Download Twisted Perspective: New Insights into Extracellular Modulation of

Cell, Vol. 104, 801–804, March 23, 2001, Copyright 2001 by Cell Press
Twisted Perspective: New Insights
into Extracellular Modulation of
BMP Signaling during Development
Robert P. Ray and Kristi A. Wharton*
Department of Molecular Biology, Cell Biology
and Biochemistry
Division of Biology and Medicine
Box G-J160
Brown University
Providence, Rhode Island 02912
During the past several decades, considerable research
efforts have elucidated the central components of many
signal transduction pathways in animal systems. More
recently attention has shifted to the importance of modulating intercellular signaling processes during development, as it is clear that fine-tuned regulation of signaling
systems in vivo results in subtle distinctions at the cellular level. Extracellular modulators have been a particular
focus of investigation, including proteins that affect receptor stability, ligand function, or ligand availability. In
these latter types of regulation, proteins such as proteoglycans, heparan sulfate modifying enzymes, and proteases have been recognized as key players, as have
a diverse class of diffusible modulators that influence
ligand activity by inhibiting or facilitating its function
(see references contained within reviews by Capdevila
and Belmonte, 1999; Christian, 2000). The topic of this
review focuses on the function of one such diffusible
modulator that influences the availability of BMP ligands,
a family of the TGF-␤ superfamily of signaling molecules.
It appears that for TGF-␤ signaling in general a major
means of signal regulation is the modulation of ligand
availability and as yet this system provides one of the
best studied paradigms for this kind of regulation in
animal systems.
In recent years, our understanding of the mechanism
of signaling by TGF-␤ superfamily ligands has increased
tremendously. A little over a decade ago, the few TGF-␤like proteins that had been characterized were “orphan”
ligands for which neither receptors nor cellular targets
had been identified. The discovery of the TGF-␤ receptors, followed shortly thereafter by the characterization
of the founding member of the Smad family of transcriptional regulators, the Drosophila gene Mothers against
dpp (Mad), initiated an intensive period of research focused on the downstream components in the pathway.
These studies revealed a relatively simple signal transduction system in which the ligand dimer binds to a
heteromeric complex of Type I and Type II transmembrane serine/threonine kinase receptors that phosphorylate cellular transcription factors of the Smad family
that directly move into the nucleus and regulate target
gene expression (Massagué, 1998).
More recently, studies on a number of model systems
have demonstrated that an important and evolutionarily
conserved mechanism of regulation for one family of
* To whom correspondence
[email protected]).
should
be
addressed
(e-mail:
Minireview
TGF-␤-like ligands, the Bone Morphogenetic Proteins
(BMPs), is the direct modulation of ligand activity by
extracellular factors (reviewed in De Robertis and Sasai,
1996; Figure 1). In amphibians, opposing activities of
BMP4 and its antagonist Chordin (Chd) are responsible
for subdivision of the embryonic dorsal–ventral axis. A
similar role for BMP4 and Chd orthologs has been implicated in dorsal–ventral patterning of the zebrafish embryo (Mullins, 1998). Thus, in the vertebrates, BMP activity specifies ventral fates in the early embryo, and is
antagonized by localized expression of Chd orthologs
dorsally. In Drosophila, the same relationship exists between the BMP homologs decapentaplegic (dpp) and
screw (scw) and the Chd ortholog short gastrulation
(sog), but the axis is inverted: dpp and scw specify dorsal
fates in the embryo and are antagonized by sog, which
is expressed ventrally. In both vertebrates and invertebrates, the activity of Chd orthologs is antagonized by a
family of metalloproteases including Drosophila Tolloid
(Tld), Xenopus Xolloid (Xol), Human BMP-1, and their
orthologs in other vertebrates. Biochemical studies on
the Xenopus, Drosophila, and mouse proteins have
demonstrated that Tld/Xol/BMP1 cleavage inactivates
Sog/Chd and promotes BMP signaling (Marqués et al.,
1997; Piccolo et al., 1997; Scott et al., 1999). Thus, Tld/
Xol proteins function as BMP agonists.
In three papers appearing this month in Nature (Chang
et al., 2001; Ross et al., 2001; Scott et al., 2001), studies
on the Drosophila gene twisted gastrulation (tsg) and
its vertebrate orthologs reveal another level of complexity in the extracellular modulation of BMP signaling during development. tsg was first identified in Drosophila as
a zygotic lethal gene that, based on its mild ventralized
phenotype and its localized requirement to dorsal blastomeres, was allied with dpp, scw, tld, and other zygotic
ventralizing genes (Ray et al., 1991 and references therein).
Subsequent molecular characterization revealed tsg encodes a secreted, cysteine-rich protein that was proposed to act in combination with dpp to specify the
dorsal-most pattern elements in the embryo (Mason et
al., 1994). This role for tsg as a dpp agonist was supported by a study on the Xenopus ortholog, xTsg, in
which overexpression of xTsg was shown to result in a
loss of neural and head structures, and a reduction in
dorsal expressed molecular markers. These results led
to the conclusion that xTsg promotes BMP signaling
(Oelgeschläger et al., 2000). In apparent contradiction
to this earlier work, the current studies instead show
that Tsg orthologs can function in combination with Chd/
Sog to antagonize BMP signaling.
Tsg orthologs have now been isolated from Drosophila, Xenopus, zebrafish, chick, mouse, and humans. Drosophila and zebrafish each have two Tsg homologs, and
based on sequence comparisons, the pairs appear to
have arisen by independent gene duplication events in
these two species (Ross et al., 2001; Scott et al., 2001).
Studies on both Drosophila and Xenopus indicate that
Tsg acts nonautonomously and is highly diffusible (Mason et al., 1997; Chang et al., 2001). Tsg does not appear
to play a role in establishing asymmetries in the early
Cell
802
Figure 1. A Conserved Mechanism of Extracellular Signals Establishes Dorsal–Ventral Pattern in Invertebrates and Vertebrates (for
references, see De Robertis and Sasai, 1996)
(Top) Schematic frontal section of a Drosophila blastoderm embryo
showing dorsal–ventral fates. Sog, expressed in the ventral ectoderm (gold) diffuses dorsally and antagonizes BMP signaling by
Dpp and Scw in the dorsal ectoderm (blue). Tld, also expressed
dorsally, antagonizes Sog’s antagonism and thus promotes BMP
signaling. (Bottom) Schematic fate map of a Xenopus embryo showing molecular markers used in the current studies. The Sog ortholog
Chd is expressed dorsally in the organizer and antagonizes BMP
signaling, confining it ventrally. As in Drosophila, Chd’s action is
antagonized by the Tld ortholog Xol. The same fate map and relationship between BMP2b and zchordin applies to the zebrafish embryo.
While the double negative regulation and the proteins involved are
conserved, the axis is inverted: Sog is expressed ventrally in Drosophila and antagonizes BMP signaling dorsally, and Chd is expressed dorsally and antagonizes BMP signaling ventrally.
embryo (i.e., like Chordin) as in most organisms the tsg
RNA is not specifically localized or does not require
specific localization. Rather, Tsg acts generally or depends on the localized activities of other components
in the pathway.
The recent loss-of-function and gain-of-function analyses support the notion that Tsg acts as an antagonist of
BMP signaling in early embryogenesis. Loss-of-function
studies in Drosophila and zebrafish indicate that Tsg
inactivation produces a phenotype resembling that of
mutations in the BMP antagonists sog and chordin, respectively, rather than that expected from a loss of BMP
function. In Drosophila, the sog and tsg mutant phenotypes exhibit the same effects on the expression patterns of a number of genes that respond to the BMP
gradient, and these phenotypes are distinct from those
produced by dpp, tld, or scw mutants (Ross et al., 2001).
Thus, given that studies on sog have demonstrated unequivocally that it functions as a dpp antagonist, it follows that tsg, which produces the same phenotype,
might also act in this capacity. In zebrafish, a similar
result is seen. Using morpholino oligonucleotides, it has
been shown that the phenotype associated with the loss
of ztsg1 activity consists of an expansion of ventral
mesodermal tissues, a loss of paraxial mesoderm, and
a mild reduction of anterior ectoderm, i.e., a ventralized
phenotype, consistent with a role for ztsg1 in antagonizing BMP signaling (Ross et al., 2001).
Gain-of-function studies in Xenopus, zebrafish, and
Drosophila are consistent with the loss-of-function data
and also support the notion that Tsg functions as a BMP
antagonist. Microinjection of Tsg RNA into Xenopus embryos results in a dorsalization of the embryo as evidenced by loss of ventral cell fate markers such as
Xhox3, Msx1, and Vent1 in gastrula and loss of blood
and the heart marker Nkx2.5 in the tadpole (Chang et
al., 2001). In ectodermal explants (animal caps), overexpression of Tsg results in suppression of epidermal
keratin and induces the expression of cement gland and
neural markers, responses that are normally associated
with an inhibition of BMP signaling (Chang et al., 2001).
Similarly, in zebrafish, injection of ztsg1 RNA results in
a phenotype resembling that of a mutant for the BMP7
homolog, snailhouse, consistent with a dorsalization of
the embryo (Ross et al., 2001). Thus, in the vertebrate
embryo, loss of Tsg results in ventralization of the embryo, while overexpression leads to a dorsalization. Significantly, overexpression of Tsg in the Drosophila embryo has no effect on embryonic development, and
misexpression of Tsg in the wing disk produces either
no phenotype or only a mild venation defect (Ross et
al., 2001; Yu et al., 2000). Thus, the vertebrate embryo
either shows a sensitivity to Tsg that is not apparent in
Drosophila, or microinjection experiments in vertebrate
embryos may reveal aspects of Tsg function that cannot
be uncovered using the genetic tools of Drosophila.
The conclusion, based on mutant phenotypes, that
Chd/Sog and Tsg both act as BMP antagonists is supported by biochemical studies. Previous studies have
shown that Chd/Sog acts to antagonize BMP signaling
by binding BMP4/Scw and preventing the ligand from
interacting with its receptor (as reviewed in Podos and
Ferguson, 1999). Coimmunoprecipitations of tagged
Tsg, Chd, and BMP proteins indicate that Chd and Tsg
are both able to bind directly to BMP, and, more importantly, that Tsg promotes the binding of Chd to BMP
(Oelgeschläger et al., 2000; Chang et al., 2001; Ross et
al., 2001; Scott et al., 2001). This effect is particularly
clear in Drosophila where it has been shown that on its
own Sog can antagonize Scw signaling but not Dpp
signaling, while, in the presence of Tsg, Sog can antagonize both Dpp and Scw (Ross et al., 2001). Furthermore,
Tsg has been shown to bind Chd/Sog and the data from
both systems support the hypothesis that Tsg, Chd,
and BMP form a ternary complex that inactivates BMP
signaling more effectively than Chd alone (Figure 2).
A number of in vivo experiments indicate that Tsg and
Chd act synergistically to antagonize BMP signaling.
Loss-of-function studies on zebrafish show that inactivation of both Tsg and Chd results in a higher penetrance
of the mutant phenotype than with either alone (Ross
et al., 2001). Similarly in Xenopus, coinjection of both
Tsg and Chd is more potent in the induction of secondary axes than either alone (Scott et al., 2001), and the
combination is able to repress expression of Xbra
(Chang et al., 2001). In Drosophila, overexpression of
either Tsg or Sog alone is unable to suppress wing
Minireview
803
Figure 2. Summary of Binding and Functional Studies on Tsg
Demonstrated in vitro binding relationships are indicated in red,
demonstrated functional relationships in black, and potential functional relationships in gray. The central triangle illustrates the “ternary complex” of Tsg/Chd/BMP (Chang et al., 2001; Ross et al.,
2001; Scott et al., 2001). Potential functions of Tsg include direct
binding and inactivation of BMP (Ross et al., 2001), cooperative
binding and inactivation of BMP in combination with Sog/Chd
(Chang et al., 2001; Ross et al., 2001; Scott et al., 2001), and enhancement of Sog/Chd cleavage by Tld (Scott et al., 2001). In the
latter case, it is not clear if the new Tsg-dependent cleavages activate Sog/Chd to produce “supersog-like” molecules, or inactivate
it as suggested by the function of Tld in vivo.
phenotypes associated with ectopic expression of Dpp,
while the combination of the two is able to suppress
these phenotypes. Similarly, overexpression of either
Sog or Tsg alone in the wing disk results in only minor
venation defects, but the two together repress the endogenous dpp functions in the disk resulting in a phenocopy of dpp mutant wing phenotypes (Yu et al., 2000;
Ross et al., 2001). Notably, this result is consistent with
the biochemical data indicating that Sog is only able to
inactivate Dpp in the presence of Tsg.
Interactions between Chd/Sog and BMP are mediated
by two of the four cysteine-rich domains in the Chd
protein, CR1 and CR3, though, in isolation, these domains bind BMP with a 10-fold lower affinity than fulllength Chd (Oelgeschläger et al., 2000, and references
therein). These domains have weak homology with the
cysteine-rich domains in Tsg, and in Xenopus have been
shown to be responsible for the interaction with BMP4
(Oelgeschläger et al., 2000). Significantly, the CR1 and
CR3 domains of Chd also mediate interactions with Tsg.
Both CR1 and CR2/CR3 fragments of mouse Chd coimmunopreciptate with Tsg, and Tsg enhances the coimmunoprecipitation of CR1 and BMP4 (Scott et al., 2001).
Taken together, these results indicate that the cysteinerich domains represent protein–protein interaction motifs that appear to mediate the associations between
Tsg, Chd, and BMP. In contrast to the results with mouse
CR1, Xenopus CR1 failed to form a ternary complex with
BMP and Tsg in crosslinking experiments (Oelgeschläger et al., 2000). Thus, while the CR domains appear to
mediate interaction between Tsg, Chd, and BMP, the
specifics of their role in ternary complex formation has
yet to be resolved.
Biochemical studies on mouse and Drosophila proteins suggest that one function of Tsg in the Tsg/Chd/
BMP complex may be to promote cleavage of Chd by
Tld. Cleavage of Chd by Tld has been documented in a
number of different species, and the data support the
model that Tld activity inactivates Chd and thus releases
the active BMP ligand dimer (Marqués et al., 1997; Piccolo et al., 1997; Scott et al., 1999). Tld cleavage sites
have been mapped roughly in Drosophila and Xenopus
and precisely in mouse, and despite the conservation
between these systems, the sites do not all fall in the
same location of the Chd/Sog proteins (Figure 3). In
mouse, two cleavage sites have been documented, one
just after the CR1 domain, and one just after the CR3
domain, producing fragments of 15, 83, and 13 kDa
(Scott et al., 1999), similar fragments have been documented in Xenopus (Piccolo et al., 1997). In Drosophila,
there is an analogous cleavage site just after CR1, but
two other sites are documented, one before the CR2
domain and one just after, producing fragments of 20,
55, 20, and 25 kDa (Marqués et al., 1997). Assuming that
the mobility of these fragments is not altered by modifications such as glycosylation, neither of the C-terminal
sites in Drosophila corresponds to the second vertebrate site. The addition of Tsg alters the pattern of Chd/
Sog cleavage in both mouse and Drosophila. In the mouse,
four Tld homologs have been identified, and these differentially affect Chd cleavage in the presence and absence of Tsg (Scott et al., 2001). BMP-1 and mTLL-1
cleave Chd at the documented sites in the absence of
Tsg, while in the presence of Tsg an additional site
becomes prominent that subdivides the 84 kDa fragment into two fragments of 29 and 65 kDa. Notably,
this Tsg-dependent site corresponds to the second Tld
cleavage site in Drosophila Sog. A third Tld-like protein,
mTLD does not cleave Chd at all in the absence of Tsg,
but if Tsg is present, Chd is cleaved in the same way
as with BMP-1 and mTLL-1. The fourth Tld-like protease,
mTLL-2, is incapable of cleaving Chd either in the presence or absense of Tsg. In Drosophila, Sog cleavage is
also altered in the presence of Tsg but the precise location of the altered cleavages is not known (Yu et al.,
2000). The mouse Tsg-dependent Tld cleavage site of
Chd just before the CR2 domain coincides with one
of the reported Tld cleavage sites in Drosophila Sog
Figure 3. Cleavage Patterns of Mouse Chd
and Drosophila Sog
The N-terminal signal sequence is shown as
a black box, and the four cysteine-rich (CR)
repeats are shown as white boxes. Tld cleavage sites in the absence of Tsg are shown in
blue, Tsg-dependent sites are shown in red.
According to the current data, vertebrates
and invertebrates share common sites just
after the CR1 (Piccolo et al., 1997; Scott et al., 1999; Marqués et al., 1997) domain and a Tsg-dependent site just before the CR2 domain
(Scott et al., 2001). The final C-terminal site is different between the two: the vertebrate site falls just after the CR3 domain, while the invertebrate
site falls just after the CR2 domain (Scott et al., 1999; Marqués et al., 1997).
Cell
804
(Marqués et al., 1997). The mapping of the Drosophila
Tld cleavage sites was done in tissue culture cells and
thus, it is possible that cleavage at the site before CR2
may have been due to endogenous Tsg expression in
this cell line.
Given the in vivo data indicating that Tsg can act as
an antagonist of BMP signaling during development, the
fact that the Tsg/Chd/BMP complex promotes cleavage
by Tld is somewhat paradoxical. How is it that Tsg can
act as a BMP antagonist, and yet do so by promoting
the activity of Tld, a known BMP agonist? While a precise
answer to this question cannot be made based on the
available data, it must be kept in mind that while the
biochemical data indicates that Tsg can form complexes
with BMP and with Chd/BMP, as yet, these complexes
have not been ascribed to a particular Tsg function. Is
BMP inactive when complexed with Tsg, indicating that
Tsg can act as an antagonist independent of Chd? Does
Tsg binding to Chd/BMP act in some cases to lock Chd
onto BMP while in other cases to promote Tld cleavage
and thus release BMP? Do the novel Chd cleavage fragments produced in the presence of Tsg actually serve
to inactivate Chd, or do they act as superrepressors,
i.e., “superchordins” by analogy with the “supersogs”
(Yu et al., 2000) that have been generated in vitro in
Drosophila? Finally, is it possible that these different
complexes and fragments have distinct and perhaps
even opposing functions depending on the context in
which they were generated?
The studies on Xenopus indicate that Tsg may be
multifunctional, as it has been proposed to be both a
BMP agonist and antagonist (Oelgeschläger et al., 2000;
Chang et al., 2001; Scott et al., 2001). In this case, similar
types of experiments done in different laboratories under slightly different conditions produced results that
suggest Tsg may exhibit opposing functions in development. While the resolution of the different results awaits
further analyses, one potentially important criterion that
arises from these data is that the relative levels of Tsg
and Chd may dictate the nature of Tsg’s behavior—at
least in Xenopus. For example, it has been shown that
the ability for Chd to induce secondary axes when injected in ventral blastomeres can be enhanced by addition of Tsg, but only if the ratio of Tsg:Chd is less than
1:1. If the ratio of Tsg:Chd is 2:1 or greater, the enhancement of secondary axis formation is inhibited (Chang et
al., 2001; Ross et al., 2001). This result seems to indicate
that the types of interactions that Tsg may be involved
in depend to a large degree on the relative levels of Tsg
and Chd and perhaps a variety of other proteins involved
in extracellular modulation of BMP signaling (e.g., Tld).
In Drosophila, issues such as relative concentrations
of Sog and Tsg or other extracellular factors may also
account for the paradox of how both sog and tsg act
as antagonists yet promote peak levels of BMP signaling
in the dorsal cells of embryo. More detailed studies are
required to resolve these issues.
These recent studies have highlighted the importance
of extracellular modulation of ligand availability, a once
underappreciated means of regulating cell fate specification. The current work clearly demonstrates that Tsg
can act as a BMP antagonist, yet the full extent of activities that Tsg possesses in the context of the developing
organism remains to be clarified. Considering the cur-
rent models for patterning events mediated by BMP
signaling, there does not appear to be a requirement
for an additional antagonist, in particular, since Tsg does
not function simply as a cofactor for Chd. Indeed this
fact alone indicates that the complete function or functions of Tsg are likely more complex. Added information
on the importance of the relative levels of Tsg and Chd/
Sog, as well as on the functions of the different Chd/
Sog cleavage products, will be required to bring us to
the end of this long and twisted road.
Selected Reading
Capdevila, J., and Belmonte, J.C.I. (1999). Curr. Opin. Genet. Dev.
9, 427–433.
Chang, C., Holtzman, D.A., Chau, S., Chickering, T., Woolf, E.A.,
Holmgren, L.M., Bodorova, J., Gearing, D.P., Holmes, W.E., and
Brivanlou, A.H. (2001). Nature 410, 483–486.
Christian, J.L. (2000). Curr. Opin. Cell Biol. 12, 244–249.
De Robertis, E.M., and Sasai, Y. (1996). Nature 380, 37–40.
Marqués, G., Masacchio, M., Shimell, M.J., Wünnenberg-Stapleton,
K., Cho, K.W.Y., and O’Connor, M.B. (1997). Cell 91, 417–426.
Mason, E.D., Konrad, K.D., Webb, C.D., and Marsh, J.L. (1994).
Genes Dev. 8, 1489–1501.
Mason, E.D., Williams, S., Grotendorst, G.R., and Marsh, J.L. (1997).
Mech. Dev. 64, 61–75.
Massagué, J. (1998). Annu. Rev. Biochem. 67, 753–791.
Mullins, M.C. (1998). Trends Genet. 14, 127–129.
Oelgeschläger, M., Larrain, J., Geissert, D., and De Robertis, E.M.
(2000). Nature 405, 757–763.
Piccolo, S., Agius, E., Lu, B., Goodman, S., Dale, L., and De Robertis,
E.M. (1997). Cell 91, 407–416.
Podos, S.D., and Ferguson, E.L. (1999). Trend Genet. 15, 396–402.
Ray, R.P., Arora, K., Nüsslein-Volhard, C., and Gelbart, W.M. (1991).
Development 113, 35–54.
Ross, J.J., Shimmi, O., Vilmos, P., Petryk, A., Kim, H., Gaudenz, K.,
Hermanson, S., Ekker, S.C., O’Connor, M.B., and Marsh, J.L. (2001).
Nature 410, 479–482.
Scott, I.C., Blitz, I.L., Pappano, W.N., Imamura, Y., Clark, T.C., Steiglitz, B.M., Thomas, C.L., Maas, S.A., Cho, K.W.Y., and Greenspan,
D.S. (1999). Dev. Biol. 213, 283–300.
Scott, I.C., Blitz, I.L., Pappano, W.N., Maas, S.A., Cho, K.W.Y., and
Greenspan, D.S. (2001). Nature 410, 475–478.
Yu, K., Srinivasan, S., Shimmi, O., Biehs, B., Rashka, K.E., Kimelman,
D., O’Connor, M.B., and Bier, E. (2000). Development 127, 2143–
2154.