum (Kusakabe et al., 2011), and that the LPM is patterned into distinct cardiac and posterior regions (Kokubo et al., 2010; Onimaru et al., 2011). Few studies, however, have characterized the lamprey LPM during stages of body wall formation (Richardson et al., 2010; Onimaru et al., 2011). Here I examine the morphological changes taking place during this process in both the Sea Lamprey Petromyzon marinus and the Japanese Lamprey Lethenteron japonicum to extend classical descriptions (Shipley, 1887; Goette, 1890; Hatta, 1901), and provide the first long term fate maps of somitic and LPM in an agnathan vertebrate. Additionally, I performed antibody labeling in the shark Scyliorhinus canicula and isotopic 17

29 mesoderm transplants in axolotl, to visualize the interface between somitic and LPM in two major gnathostomes lineages. RESULTS Histological Anatomy in Lamprey Embryos Histological anatomy during body wall formation is very similar in Petromyzon marinus and Lethenteron japonicum (Fig. 1A,B; Fig. 2; Fig. S1). In plastic sections of early stage embryos (St.23) (Tahara, 1988), the presumptive lateral plate mesoderm (PLPM) forms a thin layer of cells that extends laterally from the somites, separating the dorsal epidermis from the underlying endoderm (Fig. 1A,B; Fig. S1A,B). During subsequent stages of development (Stages 24 30), the embryo straightens, the yolk ball elongates into a tube, and new myotomes form from somitic mesoderm in an anterior to posterior direction (Fig. S2). The PLPM in these later stages first surrounds the yolk tube (Fig. 2A,D; Fig. S1C,D), and then splits into distinct somatic and splanchnic layers, well before myotome closure of the body wall (Fig. 2E,F,H; Fig.S1E H). The dermomyotome (DM), a thin layer of undifferentiated cells on the external surface of the myotome, forms distinct lip-like folds both dorsally and ventrally (Fig. 2A,B,E,F; Fig. S1C F). As the ventral lip of the DM advances, it appears to wedge between the ectoderm and the somatic layer of the PLPM (Fig. 2E,F; Fig. S1E,F). 18

30 Molecular Identification of Embryonic Populations in Lamprey The transcription factor dhand has been used as a molecular marker of early LPM in a number of vertebrate taxa (Angelo et al., 2000; Fernandez-Teran et al., 2000; Tanaka et al., 2002; Onimaru et al., 2011). As previously reported, LjHandA is expressed in the branchial region, heart, and PLPM of whole mount Lethenteron embryos between stages (Fig. S3). Cryosections through these whole mounts confirm the position of label within the PLPM (Fig. 1C; Fig. S3E,G,I), supporting the lateral plate (LPM) identity of this tissue. As development proceeds, the expression domain of LjHandA extends ventrally to surround the yolk tube (stage 25, Fig. S3B D,F), consistent with the early ventral migration of LPM (compare Fig. 2A,D and Fig. S3F). LjHandA in the LPM is then down-regulated in an A-P sweep as the myotomes extend ventrally, and is limited to the posterior-most region of the embryo by Stage 27. In the absence of gene expression at stages of myotome extension, the distribution of LPM cells is difficult to follow, though as mentioned above, a thin layer of cells is visible in plastic section deep to the myotome and between the ectoderm and yolk (Fig. 2A,B,D F and Fig. S1C F). Probes generated from partial sequence of Petromyzon Hand labeled heart and branchial regions, but did not show strong LPM label. Pax3/7 transcription factors are established markers for the dermomyotome (DM) in vertebrates (Kusakabe and Kuratani, 2005; Devoto et al., 2006; Kusakabe et al., 2011). We used the antibody DP312, which recognizes Pax3/7 proteins (Davis et al., 2005; Hammond et al., 2007), to visualize changes in the position of the DM in 19

31 lamprey embryos at different stages of development (Fig. S4). In Petromyzon, DP312 labeling of the ventral lip of the DM (Fig. 2B,C,F,G and Fig. S4C F) makes visible the ventral migration of the somitic tissue as it extends between the ectoderm and underlying tissue during body wall closure. DiI Labeling of Mesoderm in the Lamprey Body Wall I further characterized the interface of somitic and lateral plate tissue in lamprey by injecting the vital dye DiI into either somites or LPM of young Petromyzon embryos, which allowed long-term fate-mapping of these lineages (Stages 22 24) (Fig.3, Fig.4, Fig.S5). In total I performed 188 somite and 209 LPM injections, which were then either collected within 12 hours of injection to evaluate DiI targeting or at developmental stages with varying degrees of body wall closure (Stages 25 30) (Table 1). I sectioned embryos with the brightest whole-mount fluorescence (18 somite injections; 29 LPM injections; Table 1) and stained sections from older embryos for skeletal muscle. Results of injections were consistent across specimens (Tables 2), and sections of embryos collected within 12 hours of injection showed localization of DiI in targeted tissue (Fig. 4A,B). In whole mounts of somite-injected embryos, the distribution of DiI is largely coincident with a single myotome (Fig. 3). Cross-sections show DiI-positive myofibers and labeled cells in the dermomyotome (DM) and presumptive sclerotome, as well as rare mesenchymal cells within the dorsal fin-fold (Fig. 3D H; Table 2B). In a majority of embryos at Stages (10/14; Table 2B), DiI-labeled cells are present in the ventral lip of the DM (compare Fig. 3F,H and Fig.2B,C,F,G). No 20

32 migratory DiI-positive cells are present ventral to this lip in any somite-injected embryos examined, indicating the ventral lip of the dermomyotome is the leading edge of somitic mesoderm during body wall closure. In LPM-injected embryos, DiI-positive cells form a band of labeled tissue largely restricted to the ventral half of each specimen (Fig. 4C E). Cross-sections through Stage 25/26 embryos reveal the movement of DiI-positive cells from the dorsum of the yolk-ball to surround the yolk tube (Fig. 4C G), consistent with changes in histology (Fig.2D) and lamprey Hand expression as described above (Fig. S3A C,F). In more advanced embryos, DiI-positive cells are observed medial to the myotome, but not within the myotome or lateral to it (Fig. 4H,I; Fig. S5A E; Table 2), with the exception of labeled ectoderm at the injection site. By the larval stage (Stage 30) DiIpositive cells contribute to the coelomic linings, typhlosole, and gut vasculature (Fig. 4J,K; Fig. S5F K). Mesoderm Distribution in Catshark Pectoral Fin and Body Wall I used the antibody DP312 to label the dermomyotome of the catshark Scyliorhinus canicula, a galeomorph shark belonging to the order Carcharhiniformes (Velez-Zuazo and Agnarsson, 2011). In Stage 27 embryos (Ballard et al., 1993), the pectoral fin buds are present as ventral expansions of the somatopleure (Fig. 5A,B). In cross-section, the DP312-labeled DM closely opposes the ectoderm dorsally. The ventral lip of the DM, however, loses contact with the modified ectoderm of the fin bud at the approximate level of the nephric duct, and enters the fin bud mesenchyme 21

33 (LPM, Fig.5A,B). At inter-fin levels of the same stage, no mesenchymal cells separate the DM and ectoderm (Fig.5C,D). In contrast, in inter-fin sections of Stage 28 embryos, the DM loses contact with the ectoderm, and loose mesenchyme separates the two tissues (Fig. 5E,F). A distinct boundary is visible at the level of the nephric duct between the DP312-positive mesenchyme generated from the DM and the DP312-negative mesenchyme (Fig. 5E,F). I interpret this label-boundary as the lateral somitic frontier, the interface of somitic and LPM tissues (Nowicki et al., 2003). The mesenchyme and ectoderm ventral to the boundary comprise the somatopleure, which persists in the external body wall. Lateral Plate Transplants in Axolotl Embryos I examined the distribution of somatopleure in axolotls (Ambystoma mexicanum) by isotopically transplanting LPM and ectoderm from GFP donors to WT hosts. Pronephros was included in a subset of grafts to ensure transplantation of the dorsal margin of LPM (Fig.6A C; Fig. S6A,B). In total, I performed 42 surgeries and sectioned 10 of these between Stages 32 and 57 (Borzilovskaya et al., 1989; Nye et al., 2003) (Table 3). Sections of chimeras collected one day post-op revealed wellincorporated graft (Fig. S6C,D). In whole mounts of chimeras reared to Stages 46 57, graft-derived cells contribute to both the fore- and inter-limb regions. Sections through the pectoral region reveal GFP-positive cells forming connective tissue of appendicular musculature (e.g., dorsalis scapulae, pectoralis) and the hypaxial myotome, as well as chondrocytes of the pectoral girdle (scapula, coracoid) (Fig. 6D 22

34 E ). At inter-limb levels, myofibers of the hypaxial myotome are also invested by GFP-positive cells. The LPM of the somatopleure is visible as GFP-positive mesenchyme lateral to the muscular body wall (Fig.6F G ). DISCUSSION Somatopleure Persistence in Gnathostomes These results provide evidence that the distribution of mesodermal lineages in the body wall of lamprey differs significantly from the body wall of gnathostomes. The histological, gene expression, and fate mapping data shown here indicate that in lamprey the LPM extends around the yolk tube prior to ventral advance of somitic mesoderm, and that the ventral lip of the dermomyotome extends along the inner surface of the ectoderm, displacing the LPM inwards during myotome closure of the body wall. In the absence of a permanent molecular marker for the somatic lateral plate, the DiI labeling experiments presented above strongly suggest that the somatopleure, which is present in the early embryo, is eliminated during this process as LPM is segregated to the coelomic linings. The resulting muscular body wall is primaxial, deriving entirely from somitic mesoderm. In model amniotes it is well established that LPM contributes connective tissue to both the musculature (e.g., latissimus dorsi, abdominal obliques) and superficial dermis of the flank, thus reflecting a somatopleure contribution to the adult body wall (Nowicki et al., 2003; Durland et al., 2008; Shearman and Burke, 2009). The phylogenetic distribution of this character has been uncertain in the absence of lineage analyses in anamniotes. Our isotopic transplants of early LPM from GFP 23

35 axolotls into wild type hosts show that the salamander body wall includes LPM cells medial and lateral to, as well as within body wall musculature. These data also indicate that an abaxial body wall is likely primitive for tetrapods. Considering the evolutionary transition from fins to limbs, it would seem possible that the persistence of the somatopleure reflects morphological changes at the base of the tetrapod radiation, or during the eventual transition to a terrestrial environment. Chondrichthyans can be used as an outgroup to test this hypothesis. They are the sister group of all other living gnathostomes, and their embryos can be used to gain insight into the development of the last common ancestor of all vertebrates with paired appendages (Cole and Currie, 2007; Dahn et al., 2007; Gillis and Shubin, 2009). During muscle development in chondrichthyans, both appendicular and body wall musculature derive from epithelial extensions of the ventral dermomyotome (Neyt et al., 2000; Cole et al., 2011). DP312 labeling presented here in Scyliorhinus shows a clean boundary between labeled mesenchyme derived from the DM, and unlabeled mesenchyme of presumed lateral plate origin at both fin and inter-fin levels (Fig 5). The LPM remains as a mesenchymal layer between the ectoderm and the advancing myotome, as in the axolotl. Studies of certain hypaxial body wall muscles in pearlfish (Windner et al., 2011) and the experimental induction of ectopic fin-field markers in zebrafish (Yonei-Tamura et al., 2008) indicate the possibility of this condition in the trunk of teleost embryos, though the LPM has not yet been mapped in an actinopterygian. The lateral somitic frontier (Burke and Nowicki, 2003; Nowicki et al., 2003) is a cryptic lineage boundary between somitic versus LPM, here revealed 24

36 by DP312 expression in shark and grafted GFP positive tissues in chimeric axolotls. There is a persistent somatopleure in the body wall of both these gnathostomes, as seen in amniotes. In contrast, the frontier in lamprey appears to be displaced to the ventral midline of the body wall as the somatopleure is disrupted by growth of the somites. Relevance to Classical Theories for Paired Fin Origins The fin/limb buds of all gnathostomes studied arise as outgrowths of the somatopleure, and the skeletal and connective tissue cells of the paired appendages derive from the LPM. My new data on the disruption of the somatopleure in lamprey provides insight into the early history of the LPM and an embryological context for focusing questions on the initial conditions necessary for the evolution of the appendicular system. If the lamprey condition is primitive for vertebrates, a comparison of lamprey, cat shark, axolotl and amniotes would suggest that a persistent somatopleure evolved along the gnathostome stem, and this innovative persistence of somatic LPM external to the myotome was a key early step in the evolution of the embryonic field that ultimately produced paired fins (Fig. 7). Comparative studies of body wall formation in myxinids will further test the polarity of this vertebrate character suite. Interestingly, the presence of pectoral fins in fossil osteostracans (Sanson, 2009) may suggest that somatopleure persistence predates the origin of jaws, and may be a synapomorphy of gnathostomes plus osteostracans (Fig. 7). 25

37 One model of paired fin-origins suggested by my data is that a persistent somatopleure arose continuously along the flank from the branchial region to the cloaca. This novel LPM domain may have carried a Hox code shared with the regionalized gut (Coates, 1998; Coates and Cohn, 1998), presaging the localization and differentiation of pectoral and pelvic potential. Although such a model invokes aspects of the fin-fold hypothesis, it does not require the presence of a continuous lateral fin, per-se. Alternatively, the somatopleure may have initially persisted in proximity to the gills, established a pectoral fin and subsequently spread posteriorly to the pelvic level. The latter model is equally supported by our data and has affinities for variations on the gill arch hypothesis (Gegenbaur, 1876; Gegenbaur, 1878; Coates, 2003). Fossil data supports the early and singular appearance of pectoral fins in osteostracans (Coates, 1994; Coates, 2003), though some authors argue for evidence of ventro-lateral fin-folds in fossil agnathans (Wilson et al., 2007). Consideration of the advent of a persistent somatopleure in the body wall of ancestral gnathostomes could suggest new interpretations of transitional forms. Many stem agnathans possess lateral fin-like structures that vary in size, number, and position along the flank, and there is debate about their homology with the paired fins of gnathostomes. Using the embryological framework (Burke and Nowicki, 2003) applied here, Johanson (2010) recently speculated that these problematic fins are primaxial, consistent with absence of a lateral plate contribution and lack of homology to gnathostome paired fins. 26

38 As noted earlier, molecular studies indicate that pre-existing gene regulatory networks were co-opted to pattern the fin/limb field. The embryological data presented here suggests that molecular events provoking mechanistic changes in the relationship between the advancing dermomyotome and the ectoderm in a vertebrate lineage with a previously primaxial body wall could generate a persistent somatopleure with the potential to harbor abaxial structures (see also pp of this thesis). I suggest that stem gnathostomes developed a localized expansion of the somatopleure (Fig. 7) in proximity to the nephric ridge at the head-trunk interface. This embryological event was likely induced by molecular signaling between ectoderm and mesoderm, provoked by the partial activation or co-option of genetic networks already established in the gill or heart fields. Such hypotheses can be pursued through comparative molecular studies aimed at testing the source, sequence, and extent of genetic re-deployment resulting in the dramatic morphological innovation of the paired appendages. MATERIALS AND METHODS Lamprey embryos were staged according to Tahara (1988). Embryos of P. marinus were reared as described by Martin et al (2009). Plastic sections for both P. marinus and L. japonicum were generated using JB-4 Plus (EM Sciences) according to manufacturer s instructions, and stained with toluidine blue or haematoxylin and eosin. 27

39 For immunohistochemistry in lamprey, embryos were fixed in 4% PFA. Cryosections were blocked in PBST with 5%NGS and 2%BSA, and incubated overnight with MF20 (DSHB, 1:20 dilution) or DP312 (1:20 dilution). AlexaFluor 647 goat anti-mouse IgG2b (Invitrogen) fluorescent secondary was used with MF20. DP312 signal was amplified using a universal biotinylated secondary (Vector Laboratories) with Streptavidin-Cy5 (SouthernBiotech). Rhodamine-Phalloidin (1:200, Invitrogen) was used to identify skeletal muscle in DP312-labeled sections. Nuclei were counterstained with Sytox Green (Invitrogen). Sections were visualized using a Zeiss LSM 510 confocal microscope or Nikon Eclipse E600. Whole mount specimens were imaged using a Nikon SMZ-U dissecting microscope. In-situ hybridization in L. japonicum was performed as described by Sugahara et al (2011), with Hand probe construct from Kuraku et al. (2010). For fate maps, fixable DiI (Molecular Probes, Invitrogen) was diluted in DMSO (1µg/µl), loaded into pulled, thin-walled glass capillaries, and pressure injected into somites or LPM (Stages 22 24) using a Parker Hannifin General Valve Picospritzer. Embryos were positioned with modeling clay and visualized using a Zeiss Stemi 2000 microscope. The margins of the somites were visible and served as landmarks for somitic injections. For LPM injections the tip of the capillary was positioned within one somite s width of the lateral somitic margin. When visible, the nephric duct was also used as a landmark. Following injection, embryos were raised to Stages and fixed in 4%PFA. Those with bright DiI-labeling in whole mount were cryosectioned and labeled with MF20 or phalloidin for muscle. 28

40 Scyliorhinus caniculi embryos were staged according to Ballard (1993), labeled with DP312 and goat anti-mouse IgG Cy5 secondary (Jackson Labs) without signal amplification. GFP and WT embryos of A. mexicanum were obtained from the Ambystoma Genetic Stock Center (University of Kentucky), and staged according to Bordzilovskaya et al (1989) and Nye et al (2003). For surgeries, embryos were positioned in 2X Steinberg solution (STN) in clay-lined Petri dishes. LPM/Ectoderm or LPM/Ectoderm/Pronephros was isotopically transplanted from stage-matched GFP donors to WT hosts (Stages 23 27) using sharpened tungsten needles and pulled glass capillaries. Grafts were held in place by glass cover slip fragments and STN was replaced with 25% Holtfreter solution. Chimeras were evaluated approximately 24 hours after surgery for graft incorporation, and reared at 18 o C. Cryosections of axolotl chimeras were labeled for GFP as described above using A6455 (Invitrogen, 1:500 dilution) and AlexaFluor 488 goat anti-rabbit IgG. Phalloidin was used to identify skeletal muscle and nuclei were counterstained with TO-PRO -3 Iodide (Invitrogen). 29

41 Figure 1. Distribution of lateral plate mesoderm in Stage 23/24 lamprey embryos. A,B. Plastic sections through the mid-yolk ball of Petromyzon (A) and Lethenteron (B) embryos stained with Toluidine blue. In both A and B, presumptive lateral plate mesoderm (plpm) extends laterally from the somites (som) along the dorsum of the yolk ball (yb). C. Cryosection of whole mount in-situ of Lethenteron embryo. Expression of LjHandA (blue) supports lateral plate identity of plpm. See Fig. S3 for additional stages. ect, ectoderm; nc, notochord; nt, neural tube. Scale bar=50µm. 30

42 31

43 Figure 2. Plastic sections and DP312 labeling of Petromyzon embryos and larvae. Plastic sections (A,B,D,E,F,H) stained with H&E. Cryosections (C,G) are labeled for Pax3/7 (DP312, orange), skeletal muscle (phalloidin, green), and nuclei (Sytox, cyan). A D. Stage 25. DP312-positive cells of the DM lie adjacent to the inner surface of the ectoderm and form an epithelial lip ventrally (vldm) (compare B and C). Elongate cells of the presumptive lpm (arrowheads in D) surround the yolk tube (yt). E G. Stage 28. The lpm has split into somatic (slp) and splanchnic (splp) layers, forming a coelom (c). The ventral lip of the dermomyotome is positioned between ectoderm and somatic lateral plate (E G). H. Stage 30. Myotomes close the body wall ventrally. See Fig. S2 for approximate planes of section. ect, ectoderm; end, endoderm; myc, myocoel; my, myotome; nc, notochord; nt, neural tube. Scale bar=50µm. 32

44 33

45 Figure 3. DiI labeling of somites in Petromyzon. A. Dorsal view of St. 23 embryo at time of injection (arrowhead indicates position of DiI). Panels B,C & D are of the same embryo. B,C. Lateral view of St.26. DiI appears coincident with a single myotome. Scale bar=500µm. D H. Cryosections through somite-injected embryos at Stages 25 (E,F) and 28 (D,G,H) labeled for skeletal muscle (MF20, green) and nuclei (Sytox, cyan). DiI (red) labels myofibers, and cells of the presumptive sclerotome ( sc ), and dermomyotome (dm), including its ventral lip (vldm). No DiI-positive cells are present ventral to the vldm. See Fig. 2 legend for abbreviations. Scale bar=50µm. 34

46 35

47 Figure 4. DiI labeling of LPM in Petromyzon. A. Dorsal view of St. 23 embryo at time of injection (arrowhead indicates position of DiI). B. Cryosection through same embryo showing DiI (red) in targeted LPM (arrowheads). C E. Dorsal view of St. 23 embryo at time of injection (C, arrowhead indicates position of DiI) and lateral view of same embryo at St.26 (D,E). F K. Cryosections through lpm-injected embryo shown in C E (F,G), and embryos at St 28 (H,I) and 30 (J,K) labeled for skeletal muscle (MF20, green) and nuclei (Sytox, cyan). DiI-positive cells (red) surround the yolk tube by Stage 26 and contribute to the coelomic linings by Stage 30. Note DiI positive cells are always medial or ventral to the myotome (my), and never lateral to or within the my. See also Fig.S5. See Fig. 2 legend for abbreviations. Scale bar=50µm. 36

48 37

49 Figure 5. DP312 labeling in Scyliorhinus. A D. Cryosections through the pectoral fin (A,B) and inter-fin trunk (C,D) of Stage 27 catshark. Mesenchyme is present between the DP312-positive dermomyotome (dm) (orange) and modified ectoderm of the fin-bud (A,B). In contrast, mesenchymal cells are not seen between the DP312- positive dm and ectoderm dorsal to the fin-bud (A,B) or along the inter-fin trunk at this stage (C,D). E,F. Cryosections through the inter-fin trunk of a Stage 28 catshark. At the level of the nephric duct (nd), DP312-positive mesenchyme from the dermomyotome forms a boundary with unlabeled mesenchyme from presumptive lateral plate. aff, apical fin-fold; lsf, lateral somitic frontier. See Fig.2 for abbreviations. Scale bar=50µm. 38

50 39

51 Figure 6. GFP to WT Isotopic transplants of LPM/Pronephros/Ect in Axolotl. A- C. Lateral view of a surgery chimera at Stages 24 (A, time of surgery), and same specimen at St. 33 (B), and 53 (C), showing distribution of GFP donor tissue (green). Note caudal migration of pronephros (arrowhead) (B) and graft-derived skeletal elements of the forelimb (C). Scale bar=500µm. D G. Cryosections through the pectoral (D E ) and inter-limb (F G ) regions of larva pictured in C labeled for GFP (green), skeletal muscle (phalloidin, red) and nuclei (TO-PRO, cyan). In the pectoral region, graft-derived cells are present in scapula (sc) and coracoid (cor), and form the connective tissue of appendicular muscles (e.g., ds, dorsalis scapulae; pct, pectoralis) and the hypaxial myotome (hp) (D E ). In inter-limb levels (F G ) graft-derived cells form mesenchyme lateral to hp, as well as connective tissue within hp. ep, epaxial myotome; pc, parietal coelomic lining. See Fig.2 legend for abbreviations. Scale bar=250µm. 40

52 41

53 Figure 7. Evolution of the persistent somatopleure. Simplified vertebrate phylogeny with schematic, mid-trunk cross-sections through lamprey (left), shark (left-middle), axolotl (right-middle) and amniote (right) embryos. Somatic LPM contributions to the body wall are purple. Somatopleure persistence is a gnathostome synapomorphy. The distributions of somatopleure in fossil taxa are hypothesized (purple stippling). In an osteostracan (ventro-lateral view), the somatopleure would correspond to the position of the pectoral fins, and may have extended posteriorly, coincident with the hypothetical ventro-lateral ridges. In a placoderm, the somatopleure would correspond to the positions of the pectoral and pelvic fins, and may have contributed to the inter-fin trunk. Fossil taxa are redrawn from Wilson et al (2007). 42

54 43

55 Figure S1. Plastic sections of Lethenteron embryos and larvae. Sections are stained with toluidine blue (A,B) or H&E (C H). A,B. Stage 23. Section through mid-yolk ball. Lateral plate mesoderm (lpm) extends laterally from somites (som) along the dorsum of yolk ball (yb). C,D. Stage 25. The ventral lip of the dermomyotome (vldm) is dorsal to the nephric duct (dashed line in D), and cells of the presumptive lpm (arrowheads in E) surround the yolk tube (yt). E,F. Stage 28. The lpm has split into somatic (slp) and splanchnic (splp) layers and a coelom (c) has formed. The vldm is positioned between ectoderm (ect) and somatic lateral plate. G,H. Stage 30. Myotomes close the body wall ventrally and the coelom has increased in size. The changes in histological anatomy summarized here for Lethenteron are highly similar to those observed in Petromyzon (see Fig. 2). dm, dermomyotome; end, endoderm; myc, myocoel; my, myotome; nc, notochord; nt, neural tube; pc, parietal coelomic lining; vc, visceral coelomic lining. Scale bar=50µm. 44

56 45

57 Figure S2. Myotome formation in Petromyzon. A E. Lateral view of a developmental series of embryos and larvae at Stages 23(A), 25(B), 26 (C), 28 (D), and 30 (E) stained with the skeletal muscle marker MF20 (brown). Trunk myotomes form dorsal to the yolk tube and extend ventrally to close the body wall. No postbranchial musculature forms discontinuously from its parent myotome. Black lines indicate approximate planes of cross-sections in Fig. 2. a, anterior; hp, head process; p, posterior; yb, yolk ball; yt, yolk tube. Scale bar=500µm. 46

58 47

59 Figure S3. LjHandA expression in Lethenteron. A D. Lateral view of embryos at Stages 24 (A), 25 (B), 26 (C), and 27 (D). E I. Cryosections through embryos shown in A, B, and D. LjHandA (blue) is expressed in the heart (ht), branchial region (br), and lateral plate mesoderm (lpm). Note that expression in the lpm initiates lateral to the somites, and then extends ventrally to surround the yolk tube in an anteriorposterior sweep before becoming down regulated. Ventral margin of expression is highlighted with dotted white line in A D. Scale bar=500µm. 48

60 49

61 Figure S4. DP312 labeling in Petromyzon. A,B. Cryosection through the mid-yolk ball of a Stage 23 embryo. DP312-positive cells are brown and Sytox-labeled nuclei are black. The dorsolateral margin of the epithelial somite labels positively with DP312 whereas cells of the ventromedial somite (adjacent to the notochord) and lateral plate are negative. C F. Cryosections through embryos at Stages 25 (C,D) and 27 (E,F). DP312-positive cells are orange; phalloidin labeled skeletal muscle is green; and Sytox-labeled nuclei are cyan. The dermomyotome, a single layer of cells superficial to the myotome, labels positively with DP312. Comparison of the DP312- positive ventral lip of the dermomyotome (D and F) with somitic DiI labeling (Fig. 2B E) indicates that DP312 can be used to identify the leading edge of somitic contributions to the body wall. See Fig. 1 legend for abbreviations. Scale bar=50µm. 50

62 51

63 Figure S5. DiI labeling of LPM in Petromyzon. A C. Dorsal view of St. 23 embryo at time of injection (C, arrowhead indicates position of DiI) and lateral view of same embryo at St.28 (B,C). D,E. Cryosections through B labeled for skeletal muscle (Phalloidin, green) and nuclei (Sytox, cyan). DiI-positive cells (red) are present medial and ventral to the myotome (my) surrounding the yolk tube. F I. Dorsal view of St. 23 embryo at time of injection (arrowhead indicates position of DiI) and right and left lateral views of same embryo at St.30 (G I). J,K. Cryosections through G at Stage 30 (D,E) labeled as above. DiI-positive cells (red) invest the gut tube and contribute to the typhlosole (ty). Note that DiI fluorescence contralateral to site of injection (I K) reflects gut rotation during development (53). See Fig. 2 legend for abbreviations. end, endoderm; lt lat, left lateral; rt, right lateral. Scale bar=50µm. 52

64 53

65 Figure S6. GFP to WT isotopic transplants of LPM/Pronephros/Ect in Ambystoma one day post-surgery. A,B. Lateral view of a single surgery chimera at Stages 24 (A, time of surgery) and 33 (B, approximately 24 hours post-surgery). GFP donor tissue is green. Note caudal migration of pronephros (arrowhead) (B). Scale bar=500µm. C,D. Cryosection through B labeled for GFP (green) and skeletal muscle (phalloidin, red) showing graft incorporation. See Fig. 2 legend for abbreviations. Scale bar=250µm. 54

66 55

67 Table 1. Total number of DiI injections in Petromyzon embryos: Mesoderm Target Total Injections Survival (%) Somite (95%) LPM (92%) LPM, lateral plate mesoderm Whole-Mount Fluorescence (%) 106 (60%) 71 (37%) Total Examined in Section Developmental Stage St.23 St St

68 Tables 2. Distribution of DiI in sectioned Petromyzon embryos: a. Collected within twelve hours of injection Injection Total Embryos Distribution of DiI-Labeled Cells Target Sectioned at St.23 Somite LPM Ectoderm Endoderm Somite LPM b. Collected between Stages Injection Target Total Embryos Sectioned at St. Distribution of DiI-Labeled Cells MY SC DFF DM/Lateral VLDM Surrounding Yolk Tube to MY ventral to VLDM Somite LPM DFF, dorsal fin fold; DM, dermomyotome; LPM, lateral plate mesoderm; MY, myotome; SC, presumptive sclerotome (cells medial to myotome); VLDM, ventral lip of dermomyotome. 57

69 Table 3. Total Number of LPM/Ectoderm and LPM/Ectoderm/Pronephros surgeries in Axolotl: Transplant Type Total Number Survival (%) Fluorescence (%) Total Sectioned Developmental Stage St.32 St St.57 LPM/Ect 11 8 (73%) 8 (100%) LPM/Ect/Pro (42%) 9 (69%) Ect, ectoderm; LPM, lateral plate mesoderm; Pro, Pronephros. 58

70 CHAPTER 2: Part B. The origin of the vertebrate appendicular system: genetic redeployment, tissue redistribution, and abaxial invention SUMMARY The presence of an appendicular system is a hallmark of the body plan of most gnathostomes. In fishes, this includes the pectoral and pelvic fins, and in tetrapods, the fore- and hind-limbs. Questions about how the appendicular system first arose require insights into the changes that took place along the gnathostome stem. The gill arch and lateral fin-fold hypotheses are undoubtedly the most famous hypotheses for the origin of paired fins, and are presented in most textbooks of comparative anatomy. However, these classical hypotheses propose anatomical homologies that are not well-supported by the fossil record or comparative anatomy. Herein, I summarize published data highlighting similarities in the developmental patterning of gill arches, median fins, and paired fins, the anatomical structures that inspired the gill arch and lateral fin-fold hypotheses. These data support the argument that preexisting, complex genetic circuits, possibly for appendage outgrowth, were recycled during the early evolution of the appendicular system, facilitating the formation of new morphologies. In the previous section of this thesis, I used comparative embryological data to hypothesize that somatopleure persistence in the lateral body wall is a shared, derived feature of gnathostomes (and possibly of gnathostomes plus osteostracans). Herein, I extend this hypothesis and propose that retention of somatic 59

71 lateral plate in the flank created a new, naïve field of cells, which provided the substrate for gene network co-option and the formation of the abaxial module. I also speculate that the intermediate mesoderm may have had a key role in directing this ontogenetic change. Classical Hypotheses and the Origin of Paired Fins Gegenbaur (1876) proposed that the skeletal elements of paired fins and limbs evolved from gill support elements (Fig. 1A). Specifically, the pectoral and pelvic girdles represent transformed homologues of the gill arches, and the bones of paired fins and limbs, homologues of the gill rays. This hypothesis in its original formulation has been largely dismissed based on paleontological and morphological evidence (Jarvik, 1980; Coates, 1994; Bemis and Grande, 1999; Coates, 2003). For example, if paired fins derived from gill arches, then pelvic fins should appear prior to pectoral fins in the fossil record, and there should be a trend for increasing branchial-topelvic distance in fin position. The fossil record, however, supports the phylogenetic appearance of pectoral fins prior to pelvic fins in stem gnathostomes (Coates, 1994; Janvier, 1996), and no branchial-to-pelvic trend has been reported. Additionally, no fossil intermediates have been discovered that exhibit structures bearing both fin-girdle and gill-arch features (e.g., a scapulocoracoid with gill ray rudiments) (Coates, 2003). Embryological data also fail to support predictions of the gill arch hypothesis. For instance, gill arches and paired fins do not derive from a common embryonic population (Coates, 1994), as gill arches form from neural crest 60

72 cells whereas the bones of the appendicular system develop from mesoderm. It is worth noting, however, that a neural crest contribution to the mouse scapula has been reported (Matsuoka et al., 2005, but see also Epperlein et al., 2012 and Kague et al., 2012) and the embryological origin of gill rays, which are only present in chondrichthyans among extant taxa, is uncertain (Gillis et al., 2009; Gillis and Shubin, 2009; Gillis et al., 2011). In a recent variant of Gegenbaur s original hypothesis, Coates (2003) proposed that the scapulocoracoid of the pectoral girdle may have originated as a serial homologue of the extra-branchial cartilages of fossil agnathans. In most gnathostomes, gill-support elements are positioned near the pharynx, medial to the gill respiratory epithelia. In lamprey, this arrangement is reversed. The gill-support cartilages are lateral to the gill epithelia, and thus extra-branchial (Mallatt, 1984; Janvier, 2007). Future work comparing the development of the extra-branchial cartilages of lamprey and the fin girdles of jawed vertebrates can be used to assess Coates s hypothesis and the generative homologies of these structures (see pp for a discussion of generative homologies). Median fins evolved early in the radiation of craniates, prior to the origin of paired fins (Coates, 1994; Janvier, 1996; Donoghue et al., 2000). In extant anamniotes, median fins form as elongate dorsal and ventral folds of tissue that become subdivided into distinct dorsal, caudal, and post-anal fins (Goodrich, 1930). This observation, together with similarities in the adult anatomy of unpaired and 61

73 paired fins, inspired the lateral fin-fold hypothesis (LFH) as an alternative to the gill arch hypothesis (Bemis and Grande, 1999). According to the LFH, paired fins arose as retained portions of a continuous fin-fold that extended the length of the trunk (Fig. 1B) (Thacher, 1877; Mivart, 1879; Balfour, 1881). Presumptive mesoderm ( somatopleuric mesomesenchyme of Jarvik, 1980:121) would have formed the core of this fin-fold (Fig. 1C), giving rise to a series of segmented, endoskeletal radials, which fused proximally to form fin-support girdles (Fig. 1C,D) (Thacher, 1877). Although the LFH has garnered fame as a concise framework for inferring how paired fins originated (Liem et al., 2001; Kardong, 2011), it, like the gill arch hypothesis, is not well-supported by empirical evidence from paleontology (Coates, 1994; Bemis and Grande, 1999). Fossil agnathans exhibit a variety of lateral appendages (Janvier, 1996; Janvier, 2007; Wilson et al., 2007; Johanson, 2010), some of which appear as continuous fin-like structures that extend along much of the length of the trunk (e.g., Haikuichthyes and Myllokunmingia: Shu et al, 1998, but see also Shu et al, 2003; Janvier, 2003; Zhang and Hou, 2004; Jamoytius: White, 1946; Ritchie, 1968; Shubin et al., 1997; and Pharyngolepis, Ritchie, 1964). Although these fin-like structures have invoked comparisons with lateral fin-folds in a hypothetical ancestral vertebrate (Fig.1B), they lack endoskeletal supports, a central tenet of the LFH (Fig. 1D) (Coates, 2003). Furthermore, similar lateral outgrowths have originated numerous times in extant aquatic gnathostomes, including the tail flukes of cetaceans, the lateral caudal keels of many fast-swimming fishes, and the lateral body folds of hellbender salamanders (Harlan and Wilkinson, 1981; Coates, 62

74 2003). Another example is provided by acanthodians, a paraphyletic group of early fossil gnathostomes (Janvier, 2007; Brazeau, 2009; Davis et al., 2012). Some acanthodians exhibit discontinuous spines along the flank, which form a line connecting the pectoral and pelvic fins (e.g., Euthacanthus: Watson, 1937). Although these spines have been interpreted as an evolutionary intermediate between a continuous lateral fin-fold and distinct paired fins, they are more likely acanthodian specializations rather than transitional morphologies (Bemis and Grande, 1999; Coates, 2003). Embryological data are similarly indirect with respect to the LFH. Balfour (1878) reported that during the development of some sharks and skates (elasmobranchs), paired fins form as specialized expansions of a linear thickening of ectoderm along the flank (Balfour, 1878). Recent SEM images, however, failed to confirm the presence of this developmental morphology in the catshark Scylliorhinus (Tanaka et al., 2002), and its phylogenetic distribution is uncertain (Bemis and Grande, 1999). In a study of sharks, Goodrich (1930) noted that epithelial buds of the myotome are transiently observed at inter-fin levels, similar to those providing musculature to the paired fins. Although Goodrich s observations have since been corroborated, along with new published data for other chondrichthyan groups (Neyt et al., 2000; Cole et al., 2011; Don et al., 2013; Tulenko et al., 2013), the presence of these morphologically similar muscle projections in the fin and flank do not directly support the evolution of paired fins from a lateral fin-fold, but instead likely reflect a primitive mode of muscle extension (Neyt et al., 2000; Hollway and Currie, 2003). 63

75 Jarvik (1980) argued that Balinsky s (1970) experimental induction of ectopic limbs in the inter-limb trunk of salamanders was consistent with predictions of the LFH (Jarvik, 1980:115). Yonei-Tamura and colleagues (Yonei-Tamura et al., 1999; Yonei-Tamura et al., 2008) used similar ectopic appendage data in chick we well as molecular data in zebrafish to conclude that flank competence for fin/limb induction was likely a primitive feature of the gnathostome body plan and may have facilitated the evolutionary transposition of appendages along the flank. In contrast to Jarvik, these authors point out that their interpretation is consistent with fossil data supporting a pectoral-origin of paired fins, and does not require the presence of an ancestral lateral fin-fold per se. Generative Homologies and the Origin of Paired Fins The problematic classical hypotheses described above are direct statements of anatomical homology, i.e., paired fins are evolutionarily transformed gill arches or remnants of an ancestral lateral fin-fold. Recent work exploring the origin of the vertebrate appendicular system has shifted focus to generative or deep homologies (Shubin et al., 1997). Unlike anatomical homology, generative homology does not require the historical continuity of particular structures, but is instead diagnosed by the use of common genetic machinery. The camera-eyes of jellyfish, squid, and humans, the paired appendages of insects and vertebrates, and the legs and horns of beetles provide now classical examples of this (Shubin et al., 1997; Moczek et al., 2006; Shubin et al., 2009). 64

76 During embryonic development, cell populations are patterned by the activation of hierarchically-organized genetic circuits, or gene regulatory networks (GRN s) (Arnone and Davidson, 1997; Davidson, 2006). In this complex process, a suite of trans-acting transcription factors bind cis-acting genomic DNA and influence the transcriptional response of the next gene within the network hierarchy. As new genes are turned on, the expression profile, or regulatory state, of a given cell changes, ultimately resulting in the production of proteins associated with differentiation (Davidson, 2006). The identification of networks shared during the development of two different traits support the generative homology of those traits and an evolutionary shift in the site of GRN function. Notably, such shifts can be heterotopic (to a new site) or heterochronic (to a new developmental time point). Studies of genetic redeployment rely on three primary lines of evidence (see also Monteiro, 2012). The first is the expression of the same patterning genes in different developmental contexts. Although gene expression provides the simplest test for predictions of network cooption, these data alone cannot be used to differentiate between generative homology or similarity due to convergent evolution. The second line of evidence comes from gene knockdown and mis-expression experiments. These data demonstrate the conservation of epistatic regulatory interactions between genes within a pathway, and can be used to argue against the convergent evolution of shared gene expression based on parsimony. The third line of evidence comes from analyses of cisregulatory elements of genomic DNA. For example, the ability of a common 65

77 genomic enhancer to drive reporter expression in different developmental contexts indicates a regulatory linkage between that enhancer element, the upstream transacting factors that bind to it, and the target gene it normally activates. It is this type of data, together with gene-knockdown and cis-regulatory assays, that are used to map GRN circuitry. Although data describing genetic interactions or the functionality of shared regulatory sequences provide the more rigorous and specific tests of gene network heterotopy, these data sets are not always obtainable, particularly in non-model organisms. Given that it was the morphological similarities between gill arches, median fins, and paired fins in certain fishes that inspired classical hypotheses for the origin of the appendicular system, it is reasonable to predict that common GRN components may underlie aspects of the development of these structures (Mabee et al., 2002; Freitas et al., 2006; Gillis et al., 2009). In this section I summarize published observations that provide support for this prediction in the form of shared gene expression and conserved regulatory interactions. Remarkably, these signatures of potential GRN co-option support generative homologies predicted by both the gillarch and lateral fin-fold hypotheses, despite their mutually exclusive statements of anatomical homology. Common patterning genes are expressed in median and paired fins As mentioned above, median fins develop from continuous folds of ectoderm and mesenchyme in extant anamniotes. Gene expression data and lineage analyses in 66

78 lamprey (Freitas et al., 2006; Haming et al., 2011; Tulenko et al., 2013), catshark (Freitas et al., 2006), teleosts (Smith et al., 1994; Kague et al., 2012; but see Shimada et al., 2013), and axolotl (Sobkow et al., 2006; Epperlein et al., 2007) support a dual embryonic origin of this mesenchyme from somites and neural crest. In contrast, the paired fins and limbs of jawed vertebrates develop as outgrowths of the somatopleure, a composite tissue consisting of somatic lateral plate mesoderm and overlying ectoderm. Somitic muscle precursors then infiltrate these buds to form appendicular musculature. Thus, the evolutionary co-option of median fin GRN s to pattern paired fins likely would have involved a heterotopic shift in expression from somitic and neural crest cells to lateral plate mesoderm (Mabee et al. 2002, Freitas et al, 2006). Studies in amniotes demonstrate that Tbx18 and HoxD genes function in limb positioning along the flank (Cohn et al., 1997; Tanaka and Tickle, 2004). Work by Freitas et al. (2006) has shown that in catsharks Tbx18 and HoxD9 13 also regionalize median fin-folds along the anterior-posterior axis, marking sites of dorsal and anal fin formation (Freitas et al, 2006). Furthermore, median fin buds exhibit nested, collinear expression of HoxD orthologues following fin-fold regression, similar to early phase HoxD expression in paired fins and limbs (Davis et al., 2007; Freitas et al., 2007). Lampreys are agnathan vertebrates that split from other vertebrates prior to the origin of the appendicular system (see below) (Janvier, 1996; Gess et al., 2006). In lamprey embryos, a median fin-fold extends along the length of the trunk, and expands posteriorly where distinct, post-metamorphic dorsal fins develop in adults (Freitas et al, 2006). Expression of Tbx15/18 and collinear 67

79 expression of Hox9y and Hox10w mark this posterior expansion, suggesting that a similar program for specifying median fin position may operate in both lamprey and catshark. Since lamprey display the primitive condition of possessing median fins in the absence of paired fins (Coates, 1994; Donoghue et al., 2000), the fin-patterning roles of Hox and Tbx genes likely arose along the midline of early vertebrates and then were redeployed to the flank, facilitating the evolution of paired fins (Freitas et al., 2006). Ephrins and their receptors provide another example of common genes expressed during the development of both paired appendages and median fins (see Holder and Klein, 1999 and Kullander and Klein, 2002 for review of ephrin-mediated signaling). In chick the receptor EphA4 exhibits two phases of expression, an early phase in the distal limb bud and a later phase along tendons within the limb (Patel et al., 1996). In catshark embryos, EphA4 displays similar biphasic expression during both paired and median fin ontogeny, initiating first in the fin buds and then localizing to the fin radials (Freitas and Cohn, 2004). It is worth noting that in mice, HoxA13 and HoxD13 have been shown to bind to the promoter region of EphA7, directly regulating its expression during limb development (Salsi and Zappavigna, 2006). In paired fins and median fins, EphA4 exhibits a similar overlap in its expression domain with Abd-B group Hox genes (Freitas and Cohn, 2004; Freitas et al., 2006; Freitas et al., 2007), and it is intriguing to speculate that the direct binding interaction observed in limbs may operate in fins as well. 68

80 During early limb development, the apical ectodermal ridge (AER) forms as a thickening of the ectoderm at the distal tip of the limb bud (Saunders, 1948). This structure expresses a battery of genes important for limb bud outgrowth and patterning along the proximo-distal axis (Capdevila and Belmonte, 2001). In zebrafish, a similar ectodermal thickening forms during the development of both the paired fins and the median fin-fold (MFF) (Dane and Tucker, 1985; Grandel and Schulte-Merker, 1998; Abe et al., 2007; Mercader, 2007). This morphological similarity is accompanied by the expression of common transcriptions factors [Dlx: (Panganiban and Rubenstein, 2002; Robledo et al., 2002; Akimeno et al., 1994; Ellies et al., 1997), Gbx: (Niss and Leutz, 1998; Su and Meng, 2002; Rhinn et al, 2003, Msx: Akimenko et al., 1995; Yonei-Tamura et al., 1999), sp: (Kawakami et al., 2004; Abe et la., 2007)], FGF signaling ligands (Furthauer et al,. 1997; Reifers et al, 1998; Yonei-Tamura et al., 1999; Ng et al., 2002; Draper et al, 2003; Fischer et al, 2003; Abe et al., 2007), and the extracellular matrix protein Lamininα5 (Webb et al., 2007). Together, these morphological and genetic parallels between the distal tips of paired and median fins suggest similar ontogenetic processes underlie their formation, consistent with predictions of GRN co-option and generative homology. Gene regulatory interactions are conserved between median and paired fins Limbs have long been used as a model for studying developmental patterning along distinct axes, and a considerable amount is known about the GRN s underlying their formation (Capdevila and Belmonte, 2001; Zeller et al., 2009; Duboc and 69

81 Logan, 2011; Rabinowitz and Vokes, 2012). Few works, however, characterize the genetic pathways that pattern median fins, thus preventing a broad scale comparison of median fin and limb GRN architecture. Several recent functional studies, however, provide evidence that at least some of the regulatory interactions governing paired and median fin development are conserved. Below I present two examples, one involving FGF-mediated signaling, and the other Shh-mediated signaling. It is well-established that FGF10 from the lateral plate mesoderm is critical for limb formation, and is an upstream regulator of several genes expressed in the AER, including FGF8 and sp9 (Min et al., 1998; Sekine et al., 1999; Capdevila and Belmonte, 2001; Kawakami et al., 2004). Furthermore, the application of exogenous FGF s, including FGF10, is sufficient to extend the AER and initiate the formation of complete ectopic limbs in the flank (Cohn et al., 1995; Crossley et al., 1996; Vogel et al., 1996; Cohn et al., 1997; Ohuchi et al., 1997; Yonei-Tamura et al., 1999). This inductive capacity is conserved during median fin development, as FGF-soaked beads have been shown to extend median fin-folds anteriorly in zebrafish (Abe et al., 2007), and induce ectopic median fin-like structures in chick, mouse, and turtle (Yonei- Tamura et al., 1999; Yonei-Tamura et al., 2008). Additionally, aspects of gene regulation appear to be conserved, as analyses in zebrafish and chick have shown that FGF10 is sufficient to upregulate sp9, fgf8, and fgf24 in ectopic median fin-folds (Abe et al, 2007; Yonei-Tamura et al., 1999). Together, these data indicate that FGF10 can initiate ontogenetic cascades for limb and median fin formation alike, and at least some the genetic interactions that comprise these cascades are conserved. 70

82 The secreted signaling factor Sonic hedgehog (Shh) is expressed posteriorly in the limb buds of tetrapods and the paired fin buds of fishes, patterning these appendages along the anterior-posterior axis (Riddle et al., 1993; Capdevila and Belmonte, 2001; Dahn et al., 2007). In tetrapods, zebrafish, and skates, exogenous retinoic acid (RA) has been shown to induce anterior expression of Shh and mirror image duplications of limb and paired fin skeletal elements (Krauss et al., 1993; Riddle et al., 1993; Akimenko and Ekker, 1995; Hoffman et al., 2002; Dahn et al., 2007). This appears to be the case during median fin development as well. In developing skates and sharks, Shh polarizes the dorsal fins, and the administration of RA induces ectopic Shh expression and mirror image duplications of dorsal fin radials (Dahn et al., 2007). Interestingly, the median fins of zebrafish do not appear to express Shh or other Hedgehog paralogues (Hadzhiev et al., 2007). The primordium that gives rise to the caudal fin, however, expresses mediators of Shh signaling, including patched, smoothened, gli3, and you. Additionally, mutant lines that lack functional Shh (syu -/- ), its co-receptor smoothened (smu -/- ), or the posterior notochord and floorplate (flh -/- ), both of which express Shh, exhibit caudal fin patterning defects (Hadzhievv et al., 2007). These observations provide an example of the requirement for Shh signaling during the formation of an unpaired fin in zebrafish, and suggest that midline structures may be the source of Shh to nearby fin primordia. The apparent absence of Shh during median fin formation in zebrafish may reflect a derived condition in teleosts, and comparative work with a basal actinopterygian such as paddlefish or gar will undoubtedly help clarify this issue. 71

83 Common patterning genes are expressed in branchial arches and paired fins The early developmental morphology of branchial arches and limbs is similar in that both initiate as buds of mesenchyme and overlying epithelium. Their embryonic origins, however, differ branchial arches derive from neural crest, whereas the bones of the limbs derive from lateral plate. Similar to median fins, branchial arches and paired fins express common patterning genes. For example, a screen for transcripts enriched in the developing branchial region of mice identified 40 genes also expressed in the limb buds (Fowles et al., 2003). Additional evidence for the developmental linkage between branchial and appendicular structures comes from human dysmorphology syndromes. These syndromes are marked by defects in the limbs as well as the portions of the cranio-facial region that derive from branchial arches. In some cases, the genetic changes underlying syndrome pathology have been identified, and include members of the FGF, Shh and BMP signaling pathways (Francis-West et al., 1998; Schneider et al., 1999). Gene regulatory interactions are conserved between gill arches and paired fins Chondrichthyans are a key group for exploring gene network heterotopy between the branchial region and paired fins because of the primitive nature of their branchial anatomy relative to the highly derived pharynx of amniotes (Gillis et al., 2009). In skates and sharks, gill rays extend outwards from the posterior margins of the hyoid and branchial arches, forming part of the gill-support basket (Gillis et al., 2009; Gillis et al., 2011). Working with skate embryos, Gillis et al (2009) 72

84 demonstrated striking parallels between the ontogeny of these structures and fin endoskeletons. As described above, RA is an upstream regulator of Shh, which polarizes median fins, paired fins, and limbs along the antero-posterior axis. In skates, Shh and its receptor patched are expressed posteriorly in the developing gill arches, marking the area of gill ray outgrowth and articulation. Similar to its effect on fins, exogenous RA induces ectopic Shh anteriorly in gill arches and mirror-image duplications of gill rays (Gillis et al., 2009). A second parallel is provided by a regulatory loop involving Shh and FGF8. During limb development, Shh from the zone of polarizing activity participates in a reciprocally inductive feedback loop with FGF8 in the AER to drive limb bud outgrowth (Laufer et al., 1994; Niswander et al., 1994; Lewandoski et al., 2000). In developing gill arches, local knockdown of FGF signaling causes a down-regulation of Shh, and the loss of gill rays at the site of inhibition. The opposite experiment confirms the interdependent nature of the Shh- FGF loop, as local knockdown of Shh results in the loss of FGF8 and gill rays. It is worth noting that the evolutionary loss of posterior gill rays in chimaeras, the sister group to sharks, rays, and skates, correlates with a downregulation of Shh signaling during gill arch development (Gillis et al., 2011). This comparative data complements the knockdown experiments described above, providing additional support for the patterning role of Shh in the formation of the branchial skeleton of chondrichthyans. Collectively, the examples presented above provide evidence that paired fins, median fins, and gill arches are patterned by similar genetic cascades. These observations are not only consistent with hypotheses of gene network co-option 73

85 during early paired fin evolution, but also the more general idea that novel structures can evolve through the redeployment of pre-existing genetic circuits, rather than the de novo assembly of new networks from scratch (Erwin and Davidson, 2009). On the surface, it seems paradoxical to stress generative homology and developmental similarity in studies of morphological change. After all, gill arches, median fins, and paired fins are morphologically unique in adults. However, identifying this level of conservation provides an important jumping off point for exploring how gene networks were redeployed into new embryonic contexts, and the relationship between subsequent genetic pathway divergence and morphological innovation (Erwin and Davidson, 2009). Clues to how gene network components were co-opted during the early evolution of the appendicular system may be provided by T-box family transcription factors. In gnathostomes, Tbx5 and Tbx4 are the earliest known genetic markers of the pectoral and pelvic fin-forming fields, respectively (Chapman et al., 1996; GibsonBrown et al., 1996; Ruvinsky et al., 2000; Ahn et al., 2002; Agarwal et al., 2003; Naiche and Papaioannou, 2003; Rallis et al., 2003; Duboc and Logan, 2011). Tbx5/4 orthologues are also expressed during heart development in gnathostomes and lamprey, as well as the contractile ventral vessels of amphioxus, a basal chordate (Horton et al., 2008; Kokubo et al., 2010; Onimaru et al., 2011). Together, these observations suggest that Tbx genes were redeployed in stem gnathostomes from an ancestral function in heart development to a novel role in paired fin initiation (Ruvinsky and Gibson-Brown, 2000; Horton et al., 2008; Duboc and Logan, 2011). 74

86 This heterotopic change was likely the result of cis-regulatory evolution, and the generation of a new enhancer, possibly a target of Hox genes, specific to the finforming field (Horton et al., 2008; Minguillon et al., 2009; Minguillon et al., 2012). Given the well-documented role of Tbx5/4 genes in paired fin/limb induction, and the conserved nature of paired fin, median fin, and gill arch patterning, it is reasonable to hypothesize that shifts in the site of Tbx5/4 activation were accompanied by the evolution of new linkages with genetic sub-circuits for appendage outgrowth. The FGF10-Shh-FGF8 signaling loop described above provides a particularly intriguing candidate for such a linkage, as Tbx5 directly binds an FGF10 enhancer initiating this loop during limb development (Rabinowitz and Vokes, 2012), and cis-regulatory changes that alter signaling ligand deployment are potent switches for GRN subcircuit activation (Davidson, 2006; Erwin and Davidson, 2009). Future characterization and comparison of the GRN s underlying diverse appendages types will help clarify this issue and the balance between genetic conservation and divergence (e.g., see Schneider et al, 2012 for an example). Changes in gene network deployment are evident in developing embryos by their effects on the behavior of cells and tissues. The phenotypes produced by these changes can be modest or dramatic, depending on the embryonic context of gene activation. For example, genetic sub-circuits redeployed early in development are more likely to have severe phenotypic consequences relative to those deployed later (Davidson and Erwin, 2006; Erwin and Davidson, 2009). Thus, the impact of genetic change is contingent on embryological context. Historically, comparative 75

87 embryology was at the center of discussions of paired fin origins (see above in Classical hypotheses for the origin of paired fins ), but such epigenetic-level phenomena have received relatively little attention in recent molecular studies. Integrating these data sets presents a key challenge in exploring the early evolution of the appendicular system, and the potential of small changes in cell behavior to alter developmental environment and facilitate morphological innovation. Embryonic Context and the Origin of Paired Appendages The fossil record and comparative anatomy indicate that gill arches, median fins, and a centralized heart predate the origin of paired fins. Given the underlying assumption that older structures were the ancestral sites of gene network function, this order of appearance can be used to predict the polarity of potential gene network co-option. Lateral plate mesoderm, the cell population that forms paired fins and limbs, was presumably the novel site of genetic redeployment in stem gnathostomes, and the body wall of these fossil taxa would have provided the embryonic context for the early evolution of the appendicular system. As mentioned above, lamprey are agnathan vertebrates that split from the lineage leading to gnathostomes prior to the origin of paired appendages. The degenerate body plan of post-metamorphic adults is specialized for a parasitic lifestyle, and has remained virtually unchanged since the late Devonian (approximately 360mya) (Gess et al., 2006). The degree of this specialization and 76

88 whether or not the developmental morphologies of lamprey embryos reflect those of ancestral vertebrates is a matter of continuing discussion (Hardisty, 1971; Shu et al., 2003; Shimeld and Donoghue, 2012). Future work with hagfish, the only other extant lineage of agnathan vertebrates, will help clarify this issue (see Donoghue et al., 2000; Donoghue et al, 2001; Delarbe et al., 2002; Forey and Janvier, 1993; Kuraku et al., 1999; Takezaki et al., 2003; Kuraku et al, 2006; Kuraku et al, 2009; Janvier, 2007 for discussions of the relationship of hagfish and lamprey). However, it is worth noting that hagfish embryos are extremely difficult to obtain (Ota and Kuratani, 2006b; Ota and Kuratani, 2006a; Kuratani and Ota, 2008; Ota and Kuratani, 2008; Shimeld and Donoghue, 2012), and lamprey present the far more tractable system for studying agnathan development. Importantly, lamprey, as agnathan vertebrates, occupy a key phylogenetic position for testing hypotheses about the distribution of lateral plate mesoderm in stem gnathostomes, and the nature of the pre-appendicular body wall. Development of the mesoderm in lamprey Classical anatomical studies describe three histologically distinct populations of post-cranial mesoderm in early-stage lamprey embryos (Fig. 2A) (Goette, 1890; Hatta, 1901). Goette (1890) referred to these as the Mesomeren, Kopfnier, and Seittenplatten, which share a common arrangement with the paraxial somites, intermediate mesoderm, and lateral plate mesoderm of other vertebrates at the phylotypic stage. Damas (1944) suggested that the lateral plate mesoderm of lamprey 77

89 embryos is transiently segmented during development, however, no other works to date have been able to confirm this observation (Richardson and Wright, 2010; FJT personal observations). Recent studies extend classical descriptions of lamprey mesoderm and use modern molecular techniques to further characterize this population during early development. Similar to other vertebrates, lamprey somites compartmentalize as they mature, forming a myotome, dermomyotome and sclerotome. During myotome formation, medial cells in close proximity to the notochord express patched and prdm, implicating Shh signaling in their specification (Hammond et al., 2009). The dermomyotome, a layer of Pax3/7-positive cells superficial to the myotome (Maurer, 1906; Nakao, 1977; Devoto et al., 2006; Scaal and Wiegreffe, 2006; Tulenko et al., 2013), also appears to be a source of muscle precursors, as Pax3/7 expression overlaps with that of the lamprey myogenic regulatory factor MRF-A (Kusakabe and Kuratani, 2005; Kusakabe et al., 2011). This overlap is particularly striking in the dorsal and ventral extremes of the dermomyotome, which have been shown to form epithelial lips (Kusakabe et al., 2011; Tulenko et al., 2013). It is worth noting that unlike gnathostomes, the lamprey myotome lacks a horizontal septum. However, it is dorso-ventrally regionalized by the expression of distinct isoforms of myosin heavy chain (Kusakabe et al., 2005), as well as the transcription factor zic1, a marker of epaxial musculature in teleosts (Ohtsuka et al., 2004; Kusakabe et al., 2011). The sclerotome extends medially from the somites to invest the neural tube and notochord (Maurer, 1906; Scaal and Wiegreffe, 2006), and has been visualized by the 78

90 expression of Parascleraxis (Freitas et al, 2006), as well as the chondrogenic markers Sox9 and Col2α (Zhang et al., 2006). Less work has been done to characterize the intermediate and lateral plate mesoderm of lamprey embryos. The earliest functional kidney, the pronephros, develops from intermediate mesoderm over the anterior yolk ball, and expresses the nephrogenic marker Pax2/5/8 (McCauley and Bronner-Fraser, 2002). Caudally, the pronephros empties into a colleting duct (the archinephric duct), which has been visualized by BMP2/4 as it extends towards the presumptive cloacal region (McCauley and Bronner-Fraser, 2004). The lateral plate lies lateral to the intermediate mesoderm and expresses the transcription factor Hand, a wellestablished marker of LPM in gnathostomes (Angelo et al., 2000; Fernandez-Teran et al., 2000; Tanaka et al., 2002; Kuraku et al., 2010; Onimaru et al., 2011; Tulenko et al., 2013). Although Hand is expressed along the length of the lamprey LPM, this population is regionalized by Tbx4/5 and Tbx20 expression just caudal to the branchial basket, marking the site of heart formation (Kokubo et al., 2010; Onimaru et al., 2011). PitxA is also expressed in the cardiogenic lpm, and is specific to the left-side, consistent with its role in regulating asymmetric organ development in gnathostomes (Mercola and Levin, 2001; Boorman and Shimeld, 2002a; Boorman and Shimeld, 2002b; Kokubo et al., 2010). Interestingly, Hox paralogues exhibit offset expression boundaries in the LPM (Takio et al., 2007; Onimaru et al., 2011), but the significance of their staggered arrangement is unclear. 79

91 Somatopleure persistence and the origin of paired fins A comparison of body wall formation between extant agnathans, chondrichthyans, and bony fishes is critical for evolutionary inferences about the nature of the primitive gnathostome trunk and the embryonic context in which paired fins first appeared. Few studies, however, have directly examined body wall formation in lamprey. In his classical monograph on the embryology of the river lamprey Lampetra fluviatilis, Goette (1890) partially addressed this issue using histological data to speculate on the fate of the LPM during muscular body wall formation. According to Goette s color-coded illustrations (Fig. 2), cells of the lateral plate extend around the gut tube prior to ventral myotome extension, and then split into distinct somatic and splanchnic layers. The somatic layer remains medial to the myotome, forming the parietal coelomic lining, whereas the splanchnic layer differentiates into the visceral coelomic lining, mesentery, and gut vasculature. In this thesis I have provided histological, molecular, and fate-mapping evidence confirming Goette s predictions (see pp.14 58). For instance, in early stage embryos the transcription factor HandA is expressed in the lateral plate mesoderm, which sits atop the dorsum of the yolk ball. As development proceeds, the expression domain of HandA extends ventrally to surround the yolk tube, consistent with an early migration of LPM prior to myotome extension. Subsequent down-regulation of HandA prevents its use as a proxy for following the distribution of LPM into later developmental stages. In order to generate the first long-term fate maps of somitic and lateral plate mesoderm in lamprey, I performed a series of vital dye injections 80

92 targeting these populations. Consistent with changes in HandA expression, cells of the LPM migrate early to invest the margins of the yolk tube. The ventral lip of the dermomyotome forms the leading edge of the somitic population during body wall closure, and no mesenchymal myoblasts migrate out of this epithelium to mix with cells of the lateral plate. As the somitic population extends ventrally, the lateral plate mesoderm is separated from the ectoderm and displaced medially, contributing to the coelomic linings, gut vasculature, and typhlosole. Thus, the somatopleure, a composite tissue composed of somatic lateral plate and ectoderm, is transiently present during the development of lamprey, but is eliminated as the LPM is sequestered to the inner body wall only. In amniotes, it is well-established that in addition to forming the limbs, somatic lateral plate contributes connective tissue to both the musculature and dermis of the flank, reflecting persistence of the somatopleure in the adult inter-limb bodywall (Nowicki et al., 2003; Durland et al., 2008). Thus, the developmental role of the somatopleure differs strikingly between agnathans and amniotes, which may not be surprising given the phylogenetic distance between these groups. In this thesis, I have also examined the distribution of post-cranial mesoderm in a representative amphibian (axolotl) and chondrichthyan (catshark) using tissue transplantation and immunohistochemistry, respectively. Importantly, cells of the lateral plate appear to be retained in the trunk body wall between the ectoderm and somitic myotome in both taxa, similar to published observations for amniotes. If the developmental morphologies of lamprey reflect those of ancestral vertebrates, the persistence of the 81

93 somatopleure in the lateral body wall is likely a derived feature of gnathostomes. This idea has implications for the evolutionary origin of the appendicular system, as redistribution of the LPM in stem gnathostomes would have been a key early step in generating a novel fin-forming field in the flank. The Birth of the Abaxial Domain As mentioned in the introduction of this thesis, mesodermal context has been used to define two domains during the development of the vertebrate musculoskeletal system (Burke and Nowicki, 2003; Nowicki et al., 2003; Shearman and Burke, 2009). In the primaxial domain, somitic cells differentiate in an entirely somitic environment, whereas in the abaxial domain, lateral plate and migratory somitic cells differentiate in the context of lateral plate connective tissue. The interface between these domains is the lateral somitic frontier. Fate-mapping studies in amniotes reveal that the primaxial and abaxial domains segregate in large part with the axial and appendicular systems, respectively (Burke and Nowicki, 2003; Nowicki et al., 2003; Durland et al., 2008; Shearman and Burke, 2009; Shearman et al., 2011). The ventrolateral dermis and the muscles of inter-limb flank, however, are abaxial. Thus the abaxial domain, most prominently expanded at the levels of the limbs, extends along the length of the inter-limb trunk. Studies of gene regulation, mutant phenotypes, and heterotopic graft transplants support the developmental modularity of the primaxial and abaxial domains, fueling hypotheses that they also may function as sites of 82

94 independent evolutionary change (e.g., Burke and Nowicki, 2003, Winslow et al., 2007, Shearman and Burke, 2009; Buchholz et al, 2012; Johansen, 2010). The muscular body wall of lamprey is entirely primaxial, and this likely represents the primitive condition for vertebrates. I propose that retention of somatic lateral plate mesoderm in the flank was a simple first step in the evolution of the abaxial domain. Initially this new, naïve field of cells would have experienced relaxed selective constraint, conceptually similar to a newly duplicated gene. This would have facilitated the co-option of pre-existing GRN s and the formation of novel abaxial morphologies, including, ultimately, paired fins and limbs. In contrast to classical hypotheses proposing the transformation of precursor structures (e.g., gill arches or a lateral fin-fold), this hypothesis for the origin of the abaxial domain focuses on early shifts in cell populations, and the potential of such shifts to create a novel developmental module with evolutionary potential. The tissue-level changes described in this thesis raise a new line of inquiry for exploring the origin of paired appendages. What are the mechanistic underpinnings of somatopleure retention in different vertebrate taxa, and can comparing these developmental data be used to make inferences about the evolutionary origin of the abaxial domain? Intriguingly, the signaling role of the intermediate mesoderm (IM) in fin and limb induction, and the alignment of the nephric duct with the persistent somatopleure in sharks, may provide the first clues to tackling these challenging and exciting questions. 83

95 The IM, or its early derivative, the pronephros, abuts the paired fin-forming field in representative chondrichthyans (Balfour, 1878), actinopterygians (Ballard and Needham, 1964; Bemis and Grande, 1992; Neto et al., 2012), amphibians (Abu-Daya et al., 2011), and amniotes (Capdevila and Belmonte, 2001), suggesting that this condition may be ancestral for gnathostomes. Several studies now implicate the IM as a critical signaling center for pectoral fin or limb development, supporting the functional significance of this tissue relationship. For instance, in zebrafish the transcription factors osr1 and osr2, as well as the signaling factor Wnt2b, are coexpressed in the IM adjacent to the pectoral fin-field and are required for fin formation (Neto et al., 2012). Fibin, a secreted protein expressed in a thin stripe between the IM and pectoral fin-field, is also necessary for fin formation, and is likely a downstream target of IM Wnt2b (Wakahara et al., 2007). Another example is provided by the integrin-ligand nephronectin. In metamorphic Xenopus froglets, nephronectin is expressed in the pronephros directly adjacent to the forelimb-forming field. When nephronectin is knocked down by either transgene insertion or morpholino injection, forelimbs fail to form (Abu-Daya et al, 2012). In amniotes, the role of the intermediate mesoderm in limb induction is also an area of active research. Although a complete summary of this work is beyond the scope of this review, it is worth noting that in chick the potent limb inducer FGF8 is expressed in the IM at fore- and hind-limb levels (Crossley et al., 1996; Vogel et al., 1996; Capdevila and Belmonte, 2001), and surgical interruption of the IM causes limb reduction (Stephens 84

96 and McNulty, 1981; Geduspan and Solursh, 1992, but see also Fernandez-Teran et al., 1997). These examples, taken collectively, are not intended to make specific inferences about the evolution of limb inductive pathways, but rather to highlight the importance of signaling between the IM and LPM during the development of abaxial structures in disparate gnathostome lineages. In catsharks, the segmental duct of the nephric system aligns with the pectoral fins anteriorly, and extends along the lateral hinge of the body cavity towards the developing cloaca (Fig. 3). This is also the precise position of the lateral somitic frontier and the persistent somatopleure at both fin and inter-fin levels (see page 41, Fig. 5 of this thesis). These observations fuel two related developmental hypotheses. In the first, derivatives of the IM mesoderm function in initiating paired fin formation in sharks, similar to other gnathostome taxa. In the second, signaling interactions between the segmental duct and adjacent tissues direct the formation of the abaxial domain at both fin and flank levels. Advances in techniques for manipulating chondrichthyan embryos will facilitate future work testing these hypotheses. The pectoral and pelvic fins of extant gnathostomes represent dramatic elaborations of the abaxial domain. The inter-fin trunk, however, may better reflect the nature of the primitive abaxial body wall prior to the advent of paired fins, i.e., stem gnathostomes likely first transitioned from a lamprey-like primaxial body wall to a shark-like abaxial body wall before the elaboration of appendicular structures. Given the role of the IM in the induction of paired fins and limbs, and the close 85

97 proximity of nephric structures with the persistent somatopleure in sharks, it is intriguing to speculate that the IM may have had an ancestral role in the formation of the pre-appendicular, abaxial body wall. For instance, a novel epithelialmesenchymal interaction in the flank of stem gnathostomes, provoked by signaling from the IM, could have resulted in early proliferation of lateral plate mesoderm, similar to the first steps of median fin, paired fin, and limb development (see also Wake, 1979; Oster and Albrech, 1981; Burke 1989; Burke, 1991, for epithelialmesenchymal interactions in development and evolution). In this scenario, expansion of the lateral plate would have altered the developmental environment encountered by somitic myotomes during body wall closure, resulting in retention of LPM in the lateral body wall. Although direct tests of this hypothesis are not possible because lineages intermediate to lamprey and chondrichthyans are long extinct, experiments with lamprey embryos can be used to assay the effects of lateral plate expansion on the organization of body wall mesoderm. For example, the application of exogenous FGF s (or even the transplantation of cranial placodes) to the flank of developing lamprey embryos can be used test the proliferative capacity of lateral plate cells, and whether an expansion of this population phenocopies the abaxial inter-fin body wall of sharks. How vertebrate paired fins first arose over 400 million years ago is one of the long-standing questions in comparative anatomy. The data presented here support the argument that the ancestral post-cranial body plan of vertebrates was primaxial, and that persistence of somatic lateral plate in the body wall of stem gnathostomes was a 86

98 key ontogenetic change fueling the birth of a new abaxial module and site of gene network redeployment. The molecular mechanisms that establish a persistent somatopleure in different taxa remain open questions, and provide a fertile ground for future research. Exploring these questions will enrich our understanding of how small changes in ontogeny can facilitate morphological evolution and anatomical invention. 87

99 Figure 1. Classical hypotheses for the origin of paired fins in verterbates. A. Gill arch hypothesis. According to Gegenbaur s proposed transformational series gill arches are homologues of the scapulacoracoid (pectoral girdle), and gill rays homologues of the bones of the paired fins. (Modified from Gegenbaur, 1876, 1878). B D. Lateral fin-fold hypothesis. B. Pectoral and pelvic fins (and the fore- and hindlimb that derive from them) are retained margins of a hypothesized continuous lateral fin-fold similar to the median fin-fold of anamniote embryos. (Modified from Widersheim and Parker, 1907). C. Hypothesized illustration of a cross-section through the lateral fin-fold (left) from Jarvik (1980). Muscle has been pseudo-colored pink for clarity. Somatic lateral plate mesoderm would have filled the fin-fold (right, top), ultimately giving rise to a fin-support girdle and endoskeleton (right, bottom). D. Hypothesized illustration of the lateral fin-fold endoskeleton from Thacher (1877). Proximal elements (top) near the body wall would have fused to form the fin support girdle (bottom). (Modified from Thacher, 1877). 88

100 89

101 Figure 2. Illustrations of cross sections of Lampetra fluviatilis from Goette (1890). A. In early stage embryos the Mesomeren (mes), Kopfnier (kn ), and Seittenplatten (spl) correspond to the somites, intermediate mesoderm, and lateral plate mesoderm, respectively. B. Distinct somitic compartments are apparent and include the Aufsenplatte der Mesomeren (ap) and Muskelplatte der Mesomeren (mp), which correspond to the dermomyotome and myotome, respectively. C,D. In older embryos and larvae, the spl and its derivatives are dark brown, the interstitial tissues light brown, the veins blue, and the arteries red. The spl has extended to invest the developing gut tube and split into distinct somatic and splanchnic layers. The body cavity, Leibeshöhle (lh), is beginning to form in C and is complete by D. ao, aorta; dl, linke Darmlebervene: left intestinal hepatic vein; dl rechte Darmlebervene: right intestinal hepatic vein; sch, Subchordalstrang: Subchordal strand; st,rechte Stammvene: right vein; sv, Subintestinalvene: Subintestinal vein; z, unterer Rand der Muskelplatte: lower edge of myotome. 90

102 91

103 Figure. 3. Illustrations of Scyliorhinus canicula embryos from Balfour (1878). A. Whole mount of an embryo shown in lateral view. B. Cross section through the abdominal region. Approximate plane of section is shown in A. The Muscle plate (mp) (pink) corresponds to the myotome and overlying dermomyotome, which directly contact the ectoderm dorsally (above black arrowhead), but become separated from the ectoderm at level of segmental duct (sd) and dorsal hinge of the body cavity. I interpret the interface of the free ventral margin of the mp as the position of the lateral somitic frontier (red dotted line on right). Note that pink shading, black arrowheads, and the red dotted line have been added to Balfour s original illustration for clarity. 92

104 93

105 CHAPTER 3: 3D Reconstructions of quail-chick chimeras provide a new fate map of the avian scapula The majority of this chapter has been published in Developmental Biology (Shearman et al, 2011), and was co-first authored with R.M. Shearman, Ph.D. SUMMARY Limbed vertebrates have functionally integrated postcranial axial and appendicular systems derived from two distinct populations of embryonic mesoderm. The axial skeletal elements arise from the paraxial somites, the appendicular skeleton and sternum arise from the somatic lateral plate mesoderm, and all of the muscles for both systems arise from the somites. Recent studies in amniotes demonstrate that the scapula has a mixed mesodermal origin. Here we determine the relative contribution of somitic and lateral plate mesoderm to the avian scapula from chick/quail chimeras. We generate 3D reconstructions of the grafted tissue in the host revealing a very different distribution of somitic cells in the scapula than previously reported. This novel 3D visualization of the cryptic border between somitic and lateral plate populations reveals the dynamics of musculoskeletal morphogenesis and demonstrates the importance of 3D visualization of chimera data. Reconstructions of chimeras make clear three significant contrasts with existing models of scapular development. First, the majority of the avian scapula is lateral plate derived and the 94

106 somitic contribution to the scapular blade is significantly smaller than in previous models. Second, the segmentation of the somitic component of the blade is partially lost; and third, there are striking differences in growth rates between different tissues derived from the same somites that contribute to the structures of the cervical thoracic transition, including the scapula. These data call for the reassessment of theories on the development, homology, and evolution of the vertebrate scapula. INTRODUCTION During more than 500 million years of evolution, the vertebrate musculoskeletal system has adapted to aquatic, terrestrial, fossorial, and arboreal lifestyles, yet many aspects of musculoskeletal development remain highly conserved (Goodrich, 1930). All limbed vertebrates have a postcranial axial system consisting of vertebrae, ribs and associated muscles that are functionally integrated with an appendicular system comprised of the muscles and bones of the paired appendages and their respective girdles. Two distinct populations of embryonic mesoderm provide the cells for axial and appendicular systems. The axial skeletal elements arise from the paraxial somites, the appendicular skeleton and sternum arise from the somatic lateral plate mesoderm, and all of the muscles of both systems arise from the somites. The pectoral girdle bridges the somite-derived axial column with the lateral plate-derived appendicular system, and recent studies in amniotes demonstrate that the scapula has a mixed mesodermal origin (Huang et al., 2000b; Durland et al., 2008; Valasek et al., 2010; Valasek et al., 2011). 95

107 During gastrulation the mesoderm segregates into spatially distinct paraxial, intermediate and lateral plate populations (Selleck and Stern, 1991; Schoenwolf et al., 1992). The somitic population has been especially well studied, and much is known about patterns and mechanisms of somitic cell differentiation (Christ et al., 1977; Ordahl and Ledouarin, 1992; Stockdale et al., 2000; Bothe et al., 2007; Christ et al., 2007). The intermediate mesoderm sinks inward, bringing the somitic and lateral plate populations together to form a continuous and uniform seam along the anterior-posterior (AP) axis (incipient frontier in Fig. 1A). Over subsequent development, there is dramatic expansion of these populations. Our earlier work clarified the distribution of the somitic and lateral plate mesoderm by mapping the changing interface between these populations during early stages of body wall formation in chick (Nowicki et al., 2003) and mouse (Durland et al., 2008). We defined two embryonic domains based on the mesodermal lineage of the contributing cells (Burke and Nowicki, 2003 and Fig. 1B, C). The primaxial domain ultimately includes the vertebrae, ribs, and peri-vertebral and intercostal muscles as well as their investing connective tissue, all of which arise exclusively from somitic cells. The abaxial domain, which includes the limbs and ventrolateral aspects of the body wall, is made up from lateral plate as well as somitic cells that differentiate around or within lateral plate-derived connective tissue. The border between the abaxial and primaxial domains is the lateral somitic frontier (LSF). This is where the original border between paraxial and lateral plate mesoderm (the incipient frontier), is 96

108 dynamically altered during development (Fig. 1, reviewed in Shearman and Burke, 2009). In this paper we use quail-chick chimeras to determine the relative contribution of somitic and lateral plate mesoderm to the avian scapula (Fig. 1D), and present a new model of scapular development. This powerful technique was first introduced five decades ago by Nicole LeDouarin (Ledouari.N and Barq, 1969), and has provided extensive and detailed fate maps of the avian embryo. Transplanting somitic mesoderm between quail-donors and chick hosts, Chevallier (1977) was the first to demonstrate that somites contribute cells to the scapular cartilage in chick. Using a similar approach, Huang et al. (2000b) concluded that only the head and neck of the scapula are derived from lateral plate whereas the majority of the scapular blade arises from the somites, mirroring the segmental pattern along the axis (redrawn here in Fig. 9A). We demonstrate that 3D reconstructions of grafted tissue substantially enrich data generated by quail-chick chimeras, and we use this technique to visualize the lateral somitic frontier in the scapula and musculature of the shoulder. Our data reveal a very different distribution of somitic cells in the scapula than reported by Huang et al. (2000). Specifically, we show the majority of the scapula is derived from the lateral plate, and somites contribute only to the distal 1/3 of the blade. The new avian skeletal fate map presented here is more consistent with data from the mouse, but also exposes interesting differences in the origin and insertions of critical shoulder musculature. Furthermore, the visualization of the otherwise cryptic border defined as 97

109 the lateral somitic frontier reveals the dynamics of morphogenesis, including dramatic differential growth in primaxial versus abaxial domains. MATERIALS AND METHODS Embryos Fertilized chick eggs (Gallus gallus) were purchased from Charles River Labs (Franklin, CT) and quail eggs (Coturnix coturnix) were purchased from AA Labs Inc. (Westminster, CA). All eggs were incubated at 37 C and staged according to Hamburger and Hamilton (H&H; 1951). Segmental plate and Somatic lateral plate transplants We performed isochronic, isotopic segmental plate (SP) transplants at several positions along the anterior-posterior axis (Fig. 1D and Table 1). Donor segmental plate adjacent to the last developing somite was removed using 0.15% trypsin in calcium- and magnesium-free Tyrode s solution and a tungsten needle. Grafts were approximately four somites long, and marked for orientation with 0.1% methylene blue. Host embryos were prepared in ovo, and SP corresponding to the donor graft was removed using 3% Pancreatin in Ringer s and a tungsten needle. Grafts were transferred to the host using a micropipette, and inserted in proper orientation into the host using an eyebrow hair. A 1% solution of penicillin/streptomycin was administered to prevent infection. Eggs were sealed and incubated for 5 7 days. 98

110 We performed isotopic transplants of somatic lateral plate following techniques similar to those outlined for SP transplants. Grafts were created from donor somatic lateral plate tissue lateral to the three most posterior somites. Corresponding somatic lateral plate in the hosts was removed and the grafts were inserted into the chick embryos in proper orientation. Eggs were sealed and incubated for 5 7 days. Determination of graft boundaries One of our primary objectives was to determine the interface between somitic and lateral plate mesoderm at late stages of musculoskeletal morphogenesis. In order to ensure that the quail-chick boundaries are somite-lateral plate (So-LP) boundaries and not graft-edge boundaries, we use 3D reconstructions (see below) to map the position of donor cells in multiple transplants at different axial levels. Reconstructions allow us to visualize the complete distribution of grafted tissue within a chimera, and the exact position of each section within a given graft (Fig. 2). Because donor-host boundaries at the antero-posterior margins of a graft can confound true So-LP boundaries, we limit our analysis of So-LP boundaries to those sections in the heart of each graft. We clarify the distribution of somitic and lateral plate cells in the scapula by reconstructing grafts from a series of segmental plate (SP) transplants that collectively include all somite levels contributing to the scapula. Specifically, in embryos ranging from H&H stages 10 through 14, the transplants span the region of the AP axis extending from somite 12 through somite 24, though 99

111 each transplant is approximately 3 4 somite-lengths. Somites of embryos were counted and donors and hosts matched. Each chimera was numbered based on the next somite to form, i.e., the first somite expected to contain grafted tissue. We retrospectively determined the exact level of each graft based on the first vertebra derived from graft in Day 7 or 9 chimeras (approximately H&H stages 33 35). The predicted levels were consistent between specimens and with the known alignment of vertebral number and somite of origin (Gumpel-Pinot, 1984; Burke et al., 1995). We standardize our numbering to reflect the first and last quail-derived vertebra, e.g. SP: represents a chimera where the 13th cervical (C13) through 3rd thoracic (T3) vertebrae contain quail cells. Immunohistochemistry Chimeras were fixed in 4% paraformaldehyde for 24 hrs, dehydrated, embedded in paraffin, and sectioned at 20 microns using a HM 340 E microtome. After sections were de-waxed in citrisolv and rehydrated, antigen unmasking was performed by autoclaving slides in citrate buffer (ph 6). Quail cells were labeled with QcPN antibody (Developmental Studies Hybridoma Bank, Iowa City, IA) at a 1:1 dilution in a blocking solution of 10% horse serum in PBS. Sections were incubated in primary antibody overnight at 4 o C, and the signal was developed and amplified using the Vectastain Elite ABC Kit (PK-6200) following manufacturer s instructions (Vector Laboratories, Burlingame, CA). Sections were counter stained with eosin and alcian blue. Slides were cover slipped using Permount. 100

112 DAPI staining Coverslips on QcPN-labeled sections were removed by soaking in Cirtisolv overnight. Slides were hydrated and rinsed with PBS, and then bathed in a 300nM DAPI solution in PBS for five minutes followed by several washes in PBS. They were temporarily cover slipped with water and photographed using a Nikon Eclipse E600 compound scope equipped with a Prior Lumen 200 Illuminator. The DAB/Nickel precipitate used to visualize the QcPN is centered on the quail nuclei so quail cells are less fluorescent than the unlabeled chick cells, which fluoresce brightly after exposure to DAPI. In situ hybridization In situ hybridization of Sox9 on 8µm paraffin sections of chimeras was performed according to Moorman et al. (2001). The Sox9 RNA probe was constructed following Burke et al. (1995) using a Sox9 containing plasmid provided by the Tabin Laboratory (Harvard Medical School, Cambridge, MA). Hybridization was performed using hybridization chambers to minimize the volume of probe needed for the reaction and to prevent desiccation during incubation. The signal was developed using alkaline phosphatase conjugated anti-digoxigenin Fab fragments (Roche) 1:2000, and visualized with NBT/BCIP in NTMT. Sections were counter stained with eosin and cover slipped. Three dimensional reconstructions 101

113 Sections were photographed using a Nikon Eclipse E600 compound scope, Spot RT3 camera, and Spot Advanced Plus Software Version 4.7. Amira 4.0 software was used to digitally reconstruct the sectioned and stained chimeras in three dimensions. Color images of serial sections for each chimera were uploaded into Amira 4.0 and aligned with the AlignSlice module using the least squares algorithm. The number of sections ranged from Aligned images were resampled and converted to gray scale images using either the Castfield or ChannelWorks Modules. In each image, anatomical structures (e.g., scapula, vertebrae and specific muscle groups) were manually traced using the Segmentation Editor and stored in Label Fields. Once the anatomical elements were outlined, three alternative segmentation approaches were used to delimit regions of QcPN staining within those structures. In the first, pixels with grey values corresponding to those of DAB-labeled quail cells were defined using the wand tool and histogram in the Segmentation Editor. Once a specified range was selected, all voxels within that range were outlined automatically. Similarly, Adobe Photoshop CS5 was used to generate reconstructable image stacks in which pixels representing QcPN positive staining were specifically highlighted. In this approach, we used the select color range function to define a palette of several hundred colors corresponding to labeled graft. In areas of fainter antibody labeling, regions of QcPN staining were traced manually. Polygonal surface models of quail and chick Label Fields were generated using the SurfaceGen module. As an alternative to surface models, three-dimensional representations of QcPN positive cells in younger specimens were generated as direct volume renderings using the 102

114 Voltex module. In addition to surface modeling and volume rendering, sectional anatomy was visualized using the Amira Orthoslice module. Specifically, these tools were used to digitally render images of sectional anatomy in any plane of section and to visualize the distribution of quail and chick cells in the interior of modeled structures. RESULTS The position and number of grafts is summarized in Table 1. Survivability was influenced by the A-P position of the graft and incubation time. Embryos receiving SP grafts at somite levels had the highest rate of survivorship. Mortality was greatest in somatic lateral plate (LP) transplants. Regardless of the type of transplant, survivorship diminished as incubation time increased. We will first describe the quail/donor contributions to the skeletal elements after SP transplants from different AP levels as visualized by 3D reconstructions of Day 7 and Day 9 chimeras. We will then describe graft contributions to scapular muscles from Day 9 chimeras, as well as specific graft/host boundaries as revealed by individual sections in SP and LP transplants. Figure 2 illustrates the placement of individual sections within a reconstructed chimera. We also describe the expression of Sox9 mrna relative to the donor-host boundary within the scapular condensation at Day

115 Vertebral elements and scapula in segmental plate transplants: SP:12 17 chimeras (surgery at H&H 10 11, n=17) Isotopic, isochronic transplants of quail SP corresponding to somites or into host chicks result in chimeras with posterior cervical vertebrae and associated musculature derived from quail at Day 7 and Day 9. Three dimensional reconstructions of SP chimeras reveal that the penultimate cervical vertebra (C13) includes graft cells (Fig. 3A). However, transplants anterior to somite 18 never show quail cells in the scapular cartilage (Fig. 3A). SP chimeras (surgery at H&H 12, n=18) Quail segmental plate grafts that include somite 18 contribute to the posterior aspect of C13 and make a contribution to the anlagen of the scapular blade at Day 7. The quail contribution to the cartilage is minor and restricted to the dorsal edge of the blade (Fig. 3B). The quail cells in the blade occur at an AP level a full vertebral segment posterior to the proximal scapular head and are offset by more than a full segment from the first labeled vertebra (Fig. 3B). In chimeras at this level, there is extensive quail contribution to the dermis and muscles forming around the scapula (Fig. 3C, D). Reconstructions of chimeras fixed at 9 days (H&H 35) show more mature tissues and a very different distribution of quail cells in the scapula than specimens fixed at 7 days (H&H 32 33). The graft derived area of the scapular blade is larger in the older specimens, but is still restricted to the distal 1/3 of the element (Fig. 4A). 104

116 The boundary between quail and chick cells within the blade is attenuated from dorsal-proximal to ventral-distal, forming a long shallow border visible in the reconstructions. In chimeras that also have a posterior host/graft boundary in the scapula, this border is abrupt, creating a steep angle between quail and chick tissue along the DV aspect of the blade in contrast to the attenuated anterior graft boundary (Fig. 4A). Consistent with what is shown at Day 7, there is considerable offset between the AP position of quail cells along the scapular blade and the quail derived vertebrae. The quail contribution to the scapula lies 2 3 segments behind the first quail-derived vertebra (Fig. 4A). Collectively, SP surgeries including presumptive somites 12 21, indicate that somite 18, which contributes to C13 and C14, is the most anterior somite to contribute to the scapula. SP chimeras (surgery at H&H 13 14, n=3) Reconstruction of chimeras where quail somites were transplanted into chick and fixed at 9 days exhibit the same attenuation of the anterior donor-host boundary within the scapula (data not shown). Specimens also have an abrupt posterior graft boundary in the scapula and the very distal tip is chick derived. Thoracic vertebrae 2 4 are quail derived. SP chimeras (surgery at H&H 14+, n= 4) Quail segmental plate of presumptive somites contribute to thoracic vertebrae 3 6 (Fig. 4B). The graft also contributes to the most posterior portion of the 105

117 scapular blade, and there is no posterior border to the graft represented in the scapula. In these specimens, the anterior donor/host boundary in the scapula lies at a steep angle, in contrast to the shallow angle formed in the scapulae of SP specimens (Fig. 4A, B). The angle is similar to the posterior donor/host border observed in specimens with more anterior grafts (e.g. SP18 21). Also, fewer vertebral segments separate the anterior graft margin in the scapula from the anterior graft margin in the vertebral column in SP transplants relative to those of SP Together, the position of the graft within the scapula of SP20 23 and indicates that S24 is the posterior-most somite to contribute to the distal portion of the scapular blade. Muscles of the scapula in 3D reconstructions There are two groups of shoulder muscles that we distinguish at Day 9: the proximal muscles that connect the shoulder to the trunk; and the distal muscles that connect the shoulder to the forelimb. We describe the contribution of grafted cells to these muscles. Rhomboid Complex We find donor cells in the rhomboids of chimeras at each AP level, indicating that somites as far anterior as somite 12 contribute to this muscle. As shown in Figure 5 A C, the SP graft contributions to the RHC generally align with graft-derived vertebrae but often appear slightly anterior to the first graft-derived vertebra. This relative alignment of graft-derived muscle and bone seen between the RHC and the vertebra it arises from is not seen in the insertion of the rhomboids on the scapula. 106

118 The donor-host boundary at the insertion site of the RHC does not line up between muscle and bone (scapula). The anterior graft margin in the RHC contacts the scapula anterior to the cranial border of the donor cells in that element (Fig. 5A C). In contrast, the posterior margin of the graft in the RHC is always aligned with the posterior border of donor cells in the scapula when that boundary is present (Fig. 5A, B). As previously described, the graft-derived region of the scapula is offset posteriorly relative to the AP position of quail cells in the vertebral column. As a result, the anterior and posterior graft-borders in the RHC are attenuated along the AP axis (Fig. 5B). The attenuated anterior border between graft and host RHC is not present in specimens with anterior transplants (SP18 20) because the entire extent of the muscle is derived from quail (Fig. 5A). The position of the donor muscle relative to the graft within the vertebrae and scapula suggests that the contact between the RHC and graft derived region of scapula occurs prior to posterior displacement of donor cells within the scapula. Serratus Complex The pattern of quail cells in the SRC relative to the scapula is similar to, although not as extreme as, that observed in the RHC. The graft boundaries in this muscle are less attenuated along the AP axis, and line up well with the boundary in the scapula (Fig. 5D F). The donor SRC does not extend further posterior than the graft in the scapula. However, as observed in the RHC, donor cells in the SRC extend further anterior than the graft in the scapula in SP (Fig. 5D, E). 107

119 Scapulohumeralis In SP specimens, the myoblasts in the scapulohumeralis muscle are quail, and in SP specimens, they are entirely chick (data not shown). Interestingly, the donor-host boundary within this muscle is never easy to define, even in specimens that bridge SP This is likely do to the presence of both somitic (quail) and LP (chick) cells populating this muscle. Segregation of somitic and lateral plate mesoderm during shoulder development The mesenchyme that will form the scapular cartilage can be visualized by Sox 9 expression between developmental stages H&H (Days 5 6), before significant condensation has begun. We examined Sox 9 expression in alternate sections of SP chimeras at Day 5 to visualize the relationship of the somitic and lateral plate contributions at early stages of scapula specification (Fig. 6). In the proximal, glenoid region, adjacent sections labeled with QcPN show the exclusion of donor/somitic cells from the Sox 9-expressing region (Fig. 6A, B). In more posterior sections, the Sox 9 domain narrows to prefigure the scapular blade. Overlap of Sox 9 and QcPN can be seen beginning in the dorsomedial portion of the Sox 9-positive area (Fig. 6C, D). The overlap increases in more posterior sections, but does not cover more than ~ 1/3 of the cross-sectional area of the Sox9 expression domain (Fig. 5E, F). This indicates the dorsoventral segregation of somitic and lateral plate contributions to the scapular blade is already established within the Sox9 expression 108

120 domain that preconfigures the scapular condensation. At Day 5, SP18 21 quail contributions to cells within the Sox9 expression domain appear to be present at the level of the presumptive C14/T1 boundary. This is consistent with observed SP18 21 quail contributions to the scapular blade at Day 7 when the scapular cartilage is well defined. Individual sections (Fig. 7A D) from the specimen reconstructed in Figure 3C and D show that quail cells 'encroach' on the condensing chick scapula, and then form a clean boundary slanting from dorsolateral to ventromedial along the cross section of the blade. This profile persists in Day 9 specimens (Fig. 7E H), where the boundary between chick and quail in the scapular cartilage also starts dorsomedial and expands across the cross-section of the blade in more posterior sections. Between stages H&H 30 and H&H 35 (Days 7 and 9, respectively) the scapula almost doubles in length (compare Figs. 3 and 4; see also Fig. 9B). At Day 7 the scapula is approximately the length of 3 4 vertebrae with the anterior margin adjacent to C13 and the posterior margin lying between T1 and T2. At Day 9, the blade is the length of 6 7 vertebrae. The anterior margin (glenoid) is still adjacent to C13 but the posterior tip of the blade lies between T4 and T5. The anterior border of SP 18 quail grafts at Day 7 are seen at the level of C14, though equivalent grafts in Day 9 chimeras lie adjacent to vertebra T3. Between Day 7 and Day 9 the grafted tissue is displaced posteriorly relative both to the vertebral axis and the glenoid, and at H&H 35, donor cells are located in the posterior third of the scapula. In order to clarify what cell populations formed the ventral portion of the 109

121 anterior scapular blade in Day 9 chicks, we transplanted quail somatic lateral plate into chick hosts (n= 4). Figure 7 (I J) shows sections though a specimen with a well incorporated graft from somatic lateral plate adjacent to somites (H&H 14 at time of surgery). The distal 20% of the blade consists of both donor and host tissue. Donor tissue forms the ventral portion of the scapular blade, a region never populated by segmental plate grafts. The host/graft boundary within the scapula makes a shallow angle that is the mirror image of the anterior host/graft boundary in SP specimens (compare Fig. 7E H and I L). Sections from mid-graft regions that include the natural boundary between somite and lateral plate illustrate that the perichondrium and the immediately adjacent chondrocytes are always derived from the same source, i.e., quail chondrocytes are always associated with quail perichondrium and chick chondrocytes are adjacent to chick perichondrium (e.g., Fig. 7G, K, respectively). This consistency holds even when graft derived muscles attach to host derived cartilage. DAPI staining of sections previously labeled for QcPN provides a negative image of bright field antibody labeling and reveals the identity of the connective tissue lineage in muscle, cartilage, and perichondrium (Fig. 8). In fluorescent images the donor derived primaxial rhomboideus complex (RHC) appears dark due to the absence of labelled chick cells (entirely somitic in origin). This muscle directly contacts the bright, chick derived perichondrium of the anterior scapular blade, (Fig. 8A, B). Further posterior in this chimera, both RHC and the dorsal aspect of the scapular cartilage are quail derived (dark) whereas the ventral aspect of the scapular blade in these sections is 110

122 chick (bright) (Fig. 8C, D). In contrast to the RHC, the SCH includes quail myoblasts (dark) and chick connective tissue (bright) (Fig.8 B, D). DISCUSSION The most recent published fate map for the avian scapula, schematically extrapolated from single-somite and LP transplants, indicates that the majority of the scapular blade is somitic, and that the somitic contribution is segmental, mirroring the segmentation of the axis (Fig. 9A redrawn from Huang et al. '00:fig3). This model (referred to hereafter as the Huang model) has been used to make evolutionary inferences regarding the homology of the scapula among amniotes (McGonnell, 2001; Vickaryous and Hall, 2006; Piekarski and Olsson, 2011), and supplies the context for work on the molecular control of scapula formation (reviewed by Huang et al. 2006, see below). We have generated 3D reconstructions of quail-chick chimeras that allow full visualization of the graft in the context of host anatomy. Our reconstructions make clear three significant contrasts with the Huang model. First, the majority of the scapula is lateral plate derived and the somitic contribution to the avian scapular blade is significantly smaller than previously reported. Second, the segmentation of the somitic component of the blade is not perfectly maintained. Third and likely related to the second, there are striking differences in growth rate between different tissues derived from common somites contributing to the structures at the cervical 111

123 thoracic transition, including the scapula. The reconstructions also allow us to locate the natural interface between somitic and lateral plate mesoderm, called the lateral somitic frontier, which defines the primaxial and abaxial domains (reviewed in Shearman and Burke, 2008). We discuss the novel aspects of our fate mapping data for understanding avian scapular development, and how these advance current ideas on the homology and evolution of the vertebrate scapula. The fate map and location the frontier We show that although six somites contribute to the distal end of the blade at Day 9, approximately two thirds of the scapula is derived from lateral plate (Fig. 9B, C). Our chimeras indicate that somite 18 is the first somite to contribute substantially to the scapula. The anterior quail/chick border within the scapula of SP 18+ chimeras is dramatically attenuated along the AP axis of the blade by Day 9. In contrast, the posterior quail/chick border in the scapula of these chimeras is abrupt and not drawnout (Fig. 4A and 5B). In chimeras with more posterior grafts (SP 21+), the anterior quail/chick border in the scapula is also abrupt (e.g. Fig. 4B). These more abrupt quail/chick borders reflect the incorporation site of the graft, thus the border between somitic mesoderm of donor and host. The attenuated donor/host border described above in SP18 grafts is the border between the somitic and lateral plate contributions to the scapula, corresponding to the lateral somitic frontier. The identity and position of the frontier in the scapula is confirmed by lateral plate transplantation. Lateral plate 112

124 grafts in the scapula have a posterior boundary similar in position and slope to that of the anterior SP grafts (Fig. 7I L). This new fate map provides an alternative view of the modularity of the primaxial and abaxial aspects of the scapula, and it is relevant to the interpretation of many experimental molecular studies of scapular development. A complete review of these studies is beyond the range of this paper, however, several studies show failure of blade formation in response to different molecular manipulations (e.g. Wnt signaling, Moeller et al., 2003; Pax 1 and 3, Ehehalt et al., 2004; BMPs, Wang et al. 2005). These results have been interpreted as acting specifically on cells of the dermomyotome. In light of our fate map it appears that the abaxial (LP) portion of the blade also fails to form in these experiments. The recognition of the extensive lateral plate contribution to the scapular blade implies that considerable molecular integration enables two mesodermal populations to form a single skeletal element. This is particularly interesting given the apparent decoupling of behavior in the somitic cells that contribute to the scapular cartilage compared with other derivatives of the parental somites as discussed below. Skewed segmentation and differential growth The lateral somitic frontier is a cryptic anatomical boundary, but visualizing its position relative to axial structures reveals the dramatic posterior displacement of the primaxial scapula during growth. The method we use here does not show boundaries between adjacent somites within the graft, but reconstructions of grafts at 113

125 different AP levels reveal the crowding of somite incorporation in the scapula. The distorted segmentation of the somitic contribution is schematized in Figure 9C, which summarizes our results. Though somites contribute to the blade in an anterior to posterior pattern, the anterior segmental boundaries are highly skewed along the length of the blade (Fig. 5). The Huang model implies that somitic cells contributing to the scapula act in general accordance with the segmental muscles that arise from those somites, maintaining both primary segmentation and segmental proportions during considerable periods of growth (Fig. 9A). In contrast, our data indicate that the somitic cells that incorporate into the scapular cartilage fail to keep pace with the overall growth of the region, including their own somites of origin (so 18 24, Fig. 9B). These growth dynamics are clearly revealed by the position of graft boundaries within the primaxial shoulder muscles. The extent of donor derived muscle and vertebra by Day 9 is always far more extensive than donor derived scapula. When graft contributes to the RHC, the graft-derived muscle initiates slightly anterior to the graft-derived vertebrae, and is in general alignment at the posterior graft border. This posterior border also aligns with the posterior graft boundary in the scapula. In contrast, donor derived RHC is always inserted on the scapula well in advance of the donor contribution to that element (e.g. Fig. 5A, B). There is clearly no "preference" of donor-muscle for donor-bone, as much of the donor derived, primaxial RHC inserts on chick derived scapula. The SP18 chimeras demonstrate that 114

126 most of the insertion site of (donor) RHC is scapula comprised of (host) lateral plate. Thus primaxial muscle inserts directly on abaxial scapular perichondrium (Fig. 8A, B). The position of the donor/host boundary in muscles makes obvious the posterior displacement and crowding of the somitic portion of the scapula. Once incorporated into the scapular condensation, the somitic cells are disengaged from the segmental environment, and are displaced by the growth of the abaxial scapula, which expands and extends posteriorly. The somitic cells that contribute to the primaxial rhomboids and serratus maintain their original segmentation but are clearly not constrained to insert on scapula of their own lineage. This lack of correspondence between muscle connective tissue and bone of attachment contrasts with the behavior of neural crest derived tissues in the craniofacial skeleton. Where neural crest contributes to skeletal elements in the head and branchial skeleton, the connective tissue of muscle attachments is derived from the same rhombomeric crest population as the bones, preserving strict alignment with the original segmental pattern (Noden, 1983; Kontges and Lumsden, 1996; Evans and Noden, 2006). Homology/Evolution of the scapula Our new model of primaxial and abaxial domains in the avian shoulder require a reassessment of current theories on the homology and evolution of the tetrapod scapula. The position of the frontier and the primaxial and abaxial domains of the mouse were mapped by Durland et al. (2008) in Prx1/Cre transgenics. The 115

127 primaxial nature of the distal edge of the scapular blade and associated muscles was confirmed with Pax3/Cre transgenics by Valasek et al (2010). The data presented here indicates the primaxial component of avian and rodent scapulae are of similar final proportions, though not necessarily from the same number of somites, and that the bulk of the scapula derives from LP in both species. However the relationship of the primaxial scapula with primaxial muscles appears to be different between the two amniotes. Muscle insertions on the scapula are a critical aspect of the integration between axial and appendicular systems. In mice, the primaxial muscles (RHC and levator scapulae in mouse) insert predominantly on the primaxial domain of the scapular blade (Valasek et al., 2010). Our data shows that the primaxial insertion is minor in the chick and there is extensive insertion of primaxial muscles on the abaxial scapula. A difference in the development of the shoulder girdle between a bird and a mammal would not be unexpected given the dramatic locomotor adaptations of birds. New developmental data from an amphibian makes it possible to assess the evolutionary polarity of pectoral development in amniotes. Piekarski and Olsson (2011) performed transplants between transgenic GFP and wild type axolotl embryos and provided a fate map of somitic derivatives in the occipital and anterior trunk region. They find that somites 3, 4 and 5 contribute cells to the distal tip of the salamander scapula (suprascapula), much like contributions to the distal edge of the mouse scapula. These data correct a preliminary result from somite extirpations (Burke, 1991b) that lead to the hypothesis that a somitic contribution to the scapula 116

128 was an innovation at the base of the amniotes (Burke, 1991a). It now appears likely that a somitic contribution to the scapula is shared by all tetrapods (Piekarski and Olsson, 2011). We have previously suggested that differences in the number of somites recruited to the scapula could result in response to selection for more efficient shoulder function in different amniotes (Burke, 1991a; Burke, 2000; Nowicki and Burke, 2000). The demonstration that avian scapular cells arise from the dermomyotome population of the somite lead to the suggestion that the avian scapula was extended by changing the fate of dermomyotomal cells from muscle to cartilage (Huang et al. 2000). The recent amphibian data indicating that the primitive tetrapod scapula included a primaxial domain, suggests an alternative hypothesis for evolutionary change in avian shoulder morphology. The compressed segmentation and apparent failure of the somitic portion of the avian scapula to grow in concert with the rest of the embryo raises the possibility that it represents a vestigial primitive character rather than the site of recent adaptation. The extreme offset of the primaxial portion of the scapula from the vertebrae and primaxial muscles points to evolutionary changes in the abaxial limb field rather than the primaxial domain patterned by the axial Hox code (Krumlauf, 1994; Burke et al., 1995) as previously suggested (Burke, 2000: Huang et al. 2006). In comparison to the mouse and salamander, the extensive insertion of primaxial muscles onto an abaxial element in birds may represent a significant adaptation for flight. We maintain that the border between primaxial and abaxial domains is the site of major evolutionary 117

129 change accomplished though integration of molecular signaling between cell populations. The role of developmental changes in these morphological adaptations can be visualized by the position of the lateral somitic frontier. 118

130 Figure 1. Illustration of the primaxial and abaxial domains of a developing chick embryo. A. Cross section through an embryo shortly after gastrulation. The incipient frontier (red arrow) is the interface between somitic (s, blue) and lateral plate (lp, yellow) mesoderm prior to somitic cell migration. B, C. Cross sections through an older embryo at the level of the forelimb (B) and thorax (C). The primaxial domain is blue, the abaxial domain green, and the lateral somitic frontier is marked in red. D. Illustration of a quail (donor) and chick (host) embryo depicting the alternative surgical manipulations performed. Lateral plate transplantation is shown in yellow, and segmental plate (sp) transplantation in blue. nt, neural tube. 119

131 120

132 Figure 2. Cross sections and three dimensional reconstruction of a SP chimera at Day 9. Quail cells are labeled with QcPN. A. Cross section through the middle of the graft. The vertebra and deep axial muscles are QcPN positive (black) on the operated side of the embryo. B. Surface model of the vertebrae, ribs, and scapula in lateral view. Graft (quail) contributions to skeleton are yellow. The position of cross sections A and C are indicated. C. Cross section through the anterior margin of the graft. Quail cells are scattered throughout the predominantly chickderived vertebra. a, anterior; p, posterior; RHC, rhomboideus complex; r, rib; sc, scapula; SCH, scapulohumeral complex; SRC, serratus complex; v, vertebra. Scale bar = 100 microns. 121

133 122

134 Figure 3.Three dimensional reconstructions of chimeras at Day 7. A, B. Surface models of the vertebrae and scapula of an SP (A) and SP17 21 (B) transplant in lateral view. The identities of cervical (C12 14) and thoracic vertebrae (T1 and 2) are indicated. SP transplants anterior to S18 do not contribute to the scapula whereas grafted cells (yellow) from SP18+ transplants contribute to the dorsal portion of the scapular blade. C, D. Volume rendered model of a single chimera (SP16 19) in frontal (C) and lateral (D) view. The total distribution of quail cells is yellow, see text for details on method. A surface model of the scapula (white) is shown within the context of the rendered model, and graft contributions to the scapular blade are shown in red. Skin is orange. a, anterior; d, dorsal; l, left; p, posterior; r, right; sc, scapula; v, ventral. 123

135 124

136 Figure 4. Surface models of the vertebrae, ribs, and scapula of Day 9 chimeras in lateral view. Donor cells shown in yellow. The identities of cervical and thoracic vertebrae (C14 T3) are indicated. A. In SP transplants, the anterior graft margin in the scapular blade is attenuated and lies approximately three vertebral segments caudal to the anterior graft margin in the vertebrae. B. In SP transplants, graft contributions to the scapula and vertebrae are offset by a single vertebral segment, and both the anterior and posterior graft margins appear abrupt. a, anterior; p, posterior. 125

137 126

138 Figure 5. Surface models of the vertebrae, ribs, scapula, and shoulder muscles of three chimeras at Day 9. Graft contributions are shown in yellow (skeletal elements) and red (muscle). The identities of cervical and thoracic vertebrae (C14 T3) are indicated. A,B,C. Lateral view of the rhomboideus complex (RHC) in SP (A), SP (B), and SP (C) transplants. Grafted muscle (red) inserts on portions of the scapular blade derived from both graft (yellow) and host (grey). D,E,F. Medial view of the serratus complex in SP (D), SP (E), and SP (F) transplants. Anterior graft borders in the serratus also show more anterior initiation in muscle than scapula in SP 18 and 19, but less so in SP22. Posterior graft boundaries (when present in scapula) align in muscle and bone. a, anterior; p, posterior. 127

139 128

140 Figure 6. Sox9 (A,C,E) or QcPN (B,D,F) labeling of alternate sections of a Day 5 chimera. Tracings (dashed lines) of the Sox9 expression domain in the scapulaforming region at different axial levels have been superimposed on adjacent sections labeled for QcPN. A, B. In the glenoid region there is no overlap between Sox9 expression and QCPN positive cells. C,D. In the proximate scapular blade, there is slight overlap between QcPN and Sox9 medially. E,F. In the distal scapular blade, QcPN positive cells occupy the medial one third of the Sox9 expression domain, prefiguring the distribution of somitic cells in the Day 9 scapula. Scale bar = 100 microns. 129

141 130

142 Figure 7. Sequence of cross sections through the scapula progressing from anterior to posterior. QCPN positive cells are black. A D. SP18 21 transplant at Day 7. E F. SP transplant at Day 9. C D. Somatic LP transplant adjacent to somites at Day 9. Lateral plate (quail) contributions to the scapular blade are a mirror image of somitic contributions in SP transplants. In all transplants donor/quail chondrocytes are adjacent to donor/quail derived perichondrium. sc, scapula. Scale bar = 100 microns. 131

143 132

144 Figure 8. QcPN labeled sections of an SP transplant counterstained with DAPI to visualize all QcPN-negative nuclei. A. QcPN (black) labeled cross section through the mid-scapular blade. B. Higher magnification of the scapula shown in A following DAPI counterstain. Note donor-derived rhomboideus complex inserts on chick-derived perichondrium. C. QcPN labeled cross section through the posterior scapular blade. Donor-derived rhomboideus complex inserts on donor-derived perichondrium. D. Higher magnification of scapula shown in C following DAPI counterstain. RHC, rhomboideus complex; SC, scapula; SCH, scapulohumeral complex. Scale bar = 100 microns. 133

145 134

146 Figure 9. Illustrations of fate maps of the avian scapula. A. Redrawn from Huang et al. (2000:fig3) suggesting the majority of the blade derives from the somites and a strict segmental organization is maintained during development. B. Schematic depiction of the extensive growth of the scapula through Day 9. Note that growth of the abaxial domain (green) displaces the primaxial domain (blue) posteriorly. The lateral somitic frontier is represented with a red line. C. 3D representation of a Day 9 avian vertebral column and scapula. Primaxial and abaxial structures are blue and green, respectively, and two shades of blue are used to indicate resegmentation. Cervical and thoracic vertebrae (C14 T5) are indicated. a, anterior; p, posterior. 135

147 136

148 Table 1. Number of segmental plate and lateral plate transplant surgeries in avians. TYPE OF SURGERY ~H&H STAGE Segmental plate Segmental plate Segmental plate 14+ Somatopleure ~AXIAL LEVEL NUMBER OF TRANSPLANTS PERCENT SURVIVAL GRAFT INCORPORATION 3D RECONS Somites %(17) 88% 4 Somites % (18) 94% 4 Somites % (4) 100% 2 Somites % (4) 50% 1 137

149 CHAPTER 4: A detailed method for isotopic transplantation of lateral plate mesoderm in axolotl and insights into the position of the lateral somitic frontier SUMMARY Understanding the fates of embryonic cell populations is a critical aspect of developmental biology. In chapter two of this thesis I introduced a technique for isotopically transplanting lateral plate mesoderm and ectoderm from GFP to WT axolotls, and used the data from these surgeries to test the persistence of somatopleure in the lateral body wall of an amphibian. Herein, I provide additional details describing my surgical approach, and include photographs of donor and host embryos at each step of transplantation. My hope is that this documentation will provide a reference for future work that incorporates lineage analyses into studies of axolotl limb, shoulder, and body wall development or regeneration. Additionally, I use fate maps from these surgeries to identify the position of the lateral somitic frontier. These fate maps provide evidence that the majority of the pectoral girdle and its associated musculature are abaxial. The dorsal margin of the scapula, however, incorporates the frontier and is primaxial, consistent with the hypothesis that a somitic contribution to the scapula is a shared feature of tetrapods. Interestingly, these fate maps also indicate that lateral plate connective tissue invests the myofibers of hypaxial body wall muscles. The abaxial nature of these muscles strongly suggests that they derive from migratory somitic cells, similar to the muscles of the limbs. 138

150 Future work examining the distribution of lateral plate mesoderm at earlier stages of body wall formation, in combination with the use of molecular markers for migratory myoblasts, will test this hypothesis. INTRODUCTION Insights into how embryonic cell populations segregate during development provide the tissue context for studies of molecular patterning and morphogenesis. Key to such lineage analyses is the ability to indelibly label and track cell populations in the face of cell division, migration, and differentiation. In 1969, LeDouarin introduced quail-chick chimeras as a developmental model for long term fatemapping in avians. This technique involved transplanting tissue from quail donors to chick hosts and later identifying grafted cells based on quail-specific nucleolar morphology (LeDouarin, 1969). More recent work has combined LeDouarin s technique with immunohistochemistry to better visualize quail-derived graft cells, as demonstrated in chapter three of this thesis. In transgenic mice, the Cre-LoxP system has been used as a proxy for lineage analyses (Joyner and Zervas, 2006; Kim and Dymecki, 2009). In this system, tissue specific activation of the bacteriophage enzyme Cre-recombinase catalyzes a genetic rearrangement that drives the expression of a reporter (e.g., alkaline phosphatase or GFP), which is inherited by all daughter cells (Lobe et al., 1999; Novak et al., 2000). Together, these techniques in avians and mice have advanced our understanding of a variety of developmental processes in 139

151 amniotes, and provide a wealth of data for generating hypotheses that can be tested in a comparative context. Amphibians exhibit tremendous morphological diversity, and occupy a critical phylogenetic position for work aimed at gaining insight into the evolution of tetrapods and the polarity of amniote characters. The generation of a transgenic axolotl that ubiquitously expresses green fluorescent protein (GFP) (Fig.1) has opened the door for lineage analyses with single cell resolution in a representative urodele (Sobkow et al., 2006). Since its introduction, transplantation experiments in axolotls (similar to those mentioned above for avians) have been used to study neural crest derivatives (Harlow and Barlow, 2007; Epperlein et al., 2012), tooth formation (Soukup et al., 2008), mesodermal boundaries in the musculoskeletal system (Piekarski and Olsson, 2011; Epperlein et al., 2012; Tulenko et al., 2013), and lineage contributions during limb (Kragl et al., 2009) and nerve regeneration (McHedlishvili et al., 2007; McHedlishvili et al., 2012). Given the well-documented amenability of salamanders to surgery (Balinsky, 1970), and the usefulness of fate-mapping data in studies of development, regeneration, and morphological evolution, detailed descriptions of transplantation techniques in axolotls will facilitate future, reproducible work with this emerging model system. Herein, I provide a detailed protocol for isotopic transplantation of embryonic lpm/ectoderm using GFP and wild-type axolotls. In this protocol, I also describe the inclusion of pronephros in lpm/ectoderm grafts, with the assumption that these grafts contain the dorsal-most lateral plate. It is worth noting that this protocol was first 140

152 introduced in chapter two of this thesis as a way to test somatopleure persistence in a representative amphibian. The fate-mapping data presented below draws from the same set of surgical chimeras (Chapter 2, Table 3), but adds substantial detail to the analysis of the position of the lateral somitic frontier. MATERIALS AND METHODS Wild-type and GFP transgenic axolotl embryos (Ambyostoma mexicanum) were obtained from the University of Kentucky Ambystoma Genetic Stock Center (AGSC), and staged according to Bordzilovskaya et al (1979) for young embryos (Stages 1 44) and Nye et al (2003) for more advanced embryos (St.45 57). For transplantation experiments, embryos were raised in 25% Holtfreter solution at 18 o C to Stage 23/24, and then moved to 4 o C to slow development. Prior to surgery, embryos were dechorionated using two sets of fine forceps, one to pierce the jelly coat and chorion, and the other to increase the size of the opening. Following dechorionation, embryos were transferred to a Petri dish lined with 2% agarose and placed back at 4 o C. Tools for transplantation experiments were prepared in advance, and nondisposable items were cleaned between surgeries with 70% Ethanol. Blunt glass probes for positioning the embryos were made by flaming the tips of glass Pasteur pipettes. Tungsten needles (A-M Systems, ) for cutting ectoderm were sharpened with a propane torch. A tool used to either manipulate or remove tissues during surgery was prepared as follows: First, multiple thin-walled glass 141

153 micropipettes (World Precision Instruments, TW120F-4) were pulled using a Shutter Instruments P-97 Flaming/Brown Micropipette Puller, and the tips opened by piercing through a Kim-wipe. A single pulled micropipette was then placed in the larger end of a P10 (0.5 10µm) pipette tip, which was inserted into one end of a 70cm piece of thin, Tygon tubing. A P1000 ( µm) tip was inserted in the other end of the tubing to serve as a mouthpiece. This tool was used to either remove tissues by manual suction or eject fresh solutions onto embryos. Importantly, embryonic tissues were not allowed into the Tygon tubing, and micropipettes were replaced between surgery steps to avoid contamination. Two-milliliter disposable plastic pipettes, trimmed at the tip to widen the opening, were used for all embryo transfers, with care taken to keep dechorionated embryos from becoming damaged by contacting the liquid/air interface. For each isotopic transplantation, the host-site was prepared prior to the donor-graft. First, a Stage 23/24 wild-type embryo was transferred to a clay-lined dish filled with 2X Steinberg solution (Fig. 2A). For most surgeries, a thin layer of plain Chapstick was applied to the surface of the clay to soften it. A blunt glass probe was used to position the embryo in left-lateral view, and then manipulate the clay to hold it in place, as shown in Figure 2B. A small, anterior-to-posterior incision was made in the ectoderm overlying the pronephros (visible next to somites 4 7) with a tungsten needle, and the ectoderm peeled dorsally just enough to make visible the ventral margin of the somites (Fig. 2B). The pronephros was then removed using suction (Fig. 2C), revealing the large endodermal cells beneath (Fig. 2D). At this point the 142

154 lateral plate mesoderm (lpm) was visible just inside the cut margin of the ectoderm (Fig. 2D). Fresh 2X Steinberg solution was gently expelled from a new micropipette to separate the lpm from the underlying endoderm. The ectoderm and lpm adjacent to somites 3 7 were removed by suction, similar to the pronephros, completing the host site (Fig. 2E). Once the host site was prepared, a stage-matched GFP donor was transferred to the same dish as the host embryo and positioned in left lateral view (Fig. 2F). A small incision was made in the ectoderm along the ventral margin of somites 3 7, and the ectoderm was peeled dorsally (Fig. 2G). The ventral margins of somites 3 7 were removed by suction (Fig. 2H) and the glass capillary discarded. At this point, the pronephros could also be removed by suction, though it was left intact in the specimen shown in Fig. 2. Additional superficial incisions in the ectoderm were made, presaging the position of a square-shaped graft. A gentle stream of fresh 2X Steinberg solution was used to create a space between the pronephros/lpm/ectoderm and underlying endoderm. While continuing to expel Steinberg solution, the tip of the micropipette was shifted to a position underneath the anterior ectodermal incision and pulled outwardly to cut through the lpm. This was repeated along the posterior ectodermal incision (Fig. 2I), allowing the graft to be reflected ventrally (Fig. 2J). Suction was used to free the ventral margin of the graft, which was then moved to the host site. In some embryos, the antero-dorsal margin of the graft was labeled with the vital dye Nile Blue to facilitate correct orientation (not shown). A fire-polished piece of a glass coverslip was used to secure the graft in place (Fig. 2K), and the 2X 143

155 Steinberg solution was replaced with 25% Holfretter s plus gentamicin (50µg/ml) through a series of dilutions. The resulting GFP/WT chimera was allowed to heal for approximately 35 minutes before the glass coverslip was gently removed and the embryo transferred to a new Petri dish lined with 2% agarose (Fig. 2L). All whole-mount chimeras were examined with fluorescence using a Nikon SMZ-U dissecting microscope equipped with a SPOT RT3 camera and Spot software. Cryosections of chimeras were also prepared to assess the distribution of grafted cells. For cryosections, chimeras were first transferred through a graded series of sucrose solutions (10 30% sucrose in PBS), before embedding in Tissue Freezing Media (Triangle Biomedical Sciences Inc). Blocks were sectioned on a Leica CM 3050 S cryostat at 12 16µm and dried for two hours with a hair dryer. In order to increase the contrast of donor cells, cryosections were immunohistochemically labeled for GFP. For this, sections were blocked in PBT (0.1% Tween in PBS) with 5% goat serum and 2% bovine serum albumin, and incubated overnight in an anti- GFP antibody (A6455, Invitrogen) diluted in block (1:500). Following incubation in primary, sections were washed with PBT and incubated for a minimum of 3.5 hours in secondary antibody (AlexaFluor 488 goat anti-rabbit IgG, Invitrogen), also diluted in block (1:500). Labeling with an anti-pax7 antibody (Developmental Studies Hybridoma Bank, 1:20) was similar to that of GFP, but with an AlexaFluor Goat anti-mouse IgG1 secondary (Invitrogen, 1:500). Actin was labeled with Rhodamine- Phalloidin (Invitrogen, 1:200) to visualize skeletal muscle. Nuclei were 144

156 counterstained with TO-PRO -3 Iodide (Invitrogen, 1:500). All sections were visualized using a Zeiss LSM 510 confocal microscope. RESULTS Graft incorporation was initially evaluated between 15 and 24 hours postsurgery (St.31 33) (Fig.3A,B,E,F). In whole mounts, GFP-positive tissue is visible caudal to the external gill bulge (Fig. 3 B,F), and the pronephros, when present, extends as a straight line towards the cloaca (Fig. 3F). In cryosections, GFP-positive LPM and ectoderm partially surround the yolk tube on the surgery side (Fig. 4A,B). The dorsal margin of GFP-labeled tissue directly abuts the somitic dermomyotome, which is made visible by Pax7 labeling (Fig. 4C). In whole mounts of older stage chimeras (e.g., Stages 53 57), grafted tissue extends from the pectoral region to the mid-trunk, and appears to be extensively incorporated into the forelimb, inter-limb region, and gut tube (Fig. 3C,D,G,H). The pronephros is also visible anteriorly due to its superficial position just deep to the skin (Fig. 3G). It is worth noting that GFP fluorescence from grafted lateral plate cannot always be distinguished from fluorescence from ectoderm in whole mount view. In order to identify how lateral plate mesoderm contributes to the structures of the musculoskeletal system in older stage chimeras, I examined the distribution of GFP-positive cells in cryosection, with the assumption that transplanted ectoderm and 145

157 pronephros contribute only to the epidermis and nephric system, respectively. The following descriptions are for Stage 53, unless otherwise noted. Similar to other salamanders, the pectoral girdle of axolotls lacks dermal components, and consists of a scapula, coracoid and procoracoid process (Duellman and Trueb, 1986; Shearman, 2008). These structures predominantly derive from GFP-positive chondrocytes, with the exception of the vertebral border of the scapula, which is GFP-negative (Fig. 5). This result is consistent with a recent report that the dorsal scapula derives from somitic mesoderm (Piekarski and Olsson, 2011). Interestingly, a GFP-positive perichondrium extends further dorsally than labeled chondrocytes within the scapula (Fig. 5B,C,G,H), indicating lateral plate partially invests somitic cartilage. In salamanders, three muscles suspend the pectoral girdle from the axial system: the cucullaris, levator scapula, and serratus anterior (Duelmann and Trueb, 1986). The cucullaris and levator scapula, both of which are neck muscles that connect the head and scapula, were not examined in this study. The serratus anterior extends from the distal tips of the first several ribs to the vertebral border of the scapula, forming the sole attachment between the pectoral girdle and post-cranial axial skeleton. Notably, this muscle is absent in Stage 53 larvae, but is present by Stage 57 (Fig. 5D H ). In Figure 5E H, the serratus anterior is compressed against the epaxial muscle, and its position is highlighted with white arrowheads for clarity. GFP-positive cells do not appear to contribute to the connective tissue of the serratus anterior or its site of origin or insertion (Fig. 5E H ). 146

158 In axolotls, several muscles connect the pectoral girdle and proximal humerus, including the deltoideus scapularis, subscapularis, procoracohumeralis, and supracoracoideus (Fig. 6) (Duellman and Trueb, 1986; Diogo et al., 2009; Abdala and Diogo, 2010). The connective tissue of all of these muscles is GFP-positive, deriving from lateral plate (Fig. 6B F,J,K). Similarly, the pectoralis, which originates on the fascia of the ventral body wall and inserts on the head of the humerus, is also invested by GFP-positive connective tissue (Fig. 6L,M). The axial musculature of salamanders consists of metamerically arranged myotomes, which are split into epaxial and hypaxial compartments by the horizontal septum and ribs. GFP-labeling within the epaxial compartment is limited to a few scattered cells that are likely vascular in origin. The hypaxial compartment is further sub-divided into three groups: the sub-vertebral muscles, flank muscles, and abdominal muscles (Duellmann and Trueb, 1986). The sub-vertebral muscles lie just ventral to the horizontal septum, and, like the epaxial muscles, are GFP-negative (Fig. 6A I). The muscles of the flank arise ventral to the sub-vertebral muscles, and extend around the coelomic cavity. In the pectoral region, a single flank muscle is present, the obliquus internus (Fig. 6A), which is separated from the sub-vertebral muscles by the pronephros (Fig.6B,C,F,G). Caudal to the forelimb, the sub-vertebral and flank muscles are continuous, and a second layer, the obliquus externus, joins the internus in the lateral body wall (Fig. 6A,D,E,H,I). Unlike the sub-vertebral muscles, GFP-positive lateral plate connective tissue invests the myofibers of the obliquus externus and internus (Fig. 6B I). Ventrally, the flank muscles are continuous with 147

159 the rectus abdominis (Fig. 6A). The myofibers of this muscle are also invested by GFP-positive connective tissue (Fig. 6D,E,M). GFP-positive cells contributing to the presumptive dermis align dorsally with the dorsal margin of the body cavity and pronephros at both pectoral and inter-limb levels (Fig.6B I). This is also the approximate position of labeling within the obliquus internus in the pectoral region (Fig.6 B,C,F,G), and the obliquus internus and externus posterior to the axilla (Fig. 6D,E,H,I). The connective tissue septum separating these muscles is also GFP-positive (Fig. 6E,I). Thus, cells of the lateral plate form the connective tissue of the lateral body wall surrounding the coelomic cavity, and invest the myofibers of hypaxial flank and rectus muscles. Notably, GFP fluorescence is also visible investing the margins of spinal nerves near their exit from the spinal cord (Fig. 6D,H and Fig. 7). Remarkably, this spinal nerve investment also roughly aligns with the dorsal extent of lateral plate-derived connective tissue in the obliques and dermis. DISCUSSION Herein, I have provided a detailed protocol for isotopically transplanting lpm and ectoderm between GFP and wild-type axolotls. The results of these surgeries indicate that somatopleure adjacent to somites 3 7 forms forelimb, shoulder, and anterior inter-limb trunk, consistent with previous reports for Ambystoma maculatum (Stocum and Fallon, 1982). In chapter two of this thesis, I used data from surgical chimeras to test the persistence of somatopleure in the body wall of a representative 148

160 amphibian. Below, I extend these analyses and describe in detail the position of the lateral somitic frontier at pectoral and inter-limb levels. In Stage 53 larvae, GFP-labeled lateral plate cells form a distinct boundary with unlabeled cells at the approximate dorsal margin of the body cavity. As demonstrated in Figure 6 F I, the position of this boundary is consistent not only between sections from different axial levels, but also between different tissue types within a given section. For example, labeled lateral plate cells of the body wall mesenchyme and intra-muscular connective tissue align at both pectoral and interlimb levels. Notably, these sections include a complete graft-derived parietal coelomic lining and nephric system on the surgery side. Together, I interpret these data to indicate that the dorsal extent of labeled lateral plate cells reflects the true position of the lateral somitic frontier rather than an artifact of the interface between labeled donor cells and unlabeled host. As described above, the axolotl pectoral girdle develops predominantly from lateral plate mesoderm and is thus abaxial. The dorsal scapula, however, incorporates the lateral somitic frontier, and is primaxial along its vertebral border, deriving from somites three, four and five (Piekarski and Olsson, 2011). The serratus anterior muscle arises along this primaxial portion of the scapula and extends to the distal tips of the first several ribs, forming a muscular sling that suspends the trunk from the limbs. This muscle, like its site of origin and insertion, is primaxial. In contrast, the muscles that rotate the forelimb in the shoulder (i.e., the deltoideus scapularis, 149

161 subscapularis, procoracohumeralis, supracoracoideus, and pectoralis) arise along the abaxial pectoral girdle, and are themselves abaxial. The body wall musculature of axolotls is segmented along the anteriorposterior axis, an arrangement common to vertebrates with a primarily undulatory mode of locomotion (Schilling, 2011). Vertebral ribs form along the horizontal skeletogenic septum, dividing each myomere into epaxial and hypaxial compartments. The epaxial compartment forms entirely from somitic mesoderm, and is primaxial. The hypaxial compartment includes three muscle groups, the subvertebral muscles, the flank muscles, and the rectus abdominis. The subvertebral muscles lie directly beneath the horizontal septum and derive entirely from somitic mesoderm. These muscles, like the epaxial muscles, are primaxial. In contrast, lateral plate cells ensheath the myofibers of the flank and rectus muscles, which like the muscles of the shoulder and limb are abaxial. The extensive mixing of somitic and lateral plate populations in the body wall muscles of axolotls raises interesting questions about how these muscles form. In amniotes, the mechanisms underlying hypaxial muscle development differ with axial level and muscle identity (Dietrich, 1999). At limb-forming levels, the ventro-lateral dermomyotome undergoes an epithelial to mesenchymal transition, producing longrange migratory cells that move into the limb-bud. Notably, the transcription factor Lbx1 is required for this process (Vasyutina and Birchmeier, 2006), and is a useful marker for visualizing migratory cells as they leave the somitic environment and form abaxial limb muscles (Vasyutina and Birchmeier, 2006). In contrast, inter-limb 150

162 somites do not express Lbx1, and give rise to body wall muscles including the intercostals and abdominal obliques by direct extension of the ventro-lateral dermomyotome (Dietrich, 1999). Notably, the intercostals form almost exclusively from somitic mesoderm and are primaxial (Nowicki et al., 2003; Durland et al, 2008). The abdominal obliques and rectus muscles, however, exhibit extensive mixing of somitic and lateral plate populations, and are thus abaxial (Christ et al., 1983; Nowicki et al., 2003; Durland et al., 2008). The trunk musculoskeletal system of axolotls is poorly regionalized along the anterior-posterior axis relative to amniotes (Duellman and Trueb, 1986). The abaxial nature of the hypaxial flank muscles strongly suggests that they derive from migratory muscle precursors that delaminate from the somites and mix with cells of the lateral plate. Additionally, the straight-line profile of the LSF in cross-section, and its high position along the lateral body wall, suggests that the primaxial domain undergoes ventral growth early in development, but does not extend past the level of the body cavity. Interestingly, in pre-metamorphic Xenopus tadpoles, hypaxial body wall muscles appear to derive from migratory streams of somitic cells that express Lbx1 (Martin and Harland, 2001; Martin and Harland, 2006). This observation together with the axolotl data presented above suggests that completely abaxial flank musculature may be a shared feature of amphibians. These new data for axolotls raise an interesting issue in the development of metamerically arranged body wall muscles. In anamniote vertebrates, segmental myomeres reflect the primary segmentation of the somites, and are generally thought 151

163 to develop as primaxial structures (Burke and Nowicki, 2003). During the development of axolotls, however, migratory muscle precursors appear to leave the somitic environment and become invested by connective tissue of the un-segmented lateral plate. Remarkably, these myoblasts give rise metameric flank muscles that are in perfect register with their parent somites. Future work investigating how this spatial relationship is maintained may provide new insights into migratory cell pathfinding and the role of lateral plate mesoderm in patterning abaxial derivatives. 152

164 Fig. 1. Transgenic Axolotl ubiquitously expressing Green Fluorescent Protein (GFP). A, B. Dorso-lateral view of same Stage 39 transgenic axolotl embryo under bright-field (A) and fluorescent (B) illumination. GFP is shown in green. 153

165 154

166 Fig. 2. Isotopic Transplantation of Lateral Plate Mesoderm, Ectoderm, and Pronephros from GFP Donor to Wild Type Host. Left lateral view of stagematched host (A E, K, L) and donor (F J). Host site (A E) was prepared prior to graft (F J). Somites 3 7 are labeled. The host embryo (A) was positioned in clay and an incision in the ectoderm (ect) was made over the pronephros (pro) (B). The pro was then removed by suction using a pulled glass capillary tube (pseudo-colored blue for contrast) (C), exposing the underlying endoderm (end) (D). The ect and lateral plate (lpm) adjacent to somites 3 7 were removed, completing the host site (E). In the donor (F), an incision in the ect was made along the dorsal pro, and the ect pulled dorsally (G), exposing the somites. The ventral margins of somites 3 7 were removed using suction (H). The anterior and posterior graft margins were then freed (I), and the graft reflected ventrally (J). The graft was positioned in the host site, and held in place with a piece of a glass coverslip (K). After the coverslip was removed, the chimera was transferred to a Petri-dish lined with agarose (L). a, anterior; p, posterior. Scale bar=500µm. 155

167 156

168 Fig. 3. Distribution of GFP-Positive Donor Cells in Whole Mount Axolotl Chimeras. A H. Whole-mount chimeras following isotopic transplantation of either lateral plate mesoderm (lpm) and ectoderm (ect) (A D) or lpm, ect, and pronephros (pro) (E H). A C and E G are left lateral view, D,H are ventral view. GFP is green. Developmental stages are indicated in the top of each panel. Within one day of surgery, well-incorporated GFP-positive graft was present along the flank caudal to the external gill bulge (exg) (B,F). In later stage chimeras (C,D,G,H), GFP-positive cells were present in the fore-limb, ventro-lateral trunk, and wall of the gut tube. Note that individual skeletal elements of the forelimb are discernible in C and G. Arrowhead in G indicates pronephros, which is positioned superficially in the shoulder region. Scale bar=500µm 157

169 158

170 Fig. 4. Evaluation of graft incorporation approximately one-day post-surgery. A C. Cross sections through the mid-graft of a Stage 33 chimera following transplantation of lateral plate mesoderm (lpm), ectoderm (ect), and pronephros (pro). Sections are labeled for GFP (green) and actin (Phalloidin, red) (A, B), or GFP (green) and Pax7 (blue) (B). Note that Pax7 labeling makes visible the dermomyotome, the outer-most compartment of the somite, which directly abuts GFP-positive tissue. Scale bar in A is 250µm; Scale bar is C is 50µm. 159

171 160

172 Figure 5. Distribution of graft in the pectoral region of axolotl chimeras following isotopic transplantation of LPM/Pronephros/Ect from GFP donor to WT host. A. Stage 53 chimera in left lateral view. GFP is green. B C. Cross section through the pectoral region of the chimera shown in A. The scapula (sc) derives from donor, GFP-positive cells except for its vertebral border, which is GFP-negative. D. Stage 57 chimera (left) and WT axolotl stained with MF20 for skeletal muscle (right) in left lateral view. Shoulder and body wall muscles are labeled as follows: ds, deltoideus scapularis; ep, epaxial muscle; ld, latissimus dorsi; oe, obliquus externus; oi, obliquus internus; pect, pectoralis; pro, procorocoideus; serr, serratus anterior; sup, supracoracoideus. Arrows for oi and oe indicate myofibers orientation. E H. Cross sections through the pectoral region of chimera shown in D. The vertebral border of the scapula is GFP-negative, and provides the site of attachment for the serratus anterior muscle (arrowheads in F and G), which extends ventrally to insert on the distal tip of the rib (r). No GFP positive cells invest the myofibers of the serratus, or its site of origin and insertion. Note that in G and H, GFP-positive connective tissue of the perichondrium extends slightly more dorsally than GFP positive chondrocytes. All sections are labeled for skeletal muscle (phalloidin, red), and nuclei (Topro, cyan). ant, anterior; post, posterior; pro, pronephros; sv, subvertebral muscle; v, vertebra. 161

173 162

174 Figure 6. Distribution of graft in the pectoral region and trunk of an axolotl chimera following isotopic transplantation of LPM/Pronephros/Ect from GFP donor to WT host (two pages). A. Stage 53 chimera in left lateral view (left), and WT axolotl stained with MF20 for skeletal muscle in left lateral view (middle) and ventral view (right). Shoulder and body wall muscles are labeled as follows: ds, deltoideus scapularis; ep, epaxial muscle; ld, latissimus dorsi; oe, obliquus externus; oi, obliquus internus; pect, pectoralis; pro, procorocoideus; ra, rectus abdominis; sup, supracoracoideus. Arrows for oi and oe indicate myofiber orientation. B M. Series of cross sections through chimera shown in A. All sections are labeled for skeletal muscle (phalloidin, red), and nuclei (Topro, cyan). B E. Low magnification of sections through pectoral (B,C), axillary (D), and inter-limb (E) levels. F M. High magnification of boxes in B E. The muscles of the shoulder include the deltoideus scapularis (ds) (F,K), procoracoideus (pro) (J,K), subscapularis (sub) (J,K), supracoracoideus (sup) (J,K), and pectoralis (pect) (L,M), all of which are completely invested with GFP-positive connective tissue. The hypaxial muscles include the subvertebral muscles (sv), flank obliquus internus (oi) and obliquus externus (oe), and rectus abdominis (ra). The subvertebral muscles (F I) are GFPnegative. In contrast, the obliquus internus (F I) obliquus externus (H,I) and rectus abdominis (L,M) are completely invested in GFP-positive connective tissue. The dorsal margin of GFP-positive cells forming the presumptive dermal mesenchyme and intra-muscular connective align with the dorsal margin of the coelomic cavity (F I). I interpret this to represent the lateral somitic frontier (lsf in F I ). Note that the 163

175 parietal coelomic lining (pc) and pronephros (pro) are completely labeled on the surgery-side (F I), and that GFP-positive cells invest a spinal nerve (sn) (H). 164

176 165

177 166

178 Figure 7. Cross sections showing the spinal nerves of a Stage 57 WT axolotl and a chimera following isotopic transplantation of LPM/Pronephros/Ect from GFP donor to WT host. A,B. Cross section through WT axolotl. Dorsal (dr) and ventral roots (vr) join to form a spinal nerve (sn) in the pectoral region. C,C. GFP positive cells (arrowheads) invest a spinal nerve shortly after its exit from the vertebral column. drg, dorsal root ganglion; ep, epaxial muscle; oi, obliquus internus; pro, pronephros 167

179 168

180 Concluding Remarks The studies presented throughout this work are united by the conceptual framework of the lateral somitic frontier and the importance of considering embryonic context in discussions of development and evolution. In chapter two of this thesis I compare the position of the frontier in lamprey, catshark, axolotl, mouse and chick, and provide the first experimental evidence that the distribution of somitic and lateral plate mesoderm differs between agnathan and jawed vertebrates. Specifically, I demonstrate that in lamprey, the somatopleure is present early in development, but is ultimately eliminated by ventral myotome growth, as cells of the somatic lateral plate are displaced medially and sequestered to the coelomic linings. In contrast, the somatopleure is retained in the lateral body wall of jawed vertebrates at both fin and inter-fin axial levels. Together, these data are consistent with the hypothesis that the ancestral post-cranial body plan of vertebrates was primaxial, and that persistence of somatic lateral plate in the body wall of stem gnathostomes was a key ontogenetic change that facilitated the birth of a new developmental module, now recognized as the abaxial domain. Importantly, this new field of cells provided the substrate for gene network redeployment, fueling the origin of appendicular morphologies. This work provides a new perspective on an old question in comparative anatomy, and my hope is that it will inspire future studies aimed at clarifying the molecular underpinnings of somatopleure persistence in diverse 169

181 vertebrate groups. I believe such data will provide new insights into how paired fins first arose and the relationship between genetic redeployment, tissue redistribution, and anatomical innovation. The origin of the appendicular system represents a major modification of the vertebrate body plan. Importantly, pectoral and pelvic girdles anchor paired fins and limbs to the trunk. In chapter three of this thesis, I demonstrate the distribution of somitic and lateral plate mesoderm in the avian shoulder using an approach that integrates three-dimensional reconstructions with surgical grafting experiments. From a practical standpoint, these models set a new standard for interpreting chimerabased fate maps. Additionally, the data generated during these experiments reveal that the majority of the avian scapula is abaxial, deriving from cells of the lateral plate. In contrast, the primaxial portion of the scapular blade is modest, and fails to keep pace with growth of other structures of the shoulder, including the abaxial scapula. This difference in growth rates suggests that a somitic contribution to the scapula may be a vestigial feature of pectoral girdle development, at least in avians. Fate-mapping studies in mice and salamanders indicate that a dual origin of the scapula from somitic and lateral plate mesoderm is a shared feature of tetrapods. It has been suggested previously that a somitic contribution to the scapula evolved early in the history of the appendicular system as a primaxial musculoskeletal anchor for abaxial paired fins (Shearman and Burke, 2009). Clarifying how mesodermal populations contribute to the pectoral girdle of basal actinopterygians (e.g., paddlefish 170

182 or gar) and chondrichthyans will provide an important test for this hypothesis and enrich our understanding of the early history of the vertebrate appendicular system. Finally, in chapter four of this thesis I use lateral plate transplants between GFP and WT axolotls to map the position of the lateral somitic frontier in a representative amphibian. These data provide the first evidence that the lateral body wall muscles of axolotls are abaxial, strongly suggesting that they derive from migratory somitic cells, similar to the muscles of the limb and shoulder (Dietrich, 1999; Evans et al., 2006; Christ and Brand-Saberi, 2002). Interestingly, this also appears to be the case during body wall muscle formation in pre-metamorphic Xenopus tadpoles (Martin and Harland, 2006). Placing these developmental data for amphibians in a comparative context suggests new hypotheses for the evolution of body wall musculature in vertebrates. The flank muscles of amphibians are homologous with the myomeres of fishes and the intercostal and abdominal muscles of amniotes. As demonstrated in chapter two of this thesis, the myomeres of lamprey are primaxial, reflecting the growth of individual somites. Similarly, the myomeres of sharks also appear to be primaxial, deriving from epithelial projections of the dermomyotome (Goodrich, 1930; Neyt et al., 2001; Cole and Currie, 2007). In amniotes, the intercostal muscles are the only body wall muscles to reflect the primary segmentation of the somites, and on this level, are comparable to the body wall muscles of anamniotes. Notably, the intercostal muscles, like the myomeres of lamprey and shark, grow ventrally as predominantly primaxial structures (Burke and Nowicki, 2003; Durland et al., 2008). Collectively, these data suggest that 171

183 amphibians may represent an evolutionary divergence in the behavior of cells at the lateral somitic frontier during body wall muscularization. I hypothesize that in amphibians the growth of the primaxial myotome arrests early in development, and all muscles of the ventro-lateral body wall derive from migratory somitic cells that mix with lateral plate. This scenario raises the possibility that in amphibians the abaxial domain evolved a novel role in organizing flank musculature. Given the experimental tractability of Xenopus and axolotls, and the evolutionary divergence of body wall formation in vertebrates, I propose that flank muscle ontogeny presents an intriguing system for studying the significance of the lateral somitic frontier in evolution and development. This work will add to a growing literature that explores the patterning autonomy of mesodermal derivatives in vertebrates, and the link between developmental modularity and morphological evolution. 172

184 Appendix A: Heterotopic transplantation experiments in axolotls provide preliminary data that trunk somites are competent to form abaxial appendicular musculature INTRODUCTION Heterotopic transplants of paraxial mesoderm have been used to test the patterning autonomy of primaxial derivatives in avians (reviewed in Burke and Nowicki, 2003). For instance, when pre-somitic mesoderm is moved from the thoracic region to the cervical region, it gives rise to normal primaxial structures such as the thoracic vertebrae, proximal ribs, and intercostal muscles (Kieny et al., 1972; Nowicki and Burke, 2000; Nowicki et al., 2003). In contrast, somitic cells that cross the frontier and develop within lateral plate connective tissue appear to be patterned by their new environment. In the above example, sternal ribs, which are abaxial somite derivatives, fail to form. Thus, unlike primaxial vertebral ribs, which are patterned autonomously, the somitic cells that form sternal ribs require critical environmental cues not present in the cervical region. The formation of limb musculature provides the most well-studied example of the role of lateral plate connective tissue in patterning migratory somitic cells (reviewed in the Introduction). Notably, somites from any axial level are capable of populating limb muscles in avians, and lateral plate connective tissue forms a pre-pattern of limb muscle architecture even in the absence of migratory somitic cells. 173

185 In Chapter Four of this thesis, I describe the position of the lateral somitic frontier in the pectoral and inter-limb regions of axolotls, and preliminarily diagnose the primaxial/abaxial identities of the muscles of the shoulder and body wall. The serratus anterior is primaxial, and extends from the primaxial vertebral margin of the scapula to the ribs. In contrast, the muscles of the shoulder and limb are abaxial. Using these new data, I set out to perform heterotopic somite transplants in axolotls to test the modularity of the primaxial and abaxial domains in a representative amphibian. My hypothesis was as follows: 1) Pectoral-level somites transplanted to inter-limb levels will form an ectopic serratus anterior muscle (primaxial derivative), but will not form shoulder muscles (abaxial derivatives); 2) Inter-limb somites transplanted to the pectoral region will populate shoulder and limb muscles (abaxial derivatives) but will not form a serratus anterior. Herein, I present preliminary data that indicate somites 3 5 form the muscles of the forelimb in axolotls, with possible contributions from somites 2 and 6. When pectoral level somites were transplanted to the inter-limb region, no ectopic muscle was apparent, contra my original predictions. However, most surgeries (5/7) exhibited grossly deformed body walls making interpretations of phenotypes challenging. Similar severe body wall defects were not present in isotopic controls. When inter-limb somites were transplanted to the pectoral region, grafted cells contributed to limb and shoulder muscles in two of five surviving surgeries. 174

186 MATERIALS AND METHODS Somite transplants were performed using surgical techniques similar to those described in Chapter four of this thesis, with the following modifications. Two to three recently formed somites were targeted for each transplant, and ectoderm was not included in grafted tissues. Isotopic, isochronic transplants were performed as a negative control for heterotopic challenge experiments. Chimeras were raised for approximately four months to Stage 57. MF20 antibody was used to immunohistochemically label skeletal muscle to evaluate muscle phenotypes. The staining protocol was as previously described in Chapter Two of this thesis. RESULTS AND DISCUSSION In order to identify which somites normally contribute to limb musculature, I performed a series of isotopic somite transplants using GFP-positive donors and wildtype hosts. Two to three somites were transplanted during each surgery, and axial levels were staggered between surgeries to collectively include somites two through twelve. Chimeras were fixed at Stage 57, approximately 4 months post-surgery, and evaluated by whole mount fluorescence and MF20 labeling for skeletal muscle. In total, 50 isotopic transplants were performed, with an overall survival rate of 48% (24/50). Of the surgeries that survived, 75% were GFP-positive (18/24). See Table 1 for the total number of surgeries at each axial level. 175

187 Isotopic, isochronic grafts of somites 2 4, 3 5, 4 6, and 5 7 contribute GFPpositive cells to appendicular musculature, whereas grafts of somites 7 9, 8 10, and 9 11 do not (Fig. 1; Table 1). Notably, transplanted somites at levels 3 5 and 4 6 populate all major limb and shoulder muscles, including the serratus anterior, deltoideus scapularis, procorocoideus, supracoracoideus, and pectoralis. This is consistent with a previous report that the serratus anterior receives contributions from pectoral level somites (Piekarski and Olsson, 2011). Together, these data indicate that paraxial mesoderm posterior to somite seven does not normally form appendicular musculature. Future work isotopically transplanting single somites will further clarify how somitic mesoderm segregates during development of the axolotl shoulder. For anterior-to-posterior heterotopic transplants, two to three recently formed somites from the pectoral region (between somites 3 6) of a donor were moved to the inter-limb region (between somites 7 12) of a host. In total, ten surgeries were performed, with a 70% survival rate (7/10), all of which had incorporated graft. Isotopic transplants of inter-limb somites were used as a negative control. Phenotypes were evaluated with MF20 labeling for skeletal muscle. See Table 2 for the total number of surgeries at different axial levels. Five of seven chimeras that survived to Stage 57 exhibit grossly deformed body walls. In the two surgeries with relatively normal body wall musculature, no ectopic muscles are discernible (e.g., Fig. 2A D). The absence of an ectopic serratus 176

188 anterior muscle does not support the developmental autonomy of primaxial muscular derivatives in axolotls, and suggests that the patterning of somites may be highly regulative. Future work to test the reproducibility of this data set and reduce surgery artifact will further clarify this issue. For posterior-to-anterior transplants, two to three recently formed somites from the inter-limb region of a donor (between somites 7 12) were transferred to the pectoral region (between somites 3 6) of a host. In total twelve surgeries were performed with a 58% survival rate (7/12). Of these, five exhibited graft incorporation. Phenotypes were again evaluated with MF20 labeling for skeletal muscle, and isotopic pectoral-to-pectoral transplants were used as a negative control. See Table three for the total number of surgeries at each axial level. GFP positive cells contribute to limb musculature in two of five trunk-topectoral chimeras (e.g., Fig. 2E H). Although these data are consistent with the hypothesis that somitic cells crossing the frontier are re-patterned by their new environment, the penetrance of graft contributions to abaxial musculature is low and additional experiments are warranted. 177

189 Figure 1. Axolotl chimeras following isotopic transplantation of somitic mesoderm from GFP donors to WT hosts. Panels A D and E H show a single chimera at time of surgery (A, E), one day post-surgery (B, F), and St. 57 (C, D, G, H). A D. Donor somites 3 5 contribute to limb and shoulder musculature. E H. Donor somites 7 9 contribute to body wall musculature, but not limb or shoulder musculature. In D and H MF20 labels skeletal muscle (purple). Brackets indicate axial levels containing GFP. 178

190 179

191 Figure 2. Axolotl chimeras following heterotopic transplantation of somitic mesoderm from GFP donors (green) to WT hosts. A D. Pectoral to trunk heterotopic transplant. A single chimera at time of surgery (A), one day post-surgery (B), and St. 57 (C, D) following transplantation of donor somites 4 6 to host positions Graft is well-incorporated into body wall musculature. No ectopic muscles are apparent. Chimera is shown in left lateral view. E H. Trunk to Pectoral heterotopic transplant. A single chimera at time of surgery (E), one day post-surgery (F) and St. 57 (G,H) following transplantation of donor somites 9 11 to host positions 3 5. Graft derived muscles are present in the limb and shoulder. Chimera is shown in left lateral view in E and F, and ventral-oblique view in G and H. In D and H MF20 labels skeletal muscle. pect, pectoralis; sup, supracoracoideus. 180

192 181