Developmental Signaling by Noggin and Wnt in the Frog Xenopus. John Joseph Young. A dissertation submitted in partial satisfaction of the

Transcription

1 Developmental Signaling by Noggin and Wnt in the Frog Xenopus By John Joseph Young A dissertation submitted in partial satisfaction of the requirements for the degree of Doctor of Philosophy in Molecular and Cell Biology in the Graduate Division of the University of California, Berkeley Committee in charge: Professor Richard M. Harland, Chair Professor Sharon Amacher Professor Michael S. Levine Professor Michael Freeling Spring 2013

2 Developmental Signaling by Noggin and Wnt in the Frog Xenopus 2013 By John Joseph Young

3 Abstract Developmental Signaling by Noggin and Wnt in the Frog Xenopus By John Joseph Young University of California, Berkeley Professor Richard M. Harland, Chair Xenopus has provided a powerful system to study cellular, developmental, and neurobiology. The availability of their embryos and the advent of modern molecular techniques allowed investigators to revisit the observations of classical embryologists and begin to determine the molecular mechanisms underlying germ layer formation and axis induction. My thesis work took advantage of the frog Xenopus to first address the developmental role of Noggin, a Bone morphogenic protein (Bmp) antagonist, and then to determine the mechanism of Wnt-induced anterior-posterior patterning of the neural plate. The frog Xenopus, an important research organism in cell and developmental biology, currently lacks tools for targeted mutagenesis. In the first part of this work, I address this problem by genome editing with zinc finger nucleases (ZFNs). ZFNs directed against an egfp transgene in X. tropicalis induced mutations that are consistent with results of non homologous end joining at the target site, resulting in mosaic loss of fluorescence phenotype at high frequencies. ZFNs directed against the noggin gene produced tadpoles and adult animals carrying up to 47% disrupted alleles. Founder animals yielded progeny that carry insertions and deletions in the noggin gene with no indication of off-target effects. Furthermore, functional tests demonstrated an allelic series of activity among three germline mutant alleles. Breeding an identified null allele to homozygosity resulted in tadpoles with deformaties in the cranial skeleton. Anatomical analysis revealed severe reductions in Meckel s cartilage with joint fusions. Gene expression analysis via in situ hybridization for chondrogenesis regulating factors in noggin mutants revealed a reduction in sox9 and col2a expression domains. Analysis of Bmp targets showed an expansion of hand2, edn1, and msx2 in the pharygeal arches (PAs) of mutants. This suggested a mechanism whereby incresed Bmp signaling inhibits chondrogenesis and ventralizes the PAs resulting in the jaw deformities observed in mutants. Neural development in amphibians occurs as a two-step process. First, ectodermal precursors adopt a neural fate in the absence of Bmp signaling. A second signal is then required to pattern the anterior posterior neuraxis. Signaling through Fibroblast growth factor (Fgf), retinoic acid (RA), and Wnt have each been demonstrated to be both necessary and sufficient for inducing 1

4 posterior fates in undifferentiated neural tissue. Wnt signaling in particular has been closely studied. However, the mechanism by which this pathway induces posterior fates remains unclear. To address this question, I used RNA-Seq to identify direct transcriptional targets in neural tissue by activating Wnt signaling in Xenopus neural explants pretreated with the translation inhibitor cycloheximide. Wnt-activated neural tissue resulted in over 200 genes with expression increased greater than 2-fold when compared to anterior neural tissue. in situ hybridization analysis of highly expressed transcription factors and RNA-binding proteins showed posterior expression. Of particular interest, the transcription factor sal-like 1 (Sall1) and sal-like 4 (Sall4) showed specific posterior neural expression suggesting a role in Wnt-induced neural patterning. The RNA-Seq screen found sall1 and sall4 expression to be induced by canonical Wnt signaling sall4 were enriched in β-catenin chromatin imunoprecipitations. Knockdown of Sall4 resulted in the loss of spinal cord marker expression and an increase in the expression of pou25, pou60 and pou91 (pouv genes), the three Xenopus homologs of the stem cell factor pou5f1/oct4. Overexpression of the pouv genes resulted in the loss of spinal cord identity, and knockdown of pouv function restored spinal cord marker expression in Sall4 morphants. Finally, knockdown of Sall4 blocked the posteriorizing effects of Fgf and retinoic acid signaling in the neurectoderm. These results suggest that Sall4, activated by Wnt signaling, represses the pouv genes to provide a permissive environment that allows for additional Wnt/Fgf/RA signals to posteriorize the neural plate. 2

5 To Nick Duesbery, Ph.D and Jeff McKelvey, Ph.D. I m grateful to have been your student, proud to be your colleague, and most of all, honored to be your friend. i

6 Table of Contents Table of Contents...ii Table of Figures...iv Acknowledgements...vi Chapter 1: General Introduction...1 Classical Embryology...1 The Molecular Era and Xenopus...2 Open questions...8 Goals of this thesis...9 Chapter 2: Materials and Methods...11 Embryo and explant culture...11 RNA and morpholino microinjections...11 Western Blotting...12 Celery Extract Preparation...12 Mutation Detection by Celery Extract...12 Cartilage staining...13 Cycloheximide and dexamethasone treatments...13 Whole-mount in situ hybridization...13 Embedding and Sectioning...14 RT-PCR and qpcr...14 RNA-seq...14 ii

7 Chromatin immunoprecipitation...14 Chapter 3: Analysis of the developmental role of noggin in Xenopus tropicalis development via Zinc-Finger Nuclease mutagenesis...16 Introduction...16 Results...17 Discussion...21 Chapter 4: Expression screen for direct targets of Wnt signaling in neural tissue...46 Introduction...46 Results and Discussion...47 Chapter 5: Spalt-like 4 mediates Wnt-induced neural patterning via repression of pouv/oct4 family members...68 Introduction...68 Results...69 Discussion...73 References...97 Appendices I: RNA-Seq results from Chapter 3: Genes with >2-fold expression (direct Wnt activation vs. anterior neural) II: List of PCR primers used in this work III: List of DNA plasmids used in this work iii

8 List of Figures Figure 3.1: Disruption of the egfp transgene in Xenopus tropicalis using ZFNs...24 Figure 3.2: Tolerance and activity of ZFNs targeting noggin in Xenopus tropicalis...26 Figure 3.3: ZFN-driven editing of the noggin locus in Xenopus tropicalis...28 Figure 3.4: ZFNs induce heritable loss-of-function noggin alleles mutations...30 Figure 3.5: Knockdown of Chordin and Follistatin results in a loss of dorsal structures in a subset of embryos produced by heterozygous noggin mutant adults...32 Figure 3.6: Stage series of representative embryos produced by heterozygous noggin mutant adults...34 Figure 3.7: Homozygous noggin mutant Xenopus tropicalis have severe lower jaw deformities...36 Figure 3.8: Expression of chondrogenic factors in wild-type and noggin mutant tadpoles...38 Figure 3.9: Expression of Bmp pathway targets in wild-type and noggin mutant tadpoles...40 Figure 4.1: Model of Wnt-induced patterning of the neural anterior-posterior axis...52 Figure 4.2: Temporal expression of anterior posterior neural markers in Xenopus tropicalis...54 Figure 4.3: TVGR activates canonical Wnt-signaling...56 Figure 4.4: TVGR efficiently posteriorizes neuralized ectodermal explants...58 Figure 4.5: Cycloheximide treatment prior to TVGR induction enriches for direct Wnt targets in neuralized ectodermal explants...60 Figure 4.6: Expression patterns of transcription factors identified in the screen for direct Wnt targets...62 Figure 4.7: Expression patterns of RNA-binding factors identified in the screen for iv

9 direct Wnt targets...64 Figure 4.8: Expression patterns of identified Wnt targets prickle1 and lmo Figure 5.1: sall4 is a direct transcriptional target of canonical Wnt-signaling...77 Figure 5.2: Injected embryos express functional FLAG-tagged β-catenin...79 Figure 5.3: sall4 is expressed in the neurectoderm...81 Figure 5.4: sall1 is directly activated by canonical Wnt signaling and expressed during early embryogenesis...83 Figure 5.5: Loss of Sall4 results in a loss of posterior neural differentiation...85 Figure 5.6: cdx2 is directly activated by canonical Wnt signaling and not affected by Sall4 knockdown...87 Figure 5.7: Knockdown of Sall4 causes an increase in expression of the pouv/oct4 homologs...89 Figure 5.8: A second non-overlaping Sall4 morpholino results in similar phenotypes...91 Figure 5.9: Loss of spinal cord gene expression in Sall4 morphants requires an increase in pouv/oct4 expression...93 Figure 5.9: FGF and retinoic acid signaling fail to posteriorize Sall4 morphants...95 v

10 Acknowledgements First off, I want to thank Nikki, I could not have done any of this without you. I m so excited for our lives and all the adventures we ll have together. Mom, you always said I could do this, especially when I didn t think I could. Thank you, I ll always remember Melinda Mae. Dad, thanks for all the encouragement and that great weekend of baseball, darts, and beer. Em, I m so lucky to have you as my sister. I have to thank my grandmother June, the first frog biologist I ever knew. I love you all dearly. My advisor Richard Harland, I knew I wanted to join your lab even before coming to Berkeley. You ve been an excellent mentor; kept me from flying too high, but propped me up when I needed it. Thank you for giving me the freedom to explore my questions. I m so proud to have been in your lab. To the members of my committee: Sharon Amacher, Mike Levine, and Mike Freeling, thank you for your guidance, insight, and support. I m entirely grateful to the past members of the Harland lab crew: Andrea and James, I learned so much about how to be a scientist from you, also how to play MarioKart, thank you for the discussions and distractions. Mike, even if it can t be Slayer, let s take the Prius to another metal show. Sara, thank you for scaring me first and then becoming a great friend and colleague. Isa and Jen, I m always ready for a beer lunch. My sincerest thanks to Jess, words truly fail here, I can t imagine grad school without you. To my current labmates: Debbie, the warm, comfortable environment of the lab is in no small part due to you. Thank you for making this an amazing place to work. Darwin, I ll always think of you when I need someplace to store a box of tubes and Dave, we ll get that pig next time. Caitlin and Sofia, you re up next, I m looking forward to hearing about all of your discoveries. Peter, a fellow PBR lover, I m so glad I got to know you. Hyeyoung and Lisa, thank you for setting high standards for science in our lab. Stefanie, Cameron, and Rachel you re part of the best lab in MCB, I know you ll do amazing science. I have been fortunate to work with some wonderful undergraduate students here at Berkeley. Gloria, Daniel, and Sofia, I have learned so much from you. Your futures are exciting and I wish you the best in your careers. I want to thank my lifelong friends Jason and Jeff. You are my real life heroes, thank you for all of our times together, good and bad. You ll always be my brothers. Finally, I couldn t have remained sane without Copsound: Brock, Justin, Dave, Alberto, Josh, Blair and Mike. Thank you for giving me the opportunity to rock with you guys. Your creativity and energy inspired me both in lab and at the space. vi

11 CHAPTER 1 General Introduction Classical Embryology How an animal develops from a fertilized egg to a recognizable multicellular organism is a fundamental question in developmental biology. This question has intrigued countless biologists for thousands of years prompting many theories, both ingenious and ridiculous. Aristotle first used the term epigenesis to describe chick development as a series of steps whereby structures progressively induce the formation of other structures, ultimately giving rise to a complex organism with multiple cell types and tissues. While this seems like basic knowledge in modern times, his interpretations were swept aside in favor of the creationism-friendly preformationist idea that tiny humunculi are present in the germ cells. Battles raged between the ovists and spermists as to which gamete truly contained the humunculus until the mid 19th century when Schleiden and Schwann s theory that the fertilized egg formed a cell that through division gave rise to all the cells in the body began to gain acceptance. Today, the combined theories of Aristotle and Schleiden and Schwann form the core of modern developmental biology. Every embryo contains the intrinsic information to build itself from within via cell division and epigenesis, however the question remains, how? Embryonic induction, where a specific tissue or cell(s) induces the fate of a different tissue or group of cells, provides a general mechanism for epigenesis as described by Aristotle. The experiments of Hans Spemann and Warren Lewis demonstrated the first embryonic induction when they found that grafting an optic cup from a frog embyo to an ectopic location was sufficient to induce lens formation in the overlying epidermis (Spemann, 1901) (Lewis, 1904). Spemann then turned his attention from specific organ development to axis formation by grafting pieces of gastrulae embryos to ectopic locations in host embryos. Spemann made his greatest discovery when Hilde Mangold, a student in his lab, grafted the dorsal lip from a gastrula of the lightly colored newt Triturus cristatus to the ventral region in a gastrula of the darkly pigmented newt Triturus taeniatus. This graft generated an ectopic axis but the important discovery here was that the secondary axis was comprised of a lightly pigmented notochord and floorplate and darkly pigmented somites and neural tube. The darkly pigmented cells must have come from the host and they made the conclusion that the grafted tissue induced the surrounding host tissue to adopt a dorsal fate rather than a ventral one (Spemann and Mangold, 1924). This result prompted the idea of regional specification where cells in the dorsal lip of the gastrula, known as the organizer, specify the fates of other cells in the region (Spemann, 1938). The discovery of mesendodermal tissue that induce surrounding cells to adopt alternate fates was not limited to amphibians. Waddington s discovery that transplantation of Hensen s node from chicks would induce ectopic neural tissue in host ectoderm showed this structure to be the avian equivalent of Spemann s organizer (Waddington, 1932; Waddington, 1933). The shield in teleosts (Oppenheimer, 1936), and finally the node in mammals (Beddington, 1994) were also found to 1

12 be capable of patterning surrounding tissue during gastrulation. These discoveries provided great insight into how an embryo generates its pattern but they opened several more questions about how these structures are themselves induced and how they exert their effects on neighboring tissue. Pieter Nieuwkoop s discovery that ectoderm will change fate and give rise to mesodermal derivatives when placed next to endoderm provided insight on how the embryo generates the different germ layers (Nieuwkoop 1969). Furthermore, he noticed that the induced mesodermal derivatives were different depending on the dorsal/ventral nature of the endoderm with which the ectoderm was combined. Ventral endoderm induced blood and smooth muscle fates whereas dorsal endoderm was capable of inducing nearly all tissues in the embryo (Nieuwkoop 1969) (Boterenbrood 1973) (Dale and Slack, 1987). This observation led to the discovery of the socalled Nieuwkoop center, a portion of the dorsal endoderm that induces formation of the organizer. Transplantation of the dorsal-most endodermal cells at the 64-cell stage from embryos of the frog Xenopus laevis to the ventral side of irradiated embryos rescued axial structures in what would otherwise develop into a ball of ventral tissue known as a belly piece (Gimlich and Gerhart, 1984). Despite these incredible advances, the field of developmental biology would have to wait for the molecular era before the precise mechanisms of cell specification and embryonic inductions could be determined. The Molecular Era and Xenopus The field of experimental embryology was revolutionized by the adoption of the South African Clawed Frog Xenopus laevis as a model system. In the early 20th century, the discovery that injection of a pregnant female s urine into the dorsal lymph sac of Xenopus induced ovulation provided a robust and reliable test for pregnancy (Crew, 1939). This resulted in the export and housing of these frogs in medical institutions worldwide (Gurdon and Hopwood, 2000). The byproduct of this discovery was that amphibian eggs and therefore embryos could be made available year round and in great numbers. Researchers no longer had to wait for the breeding season to obtain their often limited experimental material. The availability of embryos and the advent of modern molecular techniques allowed investigators to revisit the observations of classical embryologists and begin to determine the molecular mechanisms underlying germ layer formation and axis induction. In amphibians, the endoderm forms mostly in the vegetal region, the mesoderm forms in the marginal zone and the ectoderm comes from the animal region. Nieuwkoop s experiments suggested that factors from the endoderm induce the overlying ectoderm to adopt a mesendodermal fate. Basic Fibroblast growth factor (bfgf) was identified as having weak mesoderm inducing activity that was enhanced when combined with Transforming Growth Factor-β (TGFβ) (Kimelman and Kirschner, 1987) (Slack et al., 1987). It was found that a pellet of Xenopus tissue culture cells had the same mesoderm inducing properties as dissected endoderm on ectodermal explants (animal caps) (Smith, 1987). This led to the discovery that the TGFβ family member ActivinA was a potent mesoderm inducer. Treatment of animal caps with 2

13 ActivinA alone was sufficient to induce mesodermal differentiation (Smith et al., 1990). Furthermore, increasing doses of ActivinA resulted in more dorso-anterior characteristics of the induced mesoderm (Green and Smith, 1990). While Fgf and ActivinA were demonstrated to induce mesoderm in animal caps, it s more likely that the Xenopus nodal-related (Xnr) factors signaling through Smad2 are the molecules that induce mesoderm in the frog embryo. The Xnrs are expressed in the marginal zone and overexpression of a Nodal-specific version of the inhibitor Cerberus, Cerberus-short, blocks mesendoderm formation in embryos (Agius et al., 2000; Joseph and Melton, 1997; Kessler and Melton, 1995; Piccolo et al., 1999). Nodal induction of mesoderm is not limited to amphibians. The most direct evidence that Nodal signaling induces mesendoderm in vertebrates was provided by forward genetic screens in the zebrafish Danio rerio. Mutations in cyclops, squint and one-eyed pinhead, genes which encode the nodal homologs and their receptor, result in a failure of mesendoderm and organizer formation and can be rescued by injection of these factors(feldman et al., 1998; Gritsman et al., 1999). In both the chick and the mouse, Nodals induce mesoderm and their antagonist Lefty restricts signaling to prevent multiple primitive streak formation (Bertocchini and Stern, 2002; Conlon et al., 1994; Perea-Gomez et al., 2002; Skromne and Stern, 2001; Skromne and Stern, 2002; Zhou et al., 1993). Nodal secretion alone from the endoderm, however, does not explain the induction of the organizer. While high doses of ActivinA were able to induce dorsal mesoderm, this was not likely the mechanism employed by the embryo. The organizer is comprised of cells that will give rise to the notochord and head mesoderm in amphibians. Nodal signaling, while required for mesoderm induction, is necessary for organizer formation but not sufficient. Therefore, a dorsal modifying signal was proposed to be present in the embryo that would give the mesoderm induced in the presumptive dorsal region the inductive activity of the organizer. Strong evidence that this signal is mediated by Wnt signaling was provided when expression of Wnt8 on the future ventral side of the Xenopus embryo was able to induce a second axis (Smith and Harland, 1991; Sokol et al., 1991). Lineage tracing the injected cells revealed cells that inherited the wnt RNA comprised the notochord, pharyngeal tissue, and portions of the somites, reminiscent of dorsal lip grafts done with different newt species by Spemann and Mangold (Spemann and Mangold, 1924). Activation of the Wnt pathway results in the nuclear accumulation of β-catenin which complexes with T-cell factor/lymphocte enhancer factor (TCF/LEF) to activate transcription of target genes, reviewed in (Logan and Nusse, 2004). Depletion of the maternal pool of β-catenin transcripts resulted in embryos that lacked dorsal derivatives of the mesoderm and ectoderm (Heasman et al., 1994; Wylie et al., 1996). Several wnt transcripts are maternally deposited including wnt8b (Cui et al., 1995), wnt5a (Moon et al., 1993), and wnt11 (Ku and Melton, 1993). Through the movements driven by cortical rotation, these molecules, initially located at the vegetal pole of the egg, get displaced to the presumptive dorsal side during the first cell cycle of the zygote. Similar to β-catenin-depleted embryos, depletion of wnt11 from oocytes prior to fertilization resulted in embryos lacking dorsal structures (Heasman et al., 2000; Tao et al., 2005). These results showed that Wnt signaling is the dorsal modifying signal in the embryo. Recently, it has been proposed that maternally-provided wnt8 is necessary for organizer induction in zebrafish (Lu et al., 2011). 3

14 The combined action of Wnt and Nodal signaling on the presumptive dorsal side is responsible for the Nieuwkoop center s inductive activity in the blastula. These signaling pathways lead to activation of specific transcription factors and signaling molecules whose expression domain will come to define the organizer. At the blastula stage of Xenopus embryos, there is a gradient of activated Smad2 that begins on the dorsal side and spreads ventrally (Lee et al., 2001). This initial burst of Smad2 signaling on the dorsal side is the result of the early Wnt signal cooperating with maternally-provided VegT. xnr3 is directly activated by Wnt/β-catenin (McKendry et al., 1997) and stabilized maternal β-catenin on the dorsal side serves to prime the promoters of xnr3,5, and 6 for activation via recruitment of Histone Methyltranferases (Blythe et al., 2010). The combined activity of Wnts and Nodals on the future dorsal side serve to activate organizer specific transcription factors in the dorsal mesoderm. The 5 region of the paired-like homeobox gene siamois (sia) contains a proximal element with TCF/LEF binding sites and the closely related gene twin also has a distal element that is regulated by Smad2 (Houston et al., 2002; Laurent et al., 1997; Nishita et al., 2000). Signaling by Wnt and Nodal synergistically activates twin expression in the organizer only on the dorsal side(nishita et al., 2000). Similarly, the homeobox gene goosecoid (gsc) has a proximal region with TCF/LEF binding sites that is also bound by Twin, which serves to mediate gsc expression in response to the Wnt signal (Laurent et al., 1997). A distal enhancer with Smad2 binding elements also mediates gsc expression but requires cooperative binding with members of the Mixer homeodomain family of transcription factors (Germain et al., 2000). Bmp, however, restricts gsc to the dorsal mesoderm; injection of noggin RNA into ventral mesoderm results in a broader gsc expression domain (Eimon and Harland, 1999). Expression of gsc or sia on the ventral side of embryos results in the induction of an ectopic organizer and the generation of twinned embryos (Cho et al., 1991; Lemaire et al., 1995). While the identification of organizer-specific transcription factors led to an understanding of how the organizer is specified, it was the discovery of secreted molecules from this specialized group of cells that provided the mechanism for axis induction observed by Spemann and Mangold in their grafting experiments. Through expression screens using cdna libraries constructed either from dissected organizers or whole embryos treated with a dorsalizing agent such as LiCl, several secreted factors were found to have organizer-like activity. The first such factor to be discovered was noggin. It is normally expressed in the organizer and when expressed in UVventralized embryos, was capable of fully rescuing anterodorsal structures (Smith and Harland, 1992; Smith et al., 1993). Furthermore, treatment of animal caps with Noggin protein was sufficient to induce a neural fate (Lamb et al., 1993). Within a short time, other molecules with the same neural inducing activity were cloned including chordin (Sasai et al., 1994) and follistatin (Hemmati-Brivanlou et al., 1994). These factors bind to and inhibit Bmps (Zimmerman et al., 1996). Upon this discovery, Piccolo (Piccolo et al., 1996) suggested a mechanism for organizer function whereby the organizer cells which give rise to the head mesoderm and the notochord secrete Bmp antagonists to inhibit Bmp signaling in nearby tissue. From these experiments, and earlier ones where the Activin receptor was inhibited (Hemmati- Brivanlou and Melton, 1994), the default model for neural induction was proposed, stating that 4

15 ectoderm will adopt a neural (dorsal) fate in the absence of any inducing signal (Hemmati- Brivanlou and Melton, 1997). Consistent with this model, genetic knockouts of noggin and chordin result in mice lacking dorsoanterior-most structures however, they still retain neural tissue(bachiller et al., 2000). This result suggested that there are other Bmp antagonists that function redundantly to induce neural tissue. Knockdown of three Bmp antagonists in Xenopus tropicalis resulted in a complete loss of dorsal structures and yielded fully ventralized embryos (Khokha et al., 2005). The presence of several Bmp antagonists that act redundantly in the early embryo has made assessing their individual role challenging. There is strong evidence for the default model. In vitro culture of excised animal caps results in the adoption of an epidermal fate, however dissociation and reaggregation of caps result in a neural fate (Godsave and Slack, 1989; Grunz and Tacke, 1989; Sato and Sargent, 1989). Neural induction following dissociation can be blocked by adding Bmp to the culture medium (Wilson and Hemmati-Brivanlou, 1995). These studies provided a model where Bmp signaling through the Smad1/5/8 transcription factors serve to induce ventral fates while cells with low Bmp adopt dorsal fates. The key experiment that supported this model was performed when a truncated form of the activin receptor (which acted as a dominant negative for Bmp signaling) was expressed in animal caps and neural tissue was induced (Hemmati-Brivanlou and Melton, 1992). Taken together, these experiments support the interpretation that neural tissue is induced by the removal of Bmp and is thereby the default differentiation pathway of the ectodermal precursors. The default model can be applied to the other germ layers. Bmp signaling induces ventral fates: blood, body wall muscle and hindgut in the mesoderm and endoderm, respectively. Dorsal fates such as the somites and pharyngeal endoderm are induced when Bmp is blocked (Harland s chapter in (Stern, 2004). If one were to carry this model to a logical conclusion, then depletion of the mesoderm and endoderm inducer along with Bmp inhibition would result in the entire embryo adopting a neural fate. This intriguing hypothesis awaits experimental testing. Despite broad acceptance of the default model for neural induction, some challenges have arisen to the interpretation that Bmp inhibition alone is sufficient to induce ectoderm to adopt neural fates. Injection of a dominant negative Fgf receptor blocked the neural inducing activities of Noggin and Chordin which prompted the hypothesis that Fgf signaling is required for neural induction (Launay et al., 1996; Sasai et al., 1996). The finding that simply cutting the animal cap leads to activation of MAPK, the transducer of Fgf, called into question the interpretation of experiments where animal caps were treated with BMP antagonists (LaBonne and Whitman, 1997). Furthermore, inhibition of Smad1 also required Fgf signaling to induce neural fates in ventral epidermis (Delaune et al., 2005). This however, is due to the activity of Smad2, as inhibition of Smad1 and Smad2 is sufficient to induce neural fates ventrally (Chang and Harland, 2007). An alternative explanation could be that Fgf-mediated activation of MAPK results in phosphorylation of the linker region of SMAD1 causing its export from the nucleus and down regulation of Bmp signaling (Fuentealba et al., 2007). Recently, it was proposed that the animal cap is unsuitable for neural induction studies because of a pre-pattern imposed by other inductive signals (Linker et al., 2009). This is unlikely to be the case, animal caps will form neural tissue in the presence of a small molecule Fgf inhibitor when treated with a Bmp antagonist (Wills et 5

16 al., 2010). The source of the discrepancy is likely due to the use of sox2 as a marker for neural tissue differentiation. While differentiated neural cells express sox2, a population of uncommitted stem cells also express sox2 which revert to epidermis in the absence of Fgf (Wills et al., 2010). In addition to the Bmp antagonists, screens also identified secreted Wnt antagonists expressed in a subset of organizer cells that give rise to the head mesoderm (Niehrs, 2004). The inhibitor Dickkopf (Dkk), which blocks Wnt signaling by binding to the Wnt co-receptor Lrp5/6 and thus preventing signal transduction (Semënov et al., 2001) is expressed in a subset of organizer cells (Glinka et al., 1998). Overexpression of Dkk results in enlarged heads and reduced trunks whereas loss-of-function experiments showed Dkk to be required for anterior neural development (Glinka et al., 1998; Mukhopadhyay et al., 2001). Similarly, ectopic expression of the Frizzled-related protein, Frzb induced anterior structures by binding to Wnt8 and preventing it from binding to its receptors Lrp5/6 and Frizzled (Wang et al., 1997). Several additional secreted Frizzled-related proteins were discovered in a screen for secreted proteins produced by the organizer (Pera and De Robertis, 2000). As expected, these factors result in enlarged anterior structures when overexpressed. The identification of the head inducer Cerberus as an inhibitor of Wnts, Bmps, and Nodals (Piccolo et al., 1999) led to a model of axis induction by the organizer where the Bmp antagonists induce a dorsal fate via repression of Bmp signaling and the anterior is induced by inhibition of Wnt, Bmp, and Nodal signaling (reviewed in (Niehrs, 2004). The inactivation of the repressor Tcf3 in zebrafish headless mutants further provided genetic evidence that Wnt inhibition is required for head formation (Kim et al., 2000). The default model, now supported by molecular evidence, is not completely novel. Johannes Holtfreter and Pieter Nieuwkoop predicted that neural patterning was a two-step process whereby the ectodermal precursors are first activated to adopt a neural fate (by default anterior in nature), and then additional signals from the mesoderm posteriorize it to create the full anteriorposterior (A-P) pattern of the neural plate (HOLTFRETER, 1947; Nieuwkoop, 1952; Xenopus receptor results in anteriorized Xenopus 6

17 posterior fates. By using hox resulted in ectopic hoxb1 and a loss of krox-20 in Raldh mutant mice, and the raldh2 neckless can all be rescued by a uniform concentration of gradient not by simple diffusion but by degradation anteriorly by the Cyp26 proteins. increasingly more posterior neural markers with increasing doses of bfgf (Kengaku and hoxb9 12 caps only induces the more anterior gene enrailed2 (Lamb and Harland, 1995). This result is consistent with a model where the posterior neurectoderm is in contact with the Fgf source early in gastrulation and diffusion of the ligand reaches more anterior regions as the process of This source/sink mechanism generates a gradient that in combination with the competence of Xenopus Finally, Wnt/β-catenin signaling is the third transforming factor involved in posterior patterning of the neural plate. Activation of this pathway during gastrulation represses anterior development in contrast to its role of inducing the dorsoanterior axis in the early embryo. Animal caps expressed hoxb9 (spinal cord) and krox20 (hindbrain) following treatment with both Noggin and Wnt3a (McGrew et al., 1995). Introducing graded amounts of Dishevelled to neuralized animal caps resulted in increasingly more posterior fates which suggested that graded Wnt-signaling patterns the neural plate (Itoh and Sokol, 1997). Additional evidence to support that a morphogen gradient of Wnt serves to pattern the neural plate emerged when overexpressing the Wnt inhibitor Dkk in Xenopus embryos resulted in an expansion of the anteriorly expressed genes bf-1 and otx2 and a posterior shift of krox20. Conversely, increasing 7

18 Wnt concentration results in the opposite phenotype: an anterior shift of krox20 and a repression of anterior neural gene expression (Kiecker and Niehrs, 2001). Consistent with a graded level of Wnt signaling in the neural plate, Kiecker and Niehrs (Kiecker and Niehrs, 2001) observed an anterior-to-posterior gradient of nuclearly localized β-catenin in the neural plate of a gastrula staged embryo. Several lines of evidence point to Wnts coming form the underlying paraxial mesoderm to posteriorize the neural plate. In zebrafish, Wnt8 is expressed in the paraxial mesoderm and knockdown results in a loss of posterior neural gene expression (Erter et al., 2001; Lekven et al., 2001). Wnt3a in the dorsal paraxial mesoderm of Xenopus directly activates meis3 in the hindbrain (Elkouby et al., 2010). Ultimately, knockout studies in mice provide genetic evidence for Wnt-signaling in A-P patterning. Mice express wnt3a and wnt5a in the paraxial mesoderm and mutants lack posterior structures (Greco et al., 1996; Yamaguchi et al., 1999). Knockout of the Wnt antagonist dkk1 results in mouse embryos that lack forebrains (Mukhopadhyay et al., 2001) while genetically increasing β-catenin in the forebrain results in transformation to more posterior fates (Paek et al., 2012). The above experiments suggested that Wnt acts as a classical morphogen to pattern the A-P axis of the neural plate. While this may indeed be the case, visualization of a graded Wnt ligand has not been reported. Furthermore, several findings have suggested that the ligand is poorly soluble. Biologically active Wnt3a requires glycosylation and palmitoylation (Komekado et al., 2007; Takada et al., 2006; Willert et al., 2003), post-translational modifications that make it hydrophobic. While co-expression with Wnt antagonists results in greater diffusion of the Wnt ligand in Xenopus (Mii and Taira, 2009), the different expression domains of these antagonists suggest that the function of Wnt antagonists is unlikely to resolve the paradox of how an insoluble protein can have long range effects. The duration of Wnt-signaling as apposed to the concentration of the ligand provides an alternative explanation to a morphogen gradient of Wnt patterning the amphibian neural plate. The more posterior regions stay in contact with the Wnt source for longer periods during gastrulation than anterior or medial regions. Certainly a uniform concentration of ligand signaling for different durations can have variable outcomes. Premature inactivation of Sonic Hedgehog (Shh) signaling in the zone of of polarizing activity (ZPA) results in a loss of posterior fates in the autopods of mice (Harfe et al., 2004). The Australian two-toed skink expresses Shh in the ZPA for a shorter duration than its five-toed relative (Shapiro et al., 2003). Therefore, it is possible that uniform doses of Wnt applied to neurectoderm for different durations may elicit a full repertoire of anterior and posterior markers. While this has yet to be carefully tested, it remains an intriguing possibility. Open questions Several questions of neural patterning remain unanswered. The lack of reverse genetic mutagenesis techniques has confounded the study of noggin in amphibian development. Lossof-function studies were restricted to the use of morpholino oligonucleotides (MOs) that, while effective in protein knockdown, are transient and not useful for examining later phenotypes. MO knockdown of Noggin in Xenopus does not result in abnormal phenotypes while noggin mutant mice have severe neural and skeletal deformities (Brunet et al., 1998; McMahon et al., 1998). 8

19 Either noggin is dispensable in amphibian development or current methods are not sensitive enough to determine a role for this gene. The controversy over Bmp antagonist and neural induction has cooled in recent years, however the contribution of individual Bmp antagonists in amphibian development remains unanswered. The development of targeted mutagenesis techniques via zinc-finger nucleases, TALENs, and CRISPR presents an opportunity to mutate these genes and study their roles using genetic nulls. The identification of Wnt, Fgf, and RA signaling as posteriorizers of the neural plate opens a series of questions. What are the targets of these pathways and how are these targets regulated? A handful of target genes have been identified, but there is yet no clear consensus on how graded levels of these morphogens elicit specific expression domains of their targets. A thorough understanding of the response elements that mediate expression will give insight to the mechanisms of A-P patterning. Modern methods such as RNA- and ChIP-seq, combined with the classical embryology techniques offered by Xenopus, make this organism an excellent model for probing these questions. Xenopus has proven to be an important model for developmental biology. Its large embryos are well-suited for cut and paste experiments which provided some of the first demonstrations of embryonic inductions. Expression screens and the identification of organizer-specific genes made Xenopus extremely useful at the advent of modern molecular biology. With the dawn of the genomics age, this genus continues to be a powerful system for understanding gene function in development. The complete sequencing of the West African species Xenopus tropicalis provided the first genome of an amphibian (Hellsten et al., 2010). This brought Xenopus tropicalis into the genomics era and made whole-genome experiments possible. The recent sequencing and assembly of the Xenopus laevis genome will provide the frog community as well as the developmental biology community at large with powerful tools to uncover developmental mechanisms and address these unanswered questions. Summary of this thesis The work in my thesis took advantage of both the classical embryology and genomic resources offered by Xenopus. My goals were two-fold: (1) to establish reverse genetic strategies in the frog at two proof-of-principle loci, including one implicated in embryo dorsalization and (2) to utilize state-of-the art genomic approaches to investigate the role of Wnt in neural posteriorization. I was fortunate to collaborate with the biotechnology company Sangamo BioSciences that specializes in zinc-finger nuclease (ZFN) and TALEN design to induce mutagenic DNA double-strand breaks at specific loci in a genome. I used this technology to first demonstrate the efficacy by inactivating a Green Fluorescent Protein transgene. Next, I used ZFNs designed to target the noggin locus. It s well known that Noggin is sufficient to dorsalize tissue, but evidence for its requirement in amphibian development was lacking. I was able to successfully mutate the locus and generate several lines of Xenopus tropicalis carrying noggin alleles with different levels of activity. Through breeding of a null allele to homozygosity, I 9

20 found noggin to be required for cranial skeletal development, specifically in the dorsoventral patterning of the pharyngeal arches. Having demonstrated a role for noggin in dorsoventral patterning, I next turned to A-P patterning of the neural plate. As mentioned above, Wnt signaling is both necessary and sufficient to induce posterior fates in neural precursors, yet the mechanism remains poorly understood. I hypothesized that neural posteriorization via Wnt is mediated through transcriptional regulation of target genes. Therefore, it was necessary to identify genes directly regulated by this pathway. To that end, I carried out an expression screen using RNA-seq aimed at discovering genes that are directly activated by Wnt in the neural plate. This screen was successful; I found over 200 genes that were upregulated in response to Wnt. In situ hybridization analysis of selected genes showed a majority to be expressed in posterior regions of the embryo. Of particular note, two Spalt-like (Sall) transcription factor family members identified in the screen, sall1 and sall4, showed robust expression in posterior neural regions. The genes identified in this screen provide a basis for understanding the link between activation of the Wnt pathway and posterior patterning of the neural tube. The final chapter of my thesis focuses on the function of Sall4, one of the Sall transcription factors identified in the screen for direct neural targets of Wnt. Sall4 s role in mammalian stem cell maintenance has been well documented but no role in A-P patterning has been described. I found that sall4 is specifically expressed in the neurectoderm and TCF/LEF sites found in the first intron of sall4 were enriched in β-catenin chromatin immunoprecipitations. Morpholino oligonucleotide knockdown of Sall4 resulted in a loss of spinal cord development and an upregulation of the pouv/oct4 homologs. Ecotopic expression of pouv/oct4 was sufficient to block neural posteriorization and reducing pouv/oct4 in Sall4 morphants rescued spinal cord development. Finally, I found that Sall4 knockdown was epistatic to posteriorization by both Fgf and RA signaling. This data presented a novel model of Wnt-induced neural patterning whereby Wnt sends a permissive signal by activating sall4 in order to repress the pouv/oct4 genes. The neural plate is then competent to respond to instructive signals from Wnt/Fgf/RA following the down-regulation of the pouv/oct4 genes. 10

21 CHAPTER 2 Materials and Methods Embryo and explant culture Xenopus laevis embryos were collected, fertilized, and cultured according to (Sive et al., 2010) and staged according to (Nieuwkoop, 1967). Xenopus tropicalis embryos were collected from natural matings. To induce mating, female Xenopus tropicalis were injected with a priming dose period. Ectodermal explants (animal caps) were cut using fine watchmaker s forceps from stage 9 embryos and cultured in ¾ NAM. RNA and morpholino microinjections All ZFN plasmids were linearized by restriction enzyme digest to produce transcripts with (AscI) or without (NotI) in vitro with a ZFN RNAs were microinjected into both blastomeres of two-cell staged Xenopus tropicalis amounts injected. Embryos tend to orient animal pole uppermost, so injection penetrated the animal hemisphere, but mrna was deposited near the center of the blastomere. Broad noggin RNA was LacZ RNA and cultured in 1/3 Sall4 CS-108, Fgf8a CS-108, noggin CS-108, and β-catenin CS-108 were linearized with Asc1 and transcribed with a Sp6 mmessage mmachine kit (Ambion). The PouV genes (a gift from Joshua Brickman), TVGR (Darken and Wilson, 2001), and nuclear β-galactosidase CS2+ were linearized with Not1 and transcribed with Sp6. All RNAs were injected in either 5 or 10 ηl bursts along with GFP and β-galactosidase RNAs to serve as tracers. All morpholinos were injected in either 5 or 10 ηl bursts along with fluorescein-labeled control morpholino (Gene Tools) to serve as a tracer. The Sall4 morpholino sequences are as follows: morpholino 1: 5 - GCCAATTATTCCCTTTCTCCACCAC-3 and morpholino 2: 5 - GGTTCGGCTGCTTTCTCCTCGACAT-3. 11

22 Western Blotting Embryos were lysed in buffer containing 20 mm Tris ph 8.0, 50 mm NaCl, 2 mm EDTA, 1% Triton X-100 and freshly supplemented with μl lysis buffer per embryo. control. Celery Extract Preparation for Cel-1 assays (Apium graveolens var. dulce) were cut into 3-4 cm 2 pieces and μ μ μ Mutation Detection by Cel-1 from Celery Extract μl µg/ml proteinase K. Lysates were μ supernatant was decanted and the pellet allowed to air dry. Once dry, the pellet was resuspended 12

23 denatured and re-annealed by incubating at 94 C for 5 minutes followed by cooling at a rate of -2 C/sec until the reaction reached 25 C, then incubated at Cartilage staining Tadpoles were fixed in 4% paraformaldehyde for 2-24 hours at room temperature in 4 ml vials. Paraformaldehyde was decanted and the embryos were suspended in a sterile filtered solution of acid/alcohol (70% ethanol and.37% HCl) containing 0.1% Alcian Blue. Vials were placed on a rotator and gently mixed for 6-12 hours at room temperature when staining of the cartilage elements becomes apparent. When staining was complete, the buffer was discarded and tadpoles were resuspended in the acid/alcohol solution without alcian blue and rotated for 20 minutes at room temperature. This was repeated until the solution no longer had any blue tint. Tadpoles were then rehydrated stepwise into water and then bleached in 1X SSC supplemented with 1.2% hydrogen peroxide and 5% formamide for 1-2 hours on a white-light table. Vial caps were removed to prevent excessive bubble formation. Following bleaching, tadpoles were resuspended in a 2% KOH solution and rotated for 1 hour. Stained tadpoles were cleared by successive 2 hour incubations in 2% KOH with increasing concentration of glycerol. Once cleared, tadpoles were either directly imaged or flat-mounted on a microscope slide by fine dissection and imaged. Cycloheximide and dexamethasone treatments Noggin RNA and an inducible Wnt agonist, TVGR RNA were injected animally into both blastomeres of two-cell embryos. Embryos were cultured until stage 9 when animal caps were excised and cultured with or without 10 µm dexamethasone (Sigma) to activate Wnt signaling. To block protein translation, animal caps were pre-treated with 5 µm cycloheximide (Sigma) for 1.5 hours prior to dexamethasone addition. Animal caps were cultured until the stage 15 equivalent and total RNA was harvested using Trizol (Invitrogen). Whole-mount in situ hybridization Embryos were stained after whole mount in situ hybridization as described in (Harland, 1991). β-galactosidase staining was carried out as described in (Fletcher et al., 2006). 13

24 Embedding and Sectioning Embryos for sectioning were first equilibrated into a 30% sucrose solution and then transferred into a PBS solution containing 20% sucrose, 30% BSA, 4.9% gelatin. Embryos for sectioning were quickly transferred to fresh buffer supplemented with 1.5% glutaraldehyde and allowed to harden in peel-away plastic molds. Embedded embryos were cut to about a 0.5cm 3 block with a razor blade and super-glued to an attack die from the game Heroscape. Mounted embryos were sectioned on a Pelco 101 vibratome while submerged in PBS. RT-PCR and qpcr RNA was isolated from either whole embryos or animal caps using Trizol according to standard protocols and 1µg total RNA was reverse transcribed with either MMLV reverse transcriptase (Promega) or iscript (BIO-RAD) for semi-quantitative or qpcr, respectively. Semi-quantitative PCRs included trace amounts of 32 P labelled dctp (Perkin-Elmer) in the reaction and were analyzed during the log-phase of amplification. qpcr reactions were amplified on a CFX96 (BIO-RAD) light cycler. ernithine decarboxylase (ODC) and eukaryotic elongation factor-1a1 (eef1α1) were used for internal controls. All primers annealed at 60 C and are listed in appendix II. RNA-seq RNA-seq was performed according to standard protocols from Illumina as described in (Dichmann and Harland, 2012). mrna was purified from 10 µg total using oligo-dt Dynabeads (Invitrogen) and fragmented using zinc ion fragmentation buffer (Ambion) for 1.5 minutes at 70 C. First strand synthesis was carried out according to standard protocols using Superscript II (Invitrogen) and second strand synthesis was performed using DNA polymerase I (NEB). cdna fragment ends were repaired and ends were adenylated using Klenow, T4 DNA polymerase and T4 PNK (NEB). Adaptors (Illumina) were ligated using T4 Ligase and Quick Ligase buffer (NEB). AMPure XP beads (NEB) were used to select for fragments larger than 200 bp. The library was amplified with Phusion HF polymerase (NEB) and single-end 76- basepair reads were sequenced on an Illumina Genome Analyzer II. All reads were mapped to an index created from a collection of full-length Xenopus laevis mrna sequences ( xgc.nci.nih.gov) using TOPHAT and BOWTIE (Langmead et al., 2009; Trapnell et al., 2009). Analysis of transcript abundance differences was done using CUFFDIFF (Trapnell et al., 2010). Chromatin immunoprecipitation FLAG-β-catenin-injected embryos for immunoprecipitation were fixed in 1% formaldehyde/pbs for 1 hr, quenched with M glycine/pbs for 15 minutes followed by three 10 minute washes in PBS. Lysis was performed according to (Blythe et al., 2009). Chromatin was sheared on ice using a Branson Model 450 digital sonifier with a Model 102C probe for 24 ten second bursts set at 30% amplitude. Immune complexes were pulled down using M2 FLAG antiboby 14

25 (Sigma) bound to anti-mouse magnetic beads (Invitrogen). Samples were washed, cross-links reversed and DNA isolated according to ChIP protocols (Blythe et al., 2009). ChIP DNA was quantified with SYBR-green PCR mix (BIO-RAD) on a CFX96 light cycler (BIO-RAD). Enrichment was calculated by comparing the %input between samples. Uninjected embryos served as a control for non-specific binding. xmlc2 (Blythe et al., 2009) and meis3 (Elkouby et al., 2010) served as negative and positive controls for β-catenin binding, respectively. 15

26 Introduction CHAPTER 3 Analysis of the developmental role of noggin in Xenopus tropicalis development via Zinc-Finger Nuclease mutagenesis Frogs of the genus Xenopus have been an important model organism for cell and developmental biologists since the 1930s (Gurdon and Hopwood, 2000). X. laevis is the standard model, but due to its allotetraploid genome, less suited for genetic approaches than the diploid X. tropicalis, whose genome sequence has been determined (Hellsten et al., 2010). Whereas embryological manipulations and gain-of-function experiments are major strengths of Xenopus, reverse genetics are currently limited to the use of antisense reagents that provide transient and often incomplete gene knock down (Eisen and Smith, 2008). The ability to introduce targeted, heritable mutations that disrupt gene function has remained elusive. This work provides a generally applicable solution to this problem: targeted gene disruption with designed zinc finger nucleases (ZFNs). ZFNs are the fusion of the non-specific cleavage domain of the Type IIS restriction enzyme FokI to a zinc-finger protein (Miller et al., 1985; Pavletich and Pabo, 1991) that is engineered to bind a specific genomic locus in order to induce a targeted double-strand break (DSB). Pioneering studies in oocytes of X. laevis (Bibikova et al., 2001) and subsequent work in Drosophila (Bibikova et al., 2002) showed the mutagenic potential of a DSB induced by ZFNs (reviewed in ref (Carroll, 2008; Urnov et al., 2010)). Resolution of ZFNinduced DSBs via non-homologous end joining (NHEJ) generates small insertions and deletions which often produce null or hypomorphic alleles (Bibikova et al., 2002; Perez et al., 2008; Santiago et al., 2008). The Bone morphogenic protein (Bmp) antagonist noggin is expressed in Spemann s organizer and is a potent dorsalizing factor (Smith and Harland, 1992; Zimmerman et al., 1996). Together with two other Bmp antagonists, Chordin and Follistatin, it serves to dorsalize the ectoderm and mesoderm to give rise to neural and somitic derivatives, respectively (Khokha et al., 2005). Targeted deletion of the noggin locus in mice results in perinatal lethality, spina bifida, and skeletal deformities (Brunet et al., 1998; McMahon et al., 1998). However, antisense morpholino oligonucletide (MO) knockdown of Noggin in Xenopus does not yield a phenotype. Whether this is due to incomplete knockdown or a role for Noggin beyond the effective window of MO activity remains unknown. Genetic inactivation of the locus is required to effectively assay the function of Noggin in amphibian development, a method previously restricted to nonspecific, forward genetic screens and laborious mapping efforts. I developed an effective protocol for gene disruption in X. tropicalis. Using ZFNs designed against a reporter transgene and the noggin locus, the delivery and expression conditions for ZFNs were optimized which resulted in high frequencies of somatic and germline mutations that 16

27 were transmissible to the next generation. Furthermore, ZFN induced mutations in noggin were bred to homozygosity which yielded an unexpected phenotype. Results Xenopus eggs are large and easily manipulated (Sive et al., 2010), offering the opportunity to deliver ZFNs via injection of mrna, a method that has been successful in driving ZFN-induced gene disruption in other organisms (Carroll, 2008; Urnov et al., 2010). To develop conditions for gene disruption in Xenopus tropicalis, transgenic animals carrying a single-copy GFP transgene were used(hamlet et al., 2006). Wild type X. tropicalis eggs were fertilized with sperm from a homozygous GFP transgenic male. The resulting heterozygous embryos were injected with mrna encoding ZFNs that target the egfp coding region (Geurts et al., 2009). Uninjected tadpoles express GFP robustly in the somites, lens, and head musculature (Figure 3.1 A-C). Injection of 20 pg egfp ZFN RNAs led to mosaic loss of fluorescence in otherwise healthy tadpoles (Figure 3.1 D-F). At a higher dose of ZFNs, most cells had lost fluorescence, suggesting efficient somatic mutation of the transgene (Figure 3.1 G-I). To determine whether loss of fluorescence resulted from a ZFN-induced mutation in the egfp transgene, the target locus was genotyped using an assay based on the mismatch-sensitive endonuclease, Cel-1 (Miller et al., 2007). This analysis (Figure 3.1 J) demonstrated that ZFNtreated, but not control, tadpoles had acquired a DNA sequence alteration in the stretch targeted by the ZFNs. Sequencing subsequently revealed that individual tadpoles often carried multiple distinct indels ranging from 5 to 20 bp centered over the ZFN recognition site (Figure 3.1 K), a signature of mutagenic NHEJ. Taken together, these experiments show that ZFN mrna injection into the two-cell embryo yields tadpoles without detectable developmental defects and exhibit both genetic and phenotypic mosaicism for the ZFN-targeted locus and trait, respectively. To determine whether this approach can be used to disrupt an endogenous gene, ZFNs that target the noggin locus were designed. Noggin is a Bmp antagonist that contributes to dorsal/ventral patterning during gastrulation in Xenopus (Smith and Harland, 1992). Although its function in later development has been studied in human patients (Marcelino et al., 2001) and mice (Bachiller et al., 2000; Brunet et al., 1998; McMahon et al., 1998; Warren et al., 2003), its developmental role in non-mammalian vertebrates remains poorly understood. Therefore, mutant alleles of the endogenous noggin gene are required to probe its role throughout amphibian development. A panel of ZFNs targeting noggin was designed by Sangamo Biosciences Inc., screened in a budding yeast proxy system (Doyon et al., 2008; McCammon et al., 2011), and cloned into an expression construct that allows the synthesis of an efficiently translated mrna with a stabilizing polyadenylation signal (Turner and Weintraub, 1994). Embryos were injected in the animal pole with the RNA deposited in the center of each blastomere at the two-cell stage, raised to stage 40, and DNA was isolated from tadpoles that exhibited broad mcherry (i.e., tracer) 17

28 expression. Use of ZFNs that carry a wild-type FokI endonuclease domain yielded a significant fraction of embryos with developmental defects (Figure 3.2 A). However, this was not observed when RNA expressing the same zinc finger DNA recognition domains fused to the obligate heterodimer forms of FokI was injected. The point mutations in FokI are made in the nuclease domain to prevent the formation of a homodimeric and functional nuclease (Miller et al., 2007). These mutations, E490K, I538K and Q486E, I499L, are made in the FokI domain of the left and right ZFN, respectively (referred to as EL+KK). Even at the highest tested doses of such ZFN mrna, greater than 60% of the injected embryos developed normally (Figure 3.2 A). To optimize the delivery of ZFNs, I also tested whether unstable, non-adenylated RNA, which would be translated early and deliver a transient burst of ZFN, might be superior to the extended expression of ZFNs from transcripts that are cleaved and polyadenylated after injection (Turner and Weintraub, 1994). As expected, the non-adenylated transcripts led to considerably less protein expression (Figure 3.2 C). Even at higher doses, they did not induce noggin gene disruption at a frequency measurable by Cel-1. Thus, the prolonged presence of ZFNs from adenylated RNA is superior for effective genome editing in Xenopus. Because it is difficult to predict a priori the extent to which ZFN overexpression may cause embryonic defects, Sangamo tested a panel of ZFNs in a yeast-based single-strand annealing assay (Doyon et al., 2008) and I tested them in embryos (Figure 3.2 B, Table 3.1). Six ZFN pairs shown to have activity in the yeast assay were chosen for testing in Xenopus. The ZFN pairs found to be active in yeast and well tolerated in tadpoles also showed efficient genome editing in Xenopus embryos as measured by sequencing noggin amplicons from injected tadpoles. Since no Xenopus strains carrying noggin mutations existed, it wasn t possible to screen for phenotypes on a heterozygous background (Doyon et al., 2008), therefore I screened the ZFNs for activity by genotyping the targeted region using the Cel-1 endonuclease (Figure 3.3 A). All of the ZFNs that were well tolerated produced targeted gene disruption. Direct sequencing of the nuclease-targeted region in tadpoles injected with ZFNs revealed a broad panel of insertions and deletions ranging in size from 5 to 195bp (Figure 3.3 B), with frequencies of mutant amplicons from 10-47%. Such high rates of somatic mutagenesis suggested that tadpoles might also carry mutations in the germline, allowing the establishment of lines carrying novel noggin alleles. I injected wild-type embryos at the two-cell stage with 100 pg of mrna encoding ZFNs that target noggin and were tolerated by greater than 60% of injected embryos (Figure 3.2 B). Successful somatic genome editing in tadpoles and froglets was confirmed by isolating genomic DNA from tail or toe clips, respectively, and genotyping the noggin locus by Cel-1 and sequencing (Table 3.1). Injected embryos were raised to sexual maturity and outcrossed to wildtype animals. Offspring from this cross were raised to stage 40, lysed and analyzed via Cel-1 for mutations in the noggin locus. Genotyping offspring from a cross using a ZFN-treated male founder revealed that three of eighteen tadpoles had inherited a ZFN-induced Δ12 allele of noggin (Figure 3.4 A,B). A second male founder produced six out of fifty embryos heterozygous for a ZFN-induced three base pair insertion allele (Figure 3.4 A,B) and a third male produced 18

29 twelve of fifty tadpoles heterozygous for a four basepair insertion. The latter ZFN-induced noggin allele induces a frameshift mutation that results in a premature stop codon at position 55. Therefore, it is likely to behave as a null allele because the resulting protein lacks most of the Bmp binding residues and all of the residues required for dimerization (Groppe et al., 2002). The recovery of these mutations in the offspring of adult animals raised from injected embryos demonstrates that ZFN-induced alleles of an endogenous gene can be transmitted to the next generation. Of significant note, the parent and heterozygous tadpoles carrying mutant alleles were indistinguishable from wild type siblings, indicating this is an effective approach to establish lines of animals carrying novel alleles of investigator-specified genes. To test whether ZFN-induced mutagenesis caused off-target mutations, gynogenotes from noggin ZFN injected females that lacked mutant noggin alleles in their germline were made. These were siblings of the germline mutated males, and from clutches that showed a high frequency of somatic mutation. Gynogenesis diploidizes activated eggs by preventing the extrusion of the second polar body and serves to homozygose recessive mutations (Khokha et al., 2009). While offtarget mutations would result in high frequencies of mutant embryos in gynogenotes produced from these females, I did not detect such mutations; indeed 87% of gynogenotes were indistinguishable from wild-type tadpoles and the remaining 13% had various mediolateral and dorsoventral defects, consistent with reported phenotypes and frequencies from young wild-type females (Grammer et al., 2005). This result demonstrates that potentially confounding off-target mutations in founder animals are negligible. Noggin induces ectopic dorsal tissue when expressed on the ventral side of Xenopus embryos (Smith and Harland, 1992), which serves as an excellent test for whether the ZFN-induced noggin alleles produce loss-of-function proteins. Full-length mutant alleles were cloned and corresponding mrnas for injection were synthesized. 5 or 10pg was injected into the ventral vegetal blastomeres of 4-cell stage Xenopus laevis embryos (Figure 3.4 C). Embryos were cultured to stage 28 and scored for the presence or absence of an ectopic axis (Figure 3.4 D). The four basepair insertion allele failed to induce any ectopic dorsal tissues in embryos (Figure 3.4 H,I,J,J ), consistent with a loss-of-function frameshift mutation. LacZ mrna was used as a tracer and showed that injected cells populated the ventral posterior of injected tadpoles. The 12 allele induced ectopic axes in 8 and 16 percent of embryos when injected with 5 pg and 10 pg mutant noggin RNA, respectively. Interestingly, the induced axes in these embryos were underdeveloped, suggesting that this allele functions as a hypomorph (Figure 3.4 M arrowheads). Finally, the three basepair insertion allele was indistinguishable in activity from wild-type noggin (Figure 3.4 N,O,P). These results demonstrate that specific loss-of-function mutant lines of X. tropicalis can be generated via targeted ZFN mutagenesis. Furthermore, the mutations in noggin form the basis of an allelic series that will be useful to examine noggin function in later development. Knocking down Noggin, Chordin, and Follistatin results in a loss of dorsal structures and yield embryos without neural or somitic derivatives whereas knocking down any two in combination results in a redution but not a complete loss of these structures (Khokha et al., 2005). To test if 19

30 the 4-basepair insertion allele is truly a null allele, heterozygous adults were crossed and the resulting embryos were injected with morpholinos targeting Chordin and Follistatin. These embryos were raised to stage 14 and assayed for sox2 and myod expression. Uninjected embryos expressed sox2 throughout the neural plate and myod in the underlying mesoderm (Figure 3.5 A,D). Chordin and Follistatin morpholino injection resulted in a reduction of both sox2 and myod in ~75% of the embryos (Figure 3.5B,E). However, in approximately 25% of injected embryos, presumably those homozygous for the 4-bp insertion allele, there was a complete loss of both sox2 and myod expression (Figure 3.5C,E). These results, taken with the failure of RNA bearing the 4-basepair insertion noggin allele to induce ectopic axes suggest that the 4-bp insertion noggin allele is a true functionally null allele. noggin null mice die at birth due to numerous defects including open neural tubes and several skeletal abnormalities including chondrocyte hypertrophy and joint formation failure (Brunet et al., 1998; McMahon et al., 1998). Conversely, morpholino knockdown of Noggin in Xenopus does not yield a phenotype (Khokha et al., 2005). This discrepancy can be due to either incomplete knockdown by the morpholino or because the requirement for Noggin occurs when morpholinos are no longer effective. To distinguish between these possibilities, heterozygous adults carrying the 4-bp insertion allele were crossed and the resulting embryos were cultured and observed for phenotypic abnomalities. Unexpectedly, I did not observe defects in neural tube closure in any of the embryos produced from heterozygous parents (Figure 3.6 A-J). No defects were observed in any embryos until stage 45 where a dorsal-rostral protuberance was noticed in a subset of the clutch. Sorting tadpoles based on the presence or absence of this dorsal horn resulted in 2533 (75.1%) wild-type and 824 (24.9%) abnormal (Figure 3.6 K-N and Figure 3.7 A-F). Genotyping 10 wild-type and 10 horned tadpoles revealed the horned tadpoles were homozygous for the noggin null allele. To determine the nature of the horn observed in noggin mutants, alcian blue staining was used to viualize the cartilage skelton of the cranium. Staining and flatmounting cartilage preparations from wild-type (Figure 3.7 G,G ) and mutant tadpoles revealed overall smaller cartilage elements, deformed ceratobranchial cartilage, and most notably severely reduced Meckel s cartilage that are fused to the palatoquadrate in mutant tadpoles (Figure 3.7 H,H ). There are no major differences in the superostral plate between the wild-type and mutant, indicating that the horn observed in the mutants is due to a deformation and loss of ventral structures (i.e. Meckel s catilage and ceratohyals) rather than an overgrowth of the superostral plate. Mutant tadpoles did not display sensory defects as they responded to gentle tapping on the culture dish. Finally, the mutants began to die off rapidly by two weeks post fertilization (Figure 3.7 I), presumably due to malnutrition because of an inability to feed effectively. Next, expression of the neural crest gene sox9 was examined because it directly activates col2a, which is required for cartilage differentiation(lefebvre et al., 1997). A reduction in cartilage differentiation could explain the overall smaller cranium size observed in mutants. Since the mutant phenotype doesn t manifest until after patterning and differentiation of the pharyngeal arches, it was necessary to genotype tadpoles prior to analyzing gene expression. To that end, 20

31 the heads from stage 33 and stage 39 tadpoles were excised and processed for in situ hybridization while the tails were genotyped. There was a marked reduction in sox9 expression (Figure 3.8 A,B) and a corresponding loss of col2a expression in the mandibular arch of mutants at stage 33 (Figure 3.8 C,D arrowhead), consistent with the observed reduction in Meckel s cartilage. However, there was no noticeable difference in col2a expression between wild-type and mutant tadpoles at stage 39 (Figure 3.8 E,F). Bmp signaling plays a major role in the dorsal-ventral patterning of the pharyngeal arches, which in turn develop into the cranial skeleton. Previous work in zebrafish has shown that Bmp signaling is sufficient to induce ventral arch fates in intermediate and dorsal regions (Alexander et al., 2011; Zuniga et al., 2011). Given that Noggin functions as a Bmp antagonist, I examined wild-type and mutant tadpoles for the expression of Bmp responsive genes that function in pharyngeal arch development. bmp7 is expressed in the ventral portion of the pharyngeal arches but did not show any appreciable differences between wild-type and mutant tadpoles at stage 33 or 39 (Figure 3.9 A-D). At stage 33, expression of the Bmp responsive gene msx2 (Hollnagel et al., 1999; Tríbulo et al., 2003) showed slight dorasl expansion (Figure 3.9 E,F) but was indistinguishable from wild-type at stage 39 (Figure 3.9 G,H). Another Bmp target, endothelin1 (edn-1) was not differently expressed at stage 33 between the two genotypes (Figure 3.9 I,J). However, it showed a more diffuse expression pattern in the mutant at stage 39 (Figure 3.9 K,L). Finally, the BMP target gene hand2 (Howard et al., 2000; Xiong et al., 2009) had a broader expression domain in the mutants at stage 33 (Figure 3.9 M,N) and was ectopically expressed in the mandibular arch of stage 39 mutants (Figure 3.9 O,P). In zebrafish, Hand2 is known to repress bapx1 which is required for joint formation between Meckel s cartilage and the palatoquadrate (Miller et al., 2003). Discussion These results show that Xenopus can be added to the growing list of important model organisms for which ZFN-encoding mrna has allowed facile reverse genetics, including Drosophila (Bibikova et al., 2001), zebrafish (Doyon et al., 2008; Meng et al., 2008), and the rat (Geurts et al., 2009; Mashimo et al., 2010). An important requirement for the use of ZFNs is a streamlined protocol to predict and implement effective gene disruption. This work confirmed that a proxy assay in budding yeast is efficient in identifying ZFNs that will function effectively in the developing embryo. Furthermore, I have determined an expression vector architecture and dose of mrna that allow optimal expression of the ZFNs to generate both tadpoles and fertile adult animals that carry disrupted alleles of target genes in both the soma and germline. Remarkably, the injection of ZFN mrna into a relatively small cohort of two-cell embryos was sufficient to raise adults carrying mutant alleles in the germline. Our results demonstrate that with optimal husbandry (Grammer et al., 2005), homozygous mutants can be generated within one year of the initial mutagenesis, and potentially 7 months if female founders are generated. 21

32 The Cys2-His2 zinc finger protein (Miller et al., 1985; Pavletich and Pabo, 1991) is the most common DNA recognition motif in metazoa, and ZFNs can be engineered against any locus of interest (Urnov et al., 2010). This work here showed that ZFNs built using an archive of prevalidated two-finger modules that either target egfp or noggin and carry high-fidelity FokI endonuclease domains (Miller et al., 2007) induced mutations at a high rate when injected at doses that were well tolerated by the majority of injected embryos. Direct sequencing of both targets showed a variety of indels that are likely to result in null alleles. Indeed, there was a ZFN-induced loss of egfp fluorescence phenotype in injected tadpoles that are heterozygous for the egfp transgene. The absence of a phenotype in founder noggin animals could be due to several factors, including the mosaic nature of the injected founders, non-autonomy of secreted Noggin, and a compensatory role of other Bmp antagonists in the early embryo (Khokha et al., 2005). However, artificial ventral expression of mrnas encoding the ZFN-induced noggin mutants demonstrated that ZFNs induced both null and hypomorphic alleles of an endogenous X. tropicalis gene. The normal development of gynogenotes derived from females with high frequencies of somatic mutations in noggin (but no detectable noggin mutations in the germline) shows that off-target mutations in the germline must be rare. Breeding the induced noggin mutant alleles to homozygosity revealed that Noggin functions in the development and differentiation of the pharyngeal arches. Specifically, Noggin appears to be required to restrict expression of the Bmp target gene hand2. Mutant tadpoles express hand2 in a broader domain, including the mandibular arch which gives rise to Meckel s cartilage. This provides a potential mechanism to explain the loss of joint formation between Meckel s cartilage and the palatoquadrate observed in mutant tadpoles. Hand2 represses the homeobox gene bapx1 (Miller et al., 2003). It is plausible that in noggin mutant tadpoles, the increase in Bmp signaling due to a loss of Noggin, results in the expansion of Hand2 and a concomitant repression of bapx1, causing the fusion of Meckel s cartilage to the palatoquadrate (Figure 3.7 H,H ). This may also explain the cause of death observed in mutants two weeks post fertilization. Meckel s cartilage forms the mandible and the severe reduction/loss of this structure in noggin mutants impedes the tadpole s ability to feed effectively and likely results in starvation. This is consistent with the inverse sigmoidal survival curve observed in the mutants. As mentioned above, noggin mutant mice show neural tube closure defects (McMahon et al., 1998). While the skeletal deformaties are consistent between mutant frogs and mice, why are there no defects in the neural tube observed in noggin mutant Xenopus tadpoles? One explanation could be the presence of other Bmp antagonists. Chordin and Follistatin are expressed in a similar domain during early embyogenesis of Xenopus (Khokha et al., 2005) and could compensate for the loss of noggin in neural patterning. Indeed, there is redundancy in the function of these Bmp antagonist during neural induction and dorsalization of the mesoderm (Khokha et al., 2005). A second explanation could be that the paralog noggin2 partially compensates for the loss of noggin (Fletcher et al., 2004). Genome editing using ZFNs has the potential to enable numerous lines of experimentation that were previously impossible with existing Xenopus methodology. Permanent, heritable mutations 22

33 will allow for the study of specific genes and later developmental processes without concern for the off-target or transient effects associated with morpholino oligonucleotides (Eisen and Smith, 2008). Furthermore, the variety and size of ZFN-induced indels can be used to generate an allelic series of mutations. We note that several noggin alleles we generated were deletions of substantial size (~200 bp), indicating that ZFNs provide an attractive method for not only disrupting specific coding sequences, but also for targeting regulatory elements in the genome. Recently, alternative methods for genome editing have been developed that have replaced the use of ZFNs. Transcription activator-like effectors (TALEs) from the plant pathogenic bacteria Xanthomonas that consist of repeated motifs, each of which binds a single nucleotide (Boch et al., 2009; Moscou and Bogdanove, 2009), have been fused with FokI nucleases to generate TALE-Nucleases (Cermak et al., 2011; Christian et al., 2010). The modular nature of TALENs allows for easy design and synthesis. Other groups have been successful in using TALENs to generate biallelic mutations in Xenopus tropicalis (Lei et al., 2012). The CRISPR-CAS method provides an additional alternative to nuclease fusion proteins for inducing targeted mutations. This system uses an RNA molecule engineered to target specific sites that flank common repeat sequences in the genome and recruit the Cas-9 endonuclease to induce double strand breaks (Hwang et al., 2013; Jinek et al., 2012). The CRISPR-CAS system relies on basepair complementarity and requires only two plasmids to synthesize targeting constructs. The zebrafish genome has CRISPR sites every basepairs thus, this system is likely applicable to any gene of interest (Blackburn et al., 2013). My thesis work demonstrates that induced mutations can result in heritable null mutations and allow for analysis of gene function in amphibian development. Due to the large size, abundance, and ready manipulation of their eggs and embryos, Xenopus has provided important insights in both cell biology (King et al., 1996) and embryology (Harland and Gerhart, 1997). Completion of the genome sequence of X. tropicalis brought this model system into the genomics age (Hellsten et al., 2010), but the genetic engineering tools essential for comprehensive study of biological mechanism were lacking. My studies add a robust method for genome editing so that gene function can be analyzed throughout development. 23

34 Figure 3.1: Disruption of the egfp transgene in Xenopus tropicalis using ZFNs. (A-C) Uninjected tadpoles (UC). (D-F) Tadpoles injected with 20 pg of egfp ZFN mrna and 200 pg mcherry RNA (to monitor injection). (G-I) Heterozygous egfp tadpoles injected with 50 pg egfp ZFN mrna and 200 pg mcherry RNA (tracer). (A,D,G) Brightfield. (B,E,H) egfp expression in tadpoles from A,D, and G, respectively. (C,F,I) Enlarged view of egfp expression in B, E, and H, respectively. (J) Cel-1 digestion of egfp amplicons. Bands migrating at 345 bp are full-length amplicons, Cel-1 cleavage products migrate at 246 bp and 99 bp. The fraction of modified chromatids detected by Cel-1 are quantified as % NHEJ. UC, uninjected control. (K) Sequence alignment of ZFN-induced mutant egfp transgene alleles from tadpoles injected with 50 pg ZFN mrna. Red nucleotides indicate insertions and dashes represent deletions. Horizontal bold lines indicate ZFN binding sites. 24

35 Figure 1 A D G B U.C. E 20 pg egfp ZFNs 50 pg egfp ZFNs H C U.C. F 20 pg egfp ZFNs I 50 pg egfp ZFNs egfp ZFN L J 20 pg 50 pg K 345 bp 246 bp UC Deletions egfp ZFN L egfp ZFN R Gln Leu Gln Gln Pro Gln Arg Leu Tyr His Gly Arg Gln WT CAACTACAACAGCCACAACGTCTATATCATGGCCGACAAG CAACTACAACAGCCACAACGT---TATCATGGCCGACAAG( 3) CAACTACA TCTATATCATGGCCGACAAG( 12) CAACTACAACAG CCGACAAG( 20) CAACTACAACAG CCGACAAG( 20) CAACTACAACAG CCGACAAG( 20) CAACTACAACAG CCGACAAG( 20) CAACTACAACAGCCACAACGT CCGACAAG( 11) CAACTACAACAGCCACAAC-----TATCATGGCCGACAAG( 5) CAACTACAACAGC CATGGCCGACAAG( 14) CAACTACAACAG CCGACAAG( 20) CAACTACAACAGC CATGGCCGACAAN( 14) CAACTACAACAG CCGACAAG( 20) CAACTACAACAG CCGACAAG( 20) CAACTACAACAG CCGACAAG( 20) CAACTACAACAGC CATGGCCGACAAG( 14) CAACTACAACAGCCACAA TCATGGCCGACAAG( 8) 99 bp % NHEJ Indels Insertions GAACTACAACAGCCACAACGTGTGT--CATGGCCGACAAG( 6,+4) CAACTACAACAGCCACAACAA------CA--GCCGACAAG( 12,+4) CAACTACAACAGT CATGGCCGACAAG( 15,+1) CAACTACAACAG----AA AGCCGACAAG( 19,+3) CAACTACAACAGCCACAAC----GGATCATGGCCGACAAG( 6,+2) CAACTACAACAGCCACAACGTCTATATATCATGGCCGACA(+2) 25

36 Figure 3.2: Tolerance and activity of ZFNs targeting noggin in Xenopus tropicalis. (A) Optimization of ZFN delivery in Xenopus tropicalis. EL+KK: ZFNs with the EL and KK modifications in the FokI domain (Miller et al., 2007), numbers represent different noggin ZFN pairs (Table 3.2). WT: ZFNs with wildtype Fok1 nuclease domains. PA-: ZFN transcripts lacking a poly-adenylation signal. UC: Uninjected control. (B) Comparison of activities of different noggin ZFN pairs in the yeast activity assay and in injected tadpoles. Tadpoles were injected with 100 pg of ZFN mrna. Yeast activity values represented as a percentage relative to expression of ZFNs targeting the human CCR5 gene (Perez et al., 2008). Activity in tadpoles as calculated by the percent of mutant amplicons sequenced from injected embryos. ND: No Data. (C) Western blot for FLAG tagged ZFN proteins. 26

37 A 100% Tadpole % Phenotype 80% 60% 40% 20% Yeast Assay % Control 0% B UC pg ZFN egfp ZFN EL+KK Noggin 26EL+28KK Noggin 60EL+66KK Noggin (WT) RNA Tadpole ND Normal Mild Axial Defects Severe Axial Defects Gastrulation Defects Lethal % Mutant Amplicons Relative Activity in Yeast ND Noggin (WT) PA- 28EL+29KK 30EL+31KK 26EL+28KK 58EL+66KK 60EL+66KK 63EL+66KK Tadpole % Phenotype C FLAG UC Noggin ZFN Pair Noggin Noggin Noggin Noggin egfp ZFN 26EL+28KK 60EL+66KK (WT) (WT)PA pg ZFN RNA 27

38 Figure 3.3: ZFN-driven editing of the noggin locus in Xenopus tropicalis. (A) Somatic mutations in noggin detected by Cel-1. Bands migrating at 450 bp are full-length noggin amplicons. Bands migrating at 300 bp and 150 bp bands are Cel-1 digest products. (B) Sequence alignment of noggin alleles induced by indicated ZFN pairs. Red nucleotides indicate insertions and dashes represent deletions. Horizontal bold lines indicate ZFN binding sites. EL +KK: ZFNs with the EL and KK modifications in the FokI domain (Miller et al., 2007), numbers represent different noggin ZFN pairs (Table 3.2). 28

39 A 100pg Noggin ZFN mrna Wild Type 58EL+66KK 60EL+66KK 63EL+66KK 450bp 300bp 150bp B WT 58EL+66KK 60EL+66KK 63EL+66KK Noggin ZFN L Noggin ZFN R Leu Ile Glu His Pro Asp Pro Ile Tyr Asp Pro Lys Glu Lys Asp TGGACCTTATTGAGCATCCGGATCCTATCTATGATCCCAAGGAGAAGGATCTTA TGGACCTTATTGAGCATCCGGATCC CAAGGAGAAGGATCTTA( 12) ATCTATGATCCCAAGGAGAAGGATCTTA( 181) TGGACCTTATTGAGCATCCGGA GAAGGATCTTA( 21) TGGACCTTATTGAGCATCCGGATCC----TATGATCCCAAGGAGAAGGATCTTA( 4) ATCTATGATCCCAAGGAGAAGGATCTTA( 195) TATGATCCCAAGGAGAAGGATCTTA( 95) GATCCCAAGGAGAAGGATCTTA( 101) TGGACCTTATTGAGCATCCGGATCC CAAGGAGAAGGATCTTA( 12) TGGACCTTATTGAGCATCCGGA TCTTA( 27) TATCTATGATCCCAAGGAGAAGGATCTTA( 80) TGGACCTTATTGAGCATCCGGATCC CAAGGAGAAGGATCTTA( 12) TGGACCTTATTGAGCATCCGGATCC CAAGGAGAAGGATCTTA( 12) TGGACCTTATTGAGC GATCCCAAGGAGAAGGATCTTA( 17) TGGACCTTATTGG AGGAGAAGGATCTTA( 26) TGGACCTTATTGAGCATCC CAAGGAGAAGGATCTTA( 18) GGAGAAGGATCTTA( 116) TGGACCTTATTGAGCATCCGGATCCTATCCTATGATCCCAAGGAGAAGGATCTT(+1) 58EL+66KK TGGACCTTATTGAGCATCCGGATCCTATATCTATGATCCCAAGGAGAAGGATCT(+2) TGGACCTTATTGAGCATCCGGATCCTATCTATCTATGATCCCAAGGAGAAGGAT(+4) 60EL+66KK TGGACCTTATTGAGCATCCGGATCCTATCCTATCTATGATCCCAAGGAGAAGGA(+5) 63EL+66KK TGGACCTTATTGAGCATCCGGATCCTATCCTATCTATGATCCCAAGGAGAGGGA(+1) 29

40 Figure 3.4: ZFNs induce heritable loss-of-function noggin alleles mutations. (A) Cel-1 digests of noggin amplicons from sibling heterozygous mutant and homozygous wild type F1 tadpoles produced from three mutant line founders. Bands migrating at 450 bp are full-length noggin amplicons. Bands migrating at 300 bp and 150 bp are Cel-1 digest products. (B) Sequence alignments of the targeted noggin locus from Cel-1-positive F1 mutants. Genomic and translated sequences are shown for each mutant line. Asterisk indicates a stop codon. Red nucleotides and amino acids indicate insertions and dashes represent deletions. (C) Schematic of synthetic RNA injections into ventral vegetal blastomeres of 4-cell stage embryos to test functionality of the induced mutant noggin alleles. (D) Quantification of secondary axis induction following wild-type or mutant noggin RNA and scored for presence of ectopic dorsal axes. Bars represent results of two (5 pg) or three (10 pg) independent experiments (±SD) White bars show 5 pg RNA injections, black bars show 10 pg RNA injections. Two asterisks indicate significantly different (p<0.01) from uninjected controls. (E-G ) Uninjected control embryos. (H-J ) Embryos injected with 10 pg of 4 bp insertion mutant noggin and 200 pg LacZ RNA. (K-M ) Embryos injected with 10 pg of 12 bp deletion mutant noggin and 200 pg LacZ RNA. (N-P) Embryos injected with 10 pg of 3 bp insertion mutant noggin and 200 pg LacZ RNA. (Q- S) Embryos injected with 10 pg wild-type noggin and 200 pg LacZ RNA. (E,H,K,N,Q) Dorsal view stage 19. (F,I,L,O,R) Dorsal view stage 28. (G,J,M,P,Q) Embryos stained with 12/101 antibody. (G.J) Lateral view. (G,J,M,P,Q) Dorsal view. Arrowheads show weak ectopic dorsal axis induction. 30

41 A Founder 1 Founder 2 Founder 3 E F G H WT H WT H WT 450bp 300bp G 150bp H I J J 4bp Insertion UC B Founder 1: 12bp Deletion WT GATCCTATCTATGATCCCAAG Mut GATCC CAAG WT EHPDPIYDPKEKDL Mut EHPD----PKEKDL K L M 12bp deletion WT CCGGATCC---TAT-CTATGA Mut CCGGATCCCAAGATCCCATGA WT EHPDPIYDPKEKDL Mut EHPDKIP* Founder 2: 3bp Insertion WT GGATC---CTATCTATGATC Mut GGATCTATCTATCTATGATC WT EHPDP-IYDPKEKDL Mut EHPDLSIYDPKEKDL C Founder 3: 4bp Insertion Mutant or Wild Type Noggin RNA D N O P 3bp Insertion pg 10 pg Q R S WT 0 UC 4bp Insertion ** 12bp Deletion ** ** ** ** % Embryos with Ectopic Axis 3bp Insertion WT 31

42 Figure 3.5: Knockdown of Chordin and Follistatin results in a loss of dorsal structures in a subset of embryos produced by heterozygous noggin mutant adults. (A-C) sox2 expression. (D-E) myod expression. (A,D) Uninjected control embryos. (B,C,E,F) Embryos injected with 20 ng Chordin morphino and 20 ng Follistatin morpholino. Dorsal views with anterior towards the top. 32

43 Uninjected Control 20 ng Chordin MO+ 20 ng Follistatin MO A B C myod sox2 9/9 8/10 2/10 D E F 6/6 11/15 4/15 33

44 Figure 3.6: Stage series of representative embryos produced by heterozygous noggin mutant adults. (A-J) Phenotypically wild-type embryos and tadpoles. (K-L) Wild-type tadpoles. Total number of tadpoles and percentage within each phenotypic class. (M-N) Abnormal tadpoles with dorsal horn. Total number of tadpoles and percentage within each phenotypic class. (A,B,D,F,I,K,M) Dorsal views. (C,E,G,H,J,L,M) Lateral views. Anterior to the left. 34

45 Stage 21/22 Stage 25/26 Stage 33/34 Stage 39 A B D F C E G Stage 41 Stage 44 Stage 45 I K M H J L N 2533 (75.1%) 825 (24.9%) 35

46 Figure 3.7: Homozygous noggin mutant Xenopus tropicalis have severe lower jaw deformities. (A-C) Wild-type stage 46 tadpoles. (D-F) Homozygous noggin mutant stage 46 tadpoles. Anterior towards the left. (A,D) lateral views. (B,E) dorsal views. (C,F) Ventral views. (G-H) Flat-mounted cartilage from wild-type (G) and mutant (H) tadpoles. (G -H ) Schematic of skeletal elements in (G-H). (I) Survival curve of 100 wild-type (closed boxes) and 100 mutant tadpoles open triangles). 36

47 A B C D E F G G H H Suprarostral Plate Infrarostral Cartilage Ceratobranchial Palatoquadrate Ceratohyal Meckel s Cartilage I Wild type Mutant Days post-fertilization Mutant Wild Type Mutant Wild Type % Survival 37

48 Figure 3.8: Expression of chondrogenic factors in wild-type and noggin mutant tadpoles. (A-B) sox9 expression in the head of a representative wild-type (A) and mutant (B) tadpole at stage 33. (C-F) col2a expression in the head of stage 33 (C-D) and stage 38 (E-F) tadpole. (C,E) Representative wild-type tadpoles and (D,F) mutant tadpoles. Anterior to the left and dorsal towards the top. 38

49 sox9 col2a Wildtype Mutant Wildtype Mutant A B C D E F 39

50 Figure 3.9: Expression of Bmp pathway targets in wild-type and noggin mutant tadpoles. (A-D) Expression of bmp7 in the head of representative wild-type (A,C) and noggin mutant (B,D) tadpoles. (E-H) Expression of msx2 in the head of representative wild-type (E,G) and noggin mutant (F,H) tadpoles. (I-L) Expression of edn-1 in the head of representative wild-type (I,K) and noggin mutant (J,L) tadpoles. (M-P) Expression of hand2 in the head of representative wild-type (M,O) and noggin mutant (N,P) tadpoles. (A-B,E-F,I-J,M-N) Stage 33. (C-D,G-H,K- L,O-P) Stage 38. Anterior to the left and dorsal towards the top. 40

51 bmp7 msx2 A Wildtype Mutant Wildtype Mutant B E F C D G H I edn1 hand2 Wildtype Mutant Wildtype Mutant J M N K L O P 41

52 Table 3.1: noggin mutation induction and generation of germline mutants by Zinc-Finger Nucleases in Xenopus tropicalis. Percent mosaic embryos in cohort calculated by Founder percent germline mutagenesis calculated by the percentage of heterozygous offspring produced by founder in an outcross. 42

53 Founder (Males) Noggin ZFN Pair Embryos injected in cohort % Mosaic embryos in cohort Adults raised in cohort Founder % germline mutagenesis Mutation 1 26EL+28KK bp Deletion 2 26EL+28KK bp Insertion 3 63EL+66KK bp Insertion 43

54 Table 3.2: noggin ZFN recognition sequences. sites. 44

55 Noggin ZFN Binding Sequence (underlined) ZFN Finger 1 Finger 2 Finger 3 Finger 4 Finger 5 Finger 6 AGCATCCGGATCCTAtctatGATCCCAAGGAGAAG 25729KK RSDNLSV RSANLTR RSDNLSV IRSTLRD TSSNLSR TCGTAGGCCTAGGATagataCTAGGGTTCCTCTTC 25728EL QSSDLSR TSANLSR RSDTLSE TSANLSR RSDYLTK ATCCGGATCCTATCTatgatCCCAAGGAGAAGGAT 25731KK TSSNLSR RSDNLSV RSANLTR RSDNLSV IRSTLRD TAGGCCTAGGATAGAtactaGGGTTCCTCTTCCTA 25730EL TSSNLSR RSDTLSE TSANLSR RSDYLST QNAHRKT TTATTGAGCATCCGGATCctatcTATGATCCCAAGGAGAAG 15026EL RSDNLSV RSANLTR RSDNLSV DNRDRIK QSSNLAR TSSNRKT AATAACTCGTAGGCCTAGgatagATACTAGGGTTCCTCTTC 15028KK RSDALST ASSNRKT QSSDLSR TSANLSR RSDTLSE TSANLSR TTATTGAGCATCCGGATCctatcTATGATCCCAAGGAGAAG 25766KK RSDNLSV RSANLTR RSDNLSV IRSTLRD TSGNLTR NRGNLVT AATAACTCGTAGGCCTAGgatagATACTAGGGTTCCTCTTC 25758EL RSDALST ASSNRKT QSSDLSR TSANLSR RSDTLSE TSANLSR TTATTGAGCATCCGGATCctatcTATGATCCCAAGGAGAAG 25766KK RSDNLSV RSANLTR RSDNLSV IRSTLRD TSGNLTR NRGNLVT AATAACTCGTAGGCCTAGgatagATACTAGGGTTCCTCTTC 25760EL RSDALST ASSNRKT QSSDLSR TSANLSR RSDTLSE TSANLSR TTATTGAGCATCCGGATCctatcTATGATCCCAAGGAGAAG 25766KK RSDNLSV RSANLTR RSDNLSV IRSTLRD TSGNLTR NRGNLVT AATAACTCGTAGGCCTAGgatagATACTAGGGTTCCTCTTC 25763EL RSDALST ASSNRKT QSSDLSR TSANLSR RSDDLSE TNSNRKR 45

56 CHAPTER 4 Expression Screen for Direct Targets of Wnt-Signaling in Neural Tissue Introduction (NieuwkoopOthers, 1952b), postulates that neural tissue (by default anterior in nature) is induced Xenopus (Lamb et al., 1993).. A role for the Bmp antagonists Chordin, Noggin (Lamb et al., 1993), and Follistatin in the induction step is well supported in Xenopus. Reduction of these three antagonists results in a complete loss of neural tissue in Xenopus induced by Bmp antagonism will adopt an anterior fate in the absence of additional signals. in Xenopus, al., 1995) and Fgf Xenopus was shown to marker hoxb1. and LRP6, leading to cytoplasmic enrichment and nuclear translocation of β-catenin. Once inside the nucleus, β hindbrain marker krox20 Xenopus. Furthermore, a gradient of nuclear localized β neural plate of late gastrula/early neurula stage frog embryos, 46

57 posteriorizes the neural plate from mesodem in the dorsal-lateral marginal zone of Xenopus embryos meis3, mesoderm, induces a more posterior neural fate such as spinal cord and intermediate doses induce more medial fates like mid and hindbrain (Figure 4.1). markers (e.g. hoxb9) posterior to hindbrain markers (e.g. krox20 Here, I describe a screen that I performed to Results and Discussion Before assaying for Wnt-induced posterior neural gene expression, it is important to determine the normal temporal and spatial expression of A-P neural markers at the onset of gastrulation through neurulation. Therefore, Xenopus tropicalis embryos were fixed and stained for the following factors: otx2, expressed in the forebrain and midbrain (Blitz and Cho, 1995), en2, expressed at the midbrain-hindbrain boundary (Hemmati-Brivanlou et al., 1990; Hemmati- Brivanlou et al., 1991), krox20, expressed in rhombomeres 3 and 5 (Bradley et al., 1993), and hoxb9, a marker for the spinal cord (Wright et al., 1990). Consistent with the two-step model for neural patterning, otx2 is expressed broadly throughout the presumptive neural plate at the onset of gastrulation but becomes more restricted to the anterior with time (Figure 4.2 A-F). The more posterior markers begin to be expressed at the onset of neurulation and the embryo exhibits a full A-P pattern by the mid neurula stage (Figure 4.2 G-X). These results show that the neural plate is fully patterned by stage 15 and therefore, an optimal stage to assay for Wnt targets in neural tissue. To allow temporal control over the activation of Wnt signaling, an inducible activator is required. I used a fusion protein with the DNA binding domain of TCF fused to the transactivating domain of VP-16 all in turn fused to a glucocorticoid receptor (TVGR) (Darken and Wilson, 2001). Canonical Wnt signaling is sufficient to specify the dorsal axis during cleavage stages of Xenopus embryos, and thus provides an excellent readout for Wnt activity (Smith and Harland, 1991; Sokol et al., 1991). To confirm the results of Darken and colleagues that TVGR efficiently activates Wnt-signaling (Darken and Wilson, 2001), TVGR RNA was injected into the ventralvegetal blastomeres of 4-cell stage embryos and the resulting tadpoles were assayed for secondary axis induction. Ventral-vegetal injection of TVGR alone or injection followed with 0.2% EtOH (vehicle) treatment resulted in normal development in 80% of injected tadpoles. The remaining 20% showed only partial secondary axes (Figure 4.3A-C). Conversely, dexamethasone (DEX) treatment of TVGR injected embryos resulted in 80% of injected tadpoles with robust secondary axes induction as measured by the presence of eyes in the ectopic axis 47

58 (Figure 4.3A,D). The results from this experiment show that TVGR is a potent activator of Wnt signaling upon DEX addition with minimal activation in the absence of the inducer. In order to discover Wnt targets in neural tissue, ectodermal explants (animal caps) were induced to become neural tissue via noggin expression followed by Wnt activation (Figure 4.4 A) Therefore, the activity of TVGR in posteriorizing neural tissue was tested by assaying for known A-P markers in animal caps under different treatments. As expected, uninjected animal caps showed high expression of the epidermal marker epidermal keratin. Animal caps treated with noggin alone expressed the anterior neural marker otx2 but not the epidermal marker epidermal keratin or any posterior neural gene, demonstrating that this tissue adopted an anterior neural fate. TVGR-injected animal caps that were neuralized by noggin showed robust expression of the posterior neural markers krox20 and hoxb9 upon induction with DEX. While EtOH treatment of TVGR-expressing animal caps express the hindbrain marker krox20, activation via DEX is required to induce spinal cord fates as assayed by hoxb9 expression (Figure 4.4 B). Consistent with these results, DEX treatment of neuralized animal caps expressing TVGR induced the caps to undergo convergent-extension like morphogenesis that is consistent with differentiation into spinal cord (Elul et al., 1997) (Figure 4.4 C). The observed morphogenesis and neurulation are not due to mesodermal contamination of the ectodermal explants as there is no expression of the mesodermal marker muscle actin. Next, I validated the use of TVGR in neuralized animal caps to find direct transcriptional targets of Wnt signaling. Detection of direct transcriptional targets requires inhibition of protein synthesis to prevent activation of secondary targets following translation of Wnt-induced transcripts. Cycloheximide (CHX) is a potent inhibitor of protein translation (Obrig et al., 1971) and provides a convenient tool to prevent the expression of secondary targets upon activation of a signaling pathway. To determine the dose of CHX I used was effective, I pretreated neuralized animal caps that were injected with TVGR with CHX, then induced with DEX, and assayed for known direct and indirect transcriptional targets of Wnt (Figure 4.5 A). CHX treatment prior to activation of TVGR did not prevent activation of the direct transcriptional target meis3 (Elkouby et al., 2010) but blocked expression of the indirect target hoxb9 (Domingos et al., 2001) (Figure 4.5 B). These results demonstrated that the conditions I used were sufficient to induce neural tissue, posteriorize it via Wnt activation and finally, enrich for direct targets. Next, total RNA from animal caps treated with noggin alone (anterior neural sample), neuralized caps with activated TVGR (posteriorized neural sample) and neuralized animal caps treated with CHX prior to TVGR activation via DEX addition (direct target sample) was harvested. The RNA from these samples were used to construct RNA-seq libraries which were then 76-basepair single-end sequenced. The resulting sequence reads were mapped to a collection of 10,088 nonredundant, full-length Xenopus laevis cdna sequences ( Data analysis was preformed by first aligning the reads using the programs TOPHAT(Trapnell et al., 2009) and BOWTIE (Langmead et al., 2009) to map the reads to cdna sequences and then using CUFFDIFF (Trapnell et al., 2010) to determine differences in read quantities between samples. This analysis identified 228 genes greater than 2-fold increased in the direct target sample when 48

59 compared to the anterior neural sample (Appendix 1). Enrichment of the direct target meis3 (Elkouby et al., 2010) as well as cdx2 (Wang and Shashikant, 2007) in the direct target sample reads provided confirmation that this screen and analysis were successful in identifying Wnt target genes and suggests that candidates are likely to be direct targets. While I found over 200 genes with enriched expression when directly activated by Wnt, this method of analysis was limited because the reads were mapped to a cdna library in lieu of a full genome. Any genes absent from this library would have been missed and it is likely that there are more potential Wnt targets to be identified. With the recent sequencing of Xenopus laevis, it is now possible to repeat the bioinformatic analysis using the genome and potentially reveal more candidate target genes. Furthermore, comparing the results from the screen described here to those obtained from β-catenin chromatin immunoprecipitation (ChIP) will provide a complementary approach for detection of direct Wnt targets. ChIP-Seq will also identify the enhancers and promoters that mediate Wnt regulation of the identified target genes allowing for insight into the mechanism of gene regulation via Wnt signaling. This screen aimed to identify novel factors that mediate Wnt signaling to posteriorize the neural plate, therefore I prioritized the candidate genes that are likely to function in patterning by first assaying expression of transcription factors, RNA binding proteins, and various signaling proteins. These genes need to be expressed at the right developmental timepoint and relevant tissue to play a role in Wnt mediated neural patterning. Accordingly, I assayed expression of selected candidates via in situ hybridization in Xenopus tropicalis embryos during gastrulation and neurulation to identify those that are expressed in the posterior neurectoderm. In addition to the known neural Wnt targets identified in the screen, several of the selected transcription factors had detectable expression in neurula stages. The transcription factors ahctf1 (Figure 4.6, A-C), foxi4.2 (Figure 4.6, D-F), and churchill (Figure 4.6 G-I) were barely expressed over background staining but show some weak staining in the posterior. zmiz1 is first detected in the posterior neural plate at the onset of neurulation and becomes restricted to two posterior stripes in the spinal cord anlage by stage 15 (Figure 4.6, J-L). In mice, ahctf1, the gene responsible for the Elys mutation, and zmiz1 are expressed in the developing spinal cord (Beliakoff et al., 2008; Kimura et al., 2002). However their roles in A-P patterning have not been established. max (Figure 4.6 M-O), originally reported to be ubiquitously expressed (Newman and Krieg, 1999), and znf384 are expressed in the presumptive neural plate at gastrulation and show enrichment in the posterior neural region at neurulae stages (Figure 4.6 P-R). The Spalt-like genes sall1 (Figure 4.6 S-U) and sall4 (Figure 4.6 V-X) are expressed at gastrulation in the presumptive neural plate and show robust expression in the posterior neural plate in later stages. The role of the sall genes in human diseases (Kohlhase et al., 2002; Kohlhase et al., 1998), stem cell maintenance (Zhang et al., 2006), and limb development (Neff et al., 2005) have been studied but a neural patterning role has not been described. Unexpectedly, sox11 was identified in the screen and is expressed in the neural plate but is absent from the most posterior region. This could be the result of secondary repression wherein the function of a different factor that normally represses sox11 expression in this region is relieved when the tissue was treated with CHX (Figure 4.6 Y-Aa). 49

60 The splicing factor fus was shown to be crucial for gastrulation via regulation of multiple developmental pathways (Dichmann and Harland, 2012). It is possible that RNA-binding proteins may play a previously unappreciated role in A-P patterning. Both heterologous nuclear ribonuclear proteins hnrnpk (Figure 4.7 A-C) and hnrnp H3 (Figure 4.7 D-F) are expressed in the posterior neural plate. The genes encoding splicing factors sf3b4 (Figure 4.7 G-I) and sap130 (Figure 4.7 J-L) show specific expression dorsally around the blastopore at stage 12 which becomes more enriched at the midneurula stage. These factors are known to interact with each other (Kotake et al., 2007; Menon et al., 2008) and sf3b4 has been implicated in bone development (Watanabe et al., 2007), a process that is also regulated by Wnt signaling (reviewed in (Williams and Insogna, 2009). Two additional genes encoding splicing factors sfrs7 and sfrs6 were also identified in this screen (Figure 4.7 M-R). sfrs7 is specifically expressed in posterior regions and sfrs6 also shows a posterior neural expression domain at stage 12. However the staining is faint and does not show specific expression at stage 15. A role for RNA-binding proteins and splicing factors in mediating neural A-P patterning has not been described or implicated. The neural expression of identified RNA binding factors is strong evidence that these proteins are important in neural patterning. It will be interesting to determine if the splicing factors identified in this screen comprise a core of Wnt-regulated splicing proteins that function in cell fate specification. In addition to transcription and RNA binding factors, the planar cell polarity gene prickle1 (Wallingford et al., 2002) is expressed in the midline and enriched in the posterior (Figure 4.8 A- C). Prickle homologs have been shown to mediate movements during gastrulation, but morpholino studies in Xenopus and zebrafish have not revealed a role in neural A-P patterning (Takeuchi et al., 2003). It remains to be tested if there is functional redundancy between the prickle genes in neural patterning or if there simply is not a role for Prickle in this process. Finally, the adhesion molecule Lmo7 (Ooshio et al., 2004)shows punctate expression in the neural plate, although no function in neural development for this gene has been established previously (Figure 4.8 D-F). Many of the candidates identified in the screen are enriched in the posterior neural region demonstrating the effectiveness of this method to identify novel targets of Wnt signaling. However, not all of the candidates identified in the screen were expressed in the neural plate or were detectable via in situ hybridization. One explanation could be that the expression of these genes is below the threshold of detection by this method. This does not explain however, why some genes found to be highly expressed in the RNA-seq data were not detected via in situ hybridization. These could be false positives that result from the nature of the analysis. The software packages used to align and quantitate reads also find splice junctions which could lead to spurious results when reads are aligned to an index of spliced cdnas as was done for this screen. An alternative explanation is that the activation of Wnt signaling in the animal caps is not physiological and results in activation of targets that are not normally expressed during early or mid-neurula stages. However, this is likely to be a minority of the data set as 17 of 22 previously unknown posterior neural genes showed specific posterior expression. Another 50

61 possibility is that some of the identified genes are subject to repression via secondary targets. Therefore, these genes would not be detected in wild-type embryos but would be highly expressed in conditions where Wnt was activated in the presence of CHX. While I prioritized transcription factors and RNA-binding proteins, there are other important classes of genes identified in the screen that could play crucial roles in meditating A-P neural patterning. For example, molecules involved in signal transduction and chromatin remodeling likely have significant effects in Wnt-signaling and downstream gene expression. Further analysis of candidates identified in this screen will provide a more complete understanding of Wnt-mediated neural patterning. 51

62 Figure 4.1: Model of Wnt-induced patterning of the neural anterior-posterior axis. Schematic of a proposed Wnt gradient and anterior-posterior patterning of the neural plate. 52

63 Forebrain Midbrain Hindbrain Spinal Cord 1. BMP antagonists induce neural tissue from ectodermal precursors. 2. Neural tissue is then posteriorized by a gradient of FGF, RA, and Wnt. Wnt Source: paraxial mesoderm Head Mesoderm Source of BMP and Wnt antagonists Notochord Source of BMP antagonists 53

64 Figure 4.2: Temporal expression of anterior posterior neural markers in Xenopus tropicalis. In situ hybridization for otx2 (A-F), engrailed2 (en2) (G-L), krox20 (M-L), and hoxb9 (S-X). Dorsal views with anterior towards the top. 54

65 NF Stage A B C D E F G H I J K L M N O P Q R S T U V W X hoxb9 krox20 en2 otx2 55

66 Figure 4.3: TVGR activates canonical Wnt signaling. (A) Quantification of secondary axis induction by ventral-vegetal injection of TVGR at the 4-cell stage. (B) Uninjected control tadpole. (C) TVGR injected tadpole. (D) TVGR injected tadpole treated with 0.4% EtOH. (E) TVGR injected tadpole treated with 10 µm DEX, red staining shows β-galactosidase (tracer). 56

67 A B TVGR/Dex Induction of Wnt Signaling 100% U.C. 80% C 60% 40% Complete 2º Axis Partial 2º Axis Normal D 20% E 0% U.C. 0.2% EtOH DEX 0.4% EtOH DEX 57

68 Figure 4.4: TVGR efficiently posteriorizes neuralized ectodermal explants. (A) Strategy to use TVGR to induce Wnt signaling in animal caps. (B) RT-PCR on 5 embryos or 25 animal caps treated with the indicated reagents. -RT: reaction done in the absence of Reverse Transcriptase, epi. ker: epidermal keratin (epidermis), mus. act.: muscle actin (mesoderm). When indicated the following doses were used: noggin 10 pg, TVGR 4 pg, Dexamethasone: 10 µm (C) Animal caps treated with the indicated reagents. 58

69 B noggin otx2 en2 krox20 hoxb9 epi. ker. mus. act. eef1a1 -RT A Inject noggin+ TVGR Excise animal caps 10 um DEX 2-cell stage 9 isolate RNA at stage 15 equivalent C UC noggin noggin/tvgr+etoh noggin/tvgr+dex 59

70 Figure 4.5: Cycloheximide treatment prior to TVGR induction enriches for direct Wnt targets in neuralized ectodermal explants. (A) Strategy to use animal caps to screen for direct transcriptional targets of Wnt signaling in neural tissue. (B) Semi-quantitative qpcr on either 5 whole embryos or 25 animal caps treated with the indicated reagents and used for RNA-seq library synthesis. meis3 and hoxb9 serve as controls for direct and indirect targets of Wnt signaling, respectively. When indicated, the following doses were used: noggin 10 pg, TVGR 4 pg, Dexamethasone 10 µm. 60

71 A Inject noggin+ TVGR Excise animal caps 5 um CHX 1.5hrs 10 um DEX 2-cell stage 9 isolate RNA at stage 15 equivalent B Whole Embryo Animal Caps noggin TVGR EtOH Dex - -RT hoxb9 hoxb9 meis3 eef1a1 5 µm Cycloheximide 61

72 Figure 4.6: Expression patterns of transcription factors identified in the screen for direct Wnt targets. In situ hybridization of selected genes in Xenopus tropicalis. Dorsal-vegetal views of stage 10.5 embryos, dorsal views of stage 12 and 15 with anterior toward the top. NF stage: Nieuwkoop Faber stage (Nieuwkoop, 1967). 62

73 NF stage sall1 max churchill ahctf A B C D E F G H I J K L M N O P Q R S T U V W X foxi4.2 zmiz1 sox11 znf384 sall4 Y X Aa 63

74 Figure 4.7: Expression patterns of RNA-binding factors identified in the screen for direct Wnt targets. In situ hybridization selected genes in Xenopus tropicalis. Dorsal-vegetal views of stage 10.5 embryos, dorsal views of stage 12 and 15 with anterior toward the top. NF stage: Nieuwkoop Faber stage (Nieuwkoop, 1967). 64

75 NF stage sfrs7 sfrs6 sf3b4 hnrnpk A B C D E F hnrnp H3 G H I J K L sap130 M N O P Q R 65

76 Figure 4.8: Expression patterns of identified Wnt targets prickle1 and lmo7. In situ hybridization in Xenopus tropicalis for indicated genes. Dorsal-vegetal views of stage 10.5 embryos, dorsal views of stage 12 and 15 with anterior toward the top. NF stage: Nieuwkoop Faber stage (Nieuwkoop, 1967). 66

77 NF stage prickle A B C D E F lmo7 67

78 CHAPTER 5 Spalt-like 4 mediates Wnt-induced neural patterning via repression of pouv/ Oct4 family members. Introduction Pieter NieuwkoopOthers, 1952a). This hypothesis states that neural tissue is induced as an anterior state by the organizer, and then posteriorized by additional signals from the mesoderm to create manipulation that blocks Bmp signaling in ectoderm results in cells adopting anterior neural fates β factors as mediators of anterior-posterior (A-P) neural patterning, the mechanism by which gene expression and discrete neural fates remains poorly understood. gbx2 is a meis3, cdx1 and cdx4 Xenopus, simultaneous Drosophila spalt result in 68

79 sall1 and sall4 cause the autosomal sall4 is the best studied of the spalt homologs because of its role in maintaining and inducing pluripotency. sall4 sall4 null embryos fail to pou5f1 cdx2 sall4 oct3/4, sox2, and klf4; sall4 along with oct3/4, sox2 and klf4 sall2, sall3, and sall4 early Xenopus sall2 of-function alleles of sall1, sall2 and sall4 result in mouse embryos with neural tube closure sall genes in neural differentiation and/or role for the sall genes in caudalization has not been elucidated. Here, I show that sall4 sall4 represses the stem cell factor pouv/oct4 in order to release cells from an undifferentiated state. Results In the previous chapter, I described a screen to identify direct targets of Wnt in neural tissue. That screen identified sall1 and sall4 as candidate Wnt targets that are involved in neural posteriorization. Since the sall genes were identified in a high-throughput screen, it was important to confirm that they were not false positives. To that end, I confirmed the results of my screen via the animal cap assay described in Figure 4.5 and quantitated transcript abundance with qpcr. Incubation with CHX prior to activation of TVGR in neuralized caps resulted in increased cdx2 (Figure 5.6A), sall1 (Figure 5.4A), sall4, and meis3, but not hoxb9, expression (Figure 5.1A). The activation of the sall genes by canonical Wnt activation in the absence of protein translation strongly suggests that they are direct targets of Wnt signaling in neural tissue. As a complementary approach to using CHX-treated animal caps, I used chromatin immunoprecipitation (ChIP) followed by qpcr to examine whether β-catenin is bound to the sall4 genomic locus. β-catenin forms an activation complex with TCF/LEF in response to Wnt signaling to activate target genes (Behrens et al., 1996). Due to a lack of effective commercial β- catenin antibodies, I overexpressed a C-terminal FLAG-tagged version of X. laevis β-catenin. 69

80 Expression of the FLAG epitope was confirmed by immunoblotting (Figure 5.2A). The activity, dosage and specificity of FLAG-tagged β-catenin were tested by rescuing the β-catenin knockdown phenotype. β-catenin morphants lack dorsal structures, resulting in ventralized belly pieces (Heasman et al., 2000). Injection of β-catenin morpholino resulted in over 60% of embryos lacking dorsal tissue. Co-injection of FLAG-tagged β-catenin RNA restored dorsal structures, demonstrating both the specificity and activity of this construct (Figure 5.2B). Finally, I assayed dorsalization resulting from overexpression of FLAG-tagged β-catenin without morpholino injection. Consistent with Yost and colleagues (Yost et al., 1996), injection of 500 pg of RNA encoding tagged β-catenin did not significantly alter dorsal structures as measured by the dorsoanterior index (Kao and Elinson, 1985) (Figure 5.1B). I assume that the injected RNA results in β-catenin that enters the normal pool of unstable protein, and whose amount is subject to regulation by the normal signaling pathway (Yost et al., 1996). The sall4 locus in Xenopus laevis contains four exons and three introns (Figure 5.1C). Scanning for consensus binding sequences within 3 kb of flanking sequence of the sall4 locus revealed six putative TCF/LEF binding sites (Elkouby et al., 2010; McKendry et al., 1997) within the first intron. Three of these sites are tightly clustered within a 150 bp span at positions +2347, +2387, and (relative to the predicted transcription start site) and are conserved in Xenopus tropicalis. The conservation and clustering of these sites suggested that this region could be regulated by β-catenin. Using FLAG antibodies for ChIP, this region was found to be significantly enriched compared to a negative control (xmlc2) region (Figure 5.1D). A 2.7 Kb region upstream of meis3 was used as a positive control for β-catenin binding (Elkouby et al., 2010). Anti-FLAG immunoprecipitations in uninjected control embryos resulted in negligible enrichment of any loci assayed. Taken together, activation via TVGR in the presence of CHX along with β-catenin binding to TCF/LEF sites in the first intron provide strong evidence that sall4 is a direct transcriptional target of Wnt signaling in the neurectoderm. During gastrulation, sall4 is initially expressed in a broad domain throughout the marginal zone and the animal pole (Figure 5.3A). At stage 10, sall4 expression is restricted to the sensorial neurectodermal cells (cells that will give rise to the central nervous system) in animal dorsal regions (Fig 3E). At the onset of neurulation, sall4 is strongly expressed throughout the sensorial neurectodermal layer of cells in the neural plate (Fig 3B,F,G). Neural expression of sall4 in stage 15 (mid-neurula) embryos is in the hindbrain and spinal cord anlage (Figure 5.3C,H,I). Later stage neurulae (stage 18) strongly express sall4 in the posterior neural tube, hindbrain, developing placodes, and epidermis (Figure 5.3D,J,K). The posterior neural expression of sall4 and regulation by β-catenin and TCF/LEF make it a candidate for mediating Wnt-induced posteriorization of the neural plate. sall1 is expressed in the dorsal ectoderm and involuting mesoderm during gastrulation (Figure 5.4 B,B ). Expression becomes restricted to the notochord and circumblastoporal collar at the early neurula stage (Figure 5.4C-C ). Similar to sall4, sall1 is expressed in the spinal cord anlage at mid and late neurula stages (Figure 5.4D-D,E-E ). 70

81 Given the neural expression of sall4 and its identification as a Wnt target in neuralized animal caps, I hypothesized that loss of Sall4 would affect neural patterning. To test this, I used morpholinos targeting sall4 and assayed for neural gene expression. Morphant embryos had neural tube closure defects and began to disintegrate and die at mid-tailbud stages. The neural tube closure defect is consistent with defects in neural patterning, so I assayed several markers of neural differentiation. The pan-neural marker sox2 was expressed in the neural plate in uninjected and Sall4 morphants, demonstrating that the dorsal ectoderm of morphants still retained a neural identity (Figure 5.5A,B). Conversely, the expression of n-tub, a marker for differentiating neurons, was markedly reduced but still present in morphants, suggesting that Sall4 is required for the second wave of neurogenesis in the tailbud tadpole (Figure 5.5C,D). Another marker for early motor neuron differentiation, nkx6.1, was expressed in the central nervous system of morphants. Neural crest cells were present as determined by the expression of snai2 (Figure 5.5E,F). Though present, these markers were expressed in a pattern more similar to early neurulae, suggesting either a delay or failure of terminal differentiation. Sall4 morphants expressed the dorsal mesoderm marker myod in a similar pattern to uninjected control embryos, therefore, neural defects were not due to a major loss of paraxial mesoderm (Fig. G,H). It is well understood that Wnt is a caudalizing factor in neural patterning, therefore if Sall4 mediates this process I predicted that Sall4 morphants would lose posterior neural identity. To test for this, Sall4 morpholinos were injected into one animal dorsal cell of 4-cell stage embryos to allow for comparison between injected and uninjected sides. Strikingly, the injected side of embryos showed a significant posterior shift in the expression of otx2 and krox20 relative to the control side (Figure 5.5I,J,Q). This shift suggested that knockdown of Sall4 results in an expansion of anterior neural identity at the expense of posterior neural differentiation, consistent with a Wnt loss-of-function phenotype. Accordingly, the injected side had a significant reduction in the expression domain of the spinal cord marker hoxb9 (Figure 5.5K,L). Furthermore, the expression of two other markers for spinal cord differentiation, hoxc10 (Figure 5.5M,N) and hoxd10 (Figure 5.5O,P) was significantly reduced following Sall4 knockdown (Figure 5.5R,S,T). However, Sall4 knockdown does not equally reduce cdx2 expression, another identified direct Wnt target (Figure 5.6B,C). These results demonstrate that Sall4 is required for specification of posterior neural tissue. The failure of Sall4 morphants to induce posterior fates in neural tissue suggested that the caudal tissue remained in an undifferentiated stem-like state. Sall4 positively regulates the stem cell factor Oct4/pou5f1 locus in the inner cell mass of mouse embryos to maintain pluripotency (Zhang et al., 2006). Since the posterior neurectoderm failed to differentiate in Sall4 morphants, I hypothesized that Oct4 prevents neural differentiation in the morphants because Sall4 was negatively regulating pou5f1 in the neural tissue of Xenopus. There are three class V Poudomain proteins in Xenopus with similar sequence and syntenic organization to mammalian pou5f1/oct4: pou25, pou60, and pou91, henceforth referred to as pouv/oct4 (Morrison and Brickman, 2006). 71

82 If Sall4 negatively regulates pouv/oct4, Sall4 morphants should have increased expression of these factors. Indeed, knockdown of Sall4 in unilateral and bilateral injections resulted in ectopic expression of pou25 (Figure 5.7A,B,C), pou60 (Figure 5.7D,E,F), and pou91 (Figure 5.7G,H,I). Accordingly, the increase in expression of pou25 and pou91 was the greatest in the neural tube where sall4 is normally expressed. pouv/oct4 expression in Sall4 morphants relative to control embryos was quantified via qpcr and displayed a significant increase in all three of the pouv/oct4 genes (Figure 5.7O,P,Q). Co-injection of Xenopus tropicalis sall4 RNA that is not targeted by the Sall4 morpholino resulted in a partial rescue of the pou25 expression level and restored pou60 and pou91 expression to wild-type levels (Figure 5.7O,P,Q). These experiments demonstrate that loss of Sall4 relieves repression of the pouv/oct4 homologs, resulting in an overexpression of pouv/oct4 stem cell factors in the neural plate. A second, non-overlapping Sall4 morpholino (MO2) was used as an additional control for specificity. If the above results are indeed due to Sall4 knockdown, then this second morpholino should result in the same phenotype. As seen with the first Sall4 morpholino, I found that injection of the Sall4 MO2 resulted in a similar loss of hoxb9 (Figure 5.8A,B) and hoxd10 (Figure5.8C,D) expression as well as a posterior shift in oxt2 and krox20 expression (Figure 5.8A,B). The Sall4 MO2 also resulting in ectopic expression of pou91 (Figure 5.8E,F). These results demonstrate that the morpholinos targeting Sall4 are specific and effective. Next, I asked whether ectopic pouv/oct4 expression is sufficient to block posterior neural differentiation by injecting RNA for the three pouv/oct4 genes unilaterally into embryos and assaying A-P neural gene expression. Neural plate cells expressing ectopic pouv/oct4 as traced by the co-injected marker β-galactosidase did not express hoxb9, whereas cells on the uninjected side expressed this spinal cord marker (Figure 5.7J,K,L,M,N). The increase in pouv/oct4 expression in Sall4 morphants, together with the result that ectopic expression of these genes inhibited posterior neural differentiation, provides evidence that the loss of caudal identity in morphants is due to pouv/oct4 overexpression. The observed pouv/oct4 increase following knockdown of Sall4 suggested a mechanism for the loss of posterior neural identity whereby the ectopic pouv/oct4 expression prevents differentiation of neural tissue into spinal cord. I therefore reasoned that depleting pouv/oct4 in Sall4 morphants would restore posterior neural identity. To that end, morpholinos targeting the three pouv/oct4 homologs(morrison and Brickman, 2006) were co-injected with Sall4 morpholinos. Consistent with results described above, knockdown of Sall4 resulted in a loss of expression of the spinal cord markers hoxb9 (Figure 5.9A,E), hoxc10 (Figure 5.9B,F), and hoxd10 (Figure 5.9C,G) but not in a loss of the pan-neural marker sox2 expression (Figure 5.9D,H). Co-injection of the pouv/oct4 morpholinos with Sall4 morpholinos restored posterior neural gene expression lost by Sall4 knockdown alone (Figure 5.9I,J,K). Though reduced, sox2 was expressed in the neural plate of Sall4-pouV/Oct4 morphants (Figure 5.9L). Finally, comparing the length of the hox gene expression domains in Sall4 and Sall4-pouV/Oct4 morphants revealed a significant rescue of all three spinal cord markers in the quadruple morphants (Figure 5.9N). pouv/oct4 knockdown resulted in anterior defects, evidenced by 72

83 mispatterning of the otx2 and krox20 expression domains and a lateral expansion of hoxb9, consistent with a role for the Oct4 homologs in repressing posterior neural identity (Figure 5.9M). My results demonstrated that Wnt-signaling induces Sall4 expression, which represses pouv/ Oct4 allowing posterior neural differentiation. Yet posterior patterning of the neural plate also occurs in response to Fgf and RA signaling. If Sall4 were required to repress pouv/oct4 in order to allow for caudal differentiation, then depletion of Sall4 should block posteriorization by Fgf and RA. Therefore, I tested whether repression of pouv/oct4 via Sall4 is required for both Fgfand RA-induced caudalization. To address the question of whether Fgf-induced caudalization requires Sall4, uninjected and Sall4 morphant embryos were injected with fgf8a RNA. Again, Sall4 knockdown resulted in a loss of hoxb9 expression (Figure 5.10A,B) without major alterations to sox2 expression (Figure 5.10E,F). Overexpression of fgf8a in the dorsal ectoderm resulted in an expansion of hoxb9 (spinal cord) expression, lateral expansion of krox20 expression, and repression of otx2 (brain) (Figure 5.10C) (Fletcher et al., 2006). Overexpressing fgf8a in Sall4 morphants resulted in otx2 repression, but also resulted in a loss of hoxb9 expression (Figure 5.10D). krox20 expression in rhombomere 5 was severely reduced in the Sall4 morphants despite fgf8a overexpression while rhombomere 3 expression remained expanded. These results support the conclusion that Sall4 is required for Fgf-induced posterior neural differentiation. To address the question of whether RA-induced caudalization requires Sall4, uninjected and Sall4 morphant embryos were incubated in RA. Increasing RA signaling results in severe loss of anterior neural tissue and expansion of posterior identities (Blumberg et al., 1997; Durston et al., 1989; Ruiz i Altaba and Jessell, 1991; Shiotsugu et al., 2004; Sive et al., 1990). Uninjected control embryos treated with 1 µm all-trans retinoic acid (ATRA) lacked otx2 and krox20 but had hoxb9 expression (Figure 5.10I). Sall4 morphant embryos treated with 1 µm ATRA failed to express otx2 and krox20, however, they also failed to express hoxb9 (Figure 5.10J). The reduction of these markers was not due to a loss of neural tissue as sox2 expression was similar among control embryos, embryos treated with ATRA, and Sall4 morphant embryos treated with ATRA (Figure 5.10E,K,L). Thus, Sall4 is required for caudalization of the neural plate via both Fgf and RA signaling. Discussion Wnt, Fgf, and RA signaling are caudalizing factors required for posteriorization of the neural plate. However, the transcription factors identified to mediate patterning signals from these pathways have largely been restricted to those specifying midbrain and hindbrain fates. In this study, I show that canonical Wnt signaling directly activates sall4, which is required for spinal cord differentiation. The primary role of Sall4 in Wnt-induced posterior patterning is the repression of the pouv/oct4 homologs. This repression is necessary for spinal cord differentiation; Sall4 knockdown as well as pouv/oct4 overexpression results in a loss of spinal 73

84 cord fate. Furthermore, the posterior defects in Sall4 morphants can be rescued via pouv/oct4 knockdown. The data presented here suggest a model whereby the repression of pouv/oct4 via Sall4 provides a permissive environment, allowing cells in the neural plate to respond to posteriorizing signals from Fgf, RA, and Wnt (Figure 5.10M). Other studies have shown interactions between Wnt signaling and Sall factors. Sall2 (published as XsalF) functions as a Wnt antagonist in anterior neural tissue (Onai et al., 2004). Our work found sall4 to be directly activated by Wnt signaling in the neurectoderm. TCF/LEF binds to and activates the sall4 gene in human cell culture lines at a conserved binding site within the promoter (Böhm et al., 2006). This conservation is restricted to mammals: the chick Gallus gallus and Xenopus lack TCF/LEF sites in the promoter regions of sall4. However, the first intron of Xenopus laevis sall4 contains six putative consensus TCF/LEF binding sites, three of which are found within a 150 bp region and conserved in Xenopus tropicalis, though not in chickens. Our finding that this region is enriched upon β-catenin ChIP demonstrates that β- catenin is bound to it during late gastrulation at the appropriate time to transduce the Wnt signal and activate sall4, resulting in the down-regulation of pouv/oct4. In the future, mutagenesis of the consensus binding sites will test whether they are indeed required for Wnt-induced sall4 expression. Our results provide a novel mechanism of neural posteriorization wherein Wnt signaling activates sall4 in order to repress inhibitors of neural differentiation. A key prediction of this mechanism is that a requirement for Sall4 in adoption of posterior fates would not be restricted to Wnt-induced posteriorization. Therefore, one would expect that the increase in pouv/oct4 expression resulting from Sall4 knockdown would inhibit differentiation that was induced by other caudalizing factors. Indeed, I found Sall4 knockdown prevented induction of hoxb9 by Fgf or RA. Future experiments will test whether overexpression of pouv/oct4 is sufficient to block posteriorization by Fgf and RA. While I show that sall4 is a target of Wnt signaling, it remains possible that sall4 is also regulated by other signaling pathways. In flies, spalt is regulated by Dpp (BMP), Hedgehog and EGF signaling, depending on tissue type (de Celis et al., 1996; Elstob et al., 2001; Sturtevant et al., 1997). sall1 is regulated by both Fgf and Wnt signaling in the chick limb (Farrell and Münsterberg, 2000) and sall4 is expressed in the Xenopus limbs during development and regeneration (Neff et al., 2011). The expression pattern of sall4 at early neurula stages is broader than what would be expected for a gene solely regulated by Wnt. This suggests possible activation via multiple pathways. One possibility could be that sall4 expression is regulated through different enhancers, each responsible for different expression domains. In the chick, sox2 is expressed throughout the neural plate but is regulated by five different enhancers, each responsible for a portion of the full expression domain (Uchikawa et al., 2003). Furthermore, Fgf and Wnt signaling converge on one such enhancer, N-1, to mediate the most posterior expression of sox2 in the neural plate (Takemoto et al., 2006). 74

85 Fgf signaling is sufficient to posteriorize neurectoderm (Christen and Slack, 1997; Fletcher et al., 2006; Kengaku and Okamoto, 1995; Lamb and Harland, 1995), and I found this activity requires Sall4. Therefore, it is possible that FGF and canonical Wnt signaling converge on an as-yet unidentified enhancer to regulate sall4 expression. As in the sox2 N-1 enhancer, Wnt- and Fgfresponsive elements in the enhancers of pax3 and zic genes cooperatively regulate their expression (Garnett et al., 2012), and both pathways are required for expression of these genes at the neural plate border (Monsoro-Burq et al., 2005). Multiple studies have identified Wnt signaling as a key factor in stem cell maintenance and differentiation (Wang and Wynshaw-Boris, 2004). In amphibians, caudalization of the neural plate via canonical Wnt signaling induces undifferentiated neural precursors to commit to posterior fates. This induction requires repression of stem cell factors and the activation of differentiating factors. In Xenopus, pouv/oct4 play a conserved role in maintaining pluripotency of uncommitted cells (Morrison and Brickman, 2006). pouv/oct4 are first expressed animally in cleavage stages and throughout the mesoderm and ectoderm of amphibian gastrulae (Frank and Harland, 1992; Morrison and Brickman, 2006). Knockdown of pouv/oct4 results in precocious cell fate commitment in the three germ layers (Morrison and Brickman, 2006; Snir et al., 2006). The spatiotemporal expression of pouv/oct4 serves to prevent these cells from adopting inappropriate fates due to premature response to differentiating signals. Accordingly, pouv/oct4 overexpression prolongs the undifferentiated state (Archer et al., 2011; Morrison and Brickman, 2006). Our results suggest that pouv/oct4 expression must be down-regulated in the neurectoderm to allow for cells to respond to instructive Wnt/Fgf/RA signals and commit to posterior fates. Wnt signaling activates cdx1 (Pilon et al., 2007; Prinos et al., 2001) and in frogs, Cdx1 represses pouv/oct4 gene expression at the onset of gastrulation (Rousso et al., 2011). However, knockdown of Cdx1 does not result in a loss of spinal cord differentiation, and combinatorial knockdown of Cdx1/2/4 is required before hoxb9 and hoxc10 expression is reduced (Faas and Isaacs, 2009). There is, however, a dramatic loss of hoxb9, hoxc10, and hoxd10 expression in Sall4 morphants. In the absence of Sall4, pouv/oct4 expression remains high, resulting in neural cells being unable to commit to a posterior neural fate and differentiate into spinal cord. Our results suggest that Sall4, and not Cdx1, acts as the main negative regulator of pouv/oct4 in the neurectoderm. My findings are consistent with described roles for the cdx genes in Wnt-mediated neural patterning. Several studies have shown the Cdx factors function to regulate posterior hox gene expression in vertebrates (Gaunt et al., 2004; Gaunt et al., 2008; Isaacs et al., 1998; van den Akker et al., 2002). Therefore, Wnt acts as an instructive signal through activation of cdx genes to induce posterior hox genes, thereby transforming the neural precursors into a posterior fate. Here, I find that canonical Wnt also induces the expression of sall4 to repress pouv/oct4, providing a parallel, permissive signal for the induction of posterior hox gene expression via Cdx factors. Wnt still signals in posterior neural regions of Sall4 morphants - presumably activating cdx genes - but the prolonged expression of pouv/oct4 prevents hox gene expression and 75

86 adoption of spinal cord fates. Conversely, it s likely that Wnt induces sall4 expression in Cdx morphants, priming the neural plate to respond to other instructive signals that induce posteriorization. This mechanism could explain the finding that knockdown of individual Cdx homologs results in minor phenotypes. The interaction between Sall4 and Oct4 has been well studied in cell culture and mouse embryos. In the ICM of mouse embryos and in ips cells, Sall4 positively regulates Oct4/pou5f1 expression allowing for self-renewal and suppression of differentiation (Tsubooka et al., 2009; Zhang et al., 2006). However, I found Sall4 knockdown resulted in increased pouv/oct4 gene expression in neural tissue, which suggests that Sall4 negatively regulates the Oct4 homologs in Xenopus neurectoderm. One explanation for this apparent discrepancy is that cellular context affects whether Sall4 will act as a transcriptional activator or repressor. The N-terminus of Sall4 recruits DNA methyltranferases and the NuRD complex resulting in transcriptional inhibition of target genes in HEK293 cells(yang et al., 2012). Conversely, in embryonic stem (ES) cells, Sall4 was found to directly bind to the promoter of pou5f1/oct4 and activate transcription (Zhang et al., 2006). Additionally, Sall4 regulates distinct circuitries between ES and extra-embryonic endoderm cells, presumably due to the presence of different co-factors (Lim et al., 2008). Therefore, it is likely that the regulation of pouv/oct4 via Sall4 is conserved but is repressive in the context of Xenopus neurectoderm. It remains unclear, however, if this regulation is direct or indirect. While the mechanistic role of Sall4 in the neural plate is likely to be complex, our experiments demonstrate Sall4 to be a pivotal factor in neural patterning. 76

87 Figure 5.1: sall4 is a direct transcriptional target of canonical Wnt-signaling. (A) qpcr on 5 whole embryos or animal caps treated according to the conditions indicated on the X-axis. The Y-axis shows relative expression to odc. meis3 and hoxb9 serve as controls for known direct and indirect targets of Wnt-signaling, respectively. (B) Quantification of dorsalization in uninjected embryos (open bars) and embryos injected animally with 500 pg FLAG-tagged β- catenin RNA (250 pg/blastomere) at the 2-cell stage (filled bars) as scored by the dorsoanterior index (DAI). Error bars show 1 standard deviation from the mean. Images show a representative uninjected (UC) embryo with an DAI of 7 (normal) and a representative embryo with a DAI of 6 (kinked axis). (C) Schematic of the sall4 genomic locus in Xenopus laevis. Blue boxes indicate exons and yellow circles indicate the location of putative TCF/LEF binding sites. Black ovals show the locations of the zinc-finger domains. Numbers indicate the position of putative binding sites relative to the transcription start site (TSS). Red octagon shows the stop codon. (D) Chromatin immunoprecipitation of FLAG-tagged β-catenin in late gastrulae/early neurulae. Open bars represent uninjected embryos and closed bars represent embryos injected with 500 pg FLAG-tagged β-catenin (250 pg/blastomere at the 2-cell stage). Error bars represent one standard deviation from the mean per cent input for each measurement. meis3 and xmlc2 serve as positive and negative controls for β-catenin binding, respectively. Means were compared to the negative control (xmlc2) by one-way ANOVA followed by Tukey post-hoc analyses (*: p<0.05). 77

88 A sall4 B DAI UC DAI:7 DAI: pg FLAG -catenin hoxb9 C sall4 +0 TSS E1 E2 E3 E meis3 D % Input UC FLAG * noggin TVGR EtOH DEX CHX WE Animal Caps 0 xmlc2 meis3 sall4 intron 1 78

89 Figure 5.2: Injected embryos express functional FLAG-tagged β-catenin. (A) Western blot for the FLAG epitope in injected embryos. Actin serves as the loading control. (B) Ventralization of embryos injected with β-catenin MO or co-injection of β-catenin MO with different doses of FLAG-tagged β-catenin RNA. F-βcat: FLAG-tagged β-catenin. 79

90 A UC 500 pg FLAG- -catenin 1000 pg FLAG- -catenin ANTI-FLAG ANTI-Actin B 100% 90% 80% 70% 60% Normal Partially ventralized Ventralized 50% 40% 30% 20% 10% 0% UC 250 pg F- cat 350 pg F- cat 500 pg F- cat + 10 ng -catenin MO 750 pg F- cat 80

91 Figure 5.3: sall4 is expressed in the neurectoderm. (A-D) Whole-mount in situ hybridizations of sall4 in Xenopus laevis embryos. (A) Whole mount stage 10 embryo stained for sall4, dorsovegetal view with the dorsal lip of the blastopore oriented towards the top. (B-D) Neurula stage embryos are shown in dorsal view, with anterior toward the top. (E) Sagittal section of stage 10 embryo stained for sall4 expression, animal pole is to the top and dorsal is to the right. (F-G) Transverse sections of stage 12 embryos stained for sall4, (F) anterior and (G) posterior. (H-I) Transverse sections of stage 15 embryos stained for sall4, (H) anterior and (I) posterior. (J-K) Transverse sections of stage 18 embryos stained for sall4, (J) anterior and (K) posterior. (E-K) 50µM sections, (F-K) all embryos are oriented with dorsal towards the top. SNE: sensorial neurectoderm, No: notochord, S: somite, PM: paraxial mesoderm, PSM: presomitic mesoderm. 81

92 A B C D F H J G I K E E F SNE H J S S No S S No G PM No PM I PSM PSM No K PSM PSM No 82

93 Figure 5.4: sall1 is directly activated by canonical Wnt signaling and expressed during early embryogenesis. (A) qpcr on 5 whole embryos or 15 to 25 animal caps treated according to the conditions indicated on the X-axis. The Y-axis shows expression relative to odc. (B-E) Whole-mount in situ hybridizations of sall1 in Xenopus laevis embryos. (B) Whole mount stage 10 embryo stained for sall1, dorso-vegetal view with the dorsal lip of the blastopore oriented towards the top. (B ) Sagittal section of stage 10.5 embryo stained for sall1 expression, animal pole is to the top and dorsal is to the right. (C-D) Dorsal views of indicated neurula stage embryos, anterior is oriented towards the top. (C -C ) Transverse sections of stage 12 embryos stained for sall1, (C ) anterior and (C ) posterior. (D -D ) Transverse sections of stage 15 embryos stained for sall1, (D ) anterior and (D ) posterior. (E -E ) Transverse sections of stage 18 embryos stained for sall1, (E ) anterior and (E ) posterior. (B, C -E ) 50 µm sections, (C - E ) dorsal oriented towards the top. No: notochord, S: somite, PSM: presomitic mesoderm. 83

94 A sall1 B B noggin TVGR EtOH DEX CHX C WE Animal Caps D E C D E C D E C No D S No S S S No E C D E PSM No PSM PSM No PSM 84

95 Figure 5.5: Loss of Sall4 results in a loss of posterior neural differentiation. (A-P) Wholemount in situ hybridization late neurula stage embryos shown in dorsal view with the anterior toward the top. (A, C, E, G) uninjected control embryos and (B, D, F, H) embryos injected bilaterally with 40 ng Sall4 MO (20 ng/blastomere at the 2-cell stage). (A-B) Expression of sox2 (C-D) Expression of n-tub. (E-F) Expression of snai2 and nkx6.1 (G-H) expression of myod. (I, K, M, O) Uninjected control embryos and (J, L, N, P) embryos injected with 20 ηg Sall4 MO into the right animal-dorsal (A/D) blastomere. (I-J) Expression of otx2/krox20/hoxb9, arrows show the relative anterior-posterior position of krox20 and the anterior limit of hoxb9. (K- L) Posterior view of hoxb9 expression. (M-N) Posterior view of hoxc10 expression. (O-P) Posterior view of hoxd10 expression. (Q-T) Quantification of A-P patterning defects associated with Sall4 knockdown. (Q) Measurement of the length between the anterior-most expression of otx2 and the first krox20 stripe in arbitrary units (AU) between the left (open bars) and right (closed bars) side of uninjected control embryos and embryos injected with 20 ng Sall4 MO into the right A/D blastomere. (R) Measurement of the length of hoxb9 expression domain in arbitrary units (AU) between the left (open bars) and right (closed bars) side of uninjected control embryos and embryos injected with 20 ng Sall4 MO into the right A/D blastomere. (S) Measurement of the length of hoxc10 expression domain in arbitrary units (AU) between the left (open bars) and right (closed bars) side of uninjected control embryos and embryos injected with 20 ng Sall4 MO into the right A/D blastomere. (T) Measurement of the length of hoxd10 expression domain in arbitrary units (AU) between the left (open bars) and right (closed bars) side of uninjected control embryos and embryos injected with 20 ng Sall4 MO into the right A/D blastomere. Error bars represent one standard deviation from the mean. Means were compared between left and right sides by student s T-test (*: p<0.05, **:p<0.01, ***:p<0.001). 85

96 UC Sall4 MO (bilateral) A B sox2 C D n-tub E F snai2/nkx6.1 G H myod UC Sall4 MO right A/D cell I J Q otx2/krox20/hoxb9 K L R hoxb9 M N S hoxc10 O P T hoxd10 Left Right * UC Sall4 MO Right A/D Cell *** UC Sall4 MO Right A/D Cell ** UC Sall4 MO Right A/D Cell * UC Sall4 MO Right A/D Cell Length (AU) Length (AU) Length (AU) Length (AU) 86

97 Figure 5.6: cdx2 is directly activated by canonical Wnt signaling and is not affected by Sall4 knockdown. (A) qpcr on 5 whole embryos or 15 to 25 animal caps treated according to the conditions indicated on the X-axis. The Y-axis shows expression relative to odc. (B-C) cdx2 expression at stage 18 shown in dorsal view with the anterior toward the top. (B) Uninjected control embryo. (C) Embryo injected with 20 g Sall4 MO in one animal-dorsal cell at the 4-cell stage. 87

98 A 450 cdx noggin TVGR EtOH DEX CHX WE Animal Caps B UC Sall4 MO right A/D cell C cdx2 8/8 5/6 88

99 Figure 5.7: Knockdown of Sall4 causes an increase in expression of the pouv/oct4 homologs. (A-L) Whole-mount in situ hybridization of late neurula staged embryos shown in dorsal view with the anterior toward the top. (A, D, G, J) Uninjected control embryos. (B, E, H) Embryos injected with 20 ng Sall4 MO into the right A/D blastomere. (C, F, I) Embryos injected bilaterally with 40 ng Sall4 MO (20 ng/blastomere at the 2-cell stage). (K-L) Embryos injected with 150 ρg PouV RNA (50 pg pou25 RNA, 50 pg pou60 RNA, and 50 pg pou91 RNA) into the right A/D blastomere. Red staining indicates β-galactosidase which is used as a tracer for RNA injection. (M-N) Higher magnification view of indicated regions in K and L, respectively. (A-C) Expression of pou25 (D-F) expression of pou60 (G-I) expression of pou91. (J-L) Expression of otx2/krox20/hoxb9. (O) qpcr for pou25 in uninjected embryos, embryos injected with 40 ng Sall4 MO (20 ng MO/blastomere at the 2-cell stage), and embryos injected with 40 ng Sall4 MO +500 pg Xtsall4 RNA (20 ng MO/blastomere at the 2-cell stage+250 pg Xtsall4 RNA/dorsal blastomere at the 4-cell stage). (P) qpcr for pou60 in uninjected embryos, embryos injected with 40 ng Sall4 MO (20ng MO/blastomere at the 2-cell stage), and embryos injected with 40 ng sall4 MO+500 pg Xtsall4 RNA (20 ng MO/blastomere at the 2-cell stage+250 pg Xtsall4 RNA/ dorsal blastomere at the 4-cell stage). (Q) qpcr for pou91 in uninjected embryos, embryos injected with 40 ng Sall4 MO (20 ng MO/blastomere at the 2-cell stage), and embryos injected with 40 ng sall4 MO+500 pg Xtsall4 RNA (20ng MO/blastomere at the 2-cell stage+250 pg Xtsall4 RNA/dorsal blastomere at the 4-cell stage). (O-Q) Expression is measured relative to odc. Error bars represent one standard deviation from the mean. Means compared to uninjected control by one-way ANOVA followed by Tukey post-hoc analyses (*: p<0.05). 89

100 A B C O pou25 D E F P pou60 UC Sall4 MO right A/D cell Sall4 MO (bilateral) G H I Q Fold Change Fold Change Fold Change pou25 * UC Sall4 MO Sall4 MO +Xtsall4 RNA * pou60 UC Sall4 MO Sall4 MO +Xtsall4 RNA pou91 * pou91 UC 150 pg pouv RNA right A/D cell J K L 0.5 UC Sall4 MO Sall4 MO +Xtsall4 RNA otx2/krox20/hoxb9 M N 90

101 Figure 5.8: A second non-overlaping Sall4 morpholino results in similar phenotypes. (A-F) Whole-mount in situ hybridization if late neurula stage embryos shown in dorsal view with the anterior toward the top. (A, C, E,) Uninjected control embryos and (B, D, F) embryos injected with 20 ng Sall4 MO2 into the right A/D blastomere. (A-B) Expression of otx2/krox20/hoxb9 (C-D) Posterior view of hoxd10 expression. (E-F) Expression of pou91. 91

102 A UC Sall4 MO2 right A/D cell B otx2/krox20/hoxb9 C D hoxd10 E F pou91 92

103 Figure 5.9: Loss of spinal cord gene expression in Sall4 morphants requires an increase in pouv/oct4 expression. (A-M) Whole-mount in situ hybridization of late neuriula stage embryos shown in dorsal view with anterior toward the top. (A-D) Uninjected control embryos. (E-H) Embryos injected with 40 ng Sall4 MO (20 ng/blastomere at the 2-cell stage). (I-L) Embryos injected with 40 ng Sall4 MO, 20 ng pou25 MO, 10 ng pou60a MO, 10 ng pou60b MO, and 20 ng pou91 MO (50 ng total MO/blastomere at the 2-cell stage). (M) Embryos injected with 20 ng pou25 MO, 10 ng pou60a MO, 10 ng pou60b MO, and 20 ng pou91 MO (30 ng total MO/ blastomere at the 2-cell stage). (A, E, I, M) Expression of otx2/krox20/hoxb9. (B, F, J) Expression of hoxc10. (C, G, K) Expression of hoxd10. (N) Quantification of posterior neural gene expression as measured by length of expression domain in arbitrary units (AU). Open bars: uninjected control embryos. Gray bars: embryos injected with 40 ng Sall4 MO (20 ng/ blastomere at the 2-cell stage). Closed bars: embryos injected with 40 ng Sall4 MO, 20 ng Xlpou25 MO, 10 ng Xlpou60a MO, 10 ng Xlpou60b MO, and 20 ng Xlpou91 MO (50 ng/ blastomere at the 2-cell stage). Error bars represent one standard deviation from the mean. Means compared to uninjected control by one-way ANOVA followed by Tukey post-hoc analyses (***: p<0.001). 93

104 A B C D Sall4 MO +pouv/oct4 MO Sall4 MO UC otx2/krox20/hoxb9 hoxc10 hoxd10 sox2 E F G H I J K L pouv/oct4 MO M N Length (AU) *** *** UC Sall4 MO Sall4 MO+pouV/Oct4 MO *** *** hoxb9 hoxc10 hoxd10 94

105 Figure 5.10: FGF and retinoic acid signaling fail to posteriorize Sall4 morphants. (A-L) Whole-mount in situ hybridization of late neurula stage embryos shown in dorsal view with anterior toward the top. (A, E) Uninjected control embryos. (B, F) Embryos injected with 40 ng sall4 MO (20 ng/blastomere at the 2-cell stage). (C, G) Embryos injected with 50 pg fgf8a RNA into the right animal dorsal (A/D) blastomere. (D, H) Embryos injected with 40 ng sall4 MO (20 ng/blastomere at the 2-cell stage) and 50 pg fgf8a RNA into the right A/D blastomere. (I, K) Embryos treated with 1 µm all-trans retinoic acid (ATRA). (J-L) Embryos injected with 40 ng sall4 MO (20 ng/blastomere at the 2-cell stage) and treated with 1 µm all-trans retinoic acid. (A- D, I, J) Expression of otx2/krox20/hoxb9 (E-H, K, L) expression of sox2. (M) A model to explain the role of Sall4 in Wnt-mediated neural posteriorization. Wnt activates sall4 then represses pouv/oct4 to allow for posteriorization via Wnt/Fgf/RA. Solid and dashed arrows indicate direct and direct or indirect regulation, respectively. 95

106 UC Sall4 MO 50 pg fgf8a RNA 1 A/D Cell Sall4 MO + 50 pg fgf8a RNA 1 A/D Cell A B C D otx2/krox20/hoxb9 E F G H sox2 I G 1 M ATRA J Sall4 MO + 1 M ATRA M Wnt/ -catenin Sall4 Fgf/RA Normal pouv/oct4 otx2/krox20/hoxb9 K L Wnt/ -catenin X Sall4 pouv/oct4 Posterior neural differentiation (Hoxb9/Hoxc10/Hoxd10) Fgf/RA Sall4 morphants sox2 Posterior neural differentiation (Hoxb9/Hoxc10/Hoxd10) X X X 96