| |
- Gli proteins mediate transcriptional control by Hedgehog signaling
in vertebrates
- In contrast to the fly, vertebrates have not one but three Ci-related
proteins, designated Gli1, Gli2, and Gli3 (after glioblastoma),
all of which have been implicated in mediating the activities of
the various Hh proteins. The situation in vertebrates is further
complicated by the fact that the genes encoding at least two of
these, Gli1 and Gli3, are themselves transcriptional targets of
Hh signaling: Gli1 is activated and Gli3 repressed in response to
Hh signals (Marigo et al. 1996b; Lee et al. 1997). As usual, genetics
has been a powerful tool in addressing the activities of these proteins,
although as in the fly, the complexity of Gli function has complicated
the interpretation of such analyses. Thus, for instance, although
studies of mutant combinations in the mouse have shown genetic interactions
between Gli1 and Gli2 in the diencephalic region of the brain (Park
and Bai 2000), Gli2 and Gli3 in skeletal development (Mo et al.
1997), Gli1 and Gli2, and Gli2 and Gli3 in lung development (Motoyama
and Liu 1998; Park and Bai 2000), and Gli2 and Gli3 in tooth development
(Hardcastle et al. 1998), the mechanistic action of the different
Gli proteins in each of these organs remains unclear. Such mutant
analyses have been complemented by studies in cell culture. Interpretation
of these data, however, is also problematic, not least because in
several cases, cells that are not responsive to Hh signals have
been used, whereas in other instances, the experiments have not
addressed the response of Gli proteins to the presence or absence
of Hh ligand. Given the many levels of posttranscriptional regulation
of Ci activity, in particular the Hh-dependent suppression of cleavage
that alters the balance between repressor (cleaved) and activator
(uncleaved) forms, conclusions from these experiments should be
treated with some caution. These studies have included the generation
of modified forms of Gli proteins that reveal potent activities.
N-terminal truncation of both Gli2 and Gli3, for example, generates
strong activating forms of these proteins (Dai et al. 1999; Sasaki
et al. 1999), but because of the paucity of antibodies against the
endogenous proteins, the in vivo relevance of these truncated forms
remains unclear. One novel approach that has been used to investigate
the regulation of Gli protein activities and their relative contributions
to transducing Hh signals, has been to express each in Drosophila
imaginal discs (von Mering and Basler 1999; Aza-Blanc et al. 2000).
Although a heterologous system, the genetics of this system can
be rigorously controlled to a degree impossible in vertebrates,
and the activities of each protein can be compared with the known
activities of Ci. These studies have led to the following predictions:
(1) Gli1 and Gli2 are Hh-dependent activators of Hh targets; (2)
Gli2 is also an Hh-independent repressor; (3) Gli3 is an Hh-dependent
repressor; and (4) similar to Ci, proteolytic cleavage of Gli2 and
Gli3 generates repressor forms of each protein. To what extent do
these predictions match up to data in vertebrate systems? [Ingham,
2001]
- Notwithstanding the caveats noted above, there is overwhelming
evidence in support of a role for Gli1 as an activator of Hh target
gene transcription, although whether or not this activity is regulated
directly by Hh signaling in vivo is not so clear. Overexpression
of Gli1 in tissue culture cells or transgenic embryos can induce
transcription of Hh target genes in the absence of Hh activity (Hynes
et al. 1997; Sasaki et al. 1997; Ruiz i Altaba 1998, 1999), but
unlike Ci, Gli1 does not appear to undergo proteolytic cleavage
and does not bind to or require CBP for its activating function
(Dai et al. 1999). Moreover, again in contrast to Ci, transcription
of Gli1 is under Hh control; indeed, up-regulation of Gli1, like
Ptc1, is diagnostic of Hh signaling within a target field. The fact
that Gli1 is activated in response to an Hh signal strongly implies
that Gli1 is not the initial transducer of the Hh signal. Indeed,
mutants that lack the zinc finger domain and show no activating
capacity are homozygous viable, fertile, and show no characteristic
Hh phenotypes (Park and Bai 2000). Nevertheless, evidence from studies
in flies and in tissue culture suggest that Gli1 activity can be
potentiated by Hh signaling (Dai et al. 1999); consistent with this
suggestion, Gli1 has been shown to interact directly with the vertebrate
Su(fu) protein, an association that can inhibit its nuclear import
and abrogate its ability to activate transcription (Ding et al.
1999; Kogerman,
1999). Full-length Gli2 can also activate transcription of Hh
target genes in vertebrate tissue culture cells, but in this case
the activation depends on Hh stimulation. This requirement can be
overridden by deletion of the N-terminal domain of the protein,
which turns Gli2 into a strong constitutive activator capable of
inducing normally Shhdependent cell types when expressed in the
mammalian brain (Sasaki et al. 1999). This apparent negative regulatory
domain of the protein corresponds to the region of Ci that mediates
the interaction with Su(fu). Direct evidence for an interaction
between Gli2 and the vertebrate Su(fu) has not been presented, nor
is there any evidence for an in vivo cleavage of the N-terminal
region of Gli2; however, the activity and nuclear import of Gli2
is reportedly stimulated by the human Fused homolog (Murone et al.
2000), which presumably could act by abrogating the inhibitory effect
of Su(fu). In vivo evidence for an activating role of Gli2 comes
from mouse mutants lacking the zinc finger regions (Ding et al.
1998; Matise et al. 1998). Absence of Gli2 leads to a complete loss
of floor plate and a reduction in V3 interneurons, cell types that
are dependent on high levels of Shh activity for their specification.
Therefore, Gli2 appears to play a critical role in mediating the
activity of Shh in the ventral neural tube (see Fig. 3). Furthermore,
transcription of both Ptc1 and Gli1 is downregulated, consistent
with Gli2 activation of these targets. Whether or not Gli2 also
has repressing activity is less certain. In contrast to Gli1, the
Gli2 protein has been shown to undergo cleavage when expressed in
tissue culture cells (Ruiz i Altaba 1999) as well as in transgenic
Drosophila, but unlike Ci, this cleavage appears to be insensitive
to Hh activity, and the cleaved form represents only a minor fraction
of the protein. Nevertheless, an artificially C-terminally truncated
form of Gli2 can efficiently repress transcription of a Gli-reporter
gene (Sasaki et al. 1999). Another line of evidence consistent with
a repressive activity of Gli2 comes from mutations in the zebrafish
gene that give rise to C-terminally truncated forms of the protein
(Karlstrom et al. 1999). These mutants have weak dominant loss-of-Hh-function
effects on somite patterning (van Eeden et al. 1996), consistent
with the mutant proteins acting as repressors of Hh target genes.
In the absence of a protein null allele, however, it remains unclear
whether or not this reflects an activity of the endogenous Gli2
protein. Ectopic expression of Gli2 in Xenopus embryos has been
reported to lead to a repression of Gli1-mediated floor plate induction,
also consistent with Gli2 playing a repressive role (Ruiz i Altaba
1998). Yet this activity is hard to reconcile with the absolute
requirement for Gli2 in induction of the floor plate revealed by
loss of Gli2 function in the mouse. Although these results may reveal
some interesting species specific difference in the role of Gli2,
they may equally well illustrate the potential pitfalls when making
inferences about gene function solely on the basis of ectopic expression
experiments. In the wild-type neural tube, Gli2 is present in cells
responding to Shh input that could modify its activity, whereas
there is no Shh input accompanying the ectopic expression of Gli2
in the Xenopus studies. In contrast to Gli1 and Gli2, there is little
doubt that Gli3 acts as a repressor of Hh targets as well as of
Shh itself (Masuya et al. 1995; Ruiz i Altaba 1998); partial loss
of this repressor activity most likely underlies the various skeletal
anomalies associated with Gli3 mutations in the human population
(for review, see Villavicencio et al. 2000). Repression of Hh targets
both in cell culture and in the embryo (Dai et al. 1999; B. Wang
et al. 2000) appears to be mediated by proteolytic cleavage of full-length
Gli3 into a repressor form (the reported size of which varies between
83 and 100 kD) that lacks amino acids C-terminal to the zinc finger
domains. This form accumulates preferentially in the nucleus of
expressing cells (Dai et al. 1999; B.Wang et al. 2000). Although
Gli3 lacks an obvious CORD and no vertebrate Costal-2 ortholog has
yet been identified, the finding that it undergoes the same Hh-dependent
cleavage when expressed in Drosophila strongly suggests that Gli3
processing occurs by a mechanism analogous to that controlling Ci.
In addition, Gli3 is itself a target of transcriptional repression
by Shh (Marigo et al. 1996b). Although some assays indicate an exclusively
repressing activity of Gli3 (Sasaki et al. 1999; B. Wang et al.
2000), others have revealed evidence of an activating activity.
For example, Gli3 activates a Gli1 promoter in an Shh-dependent
mechanism in cell culture, and activation is further potentiated
by association with CBP (Dai et al. 1999). Consistent with this,
Gli3 has been shown to interact directly with the vertebrate Su(fu)
protein (Pearse,
1999), as well as with a human homolog of Fused (Murone et al.
2000), as would be expected if the nuclear import of an activating
form of Gli3 is regulated by Shh. Given that Gli3 is initially present
within the embryo prior to production of an Hh signal, such an activating
form of Gli3 could play a role in the initial activation of Hh targets
such as Gli1. Consistent with this view, Gli1 expression is greatly
diminished in Gli3/Shh compound mutants (Litingtung and Chiang 2000).
In contrast, Ptc1 expression is significantly up-regulated, which
raises the possibility that different promoters in the same cells
may respond differently to Gli3 input. The continued presence of
Gli2 in these experiments further complicates their interpretation.
The general picture that emerges from these studies is one in which
the different Gli proteins execute subsets of the functions that
in Drosophila are subsumed by Ci. In the neural tube, a Gli2 activator
function correlates with cells that require the highest levels of
Shh signal. Gli1/Gli2 compound mutants are reported to have a similar
phenotype to Gli2 mutants (Park and Bai 2000). Therefore, Gli2 plays
the major role at spinal cord levels although some patchy loss of
floor plate cells is observed in Gli1 mutants heterozygous for a
Gli2 mutation, indicating that a Gli1 activator may play a minor
role (Park and Bai 2000). The nature of the activated form of Gli2
in cells exposed to high levels of Shh remains unclear. Studies
in Drosophila suggest that Gli2 is not subject to an Hh-dependent
inhibition of cleavage (Aza-Blanc et al. 2000); by analogy with
the other levels of Ci regulation, however, it seems likely that
the phosphorylation state and/or nuclear import of Gli2 may be regulated
by Shh. Notably, whereas floor plate fails to form and V3 interneurons
are reduced in number in the absence of Gli2 activity, other Shh
dependent ventral cell types are unaffected, although their position
changes because of the absence of midline cells. On the other hand,
specification of these same Gli2 independent cell types (V2 interneurons
and motor neurons)—but not of floor plate cells—can be
restored in Shh mutant embryos by the removal of Gli3 activity (Litingtung
and Chiang 2000). Therefore, Gli3 appears to act in an opposite
manner to Gli2, antagonizing the ventralizing activity of Shh presumably
by repressing Shh targets, much as Ci acts to repress dpp expression
in the anterior compartment of the wing imaginal disc (see Fig.
3). These data do not exclude the possibility that activator forms
of Gli3 may also play some role. For example, there are no data
ruling out an activator form of Gli3 acting together with Gli2 to
specify floor plate identity. It should also be noted that Glis
may not be the exclusive mediators of Shh signaling in the ventral
neural tube. For example, there is evidence that expression of COUPTFII
in motor neurons is Shhdependent, but no consensus Gli-binding site
is present within its regulatory region (Krishnan et al. 1997).
In the limb, in contrast, the effects of Shh may be mediated exclusively
by Gli3. Gli1/Gli2 compound mutants have normal digit number and
pattern, whereas Gli3 homozygous mutants show a dramatic polydactylous
phenotype (Hui and Joyner 1993), consistent with Gli3-mediated repression
as an important regulator of digit number. A key finding is the
demonstration of a gradient of Gli3 processing across the limb bud
(B. Wang et al. 2000). The ratio of full-length Gli3 to its processed
repressor form decreases from posterior to anterior. In addition,
there is an increase in the overall levels of protein that presumably
reflects the inverse transcriptional gradient that forms in response
to long-range Shh signaling. Therefore, one could imagine two simple
models to explain the patterning activity of Gli3 in response to
a Shh gradient: cells may either sense the absolute levels of Gli3
repressor, or alternatively the ratio of repressor to activator
forms (assuming the full-length protein has activating activity),
as proposed for Ci (von Mering and Basler 1999). What is not known
is whether this phenotype is completely independent of Shh signaling.
Ectopic activation of Shh at the anterior margin is observed in
mice heterozygous for Gli3 mutations (Masuya et al. 1995; Buscher
et al. 1997); presumably this is also true in homozygous mutants,
but this has not been reported. If so, this leaves open the possibility
that ectopic Shh signaling could act positively through Gli2. Analysis
of Gli3/Shh and Gli2/ Gli3 compound mutants would be informative
[Ingham, 2001]
- There are three Gli proteins in vertebrates (Gli1, Gli2 and Gli3).
These proteins have several regions with sequence homology, including
a centrally located DNA-binding domain with five C2-H2 zinc fingers
and a C-terminal transcription activation domain. These proteins
have distinct activities and are not functionally equivalent. Nevertheless,
their partial redundancy and often overlapping domains of expression
has made it difficult to define precisely their individual features
and functions. The Glis have been characterized most extensively
in the developing nervous system and limb buds. Their patterns of
expression are dynamic and appear to be partly orchestrated by Shh
(Hui et al., 1994; Lee et al., 1997; Marigo et al., 1996; Ruiz i
Altaba, 1998; Sasaki et al., 1999), suggesting that the polarizing
activity of Shh could derive from the sum of the particular distribution
and functional properties of each of Gli protein. For example, during
neural tube development, Shh secreted from the floor plate defines
the domains of expression of each Gli protein in the ventricular
zone of the neural tube (Ruiz i Altaba, 1998). It restricts Gli1
to the most ventral regions, places Gli2 expression in a domain
that extends from the ventral region above the Gli1 domain to the
most dorsal region where it overlaps with Gli3, and creates a graded
pattern of Gli3 expression in the most dorsal region. Other signals
are thought to positively regulate the dorsal expression of Gli2
and Gli3. There is a similar arrangement of domains of Gli expression
in developing limb buds (Büscher and Rüther, 1998; Marigo
et al., 1996). Although it is not known how these patterns are established,
Gli1 expression can be regulated by Gli3 and possibly by Gli2 as
well (Dai et al., 1999; Ding et al., 1998; Matise et al., 1998),
suggesting that the Gli proteins may constitute a Hh-driven regulatory
network. The Gli proteins are thought to be similar to Ci and to
have transcriptional activation and repression activities. In the
neural tube, Gli1 induces the expression of ptc1 and leads to the
specification of ventral neural fates (Hynes et al., 1997; Lee et
al., 1997; Ruiz i Altaba, 1998). Both its C terminal activation
domain and its N-terminal region are required for these functions;
there has been no indication that Gli1 functions as a repressor
in these contexts. The role of Gli2 appears to be more complex,
as available evidence suggests that Gli2 is both an activator and
a repressor. Ectopic expression of Gli2 in the floor plate can inhibit
floor plate induction by Gli1 (Ruiz i Altaba, 1998), C-terminally
truncated forms act as dominant negatives (Ruiz i Altaba, 1999),
and N-terminal regions of Gli2 can repress transcription when they
are fused to a heterologous DNA-binding domain (Sasaki et al., 1999).
Conversely, Gli2 can be induced by Shh and can itself induce motorneuron
differentiation (Ruiz i Altaba, 1998). Gli2 deletion mutants that
lack the N-terminal region can induce HNF3b, a function that is
normally carried out by Gli1 (Hynes et al., 1997; Lee et al., 1997;
Sasaki et al., 1997). Gli3 also has both inductive and repressive
activities (Brewster et al., 1998), and its relationship to Shh
is complex. Shh inhibits Gli3 expression and Gli3 represses Shh
expression, and the induction activity of Gli3 may be independent
of Shh (Büscher et al., 1997; Marigo et al., 1996; Ruiz i Altaba,
1998). C-terminally truncated forms that mimic mutations found in
human syndromes function as dominant negatives (Biesecker, 1997;
Ruiz i Altaba, 1999; Shin et al., 1999). Together, these results
indicate that the Gli proteins form a regulatory network that responds
to Hh signaling and that the Gli proteins have distinct properties.
The parallels between Ci and the Gli proteins with respect to their
primary structure, functions as transcription regulators and regulation
by Hh signaling are both striking and extensive. Although significant
questions remain to be answered before we understand key aspects
of Ci (for instance, the form that represents CiAct remains unidentified),
work in the Drosophila system has established the conceptual basis
for the role of these proteins as the conduit of Hh signaling. Cells
change both the structure and intracellular address of Ci in order
to orchestrate a programmed response to different concentrations
of Hh. How the functions of the three Gli proteins correspond to
the multiple functions of Ci and how they mediate responses to Hh
have yet to be elucidated. In order to address these issues, we
expressed the three Gli proteins in Drosophila and monitored them
in the presence or absence of Hh signaling. We found that their
behavior in Drosophila is consistent with the properties of the
Gli proteins that have been observed in their native environment.
Moreover, their behavior was also remarkably similar to Ci, suggesting
that the essential aspects of the mechanisms involved in Ci/Gli-mediated
Hh signaling have been tightly constrained during evolution. Interestingly,
thevarious activities of Ci appear to have been distributed among
the different Glis. [Aza-Blanc,
2000]
|
|
