The Plk1 kinase negatively regulates the Hedgehog signaling pathway by phosphorylating Gli1 Free

Hedgehog (Hh) signaling is a highly conserved pathway important for cell life, metabolism, individual development and tumorigenesis (Corbit et al., 2005; Corrales et al., 2004; Forbes et al., 1996; Rohatgi et al., 2007; Tukachinsky et al., 2010; Wang et al., 2010). The Hh signaling pathway consists of several key components, including patched (Ptc), smoothened (Smo), suppressor of fused (Sufu) and glioma-associated oncogene (Gli) proteins. Ptc1 (also known as PTCH1) is a cell membrane-localized 12-transmembrane protein that has been extensively studied as an Hh ligand receptor. Smo is a G protein-coupled receptor-like seven-transmembrane protein that functions as a transducer of Hh signaling (Corbit et al., 2005; Møller et al., 2017; Yang et al., 2012). Most Smo protein is stored in membrane vesicles within the cell. In the absence of Hh ligands, Ptc1 prevents and inhibits cell surface accumulation and activation of Smo. This inhibition is lifted by ligand binding to the receptor, which promotes endocytosis of Ptc from the cell membrane, and enhances the accumulation of Smo on the cell membrane and its activation (Hui and Angers, 2011). The activated Smo then transduces the upstream signal to the downstream Gli family transcription factors. Once activated, these Gli transcription factors translocate to the nucleus and initiate the transcription of their target genes (Ingham et al., 2011; Lum and Beachy, 2004; Robbins et al., 2012; Youn et al., 2016). Three members of the Gli family, Gli1, Gli2 and Gli3, have been identified in mammals, and only one Gli member, Cubitus interruptus (Ci), was identified in the fly Drosophila melanogaster, with this protein mediating all aspects of the transcription of Drosophila Hh target genes (Méthot and Basler, 2001). Among the three Gli members in mammals, Gli2 shows both transcriptional activator and repressor activities, Gli3 serves mainly as a transcriptional repressor, and Gli1 is thought to be driven by Hh signaling as a sensitive readout of this pathway activation (Ahn and Joyner, 2004; Altaba, 1999; Dessaud et al., 2008; Samanta et al., 2015; Sasaki et al., 1999).

Hh signaling is mainly transduced in the primary cilium, a slender organelle emanating from the cell surface, in ciliated cells of vertebrates. The primary cilium consists of a basal body, a microtubule-based axoneme generated from the basal body, and a signaling-receptor-enriched ciliary membrane sheet extending from the cell membrane. Between the ciliary membrane sheet and the cell membrane is a periciliary diffusion barrier (PDB), a transition zone that forms a selective barrier to prevent the membrane proteins from undertaking lateral transportation between the ciliary and cell membranes. Under the resting conditions of Hh signaling, Ptc1 localizes to the primary cilium and prevents the localization of Smo within the cilia or limits its residency time (Hui and Angers, 2011). Once bound with Hh ligands, Ptc1 is removed from the membrane, and its inhibitory effect on the relocation of Smo to the primary cilia membrane is lifted. This leads to the relocation of Smo into the primary cilia membrane and the activation of Hh signaling.

Transportation of the ciliary proteins between the cytoplasm and the cilium is bi-directional along the microtubule-based axoneme and is mediated by a multiprotein complex, the intraflagellar transport (IFT) complex. Once translocated to the cilia and activated, Smo modulates the activity of the Gli family transcription factors for their target genes.

The primary cilium is dynamic, and undergoes disassembly or resorption during G2-M phase transition and reassembly at the end of mitosis in a cell cycle regulation-dependent manner (Plotnikova et al., 2008; Pugacheva et al., 2007; Wang et al., 2013; Zhang et al., 2017, 2015). This not only indicates that ciliogenesis is regulated during the cell cycle but also suggests that Hh signaling is linked to cell cycle control. In this work, we investigated Hh signaling during the cell cycle. We found that Hh signaling turns off in G2 phase upon the phosphorylation and degradation of Gli1, turns on again during the G1 and S phase of the next cell cycle, and participates in the regulation of expression of G1/S transition regulators, such as cyclin E.

To investigate the coupling of cell cycle control with Hh signaling, we first examined the expression of Gli1, Smo and Sufu in several cell lines. Western blotting analysis revealed their existence in these cell lines (Fig. S1A). Since downstream members of the Hh signaling pathway have strong effects on the activity of this pathway, we focused on the regulation of the Gli transcription factors. When we expressed GFP-tagged exogenous Gli1 protein, we observed that the exogenous Gli1 proteins localized mostly in the nucleus and modestly in the cytoplasm, while exogenous Gli2 was predominantly localized in the nucleus (Fig. S1B), indicating that the Gli1 nuclear–cytoplasm transduction might be more active. Then, we treated cells with the Hh signaling agonist SAG and the antagonist cyclopamine, and analyzed the expression and localization of endogenous Gli proteins. We found that Gli1 was highly expressed and enriched in the nucleus upon Hh signaling stimulation by SAG, whereas the expression and nuclear localization of Gli1 were reduced upon Hh signaling inhibition upon treatment with cyclopamine (Fig. 1A,B). Through cell cycle synchronization followed by western blotting analysis, we further revealed that Gli1 expression was increased from the G1 to S phase and decreased during the G2 and M phases, and this expression was restored in the next G1 phase (Fig. 1C–E). Furthermore, we found that Gli1 shuttled between the nucleus and the cytoplasm with more in the nucleus during the G1–S phases and more in the cytoplasm at the G2 phase (Fig. 1D,E). To further verify the change of Gli1 protein levels during the cell cycle, we synchronized NIH 3T3 cells via serum starvation for 48 h followed by western blotting analysis. We found that Gli1 protein levels were also downregulated at the G2/M phase in NIH 3T3 cells (Fig. 1F). Taken together, these results show that the expression and localization of Gli1 is dependent on cell cycle control, and its downregulation occurs during the G2/M phases.

Next, we investigated how the Gli1 protein level is downregulated during the G2/M phase. Among the cell cycle control proteins, Plk1 and Aurora A are two vital protein kinases that function during the G2/M phase and both take part in regulating primary cilium disassembly (Pugacheva et al., 2007; Wang et al., 2013). As the Hh pathway is transduced mainly in the primary cilium, we supposed that these two kinases might be responsible for the downregulation of Hh signaling pathway during the G2 and M phase. To verify their effects on Hh pathway activity, we treated cells with Bi2536, an inhibitor of Plk1, and MLN8237 (MLN), an inhibitor of Aurora A, and monitored changes in Hh pathway activity with a widely used dual-luciferase reporter system (Taipale et al., 2000). With this reporter system, we showed that Hh pathway activity increases under the stimulation with the Hh ligand, the agonist SAG and upon treatment with exogenous Gli1 (Fig. S2). Interestingly, we found that, when the cells were treated with the Plk1 inhibitor Bi2536, Hh signaling was also significantly increased whereas treatment with the Aurora A inhibitor MLN had no effect (Fig. 2A). These results suggest that Plk1 but not Aurora A inhibits Hh signaling activity. By using immunofluorescence labelling, we revealed more nuclear retention of Gli1 in SAG- or BI2536-treated cells than in the control cells (Fig. 2B). To better analyze the localization disparities, we treated cells with Plk1 inhibitor or Hh signaling agonist then separated the nuclear and cytoplasmic fractions, followed by western blotting analysis to determine the Gli1 protein levels. In cells treated with Bi2536 and SAG, we found that levels of Gli1 protein in the nucleus were significantly increased compared to that of the control (Fig. 2C). Furthermore, we found that the Gli1 bands in Bi2536- and SAG-treated samples were significantly down-shifted compared with the control sample (Fig. 2C), indicating that Gli1 was phosphorylated in Bi2536- and SAG-untreated cells. Moreover, we synchronized cells at the G1/S or G2 phase, and treated them with Bi2536 or MLN. We found that Bi2536 treatment could increase Hh signaling activity only in the G2 phase (Fig. 2D). Moreover, we revealed that Plk1 inhibition or expression of the constitutively active mutant Plk1 T210D had no effect on the nuclear localization of Gli2 (Fig. S3A,B). Collectively, these data demonstrate that the phosphorylation of Gli1 by Plk1 reduces its nuclear retention and Hh signaling activity.

Next, we investigated whether Gli1 interacts with Plk1. Through co-immunoprecipitation (IP), we found that Plk1 interacted with Gli1 via its polo-box domain (PBD), which serves as a phosphorylation-dependent binding site (Fig. 3A,B), and this interaction was much weaker between Plk1 and Gli2 (Fig. S3C). Given that double mutation of histidine 538 (H538) and lysine 540 (K540), which are located within the PBD, to alanine residues (2A, H538A/K540A) abolishes the interaction between Plk1 and its substrates (Barr et al., 2004), we investigated the interaction between Plk1-2A and Gli1 and found no interaction between them (Fig. 3C). These findings indicated that Gli1 might be a substrate of Plk1. By synchronizing GFP- or GFP–Gli1-expressing cells at the G1/S transition and S phase by double-thymidine treatment and release, or at G2/M transition and M phase by treatment with S-trityl-L-cysteine (STLC), followed by IP, we found that the interaction between Plk1 and Gli1 was significantly increased during the G2/M transition and M phase (Fig. 3D), which is consistent with the time window for Gli1 protein level decrease. Immunostaining also revealed the colocalization between Gli1 and Plk1. In cells treated with Bi2536, Gli1 was still localized in nucleus even with exogenous expression of Plk1 (Fig. 3E). Taken together, these results indicate that Gli1 is a Plk1 substrate and that Plk1 is involved in the downregulation of Gli1.

We further tested whether Gli1 is phosphorylated by Plk1. Through sequence analysis, we found that serine 481 in the Mus musculus Gli1 molecule may be a Plk1 phosphorylation site, and it is conserved at serine 479 in human Gli1 (Fig. 4A). Then, we used a Phos-tag acrylamide gel and western blotting of SAG- and Bi2536-treated cells to analyze the modification of Gli1 proteins. We found that the Gli1 band was upshifted in the control and SAG treatment samples, but that this shift was inhibited upon the treatment with the Plk1-specific inhibitor Bi2536 (Fig. 4B), indicating that Plk1 phosphorylates Gli1. To confirm this, we mutated serine 481 to an alanine residue and performed an in vitro phosphorylation assay. We found that the [³²P]S481A mutant band was much weaker than the wild-type Gli1 band (Fig. 4C). We further performed in vivo phosphorylation mass spectrometry (MS) by constructing a Gli1 2R mutant, in which both asparagine 472 and cysteine 488 were mutated into arginine residues, to generate a cleavage site for trypsin (Fig. S4A). We confirmed that Gli1 2R showed no differences in localization and interaction with other proteins compared with wild-type Gli1 (Gli1 WT) (Fig. S4B,C). The MS results showed that Gli1 was indeed phosphorylated at S481 in vivo (Fig. 4D). By expressing the GFP-tagged phosphorylation-mimic mutant S481D and the non-phosphorylatable mutant S481A, followed by microscopy and nucleus–cytoplasm separation, we found that, compared with Gli1 WT, S481A showed increased nuclear localization, whereas S481D tended to localize in the cytoplasm (Fig. 4E–G). Taken together, these data showed that Gli1 is phosphorylated at S481 by Plk1, and this phosphorylation status may be linked with the cellular localization of Gli1 proteins.

Next, we investigated why Plk1 affects Hh signaling activity. Since Sufu is a well-known antagonistic factor of Gli proteins, we first tested whether Plk1 affects the interaction of Sufu and Gli1. By expressing Flag-tagged Sufu in cells, followed by immunostaining of endogenous Gli1, we found that nuclear localization of Gli1 was significantly downregulated in Flag–Sufu-expressing cells (Fig. 5A). By co-expressing both Sufu and Gli1 and undertaking co-IP assays, we found that these proteins interacted with each other. Interestingly, we confirmed that the interaction between Sufu and Gli2 was even stronger than the interaction between Sufu and Gli1, and, when GFP–Gli2 and Flag–Sufu were co-expressed in cells, Sufu could efficiently expel most Gli2 from nucleus (Fig. S5B). Then, we treated cells with the Plk1 inhibitor Bi2536 and performed co-IP assays. The results showed that, compared with the control, the Gli1 binding with both exogenous and endogenous Sufu was significantly reduced by the Plk1 inhibitor treatment (Fig. 5B,C). We further expressed Gli1 WT, S481A and S481D and performed co-IP assays. We found that the phosphorylation-mimicking mutant S481D as well as the wild-type Gli1 strongly bound endogenous Sufu, whereas the non-phosphorylatable mutant S481A did not bind (Fig. 5D). Immunostaining also revealed that S481D colocalized with Sufu (Fig. S5C). Taken together, these data demonstrate that phosphorylation of Gli1 at S481 by Plk1 enhances its binding with Sufu and mediates its cytoplasmic retention.

We further explored how phosphorylation of Gli1 by Plk1 enhances the binding of Gli1 with Sufu and its cytoplasmic retention. We transiently expressed Flag–Sufu in cells, treated the cells with the Plk1 inhibitor and examined cells for Gli1 by immunostaining. We found that nuclear localized Gli1 was decreased in Flag–Sufu-expressing cells compared to non-Flag–Sufu-expressing cells. However, when the cells were treated with Bi2536, we observed more nuclear localization of Gli1 in both the Flag–Sufu-expressing and non-Flag-Sufu-expressing cells (Fig. 6A). Statistical results showed that the percentage of cells with nuclear Gli1 in Bi2536-treated cells was approximately threefold higher than that in untreated control cells (Fig. 6B). These results were also confirmed by nucleus and cytoplasm separation followed by western blotting analysis (Fig. 6C). More straightforwardly, we observed that Gli1 vanished from the nucleus in cells expressing the constitutively active Plk1 T210D mutant (Fig. 6D). We also investigated the localization of Gli2 under Plk1 inhibitor treatment, and found that, regardless of the presence or absence of Bi2536, the localization of Gli2 in Flag–Sufu-expressing and non-Flag–Sufu-expressing cells was not changed (Fig. S6). Taken together, these results indicate that phosphorylation of Gli1 by Plk1 promotes the nuclear export of Gli1 by enhancing formation of the Gli1–Sufu complex.

To verify whether Hh signaling is negatively regulated by Plk1 through phosphorylation of Gli1, we tested the expression status of cyclin E1, a known transcriptional target of Gli1 and Hh signaling (Alvarez-Medina et al., 2009; Duman-Scheel et al., 2002). We found that the mRNA and protein levels of cyclin E1 were increased when the cells were transfected with exogenous Gli1 (Fig. 7A,B), whereas the cyclin E1 protein level was significantly decreased when the cells were transfected with exogenous Sufu (Fig. 7F). We observed that, while the mRNA and protein levels of cyclin E1 were increased in cells treated with the Plk1 inhibitor Bi2536 (Fig. 7C,D), this effect of Bi2536 treatment on cyclin E1 expression was abolished in cells expressing siRNA targeting Gli1 (Fig. 7E; Fig. S7). These findings reveal the indispensable role of Gli1 in the regulation of cyclin E1 expression with Bi2536 treatment. Furthermore, while exogenous Flag–Sufu in cells suppressed cyclin E1 expression, Bi2536 treatment could restore the cyclin E1 expression in these Flag–Sufu-expressing cells (Fig. 7F,G). Moreover, mutation of Gli1 into the Plk1 non-phosphorylatable form S481A, but not the phosphorylation-mimicking S481D form, could also enhance the mRNA and protein levels of cyclin E1 (Fig. 7H,I). Taken together, these results demonstrate that Plk1 negatively regulates Hh signaling by phosphorylating Gli1 to promote its nuclear exit.

Hh signaling regulates not only cell metabolism, individual development and tumorigenesis but also cell proliferation, although the underlying mechanisms remain unclear and even controversial (Agathocleous et al., 2007; Lum and Beachy, 2004; Neumann, 2005; Phua et al., 2017; Wang et al., 2010). Hh signaling functions mainly through the primary cilium in mammals (Goetz and Anderson, 2010). The primary cilium shows a dynamic behavior in a cell cycle regulation-dependent manner, with disassembly or resorption during the mitotic entry and reassembly after mitosis (Plotnikova et al., 2008; Pugacheva et al., 2007; Zhang et al., 2017, 2015). All these findings indicate a link between cell cycle control and Hh signaling, and that both cell cycle control and Hh signaling might precisely regulate each other. In this work, we studied the relationship between cell cycle control and Hh signaling. We demonstrated that one of the key cell cycle regulators, Plk1 kinase, negatively regulates Hh signaling by phosphorylating and inhibiting the transcriptional activity of Gli1.

Gli1 is the main effector of Hh signaling activation and plays crucial roles in the expression of the downstream target genes of this signaling pathway by serving as a transcriptional activator. The members of the Gli protein family can act as both transcriptional activators and repressors of Hh signaling (Hui and Angers, 2011). Of these Gli family members, Gli1 actively regulates the G1-S transition during neural progenitor proliferation in Drosophila (Alvarez-Medina et al., 2009) and the S phase checkpoint in tumor cells, promoting tumor progression and resistance to chemotherapy (Tripathi et al., 2014). In this work, we found that Gli1 activity is negatively regulated by Plk1 kinase, a putative target gene product modulated by Gli1 (Lee et al., 2010). Thus, our findings very likely reveal a feedback mechanism of Hh signaling for the G1-S phase transition, in which Gli1 regulates expression of its target genes, including Plk1, which in turn limits the activity of Gli1 to prevent overexpression of its downstream genes.

Plk1 kinase regulates many aspects of the cell cycle, especially the centrosomal cycle, mitosis and cytokinesis (Wang et al., 2014). Knockdown of Plk1 leads to G2/M arrest in cells, and this might have a negative effect on Hh signaling activity (Evangelista et al., 2008). Moreover, Plk1 has crucial roles in the regulation of ciliary resorption before mitotic entry (Wang et al., 2013; Zitouni et al., 2014). In this work, by controlling the duration of Plk1 inhibition (Bi2536, 2–3 h), we further show that Plk1 phosphorylation of Gli1 also promotes cell cycle progression into the G2/M phase by turning off the transcriptional activity of Gli1, in addition to the previous finding that Plk1 promotes DNA replication by phosphorylating Hbo1 to regulate pre-replicative complex (pre-RC) formation and DNA replication licensing (Wu and Liu, 2008).

Based on our findings, we propose a working model of the regulation of Hh signaling by Plk1 during the cell cycle (Fig. 7J). During mitotic exit and the G1 phase, Gli1 proteins are progressively expressed along with Hh signaling activation and move into the nucleus to function as transcriptional activators to induce target gene expression. These target gene products may include the regulators for the G1/S phase transition and DNA replication, such as cyclin E and Plk1, which then phosphorylate and inhibit Gli1 as a feedback regulation. Along with the progression of the cell cycle from G1 into S and G2/M phases, Plk1 accumulates, which, on the one hand, promotes cell cycle progression and, on the other hand, phosphorylates and inhibits Gli1 by enhancing the nuclear export of Gli1 and the interaction of Gli1 with Sufu, which sequesters Gli1 in the cytoplasm, leading to the switch-off of Hh signaling. With a unique structure of PBD domain, Plk1 normally requires a priming phosphorylation in its substrates for it to bind to them. Unfortunately, so far we do not know this priming kinase. Furthermore, we show that it is possible that Plk1 could also phosphorylate Gli2 (data not shown), but the phosphorylation sites and functions are still unknown. In general, our findings provide an insight into the crosstalk between Hh signaling pathway and cell cycle control, and may have important implications in the understanding of the link between cell signaling and cell proliferation.

Human Plk1 was cloned from a HeLa cell cDNA library. Mouse Gli1 was subcloned from the Flag–mGli1 plasmid, a kind gift from Dr Qing Zhang (Nanjing University, China). cDNAs encoding Plk1, Plk1-PBD, Plk1-PBD-2A and Gli1 WT, S481A and S481D were subcloned into the pEGFP-C2 (Clontech, 6083-1), pCMV-Myc (Clontech, 635689) or pET-28a (Novagen, 69864) vectors. The GFP–mGli2 and Flag–hSufu plasmids were kind gifts from Dr Yun Zhao (Institute of Biochemistry and Cell Biology, SIBS, CAS, Shanghai, China). The siRNA sequence targeting Gli1 was 5′-CCAGGAAUUUGACUCCCAA-3′. Three additional siRNAs targeting Gli1 were used in this project: siRNA1: 5′-CCGAGUAUCCAGGAUACAATT-3′, siRNA2: 5′-CCGAAGGACAGGUAUGUAATT-3′, siRNA3: 5′-CUUCCCACCUACUGAUACUTT-3′.

HeLa (ATCC, CCL-2), HEK 293T (ATCC, ACS-4500), HEK 293 (ATCC, CRL-1573), murine embryonic fibroblast (MEF; ATCC, SCRC-1008) and NIH 3T3 (ATCC, CRL-1658) cells were cultured at 37°C and 5% CO₂ in DMEM (Gibco) with 10% FBS (HyClone Laboratories, Inc.). Resting (G0 phase) cells were obtained by serum starvation. G1/S phase synchronization was achieved by double-thymidine (Sigma-Aldrich) treatment. Briefly, the cells were treated with 2.5 mM thymidine for 18 h twice with a 9 h internal release. The cells were harvested (500 g for 5 min) at the indicated time point after they were released to fresh medium. G2 phase cells were obtained with a 6 h release after the double thymidine treatment. G2/M phase cells were obtained by adding 100 ng/ml STLC (Sigma-Aldrich) for 15 h after release from the thymidine block. Bi2536 (Axon Medchem) (100 nM) was added to the medium for 2–3 h before harvest to manipulate the kinase activity of Plk1. Transient cDNA transfections were carried out on cells using MegaTrans transfection reagent (Origene) according to the manufacturer's instructions, and siRNA transfection was performed with Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions.

NIH 3T3 cells were pre-plated into 24-well plates. Then, the cells were transfected with a Gli-Luc reporter and Renilla luciferase pRL-SV40 (kind gifts from Dr Yun Zhao, Institute of Biochemistry and Cell Biology, Shanghai, China). We treated the cells with SAG (3 µM, 12 h), Bi2536 (100 nM, 2–3 h), MLN (250 nM, 2–3 h) or starved the cells to simulate their transition to G0 phase by using medium containing only 0.5% FBS for 24 h, the next day. Then, the cells were collected for luminometric detection using dual luciferase reagents (Promega). Transfection efficiency was normalized via the Renilla luciferase activity.

The cells were fixed in 4% PFA for 15 min and then permeabilized with PBS/0.2% Triton X-100 for 30 s, followed by immunostaining with the indicated primary [rabbit anti-Gli1 (1:100, ab49314, Abcam), rabbit anti-Gli2 (1:100, ab7195, Abcam), mouse anti-Plk1 (1:200, 06-813, EMD Millipore), mouse anti-γ-tubulin (1:200, T6557, Sigma-Aldrich) and mouse anti-Flag (1:200, PM020, MBL International)] and secondary antibodies [Alexa Fluor 488-conjugated donkey anti-mouse-IgG, Alexa Fluor 594-conjugated donkey anti-rabbit-IgG, Alexa Fluor 488-conjugated donkey anti-rabbit-IgG or Alexa Fluor 594-conjugated donkey anti-mouse-IgG (Invitrogen)]. DNA was stained with DAPI (Sigma-Aldrich). Images were analyzed under a 63×/1.4 NA oil objective of a microscope (Axiovert 200M; Carl Zeiss) and captured with a charge-coupled device (CCD) camera (MRM; Carl Zeiss) and Axiovert image acquisition software.

NIH 3T3 or HEK 293 cells transfected with indicated plasmids were lysed in cell lysis buffer (20 mM Tris-HCl pH 8.0, 150 mM NaCl, 2 mM EGTA, 0.5 mM EDTA, 0.5% NP-40, 5 mM NaF, 1 mM Na₃VO₄, 1 mM PMSF, and 500× protease inhibitor cocktail; Calbiochem) for 15 min on ice. The lysates were centrifuged at 9000 g for 15 min, and the supernatants were incubated with the rabbit anti-GFP polyclonal antibody-coated protein A–Sepharose beads for 2 h at 4°C. The rabbit anti-GFP polyclonal antibodies were raised against GFP protein in rabbit in our laboratory. Then, the beads were washed three times with lysis buffer and suspended in Laemmli sample buffer before resolution on SDS-PAGE gels. For GST pulldown assays, cell lysates were incubated with 5 µg of soluble GST or GST-fused proteins bound to 15 µl glutathione–Sepharose beads for 2 h at 4°C. Then, the beads were washed three times with lysis buffer and suspended in Laemmli sample buffer before resolution on SDS-PAGE gels.

After separation on SDS-PAGE gels, the proteins were transferred to nitrocellulose membranes that were then blocked in TTBS (20 mM Tris-HCl, pH 7.5, 500 mM NaCl and 0.1% Tween 20) containing 2% nonfat milk at room temperature for 1 h. Then, they were probed with primary antibodies diluted in TTBS containing 2% nonfat milk at 4°C overnight. The membranes were washed three times with TTBS before they were incubated with HRP-conjugated secondary antibody at room temperature for 1 h. After that, the membranes were washed three times with TTBS again. The membranes were developed for visualization by enhanced chemiluminescence (Sigma-Aldrich) and X-ray film. The following primary antibodies were used for immunoblotting: rabbit anti-Gli1 (1:500, V812, Cell Signaling Technology), rabbit anti-Smo (1:500, ab38686, Abcam), rabbit anti-cyclin B1 (1:1000, sc-25764, Santa Cruz Biotechnology, Inc.), rabbit anti-cyclin E1 (1:1000, A0112, ABclonal Technology), rabbit anti-Lamin A/C (1:1000, raised in our laboratory with Lamin A/C protein), mouse anti-Plk1 (1:1000, 06-813, EMD Millipore), mouse anti-Sufu (1:1000, sc-137014, Santa Cruz Biotechnology, Inc.), mouse anti-α-Tubulin (1:5000, T6074, Sigma-Aldrich), mouse anti-GFP (1:5000, M048-3, MBL International), mouse anti-Flag (1:1000, PM020, MBL International), mouse anti-Myc (1:1000, M4439-100UL, Sigma-Aldrich), mouse anti-H3pS10 (1:1000, ab47297, Abcam) and mouse anti-GAPDH (1:5000, 60004-1-lg, Proteintech). The western blot results for distinct proteins came from the same samples processed simultaneously in separate gels. All animal experiments were performed according to approved guidelines.

GFP–Gli1-WT or GFP–Gli1-481A proteins were purified from HEK 239T cells by immunoprecipitation. Beads coated with equal amounts of both proteins were combined with Plk1 kinase (PV3501, Life Technologies), 50 mM Tris-HCl pH 7.5, 10 mM MgCl₂, 2 mM EGTA, 5 mM DTT, 100 mM ATP and 1 µCi γ-[³²P]ATP (10 mCi/ml, 6000 Ci/mmol; GE Healthcare) for 30 min at 30°C. Loading buffer was added to stop the reaction. SDS-PAGE samples underwent electrophoresis. The gel was exposed to X-ray film for 6 h or overnight at 4°C. For mass spectrometry, GFP–Gli1-2R was purified from HEK 293T cells by immunoprecipitation and was electrophoresed by SDS-PAGE. The gel was stained with Coomassie Brilliant Blue to visualize the protein bands. The GFP–Gli1-2R bands were sliced and subjected to MS analysis (Zhang et al., 2015).

NIH 3T3 or HEK 293T cells were washed by low-permeability buffer A (20 mM HEPES-K pH 7.8, 5 mM KAc, 0.5 mM MgCl₂ and 0.5 mM DTT) and treated with the same buffer A at 2 ml/dish for 10 min on ice. Then, the cells were scraped from the dishes and homogenized using a homogenizer more than 20 times. Samples were then centrifuged at 1000 g for 5 min. The supernatants were taken as cytosol segments, and the sediments were taken as nuclei. The cytosol segments were centrifuged at 20,000 g for 30 min, and the supernatants were mixed with Laemmli sample buffer before resolution on SDS-PAGE gels. The nuclei were washed three times with PBS and centrifuged at 1500 g for 5 min each time. The nuclei were lysed with high permeability buffer B (10 ml buffer A and 0.234 g NaCl) at ∼1.5×10⁸ nuclei/ml for 9 min at 4°C. The lysates were centrifuged at 12,000 g for 30 min, and the supernatants were mixed with Laemmli sample buffer before separation on SDS-PAGE gels.

Total RNA was extracted and 1 µg RNA was reverse transcribed using PrimeScript™ RT reagent kit with gDNA Eraser (TaKaRa). Quantitative PCR (qPCR) was performed with a FastStart Universal SYBR Green Master (Rox) kit (Roche) on a Roche Light Cycler 96 machine using a SYBR Green qPCR template. The synthesis of cyclin E1 was detected using sense 5′-GGAGTTCTCGGCTCGCTCC-3′ and antisense 5′-CGTCCTGTCGATTTTGGCC-3′ primers. Cyclin E1 gene levels were normalized to GAPDH levels. The qPCR results were analyzed by Roche LightCycler 96 software.

We thank Drs Qing Zhang (Nanjing University, China) and Yun Zhao (Institute of Biochemistry and Cell Biology, SIBS, CAS, Shanghai, China) for reagents. We are grateful to other members of the laboratory for critical reading of this manuscript and Drs Hongxia Lv, Liying Du, and Dong Cao (Peking University, China) for technical support, and Dr John Olson (Peking University, China) for language editing.