Gli2 Acetylation at Lysine 757 Regulates Hedgehog-Dependent Transcriptional Output by Preventing Its Promoter Occupancy

The morphogenic Hedgehog (Hh) signaling regulates postnatal cerebellar development and its aberrant activation leads to medulloblastoma. The transcription factors Gli1 and Gli2 are the activators of Hh pathway and their function is finely controlled by different covalent modifications, such as phosphorylation and ubiquitination. We show here that Gli2 is endogenously acetylated and that this modification represents a key regulatory step for Hedgehog signaling. The histone acetyltransferase (HAT) coactivator p300, but not other HATs, acetylates Gli2 at the conserved lysine K757 thus inhibiting Hh target gene expression. By generating a specific anti acetyl-Gli2(Lys757) antisera we demonstrated that Gli2 acetylation is readily detectable at endogenous levels and is attenuated by Hh agonists. Moreover, Gli2 K757R mutant activity is higher than wild type Gli2 and is no longer enhanced by Hh agonists, indicating that acetylation represents an additional level of control for signal dependent activation. Consistently, in sections of developing mouse cerebella Gli2 acetylation correlates with the activation status of Hedgehog signaling. Mechanistically, acetylation at K757 prevents Gli2 entry into chromatin. Together, these data illustrate a novel mechanism of regulation of the Hh signaling whereby, in concert with Gli1, Gli2 acetylation functions as a key transcriptional checkpoint in the control of morphogen-dependent processes.


Introduction
The Hedgehog (Hh) pathway regulates development and stem/ progenitor cell fate and its deregulation is a major cause of some malignancies, such as medulloblastoma, the most frequent pediatric solid tumor [1].
In mammals, Hedgehog signaling is activated by the interaction of the ligand with the inhibitory receptor Patched (Ptch). This alleviates the repression upon the transmembrane transducer Smoothened (Smo), which promptly migrates to the tip of the primary cilium, a microtubule-based organelle indispensable for Hedgehog function [2]. Once at the cilium, Smo triggers a cascade of events which leads to the activation of the three Gli transcription factors: Gli1, Gli2 and Gli3 [2]. Functionally, Gli1 and Gli2 are both activators, whereas Gli3 functions mainly as repressor of Hedgehog-dependent transcription.
Genetic studies have revealed that Gli2 and Gli3 are the primary mediators of Hh signaling and are essential for embryogenesis. Conversely, Gli1 is dispensable for development but plays a key role in tumorigenesis [3]. Indeed Gli1 and Gli2 possess transforming activity [4][5][6] and their levels are found elevated in Hedgehog-dependent tumors and other malignancies [7].
A key mechanism regulating Gli transcriptional activity is represented by post-translational modifications [3]. All three Glis are subjected to sequential phosphorylation and ubiquitination, but the consequences of these modifications differ among the three transcription factors. Gli3 and Gli2 are sequentially phosphorylated by protein kinase A (PKA), glycogen synthase kinase 3b (GSK3b) and casein kinase 1 (CK1). Once phosphorylated the two transcription factors are recruited by the F-box subunit of an SCF E3 ubiquitin ligase, bTrCP, which targets Gli2 and Gli3 to the proteasome, thus generating truncated N-terminal isoforms provided of repressive activity (GliR) over full length, active Gli (GliA). Activation of the Hh signaling prevents this partial proteolysis and favors the formation of the GliA isoform. The balance between GliA and GliR is finely regulated by the extracellular concentration of Hh ligands and represents a crucial mechanism to modulate the strength of Hh response.
Regulation of Gli2 and Gli3 processing is also regulated by interaction with Sufu that protects them from cullin/SPOPmediated degradation [8]. Conversely, Gli1 cannot be cleaved into a repressor form and is not degraded by cullin3/SPOP, but is ubiquitinated and degraded by different ubiquitin ligases, in response to different conditions [9,10].
Therefore, despite the three Glis share the regulation by phosphorylation and ubiquitination, these covalent modifications appear to be modulated by distinct kinases and ligases and to bring about different outcomes.
In our recent work, we have found that Gli1 is acetylated at a single conserved lysine, and that this modification inhibits its transcriptional activity [5]. Conversely, removal of acetylation by class I HDACs enhances Gli1-dependent gene expression, a process turned on by Hedgehog activation and limited by REN KCTD11 [5,11] and other members of the KCASH family [12].
The mechanisms underlying the acetylation-dependent inhibition and the physiological conditions where this modification occurs are still not understood.
Given the relevance of Gli2 in mediating the transcriptional output of Hedgehog activation, here we have sought to understand the function and regulation of Gli2 acetylation.
We show that Gli2 is endogenously acetylated at a single conserved lysine and that this modification inhibits the transcriptional activity by preventing its promoter recruitment. Importantly, we provide evidence that Gli2 acetylation is a key conserved step, which regulates signal-dependent transcriptional activation and can be monitored during Hedgehog-mediated tissue development. Thus, in contrast to phosphorylation and ubiquitination, the acetylation/deacetylation checkpoint seems to operate though a conserved mechanism, involving an interplay between early and late activatory events.

Cell cultures and treatments
HEK293T and NIH3T3 cells were cultured as previously described [5]. For SAG treatment, NIH3T3 cells were incubated in low serum (0.5% bovine serum, BS) overnight, to allow a full Hedgehog response and then exposed to 200 nM SAG (Enzo Life Sciences) for 24 hours.

Plasmid and site-directed mutagenesis
The following plasmids were provided by other laboratories: 8xGli-Luc reporter vector was provided by H. Sasaki, RIKEN Center for Developmental Biology, Japan; 12xGli-Luc reporter was from R.

Transfection and Luciferase assay
HEK293T and NIH3T3 cells were transfected by using Lipofectamine 2000 Reagent or Lipofectamine Reagent and Plus reagent (Invitrogen) respectively, according to the manufacturer's protocol. Luciferase assays were performed as described previously [5,13].
Protein extracts were analyzed by SDS-PAGE and blotted onto a nitrocellulose membrane (Perkin Elmer). Membranes were blocked by 5% milk in Tris buffered saline with 0.1% Tween20, and incubated with primary antibodies.

Generation of Acetyl Gli1 and Gli2 antisera
Acetyl-Gli2(Lys757) and Acetyl-Gli1(Lys518) antisera were generated by Eurogentec by rabbit immunization with specific peptide: The specificity of the antibodies was validated by competition assays with the immunogenic peptides, with or without lysine acetylation.

RNA analysis and quantitative PCR
Total RNA was extracted from NIH3T3 cells, reverse transcribed and analized by quantitative Real Time PCR as previously described [5], using the following ABI TaqMan probe:
Data are shown as fold difference vs Empty vector.  Immunohistochemistry Immunohistochemistry was performed as described previously [5].

Statistical analysis
Statistical analysis was performed using StatView 4.1 software (Abacus Concepts, Berkeley, CA). Results are expressed as mean 6 s.d. of at least three separate experiments, each performed in triplicate. Statistical differences were analyzed with the Mann-Whitney U test for non-parametric values and a p,0.05 was considered significant.

p300 inhibits Gli2 activity by acetylating lysine 757
We first tested the ability of different histone acetyltransferases (HATs) to acetylate Gli2. We selected three different HATs, each representing a member of the main HAT families [15]: p300 (p300/CBP family), TIP60 (MYST family) and pCAF (GNAT family). Overexpression of p300 induced a robust acetylation of the exogenous Gli2 (Fig. 1A), whereas Tip60 and pCAF did not change the acetylation status of Gli2, indicating the specificity of p300 acetyltransferase in this process.
To study the functional consequence of Gli2 acetylation, we performed luciferase assays with a Gli responsive reporter (126 Gli-Luc) and the different HATs: exogenous p300 caused a strong inhibition of Gli2 mediated reporter activity, while TIP60 and pCAF did not display any effect (Fig. 1B). Thus, p300 induces acetylation of Gli2, accompanied by a decrease of its transcriptional activity.
To determine whether lysine 757 is acetylated, we mutated it into arginine and performed in vivo acetylation assay with the mutant Gli2K757R. As shown in Figure 1C, mutation of K757 abrogated the ability of Gli2 to be acetylated by p300, demonstrating that this is the only acetylatable residue of Gli2.
It has been shown that substitution of a lysine with a glutamine mimics a status of constitutive acetylation [16]. Thus, to verify the effect of acetylation at K757 we generated a K757Q mutant. Supporting the hypothesis that acetylation of Gli2 is an inhibitory modification, K757Q substitution prevented almost completely the Gli-dependent transcriptional activation, despite the levels of the expressed proteins were identical (Fig. 1D).

Endogenous Gli2 is deacetylated in response to Hedgehog activation
Having observed that Gli2 acetylation at K757 inhibits its transcriptional activity, we next wondered whether this modification occurs at endogenous level and how it is connected with the activation status of the Hedgehog pathway.
Toward this end, we generated a rabbit polyclonal antibody against the acetylated K757 of Gli2, which was able to readily detect this modification and did not react against K757R mutant Gli2 ( Fig. 2A). To determine whether acetylated Gli2 is detectable at endogenous levels and to study if this modification is perturbed in response to activated Hedgehog signaling, we performed acetylation assays in NIH3T3 cells treated with the Smo agonist SAG. As shown in Figure 2B, following immunoprecipitation, endogenous acetylated Gli2 was promptly detected with anti acetyl-Gli2(Lys757) and exposure of cells to the Smo agonist SAG caused a significant decrease of K757 acetylation.
To study the contribution of K757 deacetylation to the Gli2dependent transcriptional activation, we analyzed the activity of K757R Gli2 mutant, mimicking a status of constitutive deacetylation in luciferase assays and by quantitative Real Time PCR (Fig. 2C). Compared to the wild type protein, ectopic expression of K757R mutant induced a stronger luciferase activity (Fig. 2C, top) and upregulation of endogenous Gli1 mRNA levels (Fig. 2C,  bottom), indicating a gain of function activity. Together, these data demonstrate that removal of Gli2 acetylation enhances Hhdependent transcription.
We next performed a quantitative Real Time PCR to assess the levels of the Hh target mRNA Ptch1, in NIH3T3 cells transfected with Gli2 WT or K757R mutant, and treated with SAG or vehicle control.
Exogenous expression of Gli2 caused a significant five fold increase of Ptch1 transcript that was further increased by four fold upon Smo activation with SAG (Fig. 2D). In keeping with the luciferase data, expression of K757R mutant showed a significantly higher transcriptional activity compared to WT Gli2. Conversely, SAG failed to further increase the activity of K757R mutant construct, indicating that abrogation of K757 acetylation prevents the ability of SAG to modulate Gli2 transcriptional activity.
We next studied if the acetylation/deacetylation balance plays a role in biological contexts regulated by the Hh signaling. We selected the developing cerebellum of 6 days old mice (P6), where Hh pathway regulates development of the cerebellar granule cell progenitors and promotes their mitotic expansion in the External Granular Layer (EGL).
Confirming the increased Hh activation status, incubation of sections with anti Gli1 antisera, strongly stained the EGL (Fig. 3). In contrast, neither acetylated Gli1 nor acetylated Gli2 were detectable in the same region, thus indicating that both these Hh transcriptional activators are deacetylated in the EGL at this stage of development.

Acetylation of Gli2 prevents its promoter occupancy
Activation of Gli2 function involves at least three main steps: i) a modification of processing/stability with alterations of its half-life; ii) changes in cellular compartmentalization; iii) interaction with target promoters and transcriptional cofactors to induce transcription. Therefore, we next addressed whether acetylation of Gli2 affects one of these events leading to transcriptional activation.
We ruled out the possibility that acetylation could interfere with Gli2 stability since the steady state levels of WT and mutant proteins did not appear to be different (Fig. 4B bottom). Thus, we studied if the inhibitory effect of acetylation could be attributed to a change in intracellular localization. To this end, we analyzed the nuclear and cytoplasmic localization of WT and mutant Gli2 plasmids. As shown in Figure 4A, Gli2 WT, K757R and K757Q mutants did not show differences in their cellular localization, which appeared to be predominantly nuclear. This data suggested that inhibition of Gli2 by acetylation is a nuclear process, which is likely to interfere with one of the transactivating steps occurring at the chromatin level. To address this issue we performed chromatin immunoprecipitation assays in cells transfected with Gli2 WT and mutant plasmids and analyzed their recruitment over the Gliresponsive element of the Ptch1 promoter [5].
Remarkably, the promoter occupancy of Gli2 K757Q was abrogated, compared to WT Gli2, whereas the K757R mutant displayed a significant increase in chromatin binding (Fig. 4B). Therefore, these data suggested that acetylation inhibits Gli2 activity by preventing its chromatin recruitment.

Discussion
Protein acetylation is a critical regulatory step for histones and many non-histone proteins, including transcription factors, enzymes or structural proteins [17].
In the present work we have characterized the acetylation of Gli2, the early transcriptional effector of Hedgehog signaling. We have demonstrated that Gli2 is acetylated at lysine 757 by the coactivator p300 and that this modification inhibits Hh-target gene expression.
While it is still unclear the type of signals that may induce this p300-mediated acetylation of Gli2, we have shown that removal of the K757 acetyl group occurs upon activation of Hh signaling. It is likely that the mechanism linking Hh activation to Gli2 deacetylation involves the activation of class I HDACs [5], which are induced by Smo agonists at transcriptional and, perhaps, posttranscriptional level.
Interestingly, the acetylation/deacetylation equilibrium seems to contribute to the typical dynamic response of Hh-dependent transcription. Indeed, a Gli2 mutant mimicking a constitutive deacetylation status is hyperactive but no longer induced by Smo agonists.
Thus, in addition to the balance between the full length GliA and the truncated GliR isoforms, the acetylation/deacetylation balance seems to be an additional regulated mechanism to switch from one status to another. This observation adds further complexity to the classical model of Gli activation and implies that acetylation represents a supplementary mechanism utilized by cells to modulate Hedgehog output. In support to this model, we show that, when monitored with a specific antibody, acetylated Gli2 cannot be found in territories of active Hedgehog signaling, such as the outer EGL, where HDAC1 levels have also been found to be upregulated [5].
It will be important to understand whether and to what extent acetylation also contributes to the graded response to ligand concentrations, thus representing a further level of control during morphogen dependent development.
What are the mechanisms of transcriptional inhibition following Gli2 acetylation? In this work we have demonstrated that acetylation of Gli2 does not affect its intracellular localization and stability. Instead, acetylation appears to prevent Gli2 recruitment to target chromatin. Since the residue K757 maps outside the zinc finger domain, it is unlikely that acetylation affects Gli2/DNA affinity. A plausible explanation is that K757 acetylation prevents the association with chromatin-bound cofactors, which are essential for the promoter entry/transactivation of Gli2. In this regard, a potential candidate for this effect could be the ATP-dependent chromatin remodeling factor Brg, which has been reported to interact with both HDACs and Glis and to participate to their Hh-induced chromatin recruitment and target gene activation [18]. Further work is needed to demonstrate these issues.
In conclusion, these data illustrate a novel mechanism of regulation of the Hh signaling, where acetylation of Gli2 at lysine 757 functions as a critical regulated step, controlling the activation status of Hh pathway. We suggest that this mechanism contributes to the early regulatory events immediately downstream of Smo (i.e. cleavage-dependent balance between Gli2A and Gli2R), and is functionally coordinated with Gli1 acetylation, which represents a late regulated step of Hedgehog pathway activation (Fig. 5).
These results further emphasize the relevance of Gli acetylation as a key regulatory epigenetic modification, with promising therapeutic potential for diseases linked to aberrant Hh pathway activation.
Author Contributions Figure 5. Model of acetylation-dependent control of Gli activity. Following Hh/Ptch interaction, Smo triggers a signaling cascade leading to Gli2 deacetylation and to the inhibition of the bTrCPregulated balance between Gli2R and full length Gli2 (Gli2A). Both events contributes to the early signal-dependent activation of the Hh pathway. Once activated, Gli2 promotes transcription of Gli1, whose activity is also regulated by Hh-induced HDAC1-mediated deacetylation, thus generating a positive feedback loop (late activation). doi:10.1371/journal.pone.0065718.g005