Dissercting the activities of Capicua, Dorsal and Groucho in Drosophila dorsoventral patterning Aikaterini Papagianni Aquesta tesi doctoral està subjecta a la llicència Reconeixement- NoComercial – CompartirIgual 4.0. Espanya de Creative Commons. Esta tesis doctoral está sujeta a la licencia Reconocimiento - NoComercial – CompartirIgual 4.0. España de Creative Commons. This doctoral thesis is licensed under the Creative Commons Attribution-NonCommercial- ShareAlike 4.0. Spain License. TESIS DOCTORAL Dissecting the activities of Capicua, Dorsal and Groucho in Drosophila dorsoventral patterning Aikaterini Papagianni Barcelona, September 2017 Departamento de Genética Programa de Doctorado de Genética Facultad de Biología Universidad de Barcelona Dissecting the activities of Capicua, Dorsal and Groucho in Drosophila dorsoventral patterning Memoria presentada por Aikaterini Papagianni Para optar al grado de Doctora en Biología por la Universidad de Barcelona Esta tesis ha sido realizada en el Departamento de Biologia del Desarrollo del Instituto de Biología Molecular de Barcelona (Consejo Superior de Investigaciones, Parc CientíOic de Barcelona), bajo la supervisión del Dr. Gerardo Jiménez Cañero. El director El alumno El tutor Dr. Gerardo Jiménez Aikaterini Papagianni Dr. Francesc Cebrià 
 INDEX Index ABBREVIATIONS .................................................................................................................................. 1 INTRODUCTION ................................................................................................................................... 3 1. EARLY DEVELOPMENT OF DROSOPHILA ................................................................................ 3 1.1 AP PATTERNING DURING EMBRYOGENESIS ............................................................................................... 5 1.2 TERMINAL PATTERNING DURING EMBRYOGENESIS ................................................................................. 5 1.3 DV PATTERNING DURING EMBRYOGENESIS ............................................................................................... 6 2. MATERNAL CONTROL OF DEVELOPMENT .............................................................................. 9 3. AXIS SPECIFICATION DURING OOGENESIS. ......................................................................... 10 3.1 DV POLARITY ESTABLISHMENT DURING OOGENESIS ............................................................................. 12 3.2 TERMINAL PATTERNING DURING OOGENESIS. ......................................................................................... 15 4. DORSAL, THE MAIN NUCLEAR EFFECTOR IN DV POLARITY ......................................... 16 4.1 STRUCTURE AND FUNCTIONS OF DL .......................................................................................................... 16 4.2 TRANSCRIPTIONAL REGULATION BY DL ................................................................................................... 17 4.3 DL-MEDIATED ACTIVATION ........................................................................................................................ 18 4.4 DL-MEDIATED REPRESSION ........................................................................................................................ 19 5. CAPICUA, A DL COREPRESSOR CANDIDATE ........................................................................ 24 5.1 STRUCTURE AND FUNCTIONAL DOMAINS OF CIC .................................................................................... 24 5.2 HUMAN HOMOLOGUES OF CIC AND THEIR FUNCTIONS .......................................................................... 25 5.3 MECHANISM OF CIC FUNCTION IN DROSOPHILA ..................................................................................... 25 5.4 REGULATION AND FUNCTIONS OF CIC DURING DROSOPHILA DEVELOPMENT…… ..................... 26 OBJECTIVES ........................................................................................................................................ 31 RESULTS .............................................................................................................................................. 33 1. FUNCTIONAL ANALYSIS OF CIC IN THE ESTABLISHMENT OF THE DV AXIS ............. 33 1.1 CIC MUTATIONS REVEAL A DOUBLE FUNCTION IN THE ESTABLISHMENT OF THE DV AXIS ............. 33 1.2 THE MAIN EMBRYONIC CIC ISOFORM IS INVOLVED IN EMBRYONIC DV PATTERNING ..................... 34 2. MOLECULAR MECHANISM OF CIC FUNCTION IN EMBRYONIC DV PATTERNING .... 39 2.1 CIC REPRESSES TYPE III GENES THROUGH THE AT-RICH SITES OF THE VRE ELEMENTS ........ 39 2.2 CIC AND DL SHOW A SYNERGISTIC FUNCTION FOR REPRESSION OF DORSAL- SPECIFIC GENES ..... 48 2.3 THE CIC N2 MOTIF RECRUITS GRO TO THE DL REPRESSOR COMPLEX ............................................... 54 3. STUDY OF CIC AS AN RTK SENSOR ELEMENT DURING THE ESTABLISHMENT OF THE EMBRYONIC DV AXIS ............................................................................................................. 61 DISCUSSION ........................................................................................................................................ 65 1. THE HMG-BOX FAMILY CIC REPRESSOR HAS A DUAL FUNCTION IN DV AXIS ESTABLISHMENT ....... 65 2. CIC BINDS, WITH LOW AFFINITY, TO THE AT-RICH SITES IN SILENCER ELEMENTS OF DORSAL-SPECIFIC GENES AND REQUIRES THE PRESENCE OF DL TO REPRESS THEM ............................................... 68 MECHANISMS OF CIC AND DL TRANSCRIPTIONAL SYNERGY ........................................................................ 71 4. INTERACTIONS BETWEEN THE CIC N2 MOTIF AND GRO ARE ESSENTIAL FOR REPRESSION OF DORSAL-SPECIFIC GENES IN THE EARLY EMBRYO ........................................................................................... 73 5. CIC IS A SENSOR OF RTK SIGNALLING DURING DV PATTERNING ........................................................... 76 CONCLUSIONS .................................................................................................................................... 79 MATERIALS AND METHODS .......................................................................................................... 81 1. SYNTNETIC DNA CONSTRUCTS ................................................................................................ 81 1.1 RECOMBINANT DNA TECHNIQUES ............................................................................................................ 81 1.2 TREATMENT OF DNA .................................................................................................................................. 81 2. RNA ANTISENSE PROBE GENERATION AND LABELLING ................................................ 92 Index 3. GENETIC ANALYSES ..................................................................................................................... 94 3.1 FLY CULTURE AND HUSBANDRY ................................................................................................................. 94 3.2 GENETIC MANIPULATION METHODS .......................................................................................................... 94 4. OVARY ANALYSIS ....................................................................................................................... 101 4.1 COLLECTION AND FIXATION OF FEMALE OVARIES ............................................................................... 101 4.2 IMMUNOHISTOCHEMISTRY OF FIXED OVARIES .................................................................................... 101 5. EMBRYO ANALYSIS .................................................................................................................... 102 5.1 PREPARATTION OF EMBRYONIC CUTICLES ........................................................................................... 102 5.2 COLLECTION AND FIXATION OF EMBRYOS. ............................................................................................ 102 5.3 IN SITU HYBRIDIZATION IN FIXED EMBRYOS ......................................................................................... 103 5.4 DOUBLE FLUORESCENT IN SITU HYBRIDIZATION IN FIXED EMBRYOS .............................................. 105 5.5 IMMUNOHISTOCHEMISTRY IN FIXED EMBRYOS .................................................................................... 107 5.6 PREABSORBING OF PRIMARY ANTIBODIES ............................................................................................ 108 6.1 EXPRESSION AND PURIFICATION OF RECOMBINANT PROTEINS ........................................................ 109 6.2 PROTEIN-PROTEIN INTERACTION ANALYSES- GST PROTEIN PULLDOWN ASSAY .......................... 111 6.3. PROTEIN-DNA INTERACTIONS-EMSA ASSAY .................................................................................... 111 BIBLIOGRAPHY ............................................................................................................................... 115 SUMMARY IN SPANISH ................................................................................................................. 129 ABBREVIATIONS Abbreviations Abbreviation (genes) aos Atx bHLH bcd bwk cact cic Cy Da Dl dpERK Dpn DSP1 Dri EGF En Grk Gro gt H hb HMG Hox Hu IκB ind kni kr MAPK mirr nos Ndl NF-κB NTF-1 osk pip rho RTK Sb sna sim sog Spz sxl tld tll trk tsl Tor Full name argos Ataxin basic Helix Loop Helix bicoid bullwinkle cactus capicua Curly Daughterless Dorsal double phospho Extracellular Related Kinase Deadpan Dorsal Switch Protein 1 Dead-Ringer Epidermal Growth Factor Engrailed Gurken Groucho giant Hairy hunchback High Mobility Group Homeobox Humeral Inhibitor of kappa beta intermediate neuroblasts defective knirps kruppel Mitogen Activated Protein Kinase mirror nanos Nudel Nuclear Factor Kappa Beta Nuclear Transcription Factor 1 oskar pipe rhomboid Receptor Tyrosine Kinase Stubble snail single-minded short gastrulation Spätzle sex-lethal tolloid tailless trunk torso-like Torso 
 1 Abbreviations Tub twi zen Other abbreviations AP ana bp Cas9 CBS ChIP CRISPR CRM CTD DAPI DFS DNA dNTP dUTP DV Flp FRT GLC GST h His kb KO M mel Min ml mol mut μg NHEJ OD Pol PCR pse RHD RNA UAS UTR vir 
 Tubuline twist zerknüllt Full name anteroposterior annanasai base pairs Caspase 9 Cic binding sites Chromatin Immunoprecipitation Clustered Regulatory Interspaced Short Palindromic Repeats cis-Regulatory Module C-terminal domain 4’,6-diamidino-2-phenylindole Dominant Female Sterile Deoxyribonucleic acid deoxyribonucleotide triphosphate deoxyuridine-5'-triphosphate dorsoventral Flipase Flipase Recognition Target Germ-line clones Glutathionine S transferase hours Histidine kilobases knock-out molar melanogaster minutes Milliliters moles mutant micrograms Non homologous end joining Oligoderndroglioma Polymerase Polymerase Chain reaction pseudooscura Rel Homology Domain Ribonucleic acid Upstream Activator Sequence Untranslated Region virilis 2 INTRODUCTION Introduction Animal development is a complex process during which genetically identical cells, originating from the division of a single diploid cell, the egg, differentiate and assemble into highly organized tissues and organs. In bilaterian animals, development also involves the formation of two asymmetry axes: the anterior to posterior (AP) and dorsal to ventral (DV). Morphogenesis and determination of the body axes are accomplished through the precise and coordinated function of molecules that constantly receive, transmit and interpret physical and chemical signals, inducing the adequate changes in the cell. Many of these changes are done at the level of the genetic material and result in its differential activation or suppression. A powerful and versatile model used to understand gene regulation is the fruit fly, Drosophila melanogaster, which due to its rapid and robust development has been established as the model of choice for genetic studies. In this thesis, we focus on the molecular mechanisms underlying the patterning of the early Drosophila embryo, specifically the establishment of its DV body axis, a process which has traditionally been of interest in the field of developmental biology. Breakage of the embryo symmetry requires a number of factors that act soon after fertilization, but also relies on information provided by the mother. For this reason, in the next sections we will introduce both the processes of embryogenesis and oogenesis. 1. Early development of Drosophila Brief overview of embryogenesis Embryogenesis of Drosophila is a process that takes roughly one day and is divided into 17 stages (figure 1). Once the fertilized Drosophila egg is laid, it undergoes a series of rapid, synchronous mitotic divisions. During the first cleavages, no cytokinesis or cellular membrane formation occur. Rather, the dividing nuclei are localized inside a large yolk, which serves as a common cytoplasm, and surrounded by a unique membrane, the plasma membrane. This structure, the syncytial blastoderm, is the form as which the embryo exists during the first three hours of development. After the first nine divisions, which take place in the center of the common yolk, the nuclei migrate to the periphery while they continue to divide, at lower rates. A group of them migrates to the posterior region and adopts a distinct fate, forming the precursors of the future germline.
 3 Introduction In the meanwhile, the rest of the nuclei, that have reached the periphery of the syncytium divide for a few rounds more, with each cycle taking about 25 minutes to be completed. During nuclear cleavage 14 (stage 5), the oocyte plasma membrane folds inwards, between the nuclei, dividing them to individual cells. At this point, the embryo enters the cellular blastoderm stage, during which mitotic divisions slow down and the genome is activated. Upon cellularization, morphogenesis also starts to occur: the embryo begins to gastrulate, germ layers are formed, body segments appear and differentiation begins to become obvious along the two orthogonal axes of the embryo. After various events of migration, invagination and cell group formation, the embryo is ready to hatch and enter the first stage of the larval phase. Since early embryogenesis, two assymetry axes, which are maintained until the adult life, are distinguished: the anterior to posterior (AP) and the dorsal to ventral (DV). Finally, the extremes of the embryo adopt a different fate, the terminal fate. In the next sections we will discuss how these asymmetries are established in the early embryo. Figure 1. Overview of Drosophila embryogenesis stages. Adapted from FlyMove and Hales et al., 2015.
 4 Introduction 1.1 AP patterning during embryogenesis The head, thorax and abdomen are formed along the embryo AP axis. The thorax and abdomen are segmented to 8 regions, some of which will give rise to structures such as the wings and the legs. The differentiation of these regions depends on complementary protein gradients, which are provided maternally as mRNA´s and translated soon after fertilization; Bicoid (Bcd) anteriorly, and the Hunchback (Hb) and Nos (Nos) proteins posteriorly, turn on the expression of zygotic genes. Of these, the first to be expressed are the gap genes. The boundaries of giant (gt), hunchback (hb), kruppel (Kr) and knirps (kni) determine where the future embryo sections (head, abdomen etc) will be, and mutations in them cause gaps in the normal body plan (Nüsslein-Volhard and Wieschaus, 1980). During nuclear cycles 11-14 gap genes mutually regulate each other and then, they turn on the expression of a second group, the pair-rule genes (Clyde et al., 2003; Nusslein-Volhard et al., 1987; Rivera-Pomar and Jäckle, 1996). These are expressed as intercalating stripes along the AP axis and their products define the parasegments of the embryo. Similarly to gap genes, the products of pair-rule genes control the expression of other pair-rule genes, as well as the next zygotic genes to be activated: segment polarity genes and homeobox genes. These two last classes of genes pattern the different segments and give rise to the epidermis and anatomical structures such as wings, legs and antennae of the developing individual (Doe et al., 1988; Nüsslein-Volhard and Wieschaus, 1980; Patel et al., 1989, Morata, 1993). 1.2 Terminal patterning during embryogenesis Cephalic structures at the anterior extreme, and features such as the tail at the posterior extreme, are patterned by the terminal system (revised in Furriols and Casanova, 2003). The key element of this system is Torso, a Tyrosine Kinase receptor, which is present ubiquitously on the plasma membrane of the embryo, but turned on only at the poles, due to the presence of the Torso-like (tsl) protein, which is responsible for the cleavage/activation of the Trunk (Trk) ligand. Activated Torso initiates an intrecellular Ras/Raf/MAPK signalling cascade, which culminates in the phosphorylation of nuclear factors that regulate gene expression. 
 5 Introduction One of the factors active MAPK phosphorylates and inactivates is the Capicua (Cic) repressor. As a result, Cic targets tailless (tll) and huckebein (hkb) are relieved and expressed at the embryo termini. Expression of tll and hkb excludes the expression of gap genes such as kni, giving a terminal fate to these nuclei, and, conversely, exclusion of tll and hkb from the central regions by active Cic permits patterning of the trunk. (Jimenez et al., 2000). Furthermore, as we will see later on, the localized activation of the Torso RTK pathway at the embryo poles also regulates the expression of genes along the DV axis. Figure 2. Terminal patterning during embryogenesis. The localized cleavage/ activation of the Trk ligand leads to the localized activation of the Torso receptor. Downstream Torso, the intracellular Ras/Raf/MAPK pathway phosphorylates/inactivates Cic, relieving the repression of the terminal genes tll and hkb. 1.3 DV patterning during embryogenesis Considering the dorsal to ventral (DV) axis, at the end of gastrulation the embryo is divided into three regions, the germ layers: the mesoderm, the endoderm and the ectoderm. The mesoderm, at the ventral side of the embryo will give rise to tissues such as muscles, gonads, fat bodies and the heart. The endoderm contains precursors of the intestinal tract. The ectoderm consists of the neurogenic ectoderm, containing neural precursors and the dorsal ectoderm that 
 6 Introduction gives rise to structures such as the trachea and the larval epidermis. Finally, an extraembryonic membrane on the dorsal side, the amnioserosa, will contribute to dorsal closure, a process needed to form the larvae epidermis. The mesoderm and ectoderm are differentiated during early embryogenesis, while the endoderm emanates later, from the folding of terminal regions. The main factor responsible for the onset of different gene expression programmes within the germ layers is Dorsal (Dl), a morphogen transcription factor that is present in the nuclei of the blastoderm embryo in a ventral to dorsal gradient (Roth, 1989; Roth, 2003; Moussian and Roth, 2005, Stathopoulos and Levine, 2002). As a definition, morphogens are molecules that can induce at least two different cell types at different concentrations (Ashe and Briscoe, 2006). As such, Dl establishes different tissues by inducing different target genes. At the ventral-most nuclei, high levels of Dl activate two main genes that define the mesoderm: twi, which encodes a basic helix-loop-helix (bHLH) factor (Pan 1991, Thisse 1991), and snail (sna), which encodes a zinc-finger transcriptional repressor, that demarcates the boundaries between the mesoderm and the overlying tissues (Gonzalez-Crespo and Levine, 1993; Kosman et al., 1991; Stathopoulos et al., 2002). Immediately above the sharp borders of sna, a transitional region is formed by the expression of single-minded (sim), a gene activated by Dl and repressed by Sna. This transitional region, the mesectoderm, is a single stripe of cells that separates the mesoderm from the ectoderm, forms the ventral midline of the animal and gives rise to neurons and glial cells. Within the ectoderm, different regions are established by intermediate Dl levels: at its ventral-most region, the lateral neuroectoderm, rhomboid (rho) and ventral neuroblast defective (vnd) are expressed as a lateral stripe of 8-10 nuclei and specify the future peripheral nervous system. Above them, intermediate neuroblast defective (ind) specifies the intermediate neuroectoderm, that will give rise to the neuroblasts (Ip et al., 1992a; Jiménez et al., 1995; Weiss et al., 1998). The upper-most part of the ectoderm, the dorsal ectoderm, that will surround the neuroectoderm during gastrulation, is specified by the broad expression of genes such as short gastrulation (sog), which are activated by the lowest Dl levels. Finally, the low levels of Dl that activate the expression of sog also repress a 
 7 Introduction last set of genes, zerknüllt (zen), decapentaplegic (dpp) and tolloid (tld) restricting them to the dorsal most part of the developing embryo (Doyle et al., 1989; Ferguson and Anderson, 1992; Kirov et al., 1994; Rushlow and Levine, 1990). These specify the amnioserosa, an extraembryonic tissue, which is critical for germ-band extension and dorsal closure (Lacy and Hutson, 2016, Rushlow and Levine 1990). Following this first round of transcriptional regulation, zygotic genes that have been regulated by Dl, such as dpp and sog, form a second patterning network, that continues the cell differentiation process along the DV axis (Ashe et al., 2000; O’Connor et al., 2006). Figure 3. The dorsoventral network of embryogenesis. (A): Scheme of cross-section of an early embryo. Green colour indicated the presence of nuclear Dl. (B) : cross section of immunostaining of early stage embryos with an anti-Dl antibody (C): cross section of in situ hybridization of early stage embryos, showing the expression of Dl targets along the DV axis (Adapted from Reeves and Stathopoulos, 2009). (D): Schematic representation of the formation of ventral and dorsal tissues during embryogenesis.
 8 Introduction 2. Maternal control of developmentShaping of the Drosophila body begins even before the egg is fertilized and is based on information provided by the mother. Maternal control dominates development for a prolonged period, during several cell cleavage cycles after fertilization, and in some cases it persists even after the zygotic genome has been activated. Four systems of maternal control exist in Drosophila (figure 4): the anterior and posterior systems instruct the formation of the perspective head/thorax and abdomen/germline respectively. The dorsoventral system determines the formation of structures along the DV axis. Finally, the terminal system controls both the anterior (acron) and posterior (telson) extremes of the embryo. Although each of these systems regulates a well-defined region of the future animal, crosstalk between them exists. For example, the anterior extremity is determined both by the anterior and terminal systems, while the DV axis is patterned under the control of both dorsoventral and terminal maternal systems. In all of these systems, information held by the mother’s germline or somatic cells is provided to its offspring, either as proteins that act during oogenesis, either as mRNA that is translated once fertilization occurs. Another mechanism of maternal control is referred to as ´´late induction´´(LeMosy et al., 2003; Roth, 1998). In this, signalling events that occur in the egg act as positional cues that are interpreted by the embryo, which translates them into distinct cell fates after fertilization. Figure 4. Maternal control of Drosophila development. Four maternal systems act in the ovary (A) and transmit the necessary information to the embryo (B) in order to pattern the different body regions of the future adult (C). 9 Introduction 3. Axis specification during oogenesis One of the functions of the Drosophila maternal systems is to provide the bases for the establishment of the embryo body axes. Therefore, to fully understand the molecular events that polarize the embryo, we will take a step back and provide a brief description of oogenesis and the symmetry breaking events occurring at this stage. Brief overview of oogenesis The development of the new Drosophila adult begins with the formation of the oocyte during oogenesis, a process that takes roughly one week, beginning with an undifferentiated cell that becomes the mature, ready for fertilization egg (reviewed in Bastock and Johnson, 2008). Oogenesis takes place in the female ovary, in a structure called ovariole, where eggs are produced and mature in an assembly line fashion. At the anterior end of the ovariole, the germarium, the stem cells exit and develop to oocytes as they move through the ovariole, so that each ovariole has 6-7 eggs at different stages of development. Initially, the germarium stem cells divide, giving rise to another stem cell and a daughter cell. The daughter cell undergoes 4 mitotic divisions with incomplete cytokinesis, resulting in the formation of a cluster of 16 cells that are interconnected by cytoplasmic bridges, known as ring canals. From these 16 cells, one will differentiate to become the oocyte, which is characterized by its unique ability to undergo meiotic division. The rest 15 cells of the syncytium adopt a polyploid fate and become the so-called nurse cells, that have the role of delivering maternal RNA´s, proteins, nutrients and organells through the ring canals to support the immature oocyte. The cluster of nurse cells and oocyte are surrounded by a somatic follicle cell epithelium and altogether this structure is called the egg chamber, and is the basic unit of Drosophila oogenesis. At stage 10 of oogenesis, the nurse cells are eliminated by apoptosis after they are contracted and dump their contents to the oocyte, as the nurse cell cytoplasms are fused to the oocyte cytoplasm (ooplasm). Then, the follicle cells migrate to enclose the oocyte and secrete the vitelline membrane and the eggshell, known as chorion, that both protect the mature egg. 10 Introduction Figure 5. Drosophila oogenesis. (A): Drawing of the female ovary structure. The ovariole (coloured blue), contains eggs at different developmental stages. (B): Drawing of the different stages of the oocyte, starting from the stem cell at the germarium (left) until the mature oocyte (right). (C): Immunocytochemistry of egg chambers at different stages of development. Nuclei are stained with DAPI (Blue) and cell membranes with LaminC (Red). nc; nurse cells, oo; oocyte, fc; follicle cells. Adapted from Ables, 2015. If we were to divide the egg in orthogonal sections, we would see that, early, since it leaves the germarium, it is not symmetrical, in the sense that distinct subpopulations of follicle cells exist. The asymmetry (polarization) of the egg chamber is a consequence of cell signalling between the germline oocyte and the somatic follicle cells, and is required for the eggshell formation and determination of the embryo body axes. Egg chamber polarization requires the correct localization of the gurken (grk), bicoid (bcd), oskar (osk) and nanos (nos) mRNA’s, which are synthesized in the nurse cells and transported to the oocyte cortex (González-Reyes and St. Johnson, 1998; Beclaska and Clavis, 2009). Then, depending on signalling events, they are deposited at different positions in the egg chamber, where they direct the overlying follicle cells to adopt distinct fates. Two axes of asymmetry exist in the egg chamber and correspond to the two body axes that the individual will have as an adult: the anterior to posterior (AP) and the dorsal to ventral (DV). The AP polarity is the first to be determined during early oogenesis with the anterior localization of bcd and posterior localization of nos and osk mRNA´s. Then, during mid-oogenesis, the egg begins to present DV asymmetry and the terminal maternal system is also turned on. Since DV polarization of the embryo depends on both dorsoventral and terminal maternal control, we will focus on these two processes. 
 11 Introduction 3.1.1 DV polarity establishment during oogenesisDuring mid-oogenesis (stage 8), the oocyte, which is localized at the dorsal-anterior region of the egg chamber, secretes the Gurken (Grk) protein, which acts a ligand for Torpedo/DER, a Tyrosine Kinase receptor, in the adjacent overlying follicle cells (Queenan et al., 1999; Peri et al., 1999; Stein and Stevens, 2014). Although Grk is secreted at the dorsal-most region, where the oocyte is localized, it acts as a long-range morphogen and activates Torpedo in a graded manner (Stein and Stevens, 2014). This graded activation of Torpedo initiates two branches of the EGF (Epidermal Growth Factor) signalling cascade, which has a double mission: the formation of the dorsal structures of the eggshell (for example the dorsal appendages) and the generation of the distinct subpopulations of dorsal and ventral follicle cells. This latter asymmetry between the follicle cells is the basis for the dorsal to ventral polarization of the future embryo. DV patterning of the egg chamber Upon the activation of Torpedo, the canonical Ras/Raf/MAPK pathway is initiated in the dorsal-most region of the follicular epithelium (Peri et al., 1999). Two main downstream factors respond to this EGF signal and define the expression boundaries of Pipe (Pip), the key factor that transmits the DV asymmetry information from the egg chamber to the embryo. On the one hand, Mirror (Mirr), a homeodomain factor, is activated by the EGF signal, forming a gradient with peak levels at the dorsal anterior follicle cells and lowest levels at the lateral follicle cells (Jordan et al., 2000). In this dorsal-lateral domain, Mirr acts as a cell-autonomous repressor of pipe, by directly binding to a specific region in its proximal promoter (Andreu et al., 2012a; Fuchs et al., 2012). On the other hand, the HMG-box (High Mobility Group-box) Cic, downregulated by the EGF signal, forms a ventral to lateral gradient, complementary to the EGF activation domain. At the ventral most follicle cells, Cic acts as a repressor of Mirr, relieving the suppression of pipe (Andreu et al., 2012b; Goff et al., 2001; Technau et al., 2012). A scheme of the DV polarity network in the egg chamber can be seen in figure 6. 12 Introduction As a combinatorial output of the above regulatory network, during stages 9-10 of oogenesis, pipe is expressed in a broad ventral domain at the follicular epithelium of the egg chamber. The product of pipe is a sulfonyltransferase enzyme that is localized in the Golgi apparatus and modifies certain components of the vitelline membrane (Zhang et al., 2009; Zhu et al., 2007). These modified components are then secreted to the perivitelline space that surrounds the egg and initiate a serine protease cascade, which transmits the ventralizing cue to the 
 13 Figure 6. DV polarization of the egg chamber. During stages 9-10 of oogenesis, the signalling cascade initiated at the dorsal-anterior region of the egg chamber by Grk results in the localized ventral expression of the Pipe protein. (A): Scheme showing the DV polarization cascade of the egg chamber. (B-D): mRNA transcripts of cic, mirr and pip in egg stage 9-10 egg chambers of wild-type females.(B´-D´) mRNA transcripts of cic, mirr and pip in stage 9-10 egg chambers of mutant cic ( cicfetE11/cicfetU6) females. Introduction future embryo (Stein and Stevens, 2014; Zhang et al., 2009). Briefly, the output of this cascade is the processing of the inactive Spätzle (Spz) ligand to its active form. Upon fertilization, active Spz binds to the Toll receptor, which is located on the embryo plasma membrane and initiates an intracellular signalling pathway. Spz is cleaved in a ventral to dorsal graded fashion, following the domain of Pipe expression and activates Toll in the same graded manner (Roth, 1994). In turn, the Toll signalling pathway culminates with the phosphorylation and degradation of an IκB homologue protein, Cactus (Cact) (Belvin et al., 1995; Roth et al., 1991), which retains the Dl in the embryo cytoplasm. As a result of the graded Toll activation and Cact degradation, Dl enters the embryo nuclei in a ventral to dorsal gradient (Anderson et al., 1985; Rushlow et al., 1989; Steward et al., 1988). It is this Dl gradient that initiates the process of DV patterning in the early embryo. 14 Figure 7. Pipe transmits the ventralizing signal during the egg to embryo transition. Pipe induces a serine protease cascade in the perivitelline space which ultimately result in activation and binding of Spz to the Toll receptor. Toll signalling culminates with phosphorylation and degradation of Cact and release of Dl from the cytoplasm. Introduction 3.2.2 Terminal patterning during oogenesisDuring mid-oogenesis, the terminal maternal system is also turned on. A complex of germ-line expressed proteins mediate the accumulation of the Torso-like (Tsl) protein at the anterior and posterior follicle cells (Furriols et al., 2007; Mineo et al., 2015; Roch et al., 2002, Stevens et al., 2003). Through a not quite understood mechanism, which involves cleavage of the Trunk (Trk) ligand to its active form, a positional cue is generated at the extremities of the egg chamber (Casali and Casanova, 2001; Henstridge et al., 2014; Johnson et al., 2015). This cue is transferred to the fertilized embryo and results in the localized activation of the Torso RTK receptor, which is maternally provided as a ubiquitous mRNA and translated along the whole circumfence of the plasma membrane of the embryo after egg activation. Likewise the localized activation of Toll by Spz determines the ventral fate, selective activation of the Torso pathway by Trk at the embryo poles confers a terminal fate (Casanova, 1995; Casanova and Struhl, 1989; Perrimon et al., 1995; Schupback and Wierchaus, 1986). Upon fertilization, the anterior structures such as the head skeleton and labium are patterned based on information from the anterior and terminal system. At the other end of the embryo, structures such as the abdominal segment 8, the posterior spiracles and the and the Filzkörper filaments are patterned based on the information provided by the posterior and terminal systems. The terminal system also participates in DV patterning of the embryo. The same protein complex that is responsible for the restricted translocalization of Tsl, is also involved in the dorsoventral maternal system. Specifically, it ensures the proper activity of the Nudel (Ndl) protein, which forms part of the serine protease cascade downstream Pipe, that initiates the Dl localization signal (Mineo et al., 2017). Furthermore, after fertilization, activation of the Torso pathway interferes with the expression of genes regulated by Dl. 15 Introduction 4. Dorsal, the main nuclear effector in DV polarityAs we have seen above, the key factor in embryonic DV patterning is Dl, a transcription factor that belongs to the Rel family, which includes the mammalian NF-κB (nuclear factor kappa-light chain enhancer of B activated cells) proteins. Specifically, it is homologous to the vertebrate proto-oncogene c-rel and its corresponding viral oncogene v-rel (Steward et al., 1989), which are involved in various cellular processes, including inflammatory response and may cause certain cancer types, such as lymphomas, when they are mutated or amplified. Dl is maternally provided to the Drosophila embryo as a uniformly expressed mRNA that is translated approximately 90 minutes after fertilization. Soon after, activation of the Toll pathway triggers its selective nuclear import, by degrading Cact. As the Toll receptor is activated in a graded fashion, Dl enters the nuclei in a gradient as well, with peak concentrations at the ventral most nuclei of the embryo. Recent studies have shown that the nuclear-cytoplasmic localization of Dl is not stable, but results from a highly dynamic equilibrium that involves import, increase and drop of nuclear levels of Dl within each mitotic cleavage (DeLotto et al., 2007; Kanodia et al., 2009). As we move upwards along the presumptive DV axis, the rate of nuclear import decreases and levels gradually drop, with the dorsal most region having essentially cytoplasmic Dl. 4.1 Structure and functions of DlAs other members of the Rel family, Dl has an N-terminal region of 300 aminoacids, the so-called Rel homology domain (RHD), which is responsible for DNA binding, dimerization and binding to inhibitory regulators, such as Cact (Govind et al., 1996; Isoda et al., 1992; Tatei and Levine, 1995; Tony Ip et al., 1991). On the other hand, its C-terminal domain (CTD), which contains characteristic polyQ stretches, is responsible for the transcriptional activities of Dl, but also for selective nuclear import (Govind et al., 1996; Isoda et al., 1992; Jia et al., 2002; Rushlow et al., 1989). Dl is known to interact with various partner proteins, both activators and repressors, to accomplish transcriptional regulation. These interactions are mediated by both the RHD and carboxyterminal regions (Flores-Saaib and Courey, 2000; Shirokawa and Courey, 1997). Finally, it has been 
 16 Introduction suggested that inter-protein interactions between the RHD and the CTD also exist and are essential for regulating the targets (Jia et al., 2002). The structural domains of Dl are illustrated in figure 8. Considering its functions, Dl is the key regulator of a number of zygotic genes which, in turn, determine distinct cell fates along the DV axis of the developing embryo. However, these are only some of the genes regulated by Dl. Recent genome-wide studies have identified approximately 100 targets, including proteins, such as N-Cadherin, and micro-RNAs (Biemar et al., 2006; Markstein et al., 2002; Stathopoulos et al., 2002). Furthermore, whole-genome ChIP-chip analysis has shown that Dl also fine-tunes the expression of the AP axis, such as kni, and regulates genes encoding signal transduction components (Zeitlinger et al., 2007). Similarly to its vertebrate analogues, Dl also has a role in immune response. Upon bacterial challenge in the larvae, Dl, that is localized in the cytoplasm of the fat bodies, responds to Toll signalling and translocates to the nuclei (Lemaitre et al., 1996; Reichhart et al., 1993). Dl homodimers induce the expression of the antifungal peptide drosomycin, while Dorsal-Relish heterodimers activate the expression of the defensin peptide which protects against bacteria, fungi and some viruses (Ip et al., 1993). Figure 8. The functional domains of the Dl protein. The first 380 residues are the conserved Rel domain, in which DNA binding, dimerization and binding to the Cact inhibitor motifs are localized. The C-terminal domain is responsible for transcriptional activities and cytoplasmic retention. The eh1-motif has been proposed to interact with the Gro corepressor. 4.2 Transcriptional regulation by DlAs already mentioned, Dl is a DNA-binding transcription factor, which regulates a large set of target genes in different regions of the developing embryo. 
 17 Introduction To achieve the multitude of responses required for the formation of the DV axis, Dl acts both in a concentration and context-dependent manner. On the one hand, Dl, due to its graded presence in the nuclei, is considered a morphogen molecule and induces different genes at different concentrations. On the other hand, independently of changes in its concentration, Dl has a dual function: depending on the target gene, it either functions as an activator or as a repressor. 4.3 Dl-mediated activation Intrinsically, Dl functions as a transcriptional activator and this occurs along the largest portion of the DV axis. Different concentrations of Dl trigger the activation of different genes which, in turn, specify distinct regions, the mesoderm and ectoderm. However, cell fates along the DV axis do not always coincide with the steep changes in Dl concentration, nor can transcriptional outputs can always be explained by differential strength of its binding to regulatory sequences (Liberman et al., 2009; Stathopoulos et al., 2002). Instead, it is well accepted that Dl is only one element of a complex network that integrates broadly distributed activators, localized repressors and modulating inputs of numerous signalling pathways (Liberman et al., 2009; Stathopoulos and Levine, 2005; von Ohlen and Doe, 2000). To turn on its targets, Dl synergizes with activator proteins belonging mainly to the bHLH (basic Helix-Loop-Helix) family, namely Daughterless (Da), T4 and, most importantly Twi, which Dl itself turns on (Gonzalez-Crespo and Levine, 1993; Ip et al., 1992a; Kosman et al., 1991). In regions such as the mesoderm, where Dl is abundant in the nuclei, its targets, nominated type I, are turned on by the sum of the independent activities of Dl and its cofactors, while no cooperative DNA binding is observed. The low levels of activation that Dl triggers are potentiated by other activators that contribute to the additive recruitment of the basal transcriptional machinery (Hong et al., 2008; Shirokawa and Courey, 1997). This is reflected in the cis-regulatory modules (CRM’s) of type I genes, which contain low-affinity sites for Dl and somehow unlinked bHLH binding sites, whose composition, arrangement and orientation is flexible. Actually, even deleting the bHLH sites still permits low expression levels of the mesoderm targets (Jiang and Levine, 1993a; Szymanski and Levine, 1995). In the neurogenic ectoderm, where the slope of the Dl gradient is steep and its levels 
 18 Introduction are intermediate, it activates targets nominated type II. To turn on type II genes, Dl synergizes with proteins of the bHLH family, and other activators as well. In this tissue, the organization of the target CRM’s is slightly different; although they still maintain a grade of flexibility in their organization, the need for proximity and correct orientation of the binding sites are indicative of the cooperative interactions between Dl and cofactors. For example, the CRM of rho contains various arrays of high and low affinity Dl sites closely linked to bHLH sites, ensuring that the intermediate levels of Dl occupy the promoter sufficiently to activate it in a broad domain (Szymanski and Levine, 1995). A last group of genes activated are type III genes, which are turned on by low levels of Dl at the dorsal ectoderm. Activation of type III genes, such as sog, is achieved through high-affinity Dl sites, which are closely linked to sites for Zelda, an ubiquitously expressed activator which synergizes cooperatively with Dl. 4.4 Dl-mediated repression Remarkably, the same low levels of Dl that activate sog, also repress a different set of genes within the type III group, which include zen, tld and dpp. As a result, these are restricted to the dorsal-most nuclei of the blastoderm embryo, where they will specify the amnioserosa (Ray 1991). At the absence of Dl or when Dl binding sites are mutated, expansion of dorsal-specific genes towards ventral regions of the embryo is observed (Jiang et al., 199b, Kirov et al., 1993; Kirov et al., 1994). How Dl switches from its intrinsic, activator function to a repressor mode in order to restrict genes from the ectoderm has been a long-standing question in the field of Drosophila developmental genetics. Extensive studies on the regulation of dorsal-specific type III genes, have shown that their repression depends on short elements found in their enhancers, nominated VRE (Ventral Repression Elements) or VRR (Ventral Repression Regions). These modules of 600-800 bp are sufficient to reproduce the endogenous pattern of the genes and contain all the sequences necessary to drive ventral repression (Doyle et al., 1989; Jiang et al., 1993a; Kirov et al., 1993; Kirov et al., 1994; Valentine et al., 1998). VRE´s include high affinity binding sites for Dl and binding sites for general activators, such as Zelda, which, in contrast to the sog CRM, are somehow separeted from the Dl sites (Kirov et al., 1993). Instead, in the VRE´s, the Dl sites are closely neighboured by characteristic 
 19 Introduction sequences rich in A and T bases (hereafter reffered to as AT-rich sites), that have been suggested to serve as binding sites for one or more repressors that form a multiprotein complex with Dl and contribute to the switch of its activity. Purification of embryonic and cell culture extracts have revealed a number of AT-rich associated repressors, including the Cut/Dead-ringer (also named Retained) complex, NTF-1 (Nuclear Transcription Factor-1) and the Dorsal Switch Protein (DSP1) (Brickman et al., 1999; Valentine et al., 1998; Huang et al., 1995; Ip, 1995). However, phenotypes caused by mutant alleles of these repressors are not very strong, especially during stages prior to cellularization. For example, Dri effects are prominent after stage 5, while Cut has been demonstrated to have a strictly zygotic function (Valentine et al, 1998). Therefore, it has been accepted that additional repressors function through the AT-rich sites. One such repressor has been proposed to be Cic, whose hypomorphic mutations result in ventral expansion of the zen transcripts in early embryos (Jimenez et al., 2000). However, the posterior characterization of its function in the DV polarization of the egg chamber, which affects the formation of the Dl gradient itself, has obscured the implication of Cic in Dl-mediated repression. Another essential element during Dl-mediated repression is the non-DNA binding Groucho (Gro) corepressor, a protein that is ubiquitously expressed in the early embryo and is involved in various developmental processes, such as sex determination, embryo segmentation, neurogenesis and bristle formation (Jennings and Ish-Horowicz, 2008; Paroush et al., 1994, Turki-Judeh and Courey, 2012). Gro belongs to the Enhancer-of-Split complex (E(spl)-C) and consists of five identifiable regions of which the Q rich N-terminal region and a C-terminal WD-repeat domain are conserved (Schrons 1992, Jennings et al., 2006, Buscarlet and Stifani, 2007). A scheme of the structural domains of Gro can be seen in figure 9. Gro is recruited by short peptide motifs of DNA-bound repressors, that fall into two main classes: the WRPW motif and its variations, and the Engrailed motif (eh1) which has the consensus sequence FSISNILS. Although these two motifs adopt different conformations, they both contact overlapping sites in the pore of the β-propeller formed by the Gro WD domain (Buscarlet and Stifani, 2007). Once Gro is recruited by repressors bound to the promoters of target genes, it interferes with 
 20 Introduction transcription by inhibiting the RNA Poll complex recruitment and chromatin remodeling. Null mutants of Gro cause ventral derepression of zen and, interestingly, a motif resembling the eh1 peptide, with the sequence PTLSNLLS (hereafter eh1-like), has been identified in Dl and found to interact with Gro in vitro (Dubnicoff et al., 1997; Flores-Saaib et al., 2001). However, this motif lacks the first phenylalaline residue, which is critical for Gro recruitment, making the interaction weak (Jimenez et al., 1999, Flores-Saaib et al., 2001). It has been proposed that interactions between AT-rich bound repressors and Dl might expose a cryptic Gro-recruiting motif (Flores-Saaib et al., 2001; Hong et al., 2008). Other theories postulate that Dl and the AT-repressors function as a high affinity platform that collectively recruits Gro (Dubnicoff et al., 1997; Ratnaparkhi et al., 2006). Figure 9. The structural domains of the Gro corepressor. The WD domain interacts with recruiting proteins such as Hairy and Engrailed. The SP domain is phosphorylated, resulting in Gro inactivation. The polyQ stretch is responsible for oligomerization, repression and protein-protein interaction. Adapted from Buscarlet and Stifani, 2007. A final aspect of Dl-mediated repression is that it receives inputs from the terminal system. Activation of the Torso RTK pathway at the termini of the embryo modulates the repressor activities of Dl and, consequently, prevents the adoption of a ventral fate at the terminal regions (Casanova, 1991; Rusch and Levine, 1994). Due to the activation of Torso, the expression domains of dorsal-specific type III genes (zen, tld and dpp) at the terminal parts of the embryo extend to ventral nuclei where they would otherwise be repressed by Dl. Mutations that impair Torso activity result in repression of zen at the embryo poles, while in gain-of-function Torso mutants zen expands towards the central regions, overcoming the presence of Dl. This expansion is similar to derepression of zen in dl mutants (Figure 10). The mechanism through which the Torso pathway exerts this anti-repression effect is not fully understood. It has been proposed that a mediator of Torso is WntD, a 
 21 Introduction Drosophila Wnt homologue, which acts as a feedback inhibitor of Dl (Ganguly et al., 2005; Gordon et al., 2005). During stage 4 of embryogenesis, wntD is turned on in a ventral domain by Dl, but in the embryo trunk it is inhibited by Cic. Torso-mediated downregulation of Cic permits the expression of WntD, which reduces Dl nuclear levels at the poles (Helman et al., 2012). A different possibile mechanism proposed is that the activation of the Torso pathway disrupts interactions between Dl and its corepressors, or modulates one or more of the corepressors (Rusch and Levine, 1994). The corepressor downregulated by Torso could be the Cut/Dri complex, Gro, or Cic, in case it is indeed a member of the Dl repressosome. Figure 10. The Torso pathway has anti-repression effects on zen at the embryo poles. (A): mRNA transcripts of zen in wild-type embryos. Expression at the termini extends ventrally, despite the presence of Dl in the nuclei. (B): mRNA transcripts of zen in tor mutant embryos (torPM51). Expression is excluded from the poles in the absence of Tor activation. (C, D): mRNA transcripts of zen in dl mutant and tor dominant mutants respectively. Lack of Dl and constitutive activation of Tor both result in derepression of zen. (Adapted from Rusch and Levine, 1994). Despite the identification of the AT-rich sites and Gro as essential elements of Dl-mediated repression, the molecular mechanism of how it switches from an activator to a repressor is still not fully understood. Indeed, the fact that the Dl and AT-richsites in VRE’s are tightly organized and are not permissive on changes regarding their composition or phasing, suggests that a multiprotein complex (hereafter referred to as Dl represssome or Dl repressor complex) is formed. Following the logic of the regulatory code along the mesoderm and ectoderm, this arrangement also suggests that DNA cooperative interactions between Dl and the 
 22 Introduction bound cofactors accounts for at least one of the regulatory mechanisms (Valentine et al., 1998, Hong et al., 2008; Jiang et al., 1992). However, full characterization of the Dl repressosome and how it functions remains an open question. Figure 11. Activation and repression along the DV axis. (A): Nuclear levels of Dl, Twi, Sna and Zelda determine gene regulation across the DV axis. (B): high levels of Dl activate mesoderm targets such as twi and sna. Low affinity Dl and Twi sites are separated and act independently. (C): Intermediate Dl levels activate type II targets at the neuroectoderm (rho, vnd) through cooperative interactions with bHLH factors such as Twi. (D): Low Dl levels bind to high affinity sites and interact with Zelda to activate sog at the dorsal ectoderm. (E): The same low Dl levels that activate sog repress type III dorsal specific genes (zen, tld, dpp), restricting them to the presumptive amnioserosa. Repression requires high affinity Dl sites and closely linked AT-rich sites. Adapted from Reeves and Stathopoulos, 2009 and Hong et al., 2008). 23 Introduction 5. Capicua, a Dl corepressor candidate One of the factors suggested to contribute to the switch of Dl to a transcriptional repressor is Cic, which, as we have mentioned in previous sections, is involved in the egg chamber polarization and the embryonic terminal system. In the next sections, we will provide a more detailed description of its structural domains and known functions. 5.1 Structure and functional domains of Cic Cic is a nuclear factor that acts downstream the RTK Torso signalling pathway in Drosophila (Jiménez et al., 2000). It is an HMG-box (High Mobility Group) protein expressed in various isoforms that fall into two main classes, the long (Cic-L) and the short (Cic-S), encoded by the bullwinkle (bwk) and cic locus respectively (Goff et al., 2001; Jiménez et al., 2000; Roch et al., 2002b). Of these, the best characterized are the Cic-S isoforms, which have various functions in different developmental stages. Considering their structure, all Cic isoforms share the HMG-box region, and the carboxyterminal C1 and C2 domains. The HMG-box accounts for the DNA-binding domain, characteristic of the family, while C1 domain is involved in repression of Cic target genes and has been recently shown to play an active role in DNA-binding (Forés et al., 2017; Jiménez et al., 2000). The C2 domain is a response element to the RTK pathways, serving as a docking site for active MAPK (rolled in Drosophila)(Astigarraga et al., 2007). Apart from these shared regions, at least some of the Cic-S isoforms have an aminoterminal motif nominated N2, which, is associates to the Gro corepressor in the context of embryogenesis (Forés et al., 2015). The Cic-L proteins have a specific aminoterminal N1 domain with an unknown function. Figure 12. Structure of Drosophila Cic proteins. Two main isoforms, Cic-S and Cic-L are expressed and have shared and isoform-specific domains.
 24 Introduction 5.2 Human homologues of Cic and their functions In mammals, a homologue of Cic that shows high conservation of the HMG and C1 domains exists. Although the molecular mechanism of its function is not fully understood, it is involved in physiological processes such as lung alveolarization and bile acid homeostasis in the liver (Kim et al., 2015; Lee et al., 2011). Moreover, Cic is involved in a series of pathological conditions. In various malignancies, such as breast cancer, oligodendroglioma (OD) brain tumours and Ewing-like sarcomas, mutations of Cic have been observed (Bettegowda et al., 2011; Kawamura-Saito et al., 2006). These include missense mutations that inactivate the HMG-box and the C1 domain, as well as a chromosomal translocation that results in the fusion with the DUX4 protein. This fusion protein activates a set of oncogenic driver genes, which suggests that, normally, Cic has a role as a tumour suppressor (Kawamura-Saito et al., 2006; Okimoto et al., 2016). Finally, human Cic associates with the Ataxin (ATX) cofactor and mutations that interfere with this interaction have been associated with neurotoxicity observed in spinocerebellar ataxia type 1 (Lam et al., 2006; Lim et al., 2008). All together, these examples of the biomedical functions of Cic, underline the importance of further elucidating its functions. 5.3 Mechanism of Cic function in Drosophila The molecular mechanism through which Cic functions in Drosophila has been well characterized for the Cic-S proteins. These isoforms recognize the canonical, octameric T(G/C)AATG(A/G)A site in the CRM´s of target genes and contact it through the HMG-box and C1 motifs. Once it is bound to DNA, Cic-S recruits Gro to repress its targets. Although it does not contain a peptide is similar to either WRPW or eh1 motifs, genetic studies have shown that it uses the N2 motif and possibly contacts the Gro β-propeller, adopting a different conformation than other Gro-associated proteins (Ajuria et al., 2011; Forés et al., 2015; Forés et al., 2017a). Notably, Gro-dependent Cic repression only occurs during embryonic patterning, it is therefore conceivable that in other tissues it functions with other, still unidentified corepressors to restrict its targets. 
 25 Introduction 5.4 Regulation and functions of Cic during Drosophila development The first function of Cic that was characterized in Drosophila is patterning of the embryo AP axis. At the blastoderm stage, a 1400 aminoacid Cic-S protein that includes the N2 motif is the predominant isoform, which is maternally provided as an ubiquitously expressed mRNA (Jiménez et al., 2000). Immunostaining of embryos with an anti-Cic antibody shows that the protein is distributed at the nuclei of the presumptive trunk region, but gradually drops at the anterior and posterior region (see figure 13). This drop of Cic levels is complementary to the graded activation of the Torso receptor. The intracellular signal transduction of activated Torso culminates in the phosphorylation/activation of MAPK, which binds to the C2 motif of Cic and phosphorylates it at yet unknown residues. Indeed, mutant forms of Cic lacking the C2 motif are expressed in the terminal nuclei of the embryo (Astigarraga et al., 2007, figure 13). The events downstream Cic phosphorylation are still not well understood. It has been suggested that upon activation of the pathway, nuclear-cytoplasmic shuttling rates change, and Cic, spending more time in the cytoplasm, is eventually degraded (Grimm et al., 2012). However, posterior studies showed that ectopic activation of Torso did not result in ectopic degradation of Cic, which remains nuclear at the central regions of the embryo (de las Heras and Casanova, 2006). The final result is that Cic repressor activities over the terminal genes tll and hkb are relieved at the embryo poles, contributing to the formation of terminal structures. Actually, the name Capicua (from cap i cua, which means head and tail in Catalan) originated from the observation that embryos proceeding from mutant cic mothers, lacked any segments, and only consisted of anterior (head) and posterior (tail) rests, due to the expansion of terminal genes at the expense of central gap genes. Cic is also involved DV patterning of the embryo. As we have seen above, it is a key element downstream EGFR signalling during the establishment of the ovary DV polarity, and, thus, the generation of the positional cue that will determine the dorsal and ventral cell fates in the embryo. Cic downregulation by the EGFR pathway also involves C2 motif docking and Cic phosphorylation. However, in contrast to the embryo, phosphorylation results in translocation of Cic to the 
 26 Introduction Figure 13. Regulation and functions of Cic in embryonic AP patterning. (A): Immunostaining of a blastoderm showing the nuclear distribution of Cic. Schematic representation of regulation of Cic in the embryo. Activation of the Torso pathway at the poles results in activated MAPK binding to the Cic C2 motif, leading to Cic downregulation and relief of its Gro-dependent repressor activities over tll and hkb. (B, B´): Immunostaining of embryos shows nuclear localization of Cic in wild-type embryos and embryos expressing CicΔC2 (MAPK insensitive). Note that CicΔC2 perists in the terminal nuclei. Cuticle phenotypes, tll and hkb expression in wild-type (C,F,I), cic1 (D,G,J), or embryos expressing the CicΔC2 protein under maternal control (E,H,K). Adapted from Astigarraga et al., 2007. cytoplasm, but not degradation (Astigarraga et al., 2007). In the ventral follicular epithelium, Cic represses mirr and this is required for the formation of the Dl gradient and for the formation of the dorsal appendages of the eggshell (Andreu et al., 2012b, Atkey et al., 2006). Furthermore, the gene products of the bwk locus have been shown to be necessary for the formation of the dorsal appendages (Dorman et al., 2004; Rittenhouse and Berg, 1995), suggesting that this feature could also require the action of Cic. 
 27 Introduction Figure 14. Cic responds to the EGFR signal during late oogenesis. Grk activates the EGFR cascade which culminatesin phosphorylation of Cic by dpERK. Ventrally restricted Cic participates in the formation of the Dl gradient. Cic contributes to DV patterning later during development as well. During stage 5 of embryogenesis, Cic represses the homeobox gene ind at the dorsal-most cells, by binding to a canonical TGAATGAA sequence in its promoter, identified as A-box. This repression is relieved at the ventrolateral domain of the embryo due to the activation of an RTK pathway. Activation of the Torpedo receptor initiates a EGF signal which downregulates Cic, permitting the expression of ind and the formation of the neuroectoderm (Ajuria et al., 2011, see figure 15). Figure 15. Cic is involved in the formation of the neuroectoderm during embryo cellularization. (A): representation of how Cic regulates the expression of ind, under the control of EGFR signalling. (B-B´´): Immunostainings of stage 5 embryos showing the domains of EGFR activation (dpERK), Cic expression (Cic-HA) and overlapping domains. (C-E): the expression of ind in wild-type, cic1 and cic1/cic2 mutant embryos. Adapted from Ajuria et al., 2011. 28 Introduction Later during development, contributes to wing patterning, by repressing wing vein specific genes (Roch et al., 2002). In the larval imaginal disc Cic is expressed in the presumptive wing pouch region (figure 16). There it represses genes such as argos (aos), by binding to typical octamerical binding sites in their regulatory sequences (Ajuria et al., 2011). This repression is relieved by an RTK signal. The EGFR pathway, which is active in the presumptive wing vein regions, downregulates Cic activity, allowing expression of wing patterning genes (Roch et al., 2002). Remarkably, this function of Cic is independent of the Gro corepressor (Forés et al., 2015). Other functions of Cic, that are beyond the scope of our study, exist, so they will be briefly mentioned. In the eye imaginal disc of the larva, Cic inhibits growth, and is under the control of Ras signalling (Tseng et al., 2007). Again under the control of EGFR downregulation, Cic inhibits stem cell proliferation in the midgut by repressing cell cycle genes, such as CyclinE and other effectors, such as Yan (Jin et al., 2015). Finally, a recent study has demonstrated that downregulation by Cic by a different kinase, Minibrain, again downstream ERK signalling, is necessary for the development of the wings, eyes and brain (Yang et al., 2016). Thus, Cic generally senses RTK signals, restricting organ growth and tissue patterning. Figure 16. Cic has a function in wing patterning. (A): Scheme of Cic-mediated downregulation of targets in the wing imaginal disc downstream EGFR signalling. (B): Immunostaining of larval wing imaginal disc showing distribution of Cic in the wing pouch. (C, D): wing vein pattern in wild-type and cic mutant flies respectively. L2-L5 indicate presumptive wing veins. Extra vein tissue and deformed wings are seen in the absence of Cic. Adapted from Roch et al., 2002. 29 Introduction 30 OBJECTIVES Objectives Patterning of the Drosophila embryo and establishment of its body axes has been a subject of studies through many years, however, despite the extensive knowledge gained, still much remains to be discovered. Roughly 30 years after the characterization of the Dl protein as the main effector of DV patterning, the exact mechanism through which it functions is not fully understood. Rather, studies constantly demonstrate that many pieces of the puzzle were missed until now. Our study aims to focus on the repression facet of DV patterning and unravel the molecular mechanisms underlying it. Specifically, the aims of this work are the following: 1. Investigate the possible role of Cic as a repressor during early embryo DV patterning and correlate it to its known function in the DV polarization of the egg chamber. 2. Study the possible relationship between Cic and Dl, as well as with other components of the Dl repressosome, such as the Gro corepressor. 3. Elucidate the mechanism of RTK-mediated modulation of Dl repressor activities over type III dorsal-specific genes and study the possible involvement of Cic as an RTK signaling sensor in this process. 31 Objectives 32 RESULTS Results 1. Functional analysis of Cic in the establishment of the DV axis 1. Cic mutations reveal a double function in the establishment of the DV axis Cic has a known role in establishing the DV polarity of the egg chamber, by responding to an EGFR signal and regulating the ventral expression of pipe in the follicular epithelium. Pipe is the determining factor for the formation of the Dl gradient, which, in turn, is the main effector of embryonic DV patterning. Therefore, Cic is a factor that participates in the Dl gradient formation and required to transmit the DV polarity positional cue to the embryo. In fact, mutations such as cicfetE11 and cicfetU6 have been shown to produce dorsalized eggs due to this requirement of Cic in the follicular epithelium (Goff et al., 2001). Previous to the characterization of the oogenic function of Cic in DV polarity, studies done in our group had suggested that it is involved in embryonic DV patterning, particularly in the repression of dorsal-specific (type III) genes. This hypothesis was based on the observation that embryos deriving from cic1 mutant females (hereafter called cic1 embryos), showed ventral derepression of zen, which is normally restricted to the dorsal-most nuclei by a complex formed by Dl. However, the characterization of the function of Cic in the follicle cells raised the question whether the effects seen on zen repression could be reflecting a deregulation of the egg chamber DV polarity, rather than a direct implication of Cic at the level of embryonic patterning. To address this question, we monitored pipe expression as a marker to measure the activity of Cic in egg chambers from cicfetE11, cicfetU6 and cic1 mutant females. To visualize pipe, we have used a previously described fusion transgene in which a minimal enhancer module of pipe, sufficient to recapitulate its endogenous pattern, drives the expression of lacZ (Andreu et al., 2012a), and immunostained stage 9-10 egg chambers. As Goff and colleagues had previously stated, the cicfetE11 and cicfetU6 mutations resulted in almost complete loss of the pipe domain in the ventral follicle cells, and this was reflected in the fertilized egg; embryos deriving from cicfetE11/cicfetU6 females, showed only a residual expression of twi at the poles, corresponding to the residual expression of pipe at the egg chamber, while the transcripts of zen, were expanded 
 33 Results towards the ventral region (see figure 17). This alteration of the Dl targets indicated that both its activator and repressor properties were impaired and was in accord with the lack of Pipe. Intriguingly, in trans-heterozygous cic1/cicfetU6 females, pipe expression was indistinguishable from the wild-type pattern and, as a consequence, the Dl gradient was correctly formed. This Dl gradient was functional considering its activator properties, as it could turn on the expression of twi along the ventral circumfence of the embryos. However, in embryos laid by these females, zen transcripts were expanded ventrally, indicating abnormality in the repressor functions of Dl. Therefore, the cic1 mutation selectively affects repression along the DV axis of the embryo, without impairing DV polarity of the egg chamber. To further confirm this observation, we used the Dominant Female Sterile technique (DFS) to generate cic1 mutant clones specifically in the female germline (GLC). We observed that in embryos deriving from cic1 clones, zen transcripts were clearly derepressed. This defect could only be attributed to the mutant maternal transcripts of Cic produced by the germline nurse cells and deposited to the embryo, since the follicular epithelium of these females was wild- type. 1.2 The main embryonic Cic isoform is involved in embryonic DV patterningOur finding that Cic has an embryonic function in DV patterning, which is independent from its oogenic function, led us to characterize the isoform that exerts this function. Upon cloning of Cic, the main isoform expressed in the embryo was described (Jiménez et al., 2000). This is a short class Cic, that contains the Gro-associated N2 motif and is maternally provided as an ubiquitous mRNA. As a first assay to test whether this same isoform is involved in repression along the DV axis, we have used full length and truncated transgenes of the embryonic Cic and measured their ability to restore the repression of zen in an otherwise cic mutant background (cic1/cicQ474X). Indeed, a full-length transgene of the embryonic Cic isoform (herafter named CicA), which we expressed under the endogenous promoter of Cic, therefore at physiological levels, was capable of repressing zen in the absence of endogenous Cic. 
 34 Results Figure 17. Different mutations of the Cic locus cause different phenotypes in DV patterning (A): Schematic representation of the Cic locus and mutant alleles. Coloured boxes represent exon regions and connecting lines intronic sequences. Triangles represent mutations caused by transposon elements while vertical lines point mutations. cicQ (an abbreviation for cicQ474X and cicfetU6 are missense mutations that cause truncation of the translated protein. (B-D´): Expression of the pipe-lacZ protein in stage 9-10 egg chambers: expression is visualized by staining with an anti-βgal antibody and nuclear staining of the respective ovarian nuclei with DAPI. (E-G): mRNA transcripts of zerknullt (zen) and twist (twi) in wild-type and cic mutant mbryos (stage late 4-early 5). (H-J): Cuticle phenotypes of wild-type and cic mutant embryos In overall, we observed that the cic1 mutation left unaffected the function of Cic in the ovarian follicle cells, however disrupted embryonic DV patterning, in particular the repression of the dorsal-specific gene zen. This observation indicated that Cic has two distinct functions in DV patterning, one at the stage of oogenesis embryogenesis. A smaller form of the protein containing essentially the conserved motifs N2, HMG-box, C2 and C1 (Cicmini) (see Astigarraga et al., 2007), was able to partially rescue zen repression in a cic1/cicQ474X background. However, a transgene lacking 60 aminoacids of the N-terminal region (CicΔN60) was uncapable of doing so (see figure 18). These findings suggest that the Cic isoform involved in the repression of genes along the DV axis in the embryo, is the main embryonic, Gro-associated Cic-S, which is also involved in AP patterning. 
 35 Results Figure 18. Integrity of the Cic-S protein is required for zen repression. (A): Diagram of the Cic transgenes used to rescue the DV defects caused by Cic mutant alleles. CicA is the full-length main embryonic Cic-S isoform. CicΔΝ60 is a variant of CicA, lacking the aminoterminal first 60 residues, while Cicmini is a synthetic CicA form in which the most conserved regions have been assembled. (B- G): expression of zen mRNA transcripts in embryos that proceed from females that are either wild-type (B), carry germline clones for the cic1 allele (C) or are trans-heterozygous for the cic1 and cicQ474X (cicQ) alleles (D). zen transcripts in embryos that proceed from females that are cic1/ cicQ474X and express as rescue transgenes the CicA, Cicmini or CicΔΝ60 proteins (E-G). Although the effects of cic1 and rescue experiments by the Cic variants strongly indicated that the CicA isoform functions in embryo DV patterning, the nature of the cic1 allele led us to an alternative approach to confirm our hypothesis. The cic1 mutation is caused by the insertion of a hobo element in the 5’ UTR region shared by all possible Cic-S proteins in the follicular epithelium and the embryo (see figure 17). Although it theoretically disrupts their expression, we could not exclude the remote possibility that the inserted element generates cryptic splicing sites or alternative promoter elements that permit the expression of some functional Cic isoform in the follicular epithelium. In this case, the wild-type expression of Pipe would be due to a rescue of the cicfetU6 effects by this functional isoform, and not because cic1 does not affect the functions of Cic in the egg chamber. To rule out the above possibility, we have used a reverse genetics approach, with the purpose to completely knock down the N2 containing Cic form.
 36 Results Using the CRISPR/Cas9 system, we introduced a frameshift deletion of 11 bp in the Cic locus, directly downstream the ATG triplet that is the start codon for the main embryonic Cic-S isoforms. This creates a premature stop codon and as a product it gives a short polypeptide, whose sequence is irrelevant to Cic-S and lacks the N2 motif. After establishing a line for the Cic-S frameshift mutation, aka named cic5, we have further characterized it, examining its effects considering N2-dependent and N2-independent Cic functions (figure 19). Homozygous cic5 females were completely sterile and laid eggs lacking all their segments and posterior spiracle, as in other strong Cic mutations. This cuticle phenotype indicated that the repressor activities of Cic in the early embryo were highly affected. As markers for Cic repressor activity, we monitored the expression of the genes tll and kni. In cic5 mutants, tll expression expanded towards the embryo trunk, at the expense of the posterior band of kni, indicating that cic5 behaves as a null allele of Cic-S, considering its functions in embryonic AP patterning. However, wing vein formation, which is independent of the N2 motif, was normal, indicating that only the N2-associated functions of Cic were affected by the cic5 mutation. Considering DV patterning, we examined pipe in egg chambers from cic5 homozygous females, and observed that it was expressed as in a wild-type animal, indicating that cic5 leaves the oogenic functions of Cic unaffected. In the embryo, we monitored twi as a Dl- activation marker and zen as a repression marker. As in cic1 mutant embryos, twi was activated normally along the ventral circumfence, but zen was expanded throughout almost the entire embryo. These effects were more prominent than in cic1 embryos, and were also observed in females carrying germline clones of cic5. In overall, the cic5 allele invalidates the set of Cic-S variants that include the N2 motif and behaves as a null allele in the embryo, while it does not affect other Cic functions, including the ones in the ovary follicle cells. In conclusion, we have selectively targeted the functions of the major embryonic Cic-S isoform and demonstrated that it is involved in repression events required for the formation of the embryo DV axis, independently of its upstream role in DV patterning during oogenesis, which is possibly accomplished through a different, N2-independent isoform of the protein.
 37 Results Figure 19. Generation and charecterization of the cic5 allele. (A): Schematic representation of the Cic-S locus and CRISPR/Cas9 targeting. cic5 effects on known Cic functions: mRNA transcripts of tll and kni (B, E), cuticle phenotype (C, F) and wing vein formation (D, G) in embryos deriving from wild-type or homozygous cic5 females. cic5 effects on DV patterning: expression of pipe-lacZ protein is monitored by staining with an anti-βgal antibody in stage 9-10 egg chambers of wild-type (H) and homozygous cic5 females (J). DAPI staining is used to visualize of the nuclei of the follicle cells of the egg chambers (H´, J´). mRNA transcripts of zen and twi in embryos deriving from females that are wild-type (I), cic5 homozygous (K), or carry germline clones of the cic5 mutation (L). 
 38 Results 2. Molecular mechanism of Cic function in embryonic DV patterningIn section 1 of the Results we have uncoupled the functions of Cic in DV patterning during oogenesis and embryogenesis. Once we had shown that Cic has a strictly embryonic function in the establishment of the DV axis, we set off to study the molecular mechanism of this function. We had shown that the main embryonic Cic-S isoform is involved in repression of genes along the DV axis during early embryogenesis (stage late 4-early 5). This same isoform represses the terminal genes tll and hkb during the same stage through a well-described mechanism: it contacts canonical octameric sites in regulatory sequences of its target genes through its HMG-box and C1 domains and recruits the Gro corepressor through its aminoterminal N2 motif (Ajuria et al., 2011a; Forés et al., 2015; Forés et al., 2017a). Furthermore, this repressor function is subjected to the control of the Torso RTK pathway through the carboxyterminal C2 motif of Cic, that serves as a docking site for the activated MAPK (dpERK), which inactivates Cic and relieves its repressor activities. Based on this knowledge, we set off to study whether any of the above mechanisms apply for Cic repression of type III dorsal-specific genes. 2.1 Cic represses type III genes through the AT-rich sites of the VRE elements Repression of the dorsal-specific type III genes in blastoderm embryos is controlled by short elements within their enhancers, nominated Ventral Repression Elements (VRE). Interestingly, the AT-rich regions of zen and tld VRE´s contain sequences that highly resemble the consensus binding site for Cic (CBS) in its other embryonic targets. In fact, in all AT-rich regions of both genes we identified octameric sites that differ from the CBS by only one nucleotide. This similarity led us to the hypothesis that Cic could be binding to them and act as one of the long-seeked components of the Dl repressosome. To explore this possibility, we have analyzed the interactions between Cic and the AT-rich elements in vitro and in vivo.
 39 Results Figure 20. Sequence (A) and schematic representation (B) of the VRE element at the zen promoter. Cic binds in vitro to the AT-rich sites of zen and tld The VRE element of zen includes four AT-rich sites, nominated AT0, AT1, AT2 and AT3. From these, AT0 and AT1 are 10 bp upstream the respective Dl sites Dl0 and Dl1, while spacing between AT2 and AT3 from their Dl site pair is 8 and 38 bp respectively (Kirov et al., 1993). As mentioned above, the AT-rich site sequences highly resemble the Cic Binding Sites (CBS) of its known targets such as tll, hkb and ind. Specifically, the AT0 and AT2 sites, that both contain the sequence TGAACGAA, present a single base change in position 5 respectively to the hkb/ind CBS site (TGAATGAA). The AT1 and AT3 sites, TTCTTTGA and TTTATTGA resemble the reverse complement sequence of the tll CBS (TCAATGAA), with a base change in position 4 and 3 respectively. To test a direct interaction between Cic and the AT-rich sequences of the zen VRE, we performed Electrophoretic Mobility Shift Assay (EMSA) experiments, using a synthetic form of Cic, containing the HMG-box and C1 domains (figure 21). 
 40 Results Figure 21. (A): Cic binds in vitro to the AT-rich sequences in the VRE element of zen and this is linked to in vivo repression. Numbers above gel indicate fusion proteins used in the binding reactions. (B,C): mRNA transcripts of zen in wild-type and cic5/cic4 embryos. (D): Diagram of Cic-His-tag fusion proteins used for in vitro binding assays. (E): wild-type and mutant labelled DNA probes containing binding sites from the promoters of ind and zen. As a positive control to validate the binding reactions, we have used a DNA probe that contains the CBS for ind, incubated with the same HMG-C1 protein. As negative controls, we have used two variants of the HMG-C1 protein: one bearing a single aminoacid change in the HMG-box domain (L524A) (HMGmut-C1) and one bearing a deletion of four aminoacids of the C1 domain (ΔRQKL) (HMG-C1mut). Both of these mutations have been shown to abolish binding to known CBS sites (Forés et al., 2017). We observed that, the HMG-C1 protein, expressed as a His-tag
 41 Results fusion, bound in vitro to short DNA probes containing the AT1 and AT2 sites of zen, with affinities that were 12,5 fold and 7,5 fold less respectively to the ind CBS. Albeit weak, interaction between Cic and the zen AT´s, was specific, since either mutation of the binding site or the protein, disrupted the interaction. When the AT1 site was changed to TCAGAGAA (G