Mechanisms in dendrite pruning of drosophila dendritic arborization neurons

i



MECHANISMS UNDERLYING DENDRITE PRUNING OF
DROSOPHILA DENDRITIC ARBORIZATION NEURONS




GU YING
(B. Sci., Sichuan University)







A THESIS SUBMITTED FOR THE DEGREE
OF DOCTOR OF PHILOSOPHY
DEPARTMENT OF BIOLOGICAL SCIENCES
NATIONAL UNIVERSITY OF SINGAPORE
2009
ii

To my parents and grandparents
iii

Acknowledgement
I am heartily thankful to my supervisor, Dr. Fengwei Yu, whose encouragement,
guidance and support enabled me to explore any possibilities in my research work. His
great enthusiasm in science is always stimulating to me. I am also grateful to Assoc. Prof.
Boon Chuan Low for his willingness to co-supervise me from the very beginning of my
study and support on my work.

I also would like to thank my graduate committee members, Prof. William Chia,
Dr. Suresh Jeuthasan, Dr. Sudipto Roy and Dr. Yih Cherng Liou for their support and
advice.

I gratefully thank all people in Yu’s group for providing a motivating, enthusiastic
and critical atmosphere for my work, especially Daniel Kirilly for his willingness to share
his bright thoughts with me and assistance in various ways.

Many thanks also go to Dr. Hongyan Wang and her group for their discussion, in
particular to Hongyan, Nick Bogard and Wei Leong Chew for their constructive
comments on this thesis.

I owe my deepest gratitude to Dr. Arash Bashirullah (UW-Madison), Prof. Alex
Kolodkin (HHMI, Johns Hopkins University) and a broader fly community for their
generosity in sharing reagents and flies.

My gratitude also goes to the supporting staffs and my friends at Temasek Life

Sciences Laboratory for their sincere help. And lastly, my parents and grandparents, for
their love.



iv

Table of Contents
Acknowledgement iii
Table of Contents iv
Summary vii
List of Publications viii
List of Tables ix
List of Figures x
List of Abbreviations xii
Chapter One: Literature Review 1
1.1 Introduction 1
1.2 Neuronal pruning 2
1.2.1 Developmental pruning in vertebrates 3
1.2.1.1 Trophic-factor dependent axon pruning 3
1.2.1.2 Axon guidance molecules in developmental axon pruning 4
1.2.1.3 Developmental dendrite pruning in vertebrates 5
1.2.2 Two systems in Drosophila to study developmentally occurring neuronal
remodeling 6
1.3 Mechanisms in regulating neuronal pruning in Drosophila 9
1.3.1 Transcriptional regulation of neuronal pruning during metamorphosis 9
1.3.2 Ubiquitin-proteasome System 14
1.3.3 Caspase and neuronal pruning 15
1.3.4 IKK-related kinase IK2, cytoskeleton and dendrite severing 16
1.4 Aim of this study 16

Chapter Two: Materials and Methods 18
2.1 Fly Strains 18
2.2 Genetic mapping 19
2.3 Microscopy and image acquisition and quantification 19
2.4 MARCM labeling 20
2.5 Fluorescence in situ Hybridization 21
2.5.1 Primer design for DNA template 21
2.5.2 In vitro transcription of the probe 21
2.5.3 In situ hybridization 22
2.6 Immunohistochemistry 23
2.7 DNA manipulations 23
2.7.1 Escherichia. coli culture and transformation 23
v

2.7.2 Molecular cloning 24
2.7.3 DNA sequencing 24
2.7.4 Genomic DNA extraction 25
2.7.5 mical promoter-lacZ reporter plasmid constructs 26
2.7.6 Mical domain deletion plasmid constructs 27
2.7.7 Mical single domain plasmid constructs 28
2.8 Preparation of whole animal lysates, SDS-PAGE and western blot 29
Chapter Three: Results 31
3.1 Dendrite remodeling of ddaC neurons during metamorphosis 31
3.2 Forward genetic screen for novel players in ddaC dendrite pruning 32
3.3 Mical regulates dendrite pruning of dendritic arborization neurons in
Drosophila 35
3.3.1 Mical is affected in l(3)15256 with strong dendrite severing defects 35
3.3.2 Mical promotes dendrite severing of ddaCs 36
3.3.3 Cell-autonomous function of Mical for dendrite severing 38
3.3.4 Time-course analysis of EcR and Mical in dendrite severing 41

3.3.5 Temporal expression pattern of EcR-B1 and Mical 43
3.3.6 Identification of ecdysone response element(s) in the mical regulatory
region ……………………………………………………………………………48
3.3.7 Mical does not affect EcR-B1 expression 54
3.3.8 Mical is a crucial factor downstream of EcR-B1 to promote dendrite
severing 56
3.3.9 Mical and EcR-B1 are insufficient at early stage to cause premature
pruning 58
3.3.10 Functional analysis of Mical domains in dendrite severing 60
3.3.11 Cytoskeleton rearrangement in mical
15256
during dendrite pruning 64
3.4 Dronc and Mical regulate different cellular responses to EcR-B1 68
3.4.1 Dronc is not required for dendrite severing of ddaCs 68
3.4.2 Mical is not required for cell death of apoptotic neurons 73
3.5 Plexin/Semaphorin pathway is not required for ddaC dendrite pruning 73
3.6 Candidate gene analysis in dendrite pruning 77
Chapter Four: Discussions 80
4.1 Questions about the developmentally regulated neuronal remodeling in
Drosophila 80
4.2 Developmental regulation of Mical expression during dendrite pruning 81
4.3 Mical or Dronc in regulating dendrite severing 84
4.4 How Mical regulates dendrite pruning and future directions 85
Chapter Five: Conclusions 91
Bibliography 93
Appendix i 104
vi

Appendix ii 105

vii

Summary
The capability of neurons to remodel existing neuronal projections and connections
confers great flexibility in response to activity-dependent processes, developmental
regulated alterations, neuronal diseases and post-injury recoveries. Although a wide range
of events lead to neuronal remodeling, the underlying mechanisms remain elusive.
Among various types of neuronal remodeling, selective removal of dendrite branches, so
called dendrite pruning, of Drosophila dendritic arborization (da) neurons occurs during
metamorphosis, a developmental process that transforms a ‘worm-like’ larva into an adult
fruit fly. To understand the mechanisms that regulate dendrite pruning of these peripheral
neurons, a forward genetic screen was carried out and identified Mical (Molecule
interacting with CasL) as a novel factor that promotes severing of dendrites at the initial
stage of dendrite pruning. Further studies suggest that destabilization of cytoskeleton
molecules, such as microtubules and actins, is suppressed in remodeling da neurons
devoid of Mical. Mical functions in da neuron pruning downstream of the steroid nuclear
hormone receptor complex EcR-B1/Ultraspirical during the larval-pupal transition;
whereas Dronc (Drosophila Nedd2-like caspase) mediates cell-death of apoptotic da
neurons and clearance of dendrite debris.

viii


List of Publications

Kirilly D*, Gu Y*, Huang Y, Wu Z, Bashirullah A, Low BC, Kolodkin AL, Wang H, Yu
F (2009) A novel pathway composed of Sox14 and Mical governs severing of dendrites
during pruning. Nature Neuroscience 12: 1497‐1505 (*as co-first author)

ix

List of Tables


Table 1. Summary in mapping results of mutant lines with dendrite pruning defects… 34

x

List of Figures
Figure 1. Drosophila dendritic arborization (da) neuron as a model system to study
dendrite remodeling during metamorphosis. 10

Figure 2. Dendrite pruning defects of 15 EMS-induced mutant lines. 33
Figure 3. Mical is required for dendrite severing of ddaCs. 37
Figure 4. Cell-autonomous function of Mical for dendrite severing. 40
Figure 5. Time-course analysis of ddaC pruning behavior in wt, EcR-B1
DN
, usp RNAi and
mical mutant 42

Figure 6. Time-course analysis of class I neuron ddaD/E pruning behavior in wt, EcR-
B1
DN
and mical mutant. 44
Figure 7. Mical expression in ddaCs is dependent on EcR-B1/Usp. 47
Figure 8. Ecdysone-responsive elements in the mical regulatory region. 51
Figure 9. Activation of ecdysone-responsive elements of the mical regulatory region in
MB γ neurons. 53
Figure 10. EcR-B1 expression is not affected by Mical. 55
Figure 11. Mical promotes dendrite severing downstream of EcR-B1. 57
Figure 12. Overexpression of EcR-B1 or Mical by itself is not sufficient to cause
precocious pruning. 59

Figure 13. Mical domain analysis with deletion constructs in mical mutant ddaCs. 61
Figure 14. Mical domain analysis with deletion constructs in wt ddaCs. 63

Figure 15. Mical domain analysis with single domain constructs in wt ddaCs. 65
Figure 16. Cytoskeleton markers in mical
15256
. 66
Figure 17. Dronc is not required for dendrite severing with Mical. 71
Figure 18. Dronc is required for apoptotic neuron cell death but not for remodeling
neuron pruning. 72

Figure 19. Mical is not required for cell death of apoptotic neuron during metamorphosis.
74

xi

Figure 20. Mical partners in axon guidance are not required for ddaC dendrite pruning. 76
Figure 21. Candidate genes for dendrite pruning. 78













xii

List of Abbreviations

APF

after puparium formation
APP

amyloid precursor protein
BDNF

brain-deprived neurotrophic factor
BR-C

broad-complex
CC

coiled-coil
CH

calponin-homolog
CNS

central nervous system
CST

corticospinal tract
da

dendritic arborization
DIAP1

Drosophila inhibitor of apoptosis protein 1
DN

dominant negative
DR6

death receptor 6
Dronc

Drosophila Nedd2-like caspase
EB3

ephrinB3
EcR

ecdysone receptor
EcRE

ecdysone response element
eL3

early 3
rd
instar larva
EMS

ethyl methanesulfonate
EphB

ephrinB receptor
FISH

fluorescent in situ hybridization
FL full length
FM

flavinmonooxygenase
GAP

GTPase-activating proteins
Grb4

growth factor receptor-bound protein 4
HRP

horseradish peroxidase
IC

inferior colliculus
IPB

infra-pyramidal bundles
kb

kilo base
MARCM mosaic analysis with a repressive cell marker
MB

mushroom body

mical

molecule interacting with casL
NMJ

neuromuscular junction
p75NTR

p75 neurotrophin receptor
PC

purkinje cell
PNs

projection neurons
PNS

peripheral nervous system
xiii

RGC

retinal ganglion cells
Rok

Rho kinase
RORα

retinoidrelated orphan receptor alpha
RXR

retinoid X receptor
SC

superior colliculus
SCG

superior cervical ganglion
SH2/3

Src homology 2/3
SH3

Src homology 3
SMC

structural maintenance of chromosome
SPB

supra-pyramidal bundles
TGF-β

transforming growth factor-β
Tv

thoracic ventral neuronsecrectory cells
UPS

ubiquitin-proteasome system
Usp

ultraspiracle
VNC

ventral nerve cord
wL3

wandering 3
rd
instar larva
WP

white pupae
β-gal

β-galactosidase





1

Chapter One: Literature Review
1.1 Introduction
Neurons are highly diversified cells with delicate morphology. After birth, neurons
extend axons and dendrites from soma to form contacts with the extracellular
environment, surrounding neurons and non-neuronal cells. These extensions
(dendrites and axons) and connections (synapses) of each neuron, together with those
of millions of others, build up the overall neuronal circuits of the nervous system.

However, after the initial setup of neuronal territory, neurons more or less keep a
certain degree of plasticity and retain the ability to remodel their dendrites, axons or
synapses. The occurrence of neuronal remodeling can be activity-dependent (Hebbar
and Fernandes, 2004; Tessier & Broadie, 2008, 2009) or developmentally
programmed (Reviewed by Williams and Truman, 2005b). Alternatively, remodeling
can be induced by external forces, i.e. injuries and neuronal disorders/diseases
(Reviewed by Luo and O’Leary, 2005; Saxena and Caroni, 2007). The mechanisms
underlying neuronal remodeling likewise inherit such diversity. It can be either
controlled by genetically programmed intrinsic machinery or induced by external
inputs. Therefore, it is a challenge to understand the relevance of the intrinsic
machinery and extrinsic machinery of neuronal remodeling since bathing in an
environment created by neighboring cells, neurons are constantly exposed to a variety
of extracellular stimuli/signals, which nevertheless interact with the intrinsic
machinery to refine neuronal morphology. Due to the diverse reasons for neuronal
remodeling and the complexity of neuronal circuits, our understanding of neuronal
remodeling is poor. Here we mainly focus on the studies of one type of neuronal
remodeling, neuronal pruning.
2

1.2 Neuronal pruning
Neuronal pruning, in which the existing dendrite/axon branches are selectively
degraded without cell death is a fundamental mechanism to sculpt the nervous system
during animal development. The removal of existing neuronal processes can be
achieved either by retraction (Bagri, et al., 2003) or local degeneration (Watts et al.,
2003), or both (Williams and Truman, 2005a; Koirala and Ko, 2004). The scale of
pruning is generally determined by the relative length of pruned processes. The small
scale events usually involve the elimination of synapse or short neuron branches;
while the large-scale pruning, also known as the stereotyped pruning (Bagri et al.,
2003), mediates an extensive removal of entire neuronal projections. Another feature
of the stereotyped pruning is that its occurrence is highly predictable and is precisely

controlled by temporal and spatial cues during neuronal development.

It is well-known that destabilization of the cytoskeleton, for example, microtubules,
occurs in nearly all types of pruning process being described. Furthermore, several
cytoskeleton regulators, such as the small GTPase RhoA and its downstream effector
Rho kinase (Rok), as well as the myosin regulatory chain and its negative regulator
p190RhoGAP (Billuart et al., 2001), have been shown to compromise the axon
stability in vivo. However, due to the pleiotropic function of small GTPases in axon
growth (Billuart et al., 2001; Ng et al., 2002; Ng and Luo 2004), little is known about
how localized activation of these molecules is achieved and what upstream signaling
pathways are, which control neuronal pruning in a precise manner that confines it to
certain types of neurons and developmental time points. However, difficulties in
identifying, recording and manipulating such a dynamic process in vivo impede us
from understanding the underlying mechanisms of developmental pruning.
3

1.2.1 Developmental pruning in vertebrates
1.2.1.1 Trophic-factor dependent axon pruning
Recent studies of the developmental axon pruning in mammals have shed light on the
mechanism underlying this process. Neuronal culture and gene knock-out animals
were utilized for some of these studies. It has been proposed that neuron trophic-
factors may be involved in axon pruning in vivo. The study by Singh, et al. revealed
an axon-competition mechanism in the sympathetic neurons of the superior cervical
ganglion (SCG) in that the activity-dependent secretion of brain-deprived
neurotrophic factor (BDNF) from the winning axons binds to the p75 neurotrophin
receptor (p75NTR) on the losing axons to cause pruning of the latter ones by
suppressing TrkA-mediated signaling that is essential for axonal maintenance (Singh,
et al., 2008). In pursuit of local signals that trigger axon pruning after neurotrophic-
factor withdrawal, the study done by Nikolaev, et al. proposed that the shedding of
cell- surface molecule β-amyloid precursor protein (APP) from axon shafts after

trophic-factor deprivation leads to its binding to the death receptor 6 (DR6) and
thereby triggers axon pruning (Nikolaev, et al, 2009). They further demonstrated that
distinct caspases are required for neuron apoptosis and axon pruning. Activation of
caspase 3 is highly enriched in the cell body of dying neurons, and its inhibition only
protects neuronal death but not axon pruning. While activation of caspase 6 by trophic
deprivation occurs in a punctate pattern in axons and its inhibition protects against
axon degeneration (Nikolaev, et al, 2009). To support the in vitro data, they further
analyzed the pruning of retinal axons in DR6
-/-
mice. During the development of the
retinotopic map of mouse superior colliculus (SC), retinal ganglion cells (RGC)
initially send exuberant axon projection into the posterior region of the SC and
overshoot their future termination zone in anterior SC. These temporal RGC axons are
4

subsequently pruned to generate a more refined map with focused projection into the
termination zone (McLaughlin et al., 2003; Luo and O’Leary, 2005; Nikolaev, et al,
2009). However, in DR6
-/-
mice RGC axons and arbors are present in areas far from
the termination zone, suggesting a defect of axon pruning in these neurons.

1.2.1.2 Axon guidance molecules in developmental axon pruning
Further progress in understanding developmental axon pruning comes from the
identification of several molecules previously known to mediate axon guidance. In
vertebrates, hippocampal mossy fibers, projecting from granule cells of the dentate
gyrus, form two distinct axon bundles, the supra- and infra-pyramidal bundles (SPB
and IPB). SPB travels above the CA3 pyramidal cell layer and makes synaptic
contacts with the apical dendrites of pyramidal cells; while IPB extends below the
CA3 pyramidal cell layer earlier in development and is later shortened/pruned (Bagri

et al., 2003; Liu et al., 2005). Axon guidance receptor Plexin-A3, together with
Neuropilin-2, has been shown to cell-autonomously mediate the stereotypical axon
pruning of IPB mossy fibers. Its ligand Sema3F is expressed in cells along the axon
track and potentially functions as the extracellular signal to initiate axon retraction at
a certain developmental time point (Bagri et al., 2003; Liu et al., 2005). The
dependence of axon pruning on Plexins is not restricted in the hippocampal mossy
fibers but has also been studied in the elimination of axon collaterals during the
refinement of subcortical processes arising from layer V cortical neurons. During
early development, layer V cortical pyramidal neurons in motor and visual regions of
the neocortex send nearly identical corticospinal tract (CST) axon branches to the
spinal cord, superior colliculus (SC) and inferior colliculus (IC). Later on, motor
neurons prune their axons from SC and IC, whereas visual neurons prune their axons
5

from IC and the spinal cord (Low et al., 2008). Surprisingly, Plexin-A3, -A4, and
Neuronpilin-2 selectively regulate the visual but not motor CST axon pruning (Low et
al., 2008), indicating cell-type specific reliance on the Plexin signaling.

Besides Plexins, the ephrin family of axon guidance molecules was proposed to
mediate axon pruning in vitro nearly one decade ago (Gao et al., 1999). However,
direct in vivo evidence comes from a recent study in the pruning of hippocampal
mossy fiber. Xu and Henkemeyer, 2009 showed that the murine EphrinB3 (EB3)
functions as a receptor; and upon tyrosine phosphorylation EB3 signals through an
SH2 (Src homology 2)/SH3 (Src homology 3) domain containing adaptor protein
Grb4 (growth factor receptor-bound protein 4) to mediate IPB axon retraction and the
EphrinB receptor (EphB) molecules serve as the ligands in CA3 postsynaptic
pyramidal neurons to stimulate the EB3 reverse signaling (Xu and Henkemeyer,
2009). However, although Plexins/Semaphorins and Ephrins have been known for a
while to function as axon guidance molecules steering axon growth direction by
collapsing growth cones, several important questions still remain. For example,

whether the axon pruning observed in either hippocampal mossy fiber or CST axons
is due to axon retraction or local degeneration is still an open question.

1.2.1.3 Developmental dendrite pruning in vertebrates
Since an early study of the superior cervical ganglion neruons suggested that dendritic
morphology is constantly changing in adult mice (Purves et al., 1986), previous
investigations of dendrite remodeling have been mainly focused on activity-dependent
changes of dendritic spines instead of large-scale dendrite pruning during animal
development. However, recent studies revealed that the dendritic

differentiation of
6

cerebellar Purkinje cells (PC) in rats involves two successive phases

of development
including both regressive and growth events (Sotelo and Dusart, 2009). And retraction
of the primitive dendritic tree during the early regressive phase requires specific
transcription factors such as retinoidrelated orphan receptor alpha (RORα)
(Boukhtouche et al., 2006). Further studies also identified SCLIP, a microtubule
destabilizing

factor of Stathmin family phosphoproteins, as a crucial factor regulating
PC dendrite retraction and inhibition of SCLIP accelerates the retraction of the
primitive process (Poulain, et al., 2008).
1.2.2 Two systems in Drosophila to study developmentally occurring neuronal
remodeling
Besides the mammalian models mentioned earlier, the Drosophila nervous system has
also proven to be an appealing model to study neuronal remodeling with higher
resolution. In Drosophila, a single neuron can be labeled (Lee and Luo, 2001) and its

morphological changes during development can be traced in real time in vivo
imaging. Moreover, easy genetic manipulation in the fruit fly confers a great
advantage in dissecting the mechanisms of neuronal remodeling.

One fascinating phenomenon of Drosophila life cycle is that the insect goes through a
complete change of its body plans. Within the life cycle, there are two distinct stages
of development: the larval stage and the adult stage. At the larval stage, the animal
takes the form of a worm-like creature, only capable of feeding and crawling;
however, when the animal reaches its adult stage, it develops the ability of flying and
looks like a ‘fruit fly’ as revealed by its well-known name. To achieve this
transformation, it is crucial for the animal to go through a developmental process
called metamorphosis, during which many larval tissues degenerate and most adult-
7

specific tissues or structures are generated from clusters of progenitor cells and the
imaginal discs (Bodenstein, 1965). Tissues undergoing histolysis during the early
stages of metamorphosis include the larval midgut, muscles and salivary gland (Jiang
et al., 1997); and they are eliminated mainly through programmed cell death (Jiang et
al., 1997) or autophagy pathways (Baehrecke, 2003). Cell death also occurs in many
neurons of the larval nervous system (Weeks and Levine, 1990). However, a major
difference between the nervous system and other tissues is that some groups of
functional neurons born at the embryonic or larval stages survive through
metamorphosis and accommodate drastic changes of surrounding tissues. It is
conceivable that these neurons need to eliminate larval connections which are no
longer adapted to the adult environment and re-establish new ones which are
integrated into the functional adult neural circuits. This remodeling process happens
extensively in the nervous system of Drosophila (Truman, 1990), thus making it the
mostly well-described system we know which undergo remodeling during
metamorphosis. Remodeling neurons that have been discovered so far belong to
different functional groups, such as the olfactory projection neurons (PNs) that relay

odor stimuli information in the olfactory circuit (Marin et al., 2005) and the thoracic
ventral neurosecrectory cells (Tv) of the neurohemal organ (Brown et al., 2006).
Besides these two types of remodeling neurons, in Drosophila, there are another two
well-established systems to study the developmentally programmed neuronal
remodeling; one being the axon remodeling of the mushroom body (MB) γ neuron in
the central nervous system (CNS) (Lee et al., 2000), and the other being the dendrite
remodeling of the dendritic arborization (da) neuron in the peripheral nervous system
(PNS) (Williams and Truman, 2005a; Kuo et al., 2005).

8

The MBs in the central brain of Drosophila are essential for learning and memory.
During the larval stages, the axon of γ neurons in MB bifurcates after the peduncle
and extends one branch dorsally and another medially (Lee et al., 1999). However,
during the larval to pupal transition, these two bifurcated axon branches become
destabilized and undergo local degeneration (Watts et al., 2003). Their axonal debris
is removed by glia-mediated phagocytosis within the first 24 h of pupal stage
(Awasaki and Ito, 2004; Awasaki et al., 2006; Watts et al., 2004). Subsequently, the
shortened axons are able to re-grow only medially during late pupal stages and adopt
their adult-specific form, distinct from their larval counterparts (Lee et al., 2000).
Because this process involves rather big changes in the overall axon structure, as well
as the number of neurons, it is referred to as the large-scale remodeling.

Large-scale remodeling of neurites also occur in the fly PNS. In contrast to the CNS γ
neurons that prune both their axons and dendrites, da neurons selectively prune their
dendrites with only minor changes at their axon terminals (Kuo et al., 2005). da
neurons are located peripherally between the epidermal layer and the muscle tissue
layer and contain a single axon and multiple dendrites. The axon projects ventrally to
the ventral nerve cord (VNC) where it slightly branches out to form axon terminals,
while dendrites elaborate mostly two-dimensionally underneath the epidermis. Based

on the complexity of the larval dendritic morphology, da neurons are subdivided into
four classes, namely Class I, II III and IV (Gruber et al., 2002, 2003a, 2003b; Fig 1B).
However, besides the differences in dendrite morphologies, the different classes also
acquire distinct cell fates. In Drosophila larval PNS, a large number of neurons
undergo apoptosis during metamorphosis. For instance, within the dorsal da neuron
cluster, only ddaC (Class IV), ddaD and ddaE (Class I) can survive to the adult stage,
9

while ddaA, ddaF (Class III) and ddaB (Class II) die within the first 4-6 h after
puparium formation (APF) (Fig 1A). For those surviving neurons, their larval
dendrites undergo the remodeling process, including severing, fragmentation and
clearance (Williams and Truman, 2005a; Fig 1D).

Although γ neurons and da neurons selectively remodel their axons or dendrites,
previous studies revealed that some intrinsic mechanisms are shared between both
systems. For instance, the transcriptional regulation mediated by ecdysone signaling
has been shown to control the axon/dendrite pruning events in fly CNS and PNS; and
some components of the protein degradation machinery, the ubiquitin-proteasome
system (UPS), were identified to be involved in axon pruning of γ neurons as well as
dendrite pruning of da neurons.

1.3 Mechanisms in regulating neuronal pruning in Drosophila
1.3.1 Transcriptional regulation of neuronal pruning during metamorphosis
It is reasonable to think that neuron pruning, as a part of the cellular processes of
metamorphosis, shares some common mechanisms with the rest of other processes
occurring during metamorphosis. One potential candidate mechanism is the steroid
hormone-mediated signaling which regulates the transcriptional level control of many
processes during metamorphosis. The steroid hormone 20-hydroxyecdysone (referred
to as ‘ecdysone’ hereafter) is the major hormone that elicits metamorphosis in
Drosophila. It acts through a nuclear receptor heterodimer complex composed of the

steroid hormone ecdysone receptor (EcR) and its co-receptor Ultraspiracle (Usp), the
Drosophila ortholog of Retinoid X receptor (RXR) (Thomas et al., 1993; Yao et al.,
10




Figure 1. Drosophila dendritic arborization (da) neuron as a model system to study
dendrite remodeling during metamorphosis.

(A) Live confocal image of the dorsal da neuron cluster, visualized by the expression of
mCD8-GFP driven by Gal4
109(2)80
. Labeled in red are neurons capable of survival to the adult
stage, labeled in blue are neurons undergoing apoptosis during metamorphosis.

(B) Distinct larval dendritic morphology of the dorsal da neurons. ddaD and E belongs to
Class I; ddaB, Class II; ddaA and ddaF, Class III; ddaC, Class IV.

(C) Live confocal images of ddaC neuron in the time-course study during dendrite pruning.
Dendrite branches of ddaCs are removed from the soma (pointed by arrow heads) within the
first 18 h of metamorphosis. By 8 h APF, most primary dendrites are severed from the soma
at the proximal region. Severed dendrites indicated by empty arrow heads show signs of
beading and fragmentation. By 18h APF, the pruning process is completed with only the
soma and the axon remained intact and all fragmented dendrites cleared. The re-grown
dendritic arbor of ddaCs at 96 h APF shows little resemblance to its WP counterpart.

(D) A diagram of dendrite pruning process. Dorsal is up. Anterior is left. Scale bar, 50 µm.
11

1992; Reviewed by Kozlova and Thummel, 2000). Previous studies indicated that
ecdysone and ecdysone-mediated transcriptional responses are responsible for
programmed cell death of the larval midgut (Jiang et al., 1997). In Drosophila CNS
programmed cell death is also ecdysone-dependent (Robinow et al., 1993). Not
surprisingly, EcRs were found to be required for neuronal remodeling, as well as for
other processes during metamorphosis (Schubiger et al., 1998). However, the EcR
gene in Drosophila encodes three isoforms (Koelle et al., 1991). EcR-A, EcR-B1 and
EcR-B2 share common DNA- and ligand-binding domains but differ in the variable
A/B domain located at the N-terminus (Talbot et al., 1993). To determine which
isoform(s) is specific for mediating neuronal remodeling, several experimental
approaches were used. First, an immunochemistry study with the isoform-specific
antibody revealed that the major isoform expressed in the remodeling neurons is EcR-
B1 (Truman et al., 1994). Secondly the phenotypic analysis of isoform-specific EcR
mutants suggested that EcR-B1/B2 control the remodeling of the nervous system
(Schubiger et al., 1998). Thirdly the ability of EcR-B1/B2 to restore normal neuronal
pruning in EcR mutant strongly supported the specific role of EcR-B1/B2 during
neuronal remodeling (Schubiger et al., 2003; Lee et al., 2000).

In the transcriptional regulation of dendrite pruning in da neurons, the same EcR-
B1/Usp complex is utilized, indicating a common upstream transcriptional control for
both dendrite pruning (Kuo et al., 2005) and axon pruning (Lee et al., 2000).
Interfering with ecdysone signaling by the neuronal overexpression of the dominant
negative form of EcR-B1 (EcR-B1
DN
) can abolish the dendrite pruning process, by
preventing the destabilization of dendritic microtubules and the severing of dendrites.
Loss of usp function also leads to dendrite pruning defects, similar to those of EcR-
12

B1

DN
overexpression (Kuo et al., 2005). Since EcR initiates a series of transcriptional
events in response to the ecdysone pulse at the larval-pupal transition, it is likely that
the alterations in the transcriptional profile in remodeling neurons confer certain
competence on these neurons to undergo pruning. Thus, it is very interesting to
understand the downstream targets of EcR-B1 in the context of remodeling neurons.
The well-known primary response genes after ecdysone induction include
transcription factors Broad-complex (BR-C), E74 and E75 (Burtis et al., 1990;
Fletcher and Thummel, 1995; Segraves and Hogness, 1990; Thummel et al., 1990).
However, the genetic analysis of these genes in MB remodeling suggested that they
are not required for axon pruning (Lee et al., 2000). Therefore, other attempts need to
be made to identify downsteam targets of EcR-B1 in remodeling γ neurons. For
example, Hoopfer et al. carried out a genome-wide analysis of neuron remodeling
using microarray and identified several groups of genes that are induced or suppressed
by ecdysone in MB neurons. Identified genes include transcriptional regulators,
cytoskeleton-binding proteins and components of the programmed cell death
machinery, autophagy as well as the ubiquitin-proteasome system (Hoopfer et al.,
2008). Some of these genes are bona fide targets of ecdysone signaling as supported
by previous studies, such as BR-C and E74, although neither is essential for axon
pruning (Lee et al., 2000; Hoopfer et al., 2008). However, some components of the
UPS were indeed found to be required for axon pruning of MBs in other studies and
will be described in later sections.

Further studies in the MB γ neurons identifed two novel complexes that modulate
axon pruning by regulating EcR-B1 expression. One is TGF-β (transforming growth
factor-β) signaling. Loss of function in components of TGF-β signaling, such as