Effects of Ciona Intestinalis Distal-Less-B Nonexpression and Overexpression on Phenotype and Downstream Targets

Ciona intestinalis can serve as a useful model for developmental studi es in the chordate lineage due to its basal position in the c hordate phylogeny. It shows a simplified chordate body plan during its developmen t, during which important genetic pathways are conserved with vertebrates, and its de velopmental gene regulation can be manipulated through the insertion of transgenic DNA by electroporation. The Distalless (Dll) genes are a family of homeobox genes in chordates that are homologous to the single Dll gene of other metazoan animal groups and code for d evelopmental factors that play a role in determining multiple de velopmental cell fates. These include broadly conserved roles in appendage development an d sensory functions of the central nervous system, as well as more novel roles such as differentiation of the epidermis in chordates. In C. intestinalis, Dll-B transcripts are expressed throughout the prospective epidermis during gastrulation. To s tudy the role of Ci-Dll-B in development, I have produced a transgenic dominant egative form of Ci-Dll-B by making use of the Drosophila engrailed repressor domain (EnR). I then examined its effects on development. Embryos electroporated with th s construct showed defects in adhesion of cells in the epidermis. Whole mount in situ hybridization analysis of known Ci-Dll-B downstream targets showed changes in gene expressi on only in certain targets, suggesting a degree of redundancy in the regulation of the epidermal development program. Phenotypic analysis and immuno fl orescent staining of epidermal markers suggest that Ci-Dll-B has a role in the regulation of cell adhesion or differentiation, since Ci-Dll-B knock-down alters the expression pattern of collage n and laminin. I also attempted to identify Ci-Dll-B gene targets through suppression subtractive hybridization, but was unsuccessful. Th ese results are consistent with earlier reports that Dll genes could have a role inducing final differentiat ion in the epidermis. This work characterizing a key gene in e pid rmal development may have implications for epidermal development in other cho rdates, such as mammals.


SCREENING OF CI-DLL-B KNOCK-DOWN
genes are a family of homeobox genes in chordates that code for developmental transcription factors (reviewed by Bendall and Abate-Shen, 2000;Panganiban and Rubenstein, 2002). The family is homologous to the Dll gene of Drosophila and is hypothesized to have arisen through several duplications (Stock et al., 1996;reviewed in Sumiyama et al., 2003). A duplication early in the chordate lineage resulted in a cluster of two genes in close proximity to the Hox gene cluster ( Fig. 1.1) (reviewed in Panganiban and Rubenstein, 2002;. In lancelets, the most basal chordate group (Delsuc et al., 2006), there is one Dll gene, (Holland et al., 1994), suggesting that this is the ancestral condition for chordates, whereas in the vertebrate lineage further duplications have resulted in additional bigene clusters. There are four genes in lampreys (Neidert et al., 2001), six in elasmobranchs (Stock, 2005) and tetrapods (reviewed in Panganiban and Rubenstein, 2002), and eight in teleost fishes (Fig 1.2) (Amores, et al., 1998).
The Dll genes code for transcription factors that play a role in determining developmental cell fates. Dll proteins have a role in diverse lineages of metazoan animals for programming outgrowths from the body wall (Cohen and Jurgens, 1989;Panganiban et al., 1997;Robledo et al., 2002). In Drosophila the Distal-less gene is expressed in the distal regions of head and thorax appendages during development; in Dll mutants these appendages do not develop distal regions normally (Cohen and Jurgens, 1989), which is the origin of the gene's name. It also has an ancestral role in the central nervous system. This role might pre-date its role in limb development as Dll homologs have been identified in basal metazoans without limbs, such as nematodes (Panganiban et al., 1997). Nervous system expression of Dll homologs is often associated with putative sensory organs, particularly olfactory and auditory (Solomon and Fritz, 2002;de Melo et al., 2003;Long et al., 2003;Perera et al., 2004;Brill et al., 2008;Winchell et al., 2010). Since the morphology of limb structures showing Dll homolog expression among different animal lineages is not homologous, and since sensory organs are frequently found on limbs, it has been speculated that this sensory role for Dll is what led to its frequent cooption for the development of limbs (Mittmann and Scholtz, 2001;Winchell et al., 2010). As Dll genes have been duplicated in the chordate lineage, they have taken on new roles in chordate development, including partitioning the ancestral role of patterning the central nervous system with different genes to pattern separate regions (Akimenko et al., 1994;Ghanem et al., 2003), expression in the craniofacial skeleton and palate (Levi et al., 2006), and determining fates along the proximo-distal axis of novel chordate facial structures arising from the pharyngeal arches (Depew et al., 2002;Park et al., 2004;. Interestingly, new roles for Dll are not unique to the chordate lineage. For example, in lepidopterans Dll helps pattern localized regions in the color scales of the wings, a structure that is unique to this lineage, where it is expressed in eyespot foci (Carroll et al., 1994;Beldade et al., 2002;Reed and Serfas, 2004).
In chordates, members of the Dll gene family are also believed to play a role in patterning ectodermal development, particularly in the epidermis (Imai et al., 2006;Irvine et al., 2007). Broad expression of a member of the Dll family has been observed in several chordates, in the developing animal hemisphere, which is the side of the embryo where cell division is more rapid and fated to give rise to the ectoderm. These genes include the sole homolog in lancelets (Holland et al., 1996). In zebrafish, dlx3b (Akimenko et al., 1994;Quint et al., 2000) and dlx4b (Ellies et al., 1997) are expressed in the rostral ectoderm located at the anterior of the developing nervous system. Dlx3, Dlx5, and Dlx6 are also expressed in the rostral ectoderm of murine embryos (Quint et al., 2000). In Xenopus, Dlx3 is expressed in putative epidermis and Dlx5 and Dlx6 are expressed in a domain at the border between ectodermal cells fated to become epidermis and those fated to become neural cells (Dirksen et al., 1994;Luo et al., 2001;Woda et al., 2003).
In summary, Dll genes are transcription factors with conserved roles in appendages, sensory organs, the central nervous system, and lineage specific roles. In the chordate lineage, where gene duplications have produced a Dlx gene family, these last include regulation of the developing epidermis. Homologs from this gene family display an ectodermal expression pattern consistent with this role in multiple chordate lineages, including teleost fishes, amphibians, and mammals. Intriguingly, the Dll homolog Ci-Dll-B has also been identified as a key regulator of epidermal development in the ascidian Ciona intestinalis (Imai et al., 2006).

The model species Ciona intestinalis.
Urochordates are the closest sister group to vertebrates (Delsuc et al., 2006) and are located basally within the chordate lineage ( Fig. 1.2). As a basal invertebrate chordate group, urochordates can provide both insight into the early evolution of vertebrates and a simpler chordate model than the more complex vertebrates. The subphylum Urochordata includes the ascidians, for which C. intestinalis is a commonly employed model species. The similarity of ascidian larvae to a simple vertebrate form has long been recognized ( Fig. 1.3A) (Foster, 1869) and has led to the idea that the ascidian larva represents a prototypical chordate body plan (Garstang, 1928). Upon hatching, free-swimming tadpole larvae display a body plan comparable to the phylotypic development stage of vertebrates ( Fig. 1.3A) including chordate-defining features such as a dorsal neural tube and a notochord; however, after less than one day the larvae attach to a substrate using rostral palps and begin a radical metamorphosis to a sessile form ( Fig. 1.3B). In recent years, molecular studies have revealed conservation of genetic pathways in developmental patterning between ascidians and vertebrates (reviewed in Lemaire et al., 2008). Even if the ascidian larva is not entirely representative of ancestral chordates, conserved genetic pathways can provide insight into what sort of morphological features must have been present in common ancestors and the derivation of modern vertebrate traits (reviewed in Hall, 2003;Shubin et al., 2009).
The C. intestinalis larva has many advantages as a model of early chordate development. It is relatively small, consisting of about 2500 cells. The early blastomeres are large and their later fates well documented (Fig 1.4) (Conklin, 1905).
Development is rapid, proceeding from fertilization to the tadpole larva in about 18 hr at 18 o C, though varying the incubation temperature by several degrees allows for somewhat faster or slower development (Hotta et al., 2007). Transgenic DNA can be transformed into fertilized eggs by electroporation (Corbo et al., 1997;Vierra and Irvine, 2012). This is typically accomplished through suspension of dechorionated fertilized eggs in a solution of supercoiled plasmid DNA, followed by an electrical pulse to drive the plasmid DNA into the embryos. Transgenes are usually not incorporated into the embryo's genomic DNA, though this has been observed with some techniques (Matsuoka et al., 2005), but instead produces extra-chromosomal arrays. Expression is transient and frequently mosaic depending on which early blastomeres incorporate the plasmid. The genome of C. intestinalis is 160 million base pairs, one of the smallest genomes for a chordate that can be easily manipulated experimentally (Dehal et al., 2002). The small genome contains fairly few redundant genes, implying that inducing alterations in genes is likely to have a phenotypic effect (Sasakura et al., 2009). The C. intestinalis genome has been sequenced (Dehal et al., 2002), as has been the genome of C. savignyi (Vinson et al., 2005), allowing for comparison of genomic sequences with those of a closely related species for potentially relevant conserved regions (Johnson et al., 2004).

The Dll gene family in Ciona intestinalis. In C. intestinalis, the Dlx homolog
Dll-B is one of the key regulators of gene expression in the developing epidermis, according to an important study which examined the regulatory connections between dozens of regulatory genes identified in C. intestinalis (Imai et al., 2006). C.
intestinalis has three Dll genes, Dll-A, Dll-B, and Dll-C (Caracciolo et al., 2000). Dll-A and Dll-B are arranged in a bigene cluster (Di Gregorio et al., 1995) 2.75 megabases downstream from the portion of the C. intestinalis Hox cluster which is present on chromosome 7 (Irvine et al, 2007). Vertebrate Dlx homologs are also typically found in bigene clusters downstream from Hox clusters, a shared gene ordering suggesting homology between the C. intestinalis Dll bigene cluster and those of vertebrates ( Fig.  1.1). There is no known cluster partner for Ci-Dll-C, suggesting that it was either formed by the duplication of a single Dll gene, or alternatively that the Dll cluster was duplicated but only one duplicated gene remained functional. It has been hypothesized that these clusters have been maintained because of a need to share common regulatory regions between the genes for their correct expression (Sumiyama et al., 2002;Irvine et al., 2007). In support of this hypothesis, regulatory regions have been observed in the intergenic regions between Dlx genes in several chordates Ghanem et al., 2003;Park et al., 2004;.
Expression of Ci-Dll-A is seen in the trunk ectoderm by the mid-tailbud stage of development. Expression continues through the larval stage and is particularly focused on the primordia of the atrial siphon (Caraciolo et al., 2000) as well as other sensory placode-like structures (Irvine et al., 2007). Ci-Dll-C expression begins during gastrulation and by hatching is specifically detectable in the adhesive organ. (Caraciolo, 2000).
In C. intestinalis, Dll-B has a chordate specific ectodermal expression pattern.
Maternal transcripts are present in the egg, but localized to the posterior vegetal hemisphere (Caraciolo et al., 2000), the side of the embryo where cell division is less rapid. Zygotic expression starts at the 64 cell stage and can be detected in all a-line and b-line animal hemisphere blastomeres ( Fig. 1.4), with expression being maintained in these cell lineages into early gastrulation. In later gastrulation Dll-B expression is confined to equatorial cells in the animal hemisphere and non-neural ectoderm. By neurulation Dll-B expression is radically down-regulated. It becomes restricted to isolated anterior neuroectodermal cells and during the tailbud stage is found in cells that are potentially precursors to the palps (Irvine et al., 2007).

Ci-Dll-A
and Ci-Dll-B expression is non-overlapping with sensory expression partitioned to Ci-Dll-A and pan-ectodermal expression partitioned to Ci-Dll-B (Irvine et al., 2007). This is unlike what has been observed in vertebrates, where there is typically overlap in the expression of members of the same Dlx bigene cluster. Since this unique partition of function is not seen in other chordate lineages, it suggests that the function of the Dll homologs in Ciona diverged after the evolutionary split from vertebrates. Dll is also restricted to anterior expression at later embryonic stages in other ascidians, suggesting it is especially important in this region .  (Imai et al., 2006), and ci-ADMP (Imai et al., 2012). These putative targets also have gene regulatory roles. Like Ci-Dll-B itself, several are also transcription factors; Emx is a homeobox transcription factor (Patarnello et al., 1997), FoxHa and FoxC are members of the forkhead box gene family of transcription factors (reviewed in Hannenhalli and Kaestner, 2009), GATA-b is a zinc finger-containing transcription factor (Molkentin, 2000), and SoxB2 is a transcription factor of the HMG family (Guth and Wegner, 2008). The remaining targets have roles in cell-cell signaling pathways; SOCS1/2/3 acts as an inhibitor of cytokine signaling between cells as part of the JAK/STAT pathway (Krebs and Hilton 2000) and ADMP is a ligand of the BMP signaling family (Imai et al., 2012). Most of these putative targets were identified by Imai et al. (2006) by knocking down gene expression of Ci-Dll-B at the post-transcriptional level using Morpholinos. Morpholino molecules consist of the standard nucleic acid nitrogenous bases and a non-biological backbone of morpholine rings in place of ribose sugars and phosphorodiomidate in place of ionic phosphate. A ~25-mer Morpholino antisense to the target mRNA can bind it in the same manner as a biological nucleic acid; however, the non-biological backbone cannot be recognized by cellular proteins, leaving translation of the mRNA sterically blocked (Summerton and Weller, 1997). Despite the identification of these putative targets, the functional role of Ci-Dll-B expression in the developing epidermis is still poorly understood.
Development of the epidermis. The initial patterning of the epidermis in chordates is still not well understood, but it is thought to begin under the influence of maternal determinants. The identities of the initial maternal determinants vary between chordate lineages. In zebrafish and Xenopus these initial maternal determinants promote signaling by Nodal in the vegetal hemisphere (Schier and Talbot, 2005;Heasman, 2006) to establish endodermal and mesodermal identities. At the animal pole, repressors of nodal signaling such as zic2 (Houston and Wylie, 2005), sox3 (Zhang et al., 2004), and ectodermin (Dupont et al., 2005) inhibit endo-mesodermal identity and position the border of the ectoderm. In C. intestinalis on the other hand, nodal signaling does not establish endodermal or mesodermal identity (Hudson and Yasuo, 2006), placing initial establishment of the ectoderm under the control of a different maternally initiated pathway. Ectodermal identity is initially established by Ci-GATA-a (Rothbächer et al., 2007). At the third cell division, the future ectoderm in the animal hemisphere and endo-mesoderm in the vegetal hemisphere divide from each other and zygotic expression of β-catenin begins, repressing GATA-a in the vegetal hemisphere. Expression of Ci-otx is repressed by an unidentified member of the Ets family until the beginning of neural induction. FGF signaling then activates Ciotx in the neural ectoderm while cells where it remains repressed develop into epidermis. Later development of the epidermis during the tailbud stage appears to be patterned by a combinatorial code of roughly ten transcription factors while dorsal and ventral midline identities are induced by FGF signaling and BMP signaling respectively (Pasini et al., 2006).
The initial factor responsible for activation of Ci-Dll-B remains unknown, but sequence analysis suggests SoxB1 and intriguingly GATA-a as possibilities (Irvine, unpublished). In Xenopus activation of Dlx3 is mediated by BMP signaling (Suzuki et al., 1994), presumably though the activation of an unknown regulator of Dlx3 (Beanan and Sargent, 2000). In addition to Ci-Dll-B, other genes that imply a shared regulatory network in the epidermis between C. intestinalis and vertebrates due to similar expression patterns include AP2 (Snape et al., 1991;Imai et al., 2004), KLF4 (Segre et al., 1999), Ash2l (Tan et al., 2008), and Hes1 (Fuchs, 2007).
Purpose of this study. This study has further examined the nature of Ci-Dll-B expression in the developing epidermis through production of a transgenic dominant negative of the Ci-Dll-B gene. This was used to examine its effects upon putative downstream target genes, and to compare its effects to those resulting from Ci-Dll-B misexpression in non-ectodermal tissues. Two knock-down strategies were attempted.
One sought to make use of a small interfering RNA (siRNA) construct to silence expression of Ci-Dll-B at the post-transcriptional level, while the other produced a transgenic construct fusing the Ci-Dll-B gene transcript with the powerful repressor domain of the Drosophila engrailed gene (EnR) (Jaynes and O'Farrell, 1991;Vickers and Sharrocks, 2002 In the notochord, defects were present in a mosaic pattern with some cells disrupted

Introduction
In order to analyze the effects of a Ci-Dll-B knock-down phenotype, it was first necessary to produce transgenes capable of reducing the expression of wild type Ci-Dll-B. Two alternate approaches were attempted based on two mechanisms used commonly to knock down gene expression. Several additional constructs were produced for control purposes, including an overexpression construct to rescue the normal phenotype.
One strategy sought to make use of RNA silencing. This involved constructing a transgene to produce a small interfering RNA (siRNA). siRNAs are short RNA molecules 20 to 25 base pairs long that are capable of silencing the expression of specific genes post-transcriptionally both as an endogenous regulatory mechanism (Hamilton and Baulcombe, 1999) and when introduced synthetically (Elbashir et al., 2001). The antisense construct produced here was complementary to an intron-exon junction of the Ci-Dll-B pre-mRNA. This was expected to degrade the pre-mRNA prior to translation (Smith and Davidson, 2008), preventing expression of Ci-Dll-B.
Like in most marine invertebrates, siRNA techniques in C. intestinalis remain poorly developed (Stolfi and Christiaen, 2012) and the method used here was recently developed and first applied in echinoderms (Smith and Davidson, 2008). Verification of the efficacy of this construct was therefore vital and was performed using transcripts using qRT-PCR (Imai et al., 2006). This raised the possibility of confounding effects in attempting to knock down expression of Ci-Dll-B in this way; therefore, an additional strategy was also employed.
The other strategy used was to create a fusion protein combining endogenous Ci-Dll-B with a repressor domain. The modular nature of proteins allows the creation of dominant negative variants of a protein by the addition of a powerful repressor domain such as that of the Drosophila engrailed gene (Jaynes and O'Farrell, 1991).
The mechanism by which such proteins typically work is to out-compete the endogenous gene for its binding sites in its downstream targets and then repressing instead of activating them. This strategy is better established in Ciona intestinalis (Spagnuolo and Di Lauro, 2002;Mita and Fujiwara, 2007) (Table 2.1) and cloned into the CiDB-1.0 vector (Irvine et al., 2011) at the NotI and BlpI restriction sites.
The CiDB-1.0::DllB/VP16 (DBOE) construct was made from Ci-Dll-B clone CiGC11g14 obtained from the Ciona Gene Collection . The Ci-Dll-B cDNA sequence was amplified with forward primer DBOEintforward with a 5' XhoI site and reverse primer DBOEintreverse with a 5' BamHI site ( from CiDB-A vector (Irvine et al., 2011) with forward primer CiDB5.0forwardA or CiDB5.0forwardB with a 5' AscI site and reverse primer CiDB5.0reverseA or CiDB5.0reverseB with a 5' NotI site (Table 2.1). The amplified products were hybridized as described in Zeng (1998) and cloned into the lacZ reporter gene construct TV13 (Irvine et al., 2008) at the AscI and NotI restriction sites.

Construction of CiDB-2.5::siRNA (DBsR).
The first strategy used to attempt a knock-down phenotype was the production of the siRNA expression construct. This construct was named CiDB-2.5::siRNA (DBsR) (Fig. 2.2). DBsR was designed to silence Ci-Dll-B through expression of an siRNA to bind the splicing site of the first intron at the junction with the second exon in the Ci-Dll-B pre-mRNA. This mechanism has been shown to be able to knock down gene expression in the sea urchin Strongylocentrotus purpuratus (Smith and Davidson, 2008). The mechanism by which this repression works is only partially understood. Smith and Davidson (2008) demonstrated that their siRNA construct caused the target pre-mRNA to be degraded after binding as opposed to sterically blocking splicing. However, since the pre-mRNA does not leave the nucleus before splicing, degradation cannot be due to classical RNA silencing pathways. Efficacy of such a construct binding the Ci-Dll-B pre-mRNA as a knock-down could therefore be tested using qRT-PCR to detect  (Imai et al., 2006). This effect was seen at both doses of DBsR used in electroporation. Attempts were also made to measure the levels of expression of the Ci-Dll-B target gene Ci-FoxC, however these were unsuccessful due to lack of priming. This construct was named CiDB-1.0::EnR/DllB (DBDN) (Fig. 2.5). Engrailed is a homeobox transcription factor identified in Drosophila as a potent repressor. Its repressor domain EnR is capable of silencing all activated expression, though not basal transcription (Han and Manley, 1993). This indicates that the mechanism of repression for EnR is a form of direct repression either disrupting the transcription pre-initiation complex after it has been formed or interfering with its interaction with other transcription activators. Due to the modular nature of proteins, it is possible to remove the domain responsible for gene activation from a transcription factor protein and convert it into a repressor by substituting a repressor domain without otherwise disrupting its function. Previous studies have demonstrated that the EnR domain produces a dominant negative phenotype used in this way (Jaynes and O'Farrell, 1991;Vickers and Sharrocks, 2002). Furthermore it has already been shown that it can be used for this purpose in ascidians Wada et al., 2002;Sawada et al., 2005), including C. intestinalis (Spagnuolo and Di Lauro, 2002;Mita and Fujiwara, 2007), as well as with Dlx vertebrate homologs (Woda et al., 2003). To determine if the effects of the dominant negative construct could be rescued, an overexpression rescue construct for the Ci-Dll-B gene was constructed.

Construction of
This construct was named CiDB-1.0::DllB/VP16 (DBOE) (Fig. 2.6). This construct was used in co-electroporation experiments with the Ci-Dll-B dominant negative construct in an attempt to restore the wild type phenotype in experimental embryos.
This strategy made use of the Herpes simplex viral protein 16 (VP16) activator domain. VP16 is a strong transcriptional activator , and its activator domain has been shown to render transcription factors constitutive activators when fused to the DNA binding domain (Sadowski et al., 1988). The mechanism by which this promiscuous activator domain activates transcription is unknown, but it has been shown to interact with components of the RNA polymerase II transcription preinitiation complex, including TBP (Shen et al., 1996) and the general transcription factor TFIIB (Jonker et al., 2005). This suggests possible mechanisms for the VP16 activator such as contributing to the recruitment of the transcription pre-initiation complex or shutting down autoinhibition of TBP (Hall and Struhl, 2002). It has already been shown that the VP16 activator can be used to produce overexpression constructs in ascidians Sawada et al., 2005) and with Dll-B vertebrate homologs (Woda et al., 2003). DBOE was designed to include a fusion

Construction of a reporter transgene. To complement the Ci-Dll-B
expression vectors already available in the Irvine lab, the CiDB-5.0 expression vector-lacZ expression reporter construct was also produced (Fig 2.11). It consisted of conserved regulatory elements from genomic sequences 5 kilobases upstream of the Alternatively, expression of DBsR may not have occurred as expected. Future studies to understand the mechanism involved here could attempt could have been made to verify transcription of DBsR, for example, by use of a Northern blot using a probe specific for the siRNA sequence.  (Sato et al., 2012) or negative autoregulation (Brend and Holley, 2009), as well as interactions with additional factors to modify the autoregulatory effect (Aota et al., 2003;Ebert et al., 2003). Alternatively, the apparent negative autoregulation detected by Imai et al. (2006) might have been an artifact of their screening method.               (Pfaffl et al., 2002). For primers, see Table 3.1. tested over a range from 5µg to 100µg when compared to control embryos (Table 3.2, Fig. 3.5). The range of severity of such phenotypes increased at doses of 40µg or above, as did the average percentage of embryos displaying a phenotypic effect (Table   3.3, Fig. 3 (Table 3.3, Fig. 3.6).

Ci-Dll-B dominant negative transgene produces a distinctive notochord
phenotype. In order to determine whether endodermal or mesodermal cell layers are affected by reducing Ci-Dll-B expression in the normal domain, confocal microscopy was performed on DBDN electroporated embryos stained with phalloidin to show cell boundaries. Phalloidin staining showed a mosaic pattern of disruption in the mesodermally derived notochord (Fig. 3.7). Most sections of the notochord formed a single row of cells as expected in wild type embryos. Other sections did not form the expected single row of cells, however. Disruption of notochord alignment could be due to the disruption of signaling from the epidermis. This is in contrast to embryos electroporated with a construct which expresses Ci-Dll-B in mesodermal tissue, where disruption of the organization of the notochord is more extensive (Fig. 3.1). and Ci-SOCS1/2/3 were chosen for further testing, in addition to Ci-Dll-B itself. All WMISH embryos were compared to stained negative control embryos to determine the level of background staining (Fig. 3.8). Compared to control embryos, DBDN electroporated embryos showed increased expression levels of transcripts hybridizing to the Ci-Dll-B probe at all stages analyzed (Fig. 3.9). Among known Ci-Dll-B targets, there was little apparent effect on the level of expression of Emx (Fig. 3.10) or Ci-SOCS-1/2/3 (Fig. 3.11), which was already being expressed by the late gastrula stage (~6 hr post fertilization at 18 o C) (Fig. 3.11A-B). However, expression of Ci-SoxB2 was reduced in DBDN embryos, particularly at later stages ( Fig. 3.12). treated with the secondary antibody only (Fig. 3.13); however, the SP1.D8 anticollagen antibody did not show apparent reaction with C. intestinalis embryos. Signal detected from DBDN electroporated embryos was more intense than from wild type embryos ( Fig. 3.13). This was potentially due to the greater visibility of endomesodermal cells expressing collagen and laminin in these embryos, or alternatively to increased production of these proteins due to the effects of the Ci-Dll-B knock-down construct. This result suggests that Ci-Dll-B attenuates expression of these ECM proteins. Interestingly, DBDN electroporated embryos showed an apparent reduction in the size of cells present in the epidermis compared to wild type embryos ( Fig.   3.13A-B), suggesting that reduction of Ci-Dll-B expression has an effect on epidermal cell growth.

Quantitative real time PCR analysis of Ci-Dll-B misexpression.
To determine if expression of Ci-Dll-B in ectopic domains affects cell type or behavior, the DBME transgene was used to drive transcription of Ci-Dll-B in the endoderm and mesoderm, where it is normally absent. qRT-PCR was then used to compare expression levels of genes of interest between embryos electroporated with DBME and those electroporated with the reporter construct DBFl, as a control. mRNA from electroporated embryos was used as a template for cDNA synthesis. The resulting cDNA was amplified using primers for target genes by qRT-PCR and relative expression levels normalized using an average of the expression ratio of two housekeeping genes, Ci-β-actin and Ci-calreticulin. Ci-Dll-B was substantially upregulated in the misexpression construct ( Fig. 3.14), while the endo-mesodermal marker Ci-FoxAa and the known Ci-Dll-B target Ci-GATA-b were down-regulated ( Fig. 3.14). This result suggests that Ci-Dll-B is capable of directly or indirectly activating Ci-FoxAa and Ci-GATA-b transcription. Ct values assigned to raw fluorescence indicated that expression levels of Ci-GATA-b were lower than the other mRNAs tested (Table 3.4). Interestingly, levels of expression for the epidermal marker gene Ci-Epi1 were similar between experimental and control embryos (Fig. 3.14).
This result suggests that misexpression of Ci-Dll-B in the endoderm or mesoderm does not broadly alter the fates of the cell types present there.

Knock-down of Ci-Dll-B disrupts normal epidermal assembly. Knock-down
of the effects of Ci-Dll-B expression using a dominant negative construct resulted in disruption of the outer epidermal cell layers of C. intestinalis embryos. This disruption was most apparent in the tail and was first detectable as the tail began lengthening.
Although endogenous pan-ectodermal Ci-Dll-B transcript expression occurs at an earlier stage, as a transcription factor Ci-Dll-B affects the expression of genes responsible for epidermal patterning at a later stage. Therefore a delay in the appearance of a phenotypic effect would be expected.
Comparison of the phenotypes produced by the DBDN and control constructs showed that DBDN was responsible for the observed phenotypic changes in the epidermis (Fig 3.3) while the EnR sequence could not produce this phenotype by itself (Fig 3.4C). When co-electroporated with DBOE, embryos showed phenotypes comparable to wild-type embryos or those electroporated with constructs known not to phenotypically affect C. intestinalis (Fig. 3.4A-B). As the repressor properties of EnR (Vickers and Sharrocks, 2002) and the activator properties of VP16 (Sadowski et al., 1988) are both well documented, these effects were consistent with the expectation that DBDN would out-compete and repress the effects of endogenous Dll-B expression, and demonstrated that the VP16 activator domain has the ability to rescue the effects of EnR. Since the expression of the EnR domain alone had no phenotypic effect upon epidermal morphology, it was concluded that the effects observed here were due to the fusion of EnR to the sequence-specific DNA binding protein Ci-Dll-B.

Ci-Dll-B dominant negative and misexpression phenotypes are distinct. The
DBDN transgene caused disruption of the epidermal epithelium of the tail. On the other hand, mesodermally derived cells such as the notochord usually retained their normal organization (Fig. 3.7). This was in contrast to the effects of misexpression of Ci-Dll-B in endo-mesodermal tissues, using the construct DBME, where the disruptions seen in endo-mesodermal tissues were more severe (Fig. 3.1). While Ci-Dll-B is normally expressed only in the ectoderm, its putative targets include genes associated with cell-cell signaling pathways including SOCS1/2/3 (Imai et al., 2006) and ADMP (Imai et al., 2012). It is therefore possible that while the primary role of Ci-Dll-B is in the epidermis, it could have a secondary role through cell-cell signaling in the correct assembly of cells in the vicinity of the epidermis such as the notochord.
It is also possible that overexpression of Ci-Dll-B in endo-mesodermal tissues may disrupt correct notochord assembly by affecting ECM proteins, as was seen in embryos electroporated with DBDN ( Fig. 3.13).
qRT-PCR analysis of DBME misexpression showed down-regulation of endomesodermal genes without up-regulation of epidermal genes, suggesting a disruption of normal endo-mesodermal patterning without respecification of these cells into ectodermal roles. The up-regulation of Ci-Dll-B expression in embryos with the DBME fusion transgene compared to control embryos indicated that this transgene was indeed functioning as a misexpression construct, increasing Ci-Dll-B mRNA levels in presumptive endo-mesoderm where Ci-Dll-B is normally inactive. As Ci-Epi1 is an epidermal marker, the lack of change in its expression level indicated that misexpression of Ci-Dll-B was not sufficient to respecify presumptive endo-mesoderm as epidermis. This suggests that expression of an additional factor is necessary for epidermis specification, or, alternatively, that an unknown factor was antagonizing Ci-Dll-B in the endo-mesoderm. The finding that Ci-GATA-b was down-regulated by the misexpression construct contradicted earlier findings that Ci-GATA-b is a Ci-Dll-B regulatory target. However, the low levels of Ci-GATA-b mRNA detected here (Table   3.4) are consistent with the possibility that this result was the sort of technical error to which qRT-PCR is sensitive. Ankyloblepharon-ectodermal dysplasia-clefting dysplasias are characterized by the reduction or absence of hair, teeth, and skin glands and are associated with alteration of DLX3 expression due to mutation of an upstream regulator (Radoja et al., 2007).

DBDN has limited effects on expression of putative
The phenotypic effects of altered Dlx expression typically limited to tissue morphogenesis rather than basic cell type specification or alteration of body plan patterning. Misexpression of Dlx family genes in vertebrates does not result in major alterations to limb morphology (Morasso et al., 1996). Although they are necessary factors for proper epidermal development, Dlx homologs in vertebrates are not sufficient to specify an epidermal cell fate (Feledy et al., 1999a;McLarren et al., 2003;Woda et al., 2003). However, malformation of the epidermis (Morasso et al., 1996;Hwang et al., 2011) and epidermally derived tissues such as hair (Hwang et al., 2008) and feathers (Rouzankina et al., 2004) is common. Moreover, loss-of-function of the Dlx3 gene in Xenopus can disrupt the fates of non-epidermal cell populations interacting with the epidermis, including the neural plate, neural crest, and cranial placodes (Woda et al., 2003).
Analysis of misexpression and loss-of-function of Dlx family genes in multiple vertebrate lineages indicates that Dlx homolog expression in the epidermis has proliferative and differentiative roles. In mice, premature differentiation of epidermal cells into keratinocytes resulting from Dlx3 misexpression has been shown to produce defects of variable severity in the terminally differentiated epidermis, characterized by the disappearance of cell layers in the stratum corneum (Morasso et al., 1996).
Furthermore, it appears Dlx3 misexpression or overexpression causes premature differentiation in the epidermis. In this case, alterations in the levels of expression of epidermal markers associated with different epidermal cell populations are consistent with premature differentiation depleting the supply of cells for later differentiating cell types. Loss-of-function results in a hyperproliferation of cells and changes in epidermal marker expression suggestive of changes in wild type cell differentiation (Hwang et al., 2011). The resulting epidermis is abnormal and fails to form a proper barrier. Dlx homologs also play roles in differentiation of hair and feathers, which are derived from the epidermis, but these roles are dissimilar. Dlx3 is necessary for the induction of hair follicle growth from the initial proliferating cell population in mice (Hwang et al., 2008), whereas Dlx2 and Dlx5 activate factors that inhibit the formation of feather buds (Rouzankina et al., 2004).  Fig. 3.13).

Comparison with vertebrate homologs suggests that phenotypic effects of
The reduction of Ci-SoxB2 expression, observed here in Ciona, is also consistent with this hypothesis, as the SoxB gene family has conserved roles in the regulation of cell proliferation and differentiation as well as of cell adhesion (Guth and Wegner, 2008). However, Ci-SoxB2 is unlikely to be involved directly in the establishment of this phenotype because a reduction in its expression is not apparent until after disruption of the epidermis is first apparent (Fig. 3.12). The phenotype observed in this study differs from the knock-down of Danio rerio sox21a, a zebrafish SoxB2 homolog, which results in ventralization of the developing embryo (Argenton et al., 2004). However, Dr-sox21a is maternally expressed, while Ci-SoxB2 expression is first seen during gastrulation. Due to the teleost fish-specific genome duplication, D.
rerio has an additional SoxB2 homolog, Dr-sox21b. This gene is not expressed until late in gastrulation, which could make it a more likely functional homolog for Ci-SoxB2. Dr-sox21b is necessary for lens development (Pauls et al., 2012). This suggests the possibility of regulation of Dr-sox21b by a Dll homolog, due to their frequent roles in sensory expression. Since Dll-B does not appear to have a sensory role in C.
intestinalis (Irvine et al., 2007), the function of Dll regulation of SoxB in tunicates may differ from that in teleost fishes, the vertebrate lineage where SoxB homologs have been most studied.
Whether the alteration of cell fates is responsible for the observed phenotype might be determined by further analysis of differentiated epidermal markers. Since the use of mouse derived antibodies to detect C. intestinalis structural proteins in this study was successful, this analysis could be accomplished through the continuation of such immunofluorescence experiments. Loricrin and filaggrin are two markers of differentiated epidermal tissue (Fuchs and Byrne, 1994) that would be strong candidates for observation. Misexpression of Xenopus Dlx3 in mice has been shown to cause ectopic production of these proteins (Morasso et al., 1996). Changes in the expression of these factors would be evidence that epidermal cell differentiation has been altered. If confirmed, this would provide new insight into the specific function of Ci-Dll-B within the differentiation of the epidermis and would imply a similar function for early ectodermal expression as that seen in other chordates.                  Embryos were electroporated with DBME to misexpress Ci-Dll-B and mRNA was then extracted to provide a qRT-PCR template to compare expression relative to wild type embryos electroporated with the reporter construct DBFl. Replicates were performed for each experiment in duplicate and the results from replicates were averaged. Error bars indicate minimum and maximum values for each gene. Differences in expression are shown on a log 2 scale. Red indicates a >2 fold increase in expression, blue a >2 fold decrease, and gray a <2 fold increase or decrease.

Introduction
This chapter describes an attempt to identify unknown targets of Ci-Dll-B.
Though unsuccessful, I gained experience with a technique called suppression subtractive hybridization that I might be able to use in the future.
Several putative downstream targets for Ci-Dll-B that have already been identified by others (Imai et al., 2006;Imai et al., 2012) were analyzed in this study; in addition, attempts were made to identify previously unknown targets. Previous studies have focused on smaller numbers of genes or looked at Ci-Dll-B as part of larger screens. To identify genes whose regulation is altered by the Ci-Dll-B dominant negative construct across the whole genome without the need to first identify candidates, this study sought to make use of the technique of suppression subtractive hybridization (SSH) (Fig. 4.1; Fig. 4.2) (Diatchenko et al., 1996;Diatchenko et al., 1999).
To perform SSH, mRNA is extracted first from tester experimental and control samples as a template for cDNA. These cDNA samples are then restriction digested with a frequent cutter such as RsaI and adaptors are ligated to the experimental sample to form the tester cDNA population at the restriction site (Fig 4.1). The adaptors consist of one of two double stranded oligonucleotides, resulting in two tester cDNA populations. The control cDNA used in the hybridization is not ligated with any adaptor and becomes the driver cDNA population without any further modification (Fig 4.1). Hybridization occurs in two rounds ( Fig. 4.2). The first hybridization serves to subtract out those cDNAs which are not differentially expressed between the samples. In the first hybridization each of the two tester cDNA populations is separately hybridized with driver cDNA in excess. Since driver cDNA will hybridize with tester cDNA, only those cDNAs more common in the tester population will not hybridize to the driver cDNA. Among the remaining tester cDNA the more common

Results
To prepare cDNA for hybridization, experimental embryos electroporated with DBDN and control wild type embryos were reared to the early tailbud stage and total RNA was extracted. First-strand cDNA was synthesized from this using an oligo-dT primer. To provide a reverse primer site, first strand cDNA was poly-G tailed and the oligo-dT primer used alongside an oligo-dC primer for cDNA enrichment. The cDNA product was then restriction digested with RsaI. The driver cDNA samples required no further preparation while a portion of each sample was ligated with one of two possible adaptors to produce the tester cDNA. To produce the cDNA library to be cloned, cDNA from the experimental embryos was used as the tester and cDNA from the wild type embryos as the driver. As a control a reverse subtractive hybridization was also performed switching the roles of experimental and wild-type cDNA as the tester and the driver.
After hybridization the enriched subtracted cDNA was cloned to form the subtracted library. Transformation of the cloned library yielded 724 colonies, of which 125 passed blue/white screening. Additional transformations yielded similar total colonies and ratios. Initial sequencing of selected colonies showed successful isolation of suppression subtractive hybridization library sequences, but these sequences consisted of non-cDNA contamination and non-differentially expressed genes.
A larger number of colonies were then screened by colony hybridization of transformants and hybridization with digoxygenin-labeled probes produced from the subtracted library, the control reverse subtracted library, or either of the two cDNA populations used to make the library. Colonies that showed hybridization to the library probe, but not to the reverse library or control driver probes were classified as strong candidates for differential expression. Clones that bound the subtracted library probe but not the reverse subtracted library probe or that showed stronger hybridization to the subtracted library probe than any other probes were classified as weak candidates which could be differentially expressed, but might have only been enriched in the library due to artifacts of the SSH method. All other probes were classified as noncandidates. Based on these criteria, out of 87 additional transformants screened eight were strong candidates for closer analysis (Table 4.2). Further sequencing of six of these clones and identification using NCBI-BLAST (Altschul et al., 1990) (http://blast.ncbi.nlm.nih.gov/Blast.cgi) failed to identify likely targets for Ci-Dll-B differential gene regulation (Table 4.3). Two colonies were identified as incompletely suppressed Ciona intestinalis housekeeping genes, three were identified as non-C.
intestinalis cDNA contaminants, and one could not be identified.

Discussion
The disruption of normal cell organization by alteration in normal Ci-Dll-B expression suggests that this gene has a role in cell adhesion mediated by targets not yet identified. Unfortunately, efforts to identify such targets using SSH were unsuccessful. Screening the subtracted library failed to identify any differentially expressed candidates due to a high degree of background contamination and incomplete suppression of non-differentially expressed genes. Many contaminant sequences were identified as being of human origin, suggesting contamination within the lab setting. Amplification of non-differential background sequences is a known issue for suppression subtractive hybridization (Rebrikov et al., 2000). Amplification of housekeeping genes as seen here frequently occurs even in more successful screenings. This is due to the failure to completely suppress these genes as a result of the high levels at which their transcripts are present. Moreover it is less effective at detecting genes as the difference in expression is reduced (Ji et al., 2002). Since alterations in cell adhesion were confined only to tissues in which Ci-Dll-B was expressed but template RNA was extracted from whole embryos, this could represent a low level of differential gene expression, reducing the ability to detect such genes using SSH.
The failure of SSH to recover potential differentially regulated targets could require the use of an alternate method. Several alternatives could be employed instead.
Microarray technology (Ali and Crawford, 2002) has previously been applied to detect differential gene expression in C. intestinalis (Ishibashi et al., 2003;Azumi et al., 2003). While arrays are available for use in C. intestinalis, they would have to be obtained from another laboratory, and reading the arrays would require equipment not available at the University of Rhode Island. Alternatively, recent advances in DNA sequencing technology could allow for the sequencing of cDNA extracted from dominant negative and wild type embryos using RNA-seq (Wilhelm and Landry, 2009;Costa et al., 2010). While an increasingly common and effective method of obtaining differential expression data, there would be several issues to consider. Only small amounts of total RNA could be obtained from the embryos available. This amount of total RNA was insufficient for purifying poly-A plus RNA. The available cDNA was amplified using rapid amplification of cDNA ends; however, this step was one potentially prone to contamination. Since RNA-seq functions most effectively with an mRNA template, this possible source for contamination would remain.
Finally, the sequencing reads produced by RNA-seq are short, ~30-40 base pairs, and would require the use of appropriate computational analysis to assemble the recovered cDNA sequences and determine which ones are differentially expressed (Pepke et al., 2009;Garber et al., 2011). As an alternative to this, chromatin immunoprecipitation sequencing (ChIP-seq) could be employed (Park, 2009). ChIP-seq functions by crosslinking the target protein to genomic DNA in experimental organisms, then lysing the cells and recovering the target protein through immunohistochemistry. The linked DNA can then be unlinked and sequenced. This technique would have the advantage of recovering sequences from genes known to be bound by Ci-Dll-B and would not require use of multiple treatments. Challenges would remain; mainly the lack of a suitable antibody and the need for appropriate analytical tools to analyze the data obtained (Pepke et al., 2009).
While differentially expressed genes detected in these types of screenings would be candidates for direct regulatory targets of Ci-Dll-B, further confirmation would be required. WMISH would initially be performed to determine if the expression pattern of the gene includes the epidermis as would be expected for a Ci-Dll-B target. Some of this data may already be available . The sequences of putative regulatory regions of these genes could then be checked for the presence of suitable binding sites for Ci-Dll-B, identified by the consensus sequence VTAATTRS (Feledy et al., 1999b). If found, they could be cloned and site mutations introduced into the cloned sequences at the presumptive locations to determine if they can still drive expression of a reporter that matches the normal expression pattern of the gene. DNase footprinting could be used to confirm the ability of Ci-Dll-B to bind the regulatory region of the target gene (Galas and Schmitz, 1978), but would have the disadvantage of only demonstrating this in vitro.
Most putative targets of Ci-Dll-B already identified are transcription factors indicating that Ci-Dll-B is not located at the end of the gene regulatory network responsible for cell differentiation and structure. However, the ability of Dll homologs to bind the sequence of the profilaggrin gene which codes for the precursor of the differentiated epidermal protein filaggrin has been demonstrated in mice (Morasso et al., 1996), raising the possibility that Ci-Dll-B could directly regulate some structural genes. For these reasons transcription factors identified by this type of an assay could be expected to be more likely targets. However, while structural proteins identified could instead be targets of the network downstream from Ci-Dll-B, the possibility of direct regulation of structural genes by Ci-Dll-B should not be excluded.

Oligonucleotide Sequence
Poly   Colonies were spotted on nylon membranes, grown overnight, lysed and cross-linked, and then probed with digoxygenin labeled DNA probes synthesized from subtracted library, reverse subtracted library, experimental driver, or control driver templates. Probes were bound with alkaline phosphatase labeled antibody, visualized and scored as showing strong hybridization (**), weak hybridization (*), or no hybridization (-). The scoring pattern was used to classify clones as potential candidates for differential expression. Clones that bound the subtracted library probe and neither control template probe were classified as strong candidates (shaded). Clones that bound the subtracted library probe but not the reverse subtracted library probe or that showed stronger hybridization to the subtracted library probe than any other probes were classified as weak candidates. All other probes were classified as non-candidates. Nine colonies that did not pass initial blue-white colony screening were included as negative controls.  Table 4.3. Sequence analysis of selected suppression subtractive hybridization library clones. The inserts of six colonies classified as strong candidates by the suppression subtractive hybridization colony screening were sequenced using standard vector primers. The inserts were identified and those that were identifiable classified as either incompletely suppressed C. intestinalis housekeeping genes or non-C. intestinalis cDNA contaminants. Green represents the tester cDNA (DBDN), red represents the driver cDNA (wild type), yellow represents primers for cDNA synthesis, and blue represents the adaptors annealed to the digested tester cDNA. Driver cDNA is ready for hybridization after RsaI digestion while tester cDNA is ready after adaptor ligation. Green represents the RsaI digested tester cDNA (DBDN), red represents the RsaI digested driver cDNA (wild type), and blue represents the adaptors annealed to the digested tester cDNA. Note that after the second hybridization, the recessed 3' ends produced by the adaptors are filled in during the initial cycle of PCR amplification and that molecules having adaptor 2 are also present but are not shown.