- Signaling by wingless in Drosophila
- John Klingensmith* and Roel Nusse
- Howard Hughes Medical Institute
- Dept. of Developmental Biology
- Stanford Univ. School of Medicine
- Stanford, CA 94305
- * Present address: Samuel Lunenfeld Research Inst.,
- Mount Sinai Hospital, Toronto, ONT M5G 1X5 CANADA
- wingless, a member of the Wnt gene family, is an essential gene for
segmentation in Drosophila, and is also involved in many other patterning
events. The gene encodes a secreted protein that can regulate gene expression
in adjacent cells. Recently, significant progress has been made in elucidating
the signal transduction pathway of wingless, mainly by genetic experiments
but increasingly also at the biochemical level. While many components of
wingless signaling, in particular a receptor, remain to be identified,
our current understanding of wingless pathway is more advanced than that
of other Wnt genes. We will give an overview of the various roles of wingless
in development; and we will then summarize the wingless signaling pathway
as it has emerged from genetic and biochemical studies. Where appropriate,
wingless signaling will be compared to the activity of vertebrate Wnt
- 1. wingless mutations
- Mutations at wingless (wg) or its mammalian counterpart Wnt-1 have
been discovered in several independent studies. A viable wg allele (Sharma,
1973; Sharma & Chopra, 1976) leading to frequent loss of the wing and
a duplication of the notum gave the gene its name. The classic genetic
screen by Nüsslein-Volhard & Wieschaus, (1980) identified the
segment polarity phenotype of embryonic lethal wg mutations. Working on
the origin of mammary cancer, Nusse & Varmus, (1982) found retroviral
insertional mutations in mouse tumors at the Wnt-1 (int-1) gene, later
to be identified as the mouse homolog of wg (Baker, 1987; Rijsewijk et
al., 1987). Finally, a known mouse mutant, swaying, also turned out to
be an allele of the mouse Wnt-1 gene (McMahon & Bradley, 1990; Thomas
& Capecchi, 1990; Thomas et al., 1991).
- 2. wg has many roles in development
- In the embryo, wg is required at multiple points in patterning the
segmented ectoderm of the trunk (Baker, 1988a; Bejsovec & Martinez-Arias,
1991; Bejsovec & Wieschaus, 1993; Dougan & Dinardo, 1992; Nüsslein-Volhard
& Wieschaus, 1980), and is necessary for development of the head (Schmidt-Ott
& Technau, 1992). Subsets of neurons in each segment of the early central
nervous system also require wg activity (Chu-Lagraff & Doe, 1993).
Internally, wg is needed for patterning the midgut (Bienz, 1994; Immerglück
et al., 1990; Reuter et al., 1990; Thüringer & Bienz, 1993) and
growth of the malpighian tubules (Skaer & Martinez Arias, 1992).
- Development of adult structures begins in the embryo with the partitioning
of cells into imaginal disc primordia, a process for which wg is necessary
(Cohen et al., 1993; Cohen, 1990; Simcox et al., 1989). In the developing
leg, wg is required for both dorsoventral and anteroposterior axial patterning
(Baker, 1988b; Campbell et al., 1993; Couso et al., 1993). Early development
in the wing disc requires wg for growth of wing blade structures (Morata
& Lawrence, 1977; Williams et al., 1993), while later it is involved
in bristle patterning along the wing margin (Phillips & Whittle, 1993).
Several other adult structures also need wg activity for proper development,
including the genitalia, antennae, and eyes (Baker, 1988b; Morata &
- Hence, while most attention with respect to wg function has traditionally
been given to patterning of the embryonic epidermis and its segment polarity
phenotype, it should be kept in mind that wg is required for a great number
of additional developmental processes. In extrapolating the molecular mechanism
of wg to the less understood functions of Wnt genes in vertebrates, this
diversity should be taken into account. Rather than the paradigm of the
regulation of engrailed (en) by wg (see below), which is specific for the
epidermis, it may well be that the most conserved function of wg is not
in segmentation. A search for conserved molecular pathways therefore might
better be based on other aspects of wg signaling, such as the midgut, for
- 3. Functional analysis of wg activity in vivo.
- 3.1 wg as an extracellular signal
- The first clue as to how wg works came from mosaic analysis, in which
clones of cells deficient for wg occurred within fields of wild type cells
in adult flies. wg was found to function non-autonomously (Baker, 1988a;
Morata & Lawrence, 1977); in other words, cells which do not make Wg
themselves can be perfectly normal, provided wild type cells were nearby.
The gene also behaves non-autonomously in embryos (Chu-Lagraff & Doe,
1993; Wieschaus & Riggleman, 1987).
- That the wg-dependent rescuing activity is Wg protein itself is strongly
suggested by examining embryos stained with anti-wg antibodies (Figure
1A, B; Figure 2A). Electron microscopy has revealed that within the embryonic
epidermis, wg antigen occurs not only in the narrow stripes of cells which
express wg RNA, but also in adjacent cells within a few diameters (González
et al., 1991; Van den Heuvel et al., 1989). In the midgut, wg RNA is transcribed
in a patch of cells in the visceral mesoderm, yet the protein can also
be found in adjoining endodermal cells (Reuter et al., 1990)
- In many cases, wg activity is necessary for gene regulation in adjacent
cells. In the developing epidermis, wg is expressed in cells just anterior
to those expressing en separated by the parasegmental border (Figure 2
E, G) (Baker, 1987; Martinez-Arias et al., 1988; Van den Heuvel et al.,
1989). This configuration is also found in short germ band insects, whose
segmentational organization is achieved by adding segments sequentially,
in a different way than in Drosophila (Nagy & Carroll, 1994). Although
wg is not necessary for the initiation of en expression, it is required
for its maintenance (DiNardo et al., 1988; Heemskerk et al., 1991; Martinez-Arias
et al., 1988). Note that in the developing CNS, wg governs expression of
even skipped and runt in adjacent neuroblasts, but en expression is independent
of wg (Chu-Lagraff & Doe, 1993; Doe, 1992)
- In addition to these planar signaling functions, wg can also convey
inductive signals, between germ layers. In the midgut, its expression
in the visceral mesoderm is necessary for proper expression of labial (lab)
across germ layers in the adjoining endoderm (Bienz, 1994; Immerglück
et al., 1990; Reuter et al., 1990). Thus Wg protein can be secreted within
and across tissue layers, and can be taken up by cells in which wg activity
- If Wg protein were truly a secreted factor conveying cell-fate instructions,
then cells overproducing wg should effect patterning in neighboring wild
type cells. Struhl and Basler (1993) have observed just this in a study
of clones expressing ectopic wg in the leg. A number of in vitro studies,
which we shall discuss later, have reinforced the conclusion that wg encodes
a secreted factor influencing the fates of responsive cells.
- In addition to paracrine signaling, wg also has autocrine functions,
since the gene is required in at least some cases by the cells which express
it. One example is regulation of armadillo (arm) protein accumulation in
the embryonic epidermis. Arm protein accumulates at high levels as a rapid
response to wg expression (Riggleman et al., 1990). High levels of Arm
occur in broad stripes centered around the narrow wg stripes, clearly including
those cells which express wg as well as those which flank them on either
side. It is not yet known whether wg must be secreted and taken up by the
expressing cell to achieve its effects on Arm. Another example of autocrine
wg action is on its own expression (see 7).
- 3.2 wg as an instructive signal.
- Several observations suggest that wg is instructive of cell fate.
In the embryonic epidermis for example, wg promotes different fates among
the en expressing cells, such that those closest to the wg domain secrete
naked cuticle while those further away secrete denticles (Dougan &
Dinardo, 1992). If the embryo is genetically perturbed such that the
en domain is flanked by wg-expressing cells on both sides, all en cells
give rise to naked cuticle. In contrast, ubiquitous wg expression - via
induction of the cDNA fused to a heatshock promoter (HS-wg) - leads to
a largely opposite phenotype, in which all cells secrete naked cuticle
(Noordermeer et al., 1992). In the absence of wg function, naked cuticle
is replaced by denticles (Bejsovec & Martinez-Arias, 1991; Bejsovec
& Wieschaus, 1993). wg seems to instruct cell fate in other circumstances
as well, such as in the leg imaginal disc. Here, the genes is expressed
in a ventral wedge, and is necessary for the development of ventral bristles
(Couso et al., 1993; Peifer et al., 1991). Dorsal clones of cells expressing
wg ectopically lead to locally ventralized patches of bristles, suggesting
that Wg instructs cells to assume ventral characteristics (Struhl &
- Nevertheless, there are data that Wg does not provide instructive
signals. When wg is expressed ubiquitously in an embryo lacking endogenous
wg, the level of wg is uniform in all cells of the ventral epidermis.
Because all such cells are capable of secreting naked cuticle, a fate apparently
promoted by wg, one would expect a naked cuticle. Surprisingly, the response
of cells is not uniform; many cells secrete denticles, and a periodic pattern
of denticles and naked cuticle occurs across the segments (Noordermeer
et al., 1994; Sampedro et al., 1993). One suggestion made to explain these
results is that wg acts to isolate or seal parasegments from hypothetical
gradients of positional information in adjacent parasegments (Sampedro
et al., 1993).
- There is another explanation which reconciles the conflicting data
on the instructive nature of Wg, namely that a second factor, either permissive
or instructive, permits or modifies responses to the Wg signal. There is
independent evidence for this notion, as follows. In addition to promoting
naked cuticle, wg also appears to promote en expression - but this can
occur only in cells at the posterior of the emerging segment; thus when
HS-wg is expressed ubiquitously with or without endogenous wg, only these
posterior cells can express en (Noordermeer et al., 1992; Noordermeer et
al., 1994; Sampedro et al., 1993). Such cells might be considered to form
an "en competence domain," made so by prior action of pair-rule
genes (Ingham & Martinez Arias, 1992). Wg may indeed have an instructive
role in cell fate determination, but in some cases this may require an
independent permissive function. The latter may include prepatterning genes
as in the case above, localized reception pathway components, or possibly
even second signaling molecules, such as Hedgehog, acting in concert with
Wg (Heemskerk & DiNardo, 1994).
- 3.3 A graded morphogen or a local signal?
- Whether or not a morphogenetic gradient of Wg protein accounts for
the effects of the gene has been hotly debated for years. Light and electron
microscopy of embryos stained with anti-wg antibodies provide no definitive
support for either case, but suggest that the protein may have a graded
distribution across several cell diameters (González et al., 1991;
van den Heuvel et al., 1993a; Van den Heuvel et al., 1989). Several results
support a gradient model of wg action. In the ventral epidermis of the
embryo, distance of a cell from the wg expressing stripe is correlated
with its cuticle phenotype, both to the anterior (Bejsovec & Martinez-Arias,
1991) and to the posterior (Dougan & Dinardo, 1992). In the leg imaginal
disc, Struhl & Basler, (1993) observed that clones of cells expressing
wg are associated with ventral bristle identities, even when they occur
dorsally. Although the most ventral bristle types were not present, this
may be because ectopic wg did not occur at sufficiently high levels; the
ventrolateral bristle types observed would reflect lower Wg concentrations.
Nonetheless, they found that the induced ventralization extended beyond
the clone itself, suggesting that wg might organize ventral pattern in
a gradient-type manner.
- On the other hand, some observations are at odds with the gradient
model for wg action. As mentioned before, ubiquitous levels of wg do not
result in a uniform cuticle (Noordermeer et al., 1994; Sampedro et al.,
- An additional conflict with morphogen models arises from the pattern
of gene expression and cuticle secretion in embryos bearing alleles of
wg which result in non-functional, non-secreted Wg proteins (van den Heuvel
et al., 1993a). Homozygous embryos show a distribution of Wg protein limited
to the cells which express it, and en expression is consequently lost (Figure
2C). In the heterozygotes, in which one allele of wg is wildtype, en is
maintained as normal yet wg appears still to be largely confined to those
cells which express it (Figure 2D, F). Later, heterozygotes secrete completely
wildtype cuticles and are perfectly viable. In these heterozygotes, no
more than half the normal amount of Wg is secreted or functional, yet there
is no deleterious consequence. Similarly, in embryos mutant for the temperature
sensitive allele wgIL114, staining for the wg protein shows a pattern intermediate
between mutant and wild type, with little detectable spread of the protein
(Figure 2B; van den Heuvel et al., 1993a), yet the embryos develop normally.
In cell culture experiments, a cloned version of this allele also behaves
as somewhat impaired in secretion, even at the permissive temperature (van
den Heuvel et al., 1993a). Similar results come from analysis of porcupine
(porc) , which is required for extracellular distribution of Wg (van den
Heuvel et al., 1993a) in that heterozygotes are perfectly wildtype despite
the fact that most Wg is restricted to the cells which express it (Siegfried
et al., 1994).
- Thus, it is not clear whether Wg protein acts via a concentration
gradient or as a local signal, affecting cells only in the immediate vicinity
of those which express it. It is possible that long range effects of wg
activity on pattern are the result of sequential local cell-cell interactions,
or alternatively, that the spread of active wg protein is under some form
of regulation, for example by binding to components of the extracellular
matrix or to its own receptor.
- 3.4 wg defines positional coordinates
- In addition to its involvement in generating larval cuticle pattern
from epidermal cells during segmentation, wg is concomitantly involved
in allocating some of these cells to form the primordia for thoracic imaginal
discs, from which adult legs, wings and halteres will develop. Along the
parasegment boundary of the emerging thoracic segments, small bilateral
clusters of cells are specified to contribute to imaginal rather than larval
tissue. These cells invaginate and remain diploid while their siblings
in the overlying larval epidermis become large and polyploid. Formation
of imaginal primordia in (at least) the thoracic segments does not occur
in wg mutants (Simcox et al., 1989). Using a conditional allele, Cohen
et al., (1993) have pinpointed the requirement for wg in this process at
mid-germband extended stages, well before any morphological signs of the
primordia can be detected. This suggests that wg is directly involved in
the allocation process per se. Intriguingly, the clusters occur at the
intersection of segmental stripes of wg expression and longitudinal streaks
of decapentaplegic (dpp) expression (Cohen et al., 1993). dpp is a Drosophila
member of the TGF-B family of secreted growth factors (reviewed by Kingsley,
(1994)). We will explore below the possibility that wg and dpp signals
synergistically convey positional information.
- In wing development, wg appears to act at second instar as the first
in a hierarchy of genes which controls dorsal-ventral pattern (Diaz-Benjumea
& Cohen, 1993; Williams et al., 1993). Expression of wg in the wing
disc is spatially complex and temporally dynamic, apparently independent
from embryonic antecedents (Couso et al., 1993; Williams et al., 1993).
In the early leg disc, which straddles the parasegmental border, the pattern
of wg and en expression is largely retained from the embryo. Superimposition
of fate maps and expression patterns shows that wg is expressed in a ventral-anterior
wedge while en is in the posterior (Couso et al., 1993; Struhl & Basler,
1993). Experiments with a conditional allele reveal a continuous requirement
for wg throughout larval stages in patterning of the ventral face of the
leg, in both the cells expressing wg as well as in flanking regions (Couso
et al., 1993). Couso and colleagues suggest that a hypothetical cross-section
of the leg at any given proximo-distal (P-D) level can be thought of as
a circular segment, and argue that wg activity forms the basis for a polar-coordinate
system of positional information (Bryant, 1993; Couso et al., 1993). The
polar coordinate model predicts that the position of a given cell in an
appendage is specified as a particular circular value in the plane defined
by the D-V and A-P axes, at some linear position along the perpendicular
P-D axis (French et al., 1976) Couso et al. (1993) envision that the function
of wg is to define circular values centered on the D-V axis, with a resulting
secondary effect on P-D axial pattern. As discussed above, Struhl &
Basler (1993) have presented evidence suggesting that the basis of this
D-V function is a gradient of ventral organizing activity encoded by wg.
- An alternative explanation for the role of wg in P-D patterning is
that it directly specifies a distal-most organizing center which brings
about growth and pattern along this axis. Campbell et al. (1993) observed
that cells expressing wg and dpp are juxtaposed at the presumptive distal
tips of both leg and wing discs, in each case associated with expression
of aristaless (al). They tested the potential of this juxtaposition to
influence pattern by inducing clones of cells overexpressing wg in the
leg. If the clone occurred near cells expressing high levels of dpp (i.e.
dorsal cells) a supernumerary leg with a duplication of the P-D axis would
occur, accompanied by an ectopic patch of al expression. Moreover, the
distal tips of the supernumerary legs show the juxtaposed pattern of wg,
dpp, and al expression. The implication is that wg promotes the establishment
of an organizer of P-D pattern, normally at the presumptive distal tip
of the developing limb (Campbell et al., 1993). Unfortunately, size and
time limitations prevent direct tests of this idea by surgical manipulations
similar to the organizer grafts of amphibian embryologists.
- 3.5 Cooperation between wg and dpp
- As noted, these potential organizers of P-D pattern occur at sites
where cells expressing wg and dpp are juxtaposed, and expression of al,
a homeobox gene, seems to follow (Campbell et al., 1993). This situation
is similar to the embryonic case of thoracic disc primordia, which occur
at the intersection of perpendicular stripes of wg and dpp (Cohen et al.,
1993). At the site of the primordia, expression of the homeobox gene Distalless
(Dll) is induced (Cohen et al., 1993; Cohen, 1990). In neither of these
cases has the role of dpp been tested directly, but it seems likely that
wg and dpp signals together specify positional information which is manifested
in the expression of specific homeobox genes and the adoption of certain
cell fates (Campbell et al., 1993). Similarly, in the embryonic midgut,
cells expressing wg and dpp are directly juxtaposed and both activities
are required for normal expression of the homeobox gene lab and formation
of the second midgut constriction (Immerglück et al., 1990; Reuter
et al., 1990) L. Matthies and M.P. Scott, personal communication). An analogous
situation may occur in Xenopus development, in which certain Wnt and TGF-ß
family members may be involved in establishing Spemann's Organizer and
inducing expression of organizer-specific homeobox genes such as goosecoid
(Kimelman et al., 1992). Perhaps these phenomena represent a generalized,
evolutionarily-conserved mechanism of pattern formation in which Wnt and
TGF-ß proteins function as synergistic signals to promote cell fate,
with homeobox gene expression a consequential manifestation involved in
subsequent patterning events.
- These combined effects of wg and dpp form a nice example of the combinatorial
action of growth factors. It is not clear at what level dpp and wg interact;
in principle that could occur between the molecules themselves, their receptors,
downstream effectors or nuclear target genes.
- 3.6 wg as a regulator of cell proliferation
- In view of the mitogenic and oncogenic role of the mammalian Wnt-1
gene (reviewed in Nusse & Varmus, 1992), the question whether wg can
act as a regulator of cell proliferation has been examined with special
interest. Two studies have directly addressed this aspect of wg function,
both initially based on a correlation between the expression of wg and
local changes in the rate of cell division.
- In the embryo, wg is expressed in the tips of the developing malpighian
tubules. Analysis of wg mutant embryos shows that wg activity is required
during the proliferative period for continued cell division (Skaer &
Martinez Arias, 1992). Moreover, ectopic wg expression results in over
proliferation. Proliferation is regulated by the tip cells of the tubule,
but wg is expressed in more proximal cells as well. It is possible that
wg plays an indirect role, for example during the normal differentiation
of tip cells which in turn regulate proliferation of more proximal cells
(Skaer & Martinez Arias, 1992), but these findings indicate that wg
can be a positive regulator of cell growth.
- In the second study, the conclusion is that wg can be a negative regulator
of cell growth. Late third instar wing discs are characterized by a zone
of non-proliferating cells (ZNC) along the presumptive wing margin, coincident
with a stripe of wg expression (Phillips & Whittle, 1993). By examination
of cell division rates under permissive and non-permissive conditions of
a wg allele, Phillips and Whittle (1993) found that wg is required to make
these cells quiescent, in a zone including and flanking the wg-expression
- Thus wg may indeed effect proliferation rates, but this can be negative
or positive and is context dependent. It should be noted that the small
size of embryos lacking wg is due to cell death rather than failure of
proliferation (Klingensmith et al., 1989; Perrimon & Mahowald, 1987),
and that embryos in which wg is expressed ubiquitously are not larger than
normal (Noordermeer et al., 1992). Thus it seems that proliferation control
is not the physiological basis of wg function in general.
- 4. The wg protein
- 4.1 Biochemical aspects
- As do all members of the Wnt gene family (Nusse & Varmus, 1992),
the wg protein contains a conserved pattern of cysteine residues, a signal
sequence and several N-linked glycosylation sites (Cabrera et al., 1987;
Rijsewijk et al., 1987). Among the cysteines, one is at a position unique
to wg and to several vertebrate Wnt-1 variants, which, in conjunction with
other amino acid identities, suggest that Wnt-1 and wg are true orthologs
within the larger Wnt gene family (Sidow, 1992). There is one major structural
difference between wg and other Wnt genes: an insert of 85 amino acids
which likely has no functional significance as it is not conserved even
within insects (Nagy & Carroll, 1994). This insert has nevertheless
been very useful in studying the properties of the Wg protein, because
it presents a strong antigenic determinant (Van den Heuvel et al., 1989).
Antibodies made to the wg protein are very specific and allow the visualization
of the protein in whole mount Drosophila embryos (Figures 1,2). From these
stainings, it appears that the wg protein is secreted, as it is present
in vesicles and in between cells (González et al., 1991; Van den
Heuvel et al., 1989) The protein can also be detected in multi vesicular
bodies, which are presumably early stage endosomes destined for the lysosome
(Van den Heuvel et al., 1989).
- Biochemical evidence for secretion of the wg protein has been obtained
from cell culture experiments. Drosophila Schneider cells transfected with
expression constructs produce the wg protein in the extracellular matrix
and in the medium (van den Heuvel et al., 1993a; Van Leeuwen et al., 1994).
A significant fraction of the protein is also seen on the cell surface,
from which it can be released by the polyanionic drug suramin. When wg
is expressed in heterologous systems, such as Xenopus oocytes or mouse
mammary gland cells, the protein behaves in a similar fashion and has biological
activity like mammalian Wnt genes (Chakrabarti et al., 1992; Ramakrishna
& Brown, 1993).
- In spite of these observations, there are uncertainties about the
way the wg protein is secreted and about the nature of the active Wg protein.
In cultured cells, the majority of wg protein remains associated with cells,
similar to what has been found for its mammalian counterpart Wnt-1, on
which much more biochemical work has been done. When the intracellular
processing of the Wnt-1 protein is examined, it appears that most of it
accumulates in the endoplasmic reticulum; is incompletely processed (Brown
et al., 1987; Papkoff et al., 1987); and is associated with the resident
ER protein BiP (Kitajewski et al., 1992). These observations suggest that
correct folding and secretion of Wnt protein is dependent on an accessory
protein, and that the limited presence of such an accessory protein in
many cultured cells would impair efficient secretion. Extracellular forms
of the mammalian Wnt proteins can only be found when the cells are incubated
in the presence of charged polymeric molecules such as suramin or heparin
(Bradley & Brown, 1990; Papkoff & Schryver, 1990).
- 4.2 The nature of wg mutations
- The molecular lesions in several existing lethal wg mutants have been
mapped in detail, all showing changes in highly conserved residues of the
wg protein (van den Heuvel et al., 1993a). Curiously, the wgCE7 allele
has sustained a splice site mutation, which in effect removes the last
exon. A non functional protein is encoded that in vivo appears nevertheless
to be secreted and even taken up by adjacent cells (A. Besjovec, personal
communication). The wgIL114 allele is temperature sensitive; at the permissive
temperature, this allele is sufficient for completely normal embryonic
development, while at the restrictive temperature it results in patterning
defects similar to protein null mutants. In this allele the aminoterminal
conserved cysteine is changed into a serine (van den Heuvel et al., 1993a).
Immunostaining experiments and expression studies in cultured cells showed
that most of these mutant wg proteins are not secreted. It should not be
concluded, however, that the mutations affect secretion as such, and that
the loss of function of these alleles is secondary, due to impaired secretion:
mutant forms of normally secreted proteins usually become folded incorrectly
and are retained in the endoplasmic reticulum. For the time being, these
mutational analyses have not provided clues as to the function of the wg
protein, although the cloned temperature sensitive allele has proved to
be a very useful reagent for in vivo and in vitro expression studies.
- 4.3 porcupine may provide an accessory function for wg secretion or
- The poor secretion of Wnt proteins and their retention in the ER
could be explained by a requirement for an accessory protein that facilitates
correct folding. If such an accessory protein were specific for wg, one
might expect that mutations in its gene have the same cuticle phenotype
as wg and would affect secretion of the wg protein in vivo. This is exactly
the phenotype of the segment polarity gene porc (porc) (van den Heuvel
et al., 1993a), as shown in Figures 1 C,D and 2H. porc mutant cells are,
like wg mutant cells, non-autonomous in mosaic animals, indicating that
porc works in presenting rather than in interpreting or receiving the wg
signal (J. K. and N. Perrimon. in preparation).An indication for an accessory
function of porc in wg secretion also comes from observations made on expression
of the armadillo gene, one of the downstream effectors of wg in the Drosophila
embryo. wg normally regulates the local accumulation of the armadillo
protein, both within wg-producing and in adjacent cells. In porc embryos,
arm accumulates only within cells expressing wg (Riggleman et al., 1990).
This may imply that wg has both autocrine and paracrine effects and that
porc is required for the paracrine functions by facilitating wg secretion
or transport. Molecular cloning of porc is necessary to further address
- 5. The wg signaling pathway as emerged from genetic studies
- 5.1 The model
- In the following section we will integrate several recent studies
pertaining to the genetic requirements for wg function, which collectively
begin to define a wg signaling pathway. It should be emphasized that the
context in which wg signaling has been addressed in depth is the embryonic
epidermis. Most research has focused on the regulation of en transcription
by the wg signal and it remains to be determined whether the same pathway
is used in other contexts of wg action. We first describe the current model
of the signaling pathway (see Figure 3 and table I) and we will then discuss
the experiments leading to this scheme in more detail.
- In this model, the wg protein is secreted or transported with the
assistance of the porc gene product. In all likelihood, wg binds to a cell
surface receptor, yet to be identified. Within the target cell, the product
of dishevelled (dsh) is the first known component in signaling, acting
through an unknown mechanism. Downstream of dsh is a protein kinase, called
zeste -white 3 (zw3) which then leads directly or indirectly to the phosphorylation
of the arm protein. The kinase activity of zw3 will inhibit arm, unless
this inhibitory effect of zw3 is relieved by the wg signal. This relief
results in higher levels of the arm protein, which is itself also necessary
to transmit the wg signal to more downstream targets, such as en (Figure
3, Table I).
- 5.2 The experiments
- Phenotypic observations suggested that a group of three other genes
function together with wg to affect patterning (Klingensmith & Perrimon,
1991). When maternal and zygotic contributions of dsh, arm or porc are
removed via germ line clones, each of these loci results in a segment
polarity phenotype indistinguishable from that of wg (Figure 4 BE; Table
I ; Klingensmith et al., 1989; Klingensmith & Perrimon, 1991; Nüsslein-Volhard
& Wieschaus, 1980; Perrimon & Mahowald, 1987). Moreover, mutations
at all four loci lead to an identical temporal and spatial pattern of en
decay, which occurs earlier than in any of the other segment polarity mutants
(Van den Heuvel et al., 1993b). As none of these genes is necessary for
transcription or translation of wg gene products (Van den Heuvel et al.,
1993b), these genes are good candidates for components of the wg signaling
pathway. All three genes are fully paternally rescuable.
- Because wg, arm, dsh and porc all have identical, recessive phenotypes,
double-mutants between null and hypomorphic alleles have not been informative.
Epistatic relationships between these genes have nevertheless been established,
with the aid of mutations giving opposite phenotypes (Figure 4 C,F): an
artificial gain of function wg allele (HS-wg) and mutations in another
gene that participates in the wg pathway, zw3.
- Mutations in zw3 lead to a naked cuticle and to an expanded en
domain (Perrimon & Smouse, 1989). Siegfried and colleagues (1992) have
presented evidence that zw3 acts as a repressor of en, possibly by inhibiting
en autoregulation, and that wg acts to relieve this repression (Siegfried
et al., 1992). This was concluded in part from the epistasis of zw3 to
wg; i.e. the double mutant looks like zw3. By combining wg null mutants
with a reduced dose of zw3 (rather than total loss of function; in embryos
lacking maternal zw3 but having a zygotically active allele), it was inferred
that zw3 is indeed in the wg pathway rather than acting independently:
in such combinations wg is necessary to give ectopic en (Siegfried et al.,
1992). These findings imply that wg acts through zw3 to effect en expression
and cuticle patterning. In other double mutant combinations, the relationship
of other wg-like mutants to zw3 was established, finding that porc and
dsh function upstream of zw3 while arm functions downstream (Siegfried
et al., 1994).
- A similar relationship between zw3 and arm was found by (Peifer et
al., 1994). These authors also used the partial activity of zw3 to link
zw3 activity to arm; embryos having no maternal zw3, but carrying one wild
type zygotically active allele do not hatch unless one also removes the
maternal contribution of arm. In other words, zw3 is required for the negative
regulation of arm, and if too little zw3 activity is present, the embryo
can still survive by having less arm protein than normal (Peifer et al.,
1994). Consistent with the epistasis between zw3 and arm is the ubiquitous
high expression of the arm protein in zw3 mutant embryos, indicating that
zw3 normally inhibits arm protein accumulation (Peifer et al., 1994).
- To order the activity of arm, dsh and porc with respect to wg itself,
a heat-shock wg transgene (HS-wg) proved to be very useful. Ubiquitous
wg expression leads to a phenotype which is approximately the opposite
of loss of function at wg and the same as zw3 (Noordermeer et al., 1992).
Two aspects of this phenotype have been very informative: the domain of
cells expressing en is doubled in width, and the cuticle is completely
naked. To determine the linear functional relationships of the other three
genes to wg, double mutants with the dominant transgene HS-wg were made
(Noordermeer et al., 1994) Double-mutants between HS-wg and any lethal
allele of dsh show the dsh pattern of en decay and the dsh segment polarity
phenotype. The en domain decays rather than broadens, and the cuticle shows
a lawn of denticles with no naked cuticle. The case of arm is more complex
in that only the strongest alleles of arm suppress HS-wg, with weaker alleles
showing partial suppression, despite both types of allele having the same
phenotype with respect to en expression and cuticle pattern. Nevertheless,
this experiment shows that arm and dsh are essential for the effects of
wg on both en expression and cuticle patterning (Noordermeer et al., 1994).
- Unlike arm and dsh, lack of porc function does not appear to effect
the outcome of ubiquitous wg expression, except for a partial suppression
of the cuticle phenotype (Noordermeer et al., 1994) Thus, although porc
mutations are identical to dsh and arm with respect to en and cuticle pattern,
they behave differently in combination with the ubiquitously expressed
wg transgene. Thus, porc is not epistatic to HS-wg; ubiquitous wg expression
bypasses the need for porc in wg signaling, possibly because the transport
function of Porc is not necessary when Wg is everywhere.
- The epistasis analyses discussed above collectively imply a linear
order of gene function. Taken together with the cellular requirements for
these genes, the signaling pathway between the wg-expressing cell and the
en-expressing cell can be sketched out. Like wg itself, porc function is
non-autonomous (J. K. and N. Perrimon, in preparation). In other words,
a cell lacking either wg or porc can be rescued by wildtype neighbors.
In contrast, a given cell will have a mutant fate unless it correctly expresses
each of the cell-autonomous genes: dsh (Klingensmith et al., 1994), zw3
(Simpson et al., 1988; (Perrimon & Smouse, 1989) and arm (Klingensmith
et al., 1989; Wieschaus & Riggleman, 1987). These results indicate
that in terms of wg signaling, porc and wg itself are involved in the signaling
mechanism while dsh, zw3 and arm function in the response to the signal
- 5.3 The signaling pathway in other contexts
- Most intercellular signaling pathways employ unique components, such
as specific ligands and receptors, as well as general signal transduction
factors, ras being a good example (Egan & Weinberg, 1993). Although
only a few links in the wg signaling chain are known, one can ask whether
they are unique to wg function, in addition to the related question of
whether the known components mediate wg signaling in all contexts. We have
discussed that wg has many targets, and that a given gene might be regulated
by wg in one tissue but not in another. The pathway could diverge at any
time during development. To determine whether a gene involved in wg signaling
in the embryo is also used later, one can induce mutant clones before the
later signaling event and analyze the phenotype of mutant cells in the
target tissue. However, such mosaic analysis is made difficult by non-autonomy
of function such as that of wg itself, and by cell-lethality, such as when
occurs when one attempts to make clones devoid of all arm activity (Peifer
et al., 1991).
- Nonetheless, it appears that all of the genes involved in wg signaling
in the embryo are also critical for patterning of the imaginal discs. Hypomorphic
alleles of wg, arm, dsh and porc all result in similar imaginal defects
(Baker, 1988b; Couso et al., 1994; Klingensmith et al., 1994; Peifer et
al., 1991; Theisen et al., 1994; J.K. and N. Perrimon, manuscript in preparation).
arm and zw3 are also required for oogenesis (Perrimon & Smouse, 1989;
Wieschaus & Noell, 1986), in which wg, dsh and porc have no role (Perrimon
et al., 1989). In imaginal discs, zw3 is involved in lateral inhibition
among presumptive sensory organ cells (Simpson, 1990), for which wg is
unnecessary. Thus, zw3 has additional roles in patterning unrelated to
wg, while arm is necessary at a more basic level for cell viability. dsh
and porc however appear to be relatively specific to wg-mediated patterning
events. On the other hand, dsh has a function in tissue polarity in conjunction
with several other genes (Theisen et al., 1994; Wong & Adler, 1993)
in which a role for wg is less clear.
- Having considered the specificity of components of the wg signaling
pathway, we can address the generality of this pathway in other contexts
of wg function. dsh appears to be required for target cells to respond
to the wg signal throughout development and in multiple tissue types. For
example, the midgut phenotype of dsh is identical to that of wg, and lab
expression in the midgut endoderm is affected in an identical way (Klingensmith
et al., 1994). Because the effect on lab expression is known to be directly
dependent on wg secreted by the visceral mesoderm across to the endoderm,
this indicates that dsh mediates wg function not only within the ectoderm,
but also in inductive signaling across germ layers. In patterning the margin
of the wing, wg also appears to act through dsh and zw3, in the same linear
relationship as in the embryonic epidermis (Couso et al., 1994).
- These results suggest that the wg reception pathway is common to its
many physiological roles. Certainly the targets of wg differ; for example,
en is not involved in the midgut functions of wg. At what point the pathway
diverges after dsh is difficult to establish, because of the more general
roles of zw3 and arm. In zw3 mutants, identifiable midgut does not form,
so the role of zw3 in transducing the wg signal in this context cannot
be easily determined. In the case of arm, levels of activity insufficient
for wg function in the epidermis appear to allow relatively normal midgut
development. This does not necessarily mean that arm is dispensable for
this function of wg; lower levels of arm might produce the same defects
as wg and dsh mutants. But conditions in which arm activity is further
reduced maternally interfere with oogenesis, so the actual role of arm
remains unclear as well.
- The next significant gains in our understanding of how dsh, zw3 and
arm transduce wg signals are likely to come from in vitro studies coupled
with transgenesis. Expression cloning of genes encoding factors which physically
interact with known components might offer a shortcut to genetic methods,
such as suppressor or enhancer screens. However, good candidates are likely
to result from further screens for maternal effect segment polarity genes.
dsh, zw3, arm and porc all have their segment polarity phenotypes when
both maternal and zygotic activities are lacking, and were revealed in
a large-scale X-chromosomal screen for maternal-effect loci with zygotic
functions (Perrimon et al., 1989). The genetic technology now exists to
extend the screen to the autosomes as well (Chou et al., 1993). Because
the X-chromosome represents only about 20% of the genome, one expects perhaps
a dozen more genes mutable to a wg-like phenotype, and a few more mutable
to a zw3-like phenotype.
- 5.4 The molecular nature of wg signaling components
- Except for porc, all known components of wg signaling have been cloned.
dsh encodes a pioneer protein, having no relatives in the database, nor
does it contain any known signaling motifs (Klingensmith et al., 1994;
Theisen et al., 1994). The only recognizable homology is with an unidentified
domain in the Drosophila tumor suppressor gene discs-large (Klingensmith
et al., 1994; Theisen et al., 1994; Woods & Bryant, 1991). The dsh
protein appears to be cytoplasmic (S. Yanagawa, F. Van Leeuwen, J. K. and
R.N., unpublished) and has no membrane spanning domains or signal sequences.
The gene is highly conserved in evolution; several vertebrate homologs
have been isolated which show extensive similarity with the Drosophila
gene throughout the coding domain (D. Sussman and J.K. , pers. comm.).
- With zw3, more familiar ground is entered as this gene encodes a protein
kinase with high homology to the mammalian glycogen synthase kinase 3ß
(GSK-3ß) enzyme (Bourouis et al., 1990; Siegfried et al., 1990).
The mammalian GSK-3ß can even substitute for zw3 in Drosophila (Siegfried
et al., 1992). There are several interesting aspects of this kinase, one
being its regulation by tyrosine phosphorylation. Unlike other kinases
regulated by this modification, GSK-3ß is normally phosphorylated
on tyrosine in resting cells (Hughes et al., 1993; Plyte et al., 1992).
Furthermore, the known functions of GSK-3 are all inhibitory, for example
its regulation of the c-jun protein (Boyle et al., 1991). zw3 appears to
have inhibitory functions as well, on the autoregulation of en and on the
arm levels. The genetic interactions show that wg counteracts these negative
effects. It is therefore attractive to postulate a model in which wg leads
to dephosphorylation of zw3, thereby inhibiting its enzymatic activity,
and activating those molecules that are normally repressed by zw3.
- arm has homology to the vertebrate catenins, proteins present in junctional
complexes (McCrea et al., 1991; Peifer & Wieschaus, 1990). Arm itself
seems to have similar properties, as it is associated with actin filaments
and a presumed Drosophila cadherin molecule, (Peifer, 1993) but a significant
fraction of the total arm protein is present in the cytoplasm. Apart from
the necessity of arm in wg signaling, the arm protein is also modified
by wg, resulting in a significant increase in its intracellular concentration.
With respect to the mechanism of arm modification by wg, it seems relevant
that arm is differentially phosphorylated, and that a hypophosphorylated
form of arm is specifically increased by wg (Peifer et al., 1994; Peifer
et al., 1994). The simplest model is that arm is directly phosphorylated
by zw3, and de-stabilized. Inhibition of zw3 by wg leads to stabilization
and increase of arm but, needless to say, additional steps could be involved.
It should be noted that the wg-induced increase in arm protein is a relatively
ubiquitous effect of wg, as it is seen wherever wg is expressed (Peifer
et al., 1991). In contrast, many other molecular effects of wg activity
are spatially restricted and require the input from additional signals.
How catenins function in signaling is not clear, perhaps they are involved
in sequestering signaling molecules or transcription factors within the
cytoplasm or perhaps they play a more permissive role in signaling by regulating
adhesive strength between cells.
- 5.5 Interactions between wg and Notch.
- Several recent studies have indicated that wg interacts with Notch,
a gene encoding a long transmembrane protein involved in many local cell
interactions (reviewed in Artavanis-Tsakonas & Simpson, 1991). Despite
the fact that wg and Notch do not have similar phenotypes, double mutant
combinations reveal strong genetic interactions. In one such study, the
phenotype of a weak Notch allele, called notchoid, was enhanced by a 50%
reduction in wg dosage, i.e., in flies heterozygous for a null allele of
wg (Hing et al., 1994). Conversely, Couso et al. (personal communication)
found that attenuated wg signaling is modified by Notch alleles. These
findings become even more intriguing by the observation that dsh also interacts
with Notch , during bristle formation in the wing margin (J. Axelrod and
N. Perrimon, personal communication). These interactions are based on genetics
and can perhaps be explained by wg and Notch acting in parallel pathways
with some shared components. On the other hand, it is possible that the
Notch and Wg proteins interact with each other and even that the Notch
protein is part of the Wg-receptor complex. Clearly, the available assays
for soluble Wg activity will shed further light on this matter.
- 6. In vitro assays for wg signaling .
- In spite of the power of genetics, biochemical and cell biological
assays are indispensable complementary tools in elucidating signal transduction
pathways. Recently, several in vitro assays for wg have been developed,
based on known downstream effectors of wg as a targets. Cumberledge &
Krasnow (1993) co-cultured wg producing cells, either isolated from embryos
or transfected S2 cells, with en positive cells from embryos and could
show that the expression of en is maintained if wg is provided. This effect
requires direct cell to cell contact, and is not brought about by the medium
from wg producing cells.
- Van Leeuwen et al (1994) studied the wg-dependent effect on arm, using
a cell line derived from Drosophila imaginal discs. They found that transfected
wg constructs, and significantly, also Wg from the culture medium, elevates
the concentration of the arm protein (Van Leeuwen et al., 1994). The accumulation
is due to an increased stability of the arm protein which was measured
to have a rapid turnover in the absence of wg. The active protein in the
medium is soluble and is in all likelihood wg itself, because antibodies
to wg can inhibit the effect. This novel assays for wg protein should allow
the purification of the protein in an active form, leading to answers to
such important questions as whether active wg protein is bound to another
molecule or made as a multimer. Such assays will also be essential in
defining the interaction between the wg protein and its receptor, whenever
receptor candidates become identified.
- An important conclusion emerging from this in vitro work is that the
wg protein acts as a growth factor. Active wg protein is secreted from
cells; a fraction of the secreted material is soluble in tissue culture
medium, and it can influence the differentiation pattern of target cells,
in all likelihood by binding to and activating a cell surface receptor.
Nonetheless, the wg protein in the extracellular matrix (ECM) is active
in the assay on imaginal disc cells (Van Leeuwen et al., 1994). In vivo,
the ECM could be a sink for wg activity, limiting its range of action by
sequestering the protein. The mammalian Wnt-1 protein retains at least
some of its activities when it is engineered to become a transmembrane
protein (Parkin et al., 1993), and it is possible that many effects of
Wnt proteins in intact animals are short ranged due to attachment of the
proteins to producing cells or to the ECM. In addition, components of the
ECM may play a more active role in wg signaling, presenting the protein
to the responding cells in a manner similar to fibroblast growth factors
and matrix molecules (c.f. Yayon et al., 1991).
- 7. Regulation of wg expression; possible autocrine regulation
- While this review is mainly concerned with events caused by the wg
protein, we will briefly summarize our current understanding of the regulation
of wg gene expression when subject to intercellular signaling (earlier,
wg expression is regulated by pair-rule genes). More extensive reviews
have been published previously by Hooper & Scott (1992) and by Klingensmith
& Perrimon (1991).
- In the extended germ band embryo, wg and en mutually regulate each
other's expression, so that in the absence of en, wg expression decays
(Martinez-Arias et al., 1988). A similar decay of wg expression is seen
in hedgehog mutant embryos (Ingham & Hidalgo, 1993). In contrast, embryos
mutant for the segment polarity gene patched (ptc) show ectopic and persistent
wg expression (DiNardo et al., 1988; Ingham et al., 1991; Martinez-Arias
et al., 1988). Based on double mutant combinations and ectopic expression
experiments, Ingham (1991) has proposed a model in which ptc act as a constitutive
negative regulator of wg expression and that hh expression from the en
cells relieves this inhibition. A particularly informative double mutant
is between ptc and hh, in which wg expression is the same as in ptc alone
(Ingham & Hidalgo, 1993; Ingham et al., 1991). ptc encodes a protein
with multiple transmembrane domains, suggesting that it acts as a receptor
(Hooper & Scott, 1989; Nakano et al., 1989). The ptc protein is indeed
present on the plasma membrane of cells (Taylor et al., 1993). The hh gene
also codes for a membrane protein, but one that can be proteolytically
processed and secreted (Lee et al., 1992; Mohler & Vani, 1992; Tabata
et al., 1992). By misexpressing Hh in imaginal discs, Basler & Struhl
(1994) found a similar relationship between hh and wg expression as in
the embryo, suggesting that the mechanism of control between these two
important signaling genes may be used throughout development.
- While the genetic interactions and the structure of the gene products
strongly suggest that hh is a ligand for patched, there is presently no
direct evidence to support this model. The hh molecule has received considerable
attention: it has been shown to act in a concentration dependent manner,
i.e. as a secreted morphogen, in the epidermis (Heemskerk & DiNardo,
1994). It is also involved in moving the morphogenetic furrow in the eye
imaginal disc (Heberlein et al., 1993; Ma et al., 1993) and has several
functions in other imaginal discs (Tabata & Kornberg, 1994). Significantly,
there is no correlation between hh expression in imaginal discs and wg
expression (Tabata & Kornberg, 1994), underscoring a point made before:
what happens in the embryo is only a subset of all possible regulatory
circuits and not necessarily the most conserved one. Several vertebrate
hh homologs have been cloned, some having spectacular expression patterns,
such as in the floor plate, the notochord and an area known in developing
limb buds as the zone of polarizing activity (Echelard et al., 1993; Krauss
et al., 1993; Riddle et al., 1993). Ectopic hh activity in limbs leads
to polarity reversals, strongly indicating that the secreted hh molecule
regulates limb polarity (Riddle et al., 1993).
- Among the intracellular effectors in the control of wg expression
are the protein kinase fused (Préat et al., 1990, which genetically
acts downstream of ptc [Ingham, 1993 #172) and the unknown product of costal-2
(Préat et al., 1993). The genes Cubitus interruptus Dominant (CiD)
(Orenic et al., 1990) and sloppy paired (slp) (Cadigan et al., 1994) may
act as target transcription factors working on the wg promoter, although
it is not clear whether the CiD gene product is a nuclear or cytoplasmic
protein (Orenic et al., 1990). slp appears to define the so-called wg competence
domain: by being expressed anterior to the en cells only it governs the
asymmetric response to the hh signal (Cadigan et al., 1994; Grossniklaus
et al., 1992). In accordance with this model, misexpression of slp leads
to ectopic expression of wg (Cadigan et al., 1994).
- Because of the signaling loop between the wg and the en cells, wg
is required for its own expression, in a paracrine way (Ingham & Hidalgo,
1993). From another study however, evidence has emerged that wg regulates
its own expression in a truly autocrine way, without the assistance of
adjacent cells (Li & Noll, 1993). This autocrine signal would act later
and is mediated by gooseberry (gsb) as a transcription factor (Li &
- 8. Other Wnt genes in Drosophila
- Besides wg, the genome of Drosophila harbors at least three other
members of the Wnt gene family, called DWnt-2 (Russell et al., 1992), DWnt-3
(or DWnt-5) (Eisenberg et al., 1992; Russell et al., 1992) and DWnt-4 (Y.
Graba and J. Pradel, pers. communication). Although true null mutants
to assess the functions of these genes are not available for any of them,
their expression patterns suggest important developmental functions. For
example, DWnt-2 has a highly dynamic expression pattern, with an early
metameric profile and later restriction to the gonads (Russell et al.,
1992), while DWnt-3/5 is expressed in the developing CNS (Russell et al.,
1992),. DWnt-3/5 encodes a rather unusual protein; compared to other Wnt
products it has long inserts and its primary translation product is 113
kD (Eisenberg et al., 1992; Russell et al., 1992). Both genes are initially
expressed in thoracic limb anlagen where DWnt-3/5 is regulated by Distalless
(Eisenberg et al., 1992), which is in turn itself under the control of
wg. One can envisage cascades of inductive factors such as Wnt proteins
governing local expression of homeobox genes, which in turn regulate the
expression of other secreted patterning molecules.
- DWnt-4 shows an expression pattern surprisingly similar to that of
wg; it is expressed in stripes just anterior to the en stripes, although
slightly later than wg. The gene maps very close (< 50 kB) to wg itself
(Y. Graba and J. Pradel, pers. communication). Presumably, wg and DWnt-4
are the product of a gene duplication, which likely has occurred early
in evolution because the two gene are significantly diverged in sequence.
The close linkage with wg complicates a thorough genetic analysis, as deficiencies
tend to take out both genes1 and complete loss of functions mutants at
DWnt-4 have not been demonstrated. A hypomorphic DWnt-4 allele nevertheless
has a segment polarity phenotype, indicating that DWnt-4 has a function
in the embryonic epidermis.
- It is tempting to speculate that DWnt-4 and wg interact, as they have
partially overlapping expression patterns, and an obvious mode of interacting
is to make a heterodimeric, disulfide linked protein. Whether this is the
case, or whether wg would make such dimers with other Wnt products has
to be resolved by further genetic and biochemical analysis.
- The existence of a Wnt gene family in Drosophila provokes many questions
as to their signaling pathway: will they interact with the same receptor;
do they utilize the known wg signaling components? As dsh and arm have
phenotypes similar to wg, it appears that they would themselves not be
involved in other signaling events, such as those potentially elicited
by other Wnt genes. That could imply that dsh and arm are also part of
a gene family, but while they both have multiple relatives in the mouse
genome, they seem to have no kin in the fly (J.K., unpublished).
- 9 Functions of wg homologs in other species
- One of the most profound biological insights obtained over the past
few years is that signaling pathways are conserved in evolution. Not only
the signaling molecules themselves show a high degree of identity between
phylogenetically distant species, but the hierarchy in which they act
can be surprisingly similar. In view of the high degree of evolutionary
conservation of Wnt molecules, it seems valid to ask whether the machinery
of Wnt signaling is conserved as well.
- The potential for a conserved pathway is present, as all of the known
components of wg signaling are present in the vertebrate genome. Up to
4 mouse dsh homologs have been cloned (D. Sussman, pers. comm); zw3 is
very similar the mammalian glycogen-synthase 3 (GSK-3ß) enzyme (Bourouis
et al., 1990; Davis & Joyner, 1988; Siegfried et al., 1990); arm has
counterparts in the form of ß catenin and plakoglobin (McCrea et
al., 1991; Peifer & Wieschaus, 1990); and two mouse en genes exist
(Davis & Joyner, 1988).
- For Wnt-1 and the two en genes, mouse null mutants have been generated,
which show a remarkably similar phenotype (McMahon & Bradley, 1990;
McMahon et al., 1992) (A. Joyner, pers. communication). In both mutants,
a significant portion of the midbrain is absent, in addition to the most
anterior part of the hindbrain. These are indeed the structures where Wnt-1
and en are normally expressed, although the genes are expressed in the
same area, unlike the adjacent expression domains in Drosophila. This suggests
that Wnt-1 and en are part of a regulatory pathway similar to that in
Drosophila (McMahon et al., 1992). It is not clear whether the other components
of wg signaling are involved in the Wnt-1/en regulation in the mouse CNS,
but the mouse Dsh-1 gene is expressed in the right site and time to act
as a signal transducer in this tissue (Sussman et al, submitted)
- Equally intriguing as these observations made in the mouse are the
phenotypes associated with Wnt expression in Xenopus embryos. Various Wnt
genes, when ectopically expressed in early Xenopus blastomers, induce the
formation of an additional body axis (McMahon & Moon, 1989; Smith &
Harland, 1991). These findings suggest that Wnt signaling is part of the
organizing activity in Xenopus. The recently isolated XWnt-11 gene would
seem to be a good candidate for an endogenous activity because it is expressed
in the expected pattern (Ku & Melton, 1993). Perhaps in line with the
supposition that Wnt signaling controls axis formation is the observation
that injection of an antibody to ß catenin in early Xenopus embryos
likewise induces a second body axis (McCrea et al., 1993). When the interaction
between wg and arm in Drosophila is recalled, this finding is somewhat
paradoxical: wg increases the levels of the arm protein in Drosophila and
inhibition of a ß catenin would therefore be expected to lead to
a phenotype opposite to that of Wnt injections. But not all Wnt genes in
Xenopus lead to axis duplications and dorsalized embryos--- some have ventralizing
effects and may inhibit dorsalizing signals (Christian & Moon, 1993)
and the particular catenin reacting with the antibody may be part of such
a ventralizing pathway.
- A further example of Wnt-arm/catenin interactions is provided by effects
of Wnt genes in cell culture. Transfection of Wnt-1 expression constructs
into several different cell lines - PC12 cells, AtT20 and C57MG cells-
leads to an increase in the intracellular amounts of ß catenin and/or
plakoglobin (Bradley et al., 1994; Hinck et al., 1994). Interestingly,
these cells become more adherent to each other, presumably due to the increase
in cadherin levels accompanying the ß catenin elevation (Bradley
et al., 1994; Hinck et al., 1994). Finally, it is of interest to note that
wg itself can act as a mammary oncogene, since expresssion of the gene
in mammary gland cell lines leads to the same morphological transformation
as Wnt-1 does (Ramakrishna & Brown, 1993).
- Thus, it seems warranted to expect that at least part of the wg signaling
mechanism is conserved in evolution and that newly identified wg signaling
components in Drosophila will have counterparts in vertebrates.
- 10. Concluding remarks
- Over the past years, Wnt genes have attracted considerable interest,
from developmental biologists and oncologists alike. In spite of the impressive
progress made in describing the biological roles of Wnt genes, deciphering
the signal transduction machinery of Wnt action has been slow. This was
mainly due to problems encountered in isolating active Wnt protein and
the few assays available for Wnt products. The homology between wg and
Wnt-1 led to expectations that the genetic tools developed in Drosophila
would solve some of the problems encountered in studying the mammalian
Wnt signaling mechanism. Did the promise of Drosophila and the genetics
of wg live up to the expectations? One the one hand, the answer must be
a resounding yes: a number of novel insights have been obtained, leading
to a framework of the wg signaling pathway. On the other hand, one of the
most desired components of wg signaling, the wg receptor has remained
an elusive goal. Additional genetic screens, for example for maternal effect
genes on the autosomes may uncover more links in the chain.
- A traditional weakness of Drosophila as a model system has always
been the limited usefulness of cultured cells. In studying the growth requirements
and biochemistry of signaling in mammalian cells, cultured cells have been
essential tools. Nonetheless, it has recently been shown that cells isolated
from the Drosophila embryo, or permanent cell lines from imaginal discs,
are able to respond to wg protein. Using these systems, the molecular events
suggested by the genetic interactions can now also be studied directly,
hopefully leading to filling the voids in the currently sketchy pathway.
- We thank the members of our lab for discussions and comments on the
manuscript and Marcel van den Heuvel for contributing to Figures 1 and
2.. Research in our lab is supported by the Howard Hughes Medical Institute,
of which RN is an investigator.
- Legends to Figures
- Figure 1.
- Spatial distribution of wingless antigen in wildtype and porcupine
- Confocal Laser Scanning Microscope views of whole mount preparations
stained with antibodies against wingless. The bars in panels B and D represent
5 mm; A and B show the same level of magnification, as do the higher-powered
details shown in C and D. Anterior is to the left and dorsal is uppermost
in all panels.(A) Wildtype embryo (late stage 10). The scan visualizes
the dorsal-lateral patches of wg staining along the germ band. Antigen
is distributed diffusely. Circular spaces devoid of staining correspond
to cell nuclei. (B) Germline-clone derived porcupine embryo (late stage
10). The scan shows the same areas as those shown in wildtype (A). Note
the confined localization of wg antigen. (C) Magnification of three stained
areas of the wildtype embryo shown in (A). The diffuse distribution of
antigen includes many strong dots of staining; the arrow points to one
such heavily labelled body. (D) Magnification of two stained areas of the
porc embryo shown in (B). The antigen seems to be confined to the cytoplasmic
area of wg-expressing cells, which occur in a domain of one to two cell
diameters in width. Note the absence of punctate staining outside the principle
cells of wg expression.
- Figure 2.
- Localization of wg protein and engrailed expression in various wild
type and mutant embryos. Some stocks contained balancer chromosomes with
a hunchback-LacZ construct and the genotypes of embryos could be determined
by the extent of ß-galactosidase staining.
- Whole mount embryos photographed with Differential Interference Contrast
- A. Wild type embryo , stage 10). Note the diffuse appearance of the
- B. Homozygous mutant (wgIL114 at 16°C) embryo, stage 10. wg staining
is restricted compared to wild type, but expression suffices for normal
- C. Homozygous mutant (stage 11) (wgIN67) embryo. wg staining is in
sharp bands, more intense than in wild type embryos.
- D. Heterozygous embryo (wgIN67/CyO hb-ß-gal) (stage 10). The
ß-gal staining is now limited to an area including only the first
abdominal segments. Note the strong, limited staining of wg .
- E. Wild type embryo , stage 10, double staining for wg and en. Note
the diffuse brown appearance of the wg staining.
- F. Heterozygous wg wgIN67/+ embryo, double staining for wg and en..
The wg staining seems confined, yet en is normally expressed.
- G. Ventral view of embryo carrying a LacZ P-element under the control
of the wg promoter. Expression of LacZ is in blue, expression of en is
detected with an antibody in brown.
- H. porc PB16 germline clone derived embryo (stage 10) stained with
anti-wg. Note the sharp appearance of the stripes, similar to the wg mutant
- Figure 3
- Model for the interactions between wg and en cells in the embryonic
epidermis. This model in based on genetic experiments and of a few biochemical
data. See the text for further explanation.
- The upper half shows two adjacent parasegments, with the positions
of the interacting wingless (wg) and engrailed (en) expressing cells. The
lower half shows these interactions in more detail.
- The wingless (Wg) protein is secreted or transported with the assistance
of the porcupine (Porc) gene product. Wg binds to a cell surface receptor
(WgR), yet to be identified. Within the target cell, the product of dishevelled
(Dsh) is the first known component in signaling, acting through an unknown
mechanism. Downstream of Dsh is a protein kinase, zeste-white 3 (Zw3),
which then leads directly or indirectly to the phosphorylation and inhibition
of the armadillo (Arm) protein. The reversal of this negative effect of
Zw3 by Wg leads to higher levels of the Arm protein, which is itself also
necessary to transmit Wg signals to more downstream targets, such as the
expression of engrailed (En).
- The expression of the secreted protein hedgehog (Hh) is necessary
to relieve the negative effect of patched (Ptc) on the expression of Wg.
Within the wg cells, fused (Fu, a protein kinase) and costal-2 (Cos-2)
relay the signal from Ptc to the nucleus. Cubitus-interruptus Dominant
(CiD) and sloppy paired (Slp) may control Wg transcription.
- Wg may also regulate its own expression in an autoregulatory loop,
via the homeobox protein gooseberry (Gsb).
- Figure 4.
- Segmentation in wild type, wg, and naked embryos.
- The embryo and its resulting cuticle are shown for wildtype (A, D),
wingless (B,E), and naked (C,F) genotypes. The wingless phenotype is observed
in wingless, porcupine, dishevelled, and armadillo mutants. The naked phenotype
is shown by naked and zeste-white 3 mutants, and by embryos expressing
wg ubiquitously (HS-wg).Panels A-C: Scanning electron micrographs of embryos
midway through development (10-11 hours), shortly after germband shortening
(lateral views, anterior at top and ventral to the left). (A) In the wildtype
embryo, each thoracic (t) and abdominal (a) segment is delimited by a deep
segmental furrow. The small arrow points to a tracheal pit, one of which
occurs in the middle of the segment on either side of the embryo. The posterior
spiracle is also indicated (Sp). (B) In the wingless embryo, segmental
furrows are completely absent. Tracheal pits fuse into large holes (arrow).
In the head, the maxillary appendage (open arrowhead) is distinct but adjacent
tissue shows signs of the necrosis which will consume most of the head
structures. Posterior spiracles are absent. (C) In the naked mutant, segment
borders are present but are often incomplete and irregular, with partial
fusion of some segments. Tracheal pits (small arrow) do not fuse and the
posterior spiracle (large arrow) is present. Panels D-F: Darkfield micrographs
of cuticle preparations from terminally differentiated embryos -ventral
views, anterior at top. (D) In wildtype, each segment in the thorax (t)
and abdomen (a) is decorated with a denticle belt, with denticles in the
anterior and naked cuticle in the posterior of each segment. The arrows
show the extent of one segment, delimited by segment furrows. The posterior
spiracles are elaborated into the Filzkorper (fk). (E) wingless embryos
display a lawn of denticles devoid of naked cuticle. There are no segmental
divisions although the denticles are not uniform in polarity. No Filzkorper
structures form. (F) naked embryos show a phenotype largely opposite to
that of wingless, with a naked cuticle showing very few if any denticles
(none in this case). Disorderly furrows crease the cuticle, and the Filzkorper
(fk) is present though abnormal.
- All embryos are shown at approximately the same magnification, except
(D) which is shown at 50% relatively to the others.
- Artavanis-Tsakonas, S., & Simpson, P. (1991). Choosing a cell fate:
a view from the Notch locus. Trends Genet. 7, 403-408.
- Baker, N. E. (1987). Molecular cloning of sequences from wingless,
a segment polarity gene in Drosophila: The spatial distribution of a transcript
in embryos. EMBO J. 6, 1765-1773.
- Baker, N. E. (1988a). Embryonic and imaginal requirements for wingless,
a segment-polarity gene in Drosophila. Dev Biol. 125, 96-108.
- Baker, N. E. (1988b). Transcription of the segment-polarity gene wingless
in the imaginal discs of Drosophila, and the phenotype of a pupal-lethal
wg mutation. Development. 102, 489-497.
- Basler, K., & Struhl, G. (1994). Compartment boundaries and the
control of Drosophila limb pattern by the hedgehog protein. Nature. 368,
- Bejsovec, A., & Martinez-Arias, A. (1991). Roles of wingless in
patterning the larval epidermis of Drosophila. Development. 113, 471-485.
- Bejsovec, A., & Wieschaus, E. (1993). Segment polarity gene interactions
modulate epidermal patterning in Drosophila embryos. Development. 119,
- Bienz, M. (1994). Homeotic genes and positional signalling in the Drosophila
viscera. Trends Genet. 10, 22-26.
- Bourouis, M., Moore, P., Ruel, L., Grau, Y., Heitzler, P., & Simpson,
P. (1990). An early embryonic product of the gene shaggy encodes a serine/threonine
protein kinase related to the CDC28/cdc2+ subfamily. EMBO J. 9, 2877-2884.
- Boyle, W. J., Smeal, T., Defize, L. H., Angel, P., Woodgett, J. R.,
Karin, M., & Hunter, T. (1991). Activation of protein kinase C decreases
phosphorylation of c-Jun at sites that negatively regulate its DNA-binding
activity. Cell. 64, 573-84.
- Bradley, R., Cowin, P., & Brown, A. (1994). Expression of Wnt-1
in PC12 cells results in modulation of plakoglobin and E-cadherin and increased
cellular adhesion. J. Cell Biol. in press,
- Bradley, R. S., & Brown, A. M. (1990). The proto-oncogene int-1
encodes a secreted protein associated with the extracellular matrix. EMBO
J. 9, 1569-1575.
- Brown, A. M., Papkoff, J., Fung, Y. K., Shackleford, G. M., & Varmus,
H. E. (1987). Identification of protein products encoded by the proto-oncogene
int-1. Mol Cell Biol. 7, 3971-3977.
- Bryant, P. J. (1993). The polar coordinate model goes molecular. Science.
- Cabrera, C. V., Alonso, M. C., Johnston, P., Phillips, R. G., &
Lawrence, P. A. (1987). Phenocopies induced with antisense RNA identify
the wingless gene. Cell. 50, 659-63.
- Cadigan, K., Grossniklaus, U., & Gehring, W. J. (1994). Localized
expression of sloppy paired maintains the polarity of Drosophila parasegments.
Genes&Dev. 8, 899-913.
- Campbell, G., Weaver, T., & Tomlinson, A. (1993). Axis specification
in the developing Drosophila appendage: The role of wingless, decapentaplegic,
and the homeobox gene aristaless. Cell. 74, 1113-1123.
- Chakrabarti, A., Matthews, G., Colman, A., & Dale, L. (1992). Secretory
and inductive properties of Drosophila wingless protein in Xenopus oocytes
and embryos. Development. 115, 355-369.
- Chou, T.-B., Noll, E., & Perrimon, N. (1993). Autosomal P[ovoD1]
dominant female-sterile insertions in Drosophila and their use in generating
germ-line chimeras. Development. 119, 1359-1369.
- Christian, J. L., & Moon, R. T. (1993). Interactions between Xwnt-8
and Spemann organizer signaling pathways generate dorsoventral pattern
in the embyronic mesoderm of Xenopus. Genes & Dev. 7, 13-28.