ft183.pdf

Insights into neural crest development

from studies of avian embryos

SHASHANK GANDHI

and

MARIANNE E. BRONNER*

Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, USA

ABSTRACT The neural crest is a multipotent and highly migratory cell type that contributes to

many of the defining features of vertebrates, including the skeleton of the head and most of the

peripheral nervous system. 150 years after the discovery of the neural crest, avian embryos remain

one of the most important model organisms for studying neural crest development. In this review,

we describe aspects of neural crest induction, migration and axial level differences, highlighting

what is known about the underlying gene regulatory mechanisms. Past and emerging technologies

continue to improve the resolution with which we can examine important questions of neural crest

development, with modern avian molecular embryology continuing to make important contributions.

KEY WORDS

neural crest, migration, genetic toolbox, multipotent, neural plate border

Introduction

The neural crest is a transient, multipotent stem cell population

unique to vertebrates. Induced at the border of the neural plate and

the non-neural ectoderm, this cell population initially resides in the

elevating neural folds as neurulation progresses and subsequently

within the dorsal neural tube (Fig. 1). Shortly after neural tube

closure in avian embryos, neural crest cells undergo an epithelial-

to-mesenchymal transition (EMT), losing their cell-cell contacts

and transforming into a migratory cell type, distinct from both the

neural tube and epidermal cells (Bronner and Simões-Costa, 2016;

Le Douarin and Kalcheim, 1999; Hall, 2009; Sauka-Spengler and

Bronner-Fraser, 2008; Steventon

et al

., 2005). Post EMT, these cells

migrate extensively throughout the embryo along distinct pathways,

exhibiting both collective and individual cell migration behavior

in response to environmental cues and morphogenetic signals.

They subsequently give rise to numerous cell types, including

melanocytes of the skin, craniofacial cartilage and bones, smooth

muscle, Schwann cells, and sensory and autonomic neurons of the

peripheral nervous system (Le Douarin and Kalcheim, 1999; Le

Douarin and Teillet, 1971; Dupin

et al

., 2006; Grenier

et al

., 2009;

Hall, 2009; Kirby and Hutson; Minoux and Rijli, 2010; Theveneau

and Mayor, 2011). Despite the fact that the neural crest is derived

from the ectoderm, it has been referred to as the fourth germ layer

due to its ability to migrate long distances and differentiate into

such a plethora of derivatives (Le Douarin and Dupin, 2014; Hall,

2000; Shyamala

et al

., 2015).

Int. J. Dev. Biol. 62:

183-194 (2018)

https://doi.org/10.1387/ijdb.180038sg

www.intjdevbiol.com

*Address correspondence to:

Marianne M. Bronner. MC 139-74, 1200 East California Boulevard, Pasadena, CA – 91125, USA. Tel: +1 (626) 395-3355.

E-mail: mbronner@caltech.edu -

http://orcid.org/0000-0003-4274-1862

Submitted:

24 January, 2018;

Accepted:

24 January, 2018.

ISSN: Online 1696-3547, Print 0214-6282

Printed in Spain

Abbreviations used in this paper:

EMT, epithelial-to-mesenchymal transition; BMP,

bone morphogenetic protein; FGF, fibroblast growth factor; NPB, neural plate

border; RA, retinoic acid.

The history of neural crest research dates back 150 years,

when the Swiss anatomist Wilhelm His, Sr., identified a region at

the border of the neural tube and the non-ectoderm in early stage

chick embryos that differentiated into the cranial and spinal ganglia

(His W., 1868; Trainor

et al

., 2003). This cell population, which he

named

Zwischenstrang

(the intermediate cord) was later referred to

as the neural crest. Early grafting experiments in amphibians gave

the first insights into the developmental potential of these cells (rev.

Hörstadius S., 1950). Subsequently, lineage tracing experiments in

avian embryos using various methods of marking the neural crest

greatly enhanced our understanding of the pathways of migration

and derivatives they formed. These initially involved radioactive

labeling (Chibon, 1967; Weston, 1963), but were revolutionized by

the advent of quail-chick chimeric grafts (Le Douarin, 1980), which

allowed Le Douarin and colleagues to map neural crest migratory

pathways, derivatives, and their axial level of origin (Le Douarin,

1973; Le Douarin and Teillet, 1974, 1973). Contribution of the

neural crest towards the craniofacial architecture of vertebrates led

to the genesis of the “new head hypothesis” (Gans and Northcutt,

1983), according to which, the possession of distinct craniofacial

derivatives enabled diversification of the brain and acquisition of

predatory behavior in vertebrates, as compared to filter feeding

chordates (Le Douarin, 2004).

184

S. Gandhi and M.E. Bronner

Their stem cell-like regenerative properties, long-range migratory

capabilities, and experimental malleability make neural crest cells

an excellent model system to study cell migration, developmental

potential and cell-fate decisions (Dupin and Sommer, 2012; Smith,

1990; Tucker, 2004). Therefore, understanding how the neural

crest develops can help us gain insights into how cells behave

differently depending on their position and function in the organ

ism. The chick embryo has been a particularly valuable organism

for addressing these questions. As an amniotic system, it shares

similarity with humans at both the morphological and nucleotide

level. In this review, we discuss mechanisms underlying neural crest

specification, migration and differentiation, and the genetic toolbox

available to study these biological processes in chick embryos.

Induction of the neural crest in avian embryos

Until the mid-2000s, the prevailing view of neural crest induction

was that interactions between the non-neural surface ectoderm,

neural plate, and/or underlying mesoderm resulted in specifica

tion of neural crest precursors via signaling interactions between

tissues (Basch

et al

., 2004; Gammill and Bronner-Fraser, 2002;

Monsoro-Burq

et al

., 2003; Takahashi

et al

., 2007). Moreover, mul

tiple juxtaposition studies performed in both avian and amphibian

embryos highlighted the role of the epidermis in specifying neural

crest cells. However, most of the early experiments in chick em

bryos were performed after the neural tube had already formed. In

2006, Basch

et al.,

presented a revised view on the specification

and induction of the neural crest precursors (Basch

et al

., 2006).

They hypothesized that regions of the epiblast were already in

duced to become presumptive neural crest during gastrulation,

better matching experimental evidence in frog and zebrafish. The

comparatively slow development of the chick embryo enabled

more refined analysis of the timing of neural crest induction. By

characterizing the expression level of the paired box transcription

factor Pax7 in gastrulating chick embryos, Basch and colleagues

concluded that the Neural Plate Border (NPB) region was already

fated to become neural crest prior to completion of gastrulation,

and that cells from this domain incorporate into the neural folds,

and ultimately, the migratory neural crest. Dissection of the Pax7

expressing domain from the epiblast followed by explantation in a

neutral environment was sufficient to give rise to migratory neural

crest, identified by staining for the migratory cell-surface antigen

HNK1. Finally, using molecular markers for mesoderm, they also

demonstrated that the neural crest was specified without any in

teraction between the Pax7-expressing NPB and the underlying

mesoderm. These results not only characterized Pax7 as an early

NPB marker, but also placed the initiation of neural crest specifica

tion at or before gastrulation. More recently, Roellig and colleagues

have expanded this viewpoint by resolving the co-expression pat

tern of transcription factors including Pax7, Sox2, and Six1 in the

NPB at a single-cell level (Roellig

et al

., 2017).

The specification of neural crest cells has been described as a

multistep process involving Fibroblast Growth Factor (FGF), Bone

Morphogenetic Protein (BMP), and the Wnt signaling cascades,

with additional contributions by Notch/Delta, Retinoic Acid (RA),

Hedgehog, and endothelin signaling (Barembaum and Bronner-

Fraser, 2005; Basch

et al

., 2004; Knecht and Bronner-Fraser, 2002;

Steventon

et al

., 2009). Of these, FGF signaling, which arises from

the surrounding mesoderm, acts in conjunction with Wnt signaling to

repress BMP signaling during the first half of neural crest induction

(Stuhlmiller and García-Castro, 2012). In the second half, inhibi

tion of FGF signaling allows for the activation of BMP signaling,

which together with Wnt signaling activates a pool of transcription

factors referred to as NPB specifiers. These NPB specifiers, that

include transcription factors Pax3/7, Msx1/2, Pax7, Zic1, Gbx2,

and Tfap2, play an important role in establishing the interface

Neural

tube

Delaminating

neur al cr est

Epider

mis

Neural

plate

Non-neu

ral

ectoder

Neural

plate

bor der

Notochor

Neural

fold

Premigrator

neur al cr est

Migratory

neur al cr est

Fig. 1. The various stages of neural crest development.

The neural crest

is a transient population of multipotent cells found in vertebrates. Neural

crest cells are specified at the border between the neural plate (blue) and

non-neural ectoderm (yellow) at the gastrula stage. During neurulation,

the neural plate border elevates to form the neural folds containing pre

migratory neural crest cells. As the neural folds fuse to form the neural

tube, neural crest cells undergo an epithelial-to-mesenchymal transition

(EMT), delaminate from the dorsal neural tube, and migrate extensively

to various parts of the developing embryo where they differentiate into a

plethora of derivatives.

Neural crest development in avian embryos

185

between the neuroepithelium and the surface ectoderm (Grocott

et al

., 2012; Khudyakov and Bronner-Fraser, 2009). Once the NPB

is established, the NPB specifier genes activate another set of

transcription factors called the neural crest specifier genes, includ

ing FoxD3, Sox9, Snail, and Sox10 (Simões-Costa and Bronner,

2013, 2015; Simões-Costa

et al

., 2014). Of these, the transcription

factor Forkhead Box protein D3 (FoxD3) appears to be expressed

first and is required for the subsequent onset of other neural crest

specifier genes (Labosky and Kaestner, 1998; Sasai

et al

., 2001;

Sauka-Spengler and Bronner-Fraser, 2008; Stewart

et al

., 2006).

It is expressed in both premigratory and some migratory neural

crest cells, regulating their EMT and thereby controlling emigration

(Fairchild

et al

., 2014; Kos

et al

., 2001). However, detailed genomic

analysis has revealed that the expression of FoxD3 in the cranial

versus the trunk axial level is tightly regulated by direct input from

different transcription factors into two different enhancers, NC1

and NC2 (Simões-Costa

et al

., 2012). Interestingly, the NC1 en

hancer governs the expression of FoxD3 in the premigratory cranial

neural crest, while the NC2 enhancer is initially active in the trunk

neural crest. In contrast, the neural crest specifier gene Sox10 is

expressed as neural crest cells prepare to emigrate and then is

maintained in the migratory population (Betancur

et al

., 2010). It

also directly or indirectly regulates the activity of proteins like Rho

GTPases that are actively involved in actin reorganization, thus

contributing to cell membrane fluidity, and ultimately, cell migration

(Liu and Jessell, 1998; Sit and Manser, 2011), as discussed in

detail below. Later, Sox10 plays an important role in differentiation

of neural crest cells into specific neuronal cell types (Carney

et al

2006; Kim

et al

., 2003).

Contrary to the induction model of neural crest specification, a

recent study in

Xenopus

embryos has proposed that neural crest

cells retain multipotency from an early blastula to the neurula stage

(Buitrago-Delgado

et al

., 2015). Interestingly, this model seems

consistent with findings in avian embryos, where expression of the

earliest neural plate border marker, Pax7, was observed in distinct

regions of epiblast between stages 3 and 4 (Basch

et al

., 2006).

Neural crest migration pathways along the body axis

of the chick embryo

Following delamination from the dorsolateral part of the neural

tube, neural crest cells undergo a change in their transcriptional

landscape, assuming a state distinct from both the neural tube

and the epidermis. This migratory state is marked by a reduction

in cell-cell adhesion and enhanced interactions with the extracel

lular matrix. In chick embryos, the process of delamination overlaps

with neural tube closure (Le Douarin and Kalcheim, 1999; Duband

and Thiery, 1982; Théveneau

et al

., 2007), and differs along the

rostral-caudal axis. In the rostral part of the embryo, the cranial

neural crest cells undergo EMT as a collective event, resulting in

the delamination of numerous cells concurrently. They then follow

distinct streams and migrate extensively with leader cells pioneer

ing the pathways followed by closely-associated follower cells. In

contrast, trunk neural crest cells delaminate by leaving the neural

tube one cell at a time in a drip-like fashion and start migrating

immediately after detaching from the tube. Interestingly, neural

crest migration in the trunk is tightly linked with differentiation of

the somites (Sela-Donenfeld and Kalcheim, 1999). As neural crest

cells first leave the neural tube, they migrate between the neural

tube and the adjacent epithelial somite. The somites subsequently

undergo an EMT to form the dermomyotomes and sclerotomes.

At this point, trunk neural crest cells invade the sclerotomes and

migrate throughout the anterior half of each, resulting in their seg

mental migration (Bronner-Fraser, 1986; Rickmann

et al

., 1985).

In the chick, neural crest cells typically begin to delaminate about

five somite lengths above the last formed somite. The caudal

trunk neural crest delaminates almost a day after the completion

of neurulation (Osorio

et al

., 2009).

Neural crest migration is a dynamic process. For example, the

chick cranial neural crest cells emigrate from the dorsal neural tube

immediately after its closure in a rostral-caudal wave, with cells at

the midline of the caudal forebrain and midbrain emerging prior

to those adjacent to the hindbrain. Not only are there differences

in the migratory properties of cells originating from different axial

levels, but the leader and trailer cells can also exhibit differences

in migration rates and directionality. At the hindbrain level, these

differences correlate with segmentation of the rhombomeres (Kulesa

and Fraser, 1998, 2000; Kulesa and Gammill, 2010; Lumsden

., 1991). Neural crest cells migrating towards branchial arch 1

and 3 exhibit collective cell migration with high directionality, where

the leader cell moves at a speed similar to the cells following in

the chain. In contrast, cells migrating into branchial arch 2 display

more directed, individual cell movement with low cell-cell contacts,

albeit at much slower migration speeds (Kulesa and Fraser, 2000).

The neural crest cells that originate from rhombomeres 3 and 5

migrate from the dorsal midline of the neural tube, following a

diagonal cellular trajectory in a caudo- or rostro-lateral manner

before migrating laterally away from the neural tube, thus avoiding

inhibitory environments immediately lateral to r3 and r5. Interestingly,

these cells end up getting stalled as they reach the lateral-most

extent of the neural tube. Eventually, they either merge with their

neighboring stream from rhombomere 4 migrating towards the

second branchial arch, or they remain stationary, suggesting that

the microenvironment adjacent to r3 is inhibitory for neural crest

migration, thus establishing a neural crest cell-free zone. However,

cells that merge with their neighboring stream maintain filipodial

contacts with the leader cells, allowing them to alter the direction

of their migration towards the second branchial arch (Kulesa and

Fraser, 1998, 2000).

Trunk neural crest migration is restricted to two major migratory

pathways – ventral, and dorsolateral. In the chick, neural crest

cells first migrate ventrally, since repressors such as Slit ligands,

endothelin-3, and ephrinB1 are initially expressed along the dor

solateral pathway, thus restricting neural crest cells to follow a

ventral migratory pathway and migrate through the inhibitor-free

anterior part of the sclerotome (Harris and Erickson, 2007). The

posterior sclerotome, on the other hand, expresses members of

ephrin and semaphorin families, along with inhibitory Extracellular

Matrix (ECM) molecules, Versicans, and F-spondin (Newgreen and

Gooday, 1985; Newgreen

et al

., 1986). The underlying mechanism

of repression arises from the interaction between the migratory

neural crest receptors

eph

and

neuropilin2

and their antagonistic

ligands EphrinB and Semaphorin, respectively. Indeed, when

neural crest cells from the trunk axial level were cultured in dishes

containing fibronectin-coated matrices with alternating ephrin

stripes, the cells moved along the regions with no ephrin, while

completely avoiding ephrin coated stripes (Davy and Soriano,

2007; Krull

et al

., 1997; Wang and Anderson, 1997). Migration of

186

S. Gandhi and M.E. Bronner

these cells is also controlled temporally in a chemotactic fashion

by the levels of Stromal cell-derived factor 1 (SDF1) along the

ventro-dorsal axis (Kasemeier-Kulesa

et al

., 2010). Neural crest

cells that differentiate within the sclerotome give rise to the dor

sal root ganglia, with high levels of Notch leading to a glia fate,

while high levels of Notch’s ligand Delta lead to a neuronal fate.

Expression of the chemokine (C-X-C motif) receptor 4 drives the

first wave of ventrally-migrating trunk neural crest cells towards

the dorsal aorta (Saito

et al

., 2012). Cells that lack the expression

of CXCR4 continue to migrate and give rise to sympathoadrenal

cells that then condense to form sympathetic ganglia as well as the

adrenal medulla. BMP signaling from the dorsal aorta influences

these cells to become sympathoadrenal while interactions with

presumptive adrenal cortical cells influence further differentiation

into the adrenal medulla (Saito and Takahashi, 2015).

The second migratory pathway, the dorsolateral pathway, is

followed by neural crest cells that are fate-restricted to become

pigment cell progenitors in a FoxD3-dependent manner (Tosney,

2004). Low levels of FoxD3 allow proper regulation of MITF, a

transcription factor necessary for the differentiation of neural crest

cells into melanocytes (Thomas and Erickson, 2009). Once the

pigment precursors are specified, ephrin mediates the increase

in levels of its receptor, EphB2, on the surface of these cells, fa

cilitating their migration through regions that initially inhibited the

migration of multipotent trunk neural crest cells.

Regionalization along the anterior-posterior axis

Neural crest cells migrate long distances in response to both

environmental cues and various signaling cascades (Kulesa and

Gammill, 2010). Once they reach their final destination, the cells

can differentiate into a wide array of cell types. The pioneering work

of Le Douarin and colleagues showed that the neural crest can be

categorized into four unique, albeit overlapping regions along the

body axis in avian embryos, based on their axial level of origin (Fig.

2). The cranial neural crest, also known as the cephalic neural crest,

includes the forebrain, midbrain and the anterior hindbrain region;

the vagal neural crest spans from the posterior hindbrain region to

For ebrain

Midbrain

Hindbrain

Car diac

neur al

cr est

Vagal

neur al

cr est

Trun k

neur al

cr est

Lumbosac

ral

neur al cr est

Phar yngeal

Ar ches

PA1

PA2

PA3

PA4

Somites

Cran ial

neur al

cr est

Cran

ial

neura

l crest:

Cran iofacial

car tilage&

bone

Chondr

ocytes

Glia and neur ons of cra nial ganglia

Odontoblasts

Melanocytes

Card

iac

neura

l crest:

Car diac out

ow tract

Myocytes

Parasympa

thetic

car diac neur ons and glia

Pericytes

for car diac septum

Vaga

l neur

al crest

Parasympa

thetic

ganglia

of the gut

Trun

k neura

l crest:

Dorsal

root ganglia

Sensory

neur ons

Sympathet

ic ganglia

Adr enal medulla

Melanocytes

Lumbo

sacr

al neura

l crest:

Parasympa

thetic

ganglia

of the gut

Fig. 2. Neural crest derivatives along the

Anterior-Posterior axis in chick embryos.

Neural crest subpopulations have been des

ignated based on their axial level of origin

along the anterior-posterior axis. Cranial

neural crest cells migrate from the cranial

neural tube into the pharyngeal arches and

give rise to craniofacial cartilage and bone, as

well as glia and neurons of the cranial gan

glia. The cardiac neural crest arise from the

neural tube adjacent to the mid-otic region

to somite 3 (marked in red), and plays a key

role in cardiac outflow tract septation. Vagal neural crest spans from adjacent to somite 1 through 7 (marked in blue), and together with the lumbosacral

neural crest, arising from the region posterior to somite 28 (marked in dark green), form the parasympathetic ganglia of the enteric nervous system.

Trunk neural crest arises from the neural tube adjacent to somite 7 through 28 (marked in orange), and contributes to dorsal roote and sympathetic

neurons and glia, melanocytes, and the adrenal medulla. A summary of the main derivatives formed by the entire neural crest population is listed.

Neural crest development in avian embryos

187

somite 7; the trunk neural crest comprises somite 8 – 28; and the

lumbosacral neural crest represents the region posterior of somite

28. Within the vagal neural crest, there is another subpopulation

called the cardiac neural crest that arises from the neural tube

adjacent to the mid otic-placode to somite 3. The cranial neural

crest contributes to the craniofacial skeleton, and the glia and

neurons of the cranial ganglia (D’amico-Martel and Noden, 1983;

Le Douarin and Kalcheim, 1982). The vagal neural crest populates

the gut, and the neurons and glia of the enteric nervous system.

The cardiac neural crest gives rise to the cardiac outflow tract and

the pharyngeal arches, and is the only neural crest subpopulation

that contributes to proper cardiovascular development (Stoller and

Epstein, 2005). The trunk neural crest cells form the dorsal root

ganglia, sympathetic ganglia, and the adrenal medulla. Finally, the

lumbosacral neural crest also contributes to the enteric nervous

system. Interestingly, regardless of the axial identity, neural crest

cells also differentiate into melanocytes, neurons, and glia (Le

Douarin and Kalcheim, 1982).

Axial level differences in differentiative ability are highlighted by

transplantation experiments performed in chick embryos, where

cranial and trunk neural crest cells were exchanged. Cranial neural

crest cells give rise to the craniofacial skeleton, and happens to be

the only neural crest population capable of doing so. When these

cells were transplanted to the trunk, they not only compensated

for the absence of trunk neural crest cells, but also differentiated

into cartilage nodules (Le Douarin and Teillet, 1974; Le Lievre

., 1980). In contrast, trunk neural crest cells failed to give rise to

any cartilage when transplanted to the cranial level, even though

they contributed to the neurons and glia of the cranial ganglia

(Nakamura and Ayer-le Lievre, 1982). Indeed, genomic analysis

revealed that the endogenous levels of the neural crest specifier

genes FoxD3 and Sox10 are controlled by different set of enhancers

at the cranial and trunk axial levels (Betancur

et al

., 2010, 2011;

Simões-Costa

et al

., 2012). Not only that, the transcriptional inputs

these enhancers receive from their upstream genes are also dif

ferent at the two axial levels (Simões-Costa

et al

., 2012). Taken

together, these results suggest that there are intrinsic differences

between the developmental fate of the two populations.

Recently, it has been possible to “reprogram” neural crest axial

identity by tweaking the underlying genetic circuitry. Simões-Costa

and Bronner revealed a cranial crest-specific sub-circuit by perform

ing transcriptional profiling of the migratory crest from the cranial

and trunk axial levels (Simões-Costa and Bronner, 2016). They

discovered that ectopic expression of a subset of this sub-circuit

in the trunk, namely the transcription factors Sox8, Tfap2b, and

Ets1, was sufficient to alter the fate and identity of trunk neural

crest into ‘cranial-like’ cells. The altered identity was assessed by

testing the ability of the trunk neural crest cells to activate

Sox10e2

an enhancer that governs the expression of Sox10 selectively in

the migratory cranial neural crest cells (Betancur

et al

., 2010). To

investigate the effect on trunk neural crest cell fate, they transplanted

cells expressing the three factors from GFP donor embryos into

wild-type chick embryos. Following incubation until embryonic

day 7, they observed that cells transfected with the cranial neu

ral crest-specific sub-circuit had successfully differentiated into

chondrocytes. The control group, on the other hand, did not form

any part of the cartilage, suggesting that expression of the three

transcription factors was sufficient to drive the population of the

trunk neural crest towards a craniofacial derivative fate.

Multipotent or fate-restricted?

The advent of vertebrates has been suggested to overlap with

the appearance of two cell types, neural crest cells and ectodermal

placodal cells (Gans and Northcutt, 1983; Gee, 1996; Glenn North

cutt, 2005), which contribute to the facial skeleton, cranial ganglia

and sense organs. As the craniofacial skeleton and peripheral

nervous system of the head are defining features of vertebrates,

this suggests that the ability of neural crest cells to differentiate into

a multitude of cell types played an important role in the evolution

of vertebrates. This raises the intriguing question – how are neural

crest cells able to differentiate into such diverse cell types? Over

the last several decades, many hypotheses have been formulated

to address this question.

Multiple studies performed over the last three decades support

the idea that the neural crest is a multipotent population with the

ability to give rise to many or all potential derivatives. The first

piece of evidence

in vivo

came from Bronner-Fraser and Fraser

when they injected fluorescent Dextran molecules into individual

trunk and cranial neural crest cells before their delamination and

migration (Bronner-Fraser and Fraser, 1989, 1988). They observed

that the progeny of the injected cells were capable of differentiating

into several cell types such as sensory neurons, Schwann cells,

and melanocytes, thus suggesting that a part of the migratory trunk

neural crest was multipotent. They also reported the possibility of

a common precursor for both the Central Nervous System (neural

tube) and the Peripheral Nervous System (neural crest). Around the

same time, Baroffio and colleagues successfully cultured cranial

neural crest cells obtained from quail embryos and reported that the

cells exhibited a wide developmental potential, corroborating the

vivo

results obtained by Bronner-Fraser and Fraser (Baroffio

et al

1988). In addition, other clonogenic culture studies of neural crest

cells reported that their lineage decision was directly influenced by

the growth factors used in the culture medium (Lahav

et al

., 1998).

On the other hand, there is some evidence suggesting that the

neural crest may be a heterogeneous mixture of fate-restricted

cells. For example, Nitzan and colleagues have proposed that

premigratory neural crest cells are unipotent and arranged in the

dorsal neural tube in a spatio-temporal pattern in chick embryos that

correlates with the derivatives they will form (Nitzan

et al

., 2013).

It is important to note that the techniques used in the above-

mentioned studies suffered from some major drawbacks. First,

vivo

lineage tracing experiments using fluorescent Dextran only

allowed for the characterization of neural crest derivatives based

on visualization of a lineage marker that is transient and diluted

with each cell division. Second, clonal analysis

in vitro

requires

growing cells in a culture medium outside the context of the devel

oping embryo. However, Baggiolini and colleagues used a confetti

mouse model to trace neural crest cells and their derivatives

vivo

, allowing them to achieve permanent lineage labeling at

single-cell resolution and in an intact endogenous environment of

the embryo (Baggiolini

et al

., 2015). Using this model, they traced

nearly 100 neural crest cells, both premigratory and migratory, and

demonstrated that almost 75% of these cells gave rise to multiple

derivatives comprised of the dorsal root ganglia, melanocytes,

sympathetic ganglia, and Schwann cells. While they observed

that a small proportion of the cells gave rise to a single derivative,

the vast majority of the trunk neural crest cells were multipotent,

suggesting that the neural crest is primarily a multipotent popula

188

S. Gandhi and M.E. Bronner

tion and thereby confirming the work in chick embryos done by

Bronner-Fraser and Fraser some 25 years earlier.

Experimental strategies to study neural crest

development

Some of the earliest experiments performed in developmental

biology relied on ablation and grafting experiments to decrypt

the puzzle that was the developing embryo. After Wilhelm Roux

first published his ablation technique, Hans Spemann used it to

investigate the role of tissue interactions in eye development. At

the time, the process of lens formation was not thoroughly under

stood. By destroying the optic vesicle before it associated with

the overlying ectoderm, Spemann demonstrated the importance

of the interactions between the optic vesicle and the ectoderm in

inducing proper lens development.

In the 20

century, the ablation technique was adapted to the

chick embryo to address a diverse array of questions. It was

through the ablation of the cardiac neural crest that the contribu

tion of this population of neural crest cells to proper cardiovascular

development was highlighted (Besson

et al

., 1986; Kirby and

Waldo, 1995; Kirby

et al

., 1985; Nishibatake

et al

., 1987). How

ever, these experiments only revealed some of the developmental

capabilities of the neural crest tissue. Moreover, it was difficult

to conclude whether the presence or absence of structures after

neural crest ablation indicated their site of origin. However, tech

nical breakthroughs in biology have made it possible to address

such questions in much finer detail. Here, we discuss some of

the experimental techniques that are currently available to study

neural crest development in chick embryos.

Neural crest lineage tracing

One of the biggest technical challenges in neural crest biology

was the ability to distinguish a neural crest cell from its neighbors

while it migrated through mesenchymal regions in the embryo. As

a result, cell marking techniques that would allow investigators to

trace individual cells were highly sought after. Radioactive label

ing of nuclei with tritiated thymidine followed by transplantation

of labeled tissue in a host embryo allowed Weston (1963) and

Chibon (1967) to follow migration of these cells in avian and am

phibian embryos, respectively (Chibon, 1967; Weston, 1963). The

technique was soon adapted in a number of studies (Johnston,

1966; Noden, 1975, 1978; Weston and Butler, 1966). However, the

levels of tritiated thymidine diluted with each cell division event,

making this technique appropriate for only short-term tracing of

neural crest cells. Moreover, the method was also prone to giving

false labeling of non-neural crest cells, as the radioactive label

from a dead cell could spread to its neighbors. Therefore, it was

important to look for alternative strategies for neural crest labeling.

In the 1960s, Le Douarin devised the elegant quail-chick chi

meras as a method to label cells and follow the long-term fate of

neural crest cells, allowing her to trace neural crest pathways of

migration and characterize their numerous derivatives (Le Douarin,

1969, 1973). This technique was based on the difference in the

heterochromatin state of chick and quail embryos, which could

be labeled with a simple histological Feulgen stain. These experi

ments not only made it possible to follow the migratory pathways

of these cells throughout the embryo, but also provided valuable

insights into the extensive nature of neural crest derivatives. Using

this approach, Douarin categorized the derivatives of the neural

crest at different axial levels in chick embryos, leading to important

observations regarding differences in properties of neural crest

cells derived from different axial levels.

To confirm these results using methods that did not require

interspecies transplantation, the next generation of labeling

methods made use of fluorescently labeled large molecules such

as Lysinated Rhodamine Dextran (LRD) and vital dyes such

as DiI (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyamide

perchlorate). While these labeling methods were still transient,

they offered several advantages over radioactive labeling. For

example, both LRD and DiI were not prone to leak, hence reduc

ing false labeling of tissues surrounding neural crest cells. They

both were transferred to their daughter cells for a few generations,

allowing tracking of migration at single cell resolution, albeit for

short intervals. Most importantly, using these dyes to label cells

and track their migration within the same embryo circumvented

the need for grafting labeled cells in a host embryo, making their

application much simpler than their predecessors. These tech

niques were instrumental in demonstrating the multipotent nature of

neural crest cells not only in avian embryos (Artinger

et al

., 1995;

Bronner-Fraser and Fraser, 1989, 1988; Lumsden

et al

., 1991;

Serbedzija

et al

., 1989, 1990), but also in mammalian systems

(Osumi-Yamashita

et al

., 1996). These labeling techniques gave

valuable insights into the developmental potential of neural crest

cells. However, given their transient nature, it was not possible to

trace a single neural crest cell to the tissue to which it contributed

later on during development. Hence, it was necessary to establish

techniques that would not only allow labeling of single cells, but

would also ensure long-range tracking of their migratory pathway.

With their ability to generate double-stranded DNA that stably

integrates into the host cell genome, retroviruses offer an easy and

effective alternative to classical cell labeling techniques described

above. Replication Competent and Incompetent Avian Retroviruses

(RCAS/RIA) have long been used to permanently label single cells

and their progeny by expressing reporter genes through the viral

promoter (Frank and Sanes, 1991; Murai

et al

., 2015; Price

et al

1987; Sanes, 1989; Sanes

et al

., 1986). Injecting the viruses at

different developmental stages at precise locations in the embryo

allows spatiotemporal control over the labeling of neural crest cells

at different axial levels. Pseuodotyping the virus envelop protein

ensures that the infection is limited to chicken cells. Moreover,

these viruses have also been used to misexpress endogenous

genes in neural crest cells (Eames

et al

., 2004).

Crestospheres

Methods employing

in vitro

culturing of neural crest cells were

already established in the 1970s. However, it wasn’t possible until

very recently to successfully culture premigratory neural crest cells

in their multipotent and self-renewal state (Cohen and Konigsberg,

1975). After they migrate from the neural tube, neural crest cells

have a limited ability to self-renew for a few cell cycles (Stemple

and Anderson, 1992; Trentin

et al

., 2004), and thus rapidly transi

tion from stem cells to progenitor cells. However, more recently,

Kerosuo and colleagues have successfully optimized conditions

to culture premigratory neural crest cells as “crestospheres,” en

abling long term self-renewal and retention of multipotency of the

premigratory neural crest cells (Kerosuo

et al

., 2015).

Premigratory neural crest cells express a suite of neural crest

Neural crest development in avian embryos

189

specifier genes including FoxD3, Sox9, Snail2, and Sox10 (Simões-

Costa and Bronner, 2015; Simões-Costa

et al

., 2014). Kerosuo

and colleagues tested various culture conditions that not only

enabled proper expression of these transcription factors, but also

promoted the maintenance of their self-renewal properties and

stemness (Kerosuo

et al

., 2015). Using a cocktail of growth factors

that included epidermal growth factor, basic FGF, and RA, they

found optimal conditions to culture premigratory neural crest cells

in their multipotent state, as judged by the expression of neural

crest marker genes in the cultured crestospheres. Interestingly,

this medium also enabled culturing of crestospheres from entire

neural tube tissue, suggesting that the correct proportion of growth

factors supported culturing of neural crest at the cost of neural fate.

Using these conditions, they successfully maintained crestosphere

cultures for as long as 5 weeks, albeit at a declining proliferation

rate. By exposing them to specific differentiation media, Kerusuo

and colleagues also demonstrated the differentiation potential of

crestospheres into multiple neural crest derivatives.

Taken together, this technique of

in vitro

culturing of premigra

tory neural crest cells offers an interesting model to study neural

crest development. Not only can the crestospheres be used to

answer questions about the stem cell-like behavior of neural

crest cells, but they can also be used to investigate the role of

certain environmental cues in cell migration and differentiation.

By culturing crestospheres from GFP transgenic chick embryos

and transplanting them back into wild type embryos, one could

study the developmental potential of neural crest cells at single-

cell resolution.

Antisense morpholinos

The concept of using antisense oligonucleotides to study gene

expression was first proposed almost 20 years ago (Baker and

Monia, 1999; Crooke, 1999). The antisense technology relied on

the understanding of nucleic acid structure and the underlying

mechanisms governing their hybridization. Synthetic oligonucle

otides designed following these principles were capable of inhibiting

protein translation via three different pathways: first, by disrupting

ribosome assembly at the targeted mRNA; second, by blocking

splice junctions through direct hindrance; third, by recruiting RNase

H enzymes that degrade the target mRNA through recognition of

the synthetic oligonucleotide-mRNA duplex. While the specificity

of these oligonucleotides was a cause for concern, subsequent

structural and chemical modifications of the oligonucleotides made

this a useful technique for knockdown studies in different model

systems including

Xenopus

(Heasman

et al

., 2000), zebrafish

(Nasevicius and Ekker, 2000; Yang

et al

., 2001), and sea urchins

(Howard

et al

., 2001).

Antisense morpholino oligonucleotides were first implemented

in avian embryos for gene knockdowns when Kos and colleagues

demonstrated the role of FoxD3 in maintaining the neural crest-

derived melanoblast lineage (Kos

et al

., 2001). Since then, the

entire chick neural crest community has relied on morpholinos to

interrogate the role of different genes in neural crest development

(Barembaum and Bronner, 2013; Basch

et al

., 2006; Betancur

et al

., 2010; Simões-Costa

et al

., 2015, 2012). Important con

trols for establishing specificity of morpholinos include rescue

experiments and validating protein knock-down. However, given

the disadvantages of using morpholinos, including rising con

cerns about non-specific effects (Gerety and Wilkinson, 2011),

cost ineffectiveness, and inability to target non-coding regions,

the community has turned to CRISPR/Cas9 technology for more

robust gene knockouts.

CRISPR/Cas9-mediated genome editing

Less than a decade ago, genome editing approaches were limited

to arbitrary mutations incorporated as a result of Non-Homologous

End Joining (NHEJ) repair mechanism or Homologous Recombi

nation by using a donor plasmid containing the insert fragment.

Specificity of the DNA binding domain found in transcription factors

led to the discovery of DNA nucleases, proteins capable of induc

ing site-specific mutagenesis. Zinc Finger Nucleases (ZFNs) and

Transcription Activator-Like Effector Nucleases (TALENs) were

used extensively in a variety of animal models (Beerli and Barbas,

2002; Beerli

et al

., 2000; Zhang

et al

., 2011). However, the chick

community relied on anti-sense morpholinos (Corey and Abrams,

2001) and dominant-negative constructs for transient knockdown

of genes of interest. Even though the efficiency of these techniques

has been documented, expensive and tedious cloning procedures

for DNA nucleases, and non-specific effects for morpholinos

and dominant negative proteins limited the applications of these

techniques (Eisen and Smith, 2008; Joung and Sander, 2013;

Schulte-Merker and Stainier, 2014).

The discovery of CRISPR (Clustered Regularly Interspaced

Short Palindromic Repeats)-Cas9 has ushered in a new era of

molecular genetics and has allowed researchers to interrogate the

role of specific genes in development of an organism. CRISPR-Cas

is an important part of the prokaryotic immune response (Bhaya

., 2011; Wiedenheft

et al

., 2012) and has now been harnessed

for targeted genome editing in a variety of model systems (Cong

et al

., 2013; Dickinson

et al

., 2013; Ren

et al

., 2014; Stolfi

et al

2014). Mechanistically, a short RNA sequence, known as guide RNA

(gRNA), guides a

Streptococcus pyogenes

-derived endonuclease

Cas9 to specific target sites on the genome. The binding of the

Cas9-gRNA complex is a two-step process: first, Cas9 identifies

a Protospacer Adjacent Motif (PAM) of the form “NGG” on the

genomic DNA; second, the protospacer domain within the gRNA

forms Watson-Crick base pairing with the target. Once bound, Cas9

induces a double stranded break 3-4 base pairs upstream of the

PAM (Jinek

et al

., 2012; Mali

et al

., 2013a). The CRISPR system

has been widely used for knocking out genes (Cong

et al

., 2013;

Jiang

et al

., 2013; Jinek

et al

., 2013; Mali

et al

., 2013a; Shalem

., 2015; Wang

et al

., 2014), and knocking in short oligonucleotides

and fragments of interest (Cong

et al

., 2013; Dickinson

et al

., 2013).

A catalytically de-active variant (dCas9), with mutated nuclease

sites, has also been used to regulate expression of specific genes

using fused activation (e.g. VP16) and repression (e.g. KRAB)

domains (Cheng

et al

., 2013; Gilbert

et al

., 2013; Maeder

et al

2013; Mali

et al

., 2013b; Perez-Pinera

et al

., 2013; Qi

et al

., 2013).

We and others have recently optimized the CRISPR/Cas9

system for genome editing in chicken embryos using a three-fold

optimization strategy (Gandhi

et al.,

2017b; Williams

et al

., 2018).

First, using an in-situ hybridization and quantitative Reverse Tran

scription PCR (qRT-PCR)-based assay, we demonstrated that a

much higher level of gRNA transcription can be achieved using

a chicken-specific U6 compared to its human counterpart (Kudo

and Sutou, 2005; Wise

et al

., 2007). Second, following strategies

proposed by Chen and colleagues (Chen

et al

., 2013), we imple

mented the ‘F+E’ (‘F’ –

flip

, ‘E’ –

extension

) modification in the gRNA