Tumor suppressor p16^INK4a inhibits cancer cell growth by downregulating eEF1A2 through a direct interaction

Malignant progression of transformed cells involves inappropriate cell division. Eukaryotic cell division occurs through a tightly regulated cell cycle, which comprises four different phases: gap I (G1), synthesis (S), gap 2 (G2) and mitosis (M) phases. Two cell-cycle checkpoints maintain genetic stability by ensuring an authentic chromosome replication and separation (Swanton, 2004). Failure of these checkpoints is a hallmark of cancer (Hartwell et al., 1994; Hartwell and Kastan, 1994). One of the checkpoints, the restriction check-point, occurs in mid-G1, after which cells become independent of growth factors and commit to cell division (Pardee, 1974). Intracellular signaling pathways comprising various cyclins, cyclin-dependent kinases (Cdks), Cdk inhibitors, and check-point kinases (Chk1 and Chk2) are critical regulators of cell cycle progression (Collins and Garrett, 2005; Gali-Muhtasib and Bakkar, 2002). The tumor suppressor protein p16^INK4a causes cell cycle arrest and inhibits tumor cell proliferation, at least partially, by acting as a specific inhibitor of Cdk4 and Cdk6 (Serrano et al., 1993). The regulation of the CDK4/cyclin D complex by p16^INK4a occurs in the nucleus (Ruas and Peters, 1998; Sherr and Roberts, 1999). However, subcellular localization of p16^INK4a has been documented in both the cytoplasm and nucleus (Evangelou et al., 2004). Choi and colleagues demonstrated that cytoplasmic p16^INK4a suppressed UV-induced cell transformation by interacting with c-Jun N-terminal kinase (JNK)-1 (Choi et al., 2005). Thus, cytoplasmic p16^INK4a may play other functional roles in cells.

Recent studies demonstrated that deregulation of translational machinery is associated with malignant transformation of cells (Bilanges and Stokoe, 2007; Ruggero et al., 2004; Ruggero and Pandolfi, 2003). Several eukaryotic translation initiation (eIF) and elongation (eEF) factors have been implicated in carcinogenesis (Sonenberg and Hinnebusch, 2009; White-Gilbertson et al., 2009). Eukaryotic elongation factor 1 alpha (eEF1A), a member of the G protein family, is one of four subunits that constitute the eukaryotic elongation factor 1 (Browne and Proud, 2002; Ejiri, 2002) and catalyzes the binding of aminoacyl-tRNA to the A-site of the ribosome in a GTP-dependent manner (Ejiri, 2002; Moldave, 1985). Although the biological roles of mammalian eEF1A, which exists as two isoforms, eEF1A1 and eEF1A2, have been extensively investigated, the necessity of the existence of two isoforms of eEF1A still remains unexplained. Despite sharing 98% amino acid homology and playing a similar role in protein synthesis, these two isoforms differ in their chromosomal location (EEF1A1 in 6q14 and EEF1A2 in 20q13) (Lund et al., 1996) and their binding affinity for GDP and GTP (Kahns et al., 1998). While eEF1A2 exhibits a stronger affinity for GDP than GTP, eEF1A1 displays the opposite preference. The two isoforms of eEF1A also differ in their tissue distribution. The eEF1A1 isoform is expressed in almost all tissues, whereas eEF1A2 is present only in those tissues, such as brain, heart and skeletal muscle, that are composed of cells locked in a state of nonproliferation (Khalyfa et al., 2001; Knudsen et al., 1993; Lee et al., 1992; Lee et al., 1995; Lee et al., 1993). Because eEF1A2 is expressed specifically in certain cells, such as neurons, cardiomyocytes and myocytes, that have permanently deviated from the cell cycle (Kahns et al., 1998; Knudsen et al., 1993; Lee et al., 1994), eEF1A2 is suggested to be involved in pathways of protein synthesis that are preferentially for nonproliferating cells or play non-translational roles. Beyond its defined role in the protein translation process, eEF1A2 has been reported to be involved in other non-translational functions of cells and has been identified as a putative oncogene (Amiri et al., 2007; Anand et al., 2002; Kulkarni et al., 2007; Lee and Surh, 2009; Li et al., 2010; Tomlinson et al., 2005). Anand and colleagues reported that eEF1A2 is overexpressed in ∼30% of ovarian tumors (Anand et al., 2002). The ectopic overexpression of eEF1A2 provides mouse fibroblasts (NIH3T3) with oncogenic potential and accelerates the growth of ovarian carcinoma (ES-2) cells as xenografts in nude mice. In addition, the results of tissue microarray for the assessment of eEF1A2 protein abundance in 500 primary ovarian tumors showed high expression levels in about 30% of all primary ovarian tumors (Pinke et al., 2008). In the present study, we demonstrate a novel mechanism whereby p16^INK4a interacts with eEF1A2 and exerts its tumor suppressor function by downregulating eEF1A2 in ovarian cancer cells.

In order to identify proteins that interact with p16^INK4a, we performed a yeast two-hybrid screening of a human HeLa cDNA-GAL4 expression plasmid library using a full-length p16^INK4a fused to the GAL4 DNA-binding domain as the ‘bait’. We selected the clone encoding eEF1A2 (supplementary material Fig. S1) for further investigation. The full-length eEF1A2 clone consists of 1392 nucleotides, encoding a protein of 463 amino acids. The predicted molecular mass of the protein is 53 kDa. For further study, we cloned the eEF1A2 gene, which was extracted from total mRNA of HeLa cells, into mammalian vectors.

To confirm the interaction between p16^INK4a and eEF1A2 in a cell system, we performed the mammalian two-hybrid assay using COS-7 cells transfected with the GAL4 binding domain containing pM-BD-p16 and VP16 containing pVP-eEF1A2. These results indicated that GAL4-p16^INK4a interacted with VP16-eEF1A2 in cells (Fig. 1A). The interaction between p16^INK4a and eEF1A2 was further confirmed by a co-immunoprecipitation assay. Plasmids encoding eEF1A2 fused to N-terminal V5 were transiently co-transfected with pcDNAp-16^INK4a into COS-7 cells. Immunoprecipitation of V5-eEF1A2 resulted in the co-precipitation of p16^INK4a (Fig. 1B). To demonstrate a direct interaction between p16^INK4a and eEF1A2, we performed a GST pull-down assay using a whole lysate and GST-conjugated- or GST–p16-conjugated Sepharose 4B beads. Immunoblot analysis revealed that p16^INK4a interacted with serum-induced eEF1A2 (Fig. 1C). Serum-induced CDK4 was used as a positive control to show the p16^INK4a interaction. Furthermore, endogenous p16^INK4a was also co-immunoprecipitated by eEF1A2 in human ovarian cancer (PA-1) cells (Fig. 1D). These results indicate that p16^INK4a physically interacts with eEF1A2 in mammalian cells.

To identify the interacting domains of p16^INK4a and eEF1A2, various deletion constructs of p16^INK4a and eEF1A2 were designed. Plasmids encoding the corresponding fragments of eEF1A2 were constructed and used to identify the p16^INK4a-binding domain of eEF1A2. The GST fusion full-length p16 protein, p16 (1–80 aa) and p16 (77–156 aa) interacted with eEF1A2, but not with eEF1A1 (Fig. 2A; supplementary material Fig. S2). In addition, each eEF1A2 fragment was produced by transfection in 293 cells, and the cell lysates were pulled down using the Sepharose-4B-bead-conjugated GST fusion protein, p16^INK4a. Results of the GST pull-down assay showed that p16^INK4a interacted with full-length eEF1A2, and FLAG-tagged eEF1A2 (1–249 aa) or eEF1A2 (328–463 aa) proteins. However, FLAG-tagged eEF1A2 (250–327 aa) did not bind to p16^INK4a (Fig. 2B). The binding between p16^INK4a and eEF1A2 was also predicted by the Fast Fourier transform-based docking algorithm ZDOCK. The model of the p16^INK4a–eEF1A2 complex obtained from the protein–protein docking experiment suggested that p16^INK4a bound to eEF1A2 (1–249 aa) and eEF1A2 (328–463 aa), but not to eEF1A2 (250–327 aa) (Fig. 2C).

In the predicted p16^INK4a/eEF1A2 complex, the hydrogen-bond networks between β-strands in domain III of eEF1A2 and the first, second and third ankyrin repeats of p16^INK4a are expected to be crucial in the binding of p16^INK4a with eEF1A2. Arg24 in p16^INK4a forms two hydrogen bonds with Glu374 and one hydrogen bond with Glu403 in eEF1A2 (supplementary material Table S1; Fig. S3A,B). In addition, the hydrogen bonds between the residues from the fourth repeat of p16^INK4a and those from domain I of eEF1A2 would also contribute to the stability of the p16^INK4a–eEF1A2 complex (supplementary material Table S1; Fig. S3C,D). Arg131 in p16^INK4a forms two hydrogen bonds with Glu217 in eEF1A2. These hydrogen bonds would explain the in vitro and ex vivo experimental findings regarding the p16^INK4a interaction with eEF1A2 (328–463 aa) and eEF1A2 (1–249 aa). To confirm this idea, we constructed a p16^INK4a protein with mutations at Arg24 and Arg131. Double mutation of Arg24 (R24A) and Arg131 (R131E) disrupted the interaction of p16^INK4a with eEF1A2 as expected (Fig. 2D). This result indicated that Arg24 and Arg131 of p16^INK4a are important for interaction with eEF1A2 domain I/III.

We then examined the functional significance of the interaction between p16^INK4a and eEF1A2. Because eEF1A2 functions as an elongation factor in peptide chain elongation during protein synthesis, the antiproliferative protein p16^INK4a might attenuate the translational activity of eEF1A2. To explore this possibility, in vitro protein synthesis was assessed in the presence of varying concentrations (0, 100, 200, 400 and 800 nM) of recombinant p16^INK4a. The in vitro transcription and translation experiments were performed using luciferase cDNA as the reporter gene and transcription and translation were detected by the luciferase activity assay and autoradiograpy. Results of the in vitro translation assay showed a significant decrease in protein synthesis with the addition of recombinant p16^INK4a (400 or 800 nM) (Fig. 3A,B). Analogous experiments performed with the same concentrations of GST as a negative control, did not show noticeable changes in the synthesis of ³⁵S-luciferase (Fig. 3A,B).

The micro-injection of mRNAs into Xenopus embryos constitutes a very sensitive method for the identification of mRNAs and the study of protein translation mechanisms (Richter et al., 1982). Therefore, we performed an in vivo Xenopus translation assay to explore the effect of p16^INK4a on eEF1A2 translational activity. When Xenopus embryos were injected with mRNAs of luciferase alone or with luciferase plus p16^INK4a, the luciferase translational activity was significantly decreased in the presence of p16^INK4a, suggesting that the expression of p16^INK4a inhibits the translation of luciferase mRNA (Fig. 3C). Moreover, the expression of eEF1A2 was decreased with injection of luciferase (1 ng) plus p16^INK4a (500 pg) compared to embryos injected with luciferase alone or left untreated (Fig. 3D). We optimized a morpholino-based eEF1A2 siRNA (MO-eEF1A2) by injecting into Xenopus embryos at different concentrations. Western blot results showed that MO-eEF1A2, but not MO-control, specifically inhibited the translation of endogenous eEF1A2. Results indicated that the expression of eEF1A2 was decreased by MO-eEF1A2 (5 or 10 µM; Fig. 3E). Injection of MO-eEF1A2 (5 µM) in Xenopus decreased the luciferase activity (Fig. 3F). To clarify the effect of p16^INK4a on eEF1A2 translational activity, Xenopus embryos were separately injected with luciferase (500 pg) alone, luciferase plus p16^INK4a (100 pg), luciferase plus p16^INK4a plus MO-eEF1A2 (2.5 µM), or luciferase plus MO-eEF1A2. The luciferase plus MO-eEF1A2 group, but not the MO-control (data not shown), showed moderate inhibition of the luciferase activity and this inhibition was further decreased in the presence of p16^INK4a (Fig. 3G). These data suggest that p16^INK4a inhibited the translational activity of eEF1A2 in vitro and in vivo.

The finding that p16^INK4a modulates the expression of eEF1A2 was confirmed by using the stably transfected p16^INK4a Tet-off system in HeLa cells (Fig. 4A). After removal of doxycycline, p16^INK4a expression was induced, whereas the expression level of eEF1A2 was decreased in a time-dependent manner. Moreover, co-transfection of CHO-KI cells with varying amounts of p16^INK4a and a unique concentration of eEF1A2 substantially reduced the expression of eEF1A2 with increasing concentrations of p16^INK4a (Fig. 4B). Furthermore, when eEF1A2-overexpressing and p16^INK4a-deficient SKOV3 cells were transfected with pcDNA-p16^INK4a, the level of eEF1A2 was decreased with increasing amounts of p16^INK4a (Fig. 4C). To rule out the involvement of the Rb pathway, Rb^−/− MEFs were co-transfected with V5-eEF1A2 and pcDNA-p16^INK4a. Immunoblot analysis showed that the expression of eEF1A2 was decreased with increasing concentrations of p16^INK4a (Fig. 4D). Furthermore, decreased eEF1A2 expression level was overcome by additional transfection of eEF1A2 (Fig. 4E). To identify the effect of mutant R24A/R131E p16^INK4a on expression of eEF1A2, we transfected SKOV3 cells with wild-type p16^INK4a or mutant R24A/R131E p16^INK4a. Wild-type p16^INK4a decreased expression of eEF1A2. However, the mutant R24A/R131E p16^INK4a had no effect on the expression level of eEF1A2 (Fig. 4F).

We examined the physiological significance of the p16^INK4a interaction with eEF1A2 and subsequent inhibition of the expression and translational activity of eEF1A2. When PA-1 cells, which have constitutively high expression of eEF1A2, were transfected with pcDNA-p16^INK4a (0, 50, 100 or 200 ng), their growth was significantly retarded by the presence of p16^INK4a (100 or 200 ng; Fig. 5A). In addition, PA-1 cells stably transfected with pcDNA-p16^INK4a showed significantly decreased colony formation compared to mock-transfected cells (Fig. 5B). We also generated stable cell lines (PA-1, SKOV3 and OVCAR8) transfected with short hairpin RNA against eEF1A2 (sh-eEF1A2) and examined the anchorage-independent colony formation ability of these cells. Reduced colony formation was observed in PA-1 (Fig. 5C), SKOV3 (Fig. 5D) and OVCAR8 (Fig. 5E) cells harboring sh-eEF1A2 compared to cells transfected with control sh-RNA (sh-c). Moreover, cell cycle analysis revealed that sh-eEF1A2-transfected PA-1 cells underwent G1 phase cell cycle arrest (Fig. 5F), whereas sh-eEF1A2-transfected SKOV3 (Fig. 5G) cells exhibited G2/M phase cell cycle arrest. In contrast, cell cycle was not affected in sh-eEF1A2-transfected OVCAR8 cells (Fig. 5H). Different phases of cell cycle arrest might be due to the different characteristics of each cell line. For example, PA-1 cells express Rb, p16^INK4a and eEF1A2. SKOV3 cells express Rb and eEF1A2, but are deficient in p16^INK4a. OVCAR8 cells express p16^INK4a and eEF1A2, but are deficient in Rb. This suggests that p16^INK4a can inhibit cell growth by downregulation of eEF1A2 in a cell-cycle-independent manner.

In this study, we showed that the tumor suppressor protein, p16^INK4a, interacted with a novel binding partner, eukaryotic translation elongation factor, eEF1A2, and the binding caused a downregulation of cancer cell growth by suppressing the expression and subsequent translational activity of eEF1A2. Tumor suppressor p16^INK4a is a well-known cell cycle inhibitor of the Cdk4 or Cdk6/cyclin D complexes in the early check-point of the G1 phase. A dysfunctional p16^INK4a protein results in the deregulation of the Rb and p53 tumor-suppressor signaling pathways in a wide variety of human cancers (Caldas et al., 1994; Krimpenfort et al., 2001; Sherr, 2001). The eEF1A1 mRNA and protein are expressed ubiquitously in many cancer cell types (Joseph et al., 2004; Scaggiante et al., 2012) (supplementary material Fig. S4). An anchorage-independent cell growth assay was performed to compare the oncogenic properties of eEF1A1 and eEF1A2. Results indicated that eEF1A2 expression induced more colony formation compared to eEF1A1 expression (supplementary material Fig. S5). The expression of eEF1A2 appears to be inversely related to the expression of p16^INK4a but not associated with p15^INK4b or p21 expression. Thus, p16^INK4a deficiency seems to be associated with an increased expression level of eEF1A2 in cancer cells. We observed interaction of p16^INK4a and eEF1A2 in vitro and ex vivo (Fig. 1). These results strongly indicate that p16^INK4a regulates eEF1A2 expression through a direct interaction. Although eEF1A1 and eEF1A2 exhibit a high sequence homology, the smaller binding cavity of eEF1A1 compared to eEF1A2 might prevent a strong binding interaction with p16^INK4a (Soares et al., 2009). In the three-dimensional structure of eEF1A2 (Fig. 2C), domain II is located behind domain I/III and therefore p16^INK4a interacts with domain I/III. Even though results indicate a low binding affinity between p16^INK4a and eEF1A2, this does not necessarily reflect activity (Marles et al., 2004). Our computational docking model (supplementary material Table S1) indicates that 12 hydrogen bonds are formed between p16^INK4a and eEF1A2. If each hydrogen bond cooperatively affects the activity of eEF1A2, the regulatory effect might become quite significant, as we showed. For the translation activity regulation of eEF1A2 by p16^INK4a, we used an in vitro rabbit reticulocyte translational system in a limited time frame (Fig. 3A,B). Because the sequence of human eEF1A2 coincides with rabbit eEF1A2 and p16^INK4a binds only to eEF1A2, the downregulatory translational activity of eEF1A2 was probably due to the inhibition by p16^INK4a in the time measured. On the other hand, whether all eEF1A isoforms correspond with eEF1A2 is not clear. However, the formation of a heterodimer of eEF1A1 and eEF1A2 has been reported (Sanges et al., 2012). This idea supports our data showing that selective regulation of eEF1A2 by p16^INK4a causes a decrease in the translational activity with the exclusion of a functional eEF1A1, which cannot act alone. If eEF1A2 and eEF1A1 work in the elongation phase of translation as a heterodimer, downregulation of eEF1A2 by p16^INK4a will affect the activity of the heterodimer. Therefore, translational activity could be decreased. To provide direct evidence supporting the regulation of the translational activity of eEF1A2 by p16^INK4a in vivo, we injected Xenopus embryos with combinations of p16^INK4a mRNA, luciferase mRNA, and MO-eEF1A2 and the results confirmed that p16^INKa regulated eEF1A2 activity (Fig. 3). Specifically, we injected one half of the amount of MO-eEF1A2 in Xenopus embryos (Fig. 3G). Thus, the rest of eEF1A2 activity was inhibited by p16^INK4a thereby decreasing the synthesis of luciferase. In contrast, p16^INK4a had no effect on the expression of eEF1A1 under the same conditions (Fig. 3D,E).

The exact mechanism explaining the decreased expression of eEF1A2 in the presence of p16^INK4a was not associated with ubiquitination and degradation (data not shown). On the other hand, the downregulation of the translational activity of eEF1A2 by p16^INK4a might be due to decreased expression of eEF1A2 (supplementary material Fig. S6). After transfection of p16^INK4a in SKOV3 and OVCAR8 cells, we examined the mRNA level of eEF1A2, eEF1A1 and p16^INK4a. The results showed that the mRNA level of eEF1A2 was decreased, but that of eEF1A1 was not changed (supplementary material Fig. S6). However, further studies, such as direct interruption of eEF1A2 and other interacting proteins, are needed to determine the specific mechanism for this relationship (Panasyuk et al., 2008) with p16^INK4a. Although many questions remain to be answered, the interaction between p16^INK4a and eEF1A2 does induce cell cycle arrest and inhibition of colony formation in various cancer cell types, which suggests a novel function of the tumor suppressor p16^INK4a.

MTT [3-(4,5-dimethyl-thiazolyl-2)-2,5-diphephenyl-tetrazolium bromide] was procured from Sigma (St. Louis, MO). Minimum essential medium (MEM), Dulbecco's modified Eagle's medium (DMEM), RPMI-1640 and fetal bovine serum (FBS) were purchased from Invitrogen (GIBCO, Grand Island, NY). Primary antibodies for p16^INK4a were from Pharmingen, Becton Dickinson Co. (Franklin Lakes, NJ). The eEF1A2 antibody was constructed by Takara Korea by changing the peptide sequence from mouse to human (Khalyfa et al., 2001). The eEF1A2 antibody does not detect eEF1A1 expression. The secondary horseradish peroxidase (HRP)-conjugated anti-rabbit and anti-mouse antibodies were purchased from Zymed Laboratory (San Francisco, CA).

pGBK-T7-p16 and pcDNA3.1-v5-HisA-eEF1A2 were generated by PCR using the pcDNA3.1-p16 and HeLa cDNA as templates. PCR fragments were purified and digested with EcoRI/BamHI and EcoRI/XbaI, respectively, and cloned into the EcoRI/BamHI and EcoRI/XbaI sites of pGBK-T7 (Clontech, Mountain View, CA) and pcDNA3.1-v5-HisA (Invitrogen, Carlsbad, CA). The pM-BD-p16 plasmid was constructed by releasing p16 from pGBK-T7 by EcoRI/BamHI digestion. The cDNA insert was subcloned into the EcoRI/BamHI site of the pM-BD vector (Clontech). The pVP16-AD-eEF1A2 was generated by PCR using pcDNA3.1-v5-HisA-eEF1A2 as a template. The PCR fragment was purified, digested with EcoRI/XbaI and cloned into the EcoRI/XbaI site of the pVP16-AD vector (Clontech). eEF1A2 (full-length), eEF1A2-D1 (1–249 aa), eEF1A2-D2 (250–327 aa), eEF1A2-D3 (328–463 aa), eEF1A2-D1+D2, eEF1A2-D2+D3 and eEF1A2-D1+D3 were subcloned into the pcDNA3-FLAG expression vector (Invitrogen) at the 5′ EcoRI and 3′ XbaI sites. p16^INK4a (1–158 aa), p16^INK4a (1–80 aa), and p16^INK4a (81–156 aa) were subcloned into the pGEX-4T-2 expression vector (Amersham, Piscataway, NJ) at the 5′ BamHI and 3′ XhoI sites. The R24A and R131E mutations were produced using the QuickChange Lightning site-directed mutagenesis kit (Stratagene, Santa Clara, CA). Various expression vectors were amplified in Escherichia coli XL1-blue bacteria and the plasmids were purified using the Jetstar midi kit (Genomed GmbH, Löhne, Germany). The DNA sequences of all plasmids were confirmed by sequencing (Dye Terminator ABI Type Seq., Bionex, Seoul, South Korea).

Saccharomyces cerevisiae strain AH109 was transformed with 0.1 µg of pGBK-T7-p16 as bait using a lithium acetate method and was plated on medium lacking –Trp DO supplement. The yeast strain AH109 (Clontech) containing pGBK-T7-p16 was mated with the pre-transformed human HeLa cDNA library in Y187 (Clontech) and plated on medium lacking −Leu/−Trp, −His/−Leu/−Trp, −Ade/−His/−Leu/−Trp DO supplements (Clontech). Colonies were detected with X-α-gal. To recover library plasmids, total DNA from Ade+/His+/Leu+/Trp+ colonies was isolated by the lyticase method and used to transform E. coli (XL-I blue strain from Stratagene, Santa Clara, CA). DNAs were identified through DNA sequencing (Dye Terminator ABI Type Seq., Bionex, Seoul, South Korea).

Human ovarian cancer (PA-1, SKOV3), human cervical cancer (HeLa), CHO-KI and COS-7 cells (Chinese hamster ovary and African green monkey kidney, respectively) were purchased from American Type Culture Collection (Manassas, VA). Rb^−/− murine embryonic fibroblasts (MEFs) and OVCAR8 cells were a kind gift from Dr Anton Berns (The Netherlands Cancer Institute, Amsterdam, The Netherlands) and Dr Shridhar Vijayalakshmi (Mayo Clinic College of Medicine, Rochester, MN), respectively. PA-1, OVCAR8 and HeLa cells were cultured in MEM containing penicillin (100 units/ml), streptomycin (100 µg/ml) sodium pyruvate (1 mM) and 10% FBS (GIBCO). SKOV3 cells were cultured in RPMI 1640 containing penicillin (100 units/ml), streptomycin (100 µg/ml), and 10% FBS. COS-7 and Rb^−/− MEFs were grown in DMEM containing penicillin (100 units/ml), streptomycin (100 µg/ml), and 10% FBS. Cells were maintained at 37°C in a humidified atmosphere of 95% air/5% CO₂. We transfected SKOV3 and COS-7 cells with the pcDNA3.1-p16^INK4a (4 µg) or pcDNA3.1-v5-hisA-eEF1A2 (4 µg) plasmid and jetPEI poly transfection reagent (84 µl; Polyplus-transfection SAS, Saint Quentin Yvelines, France) in 60-mm dishes to generate SKOV3 cells transiently expressing p16^INK4a or COS-7 cells expressing p16^INK4a and eEF1A2. For the mammalian two-hybrid assay, we seeded COS-7 cells in 24-well plates for 24 hours and then cells were transfected with pM-BD-p16^INK4a (0.4 µg), pVP16-AD-eEF1A2 (0.4 µg), pG5/luc reporter (0.4 µg) plasmids, and pRL-TK internal control reporter (50 ng) plasmid and jetPEI poly transfection reagent according to the manufacturer's instructions. Cells were harvested 24 hours after transfection and disrupted with 5× lysis buffer. Protein–protein interaction activity was determined by luciferase activity and was normalized to pRL-TK activity (Promega, Madison, WI).

Transfected PA-1 and COS-7 cells were harvested in NET-NL lysis buffer containing 50 mM Tris (pH 7.5), 5 mM EDTA, 150 mM NaCl, 1 mM DTT, 0.01% NP-40, 0.2 mM PMSF, and a mixture of protease inhibitors (Complete; Roche, Mannheim, Germany). Cell lysates (1000 µg) were clarified by centrifugation before incubation overnight at 4°C with a monoclonal antibody against p16^INK4 (4 µg; Pharmingen, Becton Dickinson Co., Franklin Lakes, NJ) in NET-NR [50 mM Tris (pH 7.5), 5 mM EDTA, 150 mM NaCl, 1 mM DTT, 0.01% NP-40, 2 µg/ml BSA, 0.2 mM PMSF, and a mixture of protease inhibitors (Complete, Roche)]. An aliquot of 50 µl pre-washed protein G–agarose (Roche; 50% slurry) was then added and the incubation continued for 2 h at 4°C. Immunoprecipitates were recovered by centrifugation, washed three times in NET-NW buffer (50 mM Tris pH 7.5, 5 mM EDTA, 150 mM NaCl, 1 mM DTT, 0.01% NP-40, and 0.2 mM PMSF) and resolved by western blotting.

Cells were disrupted on ice for 30 min in cell lysis buffer (20 mM Tris pH 7.5, 150 mM NaCl, 1 mM Na₂EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, 1 mM sodium vanadate, 1 µg/ml leupeptin and 1 mM phenylmethylsulfonyl fluoride (PMSF). After centrifugation at 20,817 g for 15 minutes, the supernatant fraction was harvested as the total cellular protein extract. The protein concentration was determined using the Bio-Rad protein assay reagent (Richmond, CA). The total cellular protein extracts were separated by SDS-PAGE and transferred to polyvinylidene fluoride (PVDF) membranes in 20 mM Tris-HCl (pH 8.0), containing 150 mM glycine and 20% (v/v) methanol. Membranes were blocked with 5% nonfat dry milk in 1× TBS containing 0.05% Tween 20 (TBS-T) and incubated with antibodies against p16, eEF1A2 or actin. Blots were washed three times in 1× TBS-T buffer, followed by the incubation with the appropriate HRP-linked IgG. The specific proteins in the blots were visualized using the enhanced chemiluminescence (ECL) detection kit (Amersham Pharmacia Biotech, Piscataway, NJ).

For expression of deletion mutant eEF1A2, the plasmids (pcDNA-FLAG-full-length eEF1A2, eEF1A2-D1 (1–249 aa), eEF1A2-D2 (250–327 aa), eEF1A2-D3 (328–463 aa), eEF1A2 (D1+D2), eEF1A2 (D2+D3), eEF1A2 (D1+D3) and mutant p16^INK4a (R24A, R131E, R24A/R131E) were transfected into 293 cells or pcDNA-full-length eEF1A1 and eEF1A2 were translated in vitro with L-[³⁵S]methionine using the TNT Quick coupled transcription/translation system (Promega). GST fusion proteins were collected on glutathione–Sepharose beads (Amersham Pharmacia Biotech) incubated at 4°C for 4 hours with 300 µg of cell lysate or 20 µl of translated proteins. The bound proteins were washed three times and boiled with 2.5× sample buffer for 3 minutes, centrifuged, and then the supernatant fraction was examined by 15% SDS-PAGE. The binding was detected by autoradiography or western blotting.

The docking structures of p16^INK4a and eEF1A2 were generated by the rigid-body global search algorithm and further refined by molecular dynamics simulations. Rigid-body docking was performed by the Fast Fourier Transform-based docking algorithm ZDOCK (Chen et al., 2003; Chen and Weng, 2003; Mintseris et al., 2007), where p16^INK4a and eEF1A2 were treated as solid objects. Subsequent energy-minimization and molecular dynamics simulations of the p16^INK4a/eEF1A2 complex were performed by using the Impref and Impact modules from the Schrödinger software package (Maestro v 9.0; Schrödinger; L.L.C.: New York, NY, 2010).

To determine growth of PA-1 cells, p16^INK4a-transfected cells were plated at a density of 5×10³ cells per well in 96-well plates for 24 hours prior to the assay. Cells were transfected with 0, 50, 100, 200 ng of pcDNA3.1-p16^INK4a and jetPEI poly transfection reagent to generate p16^INK4a transient expression. Cells were then incubated for 6, 12, 24, 48 or 72 hours at 37°C in a humidified atmosphere of 95% air/5% CO₂. At the various time points indicated, 10 µl of MTT (5 mg/ml) were added to each well. After 2 hours of incubation, 100 µl of DMSO were added and the optical density (O.D.) was read at 570 nm with a Microplate reader (Molecular Devices, Downingtown, PA). For the anchorage-independent colony formation assay, each cell line (8×10⁴) was suspended in 1 ml of 0.33% basal medium Eagle (BME) agar with 10% FBS and plated over a layer of solidified BME/10% FBS/0.5% agar (3 ml). Cultures were maintained in a 5% CO₂ incubator at 37°C for ∼7 days. Colonies were then counted using a microscope and the Image-Pro PLUS computer software program (v.6.2, Media Cybernetics, Bethesda, MD).

An aliquot (1 µg) of luciferase DNA (Promega) was added to 100 µl of a rabbit reticulocyte lysate in vitro translation reaction (Promega) in the presence of eEF1A2 or GST purified recombinant p16^INK4a protein at different concentrations (100, 200, 400 or 800 nM). The reaction mixture was incubated at 30°C for 2 hours. Luciferase protein activity was measured by adding substrate (Promega). The newly synthesized ³⁵S-labeled luciferase protein was analyzed by 12% SDS-PAGE and detected by autoradiography. Coomassie staining was performed to verify equal loading. To measure the translational activity, Xenopus laevis embryos were obtained by artificial fertilization. Vitelline membranes were removed by immersing embryos in a 2% cysteine solution (pH 8.0, Spectrum Chemical Mfg. Corp., New Brunswick, NJ). Embryos at the one- or two-cell stage were injected with mRNA and morpholinos as described in the figure legends and incubated for 7–8 hours. Luciferase activities were measured using a luciferase assay system according to manufacturer's instructions. Four or five embryos per group were pooled and homogenized in 15 µl of lysis buffer per embryo. All experiments were repeated at least three times using independent samples. All synthetic mRNAs used for microinjection were produced by in vitro transcription. Each of the cDNAs (luciferase and p16^INK4a) was linearized and used for the in vitro synthesis of capped mRNAs using an in vitro transcription kit (Ambion, Austin, TX) in accordance with the manufacturer's instructions. The morpholino antisense oligonucleotide directed against Xenopus eEF1A2 (MO-eEF1A2) was 5′-ATGTGTGTCTTCTCTTTCCCCATCC (GeneTools, Philomath, OR). Control MO (5′-CCTCTTACCTCAGTTACAATTTATA-3′; Gene Tools) was used as a toxicity control. Oligos were re-suspended in sterile water and injected into embryos. The embryo lysates were homogenized in lysis buffer (50 mM Tris pH 7.4, 150 mM NaCl, 1% NP-40, 0.25% sodium deoxycholate, 0.1% SDS, 50 mM NaF and 1 mM Na₃VO₄) containing of 1 mM PMSF, 15 mM β-glycerophosphate, 1× proteinase inhibitor cocktail (Calbiochem, Darmstadt, Germany) and then used for western blotting.

Values are expressed as means ± s.e.m. of at least three independent experiments. Statistical significance was determined by the Student's t-test and a P-value of less than 0.05 was considered to be statistically significant.