The RA signaling pathway in early neural tube development

businessunknownInternet et le développement Web

12 nov. 2013 (il y a 7 années et 11 mois)

404 vue(s)


Rianne Jansen, 3180840

RIVM, Bilthoven











August, 2013


The
RA signaling

pathway

in
early
neural tube development



A potentially interesting pathway

for
toxic risk assessment

and

AOP construction











1










































Master thesis

for

the master Cancer Genomics and Developmental Biology

at the University of
Utrecht
.

Supervisors:


Prof. Dr. Aldert Piersma, Dr. Ilse Tonk

UU e
xaminer:

Prof. Dr. Blaauboer

2


Table of Contents
















Page

Introduction

3

Neural tube formation and patterning

4

Patterning along the antero
-
posterior and dorso
-
ventral axis

6

Retinoic acid

8

Retinoic acid synthesis and degradation

9

Retinoic acid and neural tube development

10

Gradients of RA and FGF control neuronal differentiation, anterior
-
posterior patterning of

the neural tube and axis elongation

10

Collinear Hox gene expression provides the neural tube with positional identity along the
anterior
-
posterior axis

13

RA and Patter
ning of the hindbrain

16

Regulation of RA activity in the hindbrain

18

RA activates Hox gene expression in the hindbrain in a sequential manner

21

Initial anterior
-
posterior identity in the hindbrain might be provided by RA induced
expression of Hox genes

22

Examples of compounds affecting neural tube development

23

Ethanol

23

Ethanol competes for RALDH2 activity during gastrulation stages

24

Ethanol exposure changes RA levels in the hippocampus and cortex during late
embr
yonic development as well as in the adult brain

26

Ethanol exposure results in increased RA synthesis and altered RA signaling in the
developing cerebellum

27

The effect of Ethanol exposure on RA signaling is stage dependent

27

Triazoles

28

Triazoles interfere with RA signaling

28

Triazoles inhibit CYP26 activity

29

A potential mechanism for RA mediated teratogenic effects of triazoles

30

Conclusions

33

Acknowledgements

36

References

36




3


Introduction



During embryonic development tight regulation of developmental processes is very important. When
the finely balanced regulation is disturbed, for example by a genetic mutation or exposure to a toxic
compound, this may result in
aberrant

development.

One ve
ry important process during development
is the formation and patterning of the neural tube, the precursor of the central nervous system. It is
therefore important to be able to assess whether a compound, for example a drug or a compound
present in the envi
ronment, has the potential to disturb the fine balance and thus may

affect

neural
tube development.


The classic way to test for possible teratogenic effects of compounds is by exposing model
organisms (animals) to a high dose of the compound to see if dev
elopment is affected (Krewski et
al., 2007). However this requires a high amount of test animals, which is ethically undesirable, it is
time consuming and expensive, and it gives only limited insight in the stage and tissue dependent
mechanisms underlying
the toxic effects. Therefore there is a demand for alternative approaches.

In modern

toxicity
research
, novel approaches are being developed for

assessment of potential risks
.
One of these approaches is based on the idea of studying
the effect of compounds on specific key
pathways, of which its distortion may be indicative of an adverse effect on (for example)
embryonic
development
.

Determining
whether a compound has an effect

on these pathways may therefore
provide information on pot
ential
teratogenic
effects
.

When

such tests are

done in cell lines, for
example in
ES cells differentiated into a certain cell lineage

(e.g. neural embryonic stem cells)
, it
provides the
opportunity to

do an initial assessment of

potential

toxic
risks

on a large scale for

many
compounds
at once

in a relatively short time (Theunissen et al., 2013)
.
If it is known which
key
pathways are altered

and
how they are altered
,

this will give an indication of which developmental
processes might be affected and w
hat the effect may be. It thus becomes possible to predict what the
potential risks are of exposure to the compound.

In other words, it
become
s

possible to
construct

a
putative

adverse outcome pathway

(AOP)

for this compound. An AOP

describes the series of events
linking
the molecular initiating event (
interaction of the compound with molecules within the cell) to
the adverse outcome
for

an individual or population (for example neural tube defects, embryonic
lethality)

(fig.
1
)

(Ankl
ey et al., 2010
)
.
Such an adverse outcome pathway may

be dependent on the
life stage during which
the individuals are exposed; a compound might have quite a different effect
on the developing embryo than on an adult individual.




Fig. 1

An

adverse outcome
pathway
describes the sequence of events that links a key molecular initiating event
to
an

adverse outcome on the organism or population level
. A toxic compound interacts with a biological target. In
turn, this may lead to an effect on cellular level and subsequently on the organ level, which may produce an adverse
outcome on the organism and population
level. After: Ankley et al., 2010
.


4


In the

case that a key pathway is used as a molecular read
-
out for a toxic effect, for example the RA
pathway as a read
-
out for potential toxic effects on neural tube development, the AOP can be
partially constructed, linking the toxic compound to the cellular r
esponses (alterations of this key
pathway and its targets) which
could then be

link
ed

to
effects on the
organ
level
(altered neural tube
development) and
to
organismal responses (lethality, impaired development, etc.).


A challenge is identifying
such key

pathways. These pathways have to be (at least) predictive of
toxic effects.
A potentially int
eresting pathway
in relation to

neural tube development is

the retinoid
signaling pathway
,

as RA signaling
plays an important role

in patterning of the embryo

and

diffe
rentiation. I
t
is involved

in

several
processes relating to

neural tube development
,

for example

anterior
-
posterior

patterning
o
f

the

neural
tissue,

patterning of the

hindbrain

and dorso
-
ventral
patterning of the spinal cord

(reviewed by Maden, 2005 and Rhinn and Dollé, 2012)
. Furthermore
aberrant RA
signaling,

either excess or

absence thereof,
is associated with neural tube defects

(Maden, 2006)
. Exposure to excess RA through maternal supplementation leads to
neural tube

defects such as exencephaly and spina bifida,

the exact effect on neural tube development is
dependent on time of exposure
. A
lso mutation of RA c
atabolizing

CYP26
enzymes
cause
s

similar

n
eural tube defects

(among other defects). Absence of RA signaling
,

e
ither through m
utation of RA
producing enzymes

or its receptors

or by a vitamin A deficiency, results in neural tube defects as
well
(exencephaly and/or spina bifida)
.
Because of the role of RA signaling in neural tube
development and its associa
tion with
neural tube defects,
studying whether a compound has an effect
on this pathway may give information

about

neurotoxicity of the compound.


In this paper, I will review the role of retinoid signaling during neural tube
development,
the
pathway(s) through
which RA exerts its effect
,

with
its

time and space specific
context, and I will
propose a set of genes
associated with
this pathway that may be used to characterize (potential)
teratogenic effects of compounds on neural tube development.

Additionally
,
I w
ill give
two

examples of compounds that are known to interfere with RA signaling
and
cause

neural tube defects
and I will propose an AOP for each of these compounds.



N
eural tube formation and patterning


The formation and patterning of the neural tube involves many different
steps and
processes. Here I
will give a short overview of these steps, as described by Wolpert et al

(2007)

and
Dias and
Partington (2004)
.

The very first step towards development of t
he nervous system is the
induction

of

neural
ectoderm from the dorsal ectoderm in response to signals from the node and notochord
(“neural induction”)

(Wolpert

et al., 2007;

Dias and Partington, 2004).

The dorsal
ectoderm

thickens
to form the neural plate
and thereby

becomes morphologically different from the surrounding
epithelium
.

In humans, this thickened neural ectoderm
becomes visible
around postovulatory day 16
(Dias and Partington, 2004).

The neural pl
ate becomes longer and narrower, by a process cal
led
convergent extension (Wallingford et al., 2013)
.

At the same time t
he lateral edges of the neural plate
start to fold upwards (dorsally
)

and e
v
entually the edges meet
at the dorsal midline
and fuse to form
the
hollow
neural tube

(fig. 2)
(Wolpert

et al., 2007; Dias and Partington, 2004
)
.

The neural
ectoderm then separates from the rest of the ectoderm, the future epidermis, whic
h now overlies the
5


neural tube.
The n
eural crest cells dissociate
from the edges of the neural plate
and migrate

to their

destination.

In humans formation of the neural folds occurs between postovulatory day 19 and 21
(Dias and Partington, 2004),
in mice around E8 (Wolpert

et al., 2007
).

C
losure of the

human
neural
tube occurs between the third and fourth week of gestation, in mice
between E8.5 and E10
(Wallingford et al., 2013;
Ybot
-
Gonzalez
et al., 2002).




Fig. 2

Neural tube closure in

the

vertebrate embryo.


A: In the ectoderm of the embryo t
he dorsal ectoderm thickens to form the neural plate. The lateral edges of the neural
plate
start to

fold upwards

and then towards the midline. When

the edges meet
,

they fuse to form a hollow tube.

The
neural crest cells dissociate and migrate to their des
tination.

B:
Closure does not occur along the whole length of the neural plate at once, but
starts around the caudal
hindbrain/rostral cervical
levels and

from there
the closing movement extends

further in anterior and posterior direction.
At the same time the neural tube elongates at the posterior end as the node regresses. The anterior and posterior
neuropore
s

have not yet closed in the
fourth picture of the embryo
.
Figure from:

Wolpert et al.,

2007 (fig. 7.34);
Wallingford et al., 2013
.


F
olding of the neural tube does not occur along the whole length
of the

neural plate at once, but
it
occurs in waves

(Dias and Partington, 2004)
.

In human

embryo
s t
he first wave of closing starts

in the
region of the caudal hindbrain or rostral spinal cord,

and then continues to extend
i
n anterior and
posterior direction

(fig. 3)
.
The ends of the neural tube
still remain open and form
the anterior and
posterior neuropore.

Later, the tube closes at

the anterior and posterior end

as well

(Dias and
Partington, 2004;

Wolpert

et al., 2007)
. Failure to properly close the neural tube can lead to various
defects such as

exencephaly (failure to close cranially)

anencephaly

(failure to close cranially)

and

m
eningomyelocele (failure to close caudally),

better known as spina bifida

(
Dias and Partington,
2004
; Wallingford

et al., 2013;
Maden, 2006)
.

Failure to close along the whole A
-
P axis is referred
6


to as craniorachischisis
, one cause of which can be disruption of the lengthening and narrowing of
the neural plate by convergent extension
(
Wallingford et al., 2013
)
.

All conditions
are
lethal, except
for meningomyelocele
(
Wallingford et al., 2013
)
.

That

process of neural tube c
losure is

a

sensitive

process
,
is
demonstrated by the fact that it is the second most common
birth

defect

(
Wallingford et
al., 2013
)
.





Fig.
3

Neural tube closure occurs in waves.

The first wave
starts in the

region of the

caudal hindbrain or rostral
spinal cord and then extends in
an
anterior and posterior direction. Failure to close at the anterior end (wave 2) results in
anencephaly, while failure to close at the posterior end results in meningo
myelo
cele (spina bifida). The last portion of the
spinal cord is formed by a different process called secondary neurulation. This part develops as a solid rod, which then
develops a cavity.
Figure

from:
Gilbert, 2000


The caud
al
-
most part of the neural tube

res
ults from a different process, called secondary
neurulation
.

This part is produced by the caudal cell mass

(also known as posterior growth zone in
the tail bud of the embryo)
, an area containing pluripotent cells
.
The caudal cell mass

produces an
initially solid rod, which later develops a cavity
(Dias and Partington, 2004).

This part fuses with the
rest

of the neural tube
that
derived from the neural plate. In humans the
portion

produced by
secondary neurulation is only small

(Dias and

Partington, 2004)
.


Patterning along the antero
-
posterior

and dorso
-
ventral axis

At many stages of neural tube development signals from other tissues are important, such as signals
from the node, notochord, overlying ectoderm, and somites (e.g. RA).


The very first step towards development of the nervous system is the induction of neural ectoderm
from the dorsal ectoderm in response to signals from the node and notochord.
Initially BMP

protein
s
are expressed throughout the
ectoderm;

inhibition of BMPs
in the dorsal ectoderm allows this tissue
to become neural ectoderm.
BMP antagonists such as
n
oggin,
f
ollistatin and
c
hordin are important

for the induction of neural tissue
, as well as FGF

(Wolpert

et al., 2007;

Dias and Partington, 2004
)
.

Signals from
the mesoderm help pattern the neural tissue along the anterior
-
posterior (A
-
P) axis,
a
fter
its induction.

It is thought that all neural tissue is
first
specified as anterior neur
ectoderm after
neural induction. P
osteriorizing signals will
then
change
the
i
dentity of the neural tissue (
Wolpe
rt

et
al., 2007;

Dias and Partington, 2004
).
WNTs
,
RA

and FGFs

are important signals
in this respect

as
they

impose

posterior identity in a dose dependent manner
,

inducing the most posterior identity in
the caudal tissue
(
Wolpert et al., 2007; Dias and Partington, 2004).
This

A
-
P patterning is reflected
by region specific expression patterns of the Hox genes an
d other homeobox genes

(Dias and
7


Partington, 2004)
, which

is

important for proper further region specific developm
ent
of the neural
tube
.

Along the A
-
P axis the neural tube
develop
s

into several distinct regions
: the forebrain
,
midbrain, hindbrain and spinal cord, which will develop

and regionalize

further
. The hindbrain for
example will se
gment, forming eight
rhombomeres

(fig
. 4
)
.
Each rhombomere will give rise to
specific cranial ganglia and/or motorneurons

(Kiecker and Lumsden, 2005
)
.




Fig.
4

The hindbrain is segmented into eight rhombomeres.

Each individual segment produces neurons innervating
distinct

areas of the embryo

(Kiecker and Lumsden, 2005).

On
the left the positions of the sensory cranial ganglia is depicted. On the right three sets of motor neurons
are depicted that
innervate the branchial arches, which will innervate the h
ead and face. A

fou
rth set contributes to the vagus nerve.
ov

=
otic vesicle, green arrow indicate migration of neural crest cells into the branchial arches. Figure from: Kiecker and
Lumsden, 2005.


The neural tube

is also patterned along the dorso
-
ventral axis

by two opposi
ng signals (fig.
5
)
.

S
onic

hedgehog

(SHH)

produced by the notochord, and later
also by the floor plate, specifies

the neural
ectoderm as ventral

(Wolpert et al., 2007; Dias and Partington, 2004)
.

SHH

also induces the
formation of the floor plate itself
(Dias and Partington, 2004
)
.

BMPs produced by the non
-
neural
dorsal (epidermal) ectoderm, which overlies the neural tube after fusion of the neural folds, specifies
the dorsal side

(Wolpert et al., 2007; Dias and Partington, 2004)
. Later, the
roof plate

produces
BMPs as well.
Different types of neurons will differentiate in different regions along this D
-
V axis
(Wolpert et al., 2007; Dias and Partington, 2004).



8



Fig.
5

Opposing signals from the notochord and surface ectoderm pattern the
neural tube

along the dorsal
-
ventral axis.
The notochord secretes
SHH

and the surface ectoderm BMPs.
Later the floor plate starts to express
SHH

as
well, and the roof plate expresses BMP4 and other TGFβ family members.
SHH

diffuses from the ventral midline and
patter
ns the ventral spinal cord, BMPs diffuses from the dorsal midline.
SHH

and BMPs antagonize
each other’s

effects.

Figure obtained from
:

Gilbert (2010
)



Retinoic acid


RA is a small, vitamin A derived molecule.

RA has many roles in many processes during embryonic
development.

The response to RA is dependent on the context
,
both in time and space: the effect can
differ between types of tissues and between diff
erent stages during development
.
In general
,

RA
signaling is involved in patterning of the embryo, and with respect to anterior posterior patterning it
is a posteriorizing factor. Furthermore it induces differentiation (as opposed to proliferation). During
embryonic development RA is involved in man
y processes, among which are
:

A
-
P patterning of the
embryo, protecting the somites from
left
-
right

specific

signaling, thus
m
aintaining bilateral
symmetry
,

and

limb development, but it is also involved in several stages of neural tube development

(reviewed

by Rhinn and Dollé, 2012)
. It is therefore a very important signaling molecule.

RA is synthesized from
gastrulation stages onward in the primitive streak and mesodermal cells and
later in presomitic and somitic mesoderm (Rhinn and Dollé, 2012). After syn
thesis it

can then diffuse
a
cross tissues, forming a gradient
.
The main source of retinoids for the embryo is maternal retinol,
transferred across the placenta, however RA
seems

able to cross the placenta as well as RA
supplementation of the mother does affect the embryo (Rhinn and Dollé, 2012).
In the target cells,
RA diffuses across the cell membrane and then binds to a nuclear receptor, transforming the receptor
from a transcr
iptional repressor into an activator. It
thereby

activates the expression of target genes.
RA can directly regulate its own activity through feedback loops. For example, it can regulate
expression of its own receptors and

of

Cyp26a1
, which encodes an RA
ca
tabolizing enzyme
(reviewed by Rhinn et al., 2012).

RA seems to diffuse into the neural tube with a higher preference
than into other tissues
(Maden, 2006)
.

There are several RA receptors
RARα
,

RARβ

and RAR
γ

which can form heterodimers with retinoid
X rece
ptors RXRα, RXRβ and RXRγ (reviewed by Rhinn and Dollé, 2012). RARα, RXRα and
RXRβ are broadly expressed in many tissues, whereas the other receptors have more tissue specific
expression patterns.

The receptors

bind to RA recognition elements (RAREs) in th
e promoters of
target genes. If no RA is bound, these receptors repress gene expression, but after binding they turn
into an activator

(Rhinn and Dollé, 2012)
. The receptors form dimers, and binding of RA to one of
9


the two

receptors

is sufficient to activa
te them.
RA might also bind to other nuclear receptors such as
PPARβ or PPARγ (Rhinn and Dollé, 2012).
Besides their role in regulation of expression,
RA
receptors
might
also have cytoplasmic roles (
Kumar et al., 2010).


Retinoic acid synthesis

and degradation

The synthesis and degradation pathways of RA
have been

well
described
. Below is a short overview,
based on the review
by

Rhinn and Dollé (2012), unless otherwise indicated.

The source of RA is

vitamin A

(retinol), which must
be obtained
through diet
. Several enzymes are
involved in the
localized
synth
esis of RA from retinol (
fig. 6
A)
.

The first step is the conversion of
retinol into retinaldehyde.

This reaction is catalyzed by alcohol dehydrogenases

(ADHs) or retinol
dehydrogenases (RDHs)
, the main enzyme catalyzing this reaction
during development being
RDH10

(Rhinn and Dollé, 2012)
. In some tissues STRA6 can facilitate retinol uptake.

In a second step retinaldehyde is oxidized (oxidation) to retinoic acid by retinaldehyde
dehydrogenases

(RALDH1, RALDH2, RALDH3). This is mainly done by RALDH2, as RALDH2 is
the earliest and most widely expressed molecule
.
RALDH2

is
first
expressed in the primitive streak,
node

and mesodermal cells. Later, it is expressed
in the
presomitic and somitic
mesod
erm and
anterior forebrain (
Rhinn and Dollé, 2012)
. In mice, RALDH1 and 3 are expressed
only
after day
E8.5 i
n the eyes and olfactory

system.
After synthesis, r
etinoic acid can diffuse to other tissues,
forming a gradient.
Intracellularly RA can be bound t
o cellular retinoic acid
-
binding proteins
CRABP1 and CRABP2.




Fig.
6

RA synthesis

and degradation.

A: Retinol dehydrogenases and
aldehyde

dehydrogenases, most importantly RDH10, catalyze the reaction of retinol into
retinaldehyde. Retinaldehyde is further oxidized to retinoic acid by retinaldehyde dehydrogenases (RALDH), mainly
by

RALDH2. B: RA is degraded by the action of CYP26 enzymes.

F
igure is adapted from: Rhinn and Dollé, 2012.


RA can be degraded by the CYP26 enzymes (CYP26A1, CYP26B1

and

CYP26C1)

(Rhinn and
Dollé, 2012)
. They convert RA into 4
-
hydroxy
-
RA and 4
-
oxo
-
RA

(fig.
6
B)
.
Tissue specific
expression of CYP26 enzymes protects
these tissues from the influence of RA
.
Cyp
26a
1

is for
exampl
e expressed in the tail bu
d which contains the stem cells. Keeping this region free

of RA

prevents induction of differentiation by RA and

ensures
that this stem cell region is

maintained

(
Wilson
et al., 2009
)
.

R
A

is able to induce

expression
of its own catabolizing enzymes.


10


Retinoic acid and

neural tube development



Retinoic acid has multiple functions during neural tube development

(reviewed by Rhinn and Dollé,
2012)
.
For instance,

it plays an important role in anterior
-
posterior patterning of the neural plate and
tube, specifically of the spinal cord and caudal hindbrain, and
in
neural differentiation.
This

anterior
-
posterior patterning

of the embryo

and the onset of differentiation are

tightly coupled to the process
of axis elongation

(fig. 7)
. Furthermore, RA is needed for dorsal
-
ventral patterning of the neural
tube.
Studies

suggest a role for RA in patterning of the forebrain

as well
.
At

later sta
ges of brain
development RA is expressed in the developing hippocampus, cortex and cerebellum.

A function of
RA in the hippocampus seems to persist into adulthood.
In the adult brain, it may also have a
function in the forebrain
/cortex
.



Gradients of RA a
nd FGF c
ontrol neuronal differentiation,
anterior
-
posterior
patterning

of the neural tube and axis elongation


Patterning of the embryo is coupled to the morphogenetic movements that elongate the embryo

(fig.
7)
.
As

the node r
egresses, tissue is left
behind

and thereby the embryonic axis gets elongated at its
caudal end
. A
s the node moves away

(posteriorly)

the cells in this tissue differentiate. The interplay
between factors produced in the caudal
-
most region in and around the node, such as
FGF8

and
W
NT

proteins, on the one hand and RA produced by the somites on the other hand are important for the
regulation of
the onset of

differentiation and
for
the patterning along the A
-
P axis of both the
mesoderm
(e.g. somites)
and neural ectoderm

(Wilson et al.,

2009)
.

FGF8
maintains the proliferative
capacity o
f the cells in the c
audal region (which allows elongation of the embryo) and
suppresses
differentiation, while RA promotes
differentiation

(Rhinn and Dollé, 2012)
.
F
gf8
mRNA
is produced
in and around the node. As
F
gf
8

is gradually degraded in the cells leaving the node region (anterior
to the node),
an
Fgf8
mRNA

gradient arises with a high concentration of
FGF8

in the node that
decreases in anterior direction
, resulting in
a
gradient of FGF8 protein (Wolpert et al., 2007)
.
RA is
formed in the somites
and anterior presomitic mesoderm
and can diffuse from there to
neighboring

tissues, such as the neural ectoderm.
As the node
moves
is further
posteriorly
, the tissue
that is left
behind
is subjected to gradually lower
FGF8

levels and higher RA levels. This drop in
FGF8

and the
presence of RA are necessary to promote differentiation of the cells in the mesoderm and neural
ectoderm.

RA seems to favor neural differentiation

above mesodermal differentiation
, as excess RA
leads to formation of neural tissue at th
e expense of paraxial mesoderm

(Wilson et al., 2009)
. It also
leads
to axis truncation
, probably

due its differentiating effec
t on the stem cells in the node
.
Furtherm
ore RA
induces differentiation

into the neural lineage in
mouse embryonic stem

cells

(Rhinn and Dollé, 2012
)
.
Coupled to the regulation of neural differentiation, is also regulation of
dorso
-
ventral patterning genes.
FGF8

suppresses
Shh

expression in the floor plate, while RA
promotes its expression, thereby controlling onset of ventral patterning

(
Diez del Corral et al., 2003;
Wilson et al., 2009
)
.

At the same time FGF
, W
NT
s

and RA are also important for patterning of the
tissue along t
he A
-
P axis
.

This patterning

is reflected by specific expression patterns of Hox genes in
11


neural and mesodermal tissue

(Deschamps and Van Nes, 2005)
.

RA is especially important for
patterning of the hindbrain and anterior spinal cord.






Fig.

7

Gradients of RA and FGF control axis elongation,
(
neuronal
)

differentiation, anterior
-
posterior
patterning and dorso
-
ventral patterning of the neural tube
.

A:

Fgf8

is produced in the node. As cells leave the node area during axis elongation, the
Fgf8

level
s

in these cells
decrease. As a result, cells further away from the node will
experience

lower FGF8 levels
.
FGF8 thus forms a gradient

in
the posterior embryo
. RA produced by the somites diffuses in anterior and posterior direction, forming a gradient as w
ell.

B:
RA and FGF8 signaling repress each other. Together RA and FGF8 regulate
the process of axis elongation
, the onset
of neural differentiation, anterior
-
posterior (A
-
P) patterning and dorso
-
ventral (D
-
V) patterning of the embryo.
Adapted
from:
D
escham
ps and Van Nes, 2005.


Fgf8

and RA

negatively
regulate each other’s activity

(fig. 8)
. Thereby the proliferative area in and
around the node is kept free of RA, while the presence of RA and the drop of
Fgf8

levels allow for
differentiation anterior to the node.

In mice the expression of
F
gf
8

is induced by W
NT
3a

(reviewed
by Wilson et al., 2009)
.
FGF8

in turn
suppresses
RA

signaling
, to prevent RA from inducing
differentiation; in chick
FGF8

has been shown to suppress
R
aldh2

expression in the presomitic
mesoderm

and also the expression of
R
arb

in the neural ectoderm

(although this could be an indirect
effect of lower RA signaling:
R
arb

has a RARE and can thus be regulated by RA)
, in mice it
s
timulat
es expression of
Cyp
26a
1

in the caudal region, which effectively clears this region of RA

(
reviewed by
Wilson et al., 2009)
.

In turn,
RA produced

by the somites inhibits
F
gf
8

and
Wnt3a

expression (Wilson et al., 2009)
.



12




Fig. 8

FGF and RA
regulate each other’s activity.

A: molecular interactions observed in chick. B: molecular interactions observed in the mouse.

FGF
8

inhi
bits RA synthesis and signaling.

RA in turn
represses
Fgf8
, but st
imulates
CYP26a

expression in the caudal
embryo.

Wnt8c

is stimulated by
Fgf8

signaling.
WNT8C mediates the transition from FGF8 to RA signaling.
W
hen
Fgf8

levels are sufficiently low, WNT8C can induce
Raldh2

expression. Eventually
Wnt8c

expression is lost due to loss of
stimulation by
Fgf8

and repression by RA.
Fgf8
inhibits
Shh
expression in the floor plate, while RA promotes it. Figure
from:
Wilson et al., 2009



The transition from FGF to RA signaling is probably mediated by
W
NT8 proteins
(Olivera
-
Martinez
and Storey, 2007)
.

In chick,
FGF
signaling
stimulates
Wnt
8c

expression

(ortholog of
Wnt8a

in the
mouse)
.

WNT
8c in turn is able to
promote

expression of
Raldh2

in vitro
.
In vivo
,

lifting the
inhibition
by FGF
of
Raldh2

by itself does not elevate RALDH2 levels, nor does stimulation of
canonical
Wnt

signaling
, but whe
n FGF is blocked and Wnt signaling is increased

Raldh2

expression
is elevated. Similarly when
Wnt
signaling is inhibited, the
onset of
Raldh2

expression is
inhibited
.
This shows that
both derepression of
Raldh2

by FGF and stimulation of
Raldh2

by
Wnt

are needed
to induce
Raldh2

expression
.
As W
NT
8c

is the only caudal Wnt signal promoted by FGF and
only
W
NT
8c is expressed near the domain of
Raldh2

expression o
nset
,

it is probable that it is indeed

W
NT
8c

that mediates these Wnt specific effects.
Thus
W
NT
8c
stimulates
Raldh2

expression
, but
only if FGF levels have sufficiently decreased (Olivera
-
Martinez and Storey, 2007)
.

WNT8c
expression declines less rapidly
than FGF8, so there is a region where

Wnt signals are present
without FGF8. Here

Raldh2

expression
can be

stimulated
.

As FGF8 is needed to maintain
WNT8c
expression, e
ventually
WNT8c
expression is lost as well w
hen FGF8 signaling is attenuated
.

Both
FGF an
d
WNT8c
are able to repress neural differentiation through different mechanisms
, as
demonstrated by their inhibitory effects on Neurogenin1 and NeuroM expression
, however

Wnt
signaling is less efficient.
In the developing embryo t
he decrease of FGF and Wnt

signaling and the
presence of RA allow for
differentiation.

13


Collinear

Hox
gene expression provides the neural tube with
positional identity

along the anterior
-
posterior axis


As discussed in
the previous section
, RA is needed for neural differentiation
throughout the
developing spinal cord and this process of differentiation is tightly coupled to the process of axis
elongation, by which the embryonic tissues

including

the spinal cord
-

are extended posteriorly. Wnt

signals, FGFs and RA together impose an
terior
-
posterior patterning of the neural plate. FGF and RA
modulate the expression patterns of the Hox genes in the neural ectoderm of the future spinal cord
and posterior hindbrain (also in the mesoderm, although the anterior expression boundaries
are
di
fferent between the mesoderm and neural ectoderm) (Young and Deschamps, 2009). The Hox
genes are activated in sequential order: the first Hox genes within a Hox cluster (paralogous group 1
at the 3’end) are expressed first and most anteriorly, while the mo
re 5’ genes are activated later in
progressively more posterior tissues (
fi
g. 9
) as these are formed cells leaving the regressing node.
Thus, Hox genes are expressed in a sequential order (1
-
13) along the A
-
P axis and in time
, which is

referred to as spatio
-
temporal
collinearity
.

Anteriorly their expressio
n domains have a sharp
boundary, but
in the posterior direction their expression gradually decrease
s. The expression domains
of the neighboring Hox

genes

in a cluster

do overlap and
th
e combination of Hox genes expressed at
a certain level along the A
-
P axis determines the axial identity of the tissue

(fig.
9
)
. The tissue will
then further develop into structures corresponding

to

the acquired identity.

Signals from the node,
mainly of F
GFs, largely determine this specific Hox gene expression. Nodes

from older embryos,
thus from a more caudal position, induce Hox gene expression typical of increasingly posterior
tissues (Liu et al. ,2001).

These specific Hox gene expression patterns provide the neuronal cells with a positional identity
along the A
-
P axis. A correct Hox gene expression pattern is essential for neural development and
positional identity of the neurons, as demonstrated by defec
ts exhibited by Hox mutant mice
(reviewed by Young and Deschamps, 2009) and the effects of modulating
Hoxc6

and
Hox
c9

expression

on motor neuron identity (Dasen et al., 2003). For instance, correct expression is
important for projection of both the afferen
ts of sensory neurons and of the axons of motor neurons
(Liu et al., 2001; Young and Deschamps, 2009).



14



Fig.
9

Collinear

Hox gene expression in the neural tube and mesoderm

provides the tissue with identity along
the anterior
-
posterior axis
.


The genes

within a Hox cluster are expressed in a spatially and temporally collinear manner: the 3’ genes are expressed
first and in the most anterior structures, 5’genes are expressed progressively later in progressively more posterior tissue
.

A: Hox expression in the neural tube and mesoderm the mouse embryo.
The anterior expression boundaries of the Hox
genes differ between the neural tube and mesoderm. The expression of the Hox genes extends more anteriorly in the
neural ectoderm, and this is

dependent on RA signaling. The combination of Hox genes expressed at a certain level along
the A
-
P axis determines the axial identity of the tissue and this results further development of this tissue according to its
acquired identity.

B:
Collinear

Hox expression of
Hox
c5
-
Hoxc10

in the spinal cord of HH stage 24 embryos. Fig
.

9
A downloaded from:
http://www.pbs.org/wgbh/nova/genes/fate
-
04.html; fig
.

9
B: Liu et al., 2001.


It is thought that FGFs, notably FGF8, produced by the node induce posterior Ho
x gene expression
in the spinal cord
in a concentration dependent manner

(fig. 10)

(Liu et al
.
, 2001; Bel
-
Vialar et al.,
2002; Martinez and Storey, 2007). This is likely in part mediated through the family of CDX proteins
(Bel
-
Vialar et al., 2002). When
the cells leave the node region, they have already acquired a caudal
Hox code under influence of FGF (Liu et al., 2001). Experiments in chick suggest that RA signaling
refines this expression pattern by inducing expression of the more anterior Hox genes in

the rostral
spinal cord (Liu et al
.
, 2001; Bel
-
Vialar et al., 2002; Martinez and Storey, 2007). However, in E7.75
Raldh2
-
/
-

mouse embryo
s early
Hoxb1

expression
is unaffected (Niederreither et al. 2000)
;

only later

its expression is disturbed, suggesting
that the early expres
sion of this 3’ Hox gene
is not dependent
on RA. Similarly, Hox4 genes are expressed in the neural tissue of the
Raldh2
-
/
-

embryo,
although
the anterior expression boundary differs from
wildtype

in
E8.25
embryos

(Niederreither et al. 2000).

Thus only at

a later stage, RA
functions to extend

the anterior boundary of the expression domains of
the
anterior and
central Hox genes (
Hoxb1
-
Hoxb8
)
in the neural ectoderm
to their definitive, more
anterior locat
ion in the p
osterior hindbrain
rather than that it is responsible for the onset of these Hox
genes

(fig. 10)

(Oosterveen et al., 2003; reviewed by Young and Deschamps, 2009).

The Hox clusters contain RAREs, so they can be directly regulated by RA (Oosterveen et al.,
2003).
At first, the induced caudal Hox genes are not sensitive to RA yet, but later they do become sensitive
in a sequential order: the first to be expressed (3’) Hox genes become sensitive to RA first

(before
E8.5)
, then the later induced, more 5’ Hox
genes become sensitive

(e.g.
Hoxb8

at E10.5) (Gould et
al., 1998; Oosterveen et al.; reviewed by Deschamps and Van Nes, 2005).

15




Fig. 10

FGF8 and RA regulate Hox gene expression in the neural tube

FGF8 establishes collinear Hox gene expression patterns in

the neural tube. Supposedly, increasing FGF8 levels in the
node induce increasingly posterior Hox gene expression during axis elongation. At later stages GDF11 enhances the
posteriorizing effects of FGF8. Later
,

RA modulates Hox gene expression patterns. In response to RA the expression
domain of the Hox genes expands anteriorly, shifting the anterior limit of expression anteriorly.

The 3’ Hox genes
become sensitive to RA first, later the more 5’ Hox genes become

responsive. The specific Hox gene expression pattern
provides the cells of the neural tube with positional identity along the rostro
-
caudal axis. This is necessary for correct
projection of efferents and afferents of neurons in the spinal cord, and for co
rrect rhombomere specification in the
hindbrain.


One way the node could be able to induce
the expression of
different Hox
genes

during axis
elongation, is by increasing FGF levels while it moves posteriorly.

Indeed, i
ncreasing levels of FGF
are able to s
timulate increasingly poster
ior Hox expression (Liu et al., 2001; Bel
-
Vialar et al., 2002)
and i
n situ hybridization experiments do suggest FGF levels in the node ris
e over time (Liu et al.,

2001). S
ince in situ hybridization experiments are not quantitati
ve
the latter

should be confirmed.
Another possibility is that the Hox gene expression is dependent on the time spent in the node, and
thus the duration of exposure to FGF signaling
1
*. Furthermore, from the 11 somite stage
Gdf11

on
becomes expressed in the
tail bud

of the chick embryo and
in vitro

GDF11 is able to enhance the



1

Both alternatives are possible: older nodes are able to induce more posterior hox gene expression in neural tissue (Liu et
al., 2001). This suggests that it is the strength of the signals from the node that give the neural tissue its identity, as t
hese
ne
ural cells themselves did not spend a longer time in the node and thus weren’t exposed to FGF for a longer time. At
first glance this might point to a model in which rising FGF levels in the node cause increasingly more posterior (5’) Hox
gene expression.
On the other hand, it could be that the neural progenitor cells in the node do acquire an older, more
caudal profile, with concomitant Hox gene expression, and signal to the neighboring grafted tissue imposing the older
identity on the tissue.


16


posteriorizing effects of FGF, but it is not able to affect Hox gene expression

on its own
. Thus, in the
caudal spinal cord the posteriorizing action of FGFs

is aided by GDF11 to cause even more posterior
Hox gene expression.




RA and
Patterning of the hindbrain


After closure of the neural tube in the hindbrain region
, the hindbrain segments
to
form
seven

(zebrafish)

to
eight

rhombomeres
(mouse, chick, human)
(
Rhinn and Dollé, 2012
)
.

Retinoic acid

produced by the somitic
mesoderm
plays

a
n important

role in the patterning of the
posterior
hindbrain
, where it is needed for correct, rhombomere specific, expression of Hox genes and other
transcription factors

(Niederreither et al., 2000; reviewed by Deschamps and Van Nes, 2005)
.
M
ore
specifically
, it
is important for

posteriori
z
ation of the caudal hindbrain rhombomeres
, as
demonstrated by the effects of either absence of RA

signaling

or ex
cess RA

(fig.
11
) (as reviewed by
Rhinn and Dollé, 2012)
.

In a
bsence of retinoic acid

signaling
, for example by vitamin A deficiency
or knocking out
Raldh2
, the tissue of the posterior hindbrain adopts a more anterior fate,

the caudal
hindbrain is
decreased

in size

and
segmentation

is impaired

(
for example
Raldh2
-
/
-

phenotype:
fig
.

12
)
(
Niederreither et al., 2000;
Rhinn

and Dollé, 2012
)
.

The vitamin A deficiency

resulted in

a loss
of cranial nerves IX, X, XI, and XII and associated sensory ganglia
,
and misguidance of nerves

at
E12.5 (White et al., 1998; White et al., 2000).
Similarly, interference with the RA receptors (by
compound knock
-
out thereof or by using a pan
-
RAR antagonist), leads to anteriori
z
ation of the
hindbrain as well

(Rhinn and Dollé.
, 2012)
.

Excess RA at the late gastrula or early neurula stages
leads to an anterior expansion of the hindbrain region at the expense of the anterior brain structures
,
reflecting a posteriorization of these tissues. If excess RA is supplied at later stages
,

the
rhombomeres within the hindbrain undergo a posterior transformation (adopt a more posterior
identity)
.

Experiments in which RA is blocked at progressively later stages suggest that RA patterns
the rhombomeres in a rostral to caudal order; for example

blocking RA at an early stage results in
misspecification of r4
-
r8, while blocking it at later stages only results in defects at a progressively
more posterior level in the hindbr
ain (reviewed by Gavalas, 2002)
. Similarly
,

it seems that
increasingly higher RA levels are required for patterning

of
more posterior rhombomeres (
reviewed
by Deschamps and Van Nes, 2005).

RA does not seem required for
early
patterning of the forebrain and formation of the mid
-
hindbrain
boundary (N
iederreither et al., 2000), as this area is correctly specified in
Raldh2
-
/
-

embryos,
although exposure to RA can impair development of these structures as excess RA leads to a
reduction of mid
-

and forebrain size due to expansion of the hindbrain area (Rh
inn and Dollé, 2012).
In other words, part of the midbrain area has adopted a more posterior fate (hindbrain). Similarly,
loss of both
Cyp26b1

and
Cyp26c1

expression in the anterior hindbrain also leads a rostral shift of
the boundary between the mid
-

and hindbrain (Uehara et al., 2007).





17



Fig.
11

Altered RA signaling in the hindbrain results in
hindbrain defects.
RA is required for posteriorization of
the
posterior hindbrain rhombomeres. Decreased RA signaling due to a vitamin deficient diet (VAD), knock
-
out of
Raldh2
,
or knock
-
out of the RA receptor leads to anteriorization of the hindbrain and a decrease of posterior hindbrain size. In
contrast, ectopic RA signaling, due to knock
-
out of
Cyp26

genes leads to a posteriorization of the hindbrain
rhombomeres. Figure from
:
Rhinn and Dollé 2012.





Fig
.

12

Due to absence of retinoic acid signaling in
Raldh2
-
/
-

mutant mice
the tissue

of the posterior hindbrain
adopts a more anterior fate, the caudal hindbrain is decreased

in size and segmentation is impaired.

RA seems required for posteriorization of the caudal hindbrain. In absence of RA, due to a knock
-
out of
Raldh2
,
specification of the anterior hindbrain rhombomeres (r1,r2) occurs relatively normally, whereas the caudal hindbrain is
anteriorized; it is expr
essing genes typical of more anterior rhombomeres and the expression of the genes typical of the
r5
-
r8 is decreased or lost (Niederreither et al., 2000). The caudal hindbrain is decreased in size and segmentation is
18


impaired.

Gene expression patterns are
depicted as observed at E8.25/E.8.5, before segmentation. The hindbrain defects
are depicted as observed at E9.5.

Dark colors: high expression, lighter colors: lower expression. The pointy ends mean
that the expression boundary is not sharp.

Figure from: N
iederreither et al., 2000.


Regulation of RA activity

in the hindbrain

Using mice with the RARE
-
LacZ reporter, it was shown that RA activity is transient
ly present up to
r2/r3 boundary;
later the
activity

front
regresses

to the level of r4/r5 boundary

(Sirbu et al., 2005)
. A
similar trend was observed in zebrafish

(
using RARE
-
fluorescent protein

reporter
)
: first RA activity
extends into the hindbrain and later it retreats again from the hindbrain (reviewed by Glover et al.,
2006).
This RA pattern is es
tablished by diffusion of RA
from
the
somites into the hindbrain and RA
degradation by
dynamic
ally expressed

Cyp26

family members
in

the anterior
part of the hindbrain.
There are however

indications that there is another source of RA intrinsic in the hindb
rain itself that
may contribute to this RA gradient (Niederreither
et al., 2002; Mic et al., 2002).

Expression of CYP
26 family members protects the anterior hindbrain
against

the posteriorizing
effects of RA (Rhinn and Dollé, 2012).

T
heir expression
changes over time and th
is

d
ynamic
expression of

CYP26
enzymes
tightly
control
s

RA
activity

within

the rhombomeres in

mouse, chick
and zebrafish. A
lthough the exact expression profiles of the three different
CYP26
enzymes differ

between the species
,
the
basic principle is the same
.
The
Cyp26

genes

become expressed first in the
anterior hindbrain and later

Cyp26

members

become expressed in more posterior rhombomeres
, with
each
Cyp26

family member having its own rhombomere specific expression pattern
.
As a
result of

this dynamic expression of these RA catabolizing enzymes,

RA is excluded

from

progressively more
posterior rhombomeres
,
shifting the
anterior
boundary of RA activity

posteriorly
.

In chick
,
Cyp26
a1

is initially

expressed in the fore
-

and midbrain
region
.

The expression domain
then regresses at the anterior end and
at the posterior end it
spreads

out

into hindbrain up to the r3/r4
border

(of chick, after stage 8) (Blentic et al., 2003
)
. Eventually its expression domain has shifted
such that it is

only expressed in r3 (stages 9 and 10), after which its expression disappears.

Cyp26c1

starts
to be expressed in prerombomeres 1 and 2 just before

stage 10, when formation of the
rhombomeres commences
,

then

expands
further
posteriorly

into the hindbrain
.
By stage 11 it

is
expressed in r2, r3 and r5
and eventually
it
becomes restricted to r5 and r6

(Reijntjes et al.,

200
4
).
Cyp26b
1

becomes

expressed
in the (prospective)

r4

and
r6

from stage 7 onward (Reijntjes

et al.,
2004
).

A similar trend
is observed for
the mouse Cyp
26

genes.

Cyp26a1

is also first expressed in anterior
neural plate

and

from E7.5

on

its
expression
domain
expands

as well (Sirbu et al., 2005). Eventually
its expression domain
gets restricted to

r2

at E8.5
(
compared to

r3 in chick
) (Sirbu et
al., 2005)
.
Cyp26c1

becomes expressed in r2 and r4 a
round E8

(Sirbu et al., 2005),

a
fter induction of
Hoxb1

in

r4
,
and around or shortly after

E8.5

expression in r4 is lost

(
Tahayato et al., 2003
; Sirbu et al., 2005
)
.

Cyp26b
1

expression is first observed at E8
in
prospective
r3 and

r5
, with strongest expression in r5

(Maclean et al., 2001). By E9.5
Cyp26b1

expression has expanded,
with
its

strongest expression

throughout r5 and r6
and additional expression i
n the ventral portio
n of r2
-
r4
.

This dynamic
Cyp26

expression controls

RA
activity

and
causes
RA
to be

cleared from
progressively more posterior
hindbrain tissue, s
hifting the anterior boundary of

its
activity

p
osteriorly

(fig.
13
) (Sirbu et al., 2005)
.
I
n zebrafish the
c
yp26

genes show dynamic expression patterns that progressively exclude RA from
the hindbrain

as well

(Hernandez et al., 2007).

19


Studies do suggest some redundancy between the function of the Cyp26 genes in the hindbrain.
Hindbrain patterning of
Cyp26c1

and
Cyp2
6b1

mutant mice was normal and
Cyp26a1

mutants only
show a mild posteriorization (reviewed by Rhinn et al., 2012). A compound mutation of Cyp26a
-
/
-

Cyp26c
-
/
-

however leads to strong
posteriorization

and loss of segmentation
(
reviewed by Rhinn et
al., 2012). Similarly, in zebrafish the subtle hindbrain patterning defects of the
cyp26a1

mutant
became progressively more severe upon knock
-
down of
cyp26c1

alone or knock
-
down of
both

cyp26c1

and
cyp26b1

(Hernandez et al., 2007)
, whe
reas k
nock
-
down of
cyp26c1

and
cyp26b1

did
not lead to patterning defects of the hindbrain, aside from a slight shortening of the hindbrain.





Fig. 13

Dynamically expressed Cyp26 enzymes shift the anterior limit of RA gradient posteriorly
over

time.

Dynamic expression of Cyp26
family members

limits the extent of the RA gradient.

In mice, RA activity is first detected
up to the r2/r3 rhombomere boundary. At this time
Cyp26a1

is expressed in the anterior rhombomeres. Later,
Cyp26c1

expression becomes ex
pressed in r4 and the RA gradient regresses up to the r4/r5 boundary. From E8 on
Cyp26b1

starts
to be expressed in r3 and r5, however RA activity

(or more precisely: galactosidase activity)

is still present in r5 and
further posteriorly at E8.5. Possibly the
Cyp26b1

activity at this time is not sufficient to clear all RA from this region, or
Cyp26b1

does not have large effect at this time.
Cyp26b1

mutant

mice do not display any hindbrain def
ects
.
Also knock
-
down of

cyp26b1

in

the

cyp26a1
-
/
-

background has only slight effect

in zebrafish
,

which is

much
weaker than the

effect
of

cyp26c1

knockdown

in this background
. Adapted from: Sirbu et al., 2005


Thus, the domain of RA activity is established by diffusion of RA from the somites into the
hindbrain in anterior direction, forming a gradient, and is restricted at the anterior end by catabolism
of RA by
CYP26

enzymes. First its activity is only excluded

from the anterior hindbrain and later
also from more posterior rhombomeres, as a result of changing domains of
CYP26
enzyme activity.
However, it cannot be excluded that regulation of RA receptor expression may also play a role in
controlling RA signaling
. RAR expression in the hindbrain is dynamic and changes during
development (Hale et al., 2006; Mollard et al., 2000; Serpente et al., 2005), thus regulation of RA
receptor presence may also be
a way to regulate
RA signaling.


20


The RA gradient in the hindb
rain
specifies rhombomere identity in

the posterior hindbrain


RA synthesized in the somites diffuses into the posterior
hindbrain, forming a posterior (high) to
anterior (low) gradient in the hindbrain. RA posteriorizes the hindbrain in a concentration (and time)
dependent manner (fig. 14), presumably through sequential activation of the Hox genes (see
next
section
). Thereb
y it provides positional identity to the future rhombomeres r3
-
r8. This
posteriorization is needed for correct, rhombomere specific gene expression in the rhombomeres and
thus for correct specification of the rhombomeres. Expression of Hox genes
,
Krox20
,
K
reisler
,
vHnf1

and
Fgf3

in the hindbrain are required for development of the rhombomeres. Indirectly, correct RA
signaling is also required for rhombomeric segmentation; the rhombomere specific expression of the
Eph receptors needed for segment boundary fo
rmation is lost in
Raldh2

mutants (Niederreither et al.,
2000).

Anteriorly, CYP26 enzymes catabolize RA, thereby limiting the extent of RA activity and protecting
the anterior hindbrain from the posteriorizing effect of RA. GBX2 and FGF8, whose genes are
expressed in the anterior hindbrain, promote specification of the anterior rhombomeres, and OTX2
expressed in the future mid
-

and forebrain promotes mid
-

and forebrain development.
Gbx2

is
required for anterior hindbrain development (of r1
-
r3) (Burroughs
-
G
arcia et al., 2011). Furthermore,
GBX2 and OTX2 together position the midbrain
-
hindbrain boundary and the isthmus located here
(an organizer); the boundary arises at the border between their two expression domains (Burroughs
-
Garcia et al., 2011; Sunmonu et

al., 2011; Rhinn and Brand, 2001).
Fgf8

is expressed on the
hindbrain side of the mid
-
hindbrain boundary, while
Wnt1

is expressed on the midbrain side of the
border (Liu and Joyner, 2001). Initially their expression domains are quite broad, with
Fgf8

bein
g
expressed in the entire r1 and
Wnt1

in the midbrain, but they become restricted to two narrow bands
on either side of the mid
-
hindbrain boundary (Liu and Joyner, 2001). Eventually, the
Fgf8

expression
domain becomes restricted to the isthmus. The signals

emanating from the isthmus help pattern both
the anterior hindbrain and posterior midbrain (Sunmonu et al., 2011) and FGF8 is the key molecule
for the function of this organizer (Sunmonu et al., 2011).

After segmentation of the hindbrain, RA induces
Hoxb
5
-
Hoxb8
gene expression in r7/r8 of the
posterior hindbrain (Oosterveen et al., 2003)

(
the authors did not specify the anterior limit of the
expression domains of each gene individually, although their anterior boundaries do seem to differ).
These
two
rhom
bomeres are less well
-
defined. This induction of
Hoxb5
-
Hoxb8

might be required for
the specification of the identity these two rhombomeres
.



21




Fig. 14

Retinoic acid establishes A
-
P patterning of the posterior hindbrain in a concentration dependent
manner.

This is a schematic overview of the role of RA in patterning of the hindbrain.
Rhombomere specific gene expression
patterns are depicted as they are observed around E8.25
-
E8.5, prior to hindbrain segmentation
.

Keep in mind however
that both the extent of the RA gradient, the RA levels and the expression domains of
Cyp26

family members,
Gbx2

and
Fgf
8

are dynamic and change during patterning and development of the hindbrain
. This is also the case for
part

of the
m
arker genes mentioned here
.

Kreisler

expression, for example, is dynamic

as well
. In chick
Kreisler

is expressed in the
prospective r5 and r6 from the 5 somite stage onward, but it extends into r7 and r8 between somite stage 6 and 10
(Giudicelli et al., 20
03)
.
In zebrafish
Fgf8

is
additionally

expressed in r4
.

Inhibitory actions are depicted with long lines ending with a bar (
---
| ), while stimulatory actions are depicted with long
arrows. The small arrows indicate that FGF8 diffuses from the anterior
hindbrain into the posterior midbrain.

*: Oosterveen et al. did not specify the anterior limit of the expression domains of each Hox gene individually, although
their anterior boundaries do seem to differ.

Marker gene expression is based on the publication
s of Niederreither et al.
2000, Gavalas and Krumlauf 2000 and Kim et al., 2005.



RA activates Hox gene expression in the hindbrain

in a sequential manner

As neural tissue is laid
down

by the regressing node
,
Hox gene expression is initiated in the

newly
formed tissue
.
Only after the neural tissue is laid down by the node,
these Hox genes become
sensitive to RA
. In response to
RA
,

the expression domains of

the anterior and central Hox genes
(Hoxb1
-
Hoxb8) expand anteriorly so that the anterior limit
of their expression reaches its
definitive,
more anterior location in the posterior hindbrain (mouse) (
Gould et al., 1998;
Oosterveen et al.,
2003;

Deschamps and Van Nes, 2005;

Young and Deschamps, 2009
)
.
This

anterior expansion is

dependent

on

RAREs

locat
ed within the Hox clusters

and this anterior expansion happens
exclusively in the neural tube, not in the mesoderm

(Oosterveen et al., 2009; Deschamps and Van
22


Nes, 2005)
.
The Hox genes become sensitive in a sequential order (3’ to 5’) with the first genes to
shift
before E8.5

and the more
5’

Hoxb8

to start shifting

at E10.5

(Gould et al., 1998; Oosterveen et
al.; reviewed by Deschamps and Van Nes, 2005)
. The anterior
-
most b
oundary of the Hox genes of
paralogous g
roup 4, and other 3’ genes
, are already met before
hindbrain
segmentation (Gould et al.,
1998; Niederreither et al., 2000)
.

Additional

regulatory inputs from rhom
bomere specific
transcription factors
, such as

K
ROX
20

(E
GR
2)
, k
reisler

(MAF
b
/V
AL
)
, vH
NF
1

(HNF
1b)

and auto
-

and cross
-
regulatory

loops
between

Hox genes themselves,

modulate
s

Hox gene expression

as well

(reviewed by Deschamps and Van Nes, 2005

and Glover et al., 2006
; Wong et al., 20
11
).
This way
the

early
Hox gene
expression patterns
in the hindbrain are
established.


The

anterior
-
most boundary of the
Hox
b5
-
Hox
b8

genes is

only

reached after rhombomere
segmentation (Oosterveen et al., 2003)
.
Hox
b5

becomes sensitive

to RA
at E9,

Hox
b6

at E9.5
and
Hoxb8

at E10.5
.
By E11.5 t
heir definitive

boundaries have been reached
.
Oosterveen et al. proposed
that th
e expansion of
Hox5
-
Hox8

gene expression into the hindbrain

might contribute

t
o patterning of
r7 and r8.



Initial anterior
-
posterior identity in the
hindbrain might be provided by RA induced expression of
Hox genes

As
transcription factors

such as
Krox20

and
k
reisler are expressed in rhombomere specific
expression domains
,

the question is what provides the hindbrain cells with their initial positional
identity along this axis

to
define these rhombomere patterns. A continuous gradient has to be
translated into discrete segmental (i.e. rhombomere specific) expression patterns.

One theory

is that
this initial
anterior
-
posterior
identity is prov
ided by sequ
ential, collinear 3’ t
o 5’

Hox gene
expression

(Deschamps and Van Nes, 2005)
,
but
only for the caudal rhombomeres r3
-
r8
,

as Hox
gene expression is prevented in r1
-
r2.

This could be a plausible theory, based on the dynamic Hox
gene expr
ession patterns in
the hindbrain
.


The Hox clusters contain RA responsive elements

and can therefore be directly regulated by RA
.
The

anterior expansion of the expression domain of the first Hox genes expression precedes

the onset of
expression of
Krox20

in r3 and r5 and the restriction

of

Fgf3
2

to r5
-
r6 (Giudicelli et al., 2001;
Makki
and Capecchi, 2011; Vendrell et al., 2013
).

In chick
,

the anterior expansion also precedes
the
expression of
k
reisler (Aragon and Pujades, 2009), but in mouse it occurs eit
her just before or
around

the

onset of

expression of
k
reisler (Cordes and Barsh, 1994
; Kim et al., 2005
)
.

Furthermore,
Hox genes

seem to be involved in
the
initiation of expression of
at least

some

rhombomere specific genes or
in
the

restriction of

expression of

such
genes to specific rhombomeres
.
Hox genes

are
for example
involved in initiating
Krox20

expression

in r3 and r5. I
n r3 the Hox genes
and M
EIS
2 together activate
Krox20

and experiments suggest that
HOXA
1
might

indirectly
activate
Krox20

in r5
through activation of
vH
nf1

(
Hnf1b
)

(Wassef et al., 2008
; Makki and Capecchi, 2011)
.

E
xperiments indicate that

the expression of

genes such as

Fgf3

and k
reisler
might be

regulat
ed by
Hox genes

as well

(Wassef et al., 2008
; Makki and Capecchi, 2011
;
Pasqualetti et al., 2001
).

However, many more regulatory actions are going on at the same time, and RA is also able to target
many
other genes, thus it is possible that initial anterior
-
posterior identity is conferred by more genes
than only the Hox genes.





2

Fgf3

is first expressed throughout the rhombencephalon
a
nd later

it becomes restricted to r5
-
r6 (Vendrell et al., 2013)

23


Establishment

of

the

collinear expression

of the Hox genes

in the hindbrain seems to be based on
differential sensitivity of the Hox genes to RA.
Several studies indicated that t
he 3’ Hox genes are
most sensitive to RA (induced at low concentrations of R
A),
whereas
subsequent Hox genes

in the
cluster

are less sensitive and thus require higher concentrations of RA to become activated

(reviewed
by Deschamps and Van Nes, 2005)
. RA forms
a
concentration gradient

as it diffuses from the
somites into

hindbrain
;

the differential sensitivity to RA

then

results in a

specific Hox gene
expression pattern along A
-
P axis
.

3’

Hox genes
become expressed
most anteriorly,
while the more
5’

genes in the cluster

are expressed at progressively more posterior levels.
This Hox
gene
expression in the hindbrain provides it with its positional identity.
Further regulatory inputs from
rhombomere

specific
transcription factors

and auto
-

and cross
-
regulatory loops
between

Hox genes
themselves, then modulate the Hox gene expression patterns to establish the
ir
rhombomere specific

expression

patterns in the hindbrain (reviewed by Deschamps and Van Nes, 2005 and Glover et

al.,
2006; Wong et al., 2011
)
.
For example,
Hoxb1

is fi
rst expressed in a broad region in the hindbrain

(and spinal cord)

and only

later it
becomes restricted to a small band of ti
ssue of the future
rhombomere r4 (it remains expressed in spinal cord)

(Wong et al., 2011
; Glover et al., 2006
)
.




Examples of compounds affecting
neural tube
development



Ethanol


One compound that is very well
-
known for its effect on brain development and the adult brain, is
ethanol, commonly known as alcohol. During ethanol detoxification ethanol is first converted to
acetaldehyde, which is very toxic, and then to acetic acid. Alc
ohol abuse during pregnancy can result
in a wide range of defects in the unborn child, referred to as Fetal Alcohol Spectrum Disorder
(FASD). Exposure to EtOH can result in embryonic lethality, stillbirth, or developmental defects
(Kot
-
Leibovich and Fainso
d, 2009). Children with FAS can have, amongst other defects, craniofacial
malformations, microcephaly,
and microphthalmia
. These children experience neurological defects.

Various studies have implicated that deregulation of RA signaling by EtOH may at lea
st in part
account for the toxic effects of EtOH (Deltour et al., 1996; McCaffery et al., 2004; Yelin.
2005;
Yelin 2007;

Kot
-
Leibovich and Fainsod, 2009; Kumar et al., 2010; Kane et al., 2010).
First of all,
FAS and the developmental phenotypes caused by ethanol exposure resemble the phenotype of
embryos with reduced RA signaling as well as to those with excess RA (Yelin. 2005). Second, in
Xenopus

and zebrafish embryos the defects caused by ethan
ol can be partially rescued by
supplementation
with

retinoids (Yelin et al., 2005; Marrs et al., 2010).

In

Xenopus

embryos
treatment

with EtOH at the gastrula stages leads to impaired head development, with a reduced forebrain,
shortening along the rostro
-
caudal axis and microphthalmia (reduced eye) (Yelin et al., 2007).
Raldh2
-
/
-

mouse embryos for example show severe impaired development as well and die
midgestation, they display
amongst other defects

shortening along the A
-
P axis, incomplete closure
of th
e neural tube and reduction of frontonasal region (Niederreither et al., 1999). Also at later stages
of development there are similarities between the effects of EtOH
exposure
and

disturbed

RA

24


signaling
. In the third trimester of humans or
murine
postnatal

days 4
-
9 the cerebellum is especially
sensitive to both EtOH exposure and disturbances in RA signaling and the
ir

effects on the
cerebellum show similarities (McCaffery et
al., 2004; Kumar et al., 2010). Experiments suggest that
at least part of the toxic
effects of EtOH may be the result of altered RA signaling. The effects of
ethanol on RA signaling, however, are very much dependent on developmental stage (and tissue).
The effects of EtOH on early neural development will be discussed in detail in the next

section. After
that, the effects of EtOH on brain development during later stages are

briefly discussed as well.


Ethanol competes for RALDH2 activity during gastrulation stages

Experiments suggest that at least part of the abnormalities observed after pr
enatal alcohol exposure
are the result of abnormally low RA levels due to competition with retinaldehyde for RALDH2
activity by ethanol (Deltour et al., 1996; Kot
-
Leibovich and Fainsod, 2009). First of all, FAS and the
developmental phenotypes caused by et
hanol exposure during embryonic development resemble
those of embryos with reduced RA signaling as well as to those with excess RA (Yelin. 2005).
Ethanol treated embryos display impaired head development, with a reduced forebrain size,
shorteni
ng along the

rostro
-
caudal axis,
microphthalmia (reduced eye)
and hindbrain defects
(Yelin
et al., 2007
; Marrs et al., 2010
).
Raldh2
-
/
-

mouse embryos for example show severe impaired
development and die midgestation, they display a.o. shortening along the A
-
P axis,
incomplete
closure of the neural tube and reduction of frontonasal region (Niederreither et al., 1999).

Second,
RA reporter levels were strongly decreased in embryos
exposed to

ethanol around gastrulation stages
in both
Xenopus

and mouse (Kot
-
Leibovich and

Fainsod, 2009; Deltour et al., 1996). In
Xenopus

it
was demonstrated that the expression of RA responsive genes

Hoxb1

and
Hoxb4

was

reduced as well
(Kot
-
Leibovich and Fainsod, 2009).

Expression of
Cyp26a1
, which is involved RA degradation
expression and c
an be regulated by RA, was

also

reduced.

In support of the notion that part of the defects of ethanol exposure are due to lower RA levels, is the
fact that retinoid supplementation was able to partially rescue the effects ethanol exposure in
zebrafish and
Xenopus (Yelin et al., 2005; Marrs et al., 2010).
It

partially rescue
d

the impaired
gastrulation movements and the concomitant shortening of the axis, hindbrain development and
craniofacial defects (Yelin et al., 2005; Marrs et al., 2010).


It seems that
the effect of EtOH on RA signaling is through competition for RALDH2 activity (Kot
-
Leibovich and Fainsod, 2009).

Suppression of RALDH2 activity in combination with ethanol
treatment aggravated the reduction of RA sig
naling, the reduction of

expression of R
A responsive
genes
, as well as the phenotype. On the other hand,

overexpression of
Raldh2

in part rescue
d

the
phenotype, RA signaling and RA responsive gene expression. Furthermore, it seems that EtOH acts
mostly through competition for RALDH2,
and no

othe
r enzymes, as EtOH treatment did not
aggravate

the phenotype of
Raldh2

knock
-
down embryos
, which would be expected if EtOH also
targets other enzymes. The embryos are most sensitive to EtOH exposure at the onset of RA
signaling (Yelin et al.,2005; Kot
-
Leib
ovich and Fainsod, 2009), when RA levels are still low, and the
availability of RALDH2 is also relatively low at this point.

T
he defects are also cumulative: defects
are more severe upon longer exposure.
Exposure of embryos at the neurula stages (stage 18
and
onward
)

did not reveal any gross morphological defects in the over
-
all embryo (Yelin et al., 2005),
however patterning of the embryo was not examined in detail
.

25


In normal
Xenopus

development, RA negatively regulates expression of organizer specific genes. In
alcohol treated embryos the expression of organizer
-
specific genes
was

upregulated or expanded,
indicating that the negative control by RA is lost. Again, overexpression of
Ra
ldh2

seems to rescue
the expression levels of these genes (Kot
-
Leibovich and Fainsod, 2009). Experiments suggest that
most of the defects observed arise because of disruption of the organizer. The rostral
-
caudal
shortening
seems

to be mediated through impa
ired convergence extension
-
movements, which may
be caused by the expansion of the expression of
Otx2

in the organizer upon EtOH exposure (Yelin et
al., 2005).

Thus it seems that
at least part of the teratogenic effects ethanol are mediated by its c
ompet
ition

for
RALDH2 activity with retinaldehyde

(fig. 15)
. As a result RA levels decrease upon ethanol
exposure, which leads to loss of negative regulation of organizer specific genes and defects that
resemble the RA deficiency phenotype. In the light of the

previous chapters, we see again that

RA
seems to function to antagonize signals from the organizer; the node. It would be interesting to test if
differentiation and rostro
-
caudal patterning are affected as well, as loss of RA signaling leads to an
expansio
n of
Fgf8

expression domains, impaired neural differentiation, reduced number of neurons
in the spinal cord and anteriorization of the hindbrain with loss of rhombomere segmentation.

Also
,

both ethanol and
aberrant

RA levels can cause malformations of the
hindbrain (McCaffery et al.,
2004)
.
This
could be tested by studying the effect of EtOH on the

expression levels and expression
patterns of
Fgf8, Wnt3a,

the Hox genes, and hindbrain patterning genes
-

such as
Krox20,
kreisler,
and

vHnf1



and of neural differentiation genes
, such as
Ngn1,Ngn2, Neurod4
.
Interestingly,
Pax6

expression in the hindbrain is reduced upon EtOH exposure (Santos
-
Ledo et al., 2013), but
unaffected in the anterior structures, this may be an indication that hindbrai
n patterning is affected
(Kayam et al., 2013).

A role for RA signaling in forebrain development or in patterning of the surrounding mesoderm may
account for the reduction of the forebrain and eye in ethanol treated embryos.
Raldh2

is transiently
expressed in the anterior neural plate and optic vesicles and later
Raldh3

is expressed in the
overlying ectoderm (reviewed by Rhinn and Dollé, 2012). Interference with RA signaling, by knock
-
out of
Raldh2

or dominant negative RA receptors,

might interfere with patterning of the forebrain,
and it

resulted in decreased cell proliferation and survival

in the forebrain
. At later stages, RA may
be involved in the development of the cortex of the forebrain (Rhinn and Dollé, 2012; Kane et al.,
20
10
). The exact effects of ethanol on RA signaling, however, is dependent on stage, and tissue, thus
a later effect of ethanol on forebrain development, may not be through downregulation of RA, but
through excess RA or altering of RA signaling (see next secti
on
s
).

Lack of RA signaling also leads to
smaller somites (Rhinn et al., 2012), thus smaller somites might also be expected in EtOH exposed
embryos.


26




Fig. 15

Proposed adverse outcome pathway of embryonic EtOH exposure during gastrulation

stages

Etha
n
o
l
(EtOH)
exposure during embryonic development can lead to developmental abnormalities, such as shortening
along the anterior
-
posterior axis, hindbrain defects, impaired head development with a reduced forebrain and
microphthalmia. In the

developing

embryo E
tOH leads to expansion of the domains of organizer specific genes. The
associated disruption of the organizer then leads to impaired convergence
-
extension movements resulting in a shortening
along the anterior
-
posterior (A
-
P) axis. Ethanol competes with re
tinol for RALDH2 activity. As a result RA synthesis is
reduced, resulting in subnormal RA levels.
A potential mechanism which may explain part of the teratogenic effects
upon EtOH exposure was extrapolated, based on

the knowledge
about

the roles o
f RA in p
atterning of the embryo, the
considerable overlap between defects observed upon EtOH treatment and RA depletion
,

the phenotypes observed in
EtOH treated embryos

and the rescuing effect of
Raldh2

overexpression or retinoid supplementation
.



Ethanol
exposure changes RA levels in the hippocampus and cortex during late embryonic
development as well as in the adult brain

The hippocampus is sensitive to both EtOH and
aberrant

RA levels. Prenatal EtOH exposure leads to
impaired formation of dendritic spine
s in the hippocampus and abnormal development (
Kane et al.,
2010
). In contrast to the effects of EtOH during early development, EtOH exposure between E13 and
E19 strongly increased RA levels in the hippocampus and cortex of E19 mouse embryos, as well as
re
tinol levels (
Kane et al., 2010
).

RA and retinol levels were also increased in the adult hippocampus
and cortex upon chronic EtOH exposure. The experiments in adult rats suggest that the increase of
RA levels is likely the result of both increased uptake o
f retinol and RA from the blood stream and
increased RA synthesis

due to increased RALDH1 activity
3
, aided by a modest increase in RDH



3

Not
Raldh1
expression; RALDH1 protein levels were decreased, whereas RALDH1 activity was increased.

27


activity
.
In the hippocampus RA is involved in neurogenesis and normal dendritic spine formation
and branching (Kane et al., 2010).
Thus

EtOH

treatment of E13
-
E19 rat embryos results in increased
RA levels in the cortex and hippocampus.
This
strong increase of RA le
vels

may impair neuronal
cell division and affect the

formation of dendritic spines
and may thereby contribute to the defects
associated with fetal alcohol syndrome.


Ethanol exposure results in increased RA synthesis and altered RA signaling in the devel
oping
cerebellum

In rats ethanol exposure inhibits differentiation and causes increased cell death in the developing
cerebellum.
In humans cerebellar development takes place in the third trimester of pregnancy, in rats
this occurs at postnatal days 4
-
9.
T
he cerebellum is one of the brain areas most sensitive to RA and
EtOH (McCaffery et al., 2004; Kumar et al., 2010).
E
thanol treatment increase
s

RA levels in the
cerebellum (McCaffery et al., 2004), presumably by increas
ing

RA synthesis by astrocytes.

McCaffery et al. suggested that
EtOH does so by

stimulat
ing

shortchain retinol dehydrogenases
, but
this was not tested
. Furthermore, ethanol exposure in rats results in altered RAR and RXR receptor
expression and activity

(Kumer et al., 2010)
, which may re
su
lt in altered responses to RA. The
increase in R
XR activity may account for the increased apoptosis, as RXR targets apoptosis related
genes (Kumar et al., 2010). RAR activity on the other hand was decreased, which may account for
the decreased neural dif
ferentiation upon EtOH treatment. In adult mice, EtOH did not affect
cerebellar RA lev
els anymore (Kane et al., 2010).
Thus the toxic effects of EtOH on cerebellar
development may be mediated by altered RA levels and signaling.


The effect of Ethanol
exposure on RA signaling is stage dependent

Thus to summarize, there
is
considerable
overlap

between the teratogenic effects of EtOH and
disturbed RA signaling on neural development and it seems that at least a part of the teratogenic
effects of EtOH on ne
ural development can be contributed to the effect of EtOH on RA signaling.
The specific effect of ethanol on RA signaling is very much dependent on developmental stage (and
tissue).

E
arly on, at the onset

of RALDH2 activity when RALDH2

and RA levels are st
ill low,
ethanol competes for RALDH2 activity and this leads to a reduction of RA levels
, which may
account for at least some of the teratogenic effects of EtOH during early development
. Later
,

EtOH
may increase the activity and expression levels of RA synthesizing enzymes and the expression of
enzymes involved in retinol uptake in specific brain areas, resulting in higher retinol and RA
concentrations in these tissues. This combined results in
a drastic increase of RA levels which may
account for the observed late developmental defects

observed resulting from EtOH exposure
.
Furthermore, EtOH may alter the response to RA through its effect on RAR and RXR receptor
expression (Kumar et al., 2010).






28


Triazoles


Triazoles are used as a
ntifungal agen
ts in agriculture and medicine and affect fungal cell wall
integrity through inhibition of the

CYP51 enzyme (Robinson et al., 2012)
.
Examples of triazoles are
flusilazole (FLU), cyproconazole (CYP) and triadimefon (TDI).
These triazoles can have t
eratogenic
effects

on development,

including neural tube

development
. Defects include
axial defects
,
craniofacial defects, and defects in hind
brain patterning and segmentation. Skeletal analysis revealed
axial abnormalities, including transformation, fusion or duplication of axial segments, suggesting
that the specification of axial identity is affected by triazoles (Menegola et al., 2005
a
). As
the
establishment of axial patterning in the mesoderm and neural ectoderm involves similar mechanisms,
it seems likely that neural patterning is affected as well
.

Exposure to triazoles
at the beginning of
somitogenesis
also
affects

rhombomere patterning
;
K
ROX20 protein expression
in the

murine

hindbrain
was reduced and seemed
scattered
in triazole treated embryos
(
Menegola et al., 2004
;

Menegola et al., 2005a;
Menegola et al., 200
5b
).

In the
in vitro

rat system
HOX
B
1 protein
expression
was reduced and
scattered as well

(
Menegola et al., 2004
)
.
This scattering
could

be due to
impaired boundary formation in the rhombomeres: correct rhombomere specific, alternating Eph
expression is needed to prevent cell mixing between rhombomeres.

The

experiments

with the rat
in
vivo

and mouse
in vitro

system

suggested that th
e

c
raniofacial defects
of triazole treated embryos
may be the result of altered neural crest migration patterns, resulting from the abnormal rhombomere
patterning.



Triazoles interfere with
RA signaling

S
everal s
tudies suggest triazoles might interfere with RA levels. This is based on
the
similarities
between phenotypes of RA and triazole treated embryos,
the
changes of expression of genes
involved in RA metabolism
upon triazole exposure
and the substantial overlap
between

gene
expression changes upon treatment with Flusilazole or RA in whole embryo cultures

(WEC)

(
Menegola et al., 2004
;
Menegola et al., 2005
a
; Robinson et al., 2012).


First of all, RA is known to be involved in patterning

along the anterior
-
posterior axis in both
mesoderm and neural
ectoderm

and t
he axial defects of TDI treated embryos were similar to those
observed in

RA treated

embryos

(Menegola et al., 2005
a
).

Also the effect of RA or triazole treatment
on hindbrain

patterning

was similar, and
the

effects were additive as treatment with subteratogenic
doses of both RA and Fluconazole led to the same phenotype as the teratogenic dose of Fluconazole
alone (
Menegola et al., 2004
)
.

Second
, genes involved in RA metabolis
m were upregulated in response to triazole treatment

and
triazoles altered the expression of genes involved in
hindbrain patterning
, a process which is
dependent on and regulated by RA
.
Cyp26a1

was strongly upregulated in the rat WEC and in
zebrafish, as w
ell as
Dhrs

and
Crabp2

(Robinson et al., 2012; Hermsen et al., 2012)
.
In zebrafish
the
expression of the major gene involved in RA synthesis in early development,
Raldh2
,

was
downregulated, while genes involved in storage and degradation of RA were upregulated (Hermsen
et al., 2012).

Hindbrain patterning genes
Krox20

(
Egr2
) and kreisler (
Mafb
) were downregulated

in
the rat WEC
, whereas
Gbx2

is upregulated

(Robinson et al.,

2012)
.
Disturbed patterning, with a
reduction of
Krox20

expression was already

shown before
in vivo

and
in vitro
.
Some
Hox genes and
vHnf1

(
Hnf1b
) showed altered expression as well in zebrafish embryos (Hermsen et al., 2012)
.
An
29


upregulation of
Gbx2
may seem contradictory with an upregulation of RA, as
Gbx2
is involved in
patterning of the anterior rhombomeres while (excess) RA

has a posteriorizing effect

the hindbrain.
However,
Gbx2

expression in the hindbrain is dynamic, and
Gbx2
expression is not l
imited to the
hindbrain; it
is also expressed in the posterior embryo

(Martinez
-
Barbera et al., 2001)
. During
headfold to presomite stages
Gbx2
is broadly expressed in a large part of the embryo, later it
becomes more restricted to t
he caudal and hindbrain areas. Because of
this
, the

change of

Gbx2

expression may not be very telling; it does not tell in which part of the embryo its expression is
affected, and if there is a (slight) delay in development due to the treatment
4
, higher le
vels of
Gbx2

may be observed as well

(
as

it is still expressed in a larger domain)
.


Third, the sets of differentially expressed (DE) genes between RA and triazole treated whole
embryos showed significant overlap (31 of the 62 DE genes in triazole treated
embryos were also DE
in RA treated embryos), and the
se genes

showed a similar response to
either of the
treatments in
terms of expression changes of these genes (Robinson et al., 2012).

Thus, the effects of
triazole
exposure

show similarities to
the effect
s of
RA excess, on both the

molecular and phenotypic level.
Additionally,

triazole exposure affect
s

the expression of
at
least
some genes involved in RA
metabolism
.
This suggests that triazoles may elicit their teratogenic effects at least in part through
deregulation of RA levels.


Triazoles inhibit CYP26 activity

Triazoles

may
deregulate RA levels

by

inhibiti
ng

of CYP26A1

(Robinson et al., 2012)
. Experiments
indica
te triazoles might interact with

the active site

of CYP26A1

(
Ren et al., 2008;
Thatcher et al.,
2011;
Gomaa et al., 2012)
.

However if teratogenic effects of triazoles are caused by the inhibition of
CYP26 enzyme
s, it
seems

more

likely that
triazoles cause
these effects

by incompletely inhibiting
the activity of multiple CYP26 members

as k
nock
-
out of
Cyp26a1

leads to much more severe
axial
defects than observed in triazole treated embryos
, but it
s complete
deletion

only results in a mild
hindbrain phenotype and
a relatively normal

craniofacial phenotype
.
Cyp26a1
deletion

causes severe
body axis truncation
, well before the caudal levels

(Sakai et al., 2001; Abu
-
Abed et al., 2001)
,
whereas triazole treated embryos did

develop caudal vertebrae
.

This suggests that CYP
26
A
1
inhibition is not complete.

Cyp26a1
deletion

did cause an anterior transformation of the first lumbar
vertebra to a thirteenth thoracic vertebra identity

(Sakai et al., 2001)
, thus the anterior
transfor
mations and other segmental phenotypes in triazole treated embryos
could

in theory be
the
result of

CYP26A1 inhibition
. On

the

cervical levels
, however,
Cyp26a1
deletion

caused posterior
transformations

(Sakai et al., 2001; Abu
-
Abed et al., 2001)
.

The effect of
complete
genetic ablation
of
Cyp26a1

on hindbrain patterning however is very mild,
with only a slight effect on
Krox20

expression patterns (Sakai et al.,2001; Abu
-
Abed et al., 2001). There also was

a relatively normal
craniofacial appearance

(Abu
-
Abed et al., 2001)
.
Compound knock
-
out of
Cyp26a1

and
Cyp26c1

does lead
to severe,

more widespread hindbrain patterning defects
, failed neural crest migration

and
stronger craniofacial defects

(Uehara et al., 2007
)
, but it is embryonic lethal in mice before
embryonic day E11
.

Cyp26
b1

single deletion is also associated with abnormal neural crest migration
and craniofacial defects, but does not seem to alter hindbrain patterning (Maclean et al., 2009).

It



4

Triazole treatment does seem able to

delay development. TMS scores as well as somite count are lower in triazole
treated whole rat culture embryos 48 hours after a single dose exposure (Robinson et al., 2012), and in triazole treated
zebrafish embryos development is delayed as well(Hermsen e
t al., 2012)

30


thus seems likely that other CYP26 members are partially inhibited as well.
The idea that the
inhibition of multiple
Cyp26

enzymes results in more widespread defects in the hindbrain is in line
with the earlier proposed model in which dynamic
Cyp26

exp
ression regulates the extent of the RA
gradient

in the hindbrain, with some functional redundancy between the
Cyp26

members.

The strong upregulation of
Cyp26a1

expression in triazole treated embryos (rat whole embryo
culture / zebrafish), is also seen in e
mbryos exposed to excess RA and is likely part of a feedback
mechanism to regulate RA levels.


A potential mechanism for RA mediated teratogenic

effects o
f triazoles

Considering the effects of RA and triazole treatment on patterning
of the embryo,
phenotype

and
gene expression a

potential mechanism

for RA mediated effects of triazole
s on development of the
neural tube

could be

the following

(fig. 16)
:

Triazoles
partially inhibit

CYP26A1,

and
likely

the other CYP26 members as well,

which then
results

in locally

increased, ectopic levels of

RA. This will especially affect the caudal region
containing the posterior growth zone and the hindbrain. In the posterior growth zone decreased
CYP26A1 activity leads to ectopic RA in this region
. RA

may then
negat
ively
affect FGF8 levels

in
this region
. As increasing
Fgf8

levels are needed for

specification of

increasingly posterior tissues
in
the spinal cord,
this
lowering of FGF8 levels
may lead to

faulty specification of axial identity
(segment identity) and thus to disturbed patterning along the A
-
P axis. More specifically, it may l
ead
to altered segment identity

such as an anterior transformation
. L
ower FGF
8

levels

in and around the
node induce a

less posterior identity

in the cells leaving this area than is normal at the particular axial
level
. This is reflected by an

a
ltered induction of
Hox gene expression
.

P
roliferation rates
might be

decreased as well in
the posterior

growth zone

due to lower

levels of FGF
8
.

In the posterior
hindbrain a decrease of CYP26 expression may result in higher

and ectopic

RA
levels

in the
hindbrain
.
This may lead to faulty

Hox gene expression
patterns and to faulty

rhombomere
specification and identity. In turn this leads to altere
d hindbrain development and disturbed
neural
crest migration patterns.

The latter could then result in craniofacial defects.

In addition
,
disturbed RA
signaling in the b
ranchial arches may

contribute

to the craniofacial deformities

as well
,

as
Cyp26
enzymes are express
ed in the branchial arches

(Rhinn and Dollé, 2012)
.





31




Fig.
16

Proposed adverse outcome pathway of embryonic triazole exposure,
focusing

on the

effects

on neural tube
patterning and the embryonic axis.

Triazole

exposure during embryonic development can lead to developmental abnormalities, such as shortening along the
anterior
-
posterior

(A
-
P)

axis,

patterning defects along this axis,
hindbrain defects

and craniofacial de
fects
.

Triazoles were
shown to affect the expression of some hindbrain patterning genes (e.g. Krox20, Hoxb1). Studies suggest that these
hindbrain patterning defects result in disturbed neural crest cell migration patterns and that this may account for the

craniofacial defects. However
Cyp26

expression is also observed in
the branchial arch mesenchyme
, thus increased or
ectopic RA levels in the branchial arches may contribute as well to these defects.

Triazoles are able to inhibit CYP26A.
The resulting
ectopic or increased RA levels may account for part of the observed developmental defects.

A potential
mechanism which may explain part of the teratogenic effects of triazole exposure was extrapolated, based on the
observed developmental effects in triazol
e treated embryos, the current knowledge of RA function in the developing
embryo, the considerable overlap between
effects

of triazole
and RA treatment on development and gene expression, and
the similarities with the phenotypes exhibited by

Cyp26

deficien
t embryos.

In addition to disruption of axial and hindbrain development, inhibition o
f CYP26 enzymes could additionally result in
higher RA levels in the growth buds of other developing organs, such as the limb buds, resulting in developmental
defects ther
e. CYP26 enzymes are for example also expressed in the limb buds and branchial arch mesenchyme
(reviewed by Rhinn and Dollé, 2012)

and t
riazoles are known to affect
development of these structures

(Robinson et al.,
2012)
.





32


With this model in mind, interesting genes to look at
in relation to assessment of potential neurotoxic
effects
would be:



Fgf8



this gene

is

involved in axis elongation and patterning, excess RA can downregulate
expression of this gene
.
Fgf8

is also inte
resting because it is expressed in the growth buds of
several developing organs
in which

RA plays a role as well.

In the case of triazoles a decrease
in
Fgf8

may be expected.

Wnt3a
might also be interesting in this respect
.



Hox

genes



these genes confer positional identity to the spinal cord and posterior hindbrain.
If FGF8 levels are lower in the posterior growth zone, it might be expected that the posterior
Hox genes are induced
less or only at more posterior levels than normal
in t
he developing
spinal cord
. The more 3’ Hox genes would be
induced

in more posterior tissues than normal
.
In the hindbrain
the Hox genes

are involved in patterning and segment identity.

Increased RA
levels might result
in an enhanced

anterior
expansion of the
expression domains of the Hox1
-
Hox5 genes
,

i.e. their domains expanding

more anteriorly than normal,

but since there are a
lot of feedback loops involved in establishing the
ir

rhombomere specific expression domains,
the effect of RA on the

Hox expression

patterns
at the

rhombomere specification
stages
might
be hard to predict.



The hind
brain patterning
genes

described previously
, such as
Krox20 (Egr2),
k
reisler (
Mafb)
,
Fgf3, Gbx2, Meis2
, Otx2
, vHnf (Hnf1b)



altered
o
f expression of these
genes reflect
disturbed

anterior
-
posterior patterning of the hindbrain
.



Neural differentiation genes, such as
Neurogenin1

(Neurog1/Ngn1)
, Neurogenin
2
(Neurog2/Ngn2)
, NeuroM

(
Neurod4)



the presence of RA and a drop in
Fgf8

are needed for
neuronal differentiation.
Shh

and
Gli2
might be interesting in this respect as well
.
The
presence of RA and the drop of
Fgf8
levels are

also

needed for activation of
Shh

in the floor
plate
and
for
activation

of
other dorsal
-
ventral
patterning
related genes.

RA dependent
repression of
Gli2

seems necessary for the cells to be able to respond to the
Shh
signals
(Ribes et al., 2009).

Additionally, studying
the
expression of the genes involved in RA synthesis (most importantly
Rdh10, Rald
h2
), catabolism (
Cyp26

family), retinol uptake (Stra6) and RA signaling (
RAR
and
RXR
receptor family) may provide additional insight into potential effects on RA levels and RA signaling
activity.
M
easurin
g RA levels could be insightful as well.



Unfortunately
,

I did not have a look at the expression data to see whether these genes did change
their expression

upon triazole exposure

in

the whole rat
embryo culture

(WEC), zebrafish embryos,
embryonic stem cell test (ESTc)

or
neural embryonic stem cel
l test (ESTn)
.
Some of the

genes listed
above were mentioned in the articles describing the rat WEC and zebrafish tests
; these
were

mentioned in the overview

of the effects of triazole treatment in the text
above. Interestingly,
Fgf8

was downregulated upon

triazole exposure in the neural Embryonic Stem Cell test (ESTn), while
Rarb

and
Cyp26b1
expression

was

increased (Theunissen et al., 2012).

Expression of
Hoxb4, Hoxc4

and
Hoxb5

was upregulated as well.





33


Conclusions



The classic way to test for possible teratogenic effects of compounds is by exposing model
organisms (animals) to a high dose of the compound to see if development is affected (Krewski et
al., 2007).
B
ecause of the high amount of test animals is needed,
the costs (time and money) and the
limited insight it provides in the mechanisms underlying the toxic effects, there is a demand for
alternative approaches.

In modern toxicity research

novel approaches are being developed for
assessment of potential risks.

One of these approaches is
based

on the idea of looking at the effects
of a compound on
specific
key pathways
, of which its distortion may be indicative of an adverse
effect on (for example) embryonic development. I
f such a key pathway is altered upon exp
osure to a
compound, this

might enable the researcher
to predict the developmental outcome

based on which
components

of the pathway
are deregulated
.
Also, it might
become

possible to

reconstruct
(part of)
the sequence of events that leads

to the developmen
tal defects

after exposure
: the adverse outcome
pathway (AOP). The RA signaling pathway was proposed
as

a candidate pathway to study in relation
to neural tube development, due to its central role in development and the association of disturbed
RA signalin
g with neural tube defects. During early neural tube development RA is especially
important for anterior
-
posterior patterning of the spinal cord and hindbrain, neural differentiation and
axis elongation. Furthermore, RA is required to allow cells to respon
d to ventralizing
Shh
signals.

In order to be able to construct an AOP based on the effects of a compound on RA signaling, the role
of RA signaling in neural tube development was reviewed.
The information of how a compound
affects RA signaling and what the effects are on the genes downstream of RA in the hindbrain and
spinal cord, might allow
the

deduction of the sequential steps that may lead to a developmental
phenotype. The effect

of a co
mpound on (embryonic) development is very dependent on the stag
e(
s)
at which the embryo is exposed, the duration of exposure and the dose. This review focused on early
neural tube development; exposure at later stages may have quite different effects.


Po
tential neurotoxic risks may be identified by studying
the effect of a compound

on RA levels and
on
signaling

downstream of RA
.

If RA levels cannot be measured directly, changes in the expression
of genes associated with RA signaling, may be indicative of disturbed RA signaling. Genes involved
in RA synthesis (most importantly
Rdh10, Raldh2
), catabolism (
Cyp26

family), retinol upta
ke
(
Stra6
) and RA signaling (
RAR
and
RXR
receptor family) may provide insight into potential effects
on RA levels and RA signaling activity. The

potential

effects on neural tube development
could

be
characterized by l
ooking at marker genes that act downstr
eam of RA
. Interesting genes to study in
relation to early spinal cord development are
Fgf8
, the Hox genes and neural differentiation genes
such as
Ngn1, Ngn2, Neurod4, Shh and Gli2
. Genes such as
Mafb (
kreisler),
Egr2 (Krox20), Meis2,
Hnf1b, Fgf3, Gbx2, E
n2, Pax2, Otx2

and the Hox genes from paralogous group 1
-
4 (Hox1
-
Hox4)
provide information on the effects on patterning of the hindbrain.

Potential

effects of
exposure to a compound on patterning of the neural tube, neural differentiation or
axis elongati
on might be revealed by in situ hybridization with probes for these marker genes.
Alternatively, gene expression levels could be compared between treated and control embryos, for
example by Q
-
PCR or micro
-
array analysis. If whole embryos are used for gene
expression level
comparisons, interpretation of the results should be done with care. Some of the genes considered in
34


this review are expressed at multiple sites in the embryo,
as they are

involved in different processes.
This may make it difficult to pinp
oint in which part of the embryo the expression is affected.
Additionally, this approach may not be sensitive enough to pick up local changes, nor will it detect a
shift (e.g. an anterior or posterior shift) of an expression domain. Also the expression of
many

genes
is dynamic
during the course of development.

The

expression domains of some of the hindbrain
patterning genes
, for example,

either expand or shrink between
the onset of

hindbrain
specification
and

the stage of

hindbrain segmentation. If there is

a slight delay in development, this may also result
in altered expression patterns in the embryo as well
, while
hindbrain specification
might be
unaltered
. It is therefore
recommended

to
match
embryos
based on

developmental stage, rather than
age.

With
the neural embryonic stem cell test

(
EST
n
)
, it may become possible to do an initial assessment
of the potential risks
of

many compounds
at once

i
n

a relatively short time
scale
(Theunissen et al.,
2013). Analysis of expression changes

of aforementioned mar
ker genes

may give an indication
whether RA signaling
is affected, h
owever the time and
region

specific context may be lost
. Thus
t
issue specific downstream effects

of RA signaling
might

not be picked up
.
The EST
n

might thus
reflect the effects

of compound

exposure
on
an
embryo less well, but it may

still

give an indication
whether RA signaling may be affected. In line with this

notion
,

is the observation
that the

gene
expression changes in

the embryonic stem cell test (ESTc)
upon flusilazole treatment only

show a
mild correlation with the gene expression changes in flusilazole treated zebrafish embryos (Robinson
et al., 2012)
. However, gene expression changes in the cells do

indicat
e

that RA signaling is affected

by triazole exposure
; expression
levels
of t
he RA signaling associated genes
Rarb

and
Cyp26b1
increase
d

after exposure to another triazole (cyproconazole) (Theunissen et al., 2012).

Also the
expression

levels

of a few

Hox genes, which can be directly regulated by RA,
were

upregulated as
well. On the

other hand,
Fgf8
expression, which
in vivo

is suppressed by RA, was downregulated
upon tria
zole exposure

(Theunissen et al., 2012). These expression changes could indicate that RA
signaling is increased upon triazole exposur
e, which is concordant
with the

suggestion

that triazoles
may increase RA signaling
in vivo
. Thus, it may be possible to detect changes in RA signaling with
the ESTn and
such a change in RA signaling

may be an indication that the compound might affect
neural tube development in embryos.



For two types of toxicants, ethanol and triazoles, potential AOPs were proposed

that

linked the
molecular initiating event

(MIE)



interaction of the compound wi
th RALDH2 or CYP26
respectively


to the observed developmental defects.

The AOPs were based

on the effects of the
toxicants on RA signaling
. The effects that
the

deregulation of RA signaling may have

had

on
(neural tube) development was then extrapolated, based on the known roles of RA signaling
described in this review and the resemblance
betwe
en

t
he developmental defects associated with
exposure to the compound

and

deregulated RA signaling
.

This led to a proposal of

a series of

potential
intermediate steps

that might link

the MIE

to the observed developmental defects on
the
organ and organismal

level
.
Additional morphological and molecular analyses (
e.g.
gene expression

analyses, in situ hybridization
) are needed t
o test the validity of the AOPs and
to refine them.

Thus in
the case of EtOH and triazoles exposure

it was possible to po
se
hypothes
is

(the AOP) on
how
compound exposure might have led to the associated

developmental defects, based on current
knowledge of the role of RA in neural tube development and axis patterning. These
AOP models
might

then be used to further investigate the mechan
isms underlying triazole and ethanol exposure.

35


G
ene expression measurements
after triazole exposure
were already done in the rat WEC at the 2
-
4
somite stages and in the neural embryonic stem cell test (Robinson et al., 2012; Theunissen et al.,
2
012). A r
ev
isit

of

these data, focusing on the previously mentioned list of genes of interest, may
already provide additional information
that might help test
the validity of the
proposed
AOP and to
refine or adjust it. Gene expression was also measured in the zebraf
ish embryo, however this was
measured at 24 hpf (Hermsen et al., 2012), at which point axis elongation is (nearly) completed and
the hindbrain has already segmented.



Thus, the

RA signaling pathway and the associated genes (biomarkers) summarized here may

in the
future be
studied

to assess potential risks

posed by exposure to
a
compound in relation to early neural
tube development
and to help building AOP networks. Changes in

expression of

these genes are
indicative of defects in anterior
-
posterior pattern
ing of the spinal cord and hindbrain
, or of disruption
of

differentiation
or

axis elongation (organizer function).
In the case of the teratogenic triazoles, gene

expression
levels after exposure were already measured

in several model systems;
reexamination

of
these data could provide new insights into the mechanisms underlying the teratogenic effects of
triazole exposure.





36


Acknowledgements

I’d like to express my special thanks to Prof. Dr. Piersma for supervising me and providing me the
opportunity to
write my thesis at the RIVM, and Dr. Ilse Tonk, for supervising me and giving me
feedback on previous versions of this thesis.



References



Ankley, G.T., Bennett, R.S., Erickson, R.J., Hoff, D.J., Hornung, M.W., Johnson, R.D., Mount, D.R., Nichols, J.W.
,
Russom, C.L. & Schmieder, P.K. 2010, "Adverse outcome pathways: a conceptual framework to support
ecotoxicology research and risk assessment",
Environmental Toxicology and Chemistry,
vol. 29, no. 3, pp. 730
-
741.

Aragon, F. & Pujades, C. 2009, "FGF signa
ling controls caudal hindbrain specification through Ras
-
ERK1/2 pathway",
BMC developmental biology,
vol. 9, pp. 61
-
213X
-
9
-
61.

Bel
-
Vialar, S., Itasaki, N. & Krumlauf, R. 2002, "Initiating Hox gene expression: in the early chick neural tube
differential se
nsitivity to FGF and RA signaling subdivides the HoxB genes in two distinct groups",
Development,
vol. 129, no. 22, pp. 5103
-
5115.

Blentic, A., Gale, E. & Maden, M. 2003, "Retinoic acid signalling centres in the avian embryo identified by sites of
express
ion of synthesising and catabolising enzymes",
Developmental dynamics : an official publication of the
American Association of Anatomists,
vol. 227, no. 1, pp. 114
-
127.

Burroughs
-
Garcia, J., Sittaramane, V., Chandrasekhar, A. &

Waters, S.T. 2011, "Evolutionarily conserved function of
Gbx2 in anterior hindbrain development",
Developmental dynamics : an official publication of the American
Association of Anatomists,
vol. 240, no. 4, pp. 828
-
838.

Cordes, S.P. & Barsh, G.S. 1994, "
The mouse segmentation gene kr encodes a novel basic domain
-
leucine zipper
transcription factor",
Cell,
vol. 79, no. 6, pp. 1025
-
1034.

Dasen, J.S., Liu, J. & Jessell, T.M. 2003, "Motor neuron columnar fate imposed by sequential phases of Hox
-
c activity",
Nature,
vol. 425, no. 6961, pp. 926
-
933.

Deltour, L., Ang, H.L. & Duester, G. 1996, "Ethanol inhibition of retinoic acid synthesis as a potential mechanism for
fetal alcohol syndrome",
FASEB journal : official publication of the Federation of American Soc
ieties for
Experimental Biology,
vol. 10, no. 9, pp. 1050
-
1057.

Deschamps, J. & van Nes, J. 2005, "Developmental regulation of the Hox genes during axial morphogenesis in the
mouse",
Development,
vol. 132, no. 13, pp. 2931
-
2942.

Dias, M.S. & Partington,
M. 2004, "Embryology of myelomeningocele and anencephaly",
Neurosurgical focus,
vol. 16,
no. 2, pp. 1
-
16.

Diez del Corral, R., Olivera
-
Martinez, I., Goriely, A., Gale, E., Maden, M. & Storey, K. 2003, "Opposing FGF and
retinoid pathways control ventral ne
ural pattern, neuronal differentiation, and segmentation during body axis
extension",
Neuron,
vol. 40, no. 1, pp. 65
-
79.

Gavalas, A. 2002, "ArRAnging the hindbrain",
Trends in neurosciences,
vol. 25, no. 2, pp. 61
-
64.

Gilbert SF. Developmental Biology. 6
th edition. Sunderland (MA): Sinauer Associates; 2000. Available from:
http://www.
ncbi.nlm.nih.gov/books/NBK9983/

Gilbert SF. Developmental Biology. 9th edition. Sunderland (MA): Sinauer Associates; 2010

Giudicelli, F., Gilardi
-
Hebenstreit, P., Mechta
-
Grig
oriou, F., Poquet, C. & Charnay, P. 2003, "Novel activities of Mafb
underlie its dual role in hindbrain segmentation and regional specification",
Developmental biology,
vol. 253, no. 1,
pp. 150
-
162.

Giudicelli, F., Taillebourg, E., Charnay, P. & Gilardi
-
H
ebenstreit, P. 2001, "Krox
-
20 patterns the hindbrain through both
cell
-
autonomous and non cell
-
autonomous mechanisms",
Genes & development,
vol. 15, no. 5, pp. 567
-
580.

37


Glover, J.C., Renaud, J.S. & Rijli, F.M. 2006, "Retinoic acid and hindbrain patterning
",
Journal of neurobiology,
vol. 66,
no. 7, pp. 705
-
725.

Gomaa, M.S., Lim, A.S.T., Wilson Lau, S.C., Watts, A., Illingworth, N.A., Bridgens, C.E., Veal, G.J., Redfern, C.P.F.,
Brancale, A., Armstrong, J.L. & Simons, C. 2012, "Synthesis and CYP26A1 inhibit
ory activity of novel methyl 3
-
[4
-
(arylamino)phenyl]
-
3
-
(azole)
-
2,2
-
dimethylpropanoates",
Bioorganic & medicinal chemistry,
vol. 20, no. 20, pp.
6080
-
6088.

Gould, A., Itasaki, N. & Krumlauf, R. 1998, "Initiation of rhombomeric Hoxb4 expression requires ind
uction by somites
and a retinoid pathway",
Neuron,
vol. 21, no. 1, pp. 39
-
51.

Hale, L.A., Tallafuss, A., Yan, Y.L., Dudley, L., Eisen, J.S. & Postlethwait, J.H. 2006, "Characterization of the retinoic
acid receptor genes raraa, rarab and rarg during zebra
fish development",
Gene expression patterns : GEP,
vol. 6,
no. 5, pp. 546
-
555.

Hermsen, S.A., Pronk, T.E., van den Brandhof, E.J., van der Ven, L.T. & Piersma, A.H. 2012, "Concentration
-
response
analysis of differential gene expression in the zebrafish em
bryotoxicity test following flusilazole exposure",
Toxicological sciences : an official journal of the Society of Toxicology,
vol. 127, no. 1, pp. 303
-
312.

Hernandez, R.E., Putzke, A.P., Myers, J.P., Margaretha, L. & Moens, C.B. 2007, "Cyp26 enzymes gener
ate the retinoic
acid response pattern necessary for hindbrain development",
Development (Cambridge, England),
vol. 134, no. 1,
pp. 177
-
187.

Kane, M.A., Folias, A.E., Wang, C. & Napoli, J.L. 2010, "Ethanol elevates physiological all
-
trans
-
retinoic acid le
vels in
select loci through altering retinoid metabolism in multiple loci: a potential mechanism of ethanol toxicity",
FASEB
journal : official publication of the Federation of American Societies for Experimental Biology,
vol. 24, no. 3, pp.
823
-
832.

Kaya
m, G., Kohl, A., Magen, Z., Peretz, Y., Weisinger, K., Bar, A., Novikov, O., Brodski, C. & Sela
-
Donenfeld, D.
2013, "A novel role for Pax6 in the segmental organization of the hindbrain",
Development,
vol. 140, no. 10, pp.
2190
-
2202.

Kiecker, C. & Lumsden
, A. 2005, "Compartments and their boundaries in vertebrate brain development",
Nature Reviews
Neuroscience,
vol. 6, no. 7, pp. 553
-
564.

Kim, F.A., Sing, l.A., Kaneko, T., Bieman, M., Stallwood, N., Sadl, V.S. &

Cordes, S.P. 2005, "The vHNF1
homeodomain protein establishes early rhombomere identity by direct regulation of Kreisler expression",
Mechanisms of development,
vol. 122, no. 12, pp. 1300
-
1309.

Kot
-
Leibovich, H. & Fainsod, A. 2009, "Ethanol induces embry
onic malformations by competing for retinaldehyde
dehydrogenase activity during vertebrate gastrulation",
Disease models & mechanisms,
vol. 2, no. 5
-
6, pp. 295
-
305.

Krewski, D., Acosta Jr, D., Andersen, M., Anderson, H., Bailar III, J.C., Boekelheide, K.,

Brent, R., Charnley, G.,
Cheung, V.G. & Green Jr, S. 2007, "Summary" in
Toxicity testing in the 21st century: a vision and a strategy

Taylor & Francis,

pp. 1
-
17.

Kumar, A., Singh, C.K., DiPette, D.D. & Singh, U.S. 2010, "Ethanol impairs activation of ret
inoic acid receptors in
cerebellar granule cells in a rodent model of fetal alcohol spectrum disorders",
Alcoholism, Clinical and
Experimental Research,
vol. 34, no. 5, pp. 928
-
937.

Liu, J., Laufer, E. & Jessell, T.M. 2001, "Assigning the positional ident
ity of spinal motor neurons: rostrocaudal
patterning of Hox
-
c expression by FGFs, Gdf11, and retinoids",
Neuron,
vol. 32, no. 6, pp. 997
-
1012.

Liu, A. & Joyner, A.L. 2001, "Early anterior/posterior patterning of the midbrain and cerebellum",
Annual Review

of
Neuroscience,
vol. 24, pp. 869
-
896.

MacLean, G., Abu
-
Abed, S., Dolle, P., Tahayato, A., Chambon, P. & Petkovich, M. 2001, "Cloning of a novel retinoic
-
acid metabolizing cytochrome P450, Cyp26B1, and comparative expression analysis with Cyp26A1 during
early
murine development",
Mechanisms of development,
vol. 107, no. 1
-
2, pp. 195
-
201.

Maclean, G., Dolle, P. & Petkovich, M. 2009, "Genetic disruption of CYP26B1 severely affects development of neural
crest derived head structures, but does not compromise

hindbrain patterning",
Developmental dynamics : an official
publication of the American Association of Anatomists,
vol. 238, no. 3, pp. 732
-
745.

Maden, M. 2006, "Retinoids and spinal cord development",
Journal of neurobiology,
vol. 66, no. 7, pp. 726
-
738.

38


Makki, N. & Capecchi, M.R. 2011, "Identification of novel Hoxa1 downstream targets regulating hindbrain, neural crest
and inner ear development",
Developmental biology,
vol. 357, no. 2, pp. 295
-
304.

Marrs, J.A., Clendenon, S.
G., Ratcliffe, D.R., Fielding, S.M., Liu, Q. & Bosron, W.F. 2010, "Zebrafish fetal alcohol
syndrome model: effects of ethanol are rescued by retinoic acid supplement",
Alcohol (Fayetteville, N.Y.),
vol. 44,
no. 7
-
8, pp. 707
-
715.

Martinez
-
Barbera, J.P., Si
gnore, M., Boyl, P.P., Puelles, E., Acampora, D., Gogoi, R., Schubert, F., Lumsden, A. &
Simeone, A. 2001, "Regionalisation of anterior neuroectoderm and its competence in responding to forebrain and
midbrain inducing activities depend on mutual antagonism

between OTX2 and GBX2",
Development (Cambridge,
England),
vol. 128, no. 23, pp. 4789
-
4800.

McCaffery, P., Koul, O., Smith, D., Napoli, J.L., Chen, N. & Ullman, M.D. 2004, "Ethanol increases retinoic acid
production in cerebellar astrocytes and in cerebel
lum",
Brain research.Developmental brain research,
vol. 153, no.
2, pp. 233
-
241.

Menegola, E., Broccia, M.L., Di Renzo, F., Massa, V. & Giavini, E. 2005a, "Craniofacial and axial skeletal defects
induced by the fungicide triadimefon in the mouse",
Birth d
efects research.Part B, Developmental and reproductive
toxicology,
vol. 74, no. 2, pp. 185
-
195.

Menegola, E., Broccia, M.L., Di Renzo, F., Massa, V. & Giavini, E. 2005b, "Study on the common teratogenic pathway
elicited by the fungicides triazole
-
derivati
ves",
Toxicology in vitro : an international journal published in
association with BIBRA,
vol. 19, no. 6, pp. 737
-
748.

Menegola, E., Broccia, M.L., Di Renzo, F., Massa, V. & Giavini, E. 2004, "Relationship between hindbrain
segmentation, neural crest cell

migration and branchial arch abnormalities in rat embryos exposed to fluconazole
and retinoic acid in vitro",
Reproductive toxicology (Elmsford, N.Y.),
vol. 18, no. 1, pp. 121
-
130.

Mic, F.A., Haselbeck, R.J., Cuenca, A.E. & Duester, G. 2002, "Novel retin
oic acid generating activities in the neural tube
and heart identified by conditional rescue of Raldh2 null mutant mice",
Development (Cambridge, England),
vol.
129, no. 9, pp. 2271
-
2282.

Mollard, R., Viville, S., Ward, S.J., Decimo, D., Chambon, P. & Dol
le, P. 2000, "Tissue
-
specific expression of retinoic
acid receptor isoform transcripts in the mouse embryo",
Mechanisms of development,
vol. 94, no. 1
-
2, pp. 223
-
232.

Niederreither, K., Vermot, J., Schuhbaur, B., Chambon, P. & Dollé, P. 2000, "Retinoic ac
id synthesis and hindbrain
patterning in the mouse embryo",
Development,
vol. 127, no. 1, pp. 75
-
85.

Niederreither, K., Subbarayan, V., Dolle, P. &

Chambon, P. 1999, "Embryonic retinoic acid synthesis is essential for
early mouse post
-
implantation development",
Nature genetics,
vol. 21, no. 4, pp. 444
-
448.

Niederreither, K., Vermot, J., Fraulob, V., Chambon, P. & Dolle, P. 2002, "Retinaldehyde dehyd
rogenase 2 (RALDH2)
-

independent patterns of retinoic acid synthesis in the mouse embryo",
Proceedings of the National Academy of
Sciences of the United States of America,
vol. 99, no. 25, pp. 16111
-
16116.

Olivera
-
Martinez, I. & Storey, K.G. 2007, "Wnt si
gnals provide a timing mechanism for the FGF
-
retinoid differentiation
switch during vertebrate body axis extension",
Development,
vol. 134, no. 11, pp. 2125
-
2135.

Oosterveen, T., Niederreither, K., Dollé, P., Chambon, P., Meijlink, F. & Deschamps, J. 2003
, "Retinoids regulate the
anterior expression boundaries of 5′ Hoxb genes in posterior hindbrain",
The EMBO journal,
vol. 22, no. 2, pp.
262
-
269.

Pasqualetti, M., Neun, R., Davenne, M. & Rijli, F.M. 2001, "Retinoic acid rescues inner ear defects in Hoxa1
deficient
mice",
Nature genetics,
vol. 29, no. 1, pp. 34
-
39.

Reijntjes, S., Gale, E. & Maden, M. 2004, "Generating gradients of retinoic acid in the chick embryo: Cyp26C1
expression and a comparative analysis of the Cyp26 enzymes",
Developmental dynamics
: an official publication of
the American Association of Anatomists,
vol. 230, no. 3, pp. 509
-
517.

Ren, J., Xiong, X., Sha, Y., Yan, M., Lin, B., Wang, J., Jing, Y., Zhao, D. & Cheng, M. 2008, "Structure prediction and
R115866 binding study of human CYP26
A1: homology modelling, fold recognition, molecular docking and MD
simulations",
Molecular Simulation,
vol. 34, no. 3, pp. 337
-
346.

Rhinn, M. & Dollé, P. 2012, "Retinoic acid signalling during development",
Development,
vol. 139, no. 5, pp. 843
-
858.

Rhin
n, M. & Brand, M. 2001, "The midbrain
--
hindbrain boundary organizer",
Current opinion in neurobiology,
vol. 11,
no. 1, pp. 34
-
42.

39


Ribes, V., Le Roux, I., Rhinn, M., Schuhbaur, B. & Dolle, P. 2009, "Early mouse caudal development relies on crosstalk
betwee
n retinoic acid, Shh and Fgf signalling pathways",
Development (Cambridge, England),
vol. 136, no. 4, pp.
665
-
676.

Robinson, J.F., Tonk, E.C., Verhoef, A. & Piersma, A.H. 2012, "Triazole induced concentration
-
related gene signatures
in rat whole embryo cu
lture",
Reproductive toxicology (Elmsford, N.Y.),
vol. 34, no. 2, pp. 275
-
283.

Sakai, Y., Meno, C., Fujii, H., Nishino, J., Shiratori, H., Saijoh, Y., Rossant, J. & Hamada, H. 2001, "The retinoic acid
-
inactivating enzyme CYP26 is essential for establishin
g an uneven distribution of retinoic acid along the anterio
-
posterior axis within the mouse embryo",
Genes & development,
vol. 15, no. 2, pp. 213
-
225.

Santos
-
Ledo, A., Cavodeassi, F., Carreño, H., Aijón, J. & Arévalo, R. 2013, "Ethanol alters gene express
ion and cell
organisation during optic vesicle evagination",
Neuroscience,
.

Serpente, P., Tumpel, S., Ghyselinck, N.B., Niederreither, K., Wiedemann, L.M., Dolle, P., Chambon, P., Krumlauf, R.
& Gould, A.P. 2005, "Direct crossregulation between retinoic
acid receptor {beta} and Hox genes during hindbrain
segmentation",
Development (Cambridge, England),
vol. 132, no. 3, pp. 503
-
513.

Sirbu, I.O., Gresh, L., Barra, J. & Duester, G. 2005, "Shifting boundaries of retinoic acid activity control hindbrain
segme
ntal gene expression",
Development (Cambridge, England),
vol. 132, no. 11, pp. 2611
-
2622.

Sunmonu, N.A., Li, K., Guo, Q. & Li, J.Y. 2011, "Gbx2 and Fgf8 are sequentially required for formation of the
midbrain
-
hindbrain compartment boundary",
Development (
Cambridge, England),
vol. 138, no. 4, pp. 725
-
734.

Tahayato, A., Dolle, P. &

Petkovich, M. 2003, "Cyp26C1 encodes a novel retinoic acid
-
metabolizing enzyme expressed
in the hindbrain, inner ear, first branchial arch and tooth buds during murine development",
Gene expression
patterns : GEP,
vol. 3, no. 4, pp. 449
-
454.

Thatcher, J.
E., Buttrick, B., Shaffer, S.A., Shimshoni, J.A., Goodlett, D.R., Nelson, W.L. & Isoherranen, N. 2011,
"Substrate specificity and ligand interactions of CYP26A1, the human liver retinoic acid hydroxylase",
Molecular
pharmacology,
vol. 80, no. 2, pp. 228
-
23
9.

Theunissen, P.T., Pennings, J.L., van Dartel, D.A., Robinson, J.F., Kleinjans, J.C. & Piersma, A.H. 2013,
"Complementary Detection of Embryotoxic Properties of Substances in the Neural and Cardiac Embryonic Stem
Cell Tests",
Toxicological Sciences,
vol
. 132, no. 1, pp. 118
-
130.

Theunissen, P.T., Robinson, J.F., Pennings, J.L., de Jong, E., Claessen, S.M., Kleinjans, J.C. & Piersma, A.H. 2012,
"Transcriptomic concentration
-
response evaluation of valproic acid, cyproconazole, and hexaconazole in the neur
al
embryonic stem cell test (ESTn)",
Toxicological sciences : an official journal of the Society of Toxicology,
vol. 125,
no. 2, pp. 430
-
438.

Uehara, M., Yashiro, K., Mamiya, S., Nishino, J., Chambon, P., Dolle, P. & Sakai, Y. 2007, "CYP26A1 and CYP26C1
c
ooperatively regulate anterior
-
posterior patterning of the developing brain and the production of migratory cranial
neural crest cells in the mouse",
Developmental biology,
vol. 302, no. 2, pp. 399
-
411.

Vendrell, V., Vazquez
-
Echeverria, C., Lopez
-
Hernandez, I., Alonso, B.D., Martinez, S., Pujades, C. & Schimmang, T.
2013, "Roles of Wnt8a during formation and patterning of the mouse inner ear",
Mechanisms of development,
vol.
130, no. 2
-
3, pp. 160
-
168.

Wa
llingford, J.B., Niswander, L.A., Shaw, G.M. & Finnell, R.H. 2013, "The Continuing Challenge of Understanding,
Preventing, and Treating Neural Tube Defects",
Science,
vol. 339, no. 6123.

Wassef, M.A., Chomette, D., Pouilhe, M., Stedman, A., Havis, E., Des
marquet
-
Trin Dinh, C., Schneider
-
Maunoury, S.,
Gilardi
-
Hebenstreit, P., Charnay, P. & Ghislain, J. 2008, "Rostral hindbrain patterning involves the direct activation
of a Krox20 transcriptional enhancer by Hox/Pbx and Meis factors",
Development (Cambridge,

England),
vol. 135,
no. 20, pp. 3369
-
3378.

White, J.C., Highland, M., Kaiser, M. & Clagett
-
Dame, M. 2000, "Vitamin A deficiency results in the dose
-
dependent
acquisition of anterior character and shortening of the caudal hindbrain of the rat embryo",
Dev
elopmental biology,
vol. 220, no. 2, pp. 263
-
284.

White, J.C., Shankar, V.N., Highland, M., Epstein, M.L., DeLuca, H.F. & Clagett
-
Dame, M. 1998, "Defects in
embryonic hindbrain development and fetal resorption resulting from vitamin A deficiency in the ra
t are prevented
by feeding pharmacological levels of all
-
trans
-
retinoic acid",
Proceedings of the National Academy of Sciences,
vol.
95, no. 23, pp. 13459
-
13464.

40


Wilson, V., Olivera
-
Martinez, I. & Storey, K.G. 2009, "Stem cells, signals and vertebrate bod
y axis extension",
Development,
vol. 136, no. 10, pp. 1591
-
1604.

Wolpert, L., Jessell, T., Lawrence, P., Meyerowitz, E., Robertson, E. & Smith, J. 2007,
Principles of development,
3rd
edn, Oxford University Press Oxford.

Wong, E.Y., Wang, X.A., Mak, S.S.
, Sae
-
Pang, J.J., Ling, K.W., Fritzsch, B. & Sham, M.H. 2011, "Hoxb3 negatively
regulates Hoxb1 expression in mouse hindbrain patterning",
Developmental biology,
vol. 352, no. 2, pp. 382
-
392.

Ybot
-
Gonzalez, P., Cogram, P., Gerrelli, D. & Copp, A.J. 2002,
"Sonic hedgehog and the molecular regulation of mouse
neural tube closure",
Development,
vol. 129, no. 10, pp. 2507
-
2517.

Yelin, R., Kot, H., Yelin, D. & Fainsod, A. 2007, "Early molecular effects of ethanol during vertebrate embryogenesis",
Differentiati
on; research in biological diversity,
vol. 75, no. 5, pp. 393
-
403.

Yelin, R., Schyr, R.B., Kot, H., Zins, S., Frumkin, A., Pillemer, G. & Fainsod, A. 2005, "Ethanol exposure affects gene
expression in the embryonic organizer and reduces retinoic acid leve
ls",
Developmental biology,
vol. 279, no. 1, pp.
193
-
204.

Young, T. & Deschamps, J. 2009, "
Hox
,
Cdx
, and Anteroposterior Patterning in the Mouse Embryo",
Current topics in
developmental biology,
vol. 88, pp. 235
-
255.