Patte rn i n g the ve rteb rate body

p l a n I I : the mesod erm a nd ea rly

n e rvous syste m

Som ite fo rmation a n d pattern i n g " Early o n , we behaved as a

T h e ro l e o f t h e o rg a n izer re gion a n d n e u ra l i n d u ction
s i n g l e u n it, as th ere were n o
ba rriers between us. O n ce o u r
pOSiti o n s were known, we

beca m e sepa rate a nd joi ned

d ifferent g ro u p s . We o n ly
spo ke to those i n o u r own
g ro u p . '

-
•. the previous chapter we saw how the body axes are set up and how the
" 'ee germ layers are initially specified in various vertebrate emblYos .
.lithough amphibian, fish, chick, and mouse embryos share some features
_- these stages, there are many significant differences. As we approach the
�:,ylotypic stage-the embryonic stage common to all vertebrates (see Fig.
.a l l-the similarity between vertebrate embryos becomes greater, and so
_ can consider the patterning of the vertebrate body plan in a general
:ay. By the phylotypic stage, the emblYo has undergone gastrulation, and
:..e main axial stmctures characteristic of vertebrate embryos-somites,
-:otochord, and neural tube-are well developed and already show signs of
�onal organization along both the antero-posterior and dorso-ventral
_-es. I n this chapter we shall look at how this patterning is achieved. Each
� ·on, such as an individual somite, now develops largely independently.

D uring gastrulation, the germ layers-mesoderm, endoderm, and

ecoderm-move to the positions in which they will develop into the struc­
..[Link] of the larval or adult body. The antero-posterior body axis of the
[Link] embryo emerges clearly, with the head at one end and the future
..nI at the other (Fig. 4 . 1 ). In this chapter, we focus mainly on the patterning
= rhe somitic mesoderm that forms the skeleton and muscles of the trunk,
Ag. 4.1 Rearrangement of the
presum ptive germ layers during
gastrulation and neurulation in

Amphibian blastula just LongitudiAal section of embryo after Xenopus. The mesoderm (pink and red),

' "', ' @

before gastrulation gastrulation and neurulation which is in an equatorial band at the
blastula stage, moves inside to give rise
Animal somitic mesoderm notochord
to the notochord. somites, and lateral
mesoderm (not shown). The endoderm
(yellow) moves inside to line the g ut.
D..�, The neural tube (dark blue) forms and
the ectoderm (light blue) covers the
whole em bryo. The antero-posterior
endoderm ectoderm
Vegetal axis emerges. with the head at the
anterior end.
 .
110 4: PATT E R N I N G T H E V E R T E B R AT E B O D Y P L A N I I : T H E M E S O D E R M A N D E A R LY N E RV O U S S Y S T E M

and o f the ectoderm that will develop into the future nervous system. The
phenomenon of gastrulation and the action of the organizer region are
crucial to establishing the vertebrate body plan (see Section 1 . 6). and will be
discussed in this chapter in relation to their role in the patterning pro­
cesses. A detailed discussion of the behavior of cells and tissues during
gastrulation will. however. be deferred to Chapter 8.
After gastrulation, the part of the mesoderm that comes to lie along the
dorsal side of the embryo. under the ectodel111 . gives rise to the notochord
and somites. and to a small amount of head mesoderm anterior to the
notochord. During gastrulation. cells of the dorsal-most mesoderm (the
organizer region) are internalized. and eventually form a rigid rod-like
notochord along the dorsal midline. flanked on each side by blocks of som­
ites. which are delived from cells lying on either side of the organizer
region in the marginal zone mesoderm of the blastula (see Fig. 3 . 1 8). In
vertebrates. the notochord is a transient structure. and its cells eventually
become incorporated into the vertebral colwun. During neurulation. the
neural tube is formed from the ectoderm overlying the notochord. and
develops into the brain and spinal cord. The somites. now positioned on
either side of the neural tube. give rise to the vertebrae and ribs. to the
muscles of the trunk and limbs. and also contribute to the dermis of the
skin. Neural crest cells migrate away from the neural tube and develop
into a variety of tissues that include skeletal elements of the head. the
sensory and autonomic nervous systems. and pigment cells.
Both the mesodermally derived structures along the antero-posterior axis
of the vertebrate trunk and the ectodermally derived nervous system have a
distinct antero-posterior organization. The vertebrae. for example. have
characteristic shapes in each of the four anatomical regions: cervical. thor­
acic. lumbar. and sacral. In this chapter. we first examine the development
of the somites and how they are patterned. We then deal with how their
positional identity along the antero-posterior axis is specified. In later sec­
tions. we consider the function of the vertebrate organizer and the move­
ments of gastrulation in establishing the antero-posterior organization of
the embryo and its coordination with the dorso-ventral organization that
we discussed in the previous chapter. Finally. the induction and early
patterning of the nervous system will be discussed.

Som ite formation and pattern ing

In Chapter 3. we discussed the early speCification of the mesoderm and its

patterning along the dorso-ventral axis. The fate maps of the valious ver­
tebrates (see Fig. 3.23) show that the notochord develops from the most
dorsal region of the mesoderm. and somites from a more ventral region on
either side. During gastrulation. the mesoderm moves inside the embryo.
The notochord then develops as a rod in the dorsal midline, neurulation
begins. and the prospective somitic mesoderm segments into blocks. which
will eventually flank the neural tube. The somites give rise to the body and
limb muscles, the cartilage that forms the vertebrae and ribs. and the der­
mis. Their patterning thus provides much of the antero-posterior organiza­
tion of the body. In this section. we will look at the initial formation of the
somites after gastrulation and how they are patterned.
 S O M I T E F O R M AT I O N A N D PATT E R N I N G 111

4. 1 Somites a re fo rmed i n a well-defi ned order a long the antero­

posterior axis

In the chick embryo, somite formation occurs in the mesodermal region

anterior to and just lateral to the regressing Hensen's node (see Fig. 2.15).
Between the node and the most recently formed somite. there is an unseg­
mented region-the pre-somitic mesoderm-which will segment into
about 12 somites, although the number differs in different vertebrates.
Changes in cell shape and intercellular contacts in the pre-somitic meso­
derm result in the formation of distinct blocks of cells-the somites. Som­
ices are formed in pairs, one on either side of the notochord, with each pair
of so mites forming simultaneously. Somite formation begins at the
;mterior 'head' end and proceeds in a posterior direction.
The sequence of somite formation in the unsegmented region is
unaffected by transverse cuts in the plate of pre-somitic mesoderm, sug­
O'esting that somite formation is an autonomous process and that, at this
:ime, no signal specifYing antero-posterior position or timing is involved.
EVen if a small piece of the unsegmented mesoderm is rotated through
80°, each somite still forms at the normal time, but with the sequence of
formation running in the opposite direction to normal in the inverted tis­
me (Fig. 4.2). So, before somite formation begins , a molecular pattern that
s pecifies the time of formation of each somite has already been laid down
the pre-somitic mesoderm, and the prospective identity of each somite is
jue to the temporal order in which they leave the pre-somitic mesoderm.
The cells that give rise to the somites originate in the epiblast on
either side of the anterior primitive streak and move into it at gastrulation
:0 form a population of somitogenic stem cells in and around Hensen's
:lode and later i n the tailbud. These stem cells divide, and those that remain
L l1 the stem-cell region continue to be self-renewing stem cells, but those
- hat leave the node region as it regresses form the pre-somitic mesoderm.
:\s new cells are being added to the pre-somitic mesoderm at the posterior
end of the chick embryo, somites are forming at the anterior end (Fig. 4.3),
:eft panel.

--
Pre-somitic mesoderm is inverted Somites form in reverse order

developing
neural tube

neural tube
Fig. 4.2 The temporal order of somite
formation is specified early in
embryonic development. Somite
formation in the chick proceeds in an

{D
1
antero-posterior direction. Somites

.
2 c form sequentially in the pre-somitic
X Y 3 .2

pre-somitic � 4
� region between the last-formed somite
L---''--V
-'

! �� !
5
mesoderm 1
.£ and Hensen's node, which moves
i Y x posteriorly. If the antero-posterior axis
Hensen's --­
of the pre-somitic mesoderm is inverted

..
node
through 1 80°, as shown by the arrow,
Q;
::,,:::: n �
the temporal order of somite formation
··· 1 3 is not altered-somite 6 still develops
·· · 14
before somite 10.
 112 4 : PATT E R N I N G T H E V E RT E B RATE B O DY P L A N I I : T H E M E S O D E R M A N D E A R LY N E RVO U S S Y S T E M

SI � SI SII SII
SO SO St SI
so
SO

pre-somitic
mesoderm

--:;:---- notochord

differentiation of somitic derivatives

somite
�---- ep ithelia l somite
formation
every 90 bordef"i6rmaiion anterior

UU
minutes segment specification -
. -
- -
. - - -
· - :
§
·

§ .
- - -
-
.
·
- - -
§ - -:-
·
.
-
==: = = = :
=

� ; ;�
--+---- pre-somltic mesoderm ·
·

tJ
somitic stem cells

�b-''----- Hensen's node

o hour
, hour
posterior 3 hours
30 min

9 hours

posterior

FIg. 4.3 Somite formation in the chick. As shown in the left-hand shows stages in one of the cycles of c-hairy / expression ( b l ue ) that
panel, the somites are gen erated successively from pre-somitic sweep from posterior to anterior of the pre-somitic mesoderm every
mesoderm, which is derived from somitic stem cells in the primiti ve 90 m i nutes. During each cycle, a given pre-somitic cell (red dot)
streak. As pre-somitic ce lls are released into the posterior pre-so mitic experiences distinct phases when c-hairy/ is expressed and when it is
mesoderm, a new pair of somites buds from the anterior end every not expressed. The lower right-hand panel shows the progress of a pre­
90 m i n utes. 5t, the most recently formed somite; 511, the last but one somitic cell (red dot) from the time it enters the pre-somitic mesoderm
somite formed; SO, somite in the process of fo rmation, whose until it is in corporated into a somite. 50mitic cells in the anterior
boundaries are not yet set; 5-1, 5-11, blocks of pre-somitic cells that will somites will have experienced fewer cycles of c-hairy! expression than
form somites. At format ion, each somite acquires a n tero-posterior those in posterior somites, and this could define a 'clock' that is both
pola rity, after which it can respond to the signals that pattern it along linked to somite segmentation and 'tells' the somite its position along
the antero-posterior and dorso-ventral axes. The top right-hand panel the a n tero-posterior axis.

A favored model is that the rate of somite formation is largely deter­

mined by an internal 'clock' in the pre-somitic mesoderm. This 'clock' is
represented by periodic cycles of gene expression, such as that of the gene
c-hairy 1 in the chick embryo, whose expression sweeps from the posterior
to the anterior end of the pre-somitic mesoderm with a period of 90 min­
utes, the time it takes for a pair of somites to form_ In a newly formed
somite, c-hailY 1 expression becomes restricted to the posterior end of the
- somite, where it persists, while a new wave of c-hairy J expression starts at
the tail end of the pre-somitic mesoderm (Fig. 4.3).
The connection between thses oscillations and somite formation is not
yet clear, but one of the proteins whose expression cycles is Lunatic fringe,
which potentiates activity of the Notch-Delta signaling pathway. This
pathway (Fig. 4.4) is widely involved in determining cell fate and delimiting
boundaries, and is involved in setting the somite boundaries. Mice mutant
for the Delta-Notch pathway often do not form somites, and if they do,
the somites valY in size and are different on each side of the body.
One idea is that the timing and position of somite formation is deter­
mined by the interaction of the segmentation 'clock' with the growth
fa ctor FCF-8. This is expressed in the posterior part of the pre-somitic
mesoderm and keeps it in a labile state in which it does not form somites.
As the node regresses, any given pre-somitic cell will eventually move out
 S O M I T E F O R M AT I O N A N D PATT E R N I N G 113

Fig_ 4_4 The core Notch signaling

pathway. Binding of a receptor protein
of the Notch family to its membrane­
bound ligand such as Delta or Serrate
activates intracellular signaling from the
receptor. This is thought to involve the
enzymatic cleavage of the cytoplasmic
tail of the receptor (Notch intracellular
domain) and its translocation to the
nucleus to bind and activate a
transcription factor of the CSL family.
The final outcome of this pathway is the
activation of specific genes. Notch
Cytoplasm Nucleus signaling is very versatile. In different
organisms and in different
developmental circumstances,
activation of Notch switches different
CSL genes on or off.

f the posterior area (see Fig. 4.3), away from the influence of FGF-8. It has
been proposed that when this transition occurs at a particular phase of a
n's 'clock' cycle, this starts the cellular changes leading to somite
:ormation.
Somites differentiate into particular axial structures depending on their
position along the a ntero-posterior axis. The anterior-most somites con-
ibute to the skull, those posterior to them will form cervical vertebrae,
md more posterior ones will develop as thoracic vertebrae with ribs. Speci­
fication by position has occurred before somite formation begins during
!!astrulation; if unsegmented somitic mesoderm from, for example, the
presumptive thoracic region is grafted to replace the presumptive meso­
derm of the neck region, it will still form thoracic vertebrae with ribs
Fig. 4.5). How then is the pre-somitic mesoderm patterned so that somites

Pre-somitlc mesoderm from a thoracic vertebrae-forming Skeleton of recipient embryo at 9 days

region Is transplanted from a stage 10 chick embryo shows cervical vertebrae developing
into the cervical region of a stage 8 embryo as thoracic vertebrae

Ag. 4.5 The pre-somitic mesoderm

has a positional identity before
somite formation. Pre-somitic
mesoderm that will give rise to thoracic
vertebrae is g rafted to an anterior
region of a younger embryo that will
develop illto cervical vertebrae. The
grafted mesoderm develops according
to its original position and forms ribs in
the cervical region.
 114 4 : PAT T E R N I N G T H E V E R T E B R AT E B O D Y P L A N I I : T H E M E SO D E R M A N D E A R LY N E RV O U S S Y S T E M

acquire their identity and form particular vertebrae? We consider this

below. but first we deal with the patterning of the individual somite with
respect to the different tissues it gives rise to.

4.2 The fate of somite ce lts is determined by signals from the

adjacent tissues

The somites of the vertebrate embryo give rise to major axial structures:
the cartilage cells of the embryonic axial skeleton-the vertebrae and ribs;
Fig. 4.6 Photograph o f quail·chick chimeric all the skeletal muscles. including those of the limbs; and the dermis. The
tissue. The quail cells are o n the left and the fa te maps of particular somites have been made by grafting somites from a
chick cells on the right. quail into a corresponding position in a chick emblYo at a s i milar stage of
Photograph courtesy of Nicole Le Douarin. development and following the fate of the quail cells. These can be dis­
tinguished from chick cells by their distinctive nuclei. which can be
detected in histological sections (Fig. 4.6). The lateral and medial parts of
chick somites are of different origins, and are brought together during
gastrulation; the medial portion comes from cells in the primitive streak
close to Hensen's node. whereas the lateral portion comes from more
posterior cells.
Cells located in the dorsal and lateral regions of a newly formed somite
make up the dermomyotome. which expresses the Pax3 gene, a
homeobox-containing gene of the paired family (see Box 4A, p. 1 1 7). The
dermomyotome is made up of the myotome , which gives rise to muscle
cells, and the dermatome, an epithelial sheet over the myotome which
gives rise to the dermis. Cells from the medial region of the somite form
mainly axial and back muscles. and express the muscle-specific transcrip­
tion factor MyoD and related proteins, whereas lateral cells migrate to
give rise to abdominal and limb muscles. The ventral part of the medial
somite contains sclerotome cells that express the Pax1 gene and migrate
ventrally to surround the notochord and develop into vertebrae and ribs
Fig. 4.7 The fate map of a somite in the chick (Fig. 4.7).
embryo. The ventral medial quadrant (blue) Which cells will form cartilage. muscle, or dermis is not yet determined
gives rise to the sclerotome cells. which migrate at the time of somite formation. Specification of these fates requires signals
to form the cartilage of the vertebrae. The rest
from tissues adjacent to the somite. This is clearly shown by experiments in
of the somite-the dermomyotome-forms the
which the dorso·ventral orientation of newly formed somites is inverted;
dermatome and myotome. which give rise to
the dermis and a l l the trunk m u scles.
they still develop normally. In the chick, determination of myotome occurs
respectively. The dermomyotome also gives rise within hours of somite formation, whereas the future sclerotome is only
to muscle cells that migrate into the l i m b bud. determined later. Both the neural tube and notochord produce signals that

2-day chick embryo 3-day embryo 4·day em bryo

Dorsal

il
Dorsal Dorsal
ePidermiS dermomyotome myotome dermatome
neural /
lube

Lateral
- migrating
m u scle
cells

Ventral � �
Ventral Ventral
EI sclerotome 0 dermomyotome
 S O M I T E F O R M A T I O N A N D PATTE R N I N G 115

pattern the somite and are required for its future development. If the noto­ Dorsal
chord and neural tube are removed, the cells in the somites undergo dermomyotome
apoptosis; neither vertebrae nor axial muscles develop, although limb
musculature still does.
The role of the notochord in specitying somitic cells has been shown by
experiments in the chick, in which an extra notochord is implanted to one
side of the neural tube, adjacent to the somite. This has a dramatic effect on
somite differentiation, provided the operation is carried out on unseg­
mented pre-somitic mesoderm: when the somite develops, there is an
almost complete conversion to cartilage precursors (Fig. 4.8), suggesting notochord floor plate
Ventral
[hat the notochord is an inducer of cartilage. The neural tube also has a
cartilage-inducing effect on somites, which is mediated by the most ventral
Fig. 4.8 A signal from the notochord induces
region of the tube, the floor plate (see Section 1 1 .6). There is also evidence sclerotome formation. A g raft of an a d d itional
for a signal from the lateral plate mesoderm, which is involved in specii),­ notochord to the dorsal region of a somite in a
ing the lateral part of the dermomyotome, and for a signal from the 1 O-somite embryo suppresses the formation of
overlying ectoderm (Fig. 4.9). the dermomyotome from the dorsal portion of
the somite, and induces the formation of
Some of the signals that pattern the somite have been identified. In the
sclerotome, which develops into cartilage. The
chick, both the notochord and the floor plate express the gene Sonic hedge­
graft also affects the shape of the neural tube.
hog, which encodes a secreted protein that seems to be a key molecule for
positional signaling in a number of developmental situations (see Section
1 0.5). (We met Sonic hedgehog in Chapter 3 as a gene involved in the asym­
metry of structures about the midline. There, this gene was being expressed
at a quite different stage of development and in different tissues.) One
model proposes that the signal generated by Sonic hedgehog specifies the
ventral region of the somite and is required for sclerotome development.
Signals from the dorsal neural tube and from the overlying non-neural
ectoderm would specify the dorsal region. The TGF-p family member bone
morphogenetic protein 4 (BMP-4) and secreted signaling proteins of the
\Vnt family (both of which we also encountered in Chapter 3), are good
candidates for the lateral and dorsal signals, respectively. BMP-4 specifies Dorsal
ectoderm
orsal cartilage. The Sonic hedgehog and BMP-4 signals are mutually '-.,

antagonistic. These are signals that are used over and over again during
development.
Regulation of the Pax homeobox genes in the somite by signals from the
dermomyotome
notochord and neural tube seems to be important in determining cell fate.
Ptlx3 is initially expressed in all cells that will form somites. Its expression .....
is then modulated by the BMP-4 and Wnt family proteins so that it ------ sclerotome
becomes confined to muscle precursors. It is then further downregulated �
Ventral
in cells that differentiate as the muscles of the back, but remains switched
on in the migrating presumptive muscle cells that populate the limbs. • ventralizing signal • lateralizing signal
'.lice that lack a functional Pax3 gene-Splotch mutants-lack limb
do rsa l izing signal
muscles. In the chick, Paxl has been implicated in the formation of the
-capula, a key element in the shoulder girdle, part of which is contributed
y somites. Unlike the Paxl -expressing cells of the vertebrae, which are of Fig. 4.9 A model for patterning of somite
sclerotomal origin, the blade of the scapula is formed from dermomyo­ d ifferentiation. The sclerotome is thought to be

:ome cells of chick somites 17-24, whereas the head of the scapula is specified by a diffusible signal, probably the

derived from lateral plate mesoderm. All the scapula-forming cells express Sonic hedgehog protein, from the notochord
and the noor plate of the neural tube (blue
"axl .
arrows). Signals from the dorsal neural tube and
Having seen how somites are formed and how their different regions are ectoderm (pink arrows) would specify the
Jatterned after gastrulation, we now discuss the patterning of the pre­ dermomyotome, together with Ilateral signals
;omitic mesoderm along the antero-posterior axis that gives each somite its (green arrows) from the lateral plate mesoderm.
'.[Link] character. After Johnson, R.L.: 1 994.
116 4 : PATT E R N I N G T H E V E RT E B RATE B O DY PLA N I I : T H E M E S O D E R M A N D E A R LY N E RVO U S SYST E M

4. 3 Positional identity of somites along the antero-posterior axis is

specified by Hox gene expression

The antero-posterior patterning of the mesoderm is most clearly seen in the

differences in the vertebrae, each vertebra having well-defined anatomical
characteristics depending on its location along the axis. The most anterior
vertebrae are specialized for attachment and articulation of the skull, while
the cervical vertebrae of the neck are followed by the rib-bearing thoracic
vertebrae and then those of the lumbar region which do not bear ribs, and
finally. those of the sacral and caudal regions. This antero-posterior pat­
terning is a process distinct from the patterning of individual somites in
respect of the cells that will produce muscle, cartilage, and dermis, and it
occurs earlier, while the pre-somitic mesoderm is still unsegmented. Pat­
terning of the skeleton along the body axis is based on the mesodermal
cells acquiring a positional value that reflects their position along the axis
and so determines their subsequent development. Mesodermal cells that
will form thoracic vertebrae, for example, have different positional values
from those that will form cervical vertebrae.
Patterning along the antero-posterior axis in all vertebrates involves the
expression of a set of genes that specify positional identity along the axis.
These are the Hox genes, members of the large family of hom eobox genes
that are involved in many aspects of development (Box 4A). The concept of
positional identity. or positional value, has important implications for
developmental strategy; it implies that a cell or a group of cells in the
embryo acquires a unique state related to its pOSition at a given time, and
that this determines its later development (see Section 1 . 1 3).
Homeobox genes that specify positional identity along the antero­
posterior axis were originally identified in the fruit fly Drosophila and it
turned out that related genes are also involved in patterning the vertebrate
axis. As we shall see in the final part of this chapter, patterning along the
antero-posterior axis by Hox genes and other homeobox genes is not con­
fined to mesodermal structures; the hindbrain, for example, is also d ivided
into distinct regions.
All the homeobox genes whose functions are known encode transcrip­
How a pattern o f gene expression can specify the
tion factors. The subset known as the Hox genes are the vertebrate coun­
identity of four distinct regions: W, X, V, and Z
terparts of a cluster of homeobox genes in Drosophila that is involved i n
gene 1 gene 2 gene 3 gene 4 specifying the identities o f the different segments o f the insect body. These
are discussed fully in Chapter 5. Most vertebrates have four separate clus­
ters of Hox genes that are thought to have arisen by duplications of the
-'1<i1i gene 1 genes within a cluster, and of the clusters themselves (see Box 4A). The
I gene 2 zebrafish is unusual in having six. A particular feature of Hox gene expres­

I IM
I gene 3 sion in both insects and vertebrates is that the genes in each cluster are
gene 4 expressed in a temporal and spatial order that reflects their order on the
chromosome.
W I X Iv � Tissue A simple idealized model illustrates the key features by which a Hox gene
cluster records positional identity. Consider four genes, 1 , 2, 3, and 4,
arranged along a chromosome in that order (Fig. 4.1 0). The genes are
Fig. 4.1 0 Gene activity can provide positional
expressed in a corresponding order along the antero-posterior axis of a
values. The model shows how the pattern of
tissue. Thus, gene 1 is expressed throughout the tissue with its anterior
gene expression along a tissue can specify the
distinct regions W. X. Y. and Z. For example. only boundary at the anterior end. Gene 2 has its anterior boundary in a more
gene 1 is expressed in region W but all four posterior position and expression continues posteriorly. The same prin­
genes are expressed in region Z. ciples apply to the two other genes. This pattern of expression defines four
 S O M I T E FO R M AT I O N A N D PATT E R N I N G 117

r
I
Box 4A Homeobox genes
•

Drosophila 3' 5'

Anlennapedia complex bilhorax complex

Mouse Hoxa, chromosome 6

3' r-----iE ftll1- 5'

Hoxb, chromosome 1 1

Hoxc, chromosome 1 5

Hoxd, chromosome 2

A p

The h omeobox gene family encodes a large group of trans­ Clusters of hom eotic genes involved in specifying segment
cription factors which all contain a similar DN,A.-binding region of identity were first discovered in the fruit fly Drosophila. Similar
a round 60 a m i n o acids called the homeodoma i n . The complexes of homeotic genes have been identified in many
homeodomain contains a helix-tum-helix DNA-binding motif animals. I n vertebrates. the related clusters are known as the Hox
.'Ihich is cha racteristic of many DNA-binding proteins. This complexes. and the homeoboxes of the genes are related to the
domai n is encoded by a DNA sequence of 1 80 base pairs termed Antennapedia homeobox of Drosophila. In the mouse. there are
:he homeobox. Many homeobox genes are i nvolved in four unlinked Hox complexes. designated Hoxa. Hoxb. Hoxc, and
development, and the hom eo box was originally identified i n Hoxd, (originally called Hox 1 . Hox2. Hox3. and Hox4) located on
�enes that control patterning in Drosophila development. chromosomes 6. 1 1 . 1 5. and 2, respectively (see figure).
The name 'homeobox' comes from the fact that mutations in The vertebrate clusters have arisen by d u p l ication of an
some of these genes result i n what is known as a homeotic ancestral cluster. possibly related to the single Hox cluster i n the
:ransformation, in which one structure replaces a n other. For la ncelet ( a mphioxus). a simple chordate. Thus. corresponding
�xa mple. in one homeotic mutation in Drosophila. a segment in genes within the four cl usters resemble each other closely. The
: e fly's body that does not normally bear wings is tra nsformed original cluster is thought to have formed by gene duplication
lto an adjacent segment that does bear wings. resulting in a fly and divergence and all Hox genes thus resem ble each other to
';ith four wings. some extent; the homology is most marked within t h e
contd
118 4 : PAT T E R N I N G T H E V E R T E B R AT E B O DY P L A N I I : T H E M E SO D E R M A N D E A R LY N E R V O U S S Y S T E M

Box 4A contd
homeobox and less marked in sequences outside it. Genes that vertebrates include the Pax genes. which contain a homeobox
have arisen by duplication and d ivergence within a species are typical of the Drosophila gene paired, All these genes encode
known as paralogs. and the corresponding genes in the different transcription factors with various functions in development and
clusters (e.g. Hoxa4. Hoxb4. Hoxc4. Hoxd4) are usually known as a cell differentiation.
paralogous subgroup. In the mouse there are 1 3 paralogous The h omeobox genes are the most striking example of a
g roups. widespread conservation of developmental genes in animals. It is
The Hox gene clusters and their role in development are of widely believed that there are co mmon mecha nisms underlying
ancient origi n . The mouse and frog genes are similar to each the development of all animals. This implies that if a gene is
other and to those of the fruit fly Drosophila. both in their coding i dentified as having a central rol e i n the development of one
sequences and in their order on the chromosome. In both animal. it is worth looking to see where it is present in another
Drosophila and vertebrates. these homeotic genes a re involved i n a n i mal and whether it has a similar fu nction. This strategy of
specifying regiona l identity a l o n g the antero-posterior axis. The comparing genes by sequence homology has proved extremely
Hox c lusters i n mice and in Drosophila (where they are called successful in identifying g enes involved in develop ment in
HOM genes) almost certai n l ,y arose by gene dupl ication in some vertebrates. Numerous genes first identified in Drosophila. in
common ancestor of vertebrates and insects. which the genetic basis of development i s far better understood
Most genes that contain a homeobox do not. however. belong than in any other a n i m a l . have proved to have counterparts
to a homeotic complex. nor are they involved in hom eotic involved in development i n vertebrates, I l l ustration after Coletta.
tra nsformations. Other su bfamilie s of hom eobox genes in .•
P etal.: 1 994.

distinct regions. coded for by the expression of different combinations of

genes. If the amount of gene product is varied within each expression
domain, for example by interactions between the genes. many more
regions can be specified.
The role of the Hox genes i n vertebrate axial patterning has been best
studied in the mouse, which has four Hox clusters as a result of duplication
of the Hox regions. As in all vertebrates. the Hox genes start to be expressed
in mesoderm cells at an early stage of gastrulation when they begin to leave
the primitive streak. The 'anterior' genes are expressed first. As the pos­
terior pattern develops later. clearly defined patterns of Hox gene expres­
sion are most easily seen in the mesoderm and neural tube. after somite
formation and neurulation. respectively (Fig. 4.1 1). Hox genes are expressed
as gastrulation proceeds. Typically. the pattern of expression of each gene is
characterized by a relatively sharp anterior border and. usually. a much less
well defined posterior border. Although there is considerable overlap in
expression. almost every region in the anterior part of the antero-posterior
axis is characterized by a particular set of expressed Hox genes (Fig. 4.12).
For example. the most anterior somites are characterized by expression of

Hoxb1
Ag. 4.1 1 Hox gene expression i n the
mouse embryo after neurulation. The
three panels show lateral views of 91-2
days post-coitum embryos
immu nostained with antibodies specific
for the protein products of the Hoxb 1 ,
Hoxb4, and Hoxb9 genes. The
arrowheads indicate the anterior
boundary o f expression of each gene
within the neural tube. The position of
the three genes within the Hoxb gene
complex is indicated (inset). Scale
bar = O. S mm.
Photographs courtesy of A. Gould.
 S O M I T E F O R M AT I O N A N D PAT T E R N I N G 119

Anterior Vertebral regions Posterior

,
, ,
cervical thoracic lumbar sacral caudal
r----'-
'----,1 ir------"---" "-'---, ,---'---, ,-----'-----,
Hox genes

bl
al
d3
d4
b4
a4
b5
a5
c5
c6
a6 Fig. 4.12 Hox gene expression along
a7 the antero-posterior axis of the
b9
mouse mesoderm. The a nterior border
b7
Anterior of each gene is shown by the dark red
cB
margins c9 blocks. Expression usually extends
of expression
dB backward some distance but the
a ID posterior margin of e xpression may be
d9 poorly defined. The pattern of Hox gene
dID expression could specify the identity of
dll the tissues at different pOsitions. For
all example. the pattern of expression is
dl2 qu ite different in anterior and posterior
dl3
regions of the body axis.

genes Hoxal and Hoxbl , and n o other H o x genes are expressed in this
�egion. By contrast, all the Hox genes are expressed in the most posterior
�egions. The Hox genes thus provide a code for regional identity. The most
2nterior expression of Hox genes is in the hindbrain; the more anterior
�egions of the vertebrate body-the anterior head, forebrain, and
:nidbra in-are characterized by expression of hom eo box genes such as Eme
2nd Otx, and not by Hox genes.
If we focus on just one set of Hox genes, those of the Hoxa complex, we
Sud that the most anterior border of expression in the mesoderm is that of
.- oxal in the posterior head mesodem1 , while Hoxal 1 , the most posterior
",ene in the Hoxa cluster, has its anterior border of expression in the sacral
�egion (see Fig. 4 . 1 2). This exceptional correspondence, or co-linearity,
Jetween the order of the genes on the chromosome and their order of
>patial and temporal expression along tne antero-posterior axis, is typical
of all the Hox complexes. The genes of each Hox complex are expressed in
_11 orderly sequence, with the gene lying most 3' in the cluster being
:."xpressed the earliest and in the most anterior position. The correct
:'.xpression of the Hox genes is dependent on their position in the cluster,
2nd anterior genes must be expressed before more posterior genes.
Evidence that the Hox genes are involved in controlling regional identity
_omes from comparing their patterns of expression in mouse and chick
.·ith the well-defined anatomical regions-cervical. thoracic, and so on
::'ig. 4 . 1 3) . Hox gene expression corresponds nicely with the different
:-egions. For example, even though the number of cervical vertebrae i n
:irds ( 1 4 ) i s twice that of mammals, the anterior boundaries of Hoxe5 and
::vxe6 gene expression in both chick and mouse lie Oil either side of
-' e cervical/thoracic boundary. A correspondence between Hox gene
1 20 4: PAT T E R N I N G T H E V E R T E B R AT E B O D Y P L A N I I : T H E M E S O D E R M A N D E A R LY N E RVO U S SYST E M

fig. 4.13 Patterns o f Hox gene

Hox genes
expression in the mesoderm of chick
a4 a9 alO. all
and mouse embryos. and their b4 b9 mesoderli
relation to regionalization. The c4 cS elO
c6 e8 c9 ell
d4 d9dl0 dl1dl2
posterior margins of expression of Hox T1 L1 S1 Ca1
genes in the mesoderm vary along the DDDDDDOODOO[]O DOQC]OO OO.. •••••••• vertebrae
axes. The vertebrae are derived from
s�mites
som ites, 40 of which are shown. The
vertebrae have cha racteristic shapes in vertebrae
each of the five regions: cervical (C),
a4 a9
thoracic (T), l u m bar (L), sacral (5), and b4 b7 mesoderrr
c4 e5 c6 e8 e9
caudal (Ca). Which somites form which
d8 d9 dlO dl1d12
vertebrae differs in chick and mouse. For Hox genes
example, thoracic vertebrae start at
C=cervical T=thoracic L=lumbar S=sacral Ca=caudal
somite 20 i n the chick, but at somite 1 2
i n the mouse. The transition from one
region to another corresponds with the
expression and region is also similarly conserved among vertebrates at
pattern of Hox gene expression, so
other anatomical boundaries.
Hoxc5 and Hoxc6 are expressed on
either side of the cervica l and thoracic It must be emphasized that the summary picture of Hox gene expression
vertebral transition in both chick a n d given in Fig. 4 . 1 2 does not represent a 'snapshot' of expression at a partiCll­
mouse. Simila rly, Hoxd9 and Hoxd 7 0 are lar time but rather the total overall pattern of expression. Some genes are
expressed at the transition between switched on early and are then downregulated, while others are expressed
lumbar and sacral regions. After Bu rke,
considerably later; the most posterior Hox genes such as Hoxd12 and
A.C.: 1 995.
Hoxd 1 3 , for example, are expressed in the post-anal tail, which develops
later. Moreover, this summary picture reflects the general expression of the
genes in emblyonic regions; not all Hox genes expressed in a region are
expressed in all the cells of that region. Nevertheless, the overall pattern
suggests that the combination of Hox genes provides pOSitional identity, In
the cervical region, for example, each somite, and thus each vertebra, could
be specified by a unique pattern of Hox gene expression.
As we saw in Fig. 4.5, grafting experiments show that the character of the
somites is already determined in the pre-somitic mesoderm and that
somitic tissue transplanted to other levels along the axis retains its original
identity. It has also been shown to retain its original pattern of Hox gene
expression. By contrast, transplanted lateral plate mesoderm takes on the
Hox expression pattern of its new site. Hox genes providing positional
specification in somites and lateral plate seem to be separate systems,
though similar mechanisms may be involved.
We still do not know what switches on the Hox genes in the mesoderm
during gastrulation, and how co-linear expression is ensured. In all ver-
. tebrates, the Hox genes begin to be expressed (the anterior-most genes
being expressed first) at an early stage of gastrulation, when the mesoder­
mal cells begin their gastrulation movements. If an 'early' Hoxd gene is
relocated to the 5' end of the Hoxd complex, its expression pattern then
resembles that of the neighboring Hoxd 1 3 . This shows that the structure
of the Hox complex is crucial in determining the pattern of Hox gene
expression.
One way the antero-posterior pattern of Hox gene expression might be
established in the somitic mesoderm is through linking gene activation to
the time spent in the region adjacent to the node. Such a mechanism could
provide the striking correspondence in the order of Hox genes on the
chromosome with their spatial and temporal expression. One could pro­
pose a mechanism that starts opening lip a Hox gene complex at its 3' end
in the stem-cell region, but which stops as the cells move into the more
 S O M I T E F O R M AT I O N AN D PATT E R N I N G 1 21

anterior mesoderm from which the somites segment. If the spread of acti­
vation of the Hox complex along the chromosome in the 5' direction ceases
when cells move out of the posterior region, then more of the complex
can be expressed i n posterior cells, which leave last, than in anterior cells,
\vhich leave first. Such a mechanism, which resembles that proposed for
the patterning of the proximo-distal axis of the vertebrate limb (see Section
1 0.2), is entirely speculative at present. There is evidence that the segmen­
cation clock coordinates a burst of Hox gene expression in cells that have
left the posterior pre-somitic mesoderm and are approaching the position
at which they will form a somite. This may be when the cells acquire their
positional value. As we saw earlier, an FGF signal in the posterior pre­
somitic mesoderm is likely to set the position of segment formation. Thus,
[he time of a somite's formation and its identity would be tightly co­
ordinated by the interactions between FGF, the segmentation clock and the
Hox genes.
If the Hox genes do provide positional values that determine a region's
subsequent development, then if their pattern of expression is altered, one
would expect morphological changes. This is indeed the case, as we see next.

4.4 Deletion or overexpression of Hox genes causes changes in

axial patterning

In order to see how Hox genes control patterning, either their expression
can be prevented by mutation, or they can be expressed ectopically, in
abnormal positions. Hox gene expression can be eliminated from the
developing mouse emblYo by gene knock-out techniques (Box 4B, p . 122).
Experiments along these lines have shown that the absence of a given Hox
gene affects patterning in a way that accords with the idea that Hox gene
activity provides the cells with positional identity. For example, mice in
[Link] the gene Hoxa3 has been deleted show structural defects in the
�egion of the head and thorax, where this gene is normally strongly
expressed, and tissues derived from both ectoderm and mesoderm are
affected. But the Hox genes seem to specify positional identity in rather
complex ways. There is undoubtedly some apparent redundancy between
me effects of some of the genes, and when one gene is removed, another
may serve in its place. This can make it difficult to interpret the results of
Hox gene inactivation. There is also interaction between the individual
§Senes, and this can further complicate results. For example, with the
:l1utated Hoxa3 gene described above, more posterior axial slImctures, where
-he inactivated gene is also normally expressed, show no evident defects.
This observation illustrates a general principle of Hox gene expression,
".vhich is that more posteriorly expressed Hox genes tend to inhibit the
action of the Hox genes normally expressed anterior to them; this phe­
nomenon is known as posterior dominance or posterior prevalence. This
:neans that a change in Hox gene expression usually affects the most
mterior regions in which the gene is expressed, leaving posterior struc­
:ures relatively unaffected. The effects of a Hox gene knock-out can also be
jssue specific, so that certain tissues in which a Hox gene is normally
expressed appear normal, while other tissues at the same position along
:he antero-posterior axis are affected. The apparent absence of an effect
:nay be due to redundancy, with paralogous genes from another complex
being able to compensate. For example, Hoxbl is expressed in the same
1 22 . 4 : PAT T E R N I N G T H E V E R T E B R AT E B O DY P L A N I I : TH E M E S O D E R M A N D E A R LY N E R V O U S S Y S T E M

Box 4B Gene targetin g : Insertional muta g enesis and g ene knockout

exon In order to study the function of a gene controlling development, it is h ighly desirable to
Cloned be a ble to introduce an altered gene into the an imal to see what effect this has. Mice into
gene
which an additiona l or altered gene has been introduced are known as transgenic mice.
Two main techniques for generating transgenic mice are currently in use. One is to inject
DNA conta ining the required gene directly into the n u clei of fertilized eggs; the other is to
drug·resislance alter or add a gene to the genome of embryonic stem cells (ES cells) in culture, and then
gene
to inject the genetically altered cells into the blastocyst, where they become part of the
I exon Targ eting inner cell mass.
-""''--
-='-'"" --
.... .. ""'""'-
--''''''' ve ctor
ES cells can be genetically a ltered by tech niques that can be used to create a mutation
i n a particular gene. A vector DNA molecule that is i ntroduced into an ES cell by
Targeting vector
i ntroduced
into ES cells
'------, ,-------' exon
Ta rg eti ng
vu'--
_ ...._
... _...._
.. _"""- vector

X X
homologous
recombination
exon Target gene in
chromosome

Mutated
chromosome

transfection will usually insert randomly in the genome. However, it is possible to tailor
the vector DNA in such a way that only those DNA molecules that insert at a specific
predetermined site by homologous recombination, and thus insert into and mutate a
particular gene, will be selected. The DNA to be introd u ced must contain enough
sequence hom ology with the target gene that it will insert within the target gene i n at
least a few cells in the culture, even though most insertions will be ra ndom. These
mutated ES cells can then be introduced i nto the blastocyst, producing a transgenic
mouse carrying a m utation in a known gene (see fig u re). The use of h omologous
recombination to ina ctivate a gene is known as gene knock-out when the animal is
homozygous for the inactivated gene. Many mutations produced by this technique do not
result in a knock-out, but the mutation alters gene functio n .
The enormous advantage o f using E S cells over microinjectio� methods for generating
transgenic mice is that it is possible to design a selection proced ure to isolate just those
rare cells in which the DNA has incorporated at the desired site. These ca n then be used to
generate the chimeric em bryo. The selection procedure is based on i ncluding certain
genes for drug-resistance and dru g-sensitivity in the DNA construct such that when the
DNA inserts at the correct site, only the cells with the correctly targeted DNA can be
selected.
The mutated ES cells a re then introduced into the cavity of an early blastocyst, which is
then returned to the uterus. They become incorporated into the inner cel l mass and thus
into the em bryo, where they can g ive rise to germ cells and gam etes. Once the mutant
ge�e has entered the germ line, strains of mice heterozygous for the a ltered gene can be
intercrossed to produce either viable homozygotes or homozygous letha Is, depending on
the gene i nvolved, and the effect of completely inactivating and so knocking-out the gene
can be examined.
A technique for targeting a gene knock-out to a specific tissue and/or at a particular
time in development is provided by the ere-lox system. The target gene is first 'Ioxed' by
inserting a loxP sequence of 34 base pairs on either side of the gene. These transgenic
mice are then crossed with another line of transgenic mice carrying the gene for the
chimera recombinase ere. loxP sequences are recogn ized by ere, which will excise all the DNA
(mutant gene in germ line) between the two loxP sites. I n the offspring, if ere is expressed i n all cells, then all cells will
excise the 'Ioxed' target, causing an ubiq uitous knock-out of the target gene. However, if
the gene for ere i s u nder the control of a tissue-specific promoter, so that, for example, it is
only expressed i n heart tissue, the target gene will only be excised i n heart tissue (see
figure opposite). If the ere gene is li nked to an inducible control region. it is also possible
to induce excision of the target gene at will by exposing the mice to the inducing stimulus.
 S O M I T E F O R M AT I O N A N D PATTER N I N G 1 23

A sign ificant n u m ber of knock-outs of a

target gene s i n g l e gene res ult in mice developing
without any obvious a b n o rmality or with
fewer and less severe a b normal ities than
might be expected from the normal pattern
loxP loxP of gene activity. A striking exa m p l e is that of
myoD, a key gene in m uscle differentiation.
In myoD knock-outs, the m i ce a re
anatomically norma l , although they do have
a reduced su rvival rate. This could mean
that other genes can substitute for some of
the functions of myoD.
However, it is unlikely that any gene is
without any val u e at all to a n animal. I t is
much more l i kely that there is a n altered
phenotype in these a p p a rently n o rmal
Heart-specific Cre Ubiquitous inducible Cre
anima l s, which is too subtle to be detected
under the a rtifiCial conditions of l i fe in a
laboratory. Red u n dancy is thus probably
a p pa rent rather than rea l . A further
complication is the possibility that, under
+ inducer
such c i rc u msta n ces, related g enes with
similar functions may increase their activity
to compensate for the mutated gene.

_. '

target gene expressed D larget gene deleted

::-egion as Hoxal (see Fig. 4.12), and so may be largely able to fulfill the
�nction of an absent Hoxal gene.
Loss of Hox gene function often results in homeotic transformation­
:: e conve rsion of one body part into another, for example. This is the case
_ a knock-out mutation of Hoxc8. In normal embryos, Hoxc8 is expressed in
-"e thoracic and more posterior regions of the embryo from late gastrula-
20 D onward. Mice homozygous for mutant Hoxc8 die within a few days of
nh, and have abnormalities in patterning between the seventh thoracic
ertebra and the first lumbar vertebra. The most obvious homeotic trans­
. !1Tlations are the attachment of an eight� pair of ribs to the sternum and
:...:e
: development of a 14th pair of ribs on the first lumbar vertebra
-::: . 4.14). Thus, the absence of Hoxc8 modifies the development of some of
.e cells that would normally express it. Its absence gives them a more

_(erior positional value, and they develop accordingly. In mice in which

.� dl l is mutated, anterior sacral vertebrae are transformed into lumbar
_�rebrae. Another example of the homeotic transformation of a structure
- -0 one normally anterior to it can be seen in knock-out mutations of
>:b4. In normal mice, Hoxb4 is expressed in the mesoderm that will give
�e to the axis (the second cervical vertebra), but not in that giving rise to
- _ atlas (the first cervical vertebra). In Hoxb4 knock-out mice, the axis is
_ llsformed into another atlas.
3y contrast, abnormal expression of Hox genes in anterior regions that
:mally do not express them can result in transformations of anterior
�Jctures into structures that are normally more posterior. For example,
1 24 4: PAT T E R N I N G TH E V E R T E B RATE B O DY P L A N I I : T H E M E S O D E R M A N D E A R LY N E RVO U S S Y S T E M

when Hoxa7, whose normal anterior border of expression is in the thoracic

Normal
region. is expressed throughout the whole antero-posterior axis. the basal
occipital bone of the skull is transformed into a pro-atlas structure. nor­
mally the next most posterior skeletal structure. Overexpression of Hoxa2

t�� �
in the first chick branchial arch leads to transformation of the first arch
cartilages such as Meckel's cartilage and the quadrate into second arch
cartilages such as those of the tongue skeleton.
'

000
vertebra vertebra
There are synergistic interactions bet\veen Hox genes of the same par­
alogous group. Thus. knock-outs of Hoxa3 do not affect the first cervical

000 vertebra-the atlas-or the basal occipital bone of the skull to which it
connects. even though Hoxa3 is expressed in the mesoderm that gives rise

000 to these bones. However. knock-outs of Hoxd3 (which is also expressed in

this region) cause a homeotic transformation of the atlas into the adjacent
basal occipital bone. A double knock-out of Hoxa3 and Hoxd3 results in com­
Hoxc8 deleted
plete deletion of the atlas. The complete absence of this bone in the absence
of Hox gene expression suggests that one target of Hox gene action is the
cell proliferation required to build such a structure from the somite cells.
Although we do not know how the pattern of Hox gene expression is speci­
fied. it has been found that the pattern of vertebrae can be anteriorized by a
secreted protein-growth/differentiation factor 1 1 (GDF· l 1 ). a member of
the TGF-� family. Deletion of the gene for GDF-l l in mice results in an
increase i n thoracic vertebrae from 13 to 17 or 1 8, a small increase i n the
number of lumbar vertebrae and a very reduced tail. This shows that a
secreted molecule can alter the pattern. and, as we see in the next section, a
small diffusible molecule-retinoic acid-can also do so.
lumba r
vertebrae

4 .5 Retinoic acid can alter positional values

Ag. 4.1 4 Homeotic transformation of
vertebrae due to deletion of Hoxe8 in the Retinoic acid. a derivative of vitamin A, is a small hydrophobic molecule
mouse. In loss-of-function hom ozygous with an important role in local signaling in vertebrate development. Like
m utants of Hoxc8, the first lumbar vertebra is
the steroid and thyroid hormones , it diffuses across the plasma membrane
transformed into a rib-bearing thoracic vertebra.
unaided and binds to intracellular receptors; the complex of receptor and
The mutation has resulted i n the transformation
of the lumbar vertebra into a more a nterior retinoic acid then functions as a transcription factor. A variety of experi­
structu reo ments has shown that retinoic acid can alter cells' positional values in
limb development (see Section 1 0 . 5), and it can also have effects on the
antero-posterior axis.
The developmental abnormalities induced by retinoic acid are appar·
ently the result of its interference with the normal establishment of the
Hox gene expression pattern. For example, treatment of early mouse
embryos with retinoic acid results in homeotic transformation of ver­
tebrae, both anterior and posterior transformations being induced.
depending on the time of treatment. It is likely that this effect of retinoic
acid is mediated in part by its actions on Hox gene expression, and studies
of cells in culture show that the Hox genes can be induced by retinoic acid
in a concentration-dependent manner. In Xenopus , the gene Xlhbox6 (related
to mouse Hoxb9) is normally expressed in posterior regions. but in embryos
treated with retinoic acid, the expression ofXlhbox6 extends anteriorly, and
dorsal mesoderm and anterior structures are defective. As there is evidence
for a gradient in retinoic acid along the antero-posterior axis in the mouse
embryo. it could be important in activating Hox genes in normal
antero-posterior patterning.
 N E U R A L I N D U C T I O N A N D T H E R O L E O F T H E O R GA N I Z E R 1 25

Summary

Somites are blocks ofmesodennal tissue that are formed after gastrulation.
They form seq uentially in pairs on either side of the notochord, starting at
the anterior end of the embryo. The so mites give rise to the vertebrae, to
the muscles of the trunk and limbs, and to the dermis of the skin. The pre­
somitic mesoderm is patterned along its antero-posterior dimension before
somite formation, and the first manifestation of this pattern is the expres­
sion of the Hox genes in the mesoderm. The somites are also patterned by
signals from the notochord, neural tube, and ectoderm, which induce par­
iicular regions of each somite to give rise to muscle, cartilage, or dermis.
The regional character of the mesoderm that gives rise to somites is
pecified even before the somites form. The positional identity of the som­
: es is specified by the combinatorial expression of genes of the Hox com­
, lexes along the antero-posterior axis, from the hindbrain to the postelior
end , with the order of expression of these genes along the axis correspond­
: ng to their order along the chromosome. Mutation or overexpression of a
'-lox gene results, in general, in localized defects in the anterior parts of the
:-egions in which the gene is expressed, and can cause homeotic transfonn­
ations. We can think of Hox genes as providing positional information that
pecifies the identity of a region and its later development.
Although we have concentrated on the expression of Hox genes in the
mesoderm, they are also expressed in a patterned way in the neural tube
after its induction, and we shall return to this aspect of antero-posterior
regionalization later. But next, we look at the role of the crucially import­
�l)t organizer region both in neural induction and in organizing the
a tero-posterior axis in vertebrate embryos.

eural i nduction and the role of the orga n izer

e Spemann organizer of amphibians, the shield in zebrafish, Hensen's

ode in the chick, and the equivalent node region in the mouse all have a
:rnilar global organizing function in vertebrate development. They can,
,-ith the exception of the mouse node, all induce a complete body axis if
:ransplanted to another embryo at an appropriate stage, and so are able to
rganize and coordinate both dorso-ventral and antero-posterior aspects of
. e body plan, as well as to induce neural tissue from ectoderm. In the
'
::lOuse, the anterior visceral endoderm is required for the induction of
:1[erior structures (see Section 3.6).
During gastrulation, the ectoderm lying along the dorsal midline of the
e:nbryo becomes specified as neural plate; during the subsequent stage of
-eurulation, it forms the neural tube, and then differentiates into the brain
�, d spinal cord (see Section 2.1). The brain and spinal cord must develop in
- e correct relationship with other body structures, particularly the meso­
_ermally derived structures that give rise to the skeleto-muscular system.
-:: :' us, patterning of the nervous system must be linked to that of the meso­
_ rm. In this part of the chapter, we consider the patterning of the neural
:::Ib e up to shortly after its closure, Consideration of the hindbrain will take
- to a later stage when that region becomes segmented and the neural
st cells have migrated.
The function of the organizer has been best studied in amphibians, and
1 26 4: PATT E R N I N G T H E V E RT E B RATE B O D Y P L A N I I : T H E M E S O D E R M A N D E A R LY N E RVO U S S Y S T E M

Organizer transplant from early gastrula Organizer transplant from late gastrula

Fig. 4 . 1 5 The inductive properties o f

t h e organizer change during
gastrulation. A graft of the organizer
region, from the dorsal lip of the
blastopore of an early frog gastrula to
the ventral side of another early
gastrula, results in the development of
an additional anterior axis at the site of
the graft (left panels). A graft from the
dorsal lip region of a late gastrula to an
early gastrula only induces formation of
tail structures (right panels).

we have already emphasized its role in the dorso-ventral patterning of the

mesoderm of Xenopus (see Sections 3 . 1 6 and 3 . 1 9) . We now discuss its
profound effects on the antero-posterior axis.

4.6 The vertebrate head is specified by signals d ifferent from those

for the tru n k

In amphibians, the action o f the organizer i s dramatically demonstrated in

what is classically known as primary embryonic ind u ction. If the
Spemann organizer from an early gastrula (corresponding to the dorsal lip
of the amphibian blastopore) is grafted to the ventral side of the marginal
zone of another gastrula, the result is a twinned embryo (Fig. 4 . 1 5 ) . This
second embryo can have a well-defined head and trunk region and even a
tail, but will be joined to the main embryo along the axis (see Fig. 1 . 1 0). A
variety of other treatments, such as grafting dorsal vegetal blastomeres
containing the Nieuwkoop center to the ventral side, produce a similar
result (see Fig. 3.5), but what all these treatments have i n common is that,
directly or indirectly, they result in the formation of a new Spemann organ­
izer. An important issue is whether or not there are separate organizers for
the trunk and tail. A graft of an organizer from an early gastrula induces a
complete body axis and central nervous system, while older organizers only
induce posterior structures (Fig. 4 . 1 5). Is this a change in the quantity of
inducing signals or are different signals involved?
The cells of the organizer region in Xenopus give rise during gastrulation
to anterior endoderm, prospective head mesoderm (the pre-chordal meso­
derm), and the notochord. Understanding how the organizer mesoderm
organizes the overall pattern of the antero-posterior axis is not so straight­
forward, however. In amphibians, the inductive properties of the Spemann
organizer region change throughout gastrulation. I n transplantation
experiments of the type described above, a dorsal lip taken from an early
 N EU R AL I N D U C T I O N A N D T H E ROLE O F TH E O RGAN I Z E R 1 27

Hensen's node grafted from quail embryo to chick host New axis induced in host

liost
axis

Fig. 4 . 1 6 Hensen's node can induce a

new axis in avian em bryos. When
Hensen's node from a quail embryo is
grafted to a pOSition lateral to the
primitive streak of a chick embryo at the
somites
same stage of development, a new axis
forms at the site of transplantation.
Histological examination shows that
although some of the somites of this
new axis are formed from the g raft itself
(quail tissue can easily be distinguished
from chick tissue, see Fig. 4.6), others
Quail embryo Chick embryo have been induced from host tissue that
does not normally form somites.

crastrula induces a complete additional embryo, a dorsal lip from a

mid-gastrula induces a trunk and tail but no head, while a dorsal lip from a
.ate gastrula induces only a tail (see Fig. 4.15). As gastrulation proceeds, the
ntero-posterior axis becomes specified, and so at later stages cells in the
. lastopore induce only posterior structures.
The avian equivalent to the Spemann organizer is Hensen's node, the
region at the anterior end of the primitive streak in the chick blastoderm. It
�ontributes to notochord, pre-chordal mesoderm, somites and gut endo­
derm, and can induce an additional axis , complete with somites, if grafted
_eneath a chick epiblast at the head-process stage of development (Fig.
�. 1 6). This stage is reached when elongation of the primitive streak is com­
?!ete, the notochord (head process) has started to form anteriorly, but
I-Iensen's node has not yet started to regress (see Fig. 2.14). Induction occurs
� the graft is placed quite close to the streak; it can then induce normally
:lon-axial mesoderm to form somites and other axial structures. When
5fafted to the area opaca, it can induce a whole neural axis. The node at the
_-aerior region of the primitive streak in the mouse can induce a similar
.:uplication ofthe mouse axis on transplantation, with the exception of the • Neurectoderm
rebrain, which requires the anterior vis �eral endodelm.
The organizer is a complex structure whose constitution changes during Trunk inducing
T<I trulation, and we have just seen that at early stages it induces heads,
-hereas later it induces tails. In the early gastrula of Xenopus, the vegetal
_alf of the organizer region will induce genes that are expressed more
• Head inducing
�[eriorly. The inducing regions in the early Xenopus embryo, as deter- and
trunk inducing
ined from the results of transplantation of small pieces of tissue from the
orsal marginal zone of the early gastrula, are shown in Fig. 4.17. In the • Head inducing
.ick, the node itself is specified by signaling, involving both TGF-� and
"nt family members, fro m a region in the middle of the primitive streak.
Fig. 4.1 7 Inducing regions i n the Xenopus
_ells move into the node at an early stage and, as the node regresses, the
organ [Link] of head-inducing and trunk­
_<" s divide and some leave the node region to give rise to the notochord
inducing activity in the early Xenopus gastrula, as
-d somites. determined from transplantation experiments.
A number of genes are specifically expressed in the organizer of both Adapted from Schneider, V.A., Mercola, M . :
_ !Opus and the mouse (Fig. 4.18) and, as we saw in Section 3.19, some of 1 999.
1 28 4: PATT E R N I N G T H E V E RT E B RATE B O D Y P L A N I I : T H E M ES O D E R M A N D E A R LY N E RVO U S S Y S T E M

Xenopus Mouse these genes have a role in patterning the mesoderm along the future
gastrula gastrula antero-posterior axis. The secreted proteins chordin and noggin antagonize
BMP and pattern the mesoderm. The gene goosecoid can cause a complete
secondary axis to develop when its mRNA is injected into the ventral side 0:
an early frog embryo. However, mice lacking goosecoid have a norma:
antero-posterior axis, so this gene cannot be solely involved in patterning
along this axis.
Genes in organizer region As we saw i n Section 3.6, the extra-embryonic anterior visceral endodern�

I Brachyury
I Brachyury
in the mouse embryo has a key role in specifying anterior structures.
In Xenopus, the cerberus protein, which is related to a protein expresse '
Genes
encoding
transcription
I goosecoid
I goosecoid in mouse anterior visceral endoderm, is produced in the anterior endo­
derm and may be involved in the induction of the head region. If
factors
II Pintal/avis
I Hn f·3 p misexpressed, cerberus can induce ectopic heads, albeit with one eye

I
only, with no trunk structures. I t acts by binding to and blocking the
I Xlim-1 Lim-1 action of BMPs, Wnts and nodal proteins; all three of these activities
appear to be necessary for ectopic head induction. Down-regulation o�

I Xnr-3
I Nodal Wnt signaling in anterior regions seems to be essential for forebrain

�
II
Genes and heart development. Other inhibitors of Wnts, such as' dickopf-l an
encoding I Ichordin, Xnot2,I �
noggm, Shh,
frizbee, only induce ectopic heads if BMPs are inhibited at the same time.
I
IIIII
secreted

I
proteins This shows that a complex set of signals is i nvolved in specifying head and
Cerberus- truck.
Cerberus related

4.7 The neural plate is induced in the ectoderm

Fig. 4.1 8 Genes expressed i n the Spemann
organizer region of the Xenopus gastrula, and
The induction of neural tissue from ectoderm was first indicated by the
in Hensen's node in the mouse gastrula. There
organizer transplant experiment in frogs, which is described in Fig. 4.15; in
is a similar pattern of gene activity i n the two
animals, with homologous genes being the secondary embryo that forms at the site of transplantation, a nervous
=
expressed. Shh Sonic hedgehog. system develops from the host ectoderm that would normally have formed
ventral epidermis. This suggests that neural tissue can be induced from as

yet u nspecified ectoderm by signals ema nating from the mesoderm of the
organizer region. The requirement for induction is confirmed by experi·
ments that exchange prospective neural plate ectoderm for prospective
epidermis before gastrulation; the transplanted prospective epidermi,
develops into neural tissue, and the transplanted prospective neural tissue
into epidermis (Fig. 4.1 9). This shows that the formation of the nervous
system is dependent on an inductive signal.
An enormous amount of effort in the 1 930s and 1940s was devoted to
trying to identify the signals involved in neural induction in amphibians.
Researchers were encouraged by the finding that a dead organizer region
could still induce neural tissue. It seemed to be merely a matter of hard
work to isolate the chemicals responsible. Alas, the search was fruitless, for
it appeared that an enormous variety of substances were capable of varying
degrees of neural induction. As it turned out, this was because newt ecto­
derm, the main experimental material used, seems to have a high pro­
pensity to develop into neural tissue on its own. This is not the case with
Xenopus ectoderm, although prolonged cultme of disaggregated ectodermal
cells can result in their differentiation as neural cells. Whatever the nature
of the signal, it seems that it is a molecule that can diffuse through a Nucleo­
pore filter (which prevents cell contact but allows the passage of quite large
molecules, such as proteins), and that contact lasting about 2 hours is
required for induction to occur. The molecules responsible for neural
induction have still not been definitively identified, although there are
 N E U R A L I N D U C T I O N AN D T H E R O L E OF TH E O R G A N I Z E R 1 29

Fig. 4 . 1 9 The nervous system of

Fragment of presumptive epidermis
Early gastrulas
transplanted from one gastrula to another Xenopus is induced during
gastrulation. The left panels show the
p resum
neural ptive
ectoderm normal development fate of ectoderm

O
at two different positions in the early
gastrula. The rig h t panels show the
V 0
transplantation of a piece of ventral
ectoderm, whose normal fate is to form
p resumplive
epidermis epidermis. from the ventral side of a n
early gastrula t o t h e dorsal side o f
another, where it replaces a piece of
Induction of neural tissue in
Normal fate of ectoderm at neurula stage dorsal ectoderm whose normal fate is to
transplanted fragment
form neural tissue. In its new location,

Q"'"'. "'�
neural lissue --r=--.""
- .;
the transplanted prospective epidermis
develops not as epidermis but as neural
tissue, and forms part of a normal

V
nervous system. This shows that the
ventral tissue has not yet been
epidermis determined a t the time of

p transplantation, and that neural tissue is

induced during gastrulation.

_orne very strong candidates. However, contralY to previous ideas, inducing

molecules do not act directly on the cells that will form neural tissue, but
act instead by releasing the inhibition of neural tissue formation mediated
by proteins such as the BMPs, in a manner reminiscent of the patterning of
-be early mesoderm.
A number of secreted proteins, including chordin, noggin, and cerberus,
'e BMP antagonists involved in neural induction. BMP-4 is pivotal in
:leural induction, as it inhibits cells from forming neural tissue; if BMP
-ignaling is inhibited, neural tissue develops. One such inhibitor is noggin.
_\s discussed in Section 3 . 1 9 , noggin is secreted by the Xenopus organizer
:-egion and is probably one of the signals that dorsalizes the presumptive
esoderm. Noggin may also be a neural induction factor: if it is added at
\igh concentration to isolated Xenopus blastula animal caps, neural mark­
dS are induced. Noggin also has the expression pattern and activity
�xpected of a neural inducer. Another secreted protein, chordin, is
�xpressed in the organ izer-derived mesoderm underlying the future neural
late; it too has neuralizing activity and acts by binding to BMP-4 and pre­
-enting its action. The antagonistic actions of BMP-4 and chordin in neural
.[Link] could be similar to those seen in the dorso-ventral patterning of
:. e mesoderm itself (see Section 3 . 1 9). The situation is somewhat different
the chick. There, BMP-4 does not inhibit neural deve lopment and, per­
'laps as a consequence, chordin cannot induce neural tissue in the chick,
_ " en though it is expressed in Hensen's node. But FGF is required for neural
:1duction and Wnt signals can block this.
Neural induction is thus a complex multistep process and may begin
fore the organizer acts_ Although the organizer contributes to neural
_ duction, it may not be necessary for neural tissue to form, as a neural
....Iate will still develop in frogs, birds, and mice if the organizer, or node, is
�xcised during gastrulation_ One might explain this effect by reformation
f the node, but this has not been established. More probably, the potential
o form neural tissue is already present in the ectoderm and is specified by
_arlier signals. Although noggin and chordin code for neural-inducing signals
1 30 4: PATT E R N I N G T H E V E R T E B R AT E B O D Y P L A N I I : T H E M E SO D E R M A N D E A R LY N E RVO U S SYST E M

in the zebrafish, a double knock-out of these genes does not prevent a

neural plate developing, although there are defects in the neural tube. The
early developmental signal that confers a neural fate on the ectoderm of
Xenopus most probably involves the Wnt/P-catenin signaling pathway, as
Wnt signaling can block BMP-4 transcription and synthesis of BMP-4 pro­
tein, and so result in neural development. Thus, s ignaling by this pathway
after cortical rotation (see Section 3.4) may result in neural development on
the dorsal side.

4.8 The nervous system can be patterned by signals from the mesoderm

Pieces of mesoderm taken from different positions along the antero­

posterior axis of a newt neurula and placed in the blastocoel of an early
newt embryo induce neural structures at the site of transplantation. Pos­
itional specificity in this induction is indicated by the fact that the struc­
tures that are formed correspond more or less to the original position of the
transplanted mesoderm. Pieces of anterior mesoderm induce a head with a
brain, whereas posterior pieces induce a trunk with a spinal cqrd (Fig. 4.20).
Another indication of positional specificity in induction comes from the
observation that pieces of the neural plate themselves induce similar
regional neural structures in adjacent ectoderm when transplanted
beneath the ectoderm of a gastrula. Indications that gene expression in the
mesoderm may be influencing gene expression in the ectoderm come from
the observation of coincident expression of several Hox genes in the noto­
chord and in the pre-somitic mesoderm and ectoderm at the same position
along the antero-posterior axis: Xlhboxl in Xenopus and Hoxbl in the mouse
are examples of genes that have coincident expression of this type.
Hox gene expression cannot be detected in the anterior-most neural tis­
sue of the mouse, but genes such as atx and Eme are expressed anterior to
the hindbrain. These genes encode homeodomain transcription factors and
specifY pattern in the anterior brain in a manner similar to the Hox genes
more posteriorly. The Drosophila orthodenticle gene and the mouse atx genes
are homologous and provide a good example of the conservation of gene
function during evolution. orthodentide is expressed in the posterior region
of the future Drosophila brain and mutations in the gene result in a greatly
reduced brain. In mice, atxl and atx2 are expressed in overlapping domains

Neurula anterior mesoderm grafted into early gastrula induces a head with eyes and forebrain
A

'Fig . 4.20 Induction of the nervous

system by the mesoderm is region
specific. Mesoderm from different
positions along the dorsal antero-
posterior axis of early newt neurulas p
ind uces structures specific to its region
of origin when transplanted to ventral
;=======-==�
Neurula posterior mesoderm grafted Into early gastrula induces a trunk and tall
regions of early gastrulas. Anterior
A
mesoderm induces a head with a b rain
(top panels), whereas posterior
mesoderm induces a posterior trunk
with a spinal cord ending in a tail
(bottom panels). After Mangold, 0.:
1 933. p
 N E U R A L I N D U C T I O N A N D TH E R O L E O F T H E O R G A N I Z E R 1 31

in the developing fo rebrain and hindbra i n , and mutation in Otx1 leads to

!brain abnorma lities and e pilepsy. Mice with a defective Otx gene can be
artly rescued by re placin g it with orthodcnticle. even though the seque nce
-i m i larity of the two proteins is confined to the homeodoma i n region.
Human Otx can even rescue orthodenticle mutants in Drosophila.
In the mouse, the node, usua l ly considered to be s i m i lar in fu nction and
properties to the organizer i n chick and frog, can induce only posterior
Structures. In mouse embryos lacking HNF-3 � , a trans cription factor essen­
:ial for node develo pment, there is no node but a full range of neural
:issues . I n mice, the extra-embryonic a n terior visceral endoderm is essen­
jal fo r induction of anterior structures. i n cl u ding the forebrain (see
Section 3 . 6 ) .
O n e model fo r neural ectoderm patterning suggests that qualitatively
ifferent mesodermal inducers are present at different positions along the
a ntero-pos terior axis. However. many experiments fit quite wel l with a
- i mpler two-s ignal model of neural patterning (Fig. 4.2 1 ) , in which d i ffer­
ences are due to quantitative rather than qualitative differences in the
: ndu cing s ignal. I n this mod e l , the first, activating. signal i s produced by the
whole mesoderm and induces the ectoderm to become anterior neural tis­
;ue . The second signal transforms part of this tissue so that it acquires a
c ore posterior identity. This latter. transforming, signal would be graded
in the mesoderm with the highest concentration at the posterior e n d .
Chordin and noggin a r e good candidates for the fi r s t activating signal and
:nay act by i nh i b i t i ng BMPs and Wnts (see Sections 4.6 and 4 . 7 ) . FCF and
. 'nt3A are candidates for posteriorizing transforming signals.
In the chick, there is some evidence fo r the activation/transformation
• odel outlined above. Young nodes induce a full central nervous system
,'he reas older ones only induce more posterior ones. As the node emits
sr rong posteriorizing signals, the pros pective anterior neu ral tissue must
"je protected from these signals. This i s achieved in the chick by the pro­
,pective forebrain terri tory moving anterior to the node under the cells of
-. e spreading hypoblast.

Induction of anterior neural tissue followed by a graded signal specifying posterior regions

I ectoderm I I neural tissue I ""'"

neural tubi'

¥¥¥¥ ¥
mesoderm I
Fig_ 4.21 Models of neural pattern ing
Qualitatively different neural ind ucers specify different parts by induction. Top panel: i n the two­
of the pattern
signal model, one Signal from the
mesoderm first induces anterior tissue

neural tu�J throughout the corresponding

ectoderm. A second, graded signal from
the mesoderm then specifies more
posterior regions. Bottom panel: in an
alternative model, qualitatively different
inducers are localized in t h e mesoderm.
After Kelly, O.G., et of.: 1995.
1 32 4: PATT E R N I N G T H E V E R T E B R AT E B O D Y P L A N I I : T H E M E S O D E R M A N D E A R LY N E RV O U S S Y S T E M

Rg. 4.22 Hensen's node from a chick

Pieces of chick epiblast are sandwiched between mRNAs expressed in Xenopus
embryo can induce gene expression
pieces of Xenopus animal cap tissue ectoderm in response to chick grafts
characteristic of neural tissue i n
Xenopus ectoderm. Tissues from
different parts of the primitive streak
stage of a chick epiblast are placed
between two fragments of animal cap
tissue (prospective ectoderm) from a
Xenopus blastula. The ind uction in the
Xenopus ectoderm of genes that
characterize the nervous system is
detected by looking for expression of
mRNAs for neural cell ad hesion
molecule (N-CAM) and neurogenic
factor-3 ( N F-3), which are expressed
specifically in neural tissue in stage-30
Xenopus embryos. Only transplants from
Hensen's node ind uce the expression of
these neural ma rkers in the Xenopus
ectoderm. (EF- 1 a is a common
transcription factor expressed in a l l
cells.) After Kintn er, C.R . e t 01.: 1 99 1 .
.

Mesoderm also induces neural tissue in chick and mouse embryos.

Neural tissue can be induced in the chick epiblast-in both the area pel­
lucid a and the area opaca-by grafts from the primitive streak. Inductive
activity is initially located in the anterior primitive streak and later, during
regression of Hensen's node, becomes confined to the region just anterior
to the node. By the four-somite stage inductive activity has disappeared:
the competence of the ectoderm to respond disappears at about the same
time.
Hensen's node from the chick embryo can induce neural gene expression .Ii
in amphibian (Xenopus) ectoderm (Fig. 4.22), which is of great interest as it e
suggests that there has been an evolutionary conservation of inducing sig­ s:
nals. Moreover, early nodes induce gene expression characteristic of v.

anterior amphibian neural structures, whereas older nodes induce gene eJ

expression typical of posterior structures. These results are in line with the c(
theory that the node is able to specifY different antero-posterior positional in
values, and they also confirm the essential similaIity of Hensen's node to i t�
the Spemann organizer. But the molecular mechanisms will be different, as sy
chordin and BMP are not involved in the chick. an

tht
4.9 Signals that pattern the neural plate may travel within the neural l ani
plate itself ter
frOJ
Development of neural tissue was originally thought to occur only if the
whJ
mesoderm came to lie immediately beneath the ectoderm and in contact
alsc
with it. It now appears that there is a second route by which the nervous
the
system can be patterned. The classical one is the traditional vertical, or
mid
transverse, route from the mesoderm to the overlying ectoderm, whereas
sion
the other is planar, the signal being generated within the neural plate itself
sign
and travelling within the ectodermal sheet.
side
Evidence that signals within the plate may be important comes from
ex plants of early Xenopus gastrulas that consist of dorsal mesoderm, includ­
ing the Spemann orga nizer, and ectoderm that will normally give rise to
 N E U R A L I N D U CT I O N A N D T H E R O LE O F T H E O RG A N I Z E R 1 33

Ag. 4.23 Induction and patterning of

Stage 1 0 Explants from two embryos the nervous system involves planar
Xenopus are removed and Fates of different
tissues signals originating within the
gastrulas sandwiched together
ectoderm. Explants containing both
mesoderm (pink and red) and
prospective neural ectoderm (blue) are
taken from the marginal zone of early
gastrulas at the time when the
ectoderm mesoderm begins to invaginate. Two
and such explants a re sandwiched together
neural (to prevent curling of the strips), and
tissue
cultured. In these explants, mesoderm
lies posterior and in the same plane as
the ectoderm, with only a small point of
contact, rather than underneath the
mesoderm
ectoderm and in contact with it along
�ions above the blastopore are removed its whole length as it is in the embryo.
from each gastrula (red outline) Nevertheless, the explants differentiate
into both neural tissue and mesoderm.
The ectoderm and mesoderm converge
- ural tissue after induction. Two of these explants are cultured as a sand­ and extend in the explants, and genes
such as engrailed-2 and Krox-20, which
I in such a way as to keep the strip flat and prevent it rounding up. Both
are normally expressed in region­
__ oderm and mesoderm in such explants converge and extend as they do
specific patterns within neural tissue,
_ _ �ng normal gastrulation (see Section 8 . 1 0), and the ectoderm differenti- are expressed. After Doniach, T., et 01.:
= i nto neural tissue, as s hown by the expression of the neural-specific 1 992.
� adhesion molecule N-CAM (Fig. 4.23).
e neural ectoderm of such explants shows clear spatial patterning, as
_

"'cated by the correct order of expression of genes that are normally

ized to specific regions along the antero-posterior axis of the neural
-: e . namely engrailed-2. Krox-20, Xlhboxl , and Xlhbox6. Moreover, two of the
_ __ e (engrailed-2 and Xlhboxl ) are also expressed in the mesoderm of the

-:: . nts, as they are in the mesoderm of intact embryos. Thus, the expres-
-:: of these two genes occurs independently in mesoderm and ectoderm
. out the close apposition of these two tissues that normally occurs in
_ os in vivo. The most likely explanation of these results is that signals
-"':ned to the plane of either the ectoderm or mesoderm are both
olved in patterning the ectoderm, perhaps in the form of a gradient with
.J..igh point at the site of the organizer. Within the developing nervous
.e m , the early patterning of the hindbrain region has been best studied

_ ",:ill be considered below.

::..JO d evidence for planar signals also cemes from zebrafish mutants. In
�bsence of the signal delivered by nodal protein (see Section 3 . 1 8), no
;::iar or trunk mesoderm develops yet the neural plate is normally pat­
-=d . This must be due to signals within the plate itself, emanating both
he shield and from an anterior signaling region in the ectoderm,
- "3 patterns the forebrain. At the midbrain-hindbrain boundary there is
a ignaling center that patterns the regions on either side; if grafted to
:o rebrain it induces a midbrain/hindbrain pattern locally. The
_ -rain-hindbrain boundary is defined by the posterior limit of expres­
- of the gene Otx2, which is expressed in the anterior neural tube. The
:ing molecules FGF-8 and Wnt-1 are subsequently expressed on either
:'" the border and pattern these regions.
1 34 4: PAT T E R N I N G T H E V E R T E B R AT E B O DY PLA N I I : T H E M ES O D E R M A N D E A R LY N E RV O U S S Y S T E M

4. 1 0 The hindbrain is segmented into rhombomeres by boundaries of

cell-lineage restriction

Patterning of the posterior region of the head and the hindbrain involves
segmentation of the neural tube along the antero-posterior axis. This does
not occur elsewhere along the spinal cord, where the pattern of dorsal root
ganglia and ventral motor nerves at regular intervals-one pair per
somite-is imposed by the somites. In the chick embryo, three segmented
systems can be seen in the posterior head region by 3 days of development:
the mesoderm on either side of the notochord is subdivided into somites,
the hindbrain (the rhombencephalon) is divided into eight rhombomeres,
and the lateral mesoderm has formed a series of branchia I arches which are
populated by neural crest cells (Fig. 4.24).
Development of the head in the hindbrain region involves several inter­
acting components. The neural tube gives rise both to the segmentally
arranged cranial nerves that innervate the face and peck, and to neural
crest cells, which in turn give rise both to peripheral nelves and to ske,letal
elements. In addition, the otic vesicle gives rise to the ear. TIle main skel­
etal elements of the head in this region develop from the first three bran­
520 chial arches, into which neural crest cells migrate (see Section 2.2). For
example, the first arch gives rise to the jaws, while the second arch
Fig. 4.24 The nervous system in a 3-day chick develops into the bony parts of the ear. This region of the head is a particu­
embryo. The hindbrain is divided into eight larly valuable model for studying patterning along the antero-posterior
rhombom eres (rl to r8). The positions of the
axis because of the presence of the numerous different structures ordered
c ranial nerves III to XII are shown i n green. b l to
along it.
b4 are the four branchial arches. b 1 gives rise to
=
the jaws. s somites. Adapted from Lumsden, Immediately after the neural tube of the chick embryo [Link] in this
A.: 1 99 1 . region, the future hindbrain becomes constricted at evenly spaced posi­
tions to define eight rhombomeres (see Fig_ 4.24). The cellular basis for
these constrictions is not understood but may involve differential cell div­
ision or changes i n cell shape. Whatever the underlying cause, it seems that
the boundaries between rhombomeres are barriers of cell-lineage restric­
tion; that is, once the boundaries form, cells and their descendants are
confined within a rhombomere and do not cross from one side of a bound­
ary to the other. Marking of individual cells shows that before the constric­
tions become visible, the descendants of a given labeled cell can populate
two adjacent rhombomeres. After the constrictions appear, however, des­
cendants of cells then within a rhombomere never cross the boundaries
and are thus confined to a single rhombomere (Fig. 4.25). It seems that the
cells of a rhombomere share some adhesive property that prevents them
mixing with those of adjacent rhombomeres, and this involves ephrins,
membrane-bound proteins that interact with cell-surface ephrin receptors
(Eph) on adjacent cells and which can generate bidirectional signals (Fig.
4.26). Ephrin receptors and ephrins are thought to be separately expressed
in alternating rhombomeres, thus preventing cell mixing at the boundar­
ies. This implies that cells in each rhombomere may be under the control of
the same genes, and that the rhombomere is a developmental unit. A
rhombomere is thus behaving like a compartment, which is a common
feature of insect development, as we see in Chapter 5, but seems to be rare
in vertebrates.
The idea that each rhombomere is a developmental unit is supported by
the observation that when an odd-numbered and an even-numbered rhom­
bomere from different positions along the antero-posterior axis are placed
 N E U R A L I N D U C T I O N A N D T H E R O L E OF T H E ORG A N I Z E R 1 35

Single neuroepithelial cells labeled at early and late stages and the labeled clones mapped at 2 days

Ear l y stage labeling Pattern of clones Late stage labeling

Fig. 4.25 Lineage restriction in

rhombomeres of the embryonic chick
hindbrain. Single cells are injected with
a label (rhodamine dextran) at a n early
stage (left panel) or a later stage (right
panel) of neurulation. and their
descendants are mapped 2 days later.
Cells injected before rhombomere
boundaries form give rise to some
clones that span two rhombomeres
(dark red) as well as those that do not

a, �
cross boundaries (red). Clones marked

in
after rhombomere formation never

: - _ .. . . 1
cross the boundary of the rhombomere
\ r7
that they originate in (blue). Adapted
- . . - - -
\- from Lumsden. A . : 1 99 1 .

- -X 1: to each other after a boundary between them has been surgically

. oved. a new boundary forms. No boundaries form when different
d -numbered rhombomeres are placed next to each other, however.
_ggesting that their cells have similar surface properties.
The division of the hindbrain into rhombomeres has functional signifi­
-'lce in that each has a unique identity and this determines how each will
velop. As we see below. their development is under the control of Hox
_ _:1 e expression. We first consider the behavior of the neural crest cells.

Llich originate from the neural tube and initially migrate over the rhom­
meres. They populate the branchial arches, subsequently giving rise to
':Ill ctures such as the lower jaw.

1 Neura l crest cells have positional values

Bidirectional signaling by ephrins
cranial neural crest that migrates out from the rhombomeres of the and their receptors
rsal region of the hindbrain also has a segmental arrangement. This has
n revealed by labeling chick neural crest cells in vivo and following their
- ;,gJ.-ation pathways. The patterns of crest emergence and migration cor-
-e:ate closely with the rhombomere from which the crest cells come.
- us. branchial arches 1 . 2. and 3 become populated by crest cells from
:.-::o mbomeres 2, 4, and 6. respectively. In the chick. crest cells from
-- _ombomeres 3 and 5 are apparently largely eliminated from further
_evelopment by programmed cell death. or apoptosis.
The crest cells have already acquired a positional value before they begin
migrate. When crest cells of rhombomere 4 are replaced by cells from
---: mbomere 2 taken from another emblYo, these cells enter the second
ranchial arch but develop into structures characteristic of the first arch.
_ which they would normally have migrated. This can result in the
_ -velopment of an additional lower jaw in the chick embryo. Fig. 4.26 Signaling by ephrins and their
receptors. Ephrins binding to their receptors
(Eph) can generate a bidirectional signal. in
which the ephrin itself also generates a signal i n
t h e cell that carries it.
1 36 4 : PATT E R N I N G T H E V E RT E B RATE B O D Y P L A N I I : T H E M E S O D E R M A N D E A R LY N E RVO U S S Y S T E M

4. 1 2 Hox genes provide positional identity i n the hind brain region

Hox gene expression provides a possible molecular basis for the positional
identity of both the rhombomeres and the neural crest. Hox genes are
expressed in the mouse emblYo hindbrain in a well defined pattern, which
closely correlates with the segmental pattern (Fig. 4.27). For example,
Hoxb3 has its most anterior region of expression at the border of rhom­
bomeres 4 and 5, while Hoxb2 has its anterior border at the border of
rhombomeres 2 and 3 (Fig. 4.28). In general, the paralogous genes of the
different Hox complexes have similar patterns of expression. It is clear that
the three paralogous groups involved have different anterior margins of
expression, paralog 1 (Le. Hoxal , Hoxbl , etc.) being most anterior, followed
by paralogs 2 and 3. The pattern of Hox gene expression in the ectoderm
and branchial arches at a particular position along the antero-posterior
axis is similar to that in the neural tube and neural crest, and it may be that
the crest cells induce their positional values in the overlying ectoderm
during their migration.
Transplantation of rhombomeres from an anterior to a more posterior
position alters the pattern of Hox gene expression so it becomes the same
as that normally expressed at the new location. The signals responsible for
this reprogramming originate from the neural tube itself and not the sur­
rounding tissues. Studies of the control of Hox gene expression at the
molecular level have provided some indication as to how their pattern of
expression is controlled. For example, although the Hoxb2 gene is expressed
in the three contiguous rhombomeres 3, 4, and 5, its expression in rhom­
bomeres 3 and 5 is controlled quite independently from its expression in
rhombomere 4. The regulatolY regions of the Hoxb2 gene carlY two separate

Four paralogous groups of Hox genes in three Hox complexes

Paralogs
I
2 3 4
Hoxa

Hoxb

Hoxd

Expression of paralogs 1 to 4 in the hindbrain (rl to ra), migrating neural crest cells, and the
surface ectoderm of the branchial arches (bl to b4)

Fig. 4.27 Expression o f H o x genes in

rhombomeres
the branchial region of the head. The
expression of genes of three paralogous
Hox complexes i n the hindbrain
(rhombomeres r1 to r8). neural crest. .
} neural crest cells
branchial arches ( b 1 to b4), and su rface
ectoderm is shown. Hoxa l and Hoxd l
are not expressed at this stage. The
arrows indicate the migration of neural surface ectoderm
crest cells into the branchial arches.
Note the absence of neural crest bl b2 b3 b4 branchial arches
migration from r3 and r5. After
Anterior Posterior
Krumla uf, R.: 1 993.
 N E U R A L I N D U C T I O N A N D T H E R O L E OF T H E O R G A N I Z E R 1 37

2hancer elements which regulate its expression in these three rhom­

- :neres. Expression in rhombomeres 3 and 5 is controlled through one of
.ese enhancers, while expression in rhombomere 4 is controlled through
- other (see Fig. 4.28). In rhombomeres 3 and 5, Hoxb2 is activated in part
he zinc finger transcription factor encoded by the gene Krox-2 0, which is
ressed in these rhombomeres but not in rhombomere 4. There are bind­
-� sites for the Krox-20 protein in the enhancer element that activates
ression of Hoxb2 in rhombomeres 3 and 5. How the spatially organized
_ _ ression of transcription factors such as Krox-20 is achieved is not yet
� wn. In the case of Hoxb4, whose anterior boundary of expression is at
- - ombomere 6, there is evidence that it is induced and localized by signal-
-'S both w:ithin the neural tube and from adjacent somites. There is also a
-;e for retinoic acid in patterning the hindbrain, as an absence of retinoic
_ �d leads to loss of hindbrain rhombomeres .
. n experiment show:ing that Hox genes determine cell behavior in the
�ombomeres comes from the misexpression of Hoxb l in an anterior
mbomere. Motor axons from a pair of rhombomeres project to a single
"uchial arch; rhombomere r2 projects to the first branchial arch while r4
-�jects to the second arch. Hoxbl is expressed in r4 but not in r2. If Hoxbl i s
"': sexpressed in an r2 rhombomere, this then sends axons to the second

-ene knock-outs in mice have also shown that the Hox genes are
.::'.-olved in patterning of the hindbrain region, though the results are not
�'ays easy to interpret; knock-out of a particular Hox gene can affect
=erent populations of neural crest cells in the same animal, such as those
..:a will form neurons and those that will form skeletal structures. Knock­
_: of the Hoxa2 gene, for example, results in skeletal defects in that region
- -he head corresponding to the nonnal domain of expression of the gene
.:jch extends from rhombomere 3 backward. Segmentation itself is not
-'<"f eted, but the skeletal elements in the second branchial arch, all of
Fig . 4.28 Gene expression in the hindbrain .
• :ich come from neural crest cells derived from rhombomere 4, are
The photograph shows a coronal section
• orma!. The usual elements, such as the stapes of the inner ear, are
through the hindbrain of a 9'12 days post-coitum
. :ent, but instead SOme of the skeletal elements normally formed by the mouse embryo, which is transgenic for two
-'- .{ arch develop, such as Meckel's cartilage, which is a precursor element
reporter constructs. The first construct contains
- we lower jaw. Thus, suppression of Hoxa2 causes a partial homeotic the lacZ gene under the control of an enhancer
,... sformation of one segment into another.
....m from Hoxb2, which directs expression in
:hese obselvations, together with those described earlier in this chapter, rhombomeres 3 and 5 (revealed a s blue
staining). The second construct contains a n
- ow that during gastrulation the cells of vertebrates acquire positional
alkaline phosphatase gene under the control of
�ues along the antero-posterior axis, and that this positional identity i s an enha ncer from Hoxb 1 , which directs
..::co ded b y the genes of the Hox complexes. Many of the anatomical differ­ expression in rhombomere 4 (revea led as brown
-ces between vertebrates are probably simply due to differences in the staini ng). A similar enhancer directing
_. sequent targets of Hox gene actions, which result in the emergence of expression i n rhombomere 4 exists for Hoxb2.
-=erent but homologous skeletal structures-the mammalian jaw or the Anterior is uppermost, and the positions of five
of the rhom bomeres are ind icated (r2 to r6).
-d's beak, for example. These principles w:ill be elaborated further in
=
Scale bar 0.1 mm .
...:.a. pter 5, which deals with the development of the fruit fly Drosophila, the
Photograph courtesy ofJ. Sharpe. from
�:mal in which Hox-like genes were first identified, and the concept of Lumsden. A.. Krumlauf, R.: 1 996 .
..:: eir role in regional specification first formulated.
1 38 4: PATT E R N I N G T H E V E R T E B R AT E B O DY P L A N I I : T H E M ES O D E R M A N D E A R LY N E R V O U S S Y S T E M

4.1 3 The embryo is patterned by the neurula stage into organ-forming

Dorsal view, neurula stage Xenopus embryo
reg ions that can sti l l reg u l ate
Anterior
At the neurula stage, the body plan has been established and the regions 0:
the embryo that will form limbs, eyes. heart. and other organs have become­
determined (Fig. 4.29).
This contrasts sharply with the blastula stage, at which time suc�
determination has not yet occurred. The basic vertebrate phylotypic bod.
plan is thus established during gastrulation. But although the positions 0 :­

various organs are fixed. there is no overt sign yet of differentiatior.

Numerous grafting experiments have shown that the potential to form
given organ is now confined to specific regions. Each of these regions has
however. considerable capacity for regulation. so that ifpart of the region i ,
Posterior removed a normal structure can still form. For example, the region of the
neurula that will form a forelimb will, when transplanted to a differen:
Fig. 4.29 The Xenopus embryo has become region. still develop into a limb. If part of a future limb region is removed
regionalized by the neurul-a stage. Various the remaining part can still regulate to develop a normal limb.
organs such as limbs. heart. and eyes will
develop from specific regions (red) of the
neurula after gastrulation is complete. Some of Summary
these regions, like the limb buds, are already
determined at this stage and will not form any Patterning along both the antero-posterior and dorso-ventral axes is close\.
other structure. The boundaries of the regions rela ted to the action of the Spemann organizer and its morphogenesis dur·
are not sharply defined, however, and within ing gastrulation. When grafted to the ventral side of an early gastrula, thE:
each region or ' field' considerable regulation is
Spemann organizer induces both a new dorso-ventral axis and a new
still possible.
antero-posterior axis, with the development of a second twinned embryo
In chick development, Hensen's node serves a function similar to that 01
the Spemann organizer. and it too can specify a new antero-posterior axis
In the mouse. the anterior visceral endoderm is necessary for forebrair_
induction.
The vertebrate nervous system. which forms from the neural plate, i,
induced both by early signals and by cells that give rise to the mesodenT'
and come to lie beneath prospective neural plate ectodern1 during gastrula­
tion. For those molecules that can induce neural tissue in Xenopus. such a,
the noggin protein. induction of neural differentiation is due to inhibitior.
of BMP activity. Patterning of the neural plate, which includes the move­
ment of signals within the plate itself, can partly be accounted for by a t\vo­
signal model: the ectoderm is first specified as anterior neural tissue and
then a second set of signals, possibly graded. s pecifies more posterior
structures.
The hindbrain is segmented into rhombomeres. with the cells of each
rhombomere respecting their boundaries. Neural crest cells from the hind·
brain populate specific mesodermal regions such as the branchial arches in
a position-dependent fashion. A Hox gene code provides positional values
for the rhombomeres and neural crest cells of the hindbrain region, while
other genes specify more anterior regions. By the neurula stage. after
gastrulation, the body plan has been established.
 S U M M A RY 1 39

S U M M A R Y TO C H A P T E R 4

-'"I e germ layers specified during blastula formation become patterned along
�e antero-posterior and dorso -ventral axes during gastrulation. The Spemann
:- gan izer in am phibians and its counterpa rt, Hensen's node, in chick and
�ouse em bryos, are involved in the initial patterning that underlies the
-egi onalization of the antero-po sterior axis. Positional identity of cells along
:i'le antero-posterior axis is encoded by the com binato rial expression of genes
-: the four Hox com plexes, which provide a code for regional identity. There is
A>th spatial and tem poral co-linearity between the order of Hox genes on the
-:'1 ro mosomes and the order in which they are expressed along the antero­
:A>sterior axis of the emb ryo from the hi ndbrain backwards. Inactivation or
. erexpression of Hox genes can lead both to localized abno rmalities and to
lO meotic tra n sformations of one 'segment' of the axis into another, indicati ng
:n at these genes are crucial in s pecifying regional identity. At the end of gastru­
.;:"ion, the basic body plan has been laid down and the nervous system
duced . Specific regions of each somite give rise to cartilage, muscle, and
::ermis, a n d these regions are specified by signals from the notochord, neural
:_ be, and epidermis. I n duction and patterni n g of the nervous system i nvolves
JOth signals in the ea rly embryo and from the underlying mesoderm, and
:; a nar signals arising within the neural plate itself. In the hind brai n , Hox gene
_ pression provides positional values for both neural tissue and neural crest
�el ls .

Summa!)': p'attern ing of the vertebrate axial body plan

gastrulation and Spemann organizer activity

�
the Hox gene complexes are expressed along the antero-posterior axis

�
Hox gene expression establishes positional identity for mesoderm, endoderm, and ectoderm

mesoderm develops into notochord, early signals and mesoderm induce

somites, and lateral plate mesoderm neural plate from ectoderm

�
somites receive signals from notochord,
�
mesoderm a nd planar ectodermal signals
neural tube, and ectoderm give regional identity to neural tube

� �
rhombomeres and neural crest in the
somite develops into sclerotome
and dermomyotome hindbrain are characterized by regional
patterns of Hox gene expression