Molecular mechanisms involved in the specification of embryonic ectoderm in the zebrafish embryo

1

Acknowledgments

I want to express my sincere gratitude to my advisor, Vladimir
Korzh, for accepting me into his laboratory. If this postgraduate program
would be likened to a journey through an exotic land, then I could not have
found a better guide. Encyclopedic in knowledge and skilled in techniques, he
provided an ideal environment for me to mature into a working scientist.
I want to thank members of my thesis advisory committee, Dr.
Thameem Dheen, Dr. Suresh Jesuthasan and Prof. Wan-Jin Hong for their
valuable advice and guidance. Dr. Dheen has laid much of the foundation for
the work described in this thesis. Dr. Suresh and Prof. Hong have influenced
the direction of this project more than they realized.
I thank members of V.K. lab.- Alexander Emelyanov, the resident
guru in all things molecular who laid the foundation for Fz work and provided
me with many of the reagents used in this study; Sergei Parinov, an expert in
transposon-mediated enhancer trap, an enthusiastic diver and a fabulous
photographer, for encouragement and distraction when I needed it; Michael
Richardson, for laying the groundwork on Fz2, and shared my appreciation of
wit, understatement and sarcasm; Cathleen Teh, an expert in zebrafish
electroporation, for making the BAC-recombinant Fz construct. I am also
grateful to Shang-Wei Chong and Kar-Lai Poon, fellow travellers in the exotic
land, for putting up with me. My warmest appreciation for Lee-Thean Chu,
another fellow traveller, for painstakingly proof-reading my manuscripts and
this thesis.
I thank Paramjeet Singh, a dear friend and mentor in my formative
years, for showing me the meaning of science and how it ought to be
conducted. I will never forget his advice to ‘Never let work get in the way of
1

2

progress!’ Many fine ideas were born over coffee or beer with him. And many
other silly ones were shot down by him even before I lifted a pipette, saving
me precious time. He showed me that it’s not how much work I do, but how
much of that work is absolutely essential to answer the question. Most
importantly, is it the right question to begin with.
I thank Chai Chou, fellow lunatic and masochist, for showing me
the meaning of staying the course. In the face of difficulties bordering on the
absurd, he stoically marched on. At times when the going got too tough for
me, I only have to look at him to renew my strength.
I have very fond memories of my time in IGBMC (Strasbourg,
France). I thank Uwe Strähle, Ferenc Müller, Patrick Blader, Thomas
Dickmeis and Nadine Fischer for making me feel at home 2000 km away. I
also thank Stephen Wilson, Miguel Concha and C.P. Heisenberg of the
University of College, London (UCL) for their hospitality and kind tutelage in
the fine art of single cell transplantation.
I am also indebted to R. Moon, S. Vriz and U. Takeda for reagents
and cDNAs. To Bill Chia and Xiao-Hang Yang for discussions during the initial
stages of this work. To Chin-Heng, Amy and everyone in the fish facility;
without them there would be no work. I would also like to thank many of my
colleagues in Temasek Life-Science Laboratory (formally Institute of
Molecular Agrobiology) for support and encouragement. This work was made
possible by the generous support from the Agency for Science, Technology
and Research of Singapore.
Finally, to Kalwant Singh (1960-1992). He’s got a Ticket To
Heaven.
This work would not be complete without the generous
contributions from Sasha, Cathleen and Mike. Thank you all, from my heart.

2


3

Table of Contents

Acknowledgments

1

Summary

5

Publications

6

Chapter 1: Introduction

7

Chapter 2: Wnt signalling mediated by Tbx2b is required for cell migration
during formation of the neural plate

25

2.1: Results

26

2.2: Discussion

50

Chapter 3: Zebrafish Frizzled 2 is required for dorsoventral specification
in the gastrula leading to the formation of posterior structures 60
3.1: Results

61

3.2: Discussion

70

Conclusions

74

Appendix

79

Experimental Procedures

86

References

99

List of Figures
Fig 1. Phenotypes of tbx2b morphants at 24 hpf

27

Fig 2. shh expression in control and tbx2bMO injected embryo

29

Fig 3A. Anti-tbx2bMO blocks transcription of tbx2b mRNA and is
distributed unevenly in the blastula

31

Fig 3B. Anti-tbx2bMO impairs morphogenesis of anterior forebrain

32

Fig 4. Tbx2b is required for cells to adopt a neural fate

34

Fig 5. Tbx2b is required for convergence cell movement

35

Fig 6. ‘Exclusion’ phenotype can be rescued by tbx2b mRNA

38

3


4

Fig 7. Inhibition of Fz7 signalling leads to the ‘exclusion’ phenotype

40

Fig 8. Tbx2b is regulated by Fz7

42

Fig 9. Early function of Tbx2b and Fz7 is required for cell adhesion

46

Fig 10. Cell movement during gastrulation is affected in knockdown
cells

49

Fig 11. Early expression pattern of fz2

63

Fig 12. Fz2 function is required to promote neural fates

65

Fig 13. Transplanted deep cells move to ventral regions

67

Fig 14. Fz2 is necessary for formation of posterior structures

69

Fig 15. Regulation of shh promoter and enhancer elements in COS7
and HeLa cells

85

4


5

Summary

During gastrulation, optimal adhesion and receptivity to signalling
cues are essential for cells to acquire new positions and identities via
coordinated cell movements. T-box transcription factors and the Wnt
signalling pathways play important roles in these processes. Here we show
that Tbx2b is required cell-autonomously for cell adhesion and cell movement
in the embryonic ectoderm. In chimeric embryos, cells deficient in Tbx2b are
defective in cell adhesion and fail to migrate to the neural plate. Using this
‘exclusion’ phenotype as a screen, we show that Tbx2b acts within the context

of Fz7 via components of the canonical Wnt pathway. Independent of cell
movement, Tbx2b is also required for neuronal differentiation. In contrast to
studies in amniotes, our screen failed to demonstrate a role for FGF signalling
in the dorsal movement of embryonic ectoderm leading to neural plate
formation. Instead, our results illustrate the importance of Tbx2b-mediated
Wnt signalling in this process.
We have also fate-mapped a previously undefined population of
deep lying vegetal cells in the zebrafish blastula, and showed that Fz2
function is essential for their migration to the ventral side of the gastrula and
their subsequent incorporation into the posterior structures of the embryo.
This result highlights the early commitment of cell fate during the blastula
period, and links the positions of cells in the gastrula to their earlier positions
in the blastula. Thus, Fz2 is required for the initial specification of the
dorsoventral axis in the gastrula. We further demonstrate that Fz2 is required
for the specification of posterior paraxial mesoderm.

5


6

Publications

Steven Fong, Alexander Emelyanov, Cathleen Teh and Vladimir Korzh. Wnt
signaling mediated by Tbx2b regulates cell migration during formation of the
neural plate (Development, submitted, joint first authorship with Emelyanov).

Sudha Puttur Mudumana, Thomas Dickmeis, Steven Fong, Inna SleptsovaFriedrich, Alexander Emelyanov, Zhiyuan Gong, Uwe Strähle, Vladimir Korzh.
colIXa2 and determination of lineage of non-notochordal precursors of axial
structures of vertebrates (Development, in preparation).

Alexander Emelyanov, Steven Fong, Cathleen Teh, Chen Sok Lam, Vladimir
Korzh Zath3 acts as a proneural determinant in multipotential neural
precursors in hindbrain (Development, in preparation).

Michael Richardson, Steven Fong, Dmitry Bessarab, Alexander Emelyanov,
Vladimir Korzh. Frizzled2 acts downstream of Wnt3a as a determinant of
lateral mesoderm (in preparation, joint first authorship).

Lee Thean Chu, Steven Fong, Dmitry Bessarab, Alexander Emelyanov,
Suresh Jesuthasan, Andrej Minin and Vladimir Korzh. Blastocystic cavity
plays a role in establishing first cell lineages in zebrafish (Developmental
Biology, in preparation).

6


7

Chapter 1: Introduction

7


8

1.1 Zebrafish as a model organism

Zebrafish (Danio rerio) has become an important model organism
in the study of vertebrate developmental biology. Its main advantage is the

availability of numerous genetic mutants (Development Vol. 123, 1996) and
the relative ease with which mutagenesis can be effected by chemical (ENU),
physical (X-ray) and genetic (transposon or viral mediated insertions) means.
This powerful genetics is not available in Xenopus (although Xenopus
tropicalis is currently being developed as a genetic system) and chick.
Although mice and rats are valuable mammalian genetic systems (natural
mutants and ‘knockouts’), the embryos of these animals are not as readily
accessible and available as the hundreds of transparent, externally fertilized
eggs of the zebrafish. Having said this, the mutants currently available for
zebrafish are by no means saturating. Many genes, when mutated, will lead to
very early developmental arrest or death, thus eluding screens designed
around observable phenotypic alterations. In addition, methods for targeted
mutagenesis (similar to ‘knockout’ mice) are currently unavailable in zebrafish.
Thus, cloning of genes from mutants is a labour intensive candidate
gene/map-based effort. To complicate matters, it is an accepted view that
there was a genome duplication event early in the evolution leading to modern
teleosts (Wittbrodt et al, 1998; Woods et al, 2000). As such, redundancy in
gene function is not uncommon. This should in no way be viewed as an
indictment against zebrafish as a model organism. Its genetics and
embryological accessibility, coupled with modern molecular tools - like
dominant negative constructs and the recently developed anti-sense
morpholino oligonucleotides which work by inhibiting translation of target
mRNAs (Nasevicius and Ekker, 2000) - for manipulating gene function, makes
8


9

for a powerful system to understand higher vertebrate biology. This is
particularly evident in the study of human diseases, where many zebrafish

mutants resemble human clinical disorders (Dooley and Zon, 2000).

1.2 General zebrafish development

The zebrafish embryo develops rapidly at 28°C. At 2.75 hours post
fertilization (hpf) or midblastula stage (512 cells), zygotic transcription begins.
This time point marks the midblastula transition (MBT). Cell divisions are very
rapid and synchronized in the first 10 cleavages. At MBT, the cell cycle begins
to lengthen and cells lose their synchrony thereafter. At around this time, the
first lineage restriction is observed, where the formation of the yolk syncytial
layer (YSL) separates the yolk cell from the rest of the blastoderm. After one
or two more cell divisions, the lineage of epithelial enveloping layer (EVL)
cells becomes restricted to exclusively generate the periderm, while the deep
layer cells (DEL) proceed to give rise to the embryo proper. As such, the YSL
is an extra embryonic structure and can thus be seen as the equivalent of the
syncytiotrophoblast of the mammalian embryo. By 4.5 hpf epiboly movements
signal the start of gastrulation. Lineage tracing by intracellular injection of dye
shows that DEL cells located near the animal pole will give rise to ectodermal
fates (including the definitive epidermis) and cells located near the blastoderm
margin will give rise to mesodermal and endodermal fates. At 6 hpf, during the
gastrula period, the dorsal organizer or shield is well defined and gastrulation
is now in full swing. It is also at this stage that the future fates of respective
regions of the gastrula are defined (Kimmel et al, 1990). It has been
suggested that the position of a cell in the blastula is unrelated to the position
of its descendants in the gastrula due to the extensive cell mixing that occurs
9


10

during the transition from blastula to gastrula (Warga and Kimmel, 1990).
Subsequently, a study demonstrated that the fate of cells could be determined
as early as the 16-cell stage (Strehlow et al, 1994). However, due to the
extremely large molecular weight of the labelling tracer employed in this
study, the finding was controversial. It is suggested that the position of
blastomeres at the 16-cell stage may indeed predispose them to a particular
region of the gastrula, and thus determine their ultimate fates, but the final
contribution that any early blastomere makes to the fate map in the gastrula
cannot be predicted because of variability in both the position of the future
dorsoventral axis with respect to the early cleavage blastomeres and the
scattering of daughter cells as the gastrula is formed (Helde et al, 1994;
Wilson et al, 1995).
By 10 hpf, primary gastrulation is complete. At this stage the three
germ layers are well defined and the neural plate is formed. The first somite
appears at about 10.5 hpf and signals the beginning of the segmentation
period. The tail bud is now formed and with it, the recently proposed
secondary gastrulation begins (Kanki and Ho, 1997; Agathon et al, 2003). At
16 hpf (14 somites stage), the notochord separates from the ventral part of
the neural keel and the yolk extension begins to protrude. At 19 hpf (20
somites stage), the lens placode appears and neurons have growing axons.
At this stage, trunk myotomes produce weak contractions. By 24 hpf,
pigments, fins and a beating heart are visible. By day 3, the larva breaks free
from the chorion which has protected it until now, and becomes free
swimming. For a thorough review and description of developmental stages,
see Kimmel et al, 1995 and Westerfield, 1995.

10


11

1.3 Cell movement during gastrualtion

During gastrulation, embryonic cells undergo large scale
movements and rearrangements, leading to the formation of the germ layers
oriented around the embryonic axes. In zebrafish, unlike Xenopus, the initial
cleavage planes of the early blastula are not aligned with the future
dorsoventral (D-V) axis and the time point of D-V axis specification is unclear
(Helde et al, 1994). However, by early-gastrula (5.3 hpf) the shield or
organizer is formed and it defines the dorsal side of the embryo. Three sets of
simultaneously occurring cell movement events can be distinguished during
gastrulation (Warga and Kimmel, 1990): 1) Epiboly, where radial cell
intercalations cause the thinning of the blastoderm and spread it across the
yolk cell. These rearrangements mix cells located deeply in the blastoderm
with the more superficial ones; 2) Deep cells undergo involution to form the
nascent hypoblast of the embryonic shield, and folds the blastoderm into the
epiblast and hypoblast. As gastrulation progresses, involuting hypoblast cells
move anteriorwards relative to cells that remain in the epiblast. This
movement shears the positions of cells that were neighbours before
gastrulation. Involuting cells eventually form endoderm and mesoderm, in an
anterior-posterior sequence according to the time of involution. The epiblast is
equivalent to the embryonic ectoderm. 3) Convergence and extension (CE)
movements toward the dorsal side of the gastrula where mediolateral cell
intercalations take place in both the epiblast and hypoblast. By this
rearrangement, cells that were initially neighbouring one another become
dispersed along the anterior-posterior axis of the embryo.
In Xenopus, all three movements are interdependent due to the
tight coupling of the ectodermal and the mesendodermal sheets (Shih et al,
11

12

1994). Although the zebrafish ectoderm, like that in Xenopus, is also
organized and moves as a sheet (Concha and Adams, 1998), the
mesodermal cells migrate as individuals or groups of cells. Lacking the tight
coupling of ectoderm to the underlying mesoderm, the three gastrulation
movements in zebrafish are thus independent of each other (Warga et al,
1990). Of the three mass cell movement events, CE movement is most
extensively investigated. In the zebrafish, individual cells lose their
independence and integrate their behaviour to achieve a coherent movement
over the yolk (Concha and Adams, 1998). However, at the molecular level,
most of these studies focused on CE within the mesoderm (reviewed in Myers
et al, 2002). Thus, the molecular mechanism governing dorsal movement of
the overlying ectoderm leading to the formation of the neural plate is not well
understood. By 10.5 hpf (yolk-plug stage), primary gastrulation is complete.
A requirement for a posterior organizer and secondary gastrulation
centred around the tailbud has been proposed for the subsequent formation of
the yolk extension and posterior structures of the embryo (Kanki and Ho,
1997; Agathon et al, 2003). Cells which were initially on the ventral side of the
gastrula (opposite the shield) now contribute to the posterior of the animal.
During the gastrula stage, the D-V axis is defined by the position of the shield.
After the yolk-plug stage, as neurulation (the formation of the neural tube from
the neural plate) proceeds, the D-V axis is now applied to the embryo proper.

1.4 Neurulation

In higher vertebrates, the beginning of neurulation is marked by the
thickening of the medial neural plate into the neural keel. Concurrent
dorsal/medial convergence of the neural plate and ventral deepening of the

12


13

neural keel will eventually fold the whole sheet into a tube. As this is being
accomplished, the placode (the border between the neural plate and
epidermis) becomes crest like and will eventually give rise to neural crest
cells. When the neural plate finally folds into a tube, what was initially the
margin of the neural plate now becomes the most dorsal portion of the neural
tube (the roof plate) and the epidermis from both halves of the embryo meet
and cover the neural tube. In higher vertebrates, the neural canal is formed as
the plate folds into a tube. In zebrafish, the progress of neurulation is slightly
different. Instead of the neural plate folding in on itself, dorsal/medial
convergence of the neural plate causes a progressive thickening of the neural
keel, culminating in the formation of a solid rod. The neural canal is then
hollowed out of this rod to give rise to the neural tube (Kimmel et al, 1994;
Papan and Campos-Ortega, 1997)

1.5 Cell fate vs cell movement

It is not entirely clear if cells move according to or in search of
identity. Two possible scenarios have been put forward: a linear pathway
model, where cell movement behaviour is a downstream consequence of cellfate specification; and a parallel pathway model, where a cell initially
assesses its position within the gastrula and interpretation of this information
leads to the parallel activation of a fate specification program and a cell
movement program (Myers et al, 2002). The zebrafish is an attractive and
convenient system to study the interaction and relationship between fate
specification and cell movement. Up to the late gastrula stage, embryonic
cells are pluripotent, and will readily adopt the fate of their new location when

transplanted (Ho and Kimmel, 1993). Since techniques for cell transplantation
13


14

and single blastomere injection are well established, chimeric embryos can
thus be generated. Recently, the novel gene ‘knockdown’ technique using
antisense morpholino oligonucleotides (MO) to inhibit translation of a target
mRNA was demonstrated to work in zebrafish (Nasevicius and Ekker, 2000)
with some limitations regarding nonspecific toxicity (Heasman, 2002). This
technique, in combination with in vivo analysis in chimeric embryos generated
by cell transplantation or single blastomere injection, enables the analysis of
factors involved in cell movement and fate specification during gastrulation.

1.6 The T-box genes

The spontaneous mutation, Brachyury (T), was first described in
mice (Dobrovolskaia-Zavadskaia, 1927) and the gene was finally cloned in
1990 (Herrmann et al, 1990). When opto-motor blind (omb) from Drosophila
was cloned (Pflugfelder, 1992) it became clear that these two genes formed a
new family of transcription factors. The highly homologous and sequence
specific DNA-binding domain was termed the T-domain, and this new family
was collectively referred to as the T-box genes. Since then, many T-box
genes were cloned from a wide variety of species (Papaioannou et al, 1998).
With the completion of a few genome sequences, it is now known that there
are no more than 18 T-box genes in mammals.
T-box genes encode proteins ranging in size from 50 kDa to 78
kDa and are comprised of a structural and a functional domain: the T-domain,
and a transcriptional activator or repressor domain. The crystallographic

structure of the T-domain of the Xenopus orthologue of Brachyury (Xbra)
(Muller and Herrmann, 1997) and the human Tbx3 (Coll et al, 2002) showed
that the T-domain is unique among DNA-binding domains. Analyses of
14


15

downstream targets and binding-site selection experiments showed that
several members bind to a core consensus sequence TCACACCT, although
different members preferentially bind sequences that contain two or more of
these core motifs in various orientations (Sinha et al, 2000; Conlon et al,
2001). In addition, murine Tbx2 has been shown to bind a variant T-site
(Lingbeek et al, 2002). However, little is known about how T-box genes select
and regulate their downstream targets (Tada and Smith, 2001). T-box proteins
have been shown to function both as transcriptional activators and repressors.
In all cases studied, this functionality requires sequences in the carboxyterminal portion of the protein. Again, little is known about how the
transactivation domain selects and binds interacting partners to effect
transcriptional regulation.
T-box transcription factors play important roles in vertebrate
development (Smith, 1999; Wilson and Conlon, 2002; Showell et al, 2004).
Brachyury (T) has been shown to be required for cell adhesion (Yanagisawa
and Fujimoto, 1977), cell migration (Hashimoto et al, 1987), and
morphogenetic cell movements during gastrulation (Wilson et al, 1995). T and
the zebrafish orthologue no tail (ntl) are known to be essential for the
specification of axial mesoderm (Smith 1999). Similar to T, Tbx16/spadetail
(spt) has been shown to be required for mesodermal cell movement (Ho and
Kane, 1990); and Xbra has been demonstrated to function as a switch
between cell migration and convergent extension during gastrulation (Kwan
and Kirschner, 2003). In addition, a number of human disorders have been

linked to mutations in T-box genes, confirming their medical importance. They
include Holt-Oram syndrome/TBX5, Ulnar-Mammary syndrome/TBX3, and
more recently DiGeorge syndrome/TBX1, ACTH deficiency/TBX19 and cleft
palate with ankyloglossia/TBX22 (Packham and Brook, 2003).
15


16

1.7 The Tbx2 family

Zebrafish tbx2b belongs to the Tbx2 family and was previously
named tbx-c. Its expression is first detected by whole-mount in situ
hybridization at 8.5 hpf in the axial mesoderm. By 12 hpf it is expressed in the
eye field, ventral diencephalon and, uniquely among the Tbx2 family, the
notochord. At 24 hpf, the domains of expression include the eyes, otic
vesicles, trigeminal ganglia, Rohon-Beard cells, cell of the epiphysis and
pronephric ducts. By this stage, the expression in the notochord has receded
to the tailbud, in the region of the chordo-neural hinge (Dheen et al, 1999).
tbx2a, another zebrafish Tbx2 gene, is highly homologous to tbx2b with
extensive overlap in their expression patterns. However, tbx2a is not
expressed in the notochord and the epiphysis at any stage, but is instead
expressed in the cells of neural crest, and subsequently in the branchial
arches. It is very likely that these two genes are the product of a chromosomal
duplication event which occurred sometime during the evolution of teleosts,
about 450 million years ago (Force et al, 1999).
Tbx2b has been demonstrated to function in a complex signalling
web downstream of TGF-β signalling, ntl and floating head (flh) for the
specification of the notochord. In brief, a loss-of-function experiment, via overexpression of a dominant negative (dn)-Tbx2b that lacked the carboxyterminal transactivation domain, resulted in embryos with reduced midline
mesoderm and expanded lateral mesoderm. In agreement, the opposite gainof-function experiment, via overexpression of the full length tbx2b, led to an

expansion of the midline mesoderm and formation of ectopic midline
structures at the expense of lateral mesodermal tissues. Animal explant
experiments showed that Tbx2b acts downstream of early mesodermal
16


17

inducers (activin and ntl) and reveal an autoregulatory feedback loop between
ntl and tbx2b. Thus, in zebrafish, two T-box factors are required for the
formation of the notochord (Dheen et al, 1999).
Tbx2 of Xenopus, chick, mouse and human have been
characterized quite extensively. None of them were shown to be expressed in
the notochord. In chick, Tbx2, together with Tbx3,4 and 5 were shown to be
involved in limb and digit identity specification (Gibson-Brown et al, 1998;
Suzuki et al, 2004). Human TBX2 is known to repress the melanocyte-specific
TRP-1 promoter (Carreira et al, 1998) and is itself a target for the
microphthalmia-associated transcription factor in melanocytes (Carreira et al,
2000). Murine Tbx2 is also known to repress the tumor suppressor gene
p14ARF via a variant T-site in its promoter (Lingbeek et al, 2002). Although
murine Tbx2 is known to contain domains for both transcriptional activation
and repression (Paxton et al, 2002) no target of Tbx2 activation has been
identified. Recently, TBX2 was shown to be linked to the development of
breast cancer (Sinclair et al, 2002), possibly via its role in the repression of
p19ARF and bypassing senescence control (Jacobs et al, 2000). TBX2 was also
implicated in cell cycle control via its repression of p21WAF1cyclin-dependent
kinase inhibitor (Prince et al, 2004). The Drosophila Tbx2-related omb is
regulated by the Wingless (Wg), Decapentaplegic (Dpp) (Grimm and
Pflugfelder, 1996) and Hedgehog (Hh) (Kopp et al, 1997) signalling pathways;
Xenopus Tbx2 is known to function within the Sonic hedgehog (Shh) pathway

(Takabatake et al, 2002); and in chick, Tbx2 is known to function through both
Shh and bone morphogenetic protein (BMP, vertebrate homologue of Dpp)
signalling (Suzuki et al, 2004). Thus, it is of interest to determine the pathways
regulating zebrafish tbx2b.

17


18

1.8 The Wnt signalling pathway

Wnt (vertebrate Wg) signalling is involved in cell movement and
fate determination (Moon et al, 1997). Based on their ability to induce a
second axis in Xenopus, Wnt proteins have been divided into two classes: the
Wnt1 class, which induces a second axis; and the Wnt5 class, which does
not. Molecular dissection of the two classes has shed light on their respective
mechanisms and roles in development. The Wnt1 class acts via β-catenin and
is shown to function in fate determination. This is equivalent to the Drosophila
Wg/Armadillo pathway and is now commonly referred to as the canonical
pathway (Moon et al 2002). The Wnt5 class is β-catenin-independent and is
known to play essential roles in cell polarity and cell movement (Veeman et al,
2003). This has been shown to be similar to the Planar Cell Polarity (PCP)
signalling pathway in Drosophila and is now referred to as the non-canonical
pathway. In Xenopus, Wnt3a and Wnt8 have been implicated in the
posteriorisation of the neural plate (Bang et al, 1999; McGrew et al, 1995 and
1997). Mouse Wnt3a is involved in the specification of paraxial mesoderm (Liu
et al., 1999). In zebrafish, mutants of wnt5, wnt8 and wnt11 have been
identified (Rauch et al, 1997; Heisenberg et al, 2000; Lekven et al, 2001).
Wnt5a and Wnt11 were shown to act through the non-canonical signalling

pathways and affect morphogenetic cell movements (Heisenberg et al, 2000;
Tada and Smith 2000; Wallingford et al, 2001); whilst Wnt8 signals through
the canonical Wnt pathway to mediate mesodermal patterning by promoting
ventral cell fates (Lekven et al, 2001). The two pathways share some common
components: the 7-pass transmembrane Frizzled (Fz) receptors and the
downstream adaptor molecule Dishevelled (Dsh).
The amino-terminal extracellular Wnt binding domain and the
18


19

carboxy-terminal PDZ-binding domain of the Fz protein are known to be
essential for Fz function. Deletion of the PDZ-binding domain creates a
dominant negative molecule that will bind the Wnt ligand but will not transduce
signal intracellularly. Dsh is modular in nature and has three functional
domains. The DIX domain is essential for β-catenin activity, whereas the DEP
domain is involved in non-canonical signalling (Topczewski et al 2001). In
Drosophila, it is known that the DEP domain is important for the differential
recruitment of Dsh to the plasma membrane by Fz upon binding of Wg ligand,
and provides signalling specificity in the PCP and Wingless signalling
pathways (Axelrod et al,1998). The situation with the PDZ domain is more
ambiguous. It has been shown to affect either (Axelrod et al, 1998), or both
(Sokol, 1996), the β-catenin and CE pathaway. The data suggest that the
PDZ domain may serve a regulatory role, rather than as an obligate
transduction module (Wharton, 2003).
Fz receptors play a key role in diverting the Wnt signals into
specific pathways. In vitro studies have demonstrated that Wnts may interact
with any Fz receptor, however the binding affinity varies among different
ligand-receptor pairs (Hsieh et al, 1999; Rulifson et al, 2000). Recently a Fz4

‘knockout’ mouse was generated (Wang et al, 2001) which implicated Fz4 in
the maintenance of the nervous system during adulthood. In Xenopus, Fz3
interacts with XWnt1 and is required for the formation of the neural crest
(Deardorff et al, 2001). Xenopus Fz7 appears to operate via the canonical βcatenin pathway for fate specification (Sumanas et al, 2000), a non-canonical
pathway essential for convergent extension (Djiane et al, 2002), and a Dshindependent pathway that controls cell-sorting behaviour in the mesoderm
(Winklbauer et al 2001). It was also demonstrated that rat Fz2 is capable of
interacting with XWnt5A in the Wnt/Ca2+ pathway (Slusarski et al, 1997b). In
19


20

addition, it has been suggested that zebrafish Fz2 may function as a receptor
for Wnt5 and may play a role in the regulation of gastrulation through the noncanonical pathway (Sumanas et al, 2001).
The signalling cascade downstream of Dsh is complex. The
proteins Axin and GSK3β form a multiprotein complex with Adenomatous
Polyposis Coli (APC). Axin regulates multiple signalling pathways by serving
as a scaffold protein, controlling diverse cellular functions in proliferation, fate
determination, and suppression of tumorigenesis (Luo and Lin, 2004).
Recently, it was demonstrated that Axin also plays an important role in a JNK
signalling pathway by utilizing discriminatory domains for its distinct roles in
the both the β-catenin pathway and the JNK pathway (Zhang et al 2000).
GSK3β is a serine/threonine kinase involved in insulin, growth factor and Wnt
signalling. In the absence of Wnt signalling, GSK3β is recruited to the APC
complex via interaction with Axin, where it hyperphosphorylates β-catenin,
marking it for ubiquitylation and destruction, thereby inhibiting the canonical
pathway. Activation of Wnt signalling leads to the inhibition of GSK3β activity
and an accumulation of stabilized cytoplasmic (signalling) β-catenin, which
becomes available to bind the TCF/LEF family of transcription factors leading
to target gene expression (Nelson and Nusse, 2004).

β-catenin, in addition to its role in Wnt signalling, is also an integral
component of the adhesion complex by linking cadherins through α-catenin to
the actin cytoskeleton. Phosphorylation-dependent release of β-catenin from
the cadherin complex not only regulates the integrity and function of the
adhesion complex, but may also be an alternative mechanism for activating βcatenin signalling. For an in-depth review of the potential interaction between
Wnt, β-catenin and cadherins, see the review by Nelson and Nusse (2004). In
20


21

contrast, the non-canonical pathway was shown to act through the DEP and
PDZ domains of Dsh but not APC, and activates the JNK/SAPK signalling
cascade instead (reviewed in Moon, 1997 and 2002). However, in view of the
role of Axin in JNK signalling (Zhang et al, 2000), this may need to be
reviewed. Below is a figure summarizing the essential components of the
human Wnt signalling pathway and their interactions with each other (courtesy
of Biocarta, />
21


22

1.9 Neural induction

As mentioned earlier, formation of the neural plate during
gastrulation involves concurrent cell fate specification and cell movement
events. The embryonic ectoderm has the potential to acquire either epidermal
or neural fates. Spemann and Mangold showed that the vertebrate organizer
produces signals to induce the neural fate (Spemann et al, 1924; Spemann,

1938). These neural-inducing signals were initially thought to play instructive
roles. Subsequent experiments showed that they function via transcriptional
repression of BMP gene expression and clearance of BMP ligands by
secreted inhibitors. Thus, neural induction depends on the suppression of
BMP signalling by organizer-derived inhibitors, and in the absence of cell-cell
signalling ectodermal cells will adopt a neural fate. This is referred to as the
‘default model’ of neural induction (reviewed in Munoz-Sanjuan et al, 2002).
In Xenopus, BMP-inhibiting factors include chordin, follistatin,
noggin, cerberus and xnr3 (reviewed in Harland, 1994; Harland, 2000). Both
the Fibroblast Growth Factor (FGF) and Wnt signalling pathways have been
implicated in this process. In chick, gain-of-function experiments (by
transplantation of neural plate tissue or implantation of beads containing
proteins of interest into the non-neural epiblast) suggest that the early events
of neural induction are regulated by FGF signalling, which could be blocked
by continued Wnt signalling (reviewed in Wilson et al, 2001). In Xenopus,
although FGF was shown to be a direct neural inducer (Lamb et al, 1995) and
a truncated FGF receptor was shown to block neural induction by
endogenous Xenopus inducers (Launay et al, 1996), gain-of-function
experiments (by injection of mRNAs) also demonstrated that neural induction
depends on Wnt/β-catenin inhibition of BMP (Baker et al, 1999; Wessely et al,
22


23

2001). In the zebrafish, Wnt/β-catenin-dependent bozozok (boz) is sufficient
to suppress expression of bmp (Fekany-Lee et al 2000). While a secondary
axis consisting of ectodermal derivatives and lateral mesoderm was induced
by FGF8 (Furthauer et al 1997), in vivo blockade of FGF signalling in
Xenopus and zebrafish caused notochord deficiency but failed to disrupt

neural development (Schulte-Merker et al, 1995). As such, there seem to be
conflicting requirements for FGF and/or Wnt signalling in neural induction
between amniotes and anamniotes. This difference in molecular mechanism
needs to be further clarified.

1.10 BMP and axis specification

It has also been demonstrated that modulations in BMP levels
affect fate specification along the D-V axis without affecting those along the
anterior-posterior (A-P) axis (Barth et al, 1999). In lower vertebrates, induction
of the D-V axis may occur as early as the cleavage stage. This is marked by
the enrichment of maternal determinants such as β-catenin in the prospective
embryonic dorsal pole (Jesuthasan and Strähle, 1996; Aanstad and Whitaker,
1999; Ober and Schulte-Merker, 1999), and this process is dependent on
maternal Wnt signalling (Miller et al, 1999). Induction of the A-P axis occurs at
the onset of gastrulation and involves a combination of several pathways,
including the Wnt signalling pathway (Christian and Moon, 1993; Glinka et al,
1998; Piccolo et al, 1999; Bradley et al, 2000; reviewed in Gamse and Sive,
2000). Wnts are capable of inducing a secondary axis if expressed vegetally
prior to midblastula transition (MBT). However, Wnts will promote ventral and
posterior fates if expressed after MBT (Christian and Moon, 1993).

23


24

1.11 Summary

In this thesis, we use the zebrafish as a model to study the

molecular mechanisms behind the specification of embryonic ectoderm.
Specifically, we attempt to isolate the role of Tbx2b during the formation of the
neural plate, a prominent event during gastrulation, where known signalling
casdades like Wnt, FGF and BMP are involved in cell fate specification and
large scale morphogenetic cell movement within the ectodermal layer. The
transparancy of the zebrafish embryo, couple with established techniques of
embryology and availability of mutants and molecular tools, allow us to
demonstrate that Tbx2b plays a role in cell adhesion and cell movement
during this critical period of development. In addition, we show that a hitherto
unmapped group of deep vegetal cells of the blastoderm that expresses Fz2
migrates to the ventral region of the gastrula, and thus establish a link
between the animal-vegetal position of the blastula and the dorso-ventral axis
of the gastrula.

24


25

Chapter 2: Wnt signalling mediated by Tbx2b is required for
cell migration during formation of the neural plate

25