Involvement of GSK-3β and DYRK1B in Differentiation-inducing Factor-3-induced Phosphorylation of Cyclin D1 in HeLa Cells*

Differentiation-inducing factors (DIFs) are putative morphogens that induce cell differentiation in Dictyostelium discoideum. We previously reported that DIF-3 activates glycogen synthase kinase-3β (GSK-3β), resulting in the degradation of cyclin D1 in HeLa cells. In this study, we investigated the effect of DIF-3 on cyclin D1 mutants (R29Q, L32A, T286A, T288A, and T286A/T288A) to clarify the precise mechanisms by which DIF-3 degrades cyclin D1 in HeLa cells. We revealed that T286A, T288A, and T286A/T288A mutants were resistant to DIF-3-induced degradation compared with wild-type cyclin D1, indicating that the phosphorylation of Thr286 and Thr288 were critical for cyclin D1 degradation induced by DIF-3. Indeed, DIF-3 markedly elevated the phosphorylation level of cyclin D1, and mutations introduced to Thr286 and/or Thr288 prevented the phosphorylation induced by DIF-3. Depletion of endogenous GSK-3β and dual-specificity tyrosine phosphorylation regulated kinase 1B (DYRK1B) by RNA interference attenuated the DIF-3-induced cyclin D1 phosphorylation and degradation. The effect of DIF-3 on DYRK1B activity was examined and we found that DIF-3 also activated this kinase. Further, we found that not only GSK-3β but also DYRK1B modulates cyclin D1 subcellular localization by the phosphorylation of Thr288. These results suggest that DIF-3 induces degradation of cyclin D1 through the GSK-3β- and DYRK1B-mediated threonine phosphorylation in HeLa cells.

Cyclin D1 is synthesized in the early G 1 phase and plays a key role in the initiation and progression of this phase (9,10). There is a destruction box-like motif in the N terminus, which is required for cyclin D1 degradation induced by genotoxic stress (11). The expression level of cyclin D1 is regulated by ubiquitindependent mechanism, which is triggered by the phosphorylation of threonine residues located near the carboxyl terminus of cyclin D1 (12). Diehl et al. (13) reported that GSK-3␤ phosphorylates cyclin D1 on Thr 286 , thereby stimulating cyclin D1 turnover in response to mitogenic signals. According to the model they proposed, cyclin D1 phosphorylation on Thr 286 by GSK-3␤ induces the exclusion of cyclin D1 from nucleus to initiate its proteasomal degradation. Recently, Zou et al. (14) reported that dual-specificity tyrosine phosphorylation regulated kinase 1B (DYRK1B), a member of the DYRK family, phosphorylates cyclin D1 on Thr 288 , also resulting in its degradation.
GSK-3␤, a member of the Wnt signaling pathway, is a serine/ threonine kinase involved in a variety of cellular processes. Although GSK-3␤ is a cytosolic protein, it is translocated into the nucleus when activated (15)(16)(17). DYRK1B, also a serine/ threonine kinase found in nucleus (18), is highly expressed in normal skeletal muscle and certain carcinoma cell lines, including HeLa cells, and is not detectably expressed in many normal tissues (19). There are some similarities between GSK-3␤ and DYRK1B, because they phosphorylate the same substrates (glycogen synthase and cyclin D1) (14,20).
In this study, we investigated the effect of DIF-3 on cyclin D1 and found that the threonine residues, but not the destruction box-like motif, play a key role in cyclin D1 degradation induced by DIF-3 in HeLa cells. Moreover, we revealed that not only GSK-3␤ but also DYRK1B was involved in the phosphorylation of cyclin D1 induced by DIF-3 in HeLa cells. Thus DIF-3 efficiently induced degradation of cyclin D1.
Cell Culture and Transfection-HeLa cells were grown in Dulbecco's modified Eagle's medium (Sigma) supplemented with 10% fetal bovine serum, 100 units/ml penicillin G, and 100 g/ml streptomycin. The wild-type human cyclin D1 cDNA was subcloned into pcDNA3 (Invitrogen) as described previously (6). Human cyclin D1 mutants R29Q, L32A, T286A, T288A, and T286A/T288A were generated by a QuikChange Site-directed Mutagenesis kit (Stratagene). The FLAG-tagged human cyclin D1 constructs were also generated. Human cyclin D1 pGL3 basic luciferase reporter construct was a generous gift from Drs. O. Tetsu and F. McCormick, University of California San Francisco. Transfection was carried out using Lipofectamine TM Plus transfection reagent (Invitrogen).
RNA Interference (RNAi)-GSK-3␤ validated Stealth TM RNAi was purchased from Invitrogen. Double-stranded Stealth TM RNAi specifically targeting human DYRK1B (5Ј-gccugguauuugagcugcuguccua-3Ј) was synthesized (Invitrogen). Transfection of RNAi was carried out according to the manufacturer protocol using Lipofectamine TM 2000 transfection reagent (Invitrogen). Stealth TM RNAi negative controls, which are the GC-matched scrambled sequence, were also purchased from Invitrogen.
Western Blot Analysis-Samples were separated by 12% SDS-PAGE and transferred to a polyvinylidene difluoride membrane using a semi-dry transfer system (1 h, 12 V). After blocking with 5% skim milk for 1 h, the membrane was probed with a first antibody. The membrane was washed three times and incubated with horseradish peroxidase-conjugated anti-rabbit or anti-mouse IgG (Cell Signaling Technology) for 1 h. Immunoreactive proteins on the membrane were visualized by treatment with a detection reagent (LumiGLO, Cell Signaling Technology). An optical densitometric scan was performed using Science Lab 99 Image Gauge Software (Fuji Photo Film).
Luciferase Reporter Assay-HeLa cells were transfected with luciferase reporter plasmid and pRL-SV40, a Renilla luciferase expression plasmid (Toyo Ink Mfg Co) for the control of transfection efficiency. Cells were cultured for 24 h after transfection and stimulated with DIF-1 (30 M) for 1 h. The luciferase activity was determined with a luminometer (Lumat LB 9507, Berthold Technologies) and normalized with respect to Renilla luciferase activity. 32 P Metabolic Labeling and Immunoprecipitation-HeLa cells were preincubated in phosphate-free medium (Invitrogen) for 1 h and ALLN (20 M) was added to prevent cyclin D1 degradation. Subsequently, cells were labeled with 100 ci/ml of [ 32 P]orthophosphate (PerkinElmer) for 1 h followed by stimulation with DIF-3. Cells were lysed on ice in the lysis buffer (150 mM NaCl, 5 mM NaF, 2 mM Na 3 VO 4 , 50 mM Tris/HCl, pH 7.4, 1 mM EDTA, 1% (v/v) Tween 20, and 2 mM phenylmethylsulfonyl fluoride) and insoluble cell debris was removed by centrifugation at 5000 rpm for 3 min. The cell lysate was precleared with protein G-Sepharose CL-4B (Amersham Biosciences) and then incubated with protein G-Sepharose CL-4B and an anticyclin D1 antibody or an anti-FLAG antibody (2 g) at 4°C for 3 h. After incubation, proteins bound to the antibody/protein G-Sepharose complex were precipitated by centrifugation at 15,000 rpm for 5 min and washed three times with the lysis buffer. The samples were denatured in SDS sample buffer and separated by 12% SDS-PAGE, followed by the transfer to a polyvinylidene difluoride membrane prior to autoradiography. The radioactive bands were quantified using Science Lab 99 Image Gauge Software (Fuji Photo Film). After autoradiography, the membrane was subjected to Western blot analysis for cyclin D1.
In Vitro DYRK1B Kinase Reaction-The in vitro kinase assay of DYRK1B was carried out according to Lim et al. (22). Briefly, cells were stimulated with or without DIF-3 and immunoprecipitation was carried out using 2 g of anti-DYRK1B antibody. The immunoprecipitated samples were washed twice with lysis buffer and twice with a kinase assay buffer (20 mM Tris/HCl, pH 7.4, 5 mM MgCl 2 , and 1 mM dithiothreitol). The kinase activities of DYRK1B were tested with 30 l of kinase assay buffer con-  Flow Cytometry-Cells harvested by the trypsin/EDTA treatment were suspended in hypotonic fluorochrome solu-tion containing 50 g/ml of propidium iodide (PI), 0.1% sodium citrate, and 0.1% Triton X-100 (6). Cells (5 ϫ 10 3 ) for each sample were analyzed for fluorescence by a Becton-Dickinson FACScalibur.
Purification of Nucleic Proteins-Nucleic proteins were purified from cells transfected with indicated plasmid using NE-PER TM nuclear and cytoplasmic extraction reagents (Pierce).
Fluorescence Microscopy-Cells plated on coverslips were transfected with FLAG-tagged cyclin D1 constructs. Immunofluorescence detection of FLAG-tagged cyclin D1 was performed as described previously using an anti-FLAG antibody (6).

Effect of DIF-3 on Cyclin D1 and
Its Mutants-There are at least three independent motifs in cyclin D1 involved in its degradation. The first one is the Arg-X-X-Leu destruction box (Arg 29 -X-X-Leu 32 ), which plays a major role in rendering cyclin D1 susceptible to degradation by ionizing radiation (11). The second one is Thr 286 , which is phosphorylated by GSK-3␤ to induce cyclin D1 degradation (12,13). The third one is Thr 288 , recently identified by Zou et al. (14), who reported that a serine/threonine kinase DYRK1B phosphorylates cyclin D1 at Thr 288 , also leading to the degradation of cyclin D1. Therefore, we investigated five different mutants of cyclin D1 (R29Q, L32R, T286A, T288A, and T286A/T288A) to determine the mechanisms by which DIF-3 induces cyclin D1 degradation in HeLa cells. As shown in Fig. 1A, the effect of DIF-3 on overexpressed wild-type cyclin D1 was similar to the effect on intrinsic cyclin D1. While R29Q and L32R mutants were fully responsive to DIF-3, T286A, T288A, and T286A/T288A mutations significantly reduced the effect of DIF-3 (Fig. 1B). Among these three mutations, T286A/ T288A most strongly resisted DIF-3. These results suggest that the phosphorylation of both Thr 286 and Thr 288 were critical for DIF-3-induced cyclin D1 degradation. To analyze the effect of DIF-3 on cyclin D1 promoter activity, we performed luciferase reporter assay using the human cyclin D1 promoter construct pGL3-basic vector (8). However, DIF-3 did not have significant effect on promoter activity after 1 h incubation (440757.8 Ϯ  DECEMBER  15929.4 (mean Ϯ S.E., n ϭ 6) for control cells and 425032.3 Ϯ 10135.5 (mean Ϯ S.E., n ϭ 6) for DIF-3-treated cells). This result is consistent with our previous report (6) which showed that DIF-3 reduced cyclin D1 mRNA amount after 3 h incubation. Because cells were treated with DIF-3 for 1 h to analyze the effect of DIF-3 on cyclin D1 mutants, we concluded that the reduction of cyclin D1 protein amount was caused by the protein degradation induced by DIF-3.

GSK-3␤ and DYRK1B Are Involved in DIF-3 Action
DIF-3-phosphorylated Cyclin D1 at Thr 286 and Thr 288 -We subsequently examined whether DIF-3 induces the phosphorylation of cyclin D1 in intact cells. We have previously reported that ALLN, which inhibits ubiquitin-proteasome-dependent degradation of cyclins, prevented the DIF-3-induced cyclin D1 degradation in HeLa cells. Therefore, cells were pretreated with ALLN to avoid cyclin D1 degradation induced by DIF-3 and metabolically labeled with [ 32 P]orthophosphate. Cyclin D1 was immunoprecipitated for analysis by autoradiography and immunoblotting. By the immunoprecipitation, 51.3 Ϯ 4.5% (mean Ϯ S.E., n ϭ 3) of cyclin D1 was cleared from the lysate. As shown in Fig. 2A, DIF-3 induced phosphorylation of cyclin D1 in a time-dependent manner. Pretreatment with LiCl, which inhibits GSK-3␤, prevented the phosphorylation of cyclin D1 induced by DIF-3 (Fig. 2B). To determine whether Thr 286 and Thr 288 residues are involved in DIF-3-induced cyclin D1 phosphorylation, mutated cyclin D1 proteins (T286A, T288A and T286A/ T288A) were overexpressed in HeLa cells. The effect of DIF-1 was greatly attenuated in the T286A mutant and significantly reduced in the T288A cyclin D1 mutant (Fig.  2C). T286A/T288A double mutation almost completely abolished the phosphorylation induced by DIF-3. This result was well correlated with the result shown in Fig.  1B. Thus, DIF-3 seemed to induce cyclin D1 phosphorylation at Thr 286 and Thr 288 , triggering cyclin D1 degradation.
Inhibition of GSK-3␤ Attenuated DIF-3-induced Cyclin D1 Degradation-We previously reported that DIF-3 activated GSK-3␤ and induced cyclin D1 degradation by the acceleration of ubiquitin-proteasomedependent proteolysis in HeLa cells (6). To elucidate the role of GSK-3␤ in DIF-3-induced cyclin D1 degradation, HeLa cells were pretreated with SB216763, a specific GSK-3␣ and ␤ inhibitor. As shown in Fig. 3A, although SB216763 (20 M) attenuated the degradation of cyclin D1 induced by DIF-3, this compound could not fully inhibit DIF-3 action, even if this concentration of SB216763 caused complete inhibition of GSK-3␤ activity in in vitro kinase assay using phospho-Glycogen Synthase Peptide-2 (Upstate Biotechnology) as substrate (data not shown). Subsequently, we attempted to deplete endogenous GSK-3␤ using RNAi to determine the involvement of GSK-3␤ in DIF-3 action. As shown in Fig. 3B, the protein level of GSK-3␤ was markedly reduced by transfection with GSK-3␤ RNAi. The depletion of GSK-3␤ by transfection with GSK-3␤ RNAi attenuated the effect of DIF-3 on cyclin D1 degradation, while control RNAi (GC-matched scrambled sequence) did not have a significant effect (Fig. 3C). These results clearly indicated the involvement of GSK-3␤ in the DIF-3-induced cyclin D1 degradation but GSK-3␤ was unlikely to be the only kinase responsible for cyclin D1 degradation induced by DIF-3.
Depletion of DYRK1B Attenuated the Cyclin D1 Degradation Induced by DIF-3-Recently, it has been reported that DYRK1B phosphorylates cyclin D1 at Thr 288 leading to the proteolysis of cyclin D1 (14). Because DIF-3 induced phosphorylation of Thr 288 , the involvement of DYRK1B in the DIF-3-induced degradation of cyclin D1 was investigated. For this purpose, endogenous DYRK1B was depleted with RNAi. As shown in Fig. 4A, the protein level of DYRK1B was markedly reduced by trans- fection with DYRK1B RNAi. Although the control RNAi (GC matched scrambled sequence) did not have a significant effect, the depletion of DYRK1B significantly attenuated the effect of DIF-1 on cyclin D1 degradation (Fig. 4B), suggesting the involvement of DYRK1B in cyclin D1 degradation induced by DIF-3.
DIF-3 Activated DYRK1B-To clarify the effect of DIF-3 on DYRK1B activity, in vitro kinase assay was carried out using MBP as substrate (22). HeLa cells were stimulated with or with-out DIF-3 for 30 min and DYRK1B was immunoprecipitated to subject to kinase assay. As shown in Fig.  5A, DIF-3 significantly activated DYRK1B by 200% of control after 30 min incubation. Since LiCl almost completely inhibited cyclin D1 phosphorylation induced by DIF-3, the effect of LiCl on DYRK1B activity was examined. Interestingly, LiCl also attenuated the DYRK1B activity but failed to complete inhibition of this kinase (Fig. 5A). This result was agreeable with previous report which observed that LiCl strongly inhibit GSK-3␤ and weakly but significantly attenuated DYRK1B (14). We subsequently examined the effect of DYRK1B RNAi. As shown in Fig. 5B, DYRK1B RNAi markedly reduced the kinase activity to 30% of control and this reduction was well correlated with the reduction of DYRK1B protein level by RNAi shown in Fig. 4A.
GSK-3␤ and DYRK1B Phosphorylated Thr 286 and Thr 288 on Cyclin D1, Respectively, and Modified Cyclin D1 Subcellular Localization-Next, we examined the effect of depletion of GSK-3␤ and DYRK1B on cyclin D1 phosphorylation induced by DIF-3. As shown in Fig. 6A, the depletion of GSK-3␤ or DYRK1B significantly reduced but did not abolish the cyclin D1 phosphorylation induced by DIF-3. Subsequently, we examined the effect of depletion of GSK-3␤ or DYRK1B on phosphorylation of cyclin D1 mutants to clarify which site(s) on cyclin D1, Thr 286 or Thr 288 , are phosphorylated by GSK-3␤ or DYRK1B. For this experiment, we used FLAG-tagged wild-type and mutated cyclin D1 expression vectors. Fig. 6B showed that DIF-3 failed to induce phosphorylation on T288A cyclin D1 mutant after GSK-3␤ depletion. On the other hand, T286A cyclin D1 mutant was not significantly phosphorylated by DIF-3 treatment after depletion of DYRK1B. These results suggested that Thr 286 was phosphorylated by GSK-3␤ and Thr 288 was phosphorylated by DYRK1B, similarly to the previous reports (13,14). We further examined the subcellular localization of wild-type and mutant cyclin D1. As shown in Fig.  6C and D, FLAG-tagged T286A and T288A cyclin D1 mutants were abundant in nucleus compared with wild-type. DIF-3

GSK-3␤ and DYRK1B Are Involved in DIF-3 Action
induced export of wild-type cyclin D1 from the nucleus, whereas T286A and T288A mutations significantly reduced the effect of DIF-3, suggesting that not only Thr 286 but also Thr 288 plays an important role to modulate the subcellular localization of cyclin D1. Taken together, Thr 286 and Thr 288 were likely to be phosphorylated by GSK-3␤ and DYRK1B, respectively, and subcellular localization of cyclin D1 seemed to be modulated by GSK-3␤ and DYRK1B.
GSK-3␤ and DYRK1B Were Independently Involved in Cyclin D1 Degradation Induced by DIF-3-Furthermore, to investigate the relationship between GSK-3␤ and DYRK1B in DIF-3 action, RNAi of GSK-3␤ and DYRK1B were co-transfected. As shown in Fig. 7A, co-transfection of GSK-3␤ and DYRK1B RNAis exhibited an additive effect for cyclin D1 degradation, suggesting that GSK-3␤ and DYRK1B independently induce the phosphorylation and the degradation of cyclin D1. Because we previously reported that cyclin D1 depletion is associated with cell cycle arrest following exposure to DIF-3 (6), we examined the effects of reduction of GSK-3␤ and/or DYRK1B on DIF-3induced cell cycle arrest. Although transfection of GSK-3␤ or DYRK1B RNAi did not have a significant effect on the DIF-3-induced increase in the number of G 0 /G 1 cells, co-transfection of GSK-3␤ and DYRK1B RNAis significantly attenuated the effect of DIF-3 (Fig. 7B). This result indicated that GSK-3␤ and DYRK1B played important roles in cell cycle arrest induced by DIF-3 in HeLa cells.   DECEMBER 15, 2006 • VOLUME 281 • NUMBER 50

DISCUSSION
Cyclin D1 degradation is facilitated by the phosphorylation of the specific threonine residues 286 and 288, according to previous reports (12)(13)(14). In this study, we found that both residues play important roles in DIF-3-induced cyclin D1 degradation.
It has been reported that GSK-3␤ and IB kinase ␣ (IKK ␣) phosphorylate Thr 286 (23) and that DYRK1B phosphorylates Thr 288 of cyclin D1. Although we examined the effect of DIF-3 on IKK␣, DIF-3 did not activate this kinase (data not shown). Therefore, GSK-3␤ seemed to be the only candidate kinase that phosphorylated cyclin D1 at Thr 286 upon DIF-3 stimulation. The activity of GSK-3␤ is increased by the dephosphorylation of Ser 9 (15)(16)(17) and we previously reported that DIF-3 induced the dephosphorylation of Ser 9 and stimulated GSK-3␤ activity (6). Akt and p90 RSK , which are activated by phosphatidylinositol 3-kinase (PI3K) and MAPK cascade, respectively, are candidate enzymes for the phosphorylation of GSK-3␤ at Ser 9 (24,25). However, as we reported previously, DIF-3 did not suppress Akt and even activated p90 RSK (6). MKK3 has been reported to enhance DYRK1B kinase activity (22) and we found that DIF-3 also activated MKK3. 3 This result might indicate that DIF-3 activates DYRK1B through MKK3 activation. However, we could not address the mechanisms how DIF-3 activates GSK-3␤, DYRK1B, and/or MKK3 at present. Further study is required to clarify this point.
Shimizu et al. (4) reported that PDE1 is a pharmacological target molecule for the DIF-1. Although they showed that DIF-1 strongly inhibited PDE1 activity, they also reported that specific PDE1 inhibitor, 8-methoxymethyl-3-isobutyl-1-methylxanthine (8-MIBMX) only weakly inhibited cell proliferation in K562 human leukemia cells at the concentration of 300 M and failed to mimic the effect of DIF-1. Moreover, we found that PDE inhibitor, 3-isobutyl-1-methylxanthine (IBMX), did not induce cyclin D1 degradation (data not shown). Therefore, it is unlikely that DIFs induce cyclin D1 degradation through the inhibition of PDE1.
Recently, we reported that phosphorylation of Thr 286 is a crucial event in DIF-1 action to induce cyclin D1 degradation in the squamous cell carcinoma cell line NA, since a T286A mutant of cyclin D1 was much more stable compared with a T288A mutant after DIF-1 treatment (7). In this study, we showed that both Thr 286 and Thr 288 residues were strongly phosphorylated by DIF-3 and T286A and T288A cyclin D1 mutants were resistant to DIF-3 treatment in HeLa cells, suggesting that not only Thr 286 but also Thr 288 plays an important role in DIF-3 action. This difference might be caused by the difference of the expression level of DYRK1B in NA cells and HeLa cells. To our knowledge, no studies on the expression of DYRK1B in NA cells have been reported, whereas HeLa cells have been reported to highly express DYRK1B (19).
The destruction box-like motif in cyclin D1 (Arg 29 -X-X-Leu 32 ) has been reported to be required for cyclin D1 degradation induced by genotoxic stress (11). To examine the involvement of this motif in DIF-3 action, the effect of DIF-3 on two different mutants (R29Q, L32A) were analyzed. We found that these mutants were fully responsive to DIF-3 treatment, indicating that the destruction box-like motif is not required for DIF-3-induced cyclin D1 degradation. This motif exists in cyclin D1 but not in cyclins D2 and D3. We reported that DIF-3 degrades not only cyclin D1 but also cyclins D2 and D3 (6). Therefore, this result is in agreement with our previous observations.
In this study, we showed that GSK-3␤ and DYRK1B, both of which phosphorylate cyclin D1 to induce its degradation, were involved in DIF-3 action. This may have an important implication in DIF-3-induced cyclin D1 degradation, since DIF-3 induces rapid and strong degradation of cyclin D1 (within 1 h). In tumor cells, genes that directly regulate the cell cycle are often damaged. Among them, cyclin D1 is one of the genes strongly implicated in oncogenesis (10). Amplification of the gene encoding cyclin D1 and overexpression of cyclin D1 protein have frequently been demonstrated in several types of human malignant neoplasms (26 -29). Moreover, DIFs have been reported to inhibit PDE1 activity (4), and some inhibitors for PDE1 are expected to be applicable to cancer (30,31). Therefore, it could be suggested that DIFs are potent antitumor agents and identification of the target molecule(s) for DIFs may offer ideas for the design of new anticancer drugs.