Cyclin-Dependent Kinase 3–Mediated Activating Transcription Factor 1 Phosphorylation Enhances Cell Transformation

Cyclin-dependent kinase (cdk)-3, a member of the cdk family of kinases, plays a critical role in cell cycle regulation and is involved in G₀-G₁ and G₁-S cell cycle transitions. However, the role of cdk3 in cell proliferation, as well as cell transformation, is not yet clearly understood. Here, we report that the protein expression level of cdk3 is higher in human cancer cell lines and human glioblastoma tissue compared with normal brain tissue. Furthermore, we found that cdk3 phosphorylates activating transcription factor 1 (ATF1) at serine 63 and enhances the transactivation and transcriptional activities of ATF1. Results also indicated that siRNA directed against cdk3 (si-cdk3) suppresses ATF1 activity, resulting in inhibition of proliferation and growth of human glioblastoma T98G cells in soft agar. Importantly, we showed that cdk3 enhances epidermal growth factor–induced transformation of JB6 Cl41 cells and si-cdk3 suppresses Ras^G12V/cdk3/ATF1–induced foci formation in NIH3T3 cells. These results clearly showed that the cdk3-ATF1 signaling axis is critical for cell proliferation and transformation. [Cancer Res 2008;68(18):7650–60]

Cyclin-dependent kinases (cdk) are serine/threonine protein kinases that play essential roles in the control of cell cycle progression by interacting with a variety of regulators and substrates (1). In eukaryotic cells, cell cycle progression is driven by the sequential and periodic activation of cyclin/cdks, and dysregulation of the cell cycle is associated with cancer development (2). Very few mutations in cdks have been noted in human cancers with the exception of a p16^Ink4a-insensitive cdk4 mutation in human familial melanoma (3). However, dysregulation of cdk activity is found in many human cancers because of mutations or epigenetic alterations in upstream regulators of cdks or their substrates (1, 4). Overexpression of cdk6 has been observed in lymphomas, leukemias, and melanomas due to chromosomal translocation (5, 6). When cotransfected with activated H-Ras, cdk4 displayed oncogenic potential by provoking foci formation in primary rat embryo fibroblasts (7) and generating malignant human epidermal tumorigenesis (8). Cdk4 kinase activity was shown to be required for proliferation of breast cancer cells and mammary tumorigenesis in cdk4-null mice (9, 10) and kinase-deficient cyclin D1 knock-in mice (11). Hemizygous disruption of cdc25 was shown to inhibit cell transformation and mammary tumorigenesis in mice (12). Activation of cdk2 was required for human fibroblast proliferation and foci formation induced by cyclin E/SV40 small T antigen coexpression (13). Knockdown of cdk2 was shown to inhibit proliferation and colony formation of human melanoma cells (14), and ablation of cdk2 decreased Ras/cdk4-dependent malignant progression in mouse skin tumorigenesis models (15). Thus, cdks are closely associated with human cancer pathogenesis.

Cdk3 is an important regulator of cell cycle. The activity of cdk3 is first observed in early G₁ phase (16) and reaches a peak in mid-G₁ (17). A dominant-negative cdk3 was shown to induce G₁ arrest, which could not be rescued by cdk2, indicating that cdk3 plays an important role for G₁ exit to S entry (18, 19). Recently, cdk3 was found to form a complex with cyclin C and phosphorylate the retinoblastoma protein (pRb) at serine 807/811, which is required for G₀-G₁ transition (20). Furthermore, cdk3 seems to be expressed in various normal human tissues and cancer cell lines including glioblastoma and neuroblastoma cells (20–22). Ectopic overexpression of cdk3, but not cdk2, enhanced the proliferation and anchorage-independent growth of Ratl cells, which was associated with activation of Myc (17). The genetic locus of cdk3 was mapped to the chromosome 17q22-qter region, where cdk3 was involved in a chromosome rearrangement in a breast cancer cell line possibly resulting in an alteration of cdk3 gene expression or an abnormal transcript (23). However, although cdk3 might have an effect on cell proliferation and transformation, the precise role of cdk3 in carcinogenesis has not been clearly elucidated.

The activating transcription factor 1 (ATF1) is a member of a well-known transcription factor family, the cyclic AMP (cAMP) response element (CRE)-binding protein (CREB) family, which includes ATF1, CREB1, and the cAMP response element modulator (CREM; ref. 24). Both ATF1 (25) and CREB1 (26) are expressed ubiquitously, whereas CREM (27) is highly expressed in neuroendocrine tissues. In response to a variety of growth factors, stress signals, neurotransmitters, and other agents that elevate intracellular cAMP or Ca²⁺ levels, CREB family members are activated and promote the expression of a large number of cellular target genes that contain CRE elements in their promoters, including proto-oncogenes such as c-fos and c-jun; cell cycle genes such as cyclin D and cyclin A; and other genes related to cell growth, proliferation, and neuronal activities (28, 29).

Phosphorylation of ATF1 at serine 63 in its kinase-inducible domain (KID) by serine/threonine kinases enhances its transactivation activity by promoting recruitment of the coactivator CREB-binding protein/p300 (30). Furthermore, overexpression of ATF1 was found in lymphomas, transformed lymphocytes (31), and may contribute to the growth of these tumor cells. ATF1 was shown to be up-regulated in metastatic melanoma cells, and inhibition of ATF1 suppressed their tumorigenicity and metastatic potential in nude mice (32). Constitutive activation of ATF1 mediates the Ewing's sarcoma protein-activating transcription factor 1 (EWS-ATF1) transforming phenotypes and unique features of clear cell sarcoma (33). However, upstream kinases and the role of ATF1 in cell proliferation and cell transformation have not been completely identified.

In this study, we provide evidence showing that cdk3 is highly expressed in human glioblastoma and that ATF1 is a novel substrate of cdk3. The phosphorylation of ATF1 at serine 63 by cdk3 induced ATF1 transactivation and transcriptional activities. Ectopic expression of cdk3 enhanced the transformation of JB6 Cl41 cells, and knockdown of endogenous cdk3 suppressed the proliferation and growth of glioblastoma T98G cells in soft agar. Importantly, knockdown of cdk3 with siRNA against cdk3 (si-cdk3) suppressed foci formation in NIH3T3 cells induced by coexpression of Ras^G12V/cdk3/ATF1, indicating that the cdk3-ATF1 signaling axis plays an important role in cell transformation.

Reagents and antibodies. Chemical reagents, including Tris, NaCl, and SDS, for molecular biology and buffer preparation were purchased from Sigma-Aldrich Corp. Restriction enzymes were obtained from Roche Diagnostics Corp. and Taq DNA polymerase was from Qiagen, Inc. The DNA ligation kit (v.2.0) was purchased from TAKARA Bio, Inc. The Checkmate Mammalian Two-hybrid System was obtained from Promega. Cell culture medium and other supplements were from Life Technologies, Inc. Antibodies against cdk3, ATF1, and cyclin C were purchased from Santa Cruz Biotechnology, Inc., and the antibody to detect phosphorylated ATF1 was from Cell Signaling Technology, Inc. Anti-Flag was from Sigma-Aldrich. and anti-hemagglutinin was purchased from Roche Diagnostics. The Human Brain Glioblastoma Tissue Microarray was purchased from US Biomax, Inc., and the Cy3-conjugated donkey anti-rabbit IgG antibody was purchased from Jackson ImmunoResearch Laboratories, Inc.

Cell culture and transfections. HEK293 cells were cultured at 37°C in a 5% CO₂ incubator in DMEM supplemented with 10% fetal bovine serum (FBS). T98G cells were cultured in Eagle's MEM with 2 mmol/L l-glutamine, 0.1 mmol/L nonessential amino acids, 1.0 mmol/L sodium pyruvate, and 10% FBS. The media for cell culture were changed every 3 d and cells were split at 80% to 90% confluence. Transfection of the expression plasmids was done using jetPEI (Polyplus-transfection, Inc.) according to the manufacturer's instructions.

Reverse transcription-PCR and subcloning of transcription factors and cdk3. The cdk3 (pBIND-cdk3) and transcription factor (pACT-TF) cDNAs were cloned by PCR-based amplification as described previously (34). Briefly, human mRNAs were reverse transcribed with an oligo-dT primer and SuperScript II RNase H Reverse Transcriptase (Life Technologies). ATF1 and cdk3 cDNAs were amplified by PCR and the cDNAs were introduced into pACT or pBIND vectors as required and confirmed by DNA sequencing.

Construction of expression vectors. The glutathione S-transferase (GST)–tagged ATF1 and the deletion GST-ATF1 (residues 58–141) expression plasmids were constructed by PCR and subcloned into the SalI/NotI site of the pGEX-5X-C vector. The ATF1 cDNA was also subcloned into pcDNA4 (Invitrogen) to generate pcDNA4-ATF1. The plasmid constructs were confirmed by restriction mapping and DNA sequencing. The pRcCMV-cdk3 (pCMV-cdk3) was a gift from Dr. Barrett J. Rollins (Department of Medical Oncology, Harvard Medical School, Boston, MA). The expression plasmids, including pET28-ATF1, pET28-ATF1S63A, p3×Flag-ATF1, and p3×Flag-ATF1S63A, were kindly provided by Dr. Ron Prywes (Department of Biological Sciences, Columbia University, New York, NY). The deletion GST-ATF1 fusion vectors (1–57, 1–141, and 142–271) were gifts from Dr. Hiroshi Handa (Faculty of Bioscience and Biotechnology, Tokyo Institute of Technology, Yokohama, Japan). The mutations of cdk3 (408T/C, 411G/A, 412T/C, and 414G/C) to generate a si-cdk3–resistant cdk3 expression vector (rescue cdk3 vector) were carried out using the QuikChange II Site-Directed Mutagenesis Kit (Strategene) according to recommended protocols.

Mammalian two-hybrid assay. HEK293 cells were seeded and transfected as described previously (34). Briefly, the various DNAs, pBIND-cdk3 or pACT-TFs, and pG5-luciferease reporter plasmids were combined in the same molar ratio (1:1:1) and the total amount of DNA was not more than 100 ng/well. At 36 h after transfection, the cells were disrupted by the addition of lysis buffer [0.1 mol/L potassium phosphate buffer (pH 7.8), 2 mmol/L EDTA, 1 mmol/L DTT, 1% Triton X-100] and luciferase activity was measured using the Luminoskan Ascent (MTX Lab, Inc.). The luciferase activity was calculated against the pG5-luciferase basal control and normalized against Renilla luciferase activity. The relative activity was expressed as luminescence units normalized to a negative control (value for cells transfected with only pG5-luciferase reporter vector/pBIND-cdk3 = 1).

Western blotting and immunoprecipitation. Cells were harvested at 80% to 90% confluence and proteins were extracted in NP40 cell lysis buffer [50 mmol/L Tris-HCl (pH 8.0), 150 mmol/L NaCl, 0.5% NP40] with freezing and thawing. Protein concentration was measured using the detergent-compatible protein assay kit (Bio-Rad) for Western blotting. Proteins were transferred onto polyvinylidene difluoride membranes, hybridized with appropriate antibodies, and then visualized using the enhanced chemiluminescence detection kit (Amersham Biosciences). To identify potential protein interactions, the cdk3 (H-45) antibody was used for immunoprecipitation of T98G cell lysates (1 mg). Antibody binding was carried out at 4°C overnight and proteins were visualized by Western blotting.

In vitro kinase assay. The plasmids for GST-tagged or His-tagged fusion proteins were each respectively transformed into BL21 cells and induced with 0.5 mmol/L isopropyl-l-thio-β-d-galactopyranoside at 25°C for 3 h. The soluble GST-fusion proteins were purified using glutathione-Sepharose 4B beads and eluted with 10 mmol/L reduced glutathione in 50 mmol/L Tris-HCl (pH 8.0) buffer. The soluble His-fusion proteins were purified with Ni-NTA agarose (Qiagen). Purified fusion proteins (200 ng) or histone H1 (2 μg; Upstate Biotechnology, Inc.) was used for the in vitro kinase assay with 100 ng of active cdk3 (Upstate Biotechnology) or endogenous cdk3 immunoprecipitated from cell lysates (1 mg). Reactions were carried out in 1× kinase buffer containing 50 μmol/L unlabeled ATP with or without 10 μCi of [γ-³²P]ATP at 30°C for 30 min. Reactions were stopped and then proteins resolved by 12% SDS-PAGE and visualized by autoradiography or Western blotting.

ATF1 transactivation assay. T98G cells (8 × 10⁴) were seeded into 12-well plates for 24 h before transfection. The p5×Gal4-luciferase reporter plasmid (p5×Gal4-luc) was transfected with pCMV-cdk3 and the expression vectors for Gal4-ATF1 or Gal4-ATF1 S63A. The cells were cultured for 36 h and then disrupted for firefly luciferase activity analysis. The p5×Gal4-luc activity was normalized against Renilla luciferase activity.

Construction of si-cdk3 plasmids. The pU6pro vector (a gift from David L. Turner, University of Michigan, Ann Arbor, MI) was used to construct si-RNA against cdk3 (si-cdk3). The synthetic primers were annealed and then introduced into the pU6pro vector digested with XbaI and BbsI following the standard protocols found online.¹

The si-cdk3 construct was confirmed by DNA sequencing. An si-general scramble (si-mock) was constructed as previously described and used as a negative control (35).

MTS assay. To assess cell proliferation, psi-mock and psi-cdk3 stably transfected T98G cells (1 × 10³) were seeded into 96-well plates in 100 μL of 10% FBS/MEM and cultured at 37°C in a 5% CO₂ incubator. After culturing for 24, 48, or 72 h, 20 μL of the CellTiter 96 Aqueous One Solution (Promega) were added to each well and then incubated for 1 h at 37°C in a 5% CO₂ incubator. Absorbance was measured at 492 nm.

Anchorage-independent cell transformation assay. Epidermal growth factor (EGF)–induced cell transformation was examined in psi-mock and psi-cdk3 stably transfected T98G cells and in mock and pCMV-cdk3 stably transfected JB6 Cl41 cells. In brief, cells (8 × 10³) were exposed to EGF (10 ng/mL) in 1 mL of 0.3% Basal Medium Eagle agar/10% FBS. The cultures were maintained at 37°C for 1 to 2 wk and colonies were scored using a microscope and the Image-Pro PLUS computer software program (v.4, Media Cybernetics) as described by Colburn and colleagues (36).

Foci forming assay. Following standard protocols (37), NIH3T3 cells were transiently transfected with various combinations of H-Ras^G12V (100 ng), cdk3 (500 ng), ATF1 (500 ng), ATF1 S63A (500 ng), psi-cdk3 (500 ng), or pcDNA3-mock (compensation for equal amount of DNA) and then cultured in 5% calf serum/DMEM for 2 wk. Foci were fixed with methanol, stained with 0.5% crystal violet, and then counted using a microscope and the Image-Pro PLUS (v.4) software program.

Immunofluorescent detection. A human glioblastoma tissue array section was deparaffinized in xylene, rehydrated in alcohol, and the antigen was retrieved in 10 mmol/L sodium citrate buffer. The tissues were first blocked in 5% normal goat serum/PBS for 1 h at room temperature and then incubated overnight at 4°C with a 1:50 dilution of anti-cdk3 in PBS containing 1% bovine serum albumin. The tissues were washed and then incubated in the dark with a 1:200 dilution of a Cy3-donkey anti-rabbit antibody for 2 h at room temperature. Image stacks were captured using laser scanning confocal microscopy (NIKON C1^si Confocal Spectral Imaging System, NIKON Instruments Co.) and analyzed with the ImageJ (version 1.37v) software program. The intensity score of fluorescence from each sample was measured and the average intensity score from duplicate scores of each case was recorded as the cdk3 expression level.

Cdk3 is highly expressed in human cancer cells and glioblastoma tissue. To investigate whether cdk3 is involved in human cancer, we examined cdk3 protein expression level in 15 different human cell lines by Western blotting and cdk3 activity by immunoprecipitation kinase assay. The results indicated higher cdk3 protein levels in most of the cancer cell lines including glioblastoma T98G cells, large cell lung cancer H460 cells, colorectal carcinoma HCT116 cells, colorectal adenocarcinoma HT29 cells, prostate cancer PC-3 and DU-145 cells, osteosarcoma Saos-2 cells, epidermoid carcinoma A431 cells, and melanoma SK-MEI-28 and RPMI-1951 cells (Fig. 1A,, top). Furthermore, cdk3 activities in these cell lines are also accordingly higher (Fig. 1A,, bottom). The protein levels of cyclin C were also detected in these cell lines, and our data indicated that the protein expression pattern of cyclin C is not the same as that of cdk3 (Fig. 1A,, top). Others have reported high levels of cdk3 in glioblastoma cells (20–22), and our results confirmed that cdk3 is highly expressed in glioblastoma T98G cells. Thus, we then examined cdk3 protein level in a human glioblastoma cancer tissue array, which contained duplicate core samples of 35 cases of glioblastoma, 3 cases of nonneoplastic normal cerebrum tissues, and 2 cases of normal cerebrum tissues adjacent to cancer tissue (US Biomax, Inc.). We found that cdk3 expression was significantly higher in glioblastoma tissues, with a median expression level of 12.09, compared with expression in nontumor tissues, which had a median score of 9.47 (Fig. 1B). These results showed that high expression of cdk3 is associated with human glioblastoma.

ATF1 is a novel binding partner of cdk3. Cdk3 is a serine/threonine protein kinase that phosphorylates a number of substrates, including pRb (20) and Cables (38, 39). We hypothesized that cdk3 might phosphorylate additional substrates, including transcription factor(s). To examine this idea, we conducted the mammalian two-hybrid assay using cdk3 (pBIND-cdk3) as bait for various candidate transcription factors (pACT-TFs; ref. 34) to identify new binding partners of cdk3. Each individual pACT-TF and the pG5-luciferase reporter plasmid were cotransfected with pBIND-cdk3 into HEK293 cells. Each interaction activity was compared against the basal level (value of 1) of pG5-luc/pBIND-cdk3/pACT (Fig. 2A^,, top, lane 1). E2F1 and E2F2 (19) were used as positive controls (Fig. 2A^,, top, lanes 2 and 3). ATF1 showed a >15-fold increase in interacting activity with pBIND-cdk3 compared with the pG5-luc/pBIND-cdk3/pACT control (Fig. 2A^,, top, lane 4 versus lane 1). To determine whether the interaction of ATF1 with cdk3 occurs endogenously, we used T98G cells to perform immunoprecipitation experiments. The results showed that ATF1 binds with cdk3 ex vivo (Fig. 2A^,, bottom). Then to examine whether cdk3 phosphorylates ATF1, a GST-ATF1 fusion protein was constructed, expressed, partially purified as described in Materials and Methods, and then subjected to a kinase assay with active cdk3. The results showed that cdk3 phosphorylated ATF1 in vitro (Fig. 2B^,, lane 6). To determine the specific sites of ATF1 that are phosphorylated by cdk3, we used the human protein reference database²

to compare the amino acid similarity among known cdk3 substrates. Results showed that the putative consensus motif for cdk3 is the SP amino acid motif. We then examined the amino acid sequence of ATF1 and found that serine 63 was a potential candidate for phosphorylation by cdk3 (Fig. 3C,, top). For confirmation, we expressed a mutant ATF1 that harbored a point mutation at serine 63 (S63A), performed an in vitro kinase assay, and detected phosphorylation of ATF1 by Western blotting with an ATF1 serine 63 specific antibody (Fig. 2C,, bottom). The results indicated that cdk3 phosphorylates wild-type ATF1 at serine 63. When we performed the in vitro kinase assay with [γ-³²P]ATP, the phospho-band intensity of the ATF1S63A mutant was decreased, but was still visible (Supplementary Fig. S1A). Furthermore, in an in vitro kinase assay using deletion mutants of GST-ATF1 proteins (1–57, 1–141, or 142–271) and active cdk3, we found that another phosphorylation target amino acid exists in the NH₂-terminal region of ATF1 spanning amino acids 1 to 57 (Supplementary Fig. S1B). We further confirmed that endogenous cdk3 phosphorylated ATF1 at serine 63 (Fig. 2D) using an immunoprecipitation kinase assay and T98G cells. Overall, these results indicated that ATF1 is a novel substrate of cdk3, which phosphorylates ATF1 at serine 63.

Cdk3 enhances the transactivation and transcriptional activities of ATF1. Our studies, thus far, have shown that cdk3 phosphorylates ATF1 at serine 63. To examine the biological significance of cdk3-mediated ATF1 phosphorylation, we constructed Gal4-ATF1 and Gal4-ATF1 S63A fusion expression vectors (Fig. 3A,, left) and analyzed ATF1 transactivation activity as described previously (40). Our results indicated that ATF1 transactivation activity was increased dose-dependently by cotransfection of increasing amounts of pCMV-cdk3 (Fig. 3A,, middle). However, the mutant Gal4-ATF1 S63A did not exhibit any transactivation activity regardless of the amount of cdk3 present (Fig. 3A,, middle). On the other hand, we found that cdk3 bound equally well with the mutant ATF1 (S63A) or wild-type ATF1 (Fig. 3A , right). These results showed that cdk3 positively regulated ATF1 transactivation activity through phosphorylation at serine 63.

Because c-Jun and c-Fos expression are regulated by ATF1, we examined the cdk3-mediated transcriptional activity of ATF1 using cells transfected with either a luciferase linked c-jun (c-jun-luc) or c-fos (c-fos-luc) promoter (41, 42). Results indicated that c-Jun-luciferase activity was increased dose-dependently by cotransfection with increasing amounts of ATF1 (Fig. 3B,, left) and even more by increasing amounts of cdk3 (Fig. 3B,, right). Similar results were obtained for c-Fos (Fig. 3C). These results indicated that cdk3-mediated ATF1 phosphorylation at serine 63 is critical for ATF1 transactivation and transcriptional activities.

Knockdown of cdk3 suppresses the proliferation and colony formation of T98G cells. To examine whether knockdown of cdk3 affects proliferation and colony formation in soft agar, we used glioblastoma T98G cells because of the high expression of cdk3 in these cells and human glioblastoma tissues (Fig. 1). To knock down cdk3 expression, we designed siRNA against cdk3 (si-cdk3; Supplementary Fig. S2) and introduced it into T98G cells. Stable cells were established and knockdown efficiency was examined by Western blotting. The results revealed that si-cdk3 suppressed the endogenous cdk3 protein level by 60% compared with si-mock control T98G cells (Fig. 4A,, bottom left). Using these cells and the MTS assay, we estimated the proliferation, and the si-cdk3 stably transfected cells showed significantly suppressed proliferation compared with the si-mock T98G control cells (Fig. 4A,, top left). Further, si-cdk3 T98G cells displayed an ∼50% reduction in EGF-induced anchorage-independent colony formation in soft agar compared with mock cells (Fig. 4A,, right). To determine whether reduction of proliferation and growth in soft agar in si-cdk3 cells is associated with ATF1 activity, we analyzed ATF1 transactivation and transcriptional activities using a pGl4.29/Luc2p/CRE reporter plasmid, which contains an ATF1 binding consensus sequence in the CRE motif. The results indicated that Gal4-ATF1 transactivation activity was significantly inhibited by transfection of cells with si-cdk3 (Fig. 4B,, top). In addition, pGl4.29/Luc2p/CRE luciferase activity was also dose-dependently suppressed by introduction of si-cdk3 (Fig. 4B,, bottom). These results indicated that knockdown of cdk3 by si-cdk3 inhibited T98G proliferation and colony formation in soft agar through inhibition of ATF1 activity. To further verify whether EGF-induced ATF1 phosphorylation is mediated through cdk3, we analyzed the ATF1 phosphorylation level in si-mock and si-cdk3 stably transfected T98G cells by Western blotting. The results indicated that phosphorylation of ATF1 at serine 63 was suppressed in si-cdk3 stably transfected cells without an effect on total ATF1 protein levels (Fig. 4C,, top left). By Western blotting, we found that EGF-induced ATF1 phosphorylation at serine 63 was increased at 15 minutes, maintained to 30 minutes, and decreased at 60 minutes in si-mock control T98G cells. However, phosphorylation of ATF1 at serine 63 was minimal in si-cdk3 stably transfected cells (Fig. 4C,, bottom left). Furthermore, EGF-induced transactivation activity of ATF1 was also suppressed in si-cdk3 stably transfected cells compared with si-mock cells (Fig. 4C,, right), indicating that the diminished proliferation and colony formation in soft agar displayed by si-cdk3 cells occurs through a suppression of EGF-induced ATF1 activity. To further confirm a role for cdk3 in proliferation and colony formation, we constructed a mutant cdk3 expression plasmid (rescue cdk3 vector), which was not affected by si-cdk3, and then performed rescue experiments by returning the rescue cdk3 vector to si-cdk3 stably transfected T98G cells (Fig. 4D). At 24 hours after transfection of the rescue cdk3 vector, cdk3 expression was confirmed by Western blotting (Fig. 4D,, bottom left) in these cells, and proliferation and colony formation in soft agar were assayed. The results indicated that return of the rescue cdk3 vector reversed the si-cdk3–induced reduction in cell proliferation (Fig. 4D,, top left compared with Fig. 4A,, top left), as well as reversed the si-cdk3–induced decrease in colony formation in soft agar (Fig. 4D,, right compared with Fig. 4A , right).

The cdk3-ATF1 signaling axis plays an important role in cell transformation. Next we examined whether the cdk3-ATF1 signaling axis plays an important role in cell transformation by establishing stable JB6 Cl41 cells that express high levels of cdk3. High expression of the cdk3 protein was confirmed in these cells by Western blotting (Fig. 5A). The ATF1 protein level was not changed. Next, we examined the effect of ectopic expression of cdk3 on EGF-induced colony formation in soft agar. The results showed that, compared with mock control cells, cdk3-expressing JB6 Cl41 cells displayed an ∼8-fold increase in EGF-induced colony formation (Fig. 5B). We also analyzed the EGF-induced transactivation activity of ATF1 and the results indicated that ATF1 transactivation activity was significantly increased in cdk3-expressing stable cells compared with mock cells (Fig. 5C). Importantly, we further found that Ras^G12V induced cell transformation (Fig. 5D,, middle, lane 4), and overexpression of Ras^G12V/cdk3 or Ras^G12V/ATF1 induced more foci formation in NIH3T3 cells (Fig. 5D,, middle, lanes 5 and 6). Moreover, cotransfection of Ras^G12V/cdk3/ATF1 induced even more foci formation (Fig. 5D,, middle, lane 8). However, cotransfection of si-cdk3 with Ras^G12V/cdk3/ATF1 blocked foci formation in NIH3T3 cells (Fig. 5D,, middle, lane 9) compared with Ras^G12V/cdk3/ATF1. In addition, cotransfection of Ras^G12V/cdk3/ATF1 S63A blocked foci formation to the same level of Ras^G12V/cdk3/ATF1/si-cdk3 compared with Ras^G12V/cdk3/ATF1 (Fig. 5D , middle, lane 10). These results indicated that cdk3-ATF1 signaling plays an important role in cell transformation.

Activation of cdks induced by overexpression of activator cyclins; inactivation of cdk inhibitors such as INK4, Cip, and Kip; or inactivation of substrates such as pRb has been observed in many human cancers (1, 4). Thus, cdks are important regulators in cell cycle progression and have long been regarded as targets for human cancer therapeutic intervention (43). Previous studies showed that cdk3 was expressed in some human tissues and cell lines (20–22). In the present study, we found that cdk3 is overexpressed in glioblastoma tissues and in numerous human cancer cell lines, including T98G, H460, HCT116, HT29, PC-3, DU-145, Saos-2, A431, SK-MEI-28, and RPMI-1951, compared with the HaCaT cell line, which is an immortalized nonmalignant human keratinocyte cell line (Fig. 1); knockdown of cdk3 suppressed proliferation (Fig. 4A) and colony formation of glioblastoma T98G cells in soft agar (Fig. 4A). Furthermore, knockdown of cdk3 suppressed foci formation induced by Ras^G12V/cdk3/ATF1 in NIH3T3 cells (Fig. 5D). These results indicated that the cdk3-ATF1 signaling axis plays an important role in cell transformation.

The well-known substrates of cdk3 include the tumor suppressor pRb and Cables/IK3-1. Cdk3 phosphorylates pRb at serine 807/811 during the G₀-G₁ transition, which is required for cells to exit G₀ efficiently (20). Cdk3 phosphorylates Cables/IK3-1 at serine 274 (38). Cables/IK3-1 can bind with cdk5 and c-Abl and induces cdk5 phosphorylation at tyrosine 15 by Abl, resulting in increased kinase activity of the p35/cdk5 complex and neurite outgrowth (44). In HeLa cells, Cables enhances cdk2 tyrosine 15 phosphorylation by Weel, leading to decreased cdk2 kinase activity and inhibition of cell growth (45). Furthermore, loss of Cables expression has been found in 50% to 60% of primary colon and head and neck cancer samples (45). However, the biological significance of Cables/IK3-1 phosphorylation by cdk3 is not clear. Here, we found that cdk3 phosphorylates ATF1 at serine 63 (Fig. 2).

The expression of ATF1 is up-regulated in lymphomas (31) and melanoma cells (32) and contributes to the growth of these cancers. The EWS-ATF1 fusion oncoprotein is typically associated with clear cell sarcoma (46) because of enhanced target gene expression induced by constitutively activated ATF1 in the absence of stimuli (33). In our study, we found that the transactivation and transcriptional activities of ATF1 were increased by cdk3 in a dose-dependent manner (Fig. 3). Furthermore, our study showed that knockdown of cdk3 suppressed cell transformation and ATF1 transactivation and transcriptional activities (Fig. 4B and C). In addition, c-jun and c-fos promoter activity was increased by ATF1 and cdk3 dose-dependently (Fig. 3B and C). ATF1 belongs to the CREB transcription factor family and influences cellular physiologic processes by regulating the expression of downstream target genes, which are related to growth, survival, and other cellular activities (24). Our data indicated that c-jun and c-fos might play an important role in cell proliferation and cell transformation induced by cdk3-ATF1 signaling.

Various stimuli, including growth factors, stress signals, neurotransmitters, and other agents, induce activation of CREB family members (24). The transactivation domain of ATF1 consists of a KID domain and a glutamine-rich constitutive activation domain (Q2). Our finding that cdk3 phosphorylates ATF1 serine 63 in the KID domain is important because this phosphorylation promotes the recruitment of CREB-binding protein/p300, a core protein in the transcriptional machinery, leading to enhancement of transactivation and transcriptional activities (30). Furthermore, ATF1 can be phosphorylated at serine 63 by PKA (47) and calmodulin-dependent protein kinase I/II (47, 48). In response to EGF or 12-O-tetradecanoylphorbol-13-acetate stimulation, ATF1 can also be phosphorylated by mitogen- and stress-activated protein kinase (MSK)-1/2 (41, 49, 50), but in MSK1/MSK2 double knockout embryonic fibroblasts, mitogen-induced phosphorylation of ATF1 at serine 63 is reduced but not totally abolished (50), suggesting phosphorylation by additional kinases. We found that ectopic expression of cdk3 in JB6 Cl41 cells stimulated EGF-induced cell transformation and ATF1 activation (Fig. 5B and C). Moreover, coexpression of cdk3 and Ras^G12V, a downstream kinase of the EGF receptor (EGFR) signaling pathway, induced cell transformation through cdk3-ATF1 signaling in NIH3T3 cells (Fig. 5D), indicating that cdk3 may participate as a downstream kinase of the EGFR-Ras pathway. In summary, we found that ATF1 is a novel substrate of cdk3 and that the cdk3-ATF1 signaling axis plays an important role in cell proliferation and cell transformation.

No potential conflicts of interest were disclosed.

Grant support: The Hormel Foundation and NIH grants CA77646, CA27502, and CA111356.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. Barrett J. Rollins for the pRcCMV-cdk3 (pCMV-cdk3); Dr. Ron Prywes for pET28-ATF1, pET28-ATF1S63A, p3×Flag-ATF1, and p3×Flag-ATF1S63A mutant; Dr. Hiroshi Handa for the deletion GST-ATF1 fusion vectors (1–57, 1–141, and 142–271); Dr. David L. Turner for the pU6pro vector; and Andria Hansen for secretarial assistance.

The University of Minnesota is an equal opportunity educator and employer.