R630 Dispatch

Developmental evolution: This side of paradise

Anthony Graham and Imelda McGonnell
It has long been appreciated that the evolution of
snakes involved the loss of limbs and axis elongation,
but their developmental basis has been obscure. It has
now been shown that alterations in the deployment of
Hox genes and an early block in the formation of
hindlimb primordia underpin these modifications.
Address: Molecular Neurobiology Group, 4
th
Floor Hodgkin Building,
Kings College London, Guys Campus, London SE1 9RT, UK.
E-mail: [email protected]
Current Biology 1999, 9:R630R632
http://biomednet.com/elecref/09609822009R0630
0960-9822/99/$ see front matter 1999 Elsevier Science Ltd.
All rights reserved.
Snakes are descended from lizards, and their evolution
clearly involved radical alterations of the body plan, the
most readily apparent being the loss of limbs [1]. Interest-
ingly, it seems that this was a two-step process, with the
forelimbs being lost first followed by the hindlimbs
(Figure 1). Support for this viewpoint primarily derives
from the cretaceous fossil, Pachyrhachis [2]. The signifi-
cant point about this fossil is that it is in fact a snake,
displaying definitive cranial snake characteristics, an
elongated body and the absence of forelimbs, but it also
has hindlimbs. This two-step loss of limbs is also reflected
in the anatomy of extant snakes: while the advanced
snakes or Colubroidae, which include the vipers, are com-
pletely limbless, the relatively primitive Booidae, which
include the pythons, do have hindlimb rudiments, with a
pelvic girdle and an attached femur being evident.
The evolution of limblessness obviously necessitated
major changes in the developmental programme under-
lying limb formation, but such modifications have not
been identified until now. Marty Cohn and Cheryll
Tickle [3] have exploited the fact that pythons represent a
transitional stage in snake evolution to investigate the
developmental basis of limblessness. In what must be the
first detailed study of the mechanisms and molecules
involved in python limb development, they have been
able to pinpoint alterations in the developmental pro-
gramme that could underlie the evolution of snakes.
In their study of python limb development Cohn and
Tickle [3] found, firstly, that the forelimb bud never
initiates and that this seems to be the result of evolution-
ary alterations in the expression of the Hox genes, which
normally delineate the sites of limb bud formation. Sec-
ondly, they found that, while the hindlimb bud does initi-
ate, it fails to establish an apical ectodermal ridge (AER),
the region of the bud which promotes outgrowth. And
finally, they have shown how one of the other characteris-
tics of snakes, the elongation of the thorax, also seems to
relate to alterations in Hox expression.
Through a large body of work, primarily on avian embryos,
we now know the critical steps, and molecules, involved in
Figure 1
Squamates
Snakes (Ophidia)
Modern snakes
Current Biology
Advanced snakes
Mosasauroidea
Forelimbs, hindlimbs,
regionalized axial skeleton
Pachyrhachis
Complete hindlimbs
only
Booidea
Hindlimb
rudiments only
Colubroidea
Limbless
The key stages in the evolution of snakes. This tree shows the
evolutionary relationships between the lizard-like ancestors of snakes,
which have complete forelimbs and hindlimbs, the fossil snake,
Pachyrhachis, which has a hindlimb, the extant snakes the Booidea,
which have only hindlimb rudiments, and the Colubroidea, which are
limbless. The hindlimb of Pachyrhachis is shown, as are the hindlimb
rudiments of a python and the corresponding, but limbless, caudal end
of the thorax of a colubrid snake.
the formation of amniote limbs (reviewed in [4]; sum-
marised in Figure 2). The first step in making a limb is the
specification of the position along the axis at which it will
arise. Previous work has shown that the differential expres-
sion of Hox genes along the lateral plate mesoderm defines
the position of the forelimb- and hindlimb-forming territo-
ries and the interlimb flank [5]. Once the position is speci-
fied, the mesenchyme of the limb territory sends signals to
the overlying ectoderm that promote morphological and
molecular changes in a narrow band of cells lying between
the dorsal and ventral limb ectoderm. This band of cells
form the AER, the main function of which is to direct limb
bud outgrowth through secretion of members of the fibro-
blast growth factor (FGF) family of signalling molecules,
which act on the underlying mesenchyme.
It is during the AER-directed outgrowth that pattern is
assigned to the elements of the limb, and this is through
the action of the zone of polarising activity, or ZPA,
which resides in the posterior mesenchyme of the limb
bud. This entity was first defined in transplantation exper-
iments; it was found that, when posterior limb mes-
enchyme was grafted to the anterior of the limb bud,
subsequent limb development was altered such that a
duplicated mirror image set of digits formed. More
recently, it has been demonstrated that the signalling mol-
ecule Sonic hedgehog (Shh) is responsible for the pattern-
ing properties of the ZPA. Ectopic application of Shh to
the anterior limb bud causes pattern duplications, and loss
of Shh function severely perturbs limb patterning.
Using these well-established facts, Cohn and Tickle [3]
investigated the basis of limblessness in python embryos.
The first notable difference they found between python
and chick embryos of similar stages is a radical alteration
in the expression of the genes involved in limb bud speci-
fication the Hox genes [5]. They analysed the expres-
sion domains of HoxB5, HoxC6 and HoxC8, whose anterior
limits in the lateral plate mesoderm of tetrapod and fish
lies at the forelimb bud/pectoral fin level. Yet in pythons
these genes are expressed throughout the lateral plate,
extending anteriorly to the base of the head [3]. The
normal Hox expression patterns indicative of the forelimb
field are thus never established in snakes, and conse-
quently a forelimb bud never forms.
By contrast, the specification of the hindlimb position and
the initial bud formation does seem to be normal in
pythons, yet the bud fails to grow out from the body [3]. To
begin to understand the basis of this developmental block,
Cohn and Tickle [3] looked at the development of the
structure that promotes limb bud outgrowth the AER.
Notably, an AER is not morphologically evident in python
embryos. This can in itself be misleading, however, as there
are species of direct developing frogs that lack an obvious
AER but that nevertheless express genes associated with
AER function in the appropriate domain [6]. Cohn and
Tickle therefore looked further for the expression of a
number of marker genes known to be indicative of the AER
in other species Dlx, FGF-2 and msx-1/2.
In each case, they found that, while these genes were
expressed elsewhere in the python embryo, they did not
highlight a presumptive AER at the distal end of the
hindlimb bud. These observations showed that this
Dispatch R631
Figure 2
Steps in the development of the limb. (a) The normal expression
patterns of HoxC6 and HoxC8 in the lateral plate mesoderm of chick
(left) and python (right); importantly, these genes are not expressed in
the limb fields. In python embryos, the expression domains for these
Hox genes extend a lot further along the lateral plate mesoderm.
(b) The AER, at the distal tip of the limb bud, signals to the underlying
mesenchyme. The site of the ZPA is also shown; it signals to, and
receives signals from, the AER. Axes: Pr, proximal; D, distal; A,
anterior; Po, posterior. (c) The AER forms in the ectoderm at the
interface between the dorsal and ventral territories of the limb bud.
Axes: Pr, proximal; D, distal; Do, dorsal; V, ventral.
H
o
x
C
8
H
o
x
C
6
Flank
Hindlimb
Python
Forelimb
Flank
Hindlimb
Chick
(a)
ZPA
AER
AER
A
Po
D Pr
Do
V
D Pr
Dorsal
Ventral
(b) (c)
Current Biology
H
o
x
C
8
H
o
x
C
6
R632 Current Biology, Vol 9 No 17
structure was absent from the python hindlimb buds [3].
The fact that FGF-2 was not expressed in this region was
particularly interesting, as it is the FGFs produced in the
AER that promote outgrowth. To determine whether the
outgrowth could still be elicited by FGFs in python
hindlimb buds, Cohn and Tickle grafted a bead soaked
in FGF-2 protein to the limb buds. They found that, a
day after application of the bead, the limbs displayed an
increase in outgrowth. So it seems that it is the lack of an
AER and its associated FGFs that prevents outgrowth of
the python hindlimb.
So why does the python hindlimb bud fail to form an
AER? One factor that is important in establishing the
AER is the interaction between the dorsal and ventral
ectodermal compartments of the presumptive limb field.
Consequently, it could be that the absence of an AER
results from a failure of dorsoventral specification. To
ascertain if this is true, Cohn and Tickle [3] looked at the
expression of engrailed and lmx-1, markers of ventral and
dorsal territories, respectively. They found, however, that
the expression domains of these genes showed spatial
restrictions engrailed ventrally and lmx-1 dorsally
similar to those seen in embryos where AERs are formed.
They therefore looked for defects in the other component
that is critical for forming an AER, the mesenchyme.
Firstly, they grafted python mesenchyme into a chick
limb to determine if it was able to induce an AER in what
was known to be competent tissue indeed it could.
Moreover, they found that, while there was no Shh expres-
sion in the posterior mesenchyme of the python hindlimb
bud, when this mesenchyme was grafted to the same
position in a chick bud, it then expressed Shh and could
act as a ZPA. But they also found that the ability to gener-
ate polarising tissue from python mesenchyme was in fact
much more widespread than has been observed in species
that develop proper limbs. Normally, the anterior mes-
enchyme of the chick limb bud cannot form polarising
tissue even after transplantation, but grafting anterior
python mesenchyme into the anterior of a chick limb bud
also resulted in digit duplications. Cohn and Tickle [3]
therefore suggest that the primary defect in the python
limb lies in the mesenchyme, and that this results in the
lack of an AER. The fact that the mesenchyme of the
python hindlimb lacks Shh expression, and that this bud
only generates a femur, is interesting, as this resembles
the phenotype of the hindlimbs of embryonic mice with
mutations in the Shh gene [7].
The evolution of snakes from lizards involved more than
the loss of the limbs, and in fact another marked difference
between these groups of animals is an alteration of axial
patterning. In snakes, much of the axial skeleton is now
thoracic, with ribs being found on all vertebrae anterior to
the hindlimb level. Interestingly, this morphological
change also seems to tie in with modifications of the Hox
expression domains, particularly of HoxC6 and HoxC8. In
most tetrapods, these two Hox genes are associated with
thoracic development, having an anterior boundary at the
end of the cervical region and a posterior boundary at the
start of the lumbar region. But in pythons these genes are
expressed from the hindlimb level all along the vertebral
column to the cranial region, and correspondingly all these
vertebrae now bear ribs. This transformation is not total,
however, and the anterior vertebrae do still preserve some
aspects of their previous cervical identity.
The Cohn and Tickle paper [3] is important as it gives us
the first insight into the developmental mechanisms that
underlie the alterations of the body plan that must have
occurred in the evolution of snakes from lizards. Further-
more, it also predicts that the evolution of the colubroid
snakes, such as vipers, in which there is absolutely no
hindlimb rudiment and the axial skeleton is even more
homogenised, is likely to have involved even more
caudal expansion of Hox gene expression. This paper is
also interesting within the burgeoning evo-devo field.
This is one of the few studies within this sphere to use
embryological and molecular approaches to define the
changes in developmental gene expression that underlie
evolutionary variation.
References
1. Caroll R: Vertebrate Paleontology and Evolution. New York:
Freeman; 1988.
2. Caldwell MW, Lee MSY: A snake with legs from the marine
cretaceous of the middle east. Nature 1997, 386:705-709.
3. Cohn MJ, Tickle C: Developmental basis of limblessness and axial
patterning in snakes. Nature 1999, 399:474-479.
4. Johnson RL, Tabin CJ: Molecular models for vertebrate limb
development. Cell 1997, 90:979-990.
5. Cohn MJ, Patel K, Krumlauf R, Wilkinson DG, Clark JDW, Tickle C:
Hox9 genes and vertebrate limb specification. Nature 1997,
387:97-101.
6. Fang H, Elinson RP: Patterns of distal- less gene expression and
inductive interactions in the head of the direct developing frog
Eleutherodactylus coqui. Dev Biol 1996, 179:160-172.
7. Chiang C, Litingtung Y, Lee E, Young KE, Cordoen JL, Westphal H,
Beachy PA: Cyclopia and axial patterning in mice lacking sonic
hedgehog gene function. Nature 1996, 383:407-413.