Dissecting Mechanisms Controlling Neural Network Formation in Drosophila Melanogaster

Dissecting mechanisms controlling neural network formation in Drosophila melanogaster

Hong Long

Department of Neurology and Neurosurgery

McGill University

Montreal, Quebec, Canada

August 2009

A thesis submitted to the Faculty of Graduate Studies and Research in partial fulfillment of the requirements for the Ph.D. degree in Neuroscience

Library and Archives Bibliothèque et Canada Archives Canada

Published Heritage Direction du Branch Patrimoine de l’édition

395 Wellington Street 395, rue Wellington Ottawa ON K1A 0N4 Ottawa ON K1A 0N4 Canada Canada

Your file Votre référence ISBN: 978-0-494-66458-2 Our file Notre référence ISBN: 978-0-494-66458-2

NOTICE: AVIS:

The author has granted a non- L’auteur a accordé une licence non exclusive exclusive license allowing Library and permettant à la Bibliothèque et Archives Archives Canada to reproduce, Canada de reproduire, publier, archiver, publish, archive, preserve, conserve, sauvegarder, conserver, transmettre au public communicate to the public by par télécommunication ou par l’Internet, prêter, telecommunication or on the Internet, distribuer et vendre des thèses partout dans le loan, distribute and sell theses monde, à des fins commerciales ou autres, sur worldwide, for commercial or non- support microforme, papier, électronique et/ou commercial purposes, in microform, autres formats. paper, electronic and/or any other formats. . The author retains copyright L’auteur conserve la propriété du droit d’auteur ownership and moral rights in this et des droits moraux qui protège cette thèse. Ni thesis. Neither the thesis nor la thèse ni des extraits substantiels de celle-ci substantial extracts from it may be ne doivent être imprimés ou autrement printed or otherwise reproduced reproduits sans son autorisation. without the author’s permission.

In compliance with the Canadian Conformément à la loi canadienne sur la Privacy Act some supporting forms protection de la vie privée, quelques may have been removed from this formulaires secondaires ont été enlevés de thesis. cette thèse.

While these forms may be included Bien que ces formulaires aient inclus dans in the document page count, their la pagination, il n’y aura aucun contenu removal does not represent any loss manquant. of content from the thesis.

Table of contents

List of figures and tables 8

List of abbreviations used 10

Abstract 21

Résumé 23

Acknowledgements 25

Contribution of co-authors 26

Original contributions to knowledge 27

Chapter 1. Introduction 28

1.1 Axonal guidance, targeting, and tiling 28

1.1.1 Growth cone 28

1.1.2 Guidepost cell, Pioneer axon, and the labelled pathway 29

1.1.3 Layer-specific targeting 31

1.1.4 Topographic Mapping 32

1.1.5 Axon tiling and self-avoidance 33

1.1.6 Drosophila visual system 35

1.1.7 Review of molecules and signalling pathways involved in R-cell axon

guidance, targeting, and tiling. 39

1.1.7.1 Nuclear proteins 39

1.1.7.1.1 Brakeless 39

1.1.7.1.2 Runt 41

1.1.7.1.3 Nuclear factors YC 42

1.1.7.1.4 Sequoia 44

1.1.7.2 Cytoplasmic signalling proteins 46

1.1.7.2.1 Dreadlocks 46

1.1.7.2.2 Misshapen 47

1.1.7.2.3 Bifocal 49

1.1.7.2.4 p21-activated kinase 51

1.1.7.2.5 Trio 53

1.1.7.2.6 Rho-GTPase Rac 57

1.1.7.3 Cell surface proteins and secreted proteins 58

1.1.7.3.1 Insulin Receptor 58

1.1.7.3.2 N-Cadherin 60

1.1.7.3.3 Flamingo 68

1.1.7.3.4 Leukocyte common antigen-related RPTP 71

1.1.7.3.5 PTP69D 74

1.1.7.3.6 Capricious 76

1.1.7.3.7 Golden Goal 78

1.1.7.3.8 Anaplastic lymphoma kinase and Jelly Belly 80

1.1.7.3.9 Off-track and Semaphorin-1a 84

1.1.7.3.10 Dfrizzled2 and DWnt4 87

1.1.7.3.11 Transforming Growth Factor β /Activin 89

1.1.8 Review of genes involved in glia and target field development 91

1.1.8.1 Genes involved in glia development, migration, and function 91

1.1.8.1.1 Decapentaplegic, Hedgehog, and Gilgamesh 91

1.1.8.1.2 Wingless, Dpp, and Glial cell missing 93

1.1.8.1.3 Non-stop, SAGA complex and Histone H2B 94

1.1.8.1.4 JAB1/CSN5 95

1.1.8.2 Genes involved in the lamina-lobula boundary formation 96

1.1.8.2.1 Robo and Slit 96

1.1.8.2.2 Egghead 98

1.1.8.3 Genes involved in target neuron development 99

1.2 Dendrite patterning and function 101

1.2.1 Dendrite development, arborisation, and patterning 101

1.2.1.1 Dendrite outgrowth 102

1.2.1.2 Dendrite guidance and targeting 104

1.2.1.3 Dendrite branching and spine formation 106

1.2.1.4 Dendrite heteroneuronal tiling 108

1.2.1.5 Dendrite isoneuronal self-avoidance 110

1.2.1.6 Dendrite maintenance 112

1.2.1.7 Dendrite remodelling 115

1.2.2 Drosophila PNS dendritic arborisation (da) neurons 117

1.3 An overview of my studies and their contributions to our knowledge of the 121

related fields

1.3.1 My work in axon guidance and tiling in the fly visual system 121

1.3.2 My analyses in dendritic patterning in the fly da neurons 124

Chapter 2. De Novo GMP Synthesis Is Required for Axon Guidance in Drosophila

126

Summary 128

Introduction 129

Materials and methods 132

Genetics 132

Molecular Biology 132

Histology and Immunohistochemistry 133

Results 134

Figures 144

Table 160

Discussion 161

Acknowledgements 165

References 166

Chapter 3. Dendrite branching and self-avoidance are controlled by Turtle, a

conserved IgSF protein in Drosophila 171

Summary 173

Introduction 174

Materials and methods 177

Fly stocks and genetics 177

Imaging and quantification 178

Immunohistochemistry 178

Results 180

Figures 189

Table 213

Discussion 214

Acknowledgements 221

References 222

Chapter 4. Characterizing gene Dnrk in Drosophila 227

Summary 229

Introduction 231

Materials and methods 243

Genetics 243

Molecular biology 243

Histology and Immunohistochemistry 244

Results 246

Figures 259

Table 278

Discussion 279

Acknowledgements 282

References 283

Chapter 5. General discussion 296

5.1 The involvement of de novo GMP synthesis in R-cell axon guidance 298

5.1.1 de novo GMP synthesis versus the salvage synthesis pathway 298

5.1.2 How GMP level is controlled within the cell? 299

5.1.3 Signaling events of GMP 301

5.1.4 bur also functions in motor axon guidance 302

5.1.5 Defasciculation, a requirement for layer-specific R1-R6 targeting? 303

5.1.6 Is the de novo GMP synthesis required for neural network formation

in mammals? 304

5.2 The function of Turtle in dendrite branching and patterning 304

5.2.1 Domain requirement and the possible homotypic function of Tutl 304

5.2.2 The dendrite patterning phenotype in class I neuron 306

5.2.3 Functional relationship between tutl and other known genes in dendrite

development 307

5.2.4 Potential role of tutl in the VNC? 307

5.2.5 Tutl functions in R7 axon tiling in the fly visual system 308

5.2.6 The function of Tutl in fly behavior 308

5.2.7 The function of the mammalian homolog of Tutl 309

5.3 The function of Dnrk in fly 310

5.3.1 The potential redundancy between Dnrk and Dror 310

5.3.2 The potential function of Dnrk in the VNC 311

5.3.3 Genetic interaction between Dnrk and DWnt5 312

References 314

List of figures and tables

Chapter 2, figure 1: Molecular characterization of the bur gene 144

Chapter 2, figure 2: Mutations in the bur gene disrupted R-cell axon pathfinding 146

Chapter 2, figure 3: bur is not required for lamina-specific termination of R2-R5… 148

Chapter 2, figure 4: R-cell differentiation and patterning remained normal in bur… 150

Chapter 2, figure 5: Rescue of the R-cell pathfinding phenotype by expressing… 152

Chapter 2, figure 6: Mutations in the ras gene caused a bur-like phenotype in R… 154

Chapter 2, figure 7: Depleting guanine from the fly food did not affect R-cell axon 156

Chapter 2, figure 8: Reducing the dosage of bur enhanced the Rac phenotype 158

Chapter 2, table 1: Transgenic rescue of R-cell axonal hyperfasciculation… 160

Chapter 3, figure 1: Tutl protein structure, tutl alleles, and Tutl expression in da… 189

Chapter 3, figure 2: tutl is required to restrain dendrite branching in class I da… 191

Chapter 3, figure 3: Class II and class III DA neurons are unaffected in tutl23… 193

Chapter 3, figure 4: tutl is required for dendrite self-avoidance in class IV da… 195

Chapter 3, figure 5: tutl mutants exhibit normal dendritic tiling among class IV… 197

Chapter 3, figure 6: Overexpression of Tutl inhibits dendrite branching in class… 199

Chapter 3, figure 7: The cytoplasmic tail of Tutl is dispensable for function in… 201

Chapter 3, figure S1: Additional examples of dendrite branching and self-… 203

Chapter 3, figure S2: Quantification of the effects of tutl mutations on additional… 205

Chapter 3, figure S3: Overexpression of Tutl is not sufficient to induce dendrite… 207

Chapter 3, figure S4: Tutl expression in mutants of ab, ss, kn, and cut 209 8

Chapter 3, figure S5: Expression of Tutl and TutlΔcyto in class I da neurons of tutl… 211

Chapter 3, table S1: Genetic Interaction Experiments for tutl and trc 213

Chapter 4, figure 1: Predicted domain structure of Ror family proteins 259

Chapter 4, figure 2: Molecular characterization of the Dnrk gene 261

Chapter 4, figure 3: Mutation in the Dnrk gene disrupted R-cell axon guidance… 263

Chapter 4, figure 4: The adult R-cell axon projection pattern in Dnrk mutant was… 266

Chapter 4, figure 5: Dnrk mutant did not display axon guidance defect in the VNC… 268

Chapter 4, figure 6: Dendrite development and tiling pattern in wild type and… 270

Chapter 4, figure 7: Potential functional redundancy between Dnrk and Dror in… 272

Chapter 4, figure 8: Overexpression of UASDnrk in the eye and in the da… 274

Chapter 4, figure 9: The overexpression of UASDnrk in the wing could change… 276

Chapter 4, table 1: Modifier screen on the loss-of-cross-vein phenotype induced… 278

List of abbreviations used

18A CadN18A

18B CadN18B

Ab Abrupt

Abl Abelson

AChR Acetylcholine receptor

ACV Anterior cross vein

ADSL Adenylosuccinate lyase

AEL After egg laying

Alk Anaplastic lymphoma kinase

A-P axis Anterior-posterior axis

AP Alkaline phosphatase

APF After puparium formation

ATP Adenosine-5'-triphosphate

BDGP Berkeley Drosophila Genome Center

BDNF Brain-derived neurotrophic factor

Bif Bifocal

Bks Brakeless

BMP Bone Morphogenetic Protein

Boss Bride of sevenless

Brn Brainiac

Bur Burgundy

CadN N-cadherin

CaMK Ca2+/calmodulin-dependent protein kinases cAMP Cyclic adenosine monophosphate

Caps Capricious

CDS Coding sequence cGMP Cyclic guanosine monophosphate

CNS Central nervous system

Co-IP Co-Immunoprecipitation

CPG 15 GPI-linked candidate plasticity gene 15

CRD Cysteine rich domain

CRIB Cdc42/Rac interactive binding

CSN COP9 signalosome

C-terminal Carboxyl-terminus

CUB Complement C1r/C1s, Uegf, Bmp1

CV Cross veins da neurons Dendritic arborisation neurons

DAB 3,3'-Diaminobenzidine

Dac Dachshund

Dasm-1 Dendrite arborization and synapse maturation 1

Dfz2 Dfrizzled2

DI Potential insulin binding site I

DInR Insulin Receptor

Dlar Leukocyte common antigen-related RPTP

Dnrk Neurospecific receptor kinase

Dock Dreadlocks

Dpp Decapentaplegic

DRG Dorsal root ganglion

DS Down‟s syndrome

Dscam Down syndrome cell adhesion molecule

DSHB Developmental Studies Hybridoma Bank

D-V axis Dorsal-ventral axis

E10 Embryonic stage 10

ECM Extracellular matrix

EGF Epidermal growth factor

Egh Egghead

Elav Embryonic lethal abnormal vision

ELF ey3.5-Gal80, lama-Gal4, and UAS-FLP

EMS Ethyl methanesulfonate

Ena Enabled

EST Expression sequence tag

EyFLP Eye-specific FLP system

F-actin Filamentous actin

Fas Fasciclin

FGF Fibroblast growth factor

FLP Flippase recombination enzyme

Fmi Flamingo

FN-III Fibronectin type III

FRT Flippase Recognition Target

Fry Furry

Fz Frizzled

G protein Guanine nucleotide-binding proteins

Gbb Glass bottom boat

Gcm Glial cell missing

GDP Guanosine diphosphate

GEF Guanine nucleotide exchange factor

GF Giant fiber neurons

GFP Green fluorescent protein

Gish Gilgamesh

GMP Guanosine monophosphate

GMR Glass multimer reporter

GOF Gain-of-function

Gogo Golden Goal

GPC Glial precursor cell area

GST glutathione S-transferase

GTP Guanosine-5'-triphosphate

Hh Hedgehog

Hpo Hippo

HRP Horseradish peroxidase

HSPG Heparan sulphate proteoglycan

Ig immunoglobulin

IgSF Immunoglobulin superfamily

IMP Inosine monophosphate

IMPDH Inosine monophosphate dehydrogenase

Imp-α3 Importin-α3

IPC Inner proliferation centre

Iro Iroquois

ISN Intersegment nerve

Jak Janus kinase

Jeb Jelly Belly

JM Juxtamembrane region

Kn Knot

L neuron Lamina neuron

LDL Low density lipoprotein receptor domain

LG Lamina glia cells

LOF Loss-of-function

LRR Leucine-rich repeats

LV Longitudinal veins

M layer Medulla layer

MAb Monoclonal antibody

MAPK Mitogen-activated protein (MAP) kinases

MARCM Mosaic analysis with a repressible cell marker

MB Mushroom bodies md neuron Multiple dendrites neuron mdGAL4 GAL4109(2)80

Msn Misshapen

MuSK Muscle specific kinase

Myr myristoylation

NDR Nuclear Dbf2-related

NF-Y Nuclear factor Y

NGF Nerve Growth Factor

NIK Nck interacting kinase

NMDA N-methyl-D-aspartic acid

NMJ Neuromuscular junction

Not Non-stop

Nrg Neuroglian

NRTK Nonreceptor PTK

NT Neurotrophin

N-terminal Amino-terminus

OPC Outer proliferation centre

Otk Off-track

Pak p21-activated kinase

PBS Phosphate buffered saline

PCP Planar cell polarity

PCR Polymerase chain reaction

PCV Posterior cross vein

PDGF Platelet-derived growth factor

PDZ PSD-95/Dlg/ZO-1-like

PI3K Phosphoinositide 3-kinases

PKA Protein kinase A

PNS Peripheral nervous system

PRD Proline-rich domain

PTB Phosphotyrosine binding domain

PTK Protein tyrosine kinase family

Ras Raspberry

RATL Rapsyn associated transmembrane linker

RBG Retina basal glial cells

R-cell Photoreceptor cell

RGC Retinal ganglion cell

RNAi RNA interference

Robo Roundabout

Ror Receptor tyrosine kinase-like orphan receptor

RPTP Receptor protein-tyrosine phosphatase

RTK Receptor PTK

Run Runt

SAGA complex Spt-Ada-Gcn5-acetyltransferase complex

Sav Salvader

Sdk Sidekick

Sema Semaphorin

Sens Senseless

Seq Sequoia

SGC Guanylate cyclase

SH SRC Homology domain

Sim Single-minded

Sina Seven in absentia

SN Segmental nerve

Ss Spineless

TGF-β Transforming growth factor β

Ti1 Tibia neuron

TK Tyrosine kinase domain

TM Transmembrane domain

Trc Tricornered

Tsp1 Thromospondin1 domain

Tutl Turtle

UAS upstream activating sequence

Unp Unplugged

UTR Untranslated region

UV Ultraviolet light

VEGF Vascular endothelia growth factor

VNC Ventral nerve cord

Wg Wingless

Wt Wild-type

Wts Warts

XMP Xanthosine monophosphate

YTH Yeast-two-hybrid

Abstract

Formation of precise neuronal networks during development is fundamental for the normal function of the nervous system. As two important domains of a neuron, the axon and the dendrite are required to grow, correctly navigate, and accurately target to their synaptic targets. Extensive studies have been done to investigate the mechanism underlying neural network formation. Accumulating evidence shows that every neuron has its own specific axon and dendrite projection blueprint shaped by both the intrinsic signals and environmental cues. In this thesis, I document my research studying mechanisms of neural network formation in the Drosophila melanogaster nervous system.

In the visual system, I characterized the important function of the GMP synthetase (Bur) in controlling the correct axon guidance of photoreceptor neurons. This study reveals the requirement of the de novo GMP synthesis in the axon guidance of the nervous system. I also provide evidence that the small GTPase Rac functions downstream of Bur in this axon guidance process.

In the PNS dendritic arborization (da) neurons, I investigated the function of one IgSF family member, Turtle, in controlling dendrite branching and self-avoidance. Turtle has different functions in different classes of da neurons. As the second identified membrane protein involved in dendritic self-avoidance, Turtle helps us to broaden our knowledge in dendrite pattern formation.

I also investigated the involvement of the Ror family receptor tyrosine kinases in controlling the photoreceptor axon guidance in the visual system. Two members of this family, Dnrk and Dror, likely function redundantly with each other in this process. My preliminary analysis suggests a possible link between Dnrk and DWnt5. This presents an excellent starting point for further dissection of the mechanism by which the Ror family members regulate axon guidance in Drosophila.

Résumé

La formation précise du réseau neuronal durant le développement est fondamentale pour la fonction normale du système nerveux. En tant que deux domaines important d`un neurone, l`axone et la dendrite sont requis pour la croissance, la navigation correcte, et la mise en mire exacte de leurs cibles synaptiques. Des études étendues on déjà été accomplies pour enquêter sur le mécanisme de la formation du réseau neuronal. Un nombre croissant de preuves démontre que chaque neurone possède un propre plan spécifique pour les projections axonales et dendritiques, façonné par des signaux à la fois intrinsèques et environnementaux. Dans cette thèse, je documente ma recherche sur l`étude du mécanisme de la formation du réseau neurologique dans le système nerveux de la mouche Drosophile melanogaster.

Dans le système visuel, j`ai caractérisé l`importante fonction de la synthétase de GMP

(Bur) en contrôlant le guidage correct des axones des neurones photoréceptrices. Cette

étude découvre l`exigence de la synthèse de novo de GMP dans le guidage d'axone du système nerveux. Je démontre aussi que la petite GTPase Rac fonctionne en aval de Bur dans ce processus de guidage d'axone.

Dans les arborisations dendritiques (ad) des neurones du PNS, j'ai étudié la fonction d'un des membres de la famille des IgSF, Turtle, dans le contrôle des branchements dendritiques et l`évasion de soi. Turtle a différentes fonctions dans différentes classes des

neurones ad. En tant que deuxième protéine membranaire impliquée dans l`évasion dendritique de soi, Turtle nous aide à élargir notre connaissance dans la formation des dessins dendritiques.

J'ai également examiné la participation de la famille Ror des récepteurs tyrosines kinases dans le contrôle du guidage d'axones des photorécepteurs dans le système visuel. Deux membres de cette famille, Dnrk et Dror, fonctionnent probablement de manière redondante lùn avec làutre dans ce processus. Mon analyse préliminaire suggère un lien possible entre Dnrk et DWnt5. Ceci présente un excellent point de départ pour des dissections ultérieures du mécanisme par lequel les membres de famille de Ror règlent le guidage dàxones dans la Drosophile.

Acknowledgements

First and foremost I want to say thank you to my supervisor Dr. Yong Rao. I am grateful for all his support, advice, open-mindedness, and enthusiasm throughout all aspects of my doctoral project. It is a privilege for me to be a part of his team.

I‟d also like to thank many other faculty members that have contributes over the course of my studies. Dr. Don van Meyel and Dr. Stefano Stifani for participating on my graduate advisory committee. Special thanks to Dr. Don van Meyel for support and advice on my dendrite project. Members of the MGH Center for Research in Neuroscience (CRN): Dr.

Sal Carbonetto, Keith Murai, and Robert Dunn for helpful discussions.

Members of the Rao lab and van Meyel lab have helped me along the way and I want to thank them: Wenjing Ruan, Patrick Cafferty, Tarek Houalla, Kerry Ferguson, Scott

Cameron, WenTzu Chang, Dac Hien Vuong, Yimiao Ou, Graham Thomas, and Jennie

Yang.

I am grateful for all the inspiration and support from my parents, my husband, my aunt and uncle.

Finally, I also need to thank McGill University Health Center, Foundation Desjardins,

Canadian Developmental Biology Conference, MGH Center for Research in

Neuroscience, and McGill University for their financial support. 25

Contributions of co-authors

Scott Cameron helped with the MARCM analysis of bur mutant and quantification of the genetic interaction between bur and Rac in Chapter 2.

Yimiao Ou helped with confocal microscopy imaging for da neurons, genetic analysis of mutants, quantification of the dendrite patterning, immunostaining with Tutl antibody, and figures assembling in Chapter 3.

Dr. Don van Meyel is the research supervisor for Yimiao Ou.

Dr. Yong Rao is the primary investigator and research supervisor for Scott Cameron and

Hong Long.

Original contributions to knowledge

Chapter 2: The data presented in this chapter report the first in vivo evidence to support an essential and specific role for de novo synthesis of guanine nucleotides in axon guidance. Bur and Ras, as two key enzymes in de novo GMP synthesis pathway, control

R-cell axon guidance in the brain. Our data support the involvement of de novo GMP synthesis pathway, but not the salvage pathway in regulating axon guidance in the fly visual system. Additionally, we provide the genetic interaction data between bur and Rac.

Chapter 3: In this chapter, we provide evidence to support the cell-autonomous function of a conserved IgSF family member, Turtle, in controlling dendrite branching and patterning in the fly PNS dendrite arborization neurons. Tutl has cell-type specific functions. In class I da neurons, Tutl inhibits excessive branch formation and in class IV da neurons, Tutl controls dendrite self-avoidance. For its function in class I da neuron branching, the cytoplasmic domain of Tutl is dispensable.

Chapter 4: This is the first mutant analysis for gene Dnrk. Dnrk has been previously proposed to function specifically in the fly nervous system. In this chapter I report the generation and phenotypic analysis in Dnrk null mutant. My data suggest that, as two members of the fly Ror family, Dnrk and Dror function redundantly in regulating R-cell axon guidance in the fly visual system. Additionally, I conducted a genetic modifier screen to identify DWnt5 as a suppressor for Dnrk gain-of-function phenotype in the wing.

Chapter 1: Introduction

1.1. Axonal guidance, targeting, and tiling

1.1.1 Growth cone

The development of the nervous system requires precise and specific connections between neurons and their synaptic targets (Goodman, 1996). During the developmental stage, neurons send out their axons towards their synaptic targets. After reaching their synaptic target field, axons from different neurons need to recognize the stop signals presented by the environment and shut down their motility. Following that, they need to recognize their own specific synaptic partners in the field and form synapses with those target cells.

The distance between the neurons and their synaptic targets can be very large, as much as several centimeters in vertebrates (Tessier-Lavige and Goodman, 1996). The navigation of the axon through this large distance is directed by a special structure, the growth cone, which is located at the leading edge of the axon (Goodman, 1996). The structure of a growth cone contains two parts: the actin-rich peripheral region, which has the filopodia and lamellipodia, and the microtubule-rich central region (Mueller, 1999). The growth cone is a highly dynamic structure and it moves quickly to explore the environment. The extracellular guidance cues presented either in substrate-bound forms or in secreted forms, can be detected by receptors present on the growth cone membrane. And the signal

transduction machinery within the growth cone converts those signals into directed movement of the growth cone.

1.1.2 Guidepost cells, Pioneer axons, and the “labelled pathway”

Sometimes a neuron is located far away from its final synaptic target. When this happens, the large distance between the neuron and its target cell is usually divided into several small navigating trajectories. Intermediate guidance cells act along different trajectories to direct the growth cone to its final target. In their pioneering work, the Bentley lab has found that the tibia (Ti1) neuron of the grasshopper embryo grows from the developing leg to the CNS, with the aid of specialized “guidepost cells”. Selective disruption of those

“guidepost cells” caused the turning away of Ti1 axon from its target (Caudy and Bentley,

1986; O‟connor et al., 1990; Legg and O‟Connor, 2003). Subsequent analysis showed that some guidepost cells use the chemorepellent Semaphorin, which forms a gradient, to guide the navigation of the Ti1 growth cone (Kolodkin et al., 1992).

In the development of the nervous system, pioneer axons often have critical roles.

Harrison first noticed that at the early developmental stage, pioneer axons extended from the PNS towards their target cells, which were located just a short distance away in the

CNS. With the development of the organism, later developing axons have to reach their targets in the same CNS by crossing through much larger distance than those early pioneer axons. Conveniently, those later axons just simply follow the tracts of those pioneer axons, by fasciculating with them, to reach their target field (Harrison, 1910). At a later time, the Goodman lab found that later growing axons selectively fasciculate with

only some specific pioneer axons. This was based on their discovery in the grasshopper embryo that later growing G and C axons selected only some specific pioneer axon tracts to fasciculate with, but ignored all the other axon tracts along their paths (Raper et al.,

1983). This finding led to the “labelled pathway” theory. This theory proposes that a particular axon should be labelled with specific cell surface molecules, and those molecules can be used in a contact-dependent manner to guide later developing growth cones.

The first piece of molecular data to support the “labelled pathway” theory came from the discovery of Fasciclin (Fas) I and II. In 1987, Bastiani et al. from the Goodman lab reported their identification of Fas I and II in a screen for specific molecules by raising antibodies against cell surface molecules on axon fascicles in the grasshopper CNS

(Bastiani et al., 1987). Fas I and II show very specific and unique labelling patterns in

CNS axonal tracts, in correlation with specific fascicules formation (Bastiani et al., 1987).

Afterwards, the experimental animal model was switched from grasshopper to

Drosophila melanogaster, the fruit fly, because Drosophila embryonic CNS is a miniature replica of that of the grasshopper embryo (Thomas et al, 1984), and the power of Drosophila genetics facilitates the identification of additional molecules involved in the labelled pathway in the nervous system. Further analysis of Fas II shows that the pathfinding of the later developing axons is simply done by adhering themselves to the pioneer axons through the homophilic binding activity of Fas II on the membrane

(Grenningloh et al., 1991; Lin and Goodman, 1994; Lin et al., 1994).

1.1.3 Layer-specific targeting

The vertebrate brain is organized into parallel layers or laminae composed of neurons

“With similar functional properties and similar sets of components” (Castellani and Bolz,

1997). In the “stripe assay”, cultured neurons had their axons extended on same cortical layers of the tissue explants as the specific layers they normally targeted to in the brain in vivo (Castellani and Bolz, 1997). This result suggests that those cortical layers have their own specific molecules, which can be selectively recognized by the surface-located molecules on those afferent axon growth cones (Castellani and Bolz, 1997). To date, several molecules have been identified to be involved in the layer-specific targeting in vertebrates, including N-cadherin (CadN) (Inoue and Sanes, 1997), Sidekick (Sdk) 1 and

2 (Yamagata et al., 2003), and Dscam (Yamagata and Sanes, 2008).

CadN, as a targeting signal, was discovered by Inoue and Sanes in their antibody perturbation assay. In their assay, the transverse chick tectal section was overlaid with retinal stripes in culture. Retinal ganglion cells could extend their axons to innervate specific tectal laminae, as they do in vivo. After the application of anti-CadN antibody to the culture, some retinal axons bypassed their normal targeting sites in the retinorecipient laminae (Inoue and Sanes, 1997). This result suggests the involvement of CadN in the layer-specific targeting of retinal axons. In the chick retina, Yamagata et al. found that

Sdk-1 and -2 mediated the layer-specific selection of retinal axons. Ectopic expression of sdk-1 in retinal Sdk-1 negative cells led to the switch of the terminal sites of those cells to the Sdk-1 positive laminae (Yamagata et al., 2003). Dscam was also found to mediate target selection of subset neurons in the chick retina (Yamagata and Sanes, 2008).

In Drosophila melanogaster, the fruit fly, there is a simple and stereotyped layer-specific targeting pattern of different R-cell axons. This provides a good system to identify molecules involved in axonal layer-specific targeting. For example, by using this system,

Garrity et al. have found that PTP69D was involved in the lamina-specific targeting of subset of photoreceptor cells (R cells) in the optic lobe of the third-instar larval brain

(Garrity et al., 1999). I will describe the details of fly visual system axon guidance and targeting in the section of Drosophila visual system.

1.1.4 Topographic Mapping

Topographic mapping describes the phenomenon that the spatial order of neurons maps smoothly onto their synaptic targets. Topographic mapping has been found in the retina to tectum/super colliculus axonal projection and the thalamocortical axonal projection in vertebrates. In invertebrates, it has been found in the fly R-cell axon projection to lamina in the optic lobe (Flanagan, 2006; Vanderhaeghen and Polleux, 2004; Clandinin and

Zipursky, 2002). Both vertebrates and invertebrates utilize topographic mapping in their visual systems for accurate spacial visual signal input. To date, the best known example is the retina-to-tectum axonal topographic mapping in Xenopus larvae, which is also known as retinotopic mapping.

In this retinotopic map, the temporo-nasal axis of the retina maps along the anterior- posterior (A-P) axis of the tectum, and the ventral-dorsal retinal axis maps to the dorsal- ventral (D-V) tectal axis. A-P axis mapping is controlled by ephrin-A and its receptor

EphA. The expression of ephrin-A in the tectum forms a low-to-high gradient along the

A-P axis. The expression level of EphA on retina axons increases from low to high along the nasal-temporo axis. High EphA expressing temporo axons terminate in the low ephrin-A expression anterior tectum because of the strong repulsion from Ephrin-A in the posterior tectum. Low EphA expressing nasal axons bearing weaker repulsion can go further to the posterior tectum. D-V axis mapping is controlled in a similar manner but by

Ephrin-B and its receptor EphB (Flanagan, 2006). Besides Ephrins and Eph receptors, some other repulsive molecules such as Repulsive guidance molecule (RGM), Wnt3, and

Engrailed-2 in the retina have been identified to function in regulating retinotopic mapping (Monnier et al., 2002; Brunet et al., 2005; Schmitt et al., 2006).

1.1.5 Axon tiling and self-avoidance

In both vertebrates and invertebrates, sensory processing centers are often organized into reiterated columns. This columnar structure contributes to organizing an internal topographic representation of the external world. Within each column, cells are arranged in a stereotyped fashion and form precise patterns of synaptic connections within discrete layers. Those connections are largely restricted to a single column to preserve spatial information from the periphery. Restricting axons to columns is conceptually similar to tiling (Millard et al., 2007). Tiling is defined as the complete and non-redundant coverage of receptive fields by neurons of a similar functional type (Grueber et al., 2002; Parrish et al., 2007). As two fundamental components of a neuron, both axon and dendrite display tiling phenomena in innervating their receptive field. I will focus on the axonal tiling phenomenon here, and discuss the dendritic tiling in the dendrite section.

Axonal tiling was first discovered in 1983, when researchers found that the peripheral axons of embryonic touch mechanosensory neurons (T-neurons) in the leech, innorvated restricted axonal territory between neighbouring T-neurons (Kramer and Kuwada, 1983;

Kuwada and Kramer, 1983). In the early 1990s, monkey retinal diffuse bipolar cells were also found to exhibit axonal tiling in a restricted territorial manner (Boycott and Wässle.,

1991). A similar tiling phenomenium has been discovered in dendrites even earlier than axonal tiling (Wässle et al., 1981). Investigations of dendrites have shown that dendritic tiling is controlled by both intrinsic and extrinsic cues (For details, please see the dendrite tiling section). But for axonal tiling, researchers have not obtained much information about the molecular mechanism underlying it until the discovery of the involvement of

Dscam2 and TGF-β/activin signalling pathway in axon tiling in the fly visual system

(Millard et al., 2007, Ting et al., 2007).

In the fly visual system, axons of photoreceptor cells, lamina neurons, and their target neurons in the optic lobe are organized into columnar structure (In the following section, I will discuss the structure of fly visual system.). Axons within each column are arranged in a stereotypical fashion to form precise connections with their target neurons. Those connections are largely confined to a single column. In this way the adjacent points in visual space are represented by a sharply defined topographic map of the columns. In this system, both lamina neuron L1 and photoreceptor cell R7 display axonal tiling. In the dscam2 mutant flies, the lamina neuron L1 could no longer restrict its axon terminal in a single column, but instead invaded the neighbouring columns (Millard et al., 2007). In R7

cells, mutations in the TGF-β/activin signalling pathway could disrupt the normal R7 axon tiling in the medulla (Ting et al., 2007).

Self-avoidance is defined as the tendency for arbors from the same neuron to avoid crossing, thereby spreading evenly over a receptive field (Kramer and Kuwada, 1983).

Both axons and dendrites display self-avoidance phenomina. Axonal self-avoidance is observed in some neurons having complex axon arborization. For example, the leech mechanosensory Pv neuron has complex axonal arborization pattern. Its axonal receptive region is composed of the major receptive field and the minor receptive field. Normally, axon branches from different fields do not overlap with each other. After surgically preventing or delaying the outgrowth of the axon branches which established only a minor receptive field, the remaining axon branches in the major field spread to cover this empty minor field (Kramer and Stent, 1985). If the major field was subject to surgical manipulations, axon branches from the minor field also extended to occupy the empty major receptive field. This result strongly suggests that there is self-avoidance between the axon branches from the same neuron. For the molecular mechanism controlling axonal self-avoidance, researchers have not gained much knowledge. But the analysis in the dendrite self-avoidance field (will be discussed in the dendrite self-avoidance section) may provide some suggestions.

1.1.6 Drosophila visual system

The Drosophila visual system consists of the compound eye and the optic lobe in the brain. An adult compound eye is made up of around 800 ommatidia. Each ommatidium

contains 8 photoreceptor cells (R cells), from R1 to R8, and twelve accessory cells. R cells are elongated sensory neurons that contain rhabdomeres, the photosensitive stacks of

Rhodopsin-bearing microvilli (Ready, 1989). R cells are subdivided into two groups according to their different location in the ommatidium. The outer R cells include R1-R6, detect green light. The inner R cells, include R7 and R8, detect UV and blue light respectively (Meinertzhagen and Hanson, 1993).

This adult visual system starts to develop at the early third (wandering) instar larval stage.

R cells in the eye-imaginal monolayer disc, project their axons towards the most posterior end of the disc, where they fasciculate into a thick bundle to enter the optic stalk. On entering the optic lobe of the brain, R-cell axons target to two different layers, the superficial layer called lamina, and the deeper layer called medulla (Meinertzhagen and

Hanson, 1993). Targeting of R-cell axons in the brain follows a topographic map. Axons from neighbouring R cells project to the neighbouring regions in the optic lobe. Dorsal R- cell axons project to the dorsal lamina, while ventral R-cell axons project to the ventral lamina. Young R cells located in the anterior eye disc project to the anterior region of lamina and older R cells located in the posterior eye disc project to the posterior lamina field (Kunes et al., 1993; Kunes and Steller., 1993; Sato et al., 2006).

In the eye imaginal disc, R cells differentiate in a sequential way. R8 differentiates first, followed by R2/5, R3/4, R1/6, and R7. In each ommatidium, R8 sends its pioneer axon towards the brain. The later developing R1-7 axons follow R8 pioneer axon simply by fasciculating with it. All 8 R-cell axons from a single ommatidium enter the brain as a

single fascicle (Tayler and Garrity, 2003). When R1-R6 axons reach the lamina, they detect stop signals presented by the epithelial/marginal glia and stop migrating (Poeck et al., 2001).

Around 30 hrs after reaching the lamina plexus, R1-6 axons within the same bundle defasciculate and each projects laterally to a unique target, thereby redistributing their inputs onto different synaptic targets (Cladinin and Zipursky, 2000). This redistribution allows R cells recognizing the same point in space to converge onto a common target.

Each target has five lamina neurons, named from L1 to L5. Those lamina neurons and the associated R-cell axons, form a highly patterned fascicle, called a cartridge. Within a single cartridge, each R-cell axon forms a synapse with a specific subset of lamina neurons, including L1-L3. Lamina cartridges are elaborated in a topographic way to reflect the relative visual view (Mast et al., 2006).

Although the stop signal for R1-6 axons is still unknown, researchers have found that the epithelial and marginal glia but not the lamina neurons located at the lamina plexus are essential for correct R1-6 axon termination (Poeck et al., 2001, Yoshida et al., 2005;

Weake et al., 2008). When those glia were removed from the lamina region by either the mutations of SAGA transcription complex components like non-stop, or by the disruption of Dpp signalling pathway, R1-6 axons bypassed the lamina layer to enter the medulla.

By contrast, when lamina neurons, which are the synaptic targets of R cells at the adult stage were removed by hedgehog (hh) mutation, R1-6 axons still terminated correctly in

the lamina (Huang and Kunes, 1996; Poeck et al., 2001). Those data suggest that the stop signal does come from lamina glia instead of lamina neurons.

Researchers also found that R-cell-to-glia signalling is important for lamina glia to migrate from their precursor cells areas (GPC) to the lamina plexus before R1-6 axons terminate in the lamina. The JAB1/CSN5 subunit of the COP9 signalosome complex is expressed in the R-cell axons to control the migration of lamina glia to their final location.

The correct migration and positioning of those glia are essential for the termination of later arriving R1-6 axons in the lamina layer (Chotard and Salecker, 2004).

At the third-instar larval stage, R7 and R8 pass through the lamina to enter the medulla.

Within the medulla, R7 and R8 axons first pause at the temporary layers (Ting et al.,

2005). Starting from ~50% APF, R7 and R8 axons proceed to their final destinations in the M6 and M3 layers of medulla, respectively. Within the medulla region, R7, R8, and lamina neuron axon afferents are arranged in columnar structures. Each column comprises the direct input from a single R7, a single R8, and the indirect input from R1-6 through the lamina neurons from a single lamina cartridge (Mast et al., 2006). Within each column, each axon forms its synaptic connection in a column-restricted manner, which is conceptually similar to the axon tiling as I have discussed in the previous section

(Millard et al., 2007; Ting et al., 2007). Medulla columns are also organized in a topographic manner to represent adjacent spatial cue inputs. Both lamina cartridge and medulla column, arranged in a topographic manner, can facilitate the recognition of different space point inputs (Mast et al., 2006).

The simple anatomy of the fly visual system and powerful genetic tools provide a good model system for researchers to identify genes involved in axon guidance and targeting.

Up to now, a number of molecules have been identified and analysed in this system. They include nuclear proteins, cell surface proteins, and cytoplasmic proteins. Their functions have been characterized in different aspects, including: guiding axons to the target region, regulating layer-specific targeting, maintaining proper fasciculation of axon bundles, target selection in the lamina and medulla, ensuring correct axon tiling, and inducing target field development. Some molecules are found to work in only one aspect, while the others are involved in more than one. I group those molecules according to their different subcellular locations. And I will describe the functions of those molecules in more details in following sections.

1.1.7 Review of molecules and signalling pathways involved in R-cell axon guidance, targeting, and tiling.

1.1.7.1 Nuclear proteins

1.1.7.1.1 Brakeless (Bks)

Bks is a nuclear protein. It was discovered independently by two research groups in their genetic screen for mutations disrupting R-cell axon guidance and targeting (Rao et al.,

2000; Senti et al., 2000). The first group discovered bks mutation through a P-element insertion induced mutant screen on the 2nd chromosome (Rao et al., 2000), while the second group found bks mutation through the eyFLP/FRT mosaic screen for EMS- induced random mutations (Senti et al., 2000; Newsome et al., 2000). In both screens, mutations of bks, generated either by the P-element insertion or the EMS mutagenesis,

could disrupt R-cell axon guidance and targeting. Expression of bks transgene in the eye in bks mutants could completely rescue the mutant phenotype (Rao et al., 2000; Senti et al., 2000). Those data indicate that Bks is required for correct R-cell axon guidance and targeting in the eye.

Bks has two isoforms, BksA and BksB, both having nuclei-targeting sequence (Senti et al., 2000; Yang et al, 2000). Consistently, Bks protein was detected in the nuclei of R cells in fly eye by immunohistochemistry (Rao et al, 2000; Senti et al., 2000). BksB includes the complete sequence of BksA and has an extra C-terminus zinc-finger-motif- containing tail, which suggests that BksB has a different function than BksA. Bks mutants disrupting both isoforms showed R-cell axon guidance and targeting phenotype, but mutants disrupting only BksB had normal R-cell axon projection pattern (Rao et al, 2000;

LaJeunesse et al., 2001). Either bksA or bksB transgene can rescue the R-cell axon phenotype in bks mutant. This suggests that BksA functions predominantly in R-cell axon guidance and targeting, while BksB is involved in some other developmental processes

(LaJeunesse et al., 2001).

In the analysis of R-cell differentiation and development, different R-cell-fate markers including Prospero for R7 (Kauffmann et al., 1996), Boss for R8 (Reinke and Zipursky,

1988), and Bar for R1 and R6 (Hayashi et al., 1998) were examined in bks mutants. The result indicated that bks mutant did not affect R-cell-fate determination (Rao et al.,

2000).This suggests that Bks controls R-cell axon guidance by regulating the expression of other genes, which may be involved in R-cell axon targeting. Overexpression of bks

was unable to retarget R7 axons to the lamina (Senti et al., 2000). This suggests that Bks needs to cooperate with other nuclear factors in order to properly target R1-R6 axons to the lamina. This hypothesis was confirmed by a later discovery that another nuclear protein, Runt was mislocalized to different R-cell types in bks mutant (Kaminker et al.,

2002).

1.1.7.1.2 Runt (Run)

Runt belongs to the Runt domain transcription factor family and it is specifically expressed in R7 and R8 in the wt eye disc (Kania et al., 1990; Canon and Benerjee, 2000;

Kaminker et al., 2002). Interestingly, in bks mutant, run was found to be ectopically expressed in R2 and R5, in addition to its normal expression in R7 and R8. This suggests that ectopic expression of run could be responsible for the bks mutant mistargeting phenotype. This hypothesis is supported by the overexpression phenotype of run in the eye. When run was specifically overexpressed either in all R cells or only in R2 and R5,

R1-R6 axons all bypassed the lamina to enter the medulla (Kaminker et al., 2002). No lamina plexus was formed in the run-overexpression mutant.

But in run mutant, R-cell differentiation and axon projection were all normal, suggesting that run probably functions redundantly with some other molecules, very likely another two run-related genes in fly (Kaminker et al., 2002). Notably, only R2 and R5 run overexpression, as in the bks mutant, could induce the R1-R6 axon-mistargeting phenotype. When overexpression of run happened in R cells other than R2 and R5, R1-

R6 axons all targeted normally. Like the bks mutant, the overexpression of run did not

cause any defect in R-cell differentiation (Kaminker et al., 2002). This again suggests the transcriptional functions of Bks and Run to control the expression of specific genes involved in R-cell axon targeting.

The analysis of the run-overexpression mutant also reveals the function of R2 and R5 axons as potential pioneer axons in guiding the later coming axons for proper lamina termination. In the outer R cells, R2 and R5 are the earliest differentiated cells. Possibly, their mistargeting into the medulla can guide the other outer R-cell axons to bypass the lamina, too. Because run is normally expressed exclusively in R7 and R8, which send their axons through the lamina to enter the medulla, this analysis also raise the possibility that Run controls the specific gene(s) either exclusively to be expressed in R7 and R8 or to be suppressed in those cells. When run was ectopically expressed in R2 and R5, by releasing the repression from Bks, R2 and R5 might obtain the identical axonal gene- expression profile as that of R7 and R8 to enter the medulla. This change led to the bypassing phenotype of those axons in both bks mutant and run-overexpression mutant.

To date, those genes under the transcriptional control of Bks and Run are still unidentified.

1.1.7.1.3 Nuclear factors YC (NF-YC)

To search for molecules involved in R7 function, the Zipursky lab conducted an eye- specific mosaic fly phototax-behavior screen with EMS-induced mutations (Clandinin et al., 2001; Lee et al., 2001). In the UV/visible light choice assay, flies are put in a T-maze with green light at the end of one arm and UV light at the end of another. Wt flies normally phototax towards the UV light. This behavior depends on the normal function of

R7 in detecting the UV light (Reinke and Zipursky, 1988). Flies with R7 dysfunction move towards the visible green light (Clandinin et al., 2001; Lee et al., 2001). In this screen, Morey discovered a NF-YC mutant severely disrupting R7 function (Morey et al.,

2008). Nuclear factor Y (NF-Y) is an evolutionarily conserved heterotrimeric transcription factor. It can bind to the CCAAT sequence in the promoter region of a gene, and interacts with different activators to regulate the transcription of this gene. NF-Y has three subunits: NF-YA, NF-YB, and NF-YC (Mantovani 1999). NF-Y has been found to function as both a transcription activator (Frontini et al., 2002) and a repressor (Peng et al.,

2002) in different organisms.

Mosaic analysis shows NF-YC specifically controls the layer-specific targeting of R7 through repressing senseless (sens) expression in those cells (Morey et al., 2008). In NF-

YC mutant, releasing the repression from NF-YC led to the misexpression of sens in R7.

This sens misexpression was sufficient for R7 to mistarget to R8 termination layer in M3.

Sens is the key regulator of R8 cell fate determination (Frankfort et al., 2001). Sens can also directly bind to the transcriptional activation site of capricious (caps) gene to control its expression in R8 (Morey et al., 2008; Shinza-Kameda et al., 2006). Caps is a R8- specific transmembrane adhesion protein and has been shown to control R8 layer-specific targeting (Shinza-Kameda et al., 2006).

In this model, NF-YC represses sens in R7. caps, as the target gene of Sens, its expression is thus suppressed. This allows R7 to distinguish itself from R8 and avoid the Caps in the

R8-termination layer, M3, to ensure R7 finally targets to M6. When NF-YC was removed

from R7, sens expression was up-regulated to ectopically promote caps expression in R7 to retarget it to R8 termination layer, M3. Additionally, Morey et al. also discovered that

Prospero (Pros) (Kauffmann et al., 1996), a R7-specific transcription factor, also controlled R7 layer-specific targeting, but likely in a parallel pathway to that controlled by NF-YC (Morey et al., 2008).

1.1.7.1.4 Sequoia (Seq)

Seq is a pan-neuronal nuclear protein, which has two zinc fingers highly homologous to the DNA binding domain of Tramtrack, which is a transcription factor involved in a variety of developmental processes (Brenman et al., 2001; Harrison and Travers, 1990;

Ready and Manley, 1992). seq was first identified in a genetic screen for mutations affecting the dendritic arborization (da) neuron dendrite development in the fly embryonic PNS (Gao et al., 1999). It was subsequently analysed for its functions in both dendrite/axon growth and neuronal cell fate determination in a subset of PNS neurons

(Brenman et al., 2001, Andrews et al., 2009). Like Tramtrack, Seq is also under the control of Notch (Artavanis-Tsakonas et al., 1999) and it is capable of regulating the expression level of potential downstream genes (Brenman et al., 2001).

In a genetic screen for genes controlling adult R7/R8 axon layer-specific targeting, seq mutants were isolated, which showed frequent loss of innervation in the M3 and M6 layers (Petrovic and Hummel, 2008). Seq controls R7 and R8 axon targeting in a cell- autonomous manner. When seq was absent in those cells, both R7 and R8 axons projected to the outer M1-M3 layers and missed their normal target layers in M6 and M3,

respectively. In a search for the expression pattern of seq, the expression level of seq in

R7 and R8 displayed an interesting pattern. Although seq is required in both R7 and R8, the onset and duration of seq expression in R7 and R8 within the same ommatidium is non-overlapping (Petrovic and Hummel, 2008). When seq expression reaches the peak in

R7, which starts to send out its axon, seq expression in R8 within the same ommatidium has been turned off. If the expression duration of seq in R8 was prolonged to overlap with that of R7, R8 axons projected into the R7 target layer in M6. This indicates the importance for the expression duration difference between R7 and R8 (Petrovic and

Hummel, 2008).

As a transcription factor, Seq can control the expression level of some genes encoding R- cell membrane proteins, such as Dlar and CadN (Petrovic and Hummel, 2008). Dlar functions in R7 layer-specific targeting (Clandinin et al., 2001; Maurel-Zaffran et al.,

2001), while CadN controls both R7 and R8 layers specific targeting (Lee et al., 2001).

For the detailed functions of Dlar and CadN in R-cell axon targeting, please see the following sections. Loss of CadN in seq mutant R cells can explain the targeting errors of

R7 and R8. But because overexpression of seq could not increase the expression level of

CadN, there must be some other molecules to counter control the expression of CadN.

Also, because the overexpression of seq could retarget R8 axon to R7 termination layer, while overexpression of both CadN and Dlar did not cause this phenotype, this raises the possibility that Seq also controls the expression of other targeting determinants.

The R7 and R8-specific onset of seq expression may also reflect that CadN, as one target gene controlled by Seq, is required only at specified time point for different R-cell axon targeting. According to the recent discovery that different CadN isoforms have different functions in R7 layer-specific targeting (Nern et al., 2005), it is also possible that Seq regulates the expression of different isoforms of CadN in different cell types at distinct time points for their interaction with target region to control the proper R-cell axon targeting.

1.1.7.2 Cytoplasmic signalling proteins

1.1.7.2.1 Dreadlocks (Dock) dock mutant was first identified in a screen for mutations disrupting larval R-cell axon guidance pattern (Garrity et al., 1996). The screen was done using the P-element insertion-induced mutations on the second chromosome, which caused larval or pupal lethality. dock mutant axons displayed defects in both axon guidance and targeting

(Garrity et al., 1996). At the third-instar larval stage, neighbouring dock mutant axons crossed the projection paths of each other. They abnormally fasciculated to form large bundles and thus caused the formation of gaps in the lamina. Many R-cell axons bypassed the lamina and hyperinnervated the medulla, with abnormal growth cone morphology

(Garrity et al., 1996). At the adult stage, in the dock mosaic mutant, abnormal crossing of

R7 and R8 axons could be frequently found in the brain. Some R1-R6 axons projected into the medulla. dock is expressed in R-cell axons, the central neuropil and weakly in the medulla neurons. Reintroducing the dock transgene into all R cells could completely rescue the dock mutant phenotype (Garrity et al., 1996). This suggests the requirement of

Dock in R cells to control R-cell axon guidance and targeting. Outside of the visual system, Dock is also found to play an important role in the motoneuron axon guidance

(Desai et al., 1999).

Dock is a SH2 and SH3 domain containing adaptor protein (Lehmann et al., 1990; Garrity et al., 1996). In the domain requirement analysis for Dock to function in R-cell axon guidance and targeting, Rao et al. found that both SH2 and SH3 domains were indispensable for its function (Rao and Zipursky, 1998). In general, SH2 domains bind to phosphotyrosine and SH3 domains bind to either PXXP or RXXK motif to mediate signal transduction (Downward, 1994; Schlessinger, 1994; Pawson 2004; Mayer and Eck, 1995).

Because Dock protein contains both SH2 and SH3 domains, and both domains are required in R-cell axon guidance and targeting, this suggests that Dock functions downstream of a receptor detecting the guidance cues and also potentially upstream of some other signalling molecule. This hypothesis led to the identification of Msn, the

Ste20-like serine/threonine kinase and the p21-activated kinase Pak.

1.1.7.2.2 Misshapen (Msn)

Nck, the vertebrate homolog of Dock was found to bind to NIK (Nck interacting kinase) in cultured cells (Su et al., 1997). In fly, the homolog of vertebrate NIK is Msn (Su et al.,

1998). msn mutation caused defects in cell shape and planar cell polarity of R cells in the fly eye (Treisman et al., 1997; Paricio et al., 1999) and abnormality in embryonic dorsal closure (Su et al., 1998). The homolog of Msn in C. elegans, Mig-15, is also critical for cell shape, cytoskeleton control, and axon navigation (Su et al., 1998; Poinat et al., 2002).

As an attractive candidate for potential interaction with Dock in R-cell axon guidance and targeting, msn mutants were analysed (Ruan et al., 1999). Similar to the dock mutant phenotype, msn mutant showed R-cell axon pathfinding errors and a mistargeting phenotype. At the third-instar larval stage, there were abnormal thick axon bundles in the medulla region and gaps in the lamina in the msn mutant (Ruan et al., 1999; Su et al.,

2000). Specific marker labelling indicated that a subset of R1-6 axons projected through the lamina layer into the medulla region. This R1-6 targeting defect was also observed at the adult stage.

Immunostaining shows that Msn is expressed in R-cell axons. Mosaic analysis and eye specific rescue data suggest that Msn functions in a cell-autonomous way to control R- cell axon targeting (Ruan et al., 1999). Another piece of convincing evidence for Msn controlling R-cell axon targeting is the overexpression phenotype of msn. In a wt background, overexpression of msn in R cells was sufficient to induce early termination of R-cell axons even before they reached the lamina termination layer (Ruan et al., 1999).

The interaction between Msn and Dock was assessed by both in vivo and in vitro analysis

(Ruan et al., 1999; Su et al., 2000). Physical association between Msn and Dock was confirmed by both the GST pulldown assay and Co-IP experiment (Ruan et al., 1999). In the yeast-two-hybrid (YTH) assay, the binding domains on Msn and Dock were mapped to the PXXP motif on Msn and the SH3-1 and SH3-2 domains of Dock (Ruan et al.,

1999). This is consistent with the early finding that both SH2 and SH3 domains are indispensible for Dock to function in R-cell axon targeting (Rao and Zipursky, 1998).

Genetic interaction between dock and msn was also observed. Decreasing the dosage of endogenous dock by 50% enhanced the msn hypomorphic mutant phenotype. While in the dock mutant, when msn was overexpressed, the non-stop phenotype of R-cell axons was suppressed. Interestingly, when dock and msn were co-overexpressed in the eye, R-cell axon termination pattern changed from the premature termination of R-cell axons induced by overexpression of msn alone back to normal (Ruan et al., 1999). So, dock is capable of controlling msn in both positive and negative ways. Those data suggest that Dock functions directly upstream of Msn to control R-cell axon targeting.

1.1.7.2.3 Bifocal (Bif)

Following the discovery of Msn as a downstream target of Dock in R-cell axon targeting,

Ruan et al. conducted a second-site mutation screen for mutations modifying the early axon termination phenotype induced by msn overexpression (Ruan et al., 2002). This led to the discovery of bif. bif was first discovered by its function in rhabdomere morphogenesis and its association with F-actin (Bahri et al., 1997). bif encodes a novel protein without any homology to known proteins. The tyrosine concentrated C-terminus of Bif contains a potential tyrosine phosphorylation site (Bahri et al., 1997). Like msn mutants, bif mutants showed defects in R-cell axon targeting. Over-expression of bif could also lead to the premature termination of R-cell growth cones, similar to the overexpression phenotype of msn. Immunostaining shows that Bif is present in R-cell growth cones. The function of Bif in the R cell is also supported by the eye-specific rescue of bif mutant phenotype by the bif transgene (Ruan et al., 2002).

The physical association between Bif and Msn was confirmed by both GST-pull down assay and Co-IP experiment. In the in vitro kinase assay, Msn could phosphorylate Bif on both C-terminus and N-terminus (Ruan et al., 2002). Those data indicates that Msn can directly bind to and phosphorylate Bif. Because previous work has shown that the intrinsic kinase activity of Msn is required for its function in fly embryonic dorsal closure

(Su et al., 2000), Ruan et al. examined whether the kinase activity of Msn was essential for its function in R-cell axon targeting. They found that a kinase defective msn could neither restore the R-cell axon projection pattern nor induce the premature targeting phenotype like the wt msn when it was overexpressed (Ruan et al., 2002). Thus, the kinase activity of Msn is indispensible from its function in R-cell axon targeting.

One interesting phenotype associated with bif overexpression in the eye was when bif was overexpressed with multiple copies, it caused the abnormal motility of R cell in the eye disc. R-cell cell bodies migrated to the most posterior end of the disc and clustered at the entry of the optic stalk (Ruan et al., 2002). Because bif mutation did not induce defects in

R cell migration, this abnormal cell motility induced by bif overexpression may be caused by the interfering of excessive bif with cytoskeletal events. This data suggests the potential role of Bif in regulating the actin cytoskeleton, which is also supported by its direct association with F-actin (Bahri et al., 1997; Sisson et al., 2000).

To address the functions of Bif and Msn in regulating the actin cytoskeleton, they were transfected into the Cos-7 cells individually or together to analyse the morphological change in transfected cells (Ruan et al., 2002). In control cells, F-actin is formed at low

level. It mainly localizes to the periphery and occasionally present as thin fibers inside of the cell. Expression of msn alone did not change either the cell morphology or the F-actin distribution. But when bif was expressed, cell morphology changed dramatically with increased F-actin level. Transfected cells displayed large number of microspikes and long thin projections resembling filopodia. Bif also colocalized with the long fine fiber-like F- actin. But when msn was co-expressed with bif, although the F-actin level remained similar, the organization of F-actin was strikingly different from that in cells expressing bif alone (Ruan et al., 2002). F-actin changed to be more like short large bundles and irregular large aggregates. Filopodia in cells were also much shorter. When a kinase defective msn was expressed together with bif, this mutant Msn didn‟t have the similar regulation on Bif to rearrange the F-actin (Ruan et al., 2002).This suggests that the kinase activity of Msn is required for it to regulate Bif in actin cytoskeleton rearrangement. And

Bif is capable of promoting actin polymerization. Msn can regulate the function of Bif in a kinase dependent manner to change cell morphology.

One working model for the action of Dock-Msn-Bif pathway is that Dock receives the signal from the cell surface receptor. It relays the signal down to its direct binding partner

Msn. Msn can directly bind to and phosphorylate the cytoskeleton regulator Bif. To respond to the signal input, Bif induces the actin cytoskeleton rearrangement and thus change the motility of the growth cone for proper R-cell axon targeting.

1.1.7.2.4 p21-activated kinase (Pak)

The identification of Pak as the interacting partner of Dock was also based on the observation of mammalian Pak binding to Nck in cultured cells (McCarty 1998). Fly has three proteins showing sequence homologies to mammalian Pak protein, including

Mushroom body tiny (Mbt) (Melzig et al., 1998), Pak, and Dpak2 (Hing et al., 1999).

Both Mbt and Dpak2 do not bind to Dock in vitro (Hing et al., 1999). Although Mbt is involved in fly CNS development, Mbt and Dpak2 have not been found to function in R- cell axon guidance and targeting (Hing et al., 1999). Only Pak displays the binding activity for Dock in the YTH assay (Hing et al., 1999). Similar to their mammalian relatives, the binding of Pak to Dock occurs between the N-terminal PXXP region of Pak and the SH3-2 domain of Dock, in both YTH assay and Co-IP experiment (McCarty 1998;

Hing et al., 1999).

Immunohistological analysis shows strong Pak staining in R-cell axons and growth cones.

Similar to dock mutants, pak mutants showed a disrupted R-cell axon projection pattern at the third-instar larval stage, such as the crossing of neighbouring axons, abnormal hyperfasciculation of R7 and R8 axons in the medulla and discontinuous lamina. Eye- specific rescue with pak transgene indicates that Pak is required in R cells to control R- cell axon guidance and targeting (Hing et al., 1999).

Pak protein has several functional domains, including the PXXP SH3 domain binding motif, the CRIB (Cdc42/Rac interactive binding) domain, the PIX guanine nucleotide exchange factor binding site, a serine/threonine kinase domain, and a Gβ binding site for the β subunit of G protein (Hing et al., 1999).

The Dock interacting PXXP domain, the CRIB domain, and the serine/threonine kinase domain are all required for pak transgene to rescue R-cell axon projection defects in pak mutants (Hing et al., 1999). To further test the requirement of kinase activity of Pak, Hing et al. generated a membrane anchored Pakmyr, which was predicted to be a constitutively kinase active form of Pak. pakmyr could fully rescue the pak mutant phenotype. But in wild-type background, it induced a dominant gain-of-function phenotype with early termination of R growth cones before the lamina layer in a dose-dependent manner. When multiple copies of pakmyr were expressed in the eye, the R-cell axon projection pattern was totally disrupted. R cell motility was also disturbed (Hing et al., 1999). This gain-of- function phenotype could not be induced if the constitutively active Pak was not anchored to the membrane. This indicates that the membrane anchoring of Pak is important for its function.

Interestingly, this membrane tethered pakmyr could also rescue the dock mutant phenotype, while wt pak, kinase inactive pak, and cytoplasmic constitutively active pak could not

(Hing et al., 1999). So, the recruitment of Pak by Dock to the membrane is important for

R-cell axon guidance and targeting. It is likely that Pak, after being activated by a Dock- mediated signal, forms a complex with Rho family GTPases and some other proteins to transduce the signal from the receptor to the actin cytoskeleton.

1.1.7.2.5 Trio

For Pak kinase to function in R-cell axon guidance and targeting, its CRIB domain is required (Hing et al., 1999). The GTP-bound active form of Rac can bind to the CRIB

domain of Pak and activate it (Newsome et al., 2000). In cells, there is the cycle of Rac

GTPase between its active GTP-bound form and its inactive GDP-bound form. Generally the transition from a GDP-bound form to a GTP-bound form is under the control of guanine nucleotide exchange factor (GEF). So theoretically, there is a specific GEF functioning to facilitate the binding of Rac to Pak in R-cell axon guidance and targeting.

In the eye-specific mosaic screen for mutations disrupting R-cell axon connectivity,

Dickson group identified a GEF called Trio, which regulates Pak activity to control R-cell axon guidance (Newsome et al., 2000).

Trio has the N-terminal conserved domain (NTD), eight Spectrin repeats, GEF domain 1, a SH3 motif, and GEF domain 2. Each GEF domain of Trio contains a Dbl homology

(DH) domain followed by a pleckstrin homology (PH) domain (Newsome et al., 2000).

This tandem DH-PH domain structure is a characteristic feature of GEFs that are specific for the Rho family GTPases. So, from the protein structure, Trio appear to be a GEF specific for the Rho family GTPases. Immunostaining shows that Trio is distributed along

R-cell axons and concentrates at the growth cones (Newsome et al., 2000).

Similar to dock and pak mutants, the trio mutant showed disrupted R-cell axon projection pattern at the third-instar larval stage with the hyperfasciculated axons and the disrupted retinotopic array in the lamina layer with the appearance of gaps and clumps (Newsome et al., 2000). At the adult stage, in addition to the highly disordered array of R7 terminals, trio mutants also show misrouting of axons from the medulla into the deeper region of the brain. Those misrouting axons passed around the posterior edge of medulla to enter the

deeper region and then turn anteriorly to project between the medulla and the underlying lobula. A similar medulla bypassing phenotype has also been discovered in either dock or pak mutants. But different from dock mutants, the trio mutant did not show any obvious deficits in R1-6 axon layer-specific targeting to the lamina at either the larval stage or the adult stage (Newsome et al., 2000). Thus, Trio is likely to control the accurate guidance of R-cell axons, while Dock functions in both R-cell axon guidance and ganglion specific targeting. trio is required in R cells but not in the target region for their proper axon guidance. The supporting evidence comes from the eye-specific mosaic mutant analysis and also the rescue of trio mutant defects by either neuron or eye-specific restoration of trio (Newsome et al., 2000).

To test that the GEF activity of Trio was involved in R-cell axon guidance Newsome et al. conducted a domain requirement analysis. They found that the GEF1 domain is indispensible for Trio to function in R-cell axon guidance (Newsome et al., 2000).

Analysis of human Trio has showed that the replacement of glutamine with alanine within the GEF1 domain drastically reduced the guanine nucleotide exchange activity of the

GEF1 domain (Liu et al., 1998). Fly trio transgene with same substitution in the GEF1 could not rescue either axon guidance defects or the lethality of the trio mutant. This suggests that the GEF activity of the Trio GEF1 domain is essential for the function of

Trio in R-cell axon guidance. By contrast, the GEF2 domain is not required for the function of trio in regulating R-cell axon guidance.

Beside the GEF1 domain, the N-terminal domain NTD of Trio is also required for Trio to function in R-cell axon guidance (Newsome et al., 2000). In other multiple GEF domain- containing proteins, removing the N-terminal domain could lead to the constitutively activation of the GEF (Katzav et al., 1991; Miki et al., 1993; Sone et al., 1997). When the

N-terminal domain and spectrin repeats were removed, this mutant Trio could induce a strong GOF phenotype. The majority of axons clumped in the optic stalk and the remaining axons followed highly aberrant trajectories in the brain. The GEF1domain alone from Trio could also induce the similar GOF phenotype. Tethering those two mutant forms of Trio to the membrane by adding a myristylation signal could further enhance this GOF phenotype. This GOF phenotype also specifically requires the GEF activity of GEF1 but not GEF2. When the glutamine within the GEF1 is replaced by alanine, the GOF phenotype was completely abolished (Newsome et al., 2000). But when the same change happened to GEF2, there was no change. Thus, the constitutively active

GEF1 of Trio can disturb the normal R-cell axon guidance. And the N-terminal domain and spectrin repeats act as the regulatory elements for proper GEF activity of the GEF1 domain in Trio, which is essential for R-cell axon guidance.

To determine if Rac was the specific GTPase target of Trio, Dickson and colleagues performed cell culture experiments. They found that in culture cells, expression of GEF1 of trio, but not GEF2, could induce membrane ruffling and lamellipodia formation similar as active Rac. The guanine nucleotide release assay also showed fly Rac proteins including Rac1, Rac2, and Mtl are Trio targets, but not other Rho family members. Direct physical binding between Rac and Pak has also been detected (Newsome et al., 2000).

Thus, Trio functions as a GEF for Rac to convert from the GDP-bound inactive form into the GTP-bound active form. This active Rac then binds to Pak and so mediate downstream signaling. This is consistent with the observation by Hing et al. that the

CRIB domain for Rac binding on Pak is essential for pak function (Hing et al., 1999).

This model is also supported by the genetic interaction between trio, dock and pak

(Newsome et al., 2000). All those data suggests that Trio acts through Rac and Pak to regulate R-cell axon guidance but not R-cell axon targeting.

1.1.7.2.6 Rho-GTPase Rac

Rho family small GTPases have been found to play important roles in regulating the actin cytoskeleton and so control the movement of the growth cone (Gallo 1998; Luo et al.,

1997; Dickson 2001). As I have described previously, some GTPase family members are found to be regulated by Dock, Pak, and Trio in R-cell axon guidance. Cdc42 and Rac can directly bind to Pak to re-organize the actin cytoskeleton (Hing et al., 1999). And Rac family members are physically associated with Trio and act as its targets to regulate R- cell axon guidance (Newsome et al., 2000).

Fly Rac family has three members, including rac1, rac2, and mtl. Double or triple mutants of Rac family members, displayed R-cell axon medulla bypassing and stalling phenotype (Hakeda-Suzuki et al., 2002). rac1 and rac2 mutations were able to suppress the GOF phenotype induced by overexpression of trio GEF1 domain in the eye (Hakeda-

Suzuki et al., 2002). This supports the idea that Rac family members Rac1 and Rac2 are targets of Trio in R-cell axon guidance. Besides the visual system, Rac family members

also function in regulating axon guidance of other systems including mushroom body,

VNC, and PNS sensory neurons (Hakeda-Suzuki et al., 2002; Ng et al., 2002). Rho

GTPase family members can modulate the activity of downstream effectors to control the actin-cytoskeleton organization and regulate the movement of the growth cone. To date, they have been found to be downstream targets of guidance receptors, such as Plexin,

Frazzled, and Robo (Luo 2002).

1.1.7.3 Cell surface proteins and secreted proteins

1.1.7.3.1 Insulin Receptor (DInR)

Insulin signalling pathway has been found to be required for viability (Fernandez et al.,

1995), longevity, general growth (Chen et al., 1996; Broqiolo et al., 2001), and female fertility (Tatar et al., 2001) in fly. dinr is expressed ubiquitously through the fly life circle.

The growth related phenotype is likely to be mediated through the fly insulin-receptor- substrate-like protein Chico (Böhni et al., 1999). To identify potential downstream binding partners of DInR, researchers used the intracellular domain of DInR as bait in the

YTH screen and identified Dock as a binding protein (Song et al., 2003). Because the function of Dock in R-cell axon guidance and targeting has been elucidated previously

(Garrity et al., 1996), Song et al. examined the possible involvement of DInR in R-cell axon guidance and targeting.

DInR is present in R-cell axon and growth cone. Loss of dinr induced similar R-cell axon guidance defects and growth cone expansion failure as that in dock mutants. Surprisingly, those defects were not associated with either abnormal R-cell growth or defect in the

target field development (Song et al., 2003). This suggests that DInR has specific function in R-cell axon guidance, which separates from its function in general growth. Consistent with this notion, mutation of chico, which normally functions as the substrate of DInR in growth controlling, did not cause defects in R-cell axon guidance.

DInR is a member of the receptor tyrosine kinase family. DInR protein has a potential insulin binding site I (DI), Cysteine rich domain (CRD), potential insulin binding site II

(DII), transmembrane domain (TM), juxtamembrane region (JM), tyrosine kinase domain, and IRS-1 (insulin receptor substrate-1) homology proline rich region (Fernandez et al.,

1995). After translation, the DInR precursor is processed proteolytically to generate an insulin-binding α subunit and a tyrosine kinase domain containing β subunit. Two α subunits and two β subunits assemble into a tetramer mature DInR complex (Fernandez et al., 1995).

Consistent with the result from the YTH assay, physical association between Dock and

DInR was shown in the Co-IP experiment with adult fly lysate (Song et al., 2003).

Genetic interaction is also observed between dinr and dock (Song et al., 2003). The binding domains are mapped to the C-terminal tyrosine phosphorylation site and the proline rich domain on DInR, and the SH2/SH3 domains on Dock. The kinase activity of

DInR is important for Dock binding because the kinase inactive form of DInR could not bind to Dock. Autophosphorylation of DInR is also required for the binding (Song et al.,

2003). Thus, the model is: ligand binding induces the autophosphorylation of tyrosine residue on DInR. The phosphorylated tyrosine can bind to the SH2 domain of Dock. The

auto-phosphorylation of DInR also induces conformation change on its C-terminus. This confirmation change provides binding site for the SH3 domain of Dock. Binding of DInR to either the SH2 or the SH3 domain of Dock can partially mediate the signal transduction to control R-cell axon guidance. This model is consistent with the previous finding that the SH2 and SH3 domains of Dock are partially redundant in R-cell axon guidance (Rao and Zipursky, 1998). Although DInR is found to act upstream of Dock in R-cell axon guidance, there is no evidence to suggest the involvement of DInR in the layer-specific targeting of R1-6 axons that is controlled by the Dock-Msn-Bif signalling pathway. So, there are perhaps some other receptors that act upstream of Dock to control the layer- specific targeting of R-cell axons.

1.1.7.3.2 N-Cadherin (CadN)

CadN mutants were identified through a series of visual behavior screens. In these screens, researchers created eye-specific mosaic mutants for EMS-induced mutations in an otherwise wild-type background to assay the visual response (Lee et al., 2001). The optomotor test was performed to assess the functions of R1-R6. In the optomotor assay, flies were placed at one end of a long clear tube. A bar of visible light moved in one direction. Wild type flies responded to this moving bar light by moving in the opposite direction. This behavior is mediated through the R1-R6-mediated visible light response

(Heisenberg and Buchner, 1977). Mutant flies having R1-R6 dysfunction could not detect the bar light and so moved randomly in the tube. Some CadN mutants were identified in this screen, suggesting the involvement of CadN in the normal functions of R1-R6 (Lee et al., 2001). The second type of assay is the UV/visible light choice test. Flies were placed

in a T-maze with green light at the end of one arm and UV light at the end of another.

Wild-type flies normally are attracted towards UV light (Reinke and Zipursky, 1988).

Mutant flies with R7 dysfunction moved to green light. CadN mutant flies were also identified in this screen, suggesting that CadN also regulates the function of R7 (Lee et al.,

2001).

CadN is a large transmembrane protein consisting of 3097amino acids. It contains 15 cadherin repeats, one fly classic cadherin box, two cysteine rich domains (CRD), one laminin A globular domain, one transmembrane domain, and a cytoplasmic tail. Iwai et al. showed that CadN is a classic Cadherin associated with α and β catenins, and capable of promoting homophilic binding in cultured S2 cells (Iwai et al., 1997). It is involved in regulating axon guidance in the fly embryonic VNC (Iwai et al., 1997).

In the fly visual system, CadN is expressed in all R cells in the eye disc (Lee et al., 2001).

At the third-instar larval stage, the R-cell axon projection pattern was severely disrupted in eye-specific CadN mosaic flies. Irregular clumps and gaps were present in the lamina.

R7 and R8 growth cones in the medulla did not properly expand and the local topographic mapping was disturbed. But the layer-specific targeting of R1-R6 axons was normal. At the adult stage, the axon terminal array of mutant R7 and R8 was disrupted, too (Lee et al.,

2001).

During the developmental process of the visual system, R1-R6 axons from the same ommatidium form a bundle to enter the optic stalk and navigate towards the brain. Once

they exit the optic stalk, axons defasciculate from each other and migrate outwards to form synapses with different lamina neurons. R1-R6 axon terminals and their target lamina neurons form lamina cartridges. Injecting fluorescent dye into single ommatidium can reveal the cartridge formation during the pupal stage (Clandinin and Zipursky, 2000;

Lee et al., 2001). CadN mutants showed severe disruption in cartridge formation with the failure of R1-R6 axons to defasciculate from each other when they reached the lamina plexus at the midpupal stage (Lee et al., 2001).

MARCM analysis revealed that wt R-cell axons normally migrate away from the original cartridges to join the target cartridges. But CadN mutant R-cell axons stayed in original cartridges and did not extend away (Prakash et al., 2005). This suggests that CadN functions in R-cell to regulate axon defasciculation and lateral extension in a cell- autonomous manner.

Because CadN functions in a homophilic fashion, CadN-expressing lamina neurons, which act as targets of R1-R6 axons, were examined for their potential involvement in cartridge formation. When CadN was removed from lamina neurons by MARCM, wt R1-

R6 axon innervations on these mutant lamina neurons were highly abnormal (Prakash et al., 2005). Further analysis also indicates that CadN is required only in target cartridges but not in original cartridges for proper R1-R6 axon targeting (Prakash et al., 2005). So the interaction between CadN proteins on R1-R6 axons and CadN proteins on target lamina neurons controls the target recognition of R1-R6 axons and lamina cartridge formation in the lamina plexus.

MARCM analysis also indicated that CadN is required cell-autonomously for the targeting of R7 axons to the M6 layer in the medulla. Loss of CadN in R7 cells induced abnormal targeting of R7 axons to the M3, which is the R8 target layer in the medulla

(Lee et al., 2001). The abnormal distribution of synaptobrevin-GFP (Estes et al., 2000) in

CadN mutant R7 axons suggests that CadN mutant R7 cells failed to elaborate correct synapses with their targets (Lee et al., 2001). In a functional analysis of R7 to control the

UV/visible light choice, PanR7-Tox, a R7-specific Tetanus Toxin light chain (Sweeney et al., 1995), was used to shut down the evoked synaptic transmission in R7 cells specifically (Lee et al., 2001). This led to the failure of R7 to respond to UV light. Wild- type mosaic flies with 15% normal R7 cells in the otherwise R7 activity-silenced background behaved normally in the UV/visible light choice assay. But when these 15%

R7 cells were deficient in CadN, flies showed abnormal behavior in the choice assay.

This also suggests that CadN is required for functional synapse formation of R7 cells (Lee et al., 2001).

To determine CadN is required at which stage for proper R7 targeting in the medulla, experiments were done to illustrate the detailed developmental process of R7 targeting during the pupal stage (Ting et al., 2005). In wt, at 17% APF, R7 and R8 growth cones already separate into two temporary layers, called R7-temporary layer and R8-temporary layer, respectively. Lamina neurons have their growth cones terminate in between these two temporary layers. At 35% APF, the termination region of lamina neurons is divided into two separated layers. R7 and R8 termination sites remain at similar locations as before. At 50% APF, R7 terminals migrate into the deeper medulla, while R8 terminals

begin to extend into the R7-temporary layer. At 70% APF, both R7 and R8 axon terminals reach their final targeting layers and complete their growth cone morphological changes (Ting et al., 2005). Thus, there are two steps for R7 and R8 target selections. The first step is the selection for the temporary layers and the second step is the selection for the final targeting layers.

At the first step, innervations of lamina neurons between the R7 and R8-temporary layers are not required for the segregation of these two layers because when lamina neurons were removed, the segregation occurred normally (Ting et al., 2005). The segregation of these two layers is also independent of each layer as the loss of R7 did not affect the formation of the R8-temporary layer (Ting et al., 2005). CadN is required at the first step for R7 axons to target to and maintain in the R7-temporary layer during the 17% to 35%

APF. Most CadN mutant R7 axons incorrectly expanded their growth cones in the R8- temporary layer. Some mutant R7 axons retracted from the R7-temporary layer to the R8- temporary layer even after they targeted to the correct layer at the first place (Ting et al.,

2005; Yonekura et al., 2006). CadN also plays an important role at the second step for R7 to migrate from the temporary layer to the final termination layer (Nern et al., 2005). I will discuss this later.

The cytoplasmic tail of CadN was found to be dispensable for its function at R7 initial target selection, although this tail is required for the maintenance of R7 axons in the target layer and their growth cone morphological changes at a later stage (Yonekura et al., 2007).

The cytoplasmic domain is also not required for the adhesive activity of CadN (Yonekura

et al., 2007). So, CadN likely functions as an adhesive protein, instead of a signalling molecule, for the initial target selection in R7 targeting. When Ncad was removed from medulla neurons by the ELF system (ey-GAL80, Lama-GAL4, UAS-FLP) (Chotard et al.,

2005), wild-type R7 axons could not reach their temporary targeting layer (Yonekura et al., 2007). These defects retained and became enhanced later in the adult stage. Thus

CadN-mediated adhesion between R7 terminals and medulla neurons likely initiate the original R7 terminal targeting. Afterwards, this targeting is maintained and stabilized through a CadN-mediated signalling pathway (Yonekura et al., 2007). In conclusion,

CadN is required both pre- and post-synaptically for correct target selection in both the lamina and medulla regions.

CadN can produce 12 different isoforms based on different transcription splicing (Ting et al., 2004). These transcription splicing sites reside in the exon 7, 13, and 18. Each site has two splicing variants as A and B. For some proteins, different isoforms possess different functions, such as the fly Brakeless proteins (LaJeunesse et al., 2001) and zebrafish

MuSK proteins (Zhang et al., 2004). For some other proteins, different isoforms may have same functions, such as the fly Dscam proteins (Millard and Zipursky, 2008).

Different CadN isoforms had similar rescue capability in lamina cartridge formation

(Prakash et al., 2005) and R7 layer-specific targeting (Ting et al., 2004). But in some other systems, different CadN isoforms may possess different functions (Nern et al.,

2005).

An isoform-specific mutation called CadN18Astop, was discovered in a behavior screen

(Nern et al., 2005). The exon 18 of CadN has two different spliced forms as the 18A and

18B. The CadN18Astop mutation disrupted all six 18A-containing isoforms, but did not affect six 18B-containing isoforms. R cells initially express CadN18B. At around 45%

APF, they shift their expressions to 18A predominantly. This is coincident with the time when R7 and R8 axons begin to migrate from their temporary layers to their final target layers (Ting et al., 2005). During the first step of targeting, CadN mutant R7 axons displayed strong defects in extension and maintenance in the temporary layer, while the

CadN18Astop showed normal R7 targeting. Interestingly, at the adult stage, CadN18Astop showed a R7 mistargeting phenotype like CadN mutants (Nern et al., 2005). This indicates that 18B is sufficient to regulate the R7-specific targeting to the temporary layer at the first step of targeting. It also suggests that 18A functions at the second step for R7 to target from the temporary layer to the final termination site. This hypothesis is also supported by the isoform expression shift from 18B at the first targeting step to 18A at the second step. Because both 18A and 18B could rescue the CadN mutant R7 phenotype, it is unclear about the importance of the isoform expression shift in R7 cells if both isoforms share similar functions. In S2 cell binding assay, 18B showed stronger adhesive activity than 18A (Ting et al., 2005). Thus, the expression shift in R7 cells from 18B to

18A may reflect the adhesion change between CadN-expressing cells at different developmental stages.

The expression pattern of CadN in lamina neurons suggests that CadN may have a function in lamina neuron connectivity formation (Nern et al., 2008). Lamina neurons are

divided into five different types, from the L1 to L5. Each type of lamina neurons targets their axons to specific medulla layer(s). MARCM analysis has shown that CadN is required cell-autonomously for the correct axon targeting of L1, L3, L4, and L5 (Nern et al., 2008). Further examination also reveals cell-type-specific and developmental-step- specific requirements of CadN in different lamina neuron subtypes. At different developmental stages, the extension and expansion of lamina neuron axons are correlated with the high CadN expressions in targeting fields. For example, L5 axons, which target to both the M2 and M5 layers, use L2 axon terminals located in the M2 layer as their targeting cues for the M2 layer-specific termination. CadN proteins in both L5 and L2 are involved in this targeting process. When wild-type L5 axons encountered CadN mutant

L2 axon terminals, the M2 layer-specific axon targeting of L5 was disrupted (Nern et al.,

2008). Here again, like its function in R-cell axon targeting in both the lamina and medulla regions, CadN is required in both projecting axons of lamina neurons and their target fields for correct axon targeting.

CadN has a close homolog in the fly genome named CadN2 (Prakash et al., 2005). The

CadN2 mutation alone did not induce any visible defects in R-cell axon projection pattern.

But it has a partially redundant function with CadN in R-cell axon targeting because the

CadN and CadN2 double mutants displayed more severe defects than the CadN single mutant in both R1-R6 lamina targeting and R7 medulla targeting (Prakash et al., 2005).

However, unlike CadN, CadN2 did not induce cell homophilic adhesion in in vitro cell aggregation assays (Yonekura et al., 2007).

1.1.7.3.3 Flamingo (Fmi)

Flamingo (Fmi) is first identified based on its function in the planar cell polarity (PCP) formation in the fly wing (Usui et al., 1999; Chae et al., 1999). Fmi is a nonclassic-type cadherin protein. It contains eight cadherin repeats, five EGF-like domains, two laminin

G domains, one hormone receptor domain, seven transmembrane domains, and a C- terminus tail (Usui et al., 1999). Different from classic cadherins, such as CadN and E-

Cadherin, Fmi does not interact with Catenins. Fmi has several homologs in some other species, including C. elegans (Pettitt 2005; Steimel et al., 2008), zebrafish (Carreira-

Barbosa et al., 2009), mouse (Curtin et al., 2003), rat (Beall et al., 2005), chick (Davies et al., 2005), and human (Wu and Maniatis, 2000). They all share the similar protein domain structures. Fmi is capable of inducing homophilic cell adhesion through its extracellular domain in cultured cells (Usui et al., 1999). It can also bind to itself in an asymmetric homotypic fashion to regulate the PCP formation in the wing (Chen et al., 2008; Strutt and Strutt, 2008). Fmi also regulates PCP formations in fly ommatidia and epidermis cells in a similar manner (Feiguin et al., 2001; Das et al., 2002 and 2004; Le Garrec and

Kerszberg, 2008; Lawrence et al., 2004). The involvement of c-Fmi-1 in PCP formation is also observed in chick inner ear hair cells (Davies et al., 2005).

In the fly peripheral nervous system, Fmi is involved in regulating dendritic growth of da neurons at the embryonic stage and mediating dendritic tiling at the late third-instar larval stage (Gao et al., 1999 and 2000; Grueber et al., 2002; Sweeney et al., 2002; Kimura et al.,

2006). It is hypothesized that to control the dendritic growth at the embryonic stage, Fmi needs to interact with an unknown ligand. But to mediate the dendritic tiling at the larval

stage, Fmi needs to recognize itself through its homophilic binding activity (Kimura et al.,

2006). Fmi also functions in PNS sensory neuron axon guidance and larval NMJ synaptogenesis (Steinel and Whitington, 2009; Bao et al., 2007). Celsr3, the mouse homolog of Fmi, also has similar functions in axon fasciculation and guidance (Tissir et al., 2005).

The involvement of fmi in R-cell axon targeting was discovered by two labs independently (Senti et al., 2003; Lee et al., 2003). Dickson and coworkers identified fmi mutations from a genetic screen for mutations disrupting R-cell axon connectivity (Senti et al., 2003). Zipursky and colleagues also isolated fmi mutations but from a behavior screen for mutations affecting the optomotor response of fly (Lee et al., 2003). fmi is expressed predominantly in R-cell growth cones and the medulla field, but not in lamina neurons. Fmi expression is transient. It is turned off at the stage of 46% APF (Lee et al.,

2003). At the third-instar larval stage, although the layer-specific targeting of R1-R6 axons appeared normal in fmi mutants, synaptic target selection of fmi mutant R1-R6 axons within the lamina was defective (Lee et al., 2003). The DiI labelling experiments showed that fmi mutant R1-R6 axons mistargeted to inappropriate cartridges. Analysis by electron microscopy showed that the lamina cartridge array in fmi mutants was disrupted.

Wild-type lamina cartridges normally contained 5-7 R-cell axon terminals. fmi mutant cartridges had variable numbers of axon terminals ranging from 3 to 15. These ectopically formed synapses had similar microstructure compared to normally formed synapses (Lee et al., 2003). These data suggest that Fmi is involved in regulating target selection but not in synaptogenesis (Lee et al., 2003).

Interestingly, MARCM analysis has revealed that fmi functions in a cell non-autonomous way to control R1-R6 lamina target selection (Chen and Clandinin, 2008). When a single fmi mutant R cell was surround by wild-type cells, its axon guidance and lamina cartridge targeting occurred normally. But if a single wild-type R cell was surrounded by fmi mutant cells, this wild-type cell showed inappropriate lamina cartridge targeting (Chen and Clandinin, 2008). So, fmi is likely to act locally in a cell non-autonomous manner to target adjacent R1-R6 axons to proper lamina cartridges. Because the cytoplasmic tail of

Fmi is dispensable for its function, Fmi probably functions as an adhesive molecule through its extracellular domain rather than acting as a signalling molecule (Chen and

Clandinin, 2008).

Fmi also functions in R8 axon projection, targeting, and the dorsolventral topographic map formation (Senti et al., 2003; Lee et al., 2003; Berger et al., 2008). Loss of fmi at the third-instar larval stage caused the abnormal hyperfasciculation of axons in the medulla region, and the disruption of the R8 axon dorsolventral topographic mapping (Senti et al.,

2003; Lee et al., 2003). At the adult stage, fmi specifically controls R8 target selection in the medulla (Senti et al., 2003). fmi mutant R8 terminals became irregularly distributed and fused with their neighbours. A complete rescue of R8 targeting defect could be achieved by an R8-specific expression of a fmi transgene, indicating that fmi is specifically required in R8 but not R7 (Senti et al., 2003).

Fmi is also expressed in the M3 layer, the target region of R8 in the medulla (Senti et al.,

2003). When fmi was removed from medulla neurons either by RNAi technique or by

ELF mosaic analysis, wild-type R8 axons could not target correctly into the M3 layer

(Bazigou et al., 2007). These data suggest that Fmi is required in both pre- and post- synaptical cells for R8 layer-specific targeting.

1.1.7.3.4 Drosophila Leukocyte common antigen-related RPTP (Dlar)

Dlar belongs to the receptor protein-tyrosine phosphatase (RPTP) superfamily (Streuli et al, 1989). It is highly expressed in the embryonic CNS and female germline cells (Tian et al., 1991; Fitzpatrick et al., 1995). Dlar is involved in regulating motor axon guidance at the embryonic stage. When Dlar was absent from the embryo, motor axons bypassed their normal target muscles (Krueger et al., 1996). Dlar is a transmembrane protein with three Ig domains and nine FN-III motifs in the extracellular domain. The intracellular portion of Dlar contains a PTPase enzymatic domain (Streuli et al., 1989). Dlar is highly homologous to other members of the fly RPTP family, including PTP4E, PTP10D,

PTP52F, PTP69D, and PTP99A.

Two groups independently identified the function of Dlar in R-cell axon targeting by examining eye-specific mosaic flies for EMS-induced random mutations in different ways

(Clandinin et al., 2001; Maurel-Zaffran et al., 2001). Clandinin et al. isolated Dlar mutants in a behavioural screen for optomotor response defective flies (Clandinin et al.,

2001). Maurel-Zaffran et al. conducted an immunohistological screen for R-cell axon projection mutants and isolated Dlar mutants (Maurel-Zaffran et al., 2001). Dlar functions cell-autonomously to stabilize R7 in its target site within the M6 layer. Loss of

Dlar induced the retraction of R7 axons from the M6 layer to the M3 layer in the medulla

(Clandinin et al., 2001).

Both the extracellular domain and the cytoplasmic tail of Dlar are required for its function in R7 cells (Maurel-Zaffran et al., 2001). The Dlar transgene with its cytoplasmic region replaced by the intracellular domain of PTP69D was fully functional in the rescue experiment (Maurel-Zaffran et al., 2001). This suggests that the PTPase activity in the intracellular region is sufficient to mediate the signaling cascade initiated by Dlar

(Maurel-Zaffran et al., 2001). The Dlar-PTP69D chimera construct could rescue both the

Dlar and PTP69D mutants. This result suggests that these RPTP family members may share a common cytoplasmic signalling pathway, but have unique ligand binding preferences. The requirement of Dlar in R7 target selection suggests that Dlar might act as a receptor cell-autonomously on R7 axons to detect an unknown ligand present in the targeting layer. Because the R8-specific expression of Dlar could partially rescue the R7 phenotype in Dlar mutants, Dlar may also function as a ligand on R8 axons to bind to its receptor on R7 axons, thus facilitating the separation of R7 terminals from R8 (Maurel-

Zaffran et al., 2001).

In addition to its function in R7 targeting, Dlar also controls the targeting specificity of

R1-6 axons to lamina neurons in the lamina plexus. Dlar mutant R1-R6 axons targeted to inappropriate lamina neurons and formed presynaptic terminals with abnormal morphology (Clandinin et al., 2001). This suggests that Dlar is also involved in synapse

formation. This discovery is consistent with the finding that Dlar functions in synapse morphogenesis of the larval NMJ (Kaufmann et al., 2002).

Two cytoplasmic proteins, Trio and Ena have been found to interact with Dlar in the visual system (Kaufmann et al., 2002). But the ligand for Dlar in the visual system is still unknown. The Zinn lab conducted a biochemical screen by using a recombinant protein

Dlar-AP to look for potential ligands for Dlar. They identified heparan sulphate proteoglycan (HSPG) Syndecan as an in vivo binding ligand for Dlar on motor axons in the embryonic VNC (Fox and Zinn, 2005). Johnson et al. also discovered that HSPGs

Syndecan and Dallylike interacted with Dlar both genetically and physically in the larval

NMJ (Johnson et al., 2006). However both HSPGs are unlikely to be the ligands for Dlar in the visual system. Although mutants of either HSPG displayed R-cell axon guidance defects (Rawson et al., 2005), their phenotypes did not resemble that of Dlar mutants.

These data suggest that in different systems, Dlar may have different binding partners.

Each of these ligands may initiate a different signalling response when interacting with

Dlar.

Some early analysis of mammalian Lar-family RPTPs showed that they all associated directly with some members of the highly conserved Liprin family at the cell-substrate interaction sites (Serra-Pages et al., 1995, 1998). Using the entire cytoplasmic domain of

Dlar as the bait, Kaufmann et al. conducted a Yeast interaction Trap screen and identified fly Liprin-α as a binding partner for fly Dlar at the larval NMJ (Kaufmann et al., 2002).

Liprin-α protein has one N-terminal coiled-coil domain and three C-terminal Steryl Alpha

motifs. The Dlar-binding site is located in the C-terminus. The physical binding between

Dlar and Liprin-α was confirmed by the Co-IP experiment (Hofmeyer et al., 2006).

Liprin-α has some similar functions as Dlar in controlling R1-R6 target selection in the lamina and R7 targeting in the medulla (Choe et al., 2006 and Hofmeyer et al., 2006). But analysis of the Dlar and liprin-α double mutant phenotype argued strongly against the notion that these two genes functioned in the same pathway (Choe et al., 2006 and

Hofmeyer et al., 2006). It is likely that they function in parallel pathways to regulate R1-

R6 and R7 axon target selections (Choe et al., 2006 and Hofmeyer et al., 2006).

1.1.7.3.5 PTP69D

PTP69D is also a member of the fly RPTP family. PTP69D was firstly identified in pulldown experiments by using an anti-Horseradish Peroxidase (HRP) polyclonal antibody to pull down target proteins. This anti-HRP antibody recognizes a carbohydrate epitope on glycoproteins selectively expressed in the insect nervous system (Desai et al., 1994 and

1996). Immunohistological analysis revealed a specific expression pattern of PTP69D in the VNC, brain, and eye disc of the fly. PTP69D protein has two Ig domains at the N- terminal, followed by three FN-III motifs, and a membrane proximal region (MPR) in the extracellular portion. The intracellular domain of PTP69D contains two functional PTP enzymatic domains. There is a proteolytic cleavage site within the MPR region, which may be required for the functional cleavage of PTP69D protein (Garrity et al., 1999).

PTP69D plays a role in embryonic motor axon guidance as the loss of PTP69D caused severe motor axon guidance defects (Desai et al., 1994 and 1996).

The requirement of the SH2/SH3 adapter protein Dock in the visual system suggests the involvement of phosphotyrosine signalling in R-cell axon guidance and targeting (Garrity et al., 1996). Garrity et al. conducted a candidate gene screen to examine previously identified genes, which were required for embryonic axon growth and involved in mediating the phosphotyrosine signalling. They found that PTP69D mutants displayed defects in R1-R6 axon targeting at the third-instar larval stage (Garrity et al., 1999). Loss of PTP69D caused a subset of R1-R6 axons to mistarget to the medulla (Garrity et al.,

1999, Newsome et al., 2000). The rescue of PTP69D mutant phenotype by eye-specific expression of a PTP69D transgene indicates that PTP69D is required in R cells for R1-6 axons to terminate in the lamina layer. The action of PTP69D requires its FN-III motif and PTPase enzymatic activity of either one of the two PTP domains at the C-terminal region (Garrity et al., 1999). So, the FN-III domain may be involved in the ligand binding, while the PTPase activity is likely involved in mediating the downstream signalling pathway.

The working model is that PTP69D, located on the surface of R1-R6 growth cones, detects a stop signal in the developing lamina plexus and subsequently converts the signal into the shutdown of growth cone motility. Only PTP69D on R1-R6 growth cones was capable of responding to the stop signal, because PTP69D overexpression in R7 cells could not induce premature termination of R7 axons in the lamina (Garrity et al., 1999).

These data suggest that PTP69D may need to interact with some R1-R6 specific proteins in order to respond properly to the stop signal. To date, the identity of the stop signal in the lamina plexus as the ligand for PTP69D is still unknown.

PTP69D also controls R7 targeting at the pupal stage. In PTP69D mosaic mutants, instead of targeting to the M6 layer in the medulla, R7 axons terminated into the R8 target layer in the M3 (Newsome et al., 2000). Based on the phenotype observed in PTP69D mutants, Dickson and colleagues have hypothesized that PTP69D is involved in reducing the adhesion of R1-R7 axons to the pioneer R8 axon and so R1-R7 axons are able to defasciculate from the R8 axon and respond independently to their own targeting cues.

This is different from the hypothesis provided by Garrity et al. that PTP69D acts as a receptor to detect the stop signal. At this moment, it is unclear which model is correct.

1.1.7.3.6 Capricious (Caps) caps was firstly identified in a P-lacZ enhancer trap (Hartenstein and Jan, 1992) screen for candidate genes potentially involved in NMJ synapse formation (Shishido et al., 1998).

Caps is a transmembrane protein, containing fourteen Leucine-rich repeats (LRR) in the extracellular region and a conserved intracellular juxtamembrane domain. LRR is widely distributed in different types of proteins and is involved in mediating protein-protein interactions. A recent discovery from a genetic screen suggests that LRR proteins are key mediators in synaptic target selection at the fly larval NMJ sites (Kurusu et al., 2008). The closest homolog of Caps in fly is Tartan, which also has 14 LRRs and shares some similar functions with Caps in NMJ synaptogenesis, wing boundary formation, tracheal morphogenesis, and retinal epithelial integrity (Chang et al., 1993; Shishido et al., 1998;

Bugga et al., 2009; Milan et al., 2001 and 2005; Krause et al., 2006; Mao et al., 2008). In addition to those functions shared with Tartan, Caps has its unique functions in motor

axon guidance (Abrell et al., 2001) and R8 layer-specific targeting in the visual system

(Shinza-Kameda et al., 2006).

At the third-instar larval stage, caps is expressed exclusively in R8 cells in the retina. It is also expressed in the medulla neuropil where R8 axon terminals target to (Shinza-

Kameda et al., 2005). Phenotypic analysis supports the idea that Caps functions in a cell- autonomous way to control the targeting of R8 axons to the M3 layer in the medulla.

When caps was absent, R8 axons bypassed their normal termination sites in the M3 and targeted to the R7 termination site in the M6 (Shinza-Kameda et al., 2005). However, at the early pupal stage, the initial targeting of R8 axons occurred normally in caps mutants.

The defects started to emerge only at 40 hrs APF. This is the stage when R7 and R8 axons start to migrate from their temporal termination layers to their final target layers (Shinza-

Kameda et al., 2005). It is likely that Caps functions specifically in R8 for its final target layer recognition.

Supporting this idea, when caps was ectopically expressed in R7, it was capable of redirecting R7 terminals to the R8 target layer in the M3 (Shinza-Kameda et al., 2005).

Because Caps could induce homophilic cell adhesion in cultured S2 cells, it may function as a homophilic adhesion molecule (Shinza-Kameda et al., 2005). So, the R7 redirecting phenotype could be induced by the ectopic adhesion of R7 terminals to R8 by caps overexpression in R7 cells. This adhesion inhibited the defasciculation of R7 terminals from R8, thus forcing R7 terminals to stop at the R8 termination layer. However, it was shown that redirected R7 terminals were able to defasciculate from R8 terminals and

target to their temporary target layer at 20 hrs APF (Shinza-Kameda et al., 2005). So, these data suggest that Caps on R8 growth cones may interact with some targeting cues in the M3 layer to direct the precise termination of R8 axons. The expression of caps in the

M3 layer and the capability of Caps to induce homophilic cell adhesion suggest that Caps proteins on R8 growth cones probably recognize Caps proteins on target neurons in the

M3 layer for R8 targeting.

1.1.7.3.7 Golden Goal (Gogo)

Gogo is a newly identified transmembrane protein (Tomasi et al., 2008). It has a GOGO domain (Gogo conserved domain), followed by a thromospondin1 (Tsp1) domain and a

CUB domain in the extracellular region (Tomasi et al., 2008). The long cytoplasmic tail of Gogo does not show any obvious homology to known protein motifs (Tomasi et al.,

2008). The association between the GOGO domain and the Tsp1 domain is well conserved in homologs of Gogo in C. elegans and mouse (Tomasi et al., 2008). But the

CUB domain is unique for fly Gogo. Both Tsp1 and CUB domains are involved in mediating protein-protein interaction and cell-cell communication (Adams and Tucker

2000; Bork and Beckmann, 1993; He and Tessier-Lavigne, 1997; Takagi et al., 1991).

gogo mutants were isolated from a large-scale eye-mosaic genetic screen for genes involved in R7 and R8 targeting (Newsome et al., 2000; Tomasi et al., 2008). gogo has a dynamic expression pattern in the visual system. At the larval stage, gogo is expressed in

R8 cells in the eye disc and in the medulla region of the brain. At the early pupal stage, gogo is expressed in all R cells and potentially in the medulla region. From the midpupal

stage, gogo expression begins to decrease to a relatively low level (Tomasi et al., 2008).

The dynamic expression of gogo may be correlated with the function of Gogo at different developmental stages.

Phenotypic analysis indicates that gogo functions cell-autonomously to control R8 axon pathfinding and layer-specific targeting at both the larval and pupal stages (Tomasi et al.,

2008). gogo mutant larvae showed tangled R8 axon bundles in the medulla region. At the adult stage, gogo mutants displayed an incomplete medulla rotation with abnormal thick axon bundles bypassing an ectopic chiasm at the posterior side of the lamina (Tomasi et al., 2008).

MARCM analysis showed that during the 3rd instar larval stage, gogo mutant R8 axons could not separate from each other when they innervated the medulla region (Tomasi et al., 2008). This occurred only when mutant R8 axons were next to each other. If a mutant

R8 axon was in between two wild-type R8 axons, it did not fuse with its wild-type neighbours (Tomasi et al., 2008). This phenotype suggests that gogo mediates the repulsive interaction between adjacent R8 axons for their proper spacing.

At the adult stage, gogo mutant R8 axons showed severe mistargeting errors. Mutant R8 axons terminated in either the outer or the deeper medulla layers with hyperfasciculated bundles. The function of gogo is likely only required in R8 cells because gogo mutant R7 axons targeted correctly to the M6 layer at the adult stage (Tomasi et al., 2008).

Consistent with this notion, gogo overexpression in all R cells could retarget R8 terminals

to the superficial M1 layer in the medulla, but allowed R7 terminals to target normally to the M6 layer (Tomasi et al., 2008). These data also indicates that at the pupal stage, gogo likely functions differently from its role at the larval stage. Gogo probably detects some unknown pathfinding cues at the superficial M1 layer when R8 axons first reach the medulla region. By detecting these cues, R8 cells turn off gogo expression and so their axons become insensitive to these guidance signals. This allows R8 axons to project through the M1 layer to enter deeper medulla layers and finally reach their target region.

When this shutdown in gogo expression was prevented by the overexpression of a gogo transgene, R8 axons became constantly sensitive to the unknown guidance cues and so stayed in the M1 superficial layer without migrating further to their final targeting region

(Tomasi et al., 2008).

The GOGO domain, Tsp1 motif, and the C-terminal tail were all required for Gogo to function in R8 layer-specific targeting (Tomasi et al., 2008). Both the heterotypic function of Gogo observed in R8 cells, and the requirement of extracellular and intracellular domains of Gogo for its normal function suggest that Gogo interacts with some unknown molecules for R8 pathfinding and targeting. Consistent with this hypothesis, Gogo did not display homophilic binding activity in cultured cells (Tomasi et al., 2008). Because the GOGO and Tsp1 domains are required for the function of gogo, it will be interesting to identify the binding partner(s) of Gogo by using these domains as baits in a YTH screen or in some other biochemical screens.

1.1.7.3.8 Anaplastic lymphoma kinase (Alk) & Jelly Belly (Jeb)

Alk is a receptor tyrosine kinase originally identified as the protein product of an oncogene (Morris et al., 1994; Pulford et al., 2004). It belongs to the Insulin receptor family and functions in fly midgut development (Lorén et al., 2001, 2003) and visceral muscle fusion (Englund et al., 2003). Homologs of Alk have been identified in human, worm, mouse, and chick (Lorén et al., 2001; Iwahara et al., 1997; Liao et al., 2004;

Hurley et al., 2006; Vernersson et al., 2006). Alk is also expressed in the developing CNS of fly (Loren et al., 2001).

Jeb has functions in mesoderm migration and differentiation (Weiss et al., 2001). It acts as the ligand for Alk in fly visceral muscle pioneer specification and visceral muscle fusion (Englund et al., 2003; Lee et al., 2003; Stute et al., 2004).

Alk is a transmembrane protein containing one N-terminal type A low density lipoprotein receptor domain (LDLa), two Meprin/A5-protein/PTPmu domains (MAM), one Glycine- rich region, a transmembrane domain, and a C-terminal receptor tyrosine kinase domain

(Lorén et al., 2001). Jeb is a secreted protein with a type A LDL receptor repeat located close to the C-terminal (Weiss et al., 2001). Similar as Alk, jeb is expressed in the developing fly CNS. The interaction between Alk and Jeb has been previously discovered in both in vitro and in vivo binding assays (Englund et al., 2003; Lee et al., 2003).

To determine the potential involvement of this Alk-Jeb signalling pathway in fly nervous system development, Bazigou et al. looked into the functions of Alk and Jeb in the fly visual system (Bazigou et al., 2007). Alk is mainly expressed in the lamina and medulla

regions, which are the target fields for R-cell axons. Removal of Alk from R cells did not induce obvious defects in R-cell axon targeting. But loss of Alk in lamina neurons caused

R1-R6 axons to target to inappropriate lamina cartridges. When Alk was absent from medulla neurons, R8 displayed severe targeting defects. Some R7 axons also showed targeting errors, which were secondary defects to the R8 mistargeting phenotype

(Bazigou et al., 2007). These genetic analyses suggest that Alk functions in lamina neurons to control the targeting of R1-R6 axons to the correct lamina cartridges (Bazigou et al., 2007). It also functions in medulla neurons to control R8 layer-specific targeting in the medulla (Bazigou et al., 2007).

In Alk mutants, R8 axons started to show targeting defects at 55 hrs APF, correlated with the migration of R8 growth cones from the temporary targeting layer to their final termination sites in the M3 layer. This suggests that Alk probably functions in the M3 layer to interact with proteins on R8 growth cones for correct targeting.

Overexpressing a cytoplasmic-domain truncated Alk specifically in R cells could induce

R8 targeting defects similar to that in Alk mosaic mutant in the target regions (Bazigou et al., 2007). These truncated Alk proteins on R-cell axon surface might bind to the endogenous ligands of Alk. This binding could affect the releasing of these ligands from

R8 axon terminals, and so to affect the binding of these ligands to Alk proteins in the M3 layer. This result supports the hypothesis that Alk functions as a receptor in lamina and medulla neurons to recognize signals from R-cell axon terminals.

The expression pattern of jeb is complementary to that of Alk in the fly visual system. jeb is expressed at a high level in R-cell growth cones during the larval and first half of pupal stages. At the midpupal stage, jeb expression decreases to a low level and persists to the adult stage (Bazigou et al., 2007). Consistent with this expression pattern, Jeb functions in

R cells to control the targeting of R1-R6 axons to lamina cartridges and direct R8 axons to the M3 layer (Bazigou et al., 2007). When jeb was removed from the eye, defects in either the lamina cartridge targeting or the R8 layer-specific termination fully resembled those of Alk mutants (Bazigou et al., 2007). Eye-specific expression of a jeb transgene could fully rescue the jeb mutant phenotype. But when jeb was removed from the target regions, R-cell axon targeting was largely normal (Bazigou et al., 2007).

Same as Alk mutants, jeb mutants started to show axon targeting defects at 55 hrs APF

(Bazigou et al., 2007). These results suggest that Alk likely acts as a receptor in target regions to recognize Jeb released from R-cell axon terminals. Signaling pathways may be initiated by them to control the targeting of R1-R6 axons to lamina cartridges and the termination of R8 axons to the M3 layer.

The Jeb-Alk signalling pathway controls R-cell axon targeting likely by regulating the expression levels of some cell surface proteins (Bazigou et al., 2007). These proteins include Fmi, Dumbfounded/ Kin of irre (Duf/Kirre), and Roughest/Irrec (Rst/Irrec)

(Bazigou et al., 2007). All three proteins showed localizations in the M3 layer, where R8 axons terminate (Bazigou et al., 2007). fmi is involved in the layer-specific targeting of

R8 axons (Senti et al., 2003). Loss of fmi specifically in medulla neurons could induce

severe R8 targeting defects (Bazigou et al., 2007). It is possible that the Jeb-Alk signaling pathway regulates R8 targeting through fmi. Although Duf/Kirre is the downstream effector of the Jeb-Alk signalling pathway in the fly visceral mesoderm (Englund et al.,

2003; Lee et al., 2003; Stute et al., 2004), no report about its function in R-cell axon targeting is available. The function of Rst/Irrec in R-cell axon targeting is also unclear.

Further works need to be done in characterizing Duf/Kirre and Rst/Irrec mutants to help us understand the complete function of the Alk-Jeb signalling pathway in controlling R8 specific targeting.

1.1.7.3.9 Off-track (Otk) & Semaphorin-1a (Sema-1a)

Otk was originally shown to function in embryonic motor axon guidance by forming a receptor complex with PlexinA (Winberg et al., 2001). Otk is a transmembrane protein with six Ig domains in the extracellular region. It also has a cytoplasmic tyrosine kinase domain, which is likely to be a non-functional kinase domain (Pulido et al., 1992). In vitro studies showed that Otk was capable of promoting homophilic cell adhesion in cultured cells (Pulido et al., 1992).

At the third-instar larval stage, Otk is expressed in R1-R6 axons and control their ganglion-specific targeting in the lamina plexus (Cafferty et al., 2004). When Otk was specifically removed from the eye, a subset of R2-R5 axons mistargeted to the medulla region and this mistargeting phenotype persisted into the adult stage (Cafferty et al.,

2004). Eye-specific rescue confirmed the requirement of Otk in R cells instead of target region or lamina glial cells to control the layer-specific targeting of R1-R6 axons

(Cafferty et al., 2004). Overexpression of Otk in R7 cells was not sufficient to retarget their axons to the lamina layer, further supporting the notion that Otk controls lamina- specific targeting only in a subset of R1-R6 axons (Cafferty et al., 2004).

There are two possibilities for the mistargeting phenotype of R1-R6 axons in Otk mutants.

The first possibility is that Otk protein functions as a receptor to recognize an unidentified stop signal in the lamina plexus to induce the termination of R1-R6 axons (Cafferty et al.,

2004). The second possibility is that Otk is involved in producing repulsive force for R1-

R6 axons to defasciculate from the pioneer R8 axons and so they can respond individually to their targeting signals. Consistent with the second possibility, analyses of the motor axon targeting showed that Otk was involved in the receptor complex formation with

PlexinA to mediate the repulsive response of motor axons to Sema-1a (Winberg et al.,

2001).

As Sema-1a acted as the ligand for the receptor complex of Plexin A and Otk in the embryonic neuromuscular system (Winberg et al., 2001), Cafferty et al. examined the potential involvement of Sema-1a in R-cell axon guidance and targeting (Cafferty et al.,

2006). Sema-1a is expressed in developing R-cell axons and concentrates in R1-R6 growth cones at the lamina layer (Cafferty et al., 2006). When sema-1a was removed from the eye, the ganglion specific targeting of R1-R6 axons was normal. This is different from the Otk mutant phenotype, suggesting that Sema-1a probably does not function in the same process as Otk to control the ganglion-specific targeting in the fly visual system

(Cafferty et al., 2004).

In their analyses, Cafferty et al. found that sema-1a mutants did show defects in the local topographic patterning within the lamina layer (Cafferty et al., 2006). Wild-type R1-R6 axons terminate sharply between two layers of lamina glial cells, the epithelial glia and the marginal glia to form a condensed terminal layer (Poeck et al., 2001). Sema-1a mutant axons failed to stop in between these two glial layers. They projected further into the medulla glia layer, which was located below the marginal glia. Some mutant axons even extended laterally in between these two glial layers (Cafferty et al., 2006). This finding suggests that Sema-1a controls the axon-axon interaction at the lamina termination site for proper spacing between axon terminals. This hypothesis was further supported by the hyperfasciculation of R-cell axons induced by the eye-specific overexpression of sema-1a

(Cafferty et al., 2006).

The cytoplasmic domain of Sema-1a was required for its function in the visual system

(Cafferty et al., 2006). The cytoplasmic-domain truncated form of sema-1a could neither rescue the mutant phenotype nor induce the hyperfasciculation of axons (Cafferty et al.,

2006). These results suggest that Sema-1a probably functions as a receptor instead of a ligand in the visual system (Winberg et al., 2001). This is consistent with the finding of the bi-directional signalling of Sema-1a during the synapse formation between the giant fiber (GF) neurons and the TTM jump motoneurons in the adult fly ventral ganglion. In this process, Sema-1a has double roles. It acts as a receptor in the presynaptic cells to detect signals. Simultaneously, it also functions as a ligand to bind to receptors in the postsynaptic cells to control the synaptogenesis (Godenschwege et al., 2002).

Thus, Otk and sema-1a are likely involved in regulating different aspects of R-cell axonal projection. Otk controls the ganglion-specific targeting of R1-R6 axons, while Sema-1a functions in the topographic mapping and axon-axon interaction within the lamina plexus.

1.1.7.3.10 Dfrizzled2 (Dfz2) & DWnt4

Wnt family is a well conserved protein family, which has been intensively studied. Its family members are present in almost all species from invertebrates to vertebrates. Wnt proteins are involved in different aspects of developmental processes, including the early embryonic polarity patterning and body axis specification, cell proliferation and differentiation, nervous system development and function, cancer formation, immune response, and stem cell cytology (Wodarz and Nusse, 1998; van Amerongen et al., 2008;

Mikels and Nusse, 2006;). Wnts are all secreted, cysteine-rich proteins with highly conserved Wnt1 domains. There are several different types of Wnt receptors, including

Frizzled (Fz) and its coreceptor Arrow/LRP5/6, Derailed/RYK, FRL1/Crypto, Ror, and

MuSK (van Amerongen et al., 2008; Gordon and Nusse, 2006; Green et al., 2008). Wnt signalling pathways mediated through the Fz proteins have been intensively investigated.

They can be divided into three classes based on their different downstream effectors, including: the canonical pathway, the planar cell polarity (PCP) pathway, and the

Wnt/Ca2+ pathway. In the fly, the first two pathways have been well characterized and are implicated in a variety of developmental processes (Strutt 2003).

The fly Wnt family has seven members, including Wingless (Wg), DWnt2, DWnt3/5,

DWnt4, DWnt6, DWnt8/D, and DWnt10 (Llimargas and Lawrence, 2001). Except

DWnt6 and DWnt10, all other fly Wnts have been genetically characterized. They function in regulating the segment polarity formation, wing and eye induction and development, synaptogenesis, muscle development, axon guidance, trachea formation, planar cell polarity formation, immunity, and cell migration (Siegfried and Perrimon,

1994; Howes and Bray, 2000; Strutt and Strutt, 1999; Kozopas et al., 1998; Kozopas and

Nusse, 2002; Llimargas and Lawrence, 2001; Yoshikawa et al., 2003; Cohen et al., 2002;

Sato et al., 2006; Inaki et al., 2007; Gordon et al., 2005; Ganguly et al., 2005 ). Two fly

Wnt proteins, DWnt4 and DWnt5 could control axon guidance in the visual system and the ventral nerve cord, respectively (Sato et al., 2006; Yoshikawa et al., 2003).

In the fly visual system, DWnt4 mRNA is expressed by the anterior-most two or three rows of lamina neurons located in only the ventral half of the lamina (Sato et al., 2006).

DWnt4 proteins bind to ventral R-cell axons immediately after they exit the optic stalk, but don‟t bind to those dorsal axons (Sato et al., 2006). This suggests that DWnt4 may have a role in controlling the retinotopic projection of ventral R-cell axons. When DWnt4 was removed from the fly, instead of projecting to the ventral lamina, ventral axons projected to the dorsal half of the lamina (Sato et al., 2006). This misprojection also occurred when DWnt4 was ectopically expressed in both the ventral and dorsal lamina regions (Sato et al., 2006). These data suggests that DWnt4 in the ventral lamina region may guide ventral R-cell axons to their target sites in the ventral half of the lamina plexus.

For its function in the ventral retinotopic patterning, DWnt4 needs its receptor Dfz2 and downstream signalling molecules including Dishevelled and Hemipterous (hep) (Sato et

al., 2006). When either Dfz2 or dsh was removed, ventral R-cell axons displayed a dorsal projecting phenotype similar as that of DWnt4 mutants. Hep, a JNK kinase (Glise et al.,

1995; Sluss et al., 1996), showed a strong genetic interaction with DWnt4 (Sato et al.,

2006). This interaction suggests the involvement of the non-canonical signaling pathway in this ventral R-cell axon topographic patterning controlled by DWnt4.

The dorsal axon topographic patterning is controlled by a transcription factor, iroquois

(iro) (Sato et al., 2006). iro is exclusively expressed in the dorsal half of the eye disc

(Sato et al., 2006). Mutations of iro caused dorsal axons to project ventrally, opposite to the mutant phenotype of DWnt4 (Sato et al., 2006). As a transcription factor, iro may regulate the expressions of some downstream genes to control the dorsal-to-dorsal topographic mapping. In a recent genetic screen for mutations affecting R-cell axon wiring in the fly visual system by eye-specific mosaic analyses, two genes enoki mushroom (enok) and Br140 have been found in regulating the topographic patterning of

R-cell axons (Berger et al., 2008). Both Enok and Br140 are involved in mediating protein acetylation and acetylated protein binding (Scott et al., 2001; Berger et al., 2008).

In future studies, it will be interesting to identify the interacting molecules of Iro, Enok, and Br140.

1.1.7.3.11 Transforming Growth Factor β (TGF-β)/Activin pathway

TGF-β signalling controls a variety of developmental processes, including cell proliferation, recognition, differentiation, and apoptosis during different developmental stages (Shi and Massagué, 2003). As ligands of the TGF-β signalling pathways, there are

two large families: the TGF-β/Activin/Nodal family and the Bone Morphogenetic Protein

(BMP)/ Growth and differentiation factor (GDF)/ Muellerian inhibiting substance (MIS) family. The fly TGF-β subfamily has four members as dActivin, Dawdle, Myoglianin, and Maverick (Lo and Frasch, 1999, Nguyen et al., 2000, Parker et al., 2006, Serpe and

O‟Connor, 2006, Zhu et al., 2008). The fly BMP subfamily has three members including

Dpp, Gbb, and Screw. As receptors of the TGF-β signalling pathways, there are also two large families, the type I and type II receptors. Both types are receptor serine/threonine kinases with small differences on their sequence. Type II receptors can phosphorylate and activate type I receptors (Shi and Massagué, 2003). In general, the binding of a ligand to the type II receptor can activate the receptor. This activation then leads to the phosphorylation of the type I receptor by the activated type II receptor. Afterwards, the ligand, type I and type II receptors form a complex to propagate the signal through the phosphorylation of Smad proteins. Smad proteins are transcription factors, which are capable of regulating expressions of specific downstream genes (Shi and Massagué,

2003). In the fly, TGF-β signalling pathways have also been intensively investigated for their involvements in the embryogenesis, wing development, synaptogenesis, and neuronal remodelling (Clarke and Liu, 2008; Marqués 2005; Tabata and Takei, 2004;

Zheng et al., 2003).

In the fly adult medulla region, axon terminals of R cells and lamina neurons are organized in column-like structures. Axons within each column restrict their terminals to this single column without invading neighbouring columns. From a large-scale

UV/visible light choice behavioral screen for mutations affecting R7 function, mutant

flies of importin-α3 (imp-α3) and baboon (babo) were isolated (Ting et al., 2007). R7 axons with the mutation of either gene failed to restrict their terminals in the same column, but instead invaded neighbouring columns (Ting et al., 2007). Detailed analyses have revealed that both imp-α3 and babo were required for R7 axon tiling when R7 axons migrated from their temporary target layer to the final termination region (Ting et al.,

2007). Imp-α3 is the nuclear import adaptor protein involved in regulating the nuclear import of the transcription factor dSmad2 (Xiao et al., 2003; Ting et al., 2007). Babo is a type I receptor located on the cell membrane to respond to Activin signals (Wrana et al.,

1994; Brummel et al., 1999). In some following analyses, similar axon tiling defects were also observed when the function of either Activin or dSmad2 was disrupted (Ting et al.,

2007).

These genetic analyses indicate that the Activin signalling pathway functions cell- autonomously to control axon motility for R7 axonal tiling. The retrograde signaling from the axon terminals to the nuclei is required in this process (Ting et al., 2007). In this model, an unknown signal, which mediates repulsive interactions between neighbouring

R7 growth cones, is needed together with this Activin signalling pathway to control R7 axonal tiling (Ting et al., 2007).

1.1.8 Review of genes involved in glia and target field development

1.1.8.1 Genes involved in glia development, migration, and function

1.1.8.1.1 Decapentaplegic (Dpp), Hedgehog (Hh), and Gilgamesh (Gish)

Glial cells have important functions during the development of the fly visual system.

Retina basal glial cells (RBG), which differentiate and migrate from the brain into the retina, have critical roles in guiding R-cell axons to enter the optic stalk (Rangarajan et al.,

1999). Lamina glia cells (LG) located in the target region of R1-R6 axons are essential for the ganglion-specific targeting of R1-R6 axons (Poeck et al., 2001).

Analyses suggest there is a bi-directional communication between potential RBG cells and differentiating R cells during the early developmental stage. RBG precursor cells are located in the glial precursor cell regions (GPC) of the brain. After their final differentiation, glial cells migrate from the brain into the optic stalk, which connects the eye disc to the brain. The eye-disc-derived diffusion protein Dpp is important for the differentiation and migration of RBG glia (Rangarajan et al., 2001). An unknown chemoattractant from the differentiating R cells is required for the final entering of RBG from the optic stalk into the eye disc (Rangarajan et al., 1999). This final entering of RBG is independent of R-cell axon projections. As long as there are some differentiating R cells, no matter they have axon projections or not, RBG can successfully enter the eye disc (Rangarajan et al., 1999).

Hh, which is also an eye-disc-derived secreted protein, controls the proper time for RBG to enter the eye disc from the optic stalk (Rangarajan et al., 2001; Hummel et al., 2002). If the Hh pathway was disrupted, RBG entered the eye disc prematurely (Hummel et al.,

2002). This is similar to the defects found in gish mutants (Hummel et al., 2002). gish encodes a Casein kinase Iγ. It functions at the most posterior margin of the eye disc to

control the time point of the entering of RBG glia from the optic stalk into the eye disc

(Hummel et al., 2002). Currently, it is still unclear how Hh and Gish regulate the entering of RBG.

RBG cells in the eye disc have critical roles in guiding R-cell axons to enter the optic stalk (Rangarajan et al., 1999 and 2001; Hummel et al., 2002). If RBG cells failed to enter the eye disc, R-cell axons could not project into the optic stalk (Hummel et al., 2002). If

RBG cells were ectopically located in the anterior part of the eye disc, they could induce

R-cell axons to project in the wrong direction (Rangarajan et al., 1999 and 2001; Hummel et al., 2002). These results suggest RBG probably serve as intermediate targets, like the

„guidepost cells‟ in the grasshopper embryonic leg disc, to guide R-cell axons to the optic stalk. Further investigations on molecules involved in the interactions between RBG cells and R-cell axons for proper axon gauidance will be necessary.

1.1.8.1.2 Wingless (Wg), Dpp, and Glial cell missing (Gcm)

At the third-instar larval stage, in the target region of R1-R6 axons, there are three different layers of glial cells: the epithelial glia, the marginal glia, and the medulla glia.

The lamina plexus is located in between the epithelial glial layer and the marginal glial layer. These glial cells have critical functions for the layer-specific targeting of R1-R6 axons. When they were removed from the target region by genetic manipulation, R1-R6 axons projected through the lamina to enter the medulla region (Poeck et al., 2001).

These glial cells all originate from the GPC areas, which are located at the most dorsal and ventral edges of the R cell projection field on the surface of the optic lobe (Poeck et al., 2001). After birth, they migrate from the GPC areas to their final position along the lamina plexus (Perez and Steller, 1996; Poeck et al., 2001; Dearborn and Kunes, 2004). wg is expressed in the most posterior domains of GPC areas to induce the differentiation and maturation of all three types of glia by regulating the expressions of some other glia- specific genes, including omb, dpp, and dachsous (Dearborn and Kunes, 2004). Wg also controls the migration of glia from the GPC regions to the lamina plexus along a Wg- positive axonal scaffold (Dearborn and Kunes, 2004).

Dpp-expressing cells are located at the dorsal and ventral margins of the GPC areas, just adjacent to these wg-expressing cells (Yoshida et al., 2005). Inhibition of Dpp signalling in GPC caused defects in glial cell development and errors in R1-R6 axon targeting

(Yoshida et al., 2005). Dpp controls the expressions of gcm and gcm2, which regulate the final specification of glial cells in the GPC areas (Chotard et al., 2005). Mutations of both genes led to the loss of lamina glia and overshooting of R1-R6 axons into the medulla, like the dpp mutant phenotype (Chotard et al., 2005). Thus, Wg, Dpp, and Gcms function together to control the differentiation, maturation, and migration of lamina glia, which are essential for the layer-specific targeting of R1-R6 axons.

1.1.8.1.3 Non-stop (Not) with SAGA complex and Histone H2B

At the third instar larval stage, epithelia glia, marginal glia, and medulla glia displayed a characteristic triple-layer localization around the lamina plexus inside of the optic lobe

(Poeck et al., 2001). The migration and localization of these glial cells to the lamina layer is controlled by a group of proteins composing the SAGA complex (Spt-Ada-Gcn5- acetyltransferase) (Martin et al., 1995; Poeck et al., 2001; Weake et al., 2008).

Non-stop (Not) is an ubiquitin-specific protease. It is the first component of the SAGA complex identified as a regulator of glia migration and positioning in the lamina region

(Martin et al., 1995; Poeck et al., 2001). Not controls the migration of glial cells from the edge of the GPC, along the lamina plexus, to their final destinations on either side of the lamina layer. When not was absent, lamina glia accumulated at the dorsal and ventral margins of the R-cell axon projection field, and R1-R6 axons bypassed the lamina to invade the medulla (Poeck et al., 2001). not is directly required for the deubiquitination of histone H2B to control the transcriptions of some downstream genes (Weake et al., 2008).

Following not, the entire SAGA complex, as a coactivator of gene transcription, was implicated in controlling glial cell migration and positioning in the lamina region (Weake et al., 2008). Other components of the SAGA complex, when mutated, also caused similar defects in glia migration and R1-R6 axon targeting like that in the not mutants. Because the SAGA complex is a coactivator of gene transcription, the downstream target genes of this complex may be involved in regulating the migration and distribution of glial cells along the lamina plexus. Further microarray analysis on mutants of the SAGA complex components may help to identify these downstream target genes (Weake et al., 2008).

1.1.8.1.4 JAB1/CSN5

Different from the function of the SAGA complex in glial cells to control glia migration and positioning, gene JAB1/CSN5 functions in R cells to regulate the same process (Suh et al., 2002). JAB1/CSN5 is a subunit of the COP9 signalosome (CSN) complex. CSN is widely distributed in plants and animals to mediate signalling pathways (Chamovitz et al.,

1996; Chamovitz and Segal, 2001; Kim et al., 2001). When CSN5 was absent from R cells, glia could not migrate from the GPC areas to their final position along the lamina plexus (Suh et al., 2002). Researchers speculate that CSN5 may participate in some signalling pathways in R cells to regulate the expressions of downstream genes. Products of these genes may mediate the communication between R cell growth cones and glial cells to instruct the migration and localization of glia along the lamina plexus (Suh et al.,

2002).

Thus in the fly visual system, the development of glial cells and R cells are mutually dependent. The communication between these two different cell types is essential for the establishment of R-cell-to-optic-lobe connections.

1.1.8.2 Genes involved in the lamina-lobula boundary formation

1.1.8.2.1 Robo and Slit

Cells located in the optic lobe begin to proliferate soon after larvae hatching. These cells divide into two different groups of progenitor cells and they are located to the inner proliferation centre (IPC) and outer proliferation centre (OPC), respectively

(Meinertzhagen and Hanson, 1993; Younossi-Hartenstein et al., 1996). The IPC generates the inner medulla, the lobula and the lobula plates. The OPC generates the outer medulla

and the lamina neurons. The lamina, medulla, lobula, and lobula plate are the four ganglia in the fly optic lobe (Meinertzhagen and Hanson, 1993). Descendents from different anlagens lie adjacent to one another to form distinct compartments in the brain without mingling with each other (Meinertzhagen and Hanson, 1993). The compartment establishment of each ganglion is essential for the proper neural network formation during different developmental processes. For example, the boundary formation between the lamina and the lobula cortex is critical for the termination of R1-R6 axons in the lamina layer (Tyler et al., 2004).

The function of the Slit-Robo signalling pathway was originally characterized in axon midline crossing in the fly embryonic VNC (Kidd et al., 1999; Brose et al., 1999; Brose and Tessier-Lavigne, 2000). Slit protein has four leucine rich repeats (LRRs), seven epidermal growth factor (EGF)-like motifs, one laminin G domain, and a C-terminal cysteine-rich domain (CRD) (Rothberg et al., 1988). In the fly visual system, slit is expressed in a subpopulation of cells in the basal lamina and medulla regions. Secreted

Slit proteins form a continuous thick layer spanning from the lamina to the medulla neuropil, as a boundary between the lamina/medulla and the lobula cortex (Tyler et al.,

2004). Distal cell neurons, as a part of the lobula cortex, are located next to the posterior edge of the lamina and so adjacent to this Slit protein field. When slit was removed from the lamina and medulla neuropil, distal cell neurons invaded the base of the lamina neuropil. This invasion caused the disorganization of lamina glia, which led to the passing through of R1-R6 axons (Tyler et al., 2004).

The fly robo family has three members as robo, robo2, and robo3. All three Robo proteins have five Ig domains and three FN-III motifs in the extracellular region. Robo has four highly conserved cytoplasmic motifs from CC0 to CC3, while Robo2 and Robo3 have only CC0 and CC1 (Kidd et al., 1998; Simpson et al., 2000a, 2000b). Robos are receptors for Slit proteins (Kidd et al., 1999; Simpson et al., 2000a, 2000b). robo family members are expressed in overlapping patterns in the brain. Interestingly, all three robo genes show expression in distal cell neurons, which are located adjacent to the lamina plexus where slit is expressed (Tyler et al., 2004). When all three robo genes were simultaneously knocked down in distal cell neurons, these neurons no longer sensed the repulsive Slit proteins in the lamina neuropil. They invaded the lamina and led to the mislocalization of lamina glial cells. As the consequence, R1-R6 axons bypassed the lamina through the gaps created by distal cell neurons to enter the medulla. This phenotype is similar to that in slit mutants (Tyler et al., 2004). All these data suggest that the proper compartment formation in the optic lobe, and the boundary formation between different ganglia, is essential for neural network formation.

1.1.8.2.2 Egghead (Egh)

Failure of R1-R6 axon termination in the lamina layer and mislocalization of glial cells were also observed in the egh mutant (Fan et al., 2005). egh gene encodes a β4- mannosyltransferase (Fan et al., 2005). Genetic analysis indicated that the egh mutant phenotype was caused by the invasion of distal cell neurons into the lamina, similar as that in the slit and robo mutants. Brainiac (Brn), a β3-N-acetylglucosaminyl-transferase, functions downstream of Egh in the glycosphingolipid biosynthesis in the fly (Wandall et

al., 2003 and 2005). Mutants of brn also displayed a similar R1-R6 axon overshooting phenotype (Fan et al., 2005). Because slit and robos expression levels did not change in the egh mutant, it is likely that Egh and Brn are involved in a different pathway to control the boundary formation in the optic lobe. Alternatively, they might act to regulate the posttranslational modification of Slit and Robo proteins for their proper functions (Fan et al., 2005).

In the wild-type brain, the satellite glia scattered above the lamina layer extend sheath- like glial processes into the posterior face of the lamina region. At the posterior margin of the lamina, those glial processes mingle with the processes of the lamina and medulla glial cells to form a continuous glial sheath between the lamina neuropil and the lobula cortex. These glial sheaths in the brain have been suggested to function in regulating the compartment formation of different neuropils in the brain (Younossi-Hartenstein et al.,

2003; Pereanu et al., 2005). In the egh mutant, this glial sheath appeared abnormal (Fan et al., 2005). Currently, researchers speculate that the glial sheath generated by the satellite glia and the lamina/medulla glia serves as a boundary to separate the lamina plexus from the lobula cortex. When this sheath was broken, distal cell neurons invaded the lamina neuropil.

1.1.8.3 Genes involved in target neuron development

Lamina neurons are not directly required for the proper R-cell axon guidance and the R1-

R6 layer-specific targeting at the third-instar larval stage (Huang et al., 1996; Poeck ea al.,

2001). But because lamina neurons serve as synaptic targets of R cells in the adult visual

system, the proper differentiation, development, and positioning of lamina neurons are essential for synapse formation during the pupal stage (Salecker et al, 1998; Clandinin and Zipursky, 2001).

Back in 1991, researchers already found that retinal innervation was necessary for the differentiation and development of lamina neurons (Selleck and Steller, 1991). In the eyeless mutant, the absence of R cells in the eye disc was associated with the loss of lamina neurons in the optic lobe (Selleck and Steller, 1991). Later studies showed that

Hedgehog (Hh) was produced by differentiating R cells, transported along their axons into the lamina target region to induce the neurogenesis of lamina precursor cells (Huang and Kunes 1996; Huang and Kunes 1998).

Hh induces the expression of dachshund (dac) and single-minded (sim), which encode a nuclear protein and a basic helix-loop-helix-PAS transcription factor, respectively

(Mardon et al., 1994; Huang and Kunes 1996, Umetsu et al., 2006). Dac controls the expression of EGFR on precursor cells and differentiating lamina neurons (Huang and

Kunes 1996). The EGFR proteins on these cells can detect the second induction signal,

Spitz, which is also produced by R-cells axon terminals (Huang et al., 1998). Spitz is a ligand of EGFR, and it can induce the final differentiation and maturation of lamina neurons (Huang et al., 1998). Sim has an important function in the integration of lamina neurons with R-cell axons to assemble lamina cartridges (Umetsu et al., 2006). The collaborative actions of above proteins thus instruct the differentiation of lamina neurons and subsequent the assembly of lamina cartridges.

100

1.2 Dendrite patterning and function

Dendritic arborisation is essential for the formation of functional neural circuits since dendrite tree is the main place for the neuron to receive synaptic input. How the receptive field is occupied by dendrite branches decides what types of input the neuron is able to receive (Scott and Luo, 2001). Normally, different types of neurons have different stereotyped dendritic arborisation patterns. In the early study of the neuroanatomy by

Golgi staining, Golgi and Cajal already noticed the cell-type specific dendritic pattern

(Cajal, 1995). And this cell-type-specific dendritic patterning phenomenon was also discovered in both vertebrates and invertebrates. For example, rabbit retina amacrine cells have more than 22 different subtypes. Each has a unique dendritic pattern (MacNeil and

Masland., 1998). While in the fly body wall, each class of the dendritic arborisation (da) neurons displayed a specific dendritic pattern (Corty et al., 2009). Defects in dendrite pattern have been found to be related to some human mental diseases, such as Down‟s syndrome and Fragile-X syndrome (Kaufmann and Moser, 2000).

For quite a long time, our knowledge about dendritic patterning came from studies using

Golgi staining. However, this technique does not allow the visualization of temporal changes in dendrite pattern. Recent discovered techniques, including biolistic transfection

(Arnold et al., 1994; Lo et al., 1994), DiI labelling (Dailey and Smith, 1996; Wu and

Cline, 1998), genetically manipulation the reporter gene expression by GAL4/UAS system (Brand and Perrimon, 1993; Gao et al., 1999) and the MARCM technique (Lee and Luo, 1999; Grueber et al., 2002), allow live-imaging of dendrite development. So,

101

our knowledge about dendrite arborisation, patterning, and function has advanced dramatically over the past few years.

1.2.1 Dendrite development, arborisation, and patterning

The development of a dendrite tree includes several steps. First, neurites grow from morphologically unpolarized young neurons, and obtain some characteristics on growth rate and protein expression profile specific for dendrite. Second, dendrites undergo directed migration under the control of guidance cues. Third, dendrites begin to branch out at defined intervals. And forth, high-order branches and spines are formed (Scott and

Luo, 2001). Dendrites undergo dynamic changes throughout life. Higher-order branches and spines keep growing out and retracting, in an activity dependent manner (Häusser et al., 2000). At specific developmental stages, such as the metamorphosis of the insect, existing dendrites undergo pruning and remodelling, such as the mushroom body (MB) neurons in the central brain and dendritic arborisation (da) neurons in the PNS of the fly

(Grueber and Jan, 2004). I will discuss each step of dendrite development in more details in following sections.

1.2.1.1 Dendrite outgrowth

To make contact with its presynaptic partner, a neuron needs to send out its dendrite away from the cell body. This process is called dendrite outgrowth (Scott and Luo, 2001).

Dendrite outgrowth is controlled by both cell-intrinsic program and extracellular factors

(Altman et al., 1972; Rakic et al., 1973). The evidence to support the importance of intrinsic factors comes from study in both vertebrates and invertebrates. Studies on

102

vertebrate Purkinje cells have shown that the initial branching pattern and spine formation occurred normally even when its presynaptic partner, the granule cells were removed

(Altman et al., 1972; Rakic et al., 1973; Corty et al., 2009). Work on several transcription factors in fly da neuron dendritic arborisation by genetic analysis, such as seq (Gao et al.,

1999; Brenman et al., 2000), cut (Grueber et al., 2003a), abrupt (Li et al., 2004; Sugimura et al., 2004), and spineless (Kim et al., 2006), also support the requirement of intrinsic factor in dendrite growth.

Extracellular factors are also important for dendrite growth. Characterization of the neurotrophin family, including BDNF (Brain-Derived Neurotrophic Factor), NGF (Nerve

Growth Factor), and Neurotrophin 3, 4/5 (NT-3, 4/5) in mammalian cortical neuron development, provides evidence for the involvement of extracellular cues (Gundersen and

Barrett, 1979; McAllister et al., 1995, 1996, 1997; Horch et al., 1999; Yacoubian and Lo,

2000). A BMP family member, Osteogenic protein 1 (OP-1), together with NGF, could induce cultured sympathetic neurons to project dendrites (Lein et al., 1995). In Xenopus tectal cells, when a GPI-linked candidate plasticity gene 15 (CPG 15), or neuritin, which is a transcription factor, was overexpressed, their neighbouring tectal projection neurons began to extend dendrites towards them (Nedivi et al., 1998, 2001; Cantallops et al.,

2000). Those results indicate that both intrinsic factors and extracellular cues play important roles in dendrite outgrowth.

In fly da neurons, when the non-classical cadherin protein Flamingo was absent, the dorsal dendrites kept extending to cross the dorsal-ventral midline, while in wild-type fly

103

they never cross (Gao et al., 1999). This result suggests that in addition to promotion of dendrite outgrowth, there is also the involvement of negative regulation machinery to limit the growth of dendrite, which is initiated by either intrinsic or extracellular cues. In the explants of chicken retina ganglion cells (Wong et al., 2000), rat hippocampal slices

(Nakayama et al., 2000), and fly CNS mushroombody neurons (Lee et al., 2000), small

GTPase RhoA has been identified as a branching limitation factor to restrict the outgrowth of dendrites.

Neuronal activity can also function to regulate the dendrite growth. To date, neuronal activity dependent dendritic growth is mainly mediated by calcium signalling and calcium responsive transcriptional activators such as CREB (Tao et al., 1998, Redmond et al.,

2002, 2005; Wayman et al., 2006). The change in the neuronal activity probably modulates the expression level of neurotrophins, which directly control the outgrowth of dendrite.

1.2.1.2 Dendrite guidance and targeting

Like the axon, dendrites navigate towards their dendritic fields. Some of the molecules involved in axon guidance have been found to play similar roles in the dendritic guidance and targeting (Polleux et al., 1998, 2000; Furrer et al., 2003; Godenschwege et al., 2002).

Dendrites can also respond to gradients of secreted ligand to either turn to or steer away from the source of the ligand.

104

One well known example is the response of the cortical pyramidal neuron dendrite to the diffusible molecule Sema3A. Sema3A has been shown to induce the collapse of the cultured dorsal root ganglion (DRG) axon growth cones (Luo et al., 1993; Messersmith et al., 1995). This repulsive response depends on the low level of cGMP within the axon of

Xenopus spinal neuron. When cGMP level is increased, instead of steering away from the source of the Sema3A, the axon navigates toward it (Song et al., 1998). For pyramidal neurons, Sema3A functions as an attractant for the apical dendrites (Polleux et al., 1998,

2000). And the Guanylate cyclase (SGC), the enzyme catalyzing the synthesis of cGMP, is also rich in those apical dendrites (Polleux et al., 2000). The high level of cGMP in apical dendrites may thus allow Sema3A to attract those dendrites to move apically.

In the ventral nerve chord of the fly, dendrite guidance of motor neurons are also regulated by extracellular cues (Landgraf et al., 1997, 2003). For axon guidance in fly ventral nerve chord, Netrin-Frazzled and Slit-Robo signalling pathway have been shown to play important roles (Mitchell et al., 1996; Kolodziej et al., 1996; Rothberg et al., 1988 and 1990; Brose et al., 1999; Kidd et al., 1999; Rajagopalan et al., 2000a and 2000b;

Simpson et al., 2000). And recently, Chiba lab implicated these two pathways in dendritic guidance of motor neurons aCC, RP2, and RP3 in the ventral nerve chord (Furrer et al.,

2003). Surprisingly, the response of motor neuron dendrites to Netrin and Slit are largely similar to that of their axons. For example, when netrin was absent, aCC and RP3 dendrites could not extend across the midline (Furrer et al., 2003). The Slit-Robo pathway also controls dendrite guidance in the giant fiber (GF) interneuron and the tergotrochantral motoneuron (TTMn) system (Godenschwege et al., 2002). Recently, an

105

overexpression phenotype screen on the dendritic targeting and morphology in fly the ventral nerve cord has identified additional players involved in dendrite guidance and targeting in the midline (Ou et al., 2008).

1.2.1.3 Dendrite branching and spine formation

After the outgrowth of the primary dendrite, extensive dendrite branches and spines are formed during development. Generally, there are two different mechanisms to form branches. The first is to simply split the growth cone of the dendrite by bifurcation. The second is to build the branches from the existing dendrite shafts (Scott and Luo, 2001).

Time-lapse recording from organic slice cultures of rat hippocampus showed that pyramidal neurons generated branches by extending protrusions from existing dendrites

(Dailey et al., 1996). The time-lapse study of da neuron growth in the fly PNS also shows the addition of branches on existing dendrites (Gao et al., 1999). Observations from both vertebrates and invertebrates suggest that the second mechanism probably is the predominant physiological way for branch and spine formation on dendrites.

For dendrite branching, small GTPase family member Rac and Rho (Li et al., 2000;

Wong et al., 2000), non-receptor tyrosine kinase Abelson (Abl) (Li et al., 2005), cytoskeleton regulator enabled (Ena) (Li et al., 2005) and Kakapo (Prokop et al., 1998;

Gao et al., 1999), and the transport protein Dynein, Kinesin, and Rab5 (Zheng et al., 2008;

Satoh et al., 2008) have been found to be essential for branching and spine formation.

Transcription factors, such as Spineless and Cut, have also been implicated in controlling dendrite branching in da neurons (Grueber et al., 2003a Kim et al., 2006). Molecules

106

involved in promoting dendrite outgrowth, like the neurotrophins, might also play a role in dendrite branching.

Rac and Rho have specific capability to regulate dendrite branch stability. When Xenopus tectal cells expressed the constitutive active form of rac, they showed an increase in the rate of both sprouting and retraction (Li et al., 2000). In the explants of chicken retina, ganglion neurons had increased number of branches and protrusions when they expressed the constitutive active form of Rac (Wong et al., 2000). While the dominant negative form of Rac could induce the dramatic loss of branches. A similar phenotype was observed when the dominant negative form of Rho is expressed (Wong et al., 2000).

In the Drosophila da neuron dendrites, non-receptor tyrosine kinase Abl, and its substrate

Ena, actively regulate branching and spine formation (Li et al., 2005). Within this system,

Abl was found to inhibit the dendritic branching of da neurons, while as its substrate, Ena promoted the formation of branch and the actin-rich spines (Li et al., 2005). Another cytoskeleton regulator, Kakapo, has been found to be involved in the dendrite branching for both motor neurons in the ventral nerve chord and the da neurons on the body wall

(Prokop et al., 1998; Gao et al., 1999). When kakapo was mutated, dendrite branch number decreased dramatically for both motoneuron and da neuron (Prokop et al., 1998;

Gao et al., 1999). All those studies suggest the changes in the cytoskeleton are underlying the branch/spine formation and retraction. Future studies will likely identify new cytoskeletal regulators in regulating dendrite branching.

107

Another grope of proteins involved in dendrite branching are proteins involved in regulating protein cargo trafficking in fly, including the motor protein Dynein and

Kinesin, and Dynein-interacting endosome protein Rab5 (Zheng et al., 2008; Satoh et al.,

2008). In fly da neurons, Golgi outposts are predominantly located in dendrites and involved in regulating dendrite growth (Ye et al., 2007). Disturbing these polarized Golgi outposts could induce changes in dendritic branching in da neurons (Ye et al., 2007).

Dynein, Kinesin, and Rab5 are involved in regulating the polarized distribution of Golgi and some other protein cargos. Mutations in those genes cause similar dendritic branching defects (Zheng et al., 2008; Satoh et al., 2008). Those findings thus support the importance of the protein and organelle cargo trafficking within of the dendrite for proper dendritic branching.

1.2.1.4 Dendrite heteroneuronal tiling

The definition of dendrite tiling comes from the original description of dendritic territory back in 1981, when researchers found that in the cat retina, each ganglion neuron has its own dendritic territory (Wässle et al., 1981). The territory of neurons of the same functional type shows largely non-overlapping pattern, while between different functional types, the territory over-lapped extensively (Wässle et al., 1981). So, dendrite tiling refers to the maximal and non-redundant coverage of receptive fields by dendrites from same- type neurons (Grueber et al., 2002; Parrish et al., 2007). If a ganglion cell was removed by surgical method, the neighbouring ganglion cells extended their dendrites into its neighbour‟s territory (Perry and Linden, 1982). This suggests strongly that existence of mutual repulsion between same-type dendrite for the formation of tiling pattern.

108

da neurons in fly PNS also show tiling between the same class (Gao et al., 2000; Grueber et al., 2002; Grueber et al., 2003; Sugimura et al., 2003; Parish et al., 2007). If the class

IV da neuron was ablated by laser during the embryonic or larval stage, the neighbouring class IV da neurons extended their dendrites beyond the original dendritic field boundary

(Gao et al., 2000; Grueber et al., 2003b; Sugimura et al., 2003). Those data also indicates the importance of mutual repulsion in da neuron dendrite tiling. Several fly genes including fmi, seq, trc, its potential adaptor fry, and its upstream kinase hippo, have been implicated in controlling class IV da neuronal tiling (Grueber et al., 2002; Kimura et al.,

2005; Emoto et al., 2004, 2006). When those genes were mutated, neighbouring class IV da neuron dendrites overlap with each other extensively at the lateral and segmental boundary (Grueber et al., 2002; Emoto et al., 2004, 2006).

Recent studies also demonstrate that intrinsic growth control is also involved in regulating dendrite tiling. When some retina ganglion cell neurons in the mouse retina were removed either by chemical depletion or by genetic ablation, the dendritic fields of the remaining neurons did not increase in size (Farajian et al., 2004; Lin et al., 2004). This observation indicates that both the intrinsic growth control and extracellular cues are involved in regulating the establishment of the tiling pattern.

Mutual repulsion may be mediated in a contact-dependent way or a contact-independent way. In fly class IV da neurons, those dendrites show dynamic movement during the development. Dendrites keep extending and retracting at the boundary by directly contacting neighbouring dendrites to establish the tiling pattern (Emoto et al., 2004). But

109

in C. elegans, the tiling between two neurons, the ALM and the PLM, is totally contact free (Gallegos and Bargmann, 2004). Those two neurons project their single dendrites to occupy the anterior and posterior half of the worm body, respectively. Their dendrites tile the field without contacting each other. But when the gene SAX1 and SAX2 were mutated in a cell-autonomous manner, overlapping of dendritic fields were found

(Gallegos and Bargmann, 2004).

1.2.1.5 Dendrite isoneuronal self-avoidance

Dendritic tiling refers to the avoidance of dendrites from adjacent same-type neurons and is also called heteroneuronal dendrite tiling. Isoneuronal dendrite tiling, or self-avoidance is defined as the tendency for dendrite branches from the same neuron to avoid crossing, thereby spreading evenly over a receptive field (Kramer and Kuwada, 1983). One well- known dendritic self-avoidance system is the isoneuronal tiling between adjacent dendritic arbors from the same class IV da neuron in the fly PNS. As I described in the previous section, the Trc/Fry signalling pathway is involved in heteroneuronal dendrite tiling (Emoto et al., 2004). This pathway is also found to be involved in dendritic self- avoidance. In either trc or fry mutant, neighbouring dendrite branches from a single class

IV neuron crossed over each other (Emoto et al., 2004). In 2007, three groups independently implicated Down‟s syndrome cell adhesion molecule (Dscam) in controlling dendritic self-avoidance in Drosophila PNS (Matthews et al., 2007; Soba et al., 2007; Hughes et al., 2007).

110

Dscam is a member of the Ig superfamily. With alternative splicing, dscam has the potential to produce more than thirty thousands of isoforms (Schmucker et al., 2000).

Each isoform has the strongest binding affinity for itself, while displaying much lower or no binding activity to other isoforms (Wojtowicz et al., 2004 and 2008, Sawaya et al.,

2008). Dscam has been implicated in regulating a variety of developmental processes, such as axon guidance and targeting in the fly ventral nerve chord (Schmucker et al.,

2000; Wojtowicz et al., 2004), mushroom body axon bifurcation (Wang et al., 2002; Zhan et al., 2004; Hattori et al., 2007), olfactory system dendritic patterning and synaptic partner matching (Hummel et al., 2003; Zhu et al., 2006), axon target recognition in the adult mechanosensory neurons (Chen et al., 2006), and in the immune system (Watson et al., 2005).

Loss of dscam in da neurons caused dendrite self-avoidance defects such as the fusion between sister branches, suggesting that Dscam mediates mutual repulsion between sister branches in a homophilic manner (Matthews et al., 2007; Soba et al., 2007; Hughes et al.,

2007). Interestingly, in class IV da neurons, loss of dscam did not cause any defect with heteroneuronal tiling. This data suggests that dendritic tiling and self-avoidance can be controlled by different signalling pathways.

When the cytoplasmic tail of Dscam was deleted, the mutated dscam could no longer rescue the dscam mutant phenotype (Matthews et al., 2007). This suggests that the cytoplasmic domain is involved in converting the signal generated by homophilic binding

111

into a repulsive response, likely through activating downstream signalling events

(Matthews et al., 2007; Hughes et al., 2007).

1.2.1.6 Dendrite maintenance

Most neurons develop their dendritic patterns during the early developmental stage.

Although their higher order branches and spines undergo dynamic changes constantly throughout the lifetime, there is no significant change in dendritic pattern, including primary dendrites orientation, branching pattern, and tiling pattern, if those neurons are not subjects to the dendritic pruning and remodelling process, which will be discussed later (Parrish et al., 2007). One important question is how dendritic pattern is maintained.

In the Down‟s syndrome (DS) adult patients, the cortical pyramidal neurons show remarkable reduction in dendritic branching, dendrite length, and spine formation. But in the early developmental stage, the DS fetus and newborn has normal or even greater dendrite branching and length. From 2.5 months of postnatal age, dendrite branches and length begin steadily decreasing. This decrease renders the adult DS phenotype (Becker et al., 1986, Vuksić et al., 2002). A similar phenotype was observed in the DS model

Ts65Dn mouse (Davisson et al., 1990, Benavides-Piccione et al., 2004). Those indicate the existence of certain mechanisms to maintain the dendrite pattern, which are disrupted in the DS patients and mouse model.

112

Recent studies showed that the intrinsic controls of specific genes expression, as well as the extracellular regulators such as neurotrophins and the extracellular matrix, play important roles in dendrite stabilization and maintenances (Perrish et al., 2007).

In fly, when the class IV da neuron was removed at either the embryonic stage or the larval stage, neighbouring same-class neurons always invade the empty space left by the removed neuron (Gao et al., 2000; Grueber et al., 2003b; Sugimura et al., 2003).

Interestingly, only relatively limited invasion was observed. Completely taking over of the space was never observed. This suggests that other mechanisms involved in maintaining dendritic field might function to restrict the dendritic territory in fly da neurons.

The work on the role of warts (wts) in da neurons in fly provides a good example in the dendrite field maintenance (Emoto et al., 2006). There are two NDR (Nuclear Dbf2- related) kinase family members in fly: trc and wts. trc is involved in dendritic tiling. wts is a tumor suppressor originally implicated in regulating cell proliferation (Justice et al., 1995, Xu et al., 1995).

Loss of wts caused the formation of large gaps between two opposing class IV da neurons in the receptive field on the body wall of the third-instar larvae (Emoto et al., 2006).

Time-lapse recording shows that at the early developmental stage 24-28 hrs AEL, the overall dendritic growth and branching in wts mutant was indistinguishable from the wild type. At 48-52 hrs AEL, the dendrites of those mutant da neurons could form correct

113

tiling pattern at the midline. But at 72-76 hrs AEL, those tiled dendrites retracted and no longer contacted each other. This left behind the large unoccupied space in between those neurons (Emoto et al., 2006). This suggests that wts is involved in maintaining the dendritic field. As the adaptor protein for Wts, Salvader (Sav) is also found to function in dendrite maintenance (Emoto et al., 2006). As the upstream regulator of both wts and trc, the ste-20-related tumor suppressor kinase Hippo (Hpo), also functions in dendrites maintenance in addition to its role in dendritic tiling (Emoto et al., 2006).

For cell surface molecules involved in dendrite maintenance, two major groups have been identified. The first group includes TrkB proteins, as high affinity receptors for the neurotrophins BDNF. Both the TkB receptor and its ligand show important roles in maintaining the cortical dendrite pattern (Gorski et al., 2003, Xu et al. 2000). As the neurotrophins also act in promoting dendrite growth, in order to separate the early growth function from the latter maintenance, late-onset cortical knockout of either the receptor or the ligand was characterized to specifically address its role in maintenance. Those experiments showed great reduction in the basal dendritic complexity (Gorski et al., 2003,

Xu et al. 2000). The second group include the mouse homolog of fly fmi gene, the celsr2.

Knocking down celsr2 by RNAi induces the progressive loss of dendrite arbors in both cortical pyramidal neurons and Purkinje cells (Shima et al. 2004).

Extracellular matrix (ECM) is also likely to play roles in dendrite stabilization and maintenance. Antagonizing Integrin-mediated adhesion by exogenously added peptides or over-expression of dominant-negative integrins induces remarkable dendrite retraction in

114

the chicken retina (Marrs et al. 2006). The conditional knockout mouse of either arg or abl, which are nonreceptor tyrosine kinases acting downstream of integrin, also shows progressive dendrite loss (Moresco et al., 2005). Those results supports the idea that integrin helps to maintain the established dendrite territory.

1.2.1.7 Dendrite remodelling

During development, dendrites need to be adjusted and refined in response to changes in the target region. Our knowledge about the dendrite remodelling process mainly comes from three sources (Parrish et al., 2007). The first is the dendritic pruning during the insect metamorphosis. The second is the dendritic remodelling after the pathological changes. The third is activity dependent remodelling.

Study from both fly and moth shows that during metamorphosis, the dendrites of some existing neurons undergo dramatic pruning to prepare for the establishment of the adult dendrite pattern (Lee et al., 2000, Truman et al., 1994, Schubiger et al., 1998). In the fly larval mushroom body, neurons send their dendrites to the calyx. Those dendrites are subjected to pruning during metamorphosis. This remodelling process is under control of the ecdysone signalling (Lee et al., 2000, Truman et al., 1994, Schubiger et al., 1998).

From a genetic mosaic screen for mutations affecting this pruning process, mutations in members of TGF-β/Activin signalling pathway were isolated, including the type I receptor babo, the type II receptor punt and wit, and the transcription effecter dSmad2

(Zheng et al., 2003). Genetic analysis shows that the TGF-β/Activin signalling pathway

115

functions through regulating the expression level of Ecdysone receptor B1 to control dendritic remodelling (Zheng et al., 2003).

In the fly PNS, during metamorphosis, most da neurons undergo programmed cell death, while the survivors undergo extensive dendritic remodelling (Williams and Truman

2005b). Similarly as in the mushroom body neurons, the ecdysone signalling pathway is also important for the remodelling process of those da neurons (Williams and Truman

2005a). Recent studies demonstrate that the caspase DRONC, working as the transcription targets of the ecdysone signalling pathway, controls the remodelling process

(Kuo et al., 2005, 2006; Williams et al., 2006). In this process, DRONC may either be activated locally to degenerate the proximal dendrites or function to recruit phagocytes to engulf dendrite fragments after the degeneration.

Dendritic remodelling at the pathological situation such as epilepsy and ischemia has been observed. In the rat model, dentate granule cells can form new basal dendrites when temporal lobe epilepsy is induced (Spigelman et al., 1998). And when global transient ischemia occurred to the rat brain, the CA1 pyramidal cells show an increase in the apical dendrites and the invasion of apical dendrite field by the basal dendrites (Ruan et al.,

2006). So far, little is known about the underlying mechanisms. But those studies indicate that neurons retain the capability to undergo large-scale dendritic remodelling in specific conditions even at the adult stage. Future study may help us to understand the exact mechanism of dendrite remodelling at pathological conditions.

116

Compared to rare cases of remodelling under pathological situations, activity-dependent dendrite remodelling has been widely observed in a variety of model systems. One well- known example of activity-dependent dendrite remodelling is dendrite remodelling induced by the sensory evoked neuronal activity. Mitral cells in the olfactory system of vertebrates originally have multiple dendrite branches contacting adjacent glomeruli.

With the experience of sensory input, over time, most of those dendrites are lost and leave only one primary dendrite to make contact with a single glomerulus (Malun and Brunjes.,

1996). Similar observations have been made in the mouse somatosensory cortex in which neurons remodel their dendrites in response to the sensory stimuli from the whiskers

(Tailby et al., 2005). It is also reported that light stimuli increase the dendrite growth in the Xenopus optic tectal cells (Sin et al., 2002). Those remodelling may be mediated through the NMDA receptor-mediated signalling pathway, which uses MAPK, CaMKs, and Rho family GTPases as downstream effectors (Redmond et al., 2002; Vaillant et al.,

2002; Wu and Cline, 1998; Wu et al., 2001). Sensory input can also induce an increase in the calcium signalling (Redmond et al., 2002). This may also affect the expression of neurotrophins to regulate the dendrite growth.

1.2.2 Drosophila PNS dendritic arborisation (da) neurons

As an excellent model to study dendrite growth and patterning, Drosophila PNS neurons elaborate their dendrites in two dimensions between the epidermis and the body wall muscles. Specific antibody such as MAb 22C10, together with the powerful GAL4/UAS-

GFP system and MARCM technique, helps to characterize dendrite pattern of those PNS sensory neurons (Kolodziej et al., 1995; Gao et al., 1999; Grueber et al, 2002). The

117

expression of GFP in embryos makes it possible to examine dendrite growth and patterning in living animals.

The fly embryonic PNS is organized in a segmental pattern. There are forty-four peripheral neurons in each hemisegment of the fly embryo (Campos-Ortega et al., 1985;

Ghysen et al., 1986). They are divided into dorsal, lateral, and ventral clusters based on their relative location on the body wall (Campos-Ortega, 1985; Ghysen et al., 1986, 1989).

Based on different dendritic morphology, especially the numbers of dendrites, those neurons are divided into two categories: neurons with single dendrite and neurons with multiple dendrites (md) (Hartenstein et al., 1988). The first group includes the chordotonal (ch) neuron and the external sensory (es) neuron. md neurons are subdivided into three groups, the tracheal dendrite (td) neuron, the bipolar dendrite (bd) neuron, and the dendritic arborisation (da) neuron (Bodmer et al., 1987; Gao et al., 1999). Da neurons are further divided into four classes, class I to class IV with the increase in the size of their dendritic territory and the complexity of their dendrites.

With the help of MARCM technique, Grueber et al. elucidated the dendritic pattern of each class at high resolution (Grueber et al., 2002). Class I da neuron has the simplest pattern with a relatively long dorsal directed primary dendrite. This primary dendrite branches at repeated interval to form the anterior-posterior oriented secondary branches.

This gives the class I neuron the “comb like” structure. The class II da neuron also has long primary branches. But compared to class I neuron, its dendrites are more sinuous and more symmetrically bifurcated. Similar to class II, Class III da neuron has long and

118

sinuous primary and secondary branches. Different from all the other classes, class III da neuron has numerous spine-like protrusions along most of its primary and secondary branches. The class IV da neuron has the most complex dendritic arborisation pattern.

They occupy largely the entire body wall, in a non-overlapping manner (Grueber et al.,

2002).

Among all the md neurons in a hemisegment, the dorsal cluster da neurons are the most well characterized (Gao et al., 1999; Grueber et al., 2002). There are eight md neurons within this cluster, one td, one bd, and six da neurons. Those six da neurons include two class I (ddaD and ddaE), one class II (ddaB), two class III (ddaA and ddaF), and one class

IV (ddaC). da neurons of the same class are distributed in a largely non-overlapping way, while different classes overlap with each other extensively (Grueber et al., 2002).

The dendrite development of dorsal cluster da neurons at different developmental stages are well documented (Gao et al., 1999). The primary dendrites of this dorsal cluster da neurons emerge at around 13-14 hrs AEL, just two hours after the PNS axons reached the

CNS. At around 13 hrs AEL, a dorsal dendrite emerges from one da neuron within the anterior of the dorsal cluster. The second dorsal dendrite projects out from one da neuron within the posterior of the same cluster shortly after. Both dorsal dendrites extend perpendicular to the anterior-posterior axis towards the dorsal midline. Normally, each md neuron within the dorsal cluster sends out only one dorsal primary dendrite, while some of them may have additional primary lateral dendrites. The dorsal extension stops at

15-16 hrs AEL, before the lateral branches start to grow. Between 15-17 hrs AEL, the

119

lateral branches start to develop. Most of those lateral branches are transient. They project out and retract constantly before subsets of them eventually stabilize at 18-20 hrs AEL to become the final lateral branches. At this stage, the dorsal and lateral branches are clearly distinguishable from each other. And the numbers of lateral branches in each segment are similar from animal to animal. Before hatching (23-24hr AEL), the lateral branches keep adding on higher order branches before and after they reach the segment boundary. Only a small number of branches cross the segmental boundary to invade the neighbouring segment (Bodmer and Jan 1987; Gao et al., 1999). At the time of hatching, those dorsal branches have not reached the dorsal midline. They keep growing until they reach the dorsal midline at around the second-instar larval stage. The length and the thickness of dendrites increase as the size of the larval body grows (Gao et al., 1999).

Functions of each class of da neurons have not been studied clearly. Md neurons, in a group, have been found to control the rhythmic locomotion behaviour (Hughes and

Thomas, 2007; Song et al., 2007) and nociception (Tracey et al., 2003). Individual class of da neurons has been found to control the thermosensation including class I and II (Liu et al., 2003), and nociception from wasp attacking by class IV (Hwang et al., 2007).

This stereotyped dendrite arborisation pattern and the easy accessibility make da neuron to be an excellent system for identifying molecules involved in regulating dendritic growth, branching, tiling, self-avoidance, and maintenance.

120

1.3 An overview of my studies and their contributions to our knowledge of the related fields

1.3.1 My work in axon guidance and tiling in the fly visual system

In the section 1.1.6, I described the structure and the development of the fly visual system, which is an excellent model for researchers to identify molecules involved in axon guidance, targeting, and tiling. In the past two decades, generations of fly neurologists have utilized this system to successfully identify many genes functioning in neural network formation in the fly visual system. I have given a brief review of those genes in section 1.1.7 and 1.1.8. From what has been discovered, we can appreciate that there are likely multiple parallel but non-redundant pathways controlling the same process. Some molecules may participate in a single pathway, while the others may function in several different pathways. Interestingly, to date, almost all pathways identified are incomplete.

To fulfill the gaps in these pathways and to identify novel players in axon guidance, we conducted genetic screens. The first screen was an unbiased immunohistological screen with P-element-induced random mutations on the second chromosome of the fly. From this screen, the gene bur was identified. My results implicated Bur, a GMP synthetase functioning in the de novo GMP synthesis, in R-cell axon guidance. Following investigations on the gene ras confirms the involvement of the de novo GMP synthesis pathway in R-cell axon guidance. I also discovered the interaction between Bur and Rac

GTPase and propose a model in which the de novo GMP synthesis regulates the Rac

GTPase activity to control axon guidance.

121

This discovery provides novel and important insights into the mechanism of R-cell axon guidance. Besides rac, I did not find interaction between bur and any other known genes involved in R-cell axon guidance. This suggests that the de novo GMP synthesis may participate in a previously unknown pathway to regulate axon guidance.

My finding also adds a new function to the de novo GMP synthesis. Previously, the de novo synthesis of guanine nucleotides is only found to be required specifically for the lymphocyte proliferation in animals. Due to the existence of the salvage GMP synthesis, it is unclear whether the de novo GMP synthesis is required for any other cellular process

(Eugui et al. 1991; Dayton et al. 1992; Hauser and Sterzel 1999; Gu et al. 2000; Gu et al.

2003). My work on Bur and Ras in fly axon guidance suggests that the de novo GMP synthesis is required for neural connectivity formation. This is a unique function, which cannot be replaced by the salvage GMP synthesis.

The second screen we did was a second-site modifier screen designed to identify interacting genes of the serine/threonine kinase Misshapen (Msn). In previous analyses, our lab discovered that the overexpression of msn in R cells could induce the early termination of R-cell axons (Ruan et al., 1999). We screened deficiency and candidate mutations for modifiers of this early termination phenotype. From this screen, we identified the mutation of the gene turtle (tutl) as a suppressor of the msn overexpression phenotype. Further analysis indicated that tutl could regulate R-cell axon targeting and R7 axonal tiling (For the detailed information about the function of tutl in the visual system, please check the chapter of general discussion.). To date, there is only one signaling

122

pathway having been implicated in R7 axon tiling. That is the Activin signaling pathway.

Our discovery of the function of tutl adds the second potential signaling pathway for controlling R7 axonal tiling. This finding expands our knowledge about axonal tiling significantly.

During my analyses of the gene tutl in the visual system, I also found that the tutl is expressed in the PNS dendritic arborization (da) neurons. This led to the characterization of tutl for its functions in regulating dendritic patterning. I will describe the importance of this finding in the following section 1.3.2.

In addition to unbiased genetic screens, I also took the testing-candidate approach to assess the potential function of some candidate genes in the visual system. In this analysis,

I focused on the gene encoding Drosophila neurospecific receptor kinase (Dnrk). Dnrk belongs to the fly Ror family, which has another member as Dror. Dnrk is specifically expressed in the CNS and PNS at the embryonic stage (Oishi et al., 1997). Its orthologs in

C. elegans and zebrafish have been implicated in axon guidance and neuron migration

(Forrester et al., 1999; Koga et al., 1999; Francis et al., 2005; Zhang et al., 2004). In the fly, due to the unavailability of mutants, the function of the fly Ror family, including

Dnrk and Dror, has not been explored previously.

Starting with the generation of Dnrk mutants, I have conducted a series of experiments and finally discovered the function of Dnrk in regulating R-cell axon guidance. As a homolog of Dnrk, Dror acts redundantly with Dnrk in this process. Although my

123

investigation on these two genes is incomplete at this moment, it highly suggests that fly

Ror family members also act in regulating axon guidance, a function similar to their orthologs in the other species. Future analyses of these genes may help to elucidate the mechanisms underlying the specific phenotype. Dnrk and Dror may function together to initiate a novel signaling pathway in controlling R-cell axon guidance in the fly visual system.

1.3.2 My analyses in dendritic patterning in the fly da neurons

As I mentioned previously, I followed the expression pattern of tutl in the fly PNS da neurons to examine the potential function of tutl in these cells. Tutl belongs to the IgSF family. Previously, tutl mutants have been identified based on their defects in coordinated movements (Bodily et al., 2001). Tutl was defined as a CNS-specific protein. Because no morphological abnormality has been discovered in tutl mutants in the original analysis, it is difficult to locate the cause of the behavioral defects (Bodily et al., 2001). Interestingly, tutl has a homolog in the mouse as Dasm-1, which has been speculated to function in dendrite arborization and synapse maturation (Shi et al., 2004a and 2004b). Based on all these information, we decided to focus on the potential function of tutl in the dendritic development of da neurons.

Genetic analyses of tutl mutants indicated that tutl had different functions in different classes of da neurons. In class I da neurons, tutl functions cell autonomously to control the branching complexity. It inhibits excessive branching at distal termini. In class IV da neurons, tutl regulates sister dendrite self-avoidance but not tiling. In the past ten years,

124

fly scientists have used the da neurons in fly PNS as a good model to look for molecules involved in dendrite development. Different types of molecules have been identified in regulating distinct aspects of da neuron dendrite development. For dendrite branching, it has been hypothesized to be controlled by both intrinsic and extrinsic cues. Currently, the majority of identified molecules affecting dendrite branching are transcription factors and cytoplasmic proteins. It is still unclear how dendrite can respond to the extrinsic cues. Our discovery of tutl in dendrite branching provides new evidence that cell surface protein can control dendrite branching. It also presents an excellent starting point to elucidate the signaling pathway from the extracellular region to the inside of the da neuron in the future.

Previously, for dendrite self-avoidance, Dscam is the only cell membrane protein identified. The involvement of Tutl, as the second identified cell surface protein, in dendrite self-avoidance indicates that there are likely multiple pathways controlling this process. Dscam proteins function in a homophilic way to repel encountering sister dendrites. There are two possibilities for Tutl to regulate dendrite self-avoidance. It may function in a similar way like Dscam or alternatively, it may act as a receptor to respond to signals from the environment. Currently, we still don‟t know in which way tutl functions. Future investigations on this will shed new light on the mechanism of dendrite self-avoidance.

125

Chapter 2

De Novo GMP Synthesis Is Required for Axon Guidance in Drosophila

This work has been published in Genetics (2006) 172, 1633-42.

126

Fly visual system provides a good model for us to identify genes involved in photoreceptor axon guidance. In a search for genes required for the establishment of R- cell connectivity, we identify the burgundy (bur) gene as an essential player in R-cell axon guidance. bur encodes the only GMP synthetase in fly that catalyzes the final reaction of de novo GMP synthesis. Loss of bur causes severe defects in axonal fasciculation, retinotopy and growth-cone morphology, but does not affect R-cell differentiation or retinal patterning. Similar defects were observed in raspberry (ras) mutants. ras encodes inosine monophosphate dehydrogenase catalyzing the IMP-to-XMP conversion in GMP de novo synthesis. Our study thus provides the first in vivo evidence to support an essential and specific role for de novo GMP synthesis in axon guidance.

127

De Novo GMP Synthesis Is Required for Axon Guidance in Drosophila

Hong Long, Scott Cameron, Li Yu, and Yong Rao

Summary

Guanine nucleotides are key players in mediating growth-cone signaling during neural development. The supply of cellular guanine nucleotides in animals can be achieved via the de novo synthesis and salvage pathways. The de novo synthesis of guanine nucleotides is required for lymphocyte proliferation in animals. Whether the de novo synthesis pathway is essential for any other cellular processes, however, remains unknown. In a search for genes required for the establishment of neuronal connectivity in the fly visual system, we identify the burgundy (bur) gene as an essential player in photoreceptor axon guidance. The bur gene encodes the only GMP synthetase in

Drosophila that catalyzes the final reaction of de novo GMP synthesis. Loss of bur causes severe defects in axonal fasciculation, retinotopy and growth-cone morphology, but does not affect photoreceptor differentiation or retinal patterning. Similar defects were observed when the raspberry (ras) gene, encoding for inosine monophosphate dehydrogenase catalyzing the IMP-to-XMP conversion in GMP de novo synthesis, was mutated. Our study thus provides the first in vivo evidence to support an essential and specific role for de novo synthesis of guanine nucleotides in axon guidance.

128

Introduction

The establishment of neuronal connections in the developing embryo involves the differentiation of neuron precursor cells, the directed growth of axons from developing neurons, and the selection of specific target cells by the growing axons. The growth and targeting of an axon is directed by its growth cone, an expanded structure located at the leading edge of the growing axon. The growth cone expresses cell-surface receptors that recognize and integrate the guidance signals present along the path of axonal projection

(Tessier-Lavigne and Goodman 1996; Guan and Rao 2003). Subsequently, the activation of the growth-cone receptors triggers a cascade of signaling events that ultimately converts guidance signals into directed motility (Tessier-Lavigne and Goodman 1996;

Guan and Rao 2003).

Accumulated evidence points to the importance of purine nucleotides in mediating growth-cone signaling during development. For instance, GTP is required for the activation of Rho-family small GTPases, which are key players in mediating the reorganization of cytoskeletal structures in the growth cone (Luo 2000; Dickson 2001).

Pharmacological studies demonstrate that changes in the levels of cAMP and cGMP can switch the response of a growth cone to an extracellular cue (Song et al. 1997; Song et al.

1998; Nishiyama et al. 2003). For instance, the ratio of cAMP to cGMP activities has been shown to determine if a Xenopus spinal neuron growth cone makes an attractive or a repulsive response to Netrin-1 in vitro (Nishiyama et al. 2003). A higher ratio of cAMP/cGMP signaling favours attraction, while a lower ratio favours repulsion. Genetic studies in Drosophila also demonstrate that cAMP/PKA signaling antagonizes the 129

Semaphorin-induced repulsive response via Nervy, a PKA anchoring protein, which binds directly to the cytoplasmic domain of the Semaphorin receptor PlexA (Terman and

Kolodkin 2004).

The cellular level of guanine nucleotides in animals can be maintained through the de novo and salvage pathways. The de novo synthesis of guanine nucleotides involves a number of reactions that assemble the guanine ring from precursors including amino acids, carbon dioxide and tetrahydrofolate derivatives (Zalkin and Dixon 1992). In contrast, the salvage pathway synthesizes guanine nucleotides by utilizing the available free guanine bases generated by the degradation of nucleic acids and nucleotides (Zollner 1982).

Energetically, the salvage pathway appears much less costly than de novo synthesis, and is generally believed to be the principal pathway for the supply of guanine nucleotides in animals. While the de novo synthesis of guanine nucleotides has been shown to play a role in the proliferation of leukocytes (Eugui et al. 1991; Dayton et al. 1992; Hauser and

Sterzel 1999; Gu et al. 2000; Gu et al. 2003), it is unknown if de novo synthesis is essential for any other biological processes when the food supply is sufficient. In this study, we show that the de novo synthesis of guanine nucleotides is specifically required for the pathfinding of photoreceptor (R cell) axons in the developing Drosophila visual system.

In the Drosophila adult visual system, R cells in the retina project axons directly into the optic lobe of the brain (Meinertzhagen and Hanson 1993). The formation of the R-cell-to- optic-lobe connection pattern begins at the third-instar larval stage when differentiating R

130

cells in the eye-imaginal disc send out axons through the optic stalk into appropriate topographic locations in the developing optic lobe (Clandinin and Zipursky 2002; Tayler and Garrity 2003). Different subclasses of R cells (i.e. R1-R6, R7 and R8) within each ommatidium establish connections with different target layers in the optic lobe: R1-R6 axons terminate within the superficial lamina layer, while R7 and R8 axons project through the lamina and innervate two distinct sub-layers (i.e. M6 and M3, respectively) in the medulla.

In a search for genes that are required for establishing R-cell connectivity, we found that the bur gene is specifically required in R cells for the guidance of R-cell axons in the developing optic lobe. bur encodes an evolutionarily conserved GMP synthetase that catalyzes the final reaction of the de novo GMP synthesis. These findings support an essential requirement for the GMP de novo synthesis in axon guidance during neural development.

131

Materials and methods

Genetics

Bur1, l(2)k07130 (burP), UAS-RacN17, UAS-RacL89, ras11, rasG0002, rasG0436 and

Df(2L)TW161 were obtained from the Bloomington Drosophila Stock Center. ES2-2e10

(bure10) was provided by Gerald Rubin (Neufeld et al. 1998). To generate single bur mutant R-cell axons, hsFLP, UAS-mCD8::GFP, elav-GAL4 (C155); burP, FRT40A/Bc flies were crossed with Tub-GAL80, FRT40A flies. The progeny were heat-shocked at 37 oC for 1 hr at larval stage to induce mitotic recombination. To express the UAS-bur under control of the eye-specific driver GMR-GAL4 in bur mutants, GMR-GAL4, bure10/Bc flies were crossed to bure10/Bc; UAS-bur flies. The R-cell projection pattern in GMR-GAL4, bure10/bure10; UAS-bur/+ was compared to that in GMR-GAL4, bure10/bure10 larvae. To express the UAS-bur under control of the neuronal-specific driver elav-GAL4 in bur mutants, elav-GAL4; bure10/Bc flies were crossed to bure10/Bc; UAS-bur flies. R-cell projection pattern in elav-GAL4/+; bure10/bure10; UAS-bur/+ was compared to that in bure10/bure10; UAS-bur larvae. Similar genetic schemes were used to examine the rescue of the phenotype in burP mutants by eye- or neuronal-specific expression of bur, and also used to examine if UAS-mcm10 or UAS-Ret rescues the R-cell guidance phenotype in bur mutants. Guanine-depleted fly food was prepared similarly as described previously

(Johnstone et al. 1985).

Molecular Biology

The 3.3 kb fragment from the bur RE18382 EST clone containing the full-length bur cDNA was subcloned into the Not I and Xba I sites of the pUAST vector. The UAS- 132

mcm10 construct was generated by subcloning the 2.9kb Xho I-Not I fragment from the

EST clone LD15957 (Research Genetics) containing the full-length mcm10 cDNA into the pUAST vector. To generate the UAS-Ret construct, the 3.06kb Xho I-Not I and 1.87kb

Xho I partial ret cDNA fragments were ligated together and subsequently subcloned into the Not I and Xho I sites of the pUAST vector. The resulting UAS constructs were introduced into flies to generate transgenic lines by using standard methods.

Histology and Immunohistochemistry

Eye-brain complexes from third-instar larvae were dissected and stained as described

(Ruan et al. 1999). Primary antibodies were used at the following dilutions: MAb 24B10

(1:200 dilution, DSHB), Bar (1:200 dilution), Boss (1:2000 dilution), Elav (1:10 dilution),

Repo (1:10 dilution, DSHB), and β-galactosidase (1:1000 dilution). The secondary antibodies (i.e. HRP-, Texas-red- and FITC-conjugated goat anti-rabbit or anti-mouse secondary antibodies) (Jackson Immunochemicals) were used at 1:200 dilution.

Epifluorescent images of Repo and β-galactosidase double staining were captured using a high-resolution fluorescence imaging system (Canberra Packard) and analyzed by 2D

Deconvolution using MetaMorph imaging software (Universal Imaging, Brandywine,

PA).

The severity of the R-cell axon hyper-fasciculation phenotype was quantified by counting the number of separate and distinct R-cell axon bundles that were located between lamina and medulla. In bur mutants, this number decreased dramatically due to the hyperfasciculation of axon bundles.

133

Results

Identification and characterization of the bur gene:

In a search for genes required for R-cell axon guidance in the fly visual system, we found that larvae homozygous for each of two bur mutant alleles l(2)k07130 (burP) and ES2-2e10

(bure10) displayed severe R-cell axon guidance defects in the optic lobe. burP is a P insertion mutation, while bure10 is an EMS-induced mutation isolated originally as an enhancer of the rough-eye phenotype caused by misexpression of sina (Neufeld et al.

1998). The detailed phenotypic analysis is described below.

Both burP and bure10 are allelic to bur1 located at the cytological region 39B1-2, which has been shown previously to be auxotrophic for guanosine (Johnstone et al. 1985), suggesting that the bur gene encodes for an enzyme required for the de novo GMP synthesis. Consistently, the BDGP/Celera Drosophila Genome Project predicts a gene encoding a GMP synthetase at 39B1-2, whose homologs in other species catalyze the amination of XMP to GMP at the final step of the de novo GMP synthesis. We performed plasmid rescue to determine the P-element insertion site in burP and found that the P element is inserted into the first exon of this GMP synthetase gene (Figure 1A), 35 base pairs downstream of the putative transcriptional start site. To determine the mutation site in bure10, we amplified the predicted coding sequence of the bur gene in bure10 by polymerase chain reaction (PCR) and subsequently sequenced the PCR fragments. A nonsense mutation was identified in the bur coding sequence, which leads to the pre-mature termination of protein translation at the amino-terminal amino-acid residue 61 in bure10.

134

Thus, bure10 appears to be a genetically null allele, while phenotypic analysis indicates that burP is a hypomorphic allele (see below).

To further characterize the bur gene, we obtained an EST cDNA clone (i.e. RE18382) from Research Genetics. The cDNA clone was completely sequenced and compared to the bur genomic sequence available from the BDGP/Celera Drosophila Genome Project.

We found that the bur gene consists of 10 exons and 9 introns (Figure 1A). Conceptual translation of the cDNA sequence reveals an open reading frame of 683 amino acids

(Figure 1B). To further determine if bur is indeed the corresponding gene for the guidance phenotype in bur mutants, we subcloned the bur cDNA clone and the full-length cDNAs from the two nearby genes mcm10 and Ret into the pUAST vector to generate transgenic lines for rescue experiments. The mcm10 and Ret genes are located 244 and

2697 base pairs upstream of the putative bur transcriptional initiation site, respectively.

The mcm10 gene encodes a protein that binds to the members of pre-replicative complex and is required for chromosome condensation (Christensen and Tye 2003), while the Ret gene encodes an evolutionarily conserved receptor tyrosine kinase (Huen et al. 2000;

Hahn and Bishop 2001). We found that the expression of the bur transgene, but not mcm10 nor Ret, in homozygous burP (data not shown) or bure10 mutants (see below), almost completely rescued the phenotype, confirming that the lesion in the bur gene is responsible for the R-cell guidance phenotype in bur mutants.

The bur gene encodes a GMP synthetase:

135

The Bur protein is highly homologous to the members of the evolutionarily conserved

GMP synthetases in other species. The amino-acid identity between Bur and the human

GMP synthetase over the entire polypeptide is 56.8% (Figure 1B), and the percentage of additional conserved amino-acid changes between the two proteins is 17%. The similarity between Bur and the Escherichia coli GMP synthetase GuaA is also highly significant with 31.2% amino-acid identity and 16% conserved changes. Like the human GMP synthetase, the Bur polypeptide contains three distinct domains, including the amino- terminal glutamine amidotransferase class-I domain (GATase-1) (aa 20-230), a central region (aa 230-290) containing ATP-binding-3 PP-loop consensus sequence, and the highly conserved carboxyl-terminal domain (aa 429-683) (Figure 1B). The GATase-1 domain in GMP synthetases is responsible for binding glutamine and catalyzing its hydrolysis. The cysteine 95 residue in this domain, conserved in all known glutamine amidotransferases, binds to glutamine and forms a glutamyl γ-thioester intermediate during the catalytic reaction. The PP-loop motif is found in many ATP pyrophosphatases

(Bork and Koonin 1994), and may play a role in the hydrolysis of ATP into AMP and inorganic pyrophosphate, which generates the energy to drive the amination reaction. To generate GMP by amination of XMP, the remaining carboxyl-terminal sequence may bind XMP and subsequently add to XMP the amino group resulting from glutamine hydrolysis (Zalkin et al. 1985).

bur mutations caused severe R-cell axon pathfinding defects:

To determine the effect of bur mutations on R-cell projections, R-cell axons in homozygous bur mutants were stained using the monoclonal antibody 24B10 that

136

recognizes Chaoptin, a R-cell-specific cell adhesion molecule (Van Vactor et al. 1988). In wild type (Figure 2A), each differentiating R-cell cluster or ommatidium sends out a single axon bundle consisting of eight axons toward the most posterior end of the eye- imaginal disc, where it converges with axon bundles from other ommatidia and subsequently enters the optic stalk. After exiting the optic stalk, R-cell axons project evenly over the superficial lamina layer. R1-6 axons terminate within the lamina, and their growth cones expand significantly in size and establish a smooth termination layer.

R7 and R8 axons project through the lamina into appropriate topographic locations in the medulla, where their growth cones also expand and display a characteristic “Y-like” morphology.

In homozygous burP mutant larvae (Figure 2B), R-cell axons migrated properly toward the posterior end of the eye disc and then entered the optic stalk normally. The axon tract within the optic stalk was morphologically indistinguishable from that in wild type. After

R-cell axons exited the optic stalk, however, they failed to maintain appropriate neighbour relationships. Many R-cell axons formed abnormally thicker bundles in the lamina and medulla. The retinotopic array of R7 and R8 growth cones within the medulla appeared much less organized than that in wild type. Crossing over of neighbouring axon bundles was frequently observed in the medulla. R7 and R8 growth cones were less expanded and displayed irregular morphology. The phenotype in homozygous bure10 mutant larvae (Figure 2C) was much more severe than that in burP (Figure 2B). In all bure10 mutant individuals examined (n>60 hemispheres), most if not all R-cell axons were present within abnormally large bundles, causing the appearance of hyper- and hypo-

137

innervated regions (Figure 2C). The R-cell terminal field in bur mutants was much smaller than that in wild type. R7 and R8 growth cones in the medulla failed to expand, and most of them could not be identified individually. These phenotypes were not enhanced when bure10 was placed over the deficiency Df(2L)TW161 (100%, n>10 hemispheres) (data not shown), consistent with that bure10 is a null allele (see above).

To determine if R-cell projection phenotype in bur mutants reflects a cell-autonomous role for bur in R-cell axons, we used the MARCM system (mosaic analysis with a repressible cell marker) (Lee and Luo 1999) to examine the projection pattern of positively labeled homozygous bur mutant axons in an otherwise heterozygous or wild- type larvae. In controls (Figure 2D), the vast majority of labeled wild-type R-cell axons project into appropriate topographic locations (40 out of 41 labeled axons examined), where their growth cones expand significantly in size (36 out of 41 labeled axons examined, Figure 2D). By contrast, many labeled bur mutant axons did not expand upon reaching the target region (~89% penetrance, n=38, Figure 2E), and some displayed abnormal topographic projection pattern (~16% penetrance, n=38, Figure 2F). These results suggest a cell-autonomous role for bur in R-cell axonal projections.

bur mutations did not affect lamina-specific termination of R2-R5 axons:

To determine if the above guidance defects affect the binary lamina-versus-medulla target selection of R-cell axons, we used the marker ro--lacZ to examine the targeting of R2-

R5, a subset of R1-R6 axons, in third-instar bure10 mutant larvae. Surprisingly, although

R-cell axons displayed severe pathfinding defects in the optic lobe (see above), we found

138

that the vast majority of R2-R5 axons in bure10 mutants (Figure 3B), like that in wild type

(Figure 3A), terminated correctly within the lamina.

Previous studies demonstrate complex interactions between R-cell axons and lamina glial cells: R-cell axons induce the differentiation and migration of lamina glial cells (Perez and Steller 1996; Suh et al. 2002; Dearborn and Kunes 2004), and conversely lamina glial cells present a stop signal for terminating R1-R6 axons within the lamina (Poeck et al.

2001). To determine the potential effect of bur mutations on the development of lamina glial cells, we stained the third-instar optic lobe using a monoclonal antibody that recognizes the glial-specific nuclear protein Repo. In wild type (Figure 3C), differentiating glial cells migrate into the lamina in response to an unknown signal from

R-cell axons (Suh et al. 2002), forming two layers of glial cells (i. e., epithelial and marginal glia), which in turn present a stop signal for terminating R1-R6 growth cones in the lamina (Poeck et al. 2001). In bure10 mutants, the differentiation and migration of lamina glial cells appeared largely normal (Figure 3D). That R2-R5 axons still terminated correctly in bur mutants (Figure 3B) indicates that these lamina glial cells were still able to present correct targeting information for R1-R6 axons in the lamina. The density of glial cells in the lamina, however, was increased significantly, which was likely due to the reduction in the R-cell terminal field.

In summary, although loss of bur caused severe defects in R-cell fasciculation, retinotopy and growth-cone morphology, it did not affect layer-specific termination of R2-R5 axons.

139

R-cell differentiation and patterning in the developing eye disc occurred normally in bur mutants:

The bur null allele bure10 was originally isolated as a strong enhancer of the sina- misexpression-induced rough-eye phenotype (Neufeld et al. 1998). However, analysis of the overall patterning of mutant ommatidia in adult bure10 mosaic animals did not reveal any defect in either the organization of ommatidia or the number and positioning of R-cell cell bodies within each ommatidium (Neufeld et al. 1998). This data argues against the possibility that the severe pathfinding phenotype in bur mutants was secondary to the defects in R-cell differentiation or ommatidial organization. To confirm this, we examined the development of the eye disc in bur mutants at the third-instar larval stage using cell-type-specific markers including Chaoptin (R1-R8 cell bodies) (Figure 4, A and

E), Elav (R1-R8 nuclei) (Figure 4, B and F), Bar (R1 and R6 nuclei) (Figure 4, C and G), and Boss (R8 cell) (Figure 4, D and H). No defect in R-cell differentiation or the organization of R-cell clusters was detected in homozygous bure10 mutant eye-imaginal discs. Thus, R-cell guidance defects in bur mutants were unlikely to be due to abnormal

R-cell differentiation or irregular spacing of ommatidial clusters in the developing eye.

Axon guidance phenotype in bur mutants could be rescued by cell-type-specific expression of a bur transgene:

Genetic mosaic analysis suggests a cell-autonomous role for bur in R-cell axon guidance

(Figure 2E and 2F). To further confirm that R-cell axonal projection defects in bur mutants were indeed caused by loss of bur in R cells, we performed cell-type-specific rescue experiments by expressing an UAS-bur transgene in homozygous bur mutants

140

under control of the neuronal-specific elav-GAL4 or eye-specific GMR-GAL4 driver. We found that both neuronal-specific and eye-specific expression of the UAS-bur transgene almost completely rescued the guidance defects in bur mutants (compare Figure 5B and

5F to 5A, table 1). As R cells are the only neuronal cell types in the developing eye, these results, taken together with that from genetic mosaic analysis (Figure 2E and 2F), indicate that bur is specifically required in R cells for the proper guidance of R-cell axons in the optic lobe.

Although many bure10 mutants can reach third-instar larval stage, all of them die at pupation, indicating that bur is also required in other tissues for viability. We examined the ability of the bur transgene to rescue the pupal lethality. Three independent UAS-bur transgenes were used under control of the neuronal-specific elav-GAL4 driver in the rescue experiments. We found that all three UAS-bur transgenes could substantially rescue the pupal lethality. The neuronal-specific expression of one UAS-bur transgene allowed ~50% of bure10 homozygous mutants to reach adult stage, while ~26% and 34% survival rate were obtained by expressing another two UAS-bur transgenes. As the compound eye is not essential for viability, the rescue of the bur pupal lethality by expressing the UAS-bur transgene in post-mitotic neuronal cells raises the interesting possibility that bur is also required in neurons in other tissues for the wiring of neural networks.

Raspberry is also required for R-cell axon pathfinding:

141

The bur phenotype may reflect a role for de novo GMP synthesis in R-cell axon guidance.

Alternatively, bur may function in other processes to regulate R-cell axonal projections.

To distinguish among these possibilities, we examined if other enzymes involved in the de novo GMP synthesis are also required for R-cell axon guidance. Mutant alleles for the raspberry (ras) gene (Slee and Bownes 1995), encoding for inosine monophosphate dehydrogenase (IMPDH) that catalyzes the first step of the IMP-XMP-GMP conversion during de novo GMP synthesis, were analyzed for potential defects in R-cell projections.

Indeed, we found that loss of ras caused a R-cell axon guidance phenotype identical to that in bur mutants using three different ras alleles (Compare Figure 6C and 6D to 6B).

These results indicate an essential role for de novo GMP synthesis in R-cell axon guidance.

de novo GMP synthesis is sufficient for maintaining the GMP level required for R- cell axon guidance:

The above results establish an essential role for de novo GMP synthesis in maintaining the necessary level of GMP for R-cell axon guidance. To further determine the contribution of the GMP salvage synthesis pathway, we examined the effect of depleting guanine from fly food on R-cell projections. Wild-type fly embryos could still survive and develop into adults when fed with guanine-depleted food. In contrast, when burP or bure10 homozygous mutants were fed with guanine-depleted food, none of them could survive to the third-instar larval stage (data not shown), confirming the role of Bur in de novo GMP synthesis. Examination of R-cell projection pattern in wild-type larvae fed with guanine-depleted food did not reveal any obvious defect (Figure 7B), suggesting that

142

the level of GMP maintained by de novo GMP synthesis is sufficient for R-cell axon guidance.

bur interacts with Rac genetically:

After synthesis, GMP can be converted into cGMP and GTP, both are important signaling molecules. Interestingly, Rac, the Rho family small GTPase whose activation requires

GTP, has been shown previously to be required for the proper fasciculation of R-cell axons in the Drosophila visual system (Hakeda-Suzuki et al. 2002). To address the possibility that de novo synthesis of GMP is required for maintaining a high level of GTP that allows the proper function of Rac in R-cell axons, we examined the potential genetic interaction between bur and Rac during R-cell axon guidance. When the Rac1 dominant- negative transgene RacN17 was expressed in R cells, we observed a weak axonal hyperfasciculation phenotype (Figure 8A). This phenotype was enhanced when the dosage of bur was reduced by half (Figure 8B). Similar enhancement was observed when the dosage of bur was reduced in larvae expressing another Rac1 dominant-negative transgene RacL89 (compare Figure 8D to 8C). These results are consistent with a requirement for de novo GMP synthesis in the up-regulation of Rac signaling in R-cell axons.

143

144

Figure 1. Molecular characterization of the bur gene.

(A) Predicted genomic organization of the bur and two nearby genes mcm 10 and Ret at

39B1-2 by the BDGP/Celera Drosophila genome project. The exon-intron boundaries in the bur gene were determined by comparing the sequence of the full-length EST cDNA clone RE18382 to the genomic sequence. The P-element insertion site for burP is located within exon 1, 35 bp downstream of the putative bur transcriptional start site. In bure10, a

C-to-T nonsense mutation in the third exon changes the codon for R61 into a stop codon.

Open boxes, noncoding regions; filled boxes, coding regions. (B) Alignment of the Bur protein sequence with GMP synthetases in human (Hirst et al. 1994) and Escherichia coli

(Tiedeman et al. 1985). Identical residues are boxed.

145

146

Figure 2. Mutations in the bur gene disrupted R-cell axon pathfinding.

(A-C) R1-R8 axons in third-instar larvae were stained with MAb 24B10. In wild type (A),

R-cell axons elaborate smooth retinotopic arrays in the lamina (la) and medulla (me). The expanded R1-R6 growth cones form a continuous layer in the lamina. R7 and R8 axons establish a smooth topographic array in the medulla, where individual expanded “Y- shaped” growth cones can be readily identified. (B) In a burP homozygote, thicker bundles (arrow) were frequently observed. Growth cones displayed irregular morphology.

The array of R7 and R8 growth cones in the medulla was disorganized. (C) In a bure10 homozygote, the phenotype was much more severe. Axon bundles were much thicker than that in the burP mutant (B). Most R7 and R8 growth cones failed to expand and could not be identified individually. Crossing over of axons (arrow) was frequently observed. (D-F) Wild-type (D) or bur mutant axons (E and F) were labeled with the

MARCM method (Lee and Luo 1999). (D) Labeled wild-type axons projected into appropriate topographic locations. Their growth cones expanded significantly when reaching the target region (arrow). (E) Labeled burP mutant axons in a mosaic larvae did not expand their growth cones (arrow). (F) A labeled burP mutant axon turned abnormally along the dorsal-ventral axis and projected into an incorrect topographic location in medulla (arrow). Scale bar: 20 μm.

147

148

Figure 3. bur is not required for lamina-specific termination of R2-R5 axons.

(A and B) R2-R5 axons in wild type (A) and bure10 mutant (B) were labeled with the larval R2-5 marker ro--lacZ. In wild type (A), the vast majority of ro--lacZ labeled R2-

R5 axons terminated within the lamina (la). Only a few (i.e. 2-5) R2-R5 axons mistarget into the medulla (me). In a bure10 homozygote (B), the average number of mistargeted

R2-R5 axons was 6 (n=22 hemispheres), which was not significantly different from that in wild type (A). (C and D) R-cell axons (red) and lamina glial cells (green) in wild-type

(C) and bure10 (D) third-instar larvae were double stained with anti-β-galactosidase antibody and anti-Repo antibodies, respectively. Both wild type and bure10 mutants carry a glass-lacZ transgene in which the expression of lacZ is under control of the eye-specific glass promoter (Mismer and Rubin 1987), which allows the visualization of all R-cell axons with anti-β-galactosidase staining. Anti-Repo recognizes a glial-specific nuclear protein. In wild type (C), glial cells (green) migrate from progenitor regions into the lamina where they are organized into two layers, the epithelial (eg) and marginal glia

(mg), presenting a stop signal for the termination of R1-R6 growth cones (red) at the lamina plexus (lp). (D) In a bure10 homozygote, although glial cells (green) migrated correctly into the lamina, they appeared less organized. Scale bar: 20 μm.

149

150

Figure 4. R-cell differentiation and patterning remained normal in bur mutants.

(A and E) All R-cell bodies in the third-instar wild-type (A) and bure10 mutant (E) eye disc were visualized with MAb 24B10 staining. In wild type (A), ommatidial clusters of

R-cell bodies are highly organized. (E) In a homozygous bure10 mutant eye disc, R-cell clusters maintained an appropriate neighbor relationship (n>30 discs). (B and F)

Differentiating R1-R8 nuclei in the third-instar wild-type (B) and bure10 mutant (F) eye disc were stained with anti-Elav antibody, which recognizes the pan-neuronal nuclear protein Elav. The organization of R-cell clusters in bure10 mutant (F) (n=44 discs) was indistinguishable from that in wild type (B). (C and G) R1 and R6 nuclei in the third- instar wild-type (C) and bure10 mutant (G) eye disc were stained with anti-Bar antibody.

No obvious defect was observed in bure10 mutant (n=16 discs). (D and H) R8 cell bodies in the third-instar wild-type (D) and bure10 mutant (H) eye disc were stained with anti-

Boss antibody. In a bure10 homozygote (H), like that in wild type (D), each ommatidium only contains a single R8 cell (n=14 discs). Scale bar: 20 μm.

151

152

Figure 5. Rescue of the R-cell pathfinding phenotype by expressing the bur transgene in R cells.

Third-instar eye-brain complexes were stained with MAb 24B10 to visualize R1-R8 axons. In a bure10 homozygote (A), R-cell axons displayed hyper-fasciculation, aberrant topographic order and growth-cone morphology. Genotype: bure10, GMR-GAL4/ bure10.

In a bure10 homozygote carrying an UAS-bur transgene under control of the eye-specific

GMR-GAL4 driver (B), R-cell projection pattern was indistinguishable from that in wild type (see Fig. 2A). Note the appearance of expanded “Y-like” growth cones in the medulla. Genotype: bure10, GMR-GAL4/ bure10; UAS-bur/+. C and D are enlarged views of the boxed regions in A and B, respectively. (E) In a bure10 homozygote expressing an

UAS-mcm 10 transgene, or an UAS-Ret transgene (data not shown), R-cell axons still displayed severe pathfinding defects. Genotype: bure10, GMR-GAL4/ bure10; UAS-mcm

10/+. (F) In a bure10 homozygote carrying an UAS-bur transgene under control of the neuronal-specific elav-GAL4 driver, R-cell axons displayed wild-type-like innervation pattern in the optic lobe. Genotype: elav-GAL4/+; bure10/ bure10; UAS-bur/+. G and H are enlarged views of the boxed regions in E and F, respectively. Scale bar, 20 μm in A, B, E, and F; 5 μm in C, D, G, and H.

153

154

Figure 6. Mutations in the ras gene caused a bur-like phenotype in R-cell axon guidance.

Third-instar eye-brain complexes were stained with MAb 24B10. (A) Wild type. (B) bure10 homozygote. (C) In a rasG0002 hemizygote, R-cell axons displayed a severe axon guidance phenotype (24 out of 26 hemispheres examined) that was very similar to that in bure10. (D) A temperature-sensitive ras11 homozygote (100% penetrance, n=92 hemispheres) grown at restrictive temperature (i.e. 29 oC). Scale bar, 20 μm.

155

156

Figure 7. Depleting guanine from the fly food did not affect R-cell axon guidance.

Third-instar eye-brain complexes were stained with MAb 24B10. (A) R-cell projection pattern in a wild-type larvae grown on normal food. (B) In a wild-type larvae grown on guanine-depleted food, R-cell projection pattern remained normal (n=50 hemispheres).

Scale bar, 20 μm.

157

158

Figure 8. Reducing the dosage of bur enhanced the Rac phenotype.

Third-instar eye-brain complexes were stained with MAb 24B10. (A) R-cell projection pattern in a wild-type larvae expressing the dominant-negative RacN17 transgene under control of the eye-specific GMR-GAL4 driver. The average number (25±5, n=12) of clearly separated axonal bundles between lamina and medulla was decreased compared to that in wild type (31±6) due to axonal hyper-fasciculation. Genotype: GMR-

GAL4/+;UAS-RacN17/+;. (B) In a RacN17-expressing larvae in which the dosage of bur was reduced by 50%, the phenotype became more severe. Compared to that in A (25±5, n=12), the average number (21±4, n=22) of axonal bundles was decreased (P<0.05).

Genotype: GMR-GAL4, bure10/+;UAS-RacN17/+. (C) In a larvae expressing the dominant-negative RacL89 transgene, R-cell projection pattern was severely disrupted.

Genotype: GMR-GAL4/UAS-RacL89. (D) In a RacL89-expressing larvae in which the dosage of bur was reduced by 50%, R-cell axonal hyper-fasciculation phenotype was enhanced. Compared to that in C (9±5, n=9), the average number (5±2, n=9) of axonal bundles was significantly decreased (P<0.05). Genotype: GMR-GAL4, bure10/UAS-

RacL89. Scale bar, 20 μm.

159

Table 1. Transgenic rescue of R-cell axonal hyperfasciculation phenotype in bur mutants

Genotype Average number of axons or axon bundles Number of

per hemispherea hemispheres

examined wt 31 ± 6 8 bure10/ bure10 8 ± 2 8 bure10,GMR-GAL4/ bure10 8 ± 1 8 bure10,GMR-bur b (1c) 30 ± 9 20 bure10,GMR-bur (2) 18 ± 2 17 bure10,GMR-bur (3) 31 ± 8 31 bure10,elav-bur d (1) 23 ± 8 36 bure10,elav-bur (2) 30 ± 5 33 bure10,elav-bur (3) 29 ± 10 15 a The number of axons or axon bundles that clearly separated from neighboring bundles at the region between lamina and medulla were counted. This number decreased significantly in bur mutants due to the formation of thicker bundles. b GMR-bur refers to GMR-GAL4+UAS-bur. c The number indicates an independent transgenic line. d elav-bur refers to elav-GAL4 (C155)+UAS-bur.

160

Discussion

In this study, we show that the bur gene is required for R-cell axon guidance in the fly visual system. In bur mutants, R-cell axons displayed hyper-fasciculation, disrupted topographic innervation and abnormal growth-cone morphology. Phenotypic analysis using a collection of cell-type-specific markers shows that the pathfinding phenotype was not caused by abnormal R-cell differentiation or retinal patterning. Genetic mosaic analysis and cell-type-specific rescue experiments demonstrate that bur is required in R cells. bur encodes a GMP synthetase that catalyzes the final reaction of the de novo GMP synthesis, the amination of XMP to GMP. Like bur, mutations in the ras gene encoding for IMPDH catalyzing the conversion of IMP to XMP in de novo GMP synthesis, also disrupted R-cell axon guidance. Bur displayed dosage-sensitive genetic interaction with the GTP-binding protein Rac. We propose that the de novo synthesis of GMP is essential for maintaining a high level of guanine nucleotides, which are required for growth-cone signaling during axon pathfinding.

bur has been shown previously to interact with sina (Neufeld et al. 1998), which encodes for a transcriptional factor required for specifying cell fate determination in the fly eye

(Carthew and Rubin 1990). Reducing the dosage of bur enhanced the sina-overexpression eye phenotype (Neufeld et al. 1998). This interaction, however, does not appear to reflect a relevant in vivo interaction between bur and sina due to two reasons. First, unlike some other enhancers (e.g. Sin3Aes2-3) of the sina-overexpression eye phenotype identified from the same screen, bur did not display a dosage-sensitive genetic interaction with a hypomorphic sina loss-of-function allele (Neufeld et al. 1998). And second, unlike loss of 161

sina, loss of bur did not affect R-cell fate determination (Figure 4, (Neufeld et al. 1998)).

These results argue strongly against that Bur is a component of the Sina signaling pathway in R cells.

In mammals, de novo GMP synthesis has been shown to be required for cell proliferation during immunoresponse. Inhibitors of IMPDH, which catalyzes the conversion of IMP into XMP during de novo GMP synthesis, could block lymphocyte proliferation leading to immunosuppression (Sollinger et al. 1992; Eugui and Allison 1993). Consistently, it has been shown that the proliferation of lymphocytes could be significantly suppressed when the genes encoding for IMPDH were disrupted in mice (Gu et al. 2000; Gu et al.

2003). Surprisingly, no defect in cell proliferation was observed in bur and ras mutants.

The size of the mutant disc and the number of differentiating R cells were indistinguishable from that in wild type. Importantly, the expression of bur in post-mitotic neurons could completely rescue the R-cell guidance phenotype and also largely rescue the pupal lethality, arguing strongly against a role for Bur in cell proliferation. One likely explanation is that in the absence of Bur, the salvage GMP synthesis pathway maintains a certain level of GMP, which may be sufficient for cell proliferation during development.

This level of GMP generated by the salvage pathway, however, is not sufficient for certain signaling events necessary for specific growth-cone guidance decisions.

How does the de novo synthesis of guanine nucleotides play a role in R-cell growth-cone guidance? We speculate that de novo GMP synthesis is required for generating a certain level of GTP necessary for R-cell growth-cone signaling. The GTP-binding protein Rac, a

162

member of the Rho-family small GTPases, has been shown to be required for R-cell axon guidance in Drosophila (Hakeda-Suzuki et al. 2002). Like loss of bur, loss of Rac or loss of its activator Trio also caused defects in R-cell fasciculation and growth-cone expansion

(Newsome et al. 2000; Hakeda-Suzuki et al. 2002). Our results showing that bur genetically interacts with Rac during R-cell axon guidance (Figure 8), are consistent with the notion that de novo GMP synthesis is required for up-regulating Rac activity.

Alternatively or additionally, de novo GMP synthesis may also be required for maintaining a certain level of cGMP in R-cell growth cones for guidance decisions.

Pharmacological studies demonstrate that an increase in the cGMP activity is necessary for the repulsive response of Xenopus spinal cord neuron growth cones to netrin-1

(Nishiyama et al. 2003). It is possible that a similar cGMP-dependent repulsive interaction exists between R-cell axons, which is necessary for maintaining an appropriate neighbor relationship between R-cell axons. If so, loss of bur could cause a decrease in the level of cGMP, which disrupts the repulsive interaction between neighboring axons resulting in the hyper-fasciculation phenotype. Future studies are necessary to determine if cGMP signaling also plays a role in R-cell axon guidance.

The human genome carries a single GMP synthetase gene (Hirst et al. 1994), which exhibits extensive homology to the fly Bur (Figure 1B). It is highly possible that the de novo GMP synthesis mediated by this enzyme is also required for the wiring of neural network during the development of the human brain. In this context, it is notable that mutations in the gene encoding for adenylosuccinate lyase (ADSL), which is required for the de novo synthesis of both AMP and GMP (Zalkin and Dixon 1992), cause severe

163

mental retardation and autism (Jaeken and Van den Berghe 1984). How the ADSL deficiency causes mental retardation and autism is still unclear. Given the specific guidance phenotype in the fly bur mutants, it will be of interest to determine if these neurological disorders in ADSL deficiency patients are caused by malformation of neuronal connections during embryonic development. It will also be interesting to determine if any other human neurological diseases are caused by mutations in the GMP synthetase gene or genes encoding for other enzymes of the de novo GMP synthesis pathway.

164

Acknowledgements

We thank Don van Meyel for critical reading of the manuscript; the members of the Rao lab for helpful discussions; the BDGP and Bloomington Stock Center for fly stocks;

DSHB at University of Iowa for MAb 24B10 and anti-Repo antibodies; David Huen for

Ret cDNAs; Gerald Rubin for bure10. This work was supported by an operating grant

(MOP-14688) awarded to Yong Rao from Canadian Institutes of Health Research.

165

References

1. Bork, P., Koonin, E. V. (1994). A P-loop-like motif in a widespread ATP

pyrophosphatase domain: implications for the evolution of sequence motifs and

enzyme activity. Proteins 20, 347-55.

2. Carthew, R. W., Rubin, G. M. (1990). seven in absentia, a gene required for

specification of R7 cell fate in the Drosophila eye. Cell 63, 561-77.

3. Christensen, T. W., Tye, B. K. (2003). Drosophila MCM10 interacts with members

of the prereplication complex and is required for proper chromosome condensation.

Mol Biol Cell 14, 2206-15.

4. Clandinin, T. R., Zipursky, S. L. (2002). Making connections in the fly visual system.

Neuron 35, 827-41.

5. Dayton, J. S., et al., (1992). Comparison of the effects of mizoribine with those of

azathioprine, 6-mercaptopurine, and mycophenolic acid on T lymphocyte

proliferation and purine ribonucleotide metabolism. Mol Pharmacol 41, 671-76.

6. Dearborn, R., Jr., and Kunes, S. (2004). An axon scaffold induced by retinal axons

directs glia to destinations in the Drosophila optic lobe. Development 131, 2291-03.

7. Dickson, B. J., (2001). Rho GTPases in growth cone guidance. Curr Opin Neurobiol

11, 103-10.

8. Eugui, E. M., and Allison, A. C. (1993). Immunosuppressive activity of

mycophenolate mofetil. Ann N Y Acad Sci 685, 309-29.

9. Eugui, E. M., et al., (1991). Lymphocyte-selective cytostatic and immunosuppressive

effects of mycophenolic acid in vitro: role of deoxyguanosine nucleotide depletion.

Scand J Immunol 33, 161-73. 166

10. Gu, J. J., et al., (2000). Inhibition of T lymphocyte activation in mice heterozygous

for loss of the IMPDH II gene. J Clin Invest 106, 599-606.

11. Gu, J. J., et al., (2003). Targeted disruption of the inosine 5'-monophosphate

dehydrogenase type I gene in mice. Mol Cell Biol 23: 6702-12.

12. Guan, K. L., Rao, Y., (2003). Signalling mechanisms mediating neuronal responses

to guidance cues. Nat Rev Neurosci 4, 941-56.

13. Hahn, M., and Bishop, J. (2001). Expression pattern of Drosophila ret suggests a

common ancestral origin between the metamorphosis precursors in insect endoderm

and the vertebrate enteric neurons. Proc Natl Acad Sci U S A 98, 1053-58.

14. Hakeda-Suzuki, S., et al., (2002). Rac function and regulation during Drosophila

development. Nature 416, 438-42.

15. Hauser, I. A., and Sterzel, R. B. (1999). Mycophenolate mofetil: therapeutic

applications in kidney transplantation and immune-mediated renal disease. Curr

Opin Nephrol Hypertens 8, 1-6.

16. Hirst, M., et al., (1994). Human GMP synthetase. Protein purification, cloning, and

functional expression of cDNA. J Biol Chem 269, 23830-37.

17. Huen, D. S., et al., (2000). The Drosophila Ret gene is transcribed in multiple

alternatively spliced forms. Mol Gen Genet 264, 335-40.

18. Jaeken, J., and Van den Berghe, G. (1984). An infantile autistic syndrome

characterised by the presence of succinylpurines in body fluids. Lancet 2, 1058-61.

19. Johnstone, M. E., et al., (1985). Three purine auxotrophic loci on the second

chromosome of Drosophila melanogaster. Biochem Genet 23, 539-55.

167

20. Lee, T., and Luo, L. (1999). Mosaic analysis with a repressible cell marker for

studies of gene function in neuronal morphogenesis. Neuron 22, 451-61.

21. Luo, L., (2000). Rho GTPases in neuronal morphogenesis. Nat Rev Neurosci 1, 173-

80.

22. Meinertzhagen, I. A., and Hanson, T. E. (1993). The development of the optic lobe,

pp. 1363-1491 in The Development of Drosophila melanogaster, edited by M. Bate

and A. Martinez-Arias. Cold Spring Harbor Press, Cold Spring Harbor, NY.

23. Mismer, D., and Rubin, G. M. (1987). Analysis of the promoter of the ninaE opsin

gene in Drosophila melanogaster. Genetics 116, 565-78.

24. Neufeld, T. P., et al., (1998). A genetic screen to identify components of the sina

signaling pathway in Drosophila eye development. Genetics 148, 277-86.

25. Newsome, T. P., et al., (2000). Trio combines with dock to regulate Pak activity

during photoreceptor axon pathfinding in Drosophila. Cell 101, 283-94.

26. Nishiyama, M., et al., (2003). Cyclic AMP/GMP-dependent modulation of Ca2+

channels sets the polarity of nerve growth-cone turning. Nature 423, 990-95.

27. Perez, S. E., and Steller, H. (1996). Migration of glial cells into retinal axon target

field in Drosophila melanogaster. J Neurobiol 30, 359-73.

28. Poeck, B., et al., (2001). Glial cells mediate target layer selection of retinal axons in

the developing visual system of Drosophila. Neuron 29, 99-113.

29. Ruan, W., et al., (1999). The SH2/SH3 adaptor protein dock interacts with the Ste20-

like kinase misshapen in controlling growth cone motility. Neuron 24, 595-605.

30. Slee, R., and Bownes, M. (1995). The raspberry locus encodes Drosophila inosine

monophosphate dehydrogenase. Mol Gen Genet 248, 755-66.

168

31. Sollinger, H. W., et al., (1992). RS-61443 (mycophenolate mofetil). A multicenter

study for refractory kidney transplant rejection. Ann Surg 216, 513-18; discussion

518-519.

32. Song, H., et al., (1998). Conversion of neuronal growth cone responses from

repulsion to attraction by cyclic nucleotides. Science 281, 1515-18.

33. Song, H. J., et al., (1997). cAMP-induced switching in turning direction of nerve

growth cones. Nature 388, 275-79.

34. Suh, G. S., et al., (2002). Drosophila JAB1/CSN5 acts in photoreceptor cells to

induce glial cells. Neuron 33, 35-46.

35. Tayler, T. D., and Garrity, P. A. (2003). Axon targeting in the Drosophila visual

system. Curr Opin Neurobiol 13, 90-95.

36. Terman, J. R., and Kolodkin, A. L. (2004). Nervy links protein kinase A to plexin-

mediated semaphorin repulsion. Science 303, 1204-07.

37. Tessier-Lavigne, M., and Goodman, C. S. (1996). The molecular biology of axon

guidance. Science 274, 1123-33.

38. Tiedeman, A. A., et al., (1985). Nucleotide sequence of the guaA gene encoding

GMP synthetase of Escherichia coli K12. J Biol Chem 260, 8676-79.

39. Van Vactor, D., Jr., et al., (1988). Analysis of mutants in chaoptin, a photoreceptor

cell-specific glycoprotein in Drosophila, reveals its role in cellular morphogenesis.

Cell 52, 281-90.

40. Zalkin, H., et al., (1985). Identification of a trpG-related glutamine amide transfer

domain in Escherichia coli GMP synthetase. J Biol Chem 260, 3350-54.

169

41. Zalkin, H., and Dixon, J. E. (1992). De novo purine nucleotide biosynthesis. Prog

Nucleic Acid Res Mol Biol 42, 259-87.

42. Zollner, N. (1982). Purine and pyrimidine metabolism. Proc Nutr Soc 41, 329-42.

170

Chapter 3

Dendrite branching and self-avoidance are controlled by Turtle, a conserved IgSF protein in Drosophila

This work has been accepted for publication in Development (2009) 136, 3475-3484.

171

In a genetic screen for genes required for R-cell axon ganglion specific targeting, we identified gene turtle. Tutl belongs to the conserved IgSF family. During my analysis of the function of tutl in the fly visual system, I generated enhancer-trap flies of tutl.

Surprisingly, besides the expression pattern of tutl in the visual system, I also observed potential tutl expression in fly PNS dendritic arborization neurons (da neurons). This suggested that tutl might have some functions in PNS da neuron development.

Interestingly, researchers studying Dasm-1, the mammalian homolog of tutl, have proposed the involvement of Dasm-1 in hippocampal neuron dendritic arborization and synapse maturation. Based on these preliminary data, we decided to explore the function of Tutl in dendrite development. We found that tutl is required to inhibit excessive dendrite branch formation in class I da neurons, which have simple dendrite arbors. In class IV da neurons, which have complex dendrite arbors, tutl promotes dendrite self- avoidance. The cytoplasmic domain of Tutl is dispensable for its function in controlling dendrite branching in class I da neurons, suggesting that Tutl acts as a ligand or coreceptor for an unknown recognition molecule to regulate dendrites patterning.

172

Dendrite branching and self-avoidance are controlled by Turtle, a conserved IgSF protein in Drosophila

Hong Long, Yimiao Ou, Yong Rao and Donald J. van Meyel

Summary

The dendritic trees of neurons result from specific patterns of growth and branching, and dendrite branches of the same neuron avoid one another to spread over a particular receptive field. Recognition molecules on the surfaces of dendrites influence these patterning and avoidance processes by promoting attractive, repulsive or adhesive responses to specific cues. The Drosophila transmembrane protein Turtle (Tutl) and its orthologs in other species are conserved members of the immunoglobulin superfamily whose functions in vivo are unknown. In Drosophila sensory neurons, we show that the turtle gene is required to restrain dendrite branch formation in neurons with simple arbors, and to promote dendrite self-avoidance in neurons with complex arbors. The cytoplasmic tail of Tutl is dispensable for control of dendrite branching, suggesting that Tutl acts as a ligand or co-receptor for an unidentified recognition molecule to influence the architecture of dendrites and their coverage of receptive territories.

173

Introduction

Developing neurons form dendritic trees with cell type-specific patterns of arborisation, ranging from simple arbors with few branches to highly elaborate arbors that cover receptive territories with many branches. In neurons with even the most complex trees, dendrite branches growing from the same neuron (isoneuronal branches) avoid one another as they spread over a territory to receive sensory or synaptic inputs. Together, dendrite branching and self-avoidance are critical for sculpting the particular architecture of a neuron's receptive field. Both processes are thought to be controlled by molecular recognition events that occur between isoneuronal branches, or between dendrites and the substrata along which they grow. However, few of the molecules participating in these recognition events have been described.

Cell surface recognition molecules that promote dendrite growth and/or branching include cadherins (Gao et al., 2000; Kimura et al., 2006; Shima et al., 2007; Sweeney et al., 2002), as well as those mediating responses to neurotrophins (Horch and Katz, 2002), B-type ephrins (Horch and Katz, 2002) and cues that direct dendritic guidance such as

Semaphorins (Komiyama et al., 2007; Polleux et al., 2000), Slits (Dimitrova et al., 2008;

Furrer et al., 2007; Godenschwege et al., 2002; Whitford et al., 2002) and Netrins (Furrer et al., 2003). In contrast to these examples which promote or guide dendrite arborisation, recognition molecules that prevent inappropriate or excessive dendrite branching in neurons with simple arbors remain unidentified.

174

Recognition mechanisms underlying dendrite self-avoidance have only recently emerged, with findings that the Dscam family of immunoglobulin superfamily (IgSF) proteins promote self-avoidance in Drosophila (Hughes et al., 2007; Matthews et al., 2007; Soba et al., 2007) and mice (Fuerst et al., 2008). It remains to be determined whether other families of cell surface proteins are also required to promote self-avoidance.

Identification of novel transmembrane proteins required for dendrite branching and self- avoidance is a key step in understanding molecular mechanisms that underlie dendrite patterning. The Drosophila protein Turtle (Tutl) and its mammalian orthologs, Dasm-

1/Igsf9 in mice and KIAA1355/Igsf9 in humans (Doudney et al., 2002; Shi et al., 2004), are type 1 transmembrane proteins with an ectodomain comprised of 5 immunoglobulin

(Ig)-like domains and two fibronectin type III repeats (Fig.1A). In Drosophila, mutations of the turtle (tutl) gene impair responses to tactile stimuli and the execution of complex coordinated behaviors (Bodily et al., 2001), but the causes of these nervous system deficits are unknown. To date, no morphological defects have been reported for tutl mutants, despite the structural similarity of Tutl to the Neogenin, Deleted in Colorectal

Carcinoma, Frazzled and Roundabout families of axon guidance receptors (Bodily et al.,

2001).

In mice, the Tutl ortholog Dasm-1 is selectively expressed in the developing hippocampus (Mishra et al., 2008; Shi et al., 2004). Dasm-1 knockout mice have no observable defects in dendrite morphogenesis in the developing hippocampus, nor have defects of neuronal differentiation, synaptogenesis or behavior been seen in these mice

175

(Mishra et al., 2008). Therefore, the role Dasm-1 in the mammalian nervous system remains uncertain, and genetic approaches to study Dasm-1 function in mice could be complicated by redundancy of Dasm-1 with IgSF9b, a closely related protein (Mishra et al., 2008).

Here we have used genetic approaches to study the effects of tutl mutations on dendritic arborisation (da) neurons in the Drosophila peripheral nervous system. We found that

Tutl is expressed on dendrites of da neurons and, through loss-of-function and gain-of- function experiments in vivo, we demonstrate that Tutl cell-autonomously controls dendrite branching and self-avoidance. Tutl restricts branching in neurons with simple arbors and promotes self-avoidance in neurons with highly complex arbors. These results demonstrate that a member of the Tutl/Dasm-1/Igsf9 family of proteins can influence dendrite morphogenesis in vivo, and that neurons of different classes employ Tutl as a common molecular component of mechanisms that sculpt dramatically different patterns of arborisation complexity.

176

Materials and methods

Fly stocks and genetics

Flies were obtained from stock centers at Bloomington (i.e. tutl01085, tutlDf (Df(2L)ed-dp)) and Harvard (i.e tutlf03096 and tutlf02770) and from published sources (i.e. GAL4109(2)80,

GAL4221, ppk1.9-GAL4, UAS-mCD8::GFP). We generated tutl23 by FLP/FRT-mediated recombination (Parks et al., 2004) to delete the intervening DNA between tutlf03096 and tutlf02770, leaving a reconstituted WH PBac element at the deletion site (Fig. 1B), as confirmed by PCR and DNA sequencing. We generated UAS-tutl by cloning into the pUAST vector a 5.2-kb EcoR1/BamH1 fragment from the EST RE40452, which encodes full-length tutl corresponding to tutl-RD in Flybase (Fig. 1B). For UAS-tutlΔcyto, we used

PCR (oligonucleotide primers: 5′-ACG ACT CAC TAT AGG GCG-3′ and 5′-CGC CTC

TAG ACT ATA CGG CAC AAA C-3′) to amplify a fragment from EST RE40452. The second primer introduced a stop codon and XbaI site (underlined). The fragment was cut with EcoRI and XbaI and introduced into EcoRI/XbaI-restricted pUAST. UAS-tutlΔcyto encodes a truncated Tutl protein comprised of Tutl amino acids 1-879 predicted by Tutl-

RD.

For tutl MARCM, virgin females of the stock elavC155-GAL4,UAS-mCD8::GFP,hs-

FLP;FRT40A,tub-GAL80, were crossed to males that were either elavC155-GAL4,UAS- mCD8::GFP,hs-FLP;FRT40A or elavC155-GAL4,UAS-mCD8::GFP,hs-

FLP;FRT40A,tutl23. For cut MARCM, flies of the stock FRT19A, tub-GAL80, hs-FLP;

GAL4109(2)80, UAS-mCD8::GFP were crossed to flies carrying FRT19A, cutc145/FM7c.

177

Embryos were collected for 2 hours, incubated at 25°C for 2-3 hours, then heat shocked at

38°C for 1 hour and incubated at 25°C until they were analyzed just prior to pupation.

Larvae mutant for tutl were cultivated on agar plates. Mutant animals were selected with the aid of balancer chromosomes CyO,twi-Gal4,UAS-GFP (for tutl, ab, and kn) or

TM3,twi-Gal4,UAS-GFP (for ss).

Imaging and quantification

Larvae were dissected in PBS and GFP-positive da neurons were imaged with confocal microscopy using a Yokogawa spinning disk system (Perkin-Elmer) on an Eclipse

TE2000-U microscope (Nikon). Z-series images were collected using Metamorph software (Molecular Devices) and prepared for publication in Photoshop by converting images to greyscale and adjusting brightness and contrast. Reconstruct software (Fiala,

2005) was used to quantify the numbers of branch termini, branch points, and crossing points, as well as dendritic length (classes I-III) and dendritic field area (polygon method

(Grueber et al., 2002). The dendritic arbors of Class IV neurons were traced and measured with Imaris software (Bitplane). The data was tested for normal distribution using the Shapiro-Wilk test, and statistical analysis was performed using Analyse-It software for Microsoft Excel.

Immunohistochemistry

Anti-Tutl polyconal antiserum was raised in rabbits to a GST-Tutl (aa1-421) fusion protein corresponding to Ig domains 1-3 of the Tutl ectodomain, then affinity-purified and pre-absorbed using standard methods. For anti-Tutl immunofluorescence, embryos or

178

third-instar larvae were dissected in PBS and fixed in 4% paraformaldehyde (Ou et al.,

2008), then anti-Tutl (1:25, 4°C) was detected with Rhodamine Red-X conjugated secondary antibody (1:300). In double labelling of embryos for Tutl and HRP, we then added Cy3-conjugated anti-HRP (1:500, 4°C). Prior to mounting samples from third- instar larvae, muscles overlying the dorsal cluster da neurons were removed by dissection for better visualization.

179

Results

Tutl expression in da neurons

Four da neuron classes (I-IV) of increasing dendritic complexity and size can be readily observed in the larval body wall of Drosophila, sandwiched in two dimensions between muscles and epidermis (Grueber et al., 2002). We focused on the dorsal-most cluster of peripheral sensory neurons, which contains at least one representative from each class.

We examined Tutl expression with immunofluorescence and found that Tutl was expressed in the cell bodies and along the dendrites of class I neurons ddaD and ddaE in third-instar larvae (Fig. 1C',D'). Class I neurons have the least complex arbors of all da neurons. Tutl was also readily observed in the cell body of the class IV neuron ddaC

(Fig.1C',D'), whose arbors are the most complex. Tutl was also expressed in the class II neuron ddaB and the class III neurons ddaA and ddaF (Fig.1C'), in addition to other sensory neurons in the dorsal cluster (Fig. S4). The specificity of the antibody for Tutl was confirmed by the absence of expression in tutl23 mutants (Fig. 1E'), which carry a novel tutl allele (Fig. 1B).

Analysis of tutl function in class I da neurons

To examine the phenotypic consequences of tutl mutations in da neurons, we studied tutl23 and other available P-element insertion (tutl01085) and deficiency (tutlDf) alleles. We began by examining dendrite morphology of the class I da neuron ddaE, using the class I driver GAL4221 and UAS-mCD8::GFP as a reporter to reveal the dendritic tree (Grueber et al., 2003). In control larvae at third-instar, ddaE neurons have a simple, comb-like

180

appearance (Fig. 2A) and 24.1±0.8 (mean ± standard error) branch termini per cell (Fig.

2C). One of the 2 or 3 primary dendrites projects dorsally and gives rise to several lengthy interstitial secondary branches that grow in a posterior direction toward the segment boundary (Fig. 2A). In contrast, the dendritic trees of ddaE neurons in homozygous or hetero-allelic tutl mutants had a number of defects, including severely shortened interstitial branches and irregular patterns of curled or tortuous growth, often lacking directed orientation (Figs. 2B, S1A-C). We scored these defects while blind to genotype, and found them in 20/20 ddaE neurons from tutl01085/tutl23 mutants, but only in

2/20 wild-type controls, indicating high penetrance of the tutl mutant phenotype. In addition, compared to wild-type ddaE neurons, we found significantly more branch termini (Figs. 2B, S1A-C), increasing to 36.7±0.9 in tutl23 homozygotes (Fig. 2C). This was as severe as tutl23/tutlDf hemizygotes (33.5±1.1, Figs. 2C, S1B), supporting our molecular and immunochemical data that tutl23 is a null allele of tutl. Heterozygotes

(tutl23/+) had a degree of branching that was intermediate between controls and homozygotes (Fig. 2C), suggesting that ddaE branching is sensitive to the levels of Tutl.

Overall, the mutant genotypes had increases in ddaE branches that ranged from 126-

152% of wild-type controls (Fig. 2C). Similar observations were made for ddaD (not shown), another class I da neuron.

Single cell analysis of tutl function in da neuron dendrite branching

Though Tutl is expressed in dendrites of class I da neurons, the phenotypes we observed could result from either cell autonomous or non-cell autonomous Tutl activity. To ask whether tutl is required cell autonomously in class I da neurons, we generated single

181

mutant neuron clones using the MARCM system (Lee and Luo, 1999). Control ddaE

MARCM clones exhibited normal morphology and branching (24.0±1.2, Fig.2C,E) but the dendrites of tutl23 ddaE MARCM clones showed increased branch termini to the same level as found in tutl mutant animals (34.7±2.6, Fig. 2C,F). Other features of the tutl homozygous mutant phenotype (i.e. shortened interstitial branches, irregular patterns and directions of growth) were not readily observed in tutl23 ddaE MARCM clones, and thus we are unable to ascribe them to cell autonomous loss of Tutl function. Therefore, we focused on the role of Tutl in the control of dendrite branching, where MARCM analysis pointed to a specific and cell autonomous role for Tutl in preventing excessive dendrite branching.

When normalized for dendritic length, ddaE neurons in tutl01085/tutl23 mutants and tutl23

MARCM clones retained increased numbers of branch termini relative to controls (Figs.

2D, S2A), suggesting that tutl mutations increase dendrite branching complexity independent of ddaE dendrite growth. We analyzed the branching defect of tutl mutants in more detail by counting 1) branch points on primary dendrites that project directly from the cell body, and 2) second or third order branch points situated more distally on the arbor (Fig. 2G). While the number of branch points on primary dendrites was unchanged in tutl mutants and tutl23 MARCM clones, there was a clear increase in the number of second order and third order branch points.

To determine whether Tutl cell-autonomously inhibits branching in other classes of da neurons, we used MARCM to examine class II (ddaB), class III (ddaA, ddaF), and class

182

IV (ddaC) neurons (Grueber et al., 2002). In tutl23 MARCM clones for neurons of class II

(Fig. 3A-D) and class III (Fig. 3E-H), we found no significant changes of branch number nor did we observe any effects on the pattern, growth or targeting of their dendritic trees.

In the class IV neuron ddaC, we observed defects in dendrite self-avoidance (see below), but branch number was unaffected (Fig. S2B, MARCM data). Together, these results indicate that tutl has class-specific effects on dendrite branching in vivo.

Effects of tutl mutation on class IV da neurons

Dendrite self-avoidance involves repulsive interactions between isoneuronal branches following transient contact (Hughes et al., 2007; Matthews et al., 2007; Soba et al., 2007).

To investigate the role of tutl in dendrite self-avoidance, we examined the complex arbor of the class IV da neuron ddaC in tutl mutants and tutl23 MARCM clones. In control third-instar larvae, the high-order branches of ddaC seldom overlap with one another (Fig.

4A',C'). However, in tutl mutants (Fig. 4B') and tutl23 MARCM clones (Fig. 4D'), there were increased numbers of crossing points between isoneuronal dendrites. This was not due to any increases in the total length of dendrites (Fig. 4E), nor was it due to any increases in number of branch termini or changes in the area of the dendritic field (Fig.

S2B,C). In fact, when the number of crossing points was normalized to total dendritic length (Fig. 4E), tutl mutants and tutl23 MARCM clones showed significant increases in isoneuronal self-crossing to 150% and 127%, respectively, of controls (Fig. 4F). The milder effect using the MARCM technique may be due to the persistence in ddaC clones of residual Tutl protein from precursor cells. The phenotypes induced in tutl mutants and tutl MARCM clones indicate that Tutl functions within class IV da neurons to prevent

183

overlap between isoneuronal dendrites.

Class IV neurons also exhibit dendritic "tiling", which is the complete and non-redundant coverage of receptive fields by neurons of a similar functional type (Grueber et al., 2002;

Parrish et al., 2007). Class IV neurons show tiling with other class IV neurons even though they overlap extensively with dendrites of class I-III neurons. Like self-avoidance, tiling is thought to be due to mutual repulsion between dendrites (Parrish et al., 2007), and the processes are related through a common requirement for the nuclear Dbf2-related

(NDR) protein kinase Tricornered (Trc) and its putative adaptor protein Furry (Emoto et al., 2004). To determine whether tutl is required for dendritic tiling between class IV neurons, we examined the borders between ddaC and another neighboring class IV neuron (v'ada) for dendritic overlap (Fig. 5A,A'). We found no evidence that tutl is required for tiling, as the branches between these different class IV neurons approached one another but did not overlap in tutl mutants (Fig. 5B,B').

Effects of tutl overexpression in da neurons

To test whether tutl is sufficient to inhibit dendrite branching, we overexpressed full- length Tutl in da neurons using a UAS-tutl transgene. In class I neurons (ddaE), branching was unaffected (not shown), suggesting that neurons with small simple arbors and substantial levels of endogenous Tutl along their dendrites are unaffected by adding more

Tutl. We then tested the effects of UAS-tutl on the large and highly complex dendritic arbors of the class IV da neuron ddaC (Fig. 6A). In contrast to class I ddaE neurons, overexpression of Tutl in ddaC neurons inhibited dendrite branching in ddaC (Fig. 6B,C).

184

One idea consistent with the branching and self-avoidance defects caused by tutl mutations is that Tutl could promote repulsion between isoneuronal dendrite branches.

Repulsion could conceivably induce branch collapse in class I da neurons, or could steer dendrites away from one another in class IV da neurons. To explore whether Tutl is sufficient to repel dendrites from one another, we exploited the fact that the dendrites of neurons of different classes overlap with one another in the body wall. We used the driver

C161-GAL4 to overexpress UAS-tutl in class I, class II and class III neurons simultaneously, but we found no evidence for Tutl-induced repulsion among dendrites of different classes (Fig. S3).

Studies of genetic interactions between tutl and mutations of trc or Dscam

Mutations in trc cause excessive branching of class I ddaE neuron dendrites (Soba et al.,

2007) and defects of dendrite self-avoidance, particularly at distal branches of the complex arbors of class IV neurons (Emoto et al., 2004). Since mutations of trc and tutl display these similar features, it is possible that they could act in a common molecular pathway to govern the processes of branch inhibition (class I) and self-avoidance (class

IV). We looked for genetic interactions in animals doubly heterozygous for the null mutations trc1 and tutl23, but there were no enhanced defects compared with controls heterozygous for tutl23 alone (Table S1).

In tutl mutants many isoneuronal dendritic branches still avoid one another appropriately

(Fig. 4B,D), suggesting that Tutl is part of a multi-component system ensuring proper distribution of dendrites over receptive territories. One component of this self-avoidance

185

system involves the diverse family of receptors encoded by the Dscam gene. Dscam mutants show defects of self-avoidance in all four da neuron classes (Hughes et al., 2007;

Matthews et al., 2007; Soba et al., 2007). We examined self-avoidance in class IV neurons, where both genes are required, but found no evidence for genetic interactions between tutl23 and Dscam33, a strong mutant allele of Dscam (Hummel et al., 2003) (data not shown).

Investigation of Tutl regulation by transcription factors

Class-specific patterns of da neuron dendrite morphogenesis are regulated by key transcription factors. For example, mutations of the genes encoding the transcription factors Abrupt (Ab) or Spineless (Ss) cause ectopic branching in class I neurons (Kim et al., 2006; Li et al., 2004; Sugimura et al., 2004), resembling the effects we have found for tutl mutants. To test whether Tutl expression in class I neurons is regulated by Ab or Ss, we examined Tutl expression in abk02807 or ssD115.7 mutants with immunochemistry. Both of these mutant alleles are known to cause defects in dendrite morphogenesis, yet we found no obvious changes of Tutl immunoreactivity (Fig. S4A-B′′). Overexpression of

Tutl in class IV neurons inhibits dendrite branching, as do mutations in genes encoding the transcription factors Knot (also known as Collier) (Crozatier and Vincent, 2008;

Hattori et al., 2007; Jinushi-Nakao et al., 2007) and Cut (Grueber et al., 2003). To explore the possibility that Tutl expression in class IV da neurons is normally suppressed to endogenous levels in class IV da neurons by Knot or Cut, we examined Tutl protein expression in ddaC neurons in knKN2 mutants and cutc145 MARCM clones. Changes of

Tutl immunoreactivity were not detected in da neurons in either case (Fig. S4C-D′′).

186

Rescue of tutl mutant phenotypes and assessment of the dispensability of the Tutl cytoplasmic tail.

Tutl and its mammalian orthologs have a conserved ectodomain comprised of 5 Ig-like and 2 Fn-III domains. They also have lengthy but divergent cytoplasmic tails. To begin to investigate the molecular basis for Tutl function in vivo, we wondered whether the cytoplasmic tail of Tutl was dispensable for its function in dendrite morphogenesis.

Testing this in class IV da neuron self-avoidance is complicated by the gain-of-function effect of Tutl in ddaC (i.e. branch inhibition). Instead, we examined class I da neurons where Tutl has no gain-of-function effect. We assessed the requirement for the cytoplasmic tail in a rescue assay for branch inhibition in the class I neuron ddaE. To do this we used GAL4221 to specifically express either full-length Tutl or a truncated form lacking the cytoplasmic tail (TutlΔcyto) in class I neurons of tutl mutants (Figs. 7, S5).

The full-length form of Tutl fully restored the number of branch termini per cell (ddaE) to wild-type levels (Fig. 7B,D), providing further confirmation that the phenotype observed in tutl mutants was specifically due to the loss of Tutl. Importantly, we found that

TutlΔcyto was equally capable of rescuing tutl mutants (Fig. 7C,D), indicating that the cytoplasmic tail of Tutl is indeed dispensable for dendrite branch inhibition in class I da neurons.

Hypomorphic mutants of tutl have behavioral defects (Bodily et al., 2001), but strong alleles are lethal. Since Tutl is expressed broadly in the nervous system (Bodily et al.,

2001), we used the neural-specific driver elavC155-Gal4 to test whether full-length Tutl could rescue viability in tutl23/tutl1085, and whether the cytoplasmic tail was also

187

dispensable for viability. Rescue with full-length Tutl was nearly complete, since 53 of an expected 57 (93%) mutants were rendered viable. The rescued larvae showed normal behavioral responses to tactile stimuli, and could right themselves when overturned (data not shown). In contrast, the rescue with TutlΔcyto was only partial, with 30 viable adults of an expected 152 (20%). Similar results for both full-length Tutl and TutlΔcyto were observed for other tutl mutant genotypes, including tutl23 homozygotes (data not shown).

Together, the data indicates that while the cytoplasmic tail is dispensable for the role of

Tutl in limiting dendrite branching in da neurons, it is required for Tutl function in other cell types within the nervous system.

188

Figure 1 189

Figure 1. Tutl protein structure, tutl alleles, and Tutl expression in da neurons.

A, Schematic diagram of Tutl protein. B, Structure of the tutl locus and position of tutl mutations. The P-element causing tutl01085 is inserted in the fifth exon of tutl. The tutl23 mutation was generated by excising the genomic DNA between the PBac elements f03096 and f02770. C, UAS-mCD8::GFP driven by GAL4109(2)80 was used to visualize the cell bodies and dendrites of dorsal da neurons of wild-type third-instar larvae. C’, Tutl immunoreactivity was observed in da neuron cell bodies (labelled) and dendrites

(arrowheads). The majority of the dendrites labelled here belong to the class I da neuron ddaD. D, class I da neurons (ddaD and ddaE) visualized by GAL4221- driven expression of mCD8::GFP; there was also weak ectopic expression of GFP in class IV ddaC. D’,

Tutl was expressed in the cell bodies of GFP-positive da neurons. E-E’, In tutl23 homozygous mutants, Tutl staining was absent from GFP-positive da neurons. C-E', scale bar is 50 μm, anterior is left, dorsal is up.

190

Figure 2

191

Figure 2. tutl is required to restrain dendrite branching in class I da neurons.

A-B, Class I ddaE neuron visualized by GAL4221 driving UAS-mCD8::GFP. A, ddaE neuron in wild type. B, ddaE neuron in tutl01085/23 mutant with dendrite defects including shortened interstitial branches (arrows) and curled growth lacking directed orientation

(arrowheads). C, Quantification of ddaE branch termini in wild type and tutl mutants.

Bars show mean ±standard error (s.e.). Asterisk indicates significant difference from wild-type (wt) control, ANOVA (p<1 X 10-4). In similar ANOVA tests, the tutl23/+ heterozygotes were also significantly different from tutl23/Df and tutl23/23. D, normalization of branch termini to dendritic length (mean±s.e.), asterisks: tutl mutants (t-test, p<4 X 10-

5), tutl MARCM (t-test, p<0.008) E-F, MARCM clones of ddaE neurons. E, wild-type ddaE clone. F, tutl23 clone showed increased numbers of branch termini and branch points.

G, Quantification of branch points. Asterisks: termini per neuron (t-test, p<4 X 10-4), termini/dendritic length (t-test, p<1 X 10-4). A,B,E,F, scale bar is 50 μm, anterior is left, dorsal is up. N: number of neurons quantified for each genotype. Genotypes: A,

GAL4221,UAS-mCD8::GFP/+. B, tutl01085/tutl23;GAL4221,UAS-mCD8::GFP/+. E, elavC155-

GAL4,UAS-mCD8::GFP,hs-FLP;FRT40A. F, elavC155-GAL4,UAS-mCD8::GFP,hs-

FLP;FRT40A,tutl23.

192

Figure 3 193

Figure 3. Class II and class III DA neurons are unaffected in tutl23 MARCM clones.

A-B, MARCM clones of class II ddaB neurons. A, wild-type ddaB clone. B, tutl23 mutant ddaB clone showing normal dendritic pattern. C, quantification of ddaB termini

(mean±s.e.), showing no significant (ns) difference between wild-type and tutl23

MARCM clones (t-test, p>0.05). D, quantification of ddaB termini normalized to dendritic field area (mean±s.e.), again showing no significant difference. E-F, MARCM clones of class III ddaA neurons. E, wild-type clone. F, tutl23 MARCM ddaA clone. G, quantification of ddaA termini (mean±s.e.) H, quantification of ddaA termini normalized to the total length of the main dendritic branches of ddaA (mean±s.e.), not including spine-like protrusions.

194

Figure 4

195

Figure 4. tutl is required for dendrite self-avoidance in class IV da neurons.

A-B’, Class IV ddaC neurons visualized by ppk1.9-GAL4 driving UAS-mCD8::GFP. A, ddaC neuron in wild type. Dotted outline marks area shown in a'. A′, branches of ddaC neurons normally show self-avoidance, with only occasional crossing points (arrows). B-

B′, ddaC neuron in tutl01085/23 mutant showed numerous dendrite crossing points (arrows).

The dendritic field is smaller than wild type because tutl mutant animals are shorter than wild type. C-D′ MARCM clones of ddaC neurons. C-C′, wild-type ddaC clone showed self-avoidance and occasional crossing points (arrows). D-D′, tutl23 clone showed increased numbers of crossing points (arrows). E, Quantification of dendrite length in ddaC neurons (mean±s.e.), asterisks: tutl mutants (t-test, p<3 X 10-7), tutl MARCM (t-test, p=0.144, not significant (ns). N: number of neurons quantified for each genotype. F,

Quantification of dendrite crossing points normalized to dendritic length in ddaC neurons

(mean±s.e.), tutl mutants (t-test, p<2 X 10-5), tutl MARCM (t-test, p<0.03). A,B,C,D, scale bar is 100 μm, anterior is left, dorsal is up. Genotypes: A-A′, UAS- mCD8::GFP/+;ppk1.9-GAL4/+. B-B′, UAS-mCD8::GFP/+;tutl01085/tutl23;ppk1.9-

GAL4/+. C-C′, elavC155-GAL4,UAS-mCD8::GFP,hs-FLP; FRT40A. D-D′, elavC155-GAL4,

UAS-mCD8::GFP, hs-FLP; FRT40A, tutl23.

196

Figure 5

197

Figure 5. tutl mutants exhibit normal dendritic tiling among class IV da neurons.

A-B’, Tiling between class IV da neurons ddaC and v'ada labelled with ppk1.9-GAL4,

UAS-mCD8::GFP. In A′ and B′, ddaC neurons were traced in red and adjacent v‟ada neurons were traced in blue. A-A′, class IV neurons in controls established normal territories. B-B′, class IV neurons in tutl mutants showed intact tiling, but did show defects of self-avoidance. A,B, scale bar is 50 μm, anterior is left, dorsal is up.

Genotypes: A-A′, UAS-mCD8::GFP/+;ppk1.9-GAL4/+. B-B′, UAS- mCD8::GFP/+;tutl01085/ tutl23; ppk1.9-GAL4/+.

198

Figure 6

199

Figure 6. Overexpression of Tutl inhibits dendrite branching in class IV da neurons.

A, Class IV neuron ddaC in wild type. B, overexpression of Tutl in ddaC reduces branching. C, Quantification of ddaC branch termini per cell, and normalized to dendritic length. Asterisks: termini per neuron (t-test, p<2 X 10-4), termini/dendritic length (t-test, p<1 X 10-5). A,B, scale bar is 100 μm, anterior is left, dorsal is up. N: number of neurons quantified for each genotype. Genotypes: A, UAS-mCD8::GFP/+;ppk1.9-GAL4/+. B,

UAS-mCD8::GFP/+; ppk1.9-GAL4/UAS-tutl.

200

Figure 7

201

Figure 7. The cytoplasmic tail of Tutl is dispensable for function in class I da neurons.

A, Class I ddaE neuron in tutl01085/23 mutant with branching and morphology defects. B, ddaE neuron in tutl01085/23 mutant expressing full-length Tutl with GAL4221. C, ddaE neuron in tutl01085/23 mutant expressing TutlΔcyto. Anti-Tutl immunochemistry demonstrating expression of Tutl and TutlΔcyto can be found in Fig. S4. D,

Quantification of ddaE branch termini in wild type, tutl mutants, and rescued animals

(mean ± s.e). Asterisk indicates significant difference from wild-type (wt) control,

ANOVA (p<1 X 10-4). N: number of neurons quantified for each genotype. Full-length

Tutl and TutlΔcyto each rescue the number of branch termini in tutl mutants to wild-type levels. A,B,C, scale bar is 50 μm, anterior is left, dorsal is up. Genotypes: A, tutl01085/tutl23;GAL4221,UAS-mCD8::GFP/+. B, tutl01085/tutl23, UAS-tutl;GAL4221,UAS- mCD8::GFP/UAS-tutl. C, tutl01085/tutl23;GAL4221,UAS-mCD8::GFP/UAS-tutlΔcyto,UAS- tutlΔcyto.

202

Figure S1

203

Figure S1. Additional examples of dendrite branching and self-avoidance defects in tutl mutants.

A-C, Class I ddaE neurons visualized by GAL4221 driving UAS-mCD8::GFP.

Representative examples from three different heteroallelic combinations are presented.

Arrows in A indicate shortened interstitial branches, and arrowheads indicate irregular patterns of growth. D, Class IV ddaC neurons visualized by ppk1.9-GAL4 driving UAS- mCD8::GFP. The dotted outline in D marks the area shown in D'. D′, Red arrows indicate 11 crossing points. Genotypes: A, tutl23/tutl23;GAL4221,UAS-mCD8::GFP/+. B, tutl23/Df(2L)ed-dp;GAL4221,UAS-mCD8::GFP/+. C, tutl01085/Df(2L)ed-dp;GAL4221,UAS- mCD8::GFP/+. D,D′, UAS-mCD8::GFP/+;tutl23/tutl23;ppk1.9-GAL4/+.

204

Figure S2 205

Figure S2. Quantification of the effects of tutl mutations on additional DA neuron dendrite parameters.

A, Total dendrite length for ddaE (mean±s.e.), asterisks: tutl mutants (t-test, p < 2 X 10-5), tutl MARCM (t-test, p=0.04). B, Number of dendrite termini for ddaC (mean±s.e.), asterisks: tutl mutants (t-test, p < 2 X 10-13), tutl MARCM (t-test, p=0.138, not significant

(ns). C, Total dendritic field area for ddaC (mean±s.e.), asterisks: tutl mutants (t-test, p <

2 X 10-8), tutl MARCM (t-test, p=0.753, not significant (ns).

206

Figure S3

207

Figure S3. Overexpression of Tutl is not sufficient to induce dendrite branch repulsion.

A-B′, Overlap among dendrites of class I da neuron ddaE and class III neuron ddaF labelled with C161-GAL4 and UAS-mCD8::GFP. C161-GAL4 drives expression in classes I-III, but not class IV da neurons. In A′ and B′, dendrites of ddaE neurons were traced in red and those of ddaF neurons were traced in blue. A-A′, control showed normal overlap among dendrites of these neurons. B-B′, overexpression of Tutl was not sufficient to induce repulsion among dendrites of different classes. A,B, scale bar is 50 μm, anterior is left, dorsal is up. Genotypes: A-A′, UAS-mCD8::GFP/+;;C161-GAL4/+. B-B′, UAS- mCD8::GFP/+;UAS-tutl/+; C161-GAL4/ UAS-tutl.

208

Figure S4

209

Figure S4. Tutl expression in mutants of ab, ss, kn, and cut.

A-C′′, Tutl protein expression in dorsal cluster neurons of abk02807 mutants (A-A′), ssD115.7 mutants (B-B′) or knKN2 mutants (C-C′) at stage 16-17 of embryogenesis. In each case, dorsal cluster neurons are labelled with anti-HRP immunochemistry (A,B,C). As in wild type, all of the dorsal cluster da neurons label with Tutl, as do other sensory neurons including the bipolar dendrite neuron, external sensory neurons, tracheal dendrite neurons and chordotonal organs. In abk02807, ssD115.7, or knKN2 mutants, there were no obvious changes of Tutl immunoreactivity relative to wild type (A′B′C′, compare with Fig. 1c′,d′).

A,B, arrows indicate a cell likely to be ddaE based on cell body position. As in other dorsal cluster neurons, Tutl expression in this cell is unaffected even though ab and ss mutants have phenotypes similar to tutl in ddaE neurons. In kn mutants, no cells in the da cluster show obvious upregulation of Tutl that could have explained why kn mutants phenocopy Tutl overexpression in class IV ddaC neurons. A′′B′′C′′, overlays with HRP immunreactivity in green and Tutl in magenta. D-D′′, since homozygous cutc145 mutants do not survive to stage 16-17 of embryogenesis, we made cutc145 MARCM clones and examined Tutl expression in third-instar larvae. D, single-cell, GFP-labelled cutc145

MARCM clone likely to be ddaC based on cell body position, the presence of many branches, and a strong mutant phenotype in which the arbor is reduced and the branches are thickened and form aberrant clusters. D′, anti-Tutl immunochemistry showing Tutl expression in the cell body (arrow) and in the shortened and clustered dendrites of the cutc145 MARCM clone. D′′, Overlay with GFP immunreactivity in green and Tutl in magenta. Scale bars = 10 μm.

210

Figure S5

211

Figure S5. Expression of Tutl and TutlΔcyto in class I da neurons of tutl mutants.

A, In a tutl01085/23 mutant carrying UAS-tutl, class I da neurons (ddaD and ddaE; cell bodies labelled) are visualized by GAL4221- driven expression of mCD8::GFP; there was also weak ectopic expression of GFP in class IV ddaC. A′, anti-Tutl immunochemistry: elevated levels of Tutl are observed in class I da neuron cell bodies and dendrites due to

UAS-tutl expression. B, In a tutl mutant, GFP labelling of class I da neurons expressing

UAS-tutlΔcyto. B′ TutlΔcyto is detectable with anti-Tutl because the antibody is directed to Ig domains 1-3 of the Tutl ectodomain. Like full-length Tutl, TutlΔcyto is also expressed in both the cell bodies and along dendrites of GFP-positive da neurons. A-B', scale bar is 50 μm, anterior is left, dorsal is up. Genotypes : A,A′, tutl01085/tutl23, UAS- tutl;GAL4221,UAS-mCD8::GFP/UAS-tutl. B,B′, tutl01085/tutl23;GAL4221,UAS- mCD8::GFP/UAS-tutlΔcyto,UAS-tutlΔcyto.

212

Genotype N Mean SE

Class I (ddaE)

Termini per neuron*

tutl23/+ 21 29.0 0.9

trc1/+ 32 26.8 0.8

tutl23/+; trc1/+ 28 27.9 0.7

Class IV (ddaC)

Crossing points per 1000

µm of dendritic length**

tutl23/+ 10 7.5 0.4

trc1/+ 11 5.8 0.4

tutl23/+; trc1/+ 14 8.1 0.5

Table S1. Genetic Interaction Experiments for tutl and trc.

Double heterozygotes for trc and tutl have no enhanced branching defects (class I) or self- avoidance defects (class IV) compared with tutl heterozygotes alone. N= number of neurons examined, SE= standard error.

* There is no significant difference between any two genotypes (ANOVA, Tukey, p=0.18).

** tutl23/+; trc1/+ double heterozygotes are not significantly different from tutl23/+, but trc1/+ vs. tutl23/+ (ANOVA, Tukey, p=0.0022). 213

Discussion

Dendrite branching and self-avoidance are two important cellular mechanisms that shape the receptive fields of neurons during development. Here we have investigated the role of

Tutl in these processes using the da sensory neurons of Drosophila, an excellent system to study dendrite arborisation at a single cell level in vivo. Tutl is a member of the

Tutl/Dasm-1/Igsf9 family of evolutionarily conserved transmembrane proteins. We have found that Tutl inhibits excessive branch formation in neurons with simple dendrites

(class I), and prevents crossing of isoneuronal dendrite branches in neurons with complex arbors (class IV), demonstrating that Tutl influences the architecture of dendrites and their coverage of receptive territories. In contrast to our results for class I and class IV neurons, our MARCM studies found no evidence of a cell-autonomous role for Tutl in class II or III neurons, despite detectable Tutl expression in their cell bodies. The reasons for a lack of apparent effects on class II or III da neuron dendrites in tutl MARCM clones remain unclear. Sufficient Tutl protein, inherited from precursors, could have remained in

MARCM clones to promote normal outgrowth. Alternatively, there may be no role for

Tutl in these cells. Nevertheless, it is clear from our results for class I and class IV da neurons that Tutl is required for the arborisation of dendritic trees with dramatically different complexity.

A role for Tutl in dendrite branching

Tutl cell-autonomously inhibits dendrite branching in vivo, providing a means by which da neurons with the simplest architecture suppress the formation or stabilization of

214

supernumerary dendrite branches during development. We observed a clear increase in the number of second order and third order branch points on tutl mutant ddaE neurons.

This finding suggests that tutl regulates branching at only certain locations along the growing arbour, perhaps by inhibiting branch additions or promoting branch retractions.

The tutl phenotype is distinct from mutants of neuroglian (nrg), which also encodes a cell surface IgSF protein with effects on dendrite branching. Loss of Nrg reduces the number of branches on the dendritic arbors of class I da neurons, and increases branching along their axons, suggesting a role for Nrg in correctly distributing neurites but not as a branching inhibitor (Yamamoto et al., 2006). The tutl mutant phenotype is also distinct from the dendrite overgrowth observed in mutants of the IgSF receptor Robo (Dimitrova et al., 2008), or the cadherin Flamingo (Gao et al., 2000; Kimura et al., 2006; Sweeney et al., 2002). In vertebrate systems, no recognition molecules have yet been shown to inhibit dendrite branching in vivo. However, it is noteworthy that inhibition of axon branching has been demonstrated in the chick visual system, where inappropriate arborisation of retinal ganglion cell (RGC) axon terminals is thought to be inhibited by

EphA (Yates et al., 2001) and Ryk (Schmitt et al., 2006) receptors. In zebrafish, RGC axons are inhibited from branching by Robo2 (Campbell et al., 2007), an IgSF protein with which Tutl shares homology.

A role for Tutl in dendrite self-avoidance

After Dscam, Tutl is the only cell surface protein shown to be required for dendrite self- avoidance in either invertebrates or vertebrates. As in Dscam mutants, the dendrites of

215

tutl mutant neurons cross one another with increased frequency, leading to uneven coverage of the receptive field. Unlike Dscam, which promotes self-avoidance in all four da neuron classes (Hughes et al., 2007; Matthews et al., 2007; Soba et al., 2007), Tutl does so only in the highly complex arbors of class IV neurons. We observed no genetic interactions between Dscam and tutl mutations, and have yet to find any evidence that

Dscam and Tutl could act in a common molecular pathway to control dendrite self- avoidance. Future studies could reveal whether and how these seemingly distinct pathways converge, but based on our findings we speculate that the molecular mechanisms ensuring dendrite self-avoidance will prove to be more complex than is appreciated currently.

Neither Tutl nor Dscam affect dendritic tiling among neurons of a similar functional type, illustrating that self-avoidance and tiling are likely mediated by distinct recognition molecules on the surfaces of dendrites.

How does Tutl regulate dendrite morphogenesis?

The full-length form of Tutl is a transmembrane protein with a 5 Ig/2 Fn-III ectodomain and cytoplasmic tail, suggesting it could act as a signalling receptor. Alternative splicing also gives rise to a membrane tethered form that lacks the cytoplasmic tail (Bodily et al.,

2001). This suggests that Tutl could also function as a membrane-bound ligand for an unknown receptor or, alternatively, a co-receptor in a multiprotein receptor complex.

These possibilities are not mutually exclusive, since Tutl could conceivably act as a ligand or co-receptor in one cellular context, and a signaling receptor in another. We

216

found that the cytoplasmic tail was completely dispensable for the inhibition of dendrite branching in class I da neurons. This is consistent with a model in which Tutl acts as a ligand or co-receptor in dendrites. In contrast, we found that the cytoplasmic tail was required to fully rescue viability in tutl mutants, suggesting that Tutl acts as a signaling receptor in this context.

It is currently unclear how Tutl controls dendrite branching and self-avoidance because our studies have not revealed a connection between Tutl and known regulators of dendrite morphogenesis such as trc. We sought evidence for genetic interactions between trc and tutl and found none. These results alone cannot exclude the possibility that Trc and Tutl act in a common pathway to govern dendrite branching or self-avoidance, but it is noteworthy that the phenotypes of trc and tutl mutants also show some differences that could suggest they work through independent molecular pathways. Unlike tutl, trc is required for dendritic tiling among different class IV neurons, and tutl mutants do not display the excessive terminal branching in class IV neurons that is characteristic of trc mutants (Emoto et al., 2004).

The transcription factors Abrupt, Spineless, Knot/Collier, and Cut each regulate patterns of dendrite branching in keeping with tutl mutations or Tutl overexpression (Crozatier and Vincent, 2008; Grueber et al., 2003; Hattori et al., 2007; Jinushi-Nakao et al., 2007;

Kim et al., 2006; Li et al., 2004; Sugimura et al., 2004). However, we found it is likely that Tutl expression is influenced by a regulatory program distinct from those involving

Abrupt, Spineless, Knot/Collier, or Cut.

217

Tutl remains somewhat enigmatic because we have yet to find evidence for a genetic or regulatory connection between tutl and genes with similar mutant phenotypes.

Nevertheless, our discovery that Tutl regulates dendrite morphogenesis and the coverage of receptive territories underscores the fact that the molecular mechanisms that underlie dendrite morphogenesis remain incompletely understood. We can only speculate as to why tutl mutants have class-specific effects on dendrite morphogenesis, despite Tutl expression in all da neuron classes. Perhaps an unidentified Tutl-interacting protein, such as a receptor required for Tutl function, may be differentially expressed among da neurons and thus account for the specificity of the phenotype. Other explanations may also exist. For example, it is possible that our MARCM experiments failed to show cell- autonomous defects in certain da neuron classes (classes II and III) because the requirement for Tutl in these cells was met by perdurance of sufficient Tutl protein inherited from the precursor cells of MARCM clones.

Does Tutl promote dendrite branch repulsion?

It is intriguing that the two processes of branching and self-avoidance are related by a common requirement for tutl. Both phenotypes are consistent with the idea that Tutl promotes repulsion, perhaps between isoneuronal dendrite branches, or between dendrites and the substrata along which they grow. However, there is no direct evidence at this time for a repulsive role for Tutl. Simultaneous overexpression of Tutl in different da neuron classes was insufficient to induce branch repulsion among their dendrites. Together with our rescue experiments showing the dispensability of the cytoplasmic tail for dendrite branching, these data suggest that Tutl could function as a ligand or co-receptor in

218

complexes with one or more unidentified proteins at the cell surface. Such proteins may not be expressed in all da neurons, which could explain why tutl mutations do not affect da neuron classes II and III, and why Tutl overexpression cannot induce repulsion when overexpressed in overlapping neurons of classes I-III. The tutl mutant phenotypes remain the strongest evidence of a repulsive role for Tutl, and it is likely that direct evidence for repulsion must await the identification of the relevant Tutl-interacting proteins.

If it is true that Tutl mediates repulsion, we speculate that the nature or degree of that repulsion could be influenced by the size of the dendritic arbor, leading to class-specific effects. Class I dendrites remain relatively small with Tutl protein distributed along the entire arbor, where Tutl-mediated repulsion could promote the collapse of transient interstitial branches that are known to extend during development (Gao et al., 1999)

(branch inhibition). In large class IV arbors where Tutl is distributed more sparingly,

Tutl-mediated repulsion could be one part of a multi-component system to redirect isoneuronal branches away from one another (self-avoidance) and thereby ensure proper distribution of dendrites over receptive territories (Hughes et al., 2007; Matthews et al.,

2007; Soba et al., 2007). In this way, neurons of different classes could employ a common repulsive mechanism involving Tutl to sculpt dramatically different patterns of arborisation complexity.

Is there an evolutionarily conserved role for Tutl-related proteins in dendrite morphogenesis in mammals?

219

Our findings that Tutl inhibits dendrite branching in Drosophila contrast with initial observations in cultured rodent neurons, in which RNAi knockdown experiments suggested that the Tutl ortholog Dasm-1 was required to promote dendritic outgrowth

(Shi et al., 2004). However, it was recently argued that the RNAi findings were due to off-target effects (Mishra et al., 2008). The role of Dasm-1 in mammalian dendrite morphogenesis is currently unclear since Dasm-1 knockout mice have no observable dendritic defects (Mishra et al., 2008). However, the possibility has been raised that

Dasm-1 function in dendrites is redundant with IgSF9b, a closely related protein that is coexpressed in the developing hippocampus and whose expression is unaltered in the brains of Dasm-1 knockout mice (Mishra et al., 2008). Loss-of function studies for both

Dasm-1 and IgSF9b should reveal whether Tutl-like proteins in mammals share with Tutl an evolutionarily conserved role in dendrite morphogenesis.

220

Acknowledgments

We thank Fen-Biao Gao, Wayne Johnson, Liqun Luo, Yuh Nung Jan, Bing Ye, Susan

Younger, Dietmar Schmucker, Adrian Moore, the Bloomington and Harvard stock centers, and the Drosophila Genomic Resource Centre for fly stocks and reagents. We also thank Michael Haber for help in quantifying dendrite parameters, and Keith Murai,

Matthias Landgraf, Catriona McDonald and members of the Rao and van Meyel laboratories for advice. The work was supported by funds from the Canadian Institutes of

Health Research (CIHR Team Grant to Y.R and D.J.vM, Operating Grant to Y.R,

International Opportunities Grant to D.J.vM), the Research Institute of the McGill

University Health Centre (D.J.vM), and the Canadian Foundation for Innovation. D.J.vM is a CIHR New Investigator.

221

References

1. Bodily, K. D., et al., (2001). A novel member of the Ig superfamily, turtle, is a CNS-

specific protein required for coordinated motor control. J Neurosci 21, 3113-25.

2. Campbell, D. S., et al., (2007). Slit1a inhibits retinal ganglion cell arborization and

synaptogenesis via Robo2-dependent and -independent pathways. Neuron 55, 231-45.

3. Crozatier, M. and Vincent, A. (2008). Control of multidendritic neuron

differentiation in Drosophila: the role of Collier. Dev Biol 315, 232-42.

4. Dimitrova, S., et al., (2008). Slit and Robo regulate dendrite branching and

elongation of space-filling neurons in Drosophila. Dev Biol 324, 18-30.

5. Doudney, K., et al., (2002). Cloning and characterization of Igsf9 in mouse and

human: a new member of the immunoglobulin superfamily expressed in the

developing nervous system. Genomics 79, 663-70.

6. Emoto, K., et al., (2004). Control of dendritic branching and tiling by the

Tricornered-kinase/Furry signaling pathway in Drosophila sensory neurons. Cell 119,

245-56.

7. Fiala, J. C. (2005). Reconstruct: a free editor for serial section microscopy. J Microsc

218, 52-61.

8. Fuerst, P. G., et al., (2008). Neurite arborization and mosaic spacing in the mouse

retina require DSCAM. Nature 451, 470-4.

9. Furrer, M. P.,et al., (2003). Robo and Frazzled/DCC mediate dendritic guidance at

the CNS midline. Nat Neurosci 6, 223-30.

222

10. Furrer, M. P., et al., (2007). Slit and Robo control the development of dendrites in

Drosophila CNS. Development 134, 3795-804.

11. Gao, F. B., et al., (1999). Genes regulating dendritic outgrowth, branching, and

routing in Drosophila. Genes Dev 13, 2549-61.

12. Gao, F. B., et al., (2000). Control of dendritic field formation in Drosophila: the

roles of flamingo and competition between homologous neurons. Neuron 28, 91-101.

13. Godenschwege, T. A., et al., (2002). Ectopic expression in the giant fiber system of

Drosophila reveals distinct roles for roundabout (Robo), Robo2, and Robo3 in

dendritic guidance and synaptic connectivity. J Neurosci 22, 3117-29.

14. Grueber, W. B., et al., (2002). Tiling of the Drosophila epidermis by multidendritic

sensory neurons. Development 129, 2867-78.

15. Grueber, W. B., et al., (2003). Different levels of the homeodomain protein cut

regulate distinct dendrite branching patterns of Drosophila multidendritic neurons.

Cell 112, 805-18.

16. Hattori, Y., et al., (2007). Selective expression of Knot/Collier, a transcriptional

regulator of the EBF/Olf-1 family, endows the Drosophila sensory system with

neuronal class-specific elaborated dendritic patterns. Genes Cells 12, 1011-22.

17. Horch, H. W. and Katz, L. C. (2002). BDNF release from single cells elicits local

dendritic growth in nearby neurons. Nat Neurosci 5, 1177-84.

18. Hughes, M. E., et al., (2007). Homophilic Dscam interactions control complex

dendrite morphogenesis. Neuron 54, 417-27.

19. Hummel, T., et al., (2003). Axonal targeting of olfactory receptor neurons in

Drosophila is controlled by Dscam. Neuron 37, 221-31.

223

20. Jinushi-Nakao, S., et al., (2007). Knot/Collier and cut control different aspects of

dendrite cytoskeleton and synergize to define final arbor shape. Neuron 56, 963-78.

21. Kim, M. D., et al., (2006). The bHLH-PAS protein Spineless is necessary for the

diversification of dendrite morphology of Drosophila dendritic arborization neurons.

Genes Dev 20, 2806-19.

22. Kimura, H., et al., (2006). Potential dual molecular interaction of the Drosophila 7-

pass transmembrane cadherin Flamingo in dendritic morphogenesis. J Cell Sci 119,

1118-29.

23. Komiyama, T., et al., (2007). Graded expression of semaphorin-1a cell-

autonomously directs dendritic targeting of olfactory projection neurons. Cell 128,

399-410.

24. Lee, T. and Luo, L. (1999). Mosaic analysis with a repressible cell marker for studies

of gene function in neuronal morphogenesis. Neuron 22, 451-61.

25. Li, W., et al., (2004). BTB/POZ-zinc finger protein abrupt suppresses dendritic

branching in a neuronal subtype-specific and dosage-dependent manner. Neuron 43,

823-34.

26. Matthews, B. J., et al., (2007). Dendrite self-avoidance is controlled by Dscam. Cell

129, 593-604.

27. Mishra, A., et al., (2008). Dendrite arborization and synapse maturation (Dasm)-1 is

dispensable for dendrite arborization. Mol Cell Biol 28, 2782-91.

28. Ou, Y., et al., (2008). Identification of genes influencing dendrite morphogenesis in

developing peripheral sensory and central motor neurons. Neural Develop 3, 16.

224

29. Parks, A. L., et al., (2004). Systematic generation of high-resolution deletion

coverage of the Drosophila melanogaster genome. Nat Genet 36, 288-92.

30. Parrish, J. Z., et al., (2007). Mechanisms that regulate establishment, maintenance,

and remodeling of dendritic fields. Annu Rev Neurosci 30, 399-423.

31. Polleux, F., et al., (2000). Semaphorin 3A is a chemoattractant for cortical apical

dendrites. Nature 404, 567-73.

32. Schmitt, A. M., et al., (2006). Wnt-Ryk signalling mediates medial-lateral

retinotectal topographic mapping. Nature 439, 31-7.

33. Shi, S. H., et al., (2004). Control of dendrite arborization by an Ig family member,

dendrite arborization and synapse maturation 1 (Dasm1). Proc Natl Acad Sci U S A

101, 13341-5.

34. Shima, Y., et al., (2007). Opposing roles in neurite growth control by two seven-pass

transmembrane cadherins. Nat Neurosci 10, 963-969.

35. Soba, P., et al., (2007). Drosophila sensory neurons require Dscam for dendritic self-

avoidance and proper dendritic field organization. Neuron 54, 403-16.

36. Sugimura, K., et al., (2004). Development of morphological diversity of dendrites in

Drosophila by the BTB-zinc finger protein abrupt. Neuron 43, 809-22.

37. Sweeney, N. T., et al., (2002). Genetic manipulation of single neurons in vivo reveals

specific roles of flamingo in neuronal morphogenesis. Dev Biol 247, 76-88.

38. Whitford, K. L., et al., (2002). Regulation of cortical dendrite development by Slit-

Robo interactions. Neuron 33, 47-61.

39. Yamamoto, M., et al., (2006). Control of axonal sprouting and dendrite branching by

the Nrg-Ank complex at the neuron-glia interface. Curr Biol 16, 1678-83.

225

40. Yates, P. A., et al., (2001). Topographic-specific axon branching controlled by

ephrin-As is the critical event in retinotectal map development. J Neurosci 21, 8548-

63.

226

Chapter 4

Characterizing gene Dnrk in Drosophila

This work is in preparation for submission.

227

Previous analysis has proposed that Drosophila neurospecific receptor kinase (Dnrk) plays a role in fly nervous system development based on its specific expression pattern

(Oishi et al., 1997). Homologs of Dnrk in C. elegans and zebrafish have been implicated in regulating axon guidance (Forrester et al., 1999; Koga et al., 1999; Francis et al., 2005;

Zhang et al., 2004). I generated Dnrk null mutation. Phenotypic analyses indicate that

Dnrk is involved in regulating R-cell axon guidance in the fly visual system. Further studies suggest that as two members of the fly Ror family, Dnrk and Dror have redundant function in controlling axon guidance. The overexpression of Dnrk is able to induce a gain-of-function phenotype in the wing. In a genetic screen for modifiers of this GOF phenotype, I identified DWnt5 as a potential interacting gene of Dnrk.

228

Characterizing gene Dnrk in Drosophila

Hong Long and Yong Rao

Summary

Drosophila neurospecific receptor kinase (Dnrk), a gene encoding a putative receptor tyrosine kinase, is highly related to the ror receptor tyrosine kinase family. Structural analysis of this gene reveals that it has a Frizzled module/CRD, a kringle domain, and a cytoplasmic tyrosine kinase domain. Previous studies of cam-1, the C. elegans ror, have revealed that cam-1 is involved in several biological processes during the embryonic development, such as cell migration, polarity formation, and axon outgrowth (Forrester et al., 1999 and 2004; Koga et al., 1999; Zinovyeva et al., 2008; Green et al., 2007 and 2008;

Francis et al., 2005; Hayashi et al., 2009). The work on the mammalian rors, mror1and mror2, has implicated them into the dendritic branching process of mouse hippocampal neurons (Paganoni and Ferreira, 2003 and 2005). These studies suggest that Dnrk, as the homolog, might have similar functions in axon guidance and dendrite patterning in fly. It is also interesting that Dnrk shares high structural similarity with Musk. Musk plays a critical role in synaptic complex formation within the neuromuscular junction of mammals. Zebrafish Musk, unplugged, also functions in the neuromuscular system by controlling axon guidance (Zhang et al., 2004). These studies, together with the high expression level of Dnrk mRNA during the embryonic stage in the fly nervous system, suggest that Dnrk may play an important role in the fly nervous system.

229

To investigate the function of Dnrk in the fly nervous system, I generated the Dnrk null mutant. This mutant is homozygous viable, which suggests that Dnrk is not essential for viability. I examined the visual system, the ventral nerve cord, the neuromuscular system, and the PNS da neurons in the Dnrk mutant. A low penetrant axon-guidance phenotype was observed in the visual system, but not in other systems. Considering the potential redundancy between the two members of the fly ror family, Dnrk and Dror, I decided to examine the potential genetic interaction between those two genes. Knocking down Dror by RNAi could dramatically enhance axon guidance defects in the Dnrk mutant. This suggests that Dror functions redundantly in R-cell axon guidance with Dnrk.

I also generated UAS-Dnrk transgenic flies. Interestingly, when Dnrk was overexpressed in the wing, it induced the loss of cross veins. This phenotype could be suppressed by overexpression of DWnt5, which is a member of the fly Wnt family. However, the Dnrk mutant did not show any defects in the wing, which might be due to the functional redundancy between Dnrk and some other genes in the wing. The over-expression of

DWnt5 may sequester Dnrk from its endogenous binding partner and thus suppress the

Dnrk overexpression phenotype. With these promising preliminary data, it will be interesting to investigate further to determine the exact function of Dnrk in the fly developmental process.

230

Introduction

PTK family

Tyrosine phosphorylation is a very important modification that occurs in multicellular organisms. It plays key roles in cell-cell communication and intracellular signal transduction during the embryonic development and adult maintenance. Enzymes catalyzing tyrosine phosphorylation reactions are the protein tyrosine kinase family

(PTKs) (Hubbard and Till, 2000). PTK catalyzes the transferring of γ phosphate from

ATP to the tyrosine residue on the protein substrates. Tyrosine phosphorylation can modulate PTK enzyme activity and also create binding motifs for downstream signalling proteins. PTKs are divided into two groups: transmembrane receptor PTK (RTK) family and the nonreceptor PTK (NRTK) family.

RTK and NRTK family

RTKs are major players in regulating cell-cell communication during development.

Members of this family are glycoproteins located on the cell surface. They can bind to their ligands and then transduce the signal to the cytoplasm by autophosphorylation at the tyrosine residues on themselves or the downstream proteins. In this way, RTK controls multiple signalling pathways in different aspects of the cell. Several members of RTK family are the receptors for growth factors, such as insulin, epidermal growth factor

(EGF), fibroblast growth factor (FGF), platelet-derived growth factor (PDGF), vascular endothelia growth factor (VEGF), nerve growth factor (NGF), and other members of the neurotrophin family. Besides above receptors, there are also some other members, such as

231

the Eph family, Ryk, Met, muscle-specific receptor tyrosine kinase (MuSK), Ror family, and Ret (van der Geer et al., 1994; Hubbard and Till, 2000). RTK usually functions through either changing the intrinsic catalytic activity or creating binding sites for downstream proteins. The majority of RTK achieve these changes by ligand-binding induced autophosphorylation (Schlessinger 2000). NRTK family has several subfamilies, including Src, Abl, Janus kinases (Jaks), and other members. They function as the catalytic substrates of some upstream receptors, such as the RTKs and G protein coupled receptors (Hubbard and Till, 2000). Then, they relay the signals to the downstream molecules.

The RTK family have different domains at the extracellular portions varying from subfamily to subfamily. Many of them have the Ig-like domain, cysteine rich motif, and fibronectin -III domain to mediate protein-protein interaction. Their intracellular regions all have the protein tyrosine kinase catalytic domain (Hubbard and Till, 2000). For the

NRTK family, they have different types of domains located at the N-terminal region, such as the SH2, SH3, and integrin binding motif. At the C-terminus, besides the PTK catalytic domains, they have domains, such as DNA binding, F-actin binding, and focal adhesion binding domains, depending on their specific functions (Hubbard and Till, 2000). For the development and function of the nervous system, RTK family has very important roles.

Some families, like the Trk family, Eph family, Insulin receptor, MuSK family, Ryk family, and Ror family, have been shown to be involved in regulating the formation of neuronal networks (Hubbard and Till, 2000). Because Dnrk has high similarity to both the

MuSK and Ror families, I will discuss them in more detail.

232

MuSK family

Musk was initially identified by screening the library of the electric organ of Torpedo with probes corresponding to conserved sequences in tyrosine kinase (Jennings et al.,

1993). Later, rat and mouse orthologs of Musk were cloned using similar strategies

(Valenzuela et al., 1995; Ganju et al., 1995). Musk protein has four Ig domains in the extracellular region (Fig. 1). Between Ig3 and Ig4, there is a Cysteine rich motif (CRD).

The intracellular domain of MuSK contains a functional tyrosine kinase domain, a phosphotyrosine binding domain (PTB), and a C-terminal PSD-95/Dlg/ZO-1-like (PDZ) domain (Zhou et al., 1999).

The similar NMJ defective phenotypes in musk and agrin mutants suggest that they function in the same pathway (Gautam et al., 1996; Glass et al., 1996). Agrin was purified from extracellular matrix extracts of Torpedo electric organ, which is capable of mediating postsynaptic AChR clustering in vitro (Nitkin et al., 1987). Agrin is present in basal lamina of developing and mature synapses in vivo (Fallon et al., 1985; Reist et al.,

1987). Agrin released from the motor neurons can induce AChR clustering at the NMJ

(Reist et al., 1992). Between the two different subtypes of Agrins, the neural Agrin and the muscle Agrin, only the neural Agrin has the capability to induce AChR cluster formation in vitro (Burgess et al., 1999). Musk is found to participate in mediating the response to Agrin stimuli for AChR clustering at the postsynaptic site. Musk can be activated by Agrin through its first Ig domain, but there is no direct binding between them

(Glass et al., 1996). Recent discovery of LRP4, a low density lipoprotein receptor (LDLR) related protein, links Agrin to Musk by acting as a co-receptor for Agrin (Zhang et al.,

233

2008). LRP4 can bind to both Agrin and MuSK directly. And the binding of Agrin to

LRP4 can increase MuSK activity dramatically. After the Agrin stimuli, Musk is activated to transduce the signal through a putative protein RATL (Rapsyn associated transmembrane linker), to the cytoplasmic protein Rapsyn, which is associated with

AChR, to induce the clustering of AChR. Besides the signalling through the Rapsyn, the cytoplasmic domain of MuSK can also interact with several other cytoplasmic proteins, such as Abl, GGT, Dvl, and Tid1 (Strochlic et al., 2005; Linnoila et al., 2008). Those interactions may modulate the actin cytoskeleton through Rac/Cdc42 and Pak, and thus facilitate AChR clustering.

In addition to its ability to induce AChR clustering at postsynaptic sites at NMJ, MuSK also has other functions. The interaction between the cytoplasmic region of MuSK with other cytoplasmic proteins can signal through Rac/Cdc42 to modulate synaptic genes expression through the JNK pathway. Potential target genes of this pathway include erbB,

AChR, and musk itself (Lacazette et al., 2003). MuSK is also able to regulate synaptic transcription through MAPK/PI3K, mediated by an adaptor protein14-3-3γ (Strochlic et al., 2004). The C-terminal PDZ domain of MuSK can interact with the scaffolding protein

MAGI-1c, which may regulate the building up of certain protein complex at the NMJ for nerve terminal differentiation and synaptic transmission (Strochlic et al., 2001). Another interesting discovery about the transcription regulation of musk is that in mammals, Wnt signalling pathway can control musk expression level at the NMJ (Kim et al., 2003).

Recently MuSK was shown to control the prepatterning of muscles and thus define the sites for motor axon to grow and form synapses (Kim and Burden, 2008).

234

Zebrafish musk, unplugged (unp), has been shown to possess some novel functions compared to its mammalian orthologs (Zhang et al., 2004). The unp gene produces two protein isoforms, UnpSV1 and UnpFL (Fig. 1). UnpFL is the full-length protein, containing three Ig domains, a CRD, a kringle domain, and a cytoplasmic tyrosine kinase domain. UnpSV1 contains the domains listed above except those three Ig domains (Zhang et al., 2004). The domain structure of UnpSV1 is very similar to the fly Dnrk protein (Fig.

1). Because in mammals, Ig domains have been found to be involved in mediating interaction with Agrin and Rapsyn, protein structure differences between UnpFL and

UnpSV1 suggest that they may have different functions.

Similar to the defects in AChR clustering at the NMJ in mammalian musk mutant, the unp mutant has largely reduced AChR clusters at both focal innervations sites and distributed innervations sites. Besides those defects, the unp mutant also shows motor axon guidance defects at specific choice points. Further analysis shows that those two Unp isoforms have non-substitutive functions. UnpFL is the one responsible for NMJ AChR clustering induced by Agrin and synaptogenesis, while UnpSV1 mainly controls motor axon guidance and the prepatterning of AChR before the arrival of axon (Zhang et al., 2004).

To fulfil the function in AChR prepatterning and control axon guidance, UnpSV1 needs to bind to Wnt family members through its CRD, and later recruit Dishevelled in the signalling pathway (Jing et al., 2009). Function in axon guidance and interaction with

Wnt ligand are two special features of the fish Musk, which have not been discovered in other existing MuSK family members. In this case, zebrafish MuSK shares the function with the worm Cam-1 in the close related Ror family. Because the UnpSV1 is very

235

similar to fly Dnrk protein, we speculate that Dnrk may have a similar function in the fly nervous system like that of UnpSV1.

Ror family

Receptor tyrosine kinase-like orphan receptor (Ror) family was originally named after its unknown binding ligands. Ror family proteins are transmembrane proteins. Most family members have an Ig domain, a CRD motif, and a kringle domain in the extracellular region (Fig. 1). The intracellular regions of Ror family all contain a tyrosine kinase domain. Some of them also have a proline-rich domain (PRD), and serine/threonine rich domains (Green et al., 2008).

Ror family members have been discovered in many species. Vertebrate Ror family normally has two members, Ror1 and Ror2. They were previously known as neurotrophic tyrosine kinase receptor NTRKR1 and NTRKR2 because they were first identified in human neuroblastoma cell line by a PCR-based screen for tyrosine kinase similar to the

Trk neurotrophic receptor (Masiakowski and Carroll 1992). In vertebrates, Ror members include human hRor1& hRor2, zebrafish Ror1& Ror2, chickens cRor1 and cRor2, frogs

XRor1 and XRor2, and mouse mRor1 and mRor2 (Green et al., 2008). In invertebrates, the Ror family generally has only one member in a single species, such as fly Dror, worm

Cam-1/lin-18, and sea slugs ApRor (Green et al., 2008). In fly, Dror has a close relative, called neurospecific receptor kinase (Dnrk) (Oishi et al., 1997). Before the detailed sequence alignments and comparison between family members of Ror and MuSK, Dnrk was listed as a member of Ror family because of the domain-structure similarity. And

236

recently, Dnrk was grouped into the Musk family based on the sequence alignment of different domains between Dnrk and MuSK family members (Sossin, 2008). However, there is still no sharp dividing line between the Ror and Musk families, which are the descendants from the same kinase ancestor. For example, the worm Ror protein Cam-1 can be listed in either family based on the sequence alignment of different domains.

Ror family members have been implicated in regulating different aspects of development.

Many of Ror studies were carried out in culture cells (Green et al., 2008). in vitro, Rors have been shown to function as receptors for Wnt family members. In vivo functions of

Rors have also been discovered in different species. Mutation of hrors in human causes skeletal defects in Robinow syndrome and dominant brachydactyly type B (Oldridge et al., 2000, Schwabe et al., 2000, van Bokhoven et al., 2000, Afzal et al., 2000, Ermakov et al., 2007). Ror also functions as a potent survival kinase in HeLa cervical carcinoma cells

(MacKeigan et al., 2005). In mouse, ror genes have similar but partial redundant function in the development of skeleton, cardiac and respiratory systems (Nomi et al., 2001).

Ror family is also found to be involved in regulating the planar cell polarity (PCP). PCP describes the phenomenon that groups of cells are polarized along the plane of the epithelium, perpendicular to the apical-basal axis. PCP pathway is shown to be controlled by the Wnt signalling. Frizzled, dishevelled (Dvl), Van Gogh, Prickle and Flamingo are core components in this PCP signalling pathway (Seifert and Mlodzik, 2007). Studies on

Xror2 reveal the importance of Wnt5a-Ror2-JNK in the Xenopus convergent extension, which is one type of PCP (Hikasa et al., 2002; Oishi et al., 2003; Schambony et al., 2007).

237

In C. elegans, Cam-1as the receptor for Wnt, controls the epithelia cell orientation polarity, which is another type of PCP, in an integration pathway with Wnt-Van Gogh

(Green et al., 2007 and 2008). The mror2 mutant mouse also shows defects in PCP of hair cells in the inner ear (Yomanoto et al., 2008). Similar to the function in PCP, ror2 also functions in the Wnt5a induced polarized cell migration in cultured cell (Nomachi et al.,

2008).

Interestingly, although abundant mror1 and mror2 mRNAs are present in the developing nervous system of mouse (Oishi et al., 1999), double knock-out mouse of mror1 and mror2 shows no visible defects in the nervous system. Double mutants only carry deficits in the lung, heart, and skeleton development (Nomi et al., 2001; Oishi et al., 2003). In the dissociated embryonic hippocampal neuron culture, both mror1 and mror2 are expressed and concentrated in axonal processes and growth cones (Paganoni and Ferreira, 2003).

Manipulating mrors by RNAi could lead to the change in neurite extension and branching

(Paganoni and Ferreira, 2005). This is different from the double mutant phenotype. So, currently the in vivo function of mror in the nervous system of mouse is still unclear.

To date, C. elegans Ror-like protein named Cam-1(canal-associated neuron migration defective 1) is the only Ror family member showing an in vivo function in the nervous system (Fig. 1). In C. elegans, cam-1 mutant shows defects in neuronal migration, axon outgrowth and guidance, cell polarity formation, synaptic transmission, and neurite pruning (Forrester et al., 1999 and 2004; Koga et al., 1999; Zinovyeva et al., 2008; Green et al., 2007 and 2008; Francis et al., 2005; Hayashi et al., 2009). In some of those

238

processes, Cam-1 functions as the receptor for Wnt ligand (Forrester et al., 2004; Green et al., 2007 and 2008). Because of its function in regulating the localization of AChR at the neuromuscular synapses and controlling synaptic transmission (Francis et al., 2005),

Cam-1 has the similar function as Unp, the zebrafish MuSK protein.

Dror

As I described previously, there are two highly homologous members in the fly Ror family, Dror and Dnrk (Fig. 1). Dror was discovered in 1993 (Wilson et al., 1993).

Researchers used the PCR strategy to screen for putative RTKs in the fly larval brain cDNA library. In the PCR reaction, degenerate oligonucleotide primers were used to hybridize to DNA sequence corresponding to well-conserved tyrosine kinase subdomains

(Wilson et al., 1993). They discovered and named this putative RTK as Dror because it is homologous to hror in the human genome (Masiakowski and Carroll, 1992; Wilson et al.,

1993). Dror shares the CRD/Frizzled domain, the Kringle domain, and the tyrosine kinase domain with its human homolog, but does not have the Ig domain present in the hRor

(Fig. 1). The Frizzled domain is firstly discovered in Frizzled protein, which can bind to the Wnt family members. Due to the unavailability of a clean Dror mutation, the function of Dror is still not clear. But based on our preliminary data and the functions of other ror family members, we hypothesize that Dror may be involved in axon guidance in the fly nervous system.

Dnrk

239

Dnrk gene was cloned in a similar way as that of Dror (Oishi et al., 1997). The sequence of Dnrk shows high similarity to Dror. Dnrk protein has one CRD/Frizzled domain, one membrane-proximal Kringle domain, the transmembrane domain, and a cytoplasmic tyrosine kinase domain (Oishi et al., 1997) (Fig. 1). The kringle domain is a characteristic feature of the Ror family (Forrester 2002). The sequence of the CRD and the tyrosine kinase domain are highly homologous to similar domains in the MuSK family proteins

(Sossin 2006). The in vitro kinase assay showed that the tyrosine kinase domain of Dnrk protein is kinase active in cultured Cos-7 cells (Oishi et al., 1997).

The Dnrk gene is mapped to the right arm of the second chromosome at the locus 49F4.

Its annotation symbol in flybase is CG4007. Dnrk gene spans around 3435 nucleotide acids on the genomic map. The mRNA of Dnrk is 3249 nucleotide acids in length. After translation, the Dnrk gene produces a transmembrane protein of 724 amino acids. In the

BDGP EST clone collections, clone RE63791 contains the full length Dnrk cDNA. Dnrk is located in a complex genome locus (Fig. 2). Its 5‟ UTR overlaps with the gene tppII,

Cap-G, and CG34439. Its 3‟UTR overlaps with gene CG15870 and is close to the gene

GLaz. TppII is tripeptidyl-peptidase II. It functions in proteolysis (Renn et al 1998). Cap-

G functions in mitotic chromosome condensation, spindle organization, and sister chromatid segregation (Dej et al., 2004; Jager et al., 2005). GLaz is the fly homolog of the human lipoprotein Apolipoprotein D. It is expressed in the fly nervous system and controls the stress response and lifespan (Sanchez et al., 2000 and 2006; Walker et al.,

2006; Muffat et al 2008).

240

Northern blot analysis shows that Dnrk is expressed highly during the embryonic stage and the pupal stage, coincident with the time points when embryonic nervous system begins to develop and when the nervous system undergoes morphogenesis (Oishi et al.,

1997). The in situ hybridization data shows a very specific tissue distribution of Dnrk transcripts during the embryogenesis. Weak expression of Dnrk is observed at the embryonic stage 10 (E10) in the ventral area of the germ band corresponding to the neurogenic ectoderm. This expression becomes stronger and clearer at the E11 between the epidermal and the mesodermal cell layer and sustains throughout the entire embryonic stage. After the germ band shortening, the distribution of Dnrk transcripts matches the profile of developing commissural and connective neuronal projection tracts. Eventually the Dnrk expression pattern is restricted to the CNS and PNS only (Oishi et al., 1997).

The specific expression pattern of Dnrk during the embryonic development highly suggests that Dnrk play a critical role in fly nervous system development.

Based on these expression patterns of Dnrk, we hypothesized that Dnrk might have important functions in the developing fly nervous system. Starting with the generation of

Dnrk mutants, I have conducted a series of examinations in both the central and peripheral nervous systems of Dnrk mutant flies at different developmental stages.

Defects were observed only in R-cell axon guidance during the larval stage with a low penetrance. This mutant phenotype could be enhanced when Dror, the homolog of Dnrk was knockdown in Dnrk mutants. This suggests that Dnrk and Dror may share redundant function in regulating R-cell axon guidance in the visual system. To search for potential interacting genes of Dnrk, I conducted a genetic screen for candidates capable of

241

modifying the Dnrk overexpression phenotype in the wing. This screen led to the discovery of Dwnt5 as a potential interacting gene. My analysis helps to illustrate the potential in vivo functions of Dnrk and Dror in the fly.

242

Materials and methods

Genetics

The exact insertion site of the P-element in the GE12548 (purchased from GenExel,

Korea) was located by the PCR amplification and sequencing. To generate the imprecise excision mutation for Dnrk, the GE12548 flies were crossed to the transposase Δ2-3 flies.

From the progeny, around 800 white-eye male flies were collected and individually crossed to balancer female flies to generate mutant lines. Those lines were screened by

PCR strategy for potential deletions in Dnrk. Seven lines were selected based on their

PCR fragment size changes. They were finally sequenced to confirm the deletions in

Dnrk gene. Lines Df(2L)41C and Df(2L)170B were kindly provided by the Wilson lab.

UAS transgenic flies for the Wnt family members were obtained from the Lawrence lab

(Llimargas and Lawrence, 2001). UAS transgenic flies of the TGF-β signalling pathway were obtained from the Haghighi lab. All the other fly stocks were obtained from the

Bloomington fly center.

Molecular biology

To generate the UASDnrk construct, the EST clone RE63791 was obtained from the

BDGP fly cDNA collection. EcoRI sites flanking the Dnrk cDNA were used to subclone the Dnrk cDNA into the pUAST vector EcoRI site to make the final construct. To generate the RNAi construct for Dror, pWIZ vector was obtained also from the BDGP.

Partial Dror cDNA was obtained from the Wilson lab. Following the protocol with the pWIZ construct, 567 bp Dror coding sequence was amplified by PCR and ligated into the

243

pWIZ construct. Oligonucleotide primers used were “5'CGCAAATTGCCGCATG3'” and

“5'CCGATGAGTTCAGGG3'”. I generated both the head-head and tail-tail orientated hair-pin RNAi constructs. Those constructs were used to transform fly germline by standard method.

Histology and Immunohistochemistry

The eye-brain complexes from the third-instar larvae were dissected and stained as described (Ruan et al., 1999). The adult brains were dissected and stained in a similar way.

Stage E16-17 embryos were fillet prepared and stained as described for the examination of the VNC and the motor axon guidance (Van Vactor et al., 1993). For the observation of da neurons in the embryonic stage, late E17 embryos with mdGAL4>UASGFP expression were lined up on the microscopy slides and directly examined under the confocal microscopy. To examine the dendrite tiling and self-avoidance phenotype, third- instar stage larvae were dissected in PBS, mounted on slides and directly imaged by confocal microscopy. The images for all embryonic dendrite analysis were taken as Z- series and merged together. For the analysis in the wing, wings from two days old adult flies were cut off by micro-scissors and mounted on slides for direct photographing under the light microscope.

Primary antibodies were used as the following dilutions: MAb 24B10 (1: 200 dilution,

DSHB), ID4 (1: 100 dilution, DSHB), BP102 (1: 50 dilution, DSHB), J17 to β- galactosidase (1:500 dilution). The secondary antibodies (i.e. HRP- and FITC-conjugated

244

goat anti-rabbit or anti-mouse secondary antibodies) (Jackson Immunochemicals) were used at 1:200 dilution.

245

Results

Generate Dnrk mutants

Because there was no clean mutant available for Dnrk, I decided to generate the Dnrk mutation. From the GenExel (Taejon, Korea) EP line collection, we found an EP line

GE12548, carrying a p{EP} p-element inserted into the 5‟ UTR of Dnrk (Fig. 2).

Sequencing data showed that the P element was inserted into the locus 284 bp away from the start codon of Dnrk. I used the imprecise excision technique to generate Dnrk mutant.

The advantage of P-element insertion induced mutant is that the transposable element can be remobilized by the transposase P [∆2-3] when they are present together in the male or female germ cells. Because the P-element is marked by the mini-white gene and the existence of this gene can be followed by the eye color, the progeny can be screened for the loss of eye color (Kaiser 1990). When the P-element moves out of the original insertion site it occasionally takes some nearby genome sequence together with it. This is called imprecise excision (Kaiser 1990; Ryder and Russell, 2003; Hummel and Klambt,

2008). After the taking place of the imprecise excision, around eight hundred individual white-eye fly lines were generated and subjected to a PCR amplification screen for deletions in the Dnrk gene. Seven lines showed obvious difference on gel shifting, as #17,

20, 88, 96, 220, 479, and 761. The sequencing analysis confirmed the deletions in all seven lines. #220 has the largest deletion of around 2100 bp. This deletion started from the insertion site of the GE12548 P-element and ended at around two thirds of the last coding exon of Dnrk, deleting most of the CDS (Fig.2). So, theoretically #220 is a

246

molecular null allele of Dnrk. Following analysis was done with this null mutation. If it is not specified, in the following description, Dnrk mutant refers to this null mutant #220.

Analyze Dnrk mutant phenotype in the third-instar larval visual system

Dnrk mutant is homozygous viable. No obvious external morphology or behavior defects were observed. We conclude that Dnrk is not essential for viability. Eye-brain complexes from the third-instar larval Dnrk mutant were stained with MAb 24B10 (Van Vactor et al,

1988) to examine R-cell development and R-cell axon projection pattern. MAb 24B10 recognizes a cell adhesion protein called Chaoptin, which is expressed specifically on all

R-cell cell bodies and their axons (Van Vactor et al, 1988). This labelling allows me to examine R-cell cell body localization and patterning, and R-cell axon guidance and targeting in the fly visual system.

In wild type (Fig. 3A), every differentiating ommatidium sends out a single axon bundle consisting of eight axons from R1-8, toward the most posterior end of the eye-imaginal disc, where it converges with axon bundles from other ommatidia to enter the optic stalk.

After exiting the optic stalk, R-cell axons project evenly over the superficial lamina layer.

R1-6 axons terminate within the lamina, and their growth cones expand to establish a smooth lamina layer. R7 and R8 axons project through the lamina to enter the medulla, in which their growth cones generate an evenly spaced topographic map. Within the medulla,

R7 and R8 growth cones expand significantly and display a characteristic inverted “Y- like” morphology.

247

In 26 out of 28 Dnrk null mutant hemispheres examined, both the R-cell development and axon projection were normal (Fig. 3B). R-cell axons in these mutants migrated normally to innervate the brain. Mutant axons projected normally in both the lamina and medulla regions, with wt-like growth cone morphologies. Only two hemispheres showed abnormal

R-cell axon projection patterns (Fig.3C). In those two eye-brain complexes, after exiting the optic stalk, before reaching the lamina, some R-cell axons crossed neighbouring axons and occasionally fused to form thicker axon bundles. Gaps were observed in the lamina.

R7 and R8 axons in the medulla had unexpanded growth cones. The overall projection field is smaller than wild type.

To determine if Dnrk was involved in regulating the lamina-versus-medulla ganglion specific targeting of R-cell axon, I used the marker ro-τ-lacZ (Garrity et al., 1999) to examine the targeting of R2-R5 axons, a subset of R1-R6 axons, in the third-instar Dnrk mutant larvae. In wild type (Fig. 3D), the vast majority of R2-5 axons labelled by the ro-

τ-lacZ marker terminate in the lamina layer, with only 2-5 axons to mistarget into the medulla region. Like wild type, Dnrk mutant R2-5 axons terminated correctly in the lamina layer (38/40 hemispheres scored) (Fig. E). This analysis suggests that Dnrk is not involved in regulating the layer-specific targeting of R1-6 axons at the third-instar larval stage.

During the third-instar larval stage, there is a complex interaction between R-cell axons and glial cells in the target region. R-cell axons induce the differentiation and migration of lamina glial cells (Perez and Steller, 1996; Suh et al., 2002; Dearborn and Kunes,

248

2004). Conversely, lamina glial cells, including epithelial glia and marginal glia, provide stop signal for R1-6 axons to terminate in the lamina layer (Poeck et al., 2001). To determine the development and migration of those lamina glia, I used a marker Omb-lacZ to label those cells in the optic lobe (Huang and Kunes, 1998). In the wild-type lamina region, Omb-lacZ labels the well organized epithelial and marginal glia surrounding the lamina plexus (Fig. 3F). In Dnrk mutants, lamina glial cells migrated correctly to the lamina. They were well organized on either side of the lamina layer (Fig. 3G). In all Dnrk mutant hemispheres examined with omb-lacZ labelling (n=10), I did not observe any abnormality in the differentiation and migration of lamina glia.

Analyze the adult fly visual system in Dnrk mutants

To investigate R1-6 axons targeting in adults, an adult marker Rh1-lacZ specific for R1-6 axons was used (Mismer and Rubin, 1987). In whole-mount wild-type adult brain, all R1-

6 axons labelled with Rh1-lacZ terminate in the lamina (Fig. 4A). No labelled axon was present in the medulla region. Like wild type, Dnrk mutant R1-6 axons all terminated correctly in the lamina (20/20 hemispheres) (Fig. 4B).

To assess the projection pattern of R7 and R8 axons in the adult brain, GMR-lacZ (Moses and Rubin, 1991; Ellis et al., 1993) was used to label both wild type and Dnrk mutants. In the GMR-lacZ labelled and whole-mount wild-type brain, R7 and R8 axons from the same ommatidium form a single axon fascicule to enter the medulla (Fig. 4C). R8 axons terminate in the M3 medulla layer and R7 axons migrate deeper to the M6 layer. R7 and

R8 axons within the same fascicule restrict their axon terminals in the same medullar

249

column. So, the projection and termination of R7 and R8 axons in the medulla display an evenly spaced pattern (Ting et al., 2007). In Dnrk mutants, the termination and patterning of R7 and R8 axons in the medulla were indistinguishable from the wild type (14/14 examined) (Fig. 4D). Mutant R7 and R8 axon terminals were evenly spaced, without any lateral extension or crossing. I speculate that abnormal axon guidance defects observed in the third-instar Dnrk mutant larvae may be corrected during the pupal stage.

Examine the embryonic ventral nerve cord development and motor axon guidance in the Dnrk mutant

The high expression level of Dnrk mRNA detected in the VNC during the embryonic stage (Oishi et al., 1997) suggests that Dnrk may play a role in the development of the

VNC and potentially in motor axon guidance. Dnrk mutant E16-E17 embryos were subject to the fillet preparation. MAb 1D4 and MAb BP102 were used to label subsets of axon connectives within the ventral neuropil. MAb 1D4 detects the membrane-bound cell adhesion protein FasII expressed on subsets of longitudinal axon connectives in the VNC

(Van Vactor et al., 1993). In wild-type E16-17 embryos, there are three obvious 1D4 positive longitudinal axon tracts continuous and evenly spaced on either side of the VNC

(Fig. 5A). In Dnrk mutant embryos, MAb 1D4 labelling within the VNC revealed a similar pattern as that in wild type (12/12) (Fig. 5B). Those 1D4 positive longitudinal axon bundles were well formed, spaced, and continuous.

In wild-type E16-17 embryos, MAb BP102 detects an unknown epitope (Seeger et al.,

1993) and highlights an overall ladder-like pattern for both longitudinal and commissural

250

axon connectives within the VNC (Fig. 5C). In all Dnrk mutants examined, the BP102 labelling patterns within the VNC were all wild-type-like (7/7). Both the longitudinal and the commissural axon connectives were well formed, continuous, and smooth (Fig. 5D).

The space between the anterior commissures and the posterior commissures was clearly and evenly defined from segment to segment. No abnormal crossing between those commissures was observed.

With the MAb 1D4 staining, motor axon guidance and termination can also be visualized.

In abdominal segments A2-A7, motor axons exit the CNS through the segmental nerve

(SN) and intersegment nerve (ISN) roots. At the exit of these junctions, motor axons resort into five major branches, ISN, SNa, SNc, ISNb, and ISNd, which innervate different body wall muscle groups (Landgraf et al., 1997; Schmid et al., 1999). ISNb branch exits the VNC at the ISN root together with ISN and ISNd. After exiting, ISNb defasciculates from the ISN and extends dorsally through the ventral musculature to innervate between muscle 6/7 and between muscle 12/13. The SNa and SNc branches constitute the SN fascicule. After exiting the CNS at the SN root, the SNa defasciculates from the SNc and continue extending dorsally to innervate muscle 5, 8, 12, 22, 23, and 24

(Landgraf et al., 1997; Schmid et al., 1999) (Fig. 5E). The overall motor axon guidance and termination of ISN, SNa, and ISNb in the Dnrk mutant occurred normally (13/13 examined) (Fig. 5F). Mutant motor axons reached their muscle targets correctly. Muscle targeting and early synapse formation were also normal.

Test the potential function of Dnrk in dendritic arborization (da) neurons

251

The in situ pattern of Dnrk at stage E16-17 shows the expression of Dnrk mRNA in the

PNS neurons on the body wall of the embryo (Oishi et al., 1997). This led us to the speculation that Dnrk might be specifically required for the development of PNS neurons.

I chose to examine dendritic arborization (da) neurons in Dnrk mutants. All da neurons in the peripheral can be highlighted by the GAL4109(2)80 (also named mdGAL4) driving a membrane-targeted GFP (UAS-CD8::GFP) (Gao et al., 1999; Brenman et al., 2001).

During the embryonic stage at the dorsal cluster, mdGAL4>UASGFP labels all eight multiple dendritic (md) neurons, including six da neurons, one bipolar neuron, and one trachea innervation neuron. da neurons are divided into four classes from class I to class

IV, with increase in dendritic complexity. Those six da neurons within the dorsal cluster contain two class I: ddaD and ddaE; one class II: ddaB; two class III: ddaA and ddaF; and one class IV: ddaC. Each class of da neurons have their characteristic feature of dendritic development and patterning. Normally the labelling of mdGAL4> UASGFP is turned on at the stage E15 (Sugimura et al., 2004).

In wild type, primary dendrites of dorsal cluster da neurons emerge at around E13-14. At the stage of E17, those six da neurons in the dorsal cluster can be easily visualized with strong membrane-bound GFP (Fig. 6A). The cell body of each da neuron has a stereotyped location. Each cluster has multiple dorsal branches and lateral branches. The primary branches and overall dendritic patterning of each class of da neurons can be distinguished from the others. Class I da neurons ddaD and ddaE already obtain their characteristic comb-like structures. The overall dendrite branch length of the dorsal cluster is largely the same from segment to segment.

252

At the stage E17, the dorsal md clusters highlighted by the mdGAL4> UASGFP in the

Dnrk mutant displayed a wild-type-like pattern (17/17 examined) (Fig. 6B). Each cluster contained correct numbered da neurons. Locations of those da neuron cell bodies were normal, too. The outgrowth and overall patterning of the dorsal and lateral branches were indistinguishable from wild type. Class I da neurons also showed their featured comb-like patterns. This observation suggests that Dnrk is not required for the da neuron development and pattern formation during the embryonic stage.

Another feature of those da neurons is the dendrite tiling and self-avoidance in some classes at the late third-instar larval stage. Tiling is defined as the complete and non- redundant coverage of receptive fields by neurons of a similar functional type (Grueber et al., 2002; Parrish et al., 2007). Self-avoidance refers to the tendency for arbors from the same neuron to avoid crossing, thereby spreading evenly over a receptive field (Kramer and Kuwada, 1983). With the mdGAL4>UASGFP labelling, in wild type, at the dorsal midline and at the segment boundary, the high-order dendrite branches of class IV da neuron ddaC normally avoid dendrite branches from the neighbouring ddaC (Fig. 6D).

They normally do not overlap with each other. Within the same ddaC neuron, dendrite branches normally avoid crossing each other to display the self-avoidance phenomenon

(Fig. 6D). In Dnrk mutants, distal dendrites of neighbouring class IV ddaC neurons did show correct tiling pattern (60/60 scored). Mutant ddaC branches terminated correctly at the dorsal midline and at the segment boundary (Fig. 6E). Within the same ddaC, dendrites displayed self-avoidance between sister branches (60/60 scored) (Fig. 6E). Thus,

253

Dnrk does not appear to function in the dendrite development processes of da neurons in the PNS.

Potential redundancy between Dnrk and Dror

The low penetrance of R-cell axon guidance defects in Dnrk mutants raised the possibility that another gene (e.g. a homologue of Dnrk) might compensate for its function. Dror is a candidate gene because it shares the highest homology and similar expression pattern in fly embryos with Dnrk (Wilson et al., 1993; Oishi et al., 1997) (Fig. 1).

Dror is located on the left arm of the second chromosome at the locus 31B1. It is close to the gene basket, DPTEN, CG5676, and chico. Basket is the fly homolog of JNK, the

MAP kinase. It is involved in multiple developmental processes of the fly nervous system.

Both DPTEN and Chico are involved in cell growth. Currently there is no specific Dror mutant fly available. We only obtained two small deficiency fly lines, Df(2L)170B and

Df(2L)41C (Goberdhan et al., 1999). Df(2L)170B deletes gene chico, basket, Dror,

CG5676, and DPTEN. Df(2L)41C deletes chico, basket, and Dror.

To test the potential redundancy between Dror and Dnrk, I reduced the endogenous dosage of Dror by 50% by crossing those deficiencies into Dnrk mutant background. In the visual system, R-cell axon projection patterns in those Dror−/+Dnrk−/− animals were still like Dnrk mutant alone (80 hemispheres examined). The majority of them (77/80) displayed the wild-type-like R-cell axon projection pattern, with smooth lamina, evenly spaced R-cell axons, and well expanded growth cones in the medulla (Fig. 7B). The da

254

neuron development and patterning during the embryonic stage in those mutants were also normal (12/12). The dendritic outgrowth and patterning were similar to that in the wild type (Fig. 6C). One possible explanation for those wild-type-like phenotypes was that the remaining 50% of Dror might still be enough to control axon guidance and dendritic patterning. The ideal way to test the redundant function between two genes is to examine the double mutant phenotype. Because the mutants we had for Dror removed several other genes together with Dror, and some of those genes function in R-cell axon guidance and the general cell growth (Berger et al., 2008), it was not practical to investigate the function of Dror and Dnrk in double mutants by using Dnrk mutant and those Dror deficiency mutants.

To solve the problem due to the unavailability of a clean Dror mutant, I decided to use the RNA interference technique. By using RNAi construct pWIZ (Lee and Carthew,

2003), I generated Dror RNAi transgenic flies. Driven by either the eye-specific driver

LongGMRGAL4 (Wernet et al., 2003) or the neuron-specific driver elavGAL4 (Long et al.,

2006), those Dror RNAi could knock down Dror expression either specifically in the eye or specifically in all neurons in the Dnrk mutant background. I found that knocking down

Dror expression either in the eye or in neurons in the Dnrk null mutant background could disrupt the R-cell axon projection pattern, while the Dror RNAi driven by those GAL4s alone yielded a wild-type-like projection pattern. When Dror was knocked down in the eye of the Dnrk mutant, R-cell axons ectopically crossed their neighbours before reaching the lamina layer and R7 and R8 growth cones within the medulla did not expand properly

(23/35 hemispheres examined) (Fig. 7C). When Dror was knocked down in neurons in

255

the Dnrk mutant, R-cell axon projection patterns displayed similar defects as that in the eye-specific Dror knocking down in the Dnrk mutant (13/22 hemispheres examined) (Fig.

7D). Those defects I observed were similar compared to the Dnrk single mutant but with a much higher penetrance. This result is encouraging because it suggests that very likely

Dnrk and Dror have redundant function in the fly visual system. Future preparation and examination of double mutants of Dnrk and Dror will be necessary to confirm this. It will also be interesting to examine the VNC and PNS neurons to see if those two genes also show redundant roles in other systems.

Generate UASDnrk transgenic fly and analyse the Dnrk overexpression phenotype

One way to determine the potential role of a gene, which may function redundantly with another gene, is to examine its overexpression phenotype (Rorth et al., 1998). The expression sequence tag (EST) clones RE63791 for Dnrk is available from the Berkeley

Drosophila Genome Center (BDGC). I cloned the complete Dnrk cDNA into the pUAST vector (Brand and Perrimon, 1993) and used the final construct to generate Dnrk transgenic flies.

Different tissue specific promoters were used to drive the overexpression of Dnrk. The

GMRGAL4 was used to overexpress Dnrk specifically in the eye. In 20 hemispheres I examined, R cell development and R-cell axon projection all displayed normal pattern

(20/20) (Fig. 8B). The mdGAL4 drove UASDnrk in multiple dendritic neurons also showed normally developed and patterned da neurons (22/22) (Fig. 8D). But outside of the nervous system, I did observe interesting overexpression phenotype of Dnrk when I

256

used the wing specific drivers ApGAL4 (O‟Keefe et al., 1998; Milan and Cohen, 1999) and 1096GAL4 (Capdevila and Guerrero, 1994) to overexpress Dnrk in the fly wing disc.

Adult wing develops from the larval wing-imaginal disc. On the wt adult wing blade, there are five longitudinal veins (LV). From the anterior to the posterior they are named as L1 to L5 (Fig. 9A). And there are also two cross veins (CV), the anterior cross vein

(ACV) and the posterior cross vein (PCV). ACV is between L3 and L4 and it is close to the wing hinge. PCV is located more distally between L4 and L5 (Bier, 2000). When

Dnrk was overexpressed in the wing, the loss of CV was frequently observed (19/20 wings examined) (Fig. 9B and 9C). Some wings missed only one CV (10/20) and some lost both (9/20). When only one CV was missing, in most cases it was the ACV (9/20)

(Fig. 9C). Because Dnrk mutant did not show any defect in wing development, the overexpression phenotype of Dnrk could be explained by two possibilities. Either another gene functions redundantly with Dnrk on the wing or this is a completely neomorphic phenotype of Dnrk. I decided to look further into this wing phenotype.

There are several major signalling pathways involved in the wing development, including the wingless/Wnt pathway, the dpp pathway, the hedgehog pathway, the EGF pathway, and the notch pathway as reviewed by Bier (Bier, 2000). Dnrk may function in one or some of those pathways. Since the wing vein loss phenotype was easy to score, I proposed a modifier screen to identify Dnrk-interacting genes in the wing. Because our lab has a collection of UAS transgenic flies of candidate genes involved in wing development, I decided to have these UAS transgenes co-overexpressed with Dnrk in fly

257

wings to look for modification. From this screen I found that Dnrk genetically interacted with DWnt5 in the wing.

Overexpression of the dpp signalling pathway members did not modify the loss of CV phenotype induced by Dnrk overexpression (table 1). But the overexpression of one Wnt family member, DWnt5 could efficiently suppress the CV loss phenotype in the Dnrk overexpression mutant (Fig. 9E and 9F). Both ACV and PCV formed normally in the vast majority of flies overexpressing both Dnrk and DWnt5 (49/57 wings scored).

Occasionally, there was even ectopic CV formed in the wing (16/57 scored) (Fig. 9F).

Interestingly, the overexpression of DWnt5 alone resulted in a completely normal wing

(77/77 examined) (Fig. 9D). For other members of fly Wnt family I examined, there was no modification on the loss of CV phenotype found in the Dnrk overexpression mutant.

This suggests the specificity of the genetic interaction between Dnrk and DWnt5 in the wing. For results of all other candidate genes I tested, please see table 1.

258

Figure 1

259

Figure 1. Predicted domain structure of Ror family proteins.

Human MuSK and zebrafish MuSKs: UnpluggedFL (UnpFL) and UnpSV1 are shown for comparison. Conserved domains are indicated. N-termini are to the left. Note the high similarity between Dnrk and Dror.

260

Figure 2

261

Figure 2. Molecular characterization of the Dnrk gene.

Predicted genomic organization of the Dnrk and several nearby genes: tpp-II, Cap-G,

GLaz, CG34339, and CG15870 at 49F4 by the BDGP Drosophila genome project. Exon- intron boundaries in the Dnrk gene were determined by comparing the sequence data of the full-length EST cDNA clone RE63791 to the genomic sequence. The P-element insertion site for GE12548 is located within the 3rd exon, 284 bp away from the start codon of Dnrk. The deletion mutant line #220 deleted 2128 bp genomic sequence as indicated. Open boxes, noncoding regions; filled boxes, coding regions.

262

Figure 3

263

Figure 3. Mutation in the Dnrk gene disrupted R-cell axon guidance but not the layer-specific targeting and lamina glia development.

(A-C) R1-8 axons in third-instar larvae were stained with MAb 24B10. (A) In wild-type

R-cell axons formed smooth retinotopic arrays in the lamina (la) and medulla (me).

Before the lamina plexus, R-cell axons were evenly spaced. In the medulla, growth cones of R7 and R8 axons displayed inverted “Y-shaped” morphology. (B) In the majority of

Dnrk homozygous mutants (26/28 hemispheres examined), R-cell axon projection patterns formed normally. (C) At a low penetrance (2/28), the Dnrk mutation induced defects in the R-cell axon projection. The overall projection field was smaller than wild type. Before reaching the lamina plexus, R-cell axons abnormally crossed and stuck with their neighbours (arrows). Gaps (arrowhead) were present in the lamina. The array of R7 and R8 growth cones in the medulla was disorganized with the failure to expand their growth cone properly. (D-E) Dnrk was not required for the lamina specific targeting of

R2-5 axons. Third-instar larval eye-brain complexes were labelled with marker ro-τ-lacZ.

In wild type (D), the vast majority of R2-5 axons terminated in the lamina layer. (E) In

Dnrk mutant, R2-5 axons targeted correctly to the lamina (n=40 hemispheres). (F and G)

Dnrk mutation did not affect the development and patterning of lamina glia. Anti-β- galactosidase antibody was used to visualize the omb-lacZ expression in lamina glia. (F)

In wild type, lamina glia differentiated and migrated into the lamina region. There were two layers of glia, the epithelial glia and marginal glia, located on either side of the lamina to provide the stop signal. (G) In Dnrk mutant, lamina glia were well

264

differentiated and organized around the lamina layer (n=10 hemispheres). Scale bar:

20μm.

265

Figure 4

266

Figure 4. The adult R-cell axon projection pattern in Dnrk mutant was normal.

(A and B) Whole-mount adult brains were stained with anti-β-galactosidase antibody to visualize the adult R1-6 marker RhI-lacZ. (A) In wild type, all R1-6 terminated in the dark stained lamina. No axon was stained within the medulla. The fiber-like structure in the medulla and the main brain was the tracheal system. (B) In the Dnrk mutant brain, all

R1-6 axons stopped in the lamina. No R1-6 mistargeting was observed (n=20 hemispheres). (C and D) Whole-mount adult brains were stained with anti-β- galactosidase antibody to visualize the adult R1-8 marker GMR-lacZ. Lamina was removed for better observation. (C) In wild type, single R7 and R8 axon from the same ommatidium formed an individual fascicule to innervate the medulla. Within the medulla,

R8 axon terminated in the M3 layer, while R7 terminated in the deeper M6 layer. R7 and

R8 axon terminals in the medulla formed an evenly-spaced smooth topographic array.

The R7 and R8 in the same fascicule restricted their terminals within the same medulla column. (D) In Dnrk mutant, R7 and R8 axons within the medulla displayed the well organized topographic array. Axon terminals maintained the normal tiling pattern (n=14 hemispheres). Scale bar: 20μm.

267

Figure 5

268

Figure 5. Dnrk mutant did not display axon guidance defect in the VNC and the neuromuscular system.

(A-F) Fillet prepared E16-17 embryos. (A and B) MAb 1D4 stained VNC to reveal the three major longitudinal connectives on each side of the CNS. (A) In wild type, each connective is tightly bundled and continuous. (B) Dnrk mutant had a wild-type-like pattern in the VNC with evenly spaced and continuous longitudinal connectives (n=12 individuals). (C and D) MAb BP102 stained CNS axons in the VNC. (C) In wild type embryo, each segment has two major longitudinal connectives and two commissural connectives as anterior commissure and posterior commissure. All commissures were well separated. Together with longitudinal connectives, they formed the ladder-like structure. (D) Dnrk mutant had a normally organized axon tract organization pattern (n=7 individuals). (E and F) MAb 1D4 stained abdominal segments to reveal motor axons.

Ventral muscles 12 and 13 and ISN are indicated. (E) Axons that made up the SNa in a wild type embryo extend through the ventral muscles as a fasciculated bundle (SNa).

After reaching the dorsal edge of muscle 12, axons within the SNa gave rise to a dorsal and a lateral branch. (F) Dnrk mutant had normally formed SNa. The branching and muscle targeting of SNa axons were correct (n=12 individuals).Scale bar, 5μm.

269

Figure 6

270

Figure 6. Dendrite development and tiling pattern in wild type and mutants.

(A-C) In the late embryonic stage 17 embryos, dendritic trees of the six da neurons in the dorsal cluster were visualized by GAL4109(2)80 driving membrane-targeted GFP (UAS- mCD8::GFP). (A) In wild type embryo, the overexposed circles at the bottom of the figure were da neuron cell bodies. Dorsal and lateral branches could be distinguished from each other. The “comb-like” structures of class I da neurons were already largely formed. (B) Dnrk mutant displayed a normal dendritic development and branching pattern

(n=17 individuals). (C) In Dnrk mutant with the dosage of Dror reduced by 50%, the da neuron development and patterning in the dorsal cluster were normal (n=12 individuals).

Genotype: Dnrk, Df(2L)41C/Dnrk. (D and E) The third-instar larval class IV da neuron ddaC in the dorsal cluster was labelled by GAL4109(2)80 driving UAS-mCD8::GFP. It was focused on the dorsal dendrite of ddaC near the dorsal midline. (D) Wild type ddaC neuron had its dorsal dendrite branches tiled at the dorsal midline with the dendrite branches from the ddaC in the facing hemisegment. (E) In Dnrk mutant, the tiling between neighbouring ddaC neurons was well developed and maintained (n=60 cells examined). Scale bar is 20μm.

271

Figure 7

272

Figure 7. Potential functional redundancy between Dnrk and Dror in the fly visual system.

R1-R8 axons in third-instar larvae were stained with MAb 24B10. (A) Wild-type formed smooth retinotopic arrays of axons in the lamina and medulla. (B) Dnrk mutant with the dosage of Dror reduced by 50% displayed a wild-type-like axon projection pattern (77/80 hemispheres). Genotype: Dnrk, Df(2L)41C/Dnrk . (C) Dnrk mutant with the eye-specific

Dror knocking down by LongGMRGAL4 driving UASDrorRNAi. The overall projection field was smaller than wild type. Before the lamina, R axons abnormally crossed neighbouring axons (arrow). Gaps (arrowhead) were present in the lamina. The array of

R7 and R8 growth cones in the medulla was disorganized. Most of them failed to expand

(23/35 hemispheres examined). Genotype: Dnrk/ Dnrk; LongGMRGAL4/UASDrorRNAi .

(D) Dnrk mutant with the neuron-specific Dror knocking down by elavGAL4 driving

UASDrorRNAi. Similar defects were observed as showed in (C) (n=22 hemispheres).

Genotype: elavGAL4/+; Dnrk/ Dnrk; UASDrorRNAi/+. Scale bar: 20μm.

273

Figure 8

274

Figure 8. Overexpression of UASDnrk in the eye and in the da neurons did not affect the R-cell axon projection and da neuron dendrite patterning.

(A and B) The third-instar larval eye-brain complexes were stained with MAb 24B10. (A)

The wild-type brain showed the well organized topographic patterning of R axons with smooth lamina. (B) Overexpression of UASDnrk specifically in the eye by GMRGAL4 produced a wild-type-like pattern (n=20 hemispheres). (C and D) In the late embryonic stage 17 embryos, dendritic trees of the six da neurons in the dorsal cluster were visualized by GAL4109(2)80 (mdGAL4) driving UAS-mCD8::GFP. (C) Dendrite outgrowth and patterning of the dorsal cluster da neurons in the wild type embryo. (D) The overexpression of UASDnrk specifically in da neurons driven by mdGAL4 resulted in normal dendrite outgrowth and patterning (n=22 individuals). Scale bar: 20μm.

275

Figure 9

276

Figure 9. The overexpression of UASDnrk in the wing could change the vein formation in adult wings.

(A) In a wild type wing, there were five longitudinal veins from L1 to L5 and two cross veins as anterior cross vein (ACV) and posterior cross vein (PCV). (B and C) When

UASDnrk was overexpressed in the wing by a wing driver ApGAL4, cross veins were missing (n=20 wings). ACV was missing in both (B) and (C). PCV was half missing in

(B) and intact in (C). (D) When UASDWnt5 was overexpressed in the wing by the

ApGAL4 driver, the wing was wild type like. (E and F) When UASDnrk and UASDWnt5 were co-overexpressed in the wing by ApGAL4, both ACV and PCV were properly restored (n=57 wings) and occasionally (F), there was an ectopic cross vein formed between L2 and L3 (arrow).

277

Table 1. Modifier screen on the loss-of-cross-vein phenotype induced by Dnrk overexpression in the wing

Candidate transgenes Modification on loss of CV in Number of wings scored Dnrk overexpression mutant UASdpp N/Aa Lb

UASfluΔarmadillo N/A L

UASmad N/A 8

UAStkv (I) N/A L

UAStkv-HA N/A L

UAStkv (III) N/A L

UASDNtkvΔGSK No modification 23

UASmed N/A L

UASnemo No modification 14

UASfrizzled2 N/A L

UASDWnt6 No modification 37

UASwg N/A L

UASDWnt5 Suppression 57

UASDWnt10 No modification 54

UASDWnt4 N/A 44

UASDWnt2 No modification 24 a N/A means that the vein formation could not be examined either because those flies were adult lethal or their wings were not properly developed. b L means that flies were adult lethal

278

Discussion

Abundant Dnrk mRNAs in the VNC versus no obvious defects in the Dnrk mutant

Previous studies showed that Dnrk mRNA was expressed at high level in the ventral nerve ganglion (Oishi et al., 1997). But I didn‟t observe any obvious defect in the Dnrk mutant VNC. The absence of defect can be explained with two possibilities. First, the resolution of our analysis may not be sufficient to detect the subtle defects. I only used the most commonly used methods to label the overall structure of the neuropil in the

VNC. The complex structure within the neuropil may prevent us from observing minor defects. It could be helpful to use single-cell labelling technique to further examine the

Dnrk mutant. The second possibility is that Dnrk may be required in other aspects of neural development and function, for example in synaptic development and transmission.

I have not done electrophysiology analysis. Further investigation in the Dnrk mutant with electrophysiology analysis and electron scanning microscopy may provide us with more details about the function of Dnrk.

Redundant function between Dnrk and Dror

From the analysis in the visual system, we found that knocking down Dror expression by

RNAi in Dnrk mutant could dramatically increase the penetrance of R-cell axon guidance defects. This suggests the redundant function between Dnrk and Dror. From the sequence similarity and the comparable expression pattern of those two genes in the embryonic nervous system, it is possible that they compensate for each other when either one is absent. In vertebrates, each ror family has two members. In mouse, it has been shown that

279

mror1 and mror2 have partially redundant function (Nomi et al., 2001). So, it is possible that Dnrk and Dror, as the two closest family members in fly, also have redundant function. The only obstacle on the way to test this hypothesis is the unavailability of a clean Dror mutant. Because Dror is located in a complex genomic locus, generating a clean mutation for Dror will be necessary. Although the RNAi knockdown experiment provided us with promising data, the only way to convincingly show the redundancy between these two genes will be to generate double null mutants for both Dnrk and Dror.

The interesting overexpression phenotype of Dnrk in the wing

The loss of CV in Dnrk overexpression mutants suggests the potential function of Dnrk outside of the nervous system. Although the Dnrk mutant did not show any obvious wing defect, it is possible that Dnrk functions redundantly with another gene in the wing. And interestingly, this loss of CV phenotype can be completely suppressed by the overexpression of DWnt5. Dnrk contains a CRD/Frizzled motif in its extracellular region.

This Frizzled motif is normally involved in mediating the binding to Wnt family members.

Thus it is possible that Dnrk can bind to DWnt5. Among all six Wnt family members I tested, I found that DWnt5 was the only one showing the genetic interaction with Dnrk.

This suggests the interaction between those two genes is specific.

DWnt5 overexpression in the wing resulted in a normal wing formation. DWnt5 mutant also has normal wings (Fradkin et al., 2004). This suggests that in the wild-type fly,

DWnt5 probably does not function in the wing development. The genetic interaction we observed between DWnt5 and Dnrk in the wing may indicate the capability of Dnrk to

280

bind to DWnt5, when those two are co-expressed in the same tissue. In flies having both genes overexpressed, DWnt5 ectopically expressed in the wing may bind to Dnrk to prevent the binding between Dnrk and its endogenous binding partner. This may induce the suppression of the Dnrk overexpression phenotype.

DWnt5 controls the embryonic commissural axon midline crossing through the Derailed receptor in the VNC (Fradkin et al., 1995; Yoshikawa et al., 2003). Dnrk mRNA is also expressed in the VNC at the same embryonic stage as the time when DWnt5 functions

(Oishi et al., 1997). The genetic interaction we observed in the wing between Dnrk and

DWnt5 suggests that those two genes may also interact in controlling axon guidance in the VNC. Future experiment in the VNC may provide more details about the function of

Dnrk.

281

Acknowledgements

We thank Yimiao Ou for helps on the dendrite analysis; Dr. Don van Meyel for advice on experiments in the embryonic VNC and adult wings; the members of the Rao lab for helpful discussions; the BDGP and Bloomington Stock Center for fly stocks; DSHB at

University of Iowa for MAb 24B10, 1D4, and BP102; Dr. Clive Wilson for the

Df(2L)41C, Df(2L)170B and partial Dror cDNA; Dr. Peter Lawrence for UAS transgenic flies for the Wnt family members; Dr. Pejmun Haghighi for UAS transgenic flies of the

TGF-β signalling pathway.

282

References

1. Afzal, A. R., et al., (2000). Recessive Robinow syndrome, allelic to dominant

brachydactyle type B, is caused by mutation of ROR2. Nat. Genet. 25, 419-22.

2. Berger, J., et al., (2008). Systematic identification of genes that regulate neuronal

wiring in the Drosophila visual system. PLoS Genet 4, e1000085.

3. Bier, E. (2000). Drawing lines in the Drosophila wing: initiation of wing vein

development. Curr Opin Genet Dev 10, 393-8.

4. Brand, A. H., Perrimon, N. (1993). Targeted gene expression as a means of altering

cell fates and generating dominant phenotypes. Development 118, 401-15.

5. Brenman, J. E., et al., (2001). Sequoia, a tramtrack-related zinc finger protein,

functions as a pan-neural regulator for dendrite and axon morphogenesis in

Drosophila. Dev Cell 1, 667-77.

6. Burgess, R. W., et al., (1999). Alternatively spliced isoforms of nerve- and muscle-

derived agrin: their roles at the neuromuscular junction. Neuron 23, 33-44.

7. Capdevila, J., Guerrero, I. (1994). Targeted expression of the signaling molecule

decapentaplegic induces pattern duplications and growth alterations in Drosophila

wings. EMBO J 13, 4459-68.

8. de Celis, J. F. (1997). Expression and function of decapentaplegic and thick veins

during the differentiation of the veins in the Drosophila wing. Development 124,

1007-18.

283

9. Dearborn, R. Jr., Kunes, S. (2004). An axon scaffold induced by retinal axons

directs glia to destinations in the Drosophila optic lobe. Development 131, 2291-303.

10. Dej, K.J., et al., (2004). Mutations in the Drosophila condensin subunit dCAP-G:

defining the role of condensin for chromosome condensation in mitosis and gene

expression in interphase. Genetics 168, 895-906

11. Ellis, M.C., et al., (1993). Expression of Drosophila glass protein and evidence for

negative regulation of its activity in non-neuronal cells by another DNA-binding

protein. Development 119, 855-65.

12. Ermakov, S., et al., (2007). Family-based association study of ROR2 poly-

morphisms with an array of radiographic hand bone strength phenotypes.

Osteoporos. Int. 18, 1683-92.

13. Fallon, J. R., et al., (1985). Acetylcholine receptor-aggregating factor is similar to

molecules concentrated at neuromuscular junctions. Nature 315, 571-4.

14. Forrester, W. C. (2002). The Ror receptor tyrosine kinase family. Cell Mol Life Sci

59, 83-96.

15. Forrester, W. C., et al., (1999). A C. elegans Ror receptor tyrosine kinase regulates

cell motility and asymmetric cell division. Nature 400, 881-5.

16. Forrester, W. C., et al., (2004). The Caenorhabditis elegans Ror RTK CAM-1

inhibits EGL-20/Wnt signaling in cell migration. Genetics 168, 1951-62.

284

17. Fradkin, L. G., et al., (1995). The Drosophila Wnt protein DWnt-3 is a secreted

glycoprotein localized on the axon tracts of the embryonic CNS. Dev Biol 168, 202-

13.

18. Fradkin, L. G., et al., (2004). The Drosophila Wnt5 protein mediates selective axon

fasciculation in the embryonic central nervous system. Dev Biol 272, 362-75.

19. Francis, M. M., et al., (2005). The Ror receptor tyrosine kinase CAM-1 is required

for ACR-16-mediated synaptic transmission at the C. elegans neuromuscular

junction. Neuron 46, 581-94.

20. Ganju, P., et al., (1995). Cloning and developmental expression of Nsk2, a novel

receptor tyrosine kinase implicated in skeletal myogenesis. Oncogene 11, 281-90.

21. Gao, F. B., et al., (1999). Genes regulating dendritic outgrowth, branching, and

routing in Drosophila. Genes Dev 13, 2549-61.

22. Garrity, P. A., et al., (1999). Retinal axon target selection in Drosophila is regulated

by a receptor protein tyrosine phosphatase. Neuron 22, 707-17.

23. Gautam, M., et al., (1996). Defective neuromuscular synaptogenesis in agrin-

deficient mutant mice. Cell 85, 525-35.

24. Glass, D. J., et al., (1996). Agrin acts via a MuSK receptor complex. Cell 85, 513-23.

25. Goberdhan, D. C., et al., (1999). Drosophila tumor suppressor PTEN controls cell

size and number by antagonizing the Chico/PI3-kinase signaling pathway. Genes

Dev 13, 3244-58.

285

26. Green, J. L., et al., (2007). The C. elegans ROR receptor tyrosine kinase, CAM-1,

non-autonomously inhibits the Wnt pathway. Development 134, 4053-62.

27. Green, J. L., et al., (2008). Opposing Wnt pathways orient cell polarity during

organogenesis. Cell 134, 646-56.

28. Green, J.L., et al., (2008). Ror receptor tyrosine kinases: orphans no more. Trends

Cell Biol 18, 536-44.

29. Grueber, W. B., et al., (2002). Tiling of the Drosophila epidermis by multidendritic

sensory neurons. Development 129, 2867-78.

30. Haerry, T. E., et al., (1998). Synergistic signaling by two BMP ligands through the

SAX and TKV receptors controls wing growth and patterning in Drosophila.

Development 125, 3977-87.

31. Hayashi, Y., et al., (2009). A trophic role for Wnt-Ror kinase signaling during

developmental pruning in Caenorhabditis elegans. Nat Neurosci 12, 981-7.

32. Hikasa, H., et al., (2002). The Xenopus receptor tyrosine kinase Xror2 modulates

morphogenetic movements of the axial mesoderm and neuroectoderm via Wnt

signaling. Development 129, 5227-39.

33. Huang, Z., Kunes, S. (1998). Signals transmitted along retinal axons in Drosophila:

Hedgehog signal reception and the cell circuitry of lamina cartridge assembly.

Development 125, 3753-64.

34. Hubbard, S. R., Till, J. H. (2000). Protein tyrosine kinase structure and function.

Annu Rev Biochem 69, 373-98.

286

35. Hummel, T., Klämbt, C. (2008). P-element mutagenesis. Methods Mol Biol 420, 97-

117.

36. Jager, H., et al., (2005). The Drosophila melanogaster condensin subunit Cap-G

interacts with the centromere-specific histone H3 variant CID. Chromosoma 113,

350-361.

37. Jennings, C. G., et al., (1993). Muscle-specific trk-related receptor with a kringle

domain defines a distinct class of receptor tyrosine kinases. Proc Natl Acad Sci U S

A. 90, 2895-9.

38. Jing, L., et al., (2009). Wnt signals organize synaptic prepattern and axon guidance

through the zebrafish unplugged/MuSK receptor. Neuron 61, 721-33.

39. Kaiser, K. (1990). From gene to phenotype in Drosophila and other organisms.

Bioessays 12, 297-301.

40. Kim, C. H., et al., (2003). Regulation of MuSK expression by a novel signaling

pathway. J Biol Chem 278, 38522-7.

41. Kim, N., Burden, S. J. (2008). MuSK controls where motor axons grow and form

synapses. Nat Neurosci 11, 19-27.

42. Koga, M., et al., (1999). Control of DAF-7 TGF-(alpha) expression and neuronal

process development by a receptor tyrosine kinase KIN-8 in Caenorhabditis elegans.

Development 126, 5387-98.

43. Kramer, A. P., Kuwada, J. Y. (1983). Formation of the receptive fields of leech

mechanosensory neurons during embryonic development. J Neurosci 3, 2474-86.

287

44. Lacazette, E., et al., (2003). A novel pathway for MuSK to induce key genes in

neuromuscular synapse formation. J Cell Biol 161, 727-36.

45. Landgraf, M., et al., (1997). The origin, location, and projections of the embryonic

abdominal motorneurons of Drosophila. J Neurosci 17, 9642-55.

46. Lee, Y. S., Carthew, R. W. (2003). Making a better RNAi vector for Drosophila:

use of intron spacers. Methods 30, 322-9.

47. Linnoila, J., et al., (2008). A mammalian homolog of Drosophila tumorous imaginal

discs, Tid1, mediates agrin signaling at the neuromuscular junction. Neuron 60, 625-

41.

48. Llimargas, M., Lawrence, P. A. (2001). Seven Wnt homologues in Drosophila: a

case study of the developing tracheae. Proc Natl Acad Sci U S A 98, 14487-92.

49. Long, H., et al., (2006). De novo GMP synthesis is required for axon guidance in

Drosophila. Genetics 172, 1633-42.

50. MacKeigan, J. P., et al., (2005). Sensitized RNAi screen of human kinases and

phosphatases identifies new regulators of apoptosis and chemoresistance. Nat Cell

Biol 7, 591-600.

51. Masiakowski, P., Carroll, R. D. (1992). A novel family of cell surface receptors with

tyrosine kinase-like domain. J Biol Chem 267, 26181-90.

52. Milán, M., Cohen, S. M. (1999). Notch signaling is not sufficient to define the

affinity boundary between dorsal and ventral compartments. Mol Cell 4, 1073-8.

288

53. Mismer, D., Rubin, G. M. (1987). Analysis of the promoter of the ninaE opsin gene

in Drosophila melanogaster. Genetics 116, 565-78.

54. Moses, K., Rubin, G. M. (1991). Glass encodes a site-specific DNA-binding protein

that is regulated in response to positional signals in the developing Drosophila eye.

Genes Dev 5, 583-93.

55. Muffat, J., et al., (2008). Human ApoD, an apolipoprotein up-regulated in

neurodegenerative diseases, extends lifespan and increases stress resistance in

Drosophila. Proc. Natl. Acad. Sci. U.S.A. 105, 7088-93.

56. Nitkin, R. M., et al., (1987). Identification of agrin, a synaptic organizing protein

from Torpedo electric organ. J Cell Biol 105, 2471-8.

57. Nomachi, A., et al., (2008). Receptor tyrosine kinase Ror2 mediates Wnt5a-induced

polarized cell migration by activating c-Jun N-terminal kinase via actin-binding

protein filamin A. J Biol Chem 283, 27973-81.

58. Nomi, M., et al., (2001). Loss of mRor1 enhances the heart and skeletal

abnormalities in mRor2-deficient mice: redundant and pleiotropic functions of

mRor1 and mRor2 receptor tyrosine kinases. Mol Cell Biol 21, 8329-35.

59. Oishi, I., et al., (1997). A novel Drosophila receptor tyrosine kinase expressed

specifically in the nervous system. Unique structural features and implication in

developmental signaling. J Biol Chem 272, 11916-23.

60. Oishi, I., et al., (2003). The receptor tyrosine kinase Ror2 is involved in non-

canonical Wnt5a/JNK signalling pathway. Genes Cells 8, 645-54.

289

61. Oishi, I., et al., (1999). Spatio-temporally regulated expression of receptor tyrosine

kinases, mRor1, mRor2, during mouse development: implications in development

and function of the nervous system. Genes Cells 4, 41-56.

62. O'Keefe, D. D., et al., (1998). Function and specificity of LIM domains in

Drosophila nervous system and wing development. Development 125, 3915-23.

63. Oldridge, M., et al., (2000). Dominant mutations in ROR2, encoding an orphan

receptor tyrosine kinase, cause brachydactyly type B. Nat. Genet. 24, 275-78.

64. Paganoni, S., Ferreira, A. (2003). Expression and subcellular localization of Ror

tyrosine kinase receptors are developmentally regulated in cultured hippocampal

neurons. J Neurosci Res 73, 429-40.

65. Paganoni, S., Ferreira, A. (2005). Neurite extension in central neurons: a novel role

for the receptor tyrosine kinases Ror1 and Ror2. J Cell Sci 118, 433-46.

66. Parrish, J. Z., et al., (2007). Mechanisms that regulate establishment, maintenance,

and remodeling of dendritic fields. Annu Rev Neurosci 30, 399-423.

67. Perez, S. E., Steller, H. (1996). Migration of glial cells into retinal axon target field

in Drosophila melanogaster. J Neurobiol 30, 359-73.

68. Poeck, B., et al., (2001). Glial cells mediate target layer selection of retinal axons in

the developing visual system of Drosophila. Neuron 29, 99-113.

69. Reist, N. E., et al., (1987). Agrin-like molecules at synaptic sites in normal,

denervated, and damaged skeletal muscles. J Cell Biol 105, 2457-69.

290

70. Reist, N. E., et al., (1992). Agrin released by motor neurons induces the aggregation

of acetylcholine receptors at neuromuscular junctions. Neuron 8, 865-8.

71. Renn, S. C. P., et al., (1998). Characterization and cloning of tripeptidyl peptidase II

from the fruit fly, Drosophila melanogaster. J. Biol. Chem 273, 19173-82.

72. Rørth, P., et al., (1998). Systematic gain-of-function genetics in Drosophila.

Development 125, 1049-57.

73. Ruan, W., et al., (1999). The SH2/SH3 adaptor protein dock interacts with the

Ste20-like kinase misshapen in controlling growth cone motility. Neuron 24, 595-

605.

74. Ryder, E., Russell, S. (2003). Transposable elements as tools for genomics and

genetics in Drosophila. Brief Funct Genomic Proteomic 2, 57-71.

75. Sanchez, D., et al., (2000). Characterization of two novel lipocalins expressed in the

Drosophila embryonic nervous system, Int. J. Dev. Biol 44, 349-59.

76. Sanchez, D., et al., (2006). Loss of glial lazarillo, a homolog of apolipoprotein D,

reduces lifespan and stress resistance in Drosophila. Curr. Biol 16, 680-86.

77. Sanes, J. R., Lichtman, J. W. (2001). Induction, assembly, maturation and

maintenance of a postsynaptic apparatus. Nat Rev Neurosci 2, 791-805.

78. Schambony, A., Wedlich, D. (2007). Wnt-5A/Ror2 regulate expression of XPAPC

through an alternative noncanonical signaling pathway. Dev Cell 12, 779-92.

79. Schlessinger, J. (2000). Cell signaling by receptor tyrosine kinases. Cell 103, 211-25.

291

80. Schmid, A., et al., (1999). Clonal analysis of Drosophila embryonic neuroblasts:

neural cell types, axon projections and muscle targets. Development 126, 4653-89.

81. Schuster, C. M., et al., (1996). Genetic dissection of structural and functional

components of synaptic plasticity. I. Fasciclin II controls synaptic stabilization and

growth. Neuron 17, 641-54.

82. Schwabe, G. C., et al., (2000). Distinct mutations in the receptor tyrosine kinase

gene ROR2 cause brachydactyly type B. Am. J. Hum. Genet. 67, 822-31.

83. Seeger, M., et al., (1993). Mutations affecting growth cone guidance in Drosophila:

genes necessary for guidance toward or away from the midline. Neuron 10, 409-26.

84. Seifert, J. R., Mlodzik, M. (2007). Frizzled/PCP signalling: a conserved mechanism

regulating cell polarity and directed motility. Nat Rev Genet 8, 126-38.

85. Serpe, M., et al., (2005). Matching catalytic activity to developmental function:

tolloid-related processes Sog in order to help specify the posterior crossvein in the

Drosophila wing. Development 132, 2645-56.

86. Shimmi, O., et al., (2005). The crossveinless gene encodes a new member of the

Twisted gastrulation family of BMP-binding proteins which, with Short gastrulation,

promotes BMP signaling in the crossveins of the Drosophila wing. Dev Biol 282,

70-83.

87. Singer, M. A., et al., (1997). Signaling through both type I DPP receptors is required

for anterior-posterior patterning of the entire Drosophila wing. Development 124,

79-89.

292

88. Sossin, W. S. (2006). Tracing the evolution and function of the Trk superfamily of

receptor tyrosine kinases. Brain Behav Evol 68, 145-56.

89. Strochlic, L., et al., (2001). MAGI-1c: a synaptic MAGUK interacting with MuSK

at the vertebrate neuromuscular junction. J Cell Biol 153, 1127-32.

90. Strochlic, L., et al., (2004). 14-3-3 gamma associates with muscle specific kinase

and regulates synaptic gene transcription at vertebrate neuromuscular synapse. Proc

Natl Acad Sci U S A 101, 18189-94.

91. Sugimura, K., et al., (2004). Development of morphological diversity of dendrites in

Drosophila by the BTB-zinc finger protein abrupt. Neuron 43, 809-22.

92. Suh, G. S., et al., (2002). Drosophila JAB1/CSN5 acts in photoreceptor cells to

induce glial cells. Neuron 33, 35-46.

93. Strochlic, L., et al., (2005). The synaptic muscle-specific kinase (MuSK) complex:

new partners, new functions. Bioessays 27, 1129-35.

94. Ting, C. Y., et al., (2007). Tiling of R7 axons in the Drosophila visual system is

mediated both by transduction of an activin signal to the nucleus and by mutual

repulsion. Neuron 56, 793-806.

95. Vactor, D. V., et al., (1993). Genes that control neuromuscular specificity in

Drosophila. Cell 73, 1137-53.

96. Valenzuela, D. M., et al., (1995). Receptor tyrosine kinase specific for the skeletal

muscle lineage: expression in embryonic muscle, at the neuromuscular junction, and

after injury. Neuron 15, 573-84.

293

97. van Bokhoven, H., et al., (2000). Mutation of the gene encoding the ROR2 tyrosine

kinase causes autosomal recessive Robinow syndrome. Nat. Genet. 25, 423-26.

98. van der Geer, P., et al., (1994). Receptor protein-tyrosine kinases and their signal

transduction pathways. Annu Rev Cell Biol 10, 251-337.

99. Van Vactor, D. Jr., et al., (1988). Analysis of mutants in chaoptin, a photoreceptor

cell-specific glycoprotein in Drosophila, reveals its role in cellular morphogenesis.

Cell 52, 281-90.

100. Walker, D. W., et al., (2006). Overexpression of a Drosophila homolog of

apolipoprotein D leads to increased stress resistance and extended lifespan. Curr.

Biol 16, 674-679.

101. Wernet, M. F., et al., (2003). Homothorax switches function of Drosophila

photoreceptors from color to polarized light sensors. Cell 115, 267-79.

102. Wilson, C., et al., (1993). Dror, a potential neurotrophic receptor gene, encodes a

Drosophila homolog of the vertebrate Ror family of Trk-related receptor tyrosine

kinases. Proc Natl Acad Sci U S A 90, 7109-13.

103. Yamamoto, S., et al., (2008). Cthrc1 selectively activates the planar cell polarity

pathway of Wnt signaling by stabilizing the Wnt-receptor complex. Dev Cell 15, 23-

36.

104. Yoshikawa, S., et al., (2003). Wnt-mediated axon guidance via the Drosophila

Derailed receptor. Nature 422, 583-8.

294

105. Zhang, J., et al., (2004). Zebrafish unplugged reveals a role for muscle-specific

kinase homologs in axonal pathway choice. Nat Neurosci 7, 1303-9.

106. Zhou, H., et al., (1999). Distinct domains of MuSK mediate its abilities to induce

and to associate with postsynaptic specializations. J Cell Biol 146, 1133-46.

107. Zinovyeva, A. Y., et al., (2008). Complex network of Wnt signaling regulates

neuronal migrations during Caenorhabditis elegans development. Genetics 179,

1357-71.

295

Chapter 5

General discussion

296

In this thesis, I described general mechanisms underlying axon guidance, target selection, and dendrite pattern formation during the development of the nervous system. I also discussed the structure and development of two systems in fruit fly, the visual system and the PNS dendritic arborization neuron (da) system. Both are excellent model systems for dissecting molecular mechanisms involved in regulating axon guidance and target selection, and dendritic growth and patterning. After the introduction, I documented our interesting findings about the function of four genes. The gene bur, encoding a GMP synthetase involved in the de novo GMP synthesis, is specifically required for R-cell axon guidance in the fly visual system. The gene tutl, encoding an IgSF superfamily protein, controls dendrite branching and self-avoidance in the fly PNS. Two members of the fly ror family, Dnrk and Dror, have potential redundant functions in controlling R-cell axon guidance. And Dnrk can genetically interact with DWnt5 for vein formation in the wing.

My work on those genes adds new information to our knowledge about axon guidance and dendritic arborization. They also present some questions and suggestions, which need to be addressed in future studies.

297

5.1 The involvement of de novo GMP synthesis in R cell axon guidance

5.1.1 de novo GMP synthesis versus the salvage synthesis pathway

I addressed the importance of Bur, as a GMP synthetase in de novo GMP synthesis, in regulating the R-cell axon guidance. This indicates the specific requirement for de novo

GMP synthesis versus the salvage GMP synthesis in specific systems at specific developmental stages. The de novo GMP synthesis involves a number of reactions which assemble the guanine ring from precursors including amino acids, carbon dioxide and tetrahydrofolate derivative (Zalkin and Dixon 1992). The salvage pathway simply synthesizes the guanine ring by utilizing available free guanine bases generated from degeneration of nucleic acids and nucleotides (Zollner 1982). Compared to de novo GMP synthesis, the salvage pathway costs much less energy and material. The salvage pathway is commonly believed to be the principle supply for GMP in organisms.

Then, why do organisms need the de novo pathway? De novo GMP synthesis has been shown to specifically function in the proliferation of leukocytes (Eugui et al., 1991;

Dayton et al., 1992, Hauser and Sterzal 1999; Gu et al., 2000, 2003). Although the exact reason for the requirement of the de novo pathway is still not clear, those results at least indicate that in certain processes the de novo pathway is non-replaceable by the salvage pathway. The action of Bur in R-cell axon guidance is consistent with this view. In the food depletion analysis, we found that without guanine intake from food, flies could still survive and develop normal axon projection patterns. In this experiment, the guanine intake from the food was cut off from the very beginning. Flies had to synthesize GMP through the de novo pathway from the beginning of development. Their normal

298

development suggests that the de novo pathway could produce enough GMP for developmental processes. The lethality and abnormal neural network formation in bur and ras mutant flies suggests that the de novo pathway is essential for viability and axon guidance.

5.1.2 How GMP level is controlled within the cell?

It is surprising that GMP synthesis by the de novo pathway is specifically required for R- cell axon guidance but not for R-cell differentiation and fate determination. One likely explanation is that the level of GMP produced by the salvage pathway is sufficient for some processes but not for the others (e.g. axon guidance).

But the product of both the de novo and salvage pathways is GMP. As small diffusible molecules, how GMPs within the cell are distinguished and controlled based on their different origin for applications in different processes? For example, how can GMP generated by the salvage pathway in R cells be utilized in cell differentiation, but not in axon guidance? I speculate that this is probably controlled by the compartment or microdomain structures within the cell.

More than twenty years ago, people already discovered that some cellular signalling pathway like the cAMP pathway is specially restricted to only some regions of the cell

(Brunton et al., 1981). After the first discovery, live-cell imaging enables the visualization of microdomains for a range of signalling molecules including cAMP (Bacskai et al.,

1993; Nikolaev et al., 2004; Zaccolo and Pozzan, 2002), Ca2+ (Cancela et al., 2002;

299

Llinas et al., 1992; Marsault et al., 1997), GTPases (Janetopoulos et al., 2001; Mochizuki et al., 2001), and protein kinases (Nagai et al., 2000; Ting et al, 2001; Wang et al., 2005;

Neves et al., 2008). So, likely, within the neuron, GMP is also confined to some specific microdomains. The local level of GMP in a subcellular compartment may be determined by the type of synthesis machinery within this microdomain, or specific transportation mechanism, or both.

A neuron, as a highly polarized cell, has three different compartments, as the axon, soma, and dendrite. For some sensory neurons such as R cells, they have only axon and soma. It has been shown that in different subcellular components, there are different distribution profiles for proteins and cellular organelles (Dotti and Poo, 2003; Horton and Ehlers,

2003). Signalling pathways have been identified for the initiation of polarity (Conde and

Caceres, 2009; Arimuta and Kaibuchi, 2007). It is also reported that there is a diffusion barrier at the axon hillock to prevent the diffusion of polarized axon membrane protein, which is required for the maintenance of neuronal polarity (Winckler et al., 1999).

Recently, some cytoplasmic signalling molecule such as cAMP and PKA were found to be predominantly localized to the dendrites of neurons (Neves et al., 2008). I speculate that GMP and its descendant GTP and cGMP, may also have a polarized distribution pattern in neurons. The de novo GMP synthesis machinery may be localized to R-cell axons, thus controling the level of GMP within R-cell growth cones for directed movement. While within the cell body, GMP synthesized by the salvage pathway may be restricted to specific microdomains. Once the de novo pathway is impaired, R-cell axons

300

display guidance defects because GMP produced by the salvage pathway is restricted to the cell body and thus cannot travel freely into the axon to regulate axon guidance. This hypothesis may be tested in future studies by examining the distribution of Bur and other components of the de novo pathway.

5.1.3 Signaling events of GMP

GMP can be converted into GTP and cGMP. Both molecules are important in regulating axon guidance. GTP level is essential for the activity of Rho-like small GTPases. It can also regulate protein synthesis (Pogson, 1974). Rho family small GTPases are downstream targets of many guidance receptors in axon guidance (Luo et al., 1997, 2002;

Dickson, 2001). In our analysis, we observed the genetic interaction between bur and Rac in R-cell axon guidance, suggesting that the de novo GMP synthesis may control R-cell axon guidance through regulating the activity of Rac.

cGMP is involved in regulating multiple cellular processes. It can regulate ion channels and modulate the activity of PKG for downstream signalling. In cultured neurons, activation of cGMP could convert the growth cone responses to Sema3 from repulsion to attraction (Song et al., 1998). The ratio of cAMP/cGMP can determine whether Netrin mediates an attractive or repulsive response for guiding axons (Nishiyama et al., 2003). It is also reported that the polarized distribution of soluble Guanylate cyclase (SGC) in the apical dendrite of pyramidal neurons controls the attractive response of dendrites to

Sema3A (Polleux et al., 2000). This response was mediated by PKG. Those results

301

suggest that SGC, cGMP, and PKG are targeted to specific compartments of the neuron for specific guidance response.

In fly, both cAMP and cGMP are involved in motor axon guidance, and both modulate

PlexA signaling in response to Sema-1a in the neuromuscular system. Fly Nervy, a PKA anchoring protein, specifically functions in motor axon guidance downstream of PlexA

(Jonathan et al., 2004). As an AKAP, Nervy controls site-specific localization of PKA, which can be activated by cAMP to regulate axon guidance. Fly receptor guanylyl cyclase

Gyc76C is also specifically required in the Sema-1a signalling pathway for motor axon guidance (Joseph et al., 2004). Gyc76C functions in catalyzing cGMP production. Its catalytic domain for cGMP synthesis is indispensible for its function in motor axon guidance. This suggests that cGMP level is important for motor axon guidance in fly.

This finding is consistent with the discovery in mammalian cells that the cGMP level is critical for growth-cone response to Sema3 (Song et al., 1998).

In the fly visual system, cGMP and SGC are also involved in mediating R-cell axon projection (Gibbs and Truman, 1998; Gibbs et al., 2001; Gibbs, 2001). One attractive model for the action of the de novo pathway is that the synthesis of GMP by this pathway increases the level of GMP within R-cell growth cones. This increase is essential for the local production of cGMP, which determines the response of R-cell growth cones to guidance cues.

5.1.4 bur also functions in motor axon guidance

302

My results show that bur and the de novo GMP synthesis is required for R-cell axon guidance in the fly visual system. De novo GMP synthesis may be exclusively required for R-cell axon guidance. Or alternatively, it may function more broadly in axon guidance in other systems. To address this, I decided to examine motor axon guidance in the fly embryonic neuromuscular system.

In wild-type embryos, motor axons display stereotyped guidance and targeting pattern

(see chapter IV). In bur mutant embryos, the segment nerves a (SNa) and intersegment nerves b (ISNb) displayed guidance defects in around 30-40% flies examined. The mutant phenotype included axon stalling and bypassing. This result suggests that the de novo

GMP synthesis is likely broadly required in different types of neurons for regulating axon guidance in fly.

5.1.5 Defasciculation, a requirement for layer-specific R1-R6 targeting?

In the analysis of bur mutants, I noticed that although R-cell axons displayed a severe hyperfasciculation phenotype, the ganglion specific targeting of R1-R6 axons was still normal. I had expected that such strong adhesion between those axons would cause some

R1-R6 axons to mistarget into the medulla because they could not defasciculate properly from the pioneer R8 axon. However, despite the formation of large bundles in bur mutants, R1-R6 axons could still detect and respond correctly to the stop signal provided by the lamina glia. My observation thus argues against the hypothesis that the failure of defasciculation of R1-R6 axons from R8 axons was the cause of R1-R6 mistargeting

303

phenotype in the PTP69D mutant suggested by Dickson and colleagues (Newsome et al.,

2000).

5.1.6 Is the de novo GMP synthesis required for neural network formation in mammals?

The human genome contains a single GMP synthetase gene (Hirst et al. 1994), which exhibits extensive homology to the fly Bur (Long et al., 2006). It is highly possible that the de novo GMP synthesis mediated by this enzyme is also required for neural network formation during the development of the human brain. Interestingly, it was reported that mutations in the gene encoding for adenylosuccinate lyase (ADSL), which is required for the de novo synthesis of both AMP and GMP (Zalkin and Dixon 1992), cause severe mental retardation and autism (Jaeken and Van den Berghe 1984). The mechanism underlying the ADSL deficiency-induced mental retardation and autism is still unclear.

Based on the axon guidance phenotype we observed in fly bur mutant, it will be interesting to examine if knocking out GMP synthetase gene in a mouse model would cause defects in neural network formation. It will also be interesting to determine if some other human neurological diseases are caused by mutations in the GMP synthetase gene or genes encoding for other enzymes of the de novo GMP synthesis pathway.

5.2 The function of Turtle in dendrite branching and patterning

5.2.1 Domain requirement and the possible homotypic function of Tutl

Our work on the gene tutl shows that tutl is required for dendrite patterning in a class- specific fashion (Long et al., 2009). In class I da neurons, Tutl controls dendrite

304

branching cell-autonomously by inhibiting excessive branch formation in the distal dendrite region. In class IV da neurons, Tutl is required for dendrite self-avoidance in a cell-autonomous manner (Long et al., 2009). However, the mechanism by which Tutl regulates those processes is still unclear. Tutl is an IgSF protein with multiple Ig domains and Fn-III motifs in its extracellular region. Ig domain and Fn-III motif have been widely shown to be involved in regulating protein-protein interaction. Some identified proteins containing those domains function in a homotypic fashion, such as Dscam (Wojtowicz et al., 2004, 2007). Some other proteins with those domains function in a heterotypic way, such as Robo and Frazzled (Kidd et al., 1998; Kolodziej et al., 1996). One important question is whether Tutl mediates homophilic binding, or it functions as a receptor or a ligand to bind to a heterophilic partner.

We have carried out experiments to gain insights into the mechanism of Tutl action. In the domain requirement analysis, I generated three mutated forms of tutl, including tutlEx, tutlΔcyto, and tutlΔEx. I generated transgenic flies and assessed their capability to rescue tutl mutant flies. tutlΔcyto was the only one showing rescue capability (Long et al., unpublished data). It could partially rescue the tutl mutant lethality and completely rescue the class I da neuron branching phenotype. This rescue data suggests that the cytoplasmic domain of Tutl is dispensable for its function in regulating class I da neuron dendrite branching, but necessary for fly viability. So, probably, for different functions of Tutl in different systems, the requirement for its cytoplasmic domain is also different. However, this result does not allow us to conclude whether Tutl functions as a ligand or a receptor for other proteins. 305

To test the possibility that Tutl functions in a homotypic manner, we transfect S2 cells with tutlΔcyto transgene. In the cell aggregation assay, we found that the expression of tutlΔcyto could induce homophilic cell adhesion (Ferguson et al., 2009). This suggests that Tutl can recognize and bind to Tutl on opposing cell surface. I also generated tagged secreted Tutl protein probes (Long, unpublished data). Staining of dissected VNC of fly embryos with those probes showed specific binding pattern. Because the staining did not completely disappear in tutl mutant, it suggests that Tutl might also bind to some other proteins. One working model for the action of Tutl is that it mediates dendrite self- avoidance in a homophilic binding manner to induce repulsion. Tutl may bind to other proteins in different systems.

5.2.2 The dendrite patterning phenotype in class I da neurons

In addition to the branching defects in class I da neurons, we also found a dendrite branch patterning abnormality in tutl mutants. Wild-type class I neuron ddaE has a comb-like structure with the long major branches project posteriorly in a parallel fashion. Loss of tutl disrupted this pattern. The phenotype was also caused by tutl mutation as the full length tutl could largely rescue this phenotype. We speculate that Tutl also functions in controlling general dendrite patterning in addition to its role in dendrite branching in

Class I neurons. The tutlΔcyto couldn‟t rescue the patterning defect although it could fully rescue the branching phenotype. This suggests that the cytoplasmic tail of Tutl is required for its function in branch patterning. So, even within the same neuron, Tutl may have different functions in controlling different aspects of dendrite development.

306

5.2.3 Functional relationship between tutl and other known genes in dendrite development

In our analysis, we wanted to know if tutl interacted with other molecules controlling dendrite branching and self-avoidance. This would help us to place tutl into known signalling pathways in dendrite development. Based on phenotype similarity, we tested several candidates including transcription factors, cytoplasmic proteins, and cell membrane proteins (For details, please see the genetic interaction analysis of tutl in the chapter 3). We examined the potential links between those genes and tutl by either immunostaining or genetic interaction analysis. Our results showed that tutl did not interact with any of those candidate genes. This suggests that Tutl may function in a novel pathway to control dendrite branching and self-avoidance.

5.2.4 Potential role of tutl in the VNC?

Our immunohistological analysis revealed specific Tutl staining pattern in the PNS da neurons. We also discovered the expression of tutl in other regions in the fly nervous system, including the visual system and the VNC. I will discuss the function of Tutl in the visual system in the following section (Ferguson et al., 2009). Within the VNC neuropil, tutl shows strong expression on commissural axon connectives and lower expression on the longitudinal axon connectives (Long, unpublished data).This is consistent with previous work showing that tutl mRNA was expressed in the VNC (Bodily et al., 2001).

Interestingly, by using Tutl protein probes to label the embryonic VNC, we also observed specific binding patterns (Long, unpublished data).

307

Those expression and binding patterns of tutl suggest strongly that Tutl is involved in regulating the development of the VNC. The high level of tutl expression on the commissural axons suggests that Tutl functions to regulate correct midline crossing in the

VNC. Future work will be needed to determine the exact role of Tutl in midline crossing.

5.2.5 Tutl functions in R7 axon tiling in the fly visual system

Our lab also discovered the function of Tutl in controlling R7 axon tiling in the fly visual system (Ferguson et al., 2009). Tutl was specifically expressed in R7 cells and the medulla target region during the third-instar larval and pupal stages. In the wild-type medulla region, neurons and their connections are organized into regular-spaced columnar structures. Same-type axon terminals in adjacent columns avoid each other.

When tutl was removed from R7 axons, R7 axonal terminals extend laterally to invade neighbouring columns. Molecular and genetic analysis support that Tutl mediates mutual repulsion between adjacent R7 terminals in a homophilic manner. Thus, the action of Tutl in R7 axonal tiling appears similar to its role in mediating dendrite self-avoidance.

5.2.6 The function of Tutl in fly behavior

Previous work shows that tutl mutant larva displayed defects in coordinated behavior.

When tutl mutant larvae were turned over, they had difficulty to roll themselves over to the normal position (Bodily et al., 2001). I observed similar defects in all tutl mutants analyzed. Interestingly, class I da neurons and bipolar cells have been shown to control the rhythmic locomotion behaviour in fly larvae (Hughes and Thomas, 2007; Song et al.,

2007). Because tutl mutant displayed defects in class I da neuron dendrite branching and

308

patterning, I speculate that the behavior defect in tutl mutant larvae was caused by defects in dendrite patterning in the PNS.

In addition to rolling-over defects, I observed another type of behavioral defects, which we named as decision-making defects. When wild-type larvae are subject to a tactile stimulus by a brush, they display escaping behavior. This behavior includes ceasing movement, pausing for a second, and turning away from the stimulus. This escaping behavior is initiated by the mechanosensory input (Kernan et al., 1994). In response to stimulus, tutl mutants were able to stop moving and pause. However, instead of turning away to escape, they frequently followed the original path. This result suggests that the mechanosensation is normal in tutl mutants, however, their ability to choose a correct path was affected.

This decision-making defect could be completely rescued by neuronal-specific expression of tutl. Jan and colleagues identified a neuronal circuit that controls the selection of egg- laying sites by female flies, which is another type of decision making (Yang et al., 2009).

It will be of interest to elucidate the nature of Tutl-dependent neuronal circuit that controls the escape decision-making behavior.

5.2.7 The function of the mammalian homolog of Tutl

The analysis on Dasm-1, the mammalian homolog of Tutl, showed controversial results

(Shi et al., 2004a and b; Mishra et al., 2008). In cultured hippocampal neurons, Dasm-1 was selectively localized to dendrites (Shi et al., 2004a). Reducing Dasm-1 expression by

309

RNAi or expressing a dominant negative form of Dasm-1, could inhibit dendrite outgrowth and excitatory synapse maturation (Shi et al., 2004a and 2004b). Based on those results, researchers hypothesized that Dasm-1 was specifically required for dendrite growth and synapse formation. But a recent study shows that when Dasm-1 was knocked out in mouse, dendrite growth and patterning in hippocampal neurons occurred normally

(Mishra et al., 2008). And previous results from RNAi experiments were suggested to be due to the off-target effect (Mishra et al., 2008). Thus, the function of Dasm-1 in the mammalian nervous system is still not clear.

One way to further address the function of Dasm-1 in mammals is to generate double mutants that delete both Dasm-1 and its close homolog IgSF9B in mice. The lack of dendrite phenotype in Dasm-1 mouse may be due to the functional redundancy between

Dasm-1 and IgSF9B. Additionally, it will be informative to examine axonal and dendritic patterns in other brain region, for example, the retina, in Dasm-1 knock-out mice. This will allow the detailed analysis of the potential role of Dasm-1 in axonal tiling and dendrite self-avoidance.

5.3 The function of Dnrk in fly

5.3.1 The potential redundancy between Dnrk and Dror

In the Dnrk mutant fly, I did not observe obvious morphological defects in the nervous system except a low penetrance R-cell axon guidance defects in the visual system. Using the RNAi technique to specifically knock down Dror either in the visual system or in all neurons in the Dnrk mutant background, I observed more severe R-cell axon guidance

310

defects with much higher penetrance. Those data suggest that Dnrk and Dror function redundantly with each other in the visual system.

Redundancy between homologous genes is frequently observed in fly and other model organisms. It has been estimated that over 2/3 of total fly genes show no obvious loss-of- function phenotype when mutated (Miklos and Rubin, 1996; Rørth et al., 1998), likely due to the functional redundancy. There are significant similarities in the embryonic expression patterns and protein structures between Dnrk and Dror. Those similarities suggest that as the two members of the same family, Dnrk and Dror may share similar function in the fly nervous system. To confirm the redundancy between Dnrk and Dror, future work needs to be done to generate and characterize Dnrk and Dror double mutants.

In addition, examining the overexpression phenotype of Dror in other systems may help us to identify the non-redundant function of Dnrk and Dror. The primary sequences of

Dror and Dnrk do show some differences. Based on the sequence alignment with some known protein domains, it was suggested that Dnrk is more related to the MuSK family, while Dror is more related to the Ror family (Sossin, 2006). Thus, it is possible that in some systems, Dnrk and Dror may have their own non-redundant function.

5.3.2 The potential function of Dnrk in the VNC

A role for Dnrk in the VNC was suggested by the high level of Dnrk mRNA expression in the embryonic ventral neuropil (Oishi et al., 1997). However in my study, when Dnrk was removed, the axon connective formation within the VNC was normal. There are three

311

possible explanations for this normal phenotype. The first possibility is the potential functional redundancy between Dnrk and Dror as I discuss in the previous section. The second possibility is that the resolution of my analysis was not high enough to reveal subtle defects within the complex structure of the VNC. And the third possibility is that

Dnrk may be required for some other aspects of neural development and function. For example, Dnrk may function in synaptic development and transmission. Further investigation of the Dnrk mutant with electrophysiological analysis and electron scanning microscopy will help us to gain insights into the exact function of Dnrk.

5.3.3 Genetic interaction between Dnrk and DWnt5

DWnt5 could specifically modify the overexpression phenotype of Dnrk in the wing. This suggests that those two genes may also interact in other systems. In some species, members of the Ror and MuSK families have been found to function as the receptors for the Wnt family ligands. For example in Xenopus larvae, the signalling pathway mediating through Wnt5a-Ror2 controls the convergent extension (Hikasa et al., 2002; Oishi et al.,

2003; Schambony et al., 2007). In C. elegans, Ror protein Cam-1functions as the receptor of Wnt to control the epithelia cell orientation polarity (Green et al., 2007 and 2008). In zebrafish, MuSK-like protein Unplugged functions as the receptor for Wnt ligand to control the synaptic prepattern and axon guidance (Jing et al., 2009). I speculate that Dnrk, as a protein listed in both Ror and MuSK families, it may also function as a receptor to bind Wnt ligands.

312

It is reported that DWnt5 controls commissural axon midline crossing at the embryonic stage (Fradkin et al., 1995; Yoshikawa et al., 2003). Consistently, Dnrk also shows high expression level in the VNC at the same stage (Oishi et al., 1997). It will be of interest to examine commissural axon connective formation in Dnrk mutant. Future biochemical analysis will help to determine if DWnt5 binds directly to Dnrk.

313

References

1. Abrell, S., Jäckle, H. (2001). Axon guidance of Drosophila SNb motoneurons depends on the cooperative action of muscular Krüppel and neuronal capricious activities. Mech Dev 109, 3-12.

2. Adams, J. C., Tucker, R. P. (2000). The thrombospondin type 1 repeat (TSR) superfamily: diverse proteins with related roles in neuronal development. Dev Dyn 218, 280-99.

3. Altman, J., Anderson, W. J. (1972). Experimental reorganization of the cerebellar cortex. I. Morphological effects of elimination of all microneurons with prolonged x- irradiation started at birth. J Comp Neurol 146, 355-406.

4. Andrews, H. K., et al., (2009). Sequoia regulates cell fate decisions in the external sensory organs of adult Drosophila. EMBO Rep 10, 636-41.

5. Arimura, N., Kaibuchi, K. (2005). Key regulators in neuronal polarity. Neuron 48, 881-4.

6. Arimura, N., Kaibuchi, K. (2007). Neuronal polarity: from extracellular signals to intracellular mechanisms. Nat Rev Neurosci 8, 194-205.

7. Arnold, D., et al., (1994). A strategy for the analysis of gene expression during neural development. Proc Natl Acad Sci U S A 91, 9970-4.

8. Artavanis-Tsakonas, S., et al. (1999). Notch signaling: cell fate control and signal integration in development. Science 284, 770-776.

9. Bacskai, B., et al., (1993). Spatially resolved dynamics of cAMP and protein kinase A subunits in Aplysia sensory neurons. Science 260, 222–226.

10. Bahri, S. M., et al., (1997). The Drosophila bifocal gene encodes a novel protein which colocalizes with actin and is necessary for photoreceptor morphogenesis. Mol Cell Biol 17, 5521-9.

11. Baines, R. A., et al., (2001). Altered electrical properties in Drosophila neurons developing without synaptic transmission. J Neurosci 21, 1523-31.

12. Bao, H., et al., (2007). The atypical cadherin flamingo regulates synaptogenesis and helps prevent axonal and synaptic degeneration in Drosophila. Mol Cell Neurosci 34, 662-78.

314

13. Bastiani, M. J., et al., (1987). Expression of fasciclin I and II glycoproteins on subsets of axon pathways during neuronal development in the grasshopper. Cell 48, 745-55.

14. Bate, M , Broadie, K. (1995). Wiring by fly: the neuromuscular system of the Drosophila embryo. Neuron 15, 513-25.

15. Bazigou, E., et al., (2007). Anterograde Jelly belly and Alk receptor tyrosine kinase signaling mediates retinal axon targeting in Drosophila. Cell 128, 961-75.

16. Beall, S. A., et al., (2005). Hybrid GPCR/cadherin (Celsr) proteins in rat testis are expressed with cell type specificity and exhibit differential Sertoli cell-germ cell adhesion activity. J Androl 26, 529-38.

17. Becker, L. E., et al., (1986). Dendritic atrophy in children with Down's syndrome. Ann. Neurol 20, 520-78.

18. Benavides-Piccione, R., et al., (2004). On dendrites in Down syndrome and DS murine models: a spiny way to learn. Prog Neurobiol 74, 111-26.

19. Berger, J., et al., (2008). Systematic identification of genes that regulate neuronal wiring in the Drosophila visual system. PLoS Genet 4, e1000085

20. Bhat, K. M., et al., (2007). Regulation of axon guidance by slit and netrin signaling in the Drosophila ventral nerve cord. Genetics 176, 2235-46.

21. Bodily, K. D., et al., (2001). A novel member of the Ig superfamily, turtle, is a CNS specific protein required for coordinated motor control. J Neurosci 21, 3113-25.

22. Bodmer, R., et al., (1987). Transformation of sensory organs by mutations of the cut locus of D. melanogaster. Cell 51, 293-307.

23. Böhni, R., et al., (1999). Autonomous control of cell and organ size by CHICO, a Drosophila homolog of vertebrate IRS1-4. Cell 97, 865-75.

24. Bork, P., Beckmann, G. (1993). The CUB domain. A widespread module in developmentally regulated proteins. J Mol Biol 231, 539-45.

25. Boycott, B. B., Wässle, H. (1991). Morphological Classification of Bipolar Cells of the Primate Retina. Eur J Neurosci 3, 1069-88.

26. Brand, A. H., Perrimon, N. (1993). Targeted gene expression as a means of altering cell fates and generating dominant phenotypes. Development 118, 401-15.

27. Brankatschk, M., Dickson, B. J. (2006). Netrins guide Drosophila commissural axons at short range. Nat Neurosci 9, 188-94.

315

28. Brenman, J. E., et al., (2001). Sequoia, a tramtrack-related zinc finger protein, functions as a pan-neural regulator for dendrite and axon morphogenesis in Drosophila. Dev Cell 1, 667-77.

29. Brogiolo, W., et al., (2001). An evolutionarily conserved function of the Drosophila insulin receptor and insulin-like peptides in growth control. Curr Biol 11, 213-21.

30. Brose, K. and Tessier-Lavigne, M. (2000). Slit proteins: key regulators of axon guidance, axonal branching, and cell migration. Curr. Opin. Neurobiol. 10, 95-102

31. Brose, K., Bland, et al., (1999). Slit proteins bind Robo receptors and have an evolutionarily conserved role in repulsive axon guidance. Cell 96,795 -806.

32. Brummel, T., et al., (1999). The Drosophila activin receptor baboon signals through dSmad2 and controls cell proliferation but not patterning during larval development. Genes Dev 13, 98-111.

33. Brunet, I., et al., (2005).The transcription factor Engrailed-2 guides retinal axons. Nature 438, 94-8.

34. Brunton, L. L., et al., (1981). Functional compartmentation of cyclic AMP and protein kinase in heart. Adv Cyclic Nucleotide Res 14, 391-7.

35. Bugga, L., et al., (2009). The cell surface receptor Tartan is a potential in vivo substrate for the receptor tyrosine phosphatase Ptp52F. Mol Cell Biol 29, 3390-400.

36. Cafferty, P., et al., (2004). The receptor tyrosine kinase Off-track is required for layer-specific neuronal connectivity in Drosophila. Development 131, 5287-95.

37. Cafferty, P., et al., (2006). Semaphorin-1a functions as a guidance receptor in the Drosophila visual system. J. Neurosci 26, 3999-4003.

38. Cajal, S. R. (1899, 1904). Histology of the Nervous System of Man and Vertebrates (translated by Swanson, N. and Swanson, L.W., 1995) Oxford University Press.

39. Campos-Ortega, J. A. (1995). Genetic mechanisms of early neurogenesis in Drosophila melanogaster. Mol Neurobiol 10, 75-89.

40. Cancela, J. M., et al., (2002). Transformation of local Ca2+ spikes to global Ca2+ transients: the combinatorial roles of multiple Ca2+ releasing messengers. EMBO J 21, 909–919

41. Canon, J., Banerjee, U. (2000). Runt and Lozenge function in Drosophila development. Semin Cell Dev Biol 11, 327-36.

42. Cantallops, I., et al., (2000). Postsynaptic CPG15 promotes synaptic maturation and presynaptic axon arbor elaboration in vivo. Nat Neurosci 3, 1004-11.

316

43. Carreira-Barbosa, F., et al., (2009). Flamingo regulates epiboly and convergence/extension movements through cell cohesive and signalling functions during zebrafish gastrulation. Development 136, 383-92.

44. Castellani, V., Bolz, J. (1997). Membrane-associated molecules regulate the formation of layer-specific cortical circuits. Proc Natl Acad Sci U S A 94, 7030-5.

45. Caudy, M., Bentley, D. (1986). Pioneer growth cone steering along a series of neuronal and non-neuronal cues of different affinities. J Neurosci 6, 1781-95.

46. Chae, J., et al., (1999). The Drosophila tissue polarity gene starry night encodes a member of the protocadherin family. Development 126, 5421-9.Chamovitz, D. A., et al., (1996). The COP9 complex, a novel multisubunit nuclear regulator involved in light control of a plant developmental switch. Cell 86, 115-21.

47. Chamovitz, D. A., Segal, D. (2001). JAB1/CSN5 and the COP9 signalosome. A complex situation. EMBO Rep 2, 96-101.

48. Chang, Z., et al., (1993). Molecular and genetic characterization of the Drosophila tartan gene. Dev. Biol 160, 315-32.

49. Chen, B. E., et al., (2006). The molecular diversity of Dscam is functionally required for neuronal wiring specificity in Drosophila. Cell 125, 607-20.

50. Chen, C., et al. (1996). The Drosophila insulin receptor is required for normal growth. Endocrinology 137, 846-56.Chen, P. L., Clandinin, T. R. (2008). The cadherin Flamingo mediates level-dependent interactions that guide photoreceptor target choice in Drosophila. Neuron 58, 26-33.

51. Chen, W. S., et al., (2008). Asymmetric homotypic interactions of the atypical cadherin flamingo mediate intercellular polarity signaling. Cell 133, 1093-105.

52. Choe, K. M., et al., (2006). Liprin-alpha is required for photoreceptor target selection in Drosophila. Proc Natl Acad Sci U S A 103, 11601-6.

53. Chotard, C., et al., (2005). glial cells missing and gcm2 cell autonomously regulate both glial and neuronal development in the visual system of Drosophila. Neuron 48, 237-51.

54. Chotard, C., Salecker, I. (2004). Neurons and glia: team players in axon guidance. Trends Neurosci 27, 655-61.

55. Clandinin, T. R., et al., (2001). Drosophila LAR regulates R1-R6 and R7 target specificity in the visual system. Neuron 32, 237-48.

56. Clandinin, T. R., Zipursky, S. L. (2000). Afferent growth cone interactions control synaptic specificity in the Drosophila visual system. Neuron 28, 427-36. 317

57. Clandinin, T. R., Zipursky, S. L. (2002). Making Connections in the Fly Visual System. Neuron 35, 827-41.

58. Clarke, D. C., Liu, X. (2008). Decoding the quantitative nature of TGF-beta/Smad signaling. Trends Cell Biol 18, 430-42.

59. Cohen, E. D., et al., (2002). DWnt4 regulates cell movement and focal adhesion kinase during Drosophila ovarian morphogenesis. Dev Cell 2, 437-48.

60. Conde , C., Cáceres, A. (2009). Microtubule assembly, organization and dynamics in axons and dendrites. Nat Rev Neurosci 10, 319-32.

61. Corty, M. M., et al., (2009). Molecules and mechanisms of dendrite development in Drosophila. Development 136, 1049-61.

62. Curtin, J. A., et al., (2003). Mutation of Celsr1 disrupts planar polarity of inner ear hair cells and causes severe neural tube defects in the mouse. Curr Biol 13, 1129-33.

63. Dailey, M. E., Smith, S. J. (1996). The dynamics of dendritic structure in developing hippocampal slices. J Neurosci 16, 2983-94.

64. Das, G., et al., (2004). Diego interacts with Prickle and Strabismus/Van Gogh to localize planar cell polarity complexes. Development 131, 4467-76.

65. Das, G.,et al., (2002). The atypical cadherin Flamingo links Frizzled and Notch signaling in planar polarity establishment in the Drosophila eye. Dev Cell 2, 655-66.

66. Davies, A., et al., (2005). Planar polarity of hair cells in the chick inner ear is correlated with polarized distribution of c-flamingo-1 protein. Dev Dyn 233, 998- 1005.

67. Davisson, M. T., et al., (1990). Segmental trisomy of murine chromosome 16: a new model system for studying Down syndrome. Prog Clin Biol Res 360, 263-80.

68. Dayton, J. S., et al., (1992). Comparison of the effects of mizoribine with those of azathioprine, 6-mercaptopurine, and mycophenolic acid on T lymphocyte proliferation and purine ribonucleotide metabolism. Mol Pharmacol 41, 671-676.

69. Dearborn, R. Jr., Kunes, S. (2004). An axon scaffold induced by retinal axons directs glia to destinations in the Drosophila optic lobe. Development 131, 2291-303.

70. Desai, C. J., et al., (1994). A Drosophila receptor tyrosine phosphatase expressed in the embryonic CNS and larval optic lobes is a member of the set of proteins bearing the "HRP" carbohydrate epitope. J Neurosci 14, 7272-83.

71. Desai, C. J., et al., (1996). Receptor tyrosine phosphatases are required for motor axon guidance in the Drosophila embryo. Cell 84, 599-609.

318

72. Desai, C. J., et al., (1999). The Drosophila SH2-SH3 adapter protein Dock is expressed in embryonic axons and facilitates synapse formation by the RP3 motoneuron. Development 126, 1527-35.

73. Dickson, B. J. (2001). Rho GTPases in growth cone guidance. Curr Opin Neurobiol 11, 103-10.

74. Dotti, C. G., Poo, M. M. (2003). Neuronal polarization: building fences for molecular segregation. Nat Cell Biol 5, 591-4.

75. Downward, J., (1994). The GRB2/Sem-5 adaptor protein. FEBS Lett 338, 113-17.

76. Duffy, J. B. (2002). GAL4 system in Drosophila: a fly geneticist's Swiss army knife. Genesis 34, 1-15.

77. Emoto, K., et al., (2004). Control of dendritic branching and tiling by the Tricornered-kinase/Furry signaling pathway in Drosophila sensory neurons. Cell 119, 245-56.

78. Emoto, K., et al., (2006). The tumour suppressor Hippo acts with the NDR kinases in dendritic tiling and maintenance. Nature 443, 210-13.

79. Englund, C., et al., (2003). Jeb signals through the Alk receptor tyrosine kinase to drive visceral muscle fusion. Nature 425, 512-6.

80. Estes, P. S., et al., (2000). Synaptic localization and restricted diffusion of a Drosophila neuronal synaptobrevin--green fluorescent protein chimera in vivo. J Neurogenet 13, 233-55.

81. Eugui, E. M., et al., (1991). Lymphocyte-selective cytostatic and immune- suppressive effects of mycophenolic acid in vitro: role of deoxyguanosine nucleotide depletion. Scand J Immunol 33, 161-173.

82. Fan, Y., et al., (2005). The egghead gene is required for compartmentalization in Drosophila optic lobe development. Dev Biol 287, 61-73.

83. Farajian, R., et al., (2004). Cellular positioning and dendritic field size of cholinergic amacrine cells are impervious to early ablation of neighboring cells in the mouse retina. Vis Neurosci 21, 13-22.

84. Feiguin, F., et al., (2001). The ankyrin repeat protein Diego mediates Frizzled- dependent planar polarization. Dev Cell 1, 93-101.

85. Ferguson et al., (2009). The conserved Ig superfamily member Turtle mediates axonal tiling in Drosophila. J. Neurosci 29, 14151-9.

319

86. Fernandez, R., et al., (1995). The Drosophila insulin receptor homolog: a gene essential for embryonic development encodes two receptor isoforms with different signaling potential. EMBO J 14, 3373-84.

87. Fitzpatrick, K.A., et al., (1995). Expression of protein tyrosine phosphatase genes during oogenesis in Drosophila melanogaster. Mech. Dev 53, 171-83.

88. Flanagan, J. G. (2006). Neural map specification by gradients. Curr Opin Neurobiol 16, 59-66.

89. Fox, A. N., Zinn, K. (2005). The heparan sulfate proteoglycan syndecan is an in vivo ligand for the Drosophila LAR receptor tyrosine phosphatase. Curr Biol 15, 1701-11.

90. Fradkin, L. G., et al., (1995). The Drosophila Wnt protein DWnt-3 is a secreted glycoprotein localized on the axon tracts of the embryonic CNS. Dev Biol 168, 202- 13.

91. Frankfort B, J., et al., (2001). senseless repression of rough is required for R8 photoreceptor differentiation in the developing Drosophila eye. Neuron 32, 403-14.

92. Frontini, M., et al., (2002). NF-Y recruitment of TFIID, multiple interactions with histone fold TAF(II)s. J Biol Chem 277, 5841-8.

93. Furrer, M. P., et al. (2003). Robo and Frazzled/DCC mediate dendritic guidance at the CNS midline. Nat Neurosci 6, 223-30.

94. Gallegos, M. E., Bargmann, C. I. (2004). Mechanosensory neurite termination and tiling depend on SAX-2 and the SAX-1 kinase. Neuron 44, 239-78.

95. Gallo, G., Letourneau, P. C. (1998). Axon guidance: GTPases help axons reach their targets. Curr Biol 8, R80-2.

96. Ganguly, A., et al., (2005). Drosophila WntD is a target and an inhibitor of the Dorsal/Twist/Snail network in the gastrulating embryo. Development 132, 3419-29.

97. Gao, F. B., et al., (1999). Genes regulating dendritic outgrowth, branching, and routing in Drosophila. Genes Dev 13, 2549-61.

98. Gao, F. B., et al., (2000). Control of dendritic field formation in Drosophila: the roles of flamingo and competition between homologous neurons. Neuron 28, 91-101.

99. Garrity, P. A., et al., (1996). Drosophila photoreceptor axon guidance and targeting requires the dreadlocks SH2/SH3 adapter protein. Cell 85, 639-50.

100. Garrity, P. A., et al., (1999). Retinal axon target selection in Drosophila is regulated by a receptor protein tyrosine phosphatase. Neuron 22, 707-17.

320

101. Ghysen, A., Dambly-Chaudiere, C. (1989). Genesis of the Drosophila peripheral nervous system. Trends Genet 5, 251-5.

102. Ghysen, A., et al., (1986). Sensory neurons and peripheral pathways in Drosophila embryos. Roux's Arch. Dev. Biol.195, 281-89.

103. Gibbs, M. S. (2001). Regulation of Drosophila visual system development by Nitric Oxide and cyclic GMP. American Zoologist 41, 268-81.

104. Gibbs, S. M., et al., Soluble (2001). Guanylate cyclase is required during development for visual system function in Drosophila. J Neurosci 21, 7705-14.

105. Gibbs, S. M., Truman, J. W. (1998). Nitric oxide and cyclic GMP regulate retinal patterning in the optic lobe of Drosophila. Neuron 20, 83-93.

106. Glise, B., et al., (1995). hemipterous encodes a novel Drosophila MAP kinase kinase, required for epithelial cell sheet movement. Cell 83, 451-461.

107. Godenschwege, T. A., et al., (2002). Ectopic expression in the giant fiber system of Drosophila reveals distinct roles for roundabout (Robo), Robo2, and Robo3 in dendritic guidance and synaptic connectivity. J. Neurosci 22, 3117-29.

108. Godenschwege, T.A., et al., (2002). Bi-directional signaling by Semaphorin 1a during central synapse formation in Drosophila. Nat. Neurosci 5, 1294-1301.

109. Golic, K. G. (1991). Site-specific recombination between homologous chromosomes in Drosophila. Science 252, 958-61.

110. Goodman, C. S., (1996). Mechanisms and molecules that control growth cone guidance. Annu Rev Neurosci 19, 341-77.

111. Gordon, M. D., et al., (2005). WntD is a feedback inhibitor of Dorsal/NF-kappaB in Drosophila development and immunity. Nature 437, 746-9.

112. Gordon, M. D., Nusse, R. (2006) Wnt signaling: multiple pathways, multiple receptors, and multiple transcription factors. J Biol Chem 281, 22429-33.

113. Gorski, J. A., et al., (2003). Brain-derived neurotrophic factor is required for the maintenance of cortical dendrites. J Neurosci 23, 6856-65.

114. Green, J. L., et al., (2007). The C. elegans ROR receptor tyrosine kinase, CAM-1, non autonomously inhibits the Wnt pathway. Development 134, 4053-62.

115. Green, J. L., et al., (2008). Opposing Wnt pathways orient cell polarity during organogenesis. Cell 134, 646-56.

321

116. Green, J., et al., (2008). Ror receptor tyrosine kinases: orphans no more. Trends Cell Biol 18, 536-44.

117. Grenningloh, G., et al., (1991). Genetic analysis of growth cone guidance in Drosophila: fasciclin II functions as a neuronal recognition molecule. Cell 67, 45-57.

118. Grueber, W. B., et al., (2003b). Dendrites of distinct classes of Drosophila sensory neurons show different capacities for homotypic repulsion. Curr. Biol 13, 618-78.

119. Grueber, W. B., Jan, L. Y. and Jan, Y. N. (2003a). Different levels of the homeodomain protein cut regulate distinct dendrite branching patterns of Drosophila multidendritic neurons. Cell 112, 805 -818.

120. Grueber, W. B., Jan, L. Y., Jan, Y. N. (2002). Tiling of the Drosophila epidermis by multidendritic sensory neurons. Development 129, 2867-78.

121. Grueber, W. B., Jan, Y. N. (2004). Dendritic development: lessons from Drosophila and related branches. Curr Opin Neurobiol 14, 74-82.

122. Gu, J. J., et al., (2000). Inhibition of T lymphocyte activation in mice heterozygous for loss of the IMPDH II gene. J Clin Invest 106, 599-606.

123. Gu, J. J., et al., (2003). Targeted disruption of the inosine 5'-monophosphate dehydrogenase type I gene in mice. Mol Cell Biol 23, 6702-6712.

124. Gundersen, R. W., Barrett, J. N. (1979). Neuronal chemotaxis: chick dorsal-root axons turn toward high concentrations of nerve growth factor. Science 206, 1079-80.

125. Hakeda-Suzuki, S., et al., (2002). Rac function and regulation during Drosophila development. Nature 416, 438-42.

126. Harrison, R. G. (1910). The outgrowth of the nerve fiber as a mode of protoplasmic movement. J Exp Zool 9, 787-846.

127. Harrison, S. D. and Travers, A. A. (1990). The tramtrack gene encodes a Drosophila finger protein that interacts with the ftz transcriptional regulatory region and shows a novel embryonic expression pattern. EMBO J. 9, 207-16.

128. Hartenstein, V. (1988). Development of Drosophila larval sensory organs: spatiotemporal pattern of sensory neurones, peripheral axonal pathways and sensilla differentiation. Development 102, 869-886.

129. Hartenstein, V., Jan, Y.N. (1992). Studying Drosophila embryogenesis with P-lacZ enhancer trap lines. Rouxs Arch. Dev. Biol 201, 194-220.

130. Hattori, D., et al., (2007). Dscam diversity is essential for neuronal wiring and self- recognition. Nature 449, 223-7.

322

131. Hauser, I. A., Sterzel, R. B. (1999). Mycophenolate mofetil: therapeutic applications in kidney transplantation and immune-mediated renal disease. Curr Opin Nephrol Hypertens 8, 1-6.

132. Häusser, M., Spruston, N., Stuart, G. J. (2000). Diversity and dynamics of dendritic signaling. Science 290, 739-44.

133. Hayashi, T., Kojima, T., Saigo, K. (1998). Specification of primary pigment cell and outer photoreceptor fates by BarH1 homeobox gene in the developing Drosophila eye. Dev Biol 200, 131-45.

134. He, Z., Tessier-Lavigne, M. (1997). Neuropilin is a receptor for the axonal chemorepellent Semaphorin III. Cell 90, 739-51.

135. Heisenberg, M., and Buchner, E. (1977). The role of retinula cell types in the visual behavior of Drosophila melanogaster. J. Comp. Physiol. 117, 127-162.

136. Hikasa, H., et al., (2002). The Xenopus receptor tyrosine kinase Xror2 modulates morphogenetic movements of the axial mesoderm and neuroectoderm via Wnt signaling. Development 129, 5227-39.

137. Hing, H., et al., (1999). Pak functions downstream of Dock to regulate photoreceptor axon guidance in Drosophila. Cell 97, 853-63.

138. Hirst, M., et al., (1994). Human GMP synthetase. Protein purification, cloning, and functional expression of cDNA. J Biol Chem 269, 23830-37.

139. Hofmeyer, K, et al., (2006). Liprin-alpha has LAR-independent functions in R7 photoreceptor axon targeting. Proc Natl Acad Sci U S A. 103, 11595-600.

140. Horch, H. W., et al., (1999). Destabilization of cortical dendrites and spines by BDNF. Neuron 23, 353-64.

141. Horton, A. C., Ehlers, M. D. (2003). Neuronal polarity and trafficking. Neuron 40, 277- 95.

142. Howes, R., Bray, S. (2000). Pattern formation: Wingless on the move. Curr Biol.10, R222-6.

143. Huang, Z. and Kunes, S. (1998). Signals transmitted along retinal axons in Drosophila: Hedgehog signal reception and the cell circuitry of lamina cartridge assembly. Development 125, 3753-64.

144. Huang, Z., Kunes, S. (1996) Hedgehog, transmitted along retinal axons, triggers neurogenesis in the developing visual centers of the Drosophila brain. Cell 86, 411- 22.

323

145. Huang, Z., Shilo, B. Z. and Kunes, S. (1998). A retinal axon fascicle uses spitz, an EGF receptor ligand, to construct a synaptic cartridge in the brain of Drosophila. Cell 95, 693-703.

146. Hughes, C. L., Thomas, J. B. (2007). A sensory feedback circuit coordinates muscle activity in Drosophila. Mol Cell Neurosci.35, 383-96.

147. Hughes, M. E., et al., (2007). Homophilic Dscam interactions control complex dendrite morphogenesis. Neuron 54, 417-27.Hummel T, et al, (2002). Temporal control of glial cell migration in the Drosophila eye requires gilgamesh, hedgehog, and eye specification genes. Neuron 33, 193-203.

148. Hummel T, et al., (2003). Axonal targeting of olfactory receptor neurons in Drosophila is controlled by Dscam. Neuron 37, 221-31.

149. Hurley, S. P., et al., (2006). Anaplastic lymphoma kinase is dynamically expressed on subsets of motor neurons and in the peripheral nervous system. J Comp Neurol. 495, 202-12.

150. Hwang, R. Y., et al., (2007). Nociceptive neurons protect Drosophila larvae from parasitoid wasps. Curr Biol. 17, 2105-16.

151. Inaki, M., et al., (2007). Wnt4 is a local repulsive cue that determines synaptic target specificity. Curr Biol. 17, 1574-9.

152. Inoue, A., Sanes, J. R. (1997). Lamina-specific connectivity in the brain: regulation by N-cadherin, neurotrophins, and glycoconjugates. Science. 276, 1428-31.

153. Iwahara, T., et al., (1997). Molecular characterization of ALK, a receptor tyrosine kinase expressed specifically in the nervous system. Oncogene. 14, 439-49.

154. Iwai, Y., et al., (1997). Axon patterning requires DN-cadherin, a novel neuronal adhesion receptor, in the Drosophila embryonic CNS. Neuron. 19, 77-89.

155. Jaeken, J., Van den Berghe, G. (1984). An infantile autistic syndrome characterised by the presence of succinylpurines in body fluids. Lancet 2, 1058-61.

156. Janetopoulos, C., et al., (2001). Receptor-mediated activation of heterotrimeric G- proteins in living cells. Science 291, 2408–11.

157. Jeon M, et al., (2008). Redundancy and compensation in axon guidance: genetic analysis of the Drosophila Ptp10D/Ptp4E receptor tyrosine phosphatase subfamily. Neural Dev. 3, 3.

158. Jing, L., et al., (2009). Wnt signals organize synaptic prepattern and axon guidance through the zebrafish unplugged/MuSK receptor. Neuron 61, 721-33.

324

159. Johnson, K. G., et al., (2006). The HSPGs Syndecan and Dallylike bind the receptor phosphatase LAR and exert distinct effects on synaptic development. Neuron. 49, 517-31.

160. Justice, R. W., et al., (1995). The Drosophila tumor suppressor gene warts encodes a homolog of human myotonic dystrophy kinase and is required for the control of cell shape and proliferation. Genes Dev. 9, 534-46.

161. Kaminker, J. S., et al., (2002). Control of photoreceptor axon target choice by transcriptional repression of Runt. Nat Neurosci. 5, 746-50.

162. Kania, M. A., et al., (1990). The Drosophila segmentation gene runt encodes a novel nuclear regulatory protein that is also expressed in the developing nervous system. Genes Dev. 4, 1701-13.

163. Katzav, S., et al., (1991). Loss of the amino-terminal helix-loop-helix domain of the vav proto-oncogene activates its transforming potential. Mol Cell Biol. 11, 1912-20.

164. Kauffmann, R. C., et al., (1996). Ras1 signaling and transcriptional competence in the R7 cell of Drosophila. Genes Dev. 10, 2167-78.

165. Kaufmann, N., et al., (2002). Drosophila liprin-alpha and the receptor phosphatase Dlar control synapse morphogenesis. Neuron. 34, 27-38.

166. Kaufmann, W. E., Moser, H. W. (2000). Dendritic anomalies in disorders associated with mental retardation. Cereb Cortex. 10, 981-91.

167. Keleman, K., et al., (2002). Comm sorts robo to control axon guidance at the Drosophila midline. Cell 110, 415-27.

168. Keleman, K., et al., (2005). Comm function in commissural axon guidance: cell- autonomous sorting of Robo in vivo. Nat Neurosci 8, 156-63.

169. Kernan, M., et al., (1994). Genetic dissection of mechanosensory transduction: mechanoreception-defective mutations of Drosophila. Neuron 12, 1195-206.

170. Keshishian, H., et al., (1996). The Drosophila neuromuscular junction: a model system for studying synaptic development and function. Annu Rev Neurosci 19, 545- 75.

171. Kidd, T., Bland, K.S., Goodman, C.S. (1999). Slit is the midline repellent for the robo receptor in Drosophila. Cell 96, 785-94.

172. Kidd, T., et al., (1998). Roundabout controls axon crossing of the CNS midline and defines a novel subfamily of evolutionarily conserved guidance receptors. Cell 92, 205-15.

325

173. Kim, M. D., Jan, L. Y. and Jan, Y. N. (2006). The bHLH-PAS protein Spineless is necessary for the diversification of dendrite morphology of Drosophila dendritic arborization neurons. Genes Dev. 20, 2806-19.

174. Kim, T., et al., (2001) PCI complexes: pretty complex interactions in diverse signaling pathways. Trends Plant Sci. 6, 379-86.

175. Kimura, H., et al., (2006). Potential dual molecular interaction of the Drosophila 7- pass transmembrane cadherin Flamingo in dendritic morphogenesis. J Cell Sci.119, 1118-29.

176. Kolodkin, A. L., et al., (1992). Fasciclin IV: sequence, expression, and function during growth cone guidance in the grasshopper embryo. Neuron. 9, 831-45.

177. Kolodziej, P. A., et al., (1996). frazzled encodes a Drosophila member of the DCC immunoglobulin subfamily and is required for CNS and motor axon guidance. Cell. 87, 197-204.

178. Kolodziej, P. A., Jan, L. Y., Jan, Y. N. (1995). Mutations that affect the length, fasciculation, or ventral orientation of specific sensory axons in the Drosophila embryo. Neuron. 15, 273-86.

179. Kozopas, K. M., Nusse, R. (2002). Direct flight muscles in Drosophila develop from cells with characteristics of founders and depend on DWnt-2 for their correct patterning. Dev Biol. 243, 312-25.

180. Kozopas, K. M., Samos, C. H., Nusse, R. (1998). DWnt-2, a Drosophila Wnt gene required for the development of the male reproductive tract, specifies a sexually dimorphic cell fate. Genes Dev. 12, 1155-65.

181. Kramer, A. P., Kuwada, J. Y. (1983). Formation of the receptive fields of leech mechanosensory neurons during embryonic development. J Neurosci. 3, 2474-86.

182. Kramer, A. P., Stent, G. S. (1985). Developmental arborization of sensory neurons in the leech Haementeria ghilianii. II. Experimentally induced variations in the branching pattern. J Neurosci. 5, 768-75.

183. Krause, C., et al., (2006). Distinct functions of the leucine-rich repeat transmembrane proteins Capricious and Tartan in the Drosophila tracheal morphogenesis. Dev. Biol. 296, 253-64.

184. Kraut, R., Menon, K., Zinn, K. (2001). A gain-of-function screen for genes controlling motor axon guidance and synaptogenesis in Drosophila. Curr Biol. 11, 417-30.

185. Krueger, N.X., et al., (1996). The transmembrane tyrosine phosphatase DLAR controls motor axon guidance in Drosophila. Cell 84, 611-622. 326

186. Kunes, S., Steller, H. (1993) Topography in the Drosophila visual system. Curr Opin Neurobiol. 3, 53-9.

187. Kunes, S., Wilson, C., Steller, H. (1993) Independent guidance of retinal axons in the developing visual system of Drosophila. J Neurosci. 13, 752-67.

188. Kuo, C. T., et al., (2006). Identification of E2/E3 ubiquitinating enzymes and caspase activity regulating Drosophila sensory neuron dendrite pruning. Neuron 51, 283-78.

189. Kuo, C. T., Jan, L. Y., Jan, Y. N. (2005). Dendrite-specific remodeling of Drosophila sensory neurons requires matrix metalloproteases, ubiquitin-proteasome, and ecdysone signaling. Proc. Natl. Acad. Sci. USA 102, 15230-78.

190. Kurusu, M., et al., (2008). A screen of cell-surface molecules identifies leucine-rich repeat proteins as key mediators of synaptic target selection. Neuron. 59, 972-85.

191. Kuwada, J. Y., Kramer, A. P. (1983). Embryonic development of the leech nervous system: primary axon outgrowth of identified neurons. J Neurosci. 3, 2098-111.

192. LaJeunesse, D. R., et al., (2001). A Systematic screen for dominant second-site modifiers of Merlin/NF2 Phenotypes reveals an interaction with blistered/DSRF and scribbler. Genetics 158, 667-79.

193. Landgraf, M., et al., (1997). The origin, location, and projections of the embryonic abdominal motorneurons of Drosophila. J Neurosci. 17, 9642-55.

194. Landgraf, M., et al., (2003). Embryonic origins of a motor system: motor dendrites form a myotopic map in Drosophila. PLoS Biol. 1, 221-30.

195. Lawrence, P. A., Casal, J., Struhl, G. (2004). Cell interactions and planar polarity in the abdominal epidermis of Drosophila. Development. 131, 4651-64.

196. Le Garrec, J. F., Kerszberg, M. (2008). Modeling polarity buildup and cell fate decision in the fly eye: insight into the connection between the PCP and Notch pathways. Dev Genes Evol. 218, 413-26.

197. Lee, C. H., et al., (2001). N-cadherin regulates target specificity in the Drosophila visual system. Neuron. 30, 437-50.

198. Lee, H. H., et al., (2003). Jelly belly protein activates the receptor tyrosine kinase Alk to specify visceral muscle pioneers. Nature. 425, 507-12.

199. Lee, R. C., et al., (2003). The protocadherin Flamingo is required for axon target selection in the Drosophila visual system. Nat Neurosci. 6, 557-63.

200. Lee, T., et al., (2000). Cell-autonomous requirement of the USP/EcR-B ecdysone receptor for mushroom body neuronal remodeling in Drosophila. Neuron 28, 807-78.

327

201. Lee, T., et al., (2000). Essential roles of Drosophila RhoA in the regulation of neuroblast proliferation and dendritic but not axonal morphogenesis. Neuron. 25, 307-16.

202. Lee, T., Luo, L. (1999). Mosaic analysis with a repressible cell marker for studies of gene function in neuronal morphogenesis. Neuron. 22, 451-61.

203. Lee, T., Luo, L. (2001). Mosaic analysis with a repressible cell marker (MARCM) for Drosophila neural development. Trends Neurosci. 24, 251-4.

204. Legg, A., O‟Connor, T. (2003). Gradients and Growth Cone Guidance of Grasshopper Neurons. Journal of Histochemistry & Cytochemistry 51, 445-54.

205. Lehmann, J. M., Riethmüller, G., Johnson, J. P. (1990). Nck, a melanoma cDNA encoding a cytoplasmic protein consisting of the src homology units SH2 and SH3. Nucleic Acids Res. 18, 1048.

206. Lein, P., et al., (1995). Osteogenic protein-1 induces dendritic growth in rat sympathetic neurons. Neuron. 15, 597-605.

207. Li, W., et al., (2004). BTB/POZ-zinc finger protein abrupt suppresses dendritic branching in a neuronal subtype-specific and dosage-dependent manner. Neuron 43, 823-34.

208. Li, W., Li, Y., Gao, F. B. (2005) Abelson, enabled, and p120 catenin exert distinct effects on dendritic morphogenesis in Drosophila. Dev Dyn. 234, 512-22.

209. Liao, E. H., et al., (2004). An SCF-like ubiquitin ligase complex that controls presynaptic differentiation. Nature. 430, 345-50.

210. Lin, B., Wang, S. W., Masland, R. H. (2004). Retinal ganglion cell type, size, and spacing can be specified independent of homotypic dendritic contacts. Neuron. 43, 475-85.

211. Lin, D. M., et al., (1994) Genetic analysis of Fasciclin II in Drosophila: defasciculation, refasciculation, and altered fasciculation. Neuron. 13, 1055-69.

212. Lin, D. M., Goodman, C. S. (1994) Ectopic and increased expression of Fasciclin II alters motoneuron growth cone guidance. Neuron. 13, 507-23.

213. Liu, L., et al., (2003). Identification and function of thermosensory neurons in Drosophila larvae. Nat Neurosci. 6, 267-73.

214. Liu, X., et al., (1998). NMR structure and mutagenesis of the N-terminal Dbl homology domain of the nucleotide exchange factor Trio. Cell. 95, 269-77.

328

215. Llimargas, M., Lawrence, P. A. (2001). Seven Wnt homologues in Drosophila: a case study of the developing tracheae. Proc Natl Acad Sci U S A. 98, 14487-92.

216. Llinas, R., et al., (1992). Microdomains of high calcium concentration in a presynaptic terminal. Science 256, 677–79.

217. Lo, D. C., McAllister, A. K., Katz, L. C. (1994). Neuronal transfection in brain slices using particle-mediated gene transfer. Neuron. 13, 1263-8.

218. Lo, P. C. and Frasch, M. (1999). Sequence and expression of myoglianin, a novel Drosophila gene of the TGF-beta superfamily. Mech. Dev. 86, 171-75.

219. Long, H., et al., (2006). De novo GMP synthesis is required for axon guidance in Drosophila. Genetics 172, 1633-42.

220. Long, H., et al., (2009). Dendrite branching and self-avoidance are controlled by Turtle, a conserved IgSF protein in Drosophila. Development 136, 3475-84.

221. Lorén, C. E., et al., (2001) Identification and characterization of DAlk: a novel Drosophila melanogaster RTK which drives ERK activation in vivo. Genes Cells. 6, 531-44.

222. Lorén, C. E., et al., (2003). A crucial role for the Anaplastic lymphoma kinase receptor tyrosine kinase in gut development in Drosophila melanogaster. EMBO Rep. 4, 781-6.

223. Luo, L. (2002). Actin cytoskeleton regulation in neuronal morphogenesis and structural plasticity. Annu Rev Cell Dev Biol. 18, 601-35.

224. Luo, L., et al., (1997). Rho family GTP-binding proteins in growth cone signalling. Curr Opin Neurobiol 7, 81-6.

225. Luo, L., Jan, L. Y., Jan, Y. N. (1997). Rho family GTP-binding proteins in growth cone signalling. Curr Opin Neurobiol. 7, 81-6.

226. Luo, Y., Raible, D., Raper, J. A. (1993). Collapsin: a protein in brain that induces the collapse and paralysis of neuronal growth cones. Cell. 75, 217-27.

227. MacNeil, M. A., Masland, R. H. (1998). Extreme diversity among amacrine cells: implications for function. Neuron. 20, 971-82.

228. Malun, D., Brunjes, P. C. (1996). Development of olfactory glomeruli: temporal and spatial interactions between olfactory receptor axons and mitral cells in opossums and rats. J. Comp. Neurol. 368, 1-78.

229. Mantovani, R. (1999). The molecular biology of the CCAAT-binding factor NF-Y. Gene. 239, 15-27.

329

230. Mao, Y., Kerr, M., Freeman, M. (2008). Modulation of Drosophila retinal epithelial integrity by the adhesion proteins capricious and tartan. PLoS One. 3, e1827.

231. Mardon, G., Solomon, N. M., Rubin, G. M. (1994). dachshund encodes a nuclear protein required for normal eye and leg development in Drosophila. Development. 120, 3473-86.

232. Marqués, G. (2005). Morphogens and synaptogenesis in Drosophila. J Neurobiol. 64, 417-34.

233. Marrs, G. S., et al., (2006). Dendritic arbors of developing retinal ganglion cells are stabilized by beta 1-integrins. Mol Cell Neurosci. 32, 230-41.

234. Marsault, R., et al., (1997). Domains of high Ca2+ beneath the plasma membrane of living A7r5 cells. EMBO J 16, 1575–81.

235. Martin, K. A., et al., (1995). Mutations disrupting neuronal connectivity in the Drosophila visual system. Neuron. 14, 229-40.

236. Martinez-Arias, A., Lawrence, P. A. (1985). Parasegments and compartments in the Drosophila embryo. Nature. 313, 639-42.

237. Mast, J. D., et al., (2006). The mechanisms and molecules that connect photoreceptor axons to their targets in Drosophila. Semin Cell Dev Biol. 17, 42-9.

238. Matthews, B. J., et al., (2007). Dendrite self-avoidance is controlled by Dscam. Cell 129, 593-604.

239. Maurel-Zaffran, C., et al., (2001). Cell-autonomous and -nonautonomous functions of LAR in R7 photoreceptor axon targeting. Neuron. 32, 225-35.

240. Mayer, B. J., Eck, M. J. (1995). SH3 domains. Minding your p's and q's. Curr Biol. 5, 364-7.

241. McAllister, A. K., Katz, L. C., Lo, D. C. (1996). Neurotrophin regulation of cortical dendritic growth requires activity. Neuron. 17, 1057-64.

242. McAllister, A. K., Katz, L. C., Lo, D. C. (1997). Opposing roles for endogenous BDNF and NT-3 in regulating cortical dendritic growth. Neuron. 18, 767-78.

243. McAllister, A. K., Lo, D. C., Katz, L. C. (1995). Neurotrophins regulate dendritic growth in developing visual cortex. Neuron. 15, 791-803.

244. McCarty, J. H. (1998). The Nck SH2/SH3 adaptor protein: a regulator of multiple intracellular signal transduction events. Bioessays. 20, 913-21.

330

245. McGuire, S. E., Roman, G., Davis, R. L. (2004). Gene expression systems in Drosophila: a synthesis of time and space. Trends Genet. 20, 384-91.

246. Meinertzhagen, I. A. and Hanson, T. E. (1993). The development of the optic lobe. In The development of Drosophila melanogaster (ed. M. Bate and A. Martinez- Arias), 1363-1491. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press.

247. Melzig, J., et al., (1998). A protein related to p21-activated kinase (PAK) that is involved in neurogenesis in the Drosophila adult central nervous system. Curr. Biol. 8, 1223-26.

248. Messersmith, E. K., et al., (1995). Semaphorin III can function as a selective chemorepellent to pattern sensory projections in the spinal cord. Neuron. 14, 949-59.

249. Mikels, A. J., Nusse, R. (2006). Wnts as ligands:processing, secretion and reception. Oncogene 25, 7461-68.

250. Miki T, et al., (1993). Oncogene ect2 is related to regulators of small GTP-binding proteins. Nature. 362, 462-5.

251. Miklos, G. L., Rubin, G. M. (1996). The role of the genome project in determining gene function: insights from model organisms. Cell 86, 521-9.

252. Milan, M., et al., (2001). The LRR proteins Capricious and Tartan mediate cell interactions during DV boundary formation in the Drosophila wing. Cell 106, 785- 94.

253. Milan, M., Perez, L., Cohen, S.M. (2005). Boundary formation in the Drosophila wing: functional dissection of Capricious and Tartan. Dev. Dyn. 233, 804-10.

254. Millard, S. S., et al., (2007). Dscam2 mediates axonal tiling in the Drosophila visual system. Nature. 447, 720-4.

255. Millard, S. S., Zipursky, S. L. (2008). Dscam-mediated repulsion controls tiling and self-avoidance. Curr Opin Neurobiol. 18, 84-9.

256. Mishra, A., et al., (2008). The protein dendrite arborization and synapse maturation 1 (Dasm-1) is dispensable for dendrite arborization. Mol Cell Biol 28, 2782-91.

257. Mitchell KJ, et al., (1996). Genetic analysis of Netrin genes in Drosophila: Netrins guide CNS commissural axons and peripheral motor axons. Neuron. 17, 203-15.

258. Mochizuki, N., et al., (2001). Spatio-temporal images of growth-factor-induced activation of Ras and Rap1. Nature 411, 1065–68.

259. Monnier, P. P., et al., (2002). RGM is a repulsive guidance molecule for retinal axons. Nature. 419, 392-5.

331

260. Moresco, E. M., et al., (2005). Integrin-mediated dendrite branch maintenance requires Abelson (Abl) family kinases. J Neurosci. 25, 6105-18.

261. Morey, M., et al., (2008). Coordinate control of synaptic-layer specificity and rhodopsins in photoreceptor neurons. Nature. 456, 795-9.

262. Morris, S. W., et al., (1994). Fusion of a kinase gene, ALK, to a nucleolar protein gene, NPM, in non-Hodgkin's lymphoma. Science. 263, 1281-4.

263. Mueller, B. K. (1999). Growth cone guidance: first steps towards a deeper understanding. Annu Rev Neurosci. 22, 351-88.

264. Nagai, Y., et al., (2000). A fluorescent indicator for visualizing cAMP-induced phosphorylation in vivo. Nat. Biotechnol 18, 313–316.

265. Nakayama, A. Y., Harms, M. B., Luo, L. (2000). Small GTPases Rac and Rho in the maintenance of dendritic spines and branches in hippocampal pyramidal neurons. J Neurosci. 20, 5329-38.

266. Nedivi, E., et al., (2001). Developmental regulation of CPG15 expression in Xenopus. J Comp Neurol. 435, 464-73.

267. Nedivi, E., Wu, G. Y., Cline, H. T. (1998) Promotion of dendritic growth by CPG15, an activity-induced signaling molecule. Science. 281, 1863-6.

268. Nern, A., et al., (2005). An isoform-specific allele of Drosophila N-cadherin disrupts a late step of R7 targeting. Proc Natl Acad Sci U S A. 102, 12944-9.

269. Nern, A., Zhu, Y., Zipursky, S. L. (2008). Local N-cadherin interactions mediate distinct steps in the targeting of lamina neurons. Neuron. 58, 34-41.

270. Neves, S. R., et al., (2008). Cell shape and negative links in regulatory motifs together control spatial information flow in signaling networks. Cell 133, 666-80.

271. Newsome, T. P., Asling, B., Dickson, B. J. (2000). Analysis of Drosophila photoreceptor axon guidance in eye-specific mosaics. Development. 127, 851-60.

272. Newsome, T. P., et al., (2000) Trio combines with dock to regulate Pak activity during photoreceptor axon pathfinding in Drosophila. Cell. 101, 283-94.

273. Ng, J., et al., (2002). Rac GTPases control axon growth, guidance and branching. Nature. 416, 442-7.

274. Nguyen, M., et al., (2000). Identification of maverick, a novel member of the TGF- beta superfamily in Drosophila. Mech. Dev. 95, 201-206.

332

275. Nikolaev, V. O., et al., (2004). Novel single chain cAMP sensors for receptor- induced signal propagation. J. Biol. Chem 279, 37215–18.

276. Nishiyama, M., et al., (2003). Cyclic AMP/GMP-dependent modulation of Ca2+ channels sets the polarity of nerve growth-cone turning. Nature 423, 990-5.

277. O‟connor, T. P., Duerr, J. S., Bentley, D. (1990). Pioneer growth cone steering decisions mediated by single filopodial contacts in situ. J Neuroscience 10, 3935-46.

278. Oishi, I., et al., (1997). A novel Drosophila receptor tyrosine kinase expressed specifically in the nervous system. Unique structural features and implication in developmental signaling. J Biol Chem 272, 11916-25.

279. Oishi, I., et al., (2003). The receptor tyrosine kinase Ror2 is involved in noncanonical Wnt5a/JNK signalling pathway. Genes Cells 8, 645-54.

280. Ou, Y., et al., (2008). Identification of genes influencing dendrite morphogenesis in developing peripheral sensory and central motor neurons. Neural Develop. 3, 16.

281. Paricio, N., et al., (1999). The Drosophila STE20-like kinase misshapen is required downstream of the Frizzled receptor in planar polarity signaling. EMBO J. 18, 4669- 78.

282. Parker, L., et al., (2006). The divergent TGF-beta ligand Dawdle utilizes an activin pathway to influence axon guidance in Drosophila. Development 133, 4981-91.

283. Parrish, J. Z., et al., (2007). Mechanisms that regulate establishment, maintenance, and remodeling of dendritic fields. Annu Rev Neurosci. 30, 399-423.

284. Pawson, T. (2004). Specificity in signal transduction: from phosphotyrosine-SH2 domain interactions to complex cellular systems. Cell. 116, 191-203.

285. Peng, Y., Jahroudi, N. (2002) The NFY transcription factor functions as a repressor and activator of the von Willebrand factor promoter. Blood. 99, 2408-17.

286. Pereanu, W., Shy, D., Hartenstein, V. (2005) Morphogenesis and proliferation of the larval brain glia in Drosophila. Dev Biol. 283, 191-203.

287. Perez, S. E., Steller, H. (1996). Migration of glial cells into retinal axon target field in Drosophila melanogaster. J Neurobiol. 30, 359-73.

288. Perry, V. H., Linden, R. (1982). Evidence for dendritic competition in the developing retina. Nature. 297, 683-5.

289. Petrovic, M., Hummel, T. (2008). Temporal identity in axonal target layer recognition. Nature. 456, 800-3.

333

290. Pettitt, J. (2005). The cadherin superfamily. WormBook. 29, 1-9.

291. Poeck, B., et al., (2001). Glial cells mediate target layer selection of retinal axons in the developing visual system of Drosophila. Neuron. 29, 99-113.

292. Pogson, T. C. (1974). Guanine nucleotides and their significance in biochemical processes. Am J Clin Nutr 27, 380-402.

293. Poinat, P., et al., (2002). A conserved interaction between beta1 integrin/PAT-3 and Nck-interacting kinase/MIG-15 that mediates commissural axon navigation in C. elegans. Curr Biol. 12, 622-31.

294. Polleux, F., et al., (1998). Patterning of cortical efferent projections by semaphorin- neuropilin interactions. Science. 282, 1904-6.

295. Polleux, F., Morrow, T., Ghosh, A. (2000). Semaphorin 3A is a chemoattractant for cortical apical dendrites. Nature. 404, 567-73.

296. Prakash, S., et al., (2005). Drosophila N-cadherin mediates an attractive interaction between photoreceptor axons and their targets. Nat Neurosci. 8, 443-50.

297. Prokop, A., et al., (1998). The kakapo mutation affects terminal arborization and central dendritic sprouting of Drosophila motorneurons. J Cell Biol. 143, 1283-94.

298. Pulford, K., Morris, S. W., Turturro, F. (2004). Anaplastic lymphoma kinase proteins in growth control and cancer. J Cell Physiol. 199, 330-58.

299. Pulido, D., et al., (1992). Dtrk, a Drosophila gene related to the trk family of neurotrophin receptors, encodes a novel class of neural cell adhesion molecule. EMBO J. 11, 391-404.

300. Rajagopalan, S., et al., (2000a). Selecting a longitudinal pathway: robo receptors specify the lateral position of axons in the Drosophila CNS. Cell 103, 1033-45.

301. Rajagopalan, S., et al., (2000b). Crossing the midline: roles and regulation of Robo receptors. Neuron 28, 767-77.

302. Rakic, P., Sidman, R. L. (1973). Organization of cerebellar cortex secondary to deficit of granule cells in weaver mutant mice. J Comp Neurol. 152, 133-61.

303. Rangarajan, R., Courvoisier, H., Gaul, U. (2001). Dpp and Hedgehog mediate neuron-glia interactions in Drosophila eye development by promoting the proliferation and motility of subretinal glia. Mech Dev. 108, 93-103.

304. Rangarajan, R., Gong, Q., Gaul, U. (1999). Migration and function of glia in the developing Drosophila eye. Development. 126, 3285-92.

334

305. Rao Y, et al., (2000). brakeless is required for photoreceptor growth-cone targeting in Drosophila. Proc Natl Acad Sci U S A. 97, 5966-71.

306. Rao, Y., Zipursky, S. L. (1998). Domain requirements for the Dock adapter protein in growth- cone signaling. Proc Natl Acad Sci U S A. 95, 2077-82.

307. Raper, J. A., Bastiani, M., Goodman, C. S. (1983) Pathfinding by neuronal growth cones in grasshopper embryos. II. Selective fasciculation onto specific axonal pathways. J Neurosci. 3, 31-41.

308. Rawson, J. M., et al., (2005). The heparan sulfate proteoglycans Dally-like and Syndecan have distinct functions in axon guidance and visual-system assembly in Drosophila. Curr Biol. 15, 833-8.

309. Read, D., Manley, J. L. (1992). Alternatively spliced transcripts of the Drosophila tramtrack gene encode zinc finger proteins with distinct DNA binding specificities, EMBO J. 11, 1035-44.

310. Ready, D. F. (1989). A multifaceted approach to neural development. Trends Neurosci. 12, 102-10.

311. Redmond, L., Ghosh, A. (2005). Regulation of dendritic development by calcium signaling. Cell Calcium 37, 411-78.

312. Redmond, L., Kashani, A. H., Ghosh, A. (2002). Calcium regulation of dendritic growth via CaM kinase IV and CREB-mediated transcription. Neuron 34, 999-78.

313. Reinke, R., Zipursky, S. L. (1988). Cell-cell interaction in the Drosophila retina: the bride of sevenless gene is required in photoreceptor cell R8 for R7 cell development. Cell. 55, 321-30.

314. Rørth, P., et al., (1998). Systematic gain-of-function genetics in Drosophila. Development 125, 1049-57.

315. Rothberg, J. M., et al., (1988). slit: an EGF-homologous locus of Drosophila melanogaster involved in the development of the embryonic central nervous system. Cell 55, 1047-59.

316. Rothberg, J. M., et al., (1990). slit: an extracellular protein necessary for development of midline glia and commissural axon pathways contains both EGF and LRR domains. Genes Dev. 4, 2169-87.

317. Ruan, W., et al., (2002). Bifocal is a downstream target of the Ste20-like serine/threonine kinase misshapen in regulating photoreceptor growth cone targeting in Drosophila. Neuron. 36, 831-42.

335

318. Ruan, W., Pang, P., Rao, Y. (1999). The SH2/SH3 adaptor protein dock interacts with the Ste20-like kinase misshapen in controlling growth cone motility. Neuron. 24, 595-605.

319. Ruan, Y. W., et al., (2006). Dendritic plasticity of CA1 pyramidal neurons after transient global ischemia. Neuroscience 140, 191-78.

320. Salecker, I., Clandinin, T. R., Zipursky, S. L. (1998) Hedgehog and Spitz: making a match between photoreceptor axons and their targets. Cell. 95, 587-90.

321. Sato, M., et al., (2006). DWnt4 regulates the dorsoventral specificity of retinal projections in the Drosophila melanogaster visual system. Nat Neurosci. 9, 67-75.

322. Satoh, D., et al., (2008). Spatial control of branching within dendritic arbors by dynein-dependent transport of Rab5-endosomes. Nat Cell Biol. 10, 1164-71.

323. Sawaya, M. R., et al., (2008). A double S shape provides the structural basis for the extraordinary binding specificity of Dscam isoforms. Cell. 134, 1007-18.

324. Schambony, A., Wedlich, D. (2007). Wnt-5A/Ror2 regulate expression of XPAPC through an alternative noncanonical signaling pathway. Dev Cell 12, 779-92.

325. Schlessinger, J., (1994). SH2/SH3 signaling proteins. Curr. Opin. Genet. Dev. 4, 25– 30.

326. Schmid, A., Chiba, A., Doe, C. Q. (1999). Clonal analysis of Drosophila embryonic neuroblasts: neural cell types, axon projections and muscle targets. Development. 126, 4653-89.

327. Schmitt, A. M., et al., (2006). Wnt-Ryk signalling mediates medial-lateral retinotectal topographic mapping. Nature. 439, 31-7.

328. Schmucker, D., et al., (2000). Drosophila Dscam is an axon guidance receptor exhibiting extraordinary molecular diversity. Cell. 101, 671-84.

329. Schubiger, M., et al., (1998). Drosophila EcR-B ecdysone receptor isoforms are required for larval molting and for neuron remodeling during metamorphosis. Development 125, 2053-78.

330. Scott, E. K., Luo, L. (2001). How do dendrites take their shape? Nat Neurosci. 4, 359-65.

331. Scott, E.K., Lee, T., Luo, L. (2001). enok encodes a Drosophila putative histone acetyltransferase required for mushroom body neuroblast proliferation. Curr. Biol. 11, 99-104.

336

332. Selleck, S. B., Steller, H. (1991). The influence of retinal innervation on neurogenesis in the first optic ganglion of Drosophila. Neuron. 6, 83-99.

333. Senti K,et al., (2000). brakeless is required for lamina targeting of R1-R6 axons in the Drosophila visual system. Development. 127, 2291-301.

334. Senti, K. A., et al., (2003). Flamingo regulates R8 axon-axon and axon-target interactions in the Drosophila visual system. Curr Biol. 13, 828-32.

335. Serpe, M., and O‟Connor M. B. (2006). The metalloprotease tolloid-related and its TGF-beta-like substrate Dawdle regulate Drosophila motoneuron axon guidance. Development 133, 4969-79.

336. Serra-Pagès, C., et al., (1995). The LAR transmembrane protein tyrosine phosphatase and a coiled-coil LAR-interacting protein co-localize at focal adhesions. EMBO J. 14, 2827-38.

337. Serra-Pagès, C., et al., (1998). Liprins, a family of LAR transmembrane protein- tyrosine phosphatase-interacting proteins. J Biol Chem. 273, 15611-20.

338. Shi, S. H., et al., (2004a). Control of dendrite arborization by an Ig family member, dendrite arborization and synapse maturation 1 (Dasm1). Proc Natl Acad Sci U S A 101, 13341-5.

339. Shi, S. H., et al., (2004b). The immunoglobulin family member dendrite arborization and synapse maturation 1 (Dasm1) controls excitatory synapse maturation. Proc Natl Acad Sci U S A 101, 13346-51.

340. Shi, Y., Massagué, J. (2003). Mechanisms of TGF-beta signaling from cell membrane to the nucleus. Cell. 113, 685-700.

341. Shima, Y., et al., (2004). Regulation of dendritic maintenance and growth by a mammalian 7-pass transmembrane cadherin. Dev Cell. 7, 205-16.

342. Shinza-Kameda, M., et al., (2006). Regulation of layer-specific targeting by reciprocal expression of a cell adhesion molecule, capricious. Neuron. 49, 205-13.

343. Shishido, E., Takeichi, M., Nose, A. (1998). Drosophila synapse formation: regulation by transmembrane protein with Leu-rich repeats, CAPRICIOUS. Science. 280, 2118-21.

344. Siegfried, E., Perrimon, N. (1994). Drosophila wingless: a paradigm for the function and mechanism of Wnt signaling. Bioessays. 16, 395-404.

345. Simpson, J. H., et al., (2000a). Short-range and long-range guidance by Slit and its Robo receptors: robo and Robo2 play distinct roles in midline guidance. Neuron 28, 753-66. 337

346. Simpson, J. H., et al., (2000b). Short-range and long-range guidance by Slit and its Robo receptors: a combinatorial code of Robo receptors controls lateral position. Cell. 103, 1019-32.

347. Sin, W. C., et al., (2002). Dendrite growth increased by visual activity requires NMDA receptor and Rho GTPases. Nature 419, 475-78.

348. Sisson, J. C., et al., (2000). Lava lamp, a novel peripheral golgi protein, is required for Drosophila melanogaster cellularization. J Cell Biol. 151, 905-18.

349. Sluss, H. K., et al., (1996). A JNK signal transduction pathway that mediates morphogenesis and an immune response in Drosophila. Genes Dev. 10, 2745-58.

350. Soba, P., et al., (2007). Drosophila sensory neurons require Dscam for dendritic self- avoidance and proper dendritic field organization. Neuron 54, 403-416.

351. Sone, M., et al., (1997). Still life, a protein in synaptic terminals of Drosophila homologous to GDP-GTP exchangers. Science. 275, 543-7.

352. Song, H., et al., (1998). Conversion of neuronal growth cone responses from repulsion to attraction by cyclic nucleotides. Science. 281, 1515-8.

353. Song, J., et al., (2003). Axons guided by insulin receptor in Drosophila visual system. Science. 300, 502-5.

354. Song, W., et al., (2007). Peripheral multidendritic sensory neurons are necessary for rhythmic locomotion behavior in Drosophila larvae. Proc Natl Acad Sci U S A.104, 5199-204.

355. Spigelman, I., et al., (1998). Dentate granule cells form novel basal dendrites in a rat model of temporal lobe epilepsy. Neuroscience 86, 109-78.

356. Steimel, A., Wacker, I., Hutter, H. (2008). Abstract: The C. elegans flamingo homologue FMI-1 is involved in pioneer-mediated axon guidance in the ventral nerve cord. Developmental Biology 319, 539-543.

357. Steinel, M. C., Whitington, P. M. (2009). The atypical cadherin Flamingo is required for sensory axon advance beyond intermediate target cells. Dev Biol. 327, 447-57.

358. Stone, M. C., Roegiers, F., Rolls, M. M. (2008). Microtubules have opposite orientation in axons and dendrites of Drosophila neurons. Mol Biol Cell. 19, 4122-9.

359. Streuli, M., et al., (1989). A family of receptor-linked protein tyrosine phosphatases in humans and Drosophila. Proc. Natl. Acad. Sci. U.S.A. 86, 8698-02.

360. Strutt, D. (2003). Frizzled signalling and cell polarisation in Drosophila and vertebrates. Development. 130, 4501-13.

338

361. Strutt, H., Strutt, D. (1999). Polarity determination in the Drosophila eye. Curr Opin Genet Dev. 9, 442-6.

362. Strutt, H., Strutt, D. (2008). Differential stability of flamingo protein complexes underlies the establishment of planar polarity. Curr Biol. 18, 1555-64.

363. Stute, C., et al., (2004). Myoblast determination in the somatic and visceral mesoderm depends on Notch signalling as well as on milliways(mili(Alk)) as receptor for Jeb signalling. Development. 131, 743-54.

364. Su, Y. C., et al., (1997). NIK is a new Ste20-related kinase that binds NCK and MEKK1 and activates the SAPK/JNK cascade via a conserved regulatory domain. EMBO J. 16, 1279-90.

365. Su, Y. C., et al., (2000). The Ste20 kinase misshapen regulates both photoreceptor axon targeting and dorsal closure, acting downstream of distinct signals. Mol Cell Biol. 20, 4736-44.

366. Su, Y., Treisman, J. E., Skolnik, E. Y. (1998). The Drosophila STE20-related kinase misshapen is required for embryonic dorsal closure and acts through a JNK MAPK module on an evolutionary conserved signaling pathway. Genes Dev, 12, 2371-80.

367. Sugimura, K., et al., (2003). Distinct developmental modes and lesion-induced reactions of dendrites of two classes of Drosophila sensory neurons. J. Neurosci. 23, 3752-78.

368. Sugimura, K., et al., (2004). Development of morphological diversity of dendrites in Drosophila by the BTB-zinc finger protein abrupt. Neuron 43, 809-22.

369. Suh, G. S., et al., (2002). Drosophila JAB1/CSN5 acts in photoreceptor cells to induce glial cells. Neuron. 33, 35-46.

370. Sun, Q., et al., (2000). Receptor tyrosine phosphatases regulate axon guidance across the midline of the Drosophila embryo. Development. 127, 801-12.

371. Sun, Q., et al., (2001). Complex genetic interactions among four receptor tyrosine phosphatases regulate axon guidance in Drosophila. Mol Cell Neurosci. 17, 274-91.

372. Sweeney, N. T., Li, W., Gao, F. B. (2002). Genetic manipulation of single neurons in vivo reveals specific roles of flamingo in neuronal morphogenesis. Dev Biol. 247, 76-88.

373. Sweeney, S. T., et al., (1995). Targeted expression of tetanus toxin light chain in Drosophila specifically eliminates synaptic transmission and causes behavioral defects. Neuron. 14, 341-51.

339

374. Tabata, T., Takei, Y. (2004). Morphogens, their identification and regulation. Development. 131, 703-12.

375. Tailby, C., et al., (2005). Activity-dependent maintenance and growth of dendrites in adult cortex. Proc. Natl. Acad. Sci. USA 102, 4631-78.

376. Takagi, S., et al., (1991). The A5 antigen, a candidate for the neuronal recognition molecule, has homologies to complement components and coagulation factors. Neuron. 7, 295-307.

377. Tao, X., et al., (1998). Ca2+ influx regulates BDNF transcription by a CREB family transcription factor-dependent mechanism. Neuron 20, 709-78.

378. Tatar, M., et al., (2001). A mutant Drosophila insulin receptor homolog that extends life-span and impairs neuroendocrine function. Science. 292, 107-10.

379. Tayler, T. D., Garrity, P. A. (2003). Axon targeting in the Drosophila visual system. Curr Opin Neurobiol. 13, 90-5.

380. Tayler, T. D., Robichaux, M. B., Garrity, P. A. (2004). Compartmentalization of visual centers in the Drosophila brain requires Slit and Robo proteins. Development. 131, 5935-45.

381. Tessier-Lavige, M., Goodman, C. S. (1996). The molecular biology of axon guidance. Science. 274, 1123-33.

382. Thomas, J. B., et al., (1984). From grasshopper to Drosophila: a common plan for neuronal development. Nature. 310, 203-7.

383. Tian, S.S., Tsoulfas, P., Zinn, K. (1991). Three receptor-linked protein-tyrosine phosphatases are selectively expressed on central nervous system axons in the Drosophila embryo. Cell 67, 675-80.

384. Ting, A. Y., et al., (2001). Genetically encoded fluorescent reporters of protein tyrosine kinase activities in living cells, Proc. Natl. Acad. Sci. USA 98, 15003–8.

385. Ting, C. Y., et al., (2005). Drosophila N-cadherin functions in the first stage of the two-stage layer-selection process of R7 photoreceptor afferents. Development. 132, 953-63.

386. Ting, C. Y., et al., (2007). Tiling of R7 axons in the Drosophila visual system is mediated both by transduction of an activin signal to the nucleus and by mutual repulsion. Neuron. 56, 793-806.

387. Tissir, F., et al., (2005). Protocadherin Celsr3 is crucial in axonal tract development. Nat Neurosci. 8, 451-7.

340

388. Tomasi, T., et al., (2008). The transmembrane protein Golden goal regulates R8 photoreceptor axon-axon and axon-target interactions. Neuron 57, 691-704.

389. Tracey, W. D. Jr., et al., (2003). painless, a Drosophila gene essential for nociception. Cell. 113, 261-73.

390. Treisman, J. E., Ito, N., Rubin, G. M. (1997). misshapen encodes a protein kinase involved in cell shape control in Drosophila. Gene 186, 119-25.

391. Truman, J. W., et al., (1994). Ecdysone receptor expression in the CNS correlates with stage-specific responses to ecdysteroids during Drosophila and Manduca development. Development 120, 219-78.

392. Umetsu, D., et al., (2006). The highly ordered assembly of retinal axons and their synaptic partners is regulated by Hedgehog/Single-minded in the Drosophila visual system. Development. 133, 791-800.

393. Usui, T., et al., (1999). Flamingo, a seven-pass transmembrane cadherin, regulates planar cell polarity under the control of Frizzled. Cell. 98, 585-95.

394. Vaillant, A. R., et al., (2002). Signaling mechanisms underlying reversible, activity- dependent dendrite formation. Neuron 34, 985-78.

395. van Amerongen, R., Mikels, A., Nusse, R. (2008). Alternative wnt signaling is initiated by distinct receptors. Sci Signal. 1, re9.

396. Van Vactor, D., et al., (1993). Genes that control neuromuscular specificity in Drosophila. Cell 73, 1137-53.

397. Vanderhaeghen, P., Polleux, F. (2004). Developmental mechanisms patterning thalamocortical projections: intrinsic, extrinsic and in between. Trends Neurosci.27, 384-91

398. Vernersson, E., et al., (2006). Characterization of the expression of the ALK receptor tyrosine kinase in mice. Gene Expr Patterns. 6, 448-61.

399. Vuksić, M., et al., (2002). Perinatal growth of prefrontal layer III pyramids in Down syndrome. Pediatr Neurol. 27, 36-8.

400. Wandall, H. H., et al., (2003). Drosophila egghead encodes a beta 1,4- mannosyltransferase predicted to form the immediate precursor glycosphingolipid substrate for brainiac. J Biol Chem. 278, 1411-4.

401. Wandall, H. H., et al., (2005). Egghead and brainiac are essential for glycosphingolipid biosynthesis in vivo. J Biol Chem. 280, 4858-63.

341

402. Wang, J., et al., (2002). Drosophila Dscam is required for divergent segregation of sister branches and suppresses ectopic bifurcation of axons. Neuron. 33, 559-71.

403. Wang, Y., et al., (2005). Visualizing the mechanical activation of Src. Nature 434, 1040-45.

404. Wässle, H., Peichl, L., Boycott, B. B. (1981). Dendritic territories of cat retinal ganglion cells. Nature. 292, 344-5.

405. Watson, F. L., et al., (2005). Extensive diversity of Ig-superfamily proteins in the immune system of insects. Science. 309, 1874-8.

406. Wayman, G. A., et al., (2006). Activity-dependent dendritic arborization mediated by CaM-kinase I activation and enhanced CREB-dependent transcription of Wnt-2. Neuron 50, 897-78.

407. Weake, V. M., et al., (2008). SAGA-mediated H2B deubiquitination controls the development of neuronal connectivity in the Drosophila visual system. EMBO J. 27, 394-405.

408. Weiss, J. B., et al., (2001). Jelly belly: a Drosophila LDL receptor repeat-containing signal required for mesoderm migration and differentiation. Cell. 107, 387-98.

409. Williams, D. W., et al., (2006). Local caspase activity directs engulfment of dendrites during pruning. Nat. Neurosci.9, 1234-6.

410. Williams, D. W., Truman, J. W. (2005a). Cellular mechanisms of dendrite pruning in Drosophila: insights from in vivo time-lapse of remodeling dendritic arborizing sensory neurons. Development 132, 3631-42.

411. Williams, D. W., Truman, J. W. (2005b). Remodeling dendrites during insect metamorphosis. J. Neurobiol. 64, 24-33.

412. Winberg, M. L., et al., (2001). The transmembrane protein off-track associates with plexins and functions downstream of semaphorin signaling during axon guidance. Neuron 32, 53-62.

413. Winckler, B., et al., (1999). A diffusion barrier maintains distribution of membrane proteins in polarized neurons. Nature 397, 698-701.

414. Wodarz, A., Nusse, R. (1998). Mechanisms of Wnt signaling in development. Annu Rev Cell Dev Biol. 14, 59-88.

415. Wojtowicz, W. M., et al., (2004). Alternative splicing of Drosophila Dscam generates axon guidance receptors that exhibit isoform-specific homophilic binding. Cell. 118, 619-33.

342

416. Wojtowicz, W. M., et al., (2007). A vast repertoire of Dscam binding specificities arises from modular interactions of variable Ig domains. Cell. 130, 1134-45.

417. Wong, W. T., et al., (2000). Rapid dendritic remodeling in the developing retina: dependence on neurotransmission and reciprocal regulation by Rac and Rho. J Neurosci. 20, 5024-36.

418. Wrana, J. L., et al., (1994). Two distinct transmembrane serine/threonine kinases from Drosophila melanogaster form an activin receptor complex. Mol. Cell. Biol. 14, 944-50.

419. Wu, G. Y., Cline, H. T. (1998). Stabilization of dendritic arbor structure in vivo by CaMKII. Science 279, 222-6.

420. Wu, G. Y., Deisseroth, K., Tsien, R. W. (2001). Activity-dependent CREB phosphorylation: convergence of a fast, sensitive calmodulin kinase pathway and a slow, less sensitive mitogen-activated protein kinase pathway. Proc. Natl. Acad. Sci. USA 98, 2808-78.

421. Wu, Q., Maniatis, T. (2000). Large exons encoding multiple ectodomains are a characteristic feature of protocadherin genes. Proc Natl Acad Sci U S A. 97, 3124-9.

422. Xiao, Z., Latek, R., Lodish, H. F. (2003). An extended bipartite nuclear localization signal in Smad4 is required for its nuclear import and transcriptional activity. Oncogene. 22, 1057-69.

423. Xu, B., et al., (2000). The role of brain-derived neurotrophic factor receptors in the mature hippocampus: modulation of long-term potentiation through a presynaptic mechanism involving TrkB. J Neurosci. 20, 6888-97.

424. Xu, T. and Rubin, G. M. (1993). Analysis of genetic mosaics in developing and adult Drosophila tissues. Development 117, 1223-37.

425. Xu, T., et al., (1995). Identifying tumor suppressors in genetic mosaics: the Drosophila lats gene encodes a putative protein kinase. Development. 121, 1053-63.

426. Yacoubian, T. A., Lo, D. C. (2000). Truncated and full-length TrkB receptors regulate distinct modes of dendritic growth. Nat Neurosci. 3, 342-9.

427. Yamagata, M., Sanes, J. R. (2008). Dscam and Sidekick proteins direct lamina- specific synaptic connections in vertebrate retina. Nature. 451, 465-9.

428. Yamagata, M., Weiner, J. A., Sanes, J. R. (2002). Sidekicks: synaptic adhesion molecules that promote lamina-specific connectivity in the retina. Cell. 110, 649-60.

429. Yang, C. H., et al., (2008). Drosophila egg-laying site selection as a system to study simple decision-making processes. Science 319, 1679-83. 343

430. Yang, P., et al., (2000). Abnormal turning behavior in Drosophila larvae. Identification and molecular analysis of scribbler (sbb). Genetics. 155, 1161-74.

431. Ye, B., et al., (2007). Growing dendrites and axons differ in their reliance on the secretory pathway. Cell 130,717-29.

432. Yonekura, S., et al., (2006). The variable transmembrane domain of Drosophila N- cadherin regulates adhesive activity. Mol Cell Biol. 26, 6598-608.

433. Yonekura, S., et al., (2007). Adhesive but not signaling activity of Drosophila N- cadherin is essential for target selection of photoreceptor afferents. Dev Biol. 304, 759-70.

434. Yoshida, S., et al., (2005). DPP signaling controls development of the lamina glia required for retinal axon targeting in the visual system of Drosophila. Development. 132, 4587-98.

435. Yoshikawa, S., et al., (2003). Wnt-mediated axon guidance via the Drosophila Derailed receptor. Nature. 422, 583-8.

436. Younossi-Hartenstein, A., et al., (1996). Early neurogenesis of the Drosophila brain. J. Comp. Neurol. 370,313-29.

437. Younossi-Hartenstein, A., Salvaterra, P. M., Hartenstein, V. (2003) Early development of the Drosophila brain: IV. Larval neuropile compartments defined by glial septa. J Comp Neurol. 455, 435-50.

438. Zaccolo, M., Pozzan, T. (2002). Discrete microdomains with high concentration of cAMP in stimulated rat neonatal cardiac myocytes. Science 295, 1711–15.

439. Zalkin, H., and Dixon, J. E. (1992). De novo purine nucleotide biosynthesis. Prog Nucleic Acid Res Mol Biol 42, 259-87.

440. Zhan, X. L., et al., (2004). Analysis of Dscam diversity in regulating axon guidance in Drosophila mushroom bodies. Neuron. 43, 673-86.

441. Zhang, J., et al., (2004). Zebrafish unplugged reveals a role for muscle-specific kinase homologs in axonal pathway choice. Nat Neurosci. 7, 1303-9.

442. Zheng, X., et al., (2003). TGF-beta signaling activates steroid hormone receptor expression during neuronal remodeling in the Drosophila brain. Cell 112,303-315.

443. Zheng, Y., et al., (2008). Dynein is required for polarized dendritic transport and uniform microtubule orientation in axons. Nat Cell Biol. 10, 1172-80.

344

444. Zhu, C. C., et al., (2008). Drosophila Activin- and the Activin-like product Dawdle function redundantly to regulate proliferation in the larval brain. Development 135, 513-21.

445. Zhu, H., et al., (2006). Dendritic patterning by Dscam and synaptic partner matching in the Drosophila antennal lobe. Nat Neurosci. 9, 349-55.

446. Zhu, S., Chiang, A. S., Lee, T. (2003). Development of the Drosophila mushroom bodies: elaboration, remodeling and spatial organization of dendrites in the calyx. Development 130, 2603-78.

447. Zollner, N. (1982). Purine and pyrimidine metabolism. Proc Nutr Soc 41, 329-342.

448. Zong, H., et al., (2005). Mosaic analysis with double markers in mice. Cell. 121, 479-92.

345