Identification of a Suppressive Mechanism for Hedgehog Signaling through a Novel Interaction of Gli with 14-3-3*

Gli transcription factors are central effectors of Hedgehog signaling in development and tumorigenesis. Using a tandem affinity purification (TAP) strategy and mass spectrometry, we have found that Gli1 interacts with 14-3-3ϵ, and that Gli2 and Gli3 also bind to 14-3-3ϵ through homologous sites. This interaction depends on their phosphorylation, and cAMP-dependent protein kinase (PKA), a known negative regulator of Hedgehog signaling serves as a responsible kinase. A Gli2 mutant engineered to eliminate this interaction exhibited increased transcriptional activity (2 ∼ 3×). Transcriptional repression by 14-3-3 binding was also observed with Gli3, when its N-terminal repressor domain was deleted. The phosphorylation sites responsible for the binding to 14-3-3 are distinct from those required for proteolysis, the known mechanism for PKA-induced repression of Hh signaling. Our data propose a novel mechanism in which PKA down-regulates Hedgehog signaling by promoting the interaction between Gli and 14-3-3 as well as proteolysis. Given the certain neuronal or malignant disorders in human caused by the abnormality of 17p13 encompassing 14-3-3ϵ overlap with increased Hh signaling, the Gli-14-3-3 interaction may have pathological significance for those human diseases.

The signaling proteins of the Hedgehog (Hh) 2 family play central roles in many developmental processes in organisms from flies to humans, particularly processes involving patterning of the neural tube, notochord, and prechordal mesoderm (1,2). Hedgehog protein, the ligand that initiates Hh signal transduction pathway, is a developmental morphogen that elicits a graded cellular response depending on the distance between the recipient cell and the ligand-secreting cell. This gradient-dependent signaling is thought to be important in the determination of cell fate. Deregulation of Hh signaling results in a variety of congenital malformations and cancers (3)(4)(5). Given the importance of the Hh signaling pathway, there has been considerable interest in understanding the mechanism by which a target cell interprets Hh signaling to yield an appropriate response.
In Drosophila, activation and repression of Hh target genes is mediated by a single Gli zinc finger transcription factor known as Cubitus interruptus (Ci). In the absence of Hh signaling, Ci is phosphorylated by cAMP-dependent protein kinase (PKA) (6), glycogen synthase kinase 3␤ (GSK3␤), and casein kinase I (CKI). After proteolytic processing into an N-terminal repressor form, Ci translocates to the nucleus, where it silences expression of its target genes. Hh signaling triggers a series of events that prevents this phosphorylation and proteolysis of Ci, to block formation of the repressor form, and allow the fulllength, activating form of Ci to enter the nucleus and promote induction of specific target genes (1,7).
In vertebrates, three Hedgehog proteins (sonic, Indian, and desert), and three Gli proteins, Gli1, Gli2, and Gli3, are involved in transcriptional control of Hh target genes (2,8). Each Gli protein has a unique role: Gli3 functions chiefly as a transcriptional repressor, Gli2 is chiefly a transcriptional activator, and Gli1 functions only as a transcriptional activator (1). Like Ci, Gli3 is phosphorylated and processed in the absence of Hh; it responds to Hh by relocalizing to the nucleus without cleavage (9). Though Gli1 knockout mice develop normally, Gli2 and Gli3 knockouts have serious developmental abnormalities (4), suggesting that Gli2 and Gli3 signaling is likely to be intensely regulated in response to Hh ligands. To better understand the molecular mechanisms underlying Gli activation and regulation, we used a tandem affinity purification (TAP) strategy involving a novel mammalian TAP tag, called MEF (10,11), in conjunction with mass spectrometry, to isolate and characterize human protein complexes containing Gli.

EXPERIMENTAL PROCEDURES
Cell Cultures and Transient Transfection-The human embryonic kidney cell line 293T, the human hepatoma cell line PLC/PRF/5, the human cervical carcinoma cell line HeLa, and the mouse fibroblast cell line NIH3T3 (Riken Cell Bank, Tsukuba Science City, Japan) were maintained in Dulbecco's modified Eagle's medium (Sigma) containing 10% heat-inactivated fetal bovine serum (293T, PLC/PRF/5, and HeLa) or bovine serum (NIH3T3). All cells were incubated at 37°C and 5% CO 2 . Cells were seeded on culture plates, and transfections were carried out using Effectene transfection reagent (Qiagen, Hilden, Germany) for PLC/PRF/5 and NIH3T3 cells or FuGENE 6 (Roche Applied Science, Indianapolis, IN) for 293T and Hela cells, as described in the manufacturers' protocol.
Plasmids-Human Gli1 and Gli3 cDNAs (from K. W. Kinzler and B. Vogelstein (2,8)) and mouse Gli2 cDNA (from H. Sasaki (12)) with deletions of their 5Ј-untranslated sequences were cloned into the vector pcDNA3-MEF, a mammalian expression plasmid for the MEF method described previously (10,11). PCR-based mutagenesis was used, and genes encoding wt and mutant ⌬N-Gli3 were prepared utilizing the unique BamHI site. ⌬PKA-Gli2 was generated by a combination of restriction digestion and PCR strategies to delete the region from Ser-781 to Leu-861. All PCR-generated constructs were verified by sequencing. The reporter plasmid GliBS-Luc, the Myc-tagged SUFU expression vector, the vector expressing the constitutively active catalytic subunit of PKA (PKA-CA), and the 14-3-3 inhibitor peptide expression vector EYFP-difopein were kindly provided by H. Sasaki (13), R. Toftgard (14), G.S. McKnight (15), and H. Fu (16), respectively.
The protein complex was incubated with anti-FLAG Abconjugated agarose beads for a second immunoprecipitation at room temperature for 1 h. After three washes with buffer A (20 mM Tris-HCl, pH 7.5, 150 mM NaCl), the beads were eluted with FLAG peptide (80 g/ml) in buffer A. The eluted proteins were concentrated, separated by 10/20% SDS-PAGE, and visualized by silver staining.
Protein bands were excised from the gel and digested with trypsin, and the resultant peptide fragments were analyzed using a direct nano-flow LC-MS/MS system equipped with an electrospray interface reversed-phase column, a nano-flow gradient device, and a high-resolution quadrupole time-of-flight hybrid mass spectrometer (Q-TOF2; Micromass, Manchester, UK). To identify the proteins, all of the MS/MS spectra were searched against the non-redundant protein sequence data base maintained at the National Center for Biotechnology Information using the Mascot program (Matrixscience, London, UK). The MS/MS signal assignments were confirmed manually.
Immunoprecipitation and Immunoblotting-Immunoprecipitation and immunoblotting were performed as previously described (11,17). After 293T cells were lysed, the cellular proteins were incubated with anti-FLAG agarose-conjugated beads for 1 h at 4°C. To detect endogenous interaction, cell lysates from PLC/PRF/5 cells were incubated for 1 h at 4°C with anti-Gli1 or anti-14-3-3⑀ Ab or normal IgG control, followed by an additional 4 h of incubation with protein G. The beads were then washed, and the immunoprecipitated proteins were eluted by boiling in SDS sample buffer. The proteins were separated by SDS-PAGE and electrophoretically transferred to a polyvinylidene difluoride membrane. The membrane was probed with primary Ab (anti-14-3-3, anti-14-3-3⑀, anti-Gli1, anti-phospho-14-3-3 binding site, anti-phospho-PKA substrate, anti-Xpress, anti-Myc, or anti-FLAG at 1:1,000 dilution, and anti-␣tubulin at 1:5000) and with secondary Ab (HRP-conjugated anti-mouse or anti-rabbit Ab at 1:1000 to 1:5000 dilution). The HRP signal was detected using an enhanced chemiluminescence system (ECL Plus, GE Healthcare).
Purified GST fusion proteins were incubated with 25 ng of recombinant PKA catalytic subunit (Upstate, Lake Placid, NY), 20 l of Assay Dilution Buffer I (Upstate) containing 10 g of bovine serum albumin, and Magnesium/ATP Mixture (Upstate) containing 10 Ci of [␥-32 P]ATP (GE Healthcare). After 30 min at 30°C, the reactions were terminated by addition of 13 l of 4ϫ SDS-PAGE sample buffer, and phosphorylated proteins were visualized by SDS-PAGE and autoradiography.
For lentivirus infection, PLC/PRF/5 or NIH3T3 cells in Dulbecco's modified Eagle's medium with 10% fetal bovine serum or BS and 8 g/ml polybrene were incubated with an equal volume of lentivirus for 24 h. Stably transfected cells were selected in puromycin (2 g/ml for PLC/PRF/5, 3 g/ml for NIH3T3) and tested for 14-3-3⑀ expression by Western blot.
Quantitative Real-time RT-PCR-Total RNA was collected with the RNeasy Mini Kit (Qiagen) and subjected to the RT reaction using oligo-dT primer and ImProm-II Reverse Transcriptase (Promega) according to the manufacturer's instruction. Quantitative real-time PCR for murine Gli1 and Ptch1 genes was performed using the cDNA of NIH3T3 cells treated with DMSO or 100 nM SAG Ϯ 1 M FSK. The following primers were used: Gli1: gga agt cct att cac gcc ttg a (sense) and caa cct tct tgc tca cac atg taa g (antisense); Ptch1: CTC TGG AGC AGA TTT CCA AGG (sense) and TGC CGC AGT TCT TTT GAA TG (antisense). All assays were performed with FastStart Universal SYBR Green Master (ROX) (Roche) and ABI Prism 7000 Sequence Detection Systems (Applied Biosystems). The standard curve method was used to calculate target gene expression, which was normalized to that of the Gapdh gene. Each sample was analyzed in duplicate.
Immunofluorescence Microscopy-Immunofluorescence microscopy was carried out as previously described (11). HeLa cells were transfected with expression vectors 24 h after seeding. After another 36 h, including 12 h exposure to 40 M FSK in the growth medium, the cells were fixed with 2% paraformaldehyde for 15 min at room temperature, rinsed with phosphate-buffered saline (PBS), and permeabilized with 0.25% Triton X-100/PBS for 15 min. After blocking, the cells were incubated with anti-Myc Ab (1 g/ml) overnight at 4°C and then with Alexa Fluor 488 goat anti-mouse IgG secondary Ab (1:100 dilution) for 1 h at room temperature. The nuclei of the cells were stained with propidium iodide (Sigma). Fluorescence images were obtained using a Leica TCS SL confocal laser-scanning microscope (Leica Microsystems, Wetzler, Germany) and photographed using Leica confocal software (Leica Microsystems).
Luciferase Assays-Luciferase assays were performed as described previously (11). Approximately 1 ϫ 10 5 293T cells were plated into each well of a 12-well tissue culture plate (Iwaki Glass, Chiba, Japan) and transiently transfected after 24 h. The transfection mixtures contained 250 ng of GliBS-Luc, 125 ng of PKA-CA, 2.5 ng of pRL-SV40 (internal control), and 125 ng of effector construct or empty vector. The cells were harvested 24 h after transfection and assayed for luciferase activity using the PicaGene Dual SeaPansy System (Toyo Ink, Tokyo, Japan) and a luminometer (Lumat LB 9507; EG & G Berthold, Bad Wildbad, Germany). Independent experiments were performed at least three times.
Electrophoretic Mobility Shift Assays-Expression vectors for constitutively active PKA-CA and for FLAG-tagged wt and S956A Gli2 were co-transfected into 293T cells in 10-cm tissue culture dishes. The cells were lysed in 1 ml of lysis buffer, and cellular proteins were immunoprecipitated with anti-FLAG Ab-conjugated agarose beads for 2 h at 4°C. After the beads were washed with lysis buffer, the bound proteins were eluted with buffer containing 100 g/ml of 3ϫ FLAG peptide (Sigma).
This eluted protein lysate was used in electrophoretic mobility-shift assays performed using a LightShift EMSA Kit (Pierce Biotechnology), according to the manufacturer's instructions. Briefly, eluted protein lysate was incubated for 30 min with biotinylated DNA probe (Gli-RE) (18) and non-biotinylated competitor DNA oligonucleotide (100-fold molar excess) in binding buffer supplemented with 2.5% glycerol, 5 mM MgCl 2 , 0.05% Nonidet P-40, and 50 ng/l poly(dI-dC). The resulting complexes were resolved in a nondenaturing 6% polyacrylamide gel, transferred onto a membrane, and detected using HRPconjugated streptavidin and a chemiluminescent substrate.

Identification of 14-3-3⑀ as a Novel Interacting Protein with
Gli1-The MEF (Myc-TEV-FLAG) technique uses two different affinity modules (Myc and FLAG) separated by a cleavage site for the TEV protease to reduce nonspecific protein binding to the complex (Fig. 1A). A mammalian expression vector encoding Gli1 with an N-terminal MEF tag was constructed and transfected into 293T cells. Tagged Gli1 was then purified from cell lysates along with its bound proteins. After the proteins were eluted with FLAG peptides, they were separated by SDS-PAGE and visualized by silver staining (Fig. 1B). Bands detected in the Gli1 lane but not in the control lane were considered to be proteins directly or indirectly bound to Gli1. These protein bands were excised from the gel and subjected to in-gel tryptic digestion. The resultant peptides were extracted from the gel slice and analyzed using a direct nano-flow LC-MS/MS system. The corresponding proteins were identified as 14-3-3⑀, semenogelin I, and 70-kDa heat-shock proteins. To confirm the interaction between Gli1 and 14-3-3⑀ in a physiological condition, the lysates from PLC/PRF/5 cells, where Hh signaling is constitutively active (19), were applied to immunoprecipitation using anti-Gli1 and anti-14-3-3⑀ Abs. As a result, Gli1 and 14-3-3⑀ co-precipitated each other (Fig. 1, C and D), which suggests the in vivo interaction of endogenous Gli1 and 14-3-3⑀.
14-3-3 Binding Site Is Phosphorylated by PKA in Vitro-The sequence surrounding Ser-640, RRXS, is identical to the consensus phosphorylation sequence for PKA. This motif is conserved in other Gli proteins, including Ci (Fig. 3A) and has already been reported as a putative site for PKA phosphorylation (6,9,28). To confirm whether PKA directly phosphorylates Gli1 Ser-640, in vitro kinase assays were performed. Wild-type and S640A mutant sequences flanking Gli1 Ser-640 were fused to GST and the recombinant fusion proteins examined for incorporation of 32 P from [ 32 P-␥]ATP by the recombinant PKA catalytic subunit. As seen in Fig. 3B, the wt fusion protein was efficiently phosphorylated by PKA whereas S640A mutation abolished this phosphorylation. Taken together, these data suggest that Gli1 associates with 14-3-3 proteins through Ser-640, and that PKA, a negative regulator of Hh signaling, can serve as a responsible kinase for Ser-640.
Gli2 and Gli3 Also Bind to 14-3-3 in a Phosphoserine-dependent Manner-We next examined whether Gli2 and Gli3, the other Hh-regulated transcription factors with unique roles distinct from Gli1 (1), are also phosphorylated by PKA and bind to A, schematic representation of the strategy for TAP using an MEF tag (10,11). The pcDNA3 vector is designed to express a protein with tandem Myc and FLAG tags at the N terminus of the bait protein. The first round of affinity purification is performed using anti-Myc Ab, followed by cleavage with TEV protease. The second round of affinity purification is performed using anti-FLAG Ab, followed by specific elution with FLAG peptide. The eluted complex was separated by SDS-PAGE and silver-stained. Specific bands were digested with trypsin and analyzed by LC-MS/MS. B, after lysis of pcDNA3-MEF-Gli1-or pcDNA3-MEF-transfected 293T cells, the expressed Gli1 was recovered using the MEF TAP procedure. The proteins bound to Gli1 were eluted from the beads, separated by SDS-PAGE (10/20%), and silver-stained. The molecular masses of the protein are indicated. The arrowhead indicates the position of Gli1. The four protein bands, indicated by arrows, were excised from the gel and subjected to in-gel tryptic digestion. LC-MS/MS analysis identified the proteins as 14-3-3⑀, semenogelin I, and 70-kDa heat-shock proteins 5, 6, and 8. C, lysates from PLC/PRF/5 cells were subjected to immunoprecipitation using anti-Gli1 Ab or control IgG followed by Western blotting using anti-Gli1 (upper blot) and anti-14-3-3⑀ (lower blot) Abs to detect endogenous interaction. D, immunoprecipitation using anti-14-3-3⑀ Ab was performed as described in C.
14-3-3 proteins. The amino acid sequence surrounding Ser-640 of Gli1 is highly conserved among Gli transcription factors; the corresponding serine residues in mouse Gli2 and human Gli3 are Ser-956 and -1006, respectively (Fig. 3A). We expressed the Gli2 and Gli3 peptides flanking these serine residues as GST fusion proteins and performed in vitro kinase assays with recombinant PKA. Both of the Gli2 and Gli3 fusion proteins were efficiently phosphorylated by PKA, whereas the Gli2 S956A and Gli3 S1006A mutant fusion proteins were not (Fig. 3B).
Next, to examine the interaction of Gli2 and Gli3 with 14-3-3, we expressed FLAG-tagged Gli2 and Gli3 in 293T cells under PKA-activated conditions, either by co-expressing the constitutively active PKA-CA (15) or by the treatment with FSK, a PKA activator. Gli2 and Gli3 were then immunoprecipitated using anti-FLAG Ab beads, and blotted for co-associated endogenous 14-3-3. As shown in Fig. 3, C and D, both the wt Gli2 and Gli3 showed strong binding to 14-3-3, whereas the Gli2 S956A and Gli3 S1006A, corresponding to Gli1 S640A, had no detectable affinity for 14-3-3, as well as the Gli3 double mutant S1006A/S1026A (Fig. 3, C and D). As an additional negative control, we also constructed Gli2 S975A and Gli3 S1026A mutants, which are analogous to the Gli1 S659A mutant. Both of these negative control mutants showed equivalent binding to 14-3-3 as the wt Gli2 and 3 proteins (Fig. 3, C and D), confirming the importance of the Gli1 Ser-640 homologs Ser-956 and Ser-1006 in Gli2 and Gli3, respectively for conferring phosphorylation-dependent 14-3-3 binding. Finally, we demonstrated that PKA induces the in vivo phosphorylation of the 14-3-3 consensus phospho-binding sequence RXXpSXP in Gli2, using the phosphorylation-specific Ab against this motif. In addition, we showed that the interaction of Gli2 and 14-3-3 increased in proportion to the level of Gli2 phosphorylation (Fig. 3E).
Binding of Gli Proteins to 14-3-3 Decreases the Hh Signaling Activity-The effect of 14-3-3 binding on the transcriptional activity of each Gli protein was examined by luciferase assay using a Gli-regulated luciferase reporter under the control of eight tandem copies of the HNF3-␤ Gli-binding site (8ϫGBSluciferase) (13). Gli expression vectors and the 8ϫGBS luciferase reporter plasmid were co-transfected into 293T cells together with PKA-CA to potentiate the interaction between Gli and 14-3-3. As shown in Fig. 4A, co-transfection of wt Gli1 with 8ϫGBS-luciferase resulted in a 30-fold increase in reporter activity compared with co-transfection of 8ϫGBS-luciferase with vector control. The Gli1 S640A mutation up-regulated the reporter activity to the comparable level to that wt Gli1 did. Gli2 also increased reporter activity nearly 3-8-fold, and importantly, the analogous 14-3-3-binding mutant Gli2 up-regulated the activity level to more than 2-fold than that wt Gli2 did. In a while, neither wt nor the 14-3-3 binding mutant of Gli3 activated this reporter, consistent with its proposed function in vivo as a transcriptional repressor (12). Because deletion of the N-terminal repressor domain from Gli3 has been reported to restore its transcriptional activity (12), we constructed an N-terminal deletion mutant of Gli3 (⌬N-Gli3). We found that ⌬N-Gli3 functions as a transcriptional activator like Gli2 (Fig. 4B) and that ⌬N-Gli3 defective in 14-3-3 binding demonstrated the higher activity compared with that of ⌬N-Gli3 (Fig. 4, A and B). These findings support that the phos-

Negative Regulation of Gli by
To confirm that the observed transcriptional repression requires the binding to 14-3-3 as well as the phosphorylation of Gli proteins, we constructed a new Gli2 mutant in which Pro-958 was mutated to Ala (P958A). Pro-958, which is at the ϩ2 position relative to the pSer-956, is an important residue for 14-3-3 binding (25) but is not involved in PKA phosphorylation. As seen in Fig. 4C, the P958A mutation abolished 14-3-3 binding to Gli2. In luciferase assays performed under PKA-CA expression, the transcriptional activities of P958A and S956A mutants of Gli2 were comparable to each other, but 2ϳ3-fold higher compared with those of wt or S975A Gli2 (Fig. 4D). These results suggest that 14-3-3 binding is essential for the phosphorylation-dependent repression of Gli transcriptional activities.
A recent report suggested that the phosphorylation of Gli2 by PKA, CKI, or GSK-3␤, influences the transcriptional activity through its ubiquitination and degradation (28). In addition, phosphorylation-dependent 14-3-3 binding of MDMX has been reported to influence its ubiquitination and degradation (29). To address whether the proteasome-dependent degradation affects the 14-3-3-dependent repressive effect, we generated the Gli2 mutant lacking the regions essential for its degradation by the E3 ubiquitin ligase ␤-TrCP (⌬PKA). The deleted region containing the four responsible PKA phosphorylation sites, Ser-789, -805, -817, and -848 for degradation does not include Ser-956. As shown in Fig. 4E, the additional mutation in  panel). C, interaction of Gli2 with endogenous 14-3-3 through Ser-956. FLAG-tagged wt and mutant Gli2 (S956A or S975A) were expressed with PKA-CA in 293T cells. Cell lysates were immunoprecipitated using anti-FLAG Ab and blotted using anti-14-3-3 Ab (upper blot) or anti-FLAG Ab (middle blot). Samples of the lysates were blotted with anti-14-3-3 Ab as a control (lower blot). D, interaction of Gli3 with endogenous 14-3-3 through Ser-1006. A similar assay was performed with FLAG-tagged wt or mutant Gli3 (S1006A and/or S1026A). Wild-type and S1026A Gli3 interacted with 14-3-3, but S1006A and S1006A/S1026A mutants lost their affinity for 14-3-3. E, phosphorylation of Gli2 Ser-956 in 293T cells increased with FSK exposure, and the affinity of Gli2 for 14-3-3 increased as this phosphorylation increased. The 293T cells were transfected with a FLAG-tagged wt or S956A Gli2 vector or the empty vector 24 h before addition of 40 M FSK to the growth medium. The cells were lysed after 0, 1, 2, and 4 h of FSK exposure. Lysates were subjected to immunoprecipitation using anti-FLAG Ab and blotted with Abs specific for the 14-3-3 binding site (RXXpSXP), FLAG, or 14-3-3. Samples of the lysates were blotted using anti-14-3-3 Ab as a control.
Ser-956 clearly enhanced the transcriptional activity of ⌬PKA Gli2, similar to the case of wt Gli2. These findings indicate that the 14-3-3-dependent repressive mechanism against Gli2 activity is independent on the degradation by the ubiquitin-proteasome system. Gli2-14-3-3 binding affects the function of various proteins by sequestering from nucleus to cytoplasm (17,20,21). The change in the subcellular localization of Gli transcription factors, principally Ci, by their processing following PKA phosphorylation has also been investigated (7). To investigate whether 14-3-3 binding influenced the subcellular localization of Gli2, Myc-tagged wt and S956A Gli2 were expressed in HeLa cells in the presence of 40 M FSK. Immunocytochemistry using anti-Myc antibody revealed that both wt and S956A Gli2 localized in the nucleus (Fig. 5A). In addition, PKA-CA and GFP-tagged Gli2 were co-expressed in 293T cells showed similar results (data not shown). Because it is reported that Gli proteins are sequestered to the cytoplasm by SUFU (14), we also exam-ined whether 14-3-3 binding affects the interaction with SUFU. SUFU repressed the transcriptional activity of Gli2 independently of its interaction with 14-3-3 (Fig. 5B). The co-expression of Myc-tagged SUFU, Gli2, and PKA-CA showed that the interaction between Gli2 and 14-3-3 had no effect on the affinity of Gli2 for SUFU (Fig. 5C). Thus, 14-3-3 binding to Gli2 does neither affect cellular localization nor interaction with SUFU.

14-3-3 Interaction Does Not Affect the Intracellular Localization and the Relationship with SUFU of
S956A Gli2 Protein Maintains the Affinity for Target Oligonucleotide in Vitro-To assess the direct effect of 14-3-3 on DNA binding by Gli2, we performed EMSA using immunopurified FLAG-tagged wt or S956A Gli2 from cell lysates co-expressing PKA-CA. Both wt and S956A Gli2 specifically decreased the electrophoretic mobility of the Gli-RE oligonucleotide (Fig. 5D). Addition of excess unlabeled Gli-RE dramatically reduced the intensity of retarded oligonucleotide bands in both the wt and S956A samples. These findings suggest that, at least in an in vitro condition, 14-3-3 does not affect the binding capacity of Gli2 to the synthetic naked oligonucleotides. were less active than their respective Ser 3 Ala mutants. B, transcriptional activity of Gli3 was almost equal to that of the empty vector, and deletion of the N-terminal repressor domain to create ⌬N-Gli3 restored the transcriptional activity. The recovered activity of the constructs with the capacity to interact with 14-3-3 (wt and S1026A) was increased by the S1006A 14-3-3 binding site mutation (mutant S1006A and double-mutant S1006A/S1026A). C, Gli2 P958A mutant also lost the capacity to interact with 14-3-3, as shown by immunoprecipitation with anti-FLAG Ab and immunoblotting. Phosphorylation status could not be confirmed; however, since the proline residue at ϩ2 relative to the putative pSer is required for detection by the anti-14-3-3 binding site Ab. D, transcriptional activity of Gli2 constructs was measured as described in A and B. The activities of the P958A and S956A Gli2 mutants were similar. E, in mutant ⌬PKA-Gli2, the four regions of Gli2 containing PKA consensus sites responsible for its degradation following ubiquitination were deleted. The effect of interaction with 14-3-3 on transcriptional activity was not affected by these mutations. The data shown are representative of three independent experiments performed in duplicate (error bars). **, p Ͻ 0.01; *, p Ͻ 0.05.

PKA Stimulation Down-regulates the Expression of Gli1, Which Is a Representative Target of Hh Signaling in a 14-3-3⑀dependent
Manner-PKA is reported as a suppressor of Hh signaling by promoting degradation of Gli proteins. Given that PKA phosphorylates Gli proteins in the sites significant for 14-3-3 binding, PKA should be one of the putative kinases responsible for the 14-3-3 dependent regulation of the transcriptional activities of Gli proteins. Then we addressed whether PKA affects the transcriptional activities of Hh signaling in a 14-3-3-dependent manner. As Gli1 is a representative target protein of Hh signaling (30), the Gli1 protein level was compared between FSK-treated PLC/PRF/5 cells and DMSOtreated cells. Overnight treatment of 40 M FSK reduced the Gli1 expression, which was consistent with the notion that PKA is a negative regulator of this signaling (6). However, in 14-3-3⑀ knockdown PLC/PRF/5 cells, FSK treatment reduced Gli1 protein level to a lesser extent, which suggested that 14-3-3⑀ played a role in PKA-dependent repression of Hh signaling (Fig. 6A).
To certify the effect of PKA-14-3-3 axis on Hh signaling, we also used mouse fibroblast NIH3T3, where Hh signaling is inac-tive. To activate Hh signaling in the cells, they were treated with SAG, which is a potent activator of Hh signaling (31). Quantitative real-time PCR showed that incubation with 100 nM SAG for 30 h induced NIH3T3 cells to express Gli1 and Ptch1 in mRNA level, and the addition of 1 M FSK suppressed it (Fig.  6B). Given that Gli1 and Ptch1 are representative targets of Hh signaling, it means that SAG is able to activate Gli transcriptional activities in NIH3T3 cells. We also found that Gli1 protein as well as its mRNA was induced by the treatment of SAG (Fig. 6C). To elucidate the role of 14-3-3⑀ in the regulation of Hh signaling, we established 14-3-3⑀ knockdown NIH3T3 cells using lentivirus vector-based shRNA system. As shown in Fig.  6D, 14-3-3⑀ knockdown clearly canceled FSK-induced decrease in the expression of Gli1.
In addition, luciferase assay was performed using a Gli-regulated luciferase reporter, 8ϫGBS-luciferase. In this assay, FSK also suppressed SAG-activated luciferase activity. Furthermore, to address the contribution of 14-3-3 in this assay condition, we prepared an expression vector for the fusion protein EYFP-difopein, a specific peptide that can compete away 14-3-3 µ FIGURE 5. Interaction of Gli2 with 14-3-3 does neither affect its subcellular localization, SUFU-binding activity, nor the affinity for target oligonucleotide. A, neither PKA-mediated phosphorylation nor 14-3-3 binding relocalize Gli2 into the cytosol. Immunofluorescence pictures of representative HeLa cells show that both wt and S956A Gli2 localize in the nucleus in FSK-treated cells. The Myc-tagged Gli2 proteins are immunostained (green); the nuclei are stained with propidium iodide (red). Scale bar, 20 m. B, co-expression of SUFU, which represses the transcriptional activity of both wt and S956A Gli2 and eliminates the effect of interaction with 14-3-3. The data shown are representative of three independent experiments performed in duplicate (error bars). **, p Ͻ 0.01; *, p Ͻ 0.05. C, immunoprecipitation using a lysate from cells co-expressing Myc-tagged SUFU, a FLAG-tagged Gli2 construct, and PKA-CA proves that 14-3-3 binding does not influence the interaction of Gli2 with SUFU. D, binding of Gli2 to target oligonucleotide. Cells were transiently co-transfected with expression plasmids encoding PKA-CA and FLAG-tagged wt or S956A Gli2 or the empty vector. Cell lysates were immunoprecipitated using anti-FLAG Ab, and bound proteins were eluted using 3ϫ FLAG peptide. Eluted lysates were subjected to electrophoretic mobility shift assay. Both wt and S956A mutant Gli2 formed specific complexes with the consensus (labeled) Gli-binding oligonucleotide (arrow). Addition of a 100-fold excess of unlabeled Gli outcompeted the labeled oligonucleotide.
from its target proteins (16). The plasmid expressing a mutated version of difopein (R18Lys) that does not inhibit 14-3-3 was used as a negative control. Interestingly, the suppressive effects of FSK treatment for Gli transcriptional activities were recognized under the co-transfection of the EYFP-difopein plasmids as well as that of EYFP-R18Lys. The level of transcriptional activities was, however, more highly induced by the co-transfection of difopein than that of R18Lys (Fig. 6E). These findings indicate that the molecular systems for PKA-induced suppression of Gli transcriptional activities include both 14-3-3-dependent and -independent mechanisms.

DISCUSSION
We used a proteomic approach to identify 14-3-3⑀ as a novel Gli-interacting protein and found that this interaction negatively regulates Hh signaling through Gli2, and an N-terminally truncated form of Gli3. This interaction depends on Gli phosphorylation, and PKA, the negative regulator of Hh signaling (32) can serve as a responsible kinase. One reported mechanism by which PKA influences Gli-dependent transcription involves the ubiquitin-dependent proteolytic processing of Gli into its repressor form (6,9,28); however, the PKA phosphorylated residues responsible for this processing event are different from the one for 14-3-3 binding. In addition, loss of binding to 14-3-3⑀ in undegradated Gli2 (⌬PKA) enhanced its transcriptional activity (Fig. 4E), indicating that PKA-dependent Gli binding to 14-3-3 is not involved in its ubiquitination and proteolysis. Thus, our results suggest the distinct PKA-dependent mechanisms involved in the repression of Hh signaling, through 14-3-3 binding or degradation, respectively.
Previously, 14-3-3 binding has been shown to directly affect the DNA binding activity of several transcription factors including Miz1, DAF16, and FKHRL1. Miz1 binds to 14-3-3 following AKT phosphorylation The phosphorylation/14-3-3 binding site on Miz1 is located directly in the DNA binding zinc finger region, explaining how 14-3-3 binding might reduce its affinity for DNA (33). In contrast, the DNA binding Forkhead domain in DAF-16 and FKHRL1 is distinct from the AKT phosphorylation/14-3-3 binding sites flanking both its N-and C-terminal ends (34). Similar to DAF-16 and FKHRL1, the DNAbinding regions (zinc fingers) of Gli transcription factors, are distinct from its 14-3-3 binding site (Fig. 3A). Thus, whereas we could not demonstrate a certain change in the binding capacity for the synthetic naked oligonucleotides by the loss of 14-3-3 binding, we do not exclude the possibility that 14-3-3 binding might reduce the affinity to target chromatin regions in vivo. This hypothesis might be supported by the finding that the binding of 14-3-3 to DAF-16/FKHRL induced a conformational change and altered the affinity to the in vivo three-dimensional DNA structures. The interaction of Gli1 and 14-3-3 was dem-FIGURE 6. FSK suppresses the expression of Gli1 and depletion of 14-3-3⑀ cancels the suppression in PLC/PRF/5 and NIH3T3. A, cell lysates from PLC/PRF/5 cells, treated overnight with DMSO or 40 M FSK, were subjected to immunoblot analysis. FSK treatment reduced Gli1 expression in control cells, but this effect was canceled in 14-3-3⑀ knockdown cells. 14-3-3⑀ knockdown was corroborated using anti-14-3-3⑀ and anti-␣-tubulin Abs (␣-tubulin: loading control). B, normalized Gli1 and Ptch1 mRNA levels in SAG Ϯ FSK-induced NIH3T3 cells were determined by quantitative PCR. Cells were treated for 30 h with DMSO or 100 nM SAG Ϯ 1 M FSK in medium with 0.5% BS. C, Gli1 protein level was determined by immunoblotting. Cell lysates from NIH3T3 cells, treated as A, were subjected. Immunoblot using anti-phospho PKA substrate Ab showed efficient PKA phosphorylation induced by FSK (␣-tubulin: loading control). D, cell lysates from 14-3-3⑀ knockdown NIH3T3 cells, treated with SAG Ϯ FSK, were subjected to immunoblotting to determine Gli1 protein levels. 14-3-3⑀ knockdown was corroborated using anti-14-3-3⑀ and anti-␣-tubulin Abs (␣-tubulin: loading control). E, NIH3T3 cells were co-transfected with Gli-BS-Luc, pRL-SV40 (internal control), and 14-3-3 inhibitor EYFP-difopein or its control EYFP-R18Lys expression vector. After treatment as A, luciferase activity were determined. The data shown are representative of three independent experiments performed in duplicate (error bars). **, p Ͻ 0.01; *, p Ͻ 0.05.

Negative Regulation of Gli by PKA and 14-3-3
FEBRUARY 5, 2010 • VOLUME 285 • NUMBER 6 onstrated; however the overexpression of the binding mutant Gli1 did not increase the transcriptional activity (Fig. 4A). The binding of Gli1 to 14-3-3 might have an alternative effect for Hh signaling, different from those of Gli2 or Gli3. The 14-3-3⑀ isoform, which we have identified here as a Gliinteracting partner, appears to play a particularly important role in neuronal development and tumorigenesis. The 14-3-3⑀ gene is located on chromosome 17p13, and heterozygous deletions in this chromosome result in human neuronal disorder, Miller-Dieker syndrome (MDS). Although mutations in the LIS1 gene PAFAH1B1, which is located in 17p13, contributes to MDS, the severity of MDS is suggested to be caused by the co-deletion of 14-3-3⑀ in addition to LIS1 (22). Toyo-oka et al. (23) showed that mice deficient in 14-3-3⑀ had defects in brain development and neuronal migration, whereas Muslin and coworkers (24) found that Xenopus embryos lacking 14-3-3⑀ had prominent axial patterning defects.
Inappropriate activation of Hh signaling occurs in ϳ60% of medulloblastomas and the most common medulloblastomaassociated genetic lesion is loss of the chromosome 17p13 region (35). This evidence supports the importance of chromosome 17p13 in neuronal development and tumorigenesis. Determining whether the novel interaction between Gli and 14-3-3, especially the 14-3-3⑀ isoform, is involved in these human neurological diseases will require further investigation.