Main Page Dynamic Development
The Foundations of Developmental Biology
From Sperm and Egg to Embryo
Genetic Regulation of Development
Organizing the Multicellular Embryo
Generating Cell Diversity
Dynamic Development at a Glance
The Developmental Biology Journal Club
Developmental Biology Tutorial
Limb development in Drosophila
by Dr. William Brook
Department of Biochemistry and Molecular Biology
University of Calgary
In Drosophila, limb development presents different
problems for the control of growth and cell-patterning than embryonic development. Most of
the patterning of the major embryonic axes during early Drosophila development
takes place in the pre-blastoderm embryo. In this syncytial environment patterning is
controlled by concentration gradients established by the diffusion of maternally localized
transcription factors and RNA binding proteins. Limb are cellular, so different mechanisms
are required to mediate limb development. However, the specification of different cellular
fates is still controlled by gradients of regulatory proteins, in the case of the limb,
the proteins are secreted signaling molecules instead of transcription factors.
Limbs develop from imaginal discs
Drosophila limbs (legs, wings, halteres, antennae, mouth
parts) derive from structures called imaginal discs. Imaginal discs begin as small
clusters of cells which are set aside during embryogenesis. These cells proliferate during
larval development to form folded, single layer, epithelial sacs. Each imaginal disc gives
rise to a separate structure so there is a disc for each leg, wing, haltere and antenna.
The cells in the disc cease dividing just prior to differentiation which begins at the
time of pupation. As the discs begin to differentiate, they evert (or unfold) and fuse to
form a continuous adult head and thoracic cuticle.
(See Browder et al., 1991, Figure 14.8; Gilbert, 1997,
Fig. 19.14; Kalthoff, 1996, Fig. 6.20; Wolpert et al., 1998, Fig. 2.34)
Establishment of discs
The cells giving rise to the thoracic imaginal disc primordia are
initially part of the embryonic ectoderm. The primordia are established straddling the
parasegmental boundaries in each of the three thoracic segments. They are specified in
response to two different secreted signals: wingless (wg), a segment polarity gene
and member of the Wnt family of secreted factors and decapentaplegic (dpp) a Drosophila
TGF-beta homologue. Wg is expressed as a stripe just anterior to the parasegmental
boundary. Dpp is expressed in a lateral stripe running perpendicular to the cells
expressing Wg. The cells in the vicinity of the intersection between the wg and dpp
stripes are exposed to both secreted signals and become specified as imaginal disc cells.
This was demonstrated by correlating the spatial pattern of genes expressed in imaginal
discs (such as the gene Distal-less) with the expression patterns of wg and dpp
.The effects of wg and dpp mutations on the formation of imaginal discs and
the expression of disc specific genes were also assessed.
Figure 1 (After Cohen, 1993)
An overview of spatial patterning of the wing
(Please refer to Blair, 1995; Brook et al., 1996) for reviews of
the following material.
The wing disc primordia begin as small clusters of ~40 cells. By the
time the discs are ready to differentiate the cells in the disc number approximately
50,000. The complex patterns of cellular differentiation seen in the wing are not
determined in the cells of the primordia, but rather the patterns of growth and
differentiation are specified during the course of disc development.
The small number of disc founder cells (~40 cells for the wing disc)
give rise to an imaginal disc that consists of ~10000 to ~50000 cells. There are several
questions we would like to answer about how the founder cells give rise to the mature
disc: Do all cells contribute equally? Are specific founder cells committed to give rise
to specific adult structures? Does growth occur in specific parts of the disc at different
times? (ie. like the zone of proliferating cells in the vertebrate limb bud).
One way to answer these questions would be to mark cells in the limb at
different times in development and observe the clones of cells that are formed by the
marked cells, the structures the clones form and the way the clones are distributed are
distributed in tissues. In Drosophila, a technique called mitotic recombination
(refer to Figure 4) is used to induce a genetically marked cell and follow the fate of the
clone of cells that derive from the marked cell. It involves generating patches of
homozygous mutant tissue in the limb primordia of animals heterozygous for an innocuous
cell marker mutation, and subsequently assessing the size, shape and distribution of
clones in adult tissues. Typically, mutations that alter bristle shape or cuticle colour
are useful cell markers for observing clones in adult cuticle, while markers such as GFP
and beta-galactosidase are often used when it is desirable to observe clones in growing
Clonal analysis of limb development has lead to many conclusions about
how the founder cells in the limb primordia give rise the mature imaginal discs. Cells in
a clone stay together, implying that there is no cell mixing in development. Growth is
constant: there is an inverse linear relationship between the log of clone size and the
time of clone induction, and a direct linear relationship between the log of clone
frequency and time of clone induction. These two relationships imply that cells in the
clone are growing continually during the development of the limb and that the number of
cells in the primordia is also continually increasing. Growth is uniform: all cells
contribute equally and all parts of the disc are probably growing at all times
The spatial distribution of clones induced at different times is random
- implying that there is no focus for growth at any given stage and that individual cells
are not determined to form specific structures until just prior to metamorphosis.
Distribution of clone sizes at different times is uniform, implying that all cells give
rise to a more or less equal proportion of the limb. One important growth restriction is
observed: cells are confined to territories on the disc called "compartments."
Compartments are observed as boundaries that cannot be crossed by
clones of cells. The first "compartmentalization" that can be observed is the
anterior-posterior compartment restriction, established at the cellular blastoderm stage.
The boundary is observed as a straight line that runs between veins L3 and L4. Remarkably,
there is no apparent morphological structure associated with the A/P boundary. The other
major boundary that occurs in wing development is the dorsal-ventral boundary formed
during the second larval instar. This boundary co-relates with the wing margin between the
dorsal and ventral surfaces of the adult wing. Clones induced prior to the middle of
second instar are not restricted and can cross the boundary freely. Clones induced after
the middle of second instar are restricted to either the dorsal or ventral surface of the
Compartmentalization of the wing indicates very early choices that cells make during
development: i.e. to be anterior or posterior or to be dorsal or ventral. Some of the
early clues that these compartments are significant came from the domains of
transformation seen in mutants of the BX-C. For example, bithorax is a mutation in one of
the Ubx cis-acting regulatory sequences, and causes a transformation of the anterior
compartment of the haltere into the anterior compartment of the wing. Postbithorax, a
mutation in another Ubx cis-element, causes a transformation of the posterior compartment
of the haltere into the posterior compartment of the wing. Contrabithorax, yet another
regulatory mutation, causes transformation of the posterior compartment of the wing into
the posterior compartment of the haltere.
Specification of Cell Fate in the Wing
The anterior-posterior decision is controlled by the expression of the
homeobox gene engrailed. Engrailed expression results in cells becoming posterior in
identity. The later subdivision between dorsal and ventral cells is controlled by the
expression of the homeobox gene apterous in dorsal cells. Engrailed and apterous are
called selector genes because they are necessary and sufficient for posterior versus
anterior and dorsal versus ventral fate, respectively.
These initial differences are used to establish the localized expression of secreted
molecules that further pattern the a/p and d/v axes. Cells that express engrailed
interact with cells that do not express engrailed leading to the expression of the
secreted factor Dpp at the boundary between them. Interaction between the dorsal and
ventral cells (ventral cells do not express apterous) leads to the expression of the
secreted protein wingless at the boundary between them. Wg and Dpp diffuse across the
field of cells forming protein concentration gradients in the dorsal/ventral and
anterior/posterior axes, respectively. Wg and Dpp each specify cell fate in a direct,
concentration dependent manner. Thus, cells are initially specified as to their fate by
whether or not they express engrailed and/or apterous and the concentration
of Wg and Dpp to which they are exposed.
(Modified from Brook et al., 1996)
Spatial patterning of the anterior-posterior axis of the wing
As an example of the kinds of mechanisms that control imaginal disc development, we
shall consider in detail how cells become specified as to their position in the anterior-posterior
(AP) axis of the wing imaginal disc The wing is divided along the anterior posterior
axis into a series of longitudinal veins and intervein regions. Each vein has
characteristic sense organs and other structures that indicate that they are distinct from
one another. Intervein regions also have characteristic patterns of differentiation. So
the problem is to understand how these differences become specified during wing
Figure 2. (Modified from Brook et al., 1996)
Anterior versus Posterior Determination
Figure 3. engrailed expression in a wing disc
(left) and an adult wing(r)
(Modified from Brook et al., 1996)
The first decision cells make during wing development is whether
they are anterior or posterior. The disc primordia are established straddling the
parasegmental border. Cells in the posterior half express the homeodomain gene engrailed
(en), one of the segment polarity genes. Cells in the two halves of the disc are
already determined as anterior or posterior at the time of disc formation. Cells from the
posterior part never make anterior structures, anterior cells never make posterior
structures. This decision is controlled by whether or not wing cells express the engrailed
gene. How was this shown? The first clues that engrailed controlled anterior versus
posterior fate came from the phenotype of en1 mutant flies. en1 is a
regulatory mutation in the engrailed gene that results in reduced engrailed
expression in the wings. In en1 mutant flies, the posterior part of the wing (which
expresses engrailed) is transformed so that it develops as an anterior wing. The result is
that wing that has a symmetric double anterior pattern.
One difficulty with the interpreting the results of the engrailed flies is that en1
is not a complete deletion of engrailed function. Ideally, we would like to know
the effects of complete loss of function for genes when we are studying developmental
processes. A problem that arises when studying limb development is that most of the
important genes cause very early lethality when mutated. In order to circumvent this
problem we must use genetic mosaics. In Drosophila the most common way of making
genetic mosaics is to use somatic mitotic crossing over. This technique uses irradiation
or site specific recombinases to induce mitotic crossovers between sister chromosomes in
somatic cells. This results in patches of homozygous mutant tissue (i.e. en-/en- or
A/A in diagram below) in heterozygous animals (en-/en+, A/+ in diagram). It is
possible to simultaneously mark the clones with an innocuous genetic marker in order to
distinguish homozygous mutant tissue from the heterozygous background. Furthermore, the
time of induction of clones can also be controlled so the effects of loss of function for
a particular gene at different developmental stages can be assessed.
Figure 4. Exchange between
homologous chromosomes heterozygous for a mutation (A) and a cell marker (m) leads to the
production of a daughter cell homozygous for both the mutant and the marker (A m/A m) and
a twin homozygous for both wild-type alleles (+ +/+ +). The homozygous mutant cell will
proliferate to produce a patch of mutant tissue surrounded by wild-type cells. (Modified
from Brook et al., 1996)
When Tetsuya Tabata made clones of cells that lacked all engrailed
function, he noticed two things (Tabata et al., 1995). First as expected, the
mutant cells became anterior in character if the clone was in the posterior part of the
wing (i.e. the engrailed expressing region) and the clones had no effect if they
were located in the anterior compartment. The second thing he noticed was very surprising.
He found that whenever a clone was induced in the posterior part of the wing and caused a
new confrontation of anterior and posterior cells, the new confrontation induced the
development of a secondary A/P axis in the wing. This suggested that in normal
development, the boundary between cells expressing engrailed and cells not
expressing engrailed acted as an organizing centre controlling the specification of
the anterior-posterior pattern.
Figure 5. (Modified from Brook et al., 1996; see also
figures 3 and 6 from Tabata et al., 1995).
What could be mediating this interaction between the anterior and
posterior cells? From studies of segment polarity genes, it was known that the protein
Hedgehog was secreted by posterior cells in the embryonic segment and was responsible for
signaling to anterior cells. hedgehog was also expressed in the posterior cells in
the wing. Konrad Basler and Gary Struhl did a beautiful experiment to show that the
interaction between A and P was mediated by Hedgehog. They developed a technique that
allowed the generation of a different kind of genetic mosaic. This technique (called the
flip-out cassette) allowed the production of patches of tissue, which constitutively
expressed hedgehog (or any other gene). These clones are the reciprocal of the
clones induced by somatic cross-over as they lead to the activation rather than the loss
of the gene's function in clones of cells.
Figure 6. Scheme for producing clones of cells
expressing a gene of interest.
(After Basler and Struhl, 1994; modified from Brook et al., 1996)
The first transgene carries a cell marker flanked by FRT sites
separating a constitutive promoter and a protein coding sequence of interest (e.g. hh
or dpp cDNA). Excision of the "flip-out" cassette is catalyzed by the
yeast FLP recombinase, provided on a second transgene. The flp recombinase is under the
control of the heat-shock promoter and recognizes the FRT sites as targets for site
specific recombination. Production of the flip recombinase under heat shock control allows
the induction of clones expressing the gene of interest at any stage of development and in
any cell in the imaginal disc. The result is a clone of cells expressing the gene of
interest and not the cell-marker surrounded by non-expressing, marked cells.
Basler and Struhl found that clones of cells expressing hedgehog
had no effect in the posterior half of the wing (as expected because hedgehog is normally
expressed there), but cells located in the anterior half were able to re-organize pattern
in a manner similar to the effects of en mutants clones (Basler and Struhl, 1994). The
duplications produced by the hh clones were symmetric and organized around the cells
expressing hedgehog. This suggested that hedgehog could induce anterior
cells to organize a new A/P axis.
Figure 7. (Modified from Brook et al., 1996)
Organizing the Anterior-Posterior axis
Hedgehog does not organize the anterior posterior axis directly
but rather it induces a second gene at the interface between the anterior and posterior
cell populations. This gene, dpp, is expressed in the anterior cells in response to
the Hedgehog signal. If the clones of cells receiving the Hedgehog signal are prevented
from producing dpp, they do not produce axis duplications, indicating that dpp
expression is necessary for Hedgehog-induced axis duplications. Dpp (remember, it is a
secreted protein of the TGF-beta family) is expressed in a stripe of cells that bisects
the wing imaginal disc into anterior and posterior halves. Its spatial expression pattern
and its function as a signaling molecule made it an excellent candidate to organize the
pattern in the a/p axis. Basler and Struhl were able to show this by making clones of
cells which constitutively expressed dpp. These clones were able to cause pattern
duplications in both the anterior and posterior halves of the wing (see Figure 5 of Zecca et
Figure 8. (Modified from Brook
et al., 1996)
The results suggested that the way the anterior posterior axis
was organized could be broken down into three steps.
- interaction between anterior and posterior cells (i.e.. engrailed
expressing and non-expressing cells)
- short range signaling from posterior to anterior by hedgehog
- signaling from dpp to control pattern in both the anterior and
posterior halves of the wing.
Direct, long range action of dpp
So, the fate of cells along the A/P axis of the wing is specified
by two factors. The expression of the engrailed gene determines the anterior versus
posterior fate and the distance from the source of dpp signaling determines what part of
the pattern of cell will make. For example, cells that are posterior (express engrailed)
and close to the source will become vein 4 cells and posterior cells that are farther away
become vein 5. Similarly, nearby anterior cells will become vein 3 and anterior cells that
are further away become vein 2. But, the AP axis of the wing is approximately 50 cells in
diameter. How is it that Dpp influences the development of cells that are so distant from
where it is expressed?
There are two likely explanations. The first is that Dpp could diffuse a short distance
and induce a second signal. This signal would induce another signal which in turn could
induce a third signal, etc. This series of cell interactions could specify different fates
in the A/P axis. This is termed serial induction or signal relay. The second
model, direct long-range action, suggests that dpp diffuses over a long range and
forms a concentration gradient. In this model, cells differentiate into different
structures depending on the concentration of dpp to which they are exposed.
In order to test these models, it is necessary to manipulate the
ability of cells to receive the dpp signal and observe the effect on cell fate and
molecular marker expression.
The direct long range, concentration dependent action of Dpp is supported by three
experiments (See Nellen et al. 1996):
Modulating the levels of expression of Dpp affects the downstream genes spalt
and omb in a concentration dependent manner. Increasing the amount of Dpp produced
by the cells at the A/P boundary results in a broader domain of spalt expression than
normal suggesting that higher concentrations of Dpp are now found in cells further away
from the domain of Dpp expression. The expression of low levels of dpp is sufficient to
activate omb but not spalt , consistent with the idea that omb requires a
lower threshold concentration of Dpp for activation.
Clones of cells mutant for the Dpp receptor that are located far away from the
Dpp-secreting cells are not influenced by the Dpp signal. This is shown by the loss of
spalt and omb. If the action of Dpp on these cells was indirect (i.e. through a relay of
signals other than Dpp), the loss of the Dpp receptor in these cells would have no effect.
In clones of cells expressing a ligand independent, activated form of the Dpp receptor,
the expression of spalt and omb is restricted to the clone. These cells have been
"tricked into thinking" they receive the Dpp signal. If a signal relay were
occurring, expression of marker genes expressed distant from the source of Dpp would be
expected to be induced outside the clone due to the induction of second signals in the
clone expressing the activated form of the receptor.
Thus, the secretion of Dpp by cells in a stripe in the centre of the disc may set up a
concentration gradient of Dpp across the wing imaginal disc. This has been shown to
establish broad domains of gene expression in a manner that may be analogous to the way
the gap gene domains are established in the embryo in response to the bicoid gradient.
Diagram of the spatial relationships between spalt,
omb and dpp-expressing cells
Drosophila limbs are derived from embryonic epithelial
sacs called imaginal discs
The cells of the imaginal disc are stably determined to give rise to a specific segmental
appendage as shown by in vivo culture experiments. Under these conditions disc cells only
rarely change their state of determination though a process termed transdetermination
The thoracic imaginal disc primordia are set aside early in embryogenesis in response to
the secreted signals wingless and decapentaplegic.
The genetic pathway for the development of the anterior posterior axis involves i)
anterior versus posterior fate determination by the engrailed gene; ii) posterior to
anterior cell signaling mediated by hedgehog that results in the expression of dpp
bisecting the A/P axis; and iii) organization of the A/P axis by dpp in a direct
and concentration dependent manner. A similar pathway controls the development of the D/V
- Describe the signals that specify ectodermal cells to become
imaginal disc cells. What prevents abdominal cells that receive wg and dpp
from becoming imaginal discs?
- Describe the methods for making clones using somatic recombination
and the "flip-out" technique.
- Describe the role of each of the following genes in the
development of the A/P axis of the wing: engrailed, hedgehog, and dpp.
What determines the fate of a cell in the wing A/P axis?
- What is the difference between serial induction (signal relay) and
direct long range action of a secreted signal? What sorts of experiments can be performed
to distinguish between these two models?
Reviews* and References
Basler, K., and Struhl, G. (1994). Compartment boundaries and the
control of Drosophila limb pattern by hedgehog protein. Nature 368, 208-14.
*Blair, S. S. (1995). Compartments and appendage development in Drosophila. Bioessays 17,
*Brook, W. J., Diaz-Benjumea, F. J., and Cohen, S. M. (1996). Organizing spatial pattern
in limb development. Ann Rev Cell Dev Biol 12, 161-180.
Browder, L.W., Erickson, C.A. and Jeffery, W.R. 1991. Developmental Biology.
Third edition. Saunders College Pub. Philadelphia.
Cohen, S. M. (1993). Imaginal disc development. In Drosophila
Development, A. Martinez-Arias and M. Bate, eds. (Cold Spring Harbor: Cold Spring Harbor
Press), pp. 747-841.
Gilbert, S.F. 1997. Developmental Biology. Fifth Edition. Sinauer. Sunderland,
Kalthoff, K. 1996. Analysis of Biological Development. McGraw-Hill. New York.
Lecuit, T., Brook, W. J., Ng, M., Calleja, M., Sun, H., and Cohen, S. M. (1996). Two
distinct mechanisms for long-range patterning by Decapentaplegic in the Drosophila wing.
Nature 381, 387-93.
Nellen, D., Burke, R., Struhl, G., and Basler, K. (1996). Direct and long-range action of
a DPP morphogen gradient. Cell 85, 357-68.
Tabata, T., Schwartz, C., Gustavson, E., Ali, Z., and Kornberg, T. B. (1995). Creating a
Drosophila wing de novo, the role of engrailed, and the compartment border hypothesis.
Development 121, 3359-69.
Wolpert, L., Beddington, R., Brockes, J., Jessell, T., Lawrence, P. and Meyerowitz, E.
1998. Principles of Development. Current Biology. London.
Zecca, M., Basler, K., and Struhl, G. (1995). Sequential organizing activities of
engrailed hedgehog and decapentaplegic in the Drosophila wing. Development 121, 2265-2278.