ELUCIDATING THE ROLE OF BNIP 2 IN ZEBRAFISH EARLY DEVELOPMENT

TABLE OF CONTENTSTable of contents ii List of Figures vi List of Abbreviations viii Chapter 1.1.2 BNIP-2 and cell dynamics 2 1.3 BCH domain containing-proteins and cell dynamics 7 1.5.6

ROLE OF BNIP-2

IN ZEBRAFISH EARLY DEVELOPMENT

CHUA SEE KIN DOREEN

(B.Sc.(Hons.), NTU

A THESIS SUBMITTED FOR THE DEGREE OF MASTER OF SCIENCE DEPARTMENT OF BIOLOGICAL SCIENCES NATIONAL UNIVERSITY OF SINGAPORE

2013

The course of my Master’s studies has been a rollercoaster ride of upsand downs - joys and disappointments, achievements and failures - but I’mthankful for all of it, for it has been a moulding process for my character and

as a researcher I would like to acknowledge important ones who have helped

me in my journey to be a better person and scientist

Firstly, I would like to express my utmost thanks and appreciation toAssoc Prof Low Boon Chuan for giving me the privilege of being his studentand for believing in me I am grateful for the warmth, enthusiasm, guidanceand encouraging pep talks I have received from a brilliant scientist and teacher

I thank also Tiweng, my teacher in the lab for his patience and time spent inguiding me through my experiments Thanks to all members of LBC lab foryour friendliness and help extended over the course of my studies

I would like to sincerely thank Prof Gong Zhiyuan for graciously andgenerously providing me space in his laboratory to perform my experiments inthe final leg of my studies Thanks to all in the GZY lab for all the helprendered to me

I thank also Lora Tan for her generosity in giving me useful tips andfor her friendship in the lab Thanks also to Mr Balan from the zebrafishaquarium for the chats, his helpfulness and his responsibility in helping tosupply zebrafish embryos

To my family and Samuel, thanks for your full support andunderstanding, and for taking good care of me throughout it all I share thisachievement with you To Him be all glory

TABLE OF CONTENTS

Table of contents ii List of Figures vi List of Abbreviations viii

Chapter

1.1.2 BNIP-2 and cell dynamics 2

1.3 BCH domain containing-proteins and cell dynamics 7

1.5.6 Convergence and Extension (C&E) movements 14

1.7.1 bnip-2 knockdown by morpholino phosphorodiamidate 20

antisense oligonucleotides

1.7.2 Investigating potential bnip-2 interacting genes – E-cadherin, 27

RhoA, Cdc42

2.2.4 Purification of DNA Fragment From Agarose Gel 30

2.2.6 Growth, preparation and transformation of competent E.coli cells 30

2.2.6.1 Growth of E coli cells in liquid and solid media 30

2.2.6.2 Preparation of competent E coli cells 31

2.2.6.3 Transformation of competent E coli cells 31

2.3.1 Whole mount in situ hybridization 342.3.2 Synthesis of digoxigenin labeled antisense RNA Probes 35

2.3.2.2 Probe Synthesis - RNA labeling by in vitro transcription 352.3.3 Collection and Preparation of zebrafish embryos 35

2.3.5.1 Preparation of pre-absorbed Digoxigenin-Alkaline 36Phosphatase (DIG-AP) antibody

2.3.5.2 Incubation with pre-absorbed anti-DIG-AP antibody 362.3.6 Washing, Staining with NBT/BCIP and Fixation 37

2.4.1 Protein extraction from zebrafish embryos 382.4.2 Sodium Dodecyl Sulphate-Polyacrylamide gel electrophoresis 39(SDS-PAGE)

2.5.1 Design and preparation of translational morpholinos 41

2.5.3.1 Construction of pCS2–bnip-2b and pCS2-bnip2c for 42

mRNA synthesis2.5.3.2 Linearisation of plasmids for mRNA synthesis 42

3.1 bnip-2 knockdown elicits defects in epiboly and C&E processes 44

3.2 bnip-2 mRNA suppresses gastrulation defects in bnip-2 knockdown 57morphants

3.3 bnip-2 knockdown causes abnormalities in epibolic mechanisms 61

3.4 bnip-2 knockdown causes increased RhoA activity 66

3.5 bnip-2 knockdown increases myosin light chain-2 phosphorylation 74

3.6 bnip-2 knockdown disrupts E-cadherin membrane localisation 75

3.7 Dominant-negative rhoA restores membrane-localised E-cadherin in 81morphants and rescues gastrulation defects

4.2 Regulation of RhoA and E-cadherin by Bnip-2 is required for epiboly 88

LIST OF FIGURES

1 1 Domain architecture and classifications of BCH-domaincontaining proteins. 6

1 2 The four domains of mesodermal C&E movements inthe zebrafish gastrula and their characteristic underlying

1 3 Description of morpholinos used for functional rescueexperiments. 26

1 4 Schematic outline of experiments performed to elucidatebnip-2’s function in zebrafish. 26

3 1 bnip-2 knockdown elicits defects in epiboly and C&Emovements. 51

3 2 bnip-2 knockdown by morpholino causes epiboly delay. 52

3 3 Control MO-injected embryos that show epiboly delaydisplay higher percentage of abnormalities at the

3 6 Control MO-injected embryos that show abnormalitiesat the 1-somite stage or epiboly arrest display. 54

3 7 bnip-2 MO1-injected embryos that show abnormalitiesat the 1-somite stage or epiboly arrest display. 55

3 8 bnip-2 MO2-injected embryos that show abnormalitiesat the 1-somite stage or epiboly arrest display. 55

3 9 Analysis of marker gene expression in control and bnip- 2 morphant zebrafish embryos. 56

3 10 bnip-2 morphants embryos are categorised according toseverity of phenotype. 59

3 11 bnip-2 knockdown by morpholino is dose-dependent. 59

3 12 bnip-2 morphant phenotype could be rescued by bnip-2mRNA. 60

3 13 bnip-2 morphant EVL cells display cell shape defects. 63

3 14 bnip-2 morphant EVL cells display actin ringabnormality. 63

3 15 bnip-2 morphant EVL marginal cells display defects incell shape changes. 64

3 16 bnip-2 morphants display separation of EVL-DELduring late epiboly 65

3 17 bnip-2 morphants have higher RhoA activity. 70

3 18 bnip-2 morphant phenotype is aggravated by constitutively active rhoA mRNA. 71

3 19 bnip-2 morphant phenotype is suppressed by dominant negative rhoA mRNA. 71

3 20 No change in bnip-2 morphant phenotype with wild type rhoA mRNA. 72

3 22 bnip-2 morphants have higher MLC-2 activity. 74

3 23 bnip-2 morphant EVL cells display reduced membrane-localised E-cadherin. 77

3 24 e-cadherin MO knockdown elicits defects in epibolyand C&E movements. 78

3 25 Synergy of bnip-2 and e-cadherin MO knockdown ineliciting epiboly defects. 79

3 26 Synergy of bnip-2 and e-cadherin MO knockdown ineliciting C&E defects. 79

3 27 Synergy of bnip-2 and e-cadherin MO knockdown inreducing EVL membrane-localised E-cadherin. 80

LIST OF ABBREVIATIONS

BCIP 5-bromo-4-chloro-3-indonyl phosphate

BNIP bcl-2/adenovirus E1B Nineteen kilo-daltons interacting

protein bmp4 Bone morphogenetic protein 4

E-YSL external yolk syncytial layer

F-actin filamentous actin

GEF guanine nucleotide exchange factor

eGFP enhanced green fluorescent protein

hpf hours post fertilisation

I-YSL internal yolk syncytial layer

mRNA messenger ribonucleic acid

NBT 4-nitroblue tetrazolium chloride

papc paraxial protocadherin

PBS phosphate buffered saline

PBT phosphate buffered saline + 1% Tween 20

PCR polymerase chain reaction

In order to explore its potential role in development to gain further insight into

its physiological function, in vivo functional studies on bnip-2 were conducted

using translational-blocking morpholino-based knockdown in zebrafish - amodel organism selected for its high fecundity, the optical transparency andease of access to embryos The process of gastrulation, in which widespreadcellular movements and behaviours occur, was examined in particular becauseBNIP-2 is known through cell culture studies to regulate cell dynamics

bnip-2 knockdown morphants displayed rescuable gastrulation defects

such as delayed or arrested epiboly, and disrupted convergence-extension(C&E) movements leading to impaired axial extension and abnormalmediolateral widening of axial and paraxial mesodermal tissues Furthermore,

at late epiboly stages, anomalies such as separation of the enveloping layer(EVL) from the deep cells, abnormal morphology of EVL marginal cells and awidened actin band in the yolk syncytial layer were observed Overexpression

of a constitutively active form of rhoA, a key regulator of actin cytoskeletal dynamics, aggravated the bnip-2 morphant phenotype, while that of dominant negative rhoA attenuated the phenotype severity of bnip-2 morphants in which

upregulated RhoA activity and phosphorylated myosin light chain 2 was found

In addition, cell membrane localisation of E-cadherin in the EVL was

disrupted, and synergy between e-cadherin and bnip-2 morpholinos in

eliciting the bnip-2 morphant phenotype was observed Dominant negative

rhoA could suppress the bnip-2 morphant phenotype caused by synergy

between e-cadherin and bnip-2 morpholinos.

In summary, these results reveal that bnip-2 is required for normal

gastrulation movements, and that its role involves, at least in part, theregulation of the membrane localisation of E-cadherin through the modulation

of RhoA activity In conclusion, this work introduces a novel molecular player

in gastrulation, bnip-2, which may also be a new link between cell dynamics

and development These findings shed some light on the genetic interactions

of bnip-2 and their possible roles in mechanisms underlying zebrafish

gastrulation, and thus contribute insight into the molecular mechanisms

underlying the regulation of cell dynamics by bnip-2.

1 Introduction

1.1 BNIP-2

1.1.1 Initial discoveries of BNIP-2

The Bcl-2/adenovirus E1B Nineteen kilo-daltons Interacting Protein-2,

or BNIP-2 in short, was initially discovered as one of three novel proteins,Nip1, Nip2 and Nip3, in a bid to identify interacting proteins of the anti-

apoptotic adenovirus E1B 19kDA protein (Boyd et al., 1994) The E1B

19kDA protein functions to prevent host cell activation of cell deathprogrammes in order to allow viral replication during viral infection BNIP-2,

or Nip-2 as it was called then, was hypothesised to be involved in thepromotion of cell survival due to their direct interactions with both the anti-apoptotic Bcl-2 and E1B 19kDa proteins More recently, the Nip proteinshave been classified as pro-apoptotic members of the Bcl-2 family of proteinsdue to their possession of the conserved Bcl-2 homology domain 3 (BH3),

which promotes dimerization of Bcl-2 family members (Zhang et al., 2003).

After its initial discovery, further investigations on BNIP-2 were madewhen it was found to transiently interact with the cytoplasmic tail of theFibroblast Growth Factor Receptor-1 (FGFR-1) via an yeast-two-hybrid assay

(Low et al., 1999) Subsequently, it was verified to be a bona fide substrate of FGFR that is tyrosine-phosphorylated upon FGFR stimulation by FGF in in

vitro and cell culture contexts Bioinformatic analyses revealed strong

homology (61% similarity) between the N-terminus of BNIP-2 and the terminus non-catalytic domain of Cdc42GAP (otherwise known as p50-RhoGAP), a GTPase-activating protein for Cdc42 This region at the N-

C-terminus of BNIP-2 was later named the BNIP-2 and Cdc42GAP homology

domain (BCH) (Low et al., 2000) Further, it was found that the BCH domain,

via its 217RRKMP221 region, mediates homo- and heterocomplex formationbetween BNIP-2 and Cdc42GAP, but the interaction is prevented by tyrosine

phosphorylation of BNIP-2 (Low et al., 2000) Between BNIP-2 and

Cdc42GAP, there is also competitive binding to Cdc42 Strikingly, via theBCH domain, BNIP-2 also binds and promotes the GTPase-activity intrinsic

to Cdc42 via a novel arginine patch motif, 235RRLRK239, similar to the

“arginine finger” employed by one contributing partner in a Cdc42 homodimer,

and this too, is inhibited by tyrosine phosphorylation of BNIP-2 (Low et al.,

2000) The BCH domain in Cdc42GAP does not have GAP activity to Cdc42

as it lacks the arginine patch

Therefore the BNIP-2 interactome discovered from these early studieshinted at BNIP-2’s involvement in a variety of pathways such as tyrosinekinase receptor signalling, GTPase-mediated signalling pathways andapoptosis, and suggested physiological significance that should be furtherlooked into

1.1.2 BNIP-2 and cell dynamics

The physiological significance of BNIP-2’s interactions with Cdc42came to light in a later study that overexpressed BNIP- 2 in MCF-7, HeLa, andCOS-1 cells Dramatic cell morphological changes were elicited by BNIP-2overexpression, including cell elongation and the formation of membrane

protrusions at the sites of its localisation (Zhou et al., 2005) Such changes

were dependent on the recruitment of Cdc42 by the BCH domain, and were

effectively suppressed by the co-expression of dominant negative mutant

forms of Cdc42 (Zhou et al., 2005).

Unpublished work by the same laboratory showed that by activation ofRho, BNIP-2 had an inhibitory effect on MDCK epithelia cell spreading andretarded collective cell migration in a wound healing assay (Pan and Low,2012) In addition, the binding of BNIP-2 to BPGAP1 potentiated the latter’sGAP activity towards Rho and reduced cell proliferation (Pan and Low, 2012)

In addition, imaging studies to measure the activity of BNIP-2 or BCH domainalone in cells showed that BNIP-2 and BCH domain are dynamicallydistributed between endosomes and cell protrusions along the microtubules,and they were most active at protrusive tips (Pan and Low, 2012) Moreover,BNIP-2 has a kinesin-binding motif which is necessary for its trafficking in

cells (Aoyama et al., 2009).

These observations strongly support the role of BNIP-2 in theregulation of GTPase signalling and cell dynamics, and the versatility ofBNIP-2 in engaging different Rho GTPases and their GAPs and GEFs suggestthat BNIP-2 is involved in regulating GTPase signalling in a context-dependent manner (Pan and Low, 2012)

1.2 Bioinformatic analyses of the BCH domain

The BCH domain was first discovered as a region of strong homologybetween BNIP-2 and Cdc42GAP but was subsequently found to have 14%sequence identity with the lipophilic CRAL_TRIO domain of the

Saccharomyces cerevisiae Sec14p protein (Bankaitis et al., 2010) The

CRAL_TRIO domain is also present in the cellular retinaldehyde binding

protein (CRALBP) and Trio, a RhoGEF (Bankaitis et al., 2010) Similar

protein domains can be found in other proteins such as tyrosine phosphatase,

α-tocopherol transfer protein and others RhoGEFs such as Duo and Dbs (Gu et

al., 1991, Min et al., 2003, Aravind et al., 1999, Pan and Low, 2012).

Although these domains in some of these proteins bind small hydrophobic

ligands, BCH domains are not known as yet to be lipophilic (Panagabko et al.,

2003, Pan and Low, 2012) More recently, through genome-wide, species bioinformatic analyses of CRAL-TRIO and similar domains, andputative BCH sequences, the BCH domains have emerged as a novel subclass

cross-under the CRAL-TRIO superfamily (Gupta et al., 2012) BCH domains have

been recognised in a large variety of proteins from diverse species including

slime molds, plants, yeasts, insects, fish to human (Gupta et al., 2012) Further

gene-structure and protein domain context analyses reveal that BCH domainsequences can undergo alternative RNA splicing, leading to, for example,

splicing variants of BNIP-2, BNIP-2-Similar and BNIP-2 Extra Long (Zhou et

al., 2002, Soh et al., 2008).

Proteins containing the BCH domains can be subdivided into threegroups: Group 1 members are defined by the presence of a single BCHdomain that has the high amino acid sequence identity to the prototypical

BNIP-2 BCH domain compared to the other groups, Group 2 members possess

a BCH domain that is associated with the macro domain, and Group 3members contain a BCH domain associated with a RhoGAP domain (Pan andLow, 2012) The list of BCH-containing proteins can be found in Figure 1.1

Figure 1.1: Domain architecture and classifications of BCH-domain

containing proteins Proteins containing the BCH domains can be subdivided

into three groups: Group 1 members are defined by the presence of a singleBCH domain that has the high amino acid sequence identity to the prototypicalBNIP-2 BCH domain compared to the other groups, Group 2 members possess

a BCH domain that is associated with the macro domain, and Group 3members contain a BCH domain associated with a RhoGAP domain (Pan andLow, 2012) The percentages indicate the degrees of amino acid sequenceidentities compared to the prototypical BNIP-2 BCH domain This figure isadapted from Pan and Low, 2004

1.3 BCH domain containing-proteins and cell dynamics

There is significant conservation in two GTPase-binding motifs found

in the BCH domains These motifs resemble the Rho-binding domain (RBD)and the Cdc42/Rac interactive binding domain found commonly in Rho andCdc42/Rac1 effector proteins, respectively (Pan and Low, 2012) In particular,the BNIP-2 BCH domain contains within the CRIB-like region anexperimentally validated novel Cdc42-binding motif,285VPMEYVGI292, whileBNIP-S, BNIP-XL and Cdc42GAP possess RBD-like motifs These GTPase-binding motifs have been found to mediate cell morphogenesis, migration anddifferentiation

BNIP-H expression is highly specific to the brain and concentrates in

the cerebellum and hippocampus (Buschdorf et al., 2006) Mutations in

BNIP-H gene, two of which are predicted to cause defects in the BCBNIP-H domain, are

associated with the Cayman cerebellar ataxia disease (Bomar et al., 2003) The

protein targets for BNIP-H include the heavy chain of kinesin-1 motor, Rabsmall GTPases, Mek and Pin1 isomerase (Pan and Low, 2012) BNIP-Hfunctions like an adaptor in transporting mitochondria in the kinesin-1 light

chain along neuritis (Aoyama et al., 2009) BNIP-H has also been shown to

bind a kidney-type phosphate-activated glutaminase (KGA) that is a metabolicenzyme responsible for glutamate production, and relocalise it to the tips of

neurons (Buschdorf et al., 2006) Unlike BNIP-2 which interacts with Cdc42,

BNIP-H targets mainly Rab GTPases and can be observed colocalising withthese GTPases in endosomes and along neurites (Pan and Low, 2012)

BNIP-XL is one of four major isoforms, isoform-4, encoded viadifferential initiation sites from the BMCC1 gene, a gene which has been

linked to human pathologies such as prostate cancer (Clarke et al., 2009), Alzheimer’s disease (Potkin et al., 2009) and leiomyosarcomas (Price et al.,

2007) Like isoforms-1 and -3, BNIP-XL contains the BCH domain It canundergo alternative splicing to generate BNIP-XLα and BNIP-XLβ (Figure1.1) (Pan and Low, 2012) Among BNIP-2, BNIP-Sα, BNIP-H and BPGAP1,BNIP-2 is the protein BNIP-XL has the closest homology to However, itsBCH domain is most similar to that of BNIP-H BNIP-XL has been proven toaffect actin cytoskeletal reorganisation (i.e formation of stress fibers) and

antagonise Rho-mediated cellular transformation (Soh et al., 2008) In that

study, it was shown that the BCH domain of BNIP-XL interacts with RhoA(as well as RhoC), and mediates association of BNIP-XL with the catalytic

domains of Lbc, a RhoA-specific GEF (Soh et al., 2008) The knockdown of

BNIP-XL increased active RhoA levels, while its overexpression reduced it.Therefore, BNIP-XL suppresses cellular transformation by restricting RhoA

and Lbc binding, thus preventing sustained Rho activation (Soh et al., 2008).

It was surmised that this could be a general mechanism for regulating Rho

GTPases and their regulators RhoGEFs (Soh et al., 2008).

BNIP-S share 72% similarity with BNIP-2 and its BCH domain has ahigh sequence homology of 86% similarity with the BCH Domain of BNIP-2

(Zhou et al., 2002) Overexpression of BNIP-S leads to BCH

domain-mediated extensive cell rounding and consequently, apoptosis independent ofthe action of caspases This apoptotic effect can be suppressed by co-expression of dominant negative RhoA, thus suggesting that the apoptotic

effect of BNIP-S is mediated by active RhoA (Zhou et al., 2006) Indeed,

BNIP-S causes cell rounding and apoptosis by sequestering Cdc42GAP, thusnegatively regulating its activity, and capturing RhoA for further activation

BNIP-S, however, does not bind Rac1, Cdc42 and GTP-bound RhoA, bindingonly GDP-bound RhoA.

Cdc42GAP and its homolog BPGAP1, are BCH domain-containingRhoGAPs which negatively regulate Rho GTPases, specifically Cdc42 andRho, by activating their intrinsic GTPase activity, thus converting them fromthe active GTP-bound state to the inactive GDP-bound state It has recentlybeen shown that the BCH domain in Cdc42GAP, which contains a novelRhoA-binding motif, serves as a local modulator to sequester RhoA to prevent

it from being inactivated by its proximal GAP domain (Zhou et.al., 2010).

BPGAP1 activates cell protrusions and cell migration, mediated bycooperation between its BCH domain, a proline-rich region (PRR) and a GAP

domain (Shang et al., 2003) The BCH domain of BPGAP1 elicits

Cdc42/Rac-mediated cellular protrusions that enable its association with cortactin, whichhelps form branching actin network, and endophilin-2, which binds to the PRR

region, for the exertion of its function (Lua et al., 2004, Lua et al., 2005).

1.4 Zebrafish, a model organism

The zebrafish is a small and hardy freashwater tropical fish native tothe waters in India As a model organism, it offers several attractive practicaladvantages It is easily available; it can be purchased in local commercialaquariums In terms of husbandry, it has a relatively low maintenance costcompared to model organisms such as the mouse and the rat, which requiremore expensive and greater infrastructure Its small size allows for easyhandling, and its short generation time of approximately three months allowsfor relatively quick generation of transgenic lines Furthermore, the zebrafishhas high fecundity, thus allowing sufficient material and a large sample sizefor statistical power in experiments

The zebrafish was the first vertebrate that proved to be tractable tolarge-scale genetic screening most often conducted using fruit flies and worms(Fishman, 2001) This is partly due to easily discerned phenotypes generated

by random chemical or radiation mutagenesis The zebrafish has a powerfuladvantage over fruit flies and worms as it is a vertebrate Invertebrates do nothave direct analogs of biological systems found in vertebrates, such as a multi-chambered heart, neural crest cells and derivatives and kidney, thus imposinglimitations on the study of embryology, neurorobiology and endocrinology(Dooley and Zon, 2000) Furthermore, the molecular components of signallingpathways discovered by genetic screening in invertebrates cannot be simplyextrapolated to vertebrate structures For example, lipids which control germcell migration in fruit fly development, control heart precursor cell migration

in the developing zebrafish Also, since vertebrate developmental programmesare similar, the zebrafish is also useful for studying human development The

mouse, despite being a vertebrate, has its own disadvantages Thedevelopment of mouse embryos within the mother’s uterus makes itinaccessible for experimental manipulation and analyses, thus causinginconvenience to the study of early development genes In contrast, zebrafishembryos develop externally, thus allowing convenient access for manipulationand observation of early development, especially since they are opticallytransparent Therefore, developmental or phenotypic real-time analyses can bemade to the level of internal organs, and even the cell, during embryogenesis.

In exploitation of the optical transparency of the zebrafish embryo,technologies such as fluorescently tagged proteins and fluorescent resonanceenergy transfer (FRET) and cellular transplantation have been developed forthe physical tracking of cells or proteins, or for the monitoring of proteinactivity in the zebrafish embryo In addition, the zebrafish is permeable tosmall molecules in its aqueous environment, thus making it useful for thestudy of interactions between gene and environment (Fishman 2001)

The zebrafish is also useful for the study of human diseases since mosthuman genes have orthologs in zebrafish, and with parallel organ systems andthe conservation of body in vertebrates, zebrafish models for human diseaseshave been possible by mutations in orthologous zebrafish genes Althoughzebrafish are tetraploid due to a genomic duplication event during evolution,there was subsequent functional specialisation of some duplicated genes andloss of other genes, such that where evaluated, duplicated genes are notredundant in function, but rather, subdivide the function of the ancestral gene(Fishman 2001)

1.5 Zebrafish Early Developmental Stages

1.5.1 Zygote period

The zygote period starts from the newly fertilised egg and ends when

the first cleavage occurs (Kimmel et al., 1995) After fertilisation, the chorion

swells away from the egg, and cytoplasmic streaming, the movement of yolky cytosplasm towards the animal pole to segregate the blastoderm fromthe yolk-granule-rich vegetal cytoplasm, is activated This segregationcontinues into early cleavage stages

non-1.5.2 Cleavage period

During the cleavage period, the blastomeres undergo divisions that aremeroblastic, i.e the cell divisions incompletely undercut the blastoderm, andthe blastomeres or a specific subset of them remain interconnected by

cytoplasmic bridges (Kimmel et al., 1995).

1.5.3 Blastula period

The blastula period is marked by the ball-like appearance of theblastoderm at the 128-cell stage, and ends at the onset of gastrulation Duringthe period, the embryo enters the midblastula transition (MBT), the stage inwhich zygotic gene transcription is activated, the yolk syncytial layer forms,

and epiboly begins (Kimmel et al., 1995) The yolk syncytial layer is formed

by the deposition of nuclei and cytoplasmic contents by the collapse of themarginal tier of blastomeres in the early blastula The new marginal tier ofblastomeres, unlike their predecessors, is non-syncytial The YSL nucleiundergo mitotic divisions but remain syncytial Initially, the YSL forms a

moves beneath the blastoderm to form the internal-YSL (I-YSL) whichremains through embryogenesis to possibly play a nutritive role A portion ofthe YSL remains external (E-YSL), and it is currently understood to play animportant role in driving epiboly.

1.5.4 Epiboly

Epiboly is the first major morphogenetic process of gastrulation to

shape the developing embryo (Kimmel et al., 1995) Just before the onset of

epiboly, the late blastula consists of three main tissue layers – an outermostsingle cell epithelial layer termed the enveloping layer (EVL) that covers theblastoderm deep cell layer (DEL), and an innermost yolk syncytial layer (YSL)which the EVL is tightly attached at its margin to Epiboly is initiated at thesphere stage and epibolic movements thin and spread all three tissue layersvegetally such that the initial mound of cells sitting atop the yolk becomes acell multi-layer of nearly uniform thickness, and the yolk cell is covered allaround completely (100% epiboly), marking the end of epiboly This thinningand spreading of the blastoderm is accomplished by the movement of deeperblastomeres of the DEL outwards to intercalate between more superficialblastomeres of the DEL Such cell movements are termed radial intercalations,and along with the I-YSL, these movements are considered to be part of thedriving force of early epiboly

Considerable progress has been made in identifying factors involved inepiboly, but there is still very little understanding on how these factorscooperate to drive the process, and many gaps in knowledge of signallingmolecules and in understanding of mechanisms remain (Lepage and Bruce,2010)

1.5.5 Gastrula period

The gastrula period is characterised by the process of gastrulation,during which cell fate specification and massive tissue rearrangements occur,driven by widespread cell movement behaviours (Jessen and Solnica-Krezel,2005) Gastrulation is required to set up the adult body plan of organisms, toorganise germ layers and establish major body axes Besides epibolicmovements, internalisation and convergence-extension movements come intoplay during this period as well

The beginning of the gastrula period is marked by the initiation ofinternalisation, the movement of prospective mesodermal and endoderm cellsall around the circumference of the blastoderm margin beneath the superficialectodermal cells (Jessen and Solnica-Krezel, 2005) The germ ring formsduring this process, and subsequently, the embryonic shield, a thickening ofthe blastoderm margin at the future dorsal side of the embryo, appears It isthought that ingression may be the main type of cell movement mediating theprocess of internalisation Following internalisation, cells migrate anteriorlytoward the animal pole and contribute to the anterior-posterior extension of theembryonic axis

1.5.6 Convergence and Extension (C&E) movements

C&E movements narrow all the germ layers mediolaterally, whilesimultaneously elongating the embryo along its anterior-posterior axis TheC&E movements of the mesoderm is well understood, and it has beenobserved that the mesoderm can be subdivided into four domains along thedorsoventral axis of the gastrula (Figure 1.2) each domain characterised by

signalling pathways that include Stat3 signalling, the non-canonical

WNT/PCP pathway and G-protein coupled receptor signalling (Yin et al.,

2009)

The most ventral region is termed the “no convergence no extension”zone where mesodermal cells are not involved in C&E movements, but

migrate along the yolk into the tailbud region (Yin et al., 2009) The lateral

region of the embryo consists of mesodermal cells undergoing slow C&E cellmovements, but which accelerate towards the dorsal midline The third C&Edomain is the region of the presomitic mesoderm located within six-celldiameters to the axial mesoderm This domain consists of cells undergoingmodest rates of C&E Lastly the most dorsal region where the axial mesoderm

is exhibits the same convergence rate as the adjacent domain, but exhibits athree-fold higher rate of extension

The ventral mesoderm and lateral mesoderm, which display slow andmodest to fast rates of C&E, are characterised by the directed migration of

mesodermal cells in these regions (Yin et al., 2009) In directed cell migration,

cells migrate in an oriented fashion as individuals or in groups withoutsignificant neighbour exchanges In the lateral mesoderm, cells undergochanges in rates and directions of cell migratory movements depending on thestage of gastrulation At midgastrulation, cells migrate in the dorsal directionalong complex trajectories and therefore give rise to slow C&E movements.During late gastrulation, the cells have reached more dorsal locations wherethey pack densely together and exhibit a mediolaterally elongated morphology.Thus they converge towards the dorsal midline collectively along more directtrajectories and at higher speeds

Mediolateral intercalation of cells is the main cell movement driving

C&E in the axial mesoderm (Yin et al., 2009) In the process of mediolateral

intercalation, cells become elongated in morphology and membrane protrusiveactivity in the mediolateral directions is activated Simultaneously, these cellsmove in between their immediate medial and lateral neighbours, therebygenerating fast rates of mediolateral narrowing and anterior-posteriorlengthening of the region

Radial intercalations, besides driving epiboly, have also a role to play

in C&E of the medial presomitic mesoderm (Yin et al., 2009) As radial

intercalations preferentially separate anterior and posterior neighbouring cells,anisotropic extension of the tissue is enabled, thus contributing to the anterior-posterior extension of the embryonic axis

As in the case of epiboly, gaps in knowledge and understanding ofmolecular mechanisms of C&E movements on the tissue and cellular levelsstill remain, and identification and analyses of signalling molecules andpathways involved in the regulation of C&E movements will continue being

an important area of research (Jessen and Solnica-Krezel, 2005)

Figure 1.2: The four domains of mesodermal C&E movements in the zebrafish

gastrula and their characteristic underlying cell movement behaviours NCEZ,

no convergence no extension zone; A, anterior; P, posterior; D, dorsal; V,

ventral, PSM, presomitic mesoderm This figure is adapted from Yin et al.,

2009

1.6 Rationale and objectives of study

Although some target proteins of BNIP-2 have been identified andinsights into its cellular functions have been gleaned from the studiesconducted on it so far, the knowledge and understanding of the molecularsignalling pathways and mechanisms mediating or mediated by BNIP-2function remain poor The ability of BNIP-2 to affect cell dynamic behaviours,

to regulate different Rho GTPases, bind different GAPs and GEFs as well asinteract with a diversity of other proteins, makes it an intriguing subject ofstudy as it may potentially fill in the gaps in knowledge on how the “3G”(GTPase, GEF, GAP)-signalome is regulated by cellular factors and how itsfunctions and regulation are linked to other signaling networks (Pan and Low,2012)

This study seeks to investigate the physiological importance of the

bnip-2 gene in an in vivo model, the zebrafish, in the context of early

development, when a well-studied plethora of signalling molecules andpathways are activated and regulated to mediate developmental processes.Given the versatility of BNIP-2 in protein interactions, it is highly plausiblethat it engages different proteins to regulate or mediate different biologicalprocesses depending on the specific context in development Thus studying

the role of bnip-2 in development facilitates the understanding of the contextual signalling ability of bnip-2.

The aim of this study is to identify the developmental processes bnip-2

is involved in, and to determine interacting genes and signalling pathways itengages to mediate the developmental process The process of gastrulation is

paid attention to in particular, as it is driven by widespread cell movementbehaviours and dynamics that inevitably involves regulation by the 3G-

signalome and therefore very possibly requires the function of bnip-2 as well

There are two main objectives to this study One is to knockdown the

function of bnip-2 by morpholino and analyse the resulting phenotype using a

variety of cell and molecular biology methods to identify the developmental

process affected, i.e the process bnip-2 has a role in After identifying the biological process affected, functionally interacting genes of bnip-2 will be determined in order to understand bnip-2 function in the context of a

signalling pathway This will be done by analysing possible aggravation or

suppression of morphant bnip-2 phenotype resulting from co-knockdown of

genes and phenotype rescue experiments

1.7 Experimental rationale and approaches

1.7.1 bnip-2 knockdown by morpholino phosphorodiamidate antisense

oligonucleotides

One of the most direct ways to discover the function of a gene or theprotein it encodes is to observe the phenotypic outcome when the gene or theprotein it encodes is removed in an organism, resulting in a mutant Inclassical or forward genetics, random mutagenesis is conducted with DNA-damaging agents to generate a large number of mutants with various mutations

in different parts of the genomes, and genetic screens are conducted to identifyand isolate a mutant with a defect or phenotype of interest Following that,molecular characterisation is carried out to identify the gene or genesresponsible for the altered phenotype

Reverse genetics has become an approach popularly used due to thelarge-scale genome sequencing conducted in numerous genome projectsundertaken and completed in recent years This has led to an influx of largeamount of new DNA sequence information into public databases and therefore,the investigation of gene function often begins with the DNA sequence of thegene In the reverse genetics approach, the starting point could simply be agenome sequence, a cloned gene, or a protein of interest from which theencoding gene or nucleotide sequence must be first identified, as had beendone for BNIP-2 For gene functional studies, a powerful approach is tomanipulate the activity of the gene to study the effect of gain-of- or loss-of-function of the gene

As have been mentioned, based on earlier findings of BNIP-2, wehypothesised that BNIP-2 is involved in GTPase-mediated signallingpathways that regulate cell dynamics and is therefore potentially involved inthe developmental process of gastrulation, in which widespread cell movementbehaviours constitute the driving force To test our hypothesis, we chose toperform BNIP-2 knockdown in zebrafish and conduct phenotypic analysesduring early developmental stages, since that is when gastrulation processesmainly occur This loss-of-function approach would provide valuableinformation about the early developmental significance of BNIP-2 in zebrafish.The accessibility and optical transparency of the zebrafish embryo will greatlyfacilitate morphant phenotypic analyses and therefore facilitate theunderstanding of BNIP-2 function on the organism, tissue, cell and evenprotein level.

The method of BNIP-2 knockdown we employed is the use ofmorpholino phosphorodiamidate antisense oligonucleotides, or in short,morpholinos (MOs), which are short synthetic oligonucleotides consisting ofabout 25 subunits The MO nucleotides are similar to conventional DNA orRNA nucleotides, except that they possess a six-membered morpholine moietyinstead of a deoxyribose or ribose ring, and that they are connected via non-ionic phosphorodiamidate linkages instead of anionic phosphodiester linkages,thereby yielding a net neutral charge for the molecule The morpholine ringallows MOs to undergo Watson-Crick base pairing, but the linkages arenuclease- and protease-resistant and therefore stable in biological systems, andthe uncharged backbone renders it less likely to interact with other cellularproteins and components in an unspecific manner and cause toxicity (Eisenand Smith, 2008) Another advantage of the MO is that it has excellent RNA-

binding affinity and anti-sense efficacy, for e.g., at 50 nM concentration of

MO, there is more than 90 per cent inhibition of a luciferase reporter gene incell-free translational systems - more efficacious than the widely usedphosphorothioate oligonucleotides (Sumanas and Larson, 2002) Furthermore,MPOs display sequence-specific inhibition over a thirty-fold widerconcentration range than phosphorothioate oligos in cell-free translationsystems (Sumanas and Larson, 2002)

Translation-blocking MOs are designed to bind to a region between the5’ cap and about 25 bases 3’ of the AUG translation start site, and in anRNaseH –independent mechanism, elicit functional knockdown by blockingtranslation of the gene mRNA into protein through sterically blocking thetranslation initiation complex (Summerton, 1999) An alternative strategy isthe use of splice-blocking MOs, which are designed to bind to splice sitesthereby inhibiting pre-mRNA splicing or causing exon skipping, resulting in adefective protein upon translation

MOs are introduced into zebrafish embryos via microinjection into thecenter of the yolk to reduce the chance of secondary effects due to amechanical disruption of the early blastomeres, and by the process of

cytoplasmic streaming, they are transported into the embryonic cells (Bill et

al., 2009) Because MOs are small in size and are neutral in charge, diffusion

is the main driving force of spread throughout the embryo, facilitated by thecytoplasmic bridges present between early embryonic cells at the 1- to 8-cell

stages (Bedell et al., 2011, Bill et al., 2009) Although it has been reported that

MOs microinjected in this fashion can be ubiquitously delivered to allembryonic cells up to the 8-cell stage (Nasevicius and Ekker, 2000), we chose

to microinject the MOs at the single cell stage to best ensure the evendistribution of MOs into blastomeres formed by successive cleavages asdevelopment progresses Also, it is currently understood that MOs are mostefficient the first three days of development and the efficacy decreasethereafter due to dilutions caused by on-going cell divisions and perhaps,excretion (Sumanas and Larson, 2002) This time-frame of MO efficacy is

acceptable for our studies on bnip-2, as with most genes involved in vertebrate

development, because most of the patterning, morphogenesis andorganogenesis in zebrafish occur during the first two to three days of zebrafishdevelopment (Sumanas and Larson, 2002)

We designed two independent and non-overlapping 25-mer

translation-blocking MOs for bnip-2: bnip-2 MO1, which targets the translational start site, and bnip-2 MO2, which is complementary to the 5’UTR of bnip-2 (Figure 1.3) Both MOs prevent the translation of all the bnip-2 splicing isoforms in

zebrafish as the isoforms share the same 5’UTR Presently, due to theunavailability of an antibody for zebrafish BNIP-2, the effect of these MOs onBNIP-2 protein level could not be verified However, using human poly-BNIP-2 antibodies and lysates obtained from 36 to 48 hpf embryos injectedwith the same MOs at the one-cell stage, our laboratory had previously shown

that both bnip-2 MO1 and MO2 resulted in a decrease in BNIP-2 protein level This demonstrated that bnip-2 MO1 and MO2 can indeed inhibit bnip-2

translation

Control experiments are essential in order to prevent spurious resultsarising from non-specific interactions or ‘off-target’ effects of MOs that affectthe production of an unrelated gene product and result in a phenotype that is

only partially a result of the gene of interest (Eisen and Smith, 2008) Acontrol MO could be the standard control MO, which is a MO that targets anexogenous gene not present in the species used which for our case is thezebrafish An example of a commercially available standard control is onedirected against human β-globin pre-mRNA However, the standard controlcontrols only for general toxicity and embryo handling; potential ‘mis-targeting’ of the MO is not addressed (Eisen and Smith, 2008) Therefore, toaddress the latter issue, we designed two non-overlapping MOs, as the chance

of two MOs having the same off-target effect is significantly lower (Eisen andSmith, 2008) In addition, if synergy occurs between the two MOs co-injected

at respective sub-optimal doses (that do not elicit phenotypes on their own)resulting in the same phenotype as when injected independently, greaterconfidence in the phenotype observed can be obtained (Eisen and Smith,2008)

Instead of a standard control, we designed a control mismatch MO that

differs from one of the bnip-2 MOs, i.e bnip-2 MO1, by five nucleotides

(termed a five-nucleotide mismatch MO) (Figure 1.3) As it more closely

resembles the bnip-2 MO, it is a more stringent control compared to the

that retain the same amino sequence) at five nucleotides were introduced into

zebrafish bnip-2 mRNA at the region targeted by bnip-2 MO1 to prevent it

from binding to bnip-2 MO1 and titrating it out The mutated bnip-2 mRNA synthesised in vitro was co-injected with bnip-2 MO1 and the phenotype produced was assessed and compared with that of independent bnip-2 MO1 injection If the morphant phenotype could be rescued with bnip-2 mRNA, it showed that the phenotype produced by bnip-2 MO1 was specific to bnip-2

knockdown

Techniques such as live imaging, whole mount in situ hybridisation(WISH), immunofluorescence and western blot were employed for phenotypeanalyses (Figure 1.4)

MO Length Sequence (5’ to 3’) Mechanism

bnip-2 MO1 25bp TAAGCTCCACCCCCTCCATCCTCAG

Figure 1.3: Description of morpholinos used for functional rescue

experiments (top) Part of bnip-2 mRNA sequence (first 108bp) targeted by

morpholinos; highlighted in yellow: bnip-2 MO2 target sequence in the 5’UTR region; highlighted in cyan: bnip-2 MO1 target sequence at ATG translational start site (bottom) Table outlines details of bnip-2 MO1, bnip-2

MO2 and the 5 base pair mismatch MO

Figure 1.4: Schematic outline of experiments performed to elucidate bnip-2’s

function in zebrafish Arrows indicate chronological order of experiments

Phenotypic analysis of morphants

• In situ analyses with specific markers

• Immunofluorescence of possible

affected genes

• Imaging of embryo morphology, cell

migration, cell shape

• G-LISA to measure active

Cdc42/RhoA GTPase

Further manipulation of morphants

• Rescue of phenotype

• Co-injection of candidate interacting gene morpholinos, mRNAs

1.7.2 Investigating potential bnip-2 interacting genes – E-cadherin, RhoA, Cdc42

Previous (unpublished) work on zebrafish bnip-2 included a pull-down

of directly or indirectly (in multi-subunit complexes) interacting proteins inzebrafish lysate by GST-tagged zebrafish BNIP-2 bound to glutathione-Sepharose beads The interacting proteins identified in this experiment wereE-cadherin, RhoA, Cdc42 and Bcl-2 The identification of interacting proteinshas traditionally been important in gene functional studies because, in abiological context, proteins do not function independently, but in interactionwith other proteins in signalling pathways Therefore a protein has to bestudied in context with its interacting partners in order to fully understand itsphysiological function Thus, hypotheses about BNIP-2’s function may beformed by extrapolation of the functions of its interacting partners

Taking into consideration that the biochemical and cellular functions

of BNIP-2 as a core regulatory protein in multiple signalling gateways havebeen delineated (mainly in cell culture models) (Pan and Low, 2012), the priorfinding of possibly interacting proteins of BNIP-2 led to the formation ofresearch questions: Does the physical interaction of BNIP-2 with theseproteins mean that BNIP-2 operates in the same zebrafish developmentalsignalling pathways as these proteins? Is BNIP-2 involved in the regulation ofthese proteins in early zebrafish developmental processes such as thehypothesised gastrulation?

To answer these questions, further manipulations, experiments and

analyses were performed on bnip-2 knockdown zebrafish morphants to assess

possible effects on E-cadherin, RhoA and Cdc42 regulation (Figure 1.4)

2 Materials and Methods

2.1 Fish Spawning and Maintenance

Wild type zebrafish were obtained from a local supplier, maintainedand bred in line with standard procedures in a controlled environment (10 hourlight and 14 hour dark cycles) (Westerfield, 2000) Embryos spawned by thewild type zebrafish were incubated in petri dishes containing egg water (30gocean salt in 1litre of water) in a 28ͦ C incubator after being subjected toexperimental manipulations Staging of embryos were performed according tomorphological criteria (Kimmel et al., 1995)

2.2 Molecular Biology Techniques

2.2.1 RT-PCR Molecular Cloning

Full length pGEMT-easy-bnip-2 (exclusive of 5’UTR) had previously been cloned in the laboratory Primers used for the cloning of full-length bnip-

2 were designed based on Genbank listed Danio rerio Bcl-2/adenovirus E1B

19kDa interacting protein (Accession number NM_201218) Sequences of

primers used for the cloning of full length bnip-2b (restriction enzyme

sequences in bold): Forward - 5’ CGGGATCCATGGAGGGGGTGGA 3’ Reverse - 5’ CCGCTCGAGTTAAGTGAAAGCGATT 3’ Sequences of

primers used for the synthesis of bnip-2b mRNA containing five point

mutations (underlined): Forward - 5’CGGGATCCATGGAAGGAGTAGAA CTCAAGGAGGAGTGGCAGGATGAGG 3’ Reverse – 5’ CCGCTCGAGT

TAAGTGAAAGCGATT Primers were purchased from Research Biolabs,Singapore